Handbook of Single-Cell Technologies [1st ed. 2022] 9811089523, 9789811089527

Citation preview

Tuhin Subhra Santra Fan-Gang Tseng Editors

Handbook of Single-Cell Technologies

Handbook of Single-Cell Technologies

Tuhin Subhra Santra • Fan-Gang Tseng Editors

Handbook of Single-Cell Technologies With 296 Figures and 38 Tables

Editors Tuhin Subhra Santra Department of Engineering Design Indian Institute of Technology Madras Chennai, India

Fan-Gang Tseng Department of Engineering and System Science National Tsing Hua University Hsinchu, Taiwan

ISBN 978-981-10-8952-7 ISBN 978-981-10-8953-4 (eBook) ISBN 978-981-10-8954-1 (print and electronic bundle) https://doi.org/10.1007/978-981-10-8953-4 © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd, part of Springer Nature. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Preface

The cell is the fundamental unit of biological organisms. It plays a significant role in coordinating with each other to perform systemic functions. Numerous bioanalytical techniques depend on the whole sample’s analysis while performing a study on bioentities existing in a very low concentration. Thus, the outcome of the study reflects merely an average. Though the approach is best suited for whole bio-samples (blood, saliva, urine, and other bodily fluids), where quantity is not a limiting factor, there is also increasing need for extreme small-quantity sample analysis on the single cell or subcellular level. Such needs in the case of limited number of cells, for example, circulating tumor cells (CTCs), early-stage embryos or rare cells, or even when differences among the cells in the cell population, are of interest. Despite the apparent synchrony in cellular systems, single cell analysis (SCA) is important due to its capability to reveal the environmental and developmental changes in the chemical content of individual cell. The fundamental principle of cell biology is cellular heterogeneity that arises from stochastic expression of genes, proteins, and metabolites. Thus, the natural cellular heterogeneity is manifested not only in the structure and composition of cells but also in their functionality. SCA plays an important role in the system biology, where the interactions of molecular components are studied at different molecular levels, from genome to cellular functions. Individual cell or cell organelle analysis can reveal an effect of different life conditions, surrounding environment on the genome, cell cycle, and also transcriptome, proteome, peptidome, metabolome, etc. In this respect, single-cell-omics promote system biology to investigate cellular heterogeneities and their reasons. Initial basic techniques mainly focus on single modalities, such as DNA sequence, RNA expression, or chromatin accessibility. These technologies have yielded transformative insights into cellular development and diversity. But the cellular segregation is driven by methodological convenience and limits the ability to derive a deep understanding of the relationship between biomolecules in single cell. To understand these interactions is the key to derive deep understanding of the cellular state and remains a challenge for the field of SCA. Moreover, the availability and scale of the data sets are rapidly growing, which requires new computational methods for normalization and joint analysis across samples, even for the presence of significant batch effects or interindividual variation. Approximately, 5 years ago, flow cytometry, patch-clamping electrophysiology methods, fluorescence in situ v

vi

Preface

hybridization, and enzyme-linked immunospot assays were among few single cell molecular analysis tools available. From a given cell, most of these methods could analyze only between 1 and 3 molecules, while multicolor flow cytometry has been successful in capturing approximately 12 cell surface protein markers. This scenario is rapidly changing. Recently, single-cell sequencing technologies have been mainly led by the recent advances observed in the field of molecular biology, microfluidics, and nanotechnology. Several new technologies have emerged for the comprehensive analysis of single molecules. Some single cell methods are capable to assay about >40 secreted proteins, elements of phosphoprotein signaling pathways, and large number of cell surface markers. Even single cells genome can be analyzed at high coverage or focused, whereas transcriptome at sparse coverage at moderate or high cell statistics. In the last two decades, due to the rapid development of sophisticated micro-/ nanofluidic devices, we now have Bio-MEMS, Lab on a Chip (LOC), and micro total analysis systems (μTAS) that enable more complex manipulations of chemicals and biological agents in fluidic environments. Microfluidics methods enable single cell or molecular analysis correlated with measurement of cellular functionality. These devices permit single cell analysis within custom environment, highly controlled, or even allow nondestructive cell analysis to identify cell of interest, for example, B cells producing specific antibodies to be harvested for further use. In situ RNA profiling via sequential hybridization and proteomic analysis via ion beam profiling are the two recent tissue staining methods. These techniques enable single cell analysis within fixed and intact tissues, with multiplexing level that significantly exceeds traditional immunohistochemical staining methods. Thus, generate new types of data and which has been integrated with new computational tools. With these novel devices, technology has become a pioneer in omics analysis and an integral part of medical biotechnology, such as diagnostics, prognostics, and cancer therapy. This book comprises eight parts broadly covering several aspects of single cell analysis using different technologies. Part one emphasizes in detail about single cell therapy and analysis using different physical methods such as optoporation or photoporation, mechanoporation, electroporation, and microinjection. Most of these techniques use micro-/nanofluidic devices to induce different physical energy such as optical, electrical, and mechanical stress. These energies can deform cellular membrane, create hydrophilic transient membrane pores, deliver exogenous biomolecules into cells, and perform different cellular analysis. Part two broadly covers micro-/nanofluidic devices design, fabrication, and their operation for cellular analysis. The devices not only perform single cell manipulation, separation, isolation, cultivation, and lysis, but also electrical, mechanical, and biochemical characterization and analysis. Part three covers different chemical methods for single cell analysis. The part covers in detail about liposome-mediated molecular delivery into cells, antibody discovery using single cell analysis, and antibody discovery for detection, diagnosis, and treatment of infectious diseases. Moreover, this part demonstrates high-throughput screening of antigen-specific antibody-secreting cells and secreted molecules from individual cells. This can offer a valid approach

Preface

vii

towards understanding and treatment of various diseases, disorders, and syndromes. Part four elaborates in detail about single cell genomics, proteomics, transcriptomics, and high-throughput transcriptome sequencing. Part five demonstrates single cell analysis in system biology and biocatalysis and covers the adult bovine intervertebral disc model system, which anatomically and histologically reflects the situation in human. The primary cell lineages were repeatedly isolated from the annulus fibrosus and the nucleus pulposus tissues of bovine intervertebral disc, and the isolation was typically heterogeneous in culture. Moreover, the part covers rapid cell process, with a focus on receptor signal transduction within the cell membrane. Also, it covers how large-scale single cell assays provide an efficient route for the identification of biocatalysts with novel or improved function. Part six broadly discusses single cell adhesion in cancer progression, single cell technologies for cancer therapy, analytical tool for single cancer cell analysis, and transmembrane receptor dynamics as biophysical markers for cancer cells analysis. Part seven briefly emphasizes flow cytometry-based high-throughput single cell electrical characterization and single cell cytometry for the application in biology and medicine. Part eight discusses spectrum analysis, methods, targets, imaging, and applications. The part broadly covers single cell electrophysiology, analytical techniques for single cell study in microbiology, microwave and mechanical resonators for sensing and sizing of single cells, molecular force spectroscopy to measure the physiological function of cell adhesion, mass spectrometry for single cell analysis, micro-tweezers and force microscopy techniques for single cell mechanobiological analysis, acoustic tweezers for single cell manipulation, and single cell pull down for characterization of protein complexes. We hope this book will be fascinating to the readers, especially undergraduate and graduate students, and it will be efficient for scientists in academic and industrial research who are performing various aspects of single cell analysis. Chennai, India Hsinchu, Taiwan November 2021

Tuhin Subhra Santra Fan-Gang Tseng Editors

Contents

Part I

....

1

1

Light-Induced Cellular Delivery and Analysis . . . . . . . . . . . . . . . . Ashwini Shinde, Srabani Kar, Moeto Nagai, Fan-Gang Tseng, and Tuhin Subhra Santra

3

2

Mechanoporation: Toward Single Cell Approaches . . . . . . . . . . . . Amogh Kumar, L. Mohan, Pallavi Shinde, Hwan-You Chang, Moeto Nagai, and Tuhin Subhra Santra

31

3

Single-Cell Electroporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mingde Zheng

61

4

Microinjection for Single-Cell Analysis and Therapy Muniesh Muthaiyan Shanmugam and Hima Manoj

..........

81

Part II

Physical Methods for Single Cell Therapy and Analysis

Fluidic System and Integration . . . . . . . . . . . . . . . . . . . . . . .

109

5

Single-Cell Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rohit Bhardwaj, Harsh Gupta, Gaurav Pandey, Sangjin Ryu, Takayuki Shibata, Tuhin Subhra Santra, and Moeto Nagai

111

6

Single Cell Manipulation Using Macro-scale Actuator Chia-Hung Dylan Tsai

.........

137

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nan Xiang and Zhonghua Ni

155

Digital Microfluidics for Single Cell Manipulation and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Long Pang, Jing Ding, and Shih-Kang Fan

185

8

9

Single-Cell Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shilpi Pandey, Ninad Mehendale, and Debjani Paul

207

ix

x

Contents

10

Technologies for Automated Single Cell Isolation Julian Riba, Stefan Zimmermann, and Peter Koltay

.............

235

11

Dual-Well Microfluidic Technique for Single Cell Isolation and Long-Term Clonal Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chuan-Feng Yeh, Hao-Chen Chang, and Chia-Hsien Hsu

263

12

Single-Cell Cultivation Utilizing Microfluidic Systems . . . . . . . . . . Dian Anggraini, Nobutoshi Ota, Yigang Shen, Yo Tanaka, Yoichiroh Hosokawa, Ming Li, and Yaxiaer Yalikun

13

Integrated Microwell Array Technologies for Single Cell Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jolien Breukers, Caroline Struyfs, Sara Horta, Karin Thevissen, Karen Vanhoorelbeke, Bruno P. A. Cammue, and Jeroen Lammertyn

287

311

14

Micro- and Nanopore Technologies for Single-Cell Analysis . . . . . Makusu Tsutsui, Takeshi Yanagida, Takashi Washio, and Tomoji Kawai

343

15

Technologies for Single-Cell Printing and Patterning . . . . . . . . . . Pranav Ambhorkar, Mahmoud Ahmed Sakr, Hitendra Kumar, and Keekyoung Kim

375

16

Microfluidic Device with Removable Electrodes for Single Cell Electrical Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . Muhammad Asraf Mansor and Mohd Ridzuan Ahmad

17

Microfluidic and Nanomaterial Approach for Virology . . . . . . . . . Reya Ganguly and Chang-Soo Lee

Part III

Chemical Methods for Single Cell Technology . . . . . . . . . .

18

Liposome-Mediated Material Transfer in Single Cells . . . . . . . . . . Mamiko Tsugane and Hiroaki Suzuki

19

The Art of Therapeutic Antibody Discovery: Finding Them One Cell at a Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Touyana Semenova, Richard Witas, Brianna L. Schroeder, Katherine Bohn, Alexandria Voigt, and Cuong Q. Nguyen

20

Screening of Antigen-Specific Antibody-Secreting Cells . . . . . . . . . Myat Noe Hsu, Zirui Matthew Tay, Weikang Nicholas Lin, and Shih-Chung Wei

21

Biochemical Analysis of Secreted Molecules by Individual Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . O. T. M. Bucheli, I. Sigvaldadóttir, and K. Eyer

397 411

433 435

449

471

495

Contents

Part IV

xi

Single Cell Omics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

519

22

Single Cell Genomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yusuke Yamamoto, Anna Sanchez Calle, and Takahiro Ochiya

521

23

Single-Cell Proteomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiangdong Xu and Shen Hu

539

24

Single Cell Pull-Down for Characterization of Protein Complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Michael Philippi, Zehao Li, Maniraj Bhagawati, and Changjiang You

563

25

Single-Cell Transcriptomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Marta Prieto-Vila, Yusuke Yamamoto, Ryou-u Takahashi, and Takahiro Ochiya

585

26

Single-Cell Transcriptome Sequencing Using Microfluidics . . . . . . Yu-Chih Chen, Seungwon Jung, Yehyun Choi, and Euisik Yoon

607

Part V Single Cell Analysis in Systems Biology and Biocatalysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

631

27

Single-Cell Phenotyping of Complex Heterogeneous Tissue . . . . . . Petra Kraus, Kangning Li, Darren Sipes, Lara Varden, Rachel Yerden, Althea Henderson, Shantanu Sur, and Thomas Lufkin

633

28

Record the Single Cell Signal Pathway . . . . . . . . . . . . . . . . . . . . . . Ya-Yu Chiang

651

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fabrice Gielen

30

Analytical Techniques for Single-Cell Studies in Microbiology . . . Evgeny Puchkov

Part VI

Single Cell Technologies in Cancer . . . . . . . . . . . . . . . . . . .

673 695

727

31

Single Cell Adhesion in Cancer Progression . . . . . . . . . . . . . . . . . . Privita Edwina Rayappan George Edwin and Saumendra Bajpai

729

32

Single-Cell Technologies for Cancer Therapy . . . . . . . . . . . . . . . . Geng-Ming Hu, Victor Daniel Lee, Hung-Yu Lin, Pu-Wei Mao, Hsin-Yi Liu, Jih-Hou Peh, and Chih-Wei Chen

767

33

Analytical Technology for Single-Cancer-Cell Analysis . . . . . . . . . Ching-Te Kuo and Hsinyu Lee

851

xii

34

Contents

Transmembrane Receptor Dynamics as Biophysical Markers for Assessing Cancer Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mirae Kim and Yen-Liang Liu

Part VII

Flow Cytometry for Single Cell Analysis . . . . . . . . . . . . . .

865

887

35

Single-Cell Impedance Flow Cytometry . . . . . . . . . . . . . . . . . . . . . Hongyan Liang, Huiwen Tan, Deyong Chen, Junbo Wang, Jian Chen, and Min-Hsien Wu

889

36

Cytometry of Single Cell in Biology and Medicine . . . . . . . . . . . . . Shunbo Li

921

Part VIII Spectrum Analysis, Methods, Targets, Imaging, and Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

943

37

Single Cell Electrophysiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Faruk Azam Shaik, Satoshi Ihida, Agnes Tixier-Mita, and Hiroshi Toshiyoshi

38

Mechanical and Microwave Resonators for Sensing and Sizing Single Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . M. Selim Hanay

973

Molecular Force Spectroscopy on Cells: Physiological Functions of Cell Adhesion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sayan Deb Dutta, Dinesh K. Patel, Keya Ganguly, and Ki-Taek Lim

997

39

945

40

Micro-tweezers and Force Microscopy Techniques for Single-Cell Mechanobiological Analysis . . . . . . . . . . . . . . . . . . . . . 1011 Lanqi Gong, Weiyi Qian, Renee-Tyler Tan Morales, Jie Tong, Apratim Bajpai, and Weiqiang Chen

41

Mass Spectrometry for Single-Cell Analysis . . . . . . . . . . . . . . . . . . 1033 Dinesh K. Patel, Sayan Deb Dutta, and Ki-Taek Lim

42

Acoustic Tweezers for Single-Cell Manipulation Adem Ozcelik and Tony Jun Huang

. . . . . . . . . . . . . . 1051

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1079

About the Editors

Tuhin Subhra Santra is Assistant Professor in the Department of Engineering Design at the Indian Institute of Technology Madras, India, from July 2016. He was a tenure track “Honorary Visiting Professor” at National Tsing Hua University” Taiwan from 2108 to 2020, and he was a “Visiting Professor” at the University of Cambridge, UK, in 2019. Dr. Santra received his Ph.D. degree in Bio-Nano Electro Mechanical Systems (Bio-NEMS) from National Tsing Hua University (NTHU), Taiwan, in 2013. Dr. Santra was a Postdoctoral Researcher at the California NanoSystems Institute (CNSI), University of California, Los Angeles (UCLA), USA, from 2015 to 2016. His main research areas are Bio-NEMS, MEMS, single cell technology, single molecule detection, biomedical micro-/nano devices, nanomedicine, etc. Currently, Dr. Santra is serving as a Guest Editor for Cells, Micromachines, MDPI Journals and Frontiers of Bioengineering and Biotechnology. He served as a Guest Editor for Cells, MDPI, in 2020; International Journal of Molecular Sciences (IJMS) in 2018, 2017, and 2015; Sensors in 2016; Molecules in 2016; and Micromachines in 2020 and 2013, among others. He was conference chair and committee member of IEEE-NEMS in 2017, 2020 and 2021. Dr. Santra has received many honors and awards such as “DBT/Wellcome Trust India Alliance Fellowship” in 2018, Honorary Research Fellow from National Tsing Hua University, Taiwan, in 2018, Bharat Bikas Award in 2017, IEEE-NEMS best conference paper award in 2014, a silver medal from Vidyasagar University in 2004, etc. He is Editor of the book entitled Nanomaterials and Their Biomedical Applications by xiii

xiv

About the Editors

Springer Nature, Singapore, in 2021; Microfluidics and Bio-NEMS: Devices and Applications by Jenny Stanford Publisher, Singapore, in 2020; and Essential of Single Cell Analysis Springer, Germany, in 2016, among others. He published more than 6 books, 35 SCI journals, 20 book chapters, 15 US/Taiwan/Indian patents, and 20 international conference proceedings in his research field. Dr. Fan-Gang (kevin) Tseng received his Ph.D. degree in Mechanical Engineering from UCLA, USA, in 1998. He joined Engineering and System Department of National Tsing Hua University in 1999 and advanced to Professor in 2006. He was the Chairman of ESS Department in NTHU (2010–2013), Associate Vice President for Global Affair in NTHU (2013), a Visiting Scholar of Koch Institute of Integrated Cancer Research in MIT USA (2014–2015), and the Dean of Nuclear Science College in NTHU (2016–2017). He is currently a distinguished professor of ESS Department as well as NEMS I., and the Vice President for R&D at NTHU (2017–present), as well as a Research Fellow with Academia Sinica Taiwan (2006–present). He was elected an ASME fellow in 2014. His research interests are in the fields of BioNEMS, biosensors, micro-fluidics, tissue chips, and fuel cells. He received 60 patents, wrote 8 book chapters, and published more than 260 SCI journal papers and 400 conference technical papers. He has received several awards, including Shakelton Scholar, twice National Innovation Award, twice Outstanding in Research Award, and Mr. Wu, Da-Yo Memorial Award from MOST, Taiwan, and more than20 best papers and other awards in various international conferences and competitions. He is among the editorial board of several journals including IJMS, Cells, Micromachines, and Applied Science, and also the general co-chair for MicroTas 2018 and a board member of CBMS from 2018 to 2022.

Contributors

Mohd Ridzuan Ahmad Micro-Nano System Engineering Research Group, Division of Control and Mechatronics Engineering, School of Electrical Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, Skudai, Johor, Malaysia Pranav Ambhorkar School of Engineering, University of British Columbia, Kelowna, BC, Canada Dian Anggraini Division of Materials Science, Nara Institute of Science and Technology, Nara, Japan Apratim Bajpai Department of Mechanical and Aerospace Engineering, New York University, New York, NY, USA Saumendra Bajpai Applied Mechanics, Biomedical Division, Indian Institute of Technology Madras, Chennai, Tamilnadu, India Maniraj Bhagawati Institute of Molecular Cell Biology, University of Münster, Münster, Germany Rohit Bhardwaj Department of Mechanical Engineering, Toyohashi University of Technology, Toyohashi, Japan Katherine Bohn Department of Infectious Diseases and Immunology, College of Veterinary Medicine, University of Florida, Gainesville, FL, USA Jolien Breukers Department of Biosystems – Biosensors Group, KU Leuven, Leuven, Belgium O. T. M. Bucheli ETH Laboratory for Functional Immune Repertoire Analysis, Institute of Pharmaceutical Sciences, D-CHAB, ETH Zürich, Zürich, Switzerland Anna Sanchez Calle Division of Molecular and Cellular Medicine, National Cancer Center Research Institute, Tokyo, Japan Bruno P. A. Cammue Centre of Microbial and Plant Genetics, KU Leuven, Leuven, Belgium VIB Center for Plant Systems Biology, Ghent, Belgium xv

xvi

Contributors

Hao-Chen Chang Institute of Biomedical Engineering and Nanomedicine, National Health Research Institutes, Miaoli, Taiwan Tissue Engineering and Regenerative Medicine, National Chung Hsing University, Taichung, Taiwan Hwan-You Chang Department of Medical Science, National Tsing Hua University, Hsinchu, Taiwan Chih-Wei Chen Department of Molecular and Cellular Biology, National Tsing Hua University, Hsinchu, Taiwan Deyong Chen State Key Laboratory of Transducer Technology, Institute of Electronics, Chinese Academy of Sciences, Beijing, People’s Republic of China University of Chinese Academy of Sciences, Beijing, People’s Republic of China Jian Chen State Key Laboratory of Transducer Technology, Institute of Electronics, Chinese Academy of Sciences, Beijing, People’s Republic of China University of Chinese Academy of Sciences, Beijing, People’s Republic of China Weiqiang Chen Department of Biomedical Engineering, New York University, New York, NY, USA Department of Mechanical and Aerospace Engineering, New York University, New York, NY, USA Yu-Chih Chen Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, USA Forbes Institute for Cancer Discovery, University of Michigan, Ann Arbor, MI, USA Ya-Yu Chiang Department of Mechanical Engineering, National Chung Hsing University, Taichung, Taiwan Yehyun Choi Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, USA Jing Ding Department of Mechanical and Nuclear Engineering, Kansas State University, Manhattan, KS, USA Sayan Deb Dutta Department of Biosystems Engineering, The Institute of Forest Science, Kangwon National University, Chuncheon, Republic of Korea K. Eyer ETH Laboratory for Functional Immune Repertoire Analysis, Institute of Pharmaceutical Sciences, D-CHAB, ETH Zürich, Zürich, Switzerland Laboratoire de Colloïdes et Matériaux Divisés (LCMD), ESPCI Paris, PSL Université, Paris, France Shih-Kang Fan Department of Mechanical and Nuclear Engineering, Kansas State University, Manhattan, KS, USA Keya Ganguly Department of Biosystems Engineering, Kangwon National University, Chuncheon, Republic of Korea

Contributors

xvii

Reya Ganguly Chungnam National University, Daejeon, Republic of Korea Fabrice Gielen University of Exeter, Living Systems Institute, University of Exeter, Exeter, UK Lanqi Gong Department of Biomedical Engineering, New York University, New York, NY, USA Harsh Gupta Department of Mechanical Engineering, Toyohashi University of Technology, Toyohashi, Japan M. Selim Hanay Department of Mechanical Engineering, Bilkent University, Ankara, Turkey Institute of Materials Science and Nanotechnology (UNAM), Bilkent University, Ankara, Turkey Althea Henderson Department of Biology, Clarkson University, Potsdam, NY, USA Sara Horta Laboratory for Thrombosis Research, IRF Life Sciences, KU Leuven Campus Kulak Kortrijk, Kortrijk, Belgium Yoichiroh Hosokawa Division of Materials Science, Nara Institute of Science and Technology, Nara, Japan Chia-Hsien Hsu Institute of Biomedical Engineering and Nanomedicine, National Health Research Institutes, Miaoli, Taiwan Institute of Nanoengineering and Microsystems, National Tsing Hua University, Hsinchu, Taiwan Tissue Engineering and Regenerative Medicine, National Chung Hsing University, Taichung, Taiwan Geng-Ming Hu Department of Physics, National Taiwan Normal University, Taipei, Taiwan Myat Noe Hsu Singapore-MIT Alliance for Research and Technology, Singapore, Singapore Shen Hu School of Dentistry, Jonsson Comprehensive Cancer Center, University of California, Los Angeles (UCLA), Los Angeles, CA, USA Tony Jun Huang Department of Mechanical Engineering and Material Science, Duke University, Durham, NC, USA Satoshi Ihida Development Group, Display Device Company, Sharp Corporation, Tokyo, Japan Seungwon Jung Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, USA

xviii

Contributors

Srabani Kar Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India Department of Electrical Engineering, University of Cambridge, Cambridge, UK Tomoji Kawai The Institute of Scientific and Industrial Research, Osaka University, Ibaraki, Japan Keekyoung Kim Department of Mechanical and Manufacturing Engineering, Schulich School of Engineering, University of Calgary, Calgary, AB, Canada Biomedical Engineering Graduate Program, University of Calgary, Calgary, AB, Canada Mirae Kim Department of Biomedical Engineering, The University of Texas at Austin, Austin, TX, USA Peter Koltay Laboratory for MEMS Applications, Department of Microsystems Engineering-IMTEK, University of Freiburg, Freiburg, Germany Petra Kraus Department of Biology, Clarkson University, Potsdam, NY, USA Amogh Kumar Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India Hitendra Kumar School of Engineering, University of British Columbia, Kelowna, BC, Canada Ching-Te Kuo Department of Mechanical and Electro-Mechanical Engineering, National Sun Yat-sen University, Kaohsiung, Taiwan Jeroen Lammertyn Department of Biosystems – Biosensors Group, KU Leuven, Leuven, Belgium Chang-Soo Lee Chungnam National University, Daejeon, Republic of Korea Hsinyu Lee Department of Life Science, National Taiwan University, Taipei, Taiwan Victor Daniel Lee Bio-Gen Inc., Tegucigalpa, Honduras Kangning Li Department of Biology, Clarkson University, Potsdam, NY, USA Ming Li School of Engineering, Macquarie University, Sydney, Australia Shunbo Li College of Optoelectronic Engineering, Chongqing University, Chongqing, China Zehao Li College of Life Sciences, Beijing University of Chemical Technology, Beijing, China Hongyan Liang State Key Laboratory of Transducer Technology, Institute of Electronics, Chinese Academy of Sciences, Beijing, People’s Republic of China University of Chinese Academy of Sciences, Beijing, People’s Republic of China

Contributors

xix

Ki-Taek Lim Department of Biosystems Engineering, The Institute of Forest Science, Kangwon National University, Chuncheon, Republic of Korea Hung-Yu Lin NeuroSky, Inc., San Jose, CA, USA Weikang Nicholas Lin Department of Biomedical Engineering, National University of Singapore, Singapore, Singapore Hsin-Yi Liu Department of Pharmacy, Hsinchu National Military Hospital, Hsinchu, Taiwan Yen-Liang Liu Master Program for Biomedical Engineering, China Medical University, Taichung, Taiwan Research Center for Cancer Biology, China Medical University, Taichung, Taiwan Thomas Lufkin Department of Biology, Clarkson University, Potsdam, NY, USA Hima Manoj Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India Muhammad Asraf Mansor Micro-Nano System Engineering Research Group, Division of Control and Mechatronics Engineering, School of Electrical Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, Skudai, Johor, Malaysia Pu-Wei Mao Genomics Research Center, Academia Sinica, Taipei, Taiwan Ninad Mehendale Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra, India L. Mohan Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India Renee-Tyler Tan Morales Department of Biomedical Engineering, New York University, New York, NY, USA Muniesh Muthaiyan Shanmugam Department of Life Science, National Tsing Hua University, Hsinchu, Taiwan Buck Institute for Research on Aging, Novato, CA, USA Moeto Nagai Department of Mechanical Engineering, Toyohashi University of Technology, Toyohashi, Japan Cuong Q. Nguyen Department of Infectious Diseases and Immunology, College of Veterinary Medicine, University of Florida, Gainesville, FL, USA Department of Oral Biology, College of Dentistry, University of Florida, Gainesville, FL, USA Center of Orphaned Autoimmune Diseases, University of Florida, Gainesville, FL, USA

xx

Contributors

Zhonghua Ni School of Mechanical Engineering, and Jiangsu Key Laboratory for Design and Manufacture of Micro-Nano Biomedical Instruments, Southeast University, Nanjing, China Takahiro Ochiya Division of Molecular and Cellular Medicine, National Cancer Center Research Institute, Tokyo, Japan Institute of Medical Science, Tokyo Medical University, Tokyo, Japan Nobutoshi Ota Center for Biosystems Dynamics Research (BDR), RIKEN, Osaka, Japan Adem Ozcelik Department of Mechanical Engineering, Aydin Adnan Menderes University, Aydin, Turkey Gaurav Pandey Department of Mechanical Engineering, Toyohashi University of Technology, Toyohashi, Japan Shilpi Pandey Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra, India Long Pang College of Basic Medical Science, The Shaanxi Key Laboratory of Brain Disorders, Xi’an Medical University, Xi’an, China Dinesh K. Patel Department of Biosystems Engineering, The Institute of Forest Science, Kangwon National University, Chuncheon, Republic of Korea Debjani Paul Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra, India Jih-Hou Peh BSL3 core facility Laboratory, National University of Singapore, Singapore, Singapore Michael Philippi Department of Biology/Chemistry, University of Osnabrück, Osnabrück, Germany Center for Cellular Nanoanalytics (CellNanOs), University of Osnabrück, Osnabrück, Germany Marta Prieto-Vila Division of Molecular and Cellular Medicine, National Cancer Center Research Institute, Tokyo, Japan Evgeny Puchkov All-Russian Collection of Microorganisms, G.K. Skryabin Institute of Biochemistry and Physiology of Microorganisms of the Russian Academy of Sciences, Pushchino Center for Biological Research of the Russian Academy of Sciences, Pushchino, Russia Weiyi Qian Department of Mechanical and Aerospace Engineering, New York University, New York, NY, USA Privita Edwina Rayappan George Edwin Applied Mechanics, Biomedical Division, Indian Institute of Technology Madras, Chennai, Tamilnadu, India

Contributors

xxi

Julian Riba Laboratory for MEMS Applications, Department of Microsystems Engineering-IMTEK, University of Freiburg, Freiburg, Germany Sangjin Ryu Department of Mechanical and Materials Engineering, University of Nebraska-Lincoln, Lincoln, NE, USA Nebraska Center for Materials and Nanoscience, University of Nebraska-Lincoln, Lincoln, NE, USA Mahmoud Ahmed Sakr School of Engineering, University of British Columbia, Kelowna, BC, Canada Tuhin Subhra Santra Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India Brianna L. Schroeder Department of Infectious Diseases and Immunology, College of Veterinary Medicine, University of Florida, Gainesville, FL, USA Touyana Semenova Department of Infectious Diseases and Immunology, College of Veterinary Medicine, University of Florida, Gainesville, FL, USA Faruk Azam Shaik Institute of Industrial Science, The University of Tokyo, Tokyo, Japan University of Lille, Lille, France Yigang Shen Center for Biosystems Dynamics Research (BDR), RIKEN, Osaka, Japan Takayuki Shibata Department of Mechanical Engineering, Toyohashi University of Technology, Toyohashi, Japan Ashwini Shinde Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India Pallavi Shinde Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India I. Sigvaldadóttir ETH Laboratory for Functional Immune Repertoire Analysis, Institute of Pharmaceutical Sciences, D-CHAB, ETH Zürich, Zürich, Switzerland Darren Sipes Department of Biology, Clarkson University, Potsdam, NY, USA Caroline Struyfs Centre of Microbial and Plant Genetics, KU Leuven, Leuven, Belgium VIB Center for Plant Systems Biology, Ghent, Belgium Shantanu Sur Department of Biology, Clarkson University, Potsdam, NY, USA Hiroaki Suzuki Faculty of Science and Engineering, Chuo University, Tokyo, Japan Ryou-u Takahashi Department of Cellular and Molecular Biology, Hiroshima University, Higashihiroshima, Japan

xxii

Contributors

Huiwen Tan State Key Laboratory of Transducer Technology, Institute of Electronics, Chinese Academy of Sciences, Beijing, People’s Republic of China University of Chinese Academy of Sciences, Beijing, People’s Republic of China Yo Tanaka Center for Biosystems Dynamics Research (BDR), RIKEN, Osaka, Japan Zirui Matthew Tay Singapore Immunology Network, Agency for Science, Technology and Research, Singapore, Singapore Karin Thevissen Centre of Microbial and Plant Genetics, KU Leuven, Leuven, Belgium Agnes Tixier-Mita Institute of Industrial Science, The University of Tokyo, Tokyo, Japan Jie Tong Department of Mechanical and Aerospace Engineering, New York University, New York, NY, USA Hiroshi Toshiyoshi Institute of Industrial Science, The University of Tokyo, Tokyo, Japan Chia-Hung Dylan Tsai Department of Mechanical Engineering, National Chiao Tung University, Hsinchu, Taiwan Fan-Gang Tseng Department of Engineering and System Science, National Tsing Hua University, Hsinchu, Taiwan Mamiko Tsugane Faculty of Science and Engineering, Chuo University, Tokyo, Japan Japan Society for the Promotion of Science (JSPS), Tokyo, Japan Makusu Tsutsui The Institute of Scientific and Industrial Research, Osaka University, Ibaraki, Japan Karen Vanhoorelbeke Laboratory for Thrombosis Research, IRF Life Sciences, KU Leuven Campus Kulak Kortrijk, Kortrijk, Belgium Lara Varden Department of Biology, Clarkson University, Potsdam, NY, USA Alexandria Voigt Department of Infectious Diseases and Immunology, College of Veterinary Medicine, University of Florida, Gainesville, FL, USA Junbo Wang State Key Laboratory of Transducer Technology, Institute of Electronics, Chinese Academy of Sciences, Beijing, People’s Republic of China University of Chinese Academy of Sciences, Beijing, People’s Republic of China Takashi Washio The Institute of Scientific and Industrial Research, Osaka University, Ibaraki, Japan Shih-Chung Wei Department of Biomedical Engineering, National University of Singapore, Singapore, Singapore

Contributors

xxiii

Richard Witas Department of Oral Biology, College of Dentistry, University of Florida, Gainesville, FL, USA Min-Hsien Wu Graduate Institute of Biochemical and Biomedical Engineering, Chang Gung University, Taoyuan City, Taiwan Nan Xiang School of Mechanical Engineering, and Jiangsu Key Laboratory for Design and Manufacture of Micro-Nano Biomedical Instruments, Southeast University, Nanjing, China Xiangdong Xu School of Dentistry, Jonsson Comprehensive Cancer Center, University of California, Los Angeles (UCLA), Los Angeles, CA, USA School of Public Health, Hebei Medical University, Shijiazhuang, China Yaxiaer Yalikun Division of Materials Science, Nara Institute of Science and Technology, Nara, Japan Center for Biosystems Dynamics Research (BDR), RIKEN, Osaka, Japan Yusuke Yamamoto Division of Molecular and Cellular Medicine, National Cancer Center Research Institute, Tokyo, Japan Takeshi Yanagida Institute for Materials Chemistry and Engineering, Kyushu University, Kasuga, Japan Chuan-Feng Yeh Institute of Biomedical Engineering and Nanomedicine, National Health Research Institutes, Miaoli, Taiwan Institute of Nanoengineering and Microsystems, National Tsing Hua University, Hsinchu, Taiwan Rachel Yerden Department of Biology, Clarkson University, Potsdam, NY, USA Euisik Yoon Department of Electrical Engineering and Computer Science, University of Michigan, Ann Arbor, MI, USA Department of Biomedical Engineering, University of Michigan, Ann Arbor, MI, USA Center for Nanomedicine, Institute for Basic Science (IBS) and Graduate Program of Nano Biomedical Engineering (Nano BME), Advanced Science Institute, Yonsei University, Seoul, South Korea Changjiang You Department of Biology/Chemistry, University of Osnabrück, Osnabrück, Germany Center for Cellular Nanoanalytics (CellNanOs), University of Osnabrück, Osnabrück, Germany College of Life Sciences, Beijing University of Chemical Technology, Beijing, China Mingde Zheng Nokia Bell Laboratories, Murray Hill, NJ, USA Stefan Zimmermann Laboratory for MEMS Applications, Department of Microsystems Engineering-IMTEK, University of Freiburg, Freiburg, Germany

Part I Physical Methods for Single Cell Therapy and Analysis

1

Light-Induced Cellular Delivery and Analysis Ashwini Shinde, Srabani Kar, Moeto Nagai, Fan-Gang Tseng, and Tuhin Subhra Santra

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Light-Induced Intracellular Delivery from Bulk to Single-Cell Photoporation . . . . . . . . . . . . . . . . . Direct Laser-Cell Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Indirect Laser-Cell Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pulsed Laser Systems for Optoporation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Applications of Single-Cell Surgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 6 7 10 12 17 24 25

Abstract

In the era of advanced nanotechnology, laser-assisted photoporation is of particular importance for the delivery of membrane impermeable biomolecules into living cells for advanced biological and biomedical research. It has the advantage of briefness and nontoxicity. Photoporation is based on the localized transient pore generation in the cell membrane using pulsed or continuous laser light. Membrane permeability can be increased directly by focused laser light, with the

Ashwini Shinde and Srabani Kar contributed equally. A. Shinde · T. S. Santra (*) Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India e-mail: [email protected] S. Kar Department of Engineering Design, Indian Institute of Technology Madras, Chennai, India Department of Electrical Engineering, University of Cambridge, Cambridge, UK M. Nagai Department of Mechanical Engineering, Toyohashi University of Technology, Toyohashi, Japan F.-G. Tseng Department of Engineering and System Science, National Tsing Hua University, Hsinchu, Taiwan © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_4

3

4

A. Shinde et al.

help of micro-fabricated devices or in combination of sensitizing nanoparticles for higher throughput delivery. Here in this chapter, we have presented an overview toward single-cell photoporation/optoporation mechanisms and their biological and therapeutic applications.

Introduction Introduction of exogenous biomolecules into existing live cells is of particular importance not only for cell-related studies but also in bioimaging and therapy (Jones et al. 2006). Various physical, chemical, and viral methods have been developed to allow the transport of foreign molecules into cells (Kim and Eberwine 2010). The viral vectors are remarkably used for somatic gene therapy, where they are used as vehicles to incorporate altered genetic materials into the targeted cells. However, this method is cell specific, limited to DNA delivery and immunity response, and cost-effective (Pfeifer and Verma 2001). To overcome this, lipid- or polymer-based chemical vectors are designed that are internalized by endocytosis (Shinde et al. 2020a). Synthesis of micelles to carry the drug into the targeted cells are functionalized with proteins to identify their target cells and are taken up by endocytosis process. Though these techniques are partially successful for cargo delivery (drugs, DNA, RNA, fluorescence molecules, quantum dots, etc.) into targeted cells, complications such as immunogenic, pathogenic responses, instability, and low transfection efficiency are observed with no dosage control (Shinde et al. 2020a). Also elimination after intended function remains one of the major setback (Raemdonck et al. 2010). Considering their limitations, different physical approaches have widely been developed including the use of physical energy such as an electric field (Santra et al. 2020a, 2016, 2014; Santra and Tseng 2013; Kar et al. 2018), magnetic field (Shinde et al. 2020a; Du et al. 2018; Liu et al. 2012), mechanical stress (▶ Chap. 2, “Mechanoporation: Toward Single Cell Approaches”; Matsumoto et al. 2016; LaPlaca et al. 2019), optical energy (Stevenson et al. 2010; Patskovsky et al. 2020), or acoustic energy (Yoon and Park 2010; Ohta et al. 2008; Wang et al. 2018) to deform membrane of the cell and leading for formation of hydrophilic transient membrane pores to deliver the cargo into cells with spatial, temporal, and qualitative dosage control. This offers standard applicability over range of cell types (Lakshmanan et al. 2014). Again, almost all the physical methods are cell specific, resulting in nonuniform transfection and their throughputs are very limited within a few cells. For example, carrier-mediated delivery is mainly cell specific, also, limited by particular types or sizes of biomolecules, every so often mutagenic to the cells which might be risky for human studies (Wang et al. 2010). In addition, these methods are usually meant for bulk transfection to permeate a substantial number of cells together. Though the microinjection technique offers dosage control and targeted transfection, it is highly time consuming, user dependent, and therefore is not advisable for broad applications (Santra et al. 2020a; Essentials of single-cell analysis 2016; Santra 2020; Munish and Santra 2016; Harshan et al. 2020). On the contrary,

1

Light-Induced Cellular Delivery and Analysis

5

exogenous biomolecules such as nucleic acids and oligonucleotides can be efficiently delivered using electroporation into targeted living cells. But remarkably, cells suspension are exposed with high external electric field (kV/cm) inside a cuvette where two large electrodes are inserted for formation of temporary membrane pores to deliver the biomolecules inside the cell cytoplasm (Zimmermann et al. 1974). Multiple drawbacks are encountered such as field distortion, pH variation, higher voltage requirement, sample contamination, and thermal effect. This occurs because electrode consists of large surface area resulting in high toxicity and low cell viability (Kim et al. 2008, 2012; Valero et al. 2005). Application of the magnetic field has been as a promising physical methodology for cellular transfection because of its noninvasive nature. Magnetoporation, also known as magnetic drug targeting (MDT), is a well-known phenomenon for its drug release at a particular location after the application of magnetic field (Du et al. 2018; Lübbe et al. 2001). Here in this technique, biomolecules are released based on the temperature gradient. Thus by inducing hyperthermia, cell death occurs. It also has low transfection efficiency (Liu et al. 2012). Most of these techniques are meant for bulk poration instead of single-cell poration which is still a challenging and open platform to explore in-depth understanding of individual cells. Deeper knowledge about individual cells, their functioning, and interactions with the microenvironment can revolutionize the world of modern medical science. It is important to study each the biological components for making clear distinction of heterogeneity of each unique individual cells or cell subpopulations. Presence of such heterogeneous cells could be an outcome of local microenvironment or series of processes. An understanding of the process of cellular transition toward distinguished behavior can aid in unfolding the mechanisms of diseases. Every human activity is an outcome of series of coordinated functions performed by individual cells, which are grouped to form tissues and organs. These cells have independent functioning mechanism governed by molecular machinery. Various cellular pathways are invoked for performing cellular functions, which get mutated in diseased cells. Individual cells within a specific cell type showing heterogeneity can influence the overall outcome of the biological system. Single-cell analysis can discover four important aspects of cellular study as in heterogeneous distribution profile of cells, collection of genetic information from rare cells, cell lineage mapping, and discovering new cell types (Single Cell 2016; Micro/Nano fluidic devices 2015; Handbook of single cell technology 2021; Shinde et al. 2018, 2021; Gupta et al. 2020). Microfluidic systems along with micro-total analysis system (uTAS) and lab-on-a-chip (LOC) technology show potential application in controlling individual cells. However, lack of microfluidic expertise by biologist has enforced them to use commercially available microfluidic solutions, along with their limitations associated with scalability. Adoption of microfluidic technology for single-cell analysis on a wide scale, customized for varying experimental needs, is the key for rapid progress in biological studies (Meister et al. 2009; Dörig et al. 2010). In single-cell technique, the major challenges faced by scientists is the lack of tools designed for specific cell analysis with statistically significant data collection. Limitation on the analyte volume, obtained from individual cell, encourages pooling of

6

A. Shinde et al.

analyte from similar samples for analysis with bulk analysis tools. This results in averaging effect causing masking of data from low frequency allele. Further, scaling of experiments for statistically significant data requires both complex and intricate system design. The development in biological sciences owes the credibility to various tools engineered for specific process studies, and further progress depends on scaling down engineering tools for cellular and subcellular studies (Chen et al. 2016). However, the challenges faced by the world with the advent of modern diseases need more understanding of the disease phenomenon along with modern techniques for its treatment. Development of tools capable of handling and manipulating cells becomes essential. The development of Bio-MEMS technology for fabrication of cellular and subcellular structures has made cell manipulation possible. Such techniques in adjunct with single-cell technology can revolutionize biomedical research and applications. Along with all these existing methodologies, photoporation aims to utilize light as the energy source to create permeation and distinctively stands out over other physical modes. High-intensity femtosecond (fs) laser pulses are focused exactly over the cell membrane and the exogenous molecules present in the medium diffuse into the cytoplasm through the created pores (He et al. 2008). This greatly finds its application in single-cell studies. In addition, this technique is user-friendly and costeffective which make optoporation one of the best techniques for biomedical research and cellular therapy. However, it is challenging to achieve high throughput optoporation for creating transient pores on large scale (Antkowiak et al. 2013). Use of plasmonic nanoparticles (NPs) is reported handy to enhance photoporation throughput (Sapsford et al. 2013). Research on low-intensity laser pulses or continuous wavelengths (CWs) for temperature-mediated poration (Sun et al. 2013), as well as generation of vapor nanobubbles (VNBs) to mediate mechanical membrane permeation to achieve higher transfection efficiency in nanomaterial-sensitized cells is widely reported (Zhang et al. 2013). However, the field is still emerging and far to explore completely. In this chapter we will discuss how light energy has been utilized in advanced level for single-cell intracellular delivery. We have discussed the mechanisms of optoporation and the current state of art in the following sections. It is an attractive alternative to the traditional methods of intracellular drug delivery. Different terms have been coined to represent laser-mediated delivery. Usually, the term “optoinjection” has been used to describe the insertion of molecules into the cell cytoplasm, “optical transfection” to describe the process that leads to the protein expression, and “optical poration” as a generic term for creating membrane pores.

Light-Induced Intracellular Delivery from Bulk to Single-Cell Photoporation The technique of perforating cell membrane for intracellular delivery by using light energy is known as optoporation. This technique offers a noncontact, contaminationfree, and targeted way to deliver membrane-impermeable molecules into the cell

1

Light-Induced Cellular Delivery and Analysis

7

cytoplasm. Different molecules and nanoparticles can be injected using this technique, and thus cells are transfected with foreign genetic materials such as plasmids and nucleic acids in different cell types. It is important to note that it has also been applied to transfect primary neurons, stem cells, and living embryo which are hard to transfect. Mechanisms of laser-cell interactions: Laser interacts with cell membrane directly or indirectly to induce transient pores into it involving continuous wave (CW) or pulsed laser beam. These methods involve membrane ablation by plasma-induced high pressure, photothermal, photochemical, and photomechanical interactions. Laser-cell interactions occur either directly with cells or by some external objects such as nanoparticles where laserinduced physical phenomena at extracellular space in the vicinity of cell membrane help to induce membrane pore, instead of directly interacting with cell membrane or intracellular environment. Thus, the optoporation mechanisms can be divided into two categories – one, direct laser-cell interaction, and two, indirect laser-cell interactions.

Direct Laser-Cell Interaction In this process, usually a laser beam is tightly focused onto the cell membrane or in close proximity to ablate the membrane and create transient pores. A schematic is illustrated in Fig. 1. At focal point, high-intensity laser light generates plasma that enhances membrane’s permeability. The pore size can be tuned in the range of a few hundreds of nanometer to micrometer by varying the light intensity. There are mainly three mechanisms such as photothermal, and photochemical interactions that play important role for generating transient pores depending on the type of laser, pulse energy, repetition rate, intensity, pulse duration, etc. (Xiong et al. 2016). Direct photothermal interaction occurs due to single photon absorption directly by cellular constituents following the exposure of light at UV and visible range. After photoexcitation, a part of energy relaxation from excited state to ground state can occur through nonradiative recombination which results in local heating effect. This heating effect increases the permeability of cell membrane though phase transition of membrane (Parasassi et al. 1991; Schneckenburger 2019) and denaturation of the membrane proteins (Xiong et al. 2016). However, cellular constituents such as flavins, carotenoids, some vitamin precursors, cytochromes lipids, and proteins show comparable low absorption in the wavelength range of 350–1100 nm (Palumbo et al. 1996). As a result, photothermal heating induced by absorption of single photon is negligible to form pores onto cell membranes (Xiong et al. 2016). For this reason, sometimes phenol red, whose absorption at 488 nm is rather high, has been added for enhancing photothermal effect (Palumbo et al. 1996). Photoporation induced by photothermal interaction is mostly reversible below a certain light dosage. For example, a CW argon ion laser at 488 nm was used with a light dose of 2.5 MJ/cm2 for 2.5 s within a small spot of 1.0 μm diameter, which generated a tiny black spot that vanished within about 5 min after the laser exposure (Palumbo

8

A. Shinde et al.

Fig. 1 A schematic illustration of direct lasermembrane interaction. (Taken from Xiong et al. 2016)

et al. 1996). Above a certain light dose ( 5 MJ/cm2), permanent changes of morphology were observed, associated to lethal damages (Palumbo et al. 1996). Direct laser-cell interactions can also occur through photochemical interactions. It occurs due to the absorption of light by endogenous photosensitizer (PS) molecules of cells at visible and near-ultraviolet wavelengths (wavelengths below 650 nm). A schematic process is demonstrated in Fig. 2. Photochemical interactions generate reactive oxygen species (ROS) and free radicals which trigger a series of chain reaction of lipid peroxidation. As a result, hydrophobicity of the cell membrane decreases, which allows the diffusion of exogenous molecules from the surrounding microenvironment to enter the cell. Photochemical interactions might be irreversible or reversible depending on the intensity of light dosage and laser wavelengths (Schneckenburger 2019). Usually, photochemical reactions have been applied for photodynamic therapy, e.g., in the cancer treatment (Schneckenburger 2019). However, photochemical damage can also be reduced by applying near-infrared light. But, infrared absorption by water and two-photon absorption by endogenous biomolecules act as dose-limiting factors (Schneckenburger 2019). In the visible or near-UV spectral range, the photochemical interactions can be triggered at low light dose around 100 J/cm2. However, near-infrared range requires higher doses. But in

Fig. 2 Photodynamic therapy mechanism. The PS in a ground state is excited by light to an activated singlet state (1PS*), and after an intersystem crossing to a triplet state (3PS*) it transfers energy to molecular oxygen, generating singlet oxygen (type II reaction), or transfers electrons to other molecules (type I reaction), generating reactive species within the tumor tissue and leading to the death of the cancer cells. (Taken from de Freitas and Hamblin 2016)

1 Light-Induced Cellular Delivery and Analysis 9

10

A. Shinde et al.

case of small areas of cell around 1 μm2 up to some hundred MJ/cm2 are irradiated that can be applied without any photochemical cell damage. However, the techniques are more applicable as bulk optoporation.

Indirect Laser-Cell Interaction In indirect laser-cell interaction processes, laser interacts with cell membrane through some intermediate processes which may involve metallic nanoparticles (NPs) or surrounding environment and induce photothermal, photochemical, and photomechanical effects to trigger cell permeation. The most popular techniques are nanoparticle-mediated optoporation. In this process, laser is irradiated on metallic nanoparticles (mostly gold nanoparticles) or nonpatterns inducing several physical processes that cause pore formation on cell membrane, such as plasmon-induced local heating, nanobubble generation, and photochemical interactions which are shown schematically in Fig. 3. In plasmon-induced process, when the wavelength and angle of incidence of irradiated light fulfill surface plasmon resonance condition,

Fig. 3 Three major mechanisms of NP-sensitized photoporation. Outlining steps involved in (a) photothermal-mediated poration; (b) bubble-mediated photoporation; and (c) photochemicalmediated poration. (Reproduced from Xiong et al. 2016)

1

Light-Induced Cellular Delivery and Analysis

11

localized surface plasmon gets generated, resulting enhanced light absorption by several times. For example, nano-corrugated mushroom-shaped gold-coated polystyrene nanoparticle showed strong and localized field enhancement of around 300 at 945 nm (Santra et al. 2020b). The plasmonic hot spots result in increase of electron and lattice temperature of the AuNPs above their initial temperature that diffuses into the surrounding environment. Several phenomena can occur due to this fast temperature rise in microenvironment. The heat from the NPs can get absorbed to the adjacent cell membrane to create perforation. Continuous-wavelength (CW) laser irradiation or low-intensity laser pulses are given to achieve a local temperature rise of ten to hundreds of degrees (Delcea et al. 2012; Hatef et al. 2015). This in turn causes localized lipid bilayer phase transition or thermal disintegration of integral glycoproteins that can contribute toward hydrophilic transient pores (Delcea et al. 2012; Zhang et al. 2011). With CW lasers being widely available and cost-effective, achieving heat-induced poration can take tens of seconds or a few minutes to form transient pores (Delcea et al. 2012; Zijlstra and Orrit 2011). Also, the temperature rise at NP/medium interface can trigger bubble generation in vicinity of cell membrane after pulsed laser exposure. Several type of bubbles can be generated by metallic NPs or microstructures such as vapor nanobubbles, cavitation bubbles, plasmonic bubbles, etc. When NP temperature rapidly increases to several hundreds to thousands of degrees that can vaporized the surrounding liquid environment. This result in formation of water vapor nanobubble, also known as thermo-mediated nanobubble. Studies on effect of laser light in different gold NPs substantiate that the laser intensity, NPs material, and size significantly involve in formulation of size of the VNBs. VNBs provide an insulating effect that prevents the transfer of heat into the environment leading to better cell viability. This implies that almost the incident light energy is converted into mechanical energy that causes expansion of VNB (Lindenberg et al. 2008; Merabia et al. 2009). Expansion and collapse of VNB creates high-pressure shockwaves or liquid jet formations that can porate the cell membrane (Xiong et al. 2014). With diffusion being a major reason for delivery of the exogenous molecule in the medium to enter through the cell membrane pores, active flow of extracellular fluid can also be a potential contributor toward biomolecule delivery. This flow is reasoned as an outcome of the asymmetrical expansion and collapse of nanobubbles forming transient nanojets (Lukianova-Hleb et al. 2012a). Even NPs can induce photochemical interactions causing generation of ROS and ionization of water molecules. Plasma formation is another mechanism that can lead to VNB formation reported in 100-nm gold nanospheres irradiated with fs pulsed laser. Localized surface plasmon resonance (LSPR) causes localized field enhancement of NP dipole edges where plasma formation occurs by multiphoton ionization of the medium. Collision and recombination of surrounding water molecules cools down the plasma, producing a quick increase in temperature and pressure. This leads to formation of water VNB around the irradiated NPs (Boulais et al. 2013; Baumgart et al. 2012). The major advantage of NP-induced optoporation is the resonant wavelength for surface plasmon generation can be tuned by varying shapes and sizes of NPs. The enhanced absorption at resonant wavelength reduces the energy requirement, and as a result,

12

A. Shinde et al.

the heating effect reduces throughout the aqueous environment. NP-induced photoporation provides high throughput with less-intensity laser beam covering of 10– 100 s of cells at a given time. Au NPs are mostly used as they are relatively nontoxic in aqueous environment and used in various biomedical contexts (Sperling et al. 2008). It is also vital to consider the aqueous environment to be a mimic of biological medium with similar optical properties especially in the milieu of laser interaction with the microenvironment (Vogel et al. 2005). Gold nanoparticles (AuNPs) and silver nanoparticles (AgNPs), carbon, and titanium-based nanostructures have also shown promises in this field. In carbon materials, carbon black (CB) and graphene oxide (GO) are reported to have similar nano-sensitizing effects making it suitable for photoporation (Sengupta et al. 2014; Chakravarty et al. 2010). Thermal-mediated membrane permeation is replaced by the cavitation shockwaves created by carbon-steam reaction in CB (Chen and Diebold 1995; Löwen and Madden 1998). GO and carbon nanotubes when given CW laser light are known to absorb laser energy that can increase surface temperature (Tian et al. 2011; Yoo et al. 2015). This heating effect is elucidated to be a key mechanism in creating cell membrane pores, however, which requires further validation (Sun et al. 2000; Link et al. 2000; O’Connell et al. 2002).

Pulsed Laser Systems for Optoporation Various laser systems have been applied for optoporation including continuouswave laser to various pulsed laser systems including nanosecond, femtosecond, and microsecond lasers with various repetition rates, wavelengths, and pulse energy. Following sections will discuss the optoporation by nanosecond and femtosecond pulsed laser.

Nanosecond Laser In 1984, Tsukakoshi et al. demonstrated for the first time the transfection of normal rat kidney cell by optoporation by using an ultraviolet nanosecond laser with wavelength of 355 nm (Tsukakoshi et al. 1984). The laser was focused on membrane of normal rat kidney cells with spot size of 0.5 μm and 1 mJ energy to transfer the Ecogpt gene. During that time, ns-optical pulse-mediated transfection was thought as the result of laser-induced plasma and the pressure of the emitted shock wave and the energy of the cavitation bubble formed by the plasma expansion (Venugopalan et al. 2002). At early stage, the transfection efficiency was quite low (only a few tens of percentage) (Venugopalan et al. 2002; Knoll et al. 2004). In this method, a large zone extending over many cells can be impacted and molecular delivery can be achieved in several cells over a longer distance from the target cell. Later on, nanoparticles of different shape and sizes have been added in the cell containing medium to induce plasmon-enhanced optical absorption and localized intracellular delivery for various cells. Santra et al. used nano-mushroom-shaped corrugated gold nanoparticles (nm-AuPNP) to perform intracellular delivery using a nanosecond pulsed laser

1

Light-Induced Cellular Delivery and Analysis

13

Fig. 4 Schematic overview of photoporation procedure for intracellular delivery (a) PEG-mediated nm-AuPNPs were added to the cells for attachment onto the cell membrane (b) to remove unbound nm-AuPNPs, cell surface was washed and then cargo molecules were introduced just prior to laser exposure (c) formation of plasmonic nanobubbles were induced by ns pulsed laser at nm-AuPNPs and cell membrane interface which resulted in transient pore generation into the cell membrane; (d) intracellular delivery of cargo molecules was achieved successfully with membrane reseal. (Reproduced from Santra et al. 2020b)

(Fig. 4). Polyethylene glycol (PEG)-mediated nanoparticles and cells were incubated together for an hour prior to the experiment. Then cell-impermeable cargo molecules (Fig. 4b) were then added into the petridish right before performing the experiment. Ns pulsed lasers were illuminated on the nm-AuPNPs (Fig. 4c) that are attached to the cell membrane, due to which increase in electromagnetic field occurs near the nano-corrugated edges giving lightning-rod effect (optical hotspots). This generates local heat, causing a rapid enhancement in temperature, and succeeded with the

14

A. Shinde et al.

formation of PNBs near the nm-AuPNPs cell membrane interface. After formation of PNBs, they grew fast, merged, and collapsed to induce a strong fluid flow at the interface. After PNBs disintegrated, the formed cavity and stress wave transferred heat to the nano-restricted region of the cell membrane from the nm-AuPNPs (Xiong et al. 2014). Furthermore, ns pulsed laser exposure results in PNBs having lifetime of about less than 1 μs, which in turn is favorable for negligible transfer of heat from gold nanoparticle to the cell membrane (Lukianova-Hleb et al. 2012a, b). After the delivery of cargo, the cell membrane resealed again to maintain the cell viability (Fig. 4). The transient pore formation on the cell membrane depends on laser fluence and the concentration of nm-AuPNPs (Santra et al. 2020b). Even for small or negligible field enhancement titanium microstructure can induce transient local heating after being irradiated with ns laser due to increase of electron and lattice temperature (Shinde et al. 2020b; Wu et al. 2015). The local heating effect generated photothermal cavitation bubbles (Shinde et al. 2020b; Wu et al. 2015). The membrane deformation is caused by bubble-induced jet flow which is then followed by the generation of temporary hydrophilic pores at the cell membrane, which helps in intracellular delivery. This method provided a highthroughput platform for intracellular delivery for parallel large size cargo delivery into mammalian cells (Wu et al. 2015). However, modification is required for achieving single-cell selectivity. Apart from metal micro /nanostructures, Mohan et al. have demonstrated how titanium oxide micro /nanostructures can be applied as efficient platform for intracellular delivery. Micro-flower structures containing TiO2 nanotubes arrays (Mohan et al. 2021a) and micro-spike (Mohan et al. 2021b, c) structures containing reduced oxidized titanium elements also appeared to be efficient platform for optoporation (Mohan et al. 2019). The mechanisms and the associated results are schematically illustrated in Fig. 5. However, all these techniques are still lack of single-cell selectivity along with high-throughput methodologies which is the main goal for next step application purposes.

Femtosecond Laser For targeted gene transfection, femtosecond pulsed laser-induced optoporation has appeared to be a powerful technique for its ability of single-cell specificity, localized delivery, low toxicity, and consistent performance. The mechanisms of cellular optoporation depend on pulse width, repetition rate, and laser intensity. Femtosecond laser with high-repetition-rate (>1 MHz) sources generates multiphoton absorption by the membrane generating free electrons (Vogel et al. 2005). A low-density electron plasma is thus formed that induces degradation of the membrane by photochemical interaction (Vogel et al. 2005). As a result, single pore is formed at the cell membrane. In such photoporation process, the optical breakdown threshold in water is notably higher than actually required pulse energies for a cavitation bubble generation (Vogel et al. 2005). Thermoelastically induced small transient cavities are formed on the membrane with an ultrashort lifetime ( Qbypass) and reduced the ratio to Qtrap/Qbypass = 0.2 for cell trapping (Fig. 2C, D) (Sauzade and Brouzes 2017). The overhang structures in the channel drove single cells and particles toward a trapping site regardless of their initial position and thus could shorten the length of the bypass channel. These overhangs enabled the use of local flow to convey the cells toward the trapping channels. As a result, A498 or HeLa cancer cells first followed the trapping pathway and then blocked the entrance of the trapping channel. Further incoming cells flowed along the bypass pathway due to the cell plugging effect. 93.8% of incoming A498 cells were trapped by the first unoccupied trap, and 5.6% were trapped in the second empty capturing cite. Sauzade and Brouzes integrated all the functions of trapping, encapsulation, and retrieval of single cells in their device. A unique live-cell printing technique, termed “Block-Cell-Printing,” was proposed, and this technique did not employ the rule of Qtrap > Qbypass for trapping (Fig. 2E, F) (Zhang et al. 2014). Instead, there were two flow paths around a trap structure: one through the wide gap (22 μm in width) for cell bypass and the other through the narrow gap (3 μm in width) for cell capture. The ratio of fluid resistances between the paths was designed to be Rtrap/Rbypass = 41. As the flow rate through the bypass pathway was larger than that through the trap, almost all cells flowed through the wide gap of the trap area at low cell number densities (106 cells/mL), the wide gap was temporarily blocked by a group of cells, and an individual cell was pushed into the narrow gap and trapped. Hydrodynamic force could immediately clear such temporary blockages and maintain continuous flow of cells because of the flexibility of the cells. This method was applied to studying gap junction intercellular communication in heterotypic cell pairs with controlled morphologies. The cells’ ability to extend cellular membranes was characterized, and primary neurons were printed in a controlled manner. Thus, the Block-Cell-Printing method enabled high-throughput printing of functional single-cell arrays.

ä Fig. 2 Single-cell trapping in pressure-driven microfluidic networks. (A) Schematic diagrams of the self-regulating microfluidic trapping system and laser-based microbubble generation. (B) Micrographs show trapping of single beads. ((A, B) are reprinted with permission from Tan and Takeuchi (2007). Copyright (2007) National Academy of Sciences, U.S.A.) (C) Schematics of the microfluidic channel with displacement overhangs (top) and all the functionalities of the device (i.e., trapping, encapsulation, and retrieval) (bottom). (D) Flow pattern through a vacant trap (top) and sequential trapping of three cells (bottom). ((C, D) are reproduced with permission from The Royal Society of Chemistry (Sauzade and Brouzes 2017).) (E) Design and operation of the Block-Cell-Printing (BloC-Printing) technique. A typical BloC-Printing device consists of a PDMS BloC-Mold and a polystyrene petri dish. (Reprinted with permission from Zhang et al. (2014))

120

R. Bhardwaj et al.

Co-culture of Single Cells Co-culture of single cells has been explored to replicate in vivo cellular conditions. Cells in the natural environment remain in contact with others, and they interact via chemical signals to carry out basic biological processes. Thus, the cell surroundings greatly influence cell behavior. Studying this influence in an artificial environment requires cells to grow alongside other cells. Such co-culture can be used for parallel experiments to test the effects of clinical drugs on multiple samples simultaneously while providing culture media to cells for survival. The idea of the self-regulating trapping principle described in section “Single-Cell Trapping” has been extended to single-cell coupling. For co-culture of two cells in contact, Frimat et al. designed two grooves on each side with a serpentine path (Frimat et al. 2011) based on the working principle shown in Fig. 3A. The linear path, through the apertures between the mirrored cell traps, had a lower fluidic resistance Rtrap than the serpentine path Rbypass (Rtrap < Rbypass). Trapping of a spherical single cell reversed the fluidic resistance ratio, i.e., Rtrap0 > Rbypass. After the cells were incubated for 6 h for cell adhesion and flattening, the condition for Qtrap > Qbypass (i.e., Rtrap < Rbypass) was restored. Reversal flow was used to introduce a second spherical cell, and single cells were coupled sequentially under the condition Rtrap < Rbypass < Rtrap00 . Figure 3B shows the successful coupling of two cells. The efficiency of single-cell trapping was over 80% at the flow ratio of 1.4  Qtrap/Qbypass and with the diameter ratio of 2.0  trap channel diameter/cell diameter. Thus, the developed microfluidic cell manipulator enabled co-culturing of heterotypic single cells with unlabeled single SW480 cells and fluorescently labeled SW480 cells. Hong et al. developed a single-cell level co-culture platform to study dynamic cellular interactions that could maintain and track single-cell pair interactions (Hong et al. 2012) (Fig. 3C, D). Heterotypic pairing on a single-cell level was achieved through sequential cell trapping and dynamic variation in fluidic resistance. They set flow rates at Qbypass < Qtrap and introduced single mouse embryonic fibroblasts (MEF) cells to the trapping junction with a 3-μm-wide gap. During a 4 h cell incubation period, the cells generally migrated away from the trapping junction, and the trapping junctions were reactivated. Then, single mouse embryonic stem cells (mESC) were loaded in the same manner. Culture chambers with diameters of 400 μm provided the trapped cells with enough space to migrate and proliferate through multiple generations. Stem cell-fibroblast pairs were cultured and tracked for several generations. The migration patterns of the paired cells depended on their initial cell-to-cell distance, and heterotypic pairing led to distinct proliferation patterns from those of a homotypic single-cell culture. In contrast to the former two cases requiring a time difference between the first and second cell trapping and an incubation time for cell adhesion and flattening of the first cells, the platform developed by Chen et al. could trap two single cells at a chamber almost simultaneously to co-culture a pair of cells (Fig. 3E, F) (Chen et al. 2014). One platform had 56 chambers, and each chamber consisted of 2 capture sites, a central path, and 2 serpentine paths for bypass, to pair cells for the interaction.

5

Single-Cell Manipulation

A (a)

121 B

(b)

RBypass

RBypass

RTrap< RBypass

RTrap

RBypass< RTrap’

RTrap

RTrap’

RBypass

(c)

(d)

RBypass RTrap°< RBypass

RTrap°

RTrap°

RBypass RTrap°< RBypass RBypass < RTrap’’

RTrap°

RTrap’’

C

E

D

F

Fig. 3 Cellular valving for single-cell coupling for co-culture. (A) Principle of single-cell coupling with two grooves. Transformation of a trapped cell to a flattened morphology opened the channel. (B) Micrographs showing the cellular valving approach for single-cell co-culture. ((A, B) are reproduced with permission from The Royal Society of Chemistry (Frimat et al. 2011).) (C) Schematics of microfluidic co-culture platform. (D) Fluorescence images show the sequence of first cell trapping, migration, and second cell trapping. Mouse embryonic fibroblasts (MEF) were stained red, and mouse embryonic stem cells (mESC) were green. ((C, D) are reprinted with permission from Oxford University Press (Hong et al. 2012).) (E) Simulated flow velocity field in a chamber during cell capture. In this hydrodynamic capture scheme, two types of flow paths were created: the central and serpentine path. (F) Cell interactions between UM-SCC-1 squamous

122

R. Bhardwaj et al.

The long winding structure of the serpentine path was designed to increase the hydrodynamic resistance (i.e., Rcentral < Rbypass), so that the flow rate in this path was lower than that of the central path (i.e., Qbypass < Qcentral). The cells were guided to the central path and captured at the opening of the central path, which was slightly smaller (height, 10 μm; width, 10 μm) than the size of typical mammalian cells. Capturing cells stopped the flow through the central path, and thus the remaining cells flowed through the serpentine path and were captured in the next chamber. A capture rate of ~90% was achieved with proper geometric design. Furthermore, Chen et al. loaded a 1:1 ratio between two cell types in the mixture to maximize the probability of 1:1 cell pairing in the chamber. Ten chambers captured various combinations of cells after cell loading, and four of the ten chambers captured a pair of one UM-SCC-1 squamous carcinoma cell and one endothelial cell. Using two capture sites in each chamber, 25% of the chambers captured exactly a pair of two cell types, and other combinations could be generated simultaneously. They found that proliferation of cancer cells was enhanced when the cancer cells were co-cultured with endothelial cells.

Deforming, Sorting, and Separating Single Cells The previous two sections introduced the principle of single-cell trapping in microfluidic devices. The circuit representation of flow in microfluidic devices has also been utilized to design channels for other single-cell manipulation applications such as cell deformation, sorting, and separation. Kim et al. used the microfluidic single-cell trapping mechanism to sort micronsized samples (Fig. 4A) (Kim et al. 2014). Their microfluidic trapping device consisted of three trap zones of different trap sizes (zones A, B, and C), which were connected sequentially from the inlet. In each zone, the sieve-like channel wall including the trap channels was between the main channel and the side channel. Fluid flow through the sieve sequentially filled the traps with particles of the trap size, and particles smaller than the trap size passed through the sieve and were trapped by smaller traps downstream. When a trap was filled with a particle, its fluidic resistance increased. As a result, the flow rate into the occupied trap was reduced, and subsequent particles bypassed the occupied trap and flowed into the next empty trap. The fluidic resistances of the main and side channels were designed to achieve the maximum flow rate through the side. They decoupled the fluidic flow in one stage from the following stages by modularizing a main and side channel. This decoupling enabled the researchers to focus on flow in each stage regardless of flow in the other stages. More than 85% of the polystyrene ä Fig. 3 (continued) carcinoma cell and endothelial cell (EC) for 3 days. (Left) A single UM-SCC-1 cell after culture in the chamber with no proliferation. (Right) After pairing of one UM-SCC-1 and one EC co-cultured for 3 days, one UM-SCC-1 cell proliferated into three cells. ((E, F) are reproduced with permission from The Royal Society of Chemistry (Chen et al. 2014))

5

Single-Cell Manipulation

123

Fig. 4 Deforming, sorting, and separation of single cells using microfluidic devices designed based on the lumped-element modeling of flow. (A) Microfluidic trapping device for three different sizes of particles: zones A, B, and C for large, medium, and small particles, respectively. Schematic diagram for the trapping and passing mechanism (left) and circuit representation of particle flow (right). RM fluidic resistance of the main channel, RS fluidic resistance of the side channel, RT and RE fluidic resistance of the trap when filled or empty, respectively. (Reproduced with permission from The Royal Society of Chemistry (Kim et al. 2014).) (B) Photograph of a fabricated cell viscoelasticity microcytometer. Micrograph of the confining microchannel. Scale bar: 100 μm. Circuit model describing the relation among fluidic resistances of different flow section (top right). Free body diagram showing cell deformation in the confining microchannel (bottom). (Reprinted with permission from Hu et al. (2016). Copyright (2016) Springer Nature.) (C) Raman-activated cell sorting (RACS) using hydrodynamic focusing and pressure switching capability. In the fluidic network, detection channel was connected with pressure dividers. (D) Circuit model of microfluidic pressure divider and (E) the RACS platform. (F) The microfluidic channel of the RACS platform. ((C–E) are reproduced with permission from The Royal Society of Chemistry (McIlvenna et al. 2016))

microspheres of three different diameters (15 μm, 6 μm, and 4 μm) were sorted in the correct segment of zones A–C. Thus, this device could sort three different species of waterborne parasites (Entamoeba, Giardia, and Cryptosporidium) by size.

124

R. Bhardwaj et al.

Hu et al. developed an elasticity microcytometer to quantitatively measure the elasticity of live single cancer cells using a lumped-circuit model (Hu et al. 2016; Hu and Lam 2017). Floating cancer cells were moved under a pressure along a microchannel and were deformed by the channel side walls. The device for viscoelasticity measurement was composed of a confining microchannel and two bypass channels with a height of 50 μm in parallel (Fig. 4B) (Hu and Lam 2017). The confining channel was 300 μm long and had an entrance (30 μm wide) and exit (4 μm wide). Cells were conveyed through the channels by applying a gauge pressure Pdrive  Patm at the device inlet, where Patm is the atmospheric pressure at the device outlet. The device structure was modeled to analyze for the flow rates along different parts of the device. The fluidic resistance of the entire device, Rtotal, is expressed by Rtotal ¼ Rin þ Rout þ

Rbypass Rtrap Rbypass þ 2Rtrap

ð14Þ

where the fluidic resistances of the confining channel, bypass microchannels, the upstream inlet channel, and the downstream outlet channel are Rtrap, Rbypass, Rin, and Rout, respectively. The fluidic resistances were obtained by simulation software as Rin = 5.2  1010 Pas/m3, Rout = 5.2  1010 Pas/m3, Rbypass = 10.1  1010 Pas/m3, and Rtrap = 8.2  1013 Pas/m3. As Rbypass  Rtrap, the total resistance can be approximated as Rtotal  Rin + Rout + Rbypass/2, which implies that the hydraulic pressure difference across the confining channel was maintained under the steady driving pressure of the device. The gauge inlet pressure Pin was obtained to calculate the drag force on a cell by Pin = [1  (Rin + Rout)/Rtotal]  Pdrive. Raman-activated cell sorting (RACS) was demonstrated based on the Raman spectra of the molecules inside the cells (McIlvenna et al. 2016). The Raman signals from the cells were measured, while the pressure was manipulated using an external pressure control system and microfabricated pressure dividers along the flow channel (Fig. 4C, D). After the cells were aligned in a straight line using hydrodynamic focusing via sheath flow, Raman spectroscopy was used to guide separation of the cells. Using this technique on the population of cyanobacteria achieved a sorting rate of 2 cells/s and an accuracy of 96.3%. The fluidic network of the RACS platform was designed based on lumpedelement modeling (Fig. 4E) so that the flow velocity in the detection chamber of 100 μm/s was ensured for the required Raman signal integration time of 100 ms. At a speed lower than 500 μm/s in the detection channel (Fig. 4E), flow should be stable for continuous RACS. In low flow rate regimes, the pressure drop across the detection region ΔPd was often close to or less than the pressure fluctuation e intrinsic to the commonly available pressure pumps. To make the pressure switching mechanism feasible, “microfluidic pressure dividers” were employed to protect regional flow stability (Fig. 4D). When two parallel large resistors with Radd were in series with a small resistor with Rd, any variation in the pressure e was distributed across each individual resistance in proportion to its value. The pressure fluctuation in the detection region ed was minimized, and ed became 0 at Rd  Radd. In contrast, ed was equal to e at Radd = 0. The relationship Rd  Radd

5

Single-Cell Manipulation

125

achieved less disturbed transport of samples to the detection point regardless of pressure variations elsewhere in the system.

Integration with Other Methods More functional manipulation of single cells is desired for gaining active control, better specificity, and higher degree of freedom than microfluidics-based manipulation. Such improvement can be achieved by integrating the microfluidic manipulation technique with other techniques such as optical, dielectrophoretic (DEP), magnetic, and acoustic approaches. Though such integration requires an additional external field to facilitate the process, which complicates the system, the benefits compensate for the increased complexity of the system. This section discusses integration of fluidic manipulation with optical, DEP, magnetic, and acoustic manipulation.

Optical Integration Wang et al. used optical force-based switching for fluorescence-activated cell sorting (FACS) in a microchannel (Wang et al. 2005). This sorting method has advantages of a high switching rate, non-contact nature, high throughput, and suitability for complex fluidic network. In their device, parental and green fluorescent protein (GFP)-expressing HeLa cells were aligned to the channel center by flow focusing with two perpendicular buffer streams (Fig. 5A). Then, the cells passed the analysis region where a visible wavelength laser (488 nm) and a photodiode detected the presence of the cell in the analysis region (Fig. 5B). A photomultiplier tube (PMT) measured the fluorescence signal from the cells and passed it to an acousto-optic modulator. This modulator controlled activation of the infrared (IR) laser, whereas optical forces manipulated the cells in the switching region. The IR laser was focused with a relatively low numerical aperture of 0.2 to facilitate deflection. A microfluidic cartridge was used to prevent cell losses even for small initial cell populations in the range of 5–25 μL. Figure 5C shows bright and fluorescent images of the collection chamber along with those of the waste chamber. This method achieved the highest throughput rate of 106 cells/s with a maximum recovery rate of 90%. Chen et al. employed pulsed laser irradiation to detach isolated and monitored single cells for retrieval (Fig. 5D) (Chen et al. 2017). To isolate single cells for lineage tracking and division monitoring prior to retrieval, they adapted their previously developed single-cell hydrodynamic capture scheme (see section “Co-culture of Single Cells”) and set flow rates to Qbypass < Qtrap (Chen et al. 2014, 2016). Figure 5D shows the schematic of the developed single-cell capture chamber. MDA-MB-231 and T47D breast cancer cells were loaded from the inlet by gravity-driven flow. Each microchamber had two fluidic paths: a central path and a serpentine path. Initially, the central path had a lower flow resistance so that the first cell entering the chamber tended to flow through the central path. The opening

126

A

R. Bhardwaj et al. Sample input

B

C

PMT Filter (525/40)

Buffer

Buffer

488-nm laser Analysis region Optical switch

IR laser

AOM Microfluidic cartridge

Output wells Sample input

Photodiode Waste

D

Collection

Sheath reservoir

Filter (488/10)

E

Fig. 5 Integration of microfluidic single-cell manipulation with optical methods. (A) Layout of the microfluidic device and (B) schematic of the microfluidic cell sorter instrument using optical force switching. (C) Bright-field (left) and fluorescent images (right) in the collection chamber (top) and the waste well (bottom). ((A–C) are reproduced with permission from Wang et al. (2005). Copyright (2005) Springer Nature.) (D) Schematic diagram of pulsed laser irradiation-based single-cell detachment. (Top) Cells were cultured in the microfluidic chamber coated with the CNT–PDMS composite. A short pulsed laser was used to detach the target cell. (Bottom) Hydrodynamic singlecell capture scheme in the microfluidic chamber. (E) Selective single-cell detachment by optical generation of shear forces on the CNT–PDMS film. Scanning electron microscope image of the CNT grown on quartz substrate (top left) and after spin-coating with PDMS (top right, scale bars: 5 μm). Example of full (middle) and partial detachment (bottom) of an MDA-MB-231 cell (scale bar: 50 μm). ((D, E) are reprinted with permission from Chen et al. (2017). Copyright (2017) American Chemical Society)

of the central path (10 μm wide and 15 μm high) captured cells larger than the opening. The cells that followed then preferentially flowed through the serpentine path and were captured in the downstream microchambers. The two types of cells were captured at high cell capture rates (>70%). This chip integrated the detachment mechanism with the single-cell capture design: the chip included either a film of a carbon nanotube and PDMS composite or 20 nm sputtered Au/Pd alloy film, as a light-absorbing layer, and focused a

5

Single-Cell Manipulation

127

nanosecond laser pulse on trapped cells to detach them by generating a high shear force. Figure 5E illustrates the detachment process of a single MDA-MB-231 cell. Initially, the cell was captured in the chamber and allowed to adhere onto the substrate for 24 h. Then, the cell was detached by the optically generated shear force. Driven by the reversed gravity flow, the detached cell traveled upward toward the outlet for retrieval. When the irradiation was focused to one side of a cell, the cell was partially detached, leaving one side anchored and the other free.

Dielectrophoretic Integration The working principle of dielectrophoresis is that when a cell is placed in a non-uniform electric-field, the field creates forces of different magnitudes on the sides of cells depending on the electric polarizability of the medium and the cell. When the polarizability of cells is greater than the medium, cells are polarized opposite to the direction of the electric field, and DEP forces are created in the direction of the field gradient. Hence, cells are drawn to the strong electric field strength. Two examples employing DEP to separate cells are given below. The first example sorts E. coli cells from inactive cells, and the second one focuses on the separation of cells using DEP gating and an open microwell. Baret et al. reported an efficient microfluidic fluorescence-activated droplet sorter (FADS) (Baret et al. 2009) (Fig. 6A). Single E. coli cells in emulsion droplets were sorted using DEP in a fluorescence-activated manner, like that used for FACS. Mixtures of E. coli cells, expressing either the reporter enzyme β-galactosidase or an inactive variant, were compartmentalized with a fluorogenic substrate and then sorted at rates of 300 droplets/s. The droplets traveled to the sorting junction of the device at a velocity of ~0.2 m/s. At the sorting junction, the main channel branched into a narrower channel (40 μm wide; positive arm) closer to the electrodes, and a wider channel (60 μm wide; negative arm) farther from electrodes (Fig. 6B). The difference in hydraulic resistance between the two arms favored the flow of liquid along the wider channel. In the absence of an electric field, therefore, all of the droplets followed the main stream along the negative arm. A chosen particular droplet was sorted by applying a pulse of high-voltage alternating current (AC) across the electrodes adjacent to the sorting junction. The resulting electric field deflected the droplet of interest into the positive arm by DEP. At the above voltage threshold (1–1.4 kV peak to peak), the DEP force was larger than the fluidic forces maintaining the droplet in its flow line, and thus the droplet was pulled across the flow lines and flowed into the narrow arm of the sorter. The inverted open microwell is a microstructure supporting the isolation and trapping of cells, analysis of cell-to-cell and cell-to-molecule interactions, and functional cell sorting (Fig. 6C–E) (Bocchi et al. 2012). An open microwell is defined as a well open at both its upper and lower ends, and the design and workflow of the inverted open microwell array is presented in Fig. 6C. K562 leukemia cells were isolated in microwells fabricated on a flexible printed circuit board. During the delivery phase, DEP was used to control cell access to the microwell and to force the

128

R. Bhardwaj et al.

A

B Q

RPositive Sort RNegative Waste RNegative < RPositive QPositive < QNegative

C

E

D

QS 2–24 µL min–1 Peristaltic Pump

Δ p = QSR

R Well 3.76 × 1011 Pa s m–3

Outlet rgh 0.5 hPa

Fig. 6 Integration of the microfluidic single-cell manipulation with dielectrophoretic (DEP) methods. (A) Microfluidic fluorescence-activated droplet sorter (FADS) with electrodes for DEP. (Reproduced with permission from The Royal Society of Chemistry (Baret et al. 2009).) (B) Fluidic circuit of FADS. (C) Workflow of the inverted open microwell system, where different DEP configurations were used to block the access of cells to the microwell (BLOCK) or to allow cell delivery and grouping (LOAD). In the OFF mode, the electric field was deactivated. (D) K562 cells were trapped on the meniscus of an inverted open microwell. Two and three cells were paired or grouped by DEP methods (top). Multiple cells delivered without DEP activation were randomly positioned on the microwell meniscus (top right). Time-lapse images of a single microwell containing two cells during on-chip calcein-AM staining (middle). Fluorescence intensity of eight cells trapped in different microwells during on-chip calcein-AM staining, where each curve corresponds to a single cell. Proliferation of a single K562 cell after being transferred onto a well of a microtiter plate from an open microwell (bottom). (E) Microfluidic circuit of the inverted open microwell system. ((C, D) are reproduced with permission from The Royal Society of Chemistry (Bocchi et al. 2012))

formation of cell aggregates so as to ensure cell-to-cell contact and interaction. Cells were trapped at the air-fluid interface at the bottom edge of the open microwell (Fig. 6D Top). Trapped cells were retained on the meniscus even after DEP

5

Single-Cell Manipulation

129

deactivation, and fluid was exchanged to enable perfusion of nutrients and the delivery of molecules to the microwell. Cell viability was assessed on trapped cells by a calcein-AM assay (Fig. 6D Middle). Cell proliferation was demonstrated after multiple cells were recovered in parallel onto standard microtiter plates (Fig. 6D Bottom). Bocchi et al. calculated the pressure resistance for preventing water leakage from the wells by using the hydraulic–electric circuit analogy. The circuit model of the open inverted microwell system is presented in Fig. 6E. When a fluid was injected in the microchannel, the microwell was filled by surface tension, the pressure was generated in the microchannel, and a meniscus was formed at the open extremity of the well. The diameter of the microwell (75 μm) was sufficiently small, and the surface tension at the air-fluid interface was strong enough to counteract both the gravity force and the pressure generated in the microchannel. The right hand side of Eq. (15) expresses the maximum pressure difference applied on the meniscus at the air-liquid interface: ρgh þ Q

12μL sin θ   < 2γ 0:63h r wh 1  w 3

ð15Þ

where γ is the surface tension at the air-liquid interface, θ is the contact angle between the liquid and the chip substrate, and r is the microwell radius. Then, the maximum pressure tolerated by this interface was calculated using Eq. (15). When γ was assumed to be 0.072 N/m and θ of the polyimide surface on the bottom side of the microwell to fall between 31 and 53 , the maximum pressure tolerated by the meniscus (2γ sin θ/r) ranged between 19.8 hPa and 30.7 hPa. The output fluid accumulated at the fluid outlet had a height of h  5 mm, which corresponded to a hydraulic head pressure of about 50 Pa. Microchannels were designed to produce a limited pressure loss at a maximum flow rate. The hydraulic resistance Rch was 3.76  1011 Pas/m3, and the flow rate Q was 2–24 μL min1, which corresponded to a pressure drop QRch between 13 Pa and 150 Pa. Hence, in the worst case, the total pressure in the microwell was the sum of 50 Pa and 150 Pa, which was consistently lower than the maximum pressure tolerated at the interface. Their experiments confirmed the absence of fluid leakage within these ranges.

Magnetic Integration The magnetic force works on the principle of an affinity of a certain element, material, or biomaterial for a magnetic field. Cells can have intrinsic (e.g., erythrocytes (RBC)) or extrinsic magnetic moments created through external means, and this magnetic moment can be deflected by using a magnetic field. Two magnetic separation methods are discussed below. A multi-target magnetic activated cell sorter (MT-MACS) was employed to separate two different types of cell population from a mixture having different magnetic behaviors (Adams et al. 2008). Adams et al. used two different magnetic tags having different radii and magnetizations: Tag 1 has magnetization

130

R. Bhardwaj et al.

M = 14 kA/m, and Tag 2 has M = 30 kA/m with specific surface markers (Fig. 7a, Step A). A microfabricated ferromagnetic strip (MFS) was placed at two different locations to separate the cells (Fig. 7a, Step B). MFS arrays were fabricated inside the microchannel to create a short-range, high magnetic field gradient and to cause deflection of the path of the cells. A microscopic view of the strips is shown in Fig. 7b. Hydrodynamic focusing was used to align cells at the bottom of channel. The cells did not stick to the surface and were transported at a total flow rate of 47 mL/h with a sample flow rate of 5 mL/h and a buffer flow rate of 42 mL/h. The difference between the magnetic force and fluidic force caused the cells of interest to deflect in the required direction (Fig. 7b). The external magnet placed beneath the device created a long-range magnetic field gradient and thus pulled the magnetic particle toward the bottom plane of the channel and magnetized tags and MFS. The recovery rates were 91.6% for target 1 cells labeled by Tag 1, 93.9% for target 2 cells labeled by Tag 2, and 99.6% for nontarget cells. This technique can be extended to sort more than two different types of cells by manipulating magnetic tags and MFS angles and by using many units of the MFS arrays arranged in parallel to achieve high throughput. Fachin et al. incorporated magnetic sorting and two different microfluidic technologies for rapid, high-throughput negative selection of circulating tumor cells (CTC) (Fachin et al. 2017) (Fig. 7c–e). Flow was completely controlled using on-chip fluidic resistors and a single pressure source. The overall symmetrically parallelized chip architecture integrated the following five microfluidic stages as schematically represented in Fig. 7c: deterministic lateral displacement (DLD), inertial focusing stage 1 (IF1), magnetically activated cell sorting stage 1 (MACS1), IF2, and MACS2. An automated monolithic chip had 128 multiplexed DLD devices containing ~1.5 million microfabricated features (12–50 μm) to deplete red blood cells and platelets. The IF units were added to line up all the nucleated cells for multistage magnetophoresis and to remove magnetically labeled white blood cells. The monolithic CTC-iChip purified CTCs from the blood cell population (Fig. 7d) and enabled debulking of blood samples at 15–20  106 cells/s while yielding an output of highly purified CTCs. Fachin et al. quantified the size and epithelial cell adhesion molecule (EpCAM) expression of over 2,500 CTCs from 38 patient samples, which were obtained from breast, prostate, and lung cancers and melanoma. In the study by Fachin et al., understanding the cascading effects of device dimensions was critical to device operation because all the fluidics were integrated and controlled solely by the device design and pressure applied at the blood and buffer inlets. As such, a lumped resistor model of the fluidic system was developed as presented in Fig. 7e. Each model resistor represents several single components in the actual device. The actual resistance of the DLD waste resistor was determined based on a combination of analytical approximations, finite element models, and empirical observations. Owing to the absence of an analytical model for wiggler-like features, the IF1 and IF2 resistances were estimated based on simulation and test structures, of which the pressure drops were confirmed via experimental observation of relative volumetric outputs from the DLD waste resistor and IF stages.

5

Single-Cell Manipulation

131

A

B

C

D

E R1A Buffer Blood

–+ –+

R2A RDLD

DLD Waste

RWaste

MACS Waste

R8A

RIF2, MACS2

RIF1, MACS1 RIF2, MACS2 R6A

R9A

R8B

MACS Waste

R9B MACS Waste

Product

Fig. 7 Integration of microfluidic single-cell manipulation with magnetic methods. (a) Multi-target magnetic activated cell sorting (MT-MACS). (b) Optical micrographs of the tags being separated at the two microfabricated ferromagnetic strip (MFS) structures at a total flow rate of 47 mL/h. ((a, b) are reprinted with permission from Adams et al. (2008). Copyright (2008) National Academy of Sciences, U.S.A.) (c) Schematics of magnetic-sorting-based negative selection of circulating tumor cells. (d) Resistor network of the fluidic system in (c). (e) WBCs and CTCs entering the inertial focusing channel were aligned in the channel exit (left). MACS separated CTCs (yellow) and WBCs (green) (right). ((c–e) are reproduced with permission from Fachin et al. (2017))

132

R. Bhardwaj et al.

The resistances for MACS 1, MACS 2, and all other IFD waste, product, and connecting channels were analytically calculated. The lumped model enabled analyzing and visualizing the interdependencies between system subcomponents.

Acoustic Integration Sound waves exert acoustic forces on any particles present in their region of influence. There are three main types of waves used to manipulate cells: bulk standing waves, standing surface acoustic waves, and traveling surface acoustic waves (Wyatt Shields et al. 2015). This section illustrates the integration of hydraulic and acoustic manipulation. In the device developed by Petersson et al., suspended particles or cells were injected through two side channels while a medium flowed from the central channel, and free-flow acoustophoresis was achieved for fractionation of the particles (Fig. 8A) (Petersson et al. 2007). A half-wavelength acoustic standing wave was generated between the side walls of the channel using a piezo ceramic actuator. This wave generated an acoustic force field perpendicular to the direction of flow. Cesium chloride (CsCl; concentration: 0.22 g/mL) was added to make the medium denser and to change the contrast factor of the acoustic forces. Samples were separated while flowing through the acoustic field (Fig. 8B). These focused streams of particles were further fractionated downstream using outlets (Fig. 8C). The device could separate particles with different sizes and particles of the same size with different densities. 92% of the red blood cells (RBCs) and 99% of the platelets were separated (Fig. 8D). Ohiri et al. demonstrated a hybrid microfluidic system that combined fluidic trapping and acoustic switching to organize an array of single cells at high density (Fig. 8E, F) (Ohiri et al. 2018). The fluidic trapping step was achieved by balancing the hydrodynamic resistances of three parallel channel segments forming a microfluidic trifurcation. They optimized the hydrodynamic trapping step by tuning the three volumetric flow rates, Qtrap, Qbypass, and Qcomp, such that an unoccupied trap had the lowest fluidic resistance, whereas an occupied trap had higher fluidic resistance than the bypass channel. This design ensured that after a single cell was trapped, subsequent cells were diverted toward the bypass channel and trapped in the next unoccupied trap. This process allowed the traps across the entire chip to be loaded within minutes. The major parameters of the device fabricated by Ohiri et al. were Qtrap/Qbypass  2.4 and Qbypass/Qcomp  3.6. The condition of Rbypass/Rtrap > 2 was maintained by setting the width and length of the trap to be 6 μm and 4 μm, respectively, and the length of the bypass channel to over 1 mm. The compartment section was designed to achieve a resistance ratio of Rcomp/Rbypass > 2 with a purpose of biasing most fluid flow toward the bypass segment. 15-μm-diameter polystyrene beads were injected into the device and flowed through the chip at a flow rate of 50 μL/min. PC9 cancer cells and beads were hydrodynamically loaded into the traps and then intentionally transferred into the compartment region when an acoustic force was activated with piezoelectric transducers. This approach relied on a

5

Single-Cell Manipulation

133

Fig. 8 Integration of the microfluidic single-cell manipulation with acoustic methods. (A) Schematic of particle suspension passing over the transducer for acoustophoresis. (B) Crosssectional illustration of a separation channel showing separated particles. (C) Fractionation of the separated particles at the end of the separation channel. (D) Separation of platelets and red cells with and without CsCl (0.22 g/mL) added to the suspending medium. ((A–D) are reprinted with permission from Petersson et al. (2007). Copyright (2007) American Chemical Society.) (E) Acoustofluidic chip for single-cell array. Image and schematic of the entire chip (left). Image of individual acoustofluidic element comprised of a trap, bypass, and compartment region (right). Scale bar: 100 μm. (F) Image sequence detailing the acoustic switching mechanism. Beads were loaded in the compartment region. Scale bar: 200 μm. ((E, F) are reproduced with permission from The Royal Society of Chemistry (Ohiri et al. 2018))

combination of hydrodynamic capture of cells in traps and subsequent transfer of the cells into more spacious compartment chambers using an array of acoustic streaming vortices as local switches. An array of single PC9 cells was generated in compartments with an arraying efficiency of ~67%.

134

R. Bhardwaj et al.

Conclusion This chapter discussed fluidic single-cell manipulation based on flow modeling, and the physics of Hagen–Poiseuille flow was introduced for single-cell manipulation. Hydrodynamic single-cell manipulation has a potential for higher throughput with performance prediction based on the hydraulic–electric circuit analogy. Integration of several techniques (i.e., optical, DEP, magnetic, and acoustic approaches) with the hydrodynamic approach was also described. Comprehensive results were obtained for either single-cell manipulation or single-cell analysis by integrating hydraulic and other methods into a microfluidics system. Although modeling of the fluidic system has been proved to be useful and successful, small number of studies employed the modeling approach, especially for integration. Most hydrodynamic manipulation is independent of optical, DEP, magnetic, and acoustic forces, and it can be modularized and designed independently from the other forces. Modeling of the fluidic part allows to predict flow and focus on target functions using different forces. This chapter helps understanding the essential parts of microfluidic singlecell manipulation. Near future will witness development in the field of single-cell analysis based on hydrodynamic manipulation and modeling.

References Adams JD, Kim U, Soh HT (2008) Multitarget magnetic activated cell sorter. Proc Natl Acad Sci USA 105(47):18165–18170. https://doi.org/10.1073/pnas.0809795105 Ajdari A (2004) Steady flows in networks of microfluidic channels: building on the analogy with electrical circuits. C R Phys 5(5):539–546. https://doi.org/10.1016/j.crhy.2004.02.012 Avior Y, Sagi I, Benvenisty N (2016) Pluripotent stem cells in disease modelling and drug discovery. Nat Rev Mol Cell Biol 17:170. https://doi.org/10.1038/nrm.2015.27 Baret J-C, Miller OJ, Taly V, Ryckelynck M, El-Harrak A, Frenz L, Rick C, Samuels ML, Hutchison JB, Agresti JJ, Link DR, Weitz DA, Griffiths AD (2009) Fluorescence-activated droplet sorting (FADS): efficient microfluidic cell sorting based on enzymatic activity. Lab Chip 9(13):1850–1858. https://doi.org/10.1039/B902504A Bocchi M, Rambelli L, Faenza A, Giulianelli L, Pecorari N, Duqi E, Gallois J-C, Guerrieri R (2012) Inverted open microwells for cell trapping, cell aggregate formation and parallel recovery of live cells. Lab Chip 12(17):3168–3176. https://doi.org/10.1039/C2LC40124J Chen J, Li Y, Yu T-S, McKay RM, Burns DK, Kernie SG, Parada LF (2012) A restricted cell population propagates glioblastoma growth after chemotherapy. Nature 488(7412):522–526. https://doi.org/10.1038/nature11287 Chen Y-C, Cheng Y-H, Kim HS, Ingram PN, Nor JE, Yoon E (2014) Paired single cell co-culture microenvironments isolated by two-phase flow with continuous nutrient renewal. Lab Chip 14(16):2941–2947. https://doi.org/10.1039/C4LC00391H Chen Y-C, Ingram PN, Fouladdel S, McDermott SP, Azizi E, Wicha MS, Yoon E (2016) Highthroughput single-cell derived sphere formation for cancer stem-like cell identification and analysis. Sci Rep 6:27301. https://doi.org/10.1038/srep27301 Chen Y-C, Baac HW, Lee K-T, Fouladdel S, Teichert K, Ok JG, Cheng Y-H, Ingram PN, Hart AJ, Azizi E, Guo LJ, Wicha MS, Yoon E (2017) Selective photomechanical detachment and retrieval of divided sister cells from enclosed microfluidics for downstream analyses. ACS Nano 11(5):4660–4668. https://doi.org/10.1021/acsnano.7b00413 Fachin F, Spuhler P, Martel-Foley JM, Edd JF, Barber TA, Walsh J, Karabacak M, Pai V, Yu M, Smith K, Hwang H, Yang J, Shah S, Yarmush R, Sequist LV, Stott SL, Maheswaran S,

5

Single-Cell Manipulation

135

Haber DA, Kapur R, Toner M (2017) Monolithic chip for high-throughput blood cell depletion to sort rare circulating tumor cells. Sci Rep 7(1):10936. https://doi.org/10.1038/s41598-01711119-x Frimat JP, Becker M, Chiang YY, Marggraf U, Janasek D, Hengstler JG, Franzke J, West J (2011) A microfluidic array with cellular valving for single cell co-culture. Lab Chip 11(2):231–237. https://doi.org/10.1039/c0lc00172d Gallaher JA, Enriquez-Navas PM, Luddy KA, Gatenby RA, Anderson ARA (2018) Spatial heterogeneity and evolutionary dynamics modulate time to recurrence in continuous and adaptive cancer therapies. Cancer Res 78(8):2127–2139. https://doi.org/10.1158/0008-5472.Can-17-2649 Gao D, Jin F, Zhou M, Jiang Y (2019) Recent advances in single cell manipulation and biochemical analysis on microfluidics. Analyst 144(3):766–781. https://doi.org/10.1039/C8AN01186A Hayt WH, Kemmerly JE, Durbin SM (2012) Engineering circuit analysis. McGraw-Hill, New York Hong S, Pan Q, Lee LP (2012) Single-cell level co-culture platform for intercellular communication. Integr Biol 4(4):374–380. https://doi.org/10.1039/C2IB00166G Hu S, Lam RHW (2017) Characterization of viscoelastic properties of normal and cancerous human breast cells using a confining microchannel. Microfluid Nanofluid 21(4):68. https://doi.org/ 10.1007/s10404-017-1903-x Hu S, Liu G, Chen W, Li X, Lu W, Lam RHW, Fu J (2016) Multiparametric biomechanical and biochemical phenotypic profiling of single cancer cells using an elasticity microcytometer. Small 12(17):2300–2311. https://doi.org/10.1002/smll.201503620 Huang N-T, Zhang H-L, Chung M-T, Seo JH, Kurabayashi K (2014) Recent advancements in optofluidics-based single-cell analysis: optical on-chip cellular manipulation, treatment, and property detection. Lab Chip 14(7):1230–1245. https://doi.org/10.1039/C3LC51211H Kim J, Erath J, Rodriguez A, Yang C (2014) A high-efficiency microfluidic device for size-selective trapping and sorting. Lab Chip 14(14):2480–2490. https://doi.org/10.1039/C4LC00219A Kobel S, Valero A, Latt J, Renaud P, Lutolf M (2010) Optimization of microfluidic single cell trapping for long-term on-chip culture. Lab Chip 10(7):857–863. https://doi.org/10.1039/ b918055a Li S-S, Cheng C-M (2013) Analogy among microfluidics, micromechanics, and microelectronics. Lab Chip 13(19):3782–3788. https://doi.org/10.1039/C3LC50732G Luo T, Fan L, Zhu R, Sun D (2019) Microfluidic single-cell manipulation and analysis: methods and applications. Micromachines 10(2):104 McIlvenna D, Huang WE, Davison P, Glidle A, Cooper J, Yin H (2016) Continuous cell sorting in a flow based on single cell resonance Raman spectra. Lab Chip 16(8):1420–1429. https://doi. org/10.1039/C6LC00251J Munson BR, Okiishi TH (2009) Fundamentals of fluid mechanics. Wiley, Hoboken Nagai M, Tanizaki K, Shibata T (2019) Batch assembly of SU-8 movable components in channel under mild conditions for dynamic microsystems: application to biohybrid systems. J Microelectromech Syst 28(3):419–428. https://doi.org/10.1109/JMEMS.2019.2907285 Nilsson J, Evander M, Hammarström B, Laurell T (2009) Review of cell and particle trapping in microfluidic systems. Anal Chim Acta 649(2):141–157. https://doi.org/10.1016/j.aca.2009. 07.017 Oh KW, Lee K, Ahn B, Furlani EP (2012) Design of pressure-driven microfluidic networks using electric circuit analogy. Lab Chip 12(3):515–545. https://doi.org/10.1039/C2LC20799K Ohiri KA, Kelly ST, Motschman JD, Lin KH, Wood KC, Yellen BB (2018) An acoustofluidic trap and transfer approach for organizing a high density single cell array. Lab Chip 18(14): 2124–2133. https://doi.org/10.1039/C8LC00196K Perdigones FA, Luque A, Quero JM (2014) Correspondence between electronics and fluids in MEMS: designing microfluidic systems using electronics. IEEE Ind Electron Mag 8(4):6–17. https://doi.org/10.1109/MIE.2014.2318062 Petersson F, Åberg L, Swärd-Nilsson A-M, Laurell T (2007) Free flow acoustophoresis: microfluidic-based mode of particle and cell separation. Anal Chem 79(14):5117–5123. https://doi.org/10.1021/ac070444e Rapp BE (2017) Chapter 17 – Hydraulic resistance. In: Microfluidics: modelling, mechanics and mathematics. Elsevier, Oxford. https://doi.org/10.1016/B978-1-4557-3141-1.50017-4

136

R. Bhardwaj et al.

Ryu S, Pepper RE, Nagai M, France DC (2017) Vorticella: a protozoan for bio-inspired engineering. Micromachines 8(1):4 Sauzade M, Brouzes E (2017) Deterministic trapping, encapsulation and retrieval of single-cells. Lab Chip 17(13):2186–2192. https://doi.org/10.1039/c7lc00283a Senturia SD (2007) Microsystem design. Springer Science & Business Media, New York Shinde P, Mohan L, Kumar A, Dey K, Maddi A, Patananan AN, Tseng F-G, Chang H-Y, Nagai M, Santra TS (2018) Current trends of microfluidic single-cell technologies. Int J Mol Sci 19(10):3143 Sitti M, Ceylan H, Hu W, Giltinan J, Turan M, Yim S, Diller E (2015) Biomedical applications of untethered mobile milli/microrobots. Proc IEEE 103(2):205–224. https://doi.org/10.1109/ JPROC.2014.2385105 Tan W-H, Takeuchi S (2007) A trap-and-release integrated microfluidic system for dynamic microarray applications. Proc Natl Acad Sci USA 104(4):1146–1151. https://doi.org/10.1073/ pnas.0606625104 Tilmans HAC (1996) Equivalent circuit representation of electromechanical transducers: I. Lumped-parameter systems. J Micromech Microeng 6(1):157–176. https://doi.org/10.1088/ 0960-1317/6/1/036 Wang MM, Tu E, Raymond DE, Yang JM, Zhang H, Hagen N, Dees B, Mercer EM, Forster AH, Kariv I, Marchand PJ, Butler WF (2005) Microfluidic sorting of mammalian cells by optical force switching. Nat Biotechnol 23(1):83–87. https://doi.org/10.1038/nbt1050 Wyatt Shields C IV, Reyes CD, Lopez GP (2015) Microfluidic cell sorting: a review of the advances in the separation of cells from debulking to rare cell isolation. Lab Chip 15(5):1230–1249. https://doi.org/10.1039/C4LC01246A Yun H, Kim K, Lee WG (2013) Cell manipulation in microfluidics. Biofabrication 5(2):022001. https://doi.org/10.1088/1758-5082/5/2/022001 Zhang K, Chou C-K, Xia X, Hung M-C, Qin L (2014) Block-cell-printing for live single-cell printing. Proc Natl Acad Sci USA 111(8):2948–2953. https://doi.org/10.1073/pnas.1313661111

6

Single Cell Manipulation Using Macro-scale Actuator Chia-Hung Dylan Tsai

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transfer Function of Macro-to-Micro Manipulation (Mizoue et al. 2017) . . . . . . . . . . . . . . . . . . . . The Driving Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . System Identification Through Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . On-Chip Transmitter for Enhancing Manipulation Speed (Monzawa et al. 2015) . . . . . . . . . . . . Advantages of Using an Actuation Transmitter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanical Model and Theoretical Basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental Validation on Manipulation with an Actuation Transmitter . . . . . . . . . . . . . . . . . Recent Works and Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

138 140 140 141 143 146 146 147 149 152 153 154

Abstract

Single cell manipulation is one of the key methods in single cell technologies. A cell can be positioned to a designation at a specified velocity for performing tasks, such as cell evaluation and cell sorting. This chapter is started with the state-of-the-art technologies of single cell manipulation, which include manipulations with both micro-scale and macro-scale actuators. Manipulation with a macro-scale actuator has advantages of low-cost and easy-access, but is challenging to control due to scale difference of several orders. Therefore, the introduction will be followed by a review of single cell manipulations using a macro-scale actuator. Transfer function of the macro-to-micro manipulation is derived based on a mechanical model incorporating the deformation of microfluidic chip. The coefficients of the transfer function are determined based on experimental results. An advanced manipulation system with closed-loop control and an on-chip C.-H. D. Tsai (*) Department of Mechanical Engineering, National Chiao Tung University, Hsinchu, Taiwan e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_23

137

138

C.-H. D. Tsai

transmitter is presented with experimental results. Finally, applications on cell evaluation using macro-scale actuator will be discussed at the end of the chapter.

Introduction Single cell manipulation provides the capability of controlling motion of single cells in a microfluidic environment. For example, Sakuma et al. (2014a) perform fatigue test of single cells by moving a target cell back and forth through a microfluidic constriction until the cell loses its deformability. Other than cell evaluation, single cell manipulation also benefits different cell-based researches, such as cell sorting and (Sakuma et al. 2017) cell cultivation (Horade et al. 2019). Figure 1 shows a setup of manipulation system and an example of cell manipulation (Monzawa et al. 2015). The setup shown in Fig. 1a includes a microfluidic chip and a macro-scale actuator. A series of continuous photos of the cell moving along a given sinusoidal positions in a microfluidic channel are shown in Fig. 1b. The maximum of the position error in Fig. 1b is less than 1 μm in this case, which demonstrates a fairly well control of the cell position in the order of micrometer. Different approaches have been developed for single cell manipulation, and based on the types of actuation they can be categorized into manipulations with microscale actuators and with macro-scale actuators. The definitions of the micro-scale

Fig. 1 An example of cell manipulation with a macro-scale actuator. (a) The arrangement of the microfluidic chip and the macro-scale actuator. (b) Continuous image frames of a red blood cell under manipulation. The image in Fig.1b is reproduced from [T. Monzawa, M. Kaneko, C. D. Tsai, S. Sakuma and F. Arai: On-Chip Actuation Transmitter for Enhancing the Dynamic Response of Cell Manipulation using a Macro-scale Pump, Biomicrofluidics, vol.9, no.1, 014114, 2015], with the permission of AIP Publishing

6

Single Cell Manipulation Using Macro-scale Actuator

139

actuators and the macro-scale actuators in this chapter are based on the physical size of the actuator. The micro-scale actuators are those built up using modern micro/ nano fabrication methods, such as photolithography and deep reactive-ion etching (DRIE), while the macro-scale actuators are those commercially available, such as motors, piezoelectric (PZT) actuators and a syringe pump. For example, single-cell manipulation using droplets can be performed using a macro-scale syringe pump as well as a micro-scale acoustic device. Cell manipulations with optical tweezers is also a popular approach for single cell studies (Lim et al. 2004; Dao et al. 2003). Since the force is applied at micro-scale points, optical tweezers are categorized as micro-scale manipulation in this chapter. There are works on single cell manipulation with micro-scale actuators. For example Avci et al. (2015) proposed a microhand with two fingers for grasping and manipulating single cells. Chronis and Lee (2005) developed electrothermally activated SU8 microgripper for single cell manipulation. Yalikun et al. (2016) proposed a hydrodynamic approach that create a circulation zone by a fabricated orifice for rotating single cells. Din et al. (2012) manipulated single cells using surface acoustic waves. Zhang and Liu (2008) summarized works of optical tweezers for single cell manipulation. Although micro-scale for micro-scale manipulation can avoid the challenging of control in order to overcome the difference in scales, it has drawback of nonreusable and difficulty in fabrication. In general, it is challenging to directly manipulate a micro-scale object using a macro-scale actuator due to the difference of scale. Figure 2 illustrates the challenges of cell manipulation using a macro-scale actuator, compared with the manipulation with a micro-scale actuator. When the manipulation is performed with a micro-scale actuator, the motion of the piston and the target cell are in the same order of the magnitude, as illustrated in Fig. 2a. That is, when the piston is moved to the right for 1 μm, the cell would also be moved to the right for 1 μm. When the manipulation is performed with a macro-scale actuator, such as a commercial syringe pump, the motions of the syringe piston and the cell are in different orders of the magnitude. As an example illustrated in Fig. 2b. If we assume that the cross-sectional areas of the syringe and the microfluidic channel are 100 mm2 and 100 μm2, respectively, the movement of 1 μm at the actuation piston in Fig. 2b would lead to approximately 1 m of displacement for the target cell due to incompressibility of liquid. The displacement from the input to the output is one million times amplified due to the scale difference. Such a scale difference makes it almost impossible to manipulate a cell in the resolution of micrometer with a macro-scale actuator. However, there are many works of cell manipulation using a macro-scale actuator (Monzawa et al. 2015; Sakuma et al. 2014b; Tsai et al. 2017; Mizoue et al. 2016, 2017; Heo et al. 2015). It indicates that there must be certain mechanism for achieving such a fine manipulation and will be revealed in this chapter. The remaining of this chapter will cover the fundamental theory of cell manipulation with a macro-scale actuator, experimental results, enhancement for the manipulation speed, and examples of application on single cell manipulation with a macro-scale actuator.

140

C.-H. D. Tsai

Transfer Function of Macro-to-Micro Manipulation (Mizoue et al. 2017) The Driving Mechanism The driving mechanism of the cell manipulation is based on the assumptions of cell moving with the fluid and constant volumetric flow in a microfluidic channel. All the volume pushed by the piston from either a micro-scale actuator or a macro-scale actuator will result in a flow of the same volume in the microfluidic channel. The target cells are suspended and moved together with the fluid due to low Reynolds number, which is often less than 1 (Sakuma et al. 2014a; Monzawa et al. 2015). Therefore, the relation between input and output displacements can be determined by the ratio of the cross-sectional areas of the input and output chamber, as an example illustrated in Fig. 2b. There are two possibilities for changing the amount of displacement of the cell. One is the leakage in the system and the other is variation of the cross-sectional areas along the fluid pathway. For the leakage, microfluidic chip is usually bonded with plasma, which means a covalent bonding between the chip and the substrate, and it is unlikely having significant leakage. Thus, the variation of cross-sectional area, which is also known as channel deformation, becomes the main consideration and for the modeling.

Fig. 2 Illustrations cell manipulation with a micro-scale actuator and a macro-scale actuator. (a) Micro-scale actuator. (b) Macro-scale actuator

6

Single Cell Manipulation Using Macro-scale Actuator

141

Mechanical Model Figure 3 shows an overview of the manipulation system with a macro-scale actuator, as well as the mechanical model of it. There are three main components in the model, including a programmable macro-scale syringe, a deformation chamber, and a microchannel where the target cell is being manipulated. The macro-scale syringe is actuated by a piezoelectric actuator from the plunger top. The fluid in the syringe is pushed out and pulled in by the extension and retraction of the actuator, respectively. The deformation chamber and the microchannel are both fabricated on a microfluidic chip. The deformation chamber is inflated when the pressure in the chamber increases due to extra fluid is pushed in from the syringe, and is shrunk when the pressure decreases due to fluid is pulled away. For the mechanical model in Fig. 3b, the plunger movement and the cross-section change of the deformation chamber are modeled by two pistons. Based on the conservation of mass and assuming the fluid is incompressible, the relation of piston movements in the mechanical model can be written as follows: A1 x1 ¼ Ac xc þ A2 x2

ð1Þ

where A1, Ac, and A2 are the cross-sectional areas of the representative piston for the syringe, deformation chamber, and the microchannel, respectively, while x1, xc, and x2 are the displacements of them. Equation (1) indicates that the volume of the fluid injected from the syringe is equal to the sum of the volume of the deformation chamber and the volume of the fluid flowing in the microchannel. Since the target cell is suspended in the fluid, x2 is also the displacement of the cell. The model in Fig. 3b has three degrees of freedom and the equation of motion for the piston of deformation chamber and the cell displacement can be written as follows:

Fig. 3 Overview of the manipulation system. (a) The plunger displacement x1 transforms to the cell displacement x2. (b) The mechanical model for the derivation of the relation between x1 and x2. (Mizoue et al. 2017)

142

C.-H. D. Tsai

pAc ¼ M x€c þ cx_ c þ kxc

ð2Þ

pA2 ¼ m x€2 þ c2 x_ 2 þ k2 x2

ð3Þ

where p is the pressure in the deformation chamber. M, c, and k are the mass, damping constant, and stiffness constant for the piston of the deformation chamber. m, c2, and k2 are the mass, damping constant, and stiffness constant for the piston in the microchannel, respectively. x€c , x€2 , x_ c , and x_ 2 are the second and first derivatives of xc and x2 with respective to the time. The transfer function for the manipulation system, where the input and output are the plunger displacement x1 and the cell displacement x2 can be derived as follows: T:F: ¼

s2 þ 2ζ1 ω1 s þ ω21 X2 ¼α 2 X1 s þ 2ζ2 ω2 s þ ω22

ð4Þ

where X1 and X2 are the Laplace transforms of x1 and x2. s is an independent variable representing the frequency. The coefficients α, ζ1, ζ2, ω1, ω2 in Eq. (4) are derived from Eqs. (1), (2), and (3) as (Mizoue et al. 2017): α¼

A1 M 
2  A2 M þ AA2c m

c ζ1 ¼ pffiffiffiffiffiffiffi 2 Mk rffiffiffiffiffi k ω1 ¼ M
2 c þ AA2c c2 ζ2 ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
2 
2  Ac 2 M þ A2 m k þ AA2c k2 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 u u A u k þ A2c k2 ω2 ¼ u
2 t M þ AA2c m The transfer function in Eq. (4) is for displacement-based pumps that the input signal is the displacement of the plunger x1. For the transfer function with a pressurebased pump, the derivation would be much simpler and Eq. (3) will be the only governing equation. That means, the deformation chamber would not affect the motion of the cell in the microchannel, and the velocity of the cell can be solely determined by the applied pressure p and the design of the microchannel.

6

Single Cell Manipulation Using Macro-scale Actuator

143

System Identification Through Experiments Figure 4 shows the experimental setup for identifying the system coefficients in Eq. (4). Figure 4a illustrates the functional diagram of the system, where the input displacement x1 is moved along with a piezoelectric actuator. The controlling signals for the actuator are from a function generator through a piezo controller. A highspeed camera is used to monitor the motion of the particles in the microchannel, and the recorded videos are analyzed using image processing method for obtaining the displacement of the particles, as x2. Both the input signal from the function generator and the output signal from the high-speed camera are acquired by the computer for system identification.

Fig. 4 Experimental setup for cell manipulation from a macro-scale actuator. (a) The functional diagram of the experimental system. (b) A photo of the system. (Mizoue et al. 2017)

144

C.-H. D. Tsai

Figure 4b shows an actual photo of the system setup. The system consists of three units, including an input unit, a monitoring unit, and data processing unit. The input unit has a function generator for controlling the input signal to the piezoelectric actuator, a piezoelectric controller for amplifying the signal from the function generator to the actuator, a piezoelectric actuator for moving the plunger of the syringe, and a glass syringe for providing the displacement input of the plunger x1. Sinusoidal signals with the frequencies of 2 Hz, 5 Hz, 10 Hz, 20 Hz, 50 Hz, 100 Hz, 120 Hz, 150 Hz, 200 Hz, and 300 Hz are used for moving the syringe plunger. The monitoring unit in Fig. 4b is for observing the particle displacement in the microchannel. The unit includes a PDMS chip as the platform for particle flow, a microscope for observing the micro-scale motion on the chip, and a high-speed camera for tracking high-speed motion. The mold of the PDMS chip is fabricated with a standard photolithography process. The chip is cast from the mold with the mixture of PDMS and curing agent at the ratio of 9:1. The ratio is important because it determines the mechanical coefficients of the deformation chamber, as the parameters c, k, and M in Fig. 3. The high-speed camera is set at the rate of 1000 frames per second (fps) for recording the particle motion. The particles for the experiments are microbeads with the diameter of 1 μm instead of actual cells in order to avoid complications of cell deformation and adhesions. The data processing unit includes a data acquisition card and a computer. Both the input and output signals are acquired through the card and analysis in the computer. Figure 5 shows examples of manipulation results using the system shown in Fig. 4 with the input of sinusoidal signals at different frequencies. Figure 5a is a series of image frames of a moving microbead under the input of sinusoidal signal at 10 Hz. The dashed line connected cells across the frames demonstrates a sinusoidal trajectory of the cell displacement x2 with respect to the time. The microbead displacements x2 at the frequencies of 2 Hz, 5 Hz, 10 Hz, 20 Hz, 50 Hz, 100 Hz, 120 Hz, 150 Hz, 200 Hz, and 300 Hz are presented as the red curves in Fig. 5b–k, respectively. The blue curves in Fig. 5b–k are the motion of the piezoelectric actuator, as the input displacement x1. For the convenience of comparison, the amplitudes of blue and red curves in Figs. 5b–k are normalized by the maximum amplitude at 2 Hz. The frequency response of the experimental results shown in Fig. 5 is plotted in a bode diagram as shown in Fig. 6. Every points in Fig. 6 is determined based on the fast Fourier transform (FFT) of the signals in Fig. 5. According to the results in Fig. 6, the gain of the macro-to-micro manipulation is consistently decreasing with the increase of input frequency in the range between 2 Hz and 300 Hz. On the other hand, the phase difference between the input and output remains 90° in the range. The system coefficients in Eq. (4) can be identified using the results shown in Fig. 6. Here, two different approaches are employed for the identification. First approach is to determine the system by direct observation. It is noted that the frequency response of the macro-to-micro system is similar to the response of an integration operator, which has a constant decreasing rate in amplitude and the phase stays at 90°. Thus, the transfer function G(s) of the system can be assumed in the form of an integration operator, and the coefficients are determined based on the

6

Single Cell Manipulation Using Macro-scale Actuator

145

Fig. 5 Experimental results of the manipulation system with the macro-scale actuator. (a) An example of cell motion in continuous image frames. (b–k) The comparison between the actuator motion (blue) and normalized cell motion (red) at frequencies of 2 Hz, 5 Hz, 10 Hz, 20 Hz, 50 Hz, 100 Hz, 120 Hz, 150 Hz, 200 Hz, and 300 Hz, respectively. (Mizoue et al. 2017)

results in Fig. 6 where the gain follows –20 dB declining line and intersecting x-axis, as G ¼ 1, at s ¼ 63 by estimation. The identified transfer function by the direct observation from Fig. 6 is: G ðsÞ 

63 s

ð5Þ

where s represents input frequency and is in a complex form. The second approach for system identification is a more general approach that the algorithm of least-square for curve fitting is employed. The data points in Fig. 6 are plugged into the curve fitting with the target function of Eq. (4). The coefficients in Eq. (4) are identified from the best fit where the coefficient of determination R2 is 0.9993. In this particular case, the identified coefficients using curve fitting are

146

C.-H. D. Tsai

α ¼ 0:01872, ω1 ¼ 1321, ω2 ¼ 0:36, ζ1 ¼ 7:7, ζ2 ¼ 104:8 Thus, the transfer function of the macro-to-micro system can be represented by the function as follows: GðsÞ ¼ 0:01872

s2 þ 20261:5s þ 1745041 s2 þ 76:3s þ 0:13

ð6Þ

On-Chip Transmitter for Enhancing Manipulation Speed (Monzawa et al. 2015) This section will focus on an on-chip transmitter of actuation for improving the performance of macro-to-micro manipulation. The section covers the advantages of such a transmitter, modeling, and experimental validation with a closed-loop controller using high-speed vision feedback. The closed-loop control effectively compensates the phase lag observed in open-loop test shown in Fig. 6.

Advantages of Using an Actuation Transmitter There are three advantages of using the on-chip actuation transmitter, and they are clean, requiring only small sample volume and fast response. The first advantage is very straightforward that because the actuation is applied to the on-chip cells via

Fig. 6 The frequency response of the experimental results in a bode diagram. (Mizoue et al. 2017)

6

Single Cell Manipulation Using Macro-scale Actuator

147

a transmitter instead of direct application, the chance of contamination to the cells is significantly reduced. This is a practical advantage that the syringe pump and connecting tubes do not need to be frequently disinfected to avoid potential contaminations during the experiments. The second advantage is that the usage of medium can be drastically reduced. When the syringe pump is directly connected to the channels on a chip, the whole pumping pathway need to be filled with the same medium as the medium for cell suspension in order to provide a stable environment. With the actuation transmitter which physically separates the syringe and chip, the syringe and tubes connecting to the chip can be filled with different fluids, and as a result, the required volume for the test sample, such as blood, can be significantly reduced. The last advantage of fast response has been reported from theoretical and experimental perspectives, and will be explained in details in this section (Monzawa et al. 2015).

Mechanical Model and Theoretical Basis The concept of the actuation transmitter is illustrated in Fig. 7. There are two physically separated circuits on the chip, as shown in Fig. 7a. One circuit is the platform of cell manipulation, as the red circuit in Fig. 7a, and the other is the circuit directly connected to the macro-scale syringe pump, as the blue circuit in Fig. 7a. When the syringe pump is pushing forward and results in an increase of the pressure in the actuation circuit, the actuation circuit would be inflated as illustrated in Fig. 7b. The chip would be deformed due to the deformation in the actuation circuit and results in the inflation in the manipulation circuit, particularly on the side closed to the actuation circuit. Consequently, the cell in the manipulation circuit would flow toward the side of the actuation circuit. On the other hand, if the pump is retracted and the pressure in the actuation circuit reduces, the actuation circuit is shrunk and the cell in the manipulation circuit would be pushed away from the actuation circuit. It is because the shrinkage of the manipulation circuit on the side of the actuation circuit is introduced due to the shrinkage of the actuation circuit, as illustrated in Fig. 7c. Figure 8 shows the mechanical models of the manipulation systems without and with an actuation transmitter. The model of the system without an actuation transmitter in Fig. 8a is similar to the one in Fig. 3b, with slight difference in the notations for the convenience of comparison in this section. The input displacement x1 causes

Fig. 7 The working principle of the actuation transmitter. (a) Two circuits are physically separated. (b) The chip is inflated due to a push from the pump side, and it results in the cell moving toward the inflation. (c) The chip is shrunk due to a pull from the pump, and it results in the cell moving away from the shrunk circuit. (Reproduced from [T. Monzawa, M. Kaneko, C. D. Tsai, S. Sakuma and F. Arai: On-Chip Actuation Transmitter for Enhancing the Dynamic Response of Cell Manipulation using a Macro-scale Pump, Biomicrofluidics, vol.9, no.1, 014114, 2015], with the permission of AIP Publishing)

148

C.-H. D. Tsai

Fig. 8 Mechanical models for the manipulation systems with and without an actuation transmitter. (a) The model without an actuation transmitter and (b) the model with an actuation transmitter. (Figures are reproduced from [T. Monzawa, M. Kaneko, C. D. Tsai, S. Sakuma and F. Arai: On-Chip Actuation Transmitter for Enhancing the Dynamic Response of Cell Manipulation using a Macro-scale Pump, Biomicrofluidics, vol.9, no.1, 014114, 2015], with the permission of AIP Publishing)

both the deformation of the chip x3 and the displacement of the cell xc. According to the results in prior section, the transfer function between x1 and xc is like a integration operator. Figure 8b illustrates the model for the manipulation system with an actuation transmitter. Instead of directly transmitting the input displacement to the cell displacement, there is an additional degree of freedom x2, which is represented by two pistons in parallel, and is connected to the ground by a spring k2. Based on the same assumptions of incompressible fluid and constant volumetric flow, the displacements of pistons in Fig. 8b can be written as: A1 x1 ðtÞ ¼ A2 x2 ðtÞ þ A3 x3 ðtÞ

ð7Þ

A4 x2 ðtÞ ¼ Ac xc ðtÞ

ð8Þ

where the subscripts 1, 2, 3, 4, and c indicate the elements of input piston, the actuation side of the transmitter, the deformation chamber, the manipulation side of the transmitter, and the target cell, while A and x are the representative crosssectional areas and displacements, respectively. Equations (7) and (8) show the constant volumetric flows in the actuation circuit and the manipulation circuit, respectively. Since the cell is suspended by the fluid and the Reynolds number is usually small, the cell is assumed to completely move with the fluid, and xc is the displacement of the cell. The equations of motion in Fig. 8b can be described by the force balancing on the two mechanical springs k1, k2 and the viscosity of the fluid c. The relations can be written as: A3 P1 ðtÞ ¼ k1 x3 ðtÞ

ð9Þ

k 2 x 2 ð t Þ ¼ A2 P1 ð t Þ þ A4 P2 ð t Þ

ð10Þ

Ac P2 ðtÞ ¼ cx_c ðtÞ

ð11Þ

6

Single Cell Manipulation Using Macro-scale Actuator

149

where P1(t) and P2(t) are the pressures in the actuation circuit and in the manipulation circuit, respectively. x_c ðtÞ is the first derivative of xc with respective to time, and represents the velocity of the manipulated cell. The displacement of the target cell xc can be derived from Eqs. (7), (8), (9), (10), and (11) and the result is as follows: xc ðtÞ ¼ 


 k 1 A1 A 2 A4 τ1 t 1  1  e Ac k 1 A 2 2 þ k 2 A3 2 

τ1 ¼

Ac

2



ð12Þ

A3 2 A4 2 c  k 1 A2 2 þ k 2 A3 2

where τ1 is the time constant of the response of the system in Fig. 8b. A smaller time constant means the system settles faster from an input. The time constant for the system without an actuation transmitter has been previously derived, (Sakuma et al. 2014b) and is represented by the notation of τ2. To compare the time constants for the system without and with an actuation transmitter, as shown in Fig. 8a, b, the ratio between τ1 and τ2 is obtained as: γ¼

τ1 k 1 A4 2 ¼ τ 2 k 1 A 2 2 þ k 2 A3 2

ð13Þ

where Υ is the ratio, and if Υ is smaller than one, it means the response of the system with an actuation transmitter is faster, and vice versa. The ratio in Eq. (13) can be simplified with the assumption of A2 ¼ A3 ¼ A4. Thus, Eq. (13) becomes: Υ¼

τ1 k1 ¼ τ2 k1 þ k2

ð14Þ

As shown in the simplified form in Eq. (14), the ratio Υ is always smaller than one when k1 > 0 and k2 > 0. That means the response of the manipulation system with an actuation transmitter responds faster than a system without an actuation transmitter. Physically, we could interpret the phenomenon as an additional spring element is added into the system, so that overall stiffness is increased. In a typical mass-springdamper system, a stiffer spring could reduce the response time and that could be how an actuation transmitter enhancing the system response.

Experimental Validation on Manipulation with an Actuation Transmitter Experiments for the system without and with an actuation transmitter have been carried out. Figure 9 shows the experimental setups and the designs of the microfluidic circuits. The test platform, as shown in Fig. 9a, includes a high-speed camera, a microscope, a pressure sensor, a microfluidic chip, and a syringe pump, which is actuated by a piezoelectric actuator. The sampling rate for the high-speed camera is

150

C.-H. D. Tsai

Fig. 9 The experimental system for transmitter test. (a) Experimental system. (b) Microfluidic circuits with an actuation transmitter, and (c) the system without an actuation transmitter. (Reproduced from [T. Monzawa, M. Kaneko, C. D. Tsai, S. Sakuma and F. Arai: On-Chip Actuation Transmitter for Enhancing the Dynamic Response of Cell Manipulation using a Macro-scale Pump, Biomicrofluidics, vol.9, no.1, 014114, 2015], with the permission of AIP Publishing)

set at 1000 frames per second while the spatial resolution of the captured images through the microscope is 240 nm per pixel. Figure 9b, c show photos of the manipulation system with and without an actuation transmitter, respectively. The actuation transmitter is shaped like a two crossing combs and is for amplifying the efficiency of the transmission. The widths of the comb-shape channels are 50 μm and 20 μm for the actuation circuit and manipulation circuit, respectively. The gap between crossing combs is 10 μm while the height of the circuits is 3.5 μm. On the other hand for the system without an actuation transmitter shown in Fig. 9c, there is only a straight channel connected between the cell inlet and the pump inlet. Target cells are loaded from the inlet of the manipulation circuit to the straight channel. Example photos of a red blood cell being manipulated in the straight channel are illustrated in Fig. 9b, c. The manipulation system is controlled by a closed-loop PID with the feedback of cell position. Cell position is extracted using image processing on the real-time captured cell image in the channel. The controller is to manipulate the target cell following a given sinusoidal signal for simple harmonic motion in the straight channel of the manipulation circuit. Sinusoidal signal with 15 different frequencies are tested for identifying the frequency response of the system, and the frequency is in the span between 1 Hz and 130 Hz. The same tests with and without an actuation transmitter are both performed in order to compare the difference between two. The experimental results of cell manipulation with and without an actuation transmitter are presented in Fig. 10. The top-left of Fig. 10 shows a series of continuous photos captured from a video clip showing the motion of an RBC being manipulated in the channel. The results are similar to Fig. 5 but with few technical differences in the control methods, and are specially noted as follows. First, the controller for Figs. 5 and

6

Single Cell Manipulation Using Macro-scale Actuator

151

Fig. 10 The experimental results. (Figures are reproduced from [T. Monzawa, M. Kaneko, C. D. Tsai, S. Sakuma and F. Arai: On-Chip Actuation Transmitter for Enhancing the Dynamic Response of Cell Manipulation using a Macro-scale Pump, Biomicrofluidics, vol.9, no.1, 014114, 2015], with the permission of AIP Publishing)

10 are different, they are vision-based closed-loop control and displacement-based open-loop control, respectively. Instead of directly controlling the displacement of the pump as in Fig. 5, a time-dependent target position is continuously calculated and given to the controller by the follow equation: x ¼ 12:0 sin ð2πωt Þ

ð15Þ

where x, ω, and t are the target position, frequency, and elapsed time, respectively. The number of 12 is the amplitude of the simple harmonic motion in micrometer, and is about 15 pixels in the captured images. The second difference is the representative position of an RBC. The representative positions in Figs. 5 and 10 are the centroid

152

C.-H. D. Tsai

and the rightmost points, respectively. The rightmost position is used in Fig. 10 because it is faster to extract from the images than calculating the centroid, and the effect of cell deformation during the manipulation can also be neglected. The experimental results at difference frequencies are shown in Fig. 10. A total of 15 different frequencies of simple harmonic motion in Eq. (15) are tested, and the frequencies are noted on the top-left corner of each chart. The blue, red, and green curves are the given target position, the cell positions with and without an actuation transmitter, respectively. According to the results, it can be found that the three curves are almost overlapped with each other when the frequency is under 10 Hz. The green curves, the cell position without an actuation transmitter, gradually move away from the target positions which indicate that the response of the controller is not fast enough to catch up. When the frequency is up to 90 Hz, the waveforms of the target position and the cell position without an actuation transmitter are about inversed. It means the phase lag is around 180° and is the highest frequency for the system to manipulate the cell. On the other hand, the waveforms of the cell position with an actuation transmitter can still catch up with the target position at 90 Hz, although there are notable delays. The waveform between the blue and red curves becomes almost inversed when the input frequency is up to 130 Hz, and it demonstrates the speed limit of the manipulation with an actuation transmitter. The speed limits of the manipulation without and with an actuation transmitter are experimentally found in Fig. 10, and the results match to the theoretical expectation in Eq. (14) that the response is faster with an actuation transmitter. The Bode diagram of the results in Fig. 10 is calculated by plugging the both the target position and obtained cell position into a discrete Fourier Transform. The gain and the phase are obtained and are plotted in Fig. 11. The tests are repeated at least five times for each frequency. The points and error bars in Fig. 11 are the average values and the standard deviations of all the results, respectively. There are two interesting observations in Fig. 11. First, the actuation transmitter pushes the peak of the gain response to a higher frequency. Second, the lag of the system with an actuation transmitter is improved as the phase declines later. This section shows both the theoretical and experimental investigations on an actuation transmitter for on-chip single cell manipulation with a macro-scale actuation. Both results indicate that the actuation transmitter enhances the response speed to the manipulation system. In addition to the advantages of clean and requiring only small sample volume, the actuation transmitter seems to have great potential for single cell manipulations.

Recent Works and Applications Different works on cell manipulation using a macro-scale actuators have recently published in literatures. For example, Mizoue et al. (2016) controlled the waterhead of the fluid line using a macro-scale linear slider along with a noise suppression system. The work shows that a macro-scale actuator is also possible to perform extremely high resolution of cell manipulation using pressure control. They

6

Single Cell Manipulation Using Macro-scale Actuator

153

Fig. 11 The frequency responses of cell manipulation with and without the actuation transmitter. (Reproduced from [T. Monzawa, M. Kaneko, C. D. Tsai, S. Sakuma and F. Arai: On-Chip Actuation Transmitter for Enhancing the Dynamic Response of Cell Manipulation using a Macro-scale Pump, Biomicrofluidics, vol.9, no.1, 014114, 2015], with the permission of AIP Publishing)

successfully achieved cell manipulation in the order of pressure of 10 mPa, which is about 104 of atmospheric pressure. Another example is that Heo et al. (2015) integrated both macro-scale and micro-scale actuators to build up a hybrid actuator system for single particle manipulation on a chip. The manipulation system also has been applied in different applications. For example, Ito et al. (2017) utilized the manipulation system to perform cell evaluation. Horade et al. (2017) observed the evolution of cell morphology by positioning single cells in a constriction for several minutes. Arakawa et al. (2011) used single cell manipulation and cell-trappingpocket to collect large number of cells on a microfluidic device. Single cell manipulation with droplets has also been advanced recently. For example, Luo et al. (2019) summarized recent progress of single cell, or droplet, manipulation on a microfluidic chip. Breslauer et al. (2006) reviewed microfluidic systems in the biology including single cell and droplet manipulations. Gao et al. (2019) showed cutting edge techniques of single cell manipulation on single cell analysis.

Summary This chapter reviews the recent works on single cell manipulation with both micro-scale and macro-scale actuators, and is particularly focused on macro-scale actuators. The main advantage of using a macro-scale actuator is that single cell manipulation can be achieved with commercially available actuators. The deformation of the microfluidic chip is found playing a key role for such a manipulation. An on-chip actuation transmitter is also covered in this chapter and is found capable of further improving the dynamic response of the manipulation while the actuation transmitter also effectively

154

C.-H. D. Tsai

separates sample solution and actuation fluid. Recent applications of the manipulation system with a macro-scale actuator are also discussed.

References Arakawa T, Noguchi M, Sumitomo K, Yamaguchi Y, Shoji S (2011) High-throughput single-cell manipulation system for a large number of target cells. Biomicrofluidics 5(1):14114 Avci E et al (2015) High-speed automated manipulation of microobjects using a two-fingered microhand. IEEE Trans Ind Electron 62(2):1070–1079 Breslauer DN, Lee PJ, Lee LP (2006) Microfluidics-based systems biology. Mol BioSyst 2(2):97 Chronis N, Lee LP (2005) Electrothermally activated SU-8 microgripper for single cell manipulation in solution. J Microelectromech Syst 14(4):857–863 Dao M, Lim CT, Suresh S (2003) Mechanics of the human red blood cell deformed by optical tweezers. J Mech Phys Solids 51(11–12):2259–2280 Ding X et al (2012) On-chip manipulation of single microparticles, cells, and organisms using surface acoustic waves. Proc Natl Acad Sci 109(28):11105–11109 Gao D, Jin F, Zhou M, Jiang Y (2019) Recent advances in single cell manipulation and biochemical analysis on microfluidics. Analyst 144(3):766–781. Heo YJ, Kang J, Kaneko M, Chung WK A hybrid actuator system for single particle manipulation on a microfluidic chip. In IEEE international conference on robotics and automation (ICRA), 2015, pp 2691–2697 Horade M, Tsai CD, Ito H, Kaneko M (2017) Red blood cell responses during a long-standing load in a microfluidic constriction. Micromachines 8:100 Horade M, Tsai CHD, Kaneko M (2019) On-chip cell incubator for simultaneous observation of culture with and without periodic hydrostatic pressure. Micromachines 10(2) Ito H et al (2017) Mechanical diagnosis of human erythrocytes by ultra-high speed manipulation unraveled critical time window for global cytoskeletal remodeling. Sci Rep 7(February):43134 Lim CT, Dao M, Suresh S, Sow CH, Chew KT (2004) Large deformation of living cells using laser traps. Acta Mater 52(7):1837–1845 Luo T, Fan L, Zhu R, Sun D (2019) Microfluidic single-cell manipulation and analysis: methods and applications. Micromachines 10(2):104 Mizoue K, Phan M, Tsai CD, Kaneko M, Kang J, Chung W (2016) Gravity-based precise cell manipulation system enhanced by in-phase mechanism. Micromachines 7:116 Mizoue K, Teramura K, Tsai CD, Kaneko M (2017) Transfer function of macro-micro manipulation on a PDMS microfluidic chip. Micromachines 8:80 Monzawa T, Kaneko M, Tsai CD, Sakuma S, Arai F (2015) On-chip actuation transmitter for enhancing the dynamic response of cell manipulation using a macro-scale pump. Biomicrofluidics 9(1):014114 Sakuma S, Kuroda K, Tsai CD, Fukui W, Arai F, Kaneko M (2014a) Red blood cell fatigue evaluation based on the close-encountering point between extensibility and recoverability. Lab Chip 14(6):1135–1141 Sakuma S et al (2014b) High resolution cell positioning based on a flow reduction mechanism for enhancing deformability mapping. Micromachines 5:1188–1201 Sakuma S, Kasai Y, Hayakawa T, Arai F (2017) On-chip cell sorting by high-speed local-flow control using dual membrane pumps. Lab Chip 17(16):2760–2767 Tsai C-HD et al (2017) 3000Hz cell manipulation in a microfluidic channel. In: IEEE/RSJ international conference on intelligent robots and systems (IROS), vol 2017, Sept, pp 2968– 2973 Yalikun Y, Kanda Y, Morishima K (2016) Hydrodynamic vertical rotation method for a single cell in an open space. Microfluid Nanofluid 20(5):1–10 Zhang H, Liu K-K (2008) Optical tweezers for single cells. J R Soc Interface 5(24):671–690

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis Nan Xiang and Zhonghua Ni

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Underlying Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inertial Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dean Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Viscoelasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Guidelines for Designing Spiral Inertial Microfluidics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dimensionless Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Design Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Improved Understandings on Spiral Inertial Microfluidics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application Guidelines for Spiral Inertial Microfluidics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Focusing/Ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concentration/Microfiltration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion and Future Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

156 157 157 159 159 160 160 162 164 165 166 166 174 177 179

Abstract

Inertial microfluidics has been widely employed as an important sample pretreatment technique for the single-cell analysis because of the distinct advantages of label-free and external field-free operation, high-throughput processing, and simple channel structure. In this chapter, the recent advances on the application of spiral inertial microfluidics as the sample pretreatment tool for single-cell analysis will be systematically summarized. Firstly, the underlying physics of the spiral inertial microfluidics will be introduced for better understanding of the working principle. Secondly, the design guidelines will be provided for the beginners or the researchers from other disciplines to efficiently design the devices of their N. Xiang (*) · Z. Ni School of Mechanical Engineering, and Jiangsu Key Laboratory for Design and Manufacture of Micro-Nano Biomedical Instruments, Southeast University, Nanjing, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_29

155

156

N. Xiang and Z. Ni

applications. Thirdly, the recent advances on the application of spiral inertial microfluidics will be reviewed. Finally, the future perspective on inertial microfluidics will be discussed.

Introduction Single-cell analysis is a newly emerging technique for acquiring the biological information at the single-cell level and has attracted increasing interests over the recent years (Yin and Marshall 2012; Oomen et al. 2019). However, the successful implementation of the single-cell analysis is still limited by the technological level of sample preparation. For example, the single-cell analysis of circulating tumor cells (CTCs) in blood provides vital information for studying the cancer metastasis and serves as the “liquid biopsy” for cancer diagnosis (Song et al. 2017). The biggest challenge in analyzing CTCs is the extremely rare number of CTCs in the peripheral blood (1~100 cells in 7.5 ml blood or even less) (Plaks et al. 2013), and thus the CTC isolation is obviously the key pretreatment step for the downstream CTC analysis. Another example is the flow cytometer in which the cells are required to be aligned into a stream before reaching the interrogation point (Heikali and Di Carlo 2010). However, the traditional flow focuser is expensive and difficult to be miniaturized. The advent of microfluidics has provided new insights to the field of sample preparation for single-cell analysis (Hosic et al. 2016). As compared with traditional techniques, the microfluidic sample preparation offers various advantages of low sample consumption, high efficiency, small device footprint, and high manipulation resolution. Up until now, the microfluidics has been widely employed for various sample preparations such as focusing (Xuan et al. 2010), separation (Sajeesh and Sen 2014), concentration (Xiang et al. 2019c), mixing (Nguyen and Wu 2004), and trapping (Nilsson et al. 2009). According to the employed working principles, the reported microfluidic sample preparation devices are commonly classified into two categories: active devices and passive devices. The active devices use the external force fields (e.g., optical (MacDonald et al. 2003), electric (Kim et al. 2019), magnetic (Chen et al. 2018), and acoustic (Yeo and Friend 2014)) to apply a force on flowing cells for pushing or attracting the cells. Although a high manipulation resolution up to nanoscale can be achieved, the throughput of active devices is relatively low, which prevents the wide application of these active devices for processing large-volume samples. To induce the external force fields, the expensive, large-volume, and energy-consuming generators (microelectrode, signal generator, and so on) are required. Instead, the passive devices directly employ the hydrodynamic effects of microfluids (e.g., inertial (Di Carlo 2009), hydrophoresis (Choi et al. 2008), and viscoelastic (Lu et al. 2017)) or the interaction between cells and specific microstructures (e.g., deterministic lateral displacement (DLD) (McGrath et al. 2014), microfilter (Zheng et al. 2007), and hydrodynamic microfiltration (Yamada and Seki 2005)) to achieve cell manipulation.

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

157

Among the reported passive techniques, the inertial microfluidics has attracted increasing interests over the recent years because of the numerous advantages of sheath-free and external field-free operation, simple channel structure, and highthroughput processing. As a newly emerging passive technique for manipulating cells or other bioparticles at the microscale, inertial microfluidics has been successfully employed as various microfluidic sample preparations (focusing, ordering, separation, concentration, and so on) for ensuring the single-cell analysis (Zhang et al. 2016; Stoecklein and Di Carlo 2019). The employed channel geometries have also been diversified, ranging from simple straight channels (Hur et al. 2010), contraction–expansion channels (Khojah et al. 2017) to more complex curving channels (Martel and Toner 2013). Among the employed channel geometries, spiral channel has been most widely used in the inertial microfluidic research due to the significantly different cell focusing positions. However, the guidelines for the beginners or the researchers from other disciplines to efficiently design and apply spiral inertial microfluidics are still lacking. In this chapter, the basic physics of spiral inertial microfluidics will be firstly introduced for better understanding of the working principle. Then, the design guideline will be summarized for the beginners or the researchers from other disciplines to design the devices of their applications. Finally, the advances on the application of spiral inertial microfluidics will be provided to tell the readers in which areas the spiral inertial microfluidics has been or can be used.

Underlying Physics Inertial Migration The microfluidic flows in microchannels of microscales are assumed to be at low channel Reynolds numbers (Rec=o(10
6)~o(10)) and thus are commonly treated as Stokes flows for which the inertia items in the Navier–Stokes equation are often neglected (Squires and Quake 2005). Actually, in finite Reynolds number flows, the inertial effects of microfluids cannot be neglected, and the inertial migration phenomenon also exists in microfluidic flows. The inertial migration was firstly observed by Segre and Silberberg (1961) in the macroscopic flow experiment in which the millimeter-sized particles migrated perpendicular to the main flow and formed a particle ring located at 0.4 times the radius away from the channel wall in a cylindrical tube. This interesting lateral particle migration was also known as the “tubular pinch effect.” After the first discovery of this phenomenon in the 1960s, a number of theoretical studies (Saffman 1965; Cox and Brenner 1968; Ho and Leal 1974; Schonberg and Hinch 1989; Asmolov 1999; Matas et al. 2004; Asmolov et al. 2018) have been carried out to uncover its underlying physics. Up until now, the most well-accepted explanation is that this lateral particle migration is caused by the inertial lift force (FL) which is actually the resultant of the shear-gradient-induced inertial lift force (FLS) and the wall-induced inertial lift force (FLW) (Di Carlo et al. 2007). In the Poiseuille flow, the parabolic

158

N. Xiang and Z. Ni

velocity profile will induce a FLS which forces the particle down the shear gradient toward the channel wall. When the particle moves toward the channel wall, the wake of the rotating particle will be broken and thus induce a FLW to push the particle away from the channel wall. Asmolov (1999) provided a force scaling for the FL as: FL ¼

U 2m a4p ρ D2h


 f L Re c , Xp ,

ð1Þ

where ap is the particle diameter, Um is the maximum flow velocity, ρ is the fluid density, fL(Rec, Xp) is the lift coefficient which is a function of the channel Reynolds number (Rec) and the particle cross-sectional position (Xp), and Dh is the hydraulic diameter which can be calculated as 2hw/(h + w) for rectangular cross sections (h and w are the channel height and width, respectively). However, this force scaling was generated from the parabolic flow between two infinite plates (Asmolov 1999). Di Carlo et al. (2009) derived a complex scaling of the FL through the finite element numerical simulation. Specifically, the FL scales as ρU 2 a3p =H at the nearcenter region and as ρU 2 a6p =H 4 at the near-wall region, where the U (U = 1/2Um) is the average flow velocity and can be approximated as Q/hw (Q is the volumetric flow rate), and H is the characteristic channel dimension. Later, Liu et al. (2016b) proposed a generalized formula of the FL which contains four items of the sheargradient-induced inertial lift force, the wall-induced inertial lift force, the slip-shear inertial lift force, and the correction of the shear-gradient-induced inertial lift force. On the basis of the two-force competition (FLS vs. FLW) theory, the randomly dispersed particles will gradually migrate and equilibrate at specific positions, as illustrated in Fig. 1a. The particle equilibrium positions are actually the crosssectional positions where the two forces balance with each other. As illustrated in Fig. 1b, similar to macroscopic flows, the particles flowing in circular channels will form a particle annulus. In square channels, the particles equilibrate at four positions centered at each channel wall. For rectangular channels, the final equilibrium

Fig. 1 (a) Particle inertial migration in a rectangular channel. The randomly dispensed particles will gradually migrate and equilibrate at specific positions under the effects of FLS and FLW. (b) Cross-sectional particle equilibrium positions in different channels

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

159

positions can be reduced to two at the centers of the long channel walls. To help understand the particle equilibrium process in rectangular channels, a two-stage migration process that the particles first migrate toward the equilibrium positions near walls under the effects of FLS and FLW, and then migrate along the channel walls into the two final wall-centered equilibrium positions under a rotation-induced lift force (FΩ) (Zhou and Papautsky 2013).

Dean Flow In curving channels, the cross-sectional secondary flow (also named as Dean flow (Dean and Chapman 1928)) will also affect the particle lateral migration. The Dean flow is generated due to the velocity mismatch between the fluids at different crosssectional regions. In the Poiseuille flow, the fluids near the channel cross-sectional center flow faster and have a larger inertia when flowing through the curving channels. Therefore, the fluids near the channel cross-sectional center will flow outward, while the fluids near the channel outer wall move inward along the top and bottom walls due to the mass conservation. The formed Dean flow is commonly illustrated as the two counter-rotating vortices within the top and bottom half regions of the channel cross section (see Fig. 2) and will induce a Dean drag force (FD) on the flowing particles. A scaling of the FD can be described as follows (Di Carlo et al. 2007): FD / ρU 2m ap D2h R
1 ,

ð2Þ

where R is the channel radius. Actually, the value and the direction of the FD are both heavily dependent on the cross-sectional position. The Dean flow brings two beneficial effects for particle focusing in curving channels: (1) speeding up the inertial focusing process for reducing the required minimum channel length and (2) modifying the equilibrium positions for possible differential separation.

Viscoelasticity Some biological fluid samples are the non-Newtonian fluids which exhibit a combination of the viscous characteristics (liquid-like) and the elastic characteristics Fig. 2 Dean flow in curving channels. (Reproduced from Di Carlo (2009) with permission from The Royal Society of Chemistry)

160

N. Xiang and Z. Ni

(solid-like) (Chhabra 2006). The nonlinear viscoelastic characteristic of the non-Newtonian fluids makes the particle migration in these carrying fluids more complex and interesting (D’Avino and Maffettone 2015). In viscoelastic flows, the first and second normal stress differences (N 1 ðγ_Þ ¼ σ xx
σ yy , N 2 ðγ_Þ ¼ σ yy
σ zz , and the σ xx, σ yy, and σ zz are, respectively, normal stresses toward the flow, the velocity gradient, and vorticity directions) force the particles to migrate laterally across the main flow and finally occupy the stable equilibrium positions even when the inertial effect is very weak (Yuan et al. 2018). For most non-Newtonian viscoelastic fluids, N 2 ðγ_Þ does not exceed 20% of the N 1 ðγ_Þ and thus can be neglected (Karimi et al. 2013). Therefore, the mechanics of the particle viscoelastic migration is that the positive N 1 ðγ_Þ induces an elastic force (FE / a3p ∇N 1 ðγ_Þ) in the direction of decreasing the shear rate (Leshansky et al. 2007). Different from the above inertial migration, the particles will occupy a single stable equilibrium position in the center of circular channels at a relative low flow rate when the shear viscosity is constant (e.g., Boger fluids (James 2008)) (Romeo et al. 2013). Meanwhile, in typical rectangular or square channels, the particles will focus at multiplex cross-sectional equilibrium positions including four corners and one center where the shear rates are at the minimum (Seo et al. 2014). Reduction of the number of equilibrium positions can be realized through appropriately increasing the operational flow rate to make the fluid inertia non-negligible. Through the coupling of the inertial effect with the viscoelastic effect, the original equilibrium positions at the four corners can be eliminated, and thus the three-dimensional (3D) focusing of particles at the channel cross-sectional center can be achieved (named as the elasto-inertial particle focusing) (Yang et al. 2011). Up until now, this 3D elasto-inertial particle focusing has been widely employed in various microfluidic applications through simply adding a small amount of elasticity additives to the original sample fluids (Lu et al. 2017; Yuan et al. 2018). Although the elasto-inertial focusing is able to focus the nanoscale objects (e.g., DNA molecules (Kim et al. 2012), exosomes (Liu et al. 2017; Zhou et al. 2019), and nanoparticles (Liu et al. 2016a)), the addition of polymer elasticity additives may cause shear thinning or shear thickening at high flow rates and the increase of fluid viscosity, which limit the processing throughput of this technique. Recently, the researchers have employed some new elasticity additives (e.g., λDNA and hyaluronic acid) at very small concentrations to induce strong viscoelasticity for high-throughput applications (Kang et al. 2013; Lim et al. 2014). The detailed elasto-inertial focusing mechanisms in spiral channels can be found in the recent studies (Xiang et al. 2016, 2018a).

Guidelines for Designing Spiral Inertial Microfluidics Dimensionless Numbers In spiral inertial microfluidics, the flowing particles may simultaneously suffer from the effects of inertial migration and secondary Dean flow. To quantitatively evaluate the inertial effect, the dimensionless channel Reynolds number (Rec) was employed.

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

161

Rec is the ratio of inertial force to viscous force and could be calculated as follows (Squires and Quake 2005): Re c ¼

ρU Dh 2ρQ , ¼ η ηðw þ hÞ

ð3Þ

where η is the fluid dynamic viscosity. The fluid inertia will become more significant with the increase of Rec. In addition to channel Reynolds number, the particle Reynolds number ρU a2 ( Re p ¼ ηDhp ) was employed to evaluate the ratio of inertial force to viscous force at the particle length scale. To achieve the particle focusing, Rep needs to be larger than 1 (Di Carlo et al. 2007). The strength of the Dean flow in spiral channels can be characterized through using the Dean number (De) (Berger et al. 1983) which can be calculated as: rffiffiffiffiffiffi Dh De ¼ Re c , 2R

ð4Þ

where R is the radius of channels. From this definition, it is shown that the Dean flow strength can be enhanced through increasing Rec or Dh or decreasing R. For spiral channels, De will gradually decrease from the innermost loop to the outermost loop. To estimate the competition between FL and FD, a dimensionless force ratio (Rf = FL/FD) is defined as (Xiang et al. 2015b):  3
 FL 1 a p Rf ¼ / f L Re c , Xp , FD δ D h

ð5Þ

where δ is the curvature ratio δ = Dh/2R. ap/Dh is the confinement ratio (β). It can be observed that Rf is proportional to β3. For low-aspect-ratio (h < w) spiral channels, Dh for calculating Rf can be approximated as the channel height h. When Rf  1, the particle migration will be dominated by FL. On the contrary, the particle migration is dominated by FD, and the defocusing caused by the Dean mixing may occur. When Rf is of order 1, the particle will migrate under the competition between these two forces. When the carrying fluids are non-Newtonian fluids, the viscoelasticity effect of fluids can be evaluated using the Weissenberg number (Wi) which is the ratio of elastic force to viscous force and can be calculated as (Lu and Xuan 2015): Wi ¼ λγ_c ¼

2λQ , hw2

ð6Þ

where γ_c is the characteristic shear rate which can be calculated as 2Q/hw2 for rectangular channels and λ is the relaxation time of non-Newtonian viscoelastic fluids. To compare the viscoelastic and inertial effects, the Elasticity number (El) is defined through dividing the Weissenberg number (Wi) by the channel Reynolds number (Rec) (Rodd et al. 2007):

162

N. Xiang and Z. Ni

El ¼

ληðh þ wÞ Wi ¼ : Re c ρhw2

ð7Þ

It should be noticed that although both Rec and Wi are proportional to the flow rate, the newly generated El is totally independent on the flow rate for non-Newtonian fluids without shear thinning and thickening (Amini et al. 2014). For Newtonian fluids, El always equals to zero.

Design Guidelines To design and operate the spiral inertial microfluidic devices, the following guidelines need to be considered.

Cross-Sectional Dimensions For the purpose of ease fabrication, nearly all the previously reported works employed low-aspect-ratio (w > h) spiral channels. When designing spiral inertial microfluidic devices with low-aspect-ratio rectangular cross sections, the most important parameter is the channel height (h). To successfully achieve the focusing of specific-sized particles (with a diameter of ap), the channel height (h) needs to satisfy the following criterion (Bhagat et al. 2008; Xiang et al. 2015b): ap ap h : 0:5 0:07

ð8Þ

For designing the channel width (w), there is no criterion at current. However, it is better to control the channel width (w) to be within 8 h on the basis of previous experimental experiences (Martel and Toner 2012). Increasing the channel width will reduce the shear gradient across the channel width and results in unsatisfactory focusing. In addition to the most widely used rectangular cross sections, the trapezoidal cross sections are also used in spiral inertial microfluidics (Guan et al. 2013). Utilization of trapezoidal cross sections can regulate the profile of Dean flow and thus is beneficial for improving the accuracy of cell separation.

Channel Length (Loop Number) Another important parameter for designing the spiral inertial microfluidics is the channel length (L ). As the lateral migration velocity is very weaker as compared with the flowing velocity along the main stream, a minimum channel length (Lmin) is required to ensure that all the particles can reach the equilibrium positions when arriving at the outlet. First, the lateral migration velocity (UL) of particles flowing in straight channels can be calculated through assuming the Stokes drag (FStokes = 3πapUL) balances the inertial lift force (FL):

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

UL ¼

ρU 2m a3p

fL 3πηD2h




 Re c , Xp :

163

ð9Þ

Then, the required minimum channel length (Lmin) can be calculated as: Lmin ¼

3πηD2h LL LL
, U¼ UL 4ρU a3p f L Re c , Xp

ð10Þ

where LL is the lateral distance from the particle’s initial position to the final equilibrium position and can be approximated using the channel width (w), U is the average flow velocity and equals 0.5Um, and fL(Rec, Xp) can be approximated as 0.5 when calculating Lmin. As the Dean flow can assist in speeding up the particle focusing process, the required minimum channel length (Lmin) for spiral inertial microfluidics can be further shortened and can be obtained through multiplying the equation for straight channels by a factor (b = 0.2~1; its value is dependent on the operational flow rate and channel geometry) (Amini et al. 2014). Of course, it is safe to use the above straight channel equation for calculating Lmin for the spiral inertial microfluidics.

Other Structural Parameters In addition to the above cross-sectional parameters and channel length, other structural parameters include the initial radius (R) and the distance between the adjacent loops (D). The smaller initial radius (R) will induce a stronger Dean flow. Therefore, it is better to design a spiral channel with small initial radiuses (R) for enhancing the Dean flow effect. The small D will be beneficial for reducing the device footprint but may cause the leakage between the adjacent loops due to the insufficient bonding strength. There is no mature design rule to follow at current for designing these two structural parameters. Other Functional Units After deigning the main channel geometry, the next step is to design the functional units (e.g., prefilter unit, outlet system, and flow resistance unit) for specific applications. For example, the prefilter unit is commonly placed in or after the inlet for filtering the large impurities from the sample. To achieve the separation or concentration of particles, a specific outlet system needs to be designed. The design of the outlet system requires the determination of the outlet number and the width of each outlet according to the differential particle focusing positions. Utilizing the sudden expansion channel before the outlet system, and the control of the flow resistance after each outlet are very beneficial for magnifying the distance between the separated particle focusing strings. For designing the outlet system, the readers can refer to a recent work on output channel design for collecting closely spaced particle strings (Yousuff et al. 2017).

164

N. Xiang and Z. Ni

Operational Parameters To achieve the particle focusing, the driving flow rate (Qmin) for driving the sample in the channel of a channel length (L ) can be estimated as: Qmin ¼

3πηD2h w2 h : 2ρa3p L

ð11Þ

In addition to the driving flow rate, another important operational parameter is the particle concentration which may affect the particle focusing due to the heavy particle–particle interactions. To identify when the concentration may affect the particle focusing, a length fraction (LF) can be employed as (Di Carlo 2009): LF ¼ 6whV f =πa2p ,

ð12Þ

where Vf is the volume fraction of the particles in the sample. For LF > 1, single-line focusing is difficult to be achieved due to the heavy interactions between particles. As the flow rate in inertial microfluidics is relatively high, one important consideration is whether the high shear stress will affect the cell viability (e.g., the cell death and the change of physiological characteristics). Previous studies proved that the shear stress has a negligible effect on cells after comparing the viabilities or gene expression profiles of cells before and after processing (Hur et al. 2011; Xiang et al. 2019b). However, special care still needs to be taken when dealing with the flowsensitive cells (e.g., platelet).

Improved Understandings on Spiral Inertial Microfluidics In this section, the studies on inertial focusing in spiral inertial microfluidics will be summarized to provide a better understanding of the device physics for guiding the design and application of spiral inertial microfluidics. The sole studies on inertial migration or Dean flow are not involved in this section. Flow rate effect and particle focusing process model: As the only operational parameter in spiral inertial microfluidics, the flow rate is directly responsible for regulating the particle focusing behaviors. Nearly all the mechanism studies on inertial focusing in spiral inertial microfluidics have explored the effects of flow rate on particle focusing. A five-stage process model was proposed to better understand the particle focusing process in spiral channels (Xiang et al. 2013b). It was found that particles first migrate toward the inner channel wall and focus into a string under the dominant FL with the increase in the flow rate. The formed particle string keeps stable at a specific position with further increasing flow rate and finally shifts away from the inner wall at higher flow rate due to the alternation of the dominant force to FD. Studies on how the randomly dispensed particles migrate to form a particle string (Xiang et al. 2013a) and the mechanism of transition between inner and outer focusing (Martel and Toner 2013) were also investigated. In addition, the focusing positions of differently sized particles were found to be dependent on the

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

165

flow rate (Xiang et al. 2013b). At low flow rates, the small particles focus closer to the inner wall as compared with large particles. At high flow rates, the focusing positions of the two-sized particles would be exchanged, and the large particles occupy the lateral positions much closer to the inner wall. The co-focusing of differently sized particles in the same string was also observed in spiral inertial microfluidics (Xiang et al. 2013b). Particle focusing modes: As can be seen from the force ratio Rf definition, Rf is found to be proportional to β3. A recent study indicated that the particles are in three different focusing modes varying the particle confinement ratios (β) (Xiang et al. 2015b). When β is larger than 0.07, the particles are in the focusing mode and can be focused into a particle string under specific flow rates and channel lengths. As β is in the range of 0.01–0.07, the particles are in the rough focusing mode, and a relatively wide particle band can be formed (two particle-free regions are generated near the both channel walls). Further decreasing β to be smaller than 0.01, the particles will be dispersed over the channel regardless of flow rate due to the dominated Brownian motion and thus can be regarded as in the non-focusing mode. Dean flow dynamics and regulation: In spiral inertial microfluidics, the induced Dean flow can assist in speeding up the particle focusing process and reducing the number of equilibrium positions. Previous understanding on Dean flow was still established on the assumption of two Dean vortices in the cross section. Recently, the Dean flow instabilities in spiral channels were explored (Nivedita et al. 2017). The present of multiple Dean flows (four vortices) was observed when De or Rec is larger than a critical value. The generation of the additional Dean vortices will shift the particle focusing position toward the outer wall. At low De, the Dean flows can also be artificially regulated through using the non-rectangular (stair-like and trapezoidal) cross sections (Ghadami et al. 2017; Guan et al. 2013). In addition, the acceleration of Dean flow at low Des can be realized through adding the microbars in spiral channels to create the geometric confinement effect (Shen et al. 2018). Channel length and curvature effects: A two-stage focusing process was observed through characterizing the particle focusing dynamics at different positions along the channel (Martel and Toner 2012; Xiang et al. 2015a). The particles will first migrate laterally and focus into a string (stage I), and then the whole formed particle string will shift toward the inner channel wall (stage II). In spiral inertial microfluidics, the effects of inertial migration and Dean flow are coupled. To decouple these two effects, the researchers designed a set of devices with different radiuses of curvatures and investigated the particle focusing dynamics in these channels (Martel and Toner 2013).

Application Guidelines for Spiral Inertial Microfluidics The spiral inertial microfluidics has been widely employed as various sample preparation techniques for single-cell analysis. After demonstrating the basic theory and design guidelines, the recent advances will be summarized regarding the

166

N. Xiang and Z. Ni

application of spiral inertial microfluidics. In this section, the recent advances on spiral inertial microfluidics will be classified according to their functions.

Focusing/Ordering Flow cytometer has been regarded as a powerful tool and the “gold standard” for single-cell analysis (Shapiro 1983). One of the most important components in the traditional flow cytometer is the nozzle fluidic focuser in which the cells flowing in the central core are pinched by the co-flow sheath flow outside. After being pinched into a single string, the cells can one by one pass through the downstream interrogation point (Heikali and Di Carlo 2010). However, the nozzle fluidic focuser is difficult to be miniaturized, which prevents its use in microflow cytometer. As the basis function of the spiral inertial microfluidics is to focus/order the particles/cells at a high throughput without using the sheath flow or external force fields, it would be the ideal flow focuser for microflow cytometer. For example, the spiral inertial microfluidic channel was integrated with the laser-induced fluorescence (LIF) setup for cell counting at a throughput of 2100 particles/s (Bhagat et al. 2010). In addition to the microflow cytometer, Huang et al. (2018) realized the lipid profiling in single cells through coupling the spiral inertial microfluidics with the mass spectrometry (see Fig. 3a). Another interesting application area of inertial focusing is the single-cell droplet encapsulation. The encapsulation of single cells within picoliter-sized droplets (cell in droplet) enables the quantitative and precise study of single-cell biology. Kemna et al. (2012) employed the spiral inertial microfluidics as the prefocusing unit for high-yield and high-speed single-cell droplet encapsulation. Later, Moon et al. (2018) used similar technique to develop a droplet-based microfluidic platform with a significantly improved throughput and low barcoding errors (see Fig. 3b) and then applied their platform for high-throughput single-cell RNA sequencing. The detailed information on the studies in this section can be found in Table 1.

Separation To realize the single-cell analysis, the separation of desired cells from the biofluids containing complex cell components is commonly required. As compared with other inertial microfluidic channels (e.g., straight, sinusoidal, and contraction–expansion channels), the spiral inertial microfluidics is obviously the most ideal candidate for achieving the high-throughput and continuous cell separation. In previous works, two working principles are mainly employed to realize the cell separation according to their size difference. The first strategy is to focus the differently sized cells at differential lateral positions and collect the separated cell strings using a specific outlet system (Kuntaegowdanahalli et al. 2009), as illustrated in Fig. 4a. This strategy is beneficial for the simultaneous separation of multi-sized cells, but the separation resolution may be heavily limited by the close distance between the cell

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

167

Fig. 3 (a) Lipid profiling in single cells through coupling the spiral inertial microfluidics with the mass spectrometry. (Reproduced from Huang et al. (2018) with permission from The Royal Society of Chemistry). (b) Inertial focusing-assisted droplet microfluidics for high-throughput single-cell RNA sequencing. (Reproduced from Moon et al. (2018) with permission from The Royal Society of Chemistry)

168

N. Xiang and Z. Ni

Table 1 Advances on focusing/ordering using spiral inertial microfluidics Application area Microflow cytometer

Single-cell lipid profiling Single-cell droplet encapsulation

Device and dimension 5-loop, Archimedean spiral, PDMS device, 500  120 μm (w  h), L = 25 cm 3-loop, capillary, 50 μm (diameter), L = 14 cm

Sample SH-SY5Y neuroblastoma cells

Throughput 2100 particles/s

Reference Bhagat et al. (2010)

HUVEC cells

30 μl/min

5-loop, Archimedean spiral, PDMS device, 50  29 μm (w  h), L = 7.2 cm 5-loop, Archimedean spiral, PDMS device, 120  100 μm (w  h)

Myeloid leukemic cells (HL60 and K562 cells)

2700 cells/s

Huang et al. (2018) Kemna et al. (2012)

K562 (human), 293 T (human), and NIH/3 T3 (mouse) cells

2700 cells/ min

Moon et al. (2018)

strings. One interesting way for addressing this limitation is to design and use the spiral channel with trapezoidal cross sections in which the distance between the focused cell strings can be expanded through regulating the Dean vortex (Wu et al. 2012; Guan et al. 2013). The second strategy is the Dean flow fractionation (DFF) in which the large cells will focus into a string and shift into the sheath flow, whereas the small cells keep unfocused and will be confined in the sample flow (Hou et al. 2013), as illustrated in Fig. 4b. Over the recent years, the spiral inertial microfluidics has been widely employed for the separation of various cells. Among these separation applications, the isolation of CTCs has attracted the most research interests. Due to its rarity (1~100 cells in 7.5 ml blood or even less (Plaks et al. 2013)), the successful isolation of CTCs is an important prerequisite for the downstream detection which has been regarded as a noninvasive “liquid biopsy” technique for the therapeutic efficacy monitoring and the potential cancer diagnosis at the early stage (Song et al. 2017; Jackson et al. 2017). Sun et al. (2012) employed the double spiral channel for the separation of the spiked tumor cells from blood cells at a recovery ratio of 88.5% and a throughput of 3.33  107 cells/min. Hou et al. (2013) used the DFF technique for the isolation of the spiked cancer cells from blood at a recovery ratio of >85%. Through combing with immunofluorescence staining, they also applied their device for the detection of CTCs in the samples from patients with metastatic lung cancer, and a detection rate of 100% was achieved (n = 20). To increase the processing throughput, multiplexing spiral channels have been stacked in the vertical direction, which enables the processing of 7.5 ml whole blood be finished within 10 min (Warkiani et al. 2014b). Later, a clinical validation of their ultrahigh-throughput spiral inertial microfluidics was performed using the blood from patients with advanced-stage metastatic breast and lung cancers (Khoo et al. 2014). On the basis of the DFF technique, a commercialized ClearCell FX1 system was developed for the isolation

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

169

Fig. 4 (a) Differential focusing-based cell separation. The differently sized cells are focused at differential lateral positions, and the separated cell strings can be collected using a specific outlet system. (Reproduced from Kuntaegowdanahalli et al. (2009) with permission from The Royal Society of Chemistry). (b) DFF-based cell separation. The large cells will focus into a string and shift into the sheath flow, whereas the small cells keep unfocused and will be confined in the sample flow. (Reprinted with permission from Hou et al. (2013))

of CTCs (Lee et al. 2018b). The DFF technique owns a high separation resolution, but a sheath flow is required. The use of sheath flow may increase the operation complex and the target sample volume. As an improvement on the Archimedean

170

N. Xiang and Z. Ni

Fig. 5 (a) Trapezoidal spiral inertial microfluidics for the separation of CTCs from WBCs. (Reprinted with permission from Warkiani et al. (2014a). Published by The Royal Society of Chemistry). (b) The device integrated the spiral inertial microfluidics with the membrane filter. (Reprinted with permission from Wang et al. (2015). Copyright (2015) American Chemical Society). (c) The i-DLD sorter that coupled the spiral inertial microfluidics with the deterministic lateral displacement (DLD). (Reprinted with permission from Xiang et al. (2019b). Copyright (2019) American Chemical Society). (d) Automated microfluidic instrument for label-free and high-throughput separation of tumor cells from blood. (Reprinted with permission from Zhang et al. (2018). Copyright (2018) American Chemical Society)

single-inlet spiral, Warkiani et al. (2014a) developed a spiral inertial microfluidics with trapezoidal cross sections (trapezoidal spiral inertial microfluidics) for the isolation of CTCs from blood (see Fig. 5a). A recovery ratio of 80% for cancer cell lines and a 100% detection ratio for the patient samples with advanced-stage metastatic breast and lung cancers were achieved (n = 10). Similar channel system was also used to enrich the CTCs in the blood samples from head and neck cancer patients (Kulasinghe et al. 2017). As the numbers of background blood cells are very large when compared with that of CTCs, the purity of the CTCs is commonly very low. To obtain high-purity CTC samples for various subsequent bioinformatic analyses, the spiral inertial microfluidics has been integrated with other separation techniques. For example, Wang et al. (2015) integrated the spiral inertial microfluidics with the membrane filter (see Fig. 5b), which allows the high-efficiency separation of CTCs from blood. A capturing efficiency of 74.4% can be achieved for the separation of the tens of A549 cells from per mL of whole blood. Through integrating with the

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

171

immunostaining and CK-19 mRNA detection, their device is able to detect 90% of metastatic lung cancer. Xiang et al. (2019b) coupled the spiral inertial microfluidics with the deterministic lateral displacement (DLD) for precise cell separation (see Fig. 5c). A separation efficiency of over 99.9% and a sample purity of 93.59% were achieved for the separation of tumor cells from blood. In addition, the multiplex spiral inertial microfluidic devices were cascaded to simply realize the improvement of separation performances (Kim et al. 2014; Miller et al. 2016) or to achieve the simultaneous separation of different types of tumor cells (Abdulla et al. 2018). For developing the low-cost cell separation instrument, Zhang et al. (2018) integrated the eight-core spiral inertial microfluidic channel and the passive flow regulators on a compact polymer-film chip. On the basis of the integrated chip, an automated microfluidic instrument (see Fig. 5d) was built to achieve the size-based separation of tumor cells in a label-free and high-throughput manner. The detailed information on the studies in this section can be found in Table 2. In addition to the CTCs, the spiral inertial microfluidics has also been widely used for the isolation of other blood components. For example, Nivedita and Papautsky (2013) employed the spiral inertial microfluidics to sort the erythrocytes from the leukocytes at a separation efficiency of ~95% (0.1% Hct) and a throughput of 1.8 ml/ min. Wu et al. (2012) realized the recovery of polymorphonuclear leukocytes (PMNs) and mononuclear leukocytes (MNLs) from blood (1~2% Hct) at an efficiency of over 80% using the trapezoidal spiral inertial microfluidics. As compared with traditional methods, the PMNs enriched by the spiral inertial microfluidics showed negligible activation. Through combining with selective cell lysis, Ramachandraiah et al. (2017) realized the fractionation of blood cells into subpopulation (purities of 86%, 43%, and 91%, respectively, for granulocyte, monocyte, and lymphocyte). The fractionation of blood components using spiral inertial microfluidics has been used as an important pretreatment for various disease diagnoses. For diabetes testing, monocytes/monocyte
platelet aggregates (MPA) were isolated from the peripheral blood mononuclear cells (PBMCs) using the DFF chip (Tay et al. 2018). An enhanced MPA detection sensitivity in the blood of type 2 diabetes mellitus patients was realized. In another work from the same group, they firstly size-fractionated the leukocytes into different subtypes (neutrophils, monocytes, lymphocytes), and the dielectric properties of the sorted neutrophil/monocyte were characterized using the single-cell impedance measurement for diabetes testing (Petchakup et al. 2018) (see Fig. 6a). For the noninvasive prenatal testing, the isolation of trophoblastic cells from maternal peripheral blood was realized using the trapezoidal spiral inertial microfluidics, and a separation efficiency of ~79% was achieved using a trophoblastic cell line (Winter et al. 2018). In addition to the blood sample, Ryu et al. (2017) employed the closed-loop spiral inertial microfluidics (see Fig. 6b) to recover 94.0% of polymorphonuclear leukocytes (PMNs) from the airway secretions obtained from mechanically ventilated patients. Stem cells are self-renewing undifferentiated cells that can differentiate into specialized types of cells. Recently, the spiral inertial microfluidics has been proven to be able to isolate the stem cells. Nathamgari et al. (2015) demonstrated the

172

N. Xiang and Z. Ni

Table 2 Advances on CTC separation using spiral inertial microfluidics Channel type Double spiral

DFF

Trapezoidal spiral

Trapezoidal spiral

Double spiral + membrane filter

Cross-sectional dimension Sample Throughput 300  50 μm MCF-7 and 20 ml/h (w  h) HeLa cells (3.33  107 cells/ (100 tumor min) cells per million blood cells) 500  160 μm MCF-7, 3 ml/h (w  h) MDA-MB231, and HeLa cells; Blood samples from patients with metastatic lung cancer 1.7 ml/min 600  80 μm, MCF-7, T24, and 130 μm MDA-MB(w  h1, h2) 231 cells; blood samples (n = 10) from patients with advancedstage metastatic breast and lung cancer NA Blood 1.7 ml/min samples (n = 24) from patients with head and neck cancers 300  40 μm A549 cells; 25 ml/h (w  h) blood samples from patients with metastatic lung cancer

Performance Recovery ratio of 88.5%

Reference Sun et al. (2012)

Recovery ratio of >85% for tumor cells; 100% detection rate (n = 20) for patient samples

Hou et al. (2013)

Recovery ratio of 80% for tumor cells; 100% detection rate (n = 10) for patient samples

Warkiani et al. (2014a)

54% detection rate (n = 24) for patient samples

Kulasinghe et al. (2017)

Capture efficiency of 74.4% (tens of A549 cells per mL blood)

Wang et al. (2015)

(continued)

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

173

Table 2 (continued) Channel type Spiral + DLD

Cross-sectional dimension Sample 150  50 μm MCF-7 (w  h) cells

Throughput 400 μl/min

Performance Sample purity of 93.59%; separation efficiency of over 99.9%

Reference Xiang et al. (2019b)

Note: h1 and h2 are, respectively, the short and long heights of the trapezoidal cross section. Trapezoidal spiral means the Archimedean spiral with the trapezoidal cross section (the same below)

Fig. 6 (a) Size fractionation and single-cell impedance measurement of leukocyte subtypes for diabetes testing. (Reprinted from Petchakup et al. (2018). Copyright (2018), with permission from Elsevier). (b) Closed-loop spiral inertial microfluidics for recovering polymorphonuclear leukocytes (PMNs) from the airway secretions. (Reprinted with permission from Ryu et al. (2017). Copyright (2017) American Chemical Society)

isolation of single neural stem cells from the mixture of single cells and clusters from chemically dissociated neurospheres. The sorted single stem cell population can successfully differentiate into neurons and astrocytes. Later, the spiral inertial microfluidics was employed for the enrichment of mesenchymal stem cells from marrow samples at a recovery ratio of 73% (Lee et al. 2018a).

174

N. Xiang and Z. Ni

In addition to the above biological cells, the spiral inertial microfluidics has also been employed for the separation of microswimmers. For example, the separation of sperm cells from the RBCs and WBCs has been reported (Son et al. 2015, 2017), which may beneficial for improving the effectiveness of the assisted reproductive technologies (ART). Microalga is one of the most promising new sources of biomass. Syed et al. (2018) used the trapezoidal spiral inertial microfluidics to separate Tetraselmis suecica (lipid-rich microalgae) from Phaeodactylum tricornutum (invasive diatom). As the cells at different stages of the cell replication cycle may have different cell diameters or shapes, the spiral inertial microfluidics can thus be used for cell cycle synchronization. Lee et al. (2011) realized the separation of bone marrow-derived human mesenchymal stem cells (hMSCs) into subpopulations of G0/G1 (>85%), S, and G2/M phases. Sofela et al. (2018) utilized the trapezoidal spiral inertial microfluidics to sort the eggs of C. elegans from the mixed-age nematode population. The sorted eggs then can be cultured to a desired developmental stage. The detailed information on the studies in this section can be found in Table 3.

Concentration/Microfiltration Concentration is an essential pretreatment for the single-cell analysis of rare cells in large-volume samples. The spiral inertial microfluidics can be employed to achieve the high-throughput and continuous cell concentration through focusing the cells and then removing the cell-free fluids (i.e., volume reduction). Seo et al. (2007a, b) developed a double spiral channel device for the membrane-free concentration of particles at the maximum flow velocity of 92 mm/s. Burke et al. (2014) developed a spiral inertial filtration (SIFT) device which is able to achieve the high-throughput (up to 1 ml/min), high-purity separation of the spiked MCF-7 cells from WBCs while removing 93% of the sample volume. Clime et al. (2016) integrated the spiral inertial microfluidics with the peristaltic microvalves for the concentration of Phytophthora ramorum pathogens with a recovery of 95% after three circulating cycles. Instead of prototype PDMS devices in laboratories, Xiang et al. (2018b) developed an inertial microfluidic syringe cell (IMSC) concentrator through assembling the upper housing, a circular gasket, and a lower housing with a concave spiral channel. The fabricated concentrator was then applied as the “centrifugation on a syringe tip” for concentrating various particles/cells at a high throughput of 3 ml/ min. Later, they further integrated a syringe-tip flow stabilizer with the IMSC concentrator for hand-powered applications (Xiang et al. 2019a), as illustrated in Fig. 7a. The integration of a stabilizer makes the performance of the IMSC concentrator totally independent on the pushing operations. Instead of removing the cell-free fluids to reduce the volume in concentration applications, the microfiltration process uses the inertial focusing to remove the

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

175

Table 3 Advances on the separation of other cells using spiral inertial microfluidics Category Blood fractionation

Channel type and dimension Archimedean spiral, 500  110 μm (w  h) Trapezoidal spiral, 500  70 μm, 100 μm (w  h1, h2) Archimedean spiral + RBC lysis, 500  115 μm (w  h) DFF, 500  115 μm (w  h)

Trapezoidal spiral, NA Immune cell enrichment

Trapezoidal spiral, NA

Stem cell isolation

Archimedean spiral, 500  150 μm (w  h)

Sperm enrichment

Archimedean spiral, 500  160 μm (w  h) Archimedean spiral, 150  50 μm (w  h) Archimedean spiral, 150  50 μm (w  h)

Sample Erythrocytes and leukocytes

Throughput 1.8 ml/min

Reference Nivedita and Papautsky (2013)

Polymorphonuclear leukocytes (PMNs) and mononuclear leukocytes (MNLs) from blood Leukocyte subpopulations (granulocyte, monocyte, and lymphocyte) Monocyte/ monocyte
platelet aggregates (MPA) from peripheral blood mononuclear cells Trophoblastic cells from maternal peripheral blood Polymorphonuclear leukocytes (PMNs) from the airway secretions Single neural stem cells from chemically dissociated neurospheres Mesenchymal stem cells from marrow samples

1 ml/min

Wu et al. (2012)

1 ml/min

Ramachandraiah et al. (2017)

130 μl/min

Tay et al. (2018)

NA

Winter et al. (2018)

4 ml/min

Ryu et al. (2017)

1 ml/min

Nathamgari et al. (2015)

1.6 ml/min

Lee et al. (2018a)

Sperm cells from RBCs

0.52 ml/ min

Son et al. (2015)

Sperm cells from WBCs

0.52 ml/ min

Son et al. (2017)

(continued)

176

N. Xiang and Z. Ni

Table 3 (continued) Category Microalgal separation

Cell cycle synchronization

Channel type and dimension Trapezoidal spiral, 600  80 μm, 130 μm (w  h1, h2) Archimedean spiral, 500  200 μm (w  h) Trapezoidal spiral, 1000  160 μm, 220 μm (w  h1, h2)

Sample Tetraselmis suecica from Phaeodactylum tricornutum

Throughput 1 ml/min

Reference Syed et al. (2018)

Bone marrowderived human mesenchymal stem cells Eggs of C. elegans from the mixed-age nematode population

1.5  107 cell/h

Lee et al. (2011)

6 ml/min

Sofela et al. (2018)

focused cell string and recovers the cell-free fluids. As compared with traditional membrane-based filtration methods, microfiltration using spiral inertial microfluidics offers advantages of high-efficiency operation without clogging and highthroughput continuous-flow processing. One of the most attracting microfiltration applications is the isolation of blood plasma for disease diagnosis. Geng et al. (2013b) used the spiral channels embedded with micropillar arrays to extract the plasma from the blood. Later, they also employed their device for the simultaneously isolation of different blood components (Geng et al. 2013a). However, the use of micropillars may still have the clogging disadvantage of the traditional membrane filter. Directly using the spiral inertial microfluidics for plasma isolation would be more attracting. Xiang and Ni (2015) employed the spiral inertial microfluidics for the isolation of plasma from blood at the flow rate of 700 μm/min (see Fig. 7b). A plasma yield of 38.5% and a purity of 100% were achieved using their device. To increase the processing throughput, Rafeie et al. (2016) developed a blood plasma separator through integrating 16 trapezoidal spiral channels for providing a high throughput of 24 ml/min. In addition to the plasma isolation, spiral inertial microfluidic microfiltration has also been applied for various industrial applications. For example, Warkiani et al. (2015) developed a membraneless microfiltration system through massively parallelizing and vertically stacking of trapezoidal spiral inertial microfluidic channels. They then applied the developed system for the filtration of CHO and yeast cells from large-scale bioreactors at an extremely high throughput of ~500 ml/min (see Fig. 7c). Later, the work from the same group further applied their system as highefficiency cell retention tool for perfusion culture of suspended mammalian cells (Kwon et al. 2017). The detailed information on the studies in this section can be found in Table 4.

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

177

2

Plasma

WBCs

1

Blood cells + Plasma RBCs

Sample collected from inner outlet

O2 and CO2 supply

Sample collected from outer outlet

Cell suspension

P1

P2

Pump

Pump

Clarified culture medium

Concentrated cell recycle Fresh feed

Spinner flask

Inertial filteration system

Fig. 7 (a) Integration of an inertial microfluidic syringe cell (IMSC) concentrator with a syringetip flow stabilizer for hand-powered cell concentrations. (Reprinted with permission from Xiang et al. (2019a). Published by The Royal Society of Chemistry). (b) Spiral inertial microfluidics for the isolation of plasma from blood. (Reprinted with permission from Xiang and Ni (2015)). (c) Membraneless microfiltration system fabricated through massively parallelizing of trapezoidal spiral inertial microfluidic channels. (Reprinted from Warkiani et al. (2015) under a Creative Commons Attribution 4.0 International License)

Conclusion and Future Perspective Spiral inertial microfluidics has attracted increasing interests over the recent years and has been widely employed as an important sample pretreatment technique for single-cell analysis. In this chapter, the underlying physics, the design guidelines, the improved understanding, and the recent advances on spiral inertial microfluidics are provided. Although great successes have been achieved, effects may still be strongly required in the following aspects. The force models on cells in spiral inertial microfluidics are still unclear. In previous studies, force scalings for macroscopic flows were commonly used, and the cell migration was still interpreted through using many hypotheses. To deepen

178

N. Xiang and Z. Ni

Table 4 Advances on concentration and microfiltration using spiral inertial microfluidics Category Concentration

Microfiltration

Channel type and dimension Double spiral, 12-loop PDMS device, 300  100 μm (w  h) Archimedean spiral, 7-loop PDMS device, 250  50 μm (w  h)

Archimedean spiral, 3-loop thermoplastic polymer device, 600  200 μm (w  h) Archimedean spiral, 3-loop 3D-printed device, 500  100 μm (w  h) Archimedean spiral, 5-loop PDMS device, 150  50 μm (w  h) Trapezoidal spiral, 4-loop PDMS device, 500  40 μm, 70 μm (wh1, h2) Trapezoidal spiral, 8-loop PDMS device, CHO cells: 600  80 μm, 130 μm (wh1, h2); yeast cells: 450  30 μm, 70 μm (w  h1, h2)

Sample 10.5 μm polystyrene particles 15 μm and 8 μm polystyrene particles; MCF-7 cells and WBCs P. ramorum sporangia

Throughput Flow velocity of 92 mm/s

Reference Seo et al. (2007a, b)

1 ml/min

Burke et al. (2014)

2 ml/min

Clime et al. (2016)

Pollen particles and MCF-7 cells Blood

3 ml/min

Xiang et al. (2018b, 2019a) Xiang and Ni (2015)

Blood

1.5 ml/min for single spiral, 24 ml/min for the integrated device ~500 ml/min

CHO and yeast cells

700 μl /min

Rafeie et al. (2016)

Warkiani et al. (2015) and Kwon et al. (2017)

the theoretical understandings and force models, the software that can accurately predict the cell migration should be developed to inform the design of new devices or the optimization of operational parameters. The manipulation resolution of the passive spiral inertial microfluidics is relatively low when comparing with active manipulation methods. To address this issue, some attempts have been made to couple the spiral inertial microfluidics with other manipulation techniques. In the future, the inertial-based hybrid techniques that inherit the advantages of different techniques are still in urgent demands. In addition, the spiral inertial microfluidics can be integrated with various other on-chip detection methods for point of care testing. There are a large number of prototype devices for various sample pretreatments being reported each year. However, the translation of these devices into commercial outcomes is still lacking. In the future, the killer channel designs and applications of spiral inertial microfluidics are still needed.

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

179

References Abdulla A, Liu W, Gholamipour-Shirazi A, Sun J, Ding X (2018) High-throughput isolation of circulating tumor cells using cascaded inertial focusing microfluidic channel. Anal Chem 90:4397–4405. https://doi.org/10.1021/acs.analchem.7b04210 Amini H, Lee W, Di Carlo D (2014) Inertial microfluidic physics. Lab Chip 14:2739–2761. https:// doi.org/10.1039/C4LC00128A Asmolov ES (1999) The inertial lift on a spherical particle in a plane Poiseuille flow at large channel Reynolds number. J Fluid Mech 381:63–87. https://doi.org/10.1017/S0022112098003474 Asmolov ES, Dubov AL, Nizkaya TV, Harting J, Vinogradova OI (2018) Inertial focusing of finitesize particles in microchannels. J Fluid Mech 840:613–630. https://doi.org/10.1017/jfm.2018.95 Berger SA, Talbot L, Yao LS (1983) Flow in curved pipes. Annu Rev Fluid Mech 15:461–512. https://doi.org/10.1146/annurev.fl.15.010183.002333 Bhagat AAS, Kuntaegowdanahalli SS, Papautsky I (2008) Continuous particle separation in spiral microchannels using Dean flows and differential migration. Lab Chip 8:1906–1914. https://doi. org/10.1039/B807107A Bhagat AAS, Kuntaegowdanahalli SS, Kaval N, Seliskar CJ, Papautsky I (2010) Inertial microfluidics for sheath-less high-throughput flow cytometry. Biomed Microdevices 12:187–195. https://doi.org/10.1007/s10544-009-9374-9 Burke JM, Zubajlo RE, Smela E, White IM (2014) High-throughput particle separation and concentration using spiral inertial filtration. Biomicrofluidics 8:024105. https://doi.org/ 10.1063/1.4870399 Chen Q, Li D, Malekanfard A, Cao Q, Lin J, Wang M, Han X, Xuan X (2018) Tunable, sheathless focusing of diamagnetic particles in ferrofluid microflows with a single set of overhead permanent magnets. Anal Chem 90:8600–8606. https://doi.org/10.1021/acs.analchem.8b01813 Chhabra RP (2006) Bubbles, drops, and particles in non-Newtonian fluids. CRC press, New York Choi S, Song S, Choi C, Park J-K (2008) Sheathless focusing of microbeads and blood cells based on hydrophoresis. Small 4:634–641. https://doi.org/10.1002/smll.200700308 Clime L, Li K, Geissler M, Hoa XD, Robideau GP, Bilodeau GJ, Veres T (2016) Separation and concentration of Phytophthora ramorum sporangia by inertial focusing in curving microfluidic flows. Microfluid Nanofluid 21:5. https://doi.org/10.1007/s10404-016-1844-9 Cox RG, Brenner H (1968) The lateral migration of solid particles in Poiseuille flow – I theory. Chem Eng Sci 23:147–173. https://doi.org/10.1016/0009-2509(68)87059-9 D’Avino G, Maffettone PL (2015) Particle dynamics in viscoelastic liquids. J Non-Newtonian Fluid Mech 215:80–104. https://doi.org/10.1016/j.jnnfm.2014.09.014 Dean WR, Chapman S (1928) Fluid motion in a curved channel. Proc R Soc Lond Ser A 121:402–420. https://doi.org/10.1098/rspa.1928.0205 Di Carlo D (2009) Inertial microfluidics. Lab Chip 9:3038–3046. https://doi.org/10.1039/ B912547G Di Carlo D, Irimia D, Tompkins RG, Toner M (2007) Continuous inertial focusing, ordering, and separation of particles in microchannels. Proc Natl Acad Sci 104:18892. https://doi.org/ 10.1073/pnas.0704958104 Di Carlo D, Edd JF, Humphry KJ, Stone HA, Toner M (2009) Particle segregation and dynamics in confined flows. Phys Rev Lett 102:094503. https://doi.org/10.1103/PhysRevLett.102.094503 Geng Z, Ju Y, Wang Q, Wang W, Li Z (2013a) Multi-component continuous separation chip composed of micropillar arrays in a split-level spiral channel. RSC Adv 3:14798–14806. https://doi.org/10.1039/C3RA41906A Geng Z, Ju Y, Wang W, Li Z (2013b) Continuous blood separation utilizing spiral filtration microchannel with gradually varied width and micro-pillar array. Sensors Actuators B Chem 180:122–129. https://doi.org/10.1016/j.snb.2012.06.064 Ghadami S, Kowsari-Esfahan R, Saidi MS, Firoozbakhsh K (2017) Spiral microchannel with stairlike cross section for size-based particle separation. Microfluid Nanofluid 21:115. https://doi. org/10.1007/s10404-017-1950-3

180

N. Xiang and Z. Ni

Guan G, Wu L, Bhagat AA, Li Z, Chen PCY, Chao S, Ong CJ, Han J (2013) Spiral microchannel with rectangular and trapezoidal cross-sections for size based particle separation. Sci Rep 3:1475. https://doi.org/10.1038/srep01475 Heikali D, Di Carlo D (2010) A niche for microfluidics in portable hematology analyzers. J Assoc Lab Autom 15:319–328. https://doi.org/10.1016/j.jala.2010.02.005 Ho BP, Leal LG (1974) Inertial migration of rigid spheres in two-dimensional unidirectional flows. J Fluid Mech 65:365–400. https://doi.org/10.1017/S0022112074001431 Hosic S, Murthy SK, Koppes AN (2016) Microfluidic sample preparation for single cell analysis. Anal Chem 88:354–380. https://doi.org/10.1021/acs.analchem.5b04077 Hou HW, Warkiani ME, Khoo BL, Li ZR, Soo RA, Tan DS-W, Lim W-T, Han J, Bhagat AAS, Lim CT (2013) Isolation and retrieval of circulating tumor cells using centrifugal forces. Sci Rep 3:1259. https://doi.org/10.1038/srep01259 Huang Q, Mao S, Khan M, Zhou L, Lin J-M (2018) Dean flow assisted cell ordering system for lipid profiling in single-cells using mass spectrometry. Chem Commun 54:2595–2598. https://doi. org/10.1039/C7CC09608A Hur SC, Tse HTK, Di Carlo D (2010) Sheathless inertial cell ordering for extreme throughput flow cytometry. Lab Chip 10:274–280. https://doi.org/10.1039/B919495A Hur SC, Henderson-MacLennan NK, McCabe ERB, Di Carlo D (2011) Deformability-based cell classification and enrichment using inertial microfluidics. Lab Chip 11:912–920. https://doi.org/ 10.1039/C0LC00595A Jackson JM, Witek MA, Kamande JW, Soper SA (2017) Materials and microfluidics: enabling the efficient isolation and analysis of circulating tumour cells. Chem Soc Rev 46:4245–4280. https://doi.org/10.1039/C7CS00016B James DF (2008) Boger Fluids. Annu Rev Fluid Mech 41:129–142. https://doi.org/10.1146/ annurev.fluid.010908.165125 Kang K, Lee SS, Hyun K, Lee SJ, Kim JM (2013) DNA-based highly tunable particle focuser. Nat Commun 4:2567. https://doi.org/10.1038/ncomms3567 Karimi A, Yazdi S, Ardekani AM (2013) Hydrodynamic mechanisms of cell and particle trapping in microfluidics. Biomicrofluidics 7:021501. https://doi.org/10.1063/1.4799787 Kemna EWM, Schoeman RM, Wolbers F, Vermes I, Weitz DA, van den Berg A (2012) High-yield cell ordering and deterministic cell-in-droplet encapsulation using Dean flow in a curved microchannel. Lab Chip 12:2881–2887. https://doi.org/10.1039/C2LC00013J Khojah R, Stoutamore R, Di Carlo D (2017) Size-tunable microvortex capture of rare cells. Lab Chip 17:2542–2549. https://doi.org/10.1039/C7LC00355B Khoo BL, Warkiani ME, Tan DS-W, Bhagat AAS, Irwin D, Lau DP, Lim AST, Lim KH, Krisna SS, Lim W-T, Yap YS, Lee SC, Soo RA, Han J, Lim CT (2014) Clinical validation of an ultra highthroughput spiral microfluidics for the detection and enrichment of viable circulating tumor cells. PLoS One 9:e99409. https://doi.org/10.1371/journal.pone.0099409 Kim YJ, Ahn WS, Lee SS, Kim MJ (2012) Lateral migration and focusing of colloidal particles and DNA molecules under viscoelastic flow. Lab Chip 12:2807–2814. https://doi.org/10.1039/ C2LC40147A Kim TH, Yoon HJ, Stella P, Nagrath S (2014) Cascaded spiral microfluidic device for deterministic and high purity continuous separation of circulating tumor cells. Biomicrofluidics 8:064117. https://doi.org/10.1063/1.4903501 Kim D, Sonker M, Ros A (2019) Dielectrophoresis: from molecular to micrometer-scale analytes. Anal Chem 91:277–295. https://doi.org/10.1021/acs.analchem.8b05454 Kulasinghe A, Tran THP, Blick T, O’Byrne K, Thompson EW, Warkiani ME, Nelson C, Kenny L, Punyadeera C (2017) Enrichment of circulating head and neck tumour cells using spiral microfluidic technology. Sci Rep 7:42517. https://doi.org/10.1038/srep42517 Kuntaegowdanahalli SS, Bhagat AAS, Kumar G, Papautsky I (2009) Inertial microfluidics for continuous particle separation in spiral microchannels. Lab Chip 9:2973–2980. https://doi.org/ 10.1039/B908271A Kwon T, Prentice H, Oliveira JD, Madziva N, Warkiani ME, Hamel J-FP, Han J (2017) Microfluidic cell retention device for perfusion of mammalian suspension culture. Sci Rep 7:6703. https:// doi.org/10.1038/s41598-017-06949-8

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

181

Lee WC, Bhagat AAS, Huang S, Van Vliet KJ, Han J, Lim CT (2011) High-throughput cell cycle synchronization using inertial forces in spiral microchannels. Lab Chip 11:1359–1367. https:// doi.org/10.1039/C0LC00579G Lee LM, Rosano Jenna M, Wang Y, Klarmann GJ, Garson CJ, Prabhakarpandian B, Pant K, Alvarez LM, Lai E (2018a) Label-free mesenchymal stem cell enrichment from bone marrow samples by inertial microfluidics. Anal Methods 10:713–721. https://doi.org/10.1039/C7AY02500A Lee Y, Guan G, Bhagat AA (2018b) ClearCell ® FX, a label-free microfluidics technology for enrichment of viable circulating tumor cells. Cytometry A 93:1251–1254. https://doi.org/ 10.1002/cyto.a.23507 Leshansky AM, Bransky A, Korin N, Dinnar U (2007) Tunable nonlinear viscoelastic “focusing” in a microfluidic device. Phys Rev Lett 98:234501. https://doi.org/10.1103/PhysRevLett.98.234501 Lim EJ, Ober TJ, Edd JF, Desai SP, Neal D, Bong KW, Doyle PS, McKinley GH, Toner M (2014) Inertio-elastic focusing of bioparticles in microchannels at high throughput. Nat Commun 5:4120. https://doi.org/10.1038/ncomms5120 Liu C, Ding B, Xue C, Tian Y, Hu G, Sun J (2016a) Sheathless focusing and separation of diverse nanoparticles in viscoelastic solutions with minimized shear thinning. Anal Chem 88:12547–12553. https://doi.org/10.1021/acs.analchem.6b04564 Liu C, Xue C, Sun J, Hu G (2016b) A generalized formula for inertial lift on a sphere in microchannels. Lab Chip 16:884–892. https://doi.org/10.1039/C5LC01522G Liu C, Guo J, Tian F, Yang N, Yan F, Ding Y, Wei J, Hu G, Nie G, Sun J (2017) Field-free isolation of exosomes from extracellular vesicles by microfluidic viscoelastic flows. ACS Nano 11:6968–6976. https://doi.org/10.1021/acsnano.7b02277 Lu X, Xuan X (2015) Continuous microfluidic particle separation via elasto-inertial pinched flow fractionation. Anal Chem 87:6389–6396. https://doi.org/10.1021/acs.analchem.5b01432 Lu X, Liu C, Hu G, Xuan X (2017) Particle manipulations in non-Newtonian microfluidics: a review. J Colloid Interface Sci 500:182–201. https://doi.org/10.1016/j.jcis.2017.04.019 MacDonald MP, Spalding GC, Dholakia K (2003) Microfluidic sorting in an optical lattice. Nature 426:421–424. https://doi.org/10.1038/nature02144 Martel JM, Toner M (2012) Inertial focusing dynamics in spiral microchannels. Phys Fluids 24:032001. https://doi.org/10.1063/1.3681228 Martel JM, Toner M (2013) Particle focusing in curved microfluidic channels. Sci Rep 3:3340. https://doi.org/10.1038/srep03340 Matas J-P, Morris JF, Guazzelli É (2004) Inertial migration of rigid spherical particles in Poiseuille flow. J Fluid Mech 515:171–195. https://doi.org/10.1017/S0022112004000254 McGrath J, Jimenez M, Bridle H (2014) Deterministic lateral displacement for particle separation: a review. Lab Chip 14:4139–4158. https://doi.org/10.1039/C4LC00939H Miller B, Jimenez M, Bridle H (2016) Cascading and parallelising curvilinear inertial focusing systems for high volume, wide size distribution, separation and concentration of particles. Sci Rep 6:36386. https://doi.org/10.1038/srep36386 Moon H-S, Je K, Min J-W, Park D, Han K-Y, Shin S-H, Park W-Y, Yoo CE, Kim S-H (2018) Inertial-ordering-assisted droplet microfluidics for high-throughput single-cell RNA-sequencing. Lab Chip 18:775–784. https://doi.org/10.1039/C7LC01284E Nathamgari SSP, Dong B, Zhou F, Kang W, Giraldo-Vela JP, McGuire T, McNaughton RL, Sun C, Kessler JA, Espinosa HD (2015) Isolating single cells in a neurosphere assay using inertial microfluidics. Lab Chip 15:4591–4597. https://doi.org/10.1039/C5LC00805K Nguyen N-T, Wu Z (2004) Micromixers – a review. J Micromech Microeng 15:R1–R16. https://doi. org/10.1088/0960-1317/15/2/r01 Nilsson J, Evander M, Hammarström B, Laurell T (2009) Review of cell and particle trapping in microfluidic systems. Anal Chim Acta 649:141–157. https://doi.org/10.1016/j.aca.2009.07.017 Nivedita N, Papautsky I (2013) Continuous separation of blood cells in spiral microfluidic devices. Biomicrofluidics 7:054101. https://doi.org/10.1063/1.4819275 Nivedita N, Ligrani P, Papautsky I (2017) Dean flow dynamics in low-aspect ratio spiral microchannels. Sci Rep 7:44072. https://doi.org/10.1038/srep44072 Oomen PE, Aref MA, Kaya I, Phan NTN, Ewing AG (2019) Chemical analysis of single cells. Anal Chem 91:588–621. https://doi.org/10.1021/acs.analchem.8b04732

182

N. Xiang and Z. Ni

Petchakup C, Tay HM, Yeap WH, Dalan R, Wong SC, Li KHH, Hou HW (2018) Label-free leukocyte sorting and impedance-based profiling for diabetes testing. Biosens Bioelectron 118:195–203. https://doi.org/10.1016/j.bios.2018.07.052 Plaks V, Koopman CD, Werb Z (2013) Circulating tumor cells. Science 341:1186–1188. https://doi. org/10.1126/science.1235226 Rafeie M, Zhang J, Asadnia M, Li W, Warkiani ME (2016) Multiplexing slanted spiral microchannels for ultra-fast blood plasma separation. Lab Chip 16:2791–2802. https://doi.org/ 10.1039/C6LC00713A Ramachandraiah H, Svahn HA, Russom A (2017) Inertial microfluidics combined with selective cell lysis for high throughput separation of nucleated cells from whole blood. RSC Adv 7:29505–29514. https://doi.org/10.1039/C7RA02992F Rodd LE, Cooper-White JJ, Boger DV, McKinley GH (2007) Role of the elasticity number in the entry flow of dilute polymer solutions in micro-fabricated contraction geometries. J Non-Newtonian Fluid Mech 143:170–191. https://doi.org/10.1016/j.jnnfm.2007.02.006 Romeo G, D’Avino G, Greco F, Netti PA, Maffettone PL (2013) Viscoelastic flow-focusing in microchannels: scaling properties of the particle radial distributions. Lab Chip 13:2802–2807. https://doi.org/10.1039/C3LC50257K Ryu H, Choi K, Qu Y, Kwon T, Lee JS, Han J (2017) Patient-derived airway secretion dissociation technique to isolate and concentrate immune cells using closed-loop inertial microfluidics. Anal Chem 89:5549–5556. https://doi.org/10.1021/acs.analchem.7b00610 Saffman PG (1965) The lift on a small sphere in a slow shear flow. J Fluid Mech 22:385–400. https://doi.org/10.1017/S0022112065000824 Sajeesh P, Sen AK (2014) Particle separation and sorting in microfluidic devices: a review. Microfluid Nanofluid 17:1–52. https://doi.org/10.1007/s10404-013-1291-9 Schonberg JA, Hinch EJ (1989) Inertial migration of a sphere in Poiseuille flow. J Fluid Mech 203:517–524. https://doi.org/10.1017/S0022112089001564 Segre G, Silberberg A (1961) Radial particle displacements in Poiseuille flow of suspensions. Nature 189:209–210. https://doi.org/10.1038/189209a0 Seo J, Lean MH, Kole A (2007a) Membrane-free microfiltration by asymmetric inertial migration. Appl Phys Lett 91:033901. https://doi.org/10.1063/1.2756272 Seo J, Lean MH, Kole A (2007b) Membraneless microseparation by asymmetry in curvilinear laminar flows. J Chromatogr A 1162:126–131. https://doi.org/10.1016/j.chroma.2007.05.110 Seo KW, Kang YJ, Lee SJ (2014) Lateral migration and focusing of microspheres in a microchannel flow of viscoelastic fluids. Phys Fluids 26:063301. https://doi.org/10.1063/1.4882265 Shapiro HM (1983) Multistation multiparameter flow cytometry: a critical review and rationale. Cytometry 3:227–243. https://doi.org/10.1002/cyto.990030402 Shen S, Kou L, Zhang X, Wang D, Niu Y, Wang J (2018) Regulating secondary flow in ultra-low aspect ratio microchannels by dimensional confinement. Adv Theory Simul 1:1700034. https:// doi.org/10.1002/adts.201700034 Sofela S, Sahloul S, Rafeie M, Kwon T, Han J, Warkiani ME, Song Y-A (2018) High-throughput sorting of eggs for synchronization of C. elegans in a microfluidic spiral chip. Lab Chip 18:679–687. https://doi.org/10.1039/C7LC00998D Son J, Murphy K, Samuel R, Gale BK, Carrell DT, Hotaling JM (2015) Non-motile sperm cell separation using a spiral channel. Anal Methods 7:8041–8047. https://doi.org/10.1039/ C5AY02205C Son J, Samuel R, Gale BK, Carrell DT, Hotaling JM (2017) Separation of sperm cells from samples containing high concentrations of white blood cells using a spiral channel. Biomicrofluidics 11:054106. https://doi.org/10.1063/1.4994548 Song Y, Tian T, Shi Y, Liu W, Zou Y, Khajvand T, Wang S, Zhu Z, Yang C (2017) Enrichment and single-cell analysis of circulating tumor cells. Chem Sci 8:1736–1751. https://doi.org/10.1039/ C6SC04671A Squires TM, Quake SR (2005) Microfluidics: fluid physics at the nanoliter scale. Rev Mod Phys 77:977–1026. https://doi.org/10.1103/RevModPhys.77.977

7

Inertial Microfluidics for Single-Cell Manipulation and Analysis

183

Stoecklein D, Di Carlo D (2019) Nonlinear microfluidics. Anal Chem 91:296–314. https://doi.org/ 10.1021/acs.analchem.8b05042 Sun J, Li M, Liu C, Zhang Y, Liu D, Liu W, Hu G, Jiang X (2012) Double spiral microchannel for label-free tumor cell separation and enrichment. Lab Chip 12:3952–3960. https://doi.org/ 10.1039/C2LC40679A Syed MS, Rafeie M, Vandamme D, Asadnia M, Henderson R, Taylor RA, Warkiani ME (2018) Selective separation of microalgae cells using inertial microfluidics. Bioresour Technol 252:91–99. https://doi.org/10.1016/j.biortech.2017.12.065 Tay HM, Yeap WH, Dalan R, Wong SC, Hou HW (2018) Multiplexed label-free fractionation of peripheral blood mononuclear cells for identification of monocyte–platelet aggregates. Anal Chem 90:14535–14542. https://doi.org/10.1021/acs.analchem.8b04415 Wang J, Lu W, Tang C, Liu Y, Sun J, Mu X, Zhang L, Dai B, Li X, Zhuo H, Jiang X (2015) Labelfree isolation and mRNA detection of circulating tumor cells from patients with metastatic lung cancer for disease diagnosis and monitoring therapeutic efficacy. Anal Chem 87:11893–11900. https://doi.org/10.1021/acs.analchem.5b03484 Warkiani ME, Guan G, Luan KB, Lee WC, Bhagat AAS, Kant Chaudhuri P, Tan DS-W, Lim WT, Lee SC, Chen PCY, Lim CT, Han J (2014a) Slanted spiral microfluidics for the ultra-fast, labelfree isolation of circulating tumor cells. Lab Chip 14:128–137. https://doi.org/10.1039/ C3LC50617G Warkiani ME, Khoo BL, Tan DS-W, Bhagat AAS, Lim W-T, Yap YS, Lee SC, Soo RA, Han J, Lim CT (2014b) An ultra-high-throughput spiral microfluidic biochip for the enrichment of circulating tumor cells. Analyst 139:3245–3255. https://doi.org/10.1039/C4AN00355A Warkiani ME, Tay AKP, Guan G, Han J (2015) Membrane-less microfiltration using inertial microfluidics. Sci Rep 5:11018. https://doi.org/10.1038/srep11018 Winter M, Hardy T, Rezaei M, Nguyen V, Zander-Fox D, Ebrahimi Warkiani M, Thierry B (2018) Isolation of circulating fetal trophoblasts using inertial microfluidics for noninvasive prenatal testing. Adv Mater Technol 3:1800066. https://doi.org/10.1002/admt.201800066 Wu L, Guan G, Hou HW, Bhagat AAS, Han J (2012) Separation of leukocytes from blood using spiral channel with trapezoid cross-section. Anal Chem 84:9324–9331. https://doi.org/10.1021/ ac302085y Xiang N, Ni Z (2015) High-throughput blood cell focusing and plasma isolation using spiral inertial microfluidic devices. Biomed Microdevices 17:110. https://doi.org/10.1007/s10544-015-0018-y Xiang N, Chen K, Sun D, Wang S, Yi H, Ni Z (2013a) Quantitative characterization of the focusing process and dynamic behavior of differently sized microparticles in a spiral microchannel. Microfluid Nanofluid 14:89–99. https://doi.org/10.1007/s10404-012-1025-4 Xiang N, Yi H, Chen K, Sun D, Jiang D, Dai Q, Ni Z (2013b) High-throughput inertial particle focusing in a curved microchannel: insights into the flow-rate regulation mechanism and process model. Biomicrofluidics 7:044116. https://doi.org/10.1063/1.4818445 Xiang N, Chen K, Dai Q, Jiang D, Sun D, Ni Z (2015a) Inertia-induced focusing dynamics of microparticles throughout a curved microfluidic channel. Microfluid Nanofluid 18:29–39. https://doi.org/10.1007/s10404-014-1395-x Xiang N, Shi Z, Tang W, Huang D, Zhang X, Ni Z (2015b) Improved understanding of particle migration modes in spiral inertial microfluidic devices. RSC Adv 5:77264–77273. https://doi. org/10.1039/C5RA13292D Xiang N, Zhang X, Dai Q, Cheng J, Chen K, Ni Z (2016) Fundamentals of elasto-inertial particle focusing in curved microfluidic channels. Lab Chip 16:2626–2635. https://doi.org/10.1039/ C6LC00376A Xiang N, Ni Z, Yi H (2018a) Concentration-controlled particle focusing in spiral elasto-inertial microfluidic devices. Electrophoresis 39:417–424. https://doi.org/10.1002/elps.201700150 Xiang N, Shi X, Han Y, Shi Z, Jiang F, Ni Z (2018b) Inertial microfluidic syringe cell concentrator. Anal Chem 90:9515–9522. https://doi.org/10.1021/acs.analchem.8b02201 Xiang N, Han Y, Jia Y, Shi Z, Yi H, Ni Z (2019a) Flow stabilizer on a syringe tip for hand-powered microfluidic sample injection. Lab Chip 19:214–222. https://doi.org/10.1039/C8LC01051J

184

N. Xiang and Z. Ni

Xiang N, Wang J, Li Q, Han Y, Huang D, Ni Z (2019b) Precise size-based cell separation via the coupling of inertial microfluidics and deterministic lateral displacement. Anal Chem 91:10328. https://doi.org/10.1021/acs.analchem.9b02863 Xiang N, Zhang R, Han Y, Ni Z (2019c) A multilayer polymer-film inertial microfluidic device for high-throughput cell concentration. Anal Chem 91:5461–5468. https://doi.org/10.1021/acs. analchem.9b01116 Xuan X, Zhu J, Church C (2010) Particle focusing in microfluidic devices. Microfluid Nanofluid 9:1–16. https://doi.org/10.1007/s10404-010-0602-7 Yamada M, Seki M (2005) Hydrodynamic filtration for on-chip particle concentration and classification utilizing microfluidics. Lab Chip 5:1233–1239. https://doi.org/10.1039/B509386D Yang S, Kim JY, Lee SJ, Lee SS, Kim JM (2011) Sheathless elasto-inertial particle focusing and continuous separation in a straight rectangular microchannel. Lab Chip 11:266–273. https://doi. org/10.1039/C0LC00102C Yeo LY, Friend JR (2014) Surface acoustic wave microfluidics. Annu Rev Fluid Mech 46:379–406. https://doi.org/10.1146/annurev-fluid-010313-141418 Yin H, Marshall D (2012) Microfluidics for single cell analysis. Curr Opin Biotechnol 23:110–119. https://doi.org/10.1016/j.copbio.2011.11.002 Yousuff CM, Hamid NHB, Basha IHK, Ho ETW (2017) Output channel design for collecting closely-spaced particle streams from spiral inertial separation devices. AIP Adv 7:085004. https://doi.org/10.1063/1.4986617 Yuan D, Zhao Q, Yan S, Tang S-Y, Alici G, Zhang J, Li W (2018) Recent progress of particle migration in viscoelastic fluids. Lab Chip 18:551–567. https://doi.org/10.1039/C7LC01076A Zhang J, Yan S, Yuan D, Alici G, Nguyen N-T, Ebrahimi Warkiani M, Li W (2016) Fundamentals and applications of inertial microfluidics: a review. Lab Chip 16:10–34. https://doi.org/10.1039/ C5LC01159K Zhang X, Zhu Z, Xiang N, Long F, Ni Z (2018) Automated microfluidic instrument for label-free and high-throughput cell separation. Anal Chem 90:4212–4220. https://doi.org/10.1021/acs. analchem.8b00539 Zheng S, Lin H, Liu J-Q, Balic M, Datar R, Cote RJ, Tai Y-C (2007) Membrane microfilter device for selective capture, electrolysis and genomic analysis of human circulating tumor cells. J Chromatogr A 1162:154–161. https://doi.org/10.1016/j.chroma.2007.05.064 Zhou J, Papautsky I (2013) Fundamentals of inertial focusing in microchannels. Lab Chip 13:1121–1132. https://doi.org/10.1039/C2LC41248A Zhou Y, Ma Z, Tayebi M, Ai Y (2019) Submicron particle focusing and exosome sorting by wavy microchannel structures within viscoelastic fluids. Anal Chem 91:4577–4584. https://doi.org/ 10.1021/acs.analchem.8b05749

8

Digital Microfluidics for Single Cell Manipulation and Analysis Long Pang, Jing Ding, and Shih-Kang Fan

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DMF Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DMF Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fabrication of DMF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . DMF Manipulation and Analysis of a Single Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Adherent Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Suspension Cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Long-Term Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Related Applications of DMF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell Sorting and Concentrating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3D Cell Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Diagnosis and Clinical Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions and Future Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

186 187 187 191 192 193 193 195 197 197 199 201 201 202

Abstract

An investigation of cells at the level of a single cell has been recently possible and widely adopted in clinical and biological studies. Microfluidic systems for singlecell research emerged as feasible approaches with optimized microfluidic techniques. A microfluidic platform can generally be employed to operate on a single L. Pang College of Basic Medical Science, The Shaanxi Key Laboratory of Brain Disorders, Xi’an Medical University, Xi’an, China J. Ding Department of Mechanical and Nuclear Engineering, Kansas State University, Manhattan, KS, USA S.-K. Fan (*) Department of Mechanical and Nuclear Engineering, Kansas State University, Manhattan, KS, USA e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_41

185

186

L. Pang et al.

cell with various means including electric and hydrodynamics forces. Digital microfluidics (DMF), capable of manipulating individual cells in droplets, is one method that can address major challenges in the analysis of a single cell. On a DMF device, the tiny volume, for example, picoliter to microliter, of droplets allows agile actuations on adjusting the electric field across a driving plane surface. This chapter focuses primarily on the manipulation of a single cell and its analysis with DMF technologies. The long-term culture of single cells on a DMF device is then discussed. DMF methods for other applications of cell manipulation are introduced. Lastly, the challenges for future DMF studies of a single cell are highlighted. Keywords

Digital microfluidics · Single cell manipulation · Cellular heterogeneity · Cell culture · 3D organoid

Introduction Cellular heterogeneity is known to exist in diverse aspects, for example, mechanical characteristic, gene expression, and growth behavior (Lawson et al. 2018). In conventional cellular studies, only an average response from a group of cells is ascertained, overlooking subtle but important information from cellular heterogeneity. With the ability to examine a single cell individually, the heterogeneity could be revealed. In general, studies of a single cell involve several fundamental steps, such as cell isolation, tracking, labeling, imaging, macromolecule amplification, as well as measurement and data analysis (Sims and Allbritton 2007; Pang et al. 2016). Conventional analysis of a single cell typically requires dilution, flow cytometry, and fluorescence activated cell sorting (FACS) (Cheng et al. 2016) that are laborintensive and time-consuming. In addition, complex optical alignments using bulky and sophisticated devices are generally necessary to position and to localize a single cell (Yalcin et al. 2016). In recent years, varied microfluidic techniques for the analysis and manipulation of a single cell have been reported, such as capture of a single cell for sample pretreatment, multiple biological detection, and clonal cultivation of a single cell, because of the features of microfluidics on the flexible design of functional microstructures, the high-throughput capacity, and a length scale comparable with that of a cell (Gao et al. 2019). Hydrodynamic forces are commonly employed to manipulate cells when using microfluidic devices consisting of microchannel and microstructures, but these microchannels and microstructures generally require complicated processes to fabricate and are potentially clogged during operation. Other than hydrodynamic means, digital microfluidic (DMF) devices (Pang et al. 2019a) provide an alternative approach to studies of a single cell. With DMF, one manipulates a single cell in an electrically actuated droplet sandwiched between two plates without microchannels, microstructures, microvalves, or other mechanical microcomponents

8

Digital Microfluidics for Single Cell Manipulation and Analysis

187

(Fair 2007). DMF devices are hence easier to manufacture and avoid cell clogging. Various methods of detection are readily integrated in DMF single-cell devices (He et al. 2015). This chapter describes the manipulation and analysis of a single cell with DMF. Specifically, Section “DMF Systems” presents the principle of DMF and introduces briefly their application. Section “DMF Manipulation and Analysis of a Single Cell” explains the spatial and temporal manipulations and analysis of single cells. Section “Related Applications of DMF” presents related applications of DMF for cell manipulation and analysis. Last, Section “Conclusions and Future Outlook” concludes and highlights the challenges that should be addressed for a DMF-based method for future studies of single cells.

DMF Systems Microfluidic devices are considered promising in biological and chemical applications due to their compact size and low cost compared to conventional laboratory equipment. Among microfluidic techniques, DMF, featuring individual droplet manipulations, has drawn intensive attention. Unlike the traditional microfluidics, the DMF platform can operate a droplet individually on an array of electrodes without pump or microchannel (Choi et al. 2012). With the DMF platform, liquid samples can be operated in the form of droplets in picoliter to microliter size. This section briefly explains the DMF technology and the application of the DMF devices.

DMF Technology Droplet actuation and particle manipulation, the most important functions of DMF devices, have been performed based on various techniques. First, electrical techniques, such as electrowetting-on-dielectric (EWOD), electrodewetting and liquid dielectrophoresis (LDEP) for droplet actuation, and dielectrophoresis (DEP) for particle and cell manipulation, are presented. Optoelectronic techniques involving optoelectrowetting (OEW) and optoelectronic tweezers (OET) are then introduced, before a magnetic technique for particle and cell manipulation is discussed.

Electrowetting-on-Dielectric Berge originated electrowetting-on-dielectric (EWOD) based on the electrocapillarity (or electrowetting) effect (Berge 1993). In electrowetting, a voltage is applied between a conductive liquid droplet and an electrode, resulting in a reduced contact angle and improved wetting of the droplet to the substrate. Lippmann successfully used mercury and an electrolyte to demonstrate electrowetting across an electric double layer (Mugele and Baret 2005). Without a robust insulating layer between mercury and electrolyte, the aqueous solution suffers from electrolytic decomposition when applying voltage beyond a few hundred millivolts. Berge

188

L. Pang et al.

introduced the concept of EWOD to resolve this problem, in which a dielectric layer was placed to avoid direct contact of the electrode and liquid (Berge 1993). The relationship between the contact angle and the applied voltage is expressed by Lippmann-Young equation: cos θðV Þ ¼ cos θ0 þ

e0 eD 2 V , 2γ LG tD

ð1Þ

in which θ0 and θ(V ) are the contact angles before and after applying voltage V, respectively, e0 is the permittivity of vacuum, eD and tD are the relative permittivity and thickness of the dielectric layer, respectively, and γ LG is the liquid-gas interfacial tension. An EWOD device typically consists of two parallel plates, as depicted in Fig. 1a: one plate comprises a common electrode covered by a hydrophobic layer; the other plate contains an array of driving electrodes coated with a dielectric layer and a hydrophobic layer. A conductive droplet is sandwiched between the plates. When a voltage is applied, the surface hydrophobicity above the activated driving electrode reduces, enabling the droplet to move toward the activated electrode (Pollack et al. 2000; Cho et al. 2003). The EWOD force exerting on the droplet can be formulated via a method of energy minimization or an electromechanical approach (Mugele and Baret 2005; Kang 2002). For a driving electrode of width W, the horizontal component of the EWOD force is expressible as FEWOD ¼

e0 eD W 2 V : 2tD

ð2Þ

Electrodewetting Electrodewetting, the reverse phenomenon of electrowetting, utilizes electric fields to induce dewetting of a liquid droplet on a hydrophilic substrate (Li et al. 2019). Electrodewetting is achievable through attachment of an ionic surfactant to a substrate with an electric field. A droplet containing an ionic surfactant is initially placed on a hydrophilic conductive substrate. Upon applying a voltage between the droplet and the substrate, the ionic surfactant molecules are electrophoretically driven toward the substrate, rendering the surface hydrophobic and allowing the droplet dewet the substrate. Electrodewetting has been successfully applied in DMF with performance comparable with that of EWOD. Liquid Dielectrophoresis Dielectrophoresis (DEP) has been widely applied to drive an electrically polarizable object with the gradient of an electric field (Pohl 1978; Morgan and Green 2003; Jones 1995). When drawing a liquid of larger permittivity into a region of smaller permittivity using a stronger electric field, the action is referred to as liquid dielectrophoresis (LDEP) (Jones 2001; Fan et al. 2009a, b). In DMF systems, LDEP is preferable for actuating dielectric droplets toward a region of strong electric field. With voltage V applied between parallel electrode plates of width W and spaced d apart, the DEP force exerted on the dielectric droplet is expressed as

8

Digital Microfluidics for Single Cell Manipulation and Analysis

189

Fig. 1 Droplet actuation and particle manipulation on DMF systems. (a) Conductive and dielectric droplets driven with EWOD and LDEP, respectively, toward the activated electrodes. (b) Red and blue particles driven in a droplet with positive and negative DEP, respectively. (c) A droplet driven with optoelectrowetting when low-frequency AC voltage is applied; particles driven with OET when high-frequency AC voltage is applied

190

L. Pang et al.

FLDEP ¼

e0 ðeL  eG ÞW 2 V , 2d

ð3Þ

in which e0 is the permittivity of vacuum, eL and eG are the relative permittivity of the liquid and the surrounding gas, respectively. When the driving electrodes are coated with a dielectric layer, as shown in Fig. 1a, the DEP force is expressed as FLDEP ¼

  e0 eD W 2 eL eG V  , 2 eL tD þ eD d eG tD þ eD d

ð4Þ

where eD and tD are the relative permittivity and thickness of the dielectric layer, respectively.

Dielectrophoresis In addition to driving liquids, DEP is a technique more intensively established to manipulate electrically polarizable particles, including nucleic acids, proteins, cells, viruses, with a nonuniform electric field in a liquid (Fig. 1b) (Pohl 1978; Morgan and Green 2003; Jones 1995). The DEP force applying on a spherical particle of radius a is described with FDEP

 ep  em ¼ 2πa em Re  ∇E2 , ep þ 2em 3



ð5Þ

in which E is the electric field, e*p and e*m are the complex permittivities of the particle and the suspension medium, respectively. e*p and e*m that depend on frequency are expressed as ep,m ¼ e0 ep,m  j

σ p,m , 2πf

ð6Þ

in which ep,m and σ p,m are the relative permittivity and conductivity of the particle or the suspension medium; f is the frequency of the electric field. The frequency associated term in Eq. (5) can be represented with the Clausius-Mossotti factor f CM ¼

ep  em : ep þ 2em

ð7Þ

When the real part of fCM, that is, Re( fCM), is greater than zero, FDEP draws the particle toward the region of strong electric field, which is referred to as a positive DEP actuation. When Re( fCM) is smaller than zero, FDEP repels the particle from the region of strong electric field, which is referred to as a negative DEP actuation.

Optoelectrowetting Optoelectrowetting combines electrowetting with the light response of a photoconductor to actuate droplets with light (Chiou et al. 2003). A photoconductive layer is placed between the dielectric layer and the electrode, forming a series

8

Digital Microfluidics for Single Cell Manipulation and Analysis

191

connection of impedance of the droplet, insulator, and photoconductor. With an AC voltage applied, the contact angle modulation can be achieved by tuning the incident light. In the absence of light (dark state), the voltage mainly remains within the highly resistive photoconductor layer and thereby the contact angle does not change. Upon illumination, the photoconductor becomes more conductive. Consequently, the majority of the voltage exists in the dielectric layer and decreases the contact angle, as shown in Fig. 1c. When employed in DMF, optoelectrowetting enables virtual electrowetting electrodes to be addressed by light without bottlenecks of electric wiring.

Optoelectronic Tweezers Optoelectronic tweezers (OET) use light to create virtual electrodes and thus nonuniform electric field for the DEP manipulation of particles (Chiou et al. 2005). Generally, an OET device consists of two parallel plates: one plate comprises an electrode layer whereas the other comprises an electrode layer and a photoconductor layer. A liquid containing particles is sandwiched between the parallel plates; an AC voltage is applied between the plates. When light shines on the photoconductor, virtual electrodes defined by the light pattern are formed; an electric field gradient is formed to manipulate particles via DEP as shown in Fig. 1c. Positive OET that attracts particles to illuminated regions or negative OET that repels particles from illuminated regions are achievable depending on the DEP forces induced. An OET device requires a conductive plate to produce an electric field within the liquid, which limits its integration with other devices. To improve the flexibility of OET, lateral-field optoelectronic tweezers (LOET) was developed (Ohta et al. 2007). A LOET device contains interdigital photoconductive electrodes on a single plate, eliminating the necessity of a conductive plate. Upon applying a voltage across the interdigital electrodes, an electric field primarily parallel to the plate is generated. A gradient of the electric field is established between the illuminated and dark regions on illuminating the electrodes with an optical pattern. Magnetic Force A magnetic force is commonly used to separate or concentrate specific cells or molecules of interest (Yi et al. 2006). In general, functional magnetic particles are used to capture specific targets in liquids and then immobilized or concentrated with a permanent magnet or electromagnet. Subsequently, the surrounding liquid containing unbound species can be removed, exchanged, or delivered for reaction and analysis. Magnetic manipulation is simple to implement as it requires neither complicated structures nor electrical circuitry; it is also stable because it is unaffected by surface charges, pH, or ionic concentration.

Fabrication of DMF A DMF device typically consists of top and bottom plates, as shown in Fig. 1. The top plate comprises a common electrode that is deposited, for example by

192

L. Pang et al.

evaporation or sputtering, onto the substrate. A hydrophobic layer is spun or deposited onto the electrode to diminish fouling and to increase the contact angle of the driven droplets. The bottom plate contains driving electrodes that are fabricated on the substrate by deposition, photolithography, and etching. Subsequently, a dielectric layer is coated on the driving electrodes, which helps to build up charges for droplet actuation. Finally, a hydrophobic layer is deposited upon the dielectric layer. The top and bottom plates are separated with spacers. DMF devices contain four pivotal components: substrate, electrode, dielectric layer, and hydrophobic layer. Device substrates are often made of glass or silicon due to their chemical inertness. Printed circuit board (PCB) substrates are also increasingly popular owing to their low cost, high throughput (Fair 2007), and flexible electric wiring between pads and electrodes (Choi et al. 2012). Electrodes are made of metals (aluminum, chromium, copper, gold) or other conductive materials (doped polysilicon, ITO). Photolithography and wet/dry etching are implemented for electrode patterning. Dielectric layers can be formed by Parylene, silicon oxide, SU-8, or other insulating materials via vapor deposition, thermal growth, or spin-coating. Hydrophobic layers, typically Teflon AF or Cytop, can facilitate the actuation of aqueous droplets.

DMF Manipulation and Analysis of a Single Cell The critical applications of single-cell experimentation are for research in cellular heterogeneity research, being pervasive for cells and bacteria (Zhang et al. 2015). The current challenge of manipulation and study of a single cell lies not just in analyzing a single cell but also in conducting high-throughput analysis of numerous single cells (Zhai et al. 2019). Besides, the volume of single-cell is ~1 pL, hindering manipulation and analysis (Gao et al. 2019; He et al. 2015). DMF, with the use of electric power to drive droplets of the size down to picoliters, refers to a critical microfluidic technique for the manipulation and study of individual cells. Thus far, DMF manipulation and analysis of a single cell primarily addressed the isolation of various single cells (i.e., adherent cell and suspension cell) and longterm culture studies (Hosic et al. 2015; Mehling and Tay 2014). As fueled by the advance of DMF-based cell manipulation technique, single-cell heterogeneity would be fostered in various aspects, for example, single cancer cells sorted with size and deformability for cancer initiation and metastasis, single-cell DNA/RNA whole-transcriptome sequencing, and studies of single-cell exosomes. It is noteworthy that the change of DNA integrity and the expression of genes of interest of the cells after normal DMF operations (electrode 5 mm  5 mm or less, low frequency 10 or 1 kHz) were found negligible using the techniques of singlecell COMET assays, DNA microarrays, and qPCR (Au et al. 2013). In this section, the DMF-based tools for analysis of a single cell as an adherent or suspension cell are presented. The long-term culture of single cells on a DMF device is also discussed.

8

Digital Microfluidics for Single Cell Manipulation and Analysis

193

Adherent Cell With a few exceptions, most human cells are adherent cells. Study of a single adherent cell is capable of exploring the relation between cell heterogeneity and physiological function, as well as the effect of environmental stimuli (Zhang et al. 2014; Gracz et al. 2015). Various DMF techniques have been proposed for the study of adherent cells. Shah et al. originated a DMF platform combining OET for the collection of HeLa cells in a droplet (Fig. 2a) (Shah et al. 2009). Specifically, a droplet is manipulated with EWOD; individual cells were manipulated with OET. This platform would enable studies of an isolated cell involving variations of multiple environmental stimuli. Lin et al. reported a two-level metal structure that could dispense picoliter droplets (Lin et al. 2010). With multiple layers of metal, the platform could be fabricated of denser and smaller electrodes to control smaller droplets with a flexible interconnect-routing scheme. The tiny droplet volume is beneficial to future single-cell studies. Witters et al. developed a DMF platform for assays of adherent cells based on biofunctionalizing that constructs cell clusters or single cells array (hPDC stem cells) as shown in Fig. 2b (Witters et al. 2011). This platform used a dry lift-off approach with an easily peeled Parylene C mask to allow regulated micropatches of biomolecules (poly-L-lysine, PLL) in a spatial manner along the Teflon AF surface of the chip. With this platform, arrays of single cells could be produced on the transport of droplets of a cell suspension by EWOD to the driving electrodes that bear local cell-adhesion micropatches. Rival et al. developed a DMF platform to achieve a single-cell assay for transcriptome study (Rival et al. 2014); the DMF device integrated the processes of isolating a single cell (human HaCaT adherent cells) in a droplet, harvesting mRNA from the cell, catalyzing the reaction in real time, and performing RT-qPCR, as shown in Fig. 2c. Ng et al. presented a combined DMF platform called DISC (digital microfluidic immunocytochemistry in single cells) as shown in Fig. 2d (Ng et al. 2015); the platform combined with a single-cell assay with multiple steps (e.g., single-cell culture, stimulation, and immunocytochemistry) was applied for research in cellular heterogeneity. The platform contained two parallel plates to automate fluid handling; the bottom plate had an array of electrodes adopted to manipulate droplets to achieve adherent-cells seeding, culture, delivering stimuli, and immunocytochemistry. The phosphorylation states of platelet-derived growth factor receptor and the subsequent signaling protein, Akt, were investigated to assess the effect of the dependence on concentration and time of PDGF stimulation of NIH-3 T3 fibroblasts and breast cancer cells (MDA-MB-231 and MCF-7).

Suspension Cell Besides adherent cells, suspension cells, for example, bacteria, algae, and yeast, have been massively researched. For the free-floating nature of the latter, spatiotemporal studies conducted on suspended cells are less sophisticated than those performed on adherent cells (Chung et al. 2011). Nevertheless, floating in the medium, suspension

Fig. 2 DMF for manipulations of adherent cells. (a) OET cell collection and depletion in EWOD-driven droplets. (Reproduced with permission from (Shah et al. 2009), copyright 2009 Royal Society of Chemistry). (b) Driving electrode design with micropatches of biomolecules for single-cell array. (i) A droplet on the array of PLL micropatches. (ii) A fluorescent image of patterned PLL-FITC features along a Teflon AF coated substrate. (iii) Single hPDC stem cells captured on circular micropatches (diameter 20 μm). (Reproduced with permission from (Witters et al. 2011), copyright 2011 Royal Society of Chemistry). (c) Protocol for cell lysis, mRNA extraction, and qRT-PCR analysis. (Reproduced with permission from (Rival et al. 2014), copyright 2014 Royal Society of Chemistry). (d) DISC of adherent cells cultured on the patterned top plate. (Reproduced with permission from (Ng et al. 2015), copyright 2015 Springer Nature)

194 L. Pang et al.

8

Digital Microfluidics for Single Cell Manipulation and Analysis

195

cells require being captured with a physically trap to achieve downstream analysis. Son et al. presented a two-plate DMF device for analysis of a single zebrafish embryo as shown in Fig. 3a (Son and Garrell 2009). With electrodes of size 5  10 mm, a single zebrafish embryo (diameter 0.5 mm) was able to undergo transport and normal hatching on this device. They also programmed the transport of live yeast (diameter 5 μm) without diminished viability after transport; the actuation left no yeast behind. Shih et al. achieved multistep single-cell studies with a hybrid droplet-to-digital (D2D) device combining DMF with droplet-in-channel microfluidics (Fig. 3b) (Shih et al. 2014). Such a D2D platform held the merits from both microfluidic systems – droplets-in-channel facilitated individual cell encapsulation in droplets, whereas DMF operated droplets on demand with droplet volumes over a wide range (pL to μL). The device was used to determine the effect of ionic liquid (IL), salt type, and concentration on the morphological structure, growth, and ethanol production of a single yeast cell for the biofuel applications. Ahmadi et al. reported an advanced integrated droplet-digital microfluidic (ID2M) platform integrated droplet operation with cell encapsulation, which can be performed for study of yeast mutants and wild-type yeast cells as shown in Fig. 3c (Ahmadi et al. 2019). The creation of droplets and volumes adjustment on demand increased the operational flexibility; trapping and incubating droplets for 24 h were demonstrated. The integrated EWOD driving-electrode array offered multiple n-ary channels and significantly improved the sorting and analysis functions comparing with typical binary channels.

Long-Term Culture Cell culture and analysis are vital to biomedical research and drug screening (He et al. 2015; Zhai et al. 2019). Because of the high-throughput genetic sequencing and drug screening, liquid-handling robotics have been applied for automated longterm culture and analysis of cells (Shoemaker 2006). Single-cell culture has been extensively employed for cellular heterogeneity research (e.g., single stem cell differentiation and cancer stem cell research) (Lee et al. 2014; Pang et al. 2019b). For their unique strengths, for example, excellent design of the specific structure, high-throughput capability, and integration of multiple functional components, microfluidic techniques have become an excellent device for culture and study of a single cell. In contrast, DMF that handles cell samples and reagents in discrete droplets on a hydrophobic surface (Pollack et al. 2000; Cho et al. 2003) exhibits the feature of positioning the droplet for cell culture. In this section, the development of techniques for DMF study of a single cell is highlighted for the adherent and suspension single-cell culture. Adherent cells require an applicable surface to attach and to proliferate; the hydrophobic coatings of DMF devices must be modified for their attachment and growth (Eydelnant et al. 2012). Barbulovic-Nad et al. developed a DMF device with additional adhesion pads (0.8 mm), consisting of extracellular matrix (ECM) proteins such as fibronectin or collagen, to achieve adherent-cell culture, applicable to long-term cell subculture and analysis as shown in Fig. 4a (Barbulovic-Nad et al.

196

L. Pang et al.

Fig. 3 DMF for manipulation of a single suspension cell and embryo. (a) (i) Zebra fish embryo driven by DMF. (ii) The embryo developed normally 48 h after transport. (iii) The hatched fish. (Reproduced with permission from (Son and Garrell 2009), copyright 2009 Royal Society of Chemistry). (b) A D2D device with a capillary connecting the droplet-in-channel to DMF device for analysis of a single yeast cell. (Reproduced with permission from (Shih et al. 2014), copyright 2014 Royal Society of Chemistry). (c) A photo of an ID2M system for study of a single yeast cell. (Reproduced with permission from (Ahmadi et al. 2019), copyright 2019 Royal Society of Chemistry)

2010). Seeding, growth, detachment, and re-seeding of mammalian cells (HeLa, CHO-K1, NIH-3 T3, and INS-1) on a fresh surface were all demonstrated with the DMF device. Transient transfection of CHO-K1 cells on demand in droplets was also achieved. As described in section “Adherent Cell,” a similar approach was demonstrated for single cell culture with an array of miniaturized adhesion sites

8

Digital Microfluidics for Single Cell Manipulation and Analysis

197

(20 μm) (Lin et al. 2010). The selection and isolation of cells before re-seeding of adherent cells is useful for long-term cell subculture; OET or DEP becomes a candidate for cell selections after detachment of adherent cells. Valley et al. built a unified optoelectronic platform capable of actuating droplets and manipulating particles/cells in the droplets with light (i.e., OET and optoelectrowetting) simultaneously as shown in Fig. 4b (Valley et al. 2011). With such a platform, an individual cell (HeLa) could be visually selected and optically picked from a cell group using OET and isolated into its own subdroplet driven by optoelectrowetting. It is noteworthy that OET or DEP typically drives cells in an isotonic buffer that has a substantially lower conductivity than that of culture media; special care needs to be taken for addressing cells in culture media (Hsu et al. 2010). Kumar et al. established a DMF platform to achieve analysis of a nonadherent single plant cell (Kumar et al. 2014). With this automated platform, single protoplasts from an Arabidopsis thaliana plant were manipulated as shown in Fig. 4c. Protoplasts were collected and immobilized on the DMF chip with Concanavalin A (ConA) and magnetic microparticles (MMP) conjunction. Water permeability coefficients were measured under varied osmotic pressure. A time-lapse fluorescence microscopy was used to monitor cell responses over time in culture. Magnetic forces or microwells facilitate the positioning, monitoring, and long-term culture of nonadherent single cells. Subsequently, Kumar et al. developed another device to achieve high-throughput manipulation, culture, and analysis of nonadherent single yeast cells using DMF (Kumar et al. 2015). A microwell array was fabricated on the DMF device to trap single yeast cells with drag forces and surface tension, which eliminates other physical forces that might damage the cells as shown in Fig. 4d.

Related Applications of DMF Apart from the manipulation and analysis of a single cell, the DMF platform has been adopted for other cell applications, in particular cell sorting and concentrating and 3D organoid culture. Furthermore, the high reconfigurability makes this DMF platform ideal to achieve complicated and multistep procedures on a simple and compact device.

Cell Sorting and Concentrating Cell sorting and concentrating has become an enabling tool for biomedical applications. For instance, the circulating tumor cells (CTC) in the blood of a cancer patient are critical for cancer diagnosis, treatment, and prognosis; the development of enrichment and detection of CTC is pivotal (Tian et al. 2019). Cell sorting is also crucial for the study of cell heterogeneity. For example, smaller and/or more deformable tumor cells were found to show greater stemness (Pang et al. 2015). As one major application, microfluidic cell sorting was established based on hydrodynamics, filtration, surface acoustic wave, magnetic force as well as 2D

198

L. Pang et al.

Fig. 4 DMF for single-cell culture. (a) Photograph and schematics of a DMF platform developed for cell culture and subculture. (Reproduced with permission from (Barbulovic-Nad et al. 2010), copyright 2010 Royal Society of Chemistry). (b) Method for single-cell selection and encapsulation with OET and optoelectrowetting. (Reproduced with permission from (Valley et al. 2011), copyright 2011 Royal Society of Chemistry). (c) Schematics describing on-chip treatment of protoplasts under varied osmotic pressure. (Reproduced with permission from (Kumar et al. 2014), copyright 2014 Spring Nature). (d) Schematic diagram of cell loading with DMF through cell sedimentation. (Reproduced with permission from (Kumar et al. 2015), copyright 2015 Royal Society of Chemistry)

electrophoresis. DMF-based sorting techniques exhibit special strengths for cell sorting and concentrating. Unlike the microchannel-based microfluidic chips, DMF sorting devices demonstrate no clogging and easily include downstream

8

Digital Microfluidics for Single Cell Manipulation and Analysis

199

analyses. Cho et al. first described an idea of concentration and binary separation of charged particles using electrophoresis and its experimental confirmations for DMF with droplets driven by EWOD (Cho et al. 2007). Zhao et al. concentrated and separated aldehyde-sulfate latex particles (5 μm), glass beads (8 μm), and ground pine spores with a traveling-wave DEP in a droplet and then split the droplet with EWOD (Zhao et al. 2007). Fan et al. presented a platform integrating EWOD and DEP. By selectively applying electric signals at appropriate frequencies, cross-scale actuations of neuroblastoma cells (Neuro-2A, 5 μm) and droplets (1 mm) were obtained as shown in Fig. 5a (Fan et al. 2008). Shah et al. showed the separation of CD8+ and CD8– lymphocytes with enhanced collection efficiency with antibodyconjugated magnetic beads and magnets on an EWOD-based DMF platform (Shah et al. 2010).

3D Cell Culture Three-dimensional (3D) cell culture offers a more physiologically relevant microenvironment for cell study in vitro, having a great potential in tissue engineering, drug screening, and precision medicine to eliminate animal testing. Together with microfluidic techniques, some organ-like functions were recapitulated on a chip to demonstrate the concept of organ-on-a-chip. Hydrogels, behaving as an extracellular matrix (ECM), have been employed with DMF for 3D cell culture. In addition, the hanging-drop method has been demonstrated on DMF devices to construct cell spheroids. Au et al. prepared arrays of individually addressable, free-floating and 3D hydrogel-based hepatic microtissues, or hepatic organoids, with NIH-3T3, HepG2, and collagen I on a DMF device for drug discovery and screening (Au et al. 2014). The fibroblast-dependent contractile behavior, albumin secretion profiles, and cytochrome P450 3A4 activities were observed from these organoids; an acetaminophen hepatotoxicity assay was also demonstrated. Chiang et al. constructed 3D heterogeneous hydrogels with a versatile DMF device providing EWOD, LDEP, and DEP to position prepolymer droplets (conductive or dielectric), cross-linked microgels and particle/cells in prepolymer droplets before cross-linking, as shown in Fig. 5b (Chiang et al. 2016). On adopting the two reciprocal electric manipulations – EWOD and DEP, DMF demonstrated a wide variety of manipulations of objects with various (i) phases (prepolymer droplets and polymerized microgels), (ii) scales (particles/cells (μm) and hydrogel architectures (mm)), and (iii) properties (electric conductivity and cross-linking methods (photo, chemical, or thermal polymerization)). Hydrogels, including gelatin methacryloyl (GelMA), poly(ethylene glycol) diacrylate (PEGDA), Matrigel, and polyacrylamide, were manipulated and assembled on the DMF device. NIH/3T3 fibroblasts and primary neonatal mouse cardiomyocytes were reorganized or patterned within or above the hydrogel complex. In cancer research, the culture of 3D tumor spheroids is a preferred model in vitro to mimic tumor physiological conditions in vivo. Bender et al. built a DMF device that enabled spheroids to form using the hanging-drop approach, to

200

L. Pang et al.

Fig. 5 DMF for related applications in cell manipulation. (a) A parallel-plate DMF platform combining square and strip electrodes for activating droplets and cell inside droplets by EWOD and DEP, respectively. (Reproduced with permission from (Fan et al. 2008), copyright 2008 Royal Society of Chemistry) (b) 3D cell culture with a programmable DMF device for forming microgel and assembling architecture. (Reproduced with permission from (Chiang et al. 2016), copyright 2016 American Association for the Advancement of Science) (c) The DMF platform for cell spheroids formation and cultivation. (Reproduced with permission from (Bender et al. 2016), copyright 2016 Royal Society of Chemistry) (d) A DMF immunoassay for measles and rubella testing. (Reproduced with permission from (Ng et al. 2018), copyright 2018 American Association for the Advancement of Science)

encapsulate in a collagen gel, and to expose to migration-modulating agents, as shown in Fig. 5c (Bender et al. 2016). Cell spheroids of HT-29 human colorectal adenocarcinoma cells or BJ human fibroblasts were formed on this platform. A migration and invasion assay of spheroids with modulating agents, bone morphogenic protein 2 (BMP-2), and prostaglandin E2 (PGE2) was reported. Automated liquid-handling protocols for both spheroid and single-cell invasion studies were demonstrated.

8

Digital Microfluidics for Single Cell Manipulation and Analysis

201

Diagnosis and Clinical Application DMF can be integrated with other microfluidic techniques. For example, with the hybrid D2D and ID2M devices described above, DMF devices can be combined with microchannel-based microfluidic functions and offer automated protocols for bioassays. The programmable actuations of precise and tiny droplets simplify the steps of sample preparation and enable the integration of other analysis instrument and detectors. For instance, DMF has been demonstrated to facilitate complex and multistep sample processing before analysis with a mass spectrometer for proteomics studies (Wheeler et al. 2004). In addition to the integration ability of hybrid microfluidic systems, DMF itself has shown its diversity of performing bioassays, diagnoses, and treating clinical samples. DMF devices have been used for enzymatic reactions to assess substances of interest or to study reaction kinetics (Fair 2007). With antibodies immobilized on a solid support (i.e., heterogeneous immunoassays), a DMF platform has been adopted for immunoassays research. Ng et al. built a DMF platform for serological immunoassays in remote settings (Fig. 5d) (Ng et al. 2018); this DMF immunoassay was used to test measles and rubella immunoglobulin G (IgG), demonstrating DMF-based point-of-care serological diagnosis with minimized visits to centralized laboratories. Other than immunoassays, molecular diagnosis or DNA analysis has also been extensively investigated with DMF and applied in many fields, such as antidrug analysis, diagnosis, and forensics (Khilko et al. 2018). Because the related procedures are conducted on delicate, valuable samples or in a highly multiplexed format, DMF is a perfect platform for molecular studies because it requires only a small volume of sample and has convenient integration with complicated and multistep experiments. In particular, DMF has been adopted to purify and to extract DNA samples, thereby producing recombinant DNA through pyrosequencing, polymerase chain reaction (PCR), DNA hybridization, and ligation (Chang et al. 2006). As a convenient and compatible testing platform, DMF has a great prospect for application in clinical diagnosis that covers sample preparation and detecting. A DMF PCR platform was used to investigate Mycoplasma pneumoniae DNA in respiratory tract samples of patients with community-acquired pneumonia (WulffBurchfield et al. 2010). Using this platform, real-time PCR was implemented with agreement more than 95% with the traditional method of detecting DNA. Moreover, the DMF method for sample analysis required a sample size 1/1000 that of the traditional approaches. In another example, estradiol extraction from 1-μL samples of human breast tissue homogenate was performed with a DMF device that could be used also for other samples such as whole blood and serum (Mousa et al. 2009). All the DMF diagnosis functions could be further applied to single-cell studies.

Conclusions and Future Outlook Microfluidic techniques have become a useful tool for studies of a single cell. Microfluidic devices manipulate single cells by means of passive and active methods. Microchannel-based microfluidic devices typically isolate single cells

202

L. Pang et al.

with passive methods (e.g., traps, filtration, microwells) driven by hydrodynamic forces. This chapter introduced several alternative active methods to manipulate a single cell, for example, electric, optical, and magnetic means, on DMF devices. In addition to the manipulations of cells, the demonstrated DMF functions of droplet dispensing, transporting, mixing, and splitting facilitate related single-cell protocols. Although DMF exhibits great abilities to manipulate individual cells and droplets, the main challenge for the future single-cell applications lies in the throughout, sometimes limited by the available control, electronic, and fabrication abilities. Reliable and mass production of DMF devices with a large number of electrodes each of the size comparable with a single cell is desirable for single cell studies. Yet, for the simple device configuration and easy integration with other sensing and actuating techniques, the functionality and possible protocols for single-cell studies will keep growing on the future DMF platform.

References Ahmadi F, Samlali K, Vo PQN, Shih SCC (2019) An integrated droplet-digital microfluidic system for on-demand droplet creation, mixing, incubation, and sorting. Lab Chip 19:524–535 Au SH, Fobel R, Desai SP, Simmons BA, Adams PD, Singh S, Singh AK (2013) Cellular bias on the microscale: probing the effects of digital microfluidic actuation on mammalian cell health, fitness and phenotype. Integr Biol 5:1014–1025 Au SH, Chamberlain MD, Mahesh S, Sefton MV, Wheeler AR (2014) Hepatic organoids for microfluidic drug screening. Lab Chip 14:3290–3299 Barbulovic-Nad I, Au SH, Wheeler AR (2010) A microfluidic platform for complete mammalian cell culture. Lab Chip 10:1536–1542 Bender BF, Aijian AP, Garrell RL (2016) Digital microfluidics for spheroid-based invasion assays. Lab Chip 16:1505–1513 Berge B (1993) Electrocapillarity and wetting of insulator films by water. C R Acad Sci II 317:157–163 Chang YH, Bin Lee G, Huang FC, Chen YY, Lin JL (2006) Integrated polymerase chain reaction chips utilizing digital microfluidics. Biomed Microdevices 8:215–225 Cheng YH, Chen YC, Brien R, Yoon E (2016) Scaling and automation of a high-throughput singlecell-derived tumor sphere assay chip. Lab Chip 16:3708–3717 Chiang MY, Hsu YW, Hsieh HY, Chen SY, Fan SK (2016) Constructing 3D heterogeneous hydrogels from electrically manipulated prepolymer droplets and crosslinked microgels. Sci Adv 2:e1600964 Chiou PY, Moon H, Toshiyoshi H, Kim CJ, Wu MC (2003) Light actuation of liquid by optoelectrowetting. Sensors Actuators A Phys 104:222–228 Chiou PY, Ohta AT, Wu MC (2005) Massively parallel manipulation of single cells and microparticles using optical images. Nature 436:370–372 Cho SK, Moon H, Kim CJ (2003) Creating, transporting, cutting, and merging liquid droplets by electrowetting-based actuation for digital microfluidic circuits. J Microelectromech Syst 12:70–80 Cho SK, Zhao Y, Kim CJ (2007) Concentration and binary separation of microparticles for dropletbased digital microfluidics. Lab Chip 7:490–498 Choi K, Ng AH, Fobel R, Wheeler AR (2012) Digital microfluidics. Annu Rev Anal Chem 5:413–440 Chung K, Kim Y, Kanodia JS, Gong E, Shvartsman SY, Lu H (2011) A microfluidic array for largescale ordering and orientation of embryos. Nat Methods 8:171–177

8

Digital Microfluidics for Single Cell Manipulation and Analysis

203

Eydelnant IA, Uddayasankar U, Liao MW, Liao MW, Wheeler AR (2012) Virtual microwells for digital microfluidic reagent dispensing and cell culture. Lab Chip 12:750–757 Fair RB (2007) Digital microfluidics: is a true lab-on-a-chip possible? Microfluid Nanofluid 3:245–281 Fan SK, Huang PW, Wang TT, Peng YH (2008) Cross-scale electric manipulations of cells and droplets by frequency-modulated dielectrophoresis and electrowetting. Lab Chip 8:1325–1331 Fan SK, Hsieh TH, Lin DY (2009a) General digital microfluidic platform manipulating dielectric and conductive droplets by dielectrophoresis and electrowetting. Lab Chip 9:1236–1242 Fan SK, Chen WJ, Lin TH, Wang TT, Lin YC (2009b) Reconfigurable liquid pumping in electricfield-defined virtual microchannels by dielectrophoresis. Lab Chip 9:1590–1595 Gao D, Jin F, Zhou M, Jiang Y (2019) Recent advances in single cell manipulation and biochemical analysis on microfluidics. Analyst 144:766–781 Gracz AD, Williamson IA, Roche KC, Johnston MJ, Wang F, WangY APJ, Balowski J, Liu XF, Laurenza RJ, Gaynor LT, Sims CE, Galanko JA, Li L, Allbritton NL, Magness ST (2015) A high-throughput platform for stem cell niche co-cultures and downstream gene expression analysis. Nat Cell Biol 17:340–349 He JL, Chen AT, Lee JH, Fan SK (2015) Digital microfluidics for manipulation and analysis of a single cell. Int J Mol Med Sci 16:22319–22332 Hosic S, Murthy SK, Koppes AN, Koppes AN (2015) Microfluidic sample preparation for single cell analysis. Anal Chem 88:354–380 Hsu HY, Ohta AT, Chiou PY, Jamshidi A, Neale SL, Wu MC (2010) Phototransistor-based optoelectronic tweezers for dynamic cell manipulation in cell culture media. Lab Chip 10:165–172 Jones TB (1995) Electromechanics of particles. Cambridge University Press, Cambridge, UK Jones TB (2001) Liquid dielectrophoresis on the microscale. J Electrost 51:290–299 Kang KH (2002) How electrostatic fields change contact angle in electrowetting. Langmuir 18:10318–10322 Khilko Y, Weyman PD, Glass JI, Adams MD, McNeil MA, Griffin PB (2018) DNA assembly with error correction on a droplet digital microfluidics platform. BMC Biotechnol 18:37 Kumar PT, Toffalini F, Witters D, Vermeir S, Rolland F, Hertog MLATM, Nicolaï BM, Puers R, Geeraerd A, Lammertyn J (2014) Digital microfluidic chip technology for water permeability measurements on single isolated plant protoplasts. Sensors Actuators B Chem 199:479–487 Kumar PT, Vriens K, Cornaglia M, Vermeir S, Rolland F, Hertog MLATM, Bart Nicolaï M, Puers R, Geeraerd A, Lammertyn J (2015) Digital microfluidics for time-resolved cytotoxicity studies on single non-adherent yeast cells. Lab Chip 15:1852–1860 Lawson DA, Kessenbrock K, Davis RT, Pervolarakis N, Werb Z (2018) Tumour heterogeneity and metastasis at single-cell resolution. Nat Cell Biol 20:1349–1360 Lee WC, Shi H, Poon Z, Nyan LM, Kaushik T, Shivashankar GV, Chan JKY, Lim CT, Han J, Vliet KJV (2014) Multivariate biophysical markers predictive of mesenchymal stromal cell multipotency. Proc Natl Acad Sci U S A 111:E4409–E4418 Li J, Ha NS, Liu T, Michael van Dam R, Kim CJ (2019) Ionic-surfactant-mediated electrodewetting for digital microfluidics. Nature 572:507–510 Lin YY, Evans RD, Welch E, Hsu BN, Madison AC, Fair RB (2010) Low voltage electrowettingon-dielectric platform using multi-layer insulators. Sensors Actuators B Chem 150:465–470 Mehling M, Tay S (2014) Microfluidic cell culture. Curr Opin Biotechnol 25:95–102 Morgan H, Green NG (2003) AC electrokinetics: colloids and nanoparticles. Research Studies Press Ltd., Baldock Mousa NA, Jebrail MJ, Yang H, Abdelgawad M, Metalnikov P, Chen J, Abdelgawad M, Metalnikov P, Chen J, Wheeler AR, Casper RF, Casper RF (2009) Droplet-scale estrogen assays in breast tissue, blood, and serum. Sci Transl Med 1:1ra2 Mugele F, Baret JC (2005) Electrowetting: from basics to applications. J Phys Condens Matter 17: R705–R774 Ng AH, Chamberlain MD, Situ H, Lee V, Wheeler AR (2015) Digital microfluidic immunecytochemistry in single cells. Nat Commun 6:7513

204

L. Pang et al.

Ng AH, Fobel R, Fobel C, Lamanna J, Rackus DG, Summers A, Dixon C, Dryden MDM, Lam C, Ho M, Mufti NS, Lee V, Asri MAM, Sykes EA, Chamberlain MD, Joseph R, Ope M, Scobie HM, Knipes A, Rota PA, Marano N, Chege PM, Njuguna M, Nzunza R, Kisangau N, Kiogora J, Karuingi M, Burton JW, Borus P, Lam E, Wheeler AR (2018) A digital microfluidic system for serological immunoassays in remote settings. Sci Transl Med 10:eaar6076 Ohta AT, Chiou PY, Phan HL, Sherwood SW, Yang JM, Lau ANK, Hsu HY, Jamshidi A, Wu MC (2007) Optically controlled cell discrimination and trapping using optoelectronic tweezers. IEEE J Sel Top Quantum Electron 13:235–243 Pang L, Shen S, Ma C, Ma T, Zhang R, Tian C, Zhao L, Liu WM, Wang JY (2015) Deformability and size-based cancer cell separation using an integrated microfluidic device. Analyst 140:7335–7346 Pang L, Liu W, Tian C, Xu J, Li T, Chen SW, Wang JY (2016) Construction of single-cell arrays and assay of cell drug resistance in an integrated microfluidic platform. Lab Chip 16:4612–4620 Pang L, Ding J, Liu XX, Fan SK (2019a) Digital microfluidics for cell manipulation. Trends Anal Chem 117:291–299 Pang L, Ding J, Ge Y, Fan J, Fan SK (2019b) Single-cell-derived tumor-sphere formation and drugresistance assay using an integrated microfluidics. Anal Chem 91:8318–8325 Pohl HA (1978) Dielectrophoresis. Cambridge University Press, Cambridge, UK Pollack MG, Fair RB, Shenderov AD (2000) Electrowetting-based actuation of liquid droplets for microfluidic applications. Appl Phys Lett 77:1725–1726 Rival A, Jary D, Delattre C, Fouillet Y, Castellan G, Bellemin-Comte A, Gidrol X (2014) An EWOD-based microfluidic chip for single-cell isolation, mRNA purification and subsequent multiplex qPCR. Lab Chip 14:3739–3749 Shah GJ, Ohta AT, Chiou EPY, Wuc MC, Kim CJ (2009) EWOD-driven droplet microfluidic device integrated with optoelectronic tweezers as an automated platform for cellular isolation and analysis. Lab Chip 9:1732–1739 Shah GJ, Veale JL, Korin Y, Reed EF, Gritsch HA, Kim JC (2010) Specific binding and magnetic concentration of CD 8+ T-lymphocytes on electrowetting- on-dielectric platform. Biomicrofluidics 4:044106 Shih SCC, Gach PC, Sustarich J, Simmons BA, Adams PD, Singha S, Singh AK (2014) A dropletto-digital (D2D) microfluidic device for single cell assays. Lab Chip 15:225–236 Shoemaker RH (2006) The NCI60 human tumour cell line anticancer drug screen. Nat Rev Cancer 6:813–823 Sims CE, Allbritton NL (2007) Analysis of single mammalian cells on-chip. Lab Chip 7:423–440 Son SU, Garrell RL (2009) Transport of live yeast and zebrafish embryo on a droplet (“digital”) microfluidic platform. Lab Chip 9:2398–2401 Tian F, Liu C, Lin L, Chen Q, Sun J (2019) Microfluidic analysis of circulating tumor cells and tumor-derived extracellular vesicles. Trends Anal Chem 117:128–145 Valley JK, Shao NP, Jamshidi A, Hsu HY, Wu MC (2011) A unified platform for optoelectrowetting and optoelectronic tweezers. Lab Chip 11:1292–1297 Wheeler AR, Moon H, Kim CJ, Loo JA, Garrell RL (2004) Electrowetting-based microfluidics for analysis of peptides and proteins by matrix-assisted laser desorption/ionization mass spectrometry. Anal Chem 76:4833–4838 Witters D, Vergauwe N, Vermeir S, Ceyssens F, Liekens S, Puers R, Lammertyn J (2011) Biofunctionalization of electrowetting-on-dielectric digital microfluidic chips for miniaturized cell-based applications. Lab Chip 11:2790–2794 Wulff-Burchfield E, Schell WA, Eckhardt AE, Pollack MG, Hua Z, Rouse JL, Pollack MG, Hua Z, Rouse JL, Pamula VK, Srinivasan V, Benton JL, Alexander BD, Wilfret DA, Kraft M, Cairns C, Perfect JR, Wilfret DA (2010) Microfluidic platform versus conventional real-time polymerase chain reaction for the detection of Mycoplasma pneumoniae in respiratory specimens. Diagn Microbiol Infect Dis 67:22–29 Yalcin D, Hakguder ZM, Otu HH (2016) Bioinformatics approaches to single-cell analysis in developmental biology. Mol Hum Reprod 22:182–192

8

Digital Microfluidics for Single Cell Manipulation and Analysis

205

Yi C, Li CW, Ji S, Yang M (2006) Microfluidics technology for manipulation and analysis of biological cells. Anal Chim Acta 560:1–23 Zhai J, Yi S, Jia Y, Mak PI, Martins RP (2019) Cell-based drug screening on microfluidics. Trends Anal Chem 117:231–241 Zhang K, Chou CK, Xia X, Hung MC, Qin L (2014) Block-cell-printing for live single-cell printing. Proc Natl Acad Sci U S A 111:2948–2953 Zhang H, Cui X, Bi J, Dai S, Ye H (2015) Single-cell analysis for bioprocessing. Eng Life Sci 15:582–592 Zhao Y, Yi UC, Cho SK (2007) Microparticle concentration and separation by traveling-wave dielectrophoresis (twDEP) for digital microfluidics. J Microelectromech Syst 16:1472–1481

9

Single-Cell Separation Shilpi Pandey, Ninad Mehendale, and Debjani Paul

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conventional Cell Separation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Centrifugation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluorescence-Activated Cell Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Magnetic-Activated Cell Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laser Capture Microdissection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Manual Cell Picking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidic Single-Cell Separation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidic Passive Separation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deterministic Lateral Displacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrodynamic Separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidic Active Separation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison Between Different Microfluidic Separation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

208 209 210 210 212 213 215 215 216 219 220 224 228 228 231

Abstract

Advances in both basic cell biology and point-of-care diagnostic technology led to the need for reliable, accurate, and fast techniques to separate single cells from a heterogeneous mixture. Conventional cell separation techniques, such as fluorescence-activated cell sorting (FACS) and magnetic-activated cell sorting (MACS), are very efficient methods. However, these techniques are generally expensive, require bulky equipment, and are not suitable for use in field settings. On the other hand, microfluidic technology matches the performance of existing techniques, while being more portable and suitable for field use. In this chapter, we discuss the working principles and illustrate some applications of various S. Pandey · N. Mehendale · D. Paul (*) Department of Biosciences and Bioengineering, Indian Institute of Technology Bombay, Mumbai, Maharashtra, India e-mail: [email protected]; [email protected]; [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_6

207

208

S. Pandey et al.

established and emerging microfluidic techniques. The emphasis of this chapter is more on the recent advancement in the field of microfluidics for rare or single-cell separation.

Introduction Isolating rare cells from complex samples is an essential requirement for many biological and clinical applications. For instance, fetal cells present in maternal blood can be used for noninvasive prenatal testing. The frequency of fetal cells compared to the maternal mononuclear cells can vary a lot but is estimated to be approximately 10–5 to 10–7 (Wachtel et al. 2001). Isolating these cells from the mother’s blood is the first step for fetal genetic testing. Cancer is another example of a disease where a capturing rare cell is an important requirement for diagnosis. Most deaths in cancer occur due to metastasis, a mechanism by which cells from the primary tumor travel to the other organs through the circulatory system. There are typically one to ten circulating tumor cells (CTC) in 1 ml of whole blood, and these need to be detected to understand the spread of the disease. Personalized medicine is another rapidly growing area, where isolation of hematopoietic stem cells from patients is required to provide autologous treatments. Single-cell separation and analysis techniques are important due to the inherent heterogeneity of cell populations. Single-cell analysis can range from the measurement of the nucleic acid levels to the study of the proteomic and metabolomic characteristics of the cells. Conventional techniques can only measure the average properties of a cell population, which leads to a loss of valuable information from the rare cells present in that population. Moreover, owing to the time dependence of different transcriptional and signaling processes, ensemble averages fail to capture the cellular dynamics. In both basic and clinical research, single-cell separation is mainly performed using conventional techniques, such as fluorescence-activated cell sorting (FACS), magnetic-activated cell sorting (MACS), laser capture microdissection (LCM), manual cell picking, etc. (Miltenyi et al. 1990; Datta et al. 2015; Hu et al. 2016). Currently, FACS is the benchmark technology for cell sorting. It is robust and automated and has a good throughput (~10,000–50,000 cells/s) for most research applications. However, the throughput of FACS is not high enough compared to a bulk sorting technique like centrifugation. Hence, the use of FACS for routine clinical applications, where millions of cells need to be processed in a relatively short period of time, is rather limited. Most conventional sorting techniques require highly skilled technicians, large volumes of samples, and well-equipped facilities. The equipment required for FACS has a large footprint and is expensive (>$100,000) (Lee et al. 2017). As a result, there has been a strong need for a next-generation technology for single-cell separation with high reliability, low cost, portability, higher speed of sorting for rare cells, and reduced risk of biohazards due to exposure to aerosols during sample preparation.

9

Single-Cell Separation

209

Microfluidics is an emerging technology which works with low sample volumes (~μL – pL) and can address most of the problems faced by the conventional techniques. Microfluidic cell separation technologies, such as hydrodynamic separation, deterministic lateral displacement, field flow fractionation, etc., have been extensively demonstrated for sorting, counting, and analyzing the contents of single cells of interest. The performance of various cell separation techniques can be described by separation purity, throughput, sample recovery, and the enrichment factor. Purity is defined as the ratio of the number of target cells to the total number of sorted cells (counted at the outlet). Throughput is measured by the total output sample volume collected per unit time. The ratio of the total number of cells collected at the outlet to the total number of cells at the inlet is called recovery. It is a measure of the loss of cells inside the device. The enrichment factor is given by the Eq. 1. Number of target cells at the outlet Enrichment factor ¼ Total number of cells at the outlet Number of target cells at the inlet Total number of cells at the inlet

(1)

Each cell separation technique has distinct advantages and disadvantages. Often, a user needs to balance different performance parameters, leading to trade-offs. This chapter begins with an overview of some of the conventional techniques for singlecell separation. The chapter then focuses on the microfluidic single-cell separation techniques. These techniques can broadly be divided into two classes: (a) passive techniques that do not require any external fields and (b) active techniques that sort cells with the help of an external field (e.g., electric, magnetic, optical, acoustic, etc.). Finally, the performances of the different passive and active microfluidic sorting techniques are compared.

Conventional Cell Separation Techniques Conventional and microfluidic cell separation techniques (Fig. 1) can be classified into two categories: (a) techniques that exploit the difference in physical properties (e.g., size, density, deformability, electric charge, etc.) of cells in a heterogeneous population and (b) labeled techniques which rely on the presence of specific biological markers on the cells (e.g., antigens, recombinant proteins, etc.). The techniques which are based on the physical properties are also known as label-free techniques. The most common label-free technique for separating the components of blood is density gradient centrifugation. Although it is strictly not a single-cell separation technique, it can be used as an enrichment step to speed up the subsequent process of single-cell isolation. Common single-cell sorting techniques like FACS and MACS require the use of labels.

210

S. Pandey et al.

Fig. 1 Classification of different cell separation techniques

Centrifugation Centrifugation is the most commonly used technique to separate the different components of blood. It is widely used in both research laboratories and clinical settings. In centrifugation (Fig. 2a), the difference in the densities of various blood components (plasma, red blood cells, etc.) is exploited. About 5–10 mL blood is loaded into a test tube and spun at a high speed. As shown in Fig. 2b, after centrifugation, dense RBCs settle down at the bottom of the test tube. Plasma settles at the topmost layer, while platelets and WBC are found in an intermediate layer (buffy coat). A secondary centrifugation of the buffy coat is needed to separate the WBCs and the platelets. For this purpose, a special density gradient separation media (Hi-sep) is added to the buffy coat prior to spinning. Although centrifugation cannot separate individual rare cells, it is often used as an enrichment step before performing single-cell separation. For instance, Nagrath et al. used centrifugation as a preliminary sample preparation step prior to capturing CTCs from lysed blood using the CTC-chip (Nagrath et al. 2007).

Fluorescence-Activated Cell Sorting FACS was first developed by Bonner, Hulett, Sweet, and Herzenberg (Bonner et al. 1972; Herzenberg and Sweet 1976; Hulett et al. 1969). It was the first cell sorting

9

Single-Cell Separation

211

Fig. 2 Centrifugation used for separation of the components of blood. (a) Photo of a small tabletop centrifuge showing the slots for holding sample tubes. (b) Schematic diagram of a test tube showing the positions of different blood components after centrifugation. RBCs settle at the bottom, while plasma settles at the top. A buffy coat containing WBCs and platelets lies between the RBC and the plasma layers

technique to be commercialized by Becton Dickinson Immunocytometry Systems in the 1970s. FACS is still the most widely used cell sorting technique. According to a review published in 2002, approximately 30,000 sorters and analyzers were estimated to be in use throughout the world (Herzenberg et al. 2002). The principle of operation of FACS is based on both fluorescence and the light scattering properties of individual cells. Light scattering is used to analyze cell parameters, such as size (forward scatter) and internal structure (side scatter). Since many cells do not have an inherent fluorescence, they are tagged with fluorophore-conjugated antibodies that bind to cell markers. Cells can also be made to express recombinant fluorescent proteins, such as green fluorescent protein (GFP). In FACS (Fig. 3), the sample is encased by a sheath fluid and then focused into a narrow stream by a nozzle. The sample flow is adjusted such that the separation between the cells is large compared to the size of the cell, and only one cell is detected at a time. In the detection region, a laser excites the fluorescently tagged antibodies. The fluorescence and scatter signals from each cell are captured by multiple detectors. Once a cell passes the detection region, the stream is broken into droplets such that each droplet contains a single cell. Electric charge (negative or positive) is applied on the droplet based on the fluorescence signal of the cell it contains. Using electrostatic deflection, the charged droplet is led to a tube for further analysis. FACS allows users to target multiple markers in the same experiment. The latest FACS systems allow up to 18 markers to be detected simultaneously with the help of 6 lasers and 20 detectors. But the technique requires extensive sample

212

S. Pandey et al.

Fig. 3 Fluorescenceactivated cell sorting (FACS), in which cells are first focused such that only one cell is detected by the laser at a time. The detected cells are then enclosed in a drop, given an electric charge, and then separated by the deflector plates into different categories

preparation. Predictably, the increase in the number of markers results in an increase in the system complexity and the equipment cost. FACS needs a relatively large number of cells (~ tens of thousands) to run an experiment effectively. In spite of these limitations, FACS is still the most widely used single-cell sorting technique.

Magnetic-Activated Cell Sorting Magnetic-activated cell sorting (MACS) (Miltenyi et al. 1990) was developed to address the throughput limitations of FACS. In MACS (Fig. 4), antibody-coated magnetic beads (~100 nm diameter) specifically bind to a particular protein on the surface of the desired cell. The magnetically tagged cells are passed through a steel wool column placed in a magnetic field gradient formed by permanent magnets. The labeled cells are captured by the column in the presence of the magnetic field, while unlabelled cells are washed away. Cells attached to the column can be eluted by turning off the magnetic field. During enrichment (also known as the positive separation), the desired cells are labeled with the coated magnetic beads, and the unlabelled cells are discarded. In case of depletion (also known as negative separation), the unwanted cells are labeled and captured by the magnetic field. As discussed by (Miltenyi et al. 1990), the size of the magnetic particles must be carefully chosen. Large magnetic particles (size >0.5 μm) can respond well to a simple (non-gradient) magnetic field, but they lead to cell aggregation and decrease

9

Single-Cell Separation

213

Fig. 4 Magnetically activated cell sorting (MACS) is achieved by attaching antibody-coated magnetic nanometer-sized beads to the cells of interest

the cell viability. It becomes difficult to detach the bound cells, and consequently, the process can only be used for cell depletion. On the other hand, smaller particles (size 300 μl and an efficiency of >70%–80% for trapping WBCs. To balance the issues of increased separation efficiency with clogging-free operation, several variations on these primary design types have been reported. These approaches are passive (i.e., they do not require external energy sources), and the devices can be fabricated in a single lithography step. Mehendale et al. (2018) developed a radial pillar device (RAPID) design for simultaneous separation of multisized particles by combining the advantages of cross-flow and dead-end pillar filters. Here the pillars were arranged in concentric circles in three zones around a central inlet. The pillar gaps were decreased progressively in each zone according to the size of the particles to be captured. An angular displacement between the successive rows of pillars in the middle zone and the presence of a cross-flow outlet prevented clogging of this device by continuously removing the undesired particles. They demonstrated simultaneous separation of 10 μm and 2 μm polystyrene beads from a mixture of 2 μm, 7 μm, and 10 μm beads with a high throughput (3 ml/min) in a single experiment. Another group (Mohamed et al. 2007) adapted the design of a dead-end pillar device by dividing the pillars into four zones, with the pillar gaps decreasing in each successive zone. This design allowed the cells to deform and regain their shapes as the cells moved from zone to zone. They isolated fetal nucleated red blood cells (fNRBC), which are less dense and stiffer than normal red blood cells, from maternal blood. The fNRBCs contained in the mononuclear cell layer, obtained after a prior centrifugation step, was used as the sample for this experiment. There is a recent report (Masuda et al. 2017) on a pillar-based open channel sorting device for isolation of CTCs from blood. In this

9

Single-Cell Separation

219

design, the pillars were arranged in a hexagonal pattern, and the distance between pillars was chosen such that the target cell could easily be trapped in the hexagonal area between pillars. A major problem with CTC separation is the loss of these rare cells in the tubing connected to the microfluidic device. The open channel design developed by this group allowed direct recovery of the captured CTCs from the microfluidic chip by micropipette aspiration. Other approaches to solve the problem of clogging in filter-based devices include integration of micropumps (Cheng et al. 2016), performing pneumatic actuation (Huang et al. 2014), or introducing mechanical vibrations using piezoelectric transducers (Yoon et al. 2016). These approaches, while effective in dislodging trapped cells, require either complex microfabrication steps or the integration of power sources and transducers into the device.

Deterministic Lateral Displacement Deterministic lateral displacement (DLD) devices were developed (Huang et al. 2004) to overcome the low throughput and high clogging issues of size-exclusion separation devices. DLD is a pillar-based cell separation technique that separates cells based on size, as shown in Fig. 7. Unlike the other pillar-based devices, here the pillars in each row are displaced horizontally with respect to the previous row. Typically, the horizontal displacement (Δλ) is by a pre-determined fraction of the center-to-center distance (λ) of the pillars, which leads to a periodic arrangement. For example, if the lateral displacement (Δλ) between successive rows is λ/3, then the 1st and the 4th rows of pillars would be exactly aligned. In this device, particles with diameters smaller and larger than a critical diameter (Dc) follow different paths through the pillar network. Smaller particles continue along the same streamline, while larger particles are bumped from pillar to pillar, resulting in a lateral displacement (Inglis et al. 2006). The critical size (Dc) is related to the side-to-side gap between the pillars and the shift fraction (Δλ/λ). The DLD design allows fabrication of pillars with large gaps to sort cells with small size differences because the sorting ability of this device is directly proportional to shift fraction. Some variations in the pillar shapes have been reported in the literature. There is a report (Zeming et al. 2013) on I-shaped pillars for sorting of non-spherical cells in the DLD design. The “I” shape of the pillars continuously rotated the cells, leading to the highest lateral migration of the biconcave RBCs. DLD devices have been extensively used for the separation of rare cells. Huang and others (Huang et al. 2008) isolated nucleated RBCs from peripheral blood with a high throughput. Liu et al. (2013) used a DLD design for cancer cell enrichment prior to affinity-based separation. Another group (Au et al. 2017) used an asymmetric DLD approach to isolate CTC clusters from the blood. Their microfluidic chip consisted of two stages. The first stage had cylindrical micro-posts to deflect large particles. The second stage had one half of the pillars shaped as ellipsoids, and the half as I-shaped pillars to induce rotation of the CTC clusters for improved separation.

220

S. Pandey et al.

Fig. 7 Schematic of the deterministic lateral displacement (DLD) device. The consecutive rows of pillars are laterally displaced with respect to the previous row by a pre-determined distance

Hydrodynamic Separation Hydrodynamic separation techniques rely on the fluid flow alone to sort particles based on size, shape, or deformability. Usually there are no obstacles placed in the flow path. In this chapter, we have divided the hydrodynamic separation techniques into two regimes: (a) when Re ffi 0 (non-inertial regime) and (b) when 1 < Re < 100 (inertial regime).

Non-inertial Hydrodynamic Separation These techniques rely on simple flow division in microfluidic channels at very low Reynolds numbers. Due to linearity of the system, particles do not move from one streamline to another in these devices. The pinched flow fractionation (PFF) is one such technique demonstrated by Yamada and others (Yamada et al. 2004). As shown in Fig. 8, there are two inlets in the device, one containing the mixture of the particles to be separated and the other containing a buffer without particles. Both the sample and the buffer flow into a narrow channel (i.e., the pinched region), which after a distance suddenly widens. The flow containing the particles is tightly focused by the buffer flow such that all particles line up against one sidewall of the pinched region. This is the most critical step in the PFF technique. At this point, the centers of the aligned small and the large particles lie on different streamlines. When the particles enter the wider channel, they continue to move along their respective streamlines in the spreading flow profile. The small particles move along

9

Single-Cell Separation

221

Fig. 8 Pinched flow fractionation. A buffer stream focuses the particle stream against the wall in the pinched channel. Due to the difference in their sizes, larger and smaller particles follow different streamlines when they are focused. When channel widens, smaller cells continue to flow near to the wall and larger cells shift near the center. As the cells of different sizes follow different streamlines in the laminar flow, they can be collected in different outlets. (Reproduced by permission of the ACS publications Yamada et al. 2004)

the wall, and the large particles move away from the wall, toward the center of the device. These particles can be detected according to their sizes in a direction perpendicular to the direction of the flow. A number of outlet channels placed at the end of the wide channel can then collect each cell fraction. In this technique, the separation efficiency is strongly dependent on the tightness of focusing of the sample flow in the pinched region. Another group (Pødenphant et al. 2015) used pinched flow fractionation to separate LS174T colon cancer cells from WBCs with a separation efficiency of 90%. A design improvement on symmetric PFF was proposed by Takagi and others (Takagi et al. 2005) by making the flow resistances of the outlet channels asymmetric (e.g., asymmetric pinched flow fractionation). This modification was carried out to allow particles with much smaller size differences (~1 μm) to be separated. They also observed that the shape of non-spherical cells plays a strong role in this separation technique. The smallest dimension of a non-spherical cell (e.g., the ~2 μm width of a discoid red blood cell) determines the extent of pinching required to separate it using this method. A problem with this method is that cells may be damaged during extreme squeezing against the sidewall. Another group (Lin et al. 2013) used a pinch-flow design to concentrate large target cells prior to sorting with ratchet-shaped pillar gaps. The separation was based on the difference in both size and deformability of the cells. They sorted UC13 bladder cancer cells from leukocytes with 97% yield and 3000-fold enrichment.

222

S. Pandey et al.

Fig. 9 Schematic diagram of hydrodynamic cell separation device. (Reproduced by permission of the Royal Society of Chemistry Yamada and Seki 2005)

While PFF relied on flow focusing to line up the particles along one sidewall of a pinched channel, another technique from the same group (Yamada and Seki 2005) used the division of flow between the main channel and a number of perpendicular side channels to concentrate the particles near the sidewall (Fig. 9). This separation technique utilized the fact that the center of a particle cannot be positioned at any distance from the sidewall which is smaller than the radius of the particle. The division of flow among the main channel and the various side channels can be estimated using hydraulic circuit equations. If the “virtual width” of the flow segment entering the side channel is less than the particle diameter, the particle does not enter the side channel, even if the “physical width” of the side channel is greater than the particle diameter. The dimensions of the side channels and the flow rates were optimized to obtain the following three flow states with different virtual widths of the side flow: (i) the side channel only draws in the liquid and concentrates the particles in the main channel, (ii) the side channel draws in the small particles, and (iii) the side channel draws in the large particles. Later this design was used to demonstrate the enrichment of leukocytes from 10X diluted blood. Recently an improvement (Yamada et al. 2017) on this design was proposed. The main and the side channels were integrated into an asymmetric lattice pattern that acted like a sizebased sieve. The main channel is slanted at an angle with respect to the macroscopic flow, which further enhances the difference in the lateral positions of the small and the large cells. The slant angle of the side channel and the density of the liquid are the key design parameters in this device. There is another class of hydrodynamic separation techniques at low Reynolds numbers where the particles in laminar flow can be made to move in directions transverse to the axial flow by introducing some nonlinear effect, such as elastic

9

Single-Cell Separation

223

deformation of soft objects in a flow. Geislinger and others (Geislinger and Franke 2013) used a purely viscous non-inertial hydrodynamic effect to sort deformable blood cells. Their device consisted of a main channel and a side channel. The separation performance depended strongly on the size difference and somewhat less strongly on the difference in the shape and the deformability of the cells. The operation of this device was later demonstrated (Geislinger and Franke 2013) for sorting of melanoma cells from RBCs with an efficiency of 100%.

Inertial Hydrodynamic Separation Inertial effects (Di Carlo et al. 2007; Nivedita and Papautsky 2013) become significant in microfluidic channels when the Reynolds number lies between 1 and 100. Under this condition, both viscosity and inertia terms must be taken into account when solving the Navier-Stokes equation to obtain the flow profile. The introduction of the inertial term allows cross-stream lateral migration of particles, which was not possible under purely Stoke’s flow (Re ffi 0). For the case of neutrally buoyant particles (e.g., cells in buffer or media) in a straight microfluidic channel, there are two kinds of inertial forces on the particles: (a) a wall lift force resulting from the interaction between the particle and walls that pushes the particles away from the wall and (b) a shear gradient lift force, resulting from the parabolic flow profile, that pushes the particles away from the center. Particles of different sizes experience different wall and shear gradient forces and, therefore, migrate away from the axis to different equilibrium positions in the lateral direction. These equilibrium positions are strongly dependent on the channel geometry. These inertial forces in straight channels lead to strong particle focusing. Hur and others (Hur et al. 2011) combined inertial focusing with microscale laminar vortices to separate spiked cancer cells from blood with a very high throughput. Another group (Parichehreh et al. 2013) combined aqueous phase partitioning and inertial focusing in a straight microchannel to enrich the population of nucleated cells in the blood. The flow of the blood sample in the microfluidic chip was flanked by the flow of a dextran phase on both sides. The RBCs preferentially migrated to the dextran layer toward the wall of the channels, while the WBCs remained near the center. This difference in the positions of RBCs and WBCs was later amplified by inertial microfluidics to achieve improved separation. In addition to the lift forces, there is another force on particles in curved microchannels. Due to the parabolic flow profile, the fluid at the center of the channel is pushed outward due to the centrifugal force. To fulfill the incompressibility condition, the fluid from the sides of the channel moves toward the center. This sets up two symmetric vortices in the top and the bottom halves of the channel, leading to a secondary flow called the Dean flow. Larger channel widths, higher curvature, and higher flow velocities lead to higher Dean flows. Microchannels with spiral and serpentine geometries have been designed to separate cells based on the equilibrium between the inertial lift and the Dean forces. The ratio (Rf) between the inertial lift and the Dean forces varies as the square of the particle diameter. Therefore, particles of different sizes move to different radial equilibrium positions, leading to a size-based separation.

224

S. Pandey et al.

Several variations of the basic spiral cell separation device have been reported in the literature. A double spiral geometry (Sun et al. 2012) was developed to improve the separation resolution. Its performance was demonstrated by separating spiked tumor cells from diluted blood. Zhou and others (Zhou et al. 2013) modulated the channel aspect ratio. An initial high aspect ratio segment was used first to focus the particles. Then a later low aspect ratio segment separated the particles according to their sizes. They spiked and isolated prostate epithelial tumor cells from the blood. A spiral device with a trapezoidal cross-section (Warkiani et al. 2014) was designed to improve the sorting performance. The height of the outer channel wall was greater than that of the inner wall, leading to a trapezoidal cross-section. The inlet to the channel was placed at the outside of the spiral. The blood with lysed RBCs and spiked with circulating tumor cells (CTCs) was introduced at the inlet and separated with an ultrahigh throughput (7.5 ml of the blood in 8 min). The large CTCs were focused near the inner wall of the spiral channel, and the smaller WBCs were focused near the outer wall. One major limitation of the inertial focusing technique for clinical applications is the need to heavily dilute blood to reduce cell-cell interactions.

Microfluidic Active Separation Techniques Unlike passive techniques, where the cell separation is carried out using microscale obstacles or the fluid flow alone, active separation techniques require the use of an external field (e.g., electric, optical, magnetic, acoustic, etc.). We discuss a few of the most common active separation techniques here.

Dielectrophoresis (DEP) Dielectrophoresis is a separation technique based on the intrinsic polarizability of the cells. Polarizable particles, when placed in a nonuniform electric field, move either toward higher or lower field regions depending on the sign of their ClausiusMossotti (CM) factor. If polarizability of the particle is higher than the polarizability of the buffer medium in which it is suspended, the particle moves toward the higher electric field. This phenomenon is known as positive dielectrophoresis (pDEP). On the other hand, if the polarizability of the particle is lower than that of the medium, the particle moves toward the weaker electric field. This phenomenon is called negative dielectrophoresis (nDEP) (Gossett et al. 2010). The DEP force depends strongly on the size of the cell and can be used for size-based sorting. The equilibrium position of a particle in the flow depends on the relative magnitudes of the DEP and the drag forces. The cells expressing negative DEP are carried away by the flow, while the cells expressing positive DEP are retained in the channel by the electric field gradient. When an AC electric field is used for DEP (Fig. 10) instead of a DC field, the electrochemical reactions at the electrodes are minimized. In addition, cells can

9

Single-Cell Separation

225

Fig. 10 Schematic diagram explaining dielectrophoresis (DEP). In positive DEP, particles move toward the higher electric field. In negative DEP, the particles move toward the electric field minima

be sorted based on both their size and frequency response. Particles have a positive CM factor in some frequency ranges and a negative CM factor in other frequencies. The frequency at which the CM factor changes its sign is called the crossover frequency. Mammalian cells show negative DEP at low frequencies in the 10 kHz to 1 MHz frequency range and positive DEP at the higher frequencies. A mixture of particles can be separated based on their crossover frequency. Vykoukal and others (Vykoukal et al. 2009) measured the specific membrane capacitances of leukocyte subpopulations using crossover frequency measurements. Recently a high-throughput DEP was demonstrated (Faraghat et al. 2017) to separate spiked cancer cells from RBCs. Another report (Antfolk et al. 2017) used DEP to trap and sort spiked prostate cancer cells (DU145) in blood. The sample was pre-concentrated by acoustophoresis prior to sorting. Elvington and others (Elvington et al. 2013) reported a contactless DEP technique to eliminate problems, such as bubble formation and electrolysis. In contactless DEP, there is no direct contact between metallic electrodes and the sample. Instead of a metallic electrode, a fluid electrode is capacitively coupled to an AC voltage source. The highest throughput obtained with DEP is still lower than other cell separating techniques, such as inertial sorting. Another disadvantage is that DEP requires the fabrication of integrated planar electrodes in the microfluidic channel. Wang and others (Wang et al. 2000) developed a DEP-FFF system for the separation of breast cancer cells from CD34+ hematopoietic stem cells and from T lymphocytes. Their device consisted of a chamber equipped with an array of microfabricated interdigitated electrodes at the bottom. Hu and others (Hu et al. 2005) demonstrated >200-fold enrichment of rare cells by DEP in a microfluidic device.

Magnetic Separation The idea of magnetic separation in a microfluidic channel originated from MACS. Magnetic separation of blood cells looked promising due to the inherent magnetic properties of the WBCs and the RBCs. WBCs are diamagnetic, and RBCs may be diamagnetic (oxyhemoglobin state) or paramagnetic (deoxyhemoglobin state). Furlani (2007) fabricated a microfluidic channel by embedding an array of permalloy

226

S. Pandey et al.

elements underneath it, which could be magnetized using an external bias field. It led to a nonuniform field distribution. Separation of WBCs and deoxygenated RBCs was demonstrated in this device. Cells that are not intrinsically magnetic can be labeled with magnetic nanoparticles. Pamme and Wilhelm (2006) magnetically labeled mouse macrophages and HeLa cells following the endocytic pathway of the cells. A magnetic field was applied perpendicular to the flow direction in a microfluidic chip. The magnetically labeled cells were deflected toward the magnet, while the non-labeled cells flowed toward the outlet.

Acoustophoresis Acoustophoresis uses ultrasound waves to separate cells. If standing waves of ultrasound are produced in a microchannel, then cells either move to a pressure node or an antinode depending on the property of the cells. The acoustic force on a cell depends on its volume, relative density between the cell and the surrounding fluid, and their relative compressibility. It also depends on the radiation pressure, the wavelength of the sound, and a contrast factor. Most mammalian cells move to the nodes under the action of this force. If the half wavelength of a standing wave matches the channel width, a pressure node is formed at the center, and antinodes are created near the walls. As a result, cells with negative contrast move to the center and are carried away by the laminar flow through a central outlet. The cell-free liquid can be collected through peripheral outlets. A similar concept has been used to separate plasma from blood (Lenshof et al. 2009). Figure 11 shows how acoustophoresis may be used for single-cell separation. The acoustic radiation force acting on particles of different sizes, but same densities, will focus the particles in the same region of the channel. As the force varies with the volume of the cell, cells of different sizes will cross the streamlines at different rates. There is a recent report (Shields IV et al. 2014) on using acoustophoresis for capture of cells using elastomeric particles. Elastomeric particles tend to move to the channel walls (antinodes). If cells can be bound to elastomeric particles by immunological labeling in such a way that the force on the particles is greater than the force on the cells, then the particles carry these cells to the antinodes, e.g., the channel walls, of acoustic standing waves. Unlabelled cells at the pressure nodes can be

Fig. 11 Single rare cell separation using surface acoustic waves from the mixed cell population

9

Single-Cell Separation

227

carried away by the flow. Another group (Augustsson et al. 2012) used microfluidic acoustophoresis to separate prostate cancer cells from blood using an additional acoustic pre-alignment procedure.

Affinity-Based Separation Affinity-based cell separation uses labels, typically based on antigen-antibody interactions. The microfluidic chip is modified with a specific antibody which is capable of binding to the antigen on the surface of the desired cell. When the sample passes through the microfluidic chip, the desired cell is captured, while the rest of the sample is collected at the outlet. Finally, the immobilized cell is eluted from the microchannels for analysis. This technique has much better sensitivity and specificity compared to the other label-free techniques due to the use of labels specific to the target cell (Hu et al. 2016). An affinity-based microfluidic chip was successfully demonstrated by the Toner group (Nagrath et al. 2007) for the separation of epithelial CTCs from whole blood. The CTC-chip consisted of 78,000 micro-posts which were coated with the anti-epithelial cell adhesion molecule (EpCAM). The large number of micro-posts increased the surface area for binding compared to a straight microfluidic channel. The chip was clinically tested on patients with metastatic prostate, lung, breast, and colon cancer. It detected CTCs in 115 of 116 (99%) samples. Later, same group developed a herringbone-chip (HB-chip) with high throughput (Stott et al. 2010). The design was chosen such that it introduces chaotic flow in the device which further increases the cell-to-surface interaction. Another group (Sarkar et al. 2016) reported a continuous flow and high-throughput microfluidic device based on combining inertial microfluidics and affinity separation. Liu and others (Liu et al. 2013) demonstrated an integrated microfluidic chip for CTC separation from blood separation combining deterministic lateral displacement and affinity-based techniques. Breast cancer cells were spiked in the blood sample and separated with an enrichment factor of 1500X, a throughput of 9.6 ml/min, capture yield of 90%, and purity of 50%. Murlidhar et al. (2014) reported an affinity-based ultrahigh-throughput CTC capture device (Onco-Bean chip). Bean-shaped pillars were designed to increase the interaction between the antibody-coated pillars and circulating tumor cells. The pillars were arranged radially, reducing the shear along the axis of cell movement and thus allowing increased affinity-based capture of CTCs. Each successive row along the radius was displaced randomly to improve the interaction between CTCs and micro-posts. This device achieved an ultrahigh throughput of 10 ml/hr. Cells were sorted with an efficiency greater than 80%. The recovered cells were tested for viability, and 93% of the cells were found to be viable. Sheng and others (Sheng et al. 2012) used aptamers in combination with a pillar-based microfluidic chip for tumor cell separation from whole blood with a capture efficiency of >95% and purity of  81%. The device consisted of >59,000 micropillars to improve the cell-to-surface interaction. The same group (Sheng et al. 2014) later implemented geometrically enhanced mixing in the microfluidic chip prior to cancer cell separation to improve the purity and the capture efficiency. This device achieved a tumor cell capture efficiency of >90% and purity of >84% by increasing the groove widths.

228

S. Pandey et al.

Table 2 Parameters to quantify the efficiency of conventional cell separation techniques

Method Centrifugation Fluorescenceactivated cell sorting (FACS) Magneticactivated cell sorting (MACS)

Flow rate (ml/min) –

Purity (%) 95–97

Recovery (%) 40–90

Enrichment (fold) –

Throughput (cells/s) –

–

90

–

–

10,000

8

–

>90

>100

–

References Boyum 1977 (Bonner et al. 1972; Lee et al. 2017) Miltenyi et al. 1990

Comparison Between Different Microfluidic Separation Techniques As shown in Table 2, non-inertial hydrodynamic separation, deterministic lateral displacement, and dielectrophoresis techniques have reported the highest capture efficiencies, nearing 100% (Geislinger and Franke 2013; Antfolk et al. 2017; Huang et al. 2008). On the other hand, deterministic lateral displacement, non-inertial hydrodynamic, and acoustophoresis techniques achieved the highest cell recovery (nearly 99%) (Au et al. 2017; Lin et al. 2013; Pødenphant et al. 2015; Augustsson et al. 2012). However, inertial hydrodynamic, membrane, and pillar-based separation techniques also reported recoveries in excess of 90% (Parichehreh et al. 2013; Tang et al. 2014; Masuda et al. 2017). The recovery in dielectrophoresis was somewhat lower (~76%), which can be improved by optimizing the flow rates (Antfolk et al. 2017). The highest purity for the separated cells was reported to be 99%, achieved by dielectrophoresis and acoustophoresis (Antfolk et al. 2017; Augustsson et al. 2012). The enrichment factor of 3000-fold and throughput 9.6 ml/min have been achieved highest by hydrodynamic and affinity based (Lin et al. 2013; Liu et al. 2013). In comparison, each individual technique appears to excel in at most one or two separation parameters. We believe that by combining two or more techniques in an experiment, all the separation parameters can be improved (Table 3).

Conclusion Single-cell separation techniques play a major role in the diagnosis of diseases. In the chapter, we have discussed the state-of-the-art conventional and microfluidic cell separation techniques. Microfluidic techniques that are based on physical properties (e.g., deformability, size, polarizability, magnetic susceptibility, etc.) do not require any labeling. Compared to label-free techniques, affinity-based rare cell separation

Size, affinity

Size, deformability

Affinity-based separation

Pillar-based separation

Techniques Dielectrophoresis

Separation criteria Size, density, polarizability

92 50 50

– 70 –

– –

0.3

0.5

60

>90 – –

0.06

5.8  10–3

0.16–0.33

6.5  10–4

98 (MLC), 97 (PBMC) –

–

95

3.6  10–3

6  10–6

84

>80

0.016

–

90.6

–

81

–

90

1.6

99 (MLC) 95 (PBMC) –

–

–

65

0.016

–

–

–

Purity (%) 100

5  103

Recovery (%) 76

Capture/ Separation efficiency (%) 100

Flow rate (mL/min) 4  103

Table 3 Comparison of microfluidic techniques

–

–

–

–

–

–

–

1500

106

35

>200

–

Enrichment (fold) –

–

–

0.04–0.08

–

0.6

0.035

0.16

9.6

0.03

–

–

–

Throughput (mL/min) 0.08

Single-Cell Separation (continued)

References Antfolk et al. 2017 Elvington et al. 2013 Hu et al. 2005 Wang et al. 2000 Nagrath et al. 2007 Liu et al. 2013 Murlidhar et al. 2014 Sheng et al. 2012 Sheng et al. 2014 Mohamed et al. 2007 Masuda et al. 2017 McFaul et al. 2012 VanDelinder and Groisman 2006

9 229

Size

Size, deformability

Size

Size, density

Non-inertial hydrodynamic separation

Membrane-based separation

Deterministic lateral displacement

Acoustophoresis

Techniques Inertial hydrodynamic separation

Separation criteria Size

Table 3 (continued)

–

>95 100 100 –

0.2–2

–

2  10–4

8.3  10–3

0.45

0.56

–

97

–

–

>95 93.6–97.9 (for fixed cells) 72.5–93.9 (for nonfixed)

100 –

97.4–98.4 (for fixed cells) 79.6–99.7 (for nonfixed)

–

–

–

–

99

–

98

–

–

100

–

–

–

>90

90

–

>99

0.1

–

Purity (%) –

5  10–4 – 8.3  10–4 0.01

88.5

96.8

–

Recovery (%) 90

Capture/ Separation efficiency (%) –

Flow rate (mL/min) 0.03

–

10–20

–

–

–

–

3000

46

–

–

15

Enrichment (fold) –

–

0.42–0.05

0.05–0.02

–

–

–

8.33  10–6

–

–

–

0.33

Throughput (mL/min) –

References Parichehreh et al. 2013 Sun et al. 2012 Zhou et al. 2013 Pødenphant et al. 2015 Geislinger and Franke 2013 Lin et al. 2013 Tang et al. 2014 Yoon et al. 2016 Zeming et al. 2013 Au et al. 2017 Huang et al. 2008 Augustsson et al. 2012

230 S. Pandey et al.

9

Single-Cell Separation

231

shows better separation performance. The interaction between the antibody-coated microfluidic channel and the antigens on the target cell surface was further strengthened by the introduction of pillars to increase the surface-to-volume ratio. Another challenge in single-cell separation is the preprocessing of clinical samples. Most of the techniques discussed here needed a dilution of blood samples, which added an extra processing step. More single-cell separation techniques that work with sample without preprocessing (e.g., whole blood) need to be developed. The microfluidic separation techniques that we reviewed are still at the laboratory level or the proofof-concept stage. Though some of the microfluidic devices have reached clinical trials, these devices are still not widely used in the settings for which these were originally developed. There is still a strong need for microfluidic single-cell separation techniques which are affordable, reliable in clinical settings, and capable of replacing the existing techniques.

References Antfolk M, Kim SH, Koizumi S et al (2017) Label-free single-cell separation and imaging of cancer cells using an integrated microfluidic system. Sci Rep 7:1–12. https://doi.org/10.1038/ srep46507 Au SH, Edd J, Stoddard AE et al (2017) Microfluidic isolation of circulating tumor cell clusters by size and asymmetry. Sci Rep 7:1–10. https://doi.org/10.1038/s41598-017-01150-3 Augustsson P, Magnusson C, Nordin M et al (2012) Microfluidic, label-free enrichment of prostate cancer cells in blood based on acoustophoresis. Anal Chem 84:7954–7962. https://doi.org/ 10.1021/ac301723s Beebe DJ, Mensing GA, Walker GM (2002) Physics and applications of microfluidics in biology. Annu Rev Biomed Eng 4:261–286. https://doi.org/10.1146/annurev.bioeng.4.112601.125916 Bonner WA, Hulett HR, Sweet RG, Herzenberg LA (1972) Fluorescence activated cell sorting. Rev Sci Instrum 43(3):404–409. https://doi.org/10.1063/1.1685647 Boyum A (1977) Separation of lymphocytes, lymphocyte subgroups and monocytes: a review. Lymphology 10(2):71–76 Brody JP, Osborn TD, Forster FK et al (1996) A planar microfabricated fluid filter. Sensors Actuators A Phys 54(1–3):704–708. https://doi.org/10.1016/S0924-4247(97)80042-8 Chen X, Cui DF, Liu CC, Li H (2008) Microfluidic chip for blood cell separation and collection based on crossflow filtration. Sensors Actuators B Chem 130:216–221. https://doi.org/10.1016/ j.snb.2007.07.126 Cheng Y, Ye X, Ma Z, Xie S, Wang W (2016) High-throughput and clogging-free microfluidic filtration platform for on-chip cell separation from undiluted whole blood. Biomicrofluidics 10(1):014118. https://doi.org/10.1063/1.4941985 Crowley TA, Pizziconi V (2005) Isolation of plasma from whole blood using planar microfilters for lab-on-a-chip applications. Lab Chip 5:922. https://doi.org/10.1039/b502930a Datta S, Malhotra L, Dickerson R et al (2015) Laser capture microdissection: big data from small samples. Histol Histopathol 30:1255–1269. https://doi.org/10.14670/HH-11-622.Laser Di Carlo D, Irimia D, Tompkins RG, Toner M (2007) Continuous inertial focusing, ordering, and separation of particles in microchannels. Proc Natl Acad Sci 104:18892–18897. https://doi.org/ 10.1073/pnas.0704958104 Elvington ES, Salmanzadeh A, Stremler MA, Davalos RV (2013) Label-free isolation and enrichment of cells through contactless dielectrophoresis. J Vis Exp:1–10. https://doi.org/10.3791/ 50634

232

S. Pandey et al.

Emmert Buck MR, Bonner RF, Smith PD et al (1996) Laser capture microdissection. Science 274(5289):998–1001. https://doi.org/10.1126/science.274.5289.998 Faraghat SA, Hoettges KF, Steinbach MK et al (2017) High-throughput, low-loss, low-cost, and label-free cell separation using electrophysiology-activated cell enrichment. Proc Natl Acad Sci 114:4591–4596. https://doi.org/10.1073/pnas.1700773114 Fend F, Raffeld M (2000) Laser capture microdissection in pathology. J Clin Pathol 53:666–672. https://doi.org/10.1136/jcp.53.9.666 Furlani EP (2007) Magnetophoretic separation of blood cells at the microscale. J Phys D 40(5):1313. https://doi.org/10.1088/0022-3727/40/5/001 Geislinger TM, Franke T (2013) Sorting of circulating tumor cells (MV3-melanoma) and red blood cells using non-inertial lift. Biomicrofluidics 7:1–9. https://doi.org/10.1063/1.4818907 Gossett DR, Weaver WM, MacH AJ et al (2010) Label-free cell separation and sorting in microfluidic systems. Anal Bioanal Chem 397:3249–3267. https://doi.org/10.1007/s00216010-3721-9 Herzenberg LA, Sweet RG (1976) Fluorescence activated cell sorting. Rev Sci Instrum 234:108–117. https://doi.org/10.1063/1.1685647 Herzenberg LA, Parks D, Sahaf B et al (2002) The history and future of the fluorescence activated cell sorter and flow cytometry: a view from Stanford. Clin Chem 48:1819–1827. https://doi.org/ 10.1038/scientificamerican0376-108 Hu X, Bessette PH et al (2005) Marker-specific sorting of rare cells using dielectrophoresis. Proc Natl Acad Sci U S A 102(44):15757–15761. https://doi.org/10.1073/pnas.0507719102 Hu P, Zhang W, Xin H, Deng G (2016) Single cell isolation and analysis. Front Cell Dev Biol 4:1–12. https://doi.org/10.3389/fcell.2016.00116 Huang LR, Cox EC, Austin RH, Sturm JC (2004) Continuous particle separation through deterministic lateral displacement. Science 304(5673):987–991. https://doi.org/10.1126/ science.1094567 Huang R, Barber TA, Schmidt MA et al (2008) A microfluidics approach for the isolation of nucleated red blood cells (NRBCs) from the peripheral blood of pregnant women. Prenat Diagn 28(10):892–899. https://doi.org/10.1002/pd.2079 Huang SB, Zhao Y, Chen D et al (2014) A clogging-free microfluidic platform with an incorporated pneumatically driven membrane-based active valve enabling specific membrane capacitance and cytoplasm conductivity characterization of single cells. Sensors Actuators B Chem 190:928–936. https://doi.org/10.1016/j.snb.2013.09.070 Hulett HR, Bonner WA, Barrett J et al (1969) Cell sorting: automated separation of mammalian cells as a function of intracellular fluorescence. Science 166(3906):747–749. https://doi.org/10.1126/ science.166.3906.747 Hur SC, Mach AJ, Di Carlo D (2011) High-throughput size-based rare cell enrichment using microscale vortices. Biomicrofluidics 5:1–10. https://doi.org/10.1063/1.3576780 Inglis DW, Davis JA, Austin RH, Sturm JC (2006) Critical particle size for fractionation by deterministic lateral displacement. Lab Chip 6:655. https://doi.org/10.1039/b515371a Ji HM, Samper V, Chen Y et al (2008) Silicon-based microfilters for whole blood cell separation. Biomed Microdevices 10:251–257. https://doi.org/10.1007/s10544-007-9131-x Lee W, Tseng P, Di Carlo D (2017) Microfluidic cell sorting and separation technology. In: Lee W, Tseng P, Di Carlo D (eds) Microtechnology for cell manipulation and sorting. Microsystems and nanosystems. Springer, Cham Lenshof A, Ahmad-Tajudin A, Järås K, Swärd-Nilsson AM et al (2009) Acoustic whole blood plasmapheresis chip for prostate specific antigen microarray diagnostics. Anal Chem 81(15):6030–6037. https://doi.org/10.1021/ac9013572 Lin BK, McFaul SM, Jin C et al (2013) Highly selective biomechanical separation of cancer cells from leukocytes using microfluidic ratchets and hydrodynamic concentrator. Biomicrofluidics 7:17–20. https://doi.org/10.1063/1.4812688

9

Single-Cell Separation

233

Liu Z, Zhang W, Huang F et al (2013) High throughput capture of circulating tumor cells using an integrated microfluidic system. Biosens Bioelectron 47:113–119. https://doi.org/10.1016/j. bios.2013.03.017 Masuda T, Song W, Nakanishi H et al (2017) Rare cell isolation and recovery on open- channel microfluidic chip. PLoS One 12:1–14. https://doi.org/10.1371/journal.pone.0174937 McFaul SM, Lin BK, Ma H (2012) Cell separation based on size and deformability using microfluidic funnel ratchets. Lab Chip 12:2369. https://doi.org/10.1039/c2lc21045b Mehendale N, Sharma O, D’Costa C, Paul D (2018) A radial pillar device (RAPID) for continuous and high-throughput separation of multi-sized particles. Biomed Microdevices 20:1–9. https:// doi.org/10.1007/s10544-017-0246-4 Miltenyi S, Müller W, Weichel W, Radbruch A (1990) High gradient magnetic cell separation with MACS. Cytometry 11:231–238. https://doi.org/10.1002/cyto.990110203 Mohamed H, Turner JN, Caggana M (2007) Biochip for separating fetal cells from maternal circulation. J Chromatogr A 1162:187–192. https://doi.org/10.1016/j.chroma.2007.06.025 Murlidhar V, Zeinali M, Grabauskiene S et al (2014) A radial flow microfluidic device for ultrahigh-throughput affinity-based isolation of circulating tumor cells. Small 10:4895–4904. https:// doi.org/10.1002/smll.201400719 Nagrath S, Sequist LV, Maheswaran S et al (2007) Isolation of rare circulating tumour cells in cancer patients by microchip technology. Nature 450:1235–1239. https://doi.org/10.1038/ nature06385.Isolation Nivedita N, Papautsky I (2013) Continuous separation of blood cells in spiral microfluidic devices. Biomicrofluidics 7(5):054101. https://doi.org/10.1063/1.4819275 Pamme N, Wilhelm C (2006) Continuous sorting of magnetic cells via on-chip free-flow magnetophoresis. Lab Chip 6(8):974–980. https://doi.org/10.1039/B604542A Parichehreh V, Medepallai K, Babbarwal K, Sethu P (2013) Microfluidic inertia enhanced phase partitioning for enriching nucleated cell populations in blood. Lab Chip 13:892. https://doi.org/ 10.1039/c2lc40663b Pødenphant M, Ashley N, Koprowska K et al (2015) Separation of cancer cells from white blood cells by pinched flow fractionation. Lab Chip 15:4598–4606. https://doi.org/10.1039/ C5LC01014D Sarkar A, Hou HW, Mahan AE et al (2016) Multiplexed affinity-based separation of proteins and cells using inertial microfluidics. Sci Rep 6:1–9. https://doi.org/10.1038/srep23589 Sheng W, Chen T, Kamath R, Xiong X et al (2012) Aptamer-enabled efficient isolation of cancer cells from whole blood using a microfluidic device. Anal Chem 84(9):4199–4206. https://doi. org/10.1021/ac3005633 Sheng W, Ogunwobi O, Chen T, Zhang J (2014) Capture, release and culture of circulating tumor cells from pancreatic cancer patients using an enhanced mixing chip. Lab Chip 14:89–98. https://doi.org/10.1039/c3lc51017d.Capture Shields CW IV, Johnson LM, Gao L, López GP (2014) Elastomeric negative acoustic contrast particles for capture, acoustophoretic transport, and confinement of cells in microfluidic systems. Langmuir 30:3923–3927. https://doi.org/10.1021/la404677w Songjaroen T, Dungchai W, Chailapakul O, Henry CS et al (2012) Blood separation on microfluidic paper-based analytical devices. Lab Chip 12(18):3392–3398. https://doi.org/10.1039/ C2LC21299D Stone HA, Kim S (2001) Microfluidics: basic issues, applications, and challenges. AICHE J 47(6):1250–1254. https://doi.org/10.1002/aic.690470602 Stott SL, Hsu C-H, Tsukrov DI et al (2010) Isolation of circulating tumor cells using a microvortexgenerating herringbone-chip. Proc Natl Acad Sci 107(43):18392–18397. https://doi.org/ 10.1073/pnas.1012539107 Sun J, Li M, Liu C et al (2012) Double spiral microchannel for label-free tumor cell separation and enrichment. Lab Chip 12:3952. https://doi.org/10.1039/c2lc40679a

234

S. Pandey et al.

Takagi J, Yamada M, Yasuda M, Seki M (2005) Continuous particle separation in a microchannel having asymmetrically arranged multiple branches. Lab Chip 5:778. https://doi.org/10.1039/ b501885d Tang Y, Shi J, Li S et al (2014) Microfluidic device with integrated microfilter of conical-shaped holes for high efficiency and high purity capture of circulating tumor cells. Sci Rep 4:1–7. https://doi.org/10.1038/srep06052 VanDelinder V, Groisman A (2006) Separation of plasma from whole human blood in a continuous cross-flow in a molded microfluidic device. Anal Chem 78:3765–3771. https://doi.org/10.1021/ ac060042r VanDelinder V, Groisman A (2007) Perfusion in microfluidic cross-flow: separation of white blood cells from whole blood and exchange of medium in a continuous flow. Anal Chem 79:2023–2030. https://doi.org/10.1021/ac061659b Vykoukal DM, Gascoyne PRC, Vykoukal J (2009) Dielectric characterization of complete mononuclear and polymorphonuclear blood cell subpopulations for label-free discrimination. Integr Biol 1:477. https://doi.org/10.1039/b906137a Wachtel S, Shulman L, Sammons D (2001) Fetal cells in maternal blood. Clin Genet 59:74–79. https://doi.org/10.1034/j.1399-0004.2001.590202.x Wang X-B et al (2000) Cell separation by dielectrophoretic field-flow-fractionation. Anal Chem 72(4):832–839. https://doi.org/10.1021/ac990922o Warkiani ME, Guan G, Luan KB et al (2014) Slanted spiral microfluidics for the ultra-fast, labelfree isolation of circulating tumor cells. Lab Chip 14:128–137. https://doi.org/10.1039/ C3LC50617G Wilding P, Kricka LJ, Cheng J, Hvichia G et al (1998) Integrated cell isolation and polymerase chain reaction analysis using silicon microfilter chambers. Anal Biochem 257(2):95–100. https://doi.org/10.1006/abio.1997.2530 Yamada M, Seki M (2005) Hydrodynamic filtration for on-chip particle concentration and classification utilizing microfluidics. Lab Chip 5:1233. https://doi.org/10.1039/b509386d Yamada M, Nakashima M, Seki M (2004) Pinched flow fractionation: continuous size separation of particles utilizing a laminar flow profile in a pinched microchannel. Anal Chem 76:5465–5471. https://doi.org/10.1021/ac049863r Yamada M, Seko W, Yanai T et al (2017) Slanted, asymmetric microfluidic lattices as size-selective sieves for continuous particle/cell sorting. Lab Chip 17:304–314. https://doi.org/10.1039/ C6LC01237J Yoon Y, Kim S, Lee J et al (2016) Clogging-free microfluidics for continuous size-based separation of microparticles. Sci Rep 6:1–8. https://doi.org/10.1038/srep26531 Zeming KK, Ranjan S, Zhang Y (2013) Rotational separation of non-spherical bioparticles using I-shaped pillar arrays in a microfluidic device. Nat Commun 4:1625–1628. https://doi.org/ 10.1038/ncomms2653 Zhou J, Giridhar PV, Kasper S, Papautsky I (2013) Modulation of aspect ratio for complete separation in an inertial microfluidic channel. Lab Chip 13:1919. https://doi.org/10.1039/ c3lc50101a

Technologies for Automated Single Cell Isolation

10

Julian Riba, Stefan Zimmermann, and Peter Koltay

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Size and Morphology of Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample Preparation: Single-Cell Suspensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Considerations, Definitions, and Classifications for Single-Cell Isolation . . . . . . . . . . . . . . Automated Single-Cell Isolation Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Limiting Dilution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluorescence-Activated Cell Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Dispensing (SCD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidic Single-Cell Isolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Automated Micromanipulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

236 237 237 238 239 244 244 246 247 250 257 257 259

Abstract

The isolation of individual cells has gained tremendous importance with the advent of new methods for highly parallel single-cell analysis. A prerequisite for effective clonal cultivation or single-cell analysis is the efficient isolation of individual cells from liquid cell suspensions. This review provides an overview of technologies that are used to automate the isolation of single cells for subsequent cultivation or analysis. First, currently available technologies are classified based on their major technical characteristics. Then, the most prominent technologies

J. Riba · S. Zimmermann · P. Koltay (*) Laboratory for MEMS Applications, Department of Microsystems Engineering-IMTEK, University of Freiburg, Freiburg, Germany e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_9

235

236

J. Riba et al.

such as limiting dilution, FACS, single-cell printing, hydrodynamic trapping, droplet microfluidics, and cell manipulation by external forces are described in detail. Furthermore, the individual features of each technology with focus on throughput, isolation efficiency, level of automation, flexibility in terms of cell types, and their suitability for specific downstream processing and analysis methods are discussed. In contrast to previous works, this review provides a classification approach for single-cell isolation technologies according to performance requirements, makes specific reference to methods for the isolation of microbial cells, and discusses sample input requirements, which is an important aspect in particular for diagnostic purposes. Abbreviations

CLD DEP FACS LD MNC OET OT SCD

Cell line development Dielectrophoresis Fluorescence-activated cell sorting Limiting dilution Mononuclear cells Optoelectronic tweezer Optical tweezer Single-cell dispensing

Introduction There is an increasing interest in deciphering the heterogeneity of cell populations on a single-cell level following the improvement and cost reduction of analytical methods such as the latest generation sequencing technologies over the last decade. Deconstruction of the complex composition of solid tumors at the single-cell level can help to better understand their clonal evolution. Beyond the rapidly evolving field of single-cell genomics (see, e.g., the review Gawad et al. 2016), single-cell isolation is a crucial step for the establishment of clonal cell lines. This is of particular interest for the pharmaceutical industry, where the stable production cell lines for therapeutic proteins should be derived from a single-cell progenitor as a regulatory requirement (European Medicines Agency 2016; Elder 2017). Figure 1 illustrates these two main application fields in which single-cell isolation is of continuously growing importance. Depending on the application, the cell type, the required number of single cells in total, as well as many other specific requirements regarding the cells and the subsequent processing steps, a variety of single-cell isolation technologies have been developed in the past years. An overview of the different classes of such technologies will be presented in the following with special emphasis on the feasibility and requirements for isolation of certain cell types such as microbial cells.

10

Technologies for Automated Single Cell Isolation

237

Fig. 1 Currently the main application branches that require automated single-cell isolation are clonal cell line development for production of therapeutic proteins and single-cell molecular analysis. Common sources for the starting material are cell cultures, liquid biopsies, tissue samples, or environmental samples. Prior to the isolation step, the cells have to be brought into suspension

Cell Samples Cells are the smallest entity of life that can replicate independently. In general, they are separated into two groups, eukaryotic cells which have a membrane-bound nucleus and prokaryotic cells which lack a nucleus and other membraneenclosed organelles. Prokaryotes can be subdivide into two domains, archaea and bacteria. Typically, prokaryotic cells are around 1 μm in size, while eukaryotic cells are 10 μm and larger. Though this holds true for many species, significant variation in size exists. Many of the cell isolation methods discussed below either require sizebased selection of cells or are tailored to a certain cell size. Thus, most technologies can only operate with certain types of cells. Therefore, the size and morphology of different cell types are briefly reviewed in the following section.

Size and Morphology of Cells Typically, prokaryotic cells are much smaller than eukaryotic cells. Although most species range from 0.8 to 3 μm in size, there are some exceptions that can be as small as 0.3 μm and as large as several hundred microns (Levin and Angert 2015). The size ranges given here refer to the average diameter. Their shapes are very diverse including spheres (cocci), rods (bacilli), and spirals. Due to their small size and the variable shapes, the isolation of prokaryotic cells is particularly challenging (Fig. 2). Among the eukaryotes, the smallest cell size which could be found in yeast are single-cell fungi (4–6 μm in size). Beyond yeasts, there is an enormous diversity in the fungus kingdom which is only understood in parts yet (Hawksworth and Lücking 2017). Typical fungal cells grow as filamentous, elongated tubules, so-called hyphae

238

J. Riba et al.

Fig. 2 Illustration of common cell types and their typical shapes and size ranges in suspension

with an average diameter of 4–6 μm, but their length often exceeds several centimeters. Most human cells are in the range of 10–25 μm. Smaller exceptions are sperm cells (head 4–5 μm in length) and red blood cells and lymphocytes (7–8 μm in diameter). Cardiomyocytes, megakaryocytes, and fat cells are frequently larger than 25 μm, and the largest human cells are female egg cells, the oocytes (>100 μm in diameter). Plant cells are usually within 10–100 μm in size, and most of them are therefore larger than the mammalian counterparts. However, some mammalian cells such as nerve and muscle cells can extend up to several meters. Clearly, the diversity of cells in terms of size and shape is huge, and the cell types of interest determine the design or even the technology that can be appropriate for single-cell isolation. This is especially true for microfluidic concepts, as the dimensions of the channels are of a similar scale as the cells to be isolated. So far, most single-cell applications focus on mammalian cells. Therefore, most single-cell isolation technologies are tailored for these higher-organized and typically larger cells. With an increasing interest in analysis of prokaryotic cells or fungi, the demand of single-cell isolation technologies for these highly diverse and often very small cells is expected to rise.

Sample Preparation: Single-Cell Suspensions The current single-cell isolation methods have rather stringent requirements regarding the input sample: most methods require (i) a suspension of single cells in (ii) a liquid with physical properties (viscosity, surface tension) within a specific range and (iii) a cell concentration within a specific range. For random isolation methods – such as limiting dilution or droplet microfluidics – the latter crucially determines the single-cell efficiency, as discussed in more detail in the sections below. For this purpose, several manual or automated cell counting methods exist (Freshney 2015) to enumerate cells within a sample volume, from which the required dilution factors can be determined.

10

Technologies for Automated Single Cell Isolation

239

In case the sample originates from cultured cells, it is usually straightforward to harvest and resuspend the cells in an appropriate physiological medium (e.g., phosphate-buffered saline, PBS). Cells from tissue samples such as biopsies from solid tumors are often difficult to dissociate and to bring into suspension, especially if the cell integrity needs to be maintained. For single-cell sequencing, this problem can be sometimes overcome by single-nucleus sequencing of the genome (Leung et al. 2015) or (nuclear) transcriptome (Lake et al. 2017), as nuclei can be more easily isolated from tissue using established protocols (Hymer and Kuff 1964). Usually, cells from liquid biopsies such as blood can be readily recovered. Commonly, blood mononuclear cells (MNCs) are separated from plasma, erythrocytes, and granulocytes by density gradient centrifugation using Ficoll-Paque™. In this context it should be emphasized that any sample preparation method could have a detrimental effect on cell integrity to varying degree which should be minimized as far as possible. Typical stress factors are an inadequate temperature, suboptimal buffer conditions, enzymatic dissociation of cells, or mechanical stress, respectively, damage by centrifugation, resuspension, or vortexing of the cells. Thereby, the impact on the cells is often depending on the exposure time to these factors as well as on process and storage times prior to the actual single-cell isolation process. In the same vein, different sources of samples (fresh, frozen, or formalinfixed paraffin-embedded (FFPE), from cell culture, tissue biopsies, etc.) should be considered carefully in regard of the respective downstream application. Besides from biofilms, environmental microbial samples are often obtained from seawater or sediments and clinical samples from feces or sputum. Due to the diversity, many different protocols for preparing single-cell suspensions from microbial samples exist, including combinations of vortexing, centrifugation, filtering, sonication, and grinding with glass beads (Rinke et al. 2014). Adjusting the suspension to exact cell concentrations can be difficult since counting methods are less reliable for such small cells. In view of these facts, the sample processing has to be considered as integral part of the isolation technology. Any technology can only work in a certain range, and in case of failures or unsatisfactory results, it has to be carefully analyzed whether issues are caused by a specific isolation technology or rather by an inappropriate sample preparation. In the following, it should be assumed that for each technology under consideration, the sample preparation was according to the requirements without going into further detail on this point.

Basic Considerations, Definitions, and Classifications for SingleCell Isolation In principle, the task of isolation of a single cell from a bulk population is quite simple: picking a cell from the population and using an appropriate technology to transfer the cell to an isolated location (a “compartment”) like the well of a microplate into a microdroplet or into a microfluidic cavity. A closer look however

240

J. Riba et al.

Box 1 Definitions of important quantitative features of single-cell isolation technologies

reveals that there are certain aspects to be considered that are independent of the technology used for the purpose but that are determining the “quality” of isolation in addition to the quantitative measures defined in Box 1. “Quality” in this context should not mean a rank between superior and inferior performance, but should rather refer to the “characteristic properties” of a specific isolation process (regardless by which technology the process is carried out). Apart from qualitative differences between the various technologies, of course there are also quantitative performance criteria that can be determined for any technology. For the most important quantitative measures, a definition is given in Box 1. Whether or not a specific quantitative performance criterion has to be met depends on the specific application under consideration. For example, high capture efficiency is crucial for rare cell applications in order not to lose any of the rare cells (e.g., circulating tumor cells) contained in the sample. For most applications also, the throughput is important in order to be able to process a large amount of cells in the given time. Very often the quantitative performance parameters cannot be maximized at the same time due to technical or conceptional reasons. Therefore, such quantitative performance parameters can be often decisive for the choice of a specific isolation technology to achieve, for example, a certain level of single-cell isolation efficiency, viability, throughput, etc. demanded by the application. Focusing on the essential isolation process for the moment and disregarding any quantitative performance parameters and technical means to carry out such process, it appears that single-cell isolation processes can be mainly qualitatively different with respect to: 1. The selection of the individual cell to be isolated 2. The control of the location where the isolated cell is transferred to The selection of a cell out of a suspension can be either targeting a subpopulation with specific properties or accepting any kind of cell, i.e., in fact not carrying out a selection process. We will refer to these two different situations as “targeted

10

Technologies for Automated Single Cell Isolation

241

Fig. 3 Four cell isolation categories can be distinguished considering whether the selection is random or targeted and whether the location of the cell can be determined during or after the isolation step or whether the isolation location remains undetermined. The box with the question mark indicates that any information gained during the selection step (if any) cannot be linked to the cell position after the isolation. Dashed lines indicate that there is a random process involved which cell respectively which isolation location is addressed or assigned, while solid lines signify a controlled or deterministic process of picking respectively transporting cells

selection” and “random selection” like also shown in Fig. 3. Random in this context means that the selection operation is not considering which particular cell to pick, neither according to a specific property nor according to a specific order of the cells. In contrast, targeted selection means that the cell to be isolated can be selected based on given criteria (i.e., according to a given “classification” of the cells). These criteria can be defined by the user of the process. Also, here, the order of the cells to be isolated is usually not controlled (except for some specific technologies referred to as “fully targeted technologies” and described later on). The second operation in the process of single-cell isolation is the “isolation” of the selected cell at a certain “isolation location” also referred to as “compartment” in the following. The isolation location can be either “determined” or “undetermined.” Determined in this context means that an individual cell selected for isolation is assigned to a specific isolation location during the isolation process or its location is determined after the isolation by some means (e.g., imaging the wells of a microplate), such that the individual cell can be unambiguously identified. This means that in case of a targeted selection process, the property measured on the cell for classification purpose can be assigned to the isolation location. That is, the connection between the measured cell property (e.g., fluorescent intensity) and the isolation location (e.g., well ID) where the cell can be found for further downstream analysis is not lost. In contrast to this, if the isolation location is undetermined, the link between the value measured for cell classification (in case of targeted selection) and

242

J. Riba et al.

the individual cell is broken. For any cell to be recovered for downstream analysis, the measured properties previously used for selection are not known and cannot be used for further evaluation in combination with any downstream single-cell analysis (e.g., NGS sequencing). Based on this differentiation regarding the specific quality of selection and isolation of the cells, in principle four categories of single-cell isolation processes can be distinguished, as illustrated in Fig. 3. In practice, of course it will be also of highest importance according to which measurable properties a classification and targeted selection of cells can take place at all. Different technologies usually feature different sensing technologies to achieve classification according to size, morphology, fluorescent label, impedance signature, or other properties that can be measured on the cell. It is worthwhile noting that such targeted selection of the cells to be isolated can take place prior to the isolation operation (i.e., only the cells of the targeted population are isolated) or after the isolation operation (i.e., all cells are isolated into compartments, but only those isolation locations populated by the targeted subpopulation are considered for further analysis). In both of these cases, the selection operation should be referred to as “targeted selection.” 1. Random Selection with Undetermined Isolation Location (RU) This class of single-cell isolation technologies is characterized by a selection process that involves no selection of the individual cells and assigns such cells randomly to locations (“compartments”) at which they are isolated. The most prominent technology of this type is random cell encapsulation into water-in-oil droplets (a detailed description of this technology is given in the next chapter). Like sketched in Fig. 3, any cell out of the initial suspension can reach any compartment (which could be a single well of microwell plate or a particular droplet in a water/oil emulsion) as well as some compartments can remain empty or might be populated with more than one cell. Another method of this type is a microfluidic chip with hydrodynamic traps that randomly captures single cells from the sample flushed through the chip. This approach provides a mechanism ensuring that a trap can be only occupied by one cell, resulting in higher singlecell efficiencies and lower multiplet rates as compared to the limiting dilution approach. Typical characteristics of technologies according to the RU type are that of course no control of the type of cell or even the selection of a particular cell from the bulk is possible and that also the exact isolation location of a cell is undetermined and remains unknown. 2. Random Selection with Determined Isolation Location (RD) There are not too many examples of single-cell isolation devices found in the literature that fall into this category. Probably, this is due to the fact that often the isolation to a determined location does not add significant value, especially, if no selection of a specific subpopulation is required but only the mere isolation of single cells. The most representative example for this category is the combination of limiting dilution (i.e., pipetting a diluted cell suspension into microwell plates) followed by direct imaging of the wells using an automated plate imager (Shaw

10

Technologies for Automated Single Cell Isolation

243

et al. 2018). This strategy allows for counting and potentially further analysis by means of imaging of the cells in each well after random isolation. Albeit the cells are selected randomly, their position will be determined, which is the main aspect of the RD type. 3. Targeted Selection with Undetermined Isolation Location (TU) This type of technology is characterized by the fact that one or more specific properties of the cells to be isolated can be measured in order to classify and select cells for the isolation process. Sometimes, in microfluidic systems, the process of classification of the cells takes place simultaneously to their isolation, like, for example, in size-selective hydrodynamic traps that only capture cells of a certain size and let all others pass. However, there have been also microfluidic technologies demonstrated, where the isolation takes place first (e.g., encapsulation of single cells into microdroplets of an emulsion) and the classification is done afterward in a separate sorting step, where the cells of interest are separated from the rest of the population by dielectrophoretic or hydrodynamic forces (Mazutis et al. 2013). All technologies of the TU type have in common that they cannot control the isolation location to which the classified cells are directed. It is characteristic for this type of technology that the isolation locations (cell-trapping sites or microdroplets in emulsion like most often used by microfluidic approaches as compartments for the single cells) are populated in a random order (Fig. 3). TU type technologies yield some improvement over technologies providing only random selection, as they enable to extract one (or sometimes even several different) subpopulation from a heterogeneous cell sample. 4. Targeted Selection with Determined Isolation Location (TD) A quite high level of control on the individual cells as well as the possibility to maintain the link between the quantitative values of the physical property used to classify the individual cell and its isolation location is maintained when the single cells are directed to the individual compartments in a controlled way. This feature is typical for the TD type technologies that usually adopt the approach to first determine a classification value by some kind of sensing technology and then release each cell to a predetermined compartment. Very often the cells are released into the wells of conventional microplates or arrayed on flat substrates; thus the location of each individual cell in the matrix enables unambiguous identification of each single cell. Therefore, the order of isolation (if of any importance) as well as the value of the classifier obtained for the specific cell (often of quite significant interest) can be recorded during the selection and isolation process. This data is then available for the further processing and analysis. The most prominent example in this category are FACS instruments with index sorting capability which provide correlation of flow cytometry parameters of sorted events with the well location on a microplate. Other technologies of TD type are single-cell dispensing or automated micromanipulators as discussed in the sections below. The main characteristic of the TD type technologies is that information derived during classification (e.g., size, fluorescent intensity, etc.) is naturally assigned to

244

J. Riba et al.

the individual compartments and can be used as valuable information during further processing and analysis. Depending on the application, this might be irrelevant (e.g., when sequencing of a subpopulation of single cells expressing a certain protein is to be carried out); in other applications such information might matter (e.g., when looking for cells with the highest antibody production in clonal cell line development). Notably, all technologies according to the four categories discussed so far and sketched in Fig. 3 rely essentially on a random supply of the bulk cell suspension. That is, none of such technologies is able to control the order of the provision of the cells to the selection or isolation process. Furthermore, the technologies introduced so far are usually not able to recover or rearrange the cells once they have been isolated and transferred into individual compartments. However, some technologies indeed allow for the assessment of the complete cell population (or at least a significant part thereof) by some sensor technology used for classification of the cells prior to the isolation process (e.g., optical microscopy). Provided there is some means to manipulate each cell within the population, also the order of single-cell isolation can be controlled in this case (e.g., the largest cell out of a population could be picked first). This is an exceptional additional feature compared to the characteristics of the TD type introduced above: once all cells of the population are classified, any single cell can be transferred to any isolation location in a given order. Therefore, the isolation can be considered to be deterministic to a maximum extent, hence to be “fully targeted.” In addition, very often the isolated cells can be rearranged after a first classification and isolation process; they might be exposed to chemical or physical stimuli, and they might be combined to controlled populations and cell assemblies (e.g., arranging two cells with a specific size or fluorescent label into a single compartment). The two most prominent technologies (see below for a detailed description) that allow to assess the whole cell population and enable the manipulation and analysis of any cell in the population are based on optoelectronic tweezers (Chiou et al. 2005) and nDEP cages (Manaresi et al. 2003). In the next section, the major technologies for automated single-cell isolation will be described in more detail. The categories introduced before will be used to classify the technologies, and finally their most prominent applications will be discussed.

Automated Single-Cell Isolation Technologies Limiting Dilution Today many laboratories and companies use handheld pipettes or pipetting robots to isolate individual cells through dilution of the cell suspension. Due to the statistical distribution of the cells in the suspension, the number of cells in a highly diluted sample can be as low as one single cell per aliquot, when the suspension is split into small volumes (aliquots). This process is termed limiting dilution (LD) and is in

10

Technologies for Automated Single Cell Isolation

245

particular considered with the onset of the generation of hybridoma cell clones for antibody production as reviewed in Goding (1980). Such seeding of cells in low concentration is indeed simple to carry out with standard pipetting tools, but it is not very efficient since the probability of achieving a single cell in an aliquot is of statistical nature. The probability to obtain a certain number of cells per aliquot (i.e., 0, 1, 2, etc.) is described by the Poisson distribution: P λ ðk Þ ¼

λk λ e , k!

where k = 0, 1, 2, 3, . . . is the resulting number of cells per well and λ the concentration of cells per aliquot in the bulk suspension. In order to achieve a sufficiently high probability for the appearance of single cells while at the same time minimizing the probability for multiple cells, the sample has to be strongly diluted. Figure 4 illustrates the Poisson-based fraction of wells that contain no cell, one cell, or multiple cells for a given concentration λ of cells per aliquot. If it is of importance which of the wells indeed contain single cells, this has to be verified after seeding the cells in a separate process (e.g., by microscopy) due to the statistical nature of the separation method. Despite its drawbacks, limiting dilution is a generic method that does not require detection of cells and can be therefore applied to all kinds of cells provided that they are in suspension. No special equipment other than standard microplates and manual pipettes is needed, and the protocol can be easily automated using conventional pipetting robots. Furthermore, it can be also applied to micro-engineered “nanowells” or other compartmentalization methods such as droplets in oil (see section below). Two recent publications present sophisticated workflows to generate clonally derived cell lines for therapeutic proteins based on limiting dilution and automated high-resolution imaging (Shaw et al. 2018; Zhou et al. 2018). Such studies attempt to increase a so-called probability of monoclonality score to assure that a clonal cell population truly stems from a single cell,

Fig. 4 In limiting dilution, the number of wells containing no cell, a single cell, or multiple cells is governed by Poisson statistics and depends on λ, the number of cells per aliquot (A). In order to keep the multiplets at an acceptable rate, a small λ needs to choosen (B)

246

J. Riba et al.

which is a regulatory prerequisite for cell lines used in commercial production of recombinant proteins.

Suitability for Microbial Cells LD can be easily applied to microbial cells since no detection of the individual cells is required for the isolation. However, due to the small and heterogeneous size, counting of microbial cells can be error-prone. Therefore, in practice it can be challenging to set up the right cell concentration for an optimum single-cell isolation efficiency. Furthermore, the capabilities to screen for single cells in microwells using automated imaging are limited. Therefore, traditional streaking (spreading a bacterial sample onto agar plates) is still the method of choice to derive clonal cultures from microbial samples.

Fluorescence-Activated Cell Sorting Fluorescence-activated cell sorting (FACS) is a technology for sorting cells from a heterogeneous cell suspension into dedicated collecting vessels according to their optical properties that are used for classification (Fig. 5). As a form of flow cytometry, a FACS instrument not only allows for the detection of light scatter and fluorescent signals from individual cells but also for their physical separation in space and time. The principle of a FACS is based on hydrodynamic focusing, where a cell suspension is forced through a small nozzle (typically 60–100 μm in diameter) by a sheath fluid forming a jet of liquid. The stream of cells is scanned by a laser beam, and the Fig. 5 Working principle of common FACS instruments. (Reprinted from Gross et al. 2015)

10

Technologies for Automated Single Cell Isolation

247

resulting scattered light and fluorescence signal from each individual cell is captured by photodetectors. A vibrator breaks the jet into free-flying droplets which are continuously generated with a frequency of 10–200 kHz (Picot et al. 2012). The droplets are electrically charged, typically, by a charge inducing ring electrode located close to the nozzle. The charged droplets – some of them containing a single cell – can then be deflected in flight by electrostatic forces provided by a pair of electrodes towards a target position. Thereby, a cell can be either directed into a waste collection vessel or into wells of a 96- or 384-well plate located on a robotic stage underneath the nozzle according to the recorded signals. Modern instruments can analyze up to 20,000 cells/sec. For single-cell cell applications, the instruments are typically operated at lower speed to optimize for single-cell efficiency and allow sorting into 96- or 384-well plates. Measurable characteristics of cells in flow cytometry that can be used for classification are size and granularity as detected by forward- and side-scattered light, respectively. In addition, fluorescent signals can be read out allowing to sort for different subpopulations based on fluorescently labeled antibodies, fluorescent protein expression, or other fluorescent labels (e.g., FISH). The choice of modern FACS instruments ranges from compact bench-top devices with 2–9 fluorescent channels to large machines with the ability to detect up to 18 colors simultaneously. Recently, the single-cell sorting capabilities and quantitative performance parameters of several systems from BD were systematically evaluated (Evans et al. 2015; Fieder et al. 2017). A single-cell deposition efficiency of up to 99.8% was demonstrated for fluorescently labeled CHO cells using an efficient doublet discrimination strategy in conjunction with appropriate settings of the flow cytometer and sample preparation. However, shear stress, laser radiation, and electric fields inherent to the FACS system could have a non-negligible impact on some cell types and their viability. Therefore, an environment as gentle as possible to the cells should be created before, during, and after the sorting process.

Suitability for Microbial Cells FACS systems are commonly used to isolate single cells from microbial cell suspensions, although, in general, this is regarded a more challenging task compared to sorting of mammalian cells. This is mainly due to the larger heterogeneity in size and morphology, the presence of cell-sized abiotic particles in the samples, nonspecific binding of fluorescent stains, and the difficulties in cell enumeration as reviewed in Müller and Nebe-Von-Caron (2010). However, by using nucleic acid dyes such as SYBR Green, a larger-scale study has recently demonstrated that archaeal and bacterial cells can be successfully isolated from environmental samples with FACS for single-cell genomic analysis (Rinke et al. 2013, 2014).

Single-Cell Dispensing (SCD) In contrast to continuous droplet generation in traditional FACS devices, single-cell dispensing (SCD) is based on drop-on-demand non-contact dispensing to produce single cells encapsulated in free-flying microdroplets. To dispense only droplets

248

J. Riba et al.

containing a single cell onto a target substrate (such as a microwell plate), the cells have to be detected and potentially classified prior or after droplet generation. Only, if a single cell that meets the required properties is present in the volume that is to be ejected in the subsequent dispensing event, a droplet containing the cell is ejected and deposited onto the target. Otherwise (e.g., if none or several cells are close to the orifice), the liquid volume is directed to a waste (several strategies exist, as explained below). This basic concept has been proposed and explored in different variations by various authors (Yusof et al. 2011; Yamaguchi et al. 2012; Tornay et al. 2007). The detection of cells and other objects inside the nozzle of such drop-on-demand dispensers is often based on microscopy and automated image processing to identify droplets that will contain the cell of interest. Such images can be stored for later analysis and can serve as proof that truly a single cell was dispensed into the target well. Compared to, e.g., transmitted light microscopy, the optical resolution of these images in current systems such as the single-cell printer presented by Gross et al. (2013) is still limited, mainly because a relatively large working distance and a large depth of focus are used to image the cells inside the dispenser. Nevertheless, the single-cell printer allows for a reliable single-cell isolation based on cell size and simple morphological features such as the roundness of the cell. The technology developed by the authors and others was further developed and is meanwhile commercialized by cytena GmbH, Germany. Recently, the company integrated a fluorescent imaging in addition to the bright-field cell imaging system. As the classification takes place before the isolation of the cells and the cells are printed in a controlled way into well plates (i.e., these technologies fall into the TD class), the optical and fluorescent images can be correlated with the data derived for each single cell by genetic analysis or the like, afterwards. Besides using imaging technologies for classification, other cell detection methods such as impedance spectroscopy have been also proposed and successfully tested with SCD technology (Schoendube et al. 2015; Tornay et al. 2007). In all of these approaches, the cells are supplied randomly to the nozzle, and there is no control in which order or timely sequence single cells or cell clusters approach the nozzle. Therefore, different methods have been proposed to ensure that droplets not meeting the sorting requirements, such as empty droplets, droplets with multiple cells, or cells that do not exhibit the required properties, are not delivered to the target substrate: as depicted in Fig. 6, droplets can be either removed after ejection by deflecting them in flight toward a waste container using vacuum suction (Gross et al. 2013), they might be dispensed into a waste reservoir located at a position different from the target (Yamaguchi et al. 2012; Yusof et al. 2011), or the volume to be ejected can be exchanged inside the dispensing device by flushing the ejection volume away from the nozzle by a crossflow configuration of micro channels (Schoendube et al. 2015). Based on such academic work, several young companies have developed instruments for single-cell isolation in the recent years, and SCD has evolved into a valuable alternative for precise isolation of single cells into well plates. Compared to FACS instruments, the throughput of current SCD systems is significantly lower, limiting the throughput to a few thousand single cells per hour. However, as the

10

Technologies for Automated Single Cell Isolation

249

Fig. 6 Different methods exist in single-cell dispensing for the removal of droplets that do not meet the selection criteria, such as empty droplets, droplets with multiple cells, or droplets with a cell that is not selected based on the measured properties. (a) Unwanted droplets can be removed in flight by vacuum suction (Gross et al. 2013). (b) Unwanted droplets can be dispensed into a waste container by either moving the dispenser or the substrate. (c) A cross-flow configuration can be used to generate a continuous sample flow, and only if a cell in the detection volume meets the selection criteria, a droplet is dispensed

comparably young technologies are still evolving, there is a continuously ongoing improvement in terms of throughput that can be expected to continue for the coming years. Despite lower throughput than FACS, the high viability of the cells and the image-based proof of monoclonality have led to a rapid adoption of these technologies. The availability of sterile single-use printing cartridges, provided by some vendors, that safely prevent contamination turns out to be of particular value for the purpose of clonal cell line development to meet the high hygienic and regulatory requirements of the pharmaceutical industry. In general, SCD can be considered to be gentler than FACS sorting. Although, high cell viabilities can be achieved with well-tuned FACS systems, the technical and personal effort required for this is significantly higher compared to SCD

250

J. Riba et al.

technologies. Whether this difference is related to the electric fields, the shear forces generated in the sheath flow intersection, or the higher-impact force when the droplets arrive in the well has not been studied in great detail, yet. Whether or not this has a considerable impact on the experimental outcome clearly depends on the robustness of the cell types to be isolated and whether a high cell viability is required at all. Often high viability of cells is only required for applications such as clonal cell line development or for single-cell analysis workflows, where the integrity of the analyte such as RNA before its isolation matters.

Suitability for Microbial Cells Recently, the authors presented an advanced single-cell printer utilizing an optical detection system with higher resolution and dispenser chips with a nozzle size of 20 μm that allows for detection of bacteria cells down to 1 μm in size in a label-free manner (Riba et al. 2016) (Fig. 7). This prototype is able to isolate individual E. coli cells with single-cell isolation efficiencies of 93%. It was further demonstrated that individual cells from various species can be directly deposited into microtiter plates and onto agar plates for subsequent clonal cultivation. Thus, such technologies also hold great promise to enable efficient cell isolation of microbial cells for subsequent cultivation of clonal strains or single-cell genomic analysis of potentially uncultivable species.

Microfluidic Single-Cell Isolation A plethora of microfluidic geometries has been proposed for cell separation, patterning, and isolation (for recent reviews, see Hümmer et al. 2015; Narayanamurthy

Fig. 7 Single-cell printer for isolation of bacterial cells. (a) A CAD drawing of the droplet selection and dispensing unit. (b) The images of the cell in the nozzle of the dispensing chip. (Reprinted from Riba et al. 2016)

10

Technologies for Automated Single Cell Isolation

251

et al. 2017; Prakadan et al. 2017). Most of the proposed devices are fabricated by soft lithography – casting of an elastomer (polydimethylsiloxane, PDMS) – which allows rapid fabrication of microfluidic channels with minimal equipment cost, suitable for biological applications (Duffy et al. 1998). For commercial purposes most companies strive to find alternative materials and production methods, as PDMS is a rather expensive material and soft lithography is difficult to scale up (Tsao 2016). Nevertheless, microfluidic technologies are intensively used in research, and the focus of this section is to introduce a selection of established microfluidic concepts (both from industry and academia) that allow for automated cell isolation from bulk samples. An overview of the most common microfluidic principles is given in Fig. 8.

Hydrodynamic Trapping Individual cells can behy drodynamically trapped by flushing a cell suspension through a microfluidic chip with an array of fairly simple U-shaped structures as depicted in Fig. 8a. Despite their simplicity, such geometries tend to have a strong (and sometimes favored) size bias (Carlo et al. 2006). In cases where they work only reliably for a narrow range of cell sizes and shapes, they readily provide a classification, respectively, selection of cells according to size, and therefore have to be considered as TU type technologies. Tan and Takeuchi (2007) came up with the idea to implement bypass channels next to a constriction which acts as cell trap as depicted in Fig. 8b. These traps are designed such that the flow path through the trap has a lower fluidic resistance than the bypass channel. Once the trap is occupied by a cell, its resistance increases, and the majority of the flow – including other cells – is directed through the bypass channel. The traps have been optimized and combined with pneumatic valving (Fig. 8d) by Unger (2000) to develop highly integrated and controllable microfluidic chips for automated single-cell isolation and cultivation or molecular analysis. This technology termed microfluidic large-scale integration (LSI) (Thorsen et al. 2002) was commercialized and pushed forward by the company Fluidigm Inc., USA. Their integrated fluidic circuits (IFCs) such as the C1 system (Szulwach et al. 2015) demonstrate nicely that LSI allows for manufacturing (disposable) microfluidic chips that can handle reagents and cells in an automated manner. The main drawbacks of such chips are the relatively high cost for manufacturing by soft lithography, the fact that a specific hydrodynamic trap design covers only a limited range of cell sizes and that the downstream processing is fully integrated into the chip, leading to little flexibility as far as the analysis is concerned. Vertical Trapping in Nanowells In contrast to hydrodynamic trapping, individual cells can be seeded into nanolitersized micro-engineered cavities (nanowells) by gravity as depicted in Fig. 8c. In case the cells are much smaller than the nanowells, this is a stochastic isolation approach governed by the Poisson distribution corresponding to the “limiting dilution” approach but in a strongly miniaturized way. To avoid large numbers of multiple occupancies, cells are highly diluted resulting in many empty wells as governed by the Poisson distribution. Higher single-cell efficiencies can be achieved if the size of

Fig. 8 Microfluidic concepts for automated single-cell isolation. (a) Lateral hydrodynamic trapping can be achieved either by simple U-shaped structures or more sophisticated geometries with microfluidic bypass channels. (b) Vertical trapping in nanowells driven by gravitation. (c) Such structures have been used to isolate cells and capture beads decorated with barcoded primers to capture eukaryotic mRNA. A semi-permeable lid enables the exchange of buffers and small molecules. (d) Pneumatic valving can be used to open and close reaction chambers on demand. This concept has been combined successfully with hydrodynamic cell trapping. (e) Cell encapsulation into water-in-oil droplets is widely adapted for high-throughput single-cell analysis

252 J. Riba et al.

10

Technologies for Automated Single Cell Isolation

253

the nanowells matches the size of the cells, such that they can be only occupied by one cell. In practice this is difficult as this limits the technology to very homogeneous cell populations. Alternatively, this concept can also be used to deposit capture beads carrying barcoded oligo(dT) primers into nanowells prior to cell loading which was used to perform single-cell mRNA capture on a very simple chip comprising 86,000 nanowells (Gierahn et al. 2017). After bead and cell seeding, the authors covered the wells with a semi-porous membrane with 10 nm pore size which allows for delivery of reagents for cell lysis while keeping macromolecules such as mRNA in place. This concept works very well for the preparation of barcoded single-cell cDNA libraries but cannot be easily adapted to other assays which often require more complex liquid-handling steps such as DNA purification, amplification, or fragmentation. Suitability for Microbial Cells Gravity-driven vertical trapping tends to be more difficult the smaller the cells are, due to less dominant gravitational forces, and many bacterial species have flagella and are motile in solution. Park et al. (2011) found that nanowells with 8 μm diameter allow for trapping of yeast cells (S. cerevisiae) which are 3–8 μm in size with 76%), whereas A549 cells has a slightly lower single cell capture efficiency of 61.6%. However, when A549 cells were tested with capture wells of different depth (30 μm in diameter), the single cell capture efficiency was significantly increased. These results show that although a standard capture dimension could be used to obtain good single cell capture results for a range of different cell types, to obtain the best result for any cell type, it is important to choose an optimal capture well dimension for the target cells.

The Effect of Device Flipping on Cell Transfer Efficiency Single cell clonal culture can be performed in the culture wells of the dual-well device. Therefore, it is essential to ensure that individual cells trapped in the capture wells can be transferred to the culture wells after flipping the device. The microchannel height between the sets of capture and culture microwells is about ~200 μm. The center of each capture well should be aligned to the center of its corresponding culture well to form the paired wells array, as shown in Fig. 8. Figure 10a shows that in both 26 μm- and 30 μm-deep capture well devices, the percentage of cell loss is lower than 2% during cell transferring. The results demonstrate that cells could be transferred from the small well to larger well by the flipping device procedure.

Single Cell-Derived Clonal Colonies and Stem Cell Differentiation in the Microwells of the Dual-Well Device For clonal cell culture, it is common to culture single cells for more than 1 week to obtain enough number of cells for various applications. The PDMS device offers good optical observation compatibility due to their transparency property, which allows straightforward analysis of cells under general microscopy during cell culture experiments (Fig. 12a–d). To demonstrate the applicability of the dual-well device for single cell clonal culture, both KT98 cells and A549 single cells have been captured and cultured for up to 1 week for stem cell differentiation and single cell-derived colony

11

Dual-Well Microfluidic Technique for Single Cell Isolation and Long-Term. . .

283

Fig. 11 Demonstration of cell proliferation and cell differentiation in culture wells. (a–d) A single KT98 (neural stem cell) and a single A549 (lung cancer cell) cell divided and formed colonies after 6–7 days. (e–g) The single KT98 cell proliferated into a six-cell colony and showed the neural morphology after differentiation induction medium treatment. (h–k) The pictures show a single KT98 cell formed a colony during its differentiation processes. The KT98 cell expressed MAP2 of neuronal lineage marker after differentiation induction medium treatment. (The data is adapted from Lin et al. (2015), with permission of Royal Society of Chemistry)

formation, respectively. The culture wells are 285 μm in diameter, which provides enough growth area for cell division for long-term culture. As shown in Fig. 11d, g, a single KT98 cell can divide up to four to six cells (i.e., a single cell-derived colony) on the device after 6–7 days. Besides, the device could also be used to perform colony differentiation of neural stem cells toward neuronal lineage as indicated by the expression of microtubule-associated protein 2 (MAP2) which is a neuronal lineage protein marker involved in the microtubule assembly essential for neurogenesis (Fig. 11j). On the other hand, the heterogeneity of A549 cells during colony formation can also be shown by culturing single A549 cells with and without EGF, which is an epidermal growth factor receptor-mediated signaling for cancer cells. The result shows only 40–55% of captured single cells survived, and they showed various growth rates after 7 days of culture. Figure 12f shows that EGF-containing medium resulted in higher colony-forming efficiency (17.56%) compared to that of nonEGF-containing medium cultured cells (12.10%). The data indicated that the dualwell technique is not only a useful tool for clonal cell culture but also applicable to studying cellular heterogeneity at the single cell level.

284

C.-F. Yeh et al.

Fig. 12 Formation of A549 single cell-derived colonies is enhanced by epithelial growth factor (EGF). (a) One A549 cell divided to form a colony under non-EGF-containing medium culture condition after 7 days. (b) An A549 cell did not divide but survived under non-EGFcontaining medium culture condition after 7 days. (c) An A549 cell divided to form a colony under EGF-containing medium culture condition after 7 days. (d) An undivided cell survived under EGF-containing medium culture condition after 7 days. (e) Both the EGF medium and non-EGF medium cultured single A549 cells exhibit heterogeneous cell dividing rates after 7 days of cell culture. (f) The effect of EGF on promoting cell proliferation is shown form population-based analysis. (The data is adapted from Lin et al. (2015), with permission of Royal Society of Chemistry)

Conclusion This chapter describes the overall background of single cell clonal culture in both applications and approaches. Today, monoclonal culture-related techniques are critical for fundamental studies and commercial products. Despite the existence of various single cell clonal culture techniques, all of the methods have their advantages and limitations depending on the intended use and the users’ available resources.

11

Dual-Well Microfluidic Technique for Single Cell Isolation and Long-Term. . .

285

Overall, simple, low cost and being able to validate single cell event is favorable for most research laboratory users. For this, the dual-well technique represents an attractive alternative method to current existing methods. However, the throughput of the dual-well technique may not be as high as that of the other automatic instruments. This limitation may be overcome by integrating the dual-well chip with other automated systems.

References Bonnefoix T, Bonnefoix P, Verdiel P, Sotto JJ (1996) Fitting limiting dilution experiments with generalized linear models results in a test of the single-hit Poisson assumption. J Immunol Methods 194(2):113–119 Chartrain M, Chu L (2008) Development and production of commercial therapeutic monoclonal antibodies in mammalian cell expression systems: an overview of the current upstream technologies. Curr Pharm Biotechnol 9(6):447–467 Chen HY, Sun J, Wolvetang E, Cooper-White J (2015) High-throughput, deterministic single cell trapping and long-term clonal cell culture in microfluidic devices. Lab Chip 15(4):1072–1083 Chen YC, Zhang ZX, Fouladdel S, Deol Y, Ingram PN, McDermott SP, Azizi E, Wicha MS, Yoon E (2016a) Single cell dual adherent-suspension co-culture micro-environment for studying tumorstromal interactions with functionally selected cancer stem-like cells. Lab Chip 16(15):2935– 2945 Chen YC, Ingram PN, Fouladdel S, McDermott SP, Azizi E, Wicha MS, Yoon E (2016b) Highthroughput single-cell derived sphere formation for Cancer stem-like cell identification and analysis. Sci Rep 6:27301 Cheng YH, Chen YC, Brien R, Yoon E (2016) Scaling and automation of a high-throughput singlecell-derived tumor sphere assay chip. Lab Chip 16(19):3708–3717 Chung J, Ingram PN, Bersano-Begey T, Yoon E (2014) Traceable cional culture and chemodrug assay of heterogeneous prostate carcinoma PC3 cells in microfluidic single cell array chips. Biomicrofluidics 8(6):064103 Cox-Muranami WA, Nelson EL, Li GP, Bachman M (2016) Large area magnetic micropallet arrays for cell colony sorting. Lab Chip 16(1):172–181 dos Santos ML, Quintilio W, Manieri TM, Tsuruta LR, Moro AM (2018) Advances and challenges in therapeutic monoclonal antibodies drug development. Brazi J Pharm Sci 54. https://doi.org/ 10.1590/s2175-97902018000001007 Gross A, Schoendube J, Zimmermann S, Steeb M, Zengerle R, Koltay P (2015) Technologies for single-cell isolation. Int J Mol Sci 16(8):16897–16919 Kokkaliaris KD, Drew E, Endele M, Loeffler D, Hoppe PS, Hilsenbeck O, Schauberger B, Hinzen C, Skylaki S, Theodorou M, Kieslinger M, Lemischka I, Moore K, Schroeder T (2016) Identification of factors promoting ex vivo maintenance of mouse hematopoietic stem cells by long-term single-cell quantification. Blood 128(9):1181–1192 Lin CH, Hsiao YH, Chang HC, Yeh CF, He CK, Salm EM, Chen CC, Chiu IM, Hsu CH (2015) A microfluidic dual-well device for high-throughput single-cell capture and culture. Lab Chip 15(14):2928–2938 Lin CH, Chang HC, Hsu CH (2016) A microfluidic platform for high-throughput single-cell isolation and culture. J Vis Exp (112). https://doi.org/10.3791/54105 Longo PA, Kavran JM, Kim MS, Leahy DJ (2014) Single cell cloning of a stable mammalian cell line. Methods Enzymol 536:165–172 Matsumura T, Tatsumi K, Noda Y, Nakanishi N, Okonogi A, Hirano K, Li L, Osumi T, Tada T, Kotera H (2014) Single-cell cloning and expansion of human induced pluripotent stem cells by a microfluidic culture device. Biochem Biophys Res Commun 453(1):131–137 McFarland DC (2000) Preparation of pure cell cultures by cloning. Methods Cell Sci 22(1):63–66

286

C.-F. Yeh et al.

Noh SM, Shin S, Lee GM (2018) Comprehensive characterization of glutamine synthetasemediated selection for the establishment of recombinant CHO cells producing monoclonal antibodies. Sci Rep 8. https://doi.org/10.1038/s41598-018-23720-9 Pai JH, Xu W, Sims CE, Allbritton NL (2010) Microtable arrays for culture and isolation of cell colonies. Anal Bioanal Chem 398(6):2595–2604 Rettig JR, Folch A (2005) Large-scale single-cell trapping and imaging using microwell arrays. Anal Chem 77(17):5628–5634 Revzin A, Tompkins RG, Toner M (2003) Surface engineering with poly(ethylene glycol) photolithography to create high-density cell arrays on glass. Langmuir 19(23):9855–9862 Reynolds BA, Weiss S (1996) Clonal and population analyses demonstrate that an EGF-responsive mammalian embryonic CNS precursor is a stem cell. Dev Biol 175(1):1–13 Shen FM, Zhu L, Ye H, Yang YJ, Pang DW, Zhang ZL (2015) A high throughput micro-chamber Array device for single cell clonal cultivation and tumor heterogeneity analysis. Sci Rep 5:11937 Sikorski DJ, Caron NJ, VanInsberghe M, Zahn H, Eaves CJ, Piret JM, Hansen CL (2015) Clonal analysis of individual human embryonic stem cell differentiation patterns in microfluidic cultures. Biotechnol J 10(10):1546–1554 Sun CH, Fan Y, Li J, Wang GC, Zhang HS, Xi JJ (2015) Efficient production of a gene mutant cell line through integrating TALENs and high-throughput cell cloning. J Lab Autom 20(1):46–50 Tsumoto K, Isozaki Y, Yagami H, Tomita M (2019) Future perspectives of therapeutic monoclonal antibodies. Immunotherapy 11(2):119–127 Zhang ZX, Chen YC, Cheng YH, Luan Y, Yoon E (2016) Microfluidics 3D gel-island chip for single cell isolation and lineage-dependent drug responses study. Lab Chip 16(13):2504–2512 Zheng CH, Chen GE, Pang YH, Huang YY (2012) An integrated microfluidic device for long-term culture of isolated single mammalian cells. Sci China Chem 55(4):502–507

Single-Cell Cultivation Utilizing Microfluidic Systems

12

Dian Anggraini, Nobutoshi Ota, Yigang Shen, Yo Tanaka, Yoichiroh Hosokawa, Ming Li, and Yaxiaer Yalikun

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Purpose of Single-Cell Cultivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Morphology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Proliferation and Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Genomics, Transcriptomics, Proteomics, and Metabolomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Epigenomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Cultivation in Microfluidic Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell Coculture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecule-Induced Cellular Behaviors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regeneration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Further Limitations and Future Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

288 289 290 290 291 291 292 292 292 297 300 306 307 308

D. Anggraini · Y. Hosokawa Division of Materials Science, Nara Institute of Science and Technology, Nara, Japan e-mail: [email protected]; [email protected] N. Ota · Y. Shen · Y. Tanaka Center for Biosystems Dynamics Research (BDR), RIKEN, Osaka, Japan e-mail: [email protected]; [email protected] M. Li (*) School of Engineering, Macquarie University, Sydney, Australia e-mail: [email protected] Y. Yalikun (*) Division of Materials Science, Nara Institute of Science and Technology, Nara, Japan Center for Biosystems Dynamics Research (BDR), RIKEN, Osaka, Japan e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_20

287

288

D. Anggraini et al.

Abstract

Single-cell analysis is essential to deepen our understanding of cellular and subcellular cells at a single-cell level. Single-cell mechanism can be immediately observed during single-cell cultivation. Moreover, single-cell cultivation also provides sufficient cell numbers and product amounts for further single-cell manipulation and analysis. Microfluidic device is a raising system that offers efficient and sensitive single-cell processing and real-time on-chip and off-chip analysis. Single-cell cultivation microfluidics has been developed for understanding numerous biological applications. Here, we introduce the importance of single-cell cultivation from the aspect of cellular morphology to omics study. Then, we discuss numerous biological applications utilizing single-cell cultivation microfluidics such as cell coculture, molecule-induced cellular behaviors, and cell regeneration. Finally, present limitations and future prospects of singlecell cultivation microfluidics are also discussed.

Introduction Cells are known to have heterogeneity even in the same group of cells, in neighboring cells, and/or in identical environment. These cells may express different characteristics and various biological molecules, which are relevant to mechanisms of cellular development, differentiation, and evolution of pathogenic state (Inada et al. 2016; Gu et al. 2010). Single-cell analysis is highly important because it supports evaluation and utilization of single cells in various fields such as cell biology, tissue engineering, medicinal science, and medical diagnosis. Although rare cells like stem cells often play key roles in both normal and diseased tissues, important molecular information on these cells is averaged out in bulk analysis for a group of cells, which masks detailed mechanisms of cellular activities. To overcome this challenge, single-cell analysis has been performed to elucidate individual cell information from the aspects of cellular behavior analysis to omics analysis. In omics technologies, by analyzing transcriptome such as mRNA with a nextgeneration sequencer, single-cells are characterized to figure out the distribution of cellular population based on transcriptome. Protein and metabolite analyses have also been employed in single-cell level to investigate the downstream of transcriptome in cellular mechanisms (Aardema and MacGregor 2002). To obtain reliable and rich information in these single-cell analyses, single-cell cultivation is required. Single-cell cultivation is an essential process to provide sufficient number of single cells of interest for analysis and utilization. In addition, single-cell manipulation becomes easier through single-cell cultivation. Before starting single-cell cultivation, single-cell isolation is required to obtain single cells of interest from a group of cells upon characterizing cells by bulky or single-cell level measurement. There are various methods for obtaining cells of interest from a large population of cells. Manual pipetting is a widely accepted method to obtain single cells without expensive equipment. However, it requires intensive work that limits throughput of single-cell isolation. In addition, transferring single cells by pipetting increases the

12

Single-Cell Cultivation Utilizing Microfluidic Systems

289

risk of contamination or damaging single cells, especially when pipetting is performed by persons without enough training. Once single cells of interest are isolated, these cells are cultivated in a desired environment. During cultivation, single cells are tracked continuously or periodically to establish their identity for their future use and measurements. Among isolated cell types, motile cells are more challenging to keep tracking their identity because they move to change their positions. If their identity is lost during cultivation, important properties of single cells cannot be recognized. To conquer these obstacles, the use of advanced systems such as the microfluidic system must be carried out to guarantee the validity and accurate assessment during single-cell cultivation. In recent decades, microfluidic system has been widely used to culture cells at the single-cell level which provides extensive manipulation and analysis. Microfluidic system2 has a number of advantages over other conventional methods. First, microfluidic devices can be fabricated in a diverse system that suits with the type and size of cells. Second, the miniaturized compartments within the devices solely require small sample size and reagent volumes, resulting in high sensitivity of the analysis (Zhang et al. 2012; Dura et al. 2016). Third, multiple analytes can be introduced into the devices, allowing the simultaneous analysis within one device. Fourth, various controllers can also be connected to the devices, minimizing the labor-intensive and contamination during the experimental process. The controllers can be either a mechanical method, electrical method, and micropattern approach (Li et al. 2017a; Robertson et al. 2014). Fifth, devices integrated with real-time measurement assay can improve the high reliability, precision, and accuracy of evaluation. The measurements are live-cell imaging, calcium imaging, and electrophysiological analysis (Robertson et al. 2014; Poli et al. 2018; Dinh et al. 2013). Further, postexperimental analysis for a single cell can also be elucidated, such as immunohistochemistry analysis, reverse transcriptional polymerase chain reaction (RT-PCR) analysis, ELISA analysis, etc. (Dinh et al. 2013; Karakas et al. 2017). There are numerous review articles about methods and applications of microfluidic devices for single-cell cultivation, manipulation, and analysis. In this chapter, we mainly focus on the utilization of microfluidic devices for single-cell cultivation in various biological applications. We discuss the purpose of single-cell cultivation in cellular and subcellular studies, including morphology, proliferation, differentiation, migration, genomics, transcriptomics, proteomics, metabolomics, and epigenomics. We also discuss the using of microfluidic systems in single-cell cultivation for observing diverse biological applications, i.e., cell coculture, molecule-induced cellular behavior, and cellular regeneration. Finally, we also discuss the present limitations and future prospects of microfluidic systems in single-cell cultivation studies.

Purpose of Single-Cell Cultivation Unlike conventional cell culture methods that allow cell population analysis, singlecell cultivation enables the periodical observation of individual cells and allows the detection of cell offspring sequentially at a single-cell level. Single-cell culturing is

290

D. Anggraini et al.

highly needed to evaluate the process in cellular and subcellular studies, such as morphology, proliferation, differentiation, migration, genomics, transcriptomics, proteomics, metabolomics, and epigenomics.

Morphology Cell morphology can be influenced by some biomolecules including proteins. For example, actin is a fundamental molecule to work as a cytoskeleton that determines the mechanical, migration, and intracellular transportation properties of a cell. As the appearance of a cell can be recognized by microscopy, current technology allows its observation not only by visible microscope, but also other microscopic techniques, such as fluorescence microscope with chemical tags. Analysis of these techniques enables subcellular localization of specific chemical species, temporal change in chemical quantity, and intracellular transportation of chemicals via vesicles and cytoskeleton (Valentijn et al. 2003; Pécot et al. 2018). Since the morphology of a cell might change upon interaction with other cells and substances, the basic properties of a cell are desired to measure during single-cell cultivation, which provides automated, periodic, and precise cell observation at a single-cell level.

Proliferation and Differentiation During development and adulthood, cells proliferate into identical cells via mitosis and also differentiate into other types of cells. The differentiated cells such as cardiac muscle cells are retained throughout the life of the organism, whereas line blood vessels and epithelial cells resume proliferate in unfavorable conditions (Ahuja et al. 2007; Naito et al. 2020). Other types of cells called stem cells have both capabilities, i.e., self-renewal and producing mature cells of various functions. For example, all the types of hematopoietic cells, i.e., lymphocyte, erythrocyte, platelets, macrophages, and granulocytes, are differentiated from hematopoietic stem cells in the bone marrow. The hematopoietic stem cells proliferate into other types of blood progenitor cells such as myeloid and lymphoid that differentiate into various types of cells and give rise to all the types of blood cells. In the terminal stage, the fully differentiated cells maintain the blood cell populations in our body, and the capability of cells to proliferate decreases (Lim et al. 2013). Cell proliferation and differentiation are induced by the multiple parameters derived from neighboring cells such as scaffolds, stiffness of adjacent cells, and excreted molecules. For example, signal upon cell death or secreted from injured cells (e.g., liver cell death during the chronic liver injury) induces the proliferation and differentiation of neighboring cells to renew and replace the damaged cells (Luedde et al. 2014). Single-cell cultivation allows the investigation of proliferation and differentiation of single cells in a controlled environment. Also, single-cell cultivation can multiply a single cell with desired characteristics. Under optimized

12

Single-Cell Cultivation Utilizing Microfluidic Systems

291

conditions, the lineage of the desired single cell can be established to increase their number to obtain clones and other single-type cells.

Migration Cell migration is highly involved to cell morphogenesis and pathological states, and it has influence on tissue formation, tissue regeneration/renewal, and wound healing. During migration, cells communicate to neighboring cells and tissues, which allow cell movement in harmonized manner to control direction and speed of migration. An example of cell migration is the migration of cerebellar granule cells along radial glia fibers. In this migration, neuronal cells move from the external granule layer to the internal granule cells. This migration is estimated to correspond to roughly 60% of all neurons in a mammalian brain (Fritzsch et al. 2019). There are three major factors involved to cell migration: chemical gradient in solution (chemotaxis), the gradient of chemoattachment (haptotaxis), and mechanical cues (mechanotaxis) such as tissue stiffness. These factors are found highly complex in in vivo condition. Hence, single cell in controlled environment contributes to accurately evaluating individual factors on cell migration. In addition, some cells migrate throughout vascular system of animal body, for example, circulating tumor cells. These cells are important to work for biomarker investigation and therapeutic targets (Yang et al. 2019). To characterize these cells, single-cell cultivation is required for their continuous tracking for periodical observations and multiplication.

Genomics, Transcriptomics, Proteomics, and Metabolomics Cellular chemicals are fundamental to characterize single cells in terms of chemical species contained in an individual cell, quantities of metabolites, and their excreted chemicals. Metabolites are generally small molecules including building blocks of cells and tissues such as saccharides, fatty acids, and amino acids, as well as signaling molecules such as steroids. Since large portion of metabolites have been studied extensively, change in quantity of these molecules works as indicators of pathogenic states and drug screening. Large biomolecules such as DNA and RNA in a cell typically work as genetic materials, and their sequences directly characterize single cells (Karakas et al. 2017). Sequence of these molecules can be amplified by PCR and RT-PCR so that genetic information of a single-cell can be measured, while this information is averaged out if multiple cells are used for PCR. Proteins, the end products of genetic coding, are also important because they participate in regulation of metabolites. Proteins also have variation by epigenetic modification. Since proteins are not easy to amplify their copies, proteins in a single cell usually require a sensitive method of detection with small background noise. A simple method for measuring chemical contents of a single cell is loading the whole single cell in a small container such as glass capillary to directly place it to a mass spectrometer.

292

D. Anggraini et al.

Based on single-cell cultivation, the cultivated cell is expected to show less perturbation of metabolites due to stimuli during single-cell isolation.

Epigenomics Epigenetic modifications are also the targets of single-cell investigation. These modifications involve DNA, RNA, and proteins. DNA can be modified by adding methyl group, called methylation, and histones of chromatin can also be modified by methylation, phosphorylation, acetylation, and ubiquitylation. These modifications are important to manipulate cells, for example, producing pluripotent stem cells artificially (Watanabe et al. 2013). Small interfering RNA (siRNA) also affects genetic encoding by silencing translation, which results in change in protein expression and cellular metabolism. Single-cell cultivation allows to observe these modifications and is helpful to manipulate them through culturing with appropriate molecules.

Single-Cell Cultivation in Microfluidic Devices Single-cell cultivation provides opportunities to measure cellular communication and its influence on individual cells at a single-cell level or even subcellular level. Culturing individual cells in microfluidic platforms typically starts from cell isolation in a designed microchannel so that the cells are confined in a designated space. The individual cell is placed in a designated spot without touching to neighboring cells. The placed cell is allowed to communicate with an adjacent cell or substances through a specific channel. During the single-cell cultivation, cells communicate with their neighbors under controlled condition. Due to complicated nature of cell communication, single-cell cultivation in microfluidic device simplifies the setup of measurement and also enables to investigate influence of various cellular factors. The measurement and investigation of biological phenomena using microfluidic devices can be characterized into cell coculture, molecule-induced cellular behaviors, and cellular regeneration.

Cell Coculture Cell communication occurs through the binding of ligands secreted by the deliver cells with receptors on the membrane of target cells, triggering the physiological responses in the target cells, such as cell division, proliferation, migration, differentiation, and apoptosis. In brain cells, the presynaptic neurons, which release neurotransmitters, are binding to the receptor of the postsynaptic neuron membrane. It influences the cross of ions and generates action potentials in the postsynaptic neuron. The action potentials occur simultaneously among neurons as well as muscle cells and glands.

12

Single-Cell Cultivation Utilizing Microfluidic Systems

293

Communication between cells can be quantified by bulk cell cultivation, such as coculture assay. This assay is performed to elicit the synergistic and antagonistic interactions between cell populations. Two cell populations are placed on the upper compartment and below compartment that are separated by porous membrane. The excreted product metabolism is typically analyzed by protein analysis, as quantitative results of cell-cell interactions. However, this method elucidated the populationwide interactions; hence, the cause and effect phenomenon between two types of cells cannot be explained. Single-cell cultivation using microfluidic devices is potential for understanding the cause and effect phenomenon within the cell population and also detailing the single phenomenon from single-cell interaction. Microfluidic device has been used to observe the relationship between neurons, (Robertson et al. 2014; Dinh et al. 2013) interaction between tumor cells to neighboring cells, (Dura et al. 2016; Karakas et al. 2017) and coculture of neurons and cancer cells (Li et al. 2017a).

Neural Coculture Naturally, the individual shape of neurons, such as axons and dendrites, is determined by the extracellular matrix. Extracellular matrix has a role in anchoring neurons via various mechanisms, resulting in an appropriate form of neurons to allow proper functions. Microfluidic systems can be used not only to provide the method for introducing the extracellular matrix, but also to design the device that is suited to the natural structure of neurons. Using this approach, neural coculture is observed in the various compartments for placing the cell body of the neuron populations that are separated by microchannels for the outgrowth areas of neurites (axons and dendrites) to evaluate the synaptic activity modulation between neurons (Robertson et al. 2014; Poli et al. 2018; Dinh et al. 2013). In another study, chemical transport inside the device was based on a hydrostatic pressure gradient that is controlled by electrical circuit analog. The electrical circuit analog comprised a resistive network in the central microchannels connected to four capacitors inside the two pairs of inlet and outlet of compartments (Fig. 1a). The difference volume of inlet and outlet did not generate the discharge effects; hence, the solution flowed from the inlet to outlet of each compartment, providing cell seeding and medium change process (Fig. 1b). Once the volume in the first compartment in an equilibrium state, however, differs from the second compartment, the solution flowed from the high volume to small volume compartment; hence, further analysis can be performed (Fig. 1c). KCl stimulated-primary hippocampal neurons in the first compartment were significantly increased by the number of neuronal calcium events in the second compartment by 41%. Moreover, synapse formation between the coculture was showed by positive synaptophysin at 13 days in vitro (DIV) (Robertson et al. 2014). Further, fluidic resistance difference was applied in the microdevice for ensuring the single-neuron interaction. The fluidic circuit produced multiple streamlines in the site between microchannels and microchambers, providing precise single-cell arraying on the outer wall of microchannels (Fig. 1d and e). The attachment of cells was also improved by meniscus-pinning micropillar to align water mask for plasma stencilling a poly-amine coating (Fig. 1f). Neurite outgrowth was showed

294

D. Anggraini et al.

Fig. 1 Microfluidics-based coculture assay. (a) Electrical circuit analog applied in the central microchannels of microfluidic device for neuron interaction (Reproduced from Robertson et al. (2014). Copyright 2014, Oxford University Press). (b) Hydrostatic pressure differences within the compartment induce the solution flow from the inlet to the outlet of the compartment (Reproduced from Robertson et al. (2014). Copyright 2014, Oxford University Press). (c) Hydrostatic pressure differences between compartments induce the solution flow from the first compartment to the second compartment (Reproduced from Robertson et al. (2014). Copyright 2014, Oxford University Press). (d) A schematic of coculture assay with neuron array chamber and neurite outgrowth channels (Reproduced from Dinh et al. (2013). Copyright 2013, Royal Society of Chemistry). (e) The SH-SY5Y cells arraying on the outer wall of neurite outgrowth channels, confining by meniscus-pinning micropillar (Reproduced from Dinh et al. (2013). Copyright 2013, Royal Society of Chemistry). (f) Poly-lysine patterning by plasma stenciling with a water mask (Reproduced from Dinh et al. (2013). Copyright 2013, Royal Society of Chemistry). (g) A microfluidic device for evaluating interaction of tumor cells and immune cells that comprise mechanical filters and cell-trap structures (Reproduced from Dura et al. (2016). Copyright 2016, National Academy of Sciences of the United States of America (NAS)). (h) Cell-trap structures consisted of single-cell trap, two-cell trap, narrow constriction, and support pillars (Reproduced from Dura et al. (2016). Copyright 2016, National Academy of Sciences of the United States of America (NAS)). (i) A schematic of microfluidic device for observing interaction between single-tumor cell and fibroblast cells (Reproduced from Karakas et al. (2017). Copyright 2017, Springer Nature). (j) A schematic of the pneumatic microvalve controlled microfluidic device comprises a fluidic layer, control layer, PDMS membrane, and glass slide. (Reproduced from Li et al. (2017a). Copyright 2017, Elsevier)

12

Single-Cell Cultivation Utilizing Microfluidic Systems

295

across the microchannels evident by positive ß(III)-tubulin in SH-SY5Y neuroblastoma cell coculture for 13 DIV. Both the devices provided high-throughput neuron communication through calcium activity imaging and immunostaining analysis (Dinh et al. 2013).

Coculture of Tumor Cells to Neighboring Cells In the tumor microenvironment, the tumor cells interact with their surrounding cells, such as immune cells, fibroblast cells, adipocytes, etc., through paracrine signaling or systemic circulatory system. The understanding of the interaction between tumor cells and immune cells is crucial to measure the possibility of immune cell-based therapy for controlling the malignancy of tumor cells. Besides, the interaction between tumor cells and nonimmune cells also underlies the progressivity of tumor cells that induce tumorigenesis; thus, the medical diagnosis can be achieved. Microfluidic system arises for measuring the single phenomenon within single-cell interaction. The interaction between tumor cells and immune cells (natural killer cells; NK cells) was evaluated using a trap array microfluidic assay. The device consisted of mechanical filters which ensure the cell distribution during cell loading, and cell-trap structure, detailing the single-cell trap, two-cell trap, narrow constriction, and support pillars (Fig. 1g and h). Formerly, the NK cells were loaded and captured in a single-cell trap with a slow infusion rate. The cells were transferred and squeezed into the two-cell trap by higher flow rate. Then, tumor cells were also loaded with the same protocol. After the cell trapping was achieved, the fluid connections were removed, and the solution can be introduced using the manual pipette, providing natural communication between cells. This device enabled the NK cell-tumor cell pairing with 95% and 85% viability of NK cells during the culture period (6 h) and even after the culture (24 h). After the pairing of cells, the live-cell imaging showed 20% cytotoxicity and 65% interferon-gamma (IFN-γ) production of NK cells, which were similar to bulk assay. Moreover, the downstream assay can also be attained with the retrieval structure of the device. This approach was successfully resolving the relationship between tumor cells and immune cells at the defined generation of single cell (Dura et al. 2016). Another study was using PDMS microdevice to reveal the interaction between heterogeneous tumor cells that induce autophagy of stroma fibroblasts. The device comprised PDMS reservoir, PDMS membrane, and PDMS-coated cover glass. The GFP-LC3 transgenic mouse embryonic fibroblasts (MEFs) were cultured on the bottom side of the PDMS membrane connected with the PDMS reservoir for 2 h. The PDMS reservoir was turned over and attached to the PDMS-coated cover glass. The breast carcinoma cells (MDA) were cultured on the upper side of the PDMS membrane and trapped in the holes of the membrane (Fig. 1i). Autophagy activation between two types of cells was identified by GFP-LC3 dot positive cells that were triggered by MDA-produced transforming growth factor beta 1 (TGFβ1). FibroblastshTGFβ1 MDA coculture can induce the differentiation of normal fibroblasts into carcinoma-associated fibroblasts (CAFs) with 81% accuracy. Further, PCR analysis can be obtained by retrieving the MDA-activated fibroblasts. This device enabled the

296

D. Anggraini et al.

quantitative analysis of the interaction between heterogenous tumor cells to obtain specific and high-throughput screening of its paracrine factor to fibroblast cells (Karakas et al. 2017).

Coculture of Neurons and Cancer Cells Neuron-cancer cell interaction can also be observed in microdevice equipped with pneumatic microvalve. The microfluidic chip consisted of four layers, i.e., fluidic layer, control layer, PDMS membrane, and glass slide. The control layer comprised three pairs of valve inlet, valve channel, and microvalve, while the fluidic layer consisted of three chambers with each inlet, outlet, and channel (Fig. 1j). All chambers in the fluidic layer either can be combined or separated depending on the microvalve in the control layer. By adjusting the pressure within the control layer, the fluidic layer can be switched in block and unblock condition. When the pressure increased, the blocking area increased linearly, resulted in 0.006 MPa as the best blocking pressure during the experiment. In unblocking condition, the neuron (SH-SY5Y) was loaded into three chambers through the central inlet. The MPP+ solution was also introduced into three chambers, inducing the cell interference. Afterward, in blocking condition, different types of cells, such as cancer cells (Hela), colon carcinoma cells (U87), and neurons (SH-SY5Y) were loaded into each chamber through each side inlet. The number of Hela cells and U87 cells increased after the nerve damage as a consequence of neural protection. This evaluation can be elucidated simultaneously within the same condition. Therefore, this microdevice can be useful for both observing cell interaction parallelly and independently within one device (Li et al. 2017a). Extensive assessment of coculture analysis during single-cell cultivation can be achieved with the following conditions: (1) mantain the cells remain in contact, (2) maintain the cells in a desirable condition, and (3) real-time analysis during the cultivation. Designing the device that is suitable for the cells, in size and structure, is useful for improving the cell contact and guarantying the cell communication. For example, the cell trap structure provides the culture area that is suitable for tumor cells and immune cells, as well in neuron study, the size of central microchannels between compartments allowed the neurite outgrowth toward the microchannels (Dura et al. 2016; Robertson et al. 2014; Dinh et al. 2013). Undesirable conditions such as backflow of reagents and solution leakage between compartments can be overcome by the arrangement of hydrostatic pressure gradient and fluidic resistance differences, which control the one-way fluid flow and selective treatment of each compartment (Robertson et al. 2014; Dinh et al. 2013). Further, the real-time analysis is also required to evaluate the phenomena during the singlecell cultivation, i.e., calcium activity imaging and electrophysiological analysis (Robertson et al. 2014; Poli et al. 2018). The posttreatment evaluation such as immunostaining analysis is also needed to get extensive explanation about cell-cell communication (Dura et al. 2016; Robertson et al. 2014; Dinh et al. 2013; Mills et al. 2018).

12

Single-Cell Cultivation Utilizing Microfluidic Systems

297

Molecule-Induced Cellular Behaviors In our bodies, molecules are very important for both intercellular and intracellular communication. Molecules that act as signals are bound specifically with receptors within the cells or receptors on the cell membranes, forming biochemical reactions within the cells, called signal transduction pathway. These signal transduction pathways resulted in various physiological responses, i.e., cell growth, division, migration, survival, proliferation, differentiation, and apoptosis. Signaling molecules can be characterized into cytokines, growth factors, chemokines, hormones, enzymes, and neurotransmitters. A single signaling molecule can play a role in multiple signal transduction pathway that triggers various simultaneous physiological responses. For example, when Wnt5a protein binds with tyrosine kinase-like orphan receptor 2 (Ror2), cell proliferation is inhibited; when it interacts with two receptors on the cell surface (Frizzled and LRP), cell proliferation is activated. The discovery of these phenomena can be used in further studies on intercellular and intracellular communication (Mikels and Nusse 2006). Interaction between cells and molecules can be performed by culturing the cells in vitro using a transwell assay, also known as the Boyden chamber. The Boyden chamber consisted of a lower compartment and well plate with a porous membrane as an upper compartment. Chemoattractant molecule is placed in the lower compartment, while the cells are cultured in the upper compartment. The molecules diffuse through the porous membrane and affecting the migration of cells to the lower compartment of the chamber. By this approach, bulk quantitative analysis can be obtained by counting the migrating cells. However, some disadvantages arise from this method, such as limited observation of cell respond in real-time and in a single level, and also uncontrolled release gradient of the molecules. The obstacles from the previous method can be overcome with the microfluidic system. The microfluidic system has been applied to observe the interaction between molecules and cells in various fields of studies including neuroscience, immunology, and microbiology.

Molecular Cue-Guided Neuron Molecular cue-guided neuron has been extensively studied for understanding the growth and development of neurons, as well as understanding the sensitivity of neurons that will be useful for drug screening. These objectives can be evaluated by the gradient generator microfluidic device which provides the controlled gradient molecular cue release and single-cell observation. The gradient generator microfluidic device typically consists of several reservoirs (two peripheral channels and a central channel) that are connected by microcapillaries (Fig. 2a) (Bhattacharjee et al. 2010; Xiao et al. 2013; Xu and Heilshorn 2013). The cells are cultured in the central channels, while the molecules are established in the source peripheral channel. The molecules pass through the microcapillaries and generate the molecule gradients that will expose and affect the cells. In a previous study, CXCL12 was introduced to the source peripheral channel using a syringe

298

D. Anggraini et al.

Fig. 2 Microfluidic devices for molecule and cell interaction assay. (a) A schematic of a gradientgenerating device comprises three reservoirs (two manifolds and a cell culture reservoir) that are connected by microjet arrays. The cells are cultured in the cell culture reservoir, while the molecules are introduced in the source manifold. The molecule gradients are generated from source manifold passed through the microjet arrays to the cell culture reservoir (Reproduced from Bhattacharjee et al. (2010). Copyright 2010, Oxford University Press). (b) A schematic of the microfluidic migration platform consists of cell traps with two reservoirs at the opposite end that are connected by cell migrating channels. The chemoattractant molecule is placed in the chemoattractant reservoir (CR), whereas the buffer is introduced in the buffer reservoir (BF). Cells migrate toward or migrate away from the molecule that is shown in the cell migrating channels (Reproduced from Boneschansker et al. (2014). Copyright 2014, Springer Nature). (c) A schematic of automatic sorting droplet microfluidic contains chemotactic cell sorter (part I) and single-cell encapsulation system (part II). Part I contains three inlets for introducing the culture of microbes, chemoeffector solution, and buffer; the main channel, and an outlet for waste the nonchemotactic cells. Part II includes two inlets for introducing culture medium and shear flow of carrier oil (FC40), T-junction droplet generator, and Teflon tubing for collecting microbial droplets. (Reproduced from Dong et al. (2016). Copyright 2016, Springer Nature)

pump at a flow rate of 3 μm/h. The CXCL12 gradients were generated in the cell culture channels, resulting in a greater number of neural stem cells (NSCs) migrated toward the molecule gradients at 4 h and 17 h exposure (Xu and Heilshorn 2013). Furthermore, the Netrin-1 molecule gradients were induced manually by establishing the molecules to the source peripheral channels using a pipette. Before then, the microcapillaries were pressurized, leading to the solution being discharged at the microchannels. Therefore, the molecules flowed manually from the source peripheral channel passed through the microchannels and exposed the primary mice neurons without any cell interference. As a result,

12

Single-Cell Cultivation Utilizing Microfluidic Systems

299

the primary neurons showed a positive axon turning angle to the source of Netrin1 gradients with an average speed higher than the post-Netrin-1 period. These devices provide stable molecules gradients that are useful to observe the growth and development analysis and determine the sensitivity of cells in response to the various molecular gradients (Bhattacharjee et al. 2010). Using a similar design of the microfluidic device, failure neural regeneration due to overexpressed GDNF can be evaluated. The device comprised three chambers, i.e., somal chamber, middle chamber, and distal chamber, and two sets of microchannels that connected all the chambers. Sensory neurons were cultured in the somal chamber, while GDNF-overexpressing Schwann cells were cultured in the middle chamber, and GDNF was introduced in the middle and distal chamber. The medium of the cultures was maintained with passive fluid transport. Less neurite outgrowth of sensory neurons was observed when the GDNF-overexpressing Schwann cells were cultured in the middle chamber compared to normal Schwann cells. At 7 DIV, the neurites showed growth toward the middle and distal chamber more than 50% at GDNF concentration of 100 ng mL-1/ , 100 ng mL-1/100 ng mL-1, and 100 ng mL-1/700 ng mL-1 (middle chamber/ distal chamber). Moreover, in the GDNF concentration of 250 ng mL-1/ and 500 ng mL-1/ , the neurite showed outgrowth to the middle and distal chamber at 40–45%. However, if the GDNF concentration in the middle chamber were 700 ng mL-1, the number of neurite outgrowth decreased lower than 25%. Therefore, the threshold concentration of GDNF that induces the axon entrapment was between 500 ng mL-1 and 700 ng mL-1. This device allows for evaluating the threshold concentration of molecules toward the cells, which are applicable to answer the unsolved in vivo problem (Wang et al. 2018).

Molecule and Immune Cell Interaction Molecular cues can induce various cellular responses of cells including immune cells. These cellular responses are classified into migration toward the molecule (chemoattractant), migration in a random direction (chemokinesis), and migration away from the molecule (chemorepulsion). The single cellular response of single cell can be quantified and observed using a microfluidic device. The interaction between chemokines and human neutrophils can be performed using a microfluidic migration platform. The platform comprised a cell trap that connected with channels to each reservoir at the opposite end. Chemoattractant chemokine (fMet-Leu-Phe (fMLP)) was placed in the first reservoir, whereas buffer was introduced in the opposite reservoir. Human neutrophils were cultured in the cell trap (Fig. 2b). As much as 90% of neutrophils was attracted to fMLP with high maximum directional persistence. The speed of migration neutrophils to fMLP in 6 μm width of channels was faster than in 50 μm width. Investigation on other chemokines resulted in low persistence of attraction and repulsion respond to Ca5, whereas high persistence showed in the respond of neutrophils migrating to IL-8. This approach enables precise quantification of cells to various molecular cues that reduce the bias analysis by providing the object molecule reservoir and neutral molecule reservoir (Boneschansker et al. 2014).

300

D. Anggraini et al.

Molecule and Microbe Interaction Besides animal and human studies, microbiology study also has enormous impacts on human life. Knowledge of microbe and molecule interaction is essential to understand the bacterial-host interactions and pathological mechanisms of infectious diseases. Microfluidic droplet system was employed to isolate and cultivate the chemotactic bacteria by introducing aspartic acid gradients. The microfluidic device consisted of chemotactic cell sorter (part I) and single-cell encapsulation system (part II). Part I contained three inlets for introducing the microbes, aspartic acid (chemoeffector solution), and buffer (nonchemoeffector solution); a main channel; and an outlet for discarding the nonchemotactic microbes. Part II was composed of two inlets for introducing medium and shear flow carrier oil (FC-40), T-junction droplet generator, and Teflon tubing for collecting microbial droplets (Fig. 2c). The culture of Escherichia coli was exposed with aspartic acid gradients in the main channel. The chemotactic microbes will be attracted toward gradients to T-junction generator, while the nonchemotactic microbes will move away from the gradients and discard through the outlet. The sorted chemotactic microbes were fused with medium and generated into droplets by the shear flow of carrier oil (FC-40). Microfluidic-generated droplets were collected in the Teflon tubing and deposited into agar plate for scale-up cultivation. The droplets were produced in the amount of ~40 droplets/min, with the density of chemotactic microbes per drop increased from 0.043 to 0.77 after exposed with aspartic acid gradients. This device contributes to the high-throughput chemotactic sorting of microbes and single-cell cultivation of targeted microbes from the bulk culture (Dong et al. 2016). In observing the interaction between molecules and cells using a microfluidic system, we must consider the following conditions, including: 1) prevent cell interference, 2) ensure the distribution of molecules in the device, and 2) real-time analysis during the cultivation. Minimizing the shear stress by introducing the solution manually using a pipette is one thing that can prevent cell interference (Bhattacharjee et al. 2010). Further, it can also guarantee the nonbias analysis of cell response to chemoattractant or repellant molecules due to the fluid flow introduction (Boneschansker et al. 2014). Impressive microfluidic device fabrication for distributing and generating molecule gradients is also essential to ensure the simultaneous cell response toward a molecule (Bhattacharjee et al. 2010; Xu and Heilshorn 2013; Dong et al. 2016). Finally, real-time analysis can answer the missing mechanisms during the cultivation such as live-cell imaging following molecule induction (Bhattacharjee et al. 2010; Xu and Heilshorn 2013; Boneschansker et al. 2014). Eventually, deeper mechanisms can be obtained by immunohistochemistry analysis and RT-PCR analysis (Xu and Heilshorn 2013; Dong et al. 2016; Wang et al. 2018).

Regeneration Cellular regeneration is the ability of cells to restore and renew the microenvironment as a physiological response to maintain the cellular homeostasis. Cellular regeneration consists of migration, proliferation, and differentiation process.

12

Single-Cell Cultivation Utilizing Microfluidic Systems

301

Cellular regeneration can be a complete or incomplete process depending on the functionality of the new cells and tissues. For example, in the process of hematopoietic cell production, the hematopoietic stem cells regularly produce new erythrocytes in the bone marrow. The functional erythrocytes deliver oxygen throughout blood circulation during their lifespan (120 days). At the time the cells undergo aging, they are removed from the circulatory and eventually undergo programmed cell death in the spleen. The cellular regeneration by producing continual new functional cell populations is called a complete process of regeneration. An example of the incomplete process of regeneration is cellular regeneration following brain injury in mammals. As we know, the brain has a limited capacity for reparative regeneration, but it still undergoes regeneration attempts, i.e., migration, proliferation, and differentiation. Neural stem cells from two neurogenic regions (e.g., subventricular zone (SVZ) and subgranular zone (SGZ)) showed the regeneration process toward the injured brain. SVZ-derived neuroblasts migrate to the injured cortex (Jin et al. 2001, 2003; Goodus et al. 2015) through the rostral migratory stream (RMS) (Saha et al. 2013) and lateral cortical stream (LCS) (Jin et al. 2003). Furthermore, the cells also migrate to the injured striatum (Jin et al. 2003; Ohab et al. 2006) through RMS (Jin et al. 2003). Some of the migrated cells in the target area can differentiate into mature neurons, (Goodus et al. 2015) but some of them shrunken and die (Jin et al. 2003). In another research, SGZ-derived neuroblasts showed abnormal cellular behavior that is called aberrant migration following brain injury (Ibrahim et al. 2016; Shapiro 2017). These evidences showed the incomplete process of regeneration that is indicated by abnormal migration, limited number of migrated and proliferated cells, and not-functional new mature neurons. Regeneration assay can be started with cell cultivation, such as in wound healing assay. Wound healing assay is a simple method to observe the regeneration process, especially the migration of monolayer cells. The monolayer cells (e.g., fibroblast cells and keratinocytes) are seeded on extracellular matrix-coated dish. After the cells are confluent, manually scraping by a pipette tip is conducted to mimic the wound condition within the cell culture. Two main areas (e.g., lacking cell area and confluent cell area) are formed inside the cell culture. Cells from confluent cell area migrate to lacking cell area as a response to the regeneration process (Rodriguez et al. 2005). The measurement can be performed by immunohistochemistry analysis and time-lapse microscopy analysis. This method has advantages to measure cell behavior in a simple way, but this method also has limitations in the term of a not controlled environment and not automated handling. Moreover, cell behavior is measured at a population level by counting the cell number. Therefore, the regeneration response event at a single-cell level cannot be elucidated. One of the techniques that provide a controlled environment, automated handling, and single-cell analysis is a microfluidic system. Microfluidic system has been used for measuring the regeneration phenomena in single-cell cultivation. Most topics that have been discussed in the regeneration process using this system are axotomy and stem cellbased regeneration.

302

D. Anggraini et al.

Axotomy Axotomy is the process of severing the axon of the organism. This process is useful for understanding the biological mechanisms of nerve regeneration in nervous system diseases. This process can be studied in a model organism that has a simple organization of nervous systems such as nematode Caenorhabditis elegans by severing its axon and then observing the recovery process. Microfluidic system has been used to minimize the undesirable environment for obtaining high precision axotomy and monitoring the axon regeneration in single-organism. Multilayer PDMS microfluidic device separated by a membrane was used for conducting axotomy. The device integrated the trap system (e.g., valves and trap chamber) for axotomy process and time-lapse imaging, and three recovery chambers for observing the axon recovery process (Fig. 3a and b). The trap chamber was equipped by several pillar-like structures on both sides of the chamber, avoiding wash of the small-sized nematode and maintaining the position of large-sized nematode inside the chamber. The recovery chambers were also integrated with pillars on the outlets to prevent the nematode from flushing out of the chambers. The position of nematode in each chamber was controlled by adjusting the pressure of a solenoid three-way valve on the upper layer of air cavities of the microfluidic device. While the nematode was inside the trap chamber, the axon surgery was performed by femtosecond laser with 220 laser pulses of 7.2 nJ at 1 kHz repetition rate and 780-nm wavelength. After the axotomy process, the nematode was flushed to the recovery chambers that enabled the food transportation, thus appropriate recovery process of the nematode can be attained. To observe the regeneration process, the nematode was transported back to the trapping chamber. The observation was obtained using time-lapse fluorescence microscopy. Finally, the nematode either can be flashed out or transported into recovery chambers for further analysis. This device provided the high-throughput speed of axotomy process (1 min per worm), high viability of worm (100% survival), and high axonal recovery time (60–90 min). The regeneration time was considerably faster compared to the agar pad method because of the nonused anesthetic reagent (Guo et al. 2008). In another study, high-throughput speed and precise regeneration process observation of C. elegans with fully automated microfluidic device were demonstrated. The basic principle of this microdevice was similar with the previous study, i.e., multilayer microdevice with applied pressure for precise positioning of the nematode in the trapping area. The device comprised 1) loading chamber for temporary place of nematodes (up to 250 worms), 2) staging area for isolating the single-nematode from the population, 3) trapping area for placing the nematode in axotomy processing spot, 4) 3D interconnects for adjusting the sealing control of valves, 5) exit outlets for flushing out the nematode from the device (Fig. 3c). In this study, there was some specialized configuration of microfluidic device compared to the last study. The nematodes were loaded into the loading chamber from side channels with all the valves in closed conditions. The worms then entered the staging area with the opened V1, the closed V2, and flow was directed out to side channels. While

12

Single-Cell Cultivation Utilizing Microfluidic Systems

303

Fig. 3 Automated microfluidic devices for observing the regeneration process in Caenorhabditis elegans. (a) Microfluidic device comprised trap system (yellow rectangles) and three recovery chambers (blue chambers) (Reproduced from Guo et al. (2008). Copyright 2008, Springer Nature). (b) The trap system consisted of trap chamber (red rectangle) for placing the nematode for axotomy process and time-lapse imaging; and 4 valves (1–4) for controlling the position of nematode inside the trap chamber and gating to the recovery chambers (Reproduced from Guo et al. (2008). Copyright 2008, Springer Nature). (c) Microfluidic device integrated loading chamber for placing the worm population, staging area for separating the single-worm from the population, trapping area for positioning the nematode for axotomy process and imaging, 3D interconnects for controlling the sealing valves, flushing inlet for introducing the flow inside the trapping area, exit outlets for ejecting the worm to the off-chip area. The V1 and V2 were used to ensure the single-worm that was injected into the trapping area. The V3 was used to immobilize the position of the worm. The V4 and V5 provided the open and close conditions of the ejecting channel, respectively (Reproduced from Gokce et al. (2014). Copyright 2014, PLOS). Scale bars, 2 mm (a), 1 mm (b), and 400 μm (c)

304

D. Anggraini et al.

the staging area was filled with single-worm, the V1 was closed, V2 was opened, and V3 was opened, resulting the worm was transported into the trapping area. Once the nematode was in the trapping area, the V3 was closed to immobilize the position of the worm, then the axotomy process was performed using femtosecond laser with a pulse energy of 7.5 nJ and pulse-width of 260 fs, and observed using image processing algorithms. After the axon surgery, the V4 was opened, then the axotomized worms were transported into the agar plates and observed 24 h postsurgical process. The compartment for recovery process did not include in this microdevice. The average lifespan of the worms in this approach did not significantly differ from the control group. Moreover, this approach exhibited the success rate of automated axotomies of 67.4% with the processing time of 17 s per worm, providing effective and efficient axotomy method and axonal reconnection observation in the controlled microenvironment (Gokce et al. 2014).

Stem Cell-Based Regeneration Exogenous stem cells have been used as cell-based therapy for end-stage degenerative diseases, including Alzheimer’s disease, Parkinson’s disease, Amyotrophic lateral sclerosis (ALS), liver cirrhosis, and osteoporosis. There are various mechanisms of exogenous stem cells that contribute to the regeneration process, i.e., migrating to the undesirable area, inducing the cellular behavior of endogenous microenvironment, secreting growth factor, and inducing the production of endogenous stem cells. To understand exogenous stem cell-based regeneration, microfluidic technique can be used for mimicking the condition of degeneration diseases and also improving the drug delivery system. Cartilage tissue regeneration was induced by encapsulated bone marrow-derived mesenchymal stem cells (BMSCs) in gelatin norbornene (GelNB) in droplet microfluidic device. They fabricated a simple pipet tip-based microfluidic device that comprised PTFE tubes, silicone tube, pipet tip, and needles. The end parts of two PTFE tubes were placed inside the pipet tip using a mold. One PTFE tube was connected with the silicone tube which lengthened to the outside of the pipet tip, while another PTFE tube remained inside the pipet tip. The 0.5 mm hole was created in the lateral wall of the silicone tube just beside the remaining PTFE tube (Fig. 4a). The inlet sides of PTFE tubes were then connected with syringes for delivery aqueous and oil phases (Fig. 4b). The mixture of hBMSCs and hydrogel precursor were loaded into aqueous phase syringe. The aqueous phase and oil phase with different flow velocities were mixed, then the GelNB microgels were formed. The fabricated microgels were dispersed in oil and exposed to a visible light source (400–500 nm, 10 mW cm 2) to attain complete gelation of microgels. The encapsulated cells were maintained in the basal medium and changed into chondro-inductive supplements after the first 24 h with medium changes every 2 days. To understand the regeneration mechanism in the defect of articular cartilage, the encapsulated cells were injected into the agarose cavity, and the regeneration process was observed (Fig. 4c and d). This simple method allowed the production of encapsulated cells in micrometer size with uniform swelling and retained integrity for up to 10 days. The regeneration mechanisms showed by increasing the number of cell viability and cell migration from the center to the

12

Single-Cell Cultivation Utilizing Microfluidic Systems

305

Fig. 4 Droplet microfluidic platforms for producing engineered stem cells as cell-based therapy. (a) The two PTFE tubes were fixed inside the pipet tip using a mold. One of PTFE tubes was connected with the silicone tube and lengthened to the outside of the pipet tip, while another tube was remaining inside the pipet tip. The 0.5 mm hole was created as a gate for the aqueous phase that was mixed with oil phase (Reproduced from Li et al. (2017b). Copyright 2017, American Chemical Society). (b) The inlet parts of PTFE tubes were connected with the syringes as aqueous inlet and oil inlet (Reproduced from Li et al. (2017b). Copyright 2017, American Chemical Society). (c) Agarose cavity for mimicking the defect of articular cartilage (Reproduced from Li et al. (2017b). Copyright 2017, American Chemical Society). (d) Encapsulated cells inside the agarose cavity (Reproduced from Li et al. (2017b). Copyright 2017, American Chemical Society). (e) Microfluidic device setup for producing the porous microsphere. The porous microspheres were fabricated from internal phase and external phase mixture, UV irradiation, and freeze-dried process. (Reproduced from Wu et al. (2020). Copyright 2020, Elsevier)

surface of the microgels after 7 days culture. Eventually, type II collagen as an indicator of chondrogenic differentiation was similar with bulk system. The successful rate of cell-based therapy with this approach will be increased because of the long life, slow degradation, and high deposit of extracellular matrix induced by stem cells inside the microgels (Li et al. 2017b). Moreover, in in vivo study, the simple droplet microfluidic device was also used for producing the engineered stem cells for bone defect therapy in mice. Instead of mixing the cell with hydrogel, this study produced the porous microspheres that allowed the adhesion and insertion of BMSCs. The platform was made of coaxially arranged needles to create the internal and external tubes. The connecting parts of the needles were covered with cone cap. The inlets parts were connected with syringes

306

D. Anggraini et al.

for transferring internal and external phases. The cone cap was then integrated with the plastic catheter, lengthened by 10 cm, and formed disc-shaped with a diameter of 10 cm, and connected with the collection device (Fig. 4e). This disc-shaped structure was aimed to enable ultraviolet (UV) irradiation from above of the platform. The internal phase contained mixture of gelatin methacrylamide (GelMA)-PBS and 0.5% photo-initiator, while the external phase consisted of mineral oil. The hydrogel droplets were produced following the mixture of internal and external phases with different flow velocities. The droplets were cross-linked by UV irradiated (365 nm, 6.9 mW cm 2) to form the solid microspheres, then freeze-dried to get the porous microspheres. Coculture of porous microspheres and BMSCs was performed for inserting the BMSs into the porous microspheres (freeze-dried cell-laden microsphere; F-CLM). The F-CLM was injected into the defect distal femur of mice. This approach permitted the production of porous microspheres in micrometer size. Cell proliferation and cell viability rates inside the porous microspheres were higher compared to those cell-encapsulated microspheres (P-CEM). Finally, the micro-CT imaging and histological analysis showed significantly increased mineral density and continuous callus formation after 3 weeks of F-CLM treatment, which was more positive than that of P-CEM treatment. With this platform, the engineered stem cellbased therapy has been succeeded to improve the healing rate of degenerative bone disease by increasing the lifespan and multiplication ability of cells. In addition, this approach also prevented the cell interference by using the nonchemical cross-linker and removing the harmful reagents with freeze-dried process (Wu et al. 2020). For improving the reliability of regeneration measurement using a microfluidic platform, there are some considerations that are supposed to be made, including 1) adjust the position of the sample inside the device that enables for experimental and observation process, 2) ensure the observation at a single-cell level/single organism level, 3) use communicable materials for revealing the phenomena of regeneration. Sample positioning is very important to deal with motile cells/motile organisms such as nematode. Pressurize adjustment and automated controlled valves assisted by computers are necessary to get the high-throughput speed and precision of experiment (Gokce et al. 2014). The observation at a single-cell level can also be achieved by computer-assisted automated controlled valves and syringe pump-controlled flow velocities (Gokce et al. 2014; Li et al. 2017b; Wu et al. 2020). Not only high technology imaging is required for observing the regeneration mechanisms inside the nematode body, (Guo et al. 2008; Gokce et al. 2014) but the transparent material such as agarose is also considerably needed to observe clearly the regeneration phenomena of stem cells inside the cavity (Li et al. 2017b).

Further Limitations and Future Prospects Despite the advantages offered by microfluidic devices for solving various biological applications, there are still many challenges rising from each particular microfluidic system. Microfluidic systems such as droplet trap, trap array, fluidic force trap, and passive well have each its limitation. The efficiency of single-cell trapping in droplet

12

Single-Cell Cultivation Utilizing Microfluidic Systems

307

microfluidics is very low ranging from 10% to 30% due to the inherent limitation of the Poisson distributions (Zhang et al. 2017; Sesen and Whyte 2020). Also, the cells tend to lose their contact with the pair cells in the confinement trap microfluidics (Dura et al. 2016; Dura et al. 2015). Thus, these will affect to nonadequate single-cell products (e.g., DNA, RNA, and proteins) for detailed analysis. Moreover, cell interferences cannot also be avoided due to the use of fluidic force-based microfluidics. Last, the use of passive well microfluidics offers high-efficiency trapping and noninvasive technique, (Lin et al. 2015) but it has limitations in low-throughput and time-consuming processing. To obtain the ultimate goals of diagnostics and therapies, microfluidic devices should fulfill the best conditions for culturing individual cells, such as label-free, noninvasive, high-throughput, and long-term culturing and analysis. Moreover, high-resolution technology also needs to be coupled to increase the accuracy and sensitivity of single-cell measurement (Ota et al. 2019; Dusny et al. 2019). Currently, microfluidic devices are still developed as a proof-of-concept rather than as broadly applicable alternatives to solving existing approaches. If some efforts have been made to combine the complexity of single-cell cultivation, the breakthrough of biological application bottlenecks can be performed. Integrative and comprehensive single-cell cultivation-based microfluidic systems will dramatically promote understanding of the fundamental mechanism of cellular and subcellular cells. Furthermore, single cells of desired properties have potential to work as a target of disease treatment and are useful to cure pathogenic states and wounds. With single-type cells, drug influence can be appropriately quantified with stable quality. Thus, therapeutic protocol can be established more reliably. Multiplication of a single desired cell can be applied to, for an instance, a cell sheet that is desired to have quality in reproducible manner. Ultimately, single-cell cultivation microfluidics is a highly potential tool that contributes to precision medicine. Besides, the example of utilizing single-cell multiplication is the production of biofuel based on a selected single microbe cell with desired properties. Through single-cell measurement and cultivation, biofuel production can be optimized by selecting and multiplying fuelrich cells.

Conclusion It is clear that microfluidics technology has made significant improvement for singlecell analysis over the past decades. High-accuracy and high-resolution data from single cells provide new insights into cell-cultivation processes. Many applications have been reported in the field of cell coculture, molecule-induced cellular behaviors, and cell regeneration using microfluidic system. The tendency of these applications has been shifted to multicompartments, high-throughput screen, multitarget analysis, and droplet-based microfluidics, which need more integrated, accurate, and controllable microfluidic systems. Recently, more research focuses on the molecular to cell assays based on the single-cell level to benefit clinical diagnosis and therapies, which indicates an important research direction. Although some advanced systems

308

D. Anggraini et al.

have been developed in microfluidic-based single-cell analysis, most of these systems are still at blueprint process rather than broadly applicable alternatives to existing approaches, and a certain time for adaptation is needed before biological laboratories routinely use microfluidic systems for cell cultivation. Efforts should be made to combine the complexity of cell cultivation and the multiplicity of microfluidic systems to break through the bottleneck of biological applications. Integrative and comprehensive single-cell cultivation-based microfluidic systems will dramatically promote understanding of the fundamental mechanism of intercellular cells in biology and greatly benefit diagnostic and therapies in medical processes.

References Aardema MJ, MacGregor JT (2002) Toxicology and genetic toxicology in the new era of ‘toxicogenomics’: impact of ‘-omics’ technologies. Mutat Res – Fundam Mol Mech Mutagen 499:13–25 Ahuja P, Sdek P, MacLellan WR (2007) Cardiac myocyte cell cycle control in development, disease, and regeneration. Physiol Rev 87:521–544 Bhattacharjee N, Li N, Keenan TM, Folch A (2010) A neuron-benign microfluidic gradient generator for studying the response of mammalian neurons towards axon guidance factors. Integr Biol 2:669–679 Boneschansker L, Yan J, Wong E, Briscoe DM, Irimia D (2014) Microfluidic platform for the quantitative analysis of leukocyte migration signatures. Nat Commun 5:1–12 Dinh ND et al (2013) Microfluidic construction of minimalistic neuronal co-cultures. Lab Chip 13:1402–1412 Dong L, Chen DW, Liu SJ, Du W (2016) Automated chemotactic sorting and single-cell cultivation of microbes using droplet microfluidics. Sci Rep 6:1–8 Dura B et al (2015) Profiling lymphocyte interactions at the single-cell level by microfluidic cell pairing. Nat Commun 6:5940 Dura B et al (2016) Longitudinal multiparameter assay of lymphocyte interactions from onset by microfluidic cell pairing and culture. Proc Natl Acad Sci U S A 113:E3599–E3608 Dusny C, Lohse M, Reemtsma T, Schmid A, Lechtenfeld OJ (2019) Quantifying a biocatalytic product from a few living microbial cells using microfluidic cultivation coupled to FT-ICR-MS. Anal Chem 91:7012–7018 Fritzsch B et al (2019) Neuronal migration generates new populations of neurons that develop unique connections, physiological properties and pathologies. Front Cell Dev Biol 7:1–8 Gokce SK et al (2014) A fully automated microfluidic femtosecond laser axotomy platform for nerve regeneration studies in C. elegans. PLoS One 9:1–28 Goodus MT, Guzman AM, Calderon F, Jiang Y, Levison SW (2015) Neural stem cells in the immature, but not the mature, subventricular zone respond robustly to traumatic brain injury. Dev Neurosci 37:29–42 Gu B et al (2010) Global expression of cell surface proteins in embryonic stem cells. PLoS One 5:1–12 Guo SX et al (2008) Femtosecond laser nanoaxotomy lab-on-a-chip for in vivo nerve regeneration studies. Nat Methods 5:531–533 Ibrahim S et al (2016) Traumatic brain injury causes aberrant migration of adult-born neurons in the hippocampus. Sci Rep 6:1–12 Inada M, Izawa G, Kobayashi W, Ozawa M (2016) 293 cells express both epithelial as well as mesenchymal cell adhesion molecules. Int J Mol Med 37:1521–1527 Jin K et al (2001) Neurogenesis in dentate subgranular zone and rostral subventricular zone after focal cerebral ischemia in the rat. Proc Natl Acad Sci U S A 98:4710–4715

12

Single-Cell Cultivation Utilizing Microfluidic Systems

309

Jin K et al (2003) Directed migration of neuronal precursors into the ischemic cerebral cortex and striatum. Mol Cell Neurosci 24:171–189 Karakas HE et al (2017) A microfluidic chip for screening individual cancer cells via eavesdropping on autophagy-inducing crosstalk in the stroma niche. Sci Rep 7:1–11 Li R et al (2017a) Microvalve controlled multi-functional microfluidic chip for divisional cell co-culture. Anal Biochem 539:48–53 Li F, Truong VX, Thissen H, Frith JE, Forsythe JS (2017b) Microfluidic encapsulation of human mesenchymal stem cells for articular cartilage tissue regeneration. ACS Appl Mater Interfaces 9:8589–8601 Lim WF, Inoue-Yokoo T, Tan KS, Lai MI, Sugiyama D (2013) Hematopoietic cell differentiation from embryonic and induced pluripotent stem cells. Stem Cell Res Ther 4:1 Lin CH et al (2015) A microfluidic dual-well device for high-throughput single-cell capture and culture. Lab Chip 15:2928–2938 Luedde T, Kaplowitz N, Schwabe RF (2014) Cell death and cell death responses in liver disease: mechanisms and clinical relevance. Gastroenterology 147:765–783.e4 Mikels AJ, Nusse R (2006) Purified Wnt5a protein activates or inhibits β-catenin-TCF signaling depending on receptor context. PLoS Biol 4:570–582 Mills R et al (2018) Neurturin is a PGC-1 a 1-controlled myokine that promotes motor neuron recruitment and neuromuscular junction formation. Mol Metab 7:12–22 Naito H, Iba T, Takakura N (2020) Mechanisms of new blood-vessel formation and proliferative heterogeneity of endothelial cells. Int Immunol 32:295–305 Ohab JJ, Fleming S, Blesch A, Carmichael ST (2006) A neurovascular niche for neurogenesis after stroke. J Neurosci 26:13007–13016 Ota N et al (2019) Isolating single Euglena gracilis cells by glass microfluidics for raman analysis of paramylon biogenesis. Anal Chem 91:9631–9639 Pécot T, Zengzhen L, Boulanger J, Salamero J, Kervrann C (2018) A quantitative approach for analyzing the spatio-temporal distribution of 3D intracellular events in fluorescence microscopy. elife 7:1–34 Poli D, Wheeler BC, Demarse TB, Brewer GJ (2018) Pattern separation and completion of distinct axonal inputs transmitted via micro-tunnels between co-cultured hippocampal dentate, CA3, CA1 and entorhinal cortex networks. J Neural Eng 15:1–26 Robertson G, Bushell TJ, Zagnoni M (2014) Chemically induced synaptic activity between mixed primary hippocampal co-cultures in a microfluidic system. Integr Biol (United Kingdom) 6:636–644 Rodriguez LG, Wu X, Guan J (2005) Wound-healing assay. Methods Mol Biol 294:23–29 Saha B, Peron S, Murray K, Jaber M, Gaillard A (2013) Cortical lesion stimulates adult subventricular zone neural progenitor cell proliferation and migration to the site of injury. Stem Cell Res 11:965–977 Sesen M, Whyte G (2020) Image-based single cell sorting automation in droplet microfluidics. Sci Rep 10:1–14 Shapiro LA (2017) Altered hippocampal neurogenesis during the first 7 days after a fluid percussion traumatic brain injury. Cell Transplant 26:1314–1318 Valentijn AJ, Metcalfe AD, Kott J, Streuli CH, Gilmore AP (2003) Spatial and temporal changes in Bax subcellular localization during anoikis. J Cell Biol 162:599–612 Wang ZZ, Wood MD, Mackinnon SE, Sakiyama-Elbert SE (2018) A microfluidic platform to study the effects of GDNF on neuronal axon entrapment. J Neurosci Methods 308:183–191 Watanabe A, Yamada Y, Yamanaka S (2013) Epigenetic regulation in pluripotent stem cells: A key to breaking the epigenetic barrier. Philos Trans R Soc Lond B Biol Sci 368:20120292 Wu J et al (2020) Stem cell-laden injectable hydrogel microspheres for cancellous bone regeneration. Chem Eng J 393:124715 Xiao RR et al (2013) Simultaneous generation of gradients with gradually changed slope in a microfluidic device for quantifying axon response. Anal Chem 85:7842–7850 Xu H, Heilshorn SC (2013) Microfluidic investigation of BDNF-enhanced neural stem cell chemotaxis in CXCL12 gradients. Small 9:585–595

310

D. Anggraini et al.

Yang C, Xia BR, Jin WL, Lou G (2019) Circulating tumor cells in precision oncology: clinical applications in liquid biopsy and 3D organoid model. Cancer Cell Int 19:1–13 Zhang Y et al (2012) Single cell analysis of yeast replicative aging using a new generation of microfluidic device. PLoS One 7:e48275 Zhang Q et al (2017) Development of a facile droplet-based single-cell isolation platform for cultivation and genomic analysis in microorganisms. Sci Rep 7:1–11

Integrated Microwell Array Technologies for Single Cell Analysis

13

Jolien Breukers, Caroline Struyfs, Sara Horta, Karin Thevissen, Karen Vanhoorelbeke, Bruno P. A. Cammue, and Jeroen Lammertyn

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfabrication of a Microwell Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material and Design Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Soft Photolithography for Fast Prototyping of Microwell Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrophilic-in-Hydrophobic Microwells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Polyethylene Glycol Microwells for Reduced Non-specific Adsorption . . . . . . . . . . . . . . . . . . . Single Cell Docking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Docking Strategies: Manual Administration of Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Docking Strategies: Microfluidics-Assisted Administration of Cells . . . . . . . . . . . . . . . . . . . . . . . Single Cell Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Magnetic Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanical Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electric Manipulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

312 313 313 315 316 318 319 320 322 326 326 328 328 329

Jolien Breukers, Caroline Struyfs and Sara Horta contributed equally with all other contributors. J. Breukers · J. Lammertyn (*) Department of Biosystems – Biosensors Group, KU Leuven, Leuven, Belgium e-mail: [email protected]; [email protected] C. Struyfs · B. P. A. Cammue Centre of Microbial and Plant Genetics, KU Leuven, Leuven, Belgium VIB Center for Plant Systems Biology, Ghent, Belgium e-mail: [email protected]; [email protected] S. Horta · K. Vanhoorelbeke Laboratory for Thrombosis Research, IRF Life Sciences, KU Leuven Campus Kulak Kortrijk, Kortrijk, Belgium e-mail: [email protected]; [email protected] K. Thevissen Centre of Microbial and Plant Genetics, KU Leuven, Leuven, Belgium e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_21

311

312

J. Breukers et al.

Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Cell Drug Screening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Cell Omics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Detection of Single Cell Secreted Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Other Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

330 330 334 335 337 338 338

Abstract

Microwell arrays are an important tool for high-throughput single cell analysis, with different approaches of microwell-based studies being described over the last years. For this, a large variety of materials and microfabrication techniques has been developed for fabrication of microwells with dimensions compatible to the size of a single cell. Cells can be administered to the microwell array by manual pipetting or in an automated fashion by microfluidics, after which they can passively sediment inside the microwells. To increase cell docking efficiency, active seeding methods such as centrifugation or dielectrophoresis can be integrated. Furthermore, several techniques can be combined with microwell arrays for targeted single cell manipulation in a microwell, such as optical tweezers or a micromanipulator. The versatility of these microwell-based single cell analyses has enabled to study different cell types, ranging from bacteria to yeast and human cells with spatiotemporal resolution.

Introduction Individual cells within a homogeneous population show cellular heterogeneity at genomic and functional levels, due to stochastic expression of genes, proteins, and metabolites (Gao et al. 2019; Kumar et al. 2015). This heterogeneity is often overlooked when performing bulk experiments since only the average response of the cell population is measured. In contrast, single cell studies allow to discriminate and investigate each individual cell within a morphologically identical population (Lindström and Andersson-Svahn 2011). Flow cytometry is often seen as the golden standard for single cell analysis. This technique enables high-throughput detection of cell size through forward scattering, cell granularity through side scattering, and/or cell response(s) based on fluorescent labeling (Hu et al. 2016). Fluorescence-activated cell sorting (FACS), an extension of flow cytometry, enables sorting of selected cell subpopulations for further downstream analyses. Although this technique is highly versatile and adaptive to different cell types, the technique also has limitations, such as its inability to measure kinetic responses (Lindström and Andersson-Svahn 2011), secreted cell products (Shirasaki et al. 2011), or cell growth (Chin et al. 2004). Several solutions based on microfabrication technologies have been proposed to overcome the limitations of flow cytometry. Some of them involve the fabrication of microwell arrays, in which single cells can be trapped, allowing single cell studies with spatiotemporal resolution. The array is covered with cells, either manually or by

13

Integrated Microwell Array Technologies for Single Cell Analysis

313

microfluidics, after which the single cells can be docked in the microwells passively by sedimentation or by using active techniques such as centrifugation or dielectrophoresis. Additionally, specific manipulation techniques can be integrated with these arrays, allowing, for example, cell pairing, cell relocation, and cell retrieval for downstream analysis. In the following paragraphs, we will review different microwell microfabrication techniques, different techniques for single cell docking in microwells, and the integration of manipulation tools with microwell arrays. Lastly, we will highlight some single cell studies from different fields to demonstrate the versatility of microwell array technology.

Microfabrication of a Microwell Array Microwells are defined as physical structures with a diameter and depth in the micrometer range in which cells can be mechanically isolated from each other. By confining the cells in a fixed location, both spatial and temporal information on the cellular response can be obtained. Different microfabrication techniques have been developed over the years to address the specific needs of an application. A selection of techniques will be discussed here, as well as material and design considerations that need to be taken into account.

Material and Design Considerations The material choice of the microwell array is crucial and depends on the chosen application. First, the material needs to be biocompatible if living cells are monitored. When using microscopy for visualization, the material should be optically transparent and have low autofluorescence. Other factors that influence material choice are the need for surface modifications, integration of the microwell array with microfluidics, or sealing of the microwells. For the latter, it is important to note that a droplet of medium or reaction buffer often covers the array during an experiment to prevent evaporation of the liquid in the small picoliter or femtoliter microwells, thus enabling cross-talk between cells through secreted analytes (Lindström and Andersson-Svahn 2011). If cross-talk between the microwells has to be avoided, for example, when the detection of secreted cellular products is of interest, the microwells can be sealed to have individual reaction chambers. This can be done mechanically by pressing a hard cover, such as a glass slide, on a microwell array made out of flexible material (Tsuda et al. 2015). Another option is chemical sealing by covering the individual microwells with a layer of oil (Yuan and Sims 2016), which is facilitated by a hydrophobic microwell surface (Witters et al. 2013). After choosing an appropriate material, a microwell array needs to be designed with the desired microwell shape, size, and number. The microwell shape is generally square or circular (Lindström and Andersson-Svahn 2011), though also triangular microwells have been used (Park et al. 2010) (Fig. 1). The microwell size can

314

J. Breukers et al.

Fig. 1 Different microwell shapes for single cell analysis. (a) Circular microwells (Adapted with permission from Rettig and Folch 2005. Copyright © 2005 American Chemical Society). (b) Square microwells (Adapted with permission from Chin et al. 2004. Copyright © 2004 Wiley InterScience). (c) Triangular microwells. (Reprinted with permission from Springer Nature Customer Service Centre GmbH: Springer Nature Microfluidics and Nanofluidics Park et al. Copyright © 2010. Copyright 2010, Springer Nature)

be chosen to accommodate just one cell (Kumar et al. 2015) or can be chosen to be larger than a cell (Park et al. 2010; Yuan and Sims 2016). The latter can be preferred (i) when multiple cells per microwell are needed, for example, to study cellular interactions (Wang et al. 2013a; Yoshimura et al. 2014), (ii) when extra space is required to study cell growth and proliferation (Park et al. 2010), or (iii) when access for mechanical manipulation tools is required (Bradshaw et al. 2008; Jin et al. 2009). The microwell number generally ranges from hundreds to hundreds of thousands and can be chosen in accordance to the desired throughput. The area comprised by the microwell array and thus the time needed for imaging are determined by the combination of the microwell number, microwell size, and interwell spacing (Lindström and Andersson-Svahn 2011). Lastly, a suitable microfabrication technique needs to be selected to produce the microwell array design in the chosen material. In the next sections, several combinations of material and microfabrication techniques are discussed, together with their features, advantages, and disadvantages.

13

Integrated Microwell Array Technologies for Single Cell Analysis

315

Soft Photolithography for Fast Prototyping of Microwell Arrays Polydimethylsiloxane (PDMS) is a commonly used material for microfabrication of microwells in research laboratories. It is highly biocompatible, gas permeable allowing the exchange of O2 and CO2, and optically transparent and has low autofluorescence. The PDMS surface is chemically inert, but surface modifications can still be implemented after activation with air or oxygen plasma. Microstructures in PDMS can be fabricated rapidly outside the clean room, which makes it an attractive choice for prototyping in research facilities (Lindström and AnderssonSvahn 2011). PDMS is a flexible elastomer, which can be an advantage or a disadvantage depending on the desired application. Indeed, the flexibility facilitates sealing of the microwells onto other materials, such as glass (Bradshaw et al. 2008), but can cause collapsing of microwells with a high aspect ratio, i.e., large depth compared to the diameter (Lindström and Andersson-Svahn 2011). Another consideration is the hydrophobicity of PDMS, which leads to air bubble formation in the microwells when applying liquid onto the microwell array. To avoid this, the array can first be placed in a vacuum chamber with liquid on top before cells are applied (Wang et al. 2013b). In case the microwell array needs to be embedded in a continuous flow microfluidic platform, microwells in PDMS offer an advantage for device integration since PDMS is often a convenient choice for microfluidic channel fabrication (Wang et al. 2013b). Microstructures in PDMS are typically manufactured by pouring liquid PDMS on a mold, which is the negative of the final structure that has to be fabricated. Thermal curing is then used to solidify the PDMS. The mold to generate the microwells contains micropillars, which can be produced on a silicon wafer by (deep) reactive ion etching or by soft photolithography using a photoresist. The latter is the most commonly used technique as prototyping is relatively straightforward (Lindström and Andersson-Svahn 2011). Figure 2 illustrates a typical protocol for fabricating PDMS microwells using SU-8 soft photolithography. SU-8 is an epoxy-based negative photoresist which cross-links when exposed to UV light. First, SU-8 is spin-coated on top of a silicon wafer. The viscosity of the selected SU-8 and the rotation speed of the spin-coater determine the thickness of the layer and thus the final depth of the microwells. After spin-coating, an opaque mask with transparent circles with desired diameter is placed in contact with the wafer. Then, the wafer is exposed to 365 nm UV light to cross-link the photoresist below the transparent area of the mask. The uncrosslinked SU-8 is washed away by submerging the wafer in developer. The result of this process is a silicon wafer with SU-8 pillars. Subsequently PDMS is poured on the mold, cured by temperature, and peeled off again from the mold resulting in PDMS microwells. Ideally, the pillars of the SU-8 mold should remain intact for reuse (Rettig and Folch 2005). Microwells for cells in PDMS from an SU-8 mold can be easily fabricated in a range of 20 μm (Park et al. 2016) to 500 μm diameter (Chin et al. 2004).

316

J. Breukers et al.

Fig. 2 Different steps in a standard SU-8 photolithography process for microwell fabrication. (a) SU-8 photoresist is spincoated on a silicon wafer. A mask with an array of transparent circles, squares, or triangles is placed on top of the wafer with SU-8. UV light is applied to transfer the mask design onto the SU-8 mask to locally cross-link the SU-8. (b) The non-exposed SU-8 is washed away using developing solution to obtain a mold with SU-8 pillars. (c) PDMS is poured onto the mold. (d) After thermal curing, the PDMS is peeled off from the wafer to result in PDMS microwells

Hydrophilic-in-Hydrophobic Microwells The need for applying vacuum before seeding cells can be avoided by using hydrophilic microwells in a hydrophobic surrounding, named hydrophilic-inhydrophobic microwells (HIH microwells). Moreover, a HIH microwell array can be chemically sealed using hydrophobic oils such as fluorocarbon oils and can be integrated with digital microfluidics using the electrowetting-on-dielectric (EWOD) approach (Witters et al. 2013), as will be discussed in more detail in section “Digital Microfluidics.” The surface of PDMS microwells can be modified to result in HIH microwells. For example, Zhang et al. (2016) submerged PDMS microwells in a dopamine solution to form a thin, hydrophilic polydopamine layer on the entire PDMS surface. The top surface of the microwell array was made hydrophobic by pressing the PDMS against an adhesive layer followed by a superhydrophobic layer. A singlestep protocol for long-lasting hydrophilic microwells in the hydrophobic PDMS polymer has also been described. For this, a mask with circular holes was placed on top of a PDMS slab, and the exposed surface was irradiated with a low-energy electron beam, which altered the chemical structure of the exposed PDMS, causing the formation of concave hydrophilic microwells (Oyama et al. 2018). Teflon-AF ® and CYTOP® are two other hydrophobic materials that have been used to fabricate HIH microwells with a hydrophilic glass bottom. Both materials are optically transparent, have low autofluorescence, and are biocompatible as they are highly inert. Teflon microwell fabrication has been described for diameters down to 5.5 μm for yeast cells (Kumar et al. 2015) and even down to 3 μm for paramagnetic

13

Integrated Microwell Array Technologies for Single Cell Analysis

317

Fig. 3 Fabrication of hydrophilic-in-hydrophobic microwells in Teflon. (a) Teflon is spincoated on a silanized glass slide. (b) A Parylene C and aluminum layer are deposited on the Teflon layer by chemical vapor deposition and thermal evaporation, respectively. Next, photoresist is spincoated on top of the chip. (c) The photoresist is exposed by UV through a mask with the microwell array design. (d) The exposed photoresist is washed away by developing solution. (e) Using aluminum etching, the exposed aluminum is washed away. (f) Reactive ion etching is used to remove the exposed Parylene C and Teflon layers, all the way through to the glass. (g) After peeling off the upper layers, a microwell array with a hydrophilic glass bottom and a hydrophobic Teflon surface remains. A detailed protocol can be found in Witters et al. (2013)

beads (Witters et al. 2013). The standard process of Teflon microwell fabrication for single cells is labor-intensive and time-consuming and requires clean room techniques such as chemical vapor deposition, thermal evaporation, and reactive ion etching (Fig. 3) (Witters et al. 2013). A more straightforward method of fabricating Teflon microwells by hot embossing has recently been developed which can be transferred to non-clean room laboratories (Tripodi et al. 2018). CYTOP microwell fabrication for single cells has been described for diameters down to 25 μm (Shirasaki et al. 2011) and can be performed outside the clean room. Moreover, both Teflon and CYTOP are suitable materials for electrowetting-on-dielectric in digital microfluidics (section “Digital Microfluidics”). Recently, a single-step imprinting method for fabricating HIH microwells was introduced in an off-stoichiometric thiol-ene-epoxy (OSTE+) polymer (Decrop et al. 2017). This dual cure polymer contains off-stoichiometric ratios of monomers with thiol, allyl, and epoxy functional groups. After a first UV cure, the allyl

318

J. Breukers et al.

functional groups cross-link with the thiol functional groups, which are present in excess compared to the allyls, enabling the excess thiols to cross-link with the epoxy functional groups in a second thermal cure (Saharil et al. 2013). In liquid state, the OSTE+ polymer is spin-coated on a glass slide, and a PDMS slab with micropillars is pressed against the OSTE+. This way, the pillars are printed in the OSTE+, resulting in microwells in OSTE+ after the UV cure and peel-off. To have HIH microwells, hydrophobic monomers are added to the OSTE+ solution. Because of these monomers, the OSTE+ can mimic the surface energy of the PDMS slab, as the monomers can diffuse in the uncured polymer and self-organize at the hydrophobic PDMS surface. The pillars of the PDMS slab are squeezed onto the glass surface, so the bottom of the microwells is hydrophilic. So far, these microwells have only been integrated with magnetic beads, but might be suitable for single cell analysis as well since OSTE+ is a biocompatible polymer. Moreover, OSTE+ is optically clear and generally has low autofluorescence. Only in the UV/Blue region, the material has significantly stronger autofluorescence as compared to Teflon and glass (Decrop et al. 2017).

Polyethylene Glycol Microwells for Reduced Non-specific Adsorption Hydrophobic surfaces such as PDMS or Teflon can suffer from non-specific adsorption of proteins and cells. When the effect of a low abundant compound needs to be analyzed, this non-specific adsorption can lead to erroneous results (Charnley et al. 2009). Also, when studying spheroid formation, cells should be repelled by the microwell surface (Karp et al. 2007). In these cases, microwells with low biofouling surfaces are preferred, i.e., surfaces that prevent the accumulation of cells and proteins. Although the non-specific adsorption of proteins on PDMS microwells can be reduced by blocking the surface with bovine serum albumin (BSA) (Kovac and Voldman 2007), this is not always sufficient. Therefore, the low biofouling molecule polyethylene glycol (PEG) is often used, since a surface covered with PEG is known to be hydrophilic and to block protein adsorption and cell adhesion (Kang et al. 2014). The mechanism of protein repellence by PEG is not completely understood, but properties such as chain density, chain conformation, and chain length seem to play an important role (Banerjee et al. 2011). A PDMS surface can be treated with PEG to have a low fouling surface. One treatment method is physical adsorption, in which air plasma is first applied to oxidize the PDMS resulting in a negatively charged surface. By applying polycationic PEG, it adsorbs to the PDMS surface due to electrostatic interactions (Lee and Vörös 2005). PEG adsorption by contact imprinting has also been performed to only cover the top surface of the microwell array with PEG. In this method, a PDMS microwell array is simply pressed into a PEG hydrogel, leaving residues on the PDMS (Charnley et al. 2009). Furthermore, PEG can be covalently bonded to a PDMS surface by first chemically modifying the PDMS and then cross-linking a PEG type with a compatible functional group (Farrell and Beaudoin 2012; Guo et al. 2007).

13

Integrated Microwell Array Technologies for Single Cell Analysis

319

Fig. 4 Fabrication of microwell arrays using PEG diacrylate hydrogel. (a) Uncross-linked PEG diacrylate or dimethacrylate hydrogel is spin-coated on a substrate. (b) A PDMS mold with micropillars is pressed into the uncross-linked PEG hydrogel, and UV is applied to cross-link the diacrylate or dimethacrylate groups. (c) The PDMS mold is peeled off. A fully cured microwell array in PEG hydrogel is obtained

A shortcoming of physical adsorption and chemical modification of PEG onto other materials is the slow recovery to the original, more hydrophobic state of the material. Therefore, protocols for the fabrication of microwells completely composed of PEG molecules have been developed with PEG diacrylate or PEG dimethacrylate hydrogels (Kang et al. 2014). A common fabrication method is UV-assisted microwell molding, in which a PEG solution is poured on a glass slide and is imprinted by a PDMS mold with micropillars (Fig. 4). After a UV cure, the PEG molecules cross-link, and a solid structure is formed, resulting in low fouling microwells. This relatively straightforward fabrication method can be performed in non-clean room environments. Furthermore, these PEG hydrogels have shown great optical clarity, low autofluorescence, and good biocompatibility, which makes the material a popular choice for single cell microwells. Additionally, these PEG microwell arrays have been combined with continuous flow microfluidic channels in both PDMS and PEG (Kang et al. 2014).

Single Cell Docking Docking of individual cells in microwells is required to conduct microwell-based single cell experiments. Therefore, a cell suspension must be applied on top of the array, which can be done manually, often by pipetting, or in an automated fashion using microfluidic techniques, such as continuous flow or digital microfluidics. Then, single cell docking itself can be achieved using different strategies, such as gravitation or centrifugation. To assess the performance of single cell docking, the docking efficiency is evaluated as the percentage of microwells containing a single cell (Rettig and Folch 2005). This section further discusses different approaches for trapping individual cells in a microwell array.

320

J. Breukers et al.

Docking Strategies: Manual Administration of Cells To obtain high single cell occupancies without favoring multiple cells in one microwell, the concentration of the cellular suspension, the microwell dimensions, and the sedimentation time must be taken into account (Rettig and Folch 2005). The most simple and straightforward approach to bring the cell suspension onto a microwell array is by manually adding it on top of the array via pipetting (Rettig and Folch 2005; Hughes et al. 2014). Single cells can be trapped in microwells by using different strategies, such as (i) gravitation, (ii) centrifugation, (iii) applying negative pressure, and (iv) dielectrophoresis. During gravity-induced single cell docking, cells sediment in the microwells solely by gravitational forces (Fig. 5a). Afterward, a washing step removes all non-seeded cells. The main advantage of gravitation-based single cell trapping is that it does not require additional equipment such as pumps or centrifuges and allows a simple chip design, resulting in relatively low manufacturing costs (Rettig and Folch 2005). Using this approach, Rettig and Folch (2005) observed that microwells with an aspect ratio of 1 yield an optimal single cell occupancy for both fibroblasts and leukemia cells (docking efficiency of 85 and 92%, respectively). Furthermore, single cell docking efficiency increased with sedimentation time. Although at some point the docking efficiency might stagnate, the overall cell occupancy will increase, as more microwells contained multiple cells (Rettig and Folch 2005). However, it has to be noted that the sedimentation time depends on the density of the cell, thus on the cell type, as compared to the density of the medium. In general, docking time for gravity-induced cell seeding remains relatively long, and highly efficient single cell occupancy is difficult to achieve (Gao et al. 2019). Numerous applications have used this gravity-based single cell seeding approach, including to study cell proliferation, cell differentiation, and cytokine production of adult hippocampal progenitor cells (docking efficiency 27%) (Chin et al. 2004) and single cell Western blotting of neuronal stem cells (docking efficiency 40–50%) (Hughes et al. 2014). Secondly, single cell docking can be achieved using centrifugation. Therefore, a cell suspension is positioned on top of the microwell array, and the cells are deposited into the wells under centrifugal force (Fig. 5b) (Huang et al. 2005). To improve single cell occupancy of the wells, multiple centrifugation cycles can be performed. After centrifugation, the arrays are washed to remove all non-seeded cells. The microwells can have different geometries, such as a truncated cone shape, which assures that cells cannot escape easily from the microwells after trapping (Fig. 5b) (Huang et al. 2005). Apart from centrifuges, this setup does not require additional devices such as pumps or complex chip designs. Compared to gravitational cell docking, capture efficiencies are increased when using centrifugationbased trapping (Gao et al. 2019). Centrifugation-based single cell capturing has been used to study, among others, the dynamic responses of single cells to drug treatments of lung cancer cells (docking efficiency 99%; Park et al. 2016) and the real-time observation of cellular apoptosis of human cervix adenocarcinoma cells (HeLa) cells (docking efficiency 90%; Huang et al. 2005).

13

Integrated Microwell Array Technologies for Single Cell Analysis

321

Fig. 5 Manual single cell docking. (a) Gravitation-based single cell docking. Single cells are solely docked using gravitation forces. Evaporation in this setup can be avoided by placing the microwell array in a medium reservoir (Adapted with permission from Rettig and Folch 2005. Copyright © 2005 American Chemical Society). (b) Centrifugation-based single cell docking. Single cells are docked in one or multiple centrifugation cycles. Microwells can have a truncated cone shape, assuring that cells cannot escape easily from the microwells after docking (Adapted with permission from Huang et al. 2014. Copyright © 2014 American Chemical Society). (c) Applying negative pressure to achieve single cell docking. Microwells have pores at the bottom, allowing fluid to leave the microwells if a negative pressure is applied, whereas cells remain docked, as their dimension is larger than the pore diameter (Adapted with permission from Swennenhuis et al. 2014. Copyright © 2014, Royal Society of Chemistry). (d) DEP-based single cell docking. A cell suspension was positioned on top of the microwell array, between the electrode and the counter electrode. Cells are polarized by exposing them to a nonuniform electric field. Therefore, a dielectrophoretic force causes the cells to move toward the high electric field region inside the microwells. (Reproduced from Kim et al. 2011a, with the permission of AIP Publishing ©)

Thirdly, single cell docking can be achieved by applying negative pressure onto the system. For this, Swennenhuis et al. (2014) used microwells with pores at the bottom, allowing the fluid to leave the microwells when a negative pressure is applied, thereby resulting in hydrodynamic forces for cell capturing. As the dimension of the cancer cell lines is larger than the pore diameter, the cells remain docked within the wells (docking efficiency 67%; Fig. 5c). When applying negative pressure for single cell docking, additional pumps are required. As compared to gravityinduced docking, docking efficiencies are increased when applying negative pressure onto the system. Lastly, single cell docking can be achieved using dielectrophoresis (DEP). In this approach, cells are polarized by exposing them to an electric field. Since the electric field is nonuniform, the electrostatic forces on the positive and negative pole are not equal. Therefore, a dielectrophoretic force will cause the cell to move toward or away from the high electric field region. The direction of movement is defined by the polarizability of the cell: if the cell is more polarizable than the medium, the cell is directed toward the stronger electric field region (positive

322

J. Breukers et al.

DEP, pDEP). If the cell is less polarizable than the medium, the cell moves away from the high electric field region (negative DEP, nDEP). Thus by controlling electrodes, cells can be specifically deposited into microwells (Bocchi et al. 2012; Kim et al. 2011b). Using DEP, Kim et al. (2011a) developed a device to electrically trap and lyse single Escherichia coli K-12 bacteria in a high-throughput fashion (docking efficiency 61%; Fig. 5d). To achieve this, an electrode is placed below the microwell array. A cell suspension is brought on top of the array, above which the counter electrode is positioned. By applying an alternating current on the electrodes, a highly localized electric field is induced, pulling the cells down into the microwells. Interestingly, by increasing the electrical field, bacteria can be subsequently lysed in the wells allowing the analysis of intracellular proteins. The main advantage of DEP-based devices is that they can be used for both cell docking and cell targeting (see section “Electric Manipulation”). However, this system does remain costly and can be complex to manufacture. Despite being relatively straightforward approaches for single cell docking, these open systems are also associated with a number of disadvantages. When a cell suspension is pipetted on top of the microwell array for seeding, at least one boundary of the system must be accessible, resulting in a large liquid-gas interface. As the fluid containing the cells is exposed to air, evaporation of the samples must be monitored. This can be avoided by placing the microwell array in a medium reservoir (Fig. 5a) (Rettig and Folch 2005). Furthermore, risk of sample contamination increases due to the open configuration. Also, docking individual cells is associated with numerous forces that are imposed upon these cells. Therefore, it is of importance to validate if the trapping process is not affecting cell viability when optimizing docking protocols. Cell viability is often monitored by fluorescent dyes, such as ethidium homodimer-1, propidium iodide, and fluorescein diacetate.

Docking Strategies: Microfluidics-Assisted Administration of Cells In contrast to manual addition of the cell suspension, the cells can be brought onto a microwell array in an automated fashion by means of microfluidics. In this section, different microfluidics-assisted docking strategies are discussed.

Continuous Flow Microfluidics Continuous flow microfluidics (CMF) is a liquid-handling technology that allows the manipulation of a continuous fluid flow through a network of enclosed microscale channels. The flow propagates in the laminar regime; thus, fluid mixing occurs solely via diffusion. Typically, a CMF device consists of an integrated or an external force, such as syringe or pressure pumps, to actuate the liquid flow through the channels (Murphy et al. 2017). Using CMF, cells are automatically delivered on top of the microwell array via the liquid flow through the channels. Docking single cells into the microwell array can be achieved via different strategies of which some are also employed for manually adding cells on top of the array, such as gravitation and

13

Integrated Microwell Array Technologies for Single Cell Analysis

323

DEP, and of which others are more typical for CMF, such as hydrodynamic pressure and fluid streamlines inside microwells. Gravity-induced single cell seeding remains the most simple and straightforward manner of trapping single cells in a microwell array using CMF. Usually, cells are allowed to sediment in the microwells after stopping the continuous flow, after which all non-docked cells are washed away by resuming the continuous flow. Using this approach, Lecault et al. (2011) used a continuous flow device to trap single hematopoietic stem cells and perform live-cell imaging studies (docking efficiency 10–30%). Likewise, Jen et al. (2012) used a microfluidic platform to capture individual HeLa cells for performing on-chip chemical and electric lysis in a high-throughput fashion (docking efficiency 83–91%; Fig. 6a).

Fig. 6 Continuous flow microfluidic platform for single cell docking. (a) Gravitation-based single cell docking. Cells are allowed to sediment in the continuous liquid flow, and all non-trapped cells are washed away by this flow (Adapted with permission from Jen et al. 2012. Copyright © 2012, MDPI). (b) Fluid streamlines inside microwell-based single cell docking. Single cells can be docked by making use of the formation of fluid streamlines inside microwells. A flow recirculation occurs within triangular wells, which were found to be the most efficient geometry for such single cell docking (Reprinted with permission from Springer Nature Customer Service Centre GmbH: Springer Nature Microfluidics and Nanofluidics Park et al. Copyright © 2010. Copyright © 2010, Springer Nature). (c) DEP-based single cell docking. Single cells in a continuous flow can be actively docked using an array of electroactive microwells. (d) Hydrodynamic pressure-based single cell docking. Conical nanoporous membrane integrated into a microfluidic chip allow to capture one bacterium per pore by using the continuous liquid flow. (Reprinted with permission from Guo et al. 2012. Copyright © 2012, Royal Society of Chemistry)

324

J. Breukers et al.

Secondly, single cell trapping via CMF has been achieved by using a combination of gravity and the formation of fluid streamlines inside microwells. Cells in a microfluidic flow follow fluid streamlines; however, due to gravitational forces, cells gradually settle, thereby descending to a lower streamline. Park et al. (2010) found that the streamlines in triangular microwells were most efficient for docking single prostate cancer cells in large microwells, providing sufficient space for growth studies (docking efficiency 62%; Fig. 6b). Thirdly, DEP can also be used in combination with CMF to capture individual cells (Fig. 6c). Bocchi et al. (2012) developed a device for high-yield controlled patterning of leukemia cells (docking efficiency almost 100%). Similarly, Kim et al. (2011b) developed a CMF/DEP-based device with an array of electroactive microwells to actively trap both adherent (HEK 293) and nonadherent (U937) cells (docking efficiency 95%) and subsequently lysed them within the wells for further measurement of intracellular β-galactosidase activity. Lastly, single cell docking via CMF was achieved using hydrodynamic pressure. Guo et al. (2012) used a conical nanoporous membrane integrated into a microfluidic chip to capture one cyanobacterium per pore by using the continuous liquid flow, thereby generating a hydrodynamic force (docking efficiency 7–28%). Subsequently, all docked bacteria could be simultaneously released by inverting the flow (Fig. 6d). An advantage of CMF is that the systems have a relatively simple channel design and fabrication. The microscale channels also assure that these platforms are closed systems, thereby reducing the risk of sample evaporation and contamination (Murphy et al. 2017). Additionally, these devices suffer less from protein adsorption to the surface, since the surfaces are not typically hydrophobic as is the case for digital microfluidic devices (section “Digital Microfluidics”). Nevertheless, since the fluid propagates in microscale channels, these platforms may suffer from channelclogging if cell-rich samples are employed (Kumar et al. 2015). Cell viability must also be taken into account since cells might experience shear stress in a continuous flow, depending on fluid velocity and viscosity. Shear stress can have an important impact on cellular behavior, such as morphological changes and altered intracellular signaling pathways. Note that the effect of shear stress strongly depends on the cell type (Inamdar et al. 2011).

Digital Microfluidics Single cell docking can also be achieved using digital microfluidics (DMF), which is a liquid-handling technology allowing precise manipulation of pico-to-microliter droplets on a two-dimensional array of electrodes by means of electrowetting-ondielectric (EWOD) (Choi et al. 2012; Samiei et al. 2016; Murphy et al. 2017). EWOD-based DMF devices consist of an actuation and a grounding plate. The former comprises a glass plate on which individually addressable electrodes are positioned covered with both an insulating dielectric (e.g., Parylene C) and a hydrophobic layer (e.g., Teflon or CYTOP) harboring the reference electrode (Fig. 7) (Ng et al. 2015). Applying a voltage onto the system results in electrical charges in the dielectric layer, consequently generating an electrical double layer at

13

Integrated Microwell Array Technologies for Single Cell Analysis

325

Fig. 7 EWOD-based DMF platform for single cell docking. A droplet with yeast cells is sandwiched between the actuation and the grounding plate surrounded by a silicon oil layer. Droplets are manipulated via the EWOD principle. Single cells are docked and retained within the microwells due to a combination of gravity, a drag force, and the surface tension. (Reprinted with permission from Kumar et al. 2015. Copyright © 2015, Royal Society of Chemistry)

the dielectric-droplet interface. As a result, the surface tension of the liquid-solid interface will decrease, thereby increasing the wettability of the hydrophobic surface, resulting in the attraction of droplets toward the activated electrodes (Samiei et al. 2016). Therefore, droplets can be independently manipulated, which allows dispensing, merging, mixing, and splitting in an automated and parallel fashion with very high accuracy, speeding up assay-processing steps (Samiei et al. 2016; Chen and Shamsi 2017). DMF-based devices have been applied in numerous applications and are suitable for biological samples, allowing single cell docking (Choi et al. 2012). To implement spatiotemporal single cell studies onto the EWOD-based DMF platform, micropatterning of the hydrophobic layer is required. This can be achieved via surface modifications or via the integration of microwell structures. The former can be done via the selective biofunctionalization of the hydrophobic layer with hydrophilic patches. Since this resembles HIH microwell structures, they are named virtual microwells. Wheeler and co-workers used adhesive poly L-lysine patches on the DMF actuation plate for seeding and culturing adherent fibroblast cells (Ng et al. 2015). The fabrication of the latter is discussed above (section “Hydrophilic-inHydrophobic Microwells”). Single cell docking via DMF in microwells is achieved by firstly bringing the cells onto the microwell array. Then, cells can sediment either inside or in between the microwells via gravitation. Initially non-trapped cells can subsequently be docked by shuttling droplets over the microwell array, as this generates a drag force that, in combination with the surface tension, allows cells to enter and reside in the microwells (Kumar et al. 2015). Kumar and colleagues used this setup to dock single yeast cells to perform high-throughput cytotoxicity assays with spatiotemporal resolution (docking efficiency 6%) (Kumar et al. 2015; Vriens et al. 2017). Digital microfluidic liquid handling is a versatile tool, as droplet actuation is flexible and highly reconfigurable. In contrast to CMF, DMF is a channel-free system thereby eliminating the risk of channel-clogging. Additionally, DMF platforms further reduce the required sample volume, reagent consumption, and dead volume as single droplets can be manipulated instead of continuous fluid flows (Kumar et al. 2015). Using DMF does not require pumps, valves, or mechanical mixers, as it mainly uses EWOD as a sole actuation force (Samiei et al. 2016).

326

J. Breukers et al.

Nevertheless, external electric modules are required for this setup, and docking efficiencies are rather limited, reaching single yeast cell docking efficiencies of up to 6% (Kumar et al. 2015). Also, the viability of the docked cells has to be taken into account. Moreover, the reusability of DMF devices remains a challenge due to biofouling of proteins and lipids on the hydrophobic surface, rendering the surface more hydrophilic. This might interfere with further droplet manipulations and can cause sample contamination (Choi et al. 2012; Samiei et al. 2016). So far, multiple approaches have been employed to overcome this issue, such as (i) the addition of amphiphilic additives, (ii) the use of plastic films instead of dielectric layers, (iii) controlling the solution pH and electrode polarity, (iv) introducing nanostructures to render the surfaces superhydrophobic, and (v) the use of water-immiscible oil as a filler material (Choi et al. 2012).

Single Cell Manipulation Microwell arrays have been integrated with many types of manipulation techniques that can be used (i) to specifically dock a target subpopulation of cells or (ii) to manipulate target cells for further analysis. In this section, different techniques for single cell manipulation that have been used in combination with microwell arrays will be discussed. For this purpose, the manipulation techniques were divided according to the externally applied forces, which are either optical, magnetic, mechanical, or electric.

Optical Manipulation In optical manipulation, light of a laser is used to apply piconewton-level forces on nano- to micrometer-sized particles, which allows contactless manipulation of single cells (Huang et al. 2014). Due to conservation of momentum, photons that are scattered by a particle exert a force on this particle in the direction of light propagation, known as the optical scattering force. Kovac and Voldman (2007) used this property of light by integrating a focused infrared laser with a microwell array-based microfluidic chip to levitate cells of interest from their wells for further collection and downstream analysis, without compromising cell viability. When this focused laser beam is combined with a microscope objective lens with high numerical aperture, a strong light gradient is formed which results in optical gradient forces that together with the scattering forces push the particle toward the focus of the beam (Gao et al. 2017). In 1987, Arthur Ashkin already described this phenomenon as optical tweezers (OT, Fig. 8a) (Ashkin and Dziedzic 1987). The integration of OT with a microwell array-based microfluidic chip was first described by Wang et al. (2013a). In their work, human embryonic stem cells and yeast cells were seeded in PDMS microwells, and, after washing the array, the cells of interest were levitated with OT from their microwell and deposited in a secondary microwell array for further analysis.

13

Integrated Microwell Array Technologies for Single Cell Analysis

327

Fig. 8 Single cell manipulation tools integrated with microwells. (a) Optical tweezers: a cell of interest is levitated and collected from the microwell by applying a highly focused laser beam. (b) Immunomagnetic sorting: cells of interest are sorted from the heterogeneous population by forming a cell-magnetic bead complex. (c) Micromanipulator: a cell of interest is collected from the microwell by generating negative pressure inside a glass capillary. (d) Optoelectronic tweezers: a laser beam projects on a photosensitive electrode, enabling localized dielectrophoresis to precisely place cells in a microwell

Optical manipulation concepts like OT are easily integrated with microwells by using a transparent material. However, because of the narrow beam waist of a focused laser, only a limited number of cells can be manipulated simultaneously. This can be resolved by generating holographic optical tweezers, in which a single input beam is converted into several beams by a computer-generated hologram. Hence, multiple tweezers are produced, therefore allowing the manipulation of multiple objects in a single setup (Liesener et al. 2000). Another disadvantage of OT is that high optical intensities are required to achieve a stable optical trap. As this

328

J. Breukers et al.

is associated with heat generated by the laser beam, interaction time of the laser with the cell should be limited to avoid photodamage. Lastly, optical systems require expensive optical equipment and skilled technicians for the optical alignment. Note that apart from integration of OT with microwells, OT have also shown great promise for single cell stretching and for manipulating intracellular structures (as reviewed in Huang et al. 2014).

Magnetic Manipulation In magnetic manipulation, a magnetic field gradient can be used to select and separate target cells that are coupled with magnetic particles such as magnetic beads (MBs). MBs vary in the nano- to micrometer range and can be functionalized with a vast spectrum of ligands, such as antibodies, aptamers, and oligonucleotides (Liu et al. 2009). To separate a specific cell subpopulation from a heterogeneous sample, MBs are typically functionalized with antibodies against cell surface epitopes to form cell-MB complexes. Kits for magnetic-activated cell sorting (MACS), providing antibody-labeled beads, are commercially available and widely used nowadays (Hu et al. 2016). In microfluidic systems, magnetism has been applied mainly for capturing rare cells, and a few studies demonstrate its compatibility with microwell arrays for single cell trapping. As an example, Huang et al. described a microwell arraybased microfluidic chip integrating a permanent magnet for on-chip immunomagnetic single cell trapping (Fig. 8b). By labeling cells with magnetic beads, human acute monocytic leukemia cells were captured with high purity (>99%) (Huang et al. 2018). In general, the integration of a magnetic sorting system in a microfluidic chip allows fast sample processing and cost-effectiveness, while using very low sample volumes. Potential drawbacks of this concept are that (i) the direct binding of MBs may lead to cell damage, (ii) beads tend to form clusters, and (iii) it is difficult to separate different subpopulations simultaneously.

Mechanical Manipulation Mechanical manipulation techniques are based on mechanical forces that involve physical contact with the target cell. As an example, micromanipulators have been highly used for microinjection of substances in single cells and aspiration of a cell of interest for downstream analysis. This tool is commercially available and consists of an L-shaped glass capillary that allows the capture of a cell of interest into the capillary by applying a negative pressure (Fig. 8c). Micromanipulator devices have been integrated with microwell arrays for collecting specific antibody-secreting cells for further single cell sequencing (Love et al. 2006; Jin et al. 2009). Apart from commercial micromanipulators, other approaches have been described in literature for the mechanical collection of target cells. Swennenhuis et al. (2014) fabricated a punch needle out of a nickel titanium wire to use as

13

Integrated Microwell Array Technologies for Single Cell Analysis

329

mechanical manipulation tool for the collection of cells of interest from a microwell array. In their setup, the needle is punched from the top through the bottom layer (silicon nitride membrane) of a microwell containing a cell of interest, causing the cell to fall into a 384-well polymerase chain reaction (PCR) plate positioned below the array. Welch et al. (2016) developed yet another mechanical manipulation technique for the collection of individual cells in microwells. In their work, a square-microwell array in PDMS was filled with magnetic polystyrene cubes on which the seeded cells could adhere. Cells of interest were then identified based on their phenotype, after which the bottom of the microwell was punched with a motorized microneedle to dislodge the magnetic polystyrene particle. The cells of interest, adhered to the dislodged magnetic polystyrene particle, could be collected and deposited in a PCR tube using a magnetic wand. Despite their relative simplicity, these mechanical manipulation techniques are not always compatible with the small microwell array dimensions due to the size limitation of commercial glass capillaries or microneedles. Furthermore, as all of these mechanical manipulation techniques are contact-based methods, physical stress can be induced leading to cell stress and/or cell death.

Electric Manipulation Electric manipulation techniques make use of forces exerted by an electric field to manipulate single cells. The technique DEP, which was described in sections “Docking Strategies: Manual Administration of Cells” and “Docking Strategies: Microfluidics-Assisted Administration of Cells” for single cell docking, can also be applied for targeted selection and sorting of subpopulations out of a heterogeneous population. For example, Morimoto et al. (2015) described a microwell array integrated with DEP for preferential entrapment of single circulating tumor cells (CTCs) in whole blood. Because CTCs are larger than blood cells, and the magnitude of the DEP force is related to the cube of the radius of the cell, CTCs are docked much more efficiently as compared to blood cells. Since CTCs are present in blood at very low concentrations, this method provides efficient enrichment and enables faster characterization of single CTCs. Furthermore, DEP has been integrated with microwells to achieve single cell pairing. In this approach, individual electrodes are implemented for each microwell, and the DEP force can be altered to have a microwell in “BLOCK” mode or in “LOAD” mode. When in “BLOCK” mode, a DEP force prevents cells from entering in the microwell, while in “LOAD” mode a DEP force allows cell access to the microwell and even aligns cells to the middle of the microwell. This alignment allows controlled cell pairing to study cell-cell interactions or cell aggregate formation (Bocchi et al. 2012). Another electric manipulation technique is optoelectronic tweezers (OET). OET is a hybrid technology that, by combining optical tweezers and DEP forces, addresses some of the drawbacks of both methods: it requires less optical power as

330

J. Breukers et al.

compared to optical tweezers, and no prefabricated electrode patterns are required as in DEP. In OET, virtual electrodes are generated by projecting a light pattern on a photoconductive surface (i.e., amorphous silicon substrate), allowing the interaction of cells through DEP (Fig. 8d) (Chiou et al. 2005). Compared to other electrical manipulation technologies, OET can be operated at a low magnification objective and requires a low-cost light source (Ohta et al. 2010). Ke et al. (2017) integrated OET with a PEG-coated microwell microfluidic chip to precisely bring natural killer cells and tumor cells in contact, allowing the real-time analysis of cell-cell contact for cancer immunotherapy studies.

Applications The variety of approaches described in previous sections are not just limited to proof of concepts. So far, many biological questions have been answered by implementing microwell array technology, of which a selection will be discussed in this section. For every application, a summary of the microwell fabrication, cell docking, and cell manipulation strategy can be found in Table 1.

Single Cell Drug Screening The heterogeneity in cell responses upon stimuli, such as drug treatments, can result in tolerant cells within a population which is of particular importance in view of treatment failure due to reinfections, pointing to the importance of single cell studies. Therefore, single cell assays can be used to study nonresponsive subpopulations upon treatment. Vriens et al. (2017) investigated the mechanism of action of the antimycotic amphotericin B (AmB) on Saccharomyces cerevisiae cells at a single cell level by seeding the cells in a Teflon microwell array using DMF. They found that different subpopulations exist with respect to AmB-induced superoxide radical production, based on timing and their intracellular levels. These superoxide radicals were important for AmB’s fungicidal action, whereas nitric oxide radicals mediated tolerance toward AmB. Hence, the fungicidal action of AmB could be increased by specifically blocking these tolerance mechanisms. When doing so, the human pathogens Candida albicans and Candida glabrata were also more susceptible to AmB in bulk, pointing to the generality of the single cell results obtained via an EWOD-based DMF platform. Huang et al. (2005) performed real-time observations of single cell apoptosis of HeLa-C3 cells, i.e., a cell type which can be used as a model for apoptosis studies, to monitor the cellular heterogeneity upon treatment with the anticancer drug doxorubicin. For this, Hela-C3 cells were manually brought onto the array and captured in PDMS microwells via centrifugation. Doxorubicin was found to induce apoptosis of HeLa-C3 cells in a dose-dependent manner, and the percentage of drug-resistant cells decreased with an increased concentration of doxorubicin. These results were

Single cell omics

Cone Shaped

Circular 20, 25, 30, or 35 μm diameter 15 μm depth Square 20x20 μm 50 μm depth

U937 cells and HEK 293 cells

HCC827-ER1 cells

Circular 20 μm diameter, 30 μm depth

Circular 50 μm diameter 58 μm depth Square 45x45 μm 45 μm depth

Neural stem cells

Primary human macrophages

Glioma neurospheres

Bottom 10 to 24 μm diameter Top 5 to 20 μm 50x50 μm 50 μm depth

Shape Size Circular 5.5 μm diameter 3 μm depth

Circulating tumor Square cells

HeLa cells and HeLa-C3 cells

Application Cell type Drug Saccharomyces screening cerevisiae cells

Microwells

PDMS

Manual

CMF

SU8 photoresist

PDMS

Manual

Manual

CMF

Polyacrylamide gel

PDMS

PDMS

Manual

Manual

Material Teflon

PDMS

Cell docking Administration strategy DMF

Centrifugation

DEP

Gravitation

Gravitation

Gravitation

Gravitation

Centrifugation

Docking strategy Gravitation, drag force, and surface tension

100%

95%

40–50%

N.A.

N.A.

N.A.

N.A.

N.A.

>50%

80%

Micromanipulator for cell retrieval

N.A.

Manipulation strategy N.A.

N.S.

90%

Docking efficiency 6%

Single cell genome and exome sequencing Single cell RNA sequencing Single cell RNA sequencing Single cell proteomics by Western blotting Single cell intracellular enzymatic activity Single cell transcriptomics and proteomics

Application Monitoring cell response upon treatment with amphotericin B Real-time observation of cellular apoptosis

Integrated Microwell Array Technologies for Single Cell Analysis (continued)

Park et al. (2016)

Kim et al. (2011b)

Hughes et al. (2014)

Gierahn et al. (2017)

Yuan and Sims (2016)

Lohr et al. (2014)

Reference Kumar et al. (2015) and Vriens et al. (2017) Huang et al. (2005)

Table 1 Overview of the different discussed microwell-based applications with information on microfabrication, cell docking, and manipulation strategy

13 331

Silicon

N.S. Membrane with inverted microwells is pressed down to encapsulate cells immobilized on a glass slide CMF

Human T cells and human hepatocarcinoma cells

N.S.

Gravitation

Manual

40 μm PDMS diameter 50 μm depth Circular Volume of ~20 PDMS pL

Ircular

Glioblastoma cells

N.S.

N.S.

Manual

Gravitation

Gravitation

N.S.

N.S.

N.S.

Docking efficiency 50–75%

PDMS

Manual

Gravitation

Gravitation

Gravitation

Docking strategy Gravitation

Square

50  50 μm 50 μm depth

Manual

Silicon

Circular 10 μm diameter 15 μm depth Circular 10 μm diameter 15 μm depth Square 50  50 μm 50 μm depth PDMS

Manual

PDMS

50  50 μm 50 μm depth

Square

Manual

Material PDMS

Size 50  50 μm 50 μm depth

Cell docking Administration strategy Manual

Shape Square

Microwells

Colorectal tumor cells

Human T cells

Human B cells

Human B cells

Application Cell type Hybridoma cells Detection of secreted products Human B cells

Table 1 (continued)

Microbeads to monitor cellular activity

N.A.

Micromanipulator for cell retrieval

N.A.

Micromanipulator for cell retrieval

Micromanipulator for cell retrieval

Micromanipulator for cell retrieval

Manipulation strategy Micromanipulator for cell retrieval

Detection of secreting cytokines and EVs

Application Detection of secreted antibodies Detection of secreted antibodies Detection of secreted antibodies Detection of secreted antibodies Detection of secreted cytokines Detection of secreted chemokines Detection of secreted EVs

Son et al. (2016)

Cai et al. (2018)

Adalsteinsson et al. (2013)

Han et al. (2011)

Tsuda et al. (2015)

Jin et al. (2009)

Tsioris et al. (2015)

Reference Love et al. (2006)

332 J. Breukers et al.

160  160 μm 160 μm depth Circular 100–130 μm diameter 50–80 μm depth Circular 100 μm diameter 50 μm depth Circular 14 μm diameter, 25 μm depth

Square

N.A. not applicable, N.S. not specified

Human embryonic stem cells Myeloma cells

Other Hematopoietic applications stem cells Hematopoietic stem cells

CMF Manual

CMF

CMF

PDMS PEG hydrogel

PDMS

SU-8

DEP

Gravitation

Gravitation

Gravitation

65–85%

N.S.

N.S.

10–30%

N.A.

Optical tweezers

Micropipette for cell retrieval Micromanipulator for cell retrieval

Cell pairing for cell fusion

Cell pairing for cell fusion

Cell proliferation Cell proliferation

Yoshimura et al. (2014)

Wang et al. (2013a)

Lecault et al. (2011) Roch et al. (2017)

13 Integrated Microwell Array Technologies for Single Cell Analysis 333

334

J. Breukers et al.

consistent with previous bulk experiments; however, heterogeneous apoptosis was observed among the population. Hence, this single cell assay is more informative as compared to the bulk assay, thereby allowing to identify drug-resistant subpopulations.

Single Cell Omics Observed heterogeneity in cell responses can often be explained by cell-to-cell differences in the genome, transcriptome, or proteome, which are studied using omics technologies. Therefore, single cell omics studies can provide novel insights in gene transcription and translation and how these processes are altered when the cell is in a diseased state or when it is treated with a drug. Single cell omics are often studied using droplet microfluidics, in which single cells are encapsulated inside water-in-oil droplets. Although this is a very promising technique, it lies beyond the scope of this book chapter, and more information can be found elsewhere (Ven et al. 2018). In recent years, microwells also aided in high-throughput single cell omics either by on-chip detection and collection of the desired cells for off-chip analysis or by performing the omics technique partially or even completely inside the microwells. Different omics levels can be reached using such single cell analyses. At first, genomics is considered as the study of the DNA containing the core information of a cell. For example, single cell genome information is particularly valuable for cancer research as it can monitor the progression and expansion of individual tumor cell clones. Specifically, the DNA content of CTCs is of high interest for evaluating disease prognosis and for developing effective and personalized cancer treatments. Since CTCs occur in the blood at very low frequencies (0–10 cells/mL), effective single cell identification and collection methods are required. Lohr et al. (2014) first enriched CTCs from blood using a conventional bulk technique and then manually seeded the cells in PDMS microwells by gravitation. Using stainings, CTCs were identified and collected with a micromanipulator for whole exome sequencing. In two prostate cancer patients, they could discover somatic single-nucleotide variants and mutations present early in the tumor evolution. Next to genomics, the transcriptomics field is of high interest since single cell mRNA expression profiles can unveil different cellular states within a population. For example, glioma neurospheres are often used as a model system for brain tumors as they match in genotypic and phenotypic features. Although the extent of the phenotypic heterogeneity is important when studying drug responses, it is still unknown at single cell level. Therefore, Yuan and Sims (2016) developed a single cell RNA sequencing platform in PDMS microwells using continuous flow for gravity-based docking of cells together with barcoded RNA capture beads. After flowing lysis buffer over the array, the microwells are sealed by a flow of oil, so the beads can capture the RNA from the cells that reside in the same microwell. Next, the beads are extracted from the chip to perform PCR. Using this method, the authors

13

Integrated Microwell Array Technologies for Single Cell Analysis

335

could obtain phenotypic subpopulations which were related to heterogeneity already observed in glioma tissue from patients. RNA profiles can also give important information on healthy and diseased behavior of single cells, as studied by Gierahn et al. (2017) using their own microwell-based single cell RNA sequencing platform. Here, barcoded beads and cells are manually added on top of the array and docked via gravitation. The PDMS microwells are then sealed using a semipermeable membrane which allows passage of lysis buffer, while retaining the RNA within the microwell, so the RNA can be captured by the bead. With this approach, shifts in RNA expression were observed between healthy cells and cells exposed to tuberculosis bacteria. More specifically, exposed cells showed enrichment of gene expression related to metabolism. The phenotype of a cell is ultimately determined by mRNA that is effectively translated into proteins, which can be studied by proteomics. By manual seeding of cells in microwells fabricated in polyacrylamide gel, Hughes et al. (2014) implemented single cell multiplex Western blotting of up to 11 proteins per cell. They discovered increased phosphorylation of signaling proteins in stem cells after stimulating them with a growth factor, which could be related to results obtained with conventional Western blotting. Their single cell approach exposed high variability in the cell responses that could not be identified using traditional bulk Western blotting. Kim et al. (2011b) used microwells integrated with DEP to seed human cell lines and monitor their variation in intracellular β-galactosidase expression. After cell lysis, a fluorogenic substrate for β-galactosidase was administered, and the microwells were sealed with a flexible membrane. They quantified the cell heterogeneity in β-galactosidase activity based on fluorescence intensity. Since there is not always a correlation between the DNA, mRNA, and protein levels within cells, multi-omics techniques are desired. Therefore, Park et al. (2016) enabled single cell transcriptomics and proteomics in one device. First, human cells were seeded in PDMS microwells by centrifugation. Expressed proteins were then identified by fluorescently labeled antibodies, whereafter cells were lysed and the target mRNA was amplified and tagged with a fluorescent probe using reverse transcription PCR. They observed decreased protein levels when a translational inhibitor was added and uniformly decreased mRNA and protein levels upon adding a transcriptional inhibitor. This proof-of-concept study indicates that the platform can be used to identify regulatory effects of molecules on transcription and translation.

Detection of Single Cell Secreted Products Monitoring secretory activities of single cells can provide new information on how individual cells respond to a stimulus, the way our immune system works, and what triggers the development of immune-related diseases. Using microwell arrays, the most studied entities secreted by single cells are antibodies, cytokines, and extracellular vesicles (EVs).

336

J. Breukers et al.

First, the study of epitope-specific antibodies helps us in understanding humoral immune responses against a pathogen or tumor cells and can thus provide information about disease severity of a patient. The detection and isolation of antigenspecific antibody producing cells play a key role in the development of better diagnostic strategies and the improvement of current therapeutics. In 2006, Love et al. developed a method known as microengraving to identify single hybridoma cells that secrete antibodies against a specific antigen. For this, single cells were seeded by gravity in PDMS microwells, which were sealed with a glass slide. This glass slide was functionalized with either (i) the antigen that captured the secreted antibodies, which could be detected with a fluorescently labeled secondary anti-IgG antibody, or (ii) an anti-IgG antibody that captured secreted antibodies, which could then be detected with a fluorescently labeled antigen. By tracing back the fluorescent spots on the glass slide to the microwell array, the specific antibody-secreting cells could be identified and collected with a micromanipulator (Love et al. 2006). Using this microengraving approach, Tsioris et al. (2015) studied B cells from patients with West Nile virus (WNV) infection. They collected B cells secreting antibodies against the virus for single cell RNA sequencing to unravel the sequence of the variable region of the antibodies and hereby characterized four new neutralizing antibodies with high potential for therapeutics of WNV infection. Jin et al. (2009) developed another method for the detection of antibody-secreting B cells, namely, immunospot array assay on a chip (ISAAC). Here, single B cells were docked by gravity in a silicon microwell array in which the surface around the microwell was functionalized with antibodies against human IgG to capture the antibodies secreted by the cell. They then applied fluorescently labeled hepatitis B antigen to the array to identify microwells containing B cells secreting antibodies specifically against hepatitis B, which could afterward be collected with a micromanipulator for single cell sequencing. Using the same approach, Tsuda et al. (2015) studied the reactivity of autoantibodies from rheumatoid arthritis (RA) patients. This study provided evidence that the generation of RA autoantibodies can be triggered by diet or previous infections. Cytokines are small proteins secreted by a broad range of cells (T cells, B cells, macrophages, etc.) that modulate cellular signaling by binding to a specific receptor. The secretion of these proteins is triggered by immune cell activation and responses, and their monitoring is crucial to understand how the immune system is modulated under inflammatory conditions. Via microengraving, Han et al. (2011) quantified the secretion rate of the cytokines IFN-γ, IL-2, and TNF-α by single human T cells over time and provided evidence about secretion dynamics related with the differentiation state of the cell. Later on, the same group described the secretion of chemokines, a subfamily of cytokines, from single cells isolated from colorectal tumor and provided a new understanding of the tumor microenvironment (Adalsteinsson et al. 2013). Extracellular vesicles are nano-sized particles that are naturally secreted by cells and function as secondary messengers. The study and evaluation of secreted EVs have captured attention because of, among others, their microRNA cargo, which has the potential to serve as a cancer biomarker and which has been correlated with

13

Integrated Microwell Array Technologies for Single Cell Analysis

337

tumor metastasis. To study EV secretion, Cai et al. (2018) described a platform that can capture either all EVs or specific EVs secreted by a single cell. First, glioblastoma cells are docked by centrifugation in a through-hole PDMS microwell array with a membrane filter as bottom. To capture all secreted EVs, the array is placed on top of another microwell array that collects all EVs that pass through the membrane. The microRNA content of these EVs was then determined using droplet digital PCR. To capture specific EVs, a plate functionalized with antibodies is placed on top of the array, on which a sandwich assay is afterward performed for EV quantification. Son et al. (2016) developed a microfluidic device for monitoring both cytokines and EV secretion from single cells. Their device consists of a glass slide patterned with cell capturing molecules and a microfluidic PDMS channel which contains inverted microwells. Cells are first immobilized on the glass after which detection antibodies and sensing beads linked with capture antibodies are flown through the channel. Then, the top of the microfluidic channel with microwells is pressed against the glass surface, encapsulating a cell, bead, and detection antibodies together in a microwell. Secreted products of interest could then be identified by an increase in fluorescent signal around the bead by means of a sandwich assay. In this study, the secretion rate of IFN-γ from single T cells and EV secretion from single cancer cells were evaluated.

Other Applications Apart from the previous applications, microwell arrays can also be used for a spectrum of other applications, including monitoring of cell growth and differentiation from single cells. Cell cultures generated from a single cell can provide information on cellular heterogeneity and understanding of stem cell differentiation. Lecault et al. (2011) described PDMS microwells for cell culture of hematopoietic stem cells (HSCs) from an individual cell. Upon medium exchange without disturbing the seeded cells, it was possible to monitor cell growth over 72 h. The influence of stem cell factor (which is a cytokine) concentration on HSCs survival was described. Roch et al. (2017) studied the differentiation of HCSs into multipotent progenitors by culturing single cells in PEG microwells. Moreover, they evaluated gene expression by collecting cells with a micromanipulator after 70 h of culturing. Cell pairing can be used to study among others somatic cell reprogramming, hybridoma engineering, and tissue regeneration. Typically, this is performed by random cell pairing with poor cell-cell contact and without control during cell fusion; hence, the pairing of rare cells, such as stem cells, is seldom achieved. To precisely control cell pairing, microfluidics-based pairing methods can be used, as these can facilitate the efficient capturing of cell pairs (Wang et al. 2013a). In the work of Wang et al. (2013a), cells were brought onto a PDMS microwell array via CMF, after which single cells were docked using gravitation. Holographic optical tweezers were used to pair human embryonic stem cells and Jurkat cells, after which a UV pulsed laser enabled fusion of the two cells. Their work demonstrated that the

338

J. Breukers et al.

fusion efficiency of homogenic pairs is higher than that of heterogenic pairs. In a second approach, CMF was used to sequentially administer differently stained mouse myeloma cells on top of a microwell array fabricated by negative photoresist. The cells were sequentially docked via DEP, resulting in a pairing efficiency of 50%. Pairs of differently stained myeloma cells were vertically aligned in this setup as this is advantageous for cell fusion using electric pulse fields, which can be generated by the same electrodes as used for DEP (Yoshimura et al. 2014).

Conclusion Due to the observed heterogeneity in cell populations, single cell analysis technologies have advanced considerably over the years and are now a well-developed field. The large variety of fabrication techniques for microwell arrays and the endless possibilities of their integration with different cell docking strategies and manipulation techniques make microwells attractive structures for single cell analysis. The value of microwell arrays has been demonstrated in numerous applications, such as the study of single cell kinetic responses to drug treatments, the detection of secreted products, the study of cellular interactions, and the implementation of different single cell omics. Thus, these integrated microwell array approaches hold great promise as life science research tools.

References Adalsteinsson VA, Tahirova N, Tallapragada N, Yao X, Campion L, Angelini A, Douce TB, Huang C, Bowman B, Williamson CA, Kwon DS, Wittrup KD, Love JC (2013) Single cells from human primary colorectal tumors exhibit polyfunctional heterogeneity in secretions of ELR+ CXC chemokines. Integr Biol 5(10):1272–1281 Ashkin A, Dziedzic JM (1987) Optical trapping and manipulation of viruses and bacteria. Science 235(4795):1517–1520 Banerjee I, Pangule RC, Kane RS (2011) Antifouling coatings: recent developments in the design of surfaces that prevent fouling by proteins, bacteria, and marine organisms. Adv Mater 23(6):690–618 Bocchi M, Rambelli L, Faenza A, Giulianelli L, Pecorari N, Duqi E, Gallois JC, Guerrieri R (2012) Inverted open microwells for cell trapping, cell aggregate formation and parallel recovery of live cells. Lab Chip 12(17):3168–3176 Bradshaw EM, Kent SC, Tripuraneni V, Orban T, Ploegh HL, Hafler DA, Love JC (2008) Concurrent detection of secreted products from human lymphocytes by microengraving: cytokines and antigen-reactive antibodies. Clin Immunol 129(1):10–18 Cai W, Chiu YJ, Ramakrishnan V, Tsai Y, Chen C, Lo YH (2018) A single-cell translocation and secretion assay (TransSeA). Lab Chip 18(20):3154–3162 Charnley M, Textor M, Khademhosseini A, Lutolf MP (2009) Integration column : microwell arrays for mammalian cell culture. Integr Biol 1(11–12):625–634 Chen S, Shamsi MH (2017) Biosensors-on-chip: a topical review. J Micromech Microeng 27:1–15 Chin VI, Taupin P, Sanga S, Scheel J, Gage FH, Bhatia SN (2004) Microfabricated platform for studying stem cell fates. Biotechnol Bioeng 88(3):399–315 Chiou PY, Ohta AT, Wu MC (2005) Massively parallel manipulation of single cells and microparticles using optical images. Nature 436(7049):370–372

13

Integrated Microwell Array Technologies for Single Cell Analysis

339

Choi K, Ng AH, Fobel R, Wheeler AR (2012) Digital microfluidics. Annu Rev Anal Chem 5:413–440 Decrop D, Pardon G, Brancato L, Kil D, Zandi Shafagh R, Kokalj T, Haraldsson T, Puers R, van der Wijngaart W, Lammertyn J (2017) Single-step imprinting of femtoliter microwell arrays allows digital bioassays with attomolar limit of detection. ACS Appl Mater Interfaces 9 (12):10418–10426 Farrell M, Beaudoin S (2012) Surface forces and protein adsorption on dextran- and polyethylene glycol-modified polydimethylsiloxane. Colloids Surf B: Biointerfaces 81(2):468–475 Gao D, Ding W, Nieto-Vesperinas M, Ding X, Rahman M, Zhang T, Lim C, Qiu CW (2017) Optical manipulation from the microscale to the nanoscale: fundamentals, advances and prospects. Light: Sci Appl 6(9):e17039 Gao D, Jin F, Zhou M, Jiang Y (2019) Recent advances in single cell manipulation and biochemical analysis on microfluidics. Analyst 144(3):766–781 Gierahn TM, Wadsworth MH, Hughes TK, Bryson BD, Butler A, Satija R, Fortune S, Love JC, Shalek AK (2017) Seq-well: portable, low-cost RNA sequencing of single cells at high throughput. Nat Methods 14(4):395–398 Guo DJ, Han HM, Wang J, Xiao SJ, Dai ZD (2007) Surface-hydrophilic and protein-resistant silicone elastomers prepared by hydrosilylation of vinyl poly(ethylene glycol) on hydrosilanespoly(dimethylsiloxane) surfaces. Colloids Surf A Physicochem Eng Asp 308(1–3):129–135 Guo P, Hall EW, Schirhagl R, Mukaibo H, Martin CR, Zare RN (2012) Microfluidic capture and release of bacteria in a conical nanopore array. Lab Chip 12(3):558–561 Han Q, Bagheri N, Bradshaw EM, Hafler DA, Lauffenburger DA, Love JC (2011) Polyfunctional responses by human T cells result from sequential release of cytokines. Proc Natl Acad Sci 109(5):1607–1612 Hu P, Zhang W, Xin H, Deng G (2016) Single-cell isolation and analysis. Front Cell Dev Biol 4(116):135–182 Huang L, Chen Y, Chen Y, Wu H (2005) Centrifugation-assisted single-cell trapping in a truncated cone-shaped microwell array chip for the real-time observation of cellular apoptosis. Anal Chem 87(24):12169–12176 Huang NT, Zhang HL, Chung MT, Seo JH, Kurabayashi K (2014) Recent advancements in optofluidics-based single-cell analysis: optical on-chip cellular manipulation, treatment, and property detection. Lab Chip 14(7):1230–1245 Huang NT, Hwong YJ, Lai RL (2018) A microfluidic microwell device for immunomagnetic single-cell trapping. Microfluid Nanofluid 22(2):1–8 Hughes AJ, Spelke DP, Xu Z, Kang CC, Schaffer DV, Herr AE (2014) Single-cell western blotting. Nat Methods 11(7):749–755 Inamdar NK, Griffith LG, Borenstein JT (2011) Transport and shear in a microfluidic membrane bilayer device for cell culture. Biomicrofluidics 5(2):1–15 Jen CP, Hsiao JH, Maslov NA (2012) Single-cell chemical lysis on microfluidic chips with arrays of microwells. Sensors 12(1):347–358 Jin A, Ozawa T, Tajir IK, Obata T, Kondo S, Kinoshita K, Kadowaki S, Takahashi K, Sugiyama T, Kishi H, Muraguchi A (2009) A rapid and efficient single-cell manipulation method for screening antigen-specific antibody-secreting cells from human peripheral blood. Nat Med 15(9):1088–1092 Kang DH, Kim HN, Kim P, Suh KY (2014) Poly(ethylene glycol) (PEG) microwells in microfluidics: fabrication methods and applications. Biochip J 8(4):241–253 Karp JM, Yeh J, Eng G, Fukuda J, Blumling J, Suh KY, Cheng J, Mahdavi A, Borenstein J, Langer R, Khademhosseini A (2007) Controlling size, shape and homogeneity of embryoid bodies using poly(ethylene glycol) microwells. Lab Chip 7(6):786–794 Ke LY, Kuo ZK, Chen YS, Yeh TY, Dong M, Tseng HW, Liu CH (2017) Cancer immunotherapy μ-environment LabChip: taking advantage of optoelectronic tweezers. Lab Chip 18(1):106–114 Kim SH, Yamamoto T, Fourmy D, Fujii T (2011a) An electroactive microwell array for trapping and lysing single-bacterial cells. Biomicrofluidics 5(2):1–7 Kim SH, Yamamoto T, Fourmy D, Fujii T (2011b) Electroactive microwell arrays for highly efficient single-cell trapping and analysis. Small 7(22):3239–3247

340

J. Breukers et al.

Kovac JR, Voldman J (2007) Intuitive, image-based cell sorting using optofluidic cell sorting. Anal Chem 79(24):9321–9330 Kumar PT, Vriens K, Cornaglia M, Gijs M, Kokalj T, Thevissen K, Geeraerd A, Cammue BP, Puers R, Lammertyn J (2015) Digital microfluidics for time-resolved cytotoxicity studies on single non-adherent yeast cells. Lab Chip 15(8):1852–1860 Lecault V, Vaninsberghe M, Sekulovic S, Knapp DJ, Wohrer S, Bowden W, Viel F, McLaughlin T, Jarandehei A, Miller M, Falconnet D, White AK, Kent DG, Copley MR, Taghipour F, Eaves CJ, Humphries RK, Piret JM, Hansen CL (2011) High-throughput analysis of single hematopoietic stem cell proliferation in microfluidic cell culture arrays. Nat Methods 8(7):581–586 Lee S, Vörös J (2005) An aqueous-based surface modification of poly(dimethylsiloxane) with poly (ethylene glycol) to prevent biofouling. Langmuir 21(25):11957–11962 Liesener J, Reicherter M, Haist T, Tiziani HJ (2000) Multi-functional optical tweezers using computer-generated holograms. Opt Commun 185(1–3):77–82 Lindström S, Andersson-Svahn H (2011) Miniaturization of biological assays – overview on microwell devices for single-cell analyses. Biochim Biophys Acta 1810(3):308–316 Liu C, Stakenborg T, Peeters S, Lagae L (2009) Cell manipulation with magnetic particles toward microfluidic cytometry. J Appl Phys 105(10):1–11 Lohr JG, Adalsteinsson VA, Cibulskis K, Choudhury AD, Rosenberg M, Cruz-Gordillo P, Francis JM, Zhang CZ, Shalek AK, Satija R, Trombetta JJ, Lu D, Tallapragada N, Tahirova N, Kim S, Blumenstiel B, Sougnez C, Lowe A, Wong B, Auclair D, Van Allen EM, Nakabayashi M, Lis RT, Lee GS, Li T, Chabot MS, Ly A, Taplin ME, Clancy TE, Loda M, Regev A, Meyerson M, Hahn WC, Kantoff PW, Golub TR, Getz G, Boehm JS, Love JC (2014) Whole exome sequencing in circulating tumour cells provides a window into prostate cancer. Nature 32(5):479–484 Love JC, Ronan JL, Grotenbreg GM, van der Veen AG, Ploegh HL (2006) A microengraving method for rapid selection of single cells producing antigen-specific antibodies. Nat Biotechnol 24(6):703–707 Morimoto A, Mogami T, Watanabe M, Iijima K, Akiyama Y, Katayama K, Futami T, Yamamoto N, Sawada T, Koizumi F, Koh Y (2015) High-density dielectrophoretic microwell array for detection, capture, and single-cell analysis of rare tumor cells in peripheral blood. PLoS One 10(6):e0130418 Murphy TW, Zhang Q, Naler LB, Ma S, Lu C (2017) Recent advances in the use of microfluidic technologies for single cell analysis. Analyst 143(1):60–80 Ng AH, Li BB, Chamberlain MD, Wheeler AR (2015) Digital microfluidic cell culture. Annu Rev Biomed Eng 17:91–112 Ohta AT, Chiou PY, Jamshidi A, Hsu HY, Valley JK, Neale SL, Wu MC (2010) Optoelectronic tweezers for the manipulation of cells, microparticles, and nanoparticles. In: Kim KY (ed) Recent optical and photonic technologies. INTECH, Croatia, pp 367–388 Oyama TG, Barba BJD, Hosaka Y, Taguchi M (2018) Single-step fabrication of polydimethylsiloxane microwell arrays with long-lasting hydrophilic inner surfaces. Appl Phys Lett 112:1–4 Park JY, Morgan M, Sachs AN, Samorezov J, Teller R, Shen Y, Pienta KJ, Takayama S (2010) Single cell trapping in larger microwells capable of supporting cell spreading and proliferation. Microfluid Nanofluid 8(2):263–268 Park SM, Lee JY, Hong S, Lee SH, Dimov IK, Lee H, Suh S, Pan Q, Li K, Wu AM, Mumenthaler SM, Mallick P, Lee LP (2016) Dual transcript and protein quantification in a massive single cell array. Lab Chip 16(19):3682–3688 Rettig JR, Folch A (2005) Large-scale single-cell trapping and imaging using microwell arrays. Anal Chem 77(17):5628–5634 Roch A, Giger S, Girotra M, Campos V, Vannini N, Naveiras O, Gobaa S, Lutolf MP (2017) Singlecell analyses identify bioengineered niches for enhanced maintenance of hematopoietic stem cells. Nat Commun 8(1):1–12

13

Integrated Microwell Array Technologies for Single Cell Analysis

341

Saharil F, Forsberg F, Liu Y, Bettotti P, Kumar N, Niklaus F, Haraldsson T, van der Wijngaart W, Gylfason KB (2013) Dry adhesive bonding of nanoporous inorganic membranes to microfluidic devices using the OSTE(+) dual-cure polymer. J Micromech Microeng 23:1–8 Samiei E, Tabrizian M, Hoorfar M (2016) A review of digital microfluidics as portable platforms for lab-on a-chip applications. Lab Chip 16(13):2376–2396 Shirasaki Y, Nakahara A, Shimura N, Yamagishi M, Mizuno J, Ohara O, Shoji S (2011) Single cell real time secretion assay using amorphous fluoropolymer microwell array. In: Xia S (ed) 2011 16th international solid-state sensors, actuators and microsystems conference. IEEE, Beijing, pp 755–758 Son KJ, Rahimian A, Shin DS, Siltanen C, Patel T, Revzin A (2016) Microfluidic compartments with sensing microbeads for dynamic monitoring of cytokine and exosome release from single cells. Analyst 141(2):679–688 Swennenhuis JF, Tibbe AG, Stevens M, Katika MR, van Dalum J, Tong HD, van Rijn CJ, Terstappen LW (2014) Self-seeding microwell chip for the isolation and characterization of single cells. Lab Chip 15(14):3039–3046 Tripodi L, Ven K, Kil D, Rutten I, Puers R, Lammertyn J (2018) Teflon-on-glass molding enables high-throughput fabrication of hydrophilic-in-hydrophobic microwells for bead-based digital bioassays. Materials 11(12):1–13 Tsioris K, Gupta NT, Ogunniyi AO, Zimnisky RM, Qian F, Yao Y, Wang X, Stern JN, Chari R, Briggs AW, Clouser CR, Vigneault F, Church GM, Garcia MN, Murray KO, Montgomery RR, Kleinstein SH, Love JC (2015) Neutralizing antibodies against West Nile virus identified directly from human B cells by single-cell analysis and next generation sequencing. Integr Biol 7(12):1587–1597 Tsuda R, Ozawa T, Kobayashi E, Hamana H, Taki H, Tobe K, Sugiyama E, Iwamoto M, Imura J, Kishi H, Muraguchi A (2015) Monoclonal antibody against citrullinated peptides obtained from rheumatoid arthritis patients reacts with numerous citrullinated microbial and food proteins. Arthritis Rheumatol 67(8):2020–2031 Ven K, Vanspauwen B, Pérez-Ruiz E, Leirs K, Decrop D, Gerstmans H, Spasic D, Lammertyn J (2018) Target confinement in small reaction volumes using microfluidic technologies: a smart approach for single-entity detection and analysis. ACS Sensors 3(2):264–284 Vriens K, Kumar PT, Struyfs C, Cools TL, Spincemaille P, Kokalj T, Sampaio-Marques B, Ludovico P, Lammertyn J, Cammue BPA, Thevissen K (2017) Increasing the fungicidal action of amphotericin B by inhibiting the nitric oxide-dependent tolerance pathway. Oxidative Med Cell Longev 2017:1–17 Wang X, Chen S, Chow YT, Kong CW, Li RA, Sun D (2013a) A microengineered cell fusion approach with combined optical tweezers and microwell array technologies. RSC Adv 3:23589–23595 Wang X, Gou X, Chen S, Yan X, Sun D (2013b) Cell manipulation tool with combined microwell array and optical tweezers for cell isolation and deposition. J Micromech Microeng 23(7):1–12 Welch JD, Williams LA, DiSalvo M, Brandt AT, Marayati R, Sims CE, Allbritton NL, Prins JF, Yeh JJ, Jones CD (2016) Selective single cell isolation for genomics using microraft arrays. Nucleic Acids Res 44(17):8292–8201 Witters D, Knez K, Ceyssens F, Puers R, Lammertyn J (2013) Digital microfluidics-enabled singlemolecule detection by printing and sealing single magnetic beads in femtoliter droplets. Lab Chip 13(11):2047–2054 Yoshimura Y, Tomita M, Mizutani F, Yasukawa T (2014) Cell pairing using microwell array electrodes based on dielectrophoresis. Anal Chem 86(14):6818–6822 Yuan J, Sims A (2016) An automated microwell platform for large-scale single cell RNA-seq. Sci Rep 6:1–10 Zhang P, Zhang J, Bian S, Chen Z, Hu Y, Hu R, Li J, Cheng Y, Zhang X, Zhou Y, Chen X, Liu P (2016) High-throughput superhydrophobic microwell arrays for investigating multifactorial stem cell niches. Lab Chip 16(16):2996–2906

Micro- and Nanopore Technologies for Single-Cell Analysis

14

Makusu Tsutsui, Takeshi Yanagida, Takashi Washio, and Tomoji Kawai

Contents Coulter Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A History Behind the Invention of Coulter Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Counting Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Analysis with Conventional Coulter Counters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solid-State Micro- and Nanopores: Structures and Fabrication Procedures . . . . . . . . . . . . . . . . . . . Electron and Ion Beam Milling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Dielectric Breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Glass Nanopipette . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lithographically Defined Cross-Membrane Nanopore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tunable Nanopore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Material Coating (ALD, SEM, Molecule Coating) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Focused Ion Beam Lithography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electron Beam Lithographed Micro-/Nanochannel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanoimprint Lithography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carbon Nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Solid-State Micro- and Nanopores: Functions Beyond Particle Sizing . . . . . . . . . . . . . . . . . . . . . . . . Single-Particle Shape Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Particle Surface Charge Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Intermolecular Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Analysis Using Micro- and Nanopores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volume Discrimination of Single Cells Using Advanced Multichannel Sensors . . . . . . . . . . Single-Cell Shape Analysis Using Low Aspect Ratio Pores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . AI-Driven Resistive Pulse Analysis for Discriminating Single Cells . . . . . . . . . . . . . . . . . . . . . .

344 344 345 347 349 350 352 352 352 353 353 353 354 354 354 355 355 356 357 357 357 359 362

M. Tsutsui (*) · T. Washio · T. Kawai The Institute of Scientific and Industrial Research, Osaka University, Ibaraki, Japan e-mail: [email protected]; [email protected]; [email protected] T. Yanagida Institute for Materials Chemistry and Engineering, Kyushu University, Kasuga, Japan e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_36

343

344

M. Tsutsui et al.

Biorecognition Pore Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Potential and Challenges for Total Cell Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

365 366 369 369

Abstract

Coulter counter is a sensor widely used for counting small objects in a sample suspension by simple impedance measurements. Advances in MEMS technology have led to resurgence of this powerful analytical tool for single-cell analyses. Extensive efforts have been devoted to make use of ultrathin membranes for achieving better spatial resolution of micropore sensors to discriminate cells from their multiple physical features including size, shape, and surface charge. Bio-recognition pores were also developed by functionalizing pore wall surface with molecular probes, which were found useful in rendering bio-specificity to the sensor for selectively detecting cells of certain biological structures. There are also increasing biological and medical interests in exploiting the sensing mechanism for studying functions of individual cells by detecting intracellular as well as extracellular particles and molecules from exosomes to polynucleotides. The chapter reviews the recent progress in this field. It begins with an explanation of a Coulter counter principle followed by a description of the historical use of the technique for counting cells. The following sections summarize reports on research and development of functional micropores and nanopores for single-cell analyses. There, cutting-edge studies that aim to incorporate machine learning in resistive pulse analyses are also presented. Finally, potential and challenge for a total cell analysis are discussed.

Coulter Principle The last decades have witnessed a remarkable progress in the field of sensors. Solid-state micro- and nanopores are one of the emerging sensors capable of probing single cells and other particles/molecules by a simple current measurement without labelling. The development has a history stretching back more than half a century when Coulter principle was invented. This section aims to first describe a brief introduction to the mechanism and applications of the classical yet powerful apparatus.

A History Behind the Invention of Coulter Principle The Coulter principle is a cytometry method applicable for counting cells in water by means of two-probe electric current measurements (Coulter 1953, 1956). The name came after the inventor Wallace H. Coulter. During the 1940s, he was working on a project for the US Department of Naval Research. Making paint with good adhesive strength was a purpose of the research where he needed to investigate and regulate

14

Micro- and Nanopore Technologies for Single-Cell Analysis

345

the size of particles included in the coating materials. Being an electrical engineer, he experimented new ways for counting small objects by a simple impedance measurement in a laboratory built in his house. It was during the day when he faced supply stop of paint samples did he thought to apply the inventions to blood. The motivation there was merely to obtain test fluid with similar viscosity to paint in a simple and cheap way, but it actually led to the great invention of Coulter principle, wherein he used a hole-punched cellophane as a sensor to measure his own blood. Noticing the laborious routine of hospital technicians to visually count the number of blood cells under optical microscope, he thought that his technique can revolutionize the blood test in those days by realizing a high-throughput automated counting of blood cells. It took, however, several years until 1953 for him to patent the innovative idea due to the fact that the sensor is just a hole. Later, he started a business by launching a company with Joseph Coulter, Jr., that sold Coulter counter over the world (Sack et al. 2008). Nowadays, the single-particle analyzer has found a vast range of applications in modern society not only as automated cell counters for clinical use but also in other industries including food, beverage, and cosmetics, as a method for quality control of the products.

Single-Cell Counting Mechanism The Coulter principle uses an apparatus consisting of a couple of chambers separated by a thin wall (Fig. 1). In the partition wall, there is a hole connecting the compartments filled with salt solution. By applying direct voltage Vb across the wall using a pair of electrochemically active electrodes, cations and anions move in the conduit of length L and cross-sectional area S that generates a constant ionic current via the reaction occurring at the liquid/electrode interfaces (this chapter will not focus on impedance flow cytometry that utilizes AC probes for single-cell analyses (Sun and Morgan 2010)). The amount of current Iion is given by Iion ¼ Vb/Rhole where the resistance at the hole Rhole can be approximately described as ρL/S with the solution resistivity ρ (the resistance at the bulk solution is ignored here, which is the case for

Fig. 1 Schematic illustration of a Coulter counter

346

M. Tsutsui et al.

Fig. 2 Single-particle detection by Coulter principle. Ionic current flowing through a fluidic channel (left) is temporarily blocked by an object during the translocation (middle). When it escapes from the conduit, the ionic current returns to the original level, thus creating a current signal called a resistive pulse

Coulter counters where the length of the channel is much longer than the diameter). Usually, NaCl and KCl are used as electrolyte whose concentration cion in water defines the resistivity as ρ ¼ (cioneμ)1 for the monovalent ions. cion needs to be adjusted to meet the current range of the measurement setup as well as for the conditions of analytes in terms of osmotic pressure, aggregation, etc. The salt solution is also often buffered to have a constant pH condition for keeping biosamples stable. Electrical detection is implemented by letting an object move from one side of the chamber to the other (Fig. 2). This can be done by leveraging the large electric field focused at the sensing zone, due to the large Rhole compared to the bulk together with relatively short channel length, which serves to induce enough electrostatic force on the surface-charged analyte to pull into and pass through the conduit (Wood et al. 2007). It may also be practiced to simply add hydropressure to transport target cells by fluid flow (Karuhn et al. 1975). The latter approach is effective when targets have little amount of surface charge for the electrophoretic transit or to have full quantity inspection of samples (Zhe et al. 2007). At the moment when the cell moves through the conduit, it tends to exclude ions inside there. Meanwhile, the ionic current one measures is solely dependent on Rhole. Therefore, it can be viewed that the cell residing in the channel virtually shrinks the space for the ion transport by its volume Vp. As a result, Iion drops to a certain level until the cell gets out of the conduit, giving rise to a so-called resistive pulse signal. It is noticeable that no matter the conductivity of the material, i.e., either plastic or metal, it acts like insulator blocking ion transport to cause ionic current decrease (Eckhoff 1969). In this way, one can count the number of objects passing through the channel by recording the resistive pulses (Coulter 1953, 1956). In addition, as the height of each signal is in proportion to Vp, due to that larger materials occupy more space in the channel, one can examine sizing of individual particles (Cooper and Parfitt 1967). Specifically, the pulse height Ip for a spherical object of diameter dp is given by the following empirical formula:

14

Micro- and Nanopore Technologies for Single-Cell Analysis

I p ¼ α β γ I 0 4dp 3 =3dch 2 Leff

347




where I0 is the ionic current flowing through the open cylindrical channel of diameter dch and length L. Leff is the effective length taking the so-called end effect into account that is given as Leff ¼ L + 0.785 dch (Blois and Bean 1970). α is a coefficient defined as follows, which is usually close to one unless dc >> dp (Qin et al. 2011):   α ¼ 1= 1  0:8ðdc =d s Þ3 β is another correction factor for the off-axis effect (Qin et al. 2011), which is influence of an inhomogeneous electric field distribution created by a particle moving along a passage closer to the channel wall on Iion. In case for analytes having non-spherical shape, the pulse height also deviates from the theoretical prediction whose contribution to the ion blockage enters as γ (Barreiros et al. 1996). While accurate sizing of analytes call for careful calibrations and fine corrections on top of a low-noise platform with good signal intensities, the analytical description manifests a high sensitivity of the pulse height on the particle size as Ip ~ dp3. Resistive pulse analysis can thus be used not only for counting particles but also for measuring their size variations. It resembles electron microscopy techniques in the sense that the spatial resolution is adjustable by the dimensions of the channel, whereas SEM and TEM have apertures, small holes like the structure of the electrical sensing zone of Coulter counters, to do a similar thing by limiting the electron beam passing through them. Furthermore, in light of the functions, they both can measure the size of materials. Meanwhile, the difference lies in that the former is viable for sizing objects in electrolyte solution, while the latter generally requires high vacuum (except the recently developed specially designed electron microscopy set up for sample imaging in liquid (De Yoreo and Sommerdijk 2016)). Coulter principle also allows one to miniaturize the sensor into portable size and weight expanding the places and methods of use. This is because the operation involves relatively small energy to put analytes into the sensing zone and detect the temporal change in the current. For instance, resistive pulse measurements involve recording of ionic current usually lower than 106 A under a biased voltage of less than 1 V that can be exhibited with a battery and some analog circuits for current amplification and AC/DC conversion. On the other hand, it is a formidable task to achieve for electron microscopes as they essentially need a high voltage source and vacuum pumps that cannot be easily downsized. These would be reasons why Coulter counter has been widely used in practice for decades and even continued to be studied in various fields of academics and industries.

Single-Cell Analysis with Conventional Coulter Counters Coulter counter is a versatile sensor viable to analyze samples of any size that can put through the sensing zone. In fact, miscellaneous analytes of size in

348

M. Tsutsui et al.

a range from millimeter to micrometer have been tested in the past including food (Morrison and Scott 1985), oil (Walstra and Oortwijn 1969), plankton (Kim and \Manden-Deuer 2013), particulate matter PM2.5 and PM10 (Cassee et al. 2002), pollen (Kelly et al. 2002), protein aggregates (Richmond et al. 1970), spermatozoa (Brotherton 1975), and so on. Nonetheless, the sensor is basically developed as automated cell analyzer. In particular, complete blood count (CBC) has been a major goal of Coulter counter to replace a part of blood tests that traditionally relied on time-consuming and prone-to-error manual counting under microscopes by technicians. Resistive pulse detection of red blood cells was found to be relatively easy by designing an optimal sensing zone for acquiring good signal-tonoise ratio for the disk-shaped bioparticles of diameter around 7 μm suspended in diluted blood as they are the major content of blood comprising about 40% of its volume (Balaz et al. 1960). This already gives wealth of information on human health. The number of pulses recorded directly provides the concentration of the cells in a specific volume of blood (depicted as RBC in millions per microliter). Moreover, the average volume of each cell (mean corpuscular volume, MCV), which is deduced from the pulse shapes under considerations of the various factors such as the off-axis effect and particle morphologies, allows one to estimate the specific volume of the red blood cells in blood. Meanwhile, standard blood test includes analyses of white blood cells and platelets (Beckman Coulter Diagnostics 2016). These cells are similar in size to red blood cells and so give resistive pulses of comparable height. What makes the analysis more difficult is the fact that the concentration of the two cells is normally scarce in blood, less than 1/1000 compared to that of red blood cells. Hence, the resistive pulse counts will be easily buried in the broad distributions of red blood cells (Beckman Coulter Diagnostics 2016). It therefore calls for additional pretreatments to achieve CBC by resistive pulse detections. Yet, white blood cell counting was found to be feasible by eliminating only red blood cell from samples through adding lysing solution and directly measure the hemolyzed blood specimen with Coulter counters. Although cell debris can be recorded together, they could be discerned from white blood cells as the cellular size is slightly larger than red blood cells. In contrast, reliable counting of platelets was more difficult due to several reasons besides the low concentration circumstances including the small size (2–3 μm) that limits the signal-to-noise ratio in whole blood tests along with inevitable interference with small cell debris (Bull et al. 1965). To overcome the issues, a protocol was developed to statistically deduce the number of platelets from the pulse height histogram profiles. Today’s Coulter counter is capable of providing reliable CBC within less than a minute per individual (Beckman Coulter Diagnostics 2016). In addition to blood tests, Coulter principle has been widely studied for variable range of single-cell analyses (Vembadi et al. 2019). Cell cycle is a fundamental process in organisms wherein Coulter counter has been utilized to investigate the size checkpoint through evaluating a correlation between the time course change in individual cells and their timing for division (Tzur et al. 2009) that offered profound insight into intercellular interactions (Conlon and Raff 2003). Besides the importance in biology, such ability is useful in medicine in particular for studying

14

Micro- and Nanopore Technologies for Single-Cell Analysis

349

development and causes of cancer by giving insight into the abnormal cell growth and division processes. The sensor sensitivity to particle sizes has also found potential applications in screening of pathogenic cells like bacteria (Allman et al. 1992). Moreover, the applicability was demonstrated to be not limited to cell sizing but also for discriminating dead and live cells (Mernier et al. 2009).

Solid-State Micro- and Nanopores: Structures and Fabrication Procedures Invention of Coulter principle and intensive research on related sensor devices have led to development of practical automated cell counter for hematology analysis. Last decades have witnessed rejuvenation of this well-established single-particle analyzer (Luo et al. 2014; Howorka and Siwy 2009; Henriquez et al. 2004). Owing to the advanced MEMS (microelectromechanical systems) and NEMS (nanoelectromechanical systems) technologies, extensive efforts have recently been devoted to build novel fluidic channel architectures aiming not only to render improved particle sizing performance but also to realize sensor capabilities beyond the size discrimination. In addition, interest and demand for applying the principle to detect analytes smaller than cells such as virus, extracellular vesicle, protein, and genome have led to a number of studies on nanoscale versions of Coulter counter called nanopore (Luo et al. 2014; Howorka and Siwy 2009; Henriquez et al. 2004; Saleh and Sohn 2003). The advanced sensor structures can be classified into two types: cross-plane and in-plane fluidic channels (Fig. 3). The former is a hole penetrated in a thin membrane suspended on a dielectric substrate, which is called a solid-state nanopore. It mimics the motif of nanoscale pores ubiquitous in biological systems playing important roles in traffic control of solutes at the cell membrane for regulating the intracellular conditions vital to maintain the life of cells. The original idea was to measure DNA with a protein nanopore via Coulter principle wherein the long biopolymer was envisaged to be electrophoretically pulled into the nanoscale sensing zone defined by the molecular structure of particular proteins suspended in a lipid membrane (Kasianowicz et al. 1996). Nowadays, it in fact became a real sensor that allows single-molecule sequencing of polynucleotides (Deamer 2016). Several research groups have sought for a way to device a similar sensor architecture with all-solid

Fig. 3 Two types of pore sensor structures: in-plane fluidic channels (left) and cross-plane analog consisting of a hole penetrated in a thin solid membrane

350

M. Tsutsui et al.

components by using nanotechnology (Li et al. 2001; Storm et al. 2003). Advantage in their concept is that while the dimensions of biological nanopore is generally at single nanometer scale and cannot be easily changed due to the fixed molecular structure, the size of solid-state nanopores can be fabricated to essentially any size from micrometer to nanometer scales by exploiting the advanced nanotechnology. Needless to say, it has also been suggested as promising strategy for mass production as long as the relevant fabrication processes are made compatible to modern semiconductor technologies. Originated from the research on nanopore sequencing, Coulter principle has now been widely applied for analyzing not only cells but also intracellular materials together with a variety of new procedures to sculpt a hole in a thin membrane. The in-plane structure consisting of conduits lithographed on an insulating substrate has also been explored for single-particle impedance sensors. Leveraging the high resolution of today’s lithography techniques, various kinds of channel patterns have been tested whose applications being not limited to the conventional cross-channel resistive pulse measurements but also extended to other approaches such as reactance analyses and four-probe designs for better discriminating cells (Vembadi et al. 2019; Vaclavek et al. 2018). They are compatible with wellestablished separation and amplification technologies in micro total analysis systems (Kovarik et al. 2012) and hence feasible to be developed as practical cell analyzers. There are actually a variety of methods reported to date for forming a cross-plane as well as an in-plane channel (Table 1). Some of the fabrication procedures are summarized in the following.

Electron and Ion Beam Milling As explained in the preceding section, the concept of solid-state nanopore aimed to fabricate single-nanometer-sized hole in a thin membrane for sensing polynucleotides. Since the size far exceeds the resolution power of lithography, the researchers pursued different approaches. Electron and ion beam milling was one of the methods proven reliable to form a DNA-sized hole in a dielectric membrane (Li et al. 2001; Storm et al. 2003). A silicon wafer is usually used as a substrate. On the substrate, a membrane made of Si3N4 is created by partially dissolving the Si layer by wet-etching in KOH solution. Beam of ions or electrons is then focused on the membrane in ultrahigh vacuum. Meanwhile, a detector was located at the back side of the substrate, which is designed to detect current under the beam irradiation. With this setup, one can find the moment when a hole is penetrated by the beam irradiation by monitoring the current at the detector. Halting the beam sculpting thereafter can impede the hole from being enlarged. By a simple feedback control, it was actually demonstrated as a reliable way to create a nanopore of diameter ranging from sub-nm to several nm. Most of the nanopore studies on translocation dynamics of DNA use this method especially for sculpting a pore in monoatomic thin membrane made of 2D materials such as graphene (Heerema and Dekker 2016), boron nitride (Gilbert et al. 2017), and molybdenum disulfide (Feng et al. 2015). Although this

14

Micro- and Nanopore Technologies for Single-Cell Analysis

351

Table 1 Fabrication procedures for solid-state micro- and nanopore structures

approach cannot be of use for analyzing cell, it can be viewed as an analytical tool viable for an ultimate way of studying single cell at genomic level, although it needs to wait for the technology to be applicable for decoding DNA stored in each cell without amplifications by polymerase chain reaction (PCR).

352

M. Tsutsui et al.

Dielectric Breakdown When excessively high electric voltage is applied across a thin insulator, the associated large electric field triggers catastrophic fracture at intrinsic defects. This phenomenon called dielectric breakdown was found to take place in the membrane of a nanopore device during the ionic current measurements (Briggs et al. 2014; Yanagi et al. 2014). This is because of the fact that it employs very thin dielectrics that bring discharge voltage large enough to cause breakdown. When it happened, small pinholes were created that serve to add leakage current through the membrane or worse malfunction the DNA-sized nanopore. While the phenomenon was naturally considered as a detrimental feature, a couple of research groups discovered a way to leverage it for creating a nanopore. The concept was to regulate the stochastic breakdown process through fine feedback control of the voltage stress with respect to the cross-membrane current, analogous to the procedure used in electromigration break junction method in the field of singlemolecule electronics. Although it is still difficult to manage precise control of the pore size, the method was demonstrated to be able to produce nanopores of a vast range of diameters from 100 nm to subnanometers. While the beam sculpting requires large and expensive facilities, the dielectric breakdown approach can be implemented only with a setup for the ionic current measurement. The simple and feasible mechanism recently attracted many researchers to develop related nanopore fabrication techniques that incorporate atomic force microscopy (Zhang et al. 2019) and laser irradiation (Ying et al. 2018).

Glass Nanopipette Carefully pulling and deforming acid-cleaned quartz capillaries, one can create a pipette of nanoscale opening (Morris et al. 2010). Choice of glass materials with appropriate softness under laser-heated conditions is important to have small diameter with a low-angle taper shape at the apex. The dimensions of capillaries also affect the quality of nanopipette where the ratio between the outer and inner diameter should be made large to ensure sharp tips (as long as the wall thinness does not cause significant inhomogeneous heating that would result in irregular-shaped pipette). There are commercial devices available to program control the complicated fabrication schemes (Morris et al. 2010).

Lithographically Defined Cross-Membrane Nanopore Electron beam lithography is a standard technique defining the resolution power of nanofabrication processes. A state-of-the-art machine that can apply high acceleration voltage to create ultra-narrow electron beam can delineate single-nanometersized patterns (Manfrinato et al. 2013). A pore is usually drilled by reactive ion etching of a solid membrane using a resist layer after lithography and development

14

Micro- and Nanopore Technologies for Single-Cell Analysis

353

as a mask. Although the lower limit of the nanopore diameter reported to date is around several tens of nanometers (Veschueren et al. 2018), which is presumably due in part to increasing difficulty in letting gas flow into the narrow hole in the resist for sculpting a small pore with the dry etching, it has many advantages including the fact that it can be utilized to fabricate a pore of vast range of size from the 10 nanometer scale up to submillimeter.

Tunable Nanopore There are not so many procedures that can finely tune the pore size after fabrications. Tunable nanopore is one of such techniques (Willmott et al. 2010). The sensor structure is constructed with a membrane made of elastic polymer wherein a pore is drilled using a needle. What is special in this apparatus is that the structure holding the membrane is designed to be deformed mechanically by pulling it in a two-dimensional fashion. The resulting elastic deformation allows arbitral expansion/shrinking of the pore. Calibration protocol allows to characterize the size of the pore after the mechanical manipulations by measuring and analyzing the open pore cross-membrane ionic current as well as the resistive pulses obtained for standard synthetic particles of known physical properties. The device is commercialized as name of q-Nano by Izon Science Ltd. Although requiring manual skill for use, it can offer various sizes of pores down to several tens of nanometer in diameter for sensing analytes of concern.

Material Coating (ALD, SEM, Molecule Coating) Coating wall surface of a pore was proven useful in finely modifying the diameter. Scanned beam in scanning electron microscopes can serve to shrink the size of pores through deposition of amorphous carbon compounds and observe in situ their actual diameter (Prabhu et al. 2011). Transmission electron microscopy is also capable of fine-tuning the pore dimensions, which can not only shrink but also enlarge the channel via electron beam-induced softening of a membrane material (Storm et al. 2003). More straightforwardly, materials can be coated on the channel wall surface with sputtering or atomic-layer deposition (Chen et al. 2004). Similar things can be implemented by forming self-assembled monolayer of organic molecules (Meng 2012). These approaches are useful not only in adjusting the pore size but also for tailoring the surface properties at the wall that govern electroosmotic flow and analyte-pore interactions.

Focused Ion Beam Lithography While the above procedures are for creating a cross-plane pore architecture, there are also several ways to fabricate a similar structure on surface of substrate (Haywood

354

M. Tsutsui et al.

et al. 2015). Focused ion beam is widely utilized to directly dig a micro- and nanoscale groove on Si wafer (Harms et al. 2015). Sealing from the top by bonding adhesive polymer or anodic bonding of glass plate, the confined space of predefined dimensions can be used as a conduit for single-particle translocation to resistive pulse detections by Coulter principle.

Electron Beam Lithographed Micro-/Nanochannel Similarly, optical and electron beam lithography can be exploited to pattern fluidic channels on flat surface (Yukimoto et al. 2013). Analogous to the fabrication process of a nanopore, the residual resist after lithography is utilized as a mask to sculpt a groove by isotropic reactive ion etching. Both the electron and focused ion beam lithography methods allow fabrication of fine channels of arbitral 2D patterns.

Nanoimprint Lithography Nanoimprinting has emerged as more practically feasible way to prepare micro- and nanochannel sensors (Liang et al. 2007). In general, hard mold consisting of a protrusion pattern lithographed on flat substrate such as Si wafer is mechanically pressed on a photosensitive material. Subsequently, the imprinted layer is hardened by UV irradiation. After lifting off the mold, one can obtain fluidic channels for the resistive pulse measurements. While this requires post channel sealing, the imprinting approach can also be implemented to prepare sealed channels by designing the setup to allow a space between the mold and the soft layer (Guo et al. 2004). The resolution was reported to be better than 10 nm and thus superior to that of photolithography while slightly deficient compared to electron beam lithography process. The advantage of this method is that it enables to pattern fine structures at large area by simple mechanical manipulations and thus an inexpensive way of fabricating practical fluidic channel sensors.

Carbon Nanotubes Carbon nanotube is one-dimensional all-carbon structure having a nanoscale hole that can act as a conduit for mass transport. The concept has been demonstrated by fabricating microchannels at both sides of a straight single-walled nanotube on an oxidized Si wafer (Liu et al. 2010). By opening the cap through oxygen plasma etching, the hollow space was actually observed to serve as a channel for not only fluid flow but also DNA transport. In this way, one can prepare nanochannels with well-defined diameter. The system also attracted attention from a viewpoint that the inner surface of the nanotubes is hydrophobic in nature that led to many anomalous fluidic phenomena (Guo et al. 2015).

14

Micro- and Nanopore Technologies for Single-Cell Analysis

355

Fig. 4 Schematic models depicting various functions of advanced micro- and nanopore sensors. (a) Surface charge density of a single particle can be estimated by estimating the time of flight across a pore from the measured resistive pulse width. (b) Single-particle shape analysis using a low-aspect-ratio pore. Arrows indicate passage for ions to move through the channel partially occluded by an irregular-shaped object. The resistive pulse waveform reflects the fine morphology of the particle due to the thin channel structure. (c) Biorecognition solid-state pore sensors. Specific particles tend to be temporarily trapped on the molecular-probe-decorated channel wall surface via intermolecular interactions. The associated resistive pulses show a wide variation in the width reflecting the affinity of the recognition molecules to the analytes passed through the conduit

Solid-State Micro- and Nanopores: Functions Beyond Particle Sizing Many types of micro- and nanopores have been tested to expand the ability of Coulter counters. Various novel functions beyond particle sizing have been realized by the emerging sensors, though the operation principle is basically the same as that of Coulter principle, i.e., two-probe ionic current measurements (Fig. 4). This section is devoted to describe the emerging sensor capabilities of the advanced pore devices.

Single-Particle Shape Analysis One of the unique features in cross-plane pore sensors is that the length of the conduit can be easily made short. Usually, several tens of nanometer-thick Si3N4

356

M. Tsutsui et al.

layer is used as a membrane. Even thinner structure is available by partially thinning it via dry etching (Wanunu et al. 2010) or utilizing 2D materials like graphene that offers ultrathin membranes of thickness down to subnanometer scale (Heerema and Dekker 2016). A pore sculpted in the thin membrane eventually possesses unusually low thickness-to-diameter aspect ratio structure. Ion transport in such a shallow channel tends to be quite different from the conventional Coulter counter designed to have relatively high aspect ratio geometries. For example, consider a disk-like pore of diameter dpore in a thin membrane of thickness t with dpore >> t. The resistance inside the pore Rpore is given by Rpore ¼ ρt/πdpore2 which quickly becomes low as the channel diameter is made larger than the membrane thickness. Meanwhile, Maxwell’s theory predicts resistance outside the channel, which is called access resistance Racc, to be described as Racc ¼ ρ/dpore (Hall 1975). As denoted by the formula, the relative importance of the two resistances depends on the aspect ratio structure of the pore (Garaj et al. 2010; Tsutsui et al. 2012). In conventional Coulter counters, t > dpore so that Rpore >> Racc to make the resistive pulses sensitive to the difference in the volume of analytes. On contrary, Racc tends to dominate the ionic current in low aspect ratio pore sensors. This simple analytical expression has proven to be valid in micro- and nanopores of vast range of dimensions: 1 μm to 1 nm for both dpore and t. One of the outcome of the shallow pore sensors is that the resistive pulses become sensitive to the local shape of particles. It can be easily understood by imagining translocation of an irregular-shaped object that causes increase (decrease) in Rpore at the moment when the thick (thin) part resides in the conduit. Meanwhile, the profound role of Racc makes the sensing zone broader, making the ionic current sensitive to dynamic motions of particles at the orifice (Tsutsui et al. 2016), which is another distinct characteristic of the low aspect ratio pores. These sensor capabilities were found helpful in discriminating cells of similar size, which is a thing that cannot be implemented with conventional Coulter counters.

Single-Particle Surface Charge Measurements As described above, resistive pulse lineshapes contain wealth of information concerning an object that passes through a pore. This is also a case for an electrophoretically driven particle in a pore. When a charged particle feels the focused electric field at the mouth of a channel, it enters there causing the ionic current to drop by certain value. The current keeps at the level until the object passes through and escape from the conduit. As such, the width of the pulse td denotes the time of flight of individual analytes to move by the distance Lch defined by the geometry of the channel. More specifically, it defines the mean translocation velocity vmean and the mobility μ as Lch/td and @vmean/@E, respectively, where E is the electric field distribution along the channel. On the other hand, electrokinetics of the particle is given by the balance between the electrophoretic and hydrodynamic drag forces in a solution of viscosity η and permittivity esol that permits a rough estimation of μ as μ ¼ esolζ / η where ζ is the particle’s zeta potential. Experimentally, meanwhile, μ can be calculated from td by assuming linear drop of the electrical potential in the pore

14

Micro- and Nanopore Technologies for Single-Cell Analysis

357

and hence constant E amounting Vb/Lch under the applied voltage Vb. As a result, one can deduce the surface charge density of each particle from the deduced ζ (Arjmandiet al. 2012; Vogel et al. 2017).

Intermolecular Interactions Biological nanopore exploits intermolecular interactions inside the protein pore for regulating the cross-membrane mass transport (Ying et al. 2014; Haque et al. 2013). In solid-state micro- and nanopores, objects also have a chance to interact with the channel wall surface. The stochastic nature of the interaction-involved translocation is, however, usually nothing but detrimental causing wide variation in the resistive pulse widths whereby veiling features intrinsic to analytes in the resistive pulse patterns (Garaj et al. 2010). To control and make use of the pore interactions, attempts have been made to decorate functional molecules on the solid wall (Hou et al. 2011). Simple carbon chains terminated with functional groups with various isoelectric points offer a means for tailoring the surface charge states (Siwy et al. 2004). Employing molecular probes with specific affinity to analytes of concern, one can also in principle render biorecognition ability to the sensor, wherein resistive pulses for particular molecules and particles tend to show peculiarly wide pulse signals (Iqbal et al. 2007). This device concept has a potential of being applied for single-molecule immunosensors.

Single-Cell Analysis Using Micro- and Nanopores The advanced solid-state micro- and nanopores led to novel sensor capabilities to measure not only the volume of analytes but also their shape, zeta potential, and even intermolecular interactions. As described in the following, they were proven useful not only for analyzing cells at better sensitivity but also as a useful platform for detecting various substances encapsulated in the cell membrane such as vesicles, proteins, and DNA (Fig. 5).

Volume Discrimination of Single Cells Using Advanced Multichannel Sensors In the past decades, there have been continuing efforts to improve the performance of Coulter counters to discriminate cells by their size. The sensitivity of the blockade current, or equivalently resistive pulse height, to the volume of objects comes at a price of limited size range of analytes measurable with a single channel: particles that are one order of magnitude smaller in the diameter will give three orders of magnitude smaller resistive pulses. This has already been a crucial issue in counting cells in blood samples with the automated counters where the 2 μm-sized platelets needed to be detected together with 7 μm-sized red blood cells (Beckman Coulter

358

M. Tsutsui et al.

Fig. 5 Single-cell analysis using solid-state micro- and nanopores. Images not to scale for the sake of clarity

Diagnostics 2016). Advanced fluidic channel structures as well as electrical circuits have been tested in order to expand the size range of detectable analytes by resistive pulse measurements. Transverse ion transport approach has been one of the concepts to achieve better sensor spatial resolution as well as single-to-noise ratio. It uses a long fluidic channel having narrower branches in the middle. While cells move along the long channel, the ionic current through the branch channels is measured. Analogous to Coulter principle, the existence of an object blocks the transverse ion transport. Yasaki et al. (2017) applied the above principle for detecting single cells at better volume sensitivity. Their sensor consisted of a micrometer-scale conduit on glass substrate with two side channels for the transverse ionic current measurements. Moreover, a bridge circuit was utilized to detect weak resistive pulses under relatively large electrophoretic voltage applied across the main channel. This ingenious

14

Micro- and Nanopore Technologies for Single-Cell Analysis

359

combination enables effective mass transport in the inlet toward the sensing zone for high-throughput detections of analytes without any fluid pumping incorporated. The analog circuit also provided low noise platform suitable for detecting small signals that would be immersed in noise in commonly used current amplifier-based setup. The length of the channel and the distance between the branch channels were also adjusted to allow long-enough translocation time for reliable resistive pulse measurements under the use of a low-pass filter for further suppression of the current noise. They demonstrated resistive pulse detections of various kinds of cells from bacteria to cancer cells. The method also allowed resistive pulse detections of 0.2 μm-sized particles flowing through a microchannel of 2.0 μm (height)  2.0 μm (width)  14 μm (length), which corresponds to a remarkable sensitivity of 0.01% of the channel volume. Liu et al. (2017) exploited similar multichannel systems to achieve better signalto-noise ratio in resistive pulse detections. Their approach was to cancel the noise by subtracting the ionic current flowing through equi-sized two branch channels, which is akin to a four-point probe method. By further incorporating an electrokinetic means to control the fluid flow, particle detection sensitivity of 0.004% was demonstrated. Even higher sensitivity of 0.0004% was reported by Wu et al. (2008) through employing a two-stage differential amplification setup that served to cancel the noise and to amplify the signal amplitudes. On-chip integrated in a multichannel form (Zhe et al. 2007) with additional microfluidic structures for separation and amplification (Fraikin et al. 2011), these advanced impedance sensors are expected to be used as useful automated label-free single-cell analyzers.

Single-Cell Shape Analysis Using Low Aspect Ratio Pores Considering the ion exclusion mechanism of Coulter principle, spatial resolution of Coulter counter is naturally anticipated to be enhanced by making the length of the sensing zone short. In a conventional long channel, the amount of ion blockage by an object, no matter whether it has irregular shape, is constant during the translocation (except when the motions involve rotation and/or radial displacements that can cause fluctuations in the blockade current by the varying shape and off-axis effects). In sharp contrast, when an analyte passes through a shallow pore having diameter much larger than the length, the cross-membrane ionic current is expected to trace its fine shape. For instance, at the moment when a thick (thin) portion of the body resides in the short conduit, it narrows (expands) the space for the ion transport that leads to larger (smaller) ionic current drops. The eventual resistive pulse lineshape renders a corrugation reflecting the morphology of the particle. The envisaged sensor capability can be regarded as microscopic version of the computed tomography scans in hospital where a 3D picture of inner human body is created in invasive fashion by assembling 2D X-ray images taken stepwise as the patient is moved through the sensing zone. Although interesting, the concept of using a low aspect ratio conduit for resistive pulse analyses has long been remained unexplored. This was presumably due to the

360

M. Tsutsui et al.

anticipated degradation of signal-to-noise ratio along with complication of signal waveform analyses as the intimate roles of access resistance on the ionic current blockade hinder straightforward interpretation of resistive pulses; they cannot be simply characterized by the height as is done in Coulter counters, but finer features need to be extracted and compared to examine particle identification. On contrary, many studies have recently been reported on ion transport through low aspect ratio channels. The research motivation was again stemmed from the field of nanopore sequencing because of the fact that the channel structure for genome sequencing is preferred to have a low aspect ratio geometry as one needs to shrink the length of the channel to be of subnanometer scale so as to identify each nucleotide chained with spacing of as small as 0.3 nm by the resistive pulse analysis, while at the same time, the diameter is required to be around 1 nm to give enough space for the biopolymer to go through by electrophoresis. It was in fact suggested experimentally on graphene nanopores (Garaj et al. 2010) that the access resistance dominates the cross-pore ionic conductance when dpore >> t by elucidating the open-pore conductance to increase linearly with dpore instead of dpore2 dependence in conventional high aspect ratio channels. The same feature was also found in micropores (Tsutsui et al. 2012). Later, detailed analytical expression of the access resistance contributions on the net pore conductance was provided by Kowalczyk and coworkers (2011). All these studies are consistent to validate the applicability of a simple serial resistance model to describe the open-pore ion transport through a low aspect ratio pore: i.e., the total resistance R to be given as summation of the resistance inside Rpore and outside the pore Racc as R ¼ Racc + Rpore. In contrast to the rather straightforward interpretation allowed for the open-pore ion transport characteristics, how the ionic current changes during translocation of analytes through a low aspect ratio pore is elucidated to be quite complicated. For instance, because of the pronounced role of the access resistance, an object starts to affect the ionic current even when it is not yet inside the channel but just being close to the orifice. The time-course change in the current thereof depends crucially on their dynamic motions, which in other words indicates that the sensor has broad sensing zone extending by a distance longer than the channel diameter (Tsutsui et al. 2016). Simulations also predicted strong sensitivity of the ionic current in a low aspect ratio pore on the surface charge of analytes that may give rise to current enhancement at a certain moment of the translocation (Wang et al. 2014). The peculiar ion blockade behaviors are ascribed to the relatively strong electric field outside the pore, which is obvious from the fact that the access resistance dominates the open-pore conductance. In the course of capture-to-translocation motions of analytes, this electric field tends to be disturbed in a complicated manner in response to their positions and so does the associated ionic current change. Due to the complex physics involved, a finite element method is commonly used under COMSOL Multiphysics that numerically solves Navier-Stokes equations for the fluid flow and Poisson-Nernst-Planck equations for the motions of ions (Garaj et al. 2010; Tsutsui et al. 2016; Rampfer et al. 2016). The shape-sensing capability of low aspect ratio pore sensors has recently been demonstrated in experiments for synthetic particles (Davenport et al. 2012; Ryuzaki

14

Micro- and Nanopore Technologies for Single-Cell Analysis

361

et al. 2017; Ryu et al. 2019). Measuring ionic current through 1 μm-sized micropore in a 50 nm-thick Si3N4 membrane, i.e., the thickness-to-diameter aspect ratio of 0.02, resistive pulses were recorded when negatively charged sub-micrometer-sized polystyrene beads were passed electrophoretically through the sensing zone. While most of the pulses had simple single-peak shapes suggestive of translocation of spherical objects, the rest showed a double-peak motif at the apexes. The peculiar ionic signatures were attributed to presence of polystyrene microbead dimers unintentionally formed in the salt solution used. In fact, the numerical simulations of cross-membrane ionic current also revealed a double-peak feature in the resistive pulse waveforms for translocation of a dumbbell-formed object in the low aspect ratio micropore. By comparing the theoretical and experimental pulses, the spatial resolution of the pore sensors was deduced, which was reported to be around several tens of nanometers for the channel structure and measurement setup used (Ryuzaki et al. 2017). These studies served to support the shape-sensing capability of low aspect ratio channel sensors. The unique sensor ability has also been examined to measure morphologies of a live cell having characteristic shapes (Tsutsui et al. 2017). Streptococcus salivarius was used as a model, which had a characteristic shape consisting of several ellipsoidal cells linked together in a form of a Buddhist rosary (Levesque et al. 2001). Resistive pulses were detected upon translocation through a low aspect ratio micropore designed to have a diameter close to that of Streptococcus. Close look into each ionic current signals revealed that some of them had a small corrugation at the apexes. There seem to be a random number of current up-downs up to 7 current bumps that were naturally ascribed to the number of cocci, i.e., the ellipsoid-shaped cells, comprising the bacteria passed through the sensing zone. When a channel of larger diameter was used, on the other hand, the shapereflecting current corrugations tended to be more faint and eventually disappeared from the pulse signals. The overall trend was reproduced in the multiphysics calculations of the blockade current for a model object mimicking the beads-inrow motif of Streptococcus. The finding suggested a crucial importance to use a pore of diameter close to the size of analytes. This can be qualitatively understood as the following. Basically, access resistance is much larger than the resistance inside the pore for low aspect ratio channels. As such, ionic blockade by an object causes current decrease associated mostly by a change in the resistance outside the pore. This makes the effects of resistance change inside the pore to be too small to observe in resistive pulse signals. It in turn means that the change in the resistance inside the pore may become observable by shrinking the gap space between an object and the channel wall as small as possible; in an extreme case, the ionic current drops to zero if one uses an ideal pore that completely fits to the shape of analytes as they completely occlude the shallow pore at some moment of translocation. In fact, it was possible to pass through the 1 μm-sized Streptococcus through a micropore of diameter down to 1.1 μm (Tsutsui et al. 2017). Although this comes at a price of higher chance to have clogging of the channel, it provides an opportunity to perform single-particle tomography with nanoscopic spatial resolution that may open new venues for analyzing single cells.

362

M. Tsutsui et al.

Being a sensor, it is of interest to see whether low aspect ratio pores are useful for discriminating cells by the shape. Bacteria such as Escherichia coli (E. coli) and Bacillus subtilis (B. subtilis) were chosen as analytes (Goyal et al. 2015; Tsutsui et al. 2016). These bioparticles are of similar size and shape constructed with a micrometer-scale rod-shaped body and long tentacles called flagella. Translocation of the bacteria through a low aspect ratio micropore generated resistive pulses of variable patterns reflecting the random orientations of the cylindrical structure. The expected flagellum-derived fine features were not discernable in the signals due presumably to the fact that the filament-like structure is only several tens of nanometer in diameter and hence has too small volume to cause notable change in the ionic current via ion blockage in the microscale channel. Conventional analyses using the pulse height and width demonstrated large overlap in the distributions (Goyal et al. 2015; Tsutsui et al. 2016). This is not surprising considering the extensive sensing zone of the low aspect ratio pores that not only deteriorates the signal-to-noise ratio but also allows many non-volume features of objects to be reflected in the resistive pulse waveforms. More specifically, the height, for instance, varies widely depending on the bacterial body to orient along or perpendicularly to the pore axis, which is not a case for high aspect ratio channels wherein the magnitude of ion blockage is mostly determined by the occluded volume in the inside let alone the shape and off-axis effects (Tsutsui et al. 2016). Thus, in attempt to discriminate analytes by resistive pulse heights, it would be not a suitable choice to use low aspect ratio pores. This, however, does not necessarily mean that the shallow channel is a bad design; it is perhaps merely due to the fact that the sensor tends to reflect too many physical parameters into a single pulse whereby requiring different data analytics to capture a fine difference in the resistive pulses for the single-cell analysis as discussed in the next section.

AI-Driven Resistive Pulse Analysis for Discriminating Single Cells Artificial intelligence (AI) is a set of algorithms designed to act like human intelligence. There has been increasing interest in using this unique computer technology in a wide variety of applications such as robot control and medical diagnosis. Machine learning is a branch of AI that has attracted many researchers in recent years to use in their studies (Jordan and Mitchell 2015). It can be used to compare and discriminate patterns in a way akin to human. There are various types of learning algorithms developed for completing specific tasks. Supervised learning, for instance, first uses a set of well-defined data, e.g., digital photo-images of entities that can be unambiguously identified as a specific species of animal such as cat and dog, is utilized to train a model. After training, the model will be given an ability to classify input data based on the information stored in the training process. What is different in machine learning from other digital data analytics is that it looks on multiple features in each pattern and uses them to judge the similarity compared to those features in the training data, which is a similar thing to what we do in our daily life with our sensory nerves and brain.

14

Micro- and Nanopore Technologies for Single-Cell Analysis

363

As described in the preceding section, the advanced low aspect ratio pore sensors are capable of sensing not only volume of analytes but also their fine shapes, orientations, surface charge, and even translocation dynamics inside and outside the channel (Ryuzaki et al. 2017; Ryu et al. 2019; Goyal et al. 2015; Tsutsui et al. 2016, 2017). All these features are reflected in a resistive pulse waveform as fine corrugations. On the other hand, conventional resistive pulse analysis only takes pulse height, and sometimes also the width, for single-particle discriminations, which is obviously not a suitable approach for analyzing the ionic current signals obtained with a shallow channel. In contrast, it is anticipated that the pattern recognition capability of machine learning best fits to evaluate and compare the complicated multiphysical effects (Fig. 6). In fact, increasing number of nanopore studies today are using machine learning in the resistive pulse analyses (Vercoutere et al. 2001; Farimani et al. 2018; Arima et al. 2018; Yusko et al. 2017). As for the application in single-cell analyses, it was first tested to discriminate bacteria of similar size and shape, i.e., the aforementioned E. coli and B. subtilis. A shallow micropore of 3 μm diameter and 50 nm depth in a Si3N4 membrane was implemented as a sensor to detect these rod-shaped cells. The measured resistive pulses demonstrated wide variation in the height and width due to the prominent role of the access resistance (Goyal et al. 2015). Machine learning was thus employed to discern the microorganisms by the signal shapes

Fig. 6 AI-driven resistive pulse analysis. (a) Extraction of feature parameters. (b) Algorithm flow for a machine learning classification in case of using ensemble learning method like Random Forest. (c) The algorithms output probability numbers telling which particles they are likely to be for each pulse measured

364

M. Tsutsui et al.

(Tsutsui et al. 2017). Multiple feature parameters were defined that included not only height and width but also pulse area, onset angle, and even inertia. In addition to these researcher-crafted parameters, randomly selected data points in each signal were also used. Several of these feature parameters were again randomly chosen to create feature vectors. Training was carried out based on the feature vectors defined in each pulse recorded. A number of algorithms including the widely used support vector machine were first screened by actually examining classification of a standard microspheres of known size and surface charge densities. Among those tested, ensemble learning methods such as Random Forest and Rotation Forest were found to work better on distinguishing the cells by the resistive pulse patterns as evaluated by the F-measure defined as F-measure ¼ 2PrecisionRecall/ (Precision + Recall). With the power of pattern analytics, the similar-sized bacteria were shown to be discriminated with F-measure score higher than 90%, which roughly means 90% accuracy to discern the cells of resembling morphologies yet different species (Tsutsui et al. 2017). Moreover, it was revealed that the F-measure decreases with increasing the pore depth up to 500 nm. Meanwhile, the score also tended in turn to increase with further thickening of the channel. The nontrivial dependence of the cell discriminability on the pore geometries suggests that when the channel is shallow enough (0.06 in the literature), it becomes an efficient sensor that can capture the multiple physical parameters of individual analytes. On the other hand, long channels are also a nice sensor (aspect ratio larger than 0.5) capable of sensing slight difference in the volume of cells as in Coulter counters (Allman et al. 1992). Nevertheless, the intermediate range of aspect ratio channel structure can probably give good sensitivity to neither the analyte volume nor the other physical properties (Tsutsui et al. 2017). This result suggests the usefulness of the machine learning approach not only for cell analyses but also as a tool that delivers a data-driven strategy for optimizing the sensor structures. The AI-driven pore sensing has recently been applied for other types of cells. Ahuja et al. (2019) used support vector machine for label-free screening of cancer cells. In the work, detections of cancer cells treated with an antibody drug were exhibited by multi-frequency impedance measurements using embedded microelectrodes in a fluidic channel. Pulse-shaped temporal impedance changes were observed that were pattern-analyzed with the amplitude and phase change as feature parameters. Using the data collected for live and dead cancer cells for training, the approach was demonstrated to be effective to assess the cell viability. Besides single-particle analysis, machine learning pattern analyses have also been widely employed in nanopore sensing for identifying viruses (Arima et al. 2018), discriminating protein shapes (Yusko et al. 2017), and sequencing DNA (Boza et al. 2017). Whereas Coulter counters replaced the routine manual blood cell counting in medical laboratories, a human-like ability of the AI-driven resistive pulse analysis to give judgments on the presence or absence of particular cells and other molecules/ particles may replace a part of roles of doctors by realizing digital disease diagnosis and prescriptions in the future.

14

Micro- and Nanopore Technologies for Single-Cell Analysis

365

Biorecognition Pore Sensors Coulter principle is a physical mechanism for measuring physical quantities to analyze physical properties of analytes such as their size and shape in electrolyte solution. Meanwhile, there are many organisms and bioparticles that have similar morphologies but different functions. Influenza virus is a good example, which is a family of more than 140 different subtypes each having a resembling spherical shape but different genomes and surface protein constitutions. Every year, it causes seasonal epidemic and sometimes serious pandemic all over the world (Zhang and Webster 2017). Bacteria species are even more diverse. Some of these are known to be pathogenic named by the types of antigens such as O157 and O111 for enterohemorrhagic E. coli. It is only very recently that there found many unidentified species exist within a sample taken in public places (Mclntyre et al. 2017). As such, these entities are ubiquitous in human life and intimately related to public health. Coulter principle can be a promising technique for screening of the pathogens with an ultimate resolution of single particle. However, the analytes of a same family cannot be easily discriminated by resistive pulse patterns as they are often quite similar in shape and size. Meanwhile, many of the today’s bioanalytical tools utilize a specificity of molecular probes such as antibodies and DNA to find analytes of concern in a physiological sample (Zhu et al. 2018; Syemoradi et al. 2017). They tend to form strong intermolecular interactions only with molecules having particular morphologies and/or molecular patterns. When deployed on a surface, the mechanism can be used to collect specific objects at a predefined location. Incorporating additional ingenious molecular labelling processes, it enables post detections of the analytes via optical methods or even with the naked eyes by change in color. This immunosensing method has been practically utilized in various bioanalytical applications including point-of-tool kits for infection diagnosis (Peeling and Mabey 2010). Recent studies showed that the concept can be applied in pore sensors (Hou et al. 2011; Perez-Mitta et al. 2017). In the works, recognition molecules were chemically absorbed on channel wall surface so as to giving a chance for particles to contact and interact with the molecular probes during the translocation. Depending on the balance between the forces exerted on the object to move through the pore and the strength of the interactions thereof, it will be temporarily or eternally trapped in the sensing zone. The intermolecular interaction-derived translocation motions can be characterized by analyzing the pattern of resistive pulses most simply by the width that denotes the time of flight of the object to pass through the conduit: stronger interaction with the functionalized wall yields longer pulse width and vice versa. The idea was first demonstrated experimentally by Iqbal et al. (2007) for identifications of DNA having specific base sequence. They functionalized a SiO2 nanopore with hairpin DNAs of predefined sequence. When single-stranded DNA complementary to the DNA probe on the wall surface was passed through the

366

M. Tsutsui et al.

pore, they showed shorter pulses compared to those having single-base mismatch whereby demonstrated the capability of the approach for identifying SNPs at singlemolecule level (Iqbal et al. 2007). The same research group also extended the approach for detecting proteins where they prepared biomimicking solid-state nanopores by covering a SiN nanopore wall with a ligand molecule-holding lipid bilayer (Yusko et al. 2011). More recently, the molecular strategy was demonstrated to identify dynamic change in the orientations of individual proteins hanged inside the conduit via specific interactions (Yusko et al. 2017). While these studies report on the use of molecular interactions in resistive pulse detections of nanometer-sized molecules, the biorecognition approach was also examined for selective detections of single cells (Tsutsui et al. 2018). Synthetic peptides were employed as molecular probes. It was found that the amino acid sequence governs the strength of the analyte-probe interactions that needs to be not too strong for allowing analytes to transit the channel without causing clogging (we note that the pore clogging can give another sensor design for specific detections of bioparticles by using multipore structures). With an optimized sequence of the molecular probe mimicking a molecular pattern of Toll-like receptor 5 (TLR5), it was attempted to discriminate flagellated wild-type E. coli and non-flagellated mutants (Kawai and Akira 2010). As these two bacteria have essentially the same rod-like shape except the narrow tentacles that are too small to cause notable change in the cross-membrane ionic current, the resistive pulses revealed little difference in non-functionalized micropores or those functionalized with short peptides of no expected functions. In contrast, the TLR5-derived peptides on the channel wall surface could temporarily trap the wild-type E. coli via the peptide-flagellum interactions and produced resistive pulses of wide width. The flagellum-less mutants, on the other hand, showed shorter pulses indicative of little interaction with the molecular probe on the wall, thereby proving the usefulness of the approach for selective detections of flagellated bacteria (Tsutsui et al. 2018), which is an important aspect of pathogen sensors considering the intimate roles of flagella on the bacterial pathogenesis (Duan et al. 2012). The biorecognition capability of functionalized solid-state micropores can be applied not only for pathogenic bacteria but also potentially for tumor cells and other biologically important cells by designing and employing molecular probes suitable for the resistive pulse analysis.

Potential and Challenges for Total Cell Analysis As described so far, advanced fabrication technologies and data analytics have spurred many researchers to expand the capability of Coulter counters for characterizing individual cells beyond sizing. Meanwhile, Coulter principle is also of potential use for analyzing objects smaller than cells down to single-nanometer scale by sculpting and utilizing a small enough pore to implement the resistive pulse detections. From different perspective, therefore, micro- and nanopore sensors can in principle open new avenues for total single-cell analysis by providing a comprehensive method to detect and identify biomaterials encapsulated in the lipid

14

Micro- and Nanopore Technologies for Single-Cell Analysis

367

membrane. Indeed, there are many experimental and theoretical studies directed to establish resistive pulse detections of the cell contents. One of them is vesicle. Led by the finding of its valuable functions in intercellular communications (Valadi et al. 2007) and accompanied importance as cancer biomarkers (Soung et al. 2017), there has been increasing interest in using resistive pulse analyses to count and identify vesicles produced in human cells. Their structure is similar to cells consisting of lipid bilayer balls wherein various biological components are contained such as proteins, nucleotides, and metabolites, while the size is smaller typically in a range of 200 nm–30 nm (Ko et al. 2017). They are called extracellular vesicles (EVs) and are ubiquitous in a body and can be found in tissues and biological fluids including blood. Resistive pulse detections of these nanoparticles have been examined by using tunable resistive pulse sensing (Van der Pol et al. 2014; De Vrij et al. 2013). Lane and coworkers (2014) further used the method to testify the influence of pretreatments such as ultracentrifugation and density gradient separation on the size of EVs. Anderson et al. (2015) demonstrated optimization of the measurement conditions. While these literatures found discrepancies in the EV size distributions with those obtained by other methods such as transmission electron microscopy observations, a standard protocol has recently been provided by Vogel et al. (2016) for reliable tests, thereby showing the effectiveness of the Coulter principle for a quantitative analysis of the biologically important cell content. Nanopores can also potentially be used for interrogating genomes. They form a one-dimensional biopolymer basically comprised of four different nucleobases chained via pentodes and phosphoric acids. There are continuing extensive research efforts made on this enthusiastic topic that aim to establish singlemolecule sequencing by a resistive pulse analysis (Branton 2008). It has already been realized by a biological nanopore by equipping polymerase to allow a ratcheting motion of polynucleotides in the single-nanometer-sized sensing zone at an astonishing resolution of subnanometers (Deamer 2016). Solid-state nanopores are still studied intensively to accomplish high-throughput sequencing. Although slowdown of the too-rapid translocation speed of DNA remains to be a crucial issue, the field is progressing where various 2D materials are now widely employed for an ultrathin membrane to render single-base resolution to the ionic current (Lee 2018). Along with polynucleotides, proteins in a cell also carry important genetic roles. These biomolecules are constructed with chains of 22 different amino acids entangled into a distinct form. Solid-state nanopores are expected to be a useful tool for not only analyzing the structure but also sequencing by the ionic current (Restrepo-Perez et al. 2018). Analogous to the case for DNA, there is a technical difficulty in accurately measure the short-lived resistive pulses as the nanoscale molecules pass through the sensing zone quite rapidly. In addition, aggregation and denature also need to be solved (Varongchayakul et al. 2018). Nonetheless, considering that there currently exists no such technology to address the structure and peptide sequence at a single-molecule level, it is truly a valuable research field promising great things for the future biology and medicine.

368

M. Tsutsui et al.

The nanopore technologies explained above guarantee their use in detecting various kinds of biomolecules residing inside a cell. This concept of total cell analysis via resistive pulse sensing, however, can only be performed with a tool for extraction and transport control of the molecules in each cell into the nanopores. As there is no such technique at present, many laborious pretreatments such as extraction, separation, and amplification of target analytes in numerous cells are indispensable in nanopore sensors for implementing the single-molecule detections, which inevitably hinders one to address the intracellular nanomaterials at a singlecell level. Meanwhile, recent studies by Kurz et al. (2014) suggested a possible use of nanopores themselves as a tool to direct the cellular content into a sensing zone. They trapped a cell on a nanopore and demonstrated that a hole can be created in the membrane by the huge electric field formed in there. After the electroporation, it was further shown to be possible to insert single-molecule DNA into the cell for transfection via electrophoresis through the nanopore to the membrane hole. Here, the nanopore worked not only as an apparatus to punch a hole in the cell membrane but also as a filter to allow mass transport of molecules that are smaller than the channel. Such procedure is expected a priori that this mechanism can also be implemented to extract intercellular molecules from a trapped cell (Fig. 7). It is also possible to integrate nanostructures (Yasui et al. 2019) or microfluidic channels (Sibbitts et al. 2018) for on-chip cell lysis nearby a nanopore. Then, it would become available to directly detect the intracellular materials from vesicles to DNA by resistive pulse measurements through making use of a multipore-on-nanopore structure. Such analytical tool is expected to greatly extend our knowledge on diversity and versatility of cells.

Fig. 7 Total cell analysis using micro- and nanopores. (a) Single-cell electroporation was reported to be feasible by exploiting the huge electric field created at a nanopore (Kurz et al. 2014). (b) Extending the ingenious mechanism, extraction, and detection of intercellular materials in individual cells is expected to be implemented by a nanopore on multipore. 3D sensor structure, wherein single-cell electroporation and subsequent molecule extraction is performed at the multipore followed by single-molecule detections by resistive pulse measurements using the nanopore

14

Micro- and Nanopore Technologies for Single-Cell Analysis

369

Conclusion After more than six decades since the invention, Coulter principle has become an essential tool for blood tests in medical laboratories worldwide. Very recently, the simple and versatile mechanism led to bionanopore devices for single-molecule sequencing by ionic current. Advance in nanotechnology and data analytics have also enabled novel sensor abilities beyond single-cell sizing for better discriminating analytes of vast range of sizes such as cell, vesicle, protein, and genome. Much efforts have also been devoted to resolve practical issues such as channel clogging and low detection throughput. Yet, research and development in this field still continue to grow rapidly. With the great potential, it may be expected to become feasible to implement quantitative measurements of intracellular molecules for the total cell analysis.

References Ahuja K et al (2019) Toward point-of-care assessment of patient response: a portable took for rapidly assessing cancer drug efficacy using multifrequency impedance cytometry and supervised machine learning. Microsys Nanoeng 5:34 Allman R et al (1992) Characterization of bacteria by multiparameter flow cytometry. J Appl Bacteriol 73:438–444 Anderson W, Lane R, Korbie D, Trau M (2015) Observations of Tunable Resistive Pulse Sensing for Exosome Analysis: Improving System Sensitivity and Stability. Langmuir 31:6577–6587 Arima A et al (2018) Selective detections of single-viruses using solid-state nanopores. Sci Rep 8:16305 Arjmandi N et al (2012) Measuring the electric charge and zeta potential of nanometer-sized objects using pyramidal-shaped nanopores. Anal Chem 84:8490–8496 Balaz T, Grice HC, Airth JM (1960) On counting the blood cell of the rat with an electronic counter. Can J Comp Med Vet Sci 24:273–275 Barreiros FM, Ferreira PJ, Figueiredo MM (1996) Calculating shape factors from particle sizing data. Part Part Syst Charact 13:368–373 Beckman Coulter Diagnostics (2016) Talk CBC. http://www.beckmancoulter.co.jp/product/prod uct02/cbc/pdf/Vol4.pdf Blois D, Bean CP (1970) Counting and sizing of submicron particles by the resistive pulse technique. Rev Sci Instrum 41:909 Boza V, Brejova B, Vinar T (2017) DeepNano: deep recurrent neural networks for base calling in MinION nanopore reads. PLoS One 12:e0178751 Branton D (2008) The potential and challenges of nanopore sequencing. Nat Biotechnol 26:1146–1153 Briggs K, Kwok H, Tabard-Cossa V (2014) Automated fabrication of 2-nm solid-state nanopores for nucleic acid analysis. Small 10:2077–2086 Brotherton J (1975) The counting and sizing of spermatozoa from ten animal species using a Coulter counter. Andrologia 7:169–185 Bull BS, Schneiderman MA, George B (1965) Platelet counts with the Coulter counter. Am J Clin Pathol 44:678–688 Cassee FR et al (2002) Effects of diesel exhaust enriched concentrated PM2.5 in ozone preexposed or monocrotaline-treated rats. Inhal Toxicol 14:721–743 Chen P et al (2004) Atomic layer deposition to fine-tune the surface properties and diameters of fabricated nanopores. Nano Lett 4:1333–1337

370

M. Tsutsui et al.

Conlon I, Raff M (2003) Differences in the way a mammalian cell and yeast cells coordinate cell growth and cell-cycle progression. J Biol 2:7 Cooper WD, Parfitt GD (1967) Comparison of particle size distributions of “monodisperse” particles from 0.8 to 3.5 μ in diameter using a Coulter counter and electron microscopy. Kolloid Z Z Polym 223:160–166 Coulter WH (1953) Means of counting particles suspended in a fluid. US Patent No. 2656508 Coulter WH (1956) High speed automatic blood cell counter and cell size analyzer. Proc Natl Electron Conf 12:1034 Davenport M et al (2012) The role of pore geometry in single nanoparticle detection. ACS Nano 6:8366–8380 De Vrij J et al (2013) Quantification of nanosized extracellular membrane vesicles with scanning ion occlusion sensing. Nanomedicine 8:1443–1458 De Yoreo JJ, Sommerdijk NAJM (2016) Investigating materials formation with liquid-phase and cryogenic TEM. Nat Rev Mat 1:16035 Deamer D (2016) Three decades of nanopore sequencing. Nat Biotechnol 34:518–524 Duan Q et al (2012) Flagella and bacterial pathogenicity. J Basic Microbiol 53:1–8 Eckhoff RK (1969) A static investigation of the Coulter principle of particle sizing. J Sci Instrum 2:973–977 Farimani AB et al (2018) Identification of amino acids with sensitive nanoporous MoS2: towards machine learning-based prediction. npj 2D Mat Appl 2:14 Feng J et al (2015) Identification of single nucleotides in MoS2 nanopores. Nat Nanotechnol 10:1070–1076 Fraikin JL et al (2011) A high throughput nanoparticle analyzer. Nat Nanotechnol 6:308–311 Garaj S et al (2010) Graphene as a subnanometre trans-electrode membrane. Nature 467:190–193 Gilbert SM et al (2017) Fabrication of subnanometer-precision nanopores in hexagonal boron nitride. Sci Rep 7:15096 Goyal G, Mulero R, Ali J, Darvish A, Kim MJ (2015) Low aspect ratio micropores for single-particle and single-cell analysis. Electrophoresis 36:1164–1171 Guo LJ, Cheng X, Chou CF (2004) Fabrication of size-controllable nanofluidic channels by nanoimprinting and its application for DNA stretching. Nano Lett 4:69–73 Guo S et al (2015) Nanofluidic transport through isolated carbon nanotube channels: advances, controversies, and challenges. Adv Mater 27:5726–5737 Hall JE (1975) Access resistance of a small circular pore. J Gen Physiol 66:531 Haque F et al (2013) Solid-state and biological nanopore for real-time sensing of single chemical and sequencing of DNA. Nano Today 8:56–74 Harms ZD et al (2015) Single-particle electrophoresis in nanochannels. Anal Chem 87:699–705 Haywood DG et al (2015) Fundamental studies of nanofluidics: nanopores, nanochannels, and nanopipets. Anal Chem 87:172–187 Heerema SJ, Dekker C (2016) Graphene nanodevices for DNA sequencing. Nat Nanotechnol 11:127–136 Henriquez R et al (2004) The resurgence of Coulter counting for analyzing nanoscale objects. Analyst 129:478–482 Hou X, Guo W, Jiang L (2011) Biomimetic smart nanopores and nanochannels. Chem Soc Rev 40:2385–2401 Howorka S, Siwy Z (2009) Nanopore analytics: sensing of single molecules. Chem Soc Rev 38:2360 Iqbal SM, Akin D, Bashir R (2007) Solid-state nanopore channels with DNA selectivity. Nat Nanotechnol 2:243–248 Jordan MI, Mitchell TM (2015) Machine learning: trends, perspectives, and prospects. Science 349:255–260 Karuhn R et al (1975) Studies on the Coulter counter Part I. Investigations into the effect of orifice geometry and flow direction on the measurement of particle volume. Powder Technol 11:157–171

14

Micro- and Nanopore Technologies for Single-Cell Analysis

371

Kasianowicz JJ et al (1996) Characterization of individual polynucleotide molecules using a membrane channel. Proc Natl Acad Sci 93:13770–13773 Kawai T, Akira S (2010) The role of pattern-recognition receptors in innate immunity: update on toll-like receptors. Nat Immun 11:373–384 Kelly JK, Rasch A, Kalisz S (2002) A method to estimate pollen viability from pollen size variation. Am J Bot 89:1021–1023 Kim H, Manden-Deuer S (2013) Reliability of rapid, semi-automated assessment of plankton abundance, biomass, and growth rate estimates: Coulter counter versus microscope measurements. Limnol Oceanogr Methods 11:382–393 Ko J et al (2017) Combining machine learning and nanofluidic technology to diagnose pancreatic cancer using exosomes. ACS Nano 11:11182–11193 Kovarik ML et al (2012) Micro total analysis systems for cell biology and biochemical assays. Anal Chem 84:516–540 Kowalczyk SW et al (2011) Modeling the conductance and DNA blockade of solid-state nanopores. Nanotechnology 22:315101 Kurz V, Tanaka T, Timp G (2014) Single cell transfection with single molecule resolution using a synthetic nanopore. Nano Lett 14:604–611 Lane RE et al (2014) Analysis of exosome purification methods using a model liposome system and tunable-resistive pulse sensing. Sci Rep 5:7639 Lee K (2018) Recent progress in solid-state nanopores. Adv Mater 30:1704680 Levesque C et al (2001) Streptococcus salivarius fimbriae are composed of a glycoprotein containing a repeated motif assembled into a filamentous nondissociable structure. J Bacteriol 183:2724–2732 Li J et al (2001) Ion-bream sculpting at nanometer length scales. Nature 412:166–169 Liang X et al (2007) Single sub-20 nm wide, centimeter-long nanofluidic channel fabricated by novel nanoimprint mold fabrication and direct imprinting. Nano Lett 7:3774–3780 Liu H et al (2010) Translocation of single-stranded DNA through single-walled carbon nanotubes. Science 327:64–67 Liu et al (2017) Improving particle detection sensitivity of a microfluidic resistive pulse sensor by a novel electrokinetic flow focusing method. Microfluid Nanofluid 21:4 Luo L et al (2014) Resistive-pulse analysis of nanoparticles. Annu Rev Anal Chem 7:513–535 Manfrinato VR et al (2013) Resolution limits of electron beam lithography toward the atomic scale. Nano Lett 13:1555–1558 Mclntyre ABR et al (2017) Comprehensive benchmarking and ensemble approaches for metagenomics classifiers. Genome Biol 18:182 Meng JQ (2012) Synthesis of antifouling nanoporous membranes having tunable nanopores via click chemistry. J Membr Sci 401–402:109–117 Mernier G et al (2009) Cell viability assessment by flow cytometry using yeast as cell model. Sensors Actuators B 154:160–163 Morris CA, Friedman AK, Baker LA (2010) Applications of nanopipettes in the analytical science. Analyst 135:2190–2202 Morrison WR, Scott DC (1985) Measurement of the dimensions of wheat starch granule populations using a Coulter counter with 100-channel analyzer. J Cereal Sci 4:13–21 Peeling RW, Mabey D (2010) Point-of-care tests for diagnosing infections in the developing world. Clin Microbiol Infect 16:1062–1069 Perez-Mitta G et al (2017) Bioinspired integrated nanosystems based on solid-state nanopores: “ionic” transduction of biological, chemical and physical stimuli. Chem Sci 8:890–913 Prabhu AS et al (2011) SEM-induced shrinking of solid-state nanopores for single molecule detection. Nanotechnology 22:425302 Qin Z, Zhe J, Wang GX (2011) Effects of particle’s off-axis position, shape, orientation and entry position on resistance changes of micro Coulter counting devices. Meas Sci Technol 22:045804 Rampfer G et al (2016) Nanoparticle translocation through conical nanopores: a finite element study of electrokinetic transport. Macromol Theory Simul 26:1600051

372

M. Tsutsui et al.

Restrepo-Perez L, Joo C, Dekker C (2018) Paving the way to single-molecule protein sequencing. Nat Nanotechnol 13:786–796 Richmond DR, Tauxe WN, Bassingthwaighte JB (1970) Albumin macroaggregates and measurements of regional blood flow: validity and application of particle sizing by Coulter counter. J Lab Clin Med 75:336–346 Ryu J et al (2019) Ultralow-aspect-ratio micropore sensor for a particle translocation: critical role of the access region. Sensors Actuators B 288:454–459 Ryuzaki S et al (2017) Rapid structural analysis of nanomaterials in aqueous solutions. Nanotechnology 28:155501 Sack U, Tarnok A, Rothe G (2008) Cellular diagnostics: basic principles, methods, and clinical applications of flow cytometry. Karger Medical and Scientific Publishers, Basel Saleh OA, Sohn LL (2003) An artificial nanopore for molecule sensing. Nano Lett 3:37–38 Sibbitts J et al (2018) Cellular analysis using microfluidics. Anal Chem 90:65–85 Siwy Z et al (2004) Conical-nanotube ion-current rectifiers: the role of surface charge. J Am Chem Soc 126:10850–10851 Soung YH et al (2017) Exosomes in cancer diagnostics. Cancers 9:8 Storm AJ et al (2003) Fabrication of solid-state nanopores with single-nanometre precision. Nat Mater 2:437–540 Sun T, Morgan H (2010) Single cell microfluidic impedance cytometry: a review. Microfluid Nanofluid 8:423–443 Syemoradi L et al (2017) Point of care testing: the impact of nanotechnology. Biosens Bioelectron 87:373–387 Tsutsui M et al (2012) Single-nanoparticle detection using a low-aspect-ratio pore. ACS Nano 6:3499–3505 Tsutsui M et al (2016) Particle trajectory-dependent ionic current blockade in low-aspect-ratio pores. ACS Nano 10:803–809 Tsutsui M et al (2017) Discriminating single-bacterial shape using low-aspect-ratio pores. Sci Rep 7:17371 Tsutsui M, Tanaka M, Marui T, Yokota K, Yoshida T, Arima A, Tonomura W, Taniguchi M, Washio T, Okochi M, Kawai T (2018) Identification of Individual Bacterial Cells through the Intermolecular Interactions with Peptide-Functionalized Solid-State Pores. Anal Chem 90:1511–1515 Tzur A et al (2009) Cell growth and size homeostasis in proliferating animal cells. Science 325:167–171 Vaclavek T, Prikryl J, Foret F (2018) Resistive pulse sensing as particle counting and sizing method in microfluidic systems: designs and applications review. J Sep Sci 42:445–457 Valadi H et al (2007) Exosome-mediated transfer of mRNAs and microRNAs is a novel mechanism of genetic exchange between cells. Nat Cell Biol 9:654–659 Van der Pol E et al (2014) Particle size distribution of exosomes and microvesicles determined by transmission electron microscopy, flow cytometry, nanoparticle tracking analysis, and resistive pulse sensing. J Thromb Haemost 12:1182–1192 Varongchayakul N et al (2018) Single molecule protein sensing in a nanopore: a tutorial. Chem Soc Rev 47:8512–8524 Vembadi A, Menachery A, Qasaimeh MA (2019) Cell cytometry: review and perspective on biotechnological advances. Front Bioeng Biotechnol 7:147 Vercoutere W et al (2001) Rapid discrimination among individual DNA hairpin molecules at singlenucleotide resolution using an ion channel. Nat Biotechnol 19:248–252 Veschueren DV, Yang W, Dekker C (2018) Lithography-based fabrication of nanopore arrays in freestanding SiN and graphene membranes. Nanotechnology 29:145302 Vogel R et al (2016) A standardized method to determine the concentration of extracellular vesicles using tunable resistive pulse sensing. J Ext Ves 5:31242 Vogel R et al (2017) High-resolution single particle zeta potential characterization of biological nanoparticles using tunable resistive pulse sensing. Sci Rep 7:17479

14

Micro- and Nanopore Technologies for Single-Cell Analysis

373

Walstra P, Oortwijn H (1969) Estimating globule-size distribution of oil-in-water emulsions by Coulter counter. J Colloid Interface Sci 29:424–431 Wang J et al (2014) Effects of access resistance on the resistive-pulse caused by translocating of a nanoparticle through a nanopore. RSC Adv 4:7601–7610 Wanunu M et al (2010) Rapid electronic detection of probe-specific microRNAs using thin nanopore sensors. Nat Nanotechnol 5:807–814 Willmott GR et al (2010) Use of tunable nanopore blockade rates to investigate colloidal dispersions. J Phys Condens Matter 22:454116 Wood DK, Requa MV, Cleland AN (2007) Microfabricated high-throughput electronic particle detector. Rev Sci Instrum 78:104301 Wu X et al (2008) Microfluidic differential resistive pulse sensors. Electrophoresis 29:2754–2759 Yanagi I et al (2014) Fabricating nanopores with diameters of sub-1 nm to 3 nm using multilevel pulse-voltage injection. Sci Rep 4:5000 Yasaki H et al (2017) Substantial expansion of detectable size range in ionic current sensing through pores by using a microfluidic bridge circuit. J Am Chem Soc 139:14137–14142 Yasui T et al (2019) Engineering nanowire-mediated cell lysis for microbial cell identification. Nano Lett 13:2262–2273 Ying YL, Cao C, Long YT (2014) Single molecule analysis by biological nanopore sensors. Analyst 139:3826–3835 Ying C et al (2018) Formation of single nanopores with diameters of 20–50 nm in silicon nitride membranes using laser-assisted controlled breakdown. ACS Nano 12:11458–11470 Yukimoto N et al (2013) Tracking single-particle dynamics via combined optical and electrical sensing. Sci Rep 3:1855 Yusko EC, Johnson JM, Majd S, Prangkio P, Rollings RC, Li J, Yang J, Mayer M (2011) Controlling protein translocation through nanopores with bio-inspired fluid walls. https:// www.nature.com/nnano. Nat Nanotechnol 6:253–260 Yusko EC et al (2017) Real-time shape approximation and fingerprinting of single proteins using a nanopore. Nat Nanotechnol 12:360–367 Zhang W, Webster RG (2017) Can we beat influenza? Science 357:111 Zhang Y et al (2019) Nanopore formation via tip-controlled local breakdown using an atomic force microscope. Small 3:1900147 Zhe J et al (2007) A micromachined high throughput Coulter counter for bioparticle detection and counting. J Micromech Microeng 17:304–313 Zhu Y et al (2018) Immuno-affinity amperometric detection of bacterial infections. Angew Chem Int Ed 57:14942–14946

Technologies for Single-Cell Printing and Patterning

15

Pranav Ambhorkar, Mahmoud Ahmed Sakr, Hitendra Kumar, and Keekyoung Kim

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acoustophoresis and Fluorescence-Activated Cell Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Laser and Vacuum Microdissection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impedance-Based Single-Cell Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical Platforms for Single-Cell Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inkjet-Based Single-Cell Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microcontact Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Droplet-Based Patterning of Single Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Microbial Cell Printing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

376 377 378 379 380 380 383 385 387 391 391

Abstract

Printing single cells after isolating them from heterogeneous cell populations has the potential to reveal several new insights into cell behavior, therapeutics, and precision medicine. Since studies conducted on individual cells are independent of background noise that results from other cells present in the vicinity as well as intercellular phenotypic heterogeneity, they can provide an accurate characterization of tissue microenvironments such as multiclonal tumors while allowing single-cell genome/proteome analysis. Additionally, the ability to print at the Pranav Ambhorkar and Mahmoud Ahmed Sakr contributed equally with all other contributors. P. Ambhorkar · M. A. Sakr · H. Kumar School of Engineering, University of British Columbia, Kelowna, BC, Canada K. Kim (*) Department of Mechanical and Manufacturing Engineering, Schulich School of Engineering, University of Calgary, Calgary, AB, Canada Biomedical Engineering Graduate Program, University of Calgary, Calgary, AB, Canada e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_28

375

376

P. Ambhorkar et al.

resolution of a single cell could also facilitate the development of threedimensional, multicell type patterns that could potentially enable highly precise tissue engineering and accurate cell-cell communication studies. A variety of single-cell printers have been developed for research and commercial applications and can be broadly categorized into either “contact” or “noncontact” dispensing techniques. These printers, however, face significant challenges surrounding accuracy and sensitivity that can be addressed by building in additional functionalities such as optical detection algorithms and impedance detection. In this chapter, we will discuss various single-cell dispensing techniques: laser printing, inkjet printing, microcontact printing, acoustic-droplet generation, electrohydrodynamic spraying, cell sorting, and cell trapping, as well as an overview of current and potential applications. Keywords

Microgels · Single-cell patterning · Cell sorting · Cell printing · Single-cell manipulation

Introduction Cell populations in the human body are heterogeneous, and this property has become relevant to a wide range of studies in the biomedical sciences. For instance, cells that are constituent of cancer tumors tend to be genetically and phenotypically heterogeneous which has made it quite difficult to administer therapies (Altschuler and Wu 2010). Cellular heterogeneity is also a property of normal cells, constituent of body tissues that the field of tissue engineering seeks to mimic. Cells are usually intricately placed in such tissues, and their organization is particularly important to the macroscale functioning (Xu et al. 2013). Developing an understanding into the behavior of an individual cell, its signaling pathways, and properties is therefore very important to studies spanning regenerative medicine, systems biology, and cancer genomics. The advent of new single-cell “omics” protocols such as single-cell RNA sequencing (Kolodziejczyk et al. 2015) and DNA sequencing has shed a new light into subcellular functions that cause variation in cell behavior at the population level. For successfully carrying out these protocols, it is important, as the first step, to isolate single cells from a larger population. For this reason, technological developments in microfluidics and cell sorting have been instrumental for success in singlecell studies. Characterization methods such as fluorescence-activated cell sorting have enabled the accurate sorting of single cells (Sutermaster and Darling 2019). It is important to note that a lot of single-cell studies begin with the dispensing of individual cells into substrates such as microarrays (Liberski et al. 2011) and microwells (Lindström and Andersson-Svahn 2012). For such tasks, recently, highly precise cell-printing mechanisms have been modified to accurately dispense single cells. This chapter starts with a brief discussion on cell-sorting systems and then moves on to a discussion on contact and noncontact printing techniques. When printing

15

Technologies for Single-Cell Printing and Patterning

377

cells for biological studies, it is important to consider their postprint environment. Since geometry has been known to have a large impact on cell behavior and expression, we have also included a discussion on printing substrates for singlecell studies.

Acoustophoresis and Fluorescence-Activated Cell Sorting Acoustophoresis is the principle of separating particles using acoustic waves. This principle requires no physical interaction between the source of the waves and the particles. This method is useful in cell sorting and analysis. Current sorting mechanisms vary from microfluidics, florescent dyes, and magnetic activated devices. The analysis of a single cell has been enabled largely due to recent developments in microfluidic technologies. In order to study single cells, they must indeed be dispensed and patterned. However, the condition for single-cell printing is the preceding isolation and capture of single cells. This capture of individual cells into a droplet from a greater subpopulation is governed by Poisson’s statistics, which means that only about 37% of droplets would consist of single cells. The remaining droplets would either be dispensed empty or contain multiple cells. In order to ensure a purer printing of droplets/units containing single cells, it is important to isolate them from a population. Several technologies have been used to sort cells and in correspondence with printing. Acoustophoresis offers for a noncontact methodology for controlling cell alignment with free interaction. In this technique, an alternating current is applied to interdigitated transducers (IDTs), which generate standing surface acoustic waves (SSAW). The waves then propagate through the device and incite a response within cells, which align and migrate in the direction of the acoustic force (Green 2019). Using this technology, single cells can be effectively sorted. Acoustophoresis has several applications in developing a therapeutic vascular tissue with collateral adhesion vessels. The study also enabled an understanding of gene expression, secretion of angiogenic and anti-inflammatory paracrine factors. Acoustophoresis has also been used for aligning human umbilical vein endothelial cells and human adipose stem cells in a hyaluronic acid hydrogel (Kang et al. 2018). Fluorescence-activated cell sorting (FACS) works to sort a mixture of different types of cells based on the light scattered from them and their fluorescent characteristics. In this method, antibodies are associated to fluorescent molecules and are placed onto the surface of the cells. After being stained with the antibodies and fluorescent molecules, cells go through a tube. Following this, a laser beam is spotted onto the cells, and the scattered light is detected by using an optical detector. This induces either a negative or a positive charge on the metal electrodes, in accordance with the charge of the cells. Based on the charges contained in both electrodes, the cells can be sorted in three different tubes; the center tube is reserved for cells with no antibody staining, while the ones on the sides are for charged cells (Carter et al. 2013). An example of a study where cell isolation was performed using FACS involved the sorting of muscle stem cells (MuSCs) from a heterogeneous cell

378

P. Ambhorkar et al.

population. MuSCs were isolated from limb and diaphragm muscles and from adult mice of all ages (Liu et al. 2015).

Laser and Vacuum Microdissection Laser capture microdissection (LCM) is a technique that is used to isolate single cells from a tissue section under a microscope. As shown in Fig. 1a, a laser beam is spotted on the tissue with a beam diameter in the range of 7.5–15 μm and a power range of 40–50 mW with pulse duration of about 650 μs to 2.5 ms. In this technique, two types of lasers are used: infrared and ultraviolet; both techniques have the same visualization characteristics and tissue selection methodology under the microscope. For collecting the cells, there is a collection tube that is designed specifically for separated cells. Although the technique is quite advantageous due to its efficiency and accuracy when it comes to isolating cells, it is quite expensive. There are also concerns regarding the postisolation quality of the sorted cells for further analysis since they are stained and exposed to various reagents. The exposure of cells to laser beams might also lead to the dehydration of the cells (Kerk et al. 2016). LCM has been used for determining the usage of (variable) V, (diversity) D, and (Joining) J gene segments in normal and malignant B cells by isolating single cells and studying the immunoglobin (Ig) variable gene regions and by using primers to recognize V and J genes in polymerase chain reaction (PCR) (Kain 2005). Kearns-Sayre syndrome (KSS) was studied using LCM by studying mitochondrial genomes in the isolated cells. The results from this work helped enable a better understanding of mitochondrial damage and cell heterogeneity (Kummer and Wilichowski 2018). LCM has played a role in single-inclusion level by isolating Chlamydia trachomatis-infected cells to study the generation of microbiological clones of C. trachomatis, and molecular assay tests to the isolated chlamydia cells to study cell features (Jordana et al. 2016). In vacuum microdissection, cells are sorted using a capillary-based, vacuum pulse-assisted technology that isolates single cells from a group of cells in a petri dish, which are then dispensed in a new medium. Two companies, NeuroInDx and

Fig. 1 Cell-sorting mechanisms. (a) Vacuum-assisted microdissection. (Reproduced with permission from Kudo et al. (2012)). (b) Laser microdissection. (Adapted from Wang et al. 2005)

15

Technologies for Single-Cell Printing and Patterning

379

Nanocollect, were founded based on the concept of vacuum-assisted isolation and have since launched two products in the market – UnipicK+ and N1 single cell as shown in Fig. 1b. Cells could be collected from any adherent culture cells and could be isolated in a small volume, 15nL, for use in single-cell analysis and sequencing. This procedure is known to be quite simple and efficient for isolating single cells from any type of adherent cells (Kudo et al. 2012).

Impedance-Based Single-Cell Printing By using impedance-detection algorithms, printers can dispense single cells captured within pico-sized droplets. As shown in Fig. 2a, cells are flown through a microchannel and pass by an electrode; this motion generates a signal within the electrode. Subsequently, when cells are flown in proximity to a second electrode, a piezoelectric material is triggered to move a diaphragm to dispense the single cell encapsulated within the pico-droplet. Using the resulting characteristic double peak, a positive peak is recorded from the first electrode and a negative peak is recorded from the second electrode as shown in Fig. 2b. The transit time is taken as the time that cells require to move between the first and second electrodes, thus ensuring that only a single cell is present in the channel. Trigger delay is the signal to the piezo actuator by using the gravimetric regression method; a pico-droplet of the liquid and a single cell are dispensed through the dispensing nozzle (Schoendube et al. 2015). Impedance labeling is another technique for isolating cells from subpopulations. In this method shown in Fig. 3a and b, cells are labeled using small antibody conjugated beads which attach to cells. By using alternating current, cells are sorted in a microchannel that has 30 μm high patterned electrodes on the top and bottom (Holmes and Morgan 2010). By measuring the impedance of cells flowing through the channel at high speeds, the authors

Fig. 2 Single-cell printing. (a) Schematic of single-cell printing. (b) Impedance of single 15 μm bead. (Reproduced with permission from Schoendube et al. (2015))

380

P. Ambhorkar et al.

Fig. 3 (a) Microfluidic impedance cytometer, and (b) impedance labeling. (Reproduced with permission from Holmes and Morgan (2010))

could identify cell type and properties with two-simultaneous frequencies. The lower frequency indicates cell size, while the higher indicates membrane properties.

Optical Platforms for Single-Cell Printing Another technique that was developed at the company Cytena sought to isolate single cells using an optical platform. The company has developed an optical particle detection mechanism that can detect the number of cells in the dispensing nozzle and an algorithm that subsequently sorts them. The workflow of this method consists of the execution of two algorithms in order. A cell suspension is added to the dispensing chip, which is then mounted onto a piezo actuator. The single cells are dispensed into each well of a 96-well plate. As cells pass through the dispensing nozzle, a chargecoupled device camera CCD detects the number of cells. Based on this data, the automated dispensing head only dispenses cells into wells when single cells are detected; meanwhile, the detection of multiple cells causes it to dispense them into a waste reservoir (Gross et al. 2015; Yusof et al. 2011). An optical nanobiosensor is used to measure cytochrome-c which is the source of cellular energy and plays a key role in apoptosis and programmed cell death (Popović 2013). In this approach, the release of cytochrome-c from a single cell is monitored using an optical sensor. A nanoprobe loaded with immobilizing antibodies is placed directly outside of the cell, and by using a microscope stage, the tip is inserted into the cell membrane and through into the cytoplasm. The released cytochrome-c is detected using enzymatic linked immunosorbent assay (ELISA).

Inkjet-Based Single-Cell Printing As a relatively inexpensive and versatile dispensing technique, inkjet printing has gained popularity in biomedical research. In this technique, a solution, “ink,” consisting of cells is dispensed onto a substrate. Broadly, inkjet printing systems

15

Technologies for Single-Cell Printing and Patterning

381

can be categorized as belonging to three dispensing mechanisms: continuous inkjet (CIJ), drop-on-demand (DoD), and electrohydrodynamic printing (Martinez-Rivas et al. 2017). Continuous inkjet printing entails the continuous dispensing of an ink as a stream, which means that unwanted drops are printed even when they are not required (Derby 2010). To solve this, on the one hand, CIJ printers are equipped with charging devices that administer small charges to some droplets, which then deflect into a “gutter” after passing through charge deflectors. This ink is then recycled to be printed again; however, the recycled ink is also prone to contamination when used with cells. On the other hand, DoD printing mechanism is well known for its controllability and accuracy in dispensing droplets with well-defined volumes and shapes. In this process, the droplet is only dispensed when needed, and the process is further divided based on the mechanism of droplet ejection. The pressure pulse is required for administering actuation through either thermal or piezoelectric modalities. In thermal printing, the thin-film heater heats up the fluid in the nozzle to boiling temperatures, and this leads to bubble formation. Upon turning the heater off, the bubble collapses leading to the generation of the pressure pulse necessary for dispensing the droplet. Although, a considerably high temperature is applied to the fluid (>300 °C), Xu et al. have demonstrated cell printing through this method with maintained viability of cells (Xu et al. 2005). Piezoelectric printing entails the direct administration of mechanical actuation using a piezoelectric transducer (Derby 2010). Nakamura et al. demonstrated the printing of individual vascular endothelial cells at a micrometer resolution to mimic the organization of biological tissues using electrostatic actuation (Nakamura et al. 2005). Electrohydrodynamic printing is a relatively new method in which the ink is “pulled out” instead of the push mechanism of conventional inkjet-printing systems (Choi et al. n.d.). A voltage is applied to the nozzle, while the collector is grounded. When the electric force overcomes the surface tension of the fluid, a jet is formed that is attracted toward the collector. Recently, Kim et al. demonstrated the precise patterning of bacterial cells using electrohydrodynamic printing (Kim et al. 2009). Inkjet printers have been instrumental for enabling applications requiring the dispensing of “one cell, one wall.” In a study by Liberski et al., up to 50 drops containing single cells were dispensed at certain locations onto a substrate followed by the addition of paraffin oil to prevent droplet evaporation (Liberski et al. 2011). Park et al. demonstrated a way to pattern single cells in a free-form manner so as to mimic complex tissue architecture by integrating a piezoelectric inkjet-printing nozzle with a computer-controlled X-Y stage and Z-axis (Park et al. 2017) (Fig. 4b). Cell heterogeneity is a significant area of research where the isolation and patterning of single cells is required. Toward this, inkjet printing has been a dominant tool. An approach by Wang et al. demonstrated the capture of K562 cells with PBS in a cell-sized well, which was designated as a primary droplet and was sealed with fluorinated oil (Wang et al. 2017). Postprinting biological analysis of single cells is a difficult task owing to the lack of space between wells. In the past, this has restricted the amount of reagent that could be used for such experiments since the well has to be the approximate size of a single cell to maintain a single-cell

Fig. 4 The microenvironment of the individual cell affects cell morphology. (a) Cells attain a polarity and spread on single when dispensed onto a flat (2D) substrate versus a natural 3D microenvironment. (Reproduced with permission from Baker and Chen (2012)); (b) describes a freeform, direct-inkjet cellprinting method. (Reproduced with permission from Park et al. (2017)); and (c) describes the three most common droplet-generation mechanisms: (L) T-junction, (M) coflow, and (R) flow-focusing. (Reproduced with permission from Li et al. 2018)

382 P. Ambhorkar et al.

15

Technologies for Single-Cell Printing and Patterning

383

colony (circumventing the Poisson restriction). In this study, the authors used a piezoelectric printer to inject the cell-lysis buffer and a fluorogenic substrate (reagent) through the sealing oil, and into the primary droplet. In the process, the authors could successfully study the intracellular β-galactosidase activity of the K562 cells at a single-cell resolution by sequentially adding reagents for the study. Recently, Yoon et al. demonstrated an automated piezoelectric inkjet printer that was used to analyze the intratumoral heterogeneity in bladder cancer (Yoon et al. 2020). This was done by retrieving organoids from patients and dissociating them into a single cell suspension for use as a bioink. Followed by this, single cells were precisely printed onto a microwell plate using a DoD printer and were studied for protein expression, mRNA expression, morphology, and drug testing. Most importantly, it was observed that the cell viability of the clonal cells remained high post printing, thus demonstrating the safety of the inkjet-printing process. While inkjet printers have been used extensively for dispensing single cells, they have also been utilized in printing biomolecular patterns for creating single cell arrays as an alternative to microcontact printing. In an approach described by Sun et al, by integrating a narrow microchannel with inkjet printing, precise protein spots were dispensed on a superhydrophobic surface (Sun et al. 2018). The superhydrophobic surface with its characteristic high-contact angle and high-contact angle hysteresis prevented the droplets from bouncing and therefore maintained the resolution. Such a method eliminates the requirement for a mask and enables the rapid manufacture of single cell arrays for studying cell phenomena. Inkjet printing offers a fast method for the isolation and dispensing single cells; however, the technique has several limitations including excessive shear forces and heat that is applied as part of piezoelectric and thermal printing methods, which can damage cells (Wang et al. 2016). Since cells are printed within viscous liquids, the nozzles are also susceptible to clogging which can interrupt with the dispensing process.

Microcontact Printing Microcontact printing is a simple but valuable technique that was introduced by the Whitesides research group in 1993 (Kumar et al. 1994). Generally, this printing protocol involves preparing a stamp made of an elastomeric material such as polydimethylsiloxane (PDMS), by casting it in a master mold. Figure 4a outlines a typical process for microcontact printing (Kane et al. 1999). Once the stamp is prepared, it is inked and left to dry. In this state, the patterned ink can be transferred onto another substrate via contact (Kumar and Whitesides 1993). Microcontact printing sought to offer a more cost- and resource-efficient alternative to photolithography-based cell patterning which requires large equipment, while also introducing the possibility of adding chemical and biological functionalities to surfaces. While this technique was initially used to pattern gold, it has been applied for patterning proteins and cells for various applications including biosensing, fundamental cell biology, and tissue engineering (Kane et al. 1999).

384

P. Ambhorkar et al.

The primary enabling factor for cell patterning using this technique is its ability to precisely place self-assembled monolayers (SAMs) on substrates. SAMs are molecular assemblies that are formed due to adsorption of active surfactants onto surfaces (Ulman 1996) and possess functional groups on their tail that serve as adhesion points for proteins. Patterning cells on a desired substrate demands that its surface has functional anchoring units onto which cells attach and proliferate. In the native ECM, this anchorage is provided to cells by extracellular matrix (ECM) proteins such as fibronectin (Frantz et al. 2010), laminin, and vitronectin. For microcontact printing of cells, therefore, the choice of molecule for determining cell attachment in a defined pattern is especially important. In an early study, investigating protein adsorption onto patterns includes the microcontact printing of patterns onto gold surfaces consisting of tails terminating with protein-resisting oligo (ethylene glycol) groups and methyl groups that allow the adsorption of proteins. Upon the immersion of these patterned SAMs in protein solutions including those of fibronectin, fibrinogen, and immunoglobins, the methyl groups were found to attach to proteins, while the protein-resisting group did not display any adsorption (López et al. 1993), thus laying the foundation for patterning cells that could attach to protein-based inks micropatterned onto these substrates. With this technique, it is possible to generate cell patterns at multiple scales, from multiple cells down to the placement of individual cells. The general workflow for realizing a single-cell pattern via microcontact printing consists of: (1) generating an adhesive pattern on a substrate by stamping the protein-solution-based “ink,” (2) rendering the nonpatterned area on the substrate, biologically inactive, and finally, (3) seeding the cells onto the substrate and washing away nonadhering cells, thus yielding the desired pattern (D’Arcangelo and McGuigan 2015). Micropatterning at the resolution of the single-cell resolution demands that the area of the adhesive pattern be restricted to be smaller than its spreading area. While the average animal (eukaryotic) cell is 10–20 μm in diameter, some cells can grow up to 100 μm (Guertin and Sabatini 2006). Therefore, in most cases requiring the patterning of single-cells, geometrical constraints of 5–40 μm are enough. Azioune et al. demonstrated patterning of single cells on a glass that was covered with a layer of poly-L-lysine-graftedpolyethylene glycol (PLL-g-PEG), onto which arrow, circle, square, and tear drop shapes were patterned using a photomask under UV light (Azioune et al. 2009). The patterned substrate was then incubated in a mixture of fibrinogen and Invitrogen, followed by the seeding of cells. As the cells were geometrically restricted to the area where the proteins were adsorbed, their growth area was confined to the respective shape of the protein pattern. The ability to yield singlecell patterns accurately, without the need for contact during dispensing which exercises stresses on cells, has enabled a wide range of applications including the studies concerning cell migration , mechanotransduction (Parker et al. 2002), cell signaling (McWhorter et al. 2013), and gene expression (Thomas et al. 2002). The technique has also been simplified through the introduction of automated and robotic platforms for microstamping that offer a greater degree of precision and accuracy (Lagraulet et al. 2015; McNulty et al. 2014).

15

Technologies for Single-Cell Printing and Patterning

385

Traditionally, using microcontact printing methods does not result in high throughput. As a solution, Wu et al. presented a negative microcontact printing approach where hydrophilic polydopamine (PDA) was patterned at a micrometer resolution on hydrophobic substrates (Wu et al. 2018). The method involves the formation of a PDA film on the hydrophobic surface, which is then followed by placing a PDMS stamp on this surface. Due to the variation in the surface energies of the stamp and the surface, the PDA comes off the hydrophobic surface in compliment to the stamp. This negatively micropatterned surface could be used to print arrays of single cells at a 94% efficiency. In order to create large scale arrays, Foncy et al. described a low-cost, automated method for microcontact printing (Foncy et al. 2018). In this process, a magnetic field is used to handle PDMS stamps which allows for the printing of defined biomolecule patterns (e.g., fibronectin). The substrate is then put through an antifouling treatment to discourage the presence of cells in nonpatterned areas, and the method is complete with the seeding of cells. Interestingly, post seeding, the microarrays were also studied for selectivity based on the shape of the pattern, where it was observed that lined patterns showed the highest adhesivity (99%) as compared to dots (91%), squares (85%), and triangles (75%). This meant that the higher the spatial confinement was, the lower the adhesivity would be. Lithography is an integral part of the microcontact technique as it is used to pattern the stamps; however, this step is usually expensive and requires a lithography setup in the lab. Khadpekar et al. devised a lithography-free method, making use of objects such as polystyrene beads and injection needles to pattern stamps for printing complex patterns for cell adhesion (Khadpekar et al. 2019). While microcontact printing has faced several challenges including throughput and cost effectiveness, due to its inherent versatility and ease of use, it remains a dominant method to create single-cell arrays for cell-based research studies.

Droplet-Based Patterning of Single Cells After discussing single-cell printing technologies, it is also important to shed some light on the substrate onto which these cells are being dispensed. Over the last few years, research has uncovered the effects the substrate environment has on cell behavior. In native in vivo structures, cells are housed in a fibrous 3D microenvironment known as the extracellular matrix (ECM). This matrix protects and supports cells by providing biochemical and mechanical cues that dictate their function (Theocharis et al. 2016). However, most contemporary cell-printing and patterning substrates are two-dimensional in nature as opposed to the three-dimensional environments which have been known to drastically affect cell phenotype, morphology, and migration (Baker and Chen 2012) (see Fig. 4a). Generally, cells extracted from tissues that are cultured on 2D surfaces have been observed to lose their differentiated phenotype and start taking on a flatter morphology as opposed to a rounder one in 3D microenvironments. This change in morphology induces and forces what is known as apical-basal polarity to the cells which can significantly alter the way signals are propagated through cells. In a study probing the effect of culturing

386

P. Ambhorkar et al.

chondrocytes in a 3D GelMA hydrogel, it was observed that cells cultured in a hydrogel with lower stiffness took on an elongated shape and demonstrated a lower chondrogenic gene expression as opposed to the rounder shape and higher expression observed in hydrogels with higher stiffness (Li et al. 2017). Given these drastic effects of dimensionality in cell culture, it is important to consider the postprinting behavior of cells when printing and patterning single cells. Furthermore, it is also important that when these cells are being dispensed, they are separated from each other to avoid contamination. The integration of microfluidic droplet generation methods such as flow-focusing (Yang et al. 2015), coflowing (Nisisako and Tonii 2007), and T-junctions (van Steijn et al. 2013) with single-cell encapsulation and printing has enabled the high throughput manufacture of cellladen microdroplets (Li et al. 2018) (see Fig. 4c). In T-junction-configured devices, the disperse phase is introduced through one inlet while the continuous phase is flown in the main channel (Mohamed et al. 2020). The two flows intersect at a junction where due to the shear forces introduced by the continuous flow, necking is observed in the disperse phase and a droplet is generated (Chakraborty et al. 2019). The size of these droplets is defined by flow rates and viscosities of the phases and can be accommodated for the purposes of encapsulating single cells. Flow-focusing devices use the principle of hydrodynamic focusing to break off the dispersed phase by a continuous phase (Mohamed et al. 2019). Similarly, in microfluidic devices using the coflowing principle, concentric glass capillaries (inner placed into an outer) are used. In this method, both phases flow in the same direction, i.e., the dispersed flow, usually containing the hydrogel precursor and cells flows through the inner capillary while the continuous flow is flown between the inner and outer capillaries (Zhao et al. 2016). In these methods, the deterministic encapsulation of single cells is difficult since the process is governed by Poisson’s distribution, which means that in a batch of cell-laden microgels, only about 37% of microgels will have encapsulated single cells while the remaining would either be empty or carry multiple cells (Collins et al. 2015). The solution to this is to integrate a sorting system such as FACS (fluorescence-activated cell sorting), which uses fluorescent antibodies as biomarkers for cells. An optical detector sorts microgels based on this signal resulting in a pure population of single-cell-laden microgels (Sutermaster and Darling 2019). Hydrogel-based microgels have recently emerged as a valuable platform for encapsulating cells for studying cell behavior. Recently, Kamperman et al. demonstrated the creation of single-cell-laden microgels as modular bioinks for tissue engineering (Kamperman et al. 2016). The microgel was tuned to mimic the 3D pericellular matrix of the cell and demonstrated immunoprotective properties. The authors integrated this novel concept with contemporary biofabrication techniques such as extrusion bioprinting and wet spinning. This emerging focus of technologies targeting the 3D microenvironment of the single cell shows much promise for the long-term performance of tissue-engineered scaffolds and more accurate applications in single-cell analysis.

15

Technologies for Single-Cell Printing and Patterning

387

Single Microbial Cell Printing Most of the research on tissue engineering emphasizes on the study of eukaryotic mammalian cells by examining their behavior in single-cell and multicellular cultures, cocultures, and carefully designed microenvironments. Another perspective on tissue engineering focuses on the role of host-associated microbiomes in guiding organ-level behavior. A recent study by Jameson et al. (2020) highlighted the microbial interactions with the nervous system along the gut-brain axis. Therefore, bioprinting of prokaryotic cells at single-cell level presents an interesting approach to study the role of microbiome. Additionally, beyond tissue engineering, single-cell systems of bacteria present an important tool in fields such as water treatment, biomass digestion, and genetic engineering. In this section, some of the single-cell printing techniques and applications involving bacteria are discussed. In 2013,Gross et al. (2013) presented an advanced single-cell printer (SCP) capable of separating and depositing single cell from a suspension by a piezoelectric drop-on-demand system. In a follow-up study, Riba et al. (2016) applied the SCP toward deposition of single bacterial cells (Fig. 5a–d). They demonstrated a labelfree isolation of single E. coli and E. faecalis cells from heterogenous samples (Fig. 5e). In their SCP system, the nozzle region of the dispenser was monitored by an optical microscope to obtain image data for cell detection and dispensing (Fig. 5c). By single-cell dispensing, a twofold higher clonal yield was shown as compared to statistical cell isolation. This could present a significant advantage in complex systems such as human gut where a diverse population of bacteria is present. Further, if SCP approach is utilized, microbiome from different hosts can be handled in a highly efficient and precise manner. Alternatively, membraneseparated coculture systems could also be designed to study the indirect interaction pathways between host-specific gut bacteria and cells isolated from the host. Generally, in a bacterial culture, different strains are present which can present a heterogenous behavior of the colony. Various applications, such as antibiotics production, require the purified metabolite produced from specific engineered microbe strains. In addition to antibiotics production, several tissue engineering applications utilize the extracellular secretion from bacteria for scaffold fabrication (Chen et al. 2015; Kirdponpattara et al. 2015; Torgbo and Sukyai 2018). In 2014, Wang et al. (2014) applied a microfluidics-based approach for high-throughput culturing of single cells. Using a sequence of microfluidic devices, first, individual cells were encapsulated in droplets suspended in an immiscible media of fluorinated oil. These droplets were compartmentalized and cultured to examine the extracellular metabolites secretion and consumption which allowed selection of specific bacteria strains. Xylose-overconsuming Saccharomyces cerevisiae and L-lactate producing E. coli strains were isolated and enriched using this system. In an earlier work, Xu et al. (2007) addressed the hurdle of high-throughput printing of bacterial cells for the systematic study of biological systems and interactions among individual cells. They used a microcontact printing technique with high-resolution and high-aspect-ratio polydimethylsiloxane (PDMS) stamps to

P. Ambhorkar et al.

Fig. 5 (continued)

388

15

Technologies for Single-Cell Printing and Patterning

389

directly print arrays of E. coli over a large area (cm2) on agarose in several seconds. E. coli bacteria were used as the model microbe to achieve micrometer-resolution single bacteria printing. Like other microcontact printing processes, this approach relied on a reproducible PDMS stamp which was obtained by casting over a UV lithography-fabricated mold. However, to minimize diffraction blurring, the standard protocol was modified to an improved reverse in situ lithography (RISL). A layer of agarose was attached on the stamp and brought in contact with the E. coli liquid broth (LB) suspension such that the adsorption of LB resulted in a stamp pad with a monolayer of E. coli as ink. This was then transferred to new agarose substrates for further culture. This single-cell printing method and platform provided an excellent opportunity for high-throughput screening of drugs, directed evolution, and interstrain variations. Laser-induced forward transfer (LIFT) is a printing method capable of producing very high-resolution structures and patterns (Fig. 6a). LIFT has been applied to cell printing and cell isolation in several studies over the years. In the LIFT process, a high energy pulsed laser is applied on top of a donor substrate which comprises a glass layer, sacrificial layer, and cells suspended on the bottom surface (Delaporte and Alloncle 2016). The sacrificial layer is usually made of metal and absorbs the energy from laser resulting a localized melting. This creates a shockwave and pushes a tiny volume of the cell suspension effectively comprising a single cell on the receptor substrate. In 2017, Deng et al. (2017) used LIFT for transferring single Hela cells and presented a comprehensive investigation of various LIFT parameters (laser pulse energy, spot size, working distance, and the thickness of the sacrificial layer). In a seminal work in 2014 reported by Ringeisen et al. (2015), the LIFT process was used with pulsed UV laser source and a soil bioink layer as the donor suspension to print soil microsamples on a 96-well culture plate. Unlike contemporary research on bioprinting-enriched samples, this work attempted to bioprint solid-phase environmental samples to isolate microbes in an unperturbed environment. Later in 2018,

ä Fig. 5 (a) shows a scheme of the bright-field optical detection system as implemented in this work. Light of a high-power blue LED passes two apertures and illuminates the dispenser chip through a 10x objective. Light rays reflected from the chip pass a half mirror and are detected by a highresolution camera to image the dispenser chip nozzle. The single-cell printhead (b) comprises a piezo-stack actuator, the optical detection system, and the disposable cartridge including a sample reservoir and the microfluidic dispenser chip. A small mirror in front of the dispenser chip allows for tilting the optical path by 90°. On actuating the piezo-stack actuator, the piston displaces a constant volume within the chip generating a single droplet of 35 pl ejected from the nozzle. A consecutive image series (c) from each single printed cell is stored on the PC. The single-cell patterning was done on agar plates and deposition of single cells into a microwell plate (d). Bacteria colony array grown from 10  10 spotted single cells on LB-agar (e). Single bacteria cells were printed from a heterogeneous culture of E. coli and E. faecalis previously mixed in a ratio of 1:1. Two clearly distinguishable colony morphologies can be found for the two different types of bacteria. Visual inspection by light microscopy revealed that shiny sharped edge colonies were grown from E. faecalis (yellow circles) while mat colonies with diffuse edges could be assigned to E. coli (blue circles). (Reproduced with permission from Riba et al. (2016))

390

P. Ambhorkar et al.

Fig. 6 (a) Scheme of the laser microsampling setup (LMS)/ laser-induced forward transfer (LIFT). (b) Soil laser microsampling (LMS) with water (i) and with an addition of gel (ii) produced by laser printing of soil containing drops at the laser-pulse energy of 20 μJ. (c) Laser printing of soil microparticle containing 6  6 droplet arrays onto agar substrate in cases of soil mix with water (i) and with the addition of 2% gel (ii) (Gorlenko et al. 2018). (d) The main differences between the laser engineering of microbial systems (LEMS) and standard method, leading to an increase in biodiversity in the isolation of microorganisms from soil. The numbers indicate microbes that, with the standard cultivation method: (1) easy to flush out of their microenvironment, (2) most actively multiply, (3) separate from those with which they exist in symbiosis, (4) remain in the “sleeping” state. (Reproduced with permission from Yusupov et al. (2018))

Gorlenko et al. (2018) adapted the LIFT process as a high-throughput laser microsampling (LMS) method to print soil sample (Fig. 6b and c). Using the laser-based printing method, the soil microsamples were printed on a receptor substrate with effectively a single microbe per microsample. This allowed isolation of several rare microbes and generation of a microbe fingerprint equivalent to the soil sample. In recent years, LIFT printing of microbes has evolved to laser engineering of microbial systems (LEMS). Yusupov et al. (2018) bioprinted a gel/soil mixture spread on a gold-coated glass plate. With comprehensive characterization of the effect of various printing parameters on the living systems, they concluded that LEMS allowed isolation and printing of diverse systems as well as rare and unculturable microorganisms. These bioprinting methods have significantly advanced to allow fabrication of structures and patterns with extremely high complexity and resolution. Along with bioprinting of microbial systems, laser-based printing systems have also been extensively applied in printing and patterning single mammalian cells (Deng et al. 2017; Minaev et al. 2018; Yusupov et al. 2017). Additionally, these bioprinting methods offer means to address the research gap currently observed in the fabrication and study of complex mammalian cells and microbe coculture system. In future, these single-cell bioprinting approaches can lead toward platforms for patientspecific personalized medicine. Moreover, the enriched samples of rare and unculturable microbes can significantly advance the fields of biomaterials and microbe-based energy systems.

15

Technologies for Single-Cell Printing and Patterning

391

Conclusion The importance of characterizing genetic and phenotypic heterogeneity in cell populations and understanding cell behavior on the single-cell level has led to a need for better single-cell printing systems. Several commercial printing methodologies have come to existence since then, and many existing printing methods have been modified to accommodate for the precise printing of individual cells. This has been achieved by characterization methods such as fluorescence-activated cell sorting (FACS) and optical trapping, which have played a big role in yielding pure populations of isolated single cells. When printing cells, it is also important to consider the cell-microenvironment post printing. Through previous studies, it has been observed that both cell morphologies and behavior vary between 2D and 3D environments and affect gene expression. Microgel technology seeks to take this property into account to encapsulate single cells within hydrogel matrices that mimic the properties of the extracellular matrix (ECM), thus yielding more realistic data. Improvements in single-cell printing have the potential to improve how single-cell “omics” studies are carried out and have also laid the foundation for bottom-up tissue biofabrication starting at the single-cell level. Acknowledgments This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grants and Canada Foundation for Innovation John R. Evans Leaders Opportunity Fund.

References Altschuler SJ, Wu LF (2010) Cellular heterogeneity: do differences make a difference? Cell 141(4):559–563. https://doi.org/10.1016/j.cell.2010.04.033 Azioune A, Storch M, Bornens M, Théry M, Piel M (2009) Simple and rapid process for single cell micro-patterning. Lab Chip 9(11):1640. https://doi.org/10.1039/b821581m Baker BM, Chen CS (2012) Deconstructing the third dimension-how 3D culture microenvironments alter cellular cues. J Cell Sci 125:3015–3024. https://doi.org/10.1242/jcs.079509 Carter AD, Bonyadi R, Gifford ML (2013) The use of fluorescence-activated cell sorting in studying plant development and environmental responses. Int J Dev Biol 57(6–8):545–552. https://doi.org/10.1387/ijdb.130195mg Chakraborty I, Ricouvier J, Yazhgur P, Tabeling P, Leshansky AM (2019) Droplet generation at Hele-Shaw microfluidic T-junction. Phys Fluids 31(2):022010. https://doi.org/10.1063/1. 5086808 Chen C, Yu Y, Li K, Zhao M, Liu L, Yang J, . . . Sun D (2015) Facile approach to the fabrication of 3D electroconductive nanofibers with controlled size and conductivity templated by bacterial cellulose. Cellulose 22(6):3929–3939 Choi K-H, Rahman K, Muhammad NM, Khan A, Kwon K-R, Doh Y-H, Kim H-C (n.d.) Electrohydrodynamic inkjet-micro pattern fabrication for printed electronics applications. Retrieved from www.intechopen.com Collins DJ, Neild A, deMello A, Liu A-Q, Ai Y (2015) The Poisson distribution and beyond: methods for microfluidic droplet production and single cell encapsulation. Lab Chip 15(17):3439–3459. https://doi.org/10.1039/c5lc00614g

392

P. Ambhorkar et al.

D’Arcangelo E, McGuigan AP (2015) Micropatterning strategies to engineer controlled cell and tissue architecture in vitro. Biotechniques 58(1):13–23. https://doi.org/10.2144/000114245 Delaporte P, Alloncle A-P (2016) Laser-induced forward transfer: a high resolution additive manufacturing technology. Opt Laser Technol 78:33–41 Deng Y, Renaud P, Guo Z, Huang Z, Chen Y (2017) Single cell isolation process with laser induced forward transfer. J Biol Eng 11(1):2 Derby B (2010) Inkjet printing of functional and structural materials: Fluid property requirements, feature stability, and resolution. Annu Rev Mater Res 40:395–414. https://doi.org/10.1146/ annurev-matsci-070909-104502 Foncy J, Estève A, Degache A, Colin C, Cau JC, Malaquin L, . . . Trévisiol E (2018) Fabrication of biomolecule microarrays for cell immobilization using automated microcontact printing. In: Methods in molecular biology, vol 1771. Springer, New York, pp 83–95. https://doi.org/10. 1007/978-1-4939-7792-5_7 Frantz C, Stewart KM, Weaver VM (2010) The extracellular matrix at a glance. J Cell Sci 123:4195–4200. https://doi.org/10.1242/jcs.023820 Gorlenko MV, Chutko EA, Churbanova ES, Minaev NV, Kachesov KI, Lysak LV, . . . Yusupov VI (2018) Laser microsampling of soil microbial community. J Biol Eng 12(1):27 Green KT (2019) Microbial and epigenetic influences on coral reefs. Retrieved November 22, 2020, from https://digitallibrary.sdsu.edu/islandora/object/sdsu%3A26434 Gross A, Schöndube J, Niekrawitz S, Streule W, Riegger L, Zengerle R, Koltay P (2013) Single-cell printer: automated, on demand, and label free. J Lab Autom 18(6):504–518. https://doi.org/10. 1177/2211068213497204 Gross A, Schoendube J, Zimmermann S, Steeb M, Zengerle R, Koltay P (2015) Technologies for single-cell isolation. Int J Mol Sci 16(8):16897–16919. https://doi.org/10.3390/ijms160816897 Guertin DA, Sabatini DM (2006) Cell size control. In: Encyclopedia of life sciences. https://doi.org/ 10.1038/npg.els.0003359 Holmes D, Morgan H (2010) Single cell impedance cytometry for identification and counting of CD4 T-cells in human blood using impedance labels. Anal Chem 82(4):1455–1461. https://doi. org/10.1021/ac902568p Jameson KG, Olson CA, Kazmi SA, Hsiao EY (2020) Toward understanding microbiome-neuronal signaling. Mol Cell 78:577–583 Jordana SJ, Bakshib RK, Brownb LD’T, Chic X, Geisler WM (2016) Stimulated peripheral blood mononuclear cells from chlamydia-infected women release predominantly Th1-polarizing cytokines. Physiol Behav 176(1):139–148. https://doi.org/10.1016/j.physbeh.2017.03.040 Kain SR (2005) Methods and protocols. Methods Biochem Anal 47:407. https://doi.org/10.1002/ 0471739499.oth1 Kamperman T, Henke S, van den Berg A, Ryon Shin S, Tamayol A, Khademhosseini A, . . . Leijten J (2016) Single cell microgel based modular bioinks for uncoupled cellular micro-and macroenvironments. Adv Healthc Mater 6. https://doi.org/10.1002/adhm.201600913 Kane RS, Takayama S, Ostuni E, Ingber DE, Whitesides GM (1999) Patterning proteins and cells using soft lithography. Biomaterials 20(23–24):2363–2376. https://doi.org/10.1016/S01429612(99)00165-9 Kang B, Shin J, Park HJ, Rhyou C, Kang D, Lee SJ, . . . Lee H (2018) High-resolution acoustophoretic 3D cell patterning to construct functional collateral cylindroids for ischemia therapy. Nat Commun, 9(1):1–13. https://doi.org/10.1038/s41467-018-07823-5 Kerk NM, Ceserani T, Tausta SL, Sussex IM, Nelson TM (2016) Laser capture microdissection of cells from plant. Plant Physiol 132:27–35. https://doi.org/10.1104/pp.102.018127.1 Khadpekar AJ, Khan M, Sose A, Majumder A (2019) Low cost and lithography-free stamp fabrication for microcontact printing. Sci Rep 9(1):1–8. https://doi.org/10.1038/s41598-01836521-x Kim JH, Lee DY, Hwang J, Jung HI (2009) Direct pattern formation of bacterial cells using microdroplets generated by electrohydrodynamic forces. Microfluid Nanofluid 7(6):829–839. https:// doi.org/10.1007/s10404-009-0441-6

15

Technologies for Single-Cell Printing and Patterning

393

Kirdponpattara S, Khamkeaw A, Sanchavanakit N, Pavasant P, Phisalaphong M (2015) Structural modification and characterization of bacterial cellulose–alginate composite scaffolds for tissue engineering. Carbohydr Polym 132:146–155 Kolodziejczyk AA, Kim JK, Svensson V, Marioni JC, Teichmann SA (2015) The technology and biology of single-cell RNA sequencing. Mol Cell 58:610–620. https://doi.org/10.1016/j.molcel. 2015.04.005 Kudo LC, Vi N, Ma Z, Fields T, Avliyakulov NK, Haykinson MJ, . . . Karsten SL (2012) Novel cell and tissue acquisition system (CTAS): microdissection of live and frozen brain tissues. PLoS ONE 7(7):e41564. https://doi.org/10.1371/journal.pone.0041564 Kumar A, Whitesides GM (1993) Features of gold having micrometer to centimeter dimensions can be formed through a combination of stamping with an elastomeric stamp and an alkanethiol “ink” followed by chemical etching. Appl Phys Lett 63(14):2002–2004. https://doi.org/10.1063/ 1.110628 Kumar A, Biebuyck HA, Whitesides GM (1994) Patterning self-assembled monolayers: applications in materials science. Langmuir 10(5):1498–1511. https://doi.org/10.1021/la00017a030 Kummer S, Wilichowski E (2018) Combination of microdissection and single cell quantitative realtime PCR revealed intercellular mitochondrial DNA heterogeneities in fibroblasts of KearnsSayre syndrome patients. Mitochondrion 43(January):37–42. https://doi.org/10.1016/j.mito. 2018.08.004 Lagraulet A, Foncy J, Berteloite B, Esteve A, Blatche M-C, Malaquin L, Vieu C (2015) InnoStamp 40TM and InnoScan 1100ALTM: a complete automated platform for microstructured cell arrays. Nat Methods 12(9):i–iii. https://doi.org/10.1038/nmeth.f.383 Li X, Chen Y, Kawazoe N, Chen G (2017) Influence of microporous gelatin hydrogels on chondrocyte functions. J Mater Chem B 5(29):5753–5762. https://doi.org/10.1039/c7tb01350g Li W, Lee H, Weitz DA, Zhang L, Ge X, Xu B, . . . Zhang A (2018) Microfluidic fabrication of microparticles for biomedical applications. Chem Soc Rev 47:5646–5683. https://doi.org/10. 1039/c7cs00263g Liberski AR, Delaney JT, Schubert US (2011) “One Cell One Well”: a new approach to inkjet printing single cell microarrays. ACS Comb Sci 13(2):190–195. https://doi.org/10.1021/co100061c Lindström S, Andersson-Svahn H (2012) Single-cell culture in microwells. Methods Mol Biol 853:41–52. https://doi.org/10.1007/978-1-61779-567-1_5 Liu L, Cheung TH, Charville GW, Rando TA (2015) Isolation of skeletal muscle stem cells by fluorescence-activated cell sorting. Nat Protoc 10(10):1612–1624. https://doi.org/10.1038/ nprot.2015.110 López GP, Biebuyck HA, Härter R, Kumar A, Whitesides GM (1993) Fabrication and imaging of two-dimensional patterns of proteins adsorbed on self-assembled monolayers by scanning electron microscopy. J Am Chem Soc 115(23):10774–10781. https://doi.org/10.1021/ ja00076a038 Martinez-Rivas A, González-Quijano GK, Proa-Coronado S, Séverac C, Dague E (2017) Methods of micropatterning and manipulation of cells for biomedical applications. Micromachines 8:347. https://doi.org/10.3390/mi8120347 McNulty JD, Klann T, Sha J, Salick M, Knight GT, Turng LS, Ashton RS (2014) High-precision robotic microcontact printing (R-μCP) utilizing a vision guided selectively compliant articulated robotic arm. Lab Chip 14(11):1923–1930. https://doi.org/10.1039/c3lc51137e McWhorter FY, Wang T, Nguyen P, Chung T, Liu WF (2013) Modulation of macrophage phenotype by cell shape. Proc Natl Acad Sci U S A 110(43):17253–17258. https://doi.org/10.1073/pnas. 1308887110 Minaev NV, Yusupov VI, Zhigarkov VS, Churbanova ES, Cheptsov VS, Gorlenko MV, Bagratashvili VN (2018) Laser printing of gel microdrops with living cells and microorganisms. KnE Energy 3:23–31 Mohamed MGA, Kheiri S, Islam S, Kumar H, Yang A, Kim K (2019) An integrated microfluidic flow-focusing platform for on-chip fabrication and filtration of cell-laden microgels. Lab Chip 19(9):1621–1632. https://doi.org/10.1039/C9LC00073A

394

P. Ambhorkar et al.

Mohamed MGA, Ambhorkar P, Samanipour R, Yang A, Ghafoor A, Kim K (2020) Microfluidicsbased fabrication of cell-laden microgels. Biomicrofluidics 14:021501. https://doi.org/10.1063/ 1.5134060 Nakamura M, Kobayashi A, Takagi F, Watanabe A, Hiruma Y, Ohuchi K, . . . Takatani S (2005) Biocompatible inkjet printing technique for designed seeding of individual living cells. Tissue Eng 11(11–12):1658–1666. https://doi.org/10.1089/ten.2005.11.1658 Nisisako T, Tonii T (2007) Formation of biphasic janus droplets in a microfabricated channel for the synthesis of shape-controlled polymer microparticles. Adv Mater 19(11):1489–1493. https:// doi.org/10.1002/adma.200700272 Park JA, Yoon S, Kwon J, Now H, Kim YK, Kim WJ, . . . Jung S (2017) Freeform micropatterning of living cells into cell culture medium using direct inkjet printing. Sci Rep, 7(1):14610. https:// doi.org/10.1038/s41598-017-14726-w Parker KK, Brock AL, Brangwynne C, Mannix RJ, Wang N, Ostuni E, . . . Ingber DE (2002) Directional control of lamellipodia extension by constraining cell shape and orienting cell tractional forces. FASEB J Off Publ Fed Am Soc Exp Biol 16(10):1195–1204. https://doi.org/ 10.1096/fj.02-0038com Popović DM (2013) Current advances in research of cytochrome c oxidase. Amino Acids 45(5):1073–1087. https://doi.org/10.1007/s00726-013-1585-y Riba J, Gleichmann T, Zimmermann S, Zengerle R, Koltay P (2016) Label-free isolation and deposition of single bacterial cells from heterogeneous samples for clonal culturing. Sci Rep 6(1):1–9 Ringeisen BR, Rincon K, Fitzgerald LA, Fulmer PA, Wu PK (2015) Printing soil: a single-step, high-throughput method to isolate micro-organisms and near-neighbour microbial consortia from a complex environmental sample. Methods Ecol Evol 6(2):209–217 Schoendube J, Wright D, Zengerle R, Koltay P (2015) Single-cell printing based on impedance detection. Biomicrofluidics 9(1):014117. https://doi.org/10.1063/1.4907896 Sun Y, Song W, Sun X, Zhang S (2018) Inkjet-printing patterned chip on sticky superhydrophobic surface for high-efficiency single-cell array trapping and real-time observation of cellular apoptosis. ACS Appl Mater Interfaces 10:31054–31060. https://doi.org/10.1021/acsami. 8b10703 Sutermaster BA, Darling EM (2019) Considerations for high-yield, high-throughput cell enrichment: fluorescence versus magnetic sorting. Sci Rep 9(1):1–9. https://doi.org/10.1038/s41598018-36698-1 Theocharis AD, Skandalis SS, Gialeli C, Karamanos NK (2016) Extracellular matrix structure. Adv Drug Deliv Rev 97:4–27. https://doi.org/10.1016/j.addr.2015.11.001 Thomas CH, Collier JH, Sfeir CS, Healy KE (2002) Engineering gene expression and protein synthesis by modulation of nuclear shape. Proc Natl Acad Sci U S A 99(4):1972–1977. https:// doi.org/10.1073/pnas.032668799 Torgbo S, Sukyai P (2018) Bacterial cellulose-based scaffold materials for bone tissue engineering. Appl Mater Today 11:34–49 Ulman A (1996) Formation and structure of self-assembled monolayers. Chem Rev 96(4):1533– 1554. https://doi.org/10.1021/cr9502357 van Steijn V, Korczyk PM, Derzsi L, Abate AR, Weitz DA, Garstecki P (2013) Block-and-break generation of microdroplets with fixed volume. Biomicrofluidics 7(2):24108. https://doi.org/10. 1063/1.4801637 Wang K, Lau TY, Morales M, Mont EK, Straus SE (2005) Laser-capture microdissection: refining estimates of the quantity and distribution of latent herpes simplex virus 1 and varicella-zoster virus DNA in human trigeminal ganglia at the single-cell level. J Virol 79(22):14079–14087. https://doi.org/10.1128/jvi.79.22.14079-14087.2005 Wang BL, Ghaderi A, Zhou H, Agresti J, Weitz DA, Fink GR, Stephanopoulos G (2014) Microfluidic high-throughput culturing of single cells for selection based on extracellular metabolite production or consumption. Nat Biotechnol 32(5):473

15

Technologies for Single-Cell Printing and Patterning

395

Wang X, Ao Q, Tian X, Fan J, Wei Y, Hou W, . . . Bai S (2016) 3D bioprinting technologies for hard tissue and organ engineering. Materials 9:802 https://doi.org/10.3390/ma9100802 Wang C, Liu W, Tan M, Sun H, Yu Y (2017) An open-pattern droplet-in-oil planar array for single cell analysis based on sequential inkjet printing technology. Biomicrofluidics 11(4):044106. https://doi.org/10.1063/1.4995294 Wu H, Wu L, Zhou X, Liu B, Zheng B (2018) Patterning hydrophobic surfaces by negative microcontact printing and its applications. Small 14(38):1802128. https://doi.org/10.1002/ smll.201802128 Xu T, Jin J, Gregory C, Hickman JJ, Boland T (2005) Inkjet printing of viable mammalian cells. Biomaterials 26(1):93–99. https://doi.org/10.1016/j.biomaterials.2004.04.011 Xu L, Robert L, Ouyang Q, Taddei F, Chen Y, Lindner AB, Baigl D (2007) Microcontact printing of living bacteria arrays with cellular resolution. Nano Lett 7(7):2068–2072 Xu T, Zhao W, Zhu J-M, Albanna MZ, Yoo JJ, Atala A (2013) Complex heterogeneous tissue constructs containing multiple cell types prepared by inkjet printing technology. Biomaterials 34(1):130–139. https://doi.org/10.1016/j.biomaterials.2012.09.035 Yang CG, Pan RY, Xu ZR (2015) A single-cell encapsulation method based on a microfluidic multistep droplet splitting system. Chin Chem Lett 26(12):1450–1454. https://doi.org/10.1016/j. cclet.2015.10.016 Yoon WH, Lee H-R, Kim S, Kim E, Ku JH, Shin K, Jung S (2020) Use of inkjet-printed single cells to quantify intratumoral heterogeneity. Biofabrication 12:035030. https://doi.org/10.1088/17585090/ab9491 Yusof A, Keegan H, Spillane CD, Sheils OM, Martin CM, O’Leary JJ, . . . Koltay P (2011) Inkjetlike printing of single-cells. Lab Chip 11(14):2447–2454. https://doi.org/10.1039/c1lc20176j Yusupov VI, Zhigar’kov VS, Churbanova ES, Chutko EA, Evlashin SA, Gorlenko MV, . . . Bagratashvili VN (2017) Laser-induced transfer of gel microdroplets for cell printing. Quantum Electron 47(12):1158 Yusupov VI, Gorlenko MV, Cheptsov VS, Minaev NV, Churbanova ES, Zhigarkov VS, . . . Bagratashvili VN (2018) Laser engineering of microbial systems. Laser Phys Lett 15(6):65604 Zhao X, Liu S, Yildirimer L, Zhao H, Ding R, Wang H, . . . Weitz D (2016) Injectable stem cellladen photocrosslinkable microspheres fabricated using microfluidics for rapid generation of osteogenic tissue constructs. Adv Funct Mater 26(17):2809–2819. https://doi.org/10.1002/ adfm.201504943

Microfluidic Device with Removable Electrodes for Single Cell Electrical Characterization

16

Muhammad Asraf Mansor and Mohd Ridzuan Ahmad

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principle of Impedance Measurement inside Microfluidic Channel . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidic Device and Impedance Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Device Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Device Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electrical Measurement Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impedance Measurement inside Microfluidic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impedance of Cell Population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Impedance of Single Particle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

398 400 401 401 401 402 402 404 406 406 407 408

Abstract

A microfluidic device for detecting a single particle has emerged as a noninvasive technique for diagnostic and prognostic patients with cancer suspected. Microfluidic impedance cytometry has been utilized to detect and measure the electrical impedance of single biological particles at high speed. The details information of single cells such as cell size, membrane capacitance, and cytoplasm conductivity also can be obtained by impedance measurement over a wide frequency range. In this work, an integrated microneedle microfluidic device to detect and discriminate 9 μm and 16 μm microbeads was developed. Two microneedles at half the height of the microfluidic device were used as a measuring electrode to determine electrical impedance when cells were present in the sensing field. This device was also capable of distinguishing M. A. Mansor · M. R. Ahmad (*) Micro-Nano System Engineering Research Group, Division of Control and Mechatronics Engineering, School of Electrical Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, Skudai, Johor, Malaysia e-mail: [email protected]; [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_19

397

398

M. A. Mansor and M. R. Ahmad

the concentration of cells in the suspension fluid. It was easy to remove the microneedles from the disposable microchannel of PDMS. In addition, with a simple cleaning process, such as ultrasonic cleaning, the microneedles can be reused. While this tool is low cost, it retains the sensor’s functionality, which can detect the single particle in the sensing region. This device is therefore suitable for cost-effective screening and testing of medical and food safety in developing countries. Keywords

Impedance · Flow cytometry · Microfluidics · Microneedle · Single cell

Introduction Single cell analysis (SCA) was emphasized to provide peer-to-peer biologists and scientists into individual cell molecular machinery. Detection of cancer cells and pathogenic cells in the blood is used as a measure of infectious disease for the application of medical diagnosis. Detection of circulating tumor cells (CTCs) in the blood is documented to have been medically essential for early stage metastasis or cancer recurrence. Rare blood CTCs range from just 1 to 100 CTCs/ml of blood (Kantara et al. 2015). Plasmodium falciparum malaria, which mainly kills children in developing countries, has infected patients’ blood samples at a concentration of approximately 1/50 μl of blood (Ciceron et al. 1999). Single cell analysis in biological measurements and medical research has now emerged as a distinct new field and has been recognized as one of life’s fundamental building blocks (Gilchrist et al. 2001). Cell impedance measurement has become an important biological measurement tool among the different single cell analyses (Jao et al. 2011). The physiological behavior of the cells and their related molecular expressions have a significant effect on the conductivity and dielectric constant of the cell membrane and cytoplasm, which in turn affect the overall impedance characteristics (Yang et al. 2011). For this purpose, single-cell impedance measurements can provide relevant information about their functional status and can be an easy and substantially less complex alternative to comprehensive molecular expression studies. Flow cytometry, which is a fast and highly accurate measuring technique, is the standard method of cell detection in suspension. Impedance flow cytometry is an indirect signal collection on the microchannel-sensing field from the single cells without direct access to the cells’ intracellular region (Sun and Morgan 2010). Coulter (1956) first documented these techniques in the microfabrication system to analyze high-sensitivity microscale particles. Flow cytometry, however, requires expensive processing and marking of fluorescent antibodies in cells (Holmes and Morgan 2010). The impedance flow cytometry (IFC) has recently gained interest in replacing and overcoming the limitations associated with flow cytometry with the important promising techniques. Due to quick, real-time, and noninvasive methods of biological detection, the IFC is preferable. This technique can be used to count

16

Microfluidic Device with Removable Electrodes for Single Cell. . .

399

cells (Holmes and Morgan 2010), detect cancer cells (Gou et al. 2011), and detect bacteria (Du et al. 2013). Some groups demonstrated cell detection and counting by using an electrode-integrated microfluidic for different methods of electrical measurement in the field of food safety (Liu et al. 2008) and bio-threat monitoring in real time (Liu et al. 2008). This measuring technique is based on altering impedance across a measuring electrode due to blocking the passage of ionic current between electrodes when the cells are present. The IFC will recognize and count lymphocytes, monocytes, and neutrophils in the entire blood of humans (Holmes and Morgan 2010). Other studies indicated that IFC was able to detect the presence of cells based on measuring the cell’s impedance at frequencies greater than 1 MHz (Gawad et al. 2001). The developed microfluidic nanoneedle probe was used to test the concentration of cells on the surface of the sensor and to make it sensitive to the dielectric properties of the solution (Esfandyarpour et al. 2014). This tool, however, includes electrode or probe patterning on the substrate resulting in higher manufacturing process costs. The time-consuming cleaning process of the system often needs to be considered as another constraint. Some groups showed the technique of reducing the cost of electrode microfabrication by using the PCB as a measuring electrode. In capillary electrophoresis manners (Wang et al. 2001) and cell manipulation using dielectrophoresis (Park et al. 2009), they demonstrated contactless detection of conductivity. The contactless impedance cytometry has recently been introduced to reduce the manufacturing cost of the cytometric impedance device (Emaminejad et al. 2016; Emaminejad et al. 2012). The electrode was manufactured on the PCB substrate (reusable component), and the thin bare dielectric substrate attached to a PDMS microchannel (disposable component) was mounted on the PCB substrate. The sensitivity of this device is the limitation since the electric field was buried in dielectric substrate and does not reach the electrolyte. Several designs and methods have been documented in IFC for the detection and analysis of a cell (Mansor and Ahmad 2016; Mansor and Ahmad 2015). This chapter describes a new integrated microneedle-microfluidic method for detecting the concentration of yeast cells in suspension and detecting a single particle based on the measurement of impedance. The device’s design is focused on reducing manufacturing costs while retaining the key features, i.e., cell detection. In this study, the significant reduction for manufacturing costs is by replacing electrode microfabrication with microneedles. This device utilized a Tungsten microneedle as a measurement electrode which can be reused and easy to be cleaned. To detect and allow impedance measurement of passing cells through the applied electric field, the two microneedles were positioned at half height disposable microchannel. Figure 1a outlined the schematic diagram of the proposed microfluidic chip consisting of two integrated microneedles on both sides of the microchannel. The length, width, and thickness of the central microchannel sensing region is 100 μm, 25 μm, and 25 μm, respectively. The device is ideal for early implementation of cancer cell detection in developing countries as it significantly reduces manufacturing costs, i.e., it does not require microelectrode manufacturing.

400

M. A. Mansor and M. R. Ahmad

Fig. 1 (a) 3D schematic diagram of the microfluidic device structure integrated with microneedle and top view of the sensing area which the impedance measurement of single particle be measured. (b) An equivalent circuit model of sensing area of microfluidic chip

Principle of Impedance Measurement inside Microfluidic Channel The fundamental principle of the detection of suspended biological cells in the media is based on Ohm’s law. To evaluate the changing media impedance value, an AC current with a frequency sweep was applied through passing cells. An equivalent circuit model was developed, as shown in Fig. 1b, to understand the interface between two microneedles and the suspension media. The microfluidic chip’s sensing area can be electrically modeled as cell impedance of cell resistance Rp and cell capacitance Cp in parallel with the impedance contributed by all materials between the two electrodes that consist of solution resistor Rm in parallel with capacitance double layer Cdl. Both impedances in series

16

Microfluidic Device with Removable Electrodes for Single Cell. . .

401

with a pair of electrodes resistance Re. The ZT measuring system’s overall impedance is given by (Sun et al. 2007) Z T ¼ 2Re þ

Rm Rp
Rm þ Rp þ jωRm Rp C dl þ C p

ð1Þ

where ω is an angular frequency of the electrical signal. Consequently, ZT varies in the sensing region depending on the presence of cells. Equation (1) has been simplified to Eq. (2) to evaluate the impedance value of single particle that occurs between the pair of electrodes: Z cell 5

2Re Z m  Z T Z m Z T  2Re  Z m

ð2Þ

where Zcell is the total impedance of cell resistance Rp and cell capacitance Cp in parallel. Meanwhile, Zm is the total impedance of the solution resistor Rm parallel to the double layer capacitance Cdl.. This work focuses on the impedance of a single particle.

Microfluidic Device and Impedance Measurement Cell Culture As a framework for proof of concepts, Saccharomyces cerevisiae cells and microparticle are used in the present study. Saccharomyces cerevisiae was grown in a 10 ml YEPD (yeast extract peptone dextrose) petri dish. The broth of the YEPD contained 1% extract of yeast, 2% peptone, and 2% glucose. For 24 h, the YEPD plates were incubated at 37  C. By centrifugation, the cells were washed with deionized (DI) water three times and then suspended at various dilutions (1:10) in sterilized deionized water. To determine the number of cells, the cells were incubated on agar plates at 37  C for 24 h. The yeast cell diameter ranges between 4 μm and 7 μm. The number of cells was 1.3  109 per milliliter (cfu/ml) colony-forming units. DI water’s conductivity is 6 mS/m. Diluted to a final concentration of 1000 beads per mL were the non-fluorescent polystyrene (PS) microbeads with diameter 15 μm and 9 μm (Polysciences, Inc.) suspended in phosphate-buffered saline (PBS) solution. Polystyrene beads have a known size and electrical properties (Spencer et al. 2014), and their impedance is constant across the frequency range used in these experiments.

Device Fabrication The microfluidic device was produced using a technique of photolithography. The fabrication begins with the design of the masks using the software of the layout

402

M. A. Mansor and M. R. Ahmad

editor. A laser lithography method (μPG501, Heidelberg Instruments, Germany) has written two masks (top and bottom) on the chromium (Cr) plates. The top-layer mold was created using SU-82025 negative photoresist (MicroChem, USA) using two-step photolithography. The first layer encompasses a thickness of 25 μm and was spin coated onto a silicon substrate. Using a mask aligner (SUSS MicroTec MA-6) and post-baking with growth, the first-layer photoresist was exposed to top-layer Cr mask after pre-baking. Then, on the first photoresist layer and pre-baking, the second layer with a thickness of 60 μm was spin coated. The bottom layer of the Cr mask was matched with the first photoresist layer’s substrate and exposed to UV light. The photoresist substrate was post-bake and developed to gain a top mold master. Following the photolithography stage for the top mold master of the SU-8 microchannel, the 60 μm thick bottom mold master was produced. PDMS was prepared using a rigorous mixing of PDMS pre-polymers (SYLGARD184A) with curing agents (SYLGARD 184B) in a weight ratio of 10:1 and poured on a SU-8 mold master (top and bottom mold master). After overnight treatment at room temperature, PDMS microchannel was obtained. In order to increase binding power, the top-side PDMS and bottom-side PDMS were washed with isopropyl alcohol and treated with oxygen plasma (Plasma Etch PE-25) for 25 s (Mansor et al. 2017). In less than 2 min, the alignment of both side PDMS channels was completed to prevent the loss of effectiveness of oxygen plasma. Lastly, the microchannel chip’s right and left sides were cut, and a square (60 μm  120 μm) hole was left to insert a microneedle. A commercially available Parylene-coated tungsten needle (Signatone) was used as an electrode of measurement (two microneedles). Tungsten needle’s tip diameter, shank diameter, and length are, respectively, 25 μm, 250 μm, and 31.7 mm.

Device Operation To monitor the sensing region, the microchannel chip system was put under a microscope (Olympus Inverted Microscopes IX71). Two microneedles held by the micromanipulator (EB-700, Everbeing) were inserted through the square hole on the left and right side of the chip into the microchannel chip. For this experiment, the distance between microneedles was set at 20 μm. Figure 2 shows the schematic of the experimental setup. The Hioki IM3570 impedance analyzer was connected as an input to two microneedles and displayed on the computer. The solution and concentration of yeast samples were introduced using 3 ml syringes operated by syringe pumps (KDS LEGATO 111, KD Scientific, USA). Two Tygon flexible tubes, connected to the syringes and waste bottle, were inserted into the PDMS layer at the inlets for the introduction of liquids and outlet for waste liquids.

Electrical Measurement Procedure The impedance analyzer was self-calibrated to conduct the measurement using short and open standard calibration. In addition, impedance of 1xPBS solution was

16

Microfluidic Device with Removable Electrodes for Single Cell. . .

403

Fig. 2 A schematic diagram of the experimental setup

measured at the electrode gap of 20 μm to calibrate the chip. All experiments were carried out at room temperature, and three microfluidic devices were used for testing on the device for reproducibility. Impedance of the medium between the microneedles was measured to validate the equivalent circuit model. Sterilized DI water and PBS with conductivities of 6 mS/m and 1.4 S/m, respectively, were the solutions. Initially, 1 ml of PBS was prepared for the chip cleaning process at concentrations of 1500 mOsm. The sample was loaded into a syringe and was kept constant (60 μl/min) by means of the syringe pump flow rate through the microchannel. The yeast cells of 1 ml of each seven different sample concentrations from 102 to 109 cfu/ml were guided through the microchannel at a flow rate of 6 μl/min after flushing with PBS solution. The microneedles associated with the impedance analyzer were used to compute each solution’s impedance. Impedance spectra (impedance and phase versus frequency) were calculated using an AC signal frequency range from 100 Hz to 5 MHz with 1 Volt applied voltage to distinguish solution sample variations. DI and PBS water flushed the microchannel chip for 1 and 2 min, respectively, between each sample measurement. The data was recorded in MATLAB (MathWorks Inc., USA) with the impedance analyzer (Hioki IM3570) GUI and post-processed. The impedance change was measured at the sensing region during the passage of yeast cells. For each sample, impedance was measured at three frequencies (100 kHz, 500 kHz, and 1 MHz) to determine the difference in impedance. Single cell detection and measurement was performed with or without a single cell at the sensing area based on impedance measurement. For this analysis and detection, two specimens of microbeads with a diameter of 15 μm and 9 μm suspended in 1 mL of PBS with a concentration of 103 per mL were used. Every sample was propelled at a flow rate of 6 μl/min through the microchannel and measured using an AC frequency range of between 100 Hz and 5 MHz.

404

M. A. Mansor and M. R. Ahmad

Impedance Measurement inside Microfluidic As proof of concepts, the dependencies of impedance are tested using this microfluidic system on the different concentrations of yeast cells and a single microbead in the suspension medium. For two microchannels filled with sterilized DI water and PBS at the frequency range 1 kHz–1 MHz, the measured impedance spectra and fitting spectra (on a log scale) of the system were presented as shown in Fig. 3. For simulation, 100 data points were used as input to the corresponding circuit on the impedance measured spectrum (see Fig. 1b), and the matching impedance spectrum was developed using MATLAB. The result shows two domains for high conductivity fluid (PBS), i.e., a region of electrical double layer (EDL) and a resistive region (Morgan et al. 2007). The EDL occurred within the low frequency range from 1 kHz to around 300 kHz, whereas the high frequency resistive area occurred from 300 kHz to 1 MHz. The agreement between the product of the measured and suitable spectra suggested that the impedance characteristics of the solution medium can be calculated by our established circuit model for this device. The yeast cell and microbeads of specific concentration were used to demonstrate the device’s cell detection capability. With a fixed flow rate of 6 μl/min and a fixed electrode gap (25 μm), yeast cell concentrations ranging from 102 to 109 cfu/ml were infused within the microchannel. In order to calculate the impedance of concentration of yeast cells in DI water, a sweeping frequency (100 kHz–5 MHz) AC signal (1 Volt) was applied to the one side of the microneedle, and the current entering at another side of microneedle was measured. Initially, the injection of 109 cfu/ml resulted in a drop-in impedance by referring DI water impedance as a control. After that, the PBS washed the microchannel chip, followed by the maximum flow rate of DI water.

Fig. 3 Impedance spectra of sample solution together with their fitting spectra (a) DI water (b) PBS

16

Microfluidic Device with Removable Electrodes for Single Cell. . .

405

Fig. 4 (a) Impedance spectra of yeast cells in water with cell concentrations ranging from 102 to 109cfu/ml, along with DI water as controls. (b) The linear relationship between the logarithmic value of the concentration of yeast cells and the impedance measured at 1 MHz

The liquid’s maximum flow rate is 300 μl/min within the microchannel without leakage. The yeast cell’s impedance spectra in DI water with the different cell concentrations in the range 104–109 cfu/ml, together with DI water as a reference, is shown in Fig. 4a. 108 cfu/ml of cell concentration was pumped into the microchannel after washing the microchannel, leading to an increase in impedance. With decreasing cell concentration (Esfandyarpour et al. 2014), the impedance spectra of yeast cells in DI water across the sensing area (two microneedles) can be seen increasing. Based on the result of the experiment, it can be said that high-concentration cell suspension is more conductive than lower-concentration suspension.

406

M. A. Mansor and M. R. Ahmad

Impedance of Cell Population The solution’s conductivity varies in proportion to the number of concentrations of cells at fixed solution volume (Yang 2008). The relative dielectric permittivity and charged polyelectrolytes within the cell may also affect solution impedance in some studies (Esfandyarpour et al. 2014). The ideal range for microneedle sensing to distinguish the concentration of cells in DI water is between 500 kHz and 5 MHz. In this frequency region, the impedance values of the suspensions differed significantly from each other. The experiment was repeated two times the cycle of measurement, and a similar result was shown. The frequency less than 100 kHz is not regarded in the cell detection test as the EDL has started to affect the measurement at frequency less than 300 kHz (Emaminejad et al. 2012; Segerink et al. 2010). We selected 1 MHz as the best representative frequency to investigate the relationship between impedance value and cell concentration. Figure 4a displays the sample impedance response containing different concentrations of yeast cells and DI water at 1 MHz frequency measurement. The solution impedance increased significantly from 207.63 kΩ to 225.42 kΩ, 247.61 kΩ, 284.48 kΩ, 314.64 kΩ, and 348.51 kΩ when the yeast concentration dropped from 109 cfu/ml to 108 cfu/ml, 107 cfu/ml, 106 cfu/ml, 105 cfu/ml, and 104 cfu/ml, respectively. After the concentrations of cells were below 104 cfu/ml, the impedance value showed no significant change between each other or DI water. Moreover, the result trend shows a linear relationship between impedance and cell concentration logarithmic value at 104 cfu/ml to 109 cfu/ml (see Fig. 4b). The result’s linear regression equation is Z = 58.3 log X (cells/ml) + 175.4 with R2 = 0.986. It was calculated that the detection limit was 1.2  104 cfu/ml. Error bars are standard deviations with five measurement time. The impedance of the yeast suspensions can be used to estimate the cell concentration in DI water suspensions based on this linear regression equation. This device can be used to measure cells for bacteria detection in suspensions other than impedance microbiology and impedance biosensors, as this method’s detection limit is comparable to other sensors. The reported sensor for detection of pathogenic bacteria is QCM immunosensors for detection of Salmonella with detection limits of 9.9  105 cfu/ml (Park et al. 2000), surface plasmon resonance (SPR) sensor for detection of E. coli O157:H7 with a detection limit of 107 cfu/ml (Fratamico et al. 1998), and SPR immunosensors for detection of Salmonella enteritidis and Listeria monocytogenes with detection limits of 106 cfu/ml (Koubová et al. 2001).

Impedance of Single Particle Two sizes of microbead flowed into the microfluidic device to demonstrate the capability of this device to detect the presence of a single cell. Inside the microfluidic device, the impedance of PBS solution as control was initially infused. Like the yeast cell concertation measurement, two samples of microbead in PBS solution were then infused inside microchannel with same flow rate and electrode

16

Microfluidic Device with Removable Electrodes for Single Cell. . .

b 10

Impedance (ohm)

a

50 µm

407

5 PBS 15 um bead 9 um bead

104

103 0

0.5

1 1.5 2 Frequency (Hz)

2.5

3 x 106

Fig. 5 (a) The single microbead with diameter of 15 μm flow through the sensing area. (b) Impedance spectrum of two different size of beads in PBS solution and PBS solution (without bead)

gap. Figure 5a indicates a 15 μm microbead flow through the sensing region, and a sweeping frequency between 100 kHz and 3 MHz AC (1 Volt) was applied to the electrode. As a result, as shown in Fig. 5b, the impedance spectrum is plotted over the field frequency. The figure shows the average electrical impedance data for two sizes of microbead and PBS solutions without presence of beads. From this average data, the electrical impedance spectrum is expected to be used to distinguish between bead sizes. The spectrum of impedance clearly discriminates against the microbeads (9 and 15 μm in diameter). Impedance increases as the particle size increases. The electrical resistance of the sensing channel was slightly increased due to the presence of a single microbead that can be considered as an insulating element. As a result, at the high frequency range between 100 kHz and 5 MHz, we conclude this device can detect the cell concentrations in the solution medium and the single microbead. In this experiment, at the frequency below 100 kHz, the detection capability was not determine. For the future work, in order to improve the performance of the device, further investigations can be conducted, e.g. measurement and detection to the human cells (normal and cancer cells) at single cell level, effect of the size of the microneedle and utilization of a non-polarizable electrode such as Ag/AgCl (to eliminate the EDL).

Conclusions In summary, we demonstrated a very simple, label-free, and low-cost microfluidic system in the suspension medium for cell concentration and single cell detection. This system has a reusable microneedle that can be inserted into a PDMS microchannel that can be removed. With the increase in cell density in the solution

408

M. A. Mansor and M. R. Ahmad

medium, we find the impedance value decreases. This device’s ability to distinguish cell concentration from 109 cfu/ml to 104 cfu/ml indicates the proposed sensor’s core functionality despite significantly lower manufacturing costs. Furthermore, the microfluidic system is capable of detecting a single cell and decimating the size of a single cell. In this research, yeast cell and microbeads were used as proof of concept, and we emphasize this sensing technique can be applied in a range from 5 μm to 25 μm to a variety of cell types with diameter width. To avoid the potential spread of contamination to samples, it is recommended to perform only one measurement period for each PDMS microchip. The system is ideal for early detection of cancer cells and the application of water contamination in developing countries as it significantly reduces manufacturing costs (about 30% of manufacturing costs are reduced based on leasing of facilities and use of raw materials).

References Ciceron L, Jaureguiberry G, Gay F, Danis M (1999) Development of a Plasmodium PCR for monitoring efficacy of antimalarial treatment. J Clin Microbiol 37:35 Coulter WH (1956) High speed automatic blood cell counter and cell size analyzer. Proc Natl Electron Conf 12:1034 Du E, Ha S, Diez-Silva M, Dao M, Suresh S, Chandrakasan AP (2013) Electric impedance microflow cytometry for characterization of cell disease states. Lab Chip 13:3903 Emaminejad S, Javanmard M, Dutton RW, Davis RW (2012) Microfluidic diagnostic tool for the developing world: contactless impedance flow cytometry. Lab Chip 12:4499 Emaminejad S, Paik K, Tabard-Cossa V, Javanmard M (2016) Portable cytometry using microscale electronic sensing. Sensors Actuators B Chem 224:275 Esfandyarpour R, Javanmard M, Koochak Z, Harris JS, Davis RW (2014) Nanoelectronic impedance detection of target cells. Biotechnol Bioeng 111:1161 Fratamico PM, Strobaugh TP, Medina MB, Gehring AG (1998) Detection of Escherichia coli 0157: H7 using a surface plasmon resonance biosensor. Biotechnol Tech 12:571 Gawad S, Schild L, Renaud PH (2001) Micromachined impedance spectroscopy flow cytometer for cell analysis and particle sizing. Lab Chip 1:76 Gilchrist KH, Barker VN, Fletcher LE, DeBusschere BD, Ghanouni P, Giovangrandi L, Kovacs GTA (2001) General purpose, field-portable cell-based biosensor platform. Biosens Bioelectron 16:557 Gou H-L, Zhang X-B, Bao N, Xu J-J, Xia X-H, Chen H-Y (2011) Label-free electrical discrimination of cells at normal, apoptotic and necrotic status with a microfluidic device. J Chromatogr A 1218:5725 Holmes D, Morgan H (2010) Single cell impedance cytometry for identification and counting of CD4 T-cells in human blood using impedance labels. Anal Chem 82:1455 Jao J-Y, Liu C-F, Chen M-K, Chuang Y-C, Jang L-S (2011) Electrical characterization of single cell in microfluidic device. Microelectron Reliab 51:781 Kantara C, O’Connell MR, Luthra G, Gajjar A, Sarkar S, Ullrich RL, Singh P (2015) Methods for detecting circulating cancer stem cells (CCSCs) as a novel approach for diagnosis of colon cancer relapse/metastasis. Lab Investig 95:100 Koubová V, Brynda E, Karasová L, Škvor J, Homola J, Dostálek J, Tobiška P, Rošický J (2001) Detection of foodborne pathogens using surface plasmon resonance biosensors. Sensors Actuators B Chem 74:100 Liu Y-S, Banada PP, Bhattacharya S, Bhunia AK, Bashir R (2008) Electrical characterization of DNA molecules in solution using impedance measurements. Appl Phys Lett 92:143902

16

Microfluidic Device with Removable Electrodes for Single Cell. . .

409

Mansor MA, Ahmad MR (2015) Single cell electrical characterization techniques. Int J Mol Sci 16:12686 Mansor MA, Ahmad MR (2016) A simulation study of single cell inside an integrated dual nanoneedle-microfluidic system. J Teknol 78:59 Mansor MA, Takeuchi M, Nakajima M, Hasegawa Y, Ahmad M (2017) Electrical impedance spectroscopy for detection of cells in suspensions using microfluidic device with integrated microneedles. Appl Sci 7:170 Morgan H, Sun T, Holmes D, Gawad S, Green NG (2007) Single cell dielectric spectroscopy. J Phys D Appl Phys 40:61 Park IS, Kim WY, Kim N (2000) Operational characteristics of an antibody-immobilized QCM system detecting Salmonella spp. Biosens Bioelectron 15:167 Park K, Suk H-J, Akin D, Bashir R (2009) Dielectrophoresis-based cell manipulation using electrodes on a reusable printed circuit board. Lab Chip 9:2224 Segerink LI, Sprenkels AJ, ter Braak PM, Vermes I, van den Berg A (2010) On-chip determination of spermatozoa concentration using electrical impedance measurements. Lab Chip 10:1018 Spencer D, Hollis V, Morgan H (2014) Microfluidic impedance cytometry of tumour cells in blood. Biomicrofluidics 8:064124 Sun T, Morgan H (2010) Single-cell microfluidic impedance cytometry: a review. Microfluid Nanofluid 8:423 Sun T, Holmes D, Gawad S, Green NG, Morgan H (2007) High speed multi-frequency impedance analysis of single particles in a microfluidic cytometer using maximum length sequences. Lab Chip 7:1034 Wang J, Chatrathi MP, Mulchandani A, Chen W (2001) Capillary electrophoresis microchips for separation and detection of organophosphate nerve agents. Anal Chem 73:1804 Yang L (2008) Electrical impedance spectroscopy for detection of bacterial cells in suspensions using interdigitated microelectrode. Talanta 74:1621 Yang L, Arias LR, Lane TS, Yancey MD, Mamouni J (2011) Real-time electrical impedance-based measurement to distinguish oral cancer cells and non-cancer oral epithelial cells. Anal Bioanal Chem 399:1823

Microfluidic and Nanomaterial Approach for Virology

17

Reya Ganguly and Chang-Soo Lee

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General Understanding of Host–Viral Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . How Do Hosts Respond to Intruders and What Is Viral Feedback? . . . . . . . . . . . . . . . . . . . . . . . Microfluidic Technique in the Field of Virology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flow-Based Channel Microfluidics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Droplet-Based Microfluidics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electric Field-Based Digital Microfluidics (DMF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidic and Nanoparticle-Based Diagnostic Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paper-Based Device . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gold Nanoparticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidics for Host–Viral Interaction Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

412 413 414 415 415 418 419 420 420 422 426 427 427 429

Abstract

Viral infectious diseases can erupt rapidly and are capable of causing massive health issues. Recent evidence on their high specificity over host species and cell type have fueled research into decoding host–viral interaction. The conventional approach relies on cell culturing-plate and multiwell plate-based analysis, which not only utilizes a large amount of reagents but are also population biased and gives ensemble measurement of biological assay. The animal model does not perfectly recapitulate human disease nor do they provide a point of care analysis. Studying the direct interaction between the virus and the host cell remain challenging. The researcher, therefore, has addressed these challenges using a microfluidic device, which promises high-throughput analysis, analysis at the singlecell level, and small consumption of reagents. These features intensify the R. Ganguly · C.-S. Lee (*) Chungnam National University, Daejeon, Republic of Korea e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_46

411

412

R. Ganguly and C.-S. Lee

understanding of various biological interactions in terms of single-cell heterogeneity in a rapid, accurate, and cost-effective manner. While conventional methods will continue to serve biology and virology well, we envisage that microfluidic platforms for viral infections at a single-cell level will be an impressive method to develop an antiviral drug, discover essential proteins for infection, and to understand viral mutants. This chapter emphasizes the various micro and nanodevice developed for the detection and capture of viruses. In addition, it also deals with an understanding of host–viral interactions and comments on how this understanding might give profound resolution to biological research as well as to the biomedical approach.

Introduction Virology, for the past few years, is overly focused on the pathogenic particle. Now it is well known that to establish a successful infection, a viral particle has to undergo battle with the host. This defines the eventual outcome of the infection. Coordinated action by a set of proteins in a network or pathway governs the cellular functionality. An in-depth understanding of this coordination up to the molecular level may broaden the therapeutic approach to combat viral infection. Various reports on vaccine failure and inadequate production of antiviral drugs suggest the requirement of a more innovative approach towards virology. Very often studies focus on single infectious strain (DaPalma et al. 2010); however, more than single viruses might be present over a host surface, providing an opportunity for viruses to interact (Roux et al. 2015; Flores et al. 2011; Munson-McGee et al. 2018; Diaz-Munoz 2017). This interaction positively or negatively affects infection by the second virus. This type of interaction might occur in homologous or heterologous viruses (Diaz-Munoz 2017). In addition, various evidence shows that increased pathogenicity, host jump, and genetic exchange is the result of coinfection. In contrast to the prospective future of viral–host interaction and virus–virus interactions studies, researchers are still in the initial phase to explore the prevalence, mechanisms, and significance of these interactions. Host–viral interaction and virus–virus interactions constitute a critical area of virology in which two or more viruses are involved. Moreover, these interactions are prone to the ecological factors, molecular system, and population within and between hosts. Variants of Influenza A virus which showed cooperation when infecting cell culture did not demonstrate a similar effect in a clinical human sample (Xue et al. 2018). Likewise, HIV exhibit complex interaction among the diverse viral population (Gerhardt et al. 2005; Li et al. 2010). High-throughput tools are urgently essential to investigate such complex social phenotypes with diverse strains. However, sequencing approaches might not be sufficiently appropriate. Some conventional approaches such as yeast-two hybrid (Y2H) system, protein fragment complementation assays (PCAs), and affinity purification followed by mass spectrometry-based protein identification (AP-MS) are proteomic approaches to realize host–viral interaction (Ben-Ari et al. 2013). Y2H is a genetic method that

17

Microfluidic and Nanomaterial Approach for Virology

413

determines protein–protein interaction, while AP-MS is a biochemical process of analysis. However, proteomics remains challenging due to difficulties in the expression and purification of these proteins in large quantities, especially membranebound proteins. Many viral proteins are membrane-associated which also tend to perform multitask, such us HIV proteome. Thus, lack in sensitivity of the platform for studying host–viral interaction will strongly affect understanding in this field of research. Various pathogenic viruses such as the Rift valley fever virus (RVFV), known to cause hemorrhagic fever, pose threats to the public health for lack of treatment or vaccines. To investigate the cellular factors involved in host–pathogen interaction, RNA interference is a well-known genetic approach. The basic mechanism in this approach involves chemically synthesized small interfering RNA (siRNA). These siRNAs are targeted for a specific gene. The introduction of these siRNAs within a cell leads to their binding with specific mRNA, thereby reducing protein expression. As viruses exploit cellular systems to initiate infection, thus genome-wide RNAi represents a robust approach for comprehensive analysis of host–pathogen interaction. However, for the study of evolving viruses that require high-level biocontainment amenity is limited. RNAi libraries and reagents are expensive, and thus to fully utilize the advantage of RNAi screening, alternative platforms need to be developed. Currently, the application of microfluidic platforms have appeared as robust method for biological research. Incorporating microarray technology in microfluidic platform empowers programming of up to several thousand experiments. In addition it overcomes various limitations such as protein denaturation, protein purification, limited capture agents, low detection quality, and special storing conditions (Talapatra et al. 2002; Mitchell 2002). Besides high-throughput screens, the microfluidic platform has key features that make it advantageous for virus research, that is, high sensitivity, quantitative data, and compatibility with membrane proteins (Ben-Ari et al. 2013). The platforms high sensitivity facilitates detection and measurement in the range of μM of low-affinity transient interactions. Furthermore, it is reported to produce quantitative data for difficult membrane-associated protein, either viral or host. Lastly, it is widely used for exploring the compound library to recognize potential antiviral inhibitors for therapeutic development. In this chapter, we focus on various microfluidic approaches that deal with virus capture. Besides, try understanding of host–viral or viral–viral interactions. In addition, we will briefly emphasize new findings that are acquired from these approaches and critically look at novel ideas that minimize defects of the conventional technique. Furthermore, many detection devices have come up with point-ofcare diagnosis of an active patient.

General Understanding of Host–Viral Interaction Viruses exploit intracellular machinery of a host cell after establishing infection. They utilize it to replicate its genome, and expressing various viral proteins. Consequently, they interact intensely with the host throughout its life cycle.

414

R. Ganguly and C.-S. Lee

Physical factors such as Brownian motion, air delivery, or biological means are the transmission mode of virus through which they encounter its host. Encountering host do not ensure infection but largely defined by successful interaction between virus and host receptor at the cell surface. In most cases, binding of virus to receptor causes conformational changes that help virus to deliver its genetic material by penetrating in the host cell. Plant viruses, however, depend on mechanical breaches to directly introduce virions into the cells; by this way, they overcome the plant cell wall barrier. Few viruses namely mycoviruses enter from intracellular routes and do not involve any receptor interaction between virus and host cell. Additionally, they can intrude during cytoplasmic exchange, cell division, mating or anastomosis or by spores. Viruses that do not deliver genome directly need to uncoat the capsid, which is a multistep process. This built-in property to disassemble coat protein is referred to as structural metastability, i.e., the coat protein of virus particles did not attend the minimum free energy confirmation. Thus, they are can assemble in infected virus producing cells and disassemble when approach for new infection.

How Do Hosts Respond to Intruders and What Is Viral Feedback? The host interacts with and responds to viruses that it encounters through various mechanisms such as an immune mechanism, MicroRNA-induced gene silencing, etc. Host responses elicit to eliminate or halt the growth of viruses. During infection, initial defense steps are taken by innate immunity and trigger pro-inflammatory responses. However, during the later stage of viral pathogenesis, adaptive immunity plays a critical role. In addition to immune response, MicroRNA is implicated in the complex cross-talk between the host and the pathogen. Generally, to reflect infection, host cells display small viral peptides in class I major histocompatibility complex (MHC class I) protein located on the cells surface. Cytotoxic T cell (a type of T cell) are specialized to detect these presented viral proteins through their T cell receptor (TCRs). On the detection of viral antigen, T cells release cytotoxic factors that trigger apoptosis in the infected cell to prevent viral replication and thus survival. Some viruses resist MHC class I to display viral peptide on the cell surface thus prevents T cell-mediated infected cell apoptosis. However, this leads to a reduced number of MHC class I on the cell surface; as a result, another host immune cell called natural killer (NK) cells identify cells displaying fewer MHC class I than normal. When NK cells come across such cells, they release toxic substances to kill the infected cell. T cells, NK cells, and other cytotoxic cells are equipped with cytotoxic factors stored in granules. Perforin, a protein that makes pores in the cell membrane, is one of the mediators of this process. These pores help the entry of granzymes that cause apoptosis and granulysis that cause cell membrane lysis. Cytokines are another type of protein synthesized newly on exposure to the infected cell. Cytokines include interferon-g and tumor necrosis factor-α that enhance the killing mechanism by signaling the infected cell and its neighboring cells.

17

Microfluidic and Nanomaterial Approach for Virology

415

High dependence of viral infection on the cellular system makes them susceptible to various other countermeasures taken by the host cell including MicroRNAs (miRNAs). Lecellier et al. (2005) demonstrated the involvement of mammalian microRNA (mir-32) in restricting the accumulation of primate foamy virus type 1 (PFV-1) in human cells. Later, mir-507 and mir-136 were reported to have a potential binding site in Polymerase B2 (PB2) and Hemagglutinin (HA) gene, respectively. These two genes PB2 (one component of ribonucleoprotein) and HA (a surface glycoprotein) are reported to be associated with virulence. HA involved in direct binding of virus and ribonucleoproteins are responsible for replication and transcription of RNA. In spite of several mechanism to stop invading pathogens, viruses still has evolved strategies to evade host defense and establish infection. Cui et al. (2006) reported novel microRNA encoded by Herpes simplex (HSV) virus. Another report from Gupta et al. showed HSV-1 latency-associated transcript (LAT) codes for microRNA that target apoptosis-associated genes and other signaling pathways (such as TGF-β, TGF-β1, and SMAD3), thus preventing apoptosis of the infected cell. Mostly large DNA viruses have several defense strategies because of their high coding capacity. HSV and poxviruses encode cytokine receptor(s) that binds cytokines and prevent from producing signaling transduction.

Microfluidic Technique in the Field of Virology Flow-Based Channel Microfluidics High-throughput and automated analysis can be achieved through conjugation of various function in microfluidic chip. Functions such as mechanical pumps which ease the control of fluid motion in a network of microchannels, within which fluids are transported and experimental interfaces are inferred. The most effective aspect of microfluidics is, it can be modified with devices that synchronize physical conditions within the platform such as temperature and pH level. Fluids within a device typically move in continuous, laminar flow. However, the curving microchannel facilitates the separation of bacteria and viruses from large particles by centrifugal force (Hong et al. 2015), electrodes. Microchip fabrication for microstructure modeling mostly uses polydimethylsiloxane (PDMS) for its compatibility with soft lithography techniques. PDMS is known to be gas permeable and transparent, thus proving to be ideal for cell culture and live imaging (Tseng et al. 2016). Irrespective of the existing technologies, virus evolution and outbreaks often challenge the effectiveness of diagnostic tools. The primary challenge towards an understanding of newly evolved strain is its effective isolation for characterization. To achieve this, many sensitive detection kits (Toh et al. 2015; Griffiths et al. 2017; Spackman et al. 2002) with proficiency to detect minimal quantity fails, because they require prior knowledge of the strain. Thus, a strong candidate for virus surveillance is a next-generation sequencing (NSG). However, in the current situation, this approach also suffers from limitations of low viral titer in clinical samples,

416

R. Ganguly and C.-S. Lee

consequently generating host genetic material dominated sequence read. Moreover, implementing such strategies for rapid detection at remote area suffer technical issues making unprepared for the epidemic. Yin-Ting Yeh et al. reported a way to enrich and optically detect emerging viruses in label-free method. In addition, viable capture empowers further characterization via detection methods. This featured device was named as VIRRION (virus capture with rapid Raman spectroscopy detection and identification) capable of performing real-time nondestructive identification using surface-enhanced Raman spectroscopy integrated with machine learning algorithm and database. Conspicuously, this device allows inspection of captured viruses through the conventional method as a follow-up analysis. VIRRION platform consists of arrays of aligned nitrogen-doped carbon nanotubes (CNxCNTs) coated with gold (Au) nanoparticles to intensify the signal-tonoise ratio in Raman spectroscopy. The fabrication technique of this nanotubes in an array is based on the stamping technique of Fe catalytic particles that initiate reaction at a specific region during chemical vapor deposition (CVD). Stamping technique allow manipulation of intertubular distance (from 22
5 to 720
64 nm) (ITDs) through changing Fe concentration allowing the device to capture a large range of different sized viruses. It expands its utility for capturing viral strains with no prior information. To leverage its efficiency, the herringbone pattern was incorporated to boost the mixing of the sample with gold nanoparticle-coated carbon nanotube (Au/CNxCNT). A reusable patterned mold was fabricated using a 3D printer that imprints liquid-based Fe precursor over the substrate at submillimeter resolution (Fig. 1a). Different concentrations of Fe precursors (iron156(III) nitrate nonahydrate, Fe(NO3)3•9H2O) were spin-coated on micromold with herringbone patterns. This precursor-coated pattern is used for depositing precursors over the Si/SiO2 substrate. This patterned area functions as a nucleation site growth of CNxCNTs under CVD in the presence of benzylamine. As a result of this approach, 3D herringbone array with different ITDs was formed (Fig. 1b). Different concentration leads to varying density and diameter of Fe particle which justify the formation of CNxCNTs with different diameter and ITDs. This report shows a linear relation of Fe precursor concentration with Fe particle density and diameter along with that of CNxCNTs (Fig. 1c,d). For optical detection of captured viruses, this device was functionalized with Au nanoparticles (~ 15 nm in diameter) to enhance surface-enhanced Raman spectroscopy (SERS) by increasing localized plasmon resonance (Fig. 1e). The detection methods could be transmission electron microscope (TEM), scanning electron microscope (SEM), or immunofluorescence assay (Fig. 1f). An algorithm of the spectral database was created with three viruses that are two strain of AIV (H5N2 and H7N2) and reovirus. The enhancement in signaling from viruses due to nanoparticles happens only at the vicinity of “hot spot” which is a 1 nm region between two adjacent Au nanoparticles. These signals were recorded and an average of 100 spectra from specific viral strain was used to construct the algorithm. This generates a reliable averaged fingerprint for scanning similar viral strain from a clinical sample with a lower limit of ~102 EID50/mL (Fig. 1g). The proliferation of captured viruses was facilitated by cell culturing directly within the VIRRIONs platform increasing the viral titer up to 106 EID50/mL. Enrichment

Microfluidic and Nanomaterial Approach for Virology

Fig. 1 Growth of aligned CNxCNT with tunable dimension using stamping technique. (a) Patterning CNxCNT arrays, process flow of the stamping technique. (b) SEM image of before and after CVD synthesis of iron-rich particles (top row) and CNxCNT (bottom row) [Scale bar: 100 nm (top row); 100 nm (bottom row)]. (c) Under different precursor concentration: density and diameter of iron-rich particles and CNxCNT. (d) Under different precursor concentrations: tunable ITD of aligned CNxCNT array growing. (e) Aligned CNTs exhibiting herringbone patterns decorated with gold nanoparticles: photograph and SEM images. (f) VIRRION for avian influenza virus surveillance and discovery: a process flow. (g) Raman spectra collected from VIRRION of H5N2, H7N2, and reovirus. (H) Before and after VIRRION enrichment, the ratio of copy number of H5N2 and 18S rRNA is represented (Yeh et al. 2020)

17 417

418

R. Ganguly and C.-S. Lee

showed an increase in the viral genome copy number with respect to host 18S ribosomal RNA (an essential housekeeping gene) (Fig. 1h). In addition, it increased viral-specific reads and genome coverage in next-generation sequencing (NSG)

Droplet-Based Microfluidics The generation of droplets in microfluidic channels is a comparatively simple process. Active and passive methods are involved in droplet generation in the microfluidic device. Active platforms have programmable control on droplet size, content, and droplet motion. However, greater ability for the on-demand generation of droplets reduces the throughput due to the involvement of valves that require timed and controlled actuation. This platform demand preprogrammed instrumentation for controlling moving parts to enable modification of generating droplets. Pneumatic valves are well-known means for mechanically control droplet production (Wu et al. 2014; Unger et al. 2000). The microvalves made from PDMS (polydimethylsiloxane) and placed above microchannel deform when pressurized to constrict the fluid flow. By selectively operating microvalves, droplets can be generated (Guo et al. 2010). In contrast, the passive platform generates highthroughput droplet by non-movable structures that disturb interfacial tension between the oil and aqueous phase at the expense of droplet controllability. Besides the generation of the droplets, it is necessary to ensure uniform mixing of fluids within droplets which is normally rapid and chaotic advection (channel geometrybased) can enhance mixing further (Chen et al. 2005). Moreover, for monitoring of biological assays over an extended period, droplet storage can be facilitated by accumulating droplets in chambers (Schmitz et al. 2009), capillaries, or tubing (Niu et al. 2009). Enhanced design aid merging of droplets for flexibility, and stand operations such as reaction initiation, reagent dosage, droplet dilution, and reaction termination (Niu et al. 2008; Solvas and DeMello 2011). Counterwise, droplets can also be split by active or passive strategies (Link et al. 2004). A relatively more critical downstream operation is droplet sorting, which is being achieved by piezoelectric actuation (Shemesh et al. 2010) or dielectrophoretic activation (Niu et al. 2007). Wide variety of area exploits droplet microfluidics, mainly PCR amplification since compartmentalization of reaction significantly enhance amplification efficiency over conventional technique as well as flowbased channel microfluidics. This is achieved by eliminating reagent dispersion and adsorption on the channel surface. Many researchers utilize this platform for biochemical assays involving DNA, protein, and enzymes to benefit from its many advantages. A similar platform named cross-interface emulsification (XiE) was developed and utilized for the quantification of H5-subtype avian influenza viruses. This simple method generates monodisperse droplets at a fixed vibrational frequency of the capillary tube with a continuous flow of disperse phase connected with a continuous phase at its interphase (Xu et al. 2016). For the fabrication of a capillary tube, fused capillaries with flat tips (40 μm inner diameter (i.d.), 127 μm outer diameter (o.d.)) were cut into 3 cm pieces. The setup of XiE system fixes the

17

Microfluidic and Nanomaterial Approach for Virology

419

capillary tube with the electromagnetic vibrator by capillary wax, which in turn connected with a waveform generator for controlling amplitude and frequency. The capillary was connected with a syringe pump via teflon tubing (20 cm long, 380 μm i. d.) and syringe. All the connection was sealed with capillary wax. The aqueous phase was loaded to the syringe and infused via a syringe pump through the capillary. The tip of the capillary was placed 1 mm above the air/oil interface of 300 μl oil-filled microwells, the composition of which was mineral oil and n-tetradecane at a ratio of 1:1 (v/v), surfactants with 3% ABIL EM 90. The vibrational frequency and peak amplitude (~2 mm) was kept constant by a waveform generator. The flow rate of infusion was made consistent by defining a target volume in a syringe pump. To define the principle behind droplet detachment at the air/oil interface, it was shown that the effect of surface tension dominated over other dimensionless parameters like capillary number (Ca), Reynolds number (Re), and Bond number (Bo). Other investigated forces such as inertial effect (Fk), gravity (G), and the buoyant force (Fb) were negligible. On lifting up of capillary tip out of the carrier oil during vibration caused “blockage” due to confinement of the air/oil interface (Fi) that played a key role against water/oil interfacial force (Fσ) which eventually caused droplet detachment. Also, the residual oil layer at the outlet of the tip offers a contamination-free flow of aqueous solution (Fig. 2a). This method generates one droplet per vibration cycle. Thus, the volume of each droplet (v) was defined by v ¼ Q/f, where Q is the flow rate (nL/s) and f is the frequency of vibration (s-1). This platform was then employed for digital loop-mediated isothermal amplification (dLAMP) assay, where H5N1 samples were detected. Furthermore, it showed high specificity and reduced effect of inhibitors over the efficiency of the device (Fig. 2b,d). Besides, it provides a simplified integrated strategy for droplet generation, reaction to detection in a single stretch (Fig. 2c). Overall, simplified sample preparation and rapid quantification for a limited amount of sample has been achieved through this method.

Electric Field-Based Digital Microfluidics (DMF) Many groups have developed strategies to utilize the electric potential as a tool to precisely operate droplets in their platform. Paper-based and microchannel-based devices are limited for their use on one patient sample at a time; however, DMF can handle multiple parallel samples. DMF is arrayed with electrode covered with a hydrophobic surface that allow droplets manipulation precisely at picoliter or microliter range (Au et al. 2013). On application of potential to these devices, the droplets can be moved, merged, split, or dispensed from the reservoir along the different directions on-chip surface (Choi et al. 2012). This technique eliminates the use of valve, pumps, or mechanical mixtures. Depending on the actuation program, they can reconfigure the droplet path for different applications making it a versatile platform to be integrated with a mass spectrometer, NMR, optical techniques, which are known for chemical and biological analysis (Lei et al. 2015; Swyer et al. 2016; Martin et al. 2009; Sista et al. 2011).

420

R. Ganguly and C.-S. Lee

Fig. 2 Droplet generation by XiE: principle and phenomenon. (a) At the air/oil interface, force analysis of the droplet detachment. (b) After the dLAMP reaction, fluorescence image of droplet array. The label shows viral cDNA of different subtypes added in each assay. Control has no template. The graph represents the specificity of dLAMP assay to detect H5N1 viruses. (c) Overall process of dLAMP analysis through this technique, and droplet arrays generation in microwells, at 60–65 °C, and fluorescence imaging. (d) Effect of inhibitors over quantification of viral genome by dLAMP, dPCR, and qPCR. Test of humic acid (HA, 50 ng) and SDS (0.05%) as representative inhibitors and use of templates of two concentrations: 3000 and 100 copies/μL. Fluorescence images of dLAMP with and without inhibitors. Scale bar ¼ 500 μm. The graph is the quantification of templates by dLAMP, dPCR, and qPCR with and without inhibitors (Hu et al. 2017)

Microfluidic and Nanoparticle-Based Diagnostic Devices Paper-Based Device Paper-based microfluidics has received much attention due to its superiority in ease of use and no necessity of well-equipped laboratory and well-trained manpower. Besides, it has a low sample volume requirement with high sensitivity and is also efficiently portable. This feature makes paper platforms a promising alternative for

17

Microfluidic and Nanomaterial Approach for Virology

421

clinical diagnosis and quantitative analysis for biochemical substances. Among the wide range of paper-based analytical devices (PADs), the wax-printed device shows high-throughput production and needs a commercial printer and a heating source to fabricate complex structures for multipurpose systems (Jeong et al. 2015). PADs is user friendly for in situ diagnosis and analysis; hence, wax-printed PADs are promising when people are limited with resources for fast diagnosis of their serious health problems. World Health Organization (WHO), however, has defined criteria for point-of-diagnostic (POC) devices, i.e., POC device should follow “ASSURED.” “ASSURED” stands for affordable, sensitive, specific, user-friendly, rapid and robust, equipment-free, and deliverable. To satisfy the required criteria, researches should come up with an ideal platform which in due course of time was taken up by paper-based devices. Paper proved to be inexpensive materials for the fabrication of PADs, and it facilitates easy manipulation of reagents and analytes. Many methods such as conventional photolithography processes, inkjet printing, wax printing, or screen printing can easily make patterns over the paper for mass production. Thus, PADs have high demand in many fields namely food production, disease diagnosis, and environmental monitoring, etc. for its proven sensitivity and specificity. Regardless of the excellent features of the papers that overlap the required criteria for POC, the paper-based diagnostic device should also efficiently immobilize biomolecules over the solid surface. Effective immobilization is a critical step in the fabrication of biosensor for the detection and quantification of target molecules. Most paper-based devices relay on physical adsorption for immobilizing biomolecules on paper. However, these physical interactions include week forces such as electrostatic, van der Waals, and hydrophobic interactions that do not assure reproducible results. Many groups consider approximately 40% loss of adsorbed antibody during washing when attachment depends exclusively on physical adsorption. Woogyeong Hong et al. reported a method to induce covalent binding between proteins and cellulose paper for a paper-based sandwich enzyme-linked immunosorbent assay (ELISA) to detect Middle East respiratory syndrome coronavirus (MERS-CoV) (Hong et al. 2018) Adobe Illustrator CS5 software provides a feature to design patterns which are then fabricated by the process of wax patterning. Whatman No. 1 chromatography paper was used for printing followed by laminating at 160 °C for 8 s and cooling to room temperature naturally. The rate of lamination is controllable by the laminator through controlling the rotation rate of the rollers. Subsequently, the melted wax at a specific patterned area makes a hydrophobic wax barrier, thus completing the fabrication process within 5 min. To immobilize the proteins, the paper was treated for periodate oxidation leading to the formation of the Schiff base followed by reductive amination. To achieve the formation of the Schiff base, the paper was first treated with 10 μl sodium periodate (NaIO4) (0.5 mol/L) at room temperature in dark for 30 min. This produce aldehyde group; after washing, the paper was treated with bovine serum albumin (BSA) (1–10 mg/mL in PBS) for 30 min at room temperature. This process makes reversible Schiff base on the surface of the paper, the excess unbound proteins are washed and finally treated with 10 μL NaCNBH3 (1.6 mg/

422

R. Ganguly and C.-S. Lee

mL) for reductive amination to obtain stable covalent bonds (secondary amines) (Fig. 3a). The confirmation of covalent binding was obtained from Fourier transform infrared spectroscopy (FT-IR) spectra of the test area (Fig. 3b). In addition, a comparison with the physical adsorption method was made by utilizing conjugate of FITC-BSA during surface modification followed by washing. Fluorescent imaging confirmed strong bonding (Fig. 3c,d). Through FT-IR spectra it was shown that, with increasing reaction time, the oxidation of paper increases, giving absorption peak at 1729 cm1 which corresponds to carbonyl bond (C¼O) of aldehyde group. Despite this, excess reaction time reduces the cellulose stiffness by breaking the glucose ring and also changes the fiber dimension leading to the formation of inter and intrafibrillar hemiacetal crosslinks. Thus, the periodate oxidation time was fixed to 30 min that give an optimum aldehyde group without destructive changes in the paper. This platform was verified and applied to detect MERS-CoV through a paperbased sandwich ELISA. MERS-CoV capture antibody (D5) was bound to surfacemodified chromatographic paper (with aldehyde group) as the primary antibody. Nonspecific binding of capture antibody was blocked with a blocking buffer followed by loading MERS-CoV antigen (1.22 fM – 7.6 μM) at the test area. MERS-CoV antigen being specific to the capture antibody, they make antigenantibody complex. To detect the formation of this complex, a secondary antibody (HRP-labeled anti-6X his-tagged antibody) was loaded. The existence of histidine tag in the MERS-CoV antigen allows it to bind to the secondary antibody (capable of histidine recognition and carries horseradish peroxidase (HRP) enzyme) giving a sandwich structure. The addition of enzyme-specific substrate (3,30 ,5,50 -tetramethylbenzidine (TMB)) reacts with the enzyme to form a blue complex, thus changing the test area color to blue from transparent. Furthermore, this platform was applied to quantitatively analyze the different concentration of antigen (Fig. 4). The detection limit was reported to be 10.4 nM which clearly indicated retention of functionality (antigen recognition activity) of antibodies even after covalent bond formation. Hence, this research has opened the opportunity to exploit paper for attaching amine-containing biomolecules without compromising reproducibility and sensitivity of PADs.

Gold Nanoparticles Nanomaterials possess numerous exclusive electronic, optical, magnetic, and mechanical properties that permit attractive application in the field of clinical diagnosis and biomedical imaging (Wang et al. 2009; Rosi and Mirkin 2005). In the late 1990s, a pioneer report showed the detection of human papillomavirus by AuNPs coupled with silver straining (Zehbe et al. 1997). In the current day, a wide range of nanomaterials is being investigated for virus detection. Inspected NPs include carbon nanotubes, silica NPs, metal NPs, quantum dots(QDs), polymeric NPs, and upconversion NPs (Wang et al. 2009; Rosi and Mirkin 2005; Alivisatos 2004). Prevalent strategy for exploiting nanostructures for viral detection is the

Microfluidic and Nanomaterial Approach for Virology

Fig. 3 (a) Covalent immobilization of protein on paper: schematics. (b) FT-IR spectra of (a) bare cellulose, (b) periodate-oxidized cellulose, (c) BSA-immobilized cellulose, and (d) NaCNBH3-treated cellulose. (c) Fluorescence images results from 1 mg/mL to 10 mg/mL of FITC-BSA. (a) Periodate oxidized paper; (b) without periodate oxidized paper, physical adsorption. (d) Enhancement of fluorescent intensity due to covalent bonding, where n ¼ 6 for each point (Hong et al. 2018)

17 423

424

R. Ganguly and C.-S. Lee

Fig. 4 Paper-based sandwich ELISA to detect MERS antigen. (a) MERS antigen detection by paper-based sandwich ELISA: schematic representation. (b) Sandwich ELISA results from 1.22 fM to 7.6 μM of MERS antigen. (c) Calibration plot of mean intensity of color (produced by the enzymatic reaction in the sandwich ELISA assay) and MERS antigen concentration. Each datapoint is the mean of 8 replica (n ¼ 5), and the error bars show the standard deviations (Hong et al. 2018)

development of nanobio hybrid system that conjugate surface of different NPs with virus-derived biomolecules (e.g., DNA, RNA, antibody peptide, pentabody, or antigen). This technology leverage the significant signal transduction function of NPs besides its labeling property (Dougan et al. 2007). Successful conjugates are being used to build up various optical, fluorometric, electrochemical, and electric assays for single or multiple diagnoses. These devices show an advantage in terms of size, performance, signal sensitivity, stability, and specificity. In addition, multiple applications such as vaccine development, antioxidant, and multifunctional drugdelivery vehicles are reported worldwide. Apart from its various utility, this

17

Microfluidic and Nanomaterial Approach for Virology

425

technology requires a highly complex level of surface modification to accomplish efficiency and limit off-target toxicity. AuNPs (gold nanoparticles) are preferentially chosen for nanomaterial-based virus detection, for instance, resonance light scattering (Neng et al. 2013; Cao et al. 2002; Wang et al. 2012, 2010; Driskell et al. 2011) by different light spectroscopy technique including localized surface plasmon resonance (LSPR), color amplification for colorimetric detection (Li and Rothberg 2004; Glynou et al. 2003; Shawky et al. 2010; Mancuso et al. 2013; Li et al. 2012; Laderman et al. 2008), and fluorescent quenching or enhancing (Wu et al. 2014; Griffin et al. 2009; Lu et al. 2013; Zeng et al. 2012; Chang et al. 2010; Ganbold et al. 2012; Draz et al. 2012). In the year 2018, David Sebba et al. reported a nanoparticle-based multiplexed detection platform for differentiating Ebola from endemic febrile diseases (Sebba et al. 2018). Accurate diagnosis is a critical requirement of effective treatment; however, various diseases show similar early symptoms. Thus, early detection exclusively based on symptoms prevent acceptable interpretation. The conventional approach deploys RT-PCR for efficient detection; despite its high efficiency, it is limited by the requirement of infrastructure and manpower. This platform detects protein via surface-enhanced Raman spectroscopy nanoparticle tags (SERS nanotags). SERS tag also known as SERS-labeled nanoparticle in general comprised of metal nanoparticle (Au or Ag) which enhances plasmon. These nanoparticles are coated with a monolayer of Raman reporter (RaR) as fingerprints covered by a protective layer. The monolayer of RaR enhance the intensity and ensure stability and reliable signal. In addition, the protective layer can be tuned with biomolecules (proteins, DNA, or RNA). SERS is a phenomenon of inelastic optical scattering, where light is scattered based on the vibrational mode of molecules undergoing scattering. In David’s research, Raman reporter was not only used as a light scattering enhancer and stabilizer but also acted as “fingerprint.” They used different Raman reporter to create unique optical “fingerprint.” Since different Raman reporter can be differentiated in a complex mixture, thus different SERS tag was functionalized to detect different targets. The advantage of this approach was to enable detection with low-cost optics and to efficiently detect different targets at a time (multiplexed assays). To create a stable signal, they encapsulated SERS-active Raman reporter between gold core and silica encapsulation. The outer silica surface was functionalized by coating with the sulfhydryl group for convenient bio-conjugation with affinity reagent. Regardless of the Raman reporter, the scattered SERS signal was read in the NIR (820–914 nm); in this range, the body fluids show minimal absorbance, thus permitting the use of body fluids directly by eliminating washing step. Multiplexed detection was conducted for Ebola, Lassa, and malaria from the mixture of three pairs of SERS nanotag/magnetic microparticle reagents. Each pair of SERS nanotag is conjugated to antibodies targeting a single disease antigen. On the appearance of a specific antigen in the sample, the SERS tag bind to it and make a sandwich complex over the surface of the particle. The nanoparticles are collected

426

R. Ganguly and C.-S. Lee

from the fluid through magnetic pull and detected. The results show high efficiency of detection with minimum cross-reactivity (Sebba et al. 2018). This excellent approach comes with few drawbacks to overcome, such as sample matrix (frozen versus fresh, serum versus whole blood). In addition to this, the performance of the device during a real outbreak is unproven. However, the report of this device usage in the COVID-19 outbreak is unavailable. Another obstacle in quickly implementing this device is its dependence on antibodies, which at status is unpredictable and requires active sampling from an infected person. In addition, finding the right antibody requires preliminary research.

Microfluidics for Host–Viral Interaction Study Investigation in recent days on viral interaction with the host immune system shows a critical role in determining the virus–host range and its virulence. Constant evolutionary impact over the interacting proteins causes viruses to encounter several mutations to improve replication or other necessary survival factors. These mutations might, chance, or enhance the ability of viral antagonists to resist the immune power of newly encountered host. Similar to every other organism, hosts also keep adapting to the prevalent environment, thus, antiviral proteins synthesized by the host cells are shaped by the multiple interactions with various viral species. In the process, multiple hosts might share a conserved sequence of interacting proteins or pathways. This gives an opportunity for many viral species to establish successful infection in a new host. These interactions initiate cross-species transmission. In addition to many possibilities, virologist recently began to appreciate the versatility of defective viral genome (DVGs) as a possible driving force to host–viral interaction. DVGs are being acknowledged for providing means for adaptation, immune escape, and virus perpetuation. Next-generation sequencing (NSG) data claims the generation of many DVGs during single viral infection, in which few DVGs are being repeatedly detected (Sun et al. 2019). This dynamic play within a host leads to a selection of DVGs with superior relative fitness in the relation of wild type and other generated DVGs. In this selection process, the vast majority of DVGs will be lost due to the anatomical barrier or during host-to-host transmission (McCrone et al. 2018). However, few DVGs can make through such barriers in the immunosuppressed host or in cases where viral particles are co-packed with DVGs and whole-genome during assemble (Loney et al. 2009). Method to understand these dynamics are limited and no single technique is sufficient to determine all details. Combining information from various sources has the potential to illuminate the complete framework with more clarity of the downstream mechanism. Few microfluidic devices are been reported for the study of host–viral interaction. One such study show neuron-to-cell spread and axonal transport of pseudorabies virus (PVR). Wendy W. Liu et al. in their research has found several key factors necessary for successful infection. PVR spreads from neuron cell to epithelial cell through intact axons, also the infection mainly depends on gB (viral glycoprotein that mediates membrane fusion) than gD (glycoprotein that essential for extracellular virions to

17

Microfluidic and Nanomaterial Approach for Virology

427

infect). In addition, Us9 (viral membrane protein essential for axonal sorting of structural protein of virus) was identified as a necessary protein for spreading (Liu et al. 2008). Another report by Yin et al. demonstrated microwell-based device to capture single-cell based on Poisson distribution and its interaction with vesicular stomatitis virus (VSV). In this study, the microwells were separated by sealing the top of the microwell with a glass slide. A dual-color fluorescent reporter system was established to track the dynamics of viral infection (Warrick et al. 2016). Likewise, the prominent role of single-cell RNA-seq (scRNA-Seq) has been realized and reported to address the variation between cell-to-cell. Incorporation of other computational tools such as Principal Component Analysis (PCA) or t-distributed Stochastic Neighbor Embedding (t-SNE) plots to upgrade detection methods. Thus, assessment of the same cell for phenotypic analysis and subsequent screening of scRNA-Seq will indicate the overlap between gene expression signature with a specific phenotype (Fig. 5).

Conclusion Methods for single-cell analysis address unique sets of questions compared to conventional methods that are capable of follow-up investigation. These methods are significantly contributing to various fields that are markedly influenced by cellular heterogeneity and dynamics. Through uninterrupted in vitro cultivation of single-cell and imaging, improve understanding of the kinetics of viral infection. Additional tracking of signaling molecules, transcriptome profiling of specific phenotypes could not only broaden understanding of host–viral interaction but could provide clues to biomarker investigation. To fully characterize cell–viral interaction, it is worth knowing and tracking all the key players for establishing successful infection.

Future Direction Research needs to focus more on newly evolving viruses. Many fields of research, however, are helping science to efficiently detect viruses, such as conventional PCR and multiplex PCR, nanomaterials, LFTs (Hwang et al. 2017; Hong et al. 2018; Bae and Kim 2010; Lee et al. 2006), but recent outbreaks of emerging viruses are causing mass destruction. These emerging viruses are considered to cross the barriers of genetic variation and move from one species to another. Thus many raised questions are yet to answer. Can microfluidics answer the mechanism of DVGs generation? How does it impact the virulence of a virus? Can screening of DVGs help to make a future prediction of the upcoming outbreak? What is the involvement of host factors in upgrading the viral antagonist? Technological renovation and innovative microfluidic platforms along with interdisciplinary research will be required to obtain the answers.

Fig. 5 (a) Live imaging of capside and VP22 transport dynamics within microfluidic chamber (Cell bodies in the stomal compartment were infected with PRV 181.). It is the time-lapse image of axons at 11–14 h postinfection using Leica SP5 confocal microscope. (b) (a) In a glass slide, region of microwells is separated using grooves and moats to allow multiple testing sites in a single device, and a droplet is maintained due to surface tension. In addition, it shows the settled cells within the microwells from the droplet. (b) Shows the schematic to either allow or block cross-talk or communication between the microwells (Warrick et al. 2016 and Liu et al. 2008)

428 R. Ganguly and C.-S. Lee

17

Microfluidic and Nanomaterial Approach for Virology

429

References Alivisatos P (2004) The use of nanocrystals in biological detection. Nat Biotechnol 22(1):47–52 Au SH et al (2013) Cellular bias on the microscale: probing the effects of digital microfluidic actuation on mammalian cell health, fitness and phenotype. Integr Biol 5(8):1014–1025 Bae JE, Kim IS (2010) Multiplex PCR for rapid detection of minute virus of mice, bovine parvovirus, and bovine herpesvirus during the manufacture of cell culture-derived biopharmaceuticals. Biotechnol Bioprocess Eng 15:1031–1037 Ben-Ari Ya et al (2013) Microfluidic large scale integration of viral–host interaction analysis. Lab Chip 13(12):2202–2209 Cao YC, Jin R, Mirkin CA (2002) Nanoparticles with Raman spectroscopic fingerprints for DNA and RNA detection. Science 297(5586):1536–1540 Chang Y-F et al (2010) Detection of swine-origin influenza A (H1N1) viruses using a localized surface plasmon coupled fluorescence fiber-optic biosensor. Biosens Bioelectron 26(3):1068– 1073 Chen DL, Gerdts CJ, Ismagilov RF (2005) Using microfluidics to observe the effect of mixing on nucleation of protein crystals. J Am Chem Soc 127(27):9672–9673 Choi K et al (2012) Digital microfluidics. 5(1):413–440 Cui C, Griffiths A, Li G, Silva LM, Kramer MF, Gaasterland T, Wang XJ, Coen DM (2006) Prediction and identification of herpes simplex virus 1-encoded microRNAs. J Virol 80(11):5499–5508 DaPalma T et al (2010) A systematic approach to virus–virus interactions. Virus Res 149(1):1–9 Diaz-Munoz SL (2017) Viral coinfection is shaped by host ecology and virus–virus interactions across diverse microbial taxa and environments. Virus Evol 3(1) Dougan JA et al (2007) Enhanced oligonucleotide–nanoparticle conjugate stability using thioctic acid modified oligonucleotides. Nucleic Acids Res 35(11):3668–3675 Draz MS et al (2012) Hybrid nanocluster plasmonic resonator for immunological detection of hepatitis B virus. ACS Nano 6(9):7634–7643 Driskell JD et al (2011) One-step assay for detecting influenza virus using dynamic light scattering and gold nanoparticles. Analyst 136(15):3083–3090 Flores CO et al (2011) Statistical structure of host–phage interactions. Proc Natl Acad Sci U S A 108(28):E288–E297 Ganbold E-O et al (2012) Aggregation effects of gold nanoparticles for single-base mismatch detection in influenza A (H1N1) DNA sequences using fluorescence and Raman measurements. Colloids Surf B: Biointerfaces 93:148–153 Gerhardt M et al (2005) In-depth, longitudinal analysis of viral quasispecies from an individual triply infected with late-stage human immunodeficiency virus type 1, using a multiple PCR primer approach. J Virol 79(13):8249–8261 Glynou K et al (2003) Oligonucleotide-functionalized gold nanoparticles as probes in a dry-reagent strip biosensor for DNA analysis by hybridization. Anal Chem 75(16):4155–4160 Griffin J et al (2009) Size-and distance-dependent nanoparticle surface-energy transfer (NSET) method for selective sensing of hepatitis C virus RNA. Chemistry 15(2):342–351 Griffiths C, Drews SJ, Marchant DJ (2017) Respiratory syncytial virus: infection, detection, and new options for prevention and treatment. Clin Microbiol Rev 30(1):277–319 Guo F et al (2010) Valve-based microfluidic device for droplet on-demand operation and static assay. Appl Phys Lett 97(23):233701 Hong SC et al (2015) Continuous aerosol size separator using inertial microfluidics and its application to airborne bacteria and viruses. Lab Chip 15(8):1889–1897 Hong W, Jeong SG, Shim G, Kim DY, Pack SP, Lee CS (2018) Improvement in the reproducibility of a paper-based analytical device (PAD) using stable covalent binding between proteins and cellulose paper. Biotechnol Bioprocess Eng 23(6):686–692 Hu Y, Xu P, Luo J, He H, Du W (2017) Absolute quantification of H5-subtype avian influenza viruses using droplet digital loop-mediated isothermal amplification. Anal Chem 89(1):745–750

430

R. Ganguly and C.-S. Lee

Hwang BH, Shin HH, Cha HJ (2017) Optimization of DNA microarray biosensors enables rapid and sensitive detection. Biotechnol Bioprocess Eng 22:469–473 Jeong S-G et al (2015) Toward instrument-free digital measurements: a three-dimensional microfluidic device fabricated in a single sheet of paper by double-sided printing and lamination. Lab Chip 15(4):1188–1194 Laderman EI et al (2008) Rapid, sensitive, and specific lateral-flow immunochromatographic pointof-care device for detection of herpes simplex virus type 2-specific immunoglobulin G antibodies in serum and whole blood. Clin Vaccine Immunol 15(1):159–163 Lecellier CH, Dunoyer P, Arar K, Lehmann-Che J, Eyquem S, Himber C, Saïb A, Voinnet O (2005) A cellular microRNA mediates antiviral defense in human cells. Science 308(5721):557–560 Lee CS, Lee SH, Kim YG, Oh MK, Hwang TS, Rhee YW, Song HM, Kim BY, Kim YK, Kim BG (2006) Fabrication of disposable protein chip for simultaneous sample detection. Biotechnol Bioprocess Eng 11:455 Lei K-M et al (2015) A palm-size μNMR relaxometer using a digital microfluidic device and a semiconductor transceiver for chemical/biological diagnosis. Analyst 140(15):5129–5137 Li H, Rothberg LJ (2004) Colorimetric detection of DNA sequences based on electrostatic interactions with unmodified gold nanoparticles. Proc Natl Acad Sci U S A 101(39):14036– 14039 Li H et al (2010) High multiplicity infection by HIV-1 in men who have sex with men. PLoS Pathog 6(5) Li X et al (2012) A fast and sensitive immunoassay of avian influenza virus based on label-free quantum dot probe and lateral flow test strip. Talanta 100:1–6 Link DR et al (2004) Geometrically mediated breakup of drops in microfluidic devices. Phys Rev Lett 92(5):054503 Liu WW, Goodhouse J, Jeon NL, Enquist LW (2008) A microfluidic chamber for analysis of neuron-to-cell spread and axonal transport of an alpha-herpesvirus. PLoS One 3(6) Loney C, Mottet-Osman G, Roux L, Bhella D (2009) Paramyxovirus ultrastructure and genome packaging: cryo-electron tomography of sendai virus. J Virol 83(16):8191–8197 Lu X et al (2013) A gold nanorods-based fluorescent biosensor for the detection of hepatitis B virus DNA based on fluorescence resonance energy transfer. Analyst 138(2):642–650 Mancuso M et al (2013) Multiplexed colorimetric detection of Kaposi’s sarcoma associated herpesvirus and Bartonella DNA using gold and silver nanoparticles. Nanoscale 5(4):1678– 1686 Martin JG et al (2009) Toward an artificial Golgi: redesigning the biological activities of heparan sulfate on a digital microfluidic chip. J Am Chem Soc 131(31):11041–11048 McCrone JT, Woods RJ, Martin ET, Malosh RE, Monto AS, Lauring AS (2018) Stochastic processes constrain the within and between host evolution of influenza virus. elife 7:e35962 Mitchell P (2002) A perspective on protein microarrays. Nature Biotech 20(3):225–229 Munson-McGee JH et al (2018) A virus or more in (nearly) every cell: ubiquitous networks of virus–host interactions in extreme environments. ISME J 12(7):1706–1714 Neng J et al (2013) Surface-enhanced Raman scattering (SERS) detection of multiple viral antigens using magnetic capture of SERS-active nanoparticles. Biosens Bioelectron 41:316–321 Niu X et al (2007) Real-time detection, control, and sorting of microfluidic droplets. Biomicrofluidics 1(4):44101–44101 Niu X et al (2008) Pillar-induced droplet merging in microfluidic circuits. Lab Chip 8(11):1837– 1841 Niu X et al (2009) Droplet-based compartmentalization of chemically separated components in two-dimensional separations. Lab Chip (41):6159–6161 Rosi NL, Mirkin CA (2005) Nanostructures in biodiagnostics. Chem Rev 105(4):1547–1562 Roux S et al (2015) Viral dark matter and virus–host interactions resolved from publicly available microbial genomes. elife 4:e08490 Schmitz CH et al (2009) Dropspots: a picoliter array in a microfluidic device. Lab Chip 9(1):44–49

17

Microfluidic and Nanomaterial Approach for Virology

431

Sebba D, Lastovich AG, Kuroda M, Fallows E, Johnson J, Ahouidi A, Honko AN, Fu H, Nielson R, Carruthers E, Diédhiou C (2018) A point-of-care diagnostic for differentiating Ebola from endemic febrile diseases. Sci Transl Med 10(471):eaat0944 Shawky SM, Bald D, Azzazy HM (2010) Direct detection of unamplified hepatitis C virus RNA using unmodified gold nanoparticles. Clin Biochem 43(13–14):1163–1168 Shemesh J et al (2010) Advanced microfluidic droplet manipulation based on piezoelectric actuation. Biomed Microdevices 12(5):907–914 Sista RS et al (2011) Digital microfluidic platform for multiplexing enzyme assays: implications for lysosomal storage disease screening in newborns. Clin Chem 57(10):1444–1451 Solvas XC, DeMello A (2011) Droplet microfluidics: recent developments and future applications. Chem Commun 47(7):1936–1942 Spackman E et al (2002) Development of a real-time reverse transcriptase PCR assay for type A influenza virus and the avian H5 and H7 hemagglutinin subtypes. J Clin Microbiol 40(9):3256–3260 Sun Y et al (2019) Correction: a specific sequence in the genome of respiratory syncytial virus regulates the generation of copy-back defective viral genomes. PLoS Pathog 15(10) Swyer I et al (2016) Interfacing digital microfluidics with high-field nuclear magnetic resonance spectroscopy. Lab Chip 16(22):4424–4435 Talapatra A, Rouse R, Hardiman GJP (2002) Protein microarrays: challenges and promises. Pharmacogenomics 3(4):527–536 Toh SY et al (2015) Aptamers as a replacement for antibodies in enzyme-linked immunosorbent assay. Biosens Bioelectron 64:392–403 Tseng Y-T et al (2016) Integrated microfluidic system for rapid detection of influenza H1N1 virus using a sandwich-based aptamer assay. Biosens Bioelectron 82:105–111 Unger M, et al (2000) Integrated elastomer fluidic Lab-on-a-chip-Surface patterning and DNA diagnostics. 288:113–116 Wang X et al (2009) Engineering nanomaterial surfaces for biomedical applications. Exp Biol Med (Maywood) 234(10):1128–1139 Wang X et al (2010) Gold nanorod-based localized surface plasmon resonance biosensor for sensitive detection of hepatitis B virus in buffer, blood serum and plasma. Biosens Bioelectron 26(2):404–410 Wang X et al (2012) Detection of hepatitis B surface antigen by target-induced aggregation monitored by dynamic light scattering. Anal Biochem 428(2):119–125 Warrick JW, Timm A, Swick A, Yin J (2016) Tools for single-cell kinetic analysis of virus-host interactions. PLoS One 11(1):e0145081 Wu J-C et al (2014) Electrophoresis-enhanced detection of deoxyribonucleic acids on a membranebased lateral flow strip using avian influenza H5 genetic sequence as the model. Sensors (Basel) 14(3):4399–4415 Xu P, Zheng X, Tao Y, Du W (2016) Cross-interface emulsification for generating size-tunable droplets. Anal Chem 88(6):3171–3177 Xue KS et al (2018) Cooperating H3N2 influenza virus variants are not detectable in primary clinical samples. mSphere 3(1):e00552–e00517 Yeh YT, Gulino K, Zhang Y, Sabestien A, Chou TW, Zhou B, Lin Z, Albert I, Lu H, Swaminathan V, Ghedin E (2020) A rapid and label-free platform for virus capture and identification from clinical samples. Proc Natl Acad Sci 117(2):895–901 Zehbe I et al (1997) Sensitive in situ hybridization with catalyzed reporter deposition, streptavidinNanogold, and silver acetate autometallography: detection of single-copy human papillomavirus. Am J Pathol 150(5):1553 Zeng Q et al (2012) Multiple homogeneous immunoassays based on a quantum dots–gold nanorods FRET nanoplatform. Chem Commun 48(12):1781–1783

Part III Chemical Methods for Single Cell Technology

Liposome-Mediated Material Transfer in Single Cells

18

Mamiko Tsugane and Hiroaki Suzuki

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preparation and Physical Characteristics of Liposomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Liposome Fusion as a Model System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Transfer of Materials from Liposomes into Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Modification Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

436 437 440 441 443 444 445

Abstract

Liposomes are artificially prepared lipid bilayer vesicles. Among these, unilamellar liposomes are reminiscent of intracellular and extracellular vesicles, which serve as vehicles that transport proteins, nucleic acids, and other signaling molecules to other organelles and cells. These materials are encapsulated or released from vesicles via the fusion and division of lipid bilayers. Simulating this phenomenon to introduce substances into cells is thought to be useful for genetic modification and the functional analysis of cells. Conventionally, the main target material to be delivered into cells is small-size DNA such as plasmids. However, with the recent advancement of biotechnology and synthetic biology, target materials have expanded to include large materials and molecular systems. Conventional transfection methods, such as electroporation, lipofection, and viral transfection, are generally inefficient for introducing large-size substances into cells or multiple species of substances. Liposomes with volumes greater than M. Tsugane Faculty of Science and Engineering, Chuo University, Tokyo, Japan Japan Society for the Promotion of Science (JSPS), Tokyo, Japan H. Suzuki (*) Faculty of Science and Engineering, Chuo University, Tokyo, Japan e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_13

435

436

M. Tsugane and H. Suzuki

a femtoliter (>1 μm3) are capable of enclosing substances of relatively large size and thus can be used for these applications. In this chapter, we review the recent trends in liposome preparation and material transfer methods and discuss future applications toward single-cell modification technologies.

Introduction The introduction of bioactive substances into cells has many applications in cell biology, such as the modification of cell functions and the introduction of probes to elicit their properties. Endogenously, cells use membrane vesicles for transferring substances between cells and organelles (Conner and Schmid 2003; Glick and Nakano 2009). Therefore, with respect to cell engineering, liposome-mediated material transfer methods could be highly biocompatible, efficient, and less stressful to both cells and substances to be delivered. Liposomes are artificial vesicles prepared from phospholipids that are capable of carrying hydrophilic molecules in their inner water phase and hydrophobic molecules in their lipid membrane. Nucleic acids (i.e., DNA and RNA), proteins (enzymes), and low-molecular-weight drugs are the most important substances that are delivered into cells (Walde and Ichikawa 2001). Given that there are many synthetic vectors for the intracellular delivery of bioactive molecules, such as cationic lipids (widely known as the lipofection method), cationic polymers, and membrane-permeating peptides (Luo and Saltzman 2000), it is often difficult to choose an appropriate method and conditions for a specific application. In addition to the conventional transfection of small and single-species molecules, there is a growing need to introduce a set of multiple-species molecules and/or large substances into cells. For example, the introduction of multiple genes is required to transform multiple properties of cells (e.g., through co-transfection) or to construct synthetic gene circuits (Lienert et al. 2014). In genome editing, a set of guiding RNA and the enzyme (Cas9 nuclease) often must be introduced together (Cong et al. 2013; Lin et al. 2018). Furthermore, to completely alter the cellular processes, one must implant synthetic genome DNA, which is much larger than 100 kb (Kazuki and Oshimura 2011; Lee and Jaenisch 1996; Martella et al. 2016). At present, all of these applications suffer from low efficiency in their delivery processes. In addition, the introduction of various nano to micro artificial substances is expected to create bio-artificial hybrid molecular systems (Hagiya et al. 2014; Murata et al. 2013). Although exploration into these directions is still at very primitive stages, the ability of lipid vesicles to hold substances in femto- to picoliter volumes should be beneficial in introducing molecular systems and larger molecules via membrane fusion. In this chapter, we emphasize past attempts and future possibilities of introducing multiple and large components into cells via liposome fusion in vitro. In the following sections, we overview liposome formation and material encapsulation techniques relevant to this topic. Moreover, we outline the characteristics and current status of liposome-mediated material transfer methods and discuss future

18

Liposome-Mediated Material Transfer in Single Cells

437

possibilities for single-cell modification technologies. Readers who are interested in traditional drug delivery systems (DDS) in vivo should refer to excellent reviews and books published elsewhere.

Preparation and Physical Characteristics of Liposomes It is easy to prepare liposomes per se. The simplest and most widely used preparation protocols are based on the hydration of lipid films (or swelling) (Fig. 1a) (Horne et al. 1963; Lasic 1988). Usually, phospholipids, which consist of a hydrophilic head group and two acyl chains (either unsaturated or saturated), are dissolved in an organic solvent (often chloroform), which are then slowly dried on a solid substrate. This dried film consists of the stack of phospholipid bilayer structures. Upon hydration, water molecules penetrate between these layers, and they then swell into balloon-like structures. Finally, the membrane is detached from the substrate (often due to the hydrodynamic shear), and the exposed edges are closed to form a mature liposome. Since there is no control factor for liposome size and morphology in this method, liposomes formed by general hydration result in multilamellar vesicles (MLVs) with widely varying sizes. The physical characteristics of the lipid bilayer allow it to be a versatile building block for nano- and microscale compartments. For instance, the lipid bilayer is an

Fig. 1 Schematic of liposome formation methods. (a) Simple hydration. (b) Electroformation. (c) Extrusion. (d) Freeze-dried empty liposome (FDEL) method. (e) W/O emulsion transfer method

438

M. Tsugane and H. Suzuki

extremely soft sheet. With the typical bending modulus of 10–20 kBT, it only cost ~250–500 kBT (~10 18 J) to bend a flat membrane into a sphere. In addition, this bending energy is scale-free, which means that there is no difference in terms of the required energy to form nano- and micro-sized vesicles. In living cells, lipid bilayers form the plasma membrane of the cell itself (>1 μm), organelles with their complicated (often wrinkled or intertwined) structures (104 cells/s and 300–500 cells/s, respectively) and allow the analysis of many different proteins in parallel, but also require serious investments in instruments, staff, and assay development. While promising for the future, the current microfluidic systems including micro- and nanowell assays, microchamber assays and droplet-based assays have not yet translated into clinical settings. All three methods allow the direct, quantitative, as well as kinetic assessment of protein secretion and offer the possibility of cell recovery, but they differ in multiplexity (1–4, 1–42, and 1–10 parameters, respectively) and throughput (104–105 cells, 103 cells, and 106 cells per experiment, respectively). Furthermore, the label-free optical methods iSCAT microscopy and plasmonic biosensors were discussed. While promising methods for the future, they remain exploratory. Current microfluidic approaches still need considerable expertise of the user, and interlaboratory variability in the results remains a serious challenge, both limiting their translatability and application. Thus, the development of a new generation of user-friendly and robust microfluidic systems remains key to their widespread application. In addition to simplification and user-friendliness, future work needs to improve sensitivity, throughput, and multiplexing capabilities of the systems further, while simultaneously reducing error and bias. While challenging, these efforts will need to result in quality-controlled, comparable data that enables researchers to evaluate and interpret the results if these technologies are to be applied widely. Indeed, the commercialization and clinical translation of microfluidic technologies has just recently started, but the barriers and challenges facing the application of these technologies in clinical settings remain high. Additionally, the integration of systems that allow the correlative measurement of cellular secretion with subsequent addition of phenotypic, genomic, transcriptomic, or functional analysis remains a field of active research. Doing so will generate more comprehensive data sets and help to put secretion into the context of the complete cellular machinery, functionality, state, and metabolism.

21

Biochemical Analysis of Secreted Molecules by Individual Cells

513

Lastly, it is of interest to note that many developed technologies use antibodies for detection. While convenient, this also links the spectrum of analyzable molecules to the spectrum of molecules against which antibodies can be generated, mostly proteins. Most methods are therefore restricted to be used for the analysis of protein secretion. Interestingly, but also encouraging for researchers focusing on cellular secretion, fewer methods are available for the investigation of secreted metabolites. Examples in literature can be found though where various methods are used to measure one or a selection of defined metabolites. However, a concise study of the secretome, i.e., the sum of all cellular secreted proteins, or the study of the secreted metabolome remains out of reach. Here, the combination of label-free, universal detection methods like nuclear magnetic resonance spectroscopy (NMR) or especially mass spectrometry (MS) look promising. Suitable approaches are being explored but have not yet been applied or published to analyze the secretion of individual cells. Nevertheless, there is great potential in the development of methods allowing studying the secretion of metabolites at the single-cell level which should be exploited.

Conclusion Secreted proteins and metabolites play key roles in almost all biological systems, organizing cells into tissues, controlling intercellular communication, cellular migration, and the body’s defense system. Single-cell resolution in the analysis of secretion provides the resolution needed to study these potent mediators in the highly heterogeneous context that they are produced in and allows correlating secretion with the secreting cell itself. The analysis of this secretion on the single-cell level takes cellular heterogeneity, subpopulations and rare but active cells into this consideration. Exploiting and advancing the correlations in a multi-omics setting is a necessary next step, and future advances that integrate additional transcriptome or functional parameters in this analysis have the potential to revolutionize biological research and precision medicine.

References Armbrecht L, Müller RS, Nikoloff J, Dittrich PS (2019) Single-cell protein profiling in microchambers with barcoded beads. Microsyst Nanoeng 5:55 Bailey RC, Kwong GA, Radu CG, Witte ON, Heath JR (2007) DNA-encoded antibody libraries: a unified platform for multiplexed cell sorting and detection of genes and proteins. J Am Chem Soc 129:1959–1967 Bandura DR, Baranov VI, Ornatsky OI, Antonov A, Kinach R, Lou X, Pavlov S, Vorobiev S, Dick JE, Tanner SD (2009) Mass cytometry: technique for real time single cell multitarget immunoassay based on inductively coupled plasma time-of-flight mass spectrometry. Anal Chem 81:6813–6822 Braesch-Andersen S, Paulie S, Smedman C, Mia S, Kumagai-Braesch M (2013) ApoE production in human monocytes and its regulation by inflammatory cytokines. PLoS One 8:e79908

514

O. T. M. Bucheli et al.

Brandi J, Dalla Pozza E, Dando I, Biondani G, Robotti E, Jenkins R, Elliott V, Park K, Marengo E, Costello E, Scarpa A, Palmieri M, Cecconi D (2016) Secretome protein signature of human pancreatic cancer stem-like cells. J Proteome 136:1–12 Brosterhus H, Brings S, Leyendeckers H, Manz RA, Miltenyi S, Radbruch A, Assenmacher M, Schmitz J (1999) Enrichment and detection of live antigen-specific CD4(+) and CD8(+) T cells based on cytokine secretion. Eur J Immunol 29:4053–4059 Chen Z, Chen JJ, Fan R (2019) Single-cell protein secretion detection and profiling. Annu Rev Anal Chem (Palo Alto, Calif) 12:431–449 Choi J, Love KR, Gong Y, Gierahn TM, Love JC (2011) Immuno-hybridization chain reaction for enhancing detection of individual cytokine-secreting human peripheral mononuclear cells. Anal Chem 83:6890–6895 Cossarizza A, Chang HD, Radbruch A, Akdis M, Andra I, Annunziato F, Bacher P, Barnaba V, Battistini L, Bauer WM, Baumgart S, Becher B, Beisker W, Berek C, Blanco A, Borsellino G, Boulais PE, Brinkman RR, Buscher M, Busch DH, Bushnell TP, Cao X, Cavani A, Chattopadhyay PK, Cheng Q, Chow S, Clerici M, Cooke A, Cosma A, Cosmi L, Cumano A, Dang VD, Davies D, De Biasi S, Del Zotto G, Della Bella S, Dellabona P, Deniz G, Dessing M, Diefenbach A, Di Santo J, Dieli F, Dolf A, Donnenberg VS, Dorner T, Ehrhardt GRA, Endl E, Engel P, Engelhardt B, Esser C, Everts B, Dreher A, Falk CS, Fehniger TA, Filby A, Fillatreau S, Follo M, Forster I, Foster J, Foulds GA, Frenette PS, Galbraith D, Garbi N, Garcia-Godoy MD, Geginat J, Ghoreschi K, Gibellini L, Goettlinger C, Goodyear CS, Gori A, Grogan J, Gross M, Grutzkau A, Grummitt D, Hahn J, Hammer Q, Hauser AE, Haviland DL, Hedley D, Herrera G, Herrmann M, Hiepe F, Holland T, Hombrink P, Houston JP, Hoyer BF, Huang B, Hunter CA, Iannone A, Jack HM, Javega B, Jonjic S, Juelke K, Jung S, Kaiser T, Kalina T, Keller B, Khan S, Kienhofer D, Kroneis T, Kunkel D, Kurts C, Kvistborg P, Lannigan J, Lantz O, Larbi A, LeibundGut-Landmann S, Leipold MD, Levings MK, Litwin V, Liu Y, Lohoff M, Lombardi G, Lopez L, Lovett-Racke A, Lubberts E, Ludewig B, Lugli E, Maecker HT, Martrus G, Matarese G, Maueroder C, McGrath M, McInnes I, Mei HE, Melchers F, Melzer S, Mielenz D, Mills K, Mirrer D, Mjosberg J, Moore J, Moran B, Moretta A, Moretta L, Mosmann TR, Muller S, Muller W, Munz C, Multhoff G, Munoz LE, Murphy KM, Nakayama T, Nasi M, Neudorfl C, Nolan J, Nourshargh S, O’Connor JE, Ouyang W, Oxenius A, Palankar R, Panse I, Peterson P, Peth C, Petriz J, Philips D, Pickl W, Piconese S, Pinti M, Pockley AG, Podolska MJ, Pucillo C, Quataert SA, Radstake T, Rajwa B, Rebhahn JA, Recktenwald D, Remmerswaal EBM, Rezvani K, Rico LG, Robinson JP, Romagnani C, Rubartelli A, Ruckert B, Ruland J, Sakaguchi S, Sala-de-Oyanguren F, Samstag Y, Sanderson S, Sawitzki B, Scheffold A, Schiemann M, Schildberg F, Schimisky E, Schmid SA, Schmitt S, Schober K, Schuler T, Schulz AR, Schumacher T, Scotta C, Shankey TV, Shemer A, Simon AK, Spidlen J, Stall AM, Stark R, Stehle C, Stein M, Steinmetz T, Stockinger H, Takahama Y, Tarnok A, Tian Z, Toldi G, Tornack J, Traggiai E, Trotter J, Ulrich H, van der Braber M, van Lier RAW, Veldhoen M, Vento-Asturias S, Vieira P, Voehringer D, Volk HD, von Volkmann K, Waisman A, Walker R, Ward MD, Warnatz K, Warth S, Watson JV, Watzl C, Wegener L, Wiedemann A, Wienands J, Willimsky G, Wing J, Wurst P, Yu L, Yue A, Zhang Q, Zhao Y, Ziegler S, Zimmermann J (2017) Guidelines for the use of flow cytometry and cell sorting in immunological studies. Eur J Immunol 47:1584–1797 Czerkinsky CC, Nilsson LA, Nygren H, Ouchterlony O, Tarkowski A (1983) A solid-phase enzyme-linked immunospot (ELISPOT) assay for enumeration of specific antibody-secreting cells. J Immunol Methods 65:109–121 Del Ben F, Turetta M, Celetti G, Piruska A, Bulfoni M, Cesselli D, Huck WT, Scoles G (2016) A method for detecting circulating tumor cells based on the measurement of single-cell metabolism in droplet-based microfluidics. Angew Chem Int Ed Engl 55:8581–8584 Dhar M, Lam JN, Walser T, Dubinett SM, Rettig MB, Di Carlo D (2018) Functional profiling of circulating tumor cells with an integrated vortex capture and single-cell protease activity assay. Proc Natl Acad Sci U S A 115:9986–9991 Dimou E, Nickel W (2018) Unconventional mechanisms of eukaryotic protein secretion. Curr Biol 28:R406–Rr10

21

Biochemical Analysis of Secreted Molecules by Individual Cells

515

Dranoff G (2004) Cytokines in cancer pathogenesis and cancer therapy. Nat Rev Cancer 4:11–22 Eyer K, Kuhn P, Hanke C, Dittrich PS (2012) A microchamber array for single cell isolation and analysis of intracellular biomolecules. Lab Chip 12:765–772 Eyer K, Doineau RCL, Castrillon CE, Briseno-Roa L, Menrath V, Mottet G, England P, Godina A, Brient-Litzler E, Nizak C, Jensen A, Griffiths AD, Bibette J, Bruhns P, Baudry J (2017) Singlecell deep phenotyping of IgG-secreting cells for high-resolution immune monitoring. Nat Biotechnol 35:977–982 Fitzgerald W, Grivel JC (2013) A universal nanoparticle cell secretion capture assay. Cytometry A 83:205–211 Gerard A, Woolfe A, Mottet G, Reichen M, Castrillon C, Menrath V, Ellouze S, Poitou A, Doineau R, Briseno-Roa L, Canales-Herrerias P, Mary P, Rose G, Ortega C, Delince M, Essono S, Jia B, Iannascoli B, Richard-Le Goff O, Kumar R, Stewart SN, Pousse Y, Shen B, Grosselin K, Saudemont B, Sautel-Caille A, Godina A, McNamara S, Eyer K, Millot GA, Baudry J, England P, Nizak C, Jensen A, Griffiths AD, Bruhns P, Brenan C (2020) Highthroughput single-cell activity-based screening and sequencing of antibodies using droplet microfluidics. Nat Biotechnol 38:715–721 Gillardon F, Schmid R, Draheim H (2012) Parkinson’s disease-linked leucine-rich repeat kinase 2 (R1441G) mutation increases proinflammatory cytokine release from activated primary microglial cells and resultant neurotoxicity. Neuroscience 208:41–48 Han Q, Bradshaw EM, Nilsson B, Hafler DA, Love JC (2010) Multidimensional analysis of the frequencies and rates of cytokine secretion from single cells by quantitative microengraving. Lab Chip 10:1391–1400 Han Q, Bagheri N, Bradshaw EM, Hafler DA, Lauffenburger DA, Love JC (2012) Polyfunctional responses by human T cells result from sequential release of cytokines. Proc Natl Acad Sci U S A 109:1607–1612 Han X, Liu K, Sun C (2019) Plasmonics for biosensing. Materials 12:1411. https://doi.org/10.3390/ ma12091411 Hovdenak N, Haram K (2012) Influence of mineral and vitamin supplements on pregnancy outcome. Eur J Obstet Gynecol Reprod Biol 164:127–132 Hughes AJ, Spelke DP, Xu Z, Kang CC, Schaffer DV, Herr AE (2014) Single-cell western blotting. Nat Methods 11:749–755 Jahnmatz P, Sundling C, Makower B, Sondén K, Färnert A, Ahlborg N (2020) Multiplex analysis of antigen-specific memory B cells in humans using reversed B-cell FluoroSpot. J Immunol Methods 478:112715 Janetzki S, Rabin R (2015) Enzyme-linked immunospot (ELISpot) for single-cell analysis. Methods Mol Biol 1346:27–46 Janetzki S, Rueger M, Dillenbeck T (2014) Stepping up ELISpot: multi-level analysis in FluoroSpot assays. Cell 3:1102–1115 Jin A, Ozawa T, Tajiri K, Obata T, Kishi H, Muraguchi A (2011) Rapid isolation of antigen-specific antibody-secreting cells using a chip-based immunospot array. Nat Protoc 6:668–676 Jung T, Schauer U, Heusser C, Neumann C, Rieger C (1993) Detection of intracellular cytokines by flow cytometry. J Immunol Methods 159:197–207 Junkin M, Kaestli AJ, Cheng Z, Jordi C, Albayrak C, Hoffmann A, Tay S (2016) High-content quantification of single-cell immune dynamics. Cell Rep 15:411–422 Kesa G, Larsson PH, Ahlborg N, Axelsson B (2012) Comparison of ELISpot and FluoroSpot in the analysis of swine flu-specific IgG and IgA secretion by in vivo activated human B cells. Cell 1:27–34 Kida A, Iijima M, Niimi T, Maturana AD, Yoshimoto N, Kuroda S (2013) Cell surface-fluorescence immunosorbent assay for real-time detection of hybridomas with efficient antibody secretion at the single-cell level. Anal Chem 85:1753–1759 Konry T, Dominguez-Villar M, Baecher-Allan C, Hafler DA, Yarmush ML (2011) Droplet-based microfluidic platforms for single T cell secretion analysis of IL-10 cytokine. Biosens Bioelectron 26:2707–2710

516

O. T. M. Bucheli et al.

Konry T, Golberg A, Yarmush M (2013) Live single cell functional phenotyping in droplet nanoliter reactors. Sci Rep 3:3179 Labib M, Kelley SO (2020) Single-cell analysis targeting the proteome. Nat Rev Chem 4:143–158 Li X, Soler M, Szydzik C, Khoshmanesh K, Schmidt J, Coukos G, Mitchell A, Altug H (2018) Label-free optofluidic nanobiosensor enables real-time analysis of single-cell cytokine secretion. Small 14:e1800698 Liechti T, Roederer M (2019) OMIP-060: 30-parameter flow cytometry panel to assess T cell effector functions and regulatory T cells. Cytometry A 95:1129–1134 Ling C, Del Guerra S, Lupi R, Ronn T, Granhall C, Luthman H, Masiello P, Marchetti P, Groop L, Del Prato S (2008) Epigenetic regulation of PPARGC1A in human type 2 diabetic islets and effect on insulin secretion. Diabetologia 51:615–622 Liu Y, Singh AK (2013) Microfluidic platforms for single-cell protein analysis. J Lab Autom 18:446–454 Love JC, Ronan JL, Grotenbreg GM, van der Veen AG, Ploegh HL (2006) A microengraving method for rapid selection of single cells producing antigen-specific antibodies. Nat Biotechnol 24:703–707 Lovelace P, Maecker HT (2018) Multiparameter intracellular cytokine staining. Methods Mol Biol 1678:151–166 Lu Y, Chen JJ, Mu L, Xue Q, Wu Y, Wu PH, Li J, Vortmeyer AO, Miller-Jensen K, Wirtz D, Fan R (2013) High-throughput secretomic analysis of single cells to assess functional cellular heterogeneity. Anal Chem 85:2548–2556 Lu Y, Xue Q, Eisele MR, Sulistijo ES, Brower K, Han L, Amir el AD, Pe’er D, Miller-Jensen K, Fan R (2015) Highly multiplexed profiling of single-cell effector functions reveals deep functional heterogeneity in response to pathogenic ligands. Proc Natl Acad Sci U S A 112:E607–E615 Ma C, Fan R, Ahmad H, Shi Q, Comin-Anduix B, Chodon T, Koya RC, Liu CC, Kwong GA, Radu CG, Ribas A, Heath JR (2011) A clinical microchip for evaluation of single immune cells reveals high functional heterogeneity in phenotypically similar T cells. Nat Med 17:738–743 Maecker HT, Harari A (2015) Immune monitoring technology primer: flow and mass cytometry. J Immunother Cancer 3:44 Manz R, Assenmacher M, Pfluger E, Miltenyi S, Radbruch A (1995) Analysis and sorting of live cells according to secreted molecules, relocated to a cell-surface affinity matrix. Proc Natl Acad Sci U S A 92:1921–1925 Mazutis L, Gilbert J, Ung WL, Weitz DA, Griffiths AD, Heyman JA (2013) Single-cell analysis and sorting using droplet-based microfluidics. Nat Protoc 8:870–891 McDonald MP, Gemeinhardt A, Konig K, Piliarik M, Schaffer S, Volkl S, Aigner M, Mackensen A, Sandoghdar V (2018) Visualizing single-cell secretion dynamics with single-protein sensitivity. Nano Lett 18:513–519 Mongersun A, Smeenk I, Pratx G, Asuri P, Abbyad P (2016) Droplet microfluidic platform for the determination of single-cell lactate release. Anal Chem 88:3257–3263 Mulukutla BC, Yongky A, Grimm S, Daoutidis P, Hu WS (2015) Multiplicity of steady states in glycolysis and shift of metabolic state in cultured mammalian cells. PLoS One 10:e0121561 Murphy TW, Zhang Q, Naler LB, Ma S, Lu C (2017) Recent advances in the use of microfluidic technologies for single cell analysis. Analyst 143:60–80 Ogunniyi AO, Story CM, Papa E, Guillen E, Love JC (2009) Screening individual hybridomas by microengraving to discover monoclonal antibodies. Nat Protoc 4:767–782 Ornatsky O, Bandura D, Baranov V, Nitz M, Winnik MA, Tanner S (2010) Highly multiparametric analysis by mass cytometry. J Immunol Methods 361:1–20 Papalexi E, Satija R (2018) Single-cell RNA sequencing to explore immune cell heterogeneity. Nat Rev Immunol 18:35–45 Qiu L, Wimmers F, Weiden J, Heus HA, Tel J, Figdor CG (2017) A membrane-anchored aptamer sensor for probing IFNgamma secretion by single cells. Chem Commun (Camb) 53:8066–8069 Rabouille C (2017) Pathways of unconventional protein secretion. Trends Cell Biol 27:230–240

21

Biochemical Analysis of Secreted Molecules by Individual Cells

517

Rugbjerg P, Sommer MOA (2019) Overcoming genetic heterogeneity in industrial fermentations. Nat Biotechnol 37:869–876 Salerno F, Engels S, van den Biggelaar M, van Alphen FPJ, Guislain A, Zhao W, Hodge DL, Bell SE, Medema JP, von Lindern M, Turner M, Young HA, Wolkers MC (2018) Translational repression of pre-formed cytokine-encoding mRNA prevents chronic activation of memory T cells. Nat Immunol 19:828–837 Saletti G, Cuburu N, Yang JS, Dey A, Czerkinsky C (2013) Enzyme-linked immunospot assays for direct ex vivo measurement of vaccine-induced human humoral immune responses in blood. Nat Protoc 8:1073–1087 Secchi G (2008) Role of protein in cosmetics. Clin Dermatol 26:321–325 Shembekar N, Chaipan C, Utharala R, Merten CA (2016) Droplet-based microfluidics in drug discovery, transcriptomics and high-throughput molecular genetics. Lab Chip 16:1314–1331 Shembekar N, Hu H, Eustace D, Merten CA (2018) Single-cell droplet microfluidic screening for antibodies specifically binding to target cells. Cell Rep 22:2206–2215 Sinha N, Subedi N, Tel J (2018) Integrating immunology and microfluidics for single immune cell analysis. Front Immunol 9:2373 Spitzer MH, Nolan GP (2016) Mass cytometry: single cells, many features. Cell 165:780–791 Stahlberg A, Kubista M, Aman P (2011) Single-cell gene-expression profiling and its potential diagnostic applications. Expert Rev Mol Diagn 11:735–740 Stiefel P, Mauerhofer S, Schneider J, Maniura-Weber K, Rosenberg U, Ren Q (2016) Enzymes enhance biofilm removal efficiency of cleaners. Antimicrob Agents Chemother 60:3647–3652 Sukovich DJ, Kim SC, Ahmed N, Abate AR (2017) Bulk double emulsification for flow cytometric analysis of microfluidic droplets. Analyst 142:4618–4622 Taylor RW, Sandoghdar V (2019) Interferometric scattering (iSCAT) microscopy and related techniques. In: Astratov V (ed) Label-free super-resolution microscopy. Springer, Cham Uhlén M, Karlsson MJ, Hober A, Svensson AS, Scheffel J, Kotol D, Zhong W, Tebani A, Strandberg L, Edfors F, Sjöstedt E, Mulder J, Mardinoglu A, Berling A, Ekblad S, Dannemeyer M, Kanje S, Rockberg J, Lundqvist M, Malm M, Volk AL, Nilsson P, Månberg A, Dodig-Crnkovic T, Pin E, Zwahlen M, Oksvold P, von Feilitzen K, Häussler RS, Hong MG, Lindskog C, Ponten F, Katona B, Vuu J, Lindström E, Nielsen J, Robinson J, Ayoglu B, Mahdessian D, Sullivan D, Thul P, Danielsson F, Stadler C, Lundberg E, Bergström G, Gummesson A, Voldborg BG, Tegel H, Hober S, Forsström B, Schwenk JM, Fagerberg L, Sivertsson Å (2019) The human secretome. Sci Signal 12. https://doi.org/10.1126/ scisignal.aaz0274 Walenta S, Mueller-Klieser WF (2004) Lactate: mirror and motor of tumor malignancy. Semin Radiat Oncol 14:267–274 Warburg O (1956) On the origin of cancer cells. Science 123:309–314 Woronoff G, Nghe P, Baudry J, Boitard L, Braun E, Griffiths AD, Bibette J (2020) Metabolic cost of rapid adaptation of single yeast cells. Proc Natl Acad Sci U S A 117:10660–10666 Wu M, Singh AK (2012) Single-cell protein analysis. Curr Opin Biotechnol 23:83–88 Yang L, George J, Wang J (2019) Deep profiling of cellular heterogeneity by emerging single-cell proteomic technologies. Proteomics 20:e1900226 Yuan Y, Brouchon J, Calvo-Calle JM, Xia J, Sun L, Zhang X, Clayton KL, Ye F, Weitz DA, Heyman JA (2020) Droplet encapsulation improves accuracy of immune cell cytokine capture assays. Lab Chip 20:1513–1520 Zhu C, Luo X, Espulgar WV, Koyama S, Kumanogoh A, Saito M, Takamatsu H, Tamiya E (2020) Real-time monitoring and detection of single-cell level cytokine secretion using LSPR technology. Micromachines (Basel) 11:107. https://doi.org/10.3390/mi11010107

Part IV Single Cell Omics

Single Cell Genomics

22

Yusuke Yamamoto, Anna Sanchez Calle, and Takahiro Ochiya

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods of Whole Genome Amplification for Single Cell Genomics . . . . . . . . . . . . . . . . . . . . . . . . . Single Cell Genomics Is Revolutionizing Cancer Biology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Somatic Mosaicisms in Development and Disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Clinical Applications of Single Cell Genomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Cell Genomics in Microbiology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Epigenetics Meets Single Cell Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

522 523 526 528 529 530 531 534 535

Abstract

Single cells are the minimum physiological and biological units of tissues and organs. It is critical to isolate individual cells from tissues and characterize them at distinct molecular levels in order to understand the status and fate of each individual cell, their role within the organism, and their cell-to-cell interactions. During the past decade, technical advances in whole genome amplification (WGA) have enabled massive DNA sequencing at the single cell level, and this approach has been extensively applied to study physiological conditions in normal tissues, disease, and development. Single cell genomics was intensively developed to Y. Yamamoto (*) · A. S. Calle Division of Molecular and Cellular Medicine, National Cancer Center Research Institute, Tokyo, Japan e-mail: [email protected] T. Ochiya Division of Molecular and Cellular Medicine, National Cancer Center Research Institute, Tokyo, Japan Institute of Medical Science, Tokyo Medical University, Tokyo, Japan © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_11

521

522

Y. Yamamoto et al.

dissect intra-tissue heterogeneity, which is a very powerful tool for uncovering genomic diversity, particularly in cancer. Likewise, epigenetic analysis has been carried out at a single cell level and has provided new insights into cellular variation and refined cell-cell interactions and provided a comprehensive epigenetic atlas of cellular states and lineages. In this chapter, technical improvements in DNA amplification methods and emerging high-dimensional bioinformatics tools will be broadly described with an eventual discussion of perspectives in single cell genomics.

Introduction Single cells are the minimum units of physiology and biology in tissues and organs. However, limitations in the sensitivity and accuracy of single cell approaches were key challenges for a long time. Most of our knowledge and understanding of the genome, transcriptome, and epigenome have been obtained from studies based on bulk tissues or populations of millions of cells. Although these studies are highly informative in both basic and clinical research, genomic information on heterogeneity in cell populations was frequently underestimated in these analyses (Macaulay and Voet 2014; Macaulay et al. 2017). Gene mutations that cause human disease were conventionally thought to be inherited from parent to child. However, it is currently well known that de novo mutations, which are not detectable in parents, may cause cancer and other diseases (Biesecker and Spinner 2013; De 2011). In multicellular organisms, the generation of cells with distinctive genotypes in an individual requires de novo mutations, which cause somatic mosaicism during early development and the aging process, and this kind of genomic alteration may contribute to the occurrence of sporadic diseases. In 1956, C. W. Cotterman mentioned the term somatic mosaicism for the first time in his seminal paper (Cotterman 1956). Subsequently, a considerable number of studies were carried out to examine somatic mosaicism in developmental biology and disease (Carlson and Southin 1963; McClellan and King 2010; Youssoufian and Pyeritz 2002). Although genomic variations were initially focused on the emergence of cancer and genetic disorders, unexpected levels of genomic variations were also found in normal tissues. Currently, it is well known that genetic mosaicism is apparent in various types of diseases, including neurodevelopmental diseases. In the past decade, technological advances in genomics using next-generation sequencing (NGS) have made revolutionary contributions to cancer biology, particularly by providing new knowledge about molecular aberrations, such as novel driver mutations and fusion genes. Genome sequencing data from a wide variety of cancer types have been comprehensively analyzed for the purpose of personalized medicine, which will help clinicians select more accurate and suitable treatments for each patient. Practical examples are the drug imatinib, which targets the BCR-ABL fusion genes in chronic myelogenous leukemia, and the breast cancer drug trastuzumab, which has a specific effect on tumors with a particular genetic profile called HER-2 positive (Dagogo-Jack and Shaw 2018). While DNA sequencing techniques have been extensively applied to discover somatic mutations that are

22

Single Cell Genomics

523

responsible for neurological and developmental diseases (Poduri et al. 2013), little is known about the intratumoral heterogeneity of genomic mutations and whether this heterogeneity is driven by the genetic background of each patient, patient lifestyle, or other extrinsic factors in the process of disease development. Therefore, obtaining defined parameters associated with pathophysiological roles during cancer development may only be possible with meticulous elucidation of genomic variation and epigenetics at a single cell level. In addition, single cell approaches may characterize rare cell types that are not observed within the bulk population and whose accessibility and identification are limited by their cell count. Improvements in high-throughput DNA sequencing technologies have enabled the translation of these tools into the single cell field. This chapter summarizes technical advances in DNA amplification using single cell DNA sequencing techniques with reduced bias that are applied in genomics and epigenetic studies and recent single cell genomic technologies that are used to investigate development and disease and provides significant insights into single cell genomics and epigenetics in basic science and their translation into cutting-edge clinical applications.

Methods of Whole Genome Amplification for Single Cell Genomics Genomic DNA amplification is an essential process to discover genetic variations, including single nucleotide variation (SNV), copy number variation (CNV), and de novo genomic DNA assembly, in single cell genomic experiments because a single normal human cell contains only 6 pg of genomic DNA, which is insufficient for NGS without amplification (Wang and Song 2017). Currently, the major whole genome amplification (WGA) strategies were basically established by degenerate oligonucleotide-primed (DOP)-PCR, multiple displacement amplification (MDA), or both in combination. DOP-PCR is a simple, rapid method that is widely used to amplify an entire genome (Telenius et al. 1992; Zhang et al. 1992). The reaction starts with an initial low-temperature annealing and extension step of partially degenerated oligonucleotides at many binding sites in the genomic DNA template and finishes with a high-temperature annealing step that provokes the priming of specific amplifications at multiple loci in the target DNA (Fig. 1a). DOP-PCR is a suitable method for the detection of CNV and aneuploidy at the single cell level; however, the drawback is that this method is not suitable for SNV detection due to its low coverage of amplified DNA regions in a single cell (Baslan and Hicks 2014). On the other hand, MDA is a non-PCR-based DNA amplification method wherein random hexamer primers are annealed to the DNA template and subsequent DNA synthesis is catalyzed by a high-fidelity enzyme, Φ29 DNA polymerase, at a constant temperature (Dean et al. 2002). In the MDA process, very small amounts of DNA can be amplified to a sufficient quantity for detection by single cell genomic analysis in a considerably short period of time (Fig. 1b). In addition, as a result of the high-fidelity enzymatic reaction, the MDA method produces much larger DNA fragments with reduced error probability (exon coverage >90%). In the comparison to the conventional PCR method, it facilitates an accurate identification of mutations at base-pair resolution.

524

Y. Yamamoto et al.

Fig. 1 Conventional WGA methods for single cell genomics. (a) DOP-PCR. (b) MDA. (These figures were modified from the ones by Wang J et al.)

Despite its improved amplification of DNA, it provides an uneven coverage, which impairs the analysis of CNVs (Yilmaz and Singh 2012). Therefore, the abovementioned approaches for the amplification of genomic DNA have disadvantages, and one should carefully select the most reliable procedure. Nonetheless, a major hurdle to overcome in single cell genomic technologies is bias because read counts are influenced by the GC content of DNA-amplified fragments; GC-rich fragments may be downsampled, affecting the final results. Hence, new strategies have been developed in order to assure uniformity across read coverage. Zong CS et al. reported multiple annealing and looping-based amplification cycles (MALBAC), which use quasilinear preamplification to reduce the genome bias caused by nonlinear amplification (Zong et al. 2012). In the initial process of amplification, genomic DNA isolated from a single human cell is mixed with a pool of random primers. Each primer has a common 27-nucleotide sequence with 8 variable nucleotides that can evenly hybridize to the templates. During the amplification process (extension at 65  C and melting at 94  C), amplicons of the

22

Single Cell Genomics

525

products (semiamplicons) become full amplicons with complementary sequences at both ends. When the reaction temperature decreases to 58  C, both ends with complementary sequences hybridize to allow the looping of full amplicons, which avoids further amplification and cross-hybridizations. After the preamplification step, the full amplicons are exponentially amplified by PCR to generate microgram levels of DNA for single cell genomic sequencing (Fig. 2a). Amplification strategies have been rapidly improved; however, the experimental procedures described above do not exclude random priming and low-fidelity amplification, which cause bias and errors. More recently, to mitigate amplification bias, the same research group reported a novel method called linear amplification via transposon insertion (LIANTI). This method uses Tn5 transposase to randomly fragment genomic DNA isolated from a single cell. For LIANTI transposition, Tn5 transposition is specifically designed to dimerize and includes a T7 promoter sequence (Chen et al. 2017a). When transposition occurs, the genomic DNA, fragmented and tagged with T7 promoter sequences, is linearly amplified into a large amount of genomic RNA copies, which are reverse transcribed and RNase treated, resulting in ready-to-use LIANTI amplicons for the preparation of DNA libraries suitable for single cell genome sequencing (Fig. 2b). Unlike conventional WGA methods for single cell approaches, LIANTI excludes non-specific priming and exponential amplification, significantly decreasing amplification bias and errors. Gole et al. established the microwell displacement amplification system (MIDAS), which was described as a massive parallel polymerase cloning method based on the MDA method; however, it uses microwells (hundreds to thousands of

Fig. 2 Recent advances in WGA methods for single cell approaches. (a) MALBAC. (b) LIANTI. (This figures were modified from the ones by Zong C et al. and Chen C et al., respectively)

526

Y. Yamamoto et al.

wells) to separate single cells spatially and randomly in a final 12-nL reaction (Gole et al. 2013). The authors claimed that by reducing the reaction volume ~1000-fold to the nanoliter level, the effective concentration of the template is increased, thereby leading to more favorable primer-annealing kinetics in the initial stages of the MDA process. Thus, MIDAS reduces amplification bias and improves the efficiency of single-copy number change detection to up to 1–2 Mb resolution. Although amplification protocols have been significantly improved to optimize the accuracy, uniformity, and coverage of genomic information, there are still inconveniences that remain elusive, such as disparities in the breadth of genomic coverage, amplification bias due to local differences in richness of guanine and cytosine bases (GC content), chimeric DNA or the prevalence of allelic dropout rates, and preferential allelic amplifications. Therefore, to minimize the loss of diversity and fidelity, it is crucial to select a proper method for each experimental setting in order to determine specific classes of genetic variation.

Single Cell Genomics Is Revolutionizing Cancer Biology Single cell genomic sequencing has great potential to reveal novel inherent properties of cancer cell biology for a better understanding of the complexity of intratumoral heterogeneity and to identify rare subpopulations. The application of single nucleus sequencing to study genomic CNVs in breast cancer tumor populations was first reported by Navin et al. In this study, which used flow-sorted single nuclei combined with WGA (DOP-PCR-based amplification) and NGS, high-resolution copy number profiles revealed that metastatic cells were almost identical to primary tumor cells, indicating that metastatic tumors developed from an advanced population of primary tumors (Navin et al. 2011). Single cell exome sequencing analyses were performed to identify SNVs in JAK2-negative myeloproliferative neoplasms and in clear cell renal cell carcinoma (ccRCC), which represents 90–95% of kidney neoplasms (Xu et al. 2012; Hou et al. 2012). Genomic DNA isolated from single cells was amplified by MDA, and whole-exome capture was conducted followed by massively parallel DNA sequencing. In sequencing analysis of JAK2-negative myeloproliferative neoplasms, 58 of 90 single cells were sequenced and passed the quality control check. The analysis identified novel mutations of SESN2 and NTRK1 that may be associated with neoplasia progression and found evidence of monoclonal evolution of the neoplasm. On the other hand, single cell exome sequencing analysis of a ccRCC tumor versus its adjacent kidney tissue revealed the genetic complexity of the patient’s tumors. In ccRCC, VHL and PBRM1 are thought to be commonly mutated; however, in this analysis, no mutations of VHL were detected, and mutations of PBRM1 were present at a very low frequency. Although the number of sequenced cells was not large in these two studies, these works paved the way for single cell genome or exome sequencing approaches in cancer biology. More recently, single cell exome sequencing was applied to analyze FACS-sorted cancer cells (Li et al. 2017). Since CD133 is thought to be a common cancer stem-like cell, CD133-positive and CD133-negative cell populations of ccRCC were separated, and subsequently, whole-exome sequencing was carried out at the single cell level to

22

Single Cell Genomics

527

identify the mutations responsible for cancer stem-like cell phenotypes. The analysis identified LOC440040 and LOC440563 mutations specific to the CD133-positive cell population of ccRCC, which were validated by CRISPR-Cas9 to promote cancer sphere formation. Nuc-seq was introduced as a whole genome and exome single cell sequencing approach in a study investigating the mutational evolution of ER+ and triple-negative breast cancer. It represented an optimization of single cell nuclei sequencing since at the time that the genomic DNA sample is collected, cells are found in G2/M phase, presenting four copies of the genome, which are easily amplified by MDA. Interestingly, MDA from G2/M nuclei, which theoretically contain double the amount of genomic DNA (approximately 12 pg in one cell), resulted in higher-coverage depth and lower false-positive error rates than G1/G0 nuclei. In addition, profiling of single cell copy numbers revealed that aneuploidy rearrangements in breast cancer genomes occurred at early stages of tumor evolution, whereas single nucleotide mutations gradually increased during tumor progression, causing genetic diversity (Wang et al. 2014). The same research group also reported a new highly multiplexed approach for single cell DNA sequencing, which included copy number profiling, bulk exome, and targeted deep sequencing, to address metastatic clonal evolution in human colorectal cancer (CRC). Single cell data from two metastatic CRC patients showed a gradual activation of mutation points from the colon epithelium to primary tumors and liver metastasis, implying a latedissemination model of metastasis (Leung et al. 2017). Moreover, WGA was applied to individual cells to later perform targeted resequencing analysis in acute lymphoblastic leukemia (ALL). Sequencing data from 1500 single cells from 6 ALL patients revealed the developmental history of each patient’s malignancy at single cell resolution (Gawad et al. 2014). Single cell genomics has also been functional in the examination of patientderived xenografts (PDX). In particular, PDX models, which are extensively established to study the tumor biology and drug response dynamics of genomic clones from original xenograft tumors and subsequent serial passaged tumors in highly immunodeficient mice (NOD SCID γ chain -/- (NSG) mice and NOD-Rag -/- γ chain -/- (NRG) mice), were investigated in breast cancer by massively parallel whole genome shotgun sequencing (WGSS) (Eirew et al. 2015). The results of 15 PDX cases revealed clonal selection on xenografted tumors that were found in both primary and metastatic breast cancer samples; however, the dynamics were not evident from histological and imaging analyses, which were thought to be basically stable and robust. Additionally, during the serial passaging process in mice, the clonal dynamics were clearly reproduced in independently transplanted samples into immunodeficient mice, indicating that the basis of selection was nonrandom. Therefore, the data noted the possibility that for the propagation of xenograft tumors, initial populations with a minor presence might selectively propagate in mice and dominate the xenografted tumor. Another relevant field in which the study of individual cells has contributed to new insights on tumor intricacy is circulating tumor cells (CTCs). Because of their easy accessibility in the bloodstream, CTCs represent ideal candidates for unmasking target biomarkers. Thus, genomic DNA sequencing of CTCs offers a new noninvasive modality of prognosis and diagnosis by liquid biopsy. Furthermore,

528

Y. Yamamoto et al.

because CTCs invade peripheral blood vessels from primary tumors and metastasize in distal organs, the characterizations extracted from sequencing data might lead to an understanding of not only the biological phenotype of CTCs but also their intrinsic heterogeneity. Since CTCs are a very rare cell population and present in a ratio of 1 cell per 109 of normal blood cells in metastatic cancer patients, there was a limited number of cells for analysis, so multiple methods of CTC isolation were established (Yu et al. 2011; Lin et al. 2017). The first comprehensive genomic profiling of CTCs from stage IV colorectal cancer patients was carried out by array-based comparative genome hybridization (aCGH) and NGS (Heitzer et al. 2013). In this analysis, cancer-associated copy number variation and typical driver gene mutations for colorectal cancer were detected in CTCs similar to primary and metastatic tumors. However, specific mutations were also found in only CTC fractions. Ni X et al. reported MALBACbased WGA for CTC sequencing analysis in lung adenocarcinoma patients (Ni et al. 2013). The authors found common CNV regions (gain in 16 CTCs and loss in 7 CTCs) linked to cancer progression, such as a gain region in chromosome 8q containing the c-MYC gene and in chromosome 5p containing the TERT gene. Of note, the SNV and INDEL (insertion/deletion) patterns in CTCs and metastatic tumors showed larger similarity to one another than to the primary tumor. In 2014, Lohr et al. demonstrated the feasibility of single cell exome sequencing of ctDNA in metastatic prostate cancer by developing a modular set of experimental and analytical protocols that conferred robust calling of somatic SNVs using a whole exome census-based sequencing strategy (Lohr et al. 2014). SNVs, together with somatic copy number alterations (CNAs) and structural variation (SV) from primary colon tumor cells and CTCs, were also studied to deduce an evolutionary history in the occurrence of cancer metastasis (Gao et al. 2017). The results indicated that SNVs in individual cells from both primary tumor cells and CTCs appeared sporadically, whereas large-scale CNAs in primary tumor cells were gradually accumulated and then converged toward the CNAs in CTCs. Based on the data from SV analysis, the authors proposed a two-step model to explain the multiregional copy number gain based on a first complex rearrangement followed by gene amplification, which led to dramatic phenotypic effects in CTCs. Alternatively, long fragment read (LFR) appeared as an accurate cost-effective technology that enables analysis of entire genomic content without extensive DNA amplification (Peters et al. 2012). LFR is a new sequencing method capable of performing comprehensive quantitative analysis of genomic variations from large structural changes to SNVs. In a study of CTCs from an ER+/HER2-metastatic breast cancer patient, LFR analysis was conducted to identify driver gene mutations, which might provide essential information for personalized cancer therapy (Gulbahce et al. 2017).

Somatic Mosaicisms in Development and Disease Significant advances in single cell genomics have contributed to research on other types of diseases and facilitated the study of cellular heterogeneity in development. Traditionally, inherited mutations, which are present in the parents and offspring,

22

Single Cell Genomics

529

were mainly thought to drive disease in development. Nonetheless, recent whole genome and exome studies have reported a few induced spontaneous mutations during cell replications. Thus, de novo mutations (so-called somatic mutations) are present in any of the cells of the body. Furthermore, accumulating evidence has shown that de novo mutations, which are only detected in the offspring and not in parents, definitely contribute to disease development, such as in autism and epilepsy (O’Roak et al. 2011; Neale et al. 2012; Poduri et al. 2013). De novo CNVs have been known to contribute to the cause of neuropsychiatric disease; therefore, CNVs of single neurons from hemimegalencephaly (HMG) patients were sequenced to further investigate the developmental process of disease at single cell resolution (Cai et al. 2014). Interestingly, single neuron CNV analysis has revealed that less than 20% of neurons harbor tetrasomy 1q, suggesting that CNVs detected in a minor population of neurons broadly affect whole brain dysfunction. LINE-1 (L1) retrotransposon is known to be frequently active in human neurons (Singer et al. 2010), and in another study, Evrony et al. carried out genomewide L1 insertion profiling of 300 single neurons from three normal individuals and single neuron genomic sequencing, which revealed that L1 was not a key contributor to neuronal diversity since 2 between IC and CL areas were considered meaningful in this study: 13 in neurons, 14 in the BBB, and 3 in both cell types. Twelve of these proteins were selected as candidates and analyzed by immunohistofluorescence in independent brains. The quantitative proteomes of neurons and the BBB (or proteotypes) after human brain ischemia presented here contribute to the knowledge regarding the molecular mechanisms of ischemic stroke pathology and highlight new proteins that might represent putative biomarkers of brain ischemia or therapeutic targets. Budnik and Slavov used Single-Cell ProtEomics by Mass Spectrometry (SCoPE-MS) to quantify over a thousand proteins in differentiating mouse

554

X. Xu and S. Hu

embryonic stem (ES) cells (Budnik et al. 2018). There are two major challenges developing a high-throughput method for single-cell proteomics by MS: (I) delivering the proteome of a mammalian cell to a MS instrument with minimal protein losses and (II) simultaneously identifying and quantifying peptides from single-cell samples. To overcome the first challenge, the researcher manually picked live single cells under a microscope and lysed them mechanically (by Covaris sonication in glass microtubes). This method was chosen to obviate chemicals that may undermine peptide separation and ionization or sample cleanup that may incur significant losses. The proteins from each cell lysate were quickly denatured at 90  C and digested with trypsin at 45  C overnight. Special care was taken to ensure that each tube contains only one cell. To overcome the second challenge, they made use of tandem mass tags, which allow multiplexing and simultaneous identification and quantification of proteins. On the other hand, optimization and modeling of quadrupole orbitrap parameters for sensitive analysis are needed to improve the sensitivity in single-cell proteomic analysis. Sun et al. investigated two factors critical to peptide sequencing and protein detection in shotgun proteomics, precursor ion isolation window (IW), and maximum precursor ion injection time (ITmax), on Q-Exactive HF mass spectrometer (Sun et al. 2017). Counterintuitive to the frequently used proteomic parameters for bulk samples (>100 ng), their experimental data and subsequent modeling suggested a universally optimal IW of 4.0 Th for sample quantity ranging from 100 ng to 1 ng and a sample quantity-dependent ITmax of more than 250 ms for 1-ng samples. Compared with the benchmark condition of IW = 2.0 Th and ITmax = 50 ms, this optimization generated up to 300% increase in the detected protein groups for 1-ng samples. The additionally identified proteins allowed deeper penetration of proteome for better understanding of crucial cellular functions such as signaling and cell adhesion. Table 1 below shows a partial list of published studies on single-cell MS-based proteomics.

Flow Cytometry Flow cytometry was invented in the late 1960s and then soon became the most established method for semiquantitative analysis of single-cell proteins. It is based on the fact that while the absolute amounts of proteins in a cell can be vanishingly small, the localized protein concentrations can be larger and measurable if the cells are kept intact. De Rosa, Perez, et al. demonstrated multiparameter analysis with a multicolor flow cytometer to measure 10–15 key proteins in signaling pathways simultaneously in single cells (De Rosa et al. 2001; Perez and Nolan 2002). The ability to perform correlated measurements of multiple proteins in single cells has turned cytometry into a useful tool to semiquantitatively analyze pathways underlying pathological conditions (Sachs et al. 2005; Irish et al. 2004). Tyramide signal amplification, which has long been used as a means to amplify nucleic acid detection in in situ hybridization protocols, has been added to traditional antibody staining techniques for analysis of low-abundance proteins (Clutter et al. 2010). While multiparameter cytometry allowed high-content screening (e.g., of multiple kinases) in cells, its throughput was still limited to be useful for drug screening. This

23

Single-Cell Proteomics

555

Table 1 A partial list of single-cell proteomics studies by MS Methodology Laser microdissection, LC-ESI MS/MS

LC-MS/MS

Intact cell MALDI-TOF MS on single oocytes and their surrounding cumulus cells, coupled to an optimized top-down high-resolution MS Laser capture microdissection, gel-liquid chromatography-tandem mass spectrometry (GeLC-MS/MS)

SP3 (sample preparation using magnetic beads), LC-MS/MS LC-ESI-MS/MS

Pick live single cells under a microscope and lysed them mechanically. TMT with LC-MS/MS Membrane-permeable activity-based probe (ABP), CE-LIF, and LC-MS/MS SP3 with LC-MS/MS

Analytes Ninety proteins were identified only in neurons, 260 proteins only in the BBB, and 261 proteins in both cell types 28 protein groups and 40 peptide groups 386 low molecular weight biomolecules (6,000 proteins, of the predicted 18,000 proteins in the Drosophila genome, from pooled embryos at 2–4 h (stages 5–7) and

Matrix Neurons and blood-brain barrier structures in postmortem brain slices from ischemic stroke patients

References GarciaBerrocoso et al. (2018)

Various quantities of Chinese hamster ovary cell protein digest Single bovine oocytes, cumulus cells, and granulosa cells

Sun et al. (2017) Labas et al. (2018)

Individual cell layers of tomato roots

Zhu et al. (2016)

Individual human oocytes

VirantKlun et al. (2016) Sun et al. (2016)

Frog (Xenopus laevis) blastomeres isolated from early stage embryos

Differentiating mouse embryonic stem (ES) cells

Budnik et al. (2018)

Individual HeLa cells

Chen et al. (2016)

1,000 HeLa cells and single Drosophila embryos

Hughes et al. (2014b)

(continued)

556

X. Xu and S. Hu

Table 1 (continued) Methodology

SDS-polyacrylamide gel electrophoresis with LC-MS/MS

Analytes 10–12 h (stages 13–15) of development >1,000 proteins

Matrix

References

Lysates from 50 rice egg cells, 3,000 rice sperm cells, callus (0.1 mg protein), seedlings (0.12 mg protein), and pollen grains (0.06 mg protein)

Abiko et al. (2013)

limitation was overcome partly by developing a barcoding method where differently treated cells are tagged with a combination of three dyes (Krutzik and Nolan 2006). Each dye, depending upon dilution, permits up to 7 intensity levels, and hence combination of the 3 dyes allows processing of up to 343 samples from one pool of cells. While flow cytometry has been most commonly used for analysis of kinases and phosphatases, it is also useful for other types of protein measurements such as glycosylation levels. Venable et al. used a panel of 14 lectins to characterize glycans present on cell surface as potential markers of pluripotency in human embryonic stem cells (Venable et al. 2005). A factor limiting the wider application of flow cytometry to glycosylated protein is that there are very few lectins available. Flow cytometry can also be used for analysis of secreted proteins such as cytokines. This requires treating cells with a vesicle formation inhibitor to trap synthesized cytokines in the Golgi, followed by fixation and permeabilization to stain the trapped cytokines with fluorescent antibodies for flow cytometric analysis (Karlsson et al. 2003). While conventional flow cytometry is useful in providing single-cell protein analysis, there are shortcomings to the method. Typically, sample preparation for conventional flow cytometry requires large numbers of cells (~106/sample), with correspondingly large fluorescent labeled antibody requirement, which can be prohibitively expensive when profiling many proteins. Furthermore, the sample preparation process involves numerous centrifugation and aspiration steps, leading to a significant sample loss. These limitations make conventional flow cytometry unsuitable for studying rare samples such as rare primary cell types and stem cells. Recent advancements in microfluidic engineering solved this problem partly. It enabled miniaturized, chip-based sample preparation and micro-flow cytometry (μFC), where small number of single cells (~100–1000) can be cultured, prepared, and analyzed in integrated microfluidic systems to provide multiparameter protein analysis at single-cell resolution (Srivastava et al. 2009; Wu et al. 2012).

Mass Cytometry Flow cytometry analysis of single cells can analyze as many as 15 proteins simultaneously as described earlier. However, system-level interrogation of biological

23

Single-Cell Proteomics

557

pathways requires the ability to do many more correlated measurements. A methodology called mass cytometry where the throughput of flow cytometry is combined with the ultra-high dimensionality and sensitivity of MS was established. In this approach, antibodies are conjugated with isotopically pure elements, and these antibodies are used to label cellular proteins. Cells are nebulized and sent through an Argon plasma, which ionizes the metal-conjugated antibodies. The metal signals are then analyzed by a time-of-flight MS. The approach overcomes limitations of spectral overlap in flow cytometry by utilizing discrete isotopes as a reporter system instead of traditional fluorophores which have broad emission spectra. Mass cytometry uniquely enables investigation of cell identity and behavior at the level of proteins and the properties (e.g., isoforms or posttranslational modifications) of proteins. Determination of cell identity is often accomplished by measuring levels of transmembrane proteins expressed on the cell surface (Ornatsky et al. 2008). When used as so-called markers, these proteins can reveal the lineage and maturation state of individual cells, as these proteins often have restricted patterns of expression across cell types. This technique should be capable of measuring even higher numbers of markers simultaneously considering that the precision of MS detection overcomes the issue of spectral overlap confounding fluorescence measurements (Wu and Singh 2012). This powerful technique was also used to simultaneously profile 34 parameters in single bone marrow cells, including binding of 31 antibodies, viability, DNA content, and relative cell size (Bendall et al. 2011). A study of B-cell development in the healthy human bone marrow identified coordination points of cell signaling, proliferation, and cell death in distinct stages of maturation. These coordination points were only discernible because of the high-dimensional data of mass cytometry. For instance, the unambiguous identification of a checkpoint between the pre-B I and pre-B II cell stages required simultaneous quantification of 19 cell-surface proteins, 2 intracellular enzymes (TdT and RAG1), a proliferation marker (Ki67), a phosphorylated cell signaling protein (p-STAT5), and a modified form of a protein involved in apoptosis (cleaved poly ADP-ribose polymerase). Recently, mass cytometry was used for the assay of predictive biomarkers (Krieg et al. 2018). As we know, immune-checkpoint blockade has revolutionized cancer therapy. In particular, inhibition of programmed cell death protein 1 (PD-1) has been found to be effective for the treatment of metastatic melanoma and other cancers. Despite a dramatic increase in progression-free survival, a large proportion of patients do not show durable responses. Therefore, predictive biomarkers of a clinical response are urgently needed. Krieg et al. used high-dimensional singlecell mass cytometry and a bioinformatics pipeline for the in-depth characterization of the immune cell subsets in the peripheral blood of patients with stage IV melanoma before and after 12 weeks of anti-PD-1 immunotherapy. Thirty leukocyte markers (most of them are membrane proteins) were assayed to identify all of the major immune cell populations and cover all stages of T-cell differentiation and activation. A significant difference in classical monocyte frequencies was observed. By using this fluorescence flow cytometry analysis, prediction of an anti-PD-1 response could be confirmed by the difference in monocyte frequency in responding versus nonresponding patients undergoing immunotherapy. Using artificial intelligence and

558

X. Xu and S. Hu

bioinformatics, researchers can create a two-dimensional mapping that allows to read test results, creating an Instagram of millions of blood cells. It is important to improve detection sensitivities of mass cytometer instruments. While a profusion of cellular processes can be simultaneously investigated using mass cytometry, this platform does have limitations to consider when designing an experiment. Because cells are atomized and ionized, it remains infeasible to recover living cells after analysis (when compared to flow cytometry). Moreover, due to the dynamics of ion flight in the mass spectrometer, the throughput of mass cytometry lags behind that of fluorescence-based instruments. Additionally, the sensitivity of mass reporters falls shy of few, more quantum-efficient fluorophores (such as phycoerythrin), making it more difficult to measure molecular features that are expressed at very low levels using mass cytometry. That said, the sensitivity range of ions across the mass range is three- to fourfold in difference, whereas fluorophores must contend with a vast range (50-fold) encumbered by serious issues in spectral output (which can be partly remedied by fluorescence compensation). As with all new technologies, standards are being established for the comparison of data across laboratories and instruments. A study compared the ionization efficiency across CyTOF2 mass cytometers, demonstrating that quantitative comparisons between instruments are significantly improved by normalization (Tricot et al. 2015). Because each instrument had a characteristic efficiency profile, comparison of data from the same instrument can be accomplished by incorporating polystyrene beads for data normalization, as described by Finck et al. in more detail (Finck et al. 2013).

Challenges, Limitations, and Outlook The future of single-cell proteomics will depend on several factors. First of all, the major technical bottleneck is the low multiplexing capacity of current single-cell proteomic methods. The limit arises because of the reliance on antibody-based detection schemes. It poses biased analyses, as extensive prior knowledge is normally required for selecting protein panel that is most relevant to the problem under study and thus precludes the utility in discovery level studies where prior knowledge is limited. This challenge might be overcome through the development of highly sensitive MS-based tools at single-cell resolution. Targeted proteomics using MS has already evolved to the extent of analyzing single mammalian cell (Virant-Klun et al. 2016). As a complementary means, a larger repertoire of high-affinity probes (other than monoclonal antibodies) may be needed to detect low-abundance proteins, posttranslational modifications, and protein interactions. Some promising candidates include aptamers, single-chain variable fragments, biorthogonal reporters, etc. Second, with the increase of multiplexing capacity and assay throughput, singlecell proteomics generates high-dimensional single-cell data sets that require new analytical strategies and computational tools for gleaning useful biological insights from these data. For example, protein fluctuations and the protein-protein correlations, uniquely resolved by single-cell measurements, allow extraction of key information of signaling networks when coupled with appropriate analytical

23

Single-Cell Proteomics

559

approaches. Single-cell analysis techniques resolve the cellular heterogeneity and clarify the mapping between signaling states and the phenotypes, enabling predictions of the change of signaling states upon environmental and drug perturbations with the help of theoretical advance on understanding single-cell proteomics data. Third, matching cellular heterogeneity with biological context remains a grand challenge for most single-cell proteomic tools as the assays typically involve removal of the cells from their native context by dissociating the tissue samples into single-cell suspension before analysis. Thus, while cellular heterogeneity is resolved, the context of that heterogeneity is lost. In order for new single-cell analytical technologies to become widely used, we also need to develop informatics and modeling tools that are customized for single cell proteomics and can integrate the protein measurements with other types of single-cell measurements such as genomics, transcriptomics, and metabolomics. It is certain, however, as the single-cell proteomics tools continue to improve in multiplexing capacity, throughput, sensitivity, and quantification, an overarching analytical framework that connects biological questions, experimental designs, to data analysis will eventually transform the practice of single-cell proteomics as well as our understanding in single-cell biology.

References Abiko M, Furuta K, Yamauchi Y, Fujita C, Taoka M, Isobe T, Okamoto T (2013) Identification of proteins enriched in rice egg or sperm cells by single-cell proteomics. PLoS One 8:e69578 Arkhipov SN, Berezovski M, Jitkova J, Krylov SN (2005) Chemical cytometry for monitoring metabolism of a Ras-mimicking substrate in single cells. Cytometry A 63A:41–47 Begley CG, Ellis LM (2012) Drug development: raise standards for preclinical cancer research. Nature 483:531–533 Bendall SC et al (2011) Single-cell mass cytometry of differential immune and drug responses across a human hematopoietic continuum. Science 332:687–696 Boardman AK, McQuaide SC, Zhu C, Whitmore CD, Lidstrom ME, Dovichi NJ (2008) Interface of an array of five capillaries with an array of one-nanoliter wells for high-resolution electrophoretic analysis as an approach to high-throughput chemical cytometry. Anal Chem 80:7631–7634 Boardman A, Chang T, Folch A, Dovichi NJ (2010) Indium-tin oxide coated microfabricated device for the injection of a single cell into a fused silica capillary for chemical cytometry. Anal Chem 82:9959–9961 Brayboy LM, Wessel GM (2016) The double-edged sword of the mammalian oocyte – advantages, drawbacks and approaches for basic and clinical analysis at the single cell level. Mol Hum Reprod 22:200–207 Budnik B, Levy E, Harmange G, Slavov N (2018) Mass-spectrometry of single mammalian cells quantifies proteome heterogeneity during cell differentiation. Genome Biology 19:161 Chen D et al (2016) Single cell chemical proteomics with membrane-permeable activity-based probe for identification of functional proteins in lysosome of tumors. Anal Chem 88:2466–2471 Clutter MR, Heffner GC, Krutzik PO, Sachen KL, Nolan GP (2010) Tyramide signal amplification for analysis of kinase activity by intracellular flow cytometry. Cytometry A 77:1020–1031 Cohen D, Dickerson JA, Whitmore CD, Turner EH, Palcic MM, Hindsgaul O, Dovichi NJ (2008) Chemical cytometry: fluorescence-based single-cell analysis. Annu Rev Anal Chem 1:165–190 Copley MR, Eaves CJ (2013) Developmental changes in hematopoietic stem cell properties. Exp Mol Med 45:e55

560

X. Xu and S. Hu

Darmanis S et al (2016) Simultaneous multiplexed measurement of RNA and proteins in single cells. Cell Rep 14:380–389 De Rosa SC, Herzenberg LA, Herzenberg LA, Roederer M (2001) 11-color, 13-parameter flow cytometry: identification of human naive T cells by phenotype, function, and T-cell receptor diversity. Nat Med 7:245–248 Di Carlo D, Aghdam N, Lee LP (2006) Single-cell enzyme concentrations, kinetics, and inhibition analysis using high-density hydrodynamic cell isolation arrays. Anal Chem 78:4925–4930 Finck R et al (2013) Normalization of mass cytometry data with bead standards. Cytometry A 83:483–494 Garcia-Berrocoso T et al (2018) Single cell immuno-laser microdissection coupled to label-free proteomics to reveal the proteotypes of human brain cells after ischemia. Mol Cell Proteomics 17:175–189 Hu S, Zhang ZR, Cook LM, Carpenter EJ, Dovichi NJ (2000) Separation of proteins by sodium dodecyl sulfate capillary electrophoresis in hydroxypropyl cellulose sieving matrix with laserinduced fluorescence detection. J Chromatogr A 894:291–296 Hu S, Jiang J, Cook LM, Richards DP, Horlick L, Wong B, Dovichi NJ (2002a) Capillary sodium dodecyl sulfate-DALT electrophoresis with laser-induced fluorescence detection for size-based analysis of proteins in human colon cancer cells. Electrophoresis 23:3136–3142 Hu S, Ye YL, Surh G, Clark JI, Dovichi NJ (2002b) Analysis of proteins by capillary SDS-DALT electrophoresis with laser-induced fluorescence detection. LC-GC Eur 15:166 Hu K, Zarrine-Afsar A, Ahmadzadeh H, Krylov SN (2004a) Single-cell analysis by chemical cytometry combined with fluorescence microscopy. Instrum Sci Technol 32:31–41 Hu S, Michels DA, Fazal MA, Ratisoontorn C, Cunningham ML, Dovichi NJ (2004b) Capillary sieving electrophoresis/micellar electrokinetic capillary chromatography for two-dimensional protein fingerprinting of single mammalian cells. Anal Chem 76:4044–4049 Huebner A, Srisa-Art M, Holt D, Abell C, Hollfelder F, Demello AJ, Edel JB (2007) Quantitative detection of protein expression in single cells using droplet microfluidics. Chem Commun 12:1218–1220 Hughes AJ, Spelke DP, Xu ZC, Kang CC, Schaffer DV, Herr AE (2014a) Single-cell western blotting. Nat Methods 11:749–U794 Hughes CS, Foehr S, Garfield DA, Furlong EE, Steinmetz LM, Krijgsveld J (2014b) Ultrasensitive proteome analysis using paramagnetic bead technology. Mol Syst Biol 10:757 Irish JM, Hovland R, Krutzik PO, Perez OD, Bruserud O, Gjertsen BT, Nolan GP (2004) Single cell profiling of potentiated phospho-protein networks in cancer cells. Cell 118:217–228 Karlsson AC et al (2003) Comparison of the ELISPOT and cytokine flow cytometry assays for the enumeration of antigen-specific T cells. J Immunol Methods 283:141–153 Krieg C et al (2018) High-dimensional single-cell analysis predicts response to anti-PD-1 immunotherapy. Nat Med 24:144–153 Krutzik PO, Nolan GP (2006) Fluorescent cell barcoding in flow cytometry allows high-throughput drug screening and signaling profiling. Nat Methods 3:361–368 Krylov SN, Starke DA, Arriaga EA, Zhang Z, Chan NW, Palcic MM, Dovichi NJ (2000) Instrumentation for chemical cytometry. Anal Chem 72(4):872–877 Labas V, Teixeira-Gomes AP, Bouguereau L, Gargaros A, Spina L, Marestaing A, Uzbekova S (2018) Intact cell MALDI-TOF mass spectrometry on single bovine oocyte and follicular cells combined with top-down proteomics: a novel approach to characterise markers of oocyte maturation. J Proteomics 175:56–74 Li SY et al (2015) An integrated platform for isolation, processing, and mass spectrometry-based proteomic profiling of rare cells in whole blood. Mol Cell Proteomics 14:1672–1683 Li QL, Chen PL, Fan YY, Wang X, Xu KH, Li L, Tang B (2016) Multicolor fluorescence detectionbased microfluidic device for single-cell metabolomics: simultaneous quantitation of multiple small molecules in primary liver cells. Anal Chem 88:8610–8616 Lillard SJ, Yeung ES (1996) Analysis of single erythrocytes by injection-based capillary isoelectric focusing with laser-induced native fluorescence detection. J Chromatogr B Biomed Appl 687: 363–369

23

Single-Cell Proteomics

561

Lillard SJ, Yeung ES, Lautamo RM, Mao DT (1995) Separation of hemoglobin variants in single human erythrocytes by capillary electrophoresis with laser-induced native fluorescence detection. J Chromatogr A 718(2):397–404 Lombard-Banek C, Moody SA, Nemes P (2016a) Single-cell mass spectrometry for discovery proteomics: quantifying translational cell heterogeneity in the 16-cell frog (Xenopus) embryo. Angew Chem Int Ed Engl 55:2454–2458 Lombard-Banek C, Reddy S, Moody SA, Nemes P (2016b) Label-free quantification of proteins in single embryonic cells with neural fate in the cleavage-stage frog (Xenopus laevis) embryo using capillary electrophoresis electrospray ionization high-resolution mass spectrometry (CE-ESI-HRMS). Mol Cell Proteomics 15:2756–2768 Love JC, Ronan JL, Grotenbreg GM, van der Veen AG, Ploegh HL (2006) A microengraving method for rapid selection of single cells producing antigen-specific antibodies. Nat Biotechnol 24:703–707 Lucy CA, MacDonald AM, Gulcev MD (2008) Non-covalent capillary coatings for protein separations in capillary electrophoresis. J Chromatogr A 1184:81–105 Ma C et al (2011) A clinical microchip for evaluation of single immune cells reveals high functional heterogeneity in phenotypically similar T cells. Nat Med 17:738–743 Mainz ER, Wang QZ, Lawrence DS, Allbritton NL (2016) An integrated chemical cytometry method: shining a light on Akt activity in single cells. Angew Chem Int Ed Engl 55: 13095–13098 Michels DA, Hu S, Dambrowitz KA, Eggertson MJ, Lauterbach K, Dovichi NJ (2004) Capillary sieving electrophoresis-micellar electrokinetic chromatography fully automated twodimensional capillary electrophoresis analysis of Deinococcus radiodurans protein homogenate. Electrophoresis 25:3098–3105 Ornatsky OI et al (2008) Study of cell antigens and intracellular DNA by identification of elementcontaining labels and metallointercalators using inductively coupled plasma mass spectrometry. Anal Chem 80:2539–2547 Perez OD, Nolan GP (2002) Simultaneous measurement of multiple active kinase states using polychromatic flow cytometry. Nat Biotechnol 20:155–162 Ramos-Payan M, Ocana-Gonzalez JA, Fernandez-Torres RM, Llobera A, Bello-Lopez MA (2018) Recent trends in capillary electrophoresis for complex samples analysis: a review. Electrophoresis 39:111–125 Sachs K, Perez O, Pe’er D, Lauffenburger DA, Nolan GP (2005) Causal protein-signaling networks derived from multiparameter single-cell data. Science 308:523–529 Shin YS et al (2010) Chemistries for patterning robust DNA microbarcodes enable multiplex assays of cytoplasm proteins from single cancer cells. Chemphyschem 11:3063–3069 Sinkala E et al (2017) Profiling protein expression in circulating tumour cells using microfluidic western blotting. Nat Commun 8:14622 Sobhani K, Fink SL, Cookson BT, Dovichi NJ (2007) Repeatability of chemical cytometry: 2-DE analysis of single RAW 264.7 macrophage cells. Electrophoresis 28:2308–2313 Srivastava N, Brennan JS, Renzi RF, Wu M, Branda SS, Singh AK, Herr AE (2009) Fully integrated microfluidic platform enabling automated phosphoprofiling of macrophage response. Anal Chem 81:3261–3269 Su Y, Shi Q, Wei W (2017a) Single cell proteomics in biomedicine: high-dimensional data acquisition, visualization, and analysis. Proteomics 17(3–4). https://doi.org/10.1002/ pmic.201600267 Su YP et al (2017b) Single-cell analysis resolves the cell state transition and signaling dynamics associated with melanoma drug-induced resistance. Proc Natl Acad Sci USA 114:13679–13684 Sun L et al (2014) Over 10,000 peptide identifications from the HeLa proteome by using single-shot capillary zone electrophoresis combined with tandem mass spectrometry. Angew Chem Int Ed Engl 53:13931–13933 Sun L, Dubiak KM, Peuchen EH, Zhang Z, Zhu G, Huber PW, Dovichi NJ (2016) Single cell proteomics using frog (Xenopus laevis) blastomeres isolated from early stage embryos, which form a geometric progression in protein content. Anal Chem 88:6653–6657

562

X. Xu and S. Hu

Sun B, Kovatch JR, Badiong A, Merbouh N (2017) Optimization and modeling of quadrupole orbitrap parameters for sensitive analysis toward single-cell proteomics. J Proteome Res 16: 3711–3721 Thorsen T, Maerkl SJ, Quake SR (2002) Microfluidic large-scale integration. Science 298:580–584 Torres AJ, Contento RL, Gordo S, Wucherpfennig KW, Love JC (2013) Functional single-cell analysis of T-cell activation by supported lipid bilayer-tethered ligands on arrays of nanowells. Lab Chip 13:90–99 Tricot S et al (2015) Evaluating the efficiency of isotope transmission for improved panel design and a comparison of the detection sensitivities of mass cytometer instruments. Cytometry A 87:357–368 Venable A, Mitalipova M, Lyons I, Jones K, Shin S, Pierce M, Stice S (2005) Lectin binding profiles of SSEA-4 enriched, pluripotent human embryonic stem cell surfaces. BMC Dev Biol 5:15 Vickerman BM, Anttila MM, Petersen BV, Allbritton NL, Lawrence DS (2018) Design and application of sensors for chemical cytometry. ACS Chem Biol 13(7):1741–1751 Virant-Klun I, Leicht S, Hughes C, Krijgsveld J (2016) Identification of maturation-specific proteins by single-cell proteomics of human oocytes. Mol Cell Proteomics 15:2616–2627 Wei W et al (2016) Single-cell phosphoproteomics resolves adaptive signaling dynamics and informs targeted combination therapy in glioblastoma. Cancer Cell 29:563–573 Wu M, Singh AK (2012) Single-cell protein analysis. Curr Opin Biotechnol 23:83–88 Wu M et al (2012) Microfluidically-unified cell culture, sample preparation, imaging and flow cytometry for measurement of cell signaling pathways with single cell resolution. Lab Chip 12:2823–2831 Xu F et al (2014) Single-cell chemical proteomics with an activity-based probe: identification of low-copy membrane proteins on primary neurons. Angew Chem Int Ed Engl 53:6730–6733 Yang MA, Nelson R, Ros A (2016) Toward analysis of proteins in single cells: a quantitative approach employing isobaric tags with MALDI mass spectrometry realized with a microfluidic platform. Anal Chem 88:6672–6679 Yeung ES (2011) Genome-wide correlation between mRNA and protein in a single cell. Angew Chem Int Ed Engl 50:583–585 Zhong X, Zhang Z, Jiang S, Li L (2014) Recent advances in coupling capillary electrophoresisbased separation techniques to ESI and MALDI-MS. Electrophoresis 35:1214–1225 Zhu Y, Li H, Bhatti S, Zhou S, Yang Y, Fish T, Thannhauser TW (2016) Development of a laser capture microscope-based single-cell-type proteomics tool for studying proteomes of individual cell layers of plant roots. Hortic Res 3:16026

Single Cell Pull-Down for Characterization of Protein Complexes

24

Michael Philippi, Zehao Li, Maniraj Bhagawati, and Changjiang You

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluorescence Methods for Characterization of Protein Complex . . . . . . . . . . . . . . . . . . . . . . . . . . . High-Throughput Screening of Protein-Protein Interactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Molecule Fluorescence Detection of Protein Complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Cell Pull-Down . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Surface Chemistry for Single Cell Pull-Down . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Live Cell Micropatterning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . High Affinity Capturing of Target Protein Complexes from a Single Cell . . . . . . . . . . . . . . . . Dissociation Kinetics of Protein Complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Determination of the Stoichiometry of Protein Complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Automated Large-Scale Single Molecule Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Cell Pull-Down for Label Free Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

564 565 567 567 569 570 571 572 572 574 575 577 579 580

M. Philippi Department of Biology/Chemistry, University of Osnabrück, Osnabrück, Germany Center for Cellular Nanoanalytics (CellNanOs), University of Osnabrück, Osnabrück, Germany Z. Li College of Life Sciences, Beijing University of Chemical Technology, Beijing, China M. Bhagawati (*) Institute of Molecular Cell Biology, University of Münster, Münster, Germany C. You (*) Department of Biology/Chemistry, University of Osnabrück, Osnabrück, Germany Center for Cellular Nanoanalytics (CellNanOs), University of Osnabrück, Osnabrück, Germany College of Life Sciences, Beijing University of Chemical Technology, Beijing, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_43

563

564

M. Philippi et al.

Abstract

Quantitative analysis of protein complexes in their native cellular environment provides unbiased information for mechanistic understanding of various cellular processes. Particularly for protein complexes formed by transient protein-protein interactions, single cell analysis emerges as a powerful method for revealing the heterogeneity that plays important roles in spatiotemporal regulation of biological functions. In this chapter, single cell pull-down (SiCPull) is introduced for characterizing protein complexes formed in live cells at a single molecule level. This chapter starts with a brief introduction on fluorescence-based methods used for detecting protein complexes. Multidisciplinary techniques required for establishing SiCPull are elaborated with respect to surface biofunctionalization for protein capturing, micropatterning, and single molecule fluorescence analysis. Applications of SiCPull for determining the stability and stoichiometry of protein complexes are demonstrated for proteins involved in type I interferon signaling. Further applications of SiCPull are highlighted for label-free protein characterization in visible to infrared (IR) and terahertz (THz) spectral regions. At the end of the chapter, future developments toward high throughput SiCPull are discussed. Keywords

Protein complex · Single cell pull-down · Single molecule fluorescence · Labelfree detection

Introduction Numerous cellular processes such as energy conversion, biomolecular transport and signal transduction rely on appropriate functioning of proteins. In living organisms, protein–protein interactions (PPI) orchestrate biological processes in a spatiotemporally regulated manner and are therefore inextricably linked to the development of diseases. To gain a mechanistic understanding of these interactions, techniques to arrest transient product protein complexes and purify them from their native complex biological context for detailed characterization have been established (Fernandez-Leiro and Scheres 2016; Kosol et al. 2013). Biochemical methods of pull-down and co-immunoprecipitation have been very successful in identifying protein complexes by combining with protein isolation and purification. These approaches require extraction of the target protein complex from cell lysates, followed by electrophoresis and antibody staining (e.g., Western blotting). However, the relatively harsh experimental conditions and the prolonged delay that occurs between cell lysis and electrophoresis due to the multiple intervening steps often damage the integrity of protein complexes. Therefore, such biochemical methods are useful for characterization of stable protein complexes (dissociation equilibrium constant, KD of less than tens of μM).

24

Single Cell Pull-Down for Characterization of Protein Complexes

565

Fluorescence Methods for Characterization of Protein Complex Fluorescence methods are very amenable to characterize protein complexes, even for those formed by transient interactions (e.g., KD > 100 μM and dissociation lifetime less than second). Various fluorescence spectroscopy and microscopy methods have been developed for exploring PPIs and the product protein complexes. A prominently implemented principle in these methods is based on Förster Resonance Energy Transfer (FRET). FRET requires labeling of the target proteins with fluorophores having a spectral overlap, i.e., the emission spectrum of one fluorophore (donor, D) has to overlap with the absorption spectrum of another (acceptor, A). Energy transfer in FRET occurs via radiationless dipole-dipole coupling between the donor and acceptor, which makes it sensitive to a donor-acceptor distance of up to 10 nm (Fig. 1a). High FRET efficiency (>50%) indicates that the distance between donor and acceptor labeled proteins is smaller than the Förster radius (~ few nm), thus corroborating the formation of protein complexes (Fig. 1b). FRET efficiency can be quantified by intensity changes, either by the decrease of donor intensity or increase of acceptor intensity (Zeug et al. 2012). Alternatively, measuring the relative changes of the donor’s fluorescence lifetime provides a robust

Fig. 1 Characterizing protein complex by fluorescence methods. (a) Scheme of using FRET to determine the distance between proteins. Typical Förster radium (R0) is ~5 nm. Significant increase of acceptor fluorescence is detected at d < R0. (b) Interaction of protein and formation of protein complex can be monitored by FRET spectroscopy and microscopy. (c) Engineered self-enzymatic protein tag “HaloTag” and its labeling via Tetramethyrhodamine-HaloTag Ligand (TMR-HTL) as the substrate. (d) GFP and antiGFP nanobody for high affinity labeling to the GFP-fused proteins. (e) Traceless labeling of protein of interest (POI) on the N- and C-terminus using split inteins through protein trans-splicing (PTS) reaction. POI fused to either the N- or C- terminal intein fragment (IntN and IntC, respectively) can be expressed either on the cell membrane or in the cytosol

566

M. Philippi et al.

way to quantify FRET efficiency. Time correlated single photon counting (TCSPC) is commonly used for lifetime determination. Technical advances in TCSPC have also been leveraged to establish techniques such as fluorescence (cross) correlation spectroscopy (Medina and Schwille 2002) and photon counting histogram analysis (Chen et al. 1999). The two methods have been successfully applied for characterizing protein complexes by monitoring the fluorescence fluctuations or molecular brightness, respectively. Advantages of fluorescence methods in sensitivity, multiplexing, and nondestructive detection motivate broad applications in live cells, for which labeling protein in live cells is a fundamental request. Genetically encoded fluorescent proteins covering the full visible spectrum are available (Marx 2017; Shaner et al. 2013). Selfenzymatic protein tags (Keppler et al. 2004; Los et al. 2008), including SNAP-tag ®, CLIP-tag ® (New England Biolabs) and HaloTag ® (Promega) (Fig. 1c), have been engineered for bio-orthogonal labeling. Recently, single chain antibodies from llamas, i.e., nanobodies (NB) (Fig. 1d), have emerged as an ideal labeling option (Helma et al. 2015). The small size of NBs (~15 kDa, 2 nm) and the sub-nanomolar affinity to the target proteins make NBs a superior tool for protein labeling when compared to conventional antibodies. A number of split inteins have also been leveraged for traceless covalent labeling of target proteins in live cells through the protein trans-splicing reaction (Fig. 1e) (Bhagawati et al. 2019, 2020). Compared to the use of large enzymatic tags such as the SNAP or HaloTag, split inteins result in minimal change in the sequence of the protein of interest. This negligible steric impact on protein function, coupled with a high bio-orthogonality and suitable kinetic properties, make split-inteins a valuable tool for site-specific modification of proteins with fluorophores. In addition to conventional organic fluorophore, photostable quantum dots (Medintz et al. 2003; Clarke et al. 2010), and lanthanide compounds or nanoparticles (Drees et al. 2016) have also been applied for labeling target proteins. Owing to the relatively long lifetime of lanthanides (~ ms), lanthanides-based Luminescence Resonance Energy Transfer (LRET) is used as a basis for time-resolved detections to separate specific FRET signals in cellular environment (Albizu et al. 2010). With the advances in labeling technology, various fluorescence microscopy methods have been developed to visualize the spatiotemporal heterogeneity of protein complexes in live cells. By positioning the focus in a confocal laser scanning microscope (cLSM) either inside a cell or at the cell surface, fluorescence correlation spectroscopy can be used to investigate the diffusion and concentration of protein complex at different loci to quantify the dynamics of PPIs (Weidemann et al. 2014; Gandhi et al. 2014). Conducting TCSPC at each pixel during cLSM imaging leads to fluorescence lifetime imaging (FLIM). The lifetime information with sub-μm resolution is then used for mapping the FRET efficiency in the whole cell (Baumdick et al. 2018). For cells with a typical dimension of tens of μm, FLIM is able to quantify the spatiotemporal heterogeneity of protein complexes in different cellular compartments. Bimolecular fluorescence complementation (BiFC), which is based on the complementation of fragments of fluorescent proteins upon interaction between fused target proteins, has found extensive use for the validation of PPIs. A well-established BiFC analysis is based on split green fluorescent protein (GFP)

24

Single Cell Pull-Down for Characterization of Protein Complexes

567

that has its β-barrier separated into two fragments (Leonetti et al. 2016). Complementation fluorescence assays have been successfully used for analysis of tau protein aggregation (Chun et al. 2011), G Protein-Coupled Receptor dimerization (Urizar et al. 2011), and cytokine signaling in live cells (Cassonnet et al. 2011).

High-Throughput Screening of Protein-Protein Interactions Proteomic research and pharmaceutical industry for drug discovery require largescale screening of PPIs and the product protein complexes (MacBeath and Schreiber 2000). Protein microarrays have been implemented broadly for high-throughput screening (HTS) of protein complexes in vitro or in cellulo (Zhu et al. 2001). Commercially available microarrays (biochips) now provide the opportunity for genome wide screening of PPIs (e.g., HuProt™ microarray, CDI Laboratories). In protein microarrays, proteins are immobilized on a substrate in a spatially resolved manner (MacBeath and Schreiber 2000; Phizicky et al. 2003). The metastable nature of three-dimensional protein structures, including those of the pharmaceutically important membrane proteins (Overington et al. 2006), as well as the diversity of substrate materials that are applied demand dedicated solutions for engineering protein-compatible surface architectures. Multifarious surface chemistries and microfabrication methods have been developed to ensure efficient screening with high reproducibility. As this is not the focus of the chapter, readers are encouraged to find the insights from the cited references (Kam et al. 2013; You and Piehler 2016). Due to excellent compatibility with optical detection techniques including highly sensitive fluorescence imaging, glass is widely used as the substrate in these assays. Detecting the changes of fluorescence properties (e.g., intensity, lifetime, polarization) on glass support provides reliable readout of protein complex formation. Using total internal reflection fluorescence (TIRF) excitation, only a thin layer of protein sample at the glass surface is detected via evanescent field excitation (10,000), causing turbulent mixing (Bleul et al. 2011) and cell membrane damage. Other methods such as fluorescent resonance energy transfer (Lakowicz 2006; Sourjik et al. 2007; Spiller et al. 2010) and fluorescence lifetime imaging (Suhling et al. 2005) require bulky labels and greatly exceed the scale of submolecular reaction sites. Moehren and Kholodenko studied early and rapid phosphorylation of specific receptors and their targets (Moehren et al. 2001). The results indicated that maximum EGFR autophosphorylation was achieved in 5 s, and downstream tyrosine phosphorylation reactions (e.g., PI3K and PLC-γ) peaked later at 10 s . Moehren’s results (in his paper; Fig. 2a) showed the progression of tyrosine phosphorylation with a time resolution of 1 s. Time course of EGF-induced tyrosine phosphorylation at 37
C and the maximum EGFR phosphorylation was achieved in 5–10 s. It could identify the specific steps in the downstream signaling pathways (i.e., conformational and autocatalytic modifications of the receptor after initial receptor signal transduction). Despite the significant convection component, bulk volume mixing involves homogenization time scales of the order of 1 s, such that assay synchronization is poor and incubation periods can be no shorter than a few seconds. For instance, a simulation of the methodology described by Moehren et al. is demonstrated in Fig. 1.

654

Y.-Y. Chiang

Fig. 1 Macroscopic mixing using a magnetically actuated stirring element (Chiang and West 2013). Simulation of the methodology described by Moehren (Moehren et al. 2001). Reagent homogenization required a few seconds

Fig. 2 The convergence of cell and ligand streams leads to the formation of a virtual interface with membrane-hosted ligand–receptor interactions delayed by diffusion time scales (a). Rapid stimulation and reaction arrest are essential for the ideal analysis system (b). The process is repeated for reaction quenching, such that those few cells that are stimulated are the farthest from the quench buffer flows, resulting in extremely lengthy times for reaction arrest (c). Cell stream width (1 mm) is equivalent to the diameter of ~70 cells (15 μm)

A sample (10 μL) of red dye, which mimicked a ligand, was added in 1.25 mL of water, and subsequently, 3 volumes of water were added as a quencher (i.e., lysis buffer). Reagent homogenization required ~1 s and even longer when the volume was increased (~3 s). In 2007, Dengjel et al. replaced the traditional macroscale mixing method and showed activation of EGFR phosphopeptides by pumping cells and reagents in a

28

Record the Single Cell Signal Pathway

655

quench flow system (Dengjel 2007). The results suggested that the autophosphorylation of three EGFR tyrosine residues required 10 s or more, which is significantly slower than the results reported by Moehren. However, microscale flows in tubing are characteristically laminar, and mixing is diffusion-limited. Consequently, receptor stimulation by combining rapidly flowing streams confines reactions to the interface between lamina, such that no synchronization and only a small subfraction of cells are (partially) stimulated. Membrane-hosted ligand–receptor interactions are therefore delayed by diffusion. As shown in Fig. 2. Similarly, the quencher reached cells by diffusion only. A longer diffusion distance further delays the reaction arrest (Fig. 2a). For some cells, the signal transduction process is unlikely to be arrested. The ideal system should therefore enable instantaneous ligand–receptor contact and defined reaction incubation of the order of milliseconds followed by instantaneous reaction quenching (Fig. 2c).

Challenges of Rapid Mixing Currently, working in miniature dimensions and understanding miniature-scale processes are no longer challenges. Microfluidics is a multidisciplinary field in which the following are studied and provides a powerful strategy that can be used by researchers to study the behavior of fluids, the control of small volumes of fluids, and design of systems. However, in microfluidics, mixing of different liquids is challenging. Mixing is the combination and homogenization of different liquids, and it is used to initiate (bio)chemical reactions in synthetic and analytical chemistry processes. Mixing is highly dependent on the scale/dimensions of the system. In contrast to macroscale flow, microscale flow in occurs microfluidic devices, which involve dimensions ranging from millimeters to micrometers, and these devices are capable of processing volumes of fluid in the range of pico- to microliters. Microscale flows are typically laminar owing to the viscous force and the very small scale. Laminar flow with a low Reynolds number (Re) and high Péclet number (Pe) is a condition in which the fluid flows in parallel layers with no disruption, except for diffusion (Fig. 3a). Without turbulent flow in microfluidic devices, mass transfer is relatively passive and slow. The absence of turbulent flow in microdevices becomes a critical issue when mixing reagents in chemistry and biology (Janasek et al. 2006). Although miniaturization can be used to reduce mass transfer distances, rapid mixing ( 70), the τstim at the stimulation STE was determined. The reaction arrest τquench was too fast for object-triggered imaging and was instead estimated to be one-fourth this value (Fig. 7c). The mean τstim was 2.22  0.15 ms. τquench was instead extrapolated from the τstim data and estimated to be 0.55 ms (a fourfold velocity increase forms the basis of our extrapolation). Faster switching times can be achieved by truncating the STE length. Nevertheless, the present milli- and sub-millisecond pinched flow deflection switching times are ~1000-fold faster than stimulation times obtained by traditional mixing methods (Moehren et al. 2001), 10–100 faster than the times achieved by the use of actuation force fields (Seger et al. 2004; Eriksson et al. 2007; Augustsson et al. 2009; Sasso et al. 2010), faster than the times achieved by hydrodynamic filtration (Yamada and Seki 2005; Toyama et al. 2012), and equivalent to the times achieved by stopped and quenched flow techniques used for the analysis of molecular kinetics (Gutfreund 1969; Schechter et al. 1970). Asymmetric flow input, parabolic flow patterns, and “no-slip” boundary cell wall interaction conditions generated forces with different vectors acting on the spherical

662

Y.-Y. Chiang

cell body, resulting in a turning force (the so-called moment or torque) that caused friction-driven rotation or caused the rolling behavior of the cell along the microchannel wall. The rolling behavior of the cell was observed (Fig. 8), and the cell could be described as a microwheel element that is enveloped by ligands through motion along the virtual interface. A second consequence was the formation of wave-like ligand patterns, which become evident as the cell was transported into the expansion channel, which is shown in Fig. 8a. The phenomenon is scalable and is readily observed with 200-μm particles (Fig. 8b). A slip-roll transport length of ~100 μm (~1 ms) can therefore be estimated to be required for surface saturation at a flow velocity of 100 mm/s (Chiang et al. 2013). The mixing characteristic also implies that ligand concentrations should be increased

Fig. 7 Schematic of the switching time τ defined as the transit time between t0 and t1 (a). Highspeed imaging frames recording the transport of a single HeLa S3 cell through an STE at 100 mm/s (b). Measured τstim (red) with Gaussian curve fitting (R2 ¼ 0.99) and extrapolated τquench (blue) values (c). Scale bar is 30 μm

Fig. 8 Cell rolling in the STE produces a self-mixing phenomenon. The rolling switching time is determined by the cell stream width and the cell radii. Disruption of a dye-doped laminar stream by rolling single cells (a) and 200-μm-diameter Cytodex particles (b). HeLa S3 cells cultured on microcarriers (Cytodex; GE Healthcare Life Sciences, Little Chalfont, UK). Cytodex particles rolling drives a ligand (red dye) envelopment mixing process

28

Record the Single Cell Signal Pathway

663

to ensure saturation of the cell surface. The previously discussed τswitch definition considered transport along the entire 200-μm length of the STE (τstim. ¼ 2.2 ms; τquench ¼ 0.55 ms). These switching times are therefore an overestimation of the cell–biochemical mixing time scales. Single-millisecond switching times are faster than those of other cell-handling microfluidic techniques (Moehren et al. 2001; Morton et al. 2008; Amini et al. 2012; Toyama et al. 2012) and microfluidic molecular mixers (Song and Ismagilov 2003) and are faster than the 30-ms dead times detected for the cryofixation-based quench flow method (Knoll et al. 1991; Plattner and Hentschel 2006). This ultrafast cell switching capability can be applied to investigate ligand-mediated cell surface processes with unprecedented temporal resolution.

Quantification of Specific Phospho-Proteins In this work, immunofluorescent staining and Western blotting were used to quantify specific phospho-proteins. For sample preparation and quantification, cells passed through the second STE were frozen with cold (4
C) fixation buffer (BD Biosciences, San Jose, CA, USA) or radioimmunoprecipitation assay (RIPA) buffer as a quencher for postanalysis such as staining and Western blotting, respectively. The use of RIPA buffer enables efficient cell lysis and avoids protein degradation. Therefore, the phosphor states are captured immediately. Cell samples were cooled in prechilled (20
C) permeabilization buffer (Perm III; BD Biosciences, Heidelberg, Germany) and incubated on ice for 90 min. Samples were then washed twice with cold (4
C) buffer. Samples were incubated overnight at 4
C with 1% IGFR or EGFR antibody in buffer, followed by five washes for further optical inspection using a motorized Olympus FV-1000 confocal laser scanning system mounted on an Olympus IX-81 inverted microscope. • IGF-IR mouse IgG1, κ primary antibody directly conjugated to Alexa Fluor ® 647 was used for the detection of IGF-IR pY1131 (BD Biosciences). • EGFR pure rabbit IgG primary antibody was used for the detection of EGFR pY1173 (Cell Signaling Technology, Danvers, MA, USA). • Cy3-conjugated anti-rabbit IgG secondary antibody (Dianova, Hamburg, Germany) To increase the protein concentration for blotting, two well-known downstream processing methods, acetone precipitation (AP) and molecular weight cut-off (MWCO), were used. Further details are provided later. To measure the total protein concentration (μg/μL), the Pierce bicinchoninic acid protein assay kit reagent set (Thermo Fisher Scientific, Waltham, MA, USA) was used (Smith et al. 1985). The tyrosine IGF-IR phosphorylation sites 1135/1135 and 1131 were examined following on-chip IGF1R incubations for 0.1, 0.5, and 2 s by using Western blotting. The bands are shown in Fig. 9. Western blot analysis elucidated the variation of protein

664

Y.-Y. Chiang

Fig. 9 Western blot analysis of extracts from HeLa S3, untreated, or insulin-treated (100 ng/mL for the indicated times), using total and phospho-IGF1R and β-actin antibody

Fig. 10 Measured mean gray values and intensities for the time-dependent signals of phosphorylation of pY1135/1136

amounts at assigned time points, which enabled the investigation of pY1135/1136 and pY1131 autophosphorylation transitions in IGF1R. Each total and phosphor-protein band was normalized to its β-actin value to eliminate individual loading difference. By comparing phosphor-protein bands with corresponding total IGF1R bands, tyrosine pY1135/1136 (shown in Fig. 10) and pY1131 (shown in Figs. 5 and 6) autophosphorylation transitions were determined as a percentage of total IGF1R. Figure 10 shows that minor phosphorylation of pY1135/1136 occurred within 100 ms, with a significant signal emerging between 500 ms and 2.0 s. Another autophosphorylation transition of pY1131 is presented in Fig. 11 In contrast to the autophosphorylation transitions of pY1135/1136, pY1131 phosphorylation clearly occurred within 100 ms and was maintained for 500 ms, followed by dephosphorylation in the 2-s time course. Tyrosine phosphorylation as a percentage of total protein significantly fluctuated between 10% and 45%. The asynchronous autophosphorylation transitions of different IGF-IR tyrosine sites indicate the possibility of the early activation of Y1131, followed by phosphate transfer to Y1135/36. These results not only revealed autocatalytic modification but also

28

Record the Single Cell Signal Pathway

665

Fig. 11 Measured band intensities of the area mean gray values and the time-dependent signals of phosphorylation of pY1131

suggested that protein conformation changes occur at a very early stage following activation (i.e., 100 ms). However, the behavior of individual cells within the same population may differ dramatically. The “noise” underlying the entire population bulky data might be triggered to alter the cell fate (Schubert 2011). Stochasticity can have important consequences for the health and function of the entire cell population. Hence, to demonstrate individual molecular dynamics with microsecond resolution, a singlecell characterization method was developed.

Single-Cell Fluorescence Intensity Quantification The near-instantaneous and high-throughput delivery of cells within the microfluidic quenched flow analysis platform is perfectly tailored for single-cell analysis. Single cells were collected from the outlet tubing and immunostained for phosphorylated receptor residues, and the area mean gray value measurements were conducted to quantify phosphorylation levels. In the study of IGF1R, single cells were collected and stained with Alexa Fluor® 647 mouse anti-IGF1R. The single-cell merged images of bright field and the fluorescent channels with assigned color (red) are provided in Fig. 12a. The populationaveraged time-dependent signals are plotted in Fig. 12b. Complete autophosphorylation of the pY1131 residue occurred within 100 ms and remained stable throughout the 2-s time course. Phosphorylation levels were lower than those of the positive control. Consistent with the literature, maximal levels were gradually attained over 5 min (Vasilcanu et al. 2004). The quantitative image data captured the rapid state switching and the natural variations and fluctuations in the system. By decreasing the length of the incubation channel, the initial 100 ms can be investigated in greater detail to fully elucidate the mechanisms underlying the sequential phosphorylation of the pY1131, 1135, and 1136 tyrosine residues. This is the origin of IGF1R signal

666

Y.-Y. Chiang

Fig. 12 Rapid autophosphorylation state switching. The onset of the IGF-1 receptor signal transduction by autophosphorylation (pY1131). Overlaid bright field and fluorescence microscopic images of single cells treated with IGF for precisely defined periods. Cells were treated for 100, 500, and 2000 ms. The negative control sample was processed in the microfluidic quenched flow system in the absence of any ligand. Positive controls were treated for 5 min in a Petri dish (p < 9.4  108) Fig. 13 Overlaid bright field and fluorescence microscopic images of single cells treated with EGF for precisely defined periods

transduction and is necessary for stabilizing the kinase activation loop, which drives substrate catalysis for downstream signaling. This result was published in Analytical Chemistry (Chiang et al. 2013). The 100-ms IGF1R Y1131 autophosphorylation state switching was highly reproducible. The limited success with the MWCO and AP sample preparation methods for Western blot analysis and low reproducibility suggests that the apparent Y1131-to-Y1135/1136 exchange was most likely a consequence of an error introduced in the lengthy and difficult sample preparation method. Whole-cell microscopy does not have this drawback and is highly reproducible, verifying that the signal is maintained following Y1131 activation. In another experiment, single cells were immunostained for Cy3-conjugated secondary antibody for EGFR pY1173, and area mean gray value measurements were conducted to quantify phosphorylation levels. Figure 13 presents selected images of single cells immunostained for pY1173 for varying 0–2.0-s on-chip EGF stimulations.

28

Record the Single Cell Signal Pathway

667

Fig. 14 Single-cell pY1173 signaling replicated data obtained from preliminary microfluidic quenched flow EGF stimulation experiments (left). Population-averaged data describing the general kinetics of pY1173 signaling (right). Absence of >2-s to 5-s time points prevents the collision probability-driven kinetics of EGFR dimerization being observed (a and b). Lengthy sample storage (3–5 days) prior to cell imaging resulted in reduced signal levels and greater signal heterogeneity within the cell populations (c and d). Single-cell pY1173 heterogeneity throughout the full time course (e and f)

Incubations for 5 s and 5 min were achieved by straightforward mixing in 2-mL volumes. Negative control cells were combined on chip with serum-free media only. Single-cell EGFR Y1173 phosphorylation replicate data from incubation-grouped subpopulations and population-averaged data describing the general kinetics of pY1173 signaling are provided in Fig. 14. The images revealed marked heterogeneity of the pY1173 expression from the different subpopulations (n > 50 per incubation

668

Y.-Y. Chiang

Fig. 15 The population-averaged autophosphorylation dynamics of EGFR Y1173 (top, right). Proposed signaling mechanism

period) and three replicates. The quantitative image data captured the rapid state switching and natural variations and fluctuations in the system. Single-cell heterogeneity resulted from the noise caused by stochastic fluctuations in the signaling networks. The noise is possibly caused by the constitutively activated EGFR homolog ErbB2 (Ladbury and Arold 2012). Despite the fluctuations within the cell population, a clear ( p < 1012) pY1173 biphasic signal was observed and was highly reproducible. Complete Y1173 phosphorylation occurred within 100 ms and was maintained for 250 ms, followed by sharp dephosphorylation toward baseline levels at 500 ms and asymptotic recovery to saturation levels over a ~ 5-s time scale. The sharp dephosphorylation transition of initial pY1173 signal was previously missed due to the poor time resolution of existing methods. The results point to a new origin of signaling involving intramolecular phosphorylation, and activation is shortly suppressed by dephosphorylation, which is most likely mediated by SHP-1, a Y1173-specific protein tyrosine phosphatase (PTP) (Keilhack et al. 1998; Tiganis 2002). The asymptotic recovery of the pY1173 signal in the approximately 5-s time frame is a hallmark of dimerization-mediated EGFR activation, with kinetics defined by diffusion-driven collision probability within the membrane (Verveer 2000). The pY1173 protection mechanism is unknown but may involve displacement of the phosphatase during EGFR dimerization and steric shielding of the pY1173 residue to prevent repeated dephosphorylation. Taken together, the results imply a novel negative feedback regulatory mechanism, involving silencing cis-phosphorylation until diffusion-mediated collision with another ligand-occupied monomer for the initiation of trans-phosphorylation necessary to initiate downstream signaling (Fig. 15). The proposed mechanism (Chiang and West 2013) can be used to the select the potential anticancer targets and desired modes of therapeutic action. Importantly, this microfluidic-base quenched-flow technology allows whole-cell investigations, providing genuinely authentic receptor and cell surface models (Bessman and Lemmon 2012).

Conclusion The research documented in this chapter describes the development of microfluidic quenched flow platform for rapid single cell switching between biochemical microenvironments. This platform was developed because of the absence of techniques for

28

Record the Single Cell Signal Pathway

669

studying rapid cell surface events. Instead of diffusion-limited mass transfer, the high-shear conditions of turbulent mixing, or deflecting cells orthogonal to the flow by the use of force fields, a passive whole cell switching platform is demonstrated. The microfluidic quenched flow platform involves transporting single cell continuously across the virtual interface between laminar flows, dwelling with reproducible and precisely defined incubation periods, and finally switching within an adjacent quenching buffer stream for reaction arrest. Furthermore, the cell rolling phenomenon observed acts to envelop the cell for rapid cell stimulation and rapid reaction arrest (~1 ms switching time). Millisecond switching was used to stimulate and preserve insulin-like growth factor 1 receptor (IGF-1R) and epidermal growth factor receptor (EGFR) tyrosine autophosphorylation transitions. Significant autophosphorylation of the IGF-1R (pY1131) residue occurred within 100 ms, and a novel biphasic EGFR (pY1173) signaling mechanism was recorded by using the whole cell quenched flow analysis platform. The technology is highly relevant to fundamental systems biology research to identify new therapeutic targets and also relevant to the pharmaceutical industry to aid the development of new drugs.

References Adams TE et al (2000) Structure and function of the type 1 insulin-like growth factor receptor. Experientia 57(7):1050–1093. https://doi.org/10.1007/pl00000744 Amini H et al (2012) Intrinsic particle-induced lateral transport in microchannels. Proc Natl Acad Sci 109(29):11593–11598. https://doi.org/10.1073/pnas.1207550109 Augustsson P, Laurell T (2012) Acoustofluidics 11: affinity specific extraction and sample decomplexing using continuous flow acoustophoresis. Lab Chip 12(10):1742–1752. https://doi.org/10.1039/c2lc40200a Augustsson P et al (2009) Buffer medium exchange in continuous cell and particle streams using ultrasonic standing wave focusing. Microchim Acta 164(3–4):269–277. https://doi.org/10.1007/ s00604-008-0084-4 Baserga R (1999) The IGF-I receptor in cancer research. Exp Cell Res 253(1):1–6. https://doi.org/ 10.1006/excr.1999.4667 Baserga R (2000) The contradictions of the insulin-like growth factor 1 receptor. Oncogene 19(49):5574–5581. https://doi.org/10.1038/sj.onc.1203854 Beech JP et al (2012) Sorting cells by size, shape and deformability. Lab Chip 12(6):1048–1051. https://doi.org/10.1039/c2lc21083e Bessman NJ, Lemmon MA (2012) Finding the missing links in EGFR. Nat Struct Mol Biol 19(1):1–3 Bleul R et al (2011) Compact, cost-efficient microfluidics-based stopped-flow device. Anal Bioanal Chem 399(3):1117–1125. https://doi.org/10.1007/s00216-010-4446-5 Busico V et al (1999) Stopped-flow polymerizations of ethene and propene in the presence of the catalyst system rac-Me2Si(2-methyl-4-phenyl-1-indenyl)2ZrCl2/methylaluminoxane. Macromol Rapid Commun 20(3):116–121. https://doi.org/10.1002/(sici)1521-3927(19990301)20:33.0.co;2-a Chiang Y-Y, West J (2013) Ultrafast cell switching for recording cell surface transitions: new insights into epidermal growth factor receptor signalling. Lab Chip 13(6):1031–1034. https:// doi.org/10.1039/c3lc41297k Chiang Y-Y et al (2013) Whole cell quenched flow analysis. Anal Chem. https://doi.org/10.1021/ ac402881h

670

Y.-Y. Chiang

Dengjel J et al (2007) Quantitative proteomic assessment of very early cellular signaling events. Nat Biotechnol 25(5):566–568. http://www.nature.com/nbt/journal/v25/n5/suppinfo/nbt1301_S1. html Elove GA et al (1994) Kinetic mechanism of cytochrome c folding: involvement of the Heme and its ligands. Biochemistry 33(22):6925–6935. https://doi.org/10.1021/bi00188a023 Eriksson E et al (2007) A microfluidic system in combination with optical tweezers for analyzing rapid and reversible cytological alterations in single cells upon environmental changes. Lab Chip 7(1):71–76. https://doi.org/10.1039/b613650h Evander M et al (2008) Acoustophoresis in wet-etched glass chips. Anal Chem 80(13):5178–5185. https://doi.org/10.1021/ac800572n Fersht AR, Jakes R (1975) Demonstration of two reaction pathways for the aminoacylation of tRNA. Application of the pulsed quenched flow technique. Biochemistry 14(15):3350–3356. https://doi.org/10.1021/bi00686a010 Fiedler S et al (1998) Dielectrophoretic sorting of particles and cells in a microsystem. Anal Chem 70(9):1909–1915. https://doi.org/10.1021/ac971063b Freedman VH, Shin S-i (1974) Cellular tumorigenicity in nude mice: correlation with cell growth in semi-solid medium. Cell 3(4):355–359 Frimat J-P et al (2011) A microfluidic array with cellular valving for single cell co-culture. Lab Chip 11(2):231–237. https://doi.org/10.1039/c0lc00172d Gutfreund H (1969) Rapid mixing: continuous flow. In: Kenneth K (ed) Methods in enzymology, vol 16. Academic, Cambridge, Massachusetts, USA, pp 229–249 Hawkes JJ et al (2004) Continuous cell washing and mixing driven by an ultrasound standing wave within a microfluidic channel. Lab Chip 4(5):446–452. https://doi.org/10.1039/b408045a Hennemeyer M et al (2008) Cell proliferation assays on plasma activated SU8. Microelectron Eng 85(5):1298–1301. https://doi.org/10.1016/j.mee.2008.01.026 Hill M (2003) The selection of layer thicknesses to control acoustic radiation force profiles in layered resonators. J Acoust Soc Am 114(5):2654–2661. https://doi.org/10.1121/1.1616581 Holm SH et al (2011) Separation of parasites from human blood using deterministic lateral displacement. Lab Chip 11(7):1326–1332. https://doi.org/10.1039/c0lc00560f Huang Y et al (1993) Electrokinetic behaviour of colloidal particles in travelling electric fields: studies using yeast cells. J Phys D Appl Phys 26(9):1528 Il D, et al (2004) A continuous cell separation chip using hydrodynamic dielectrophoresis process. In: Micro electro mechanical systems, 2004. 17th IEEE international conference on. (MEMS) Janasek D et al (2006) Scaling and the design of miniaturized chemical-analysis systems. Nature 442(7101):374–380 Kakuta M et al (2003) Time-resolved Fourier transform infrared spectrometry using a microfabricated continuous flow mixer: application to protein conformation study using the example of ubiquitin. Lab Chip 3(2):82–85. https://doi.org/10.1039/b302295a Keilhack H et al (1998) Phosphotyrosine 1173 mediates binding of the protein-tyrosine phosphatase SHP-1 to the epidermal growth factor receptor and attenuation of receptor signaling. J Biol Chem 273(38):24839–24846. https://doi.org/10.1074/jbc.273.38.24839 Knoll G et al (1991) Quenched flow analysis of exocytosis in Paramecium cells: time course, changes in membrane structure, and calcium requirements revealed after rapid mixing and rapid freezing of intact cells. J Cell Biol 113(6):1295–1304. https://doi.org/10.1083/jcb.113.6.1295 Ladbury JE, Arold ST (2012) Noise in cellular signaling pathways: causes and effects. Trends Biochem Sci 37(5):173–178 Lakowicz JR (2006) Principles of fluorescence spectroscopy. Springer, New York Laurell T et al (2007) Chip integrated strategies for acoustic separation and manipulation of cells and particles. Chem Soc Rev 36(3):492–506. https://doi.org/10.1039/b601326k Lenshof A, Laurell T (2010) Continuous separation of cells and particles in microfluidic systems. Chem Soc Rev 39(3):1203–1217. https://doi.org/10.1039/b915999c Leroith D et al (1995) Molecular and cellular aspects of the insulin-like growth factor I receptor. Endocr Rev 16(2):143–163. https://doi.org/10.1210/edrv-16-2-143

28

Record the Single Cell Signal Pathway

671

Lopaczynski W et al (2000) Autophosphorylation of the insulin-like growth factor I receptor cytoplasmic domain. Biochem Biophys Res Commun 279(3):955–960. https://doi.org/ 10.1006/bbrc.2000.4046 Maenaka H et al (2008) Continuous and size-dependent sorting of emulsion droplets using hydrodynamics in pinched microchannels. Langmuir 24(8):4405–4410. https://doi.org/ 10.1021/la703581j McLimans WF et al (1957) The submerged culture of mammalian cells: the spinner culture. J Immunol 79(5):428–433 Moehren G et al (2001) Temperature dependence of the epidermal growth factor receptor signaling network can be accounted for by a kinetic model†. Biochemistry 41(1):306–320. https://doi.org/ 10.1021/bi011506c Morton KJ et al (2008) Crossing microfluidic streamlines to lyse, label and wash cells. Lab Chip 8(9):1448–1453. https://doi.org/10.1039/b805614e Oda K et al (2005) A comprehensive pathway map of epidermal growth factor receptor signaling. Mol Syst Biol 1:1–17. http://www.nature.com/msb/journal/v1/n1/suppinfo/msb4100014_S1.html Pamme N, Manz A (2004) On-Chip free-flow Magnetophoresis: continuous flow separation of magnetic particles and agglomerates. Anal Chem 76(24):7250–7256. https://doi.org/10.1021/ ac049183o Panine P et al (2006) Probing fast kinetics in complex fluids by combined rapid mixing and smallangle X-ray scattering. Adv Colloid Interf Sci 127(1):9–18. https://doi.org/10.1016/j. cis.2006.06.003 Paterson L et al (2001) Controlled rotation of optically trapped microscopic particles. Science 292(5518):912–914. https://doi.org/10.1126/science.1058591 Plattner H, Hentschel J (2006) Sub-second cellular dynamics: time-resolved electron microscopy and functional correlation. International Review Cytology. Kwang WJ. Academic. 255:133–176 Sai Yet al (2006) Continuous separation of particles using a microfluidic device equipped with flow rate control valves. J Chromatogr A 1127(1):214–220. https://doi.org/10.1016/j.chroma.2006.05.020 Sancho M et al (2010) Interaction between cells in dielectrophoresis and electrorotation experiments. Biomicrofluidics 4(2):022802–022811 Sasso LA et al (2010) Autonomous magnetically actuated continuous flow microimmunofluorocytometry assay. Microfluid Nanofluid 9(2):253–265. https://doi.org/10.1007/ s10404-009-0543-1 Sasso LA et al (2013) Continuous monitoring of inflammation biomarkers during simulated cardiopulmonary bypass using a microfluidic immunoassay device – A pilot study. Artif Organs 37(1):E9–E17. https://doi.org/10.1111/aor.12021 Schechter AN et al (1970) Kinetics of folding of staphylococcal nuclease. Science 167(3919):886–887. https://doi.org/10.1126/science.167.3919.886 Scheidegger KJ (2000) Estradiol decreases IGF-1 and IGF-1 receptor expression in rat aortic smooth muscle cells. Mechanisms for its atheroprotective effects. J Biol Chem 275(49):38921–38928. https://doi.org/10.1074/jbc.M004691200 Schubert C (2011) Single-cell analysis: the deepest differences. Nature 480(7375):133–137 Seger U et al (2004) Cell immersion and cell dipping in microfluidic devices. Lab Chip 4(2):148–151. https://doi.org/10.1039/b311210a Shevkoplyas SS et al (2007) The force acting on a superparamagnetic bead due to an applied magnetic field. Lab Chip 7(10):1294–1302. https://doi.org/10.1039/b705045c Shi J et al (2009a) Acoustic tweezers: patterning cells and microparticles using standing surface acoustic waves (SSAW). Lab Chip 9(20):2890–2895. https://doi.org/10.1039/b910595f Shi J et al (2009b) Continuous particle separation in a microfluidic channel via standing surface acoustic waves (SSAW). Lab Chip 9(23):3354–3359. https://doi.org/10.1039/b915113c Shin S-I et al (1975) Tumorigenicity of virus-transformed cells in nude mice is correlated specifically with Anchorage independent growth in vitro. Proc Natl Acad Sci 72(11):4435–4439. https://doi.org/10.1073/pnas.72.11.4435

672

Y.-Y. Chiang

Smith PK et al (1985) Measurement of protein using bicinchoninic acid. Anal Biochem 150(1):76–85. https://doi.org/10.1016/0003-2697(85)90442-7 Snow CD et al (2002) Absolute comparison of simulated and experimental protein-folding dynamics. Nature 420(6911):102–106. http://www.nature.com/nature/journal/v420/n6911/suppinfo/ nature01160_S1.html Song H, Ismagilov RF (2003) Millisecond kinetics on a microfluidic chip using nanoliters of reagents. J Am Chem Soc 125(47):14613–14619. https://doi.org/10.1021/ja0354566 Sourjik V, et al (2007) In vivo measurement by FRET of pathway activity in bacterial chemotaxis. Methods in enzymology. B. R. C. Melvin I. Simon and C. alexandrine, academic press. 423: 363–391 Spiller DG et al (2010) Measurement of single-cell dynamics. Nature 465(7299):736–745. http:// www.nature.com/nature/journal/v465/n7299/abs/nature09232.html#supplementaryinformation Suhling K et al (2005) Time-resolved fluorescence microscopy. Photochem Photobiol Sci 4(1):13–22. https://doi.org/10.1039/b412924p Taniike T et al (2012) Development of a large-scale stopped-flow system for heterogeneous olefin polymerization kinetics. Macromol React Eng 6(6–7):275–279. https://doi.org/10.1002/ mren.201200003 Tiganis T (2002) Protein tyrosine phosphatases: dephosphorylating the epidermal growth factor receptor. IUBMB Life 53(1):3–14. https://doi.org/10.1080/15216540210811 Toyama K et al (2012) Isolation of cell nuclei in microchannels by short-term chemical treatment via two-step carrier medium exchange. Biomed Microdevices 14(4):751–757. https://doi.org/ 10.1007/s10544-012-9653-8 Vasdekis AE, Laporte GPJ (2011) Enhancing single molecule imaging in optofluidics and microfluidics. Int J Mol Sci 12(8):5135–5156 Vasilcanu D et al (2004) The cyclolignan PPP induces activation loop-specific inhibition of tyrosine phosphorylation of the insulin-like growth factor-1 receptor. Link to the phosphatidyl inositol-3 kinase/Akt apoptotic pathway. Oncogene 23(47):7854–7862. https://doi.org/10.1038/sj.onc.1208065 Verveer PJ (2000) Quantitative imaging of lateral ErbB1 receptor signal propagation in the plasma membrane. Science 290(5496):1567–1570. https://doi.org/10.1126/science.290.5496.1567 Wang W et al (1995) Optical trapping and fluorescence detection in laminar flow streams. Appl Phys Lett 67(8):1057–1059. https://doi.org/10.1063/1.114463 Yamada M, Seki M (2005) Hydrodynamic filtration for on-chip particle concentration and classification utilizing microfluidics. Lab Chip 5(11):1233–1239. https://doi.org/10.1039/b509386d Yamada M et al (2004) Pinched flow fractionation: continuous size separation of particles utilizing a laminar flow profile in a pinched microchannel. Anal Chem 76(18):5465–5471. https://doi.org/ 10.1021/ac049863r Yoda T et al (2001) Folding-unfolding of goat α-lactalbumin studied by stopped-flow circular dichroism and molecular dynamics simulations. Proteins Struct Funct Bioinf 42(1):49–65. https://doi.org/10.1002/1097-0134(20010101)42:13.0.co;2-z Yu H, Rohan T (2000) Role of the insulin-like growth factor family in cancer development and progression. J Natl Cancer Inst 92(18):1472–1489. https://doi.org/10.1093/jnci/92.18.1472 Zellner M et al (2005) Quantitative validation of different protein precipitation methods in proteome analysis of blood platelets. Electrophoresis 26(12):2481–2489. https://doi.org/10.1002/ elps.200410262

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

29

Fabrice Gielen

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ultrahigh-Throughput Enzyme Screens: Directed Evolution and Discovery . . . . . . . . . . . . . . . . . . Single-Cell Encapsulation in Microfluidic Droplets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stochastic Encapsulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deterministic Single-Cell Encapsulation Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Microfluidic Workflow for Directed Evolution of Enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Construction of Multistep Workflows for Single-Cell Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Considerations for Setting Up Single Cell Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Operation of a Microdroplet Sorter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Formats for Single-Cell Biocatalyst Functional Screens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Lysate Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Internal Expression and Surface Display . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Secretion Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functional Metagenomics and Bioprospecting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . In Vitro Workflows as Artificial Single Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Limitations and Future Prospects of Single-Cell Microencapsulation . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

674 677 677 678 679 679 680 680 682 683 683 685 686 686 687 689 689 690

Abstract

Enzymes are responsible for many chemical reactions which support life and are used in numerous industries. Laboratory evolution experiments can mimic the Darwinian process of mutation and selection pressure, yielding newly evolved catalysts. Recently developed single-cell microencapsulation techniques enable the screening of ultralarge enzyme-mutant pools, greatly increasing the odds of isolating improved catalysts. The encapsulation process uses droplet microfluidics, a groundbreaking technology for the miniaturization and automation of biochemical assays which has already demonstrated its effectiveness at evolving F. Gielen (*) University of Exeter, Living Systems Institute, University of Exeter, Exeter, UK e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_47

673

674

F. Gielen

enzymes and identifying ultrarare catalysts from the environment. This chapter reviews how large-scale single-cell assays provide an efficient route toward the identification of biocatalysts with novel or improved function. Single-cell microencapsulation enables retention of genotype (gene sequence) and phenotype (enzyme function) within a stable microcompartment. In practice, single-gene mutants expressed by single cells with their corresponding enzyme are encapsulated into monodisperse water-in-oil microdroplets where the assay takes place. This enables the screening of ultralarge 108 members libraries in a day with a concomitant reduction of 1000-fold in assay-reagent costs and a 1000-fold increase in throughput compared to conventional tools. This technology has already been widely adopted for the accelerated discovery and evolution of biocatalysts used in many application domains from the decontamination of polluted environments, bioenergy production, to the degradation of plastics and the discovery of antimicrobials.

Introduction As the world’s major catalysts synthesized by living cells, enzymes are essential for supporting life and most biological processes. They display exquisite properties such as catalytic efficiency, regio- and stereoselectivity, or biodegradability, which can be harnessed to provide green, sustainable chemistry alleviating the need for petroleumbased chemicals and are expected to occupy a central role in the circular bioeconomy. Application fields range from biofuel production, pharmaceuticals, detergents, cosmetics, or fragrances where they have been incorporated in several manufacturing processes for many years. A major challenge is to tailor and optimize a given catalyst for an end application as enzymes will have to display a specific set of properties (stability, activity, pH range, melting temperature, and substrate specificity) suited to an industrial bioprocess. For instance, altering substrate specificity can produce catalysts able to act on several substrates simultaneously, increased melting temperature makes enzymes more compatible with large fermenters-based production processes, and changing their stereospecificity can prove invaluable for drug manufacturing. “Offthe-shelf” enzymes from Nature’s catalytic repertoire provide catalysts optimized in a specific evolutionary context but suboptimal for a different application. This has spurred protein engineers to devise tools and methods to improve enzymes artificially (also known as “recombinant” enzymes). Indeed, most enzymes are proteins, and as such have their function encoded by a unique genetic DNA sequence. With the advent of molecular and synthetic biology, DNA synthesis, and sequencing technologies, our understanding and control over genes allows to introduce alterations to any initial sequence, therefore altering and possibly improving enzyme function. This process of artificial evolution and selection has been widely used with classical colony-based screens, but new enzymes still find major hurdles on the way from laboratory discovery to industrial application: low success rates during screening and engineering campaigns, the need for expensive robotic equipment for

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

675

significant diversity exploration, time- and resource-consuming sequence analyses, and limited activity and stability in final product formulations. Recent progress made in the field of single-cell technologies has partially lifted these hurdles as they enable to rapidly fine-tune enzymes through evolution campaigns and discover novel biocatalysts from the environment. A proven means to evolve biocatalysts is through directed evolution (DE), which is a man-made approach that mimics natural evolution by artificial selection. It is, to date, one of the best routes for improving enzymes, because of our current deep inability to predict function from gene sequence. In DE, engineered mutations are accumulated from a starting gene in an iterative process followed by functional screening and selection of improved variants. This is illustrated in Fig. 1. Such adaptive walks can lead to rapidly improved enzyme properties (or “fitness”) (Chen and Arnold 1993; Cuesta et al. 2015). However, the odds of finding significantly improved variants are very low because most mutations destabilize proteins or disrupt the catalytic machinery. One solution to beat such poor odds is to screen the largest possible number of unique variants against specific desired functions (Romero and Arnold 2009; Colin et al. 2015). Even then, only a tiny fraction of the enormous combinatorial diversity around starting sequences can be accessed: for a 300-amino-acid protein, there are 5700 possible single-amino-acid substitutions and 32,381,700 ways to make just 2 substitutions with the 20 canonical amino acids (Yang et al. 2019). Apart from evolving known enzymes, Nature’s catalytic reservoir can be mined for discovering unknown enzymes by exploration of the microbiome, metagenome, or metatranscriptome (Uchiyama and Miyazaki 2009). Here again, among the vast

Fig. 1 Directed evolution of biocatalysts expressed in single cells. A gene library is constructed to introduce mutations in the original wild-type DNA sequence (1). Single cells are subsequently used to express these sequences which are assayed for their function (2). A quantitative analytical read-out is used to assess and select enzymes with improved function. Larger product formation is represented by bigger “P” symbols (3). Iterations of the process lead to enzymes with gradually improving functions

676

F. Gielen

amount of genetic material from the environment, a tiny fraction will correspond to genes-encoding active enzymes (Fig. 2). There is thus a need for high-throughput screening technology platforms able to isolate and annotate these catalysts in a cost- and time-efficient manner. Classical screening formats such as in vivo colony screens in Petri dishes and microtiter plates can be fully automated with robotic equipment (Dorr et al. 2016). These methods can reach a throughput of 105 individual enzyme mutant per day. Alternative formats are flow cytometry (fluorescence-activated cell sorting, FACS) and microencapsulation

Fig. 2 Microdroplet generation in microfluidic devices. (a) General setup used for generation of monodisperse water-in-oil droplets in microfluidic devices: Solutions loaded in syringes are pushed by syringe pumps into tubing connected to microfluidic channels. The resulting droplets are collected in a tube and can be manipulated further. (b) Snapshot of a microfluidic flow-focusing junction generating water-in-oil microdroplets where single cells get encapsulated. A carrier oil phase (labeled “2”) splits an aqueous phase with cells (labeled “1”) into microcompartments. (c) Schematic of a microdroplet used for biocatalyst screen: Single-cell-expressing enzymes are coencapsulated with substrate which is turned over into product

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

677

Table 1 Comparison between conventional and droplet microfluidic screens. (Adapted from Agresti et al. (2010))

Volume per assay Protein per assay Number of assays Size of apparatus Cost (1 million assays)

Conventional screen 1–200 μL

Single-cell microdroplet screen 1 nL–1 fL

100–200 ng 1 per second 50
10 m $1 million

10 fg–10 ag >1000 per second 10
5 cm (chip) $1–$1000

Advantage 103–106 times smaller 103–106 times less >1000 times faster 500 times smaller 103–108 times cheaper

(fluorescence-activated droplet sorting, FADS), which provide a higher throughput of 108 mutants per day (Tu et al. 2011). FACS methods require internalization of both substrates and reaction products in a cell or confinement to the cell surface, a condition rarely met. In contrast, microencapsulation methods retain enzyme-producing cells and products in the same microcompartment-linking genotype (the nucleic acid sequence encoding a potential catalyst) with phenotype (a functional trait, assessed through product formation). They are thus more widely applicable. Historically, polydisperse water-in-oil emulsions have been used to demonstrate selection of genes-encoding catalysts (Tawfik and Griffiths 1998; Miller et al. 2006). This format has since been superseded by microfluidic-based water-in-oil microdroplets. Table 1 compares conventional colony screens and single-cell microdroplet screens, highlighting the advantage of microdroplets formats for throughput (1000 times faster) and costs (1000 times cheaper).

Ultrahigh-Throughput Enzyme Screens: Directed Evolution and Discovery Single-Cell Encapsulation in Microfluidic Droplets The ability to encapsulate single cells in water-in-oil microfluidic droplets has been spurred by engineering advances borrowed from the semiconductor industry which has enabled the patterning of microchannels down to a few microns and the construction of cheap, customizable microfluidic channel networks. This has led to the possibility of building custom functionalities tailored to one’s specific application. Specifically, water-in-oil microdroplets (10–100 μm in diameter, corresponding to picoliter to nanoliter volumes) are generated at kHz rates using a carrier-oil phase, typically fluorinated oils. The oil further contains surfactant molecules that get adsorbed at the oil/water interface, ensuring the stability of the emulsion for up to days even while being manipulated and injected through successive devices. Microfluidic modules can be used sequentially to include droplet generation, incubation,

678

F. Gielen

splitting, reinjection, or sorting, all operating at high throughput so that the same monodisperse emulsion can be subjected to multiple manipulation steps, lending the possibility to perform complex assays in tightly controlled conditions for up to several million reactors at a time (Koster et al. 2008; van Vliet et al. 2015).

Stochastic Encapsulation The number of cells per compartment depends mainly on the initial cell density and droplet sizes. The volume of the droplets is determined in turn by the dimensions and geometries of the microfluidic device and the flow rates at which the continuous and disperse phases are delivered, usually using pressure or syringe pumps. For cells uniformly suspended in a buffer, their times of arrival at the droplet formation site are random so that the distribution of cells can be predicted by Poisson distribution (Fig. 3). The Poisson distribution is given by: Pð k Þ ¼

λk eλ k!

where k is the actual number of cells in a droplet and λ is the target average number of cells per droplet. In practice, the number of droplets either empty or filled with one or more than one cell can be calculated. Enzyme screens typically require avoiding cell doublets that could result in falsepositive hits. To ensure monoclonality, the typical target cell number used for singlecell biocatalysts screens is between 0.1 (an average of 1 cell per every 10 droplets for fine enrichment of smaller libraries) and 1 (an average of 1 cell per droplet for initial large-scale library screens).

Fig. 3 Theoretical Poisson distributions with target occupancy of 0.1 (green), 0.5 (red), and 1 (blue). To minimize cell doublets, concentrated cell solutions must be diluted to reach low-target occupancies (e.g., below 0.1)

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

679

Deterministic Single-Cell Encapsulation Techniques There are multiple methods that have been developed to circumvent the limitations of the Poisson distribution, many of which are reviewed in (Collins et al. 2015). Such methods are deterministic as they enable accurate control over the number of cells per microdroplet and overcome purely stochastic processes resulting in many droplets being either empty or containing more than a single cell. This can be achieved with passive or active methods. Passive approaches include the use of inertial microfluidics in high-aspect ratio geometries or spirals where cells self-order before encapsulation (Collins et al. 2015; Kemna et al. 2012). Such strategies can be applied to achieve double Poisson distribution with cells (Lagus and Edd 2013). On the whole, passive approaches can perform well on single objects or objects of the same type. However, they require adaptation for selections with any given micro-objects according to their specific hydrodynamics properties. An alternative approach to address on-demand cell loading is to perform an active droplet sort to extract the subpopulation of droplets with the right content. Droplet selection has historically relied on fluorescence readouts but has since been expanded to absorbance, image analysis, or scattering (Gielen et al. 2016; Hasan et al. 2019; Liu et al. 2016).

Microfluidic Workflow for Directed Evolution of Enzymes A typical enzyme-mutant library-screening workflow is made up of four main steps summarized in Fig. 4 which translates the essential steps shown in Fig. 1 into microfluidic manipulation modules: for diversity generation and single-cell expression, single cells expressing enzyme mutants are encapsulated with substrate molecules (1), for functional assay, droplets are incubated to allow the reactions to proceed (2), and for selection, a high-throughput sorting device allows isolation of the best enzyme mutants (3) which enables sequencing of genes encoding for

Fig. 4 Microfluidic workflow for the screening of single-cell-expressing enzymes. (1) Cells and substrate molecules (marked “S”) are coencapsulated in a flow-focusing device. (2) Droplets are incubated to allow reactions to proceed, and (3) All droplets are screened, and the ones containing the highest amount of reaction product (marked “P”) are isolated and collected. Flow direction is indicated by black arrows.

680

F. Gielen

improved catalysts (4). The number of microdroplets produced in step 1 is typically many millions (e.g., 10 million from 100 μL solution with cells) so that millions of single cells (i.e., enzyme mutants) can be evaluated in a single experiment. Practical aspects to build and implement such workflows are reviewed in detail in (Neun et al. 2019). It is worth pointing out that such screens are becoming increasingly routine in both academic and industrial settings since the first demonstration of single-cell enzyme evolution a decade ago (Agresti et al. 2010).

Construction of Multistep Workflows for Single-Cell Assays There are numerous cases where addition of reagents to single cells is required. For instance, for an enzyme-screening workflow, substrates must be added. This is usually done by coflowing a solution in which the substrate is dissolved alongside a cell suspension (Fig. 3-1). This way, enzymatic reactions are initiated immediately after cell encapsulation. However, if successive steps are required, e.g., for coupled reactions or to enable cell growth prior to a reaction, substrate can be added using a pico-injection module which adds defined picoliter amounts of reagents to preformed droplets (Abate et al. 2010). This module operates at very high throughput (kHz) so that all reactions can be initiated within minutes (Fig. 5).

Considerations for Setting Up Single Cell Assays Library Construction and DNA Transformation Directed evolution requires the construction of libraries of protein variants. The most straightforward method of constructing such libraries is to construct a library of nucleic acid molecules from which the protein library can be translated. There are three main approaches to generate sequence diversity: point mutations, insertions, or deletions; controlled level of randomization to specific positions within the DNA sequence; and combining existing diversity in new ways with recombination techniques, such as DNA shuffling. Randomization methods like error-prone PCR have been a technique of choice for creating cheap, ultralarge libraries with known

Fig. 5 Controlled addition of a reagent to droplets by pico-injection: A channel containing the solution to inject (appearing in black at the top of the figure) is connected to a second channel where droplets flow at high rates. Electrodes (lines in black at the bottom of the figure) are activated to promote coalescence between the solution to be injected and the droplets. After injection, the droplets flow away. Scalebar ¼25 microns. (Permission from PNAS (Abate et al. 2010))

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

681

average number of mutations per gene. In this technique, a low fidelity polymerase introduces random mutations during DNA amplification. Mutated gene libraries are subsequently inserted into plasmids or fosmids (extrachromosomal DNA) that contain the necessary genetic elements for protein induction. Plasmids with high-copy number (average number of copies of a certain plasmid inside a host cell) are chosen to enable efficient recovery of all the improved sequences (Neylon 2004). After plasmid construction, transformation into host cells allows the genes to be expressed. E. coli has been historically the most used host in which DNA can be internalized efficiently (>109 per μg of DNA) using electroporation or heat-shock methods (Olsen et al. 2000). This results in the uptake of individual plasmids by millions of individual cells. After internalization, the cell protein transcription/translation machinery can express the genes encoded by the foreign plasmids, typically under the control of transcriptional induction elements (e.g., LacI repressor). Expression levels need to be optimized so that yields of folded, active enzymes are high (e.g., over a million copies per cell). This is especially important for single-cell assays as outlined in section “Single-Cell Phenotypic Variations.” This way, libraries with up to 108 cells expressing unique enzymes can be created for screening.

Single-Cell Phenotypic Variations Expression levels of active, correctly folded enzymes can vary from cell to cell in a cell population due to stochastic biological processes. These variations are observed during single-cell screens, especially for short-reaction times, not allowing to reach completion. High levels of variance have been reported for expression of single genes in vivo and in vitro; for example, a coefficient of variance of 44% was reported for the expression of single GFP genes in picoliter droplets (Courtois et al. 2008). This can be highly problematic for the detection of weakly active or poorly expressed enzymes because such phenotypic variations may bring product formation below a selection threshold leading to missed hits. One solution is to overscreen a library (i.e., by rescreening the same mutants multiple times or by constructing libraries with multiple copies of each mutant) to maximize the chances to recover all the best enzymes (Kintses et al. 2012). Sensitivity of Fluorescence Detection The integration of high-sensitivity optical detection of individual microfluidic droplets has enabled monitoring of enzyme activity across ultralarge numbers of individual reactions. Fluorescence detection has been historically popular because of its high sensitivity (down to 2 nM product) and low integration times (in microseconds) which enable real-time measurement of low levels of products (~10,000 molecules in a 10 pL droplet). In a classical fluorescence detection setup, a laser is tightly focused by a microscope objective onto flowing microdroplets containing a fluorescent product. The fluorescence photons are then collected through the same objective and quantified by photodetectors. For enzyme where a fluorogenic substrate is not available or not suitable for a specific application, other modes of detection allowing detection of alternative types

682

F. Gielen

of products have been developed including absorbance, mass spectrometry, or Raman scattering (Gielen et al. 2016; Wang et al. 2017b; Holland-Moritz et al. 2020). In all cases, an expression system with high yields of active mutants is desirable. This ensures signal-to-noise is maximized and enhances the quality of the selection step. An alternative to increase product formation includes the addition of a cellular growth step during which bacteria/yeasts can divide and form a clonal population in every droplet. For example, E. coli can reach an optical density of 30 (i.e., corresponding to several thousand cells in 170 pL droplets) (Mahler et al. 2015). Growing cell populations within droplets results in a multiplied number of produced enzymes and therefore to an amplification in product formation which facilitates detection, which is especially useful to identify weakly active mutants.

Operation of a Microdroplet Sorter Isolation of improved mutants at high throughput enables the retrieval and sequencing of their genes and their further functional characterization. It is therefore an essential step in a directed evolution workflow. Conventional flow cytometers can be used in conjunction with microdroplets by converting them into water-in-oil-inwater (double) emulsions, effectively transforming them into artificial cells. This only necessitates the injection of the first emulsion into a flow-focusing device with a hydrophilic surface coating where the second aqueous phase gets added. Using such technique, a study has demonstrated extraction of an active arylsulfatase diluted in a million-fold excess of inactive mutants (Zinchenko et al. 2014). Other microfluidic formats amenable to FACS sort include gel-shell beads in which enzyme, DNA, and fluorescent products are enclosed in hydrogels (Fischlechner et al. 2014). Alternatively, custom-built single emulsion sorters are used but currently necessitate specialist knowledge in optics, electronics, and software engineering as they are not widely commercialized to date. They mostly rely on dielectrophoretic forces created by high-voltage electric pulses applied on to flowing droplets (Fig. 6).

Fig. 6 A high-throughput fluorescence-activated microdroplet sorter (FADS): A preformed microfluidic emulsion is injected into the sorting device where they reach a “Y” shaped junction. In the absence of actuation, droplets flow to a waste (bottom) channel based on local channel geometries. The fluorescence content of every droplet is monitored real-time by a software interfaced with a pulse generator that sends a high voltage pulse (10 kHz, 500 Vpp) to electrodes embedded in the microfluidic device (appearing black in the figure). This short pulse can deflect a single droplet (indicated by a red arrow) to the top sorting channel that can be collected and recovered in a tube. (Adapted from Kintses et al. (2012) with permission from Elsevier Ltd.)

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

683

The high speed of electronic and electric field propagation enables sorting of single droplets at up to 30 kHz (Sciambi and Abate 2015). The key challenge is to link a functional read-out (e.g., a fluorescence signal arising from a reaction product) to the actuation of electrodes located close to a sorting Y-junction (Fig. 3-3). Importantly, the accurate quantification of product formation by fluorescence signals typically requires uniform distribution of a fluorophore within a droplet (Mazutis et al. 2013).

Formats for Single-Cell Biocatalyst Functional Screens Typically, single bacterial or fungal cells expressing enzymes are compartmentalized in individual droplets of ~10 pL volume (Ø ~30 microns), allowing screening of enzymes expressed intracellularly (in the cytoplasm or periplasm of gram-negative bacteria), on the surface of cells or secreted from cells. These main types of singlecell assays are illustrated in Fig. 7.

Single-Cell Lysate Assays Accessibility of the enzymes with enzymatic substrate molecules is necessary to achieve chemical transformation and release products. This challenge can be solved by releasing internally expressed enzymes by cell lysis. Classical agents for cell lysis include surfactants, antibiotics, chelating agents, or combinations (Obexer et al. 2017). It is imperative that such reagents should not induce droplet coarsening by interaction with the protective surfactant molecules at the oil/water interface nor be detrimental to enzyme stability. The lysis reagents typically include a mixture of enzymes (lysozyme), antibiotics (polymyxin B,), chelating agents (EDTA or EGTA), and mild detergents. Lytic agents are then mixed with substrate in the droplet generation device (Fig. 4-1) so that lysis is initiated as soon as droplets are made.

Fig. 7 Four major types of single-cell assays for biocatalyst screens in microdroplets: For each format, substrate and enzyme are brought into contact by different means. (i) Cell-lysate assay: Internally expressed enzymes are released into droplets; (ii) The substrate can diffuse passively through cellular membranes resulting in its intake and conversion by internalized enzymes; (iii) cell -surface display, in which enzymes remain attached to the outer membrane of cells; and (iv) cellsecretion assays, in which enzymes are released into the droplet

684

F. Gielen

Mixing within microdroplets is efficient and typically occurs in millisecond timescales (Jiang et al. 2012). Examples of evolution campaigns using the cell-lysate format are given below in chronological order, and detailed protocols used for screening single-cell lysates are available from reference (Gielen et al. 2018). • In reference (Kintses et al. 2012), an arylsulfatase is evolved toward nonnative phosphonate substrate with improvement of sixfold in expression and sixfold in activity. These improvements occur over three rounds of evolution where 107 droplets were screened in each. • Romero et al. have combined deep sequencing with microfluidic screening to perform deep mutational scanning, which allows to map sequence-function relationships for a very large number of mutants (Romero et al. 2015). A single-cell lysate format was used to isolate glycosidases with improved activity from an error-prone library with an average of 3.8 amino acid substitutions per gene. 3.4 million droplets with measurable enzymatic activity were sorted and deep sequenced to identify site-specific mutational tolerance. They applied the same method to probe enzyme thermostability. • A microdroplet absorbance detector and sorter has been developed to evolve a phenylalanine dehydrogenase, improve it fourfold for cell-lysate activity, and increase its expression levels by 60% and its melting temperature by 12 °C with two rounds of evolution (Gielen et al. 2016). A first round enabled enrichment from an error-prone library, and a second round identified beneficial combinations of the first-round mutations with a shuffling library. • A cell lysate in double emulsion strategy has been used to engineer polymerases active on unnatural genetic polymer (Larsen et al. 2016). The authors first engineered a version of a DNA polymerase isolated from Thermococcus sp. 9° N (9n-GLK), which, together with a wild-type polymerase, was challenged to extend a DNA primer–template complex with deoxyribonucleoside triphosphates (dNTP) and ribonucleoside triphosphates (NTP). They developed a fluorescent reporter system that produces an optical signal when a primer–template complex is extended to full-length product based on a donor-quencher pair. They used a target occupancy of 0.1 for E. coli expressing the polymerases, and cell lysis was promoted by heating the droplets for 3 h at 55 °C. Droplets containing the 9n-GLK E. coli strain produced a highly fluorescent signal and could be enriched 1200-fold when using a 1/10,000th active to inactive enzyme dilution. The method was used to evolve polymerases that can synthesize artificial genetic polymers with a backbone structure unrelated to natural DNA and RNA. • Obexer et al. used droplet-based microfluidic screening platform to improve a previously optimized artificial aldolase by an additional factor of 30 (Obexer et al. 2017). The aldolase had been extensively modified over 13 rounds of conventional-directed evolution, but further optimization proved difficult and microfluidic screening was shown to rescue the stalled evolution process. Libraries were created by a combination of error-prone PCR, cassette mutagenesis, and

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

685

DNA shuffling. The authors used in-line droplet generation, on-chip incubation, and fluorescence-activated sorting in a single device.

Single-Cell Internal Expression and Surface Display The ability of some substrates/products to cross bacterial cell membranes means that enzymes expressed within a cell can be brought into contact with substrate molecules without the need to release enzymes by cellular lysis (as depicted in Fig. 7-ii). Small reaction products can subsequently diffuse outside the cell and be retained in the droplet. One of the first examples of enzymatic microbial assay performed in picoliter volumes has been shown by Huebner et al. in 2008 (Huebner et al. 2008). In this study, an alkaline phosphatase enzyme was expressed in the periplasm of E.coli (alkaline phosphatase has a leader peptide sequence that naturally leads to its export to the periplasm), and the fluorogenic substrate 3-O-methylfluorescein phosphate was used. It has to be pointed out that such strategies leave uncertainty about the partitioning of substrate between cellular compartments and extracellular space. Alternatively, methods have been developed to display recombinant proteins on the surface of cells (Fig. 7-iii). Using Saccharomyces cerevisiae yeasts as host, proteins in controlled numbers can be displayed (this is generically referred to as “yeast display”) via genetic fusion to an abundant cell-wall protein and is a widely adopted tool for protein engineering. Given the practicality of enzyme/substrate accessibility in such formats, it is unsurprising the first microdroplet-based evolution campaign was done using yeast surface display of horseradish peroxidase (Agresti et al. 2010). Using high-stringency screening, the authors identified several significantly improved mutants, some approaching diffusion-limited efficiency from 108 individual enzyme reactions in 10 h. Other display systems have been developed from ribosomes, mRNA, the surface of phage, bacteria, mammalian, insect, and cells (Cherf and Cochran 2015). An E. coli autodisplay system has been recently developed by van Loo et al. and enables enzyme display on the cell surface. The authors demonstrated evolution of an arylsulfatase for sulfatase activity by over sixfold for 4-nitrophenyl sulfate and 30-fold for fluorescein disulfate (van Loo et al. 2019). A side-by-side comparison of cytoplasmic, periplasmic, and inner membrane display systems with an L-asparagine substrate has been done for an L-Asparaginase, showing that periplasmic expression produced good yields and good accessibility to the substrate L-asparagine (Karamitros et al. 2020). The authors implemented a coupled assay where bacterial L-asparaginase catalyzes the hydrolysis of L-asparagine to L-aspartate and ammonium. L-aspartate is then oxidized by aspartate oxidase and the coenzyme flavin adenine dinucleotide (FAD) to form oxaloacetate and hydrogen peroxide. Hydrogen peroxide is finally used by horseradish peroxidase and coupled to the production of the fluorescent resorufin. Both substrate and resorufin can diffuse in and out the cells so that the fluorescent product gets uniformly spread in the droplets. Importantly, activity for enzymes expressed in

686

F. Gielen

the cytoplasm was estimated to be 10–40 times slower compared to the expression in the periplasm suggesting substrate/enzyme diffusion was greatly hampered by the presence of internal membranes.

Single-Cell Secretion Assays Certain microorganisms are able to efficiently secrete proteins, and this property can be used to controllably export enzymes. This type of assay, illustrated in Fig. 7-iv, conveniently circumvents lysis or substrate diffusion limitations. One such organism is filamentous fungi naturally secreting large amounts of hydrolytic enzymes, a trait necessary to their role as decomposers. Enzyme examples include amylases, cellulases, or proteases that are involved in the biochemical degradation of biomass (Su et al. 2012). For example, filamentous fungi Aspergillus niger have been used to enrich 200-fold for active alpha-amylases (Beneyton et al. 2016). Because of germination and mycelium growth, single fungi have to be encapsulated in relatively large 10 nL (Ø 260 microns) droplets to prevent hyphal tips from triggering droplet coalescence. Yeasts have the additional advantage of being able to perform eukaryotic posttranslational processing (e.g., glycosylation, folding, sequence cleavage, etc.) and can therefore produce and secrete heterologous eukaryotic proteins in their active form. Some other nonconventional organisms have been tested, e.g., the yeast Yarrowia lipolytica which can use unusual carbon sources (Beneyton et al. 2017). Screening of Cell Populations Identifying microorganisms overproducing active enzymes is an important goal for white biotechnology. Cell populations can be assessed individually for their biocatalyst production proficiency using single cell microencapsulation. A study isolated xylose-consuming yeast strains diluted in an excess of nonconsumers (Wang et al. 2014) and enriched L-lactate-producing Escherichia coli clones 5800 times from a population containing one L-lactate producer per 104 D-lactate producers. A miniaturized assay for screening cellulose-producing yeast cells has also been developed (Ostafe et al. 2014). Enzyme mutants can be produced by UV irradiation of whole cells that introduces mutations randomly throughout whole genomes. Using such method, Sjostrom et al. have demonstrated 14-fold enrichment of a yeast strain for alpha-amylase activity (Sjostrom et al. 2014). In this study, the authors used a 105 members-yeast-cell library and isolated a yeast clone with a more than twofold increase in alpha-amylase production by screening 3
106 droplets. In contrast to evolution studies focusing on improving enzymatic fitness, such studies evolve organismal fitness for improved production yields.

Functional Metagenomics and Bioprospecting Aside from screening enzyme-mutant libraries, novel enzymes can be discovered from environmental sources. This is possible thanks to the ability of enzymes to

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

687

catalyze multiple reactions (substrate promiscuity). Substrate promiscuity is a common trait for most enzymes and thought to be a key to enzyme evolvability and improvement of suboptimal functions (Pandya et al. 2014; Khersonsky and Tawfik 2010). This also represents an opportunity to discover new catalysts from the environment for which it is difficult to know their native substrate. As the vast majority (99%) of bacterial communities is not culturable, one way to tap into their biocatalytic repertoire is to perform functional metagenomic screens. Such screens can be miniaturized and performed at the single-cell level where environmental genes are expressed within a host as described in the previous paragraphs (Colin et al. 2015). The general procedure for screening environmental DNA samples is shown in Fig. 8. The main difference is that environmental samples must be processed to be transformed into purified plasmids for use in model-host organisms. Alternatively, it is possible to use large insert vectors such as fosmids as carriers that allow the construction of metagenomic libraries with lower number of clones and higher success rate in isolation of genes encoding for new enzymes, complete metabolic pathways, bioactive compounds, and polycistronic genes (Daniel 2005). Transformation from environmental samples to a format ready for single-cell screening includes DNA fragmentation, amplification, and purification techniques. Insert sizes can be selected to be a certain length (e.g., >3 kb) to maximize the probabilities of encoding for full-size proteins (Schmitz et al. 2008). Expression of heterologous genes remains a challenge which is currently partially alleviated by the use of long-induction times (days) for maximizing enzyme yields or using improvedexpression plasmids (Westmann et al. 2018). Bacteria can be directly prospected for the environment and individually screened in microdroplets without the need to culture them beforehand. For example, Najah et al. screened uncultured bacteria from a wheat stubble field for their ability to hydrolyze cellulosic biomass (Najah et al. 2014). After selection based on cellobiohydrolase activity, the authors isolated bacteria with respectively 17- and 7-fold higher cellobiohydrolase and endogluconase activity. They showed that this type of screen resulted in very different taxonomic diversity compared to growth-based selections.

In Vitro Workflows as Artificial Single Cells Instead of relying on microbial factories, individual genes can be expressed in vitro using cell-free protein synthesis reactions where all necessary components needed for in vitro transcription and translation (IVTT) are present, usually in the form of cell-free extracts or purified components. Cell-free expression systems can find use in the screening of proteins toxic to host cells, the addition of nonnatural amino acids, and in general, to apply a more controlled selection pressure. IVTT enables bypassing of cloning and DNA transformation steps needed when using host cells. In vitro workflows usually involve a DNA amplification step from single genes encapsulated in individual droplets followed by IVTT. In one study, a single DNA template encoding for green fluorescence protein (GFP) generated 40,000 GFP copies in a single 1.8 pL droplet (Courtois et al. 2008). Another study has demonstrated a 500-fold enrichment for a β-galactosidase enzyme by coupling PCR

Fig. 8 Discovery of novel enzymes with functional metagenomic single-cell screens in microfluidic droplets. General procedure. (1) Environmental DNA (eDNA) is cloned into a plasmid and transformed into a cell host. (2) Single-host cells are encapsulated into water-in-oil droplets together with substrate and lysis agents. (3) Emulsion droplets are incubated off-chip; after single cell lysis, cytoplasmically expressed protein catalysts are able to turn over substrate. The arrow designates the droplets accumulating above the fluorous oil phase. (4) Emulsion droplets are reinjected into a sorting chip and strongly fluorescent droplets (“+” channel) are separated from those with fluorescence below the threshold (““ channel) by dielectrophoresis. (5) Selected droplets are demulsified and plasmid DNA recovered following by retransformation into the host cell. For further enrichment, iterative selections can be performed. (6) Plasmids containing eDNA coding for active catalysts are sequenced. (Reproduced from Colin et al. (2015))

688 F. Gielen

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

689

amplification, cell-free expression, and fluorescence-activated droplet sorting (Fallah-Araghi et al. 2012). Selection of lacZ genes encoding beta-galactosidase and lacZmut genes encoding an inactive variant was demonstrated at 2000 droplets per second with a 502-fold enrichment. IVTT addition relies on a droplet-to-droplet addition process, and the sorting module operates based on electro-coalescence with a continuous aqueous stream.

Limitations and Future Prospects of Single-Cell Microencapsulation Beyond applications in biocatalysis, microencapsulation technologies are enabling powerful analyses of single cells across large-cell populations, from -omics studies to the screening of drug libraries. Concomitantly, increasingly complex reagent mixtures must be added to single cells during or postencapsulation. A prominent example is single-cell transcriptomics studies requiring the capture of mRNA molecules on microbeads coencapsulated with single cells (Macosko 2015 #74). Likewise, small molecule drug libraries can be encoded with DNA-tagging techniques on compound beads so that high-throughput cellular encapsulation becomes compatible with drug-discovery formats (MacConnell 2017 #76). Invariably, controlled and quantitative dosage of multielement mixtures remains challenging. Stochastic encapsulation is still limiting the biological throughput (the actual number of useful microreactors) compared to the physical throughput (the overall number of microreactors). A second hurdle is ensuring the homogeneity and integrity of suspensions of cells and other reagents. Indeed, one should avoid the presence of lysed cells in any given sample to encapsulate that would otherwise contribute to biological noise. This requires the development of more robust cell-handling techniques that preserve cells when transferred away from culturing flasks. Preserving longer-term cellular viability (over days) will be crucial to unlock applications involving slow reactions or biological processes. Looking forward, the encoding of single cells during encapsulation with genetic or optical barcodes will enable the deciphering of the makeup of entire cell populations, enabling the detailed mapping of phenotypic heterogeneities, including the identification of rare biological events. In a postgenomic era, the dissection of gene analyses of single cells from ex vivo biological samples will become increasingly routine, e.g., for use in cancer biopsies. Another area of application for microencapsulation is in vitro disease modelling with the growth of isogenic 3D cell cultures originating from single cells trapped in hydrogel scaffold. These models are becoming increasingly complex in the makeup of cellular microniches (e.g., cocultures of multiple cell types) that aim to ultimately faithfully mimic physiological conditions and close the gap with in vivo experiments.

Conclusions Single-cell microencapsulation technology holds the promise to become a generic biocatalyst screening platform. However lingering limitations remain. For instance, the range of assays and expression hosts that have been up to now implemented in

690

F. Gielen

miniature format remains limited. One major hurdle is to develop assays for which components do not diffuse in between droplets (e.g., leading to an equilibrium where all droplets have equimolar amounts of products). This is being addressed by the development of optimized chemical structure for surfactants and chemically altered substrates. Furthermore, to unlock the full potential of microfluidics for enzyme discovery and engineering, assays need to be performed in conditions relevant to industrial application, such as unlabeled, natural substrates and the use of realistic assay matrices. To this end, development of enzymatic cascades and coupled reactions is well underway to study series of biotransformations (Karamitros et al. 2020). Large efforts have been expanded to develop the technology toolbox in terms of fluidic manipulation and breadth of analytical read-out. The latter has driven the development of various types of readouts beyond fluorescence for droplet microfluidics such as fluorescence lifetime, anisotropy, bright-field imaging, Raman scattering, or mass spectrometry (Mair et al. 2017; Holland-Moritz et al. 2019; Wang et al. 2017a; Hung et al. 2020; Gielen et al. 2017). Simultaneous measurements are also becoming mainstream, lending the possibility to evolve catalysts for several phenotypic traits at the same time (coevolution) (Tovar et al. 2019). From a bioprospecting perspective, there is currently a trove of sequencing data being harvested from the environment, but efforts to characterize the proteins are lagging: functional tests are needed to quantify their proficiency at catalyzing specific reactions. This can be done at scale using microencapsulation methods. As an ever-increasing number of enzymes can be subjected to ultrahigh-throughput screening, single-cell microdroplet screens will become an increasingly important technique for protein engineering, enzymology, synthetic biology, and biochemical research.

References Abate AR, Hung T, Mary P, Agresti JJ, Weitz DA (2010) High-throughput injection with microfluidics using picoinjectors. Proc Natl Acad Sci U S A 107(45):19163–19166. https://doi.org/10. 1073/pnas.1006888107 Agresti JJ, Antipov E, Abate AR, Ahn K, Rowat AC, Baret JC, Marquez M, Klibanov AM, Griffiths AD, Weitz DA (2010) Ultrahigh-throughput screening in drop-based microfluidics for directed evolution. Proc Natl Acad Sci U S A 107(9):4004–4009. https://doi.org/10.1073/pnas. 0910781107 Beneyton T, Wijaya IPM, Postros P, Najah M, Leblond P, Couvent A, Mayot E, Griffiths AD, Drevelle A (2016) High-throughput screening of filamentous fungi using nanoliter-range droplet-based microfluidics. Sci Rep 6:27223. https://doi.org/10.1038/srep27223 Beneyton T, Thomas S, Griffiths AD, Nicaud JM, Drevelle A, Rossignol T (2017) Droplet-based microfluidic high-throughput screening of heterologous enzymes secreted by the yeast Yarrowia lipolytica. Microb Cell Fact 16:18. https://doi.org/10.1186/s12934-017-0629-5 Chen KQ, Arnold FH (1993) Tuning the activity of an enzyme for unsual environments - sequential random mutagenesis of subtilisin-E for catalysis in dimethylformamide. Proc Natl Acad Sci U S A 90(12):5618–5622. https://doi.org/10.1073/pnas.90.12.5618

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

691

Cherf GM, Cochran JR (2015) Applications of yeast surface display for protein engineering. In: Liu B (ed) Yeast surface display: methods, protocols, and applications, Methods in molecular biology, vol 1319. Humana Press Inc, Totowa, pp 155–175. https://doi.org/10.1007/978-14939-2748-7_8 Colin PY, Kintses B, Gielen F, Miton CM, Fischer G, Mohamed MF, Hyvonen M, Morgavi DP, Janssen DB, Hollfelder F (2015) Ultrahigh-throughput discovery of promiscuous enzymes by picodroplet functional metagenomics. Nat Commun 6:10008. https://doi.org/10.1038/ ncomms10008 Collins DJ, Neild A, deMello A, Liu AQ, Ai Y (2015) The Poisson distribution and beyond: methods for microfluidic droplet production and single cell encapsulation. Lab Chip 15(17):3439–3459. https://doi.org/10.1039/c5lc00614g Courtois F, Olguin LF, Whyte G, Bratton D, Huck WTS, Abell C, Hollfelder F (2008) An integrated device for monitoring time-dependent in vitro expression from single genes in picolitre droplets. ChemBioChem 9(3):439–446. https://doi.org/10.1002/cbic.200700536 Cuesta SM, Rahman SA, Furnham N, Thornton JM (2015) The classification and evolution of enzyme function. Biophys J 109(6):1082–1086. https://doi.org/10.1016/j.bpj.2015.04.020 Daniel R (2005) The metagenomics of soil. Nat Rev Microbiol 3(6):470–478. https://doi.org/10. 1038/nrmicro1160 Dorr M, Fibinger MPC, Last D, Schmidt S, Santos-Aberturas J, Bottcher D, Hummel A, Vickers C, Voss M, Bornscheuer UT (2016) Fully automatized high-throughput enzyme library screening using a robotic platform. Biotechnol Bioeng 113(7):1421–1432. https://doi.org/10.1002/bit. 25925 Fallah-Araghi A, Baret JC, Ryckelynck M, Griffiths AD (2012) A completely in vitro ultrahighthroughput droplet-based microfluidic screening system for protein engineering and directed evolution. Lab Chip 12(5):882–891. https://doi.org/10.1039/c2lc21035e Fischlechner M, Schaerli Y, Mohamed MF, Patil S, Abell C, Hollfelder F (2014) Evolution of enzyme catalysts caged in biomimetic gel-shell beads. Nat Chem 6(9):791–796. https://doi.org/ 10.1038/nchem.1996 Gielen F, Hours R, Emond S, Fischlechner M, Schell U, Hollfelder F (2016) Ultrahigh-throughput– directed enzyme evolution by absorbance-activated droplet sorting (AADS). Proc Natl Acad Sci 113(47):E7383–E7389. https://doi.org/10.1073/pnas.1606927113 Gielen F, Butz M, Rees EJ, Erdelyi M, Moschetti T, Hyvonen M, Edel JB, Kaminski CF, Hollfelder F (2017) Quantitative affinity determination by fluorescence anisotropy measurements of individual nanoliter droplets. Anal Chem 89(2):1092–1101. https://doi.org/10.1021/acs. analchem.6b02528 Gielen F, Colin P-Y, Mair P, Hollfelder F (2018) Ultrahigh-throughput screening of single-cell lysates for directed evolution and functional metagenomics. Methods Mol Biol (Clifton, NJ) 1685:297–309. https://doi.org/10.1007/978-1-4939-7366-8_18 Hasan S, Geissler D, Wink K, Hagen A, Heiland JJ, Belder D (2019) Fluorescence lifetimeactivated droplet sorting in microfluidic chip systems. Lab Chip 19(3):403–409. https://doi. org/10.1039/c8lc01278d Holland-Moritz DA, Wismer M, Mann B, Farasat I, Guetschow E, Sun S, Moore J, Kennedy R, Mangion I, Welch CJ, Devine P (2019) Mass activated droplet sorting (MADS) enables high throughput screening of enzymatic reactions at nanoliter scale. Angew Chem Int Ed Engl. https://doi.org/10.1002/anie.201913203 Holland-Moritz DA, Wismer MK, Mann BF, Farasat I, Devine P, Guetschow ED, Mangion I, Welch CJ, Moore JC, Sun SW, Kennedy RT (2020) Mass activated droplet sorting (MADS) enables high-throughput screening of enzymatic reactions at nanoliter scale. Angew Chem-Int Edit:59, 4470. https://doi.org/10.1002/anie.201913203 Huebner A, Olguin LF, Bratton D, Whyte G, Huck WTS, de Mello AJ, Edel JB, Abell C, Hollfelder F (2008) Development of quantitative cell-based enzyme assays in microdroplets. Anal Chem 80(10):3890–3896. https://doi.org/10.1021/ac800338z

692

F. Gielen

Hung S-T, Mukherjee S, Jimenez R (2020) Enrichment of rare events using a multi-parameter high throughput microfluidic droplet sorter. Lab Chip 20:834. https://doi.org/10.1039/c9lc00790c Jiang LG, Zeng Y, Zhou HB, Qu JAY, Yao SH (2012) Visualizing millisecond chaotic mixing dynamics in microdroplets: a direct comparison of experiment and simulation. Biomicrofluidics 6(1):12. https://doi.org/10.1063/1.3673254 Karamitros CS, Morvan M, Vigne A, Lim J, Gruner PJ, Beneyton T, Vrignon J, Konrad M, Baret JC (2020) Bacterial expression systems for enzymatic activity in droplet-based microfluidics. Anal Chem. https://doi.org/10.1021/acs.analchem.9b04969 Kemna EW, Schoeman RM, Wolbers F, Vermes I, Weitz DA, Van Den Berg A (2012) High-yield cell ordering and deterministic cell-in-droplet encapsulation using Dean flow in a curved microchannel. Lab Chip 12(16):2881–2887 Khersonsky O, Tawfik DS (2010) Enzyme promiscuity: a mechanistic and evolutionary perspective. In: Kornberg RD, Raetz CRH, Rothman JE, Thorner JW (eds) Annual review of biochemistry, Annual review of biochemistry, vol 79. Annual Reviews, Palo Alto, pp 471–505. https:// doi.org/10.1146/annurev-biochem-030409-143718 Kintses B, Hein C, Mohamed MF, Fischlechner M, Courtois F, Leine C, Hollfelder F (2012) Picoliter cell lysate assays in microfluidic droplet compartments for directed enzyme evolution. Chem Biol 19(8):1001–1009. https://doi.org/10.1016/j.chembiol.2012.06.009 Koster S, Angile FE, Duan H, Agresti JJ, Wintner A, Schmitz C, Rowat AC, Merten CA, Pisignano D, Griffiths AD, Weitz DA (2008) Drop-based microfluidic devices for encapsulation of single cells. Lab Chip 8(7):1110–1115. https://doi.org/10.1039/b802941e Lagus TP, Edd JF (2013) High-throughput co-encapsulation of self-ordered cell trains: cell pair interactions in microdroplets. RSC Adv 3(43):20512–20522. https://doi.org/10.1039/ c3ra43624a Larsen AC, Dunn MR, Hatch A, Sau SP, Youngbull C, Chaput JC (2016) A general strategy for expanding polymerase function by droplet microfluidics. Nat Commun 7(9). https://doi.org/10. 1038/ncomms11235 Liu X, Painter RE, Enesa K, Holmes D, Whyte G, Garlisi CG, Monsma FJ, Rehak M, Craig FF, Smith CA (2016) High-throughput screening of antibiotic-resistant bacteria in picodroplets. Lab Chip 16(9):1636–1643. https://doi.org/10.1039/c6lc00180g MacConnell AB, Price AK, Paegel BM (2017) An Integrated Microfluidic Processor for DNAEncoded Combinatorial Library Functional Screening. ACS Comb Sci 19:181–192. https://doi. org/10.1021/acscombsci.6b00192 Macosko EZ, et al. (2015) Highly parallel genome-wide expression profiling of individual cells using nanoliter droplets. Cell 161:1202–1214. https://doi.org/10.1016/j.cell.2015.05.002 Mahler L, Tovar M, Weber T, Brandes S, Rudolph MM, Ehgartner J, Mayr T, Figge MT, Roth M, Zang E (2015) Enhanced and homogeneous oxygen availability during incubation of microfluidic droplets. RSC Adv 5(123):101871–101878. https://doi.org/10.1039/c5ra20118g Mair P, Gielen F, Hollfelder F (2017) Exploring sequence space in search of functional enzymes using microfluidic droplets. Curr Opin Chem Biol 37:137–144. https://doi.org/10.1016/j.cbpa. 2017.02.018 Mazutis L, Gilbert J, Ung WL, Weitz DA, Griffiths AD, Heyman JA (2013) Single-cell analysis and sorting using droplet-based microfluidics. Nat Protoc 8(5):870–891. https://doi.org/10.1038/ nprot.2013.046 Miller OJ, Bernath K, Agresti JJ, Amitai G, Kelly BT, Mastrobattista E, Taly V, Magdassi S, Tawfik DS, Griffiths AD (2006) Directed evolution by in vitro compartmentalization. Nat Methods 3(7):561–570. https://doi.org/10.1038/nmeth897 Najah M, Calbrix R, Mahendra-Wijaya IP, Beneyton T, Griffiths AD, Drevelle A (2014) Dropletbased microfluidics platform for ultra-high-throughput bioprospecting of cellulolytic microorganisms. Chem Biol 21(12):1722–1732. https://doi.org/10.1016/j.chembiol.2014.10.020 Neun S, Kaminski TS, Hollfelder F (2019) Single-cell activity screening in microfluidic droplets. Methods Enzymol 628:95–112. https://doi.org/10.1016/bs.mie.2019.07.009

29

Single-Cell Microencapsulation for Evolution and Discovery of Biocatalysts

693

Neylon C (2004) Chemical and biochemical strategies for the randomization of protein encoding DNA sequences: library construction methods for directed evolution. Nucleic Acids Res 32(4):1448–1459. https://doi.org/10.1093/nar/gkh315 Obexer R, Godina A, Garrabou X, Mittl PRE, Baker D, Griffiths AD, Hilvert D (2017) Emergence of a catalytic tetrad during evolution of a highly active artificial aldolase. Nat Chem 9(1):50–56. https://doi.org/10.1038/nchem.2596 Olsen MJ, Stephens D, Griffiths D, Daugherty P, Georgiou G, Iverson BL (2000) Function-based isolation of novel enzymes from a large library. Nat Biotechnol 18(10):1071–1074. https://doi. org/10.1038/80267 Ostafe R, Prodanovic R, Ung WL, Weitz DA, Fischer R (2014) A high-throughput cellulase screening system based on droplet microfluidics. Biomicrofluidics 8(4):4. https://doi.org/10. 1063/1.4886771 Pandya C, Farelli JD, Dunaway-Mariano D, Allen KN (2014) Enzyme promiscuity: engine of evolutionary innovation. J Biol Chem 289(44):30229–30236. https://doi.org/10.1074/jbc.R114. 572990 Romero PA, Arnold FH (2009) Exploring protein fitness landscapes by directed evolution. Nat Rev Mol Cell Biol 10(12):866–876. https://doi.org/10.1038/nrm2805 Romero PA, Tran TM, Abate AR (2015) Dissecting enzyme function with microfluidic-based deep mutational scanning. Proc Natl Acad Sci U S A 112(23):7159–7164. https://doi.org/10.1073/ pnas.1422285112 Schmitz JE, Daniel A, Collin M, Schuch R, Fischetti VA (2008) Rapid DNA library construction for functional genomic and metagenomic screening. Appl Environ Microbiol 74(5):1649–1652. https://doi.org/10.1128/aem.01864-07 Sciambi A, Abate AR (2015) Accurate microfluidic sorting of droplets at 30 kHz. Lab Chip 15(1):47–51. https://doi.org/10.1039/c4lc01194e Sjostrom SL, Bai YP, Huang MT, Liu ZH, Nielsen J, Joensson HN, Svahn HA (2014) Highthroughput screening for industrial enzyme production hosts by droplet microfluidics. Lab Chip 14(4):806–813. https://doi.org/10.1039/c3lc51202a Su X, Schmitz G, Zhang M, Mackie RI, Cann IK (2012) Heterologous gene expression in filamentous fungi. Adv Appl Microbiol 81:1–61. https://doi.org/10.1016/B978-0-12-3943828.00001-0 Tawfik DS, Griffiths AD (1998) Man-made cell-like compartments for molecular evolution. Nat Biotechnol 16(7):652–656. https://doi.org/10.1038/nbt0798-652 Tovar M, Hengoju S, Weber T, Mahler L, Choudhary M, Becker T, Roth M (2019) One sensor for multiple colors: fluorescence analysis of microdroplets in microbiological screenings by frequency-division multiplexing. Anal Chem 91(4):3055–3061. https://doi.org/10.1021/acs. analchem.8b05451 Tu R, Martinez R, Prodanovic R, Klein M, Schwaneberg U (2011) A flow cytometry-based screening system for directed evolution of proteases. J Biomol Screen 16(3):285–294. https:// doi.org/10.1177/1087057110396361 Uchiyama T, Miyazaki K (2009) Functional metagenomics for enzyme discovery: challenges to efficient screening. Curr Opin Biotechnol 20(6):616–622. https://doi.org/10.1016/j.copbio.2009.09.010 van Loo B, Heberlein M, Mair P, Zinchenko A, Schuurmann J, Eenink BDG, Holstein JM, Dilkaute C, Jose J, Hollfelder F, Bornberg-Bauer E (2019) High-throughput, lysis-free screening for sulfatase activity using escherichia coli autodisplay in microdroplets. ACS Synth Biol 8(12):2690–2700. https://doi.org/10.1021/acssynbio.9b00274 van Vliet LD, Colin PY, Hollfelder F (2015) Bioinspired genotype-phenotype linkages: mimicking cellular compartmentalization for the engineering of functional proteins. Interface Focus 5(4). https://doi.org/10.1098/rsfs.2015.0035 Wang BL, Ghaderi A, Zhou H, Agresti J, Weitz DA, Fink GR, Stephanopoulos G (2014) Microfluidic high-throughput culturing of single cells for selection based on extracellular metabolite production or consumption. Nat Biotechnol 32(5):473–U194. https://doi.org/10.1038/nbt.2857

694

F. Gielen

Wang X, Ren L, Su Y, Ji Y, Liu Y, Li C, Li X, Zhang Y, Wang W, Hu Q, Han D, Xu J, Ma B (2017a) Raman-activated droplet sorting (RADS) for label-free high-throughput screening of microalgal single-cells. Anal Chem 89(22):12569–12577. https://doi.org/10.1021/acs.analchem.7b03884 Wang XX, Ren LH, Su YT, Ji YT, Liu YP, Li CY, Li XR, Zhang Y, Wang W, Hu Q, Han DX, Xu J, Ma B (2017b) Raman-activated droplet sorting (RADS) for label-free high-throughput screening of microalgal single-cells. Anal Chem 89(22):12569–12577. https://doi.org/10.1021/acs. analchem.7b03884 Westmann CA, Alves LD, Silva-Rocha R, Guazzaroni ME (2018) Mining novel constitutive promoter elements in soil metagenomic libraries in escherichia coli. Front Microbiol 9:15. https://doi.org/10.3389/fmicb.2018.01344 Yang KK, Wu Z, Arnold FH (2019) Machine-learning-guided directed evolution for protein engineering. Nat Methods 16(8):687–694. https://doi.org/10.1038/s41592-019-0496-6 Zinchenko A, Devenish SRA, Kintses B, Colin PY, Fischlechner M, Hollfelder F (2014) One in a million: flow cytometric sorting of single cell-lysate assays in monodisperse picolitre double emulsion droplets for directed evolution. Anal Chem 86(5):2526–2533. https://doi.org/10.1021/ ac403585p

Analytical Techniques for Single-Cell Studies in Microbiology

30

Evgeny Puchkov

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flow Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Imaging Flow Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scanning Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantitative Optical Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Brightfield Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluorescence Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scanning Probe Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scanning Atomic Force Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scanning Electrochemical Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanoscale Secondary Ion Mass Spectrometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rotational-Vibrational Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

696 697 697 697 698 699 701 703 709 709 711 713 715 718 718

Abstract

Technological advances over the past 30 years have led to the improvement of existing and the development of new analytical techniques, which have very high sensitivity and/or spatial-temporal resolution. Moreover, many of them can be implemented on commercially available devices. All this has created the conditions for obtaining fundamentally new quantitative information about the physical and chemical properties of single cells of various origins, including such tiny ones as yeasts and bacteria, the size of which can be about 0.5 μm. A basic toolbox of methods that is used in microbiology for research at the level of single E. Puchkov (*) All-Russian Collection of Microorganisms, G.K. Skryabin Institute of Biochemistry and Physiology of Microorganisms of the Russian Academy of Sciences, Pushchino Center for Biological Research of the Russian Academy of Sciences, Pushchino, Russia e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_17

695

696

E. Puchkov

cells includes flow and laser scanning cytometry, optical microscopy coupled with computer image analysis, scanning probe microscopy, vibrational spectroscopy, nanoscale secondary ion mass spectrometry, and some of their variations. The studies using quantitative analytical methods of single cells of microorganisms have opened up prospects for solving such problems as genotypic and phenotypic heterogeneity of microbial populations, the nature of unculturable and persistent forms, the development of biofilms, the interaction of microbial pathogens and host cells, the relationship of structure and function in the metabolism, and others.

Introduction The size of microorganisms such as bacteria and yeasts is so small that until recently the study of the nature of activity of each individual cell at the molecular level using physical and chemical methods was impossible. To overcome limitations on the sensitivity of traditional analytical methods, analysis of various preparations containing many cells was used to quantitatively study the structural and functional organization of microbial cells. With this approach, the results reflected the average properties of all cells in the population. However, there are a number of problems that can be solved only by analyzing individual microbial cells. These problems are the heterogeneity of microbial populations by genotypic and phenotypic traits; the nature of the microorganisms unculturable in laboratory conditions and of persistent states; the development of biofilms; the interaction of microorganisms in consortia, as well as with plant and animal cells; the relationship of structure and function in metabolism; and a number of others. Over the past 30 years, some analytical methods have been improved to such an extent that the prospect of quantifying many of the physicochemical and morphological properties of individual microbial cells has opened up. This made it possible to approach the above problems at the cellular and subcellular level. The spectrum of studies on the development and application of the methods of quantitative analysis of single microbial cells at the cellular and subcellular level appeared so wide that it became considered as a specific research and development field, “single-cell microbiology” (Brehm-Stecher and Johnson 2004). Since the publication of comprehensive reviews on this topic (Shapiro 2000; Brehm-Stecher and Johnson 2004), modifications of a number of methods have appeared, their application has expanded, and new methods have been developed. Therefore the need for updating information in this area arose. The goal of this chapter is to briefly review publications after 2004 on the main methods of quantitative analysis of single cells of bacteria and yeasts at the cellular and subcellular level with the examples of their application. The number of publications on this topic even during this period is so large that consideration and citation of all of them cannot fit into the framework of the chapter; therefore, the review papers on specific aspects of the topic under consideration will be cited as much as possible in the relevant sections.

30

Analytical Techniques for Single-Cell Studies in Microbiology

697

Cytometry Flow Cytometry Flow cytometry (FC) is the measurement of optical characteristics of individual cells in the flow of fluid. The cell suspension is passed through a narrow capillary in such a way that single cells move one after another in the flow of its central part. The flow of single cells passes through the optical section of the capillary, where it is illuminated by a focused beam of light. The resulting light scattering at different angles and fluorescence of the cells, if the cells were preliminarily treated with certain fluorochromes, are recorded. FC is perhaps the most popular method of analysis of individual cells in suspensions. This is due to the availability of commercially produced devices and a wide range of fluorochromes, applicable for detection of various properties of the cells, as well as the possibility of quick analysis of large populations. Typical commercially available instruments allow for the analysis of 5–10 optical parameters of microbial suspensions at flow rates of up to 1000 cells per second; specialized devices can process up to 50,000 cells per second and up to 20 parameters simultaneously. FC has been used in microbiology for a relatively long time. Various applications of FC in the microbiology studies before 2004 are well presented in the reviews (Shapiro 2000; Brehm-Stecher and Johnson 2004). After that there were also published a number of comprehensive reviews covering, in particular, quantitative analysis of heterogeneity of microbial populations (Avery 2006), assessment of microbial viability (Emerson et al. 2017), characterization of various physiological responses in bacteria (Ambriz-Aviña et al. 2014), and the study of ecology of microorganisms (Wang et al. 2010b; Emerson et al. 2017). Since this huge volume of information cannot be covered in the format of this publication, interested readers are encouraged to refer to these reviews. FC is the basis of a preparative technique, fluorescence activated cell sorting (FACS), for selective isolation of the cells with certain fluorescence characteristics. This technique is widely used in the studies of human and animal cells (Herzenberg et al. 2002). There are examples of the use of FACS in microbiology (on other preparative methods for obtaining single cells of microorganisms, see Ishii et al. 2010). For instance, the cells of freshwater bacteria with low nucleic acid content were sorted by FACS; they were identified as members of the genus Polynucleobacter by the analysis of the 16S rRNA gene sequences (Wang et al. 2009). Another approach based on the simultaneous use of the method of fluorescent in situ hybridization (FISH) and FACS should be mentioned. The possibility of using this approach to isolate the target bacteria in conditions of contamination by foreign genetic material was demonstrated with a mixture of the Escherichia coli cells and phage λ DNA (Chen et al. 2011).

Imaging Flow Cytometry Imaging flow cytometry (IFC) combines the capabilities of FC and fluorescence microscopy that enables simultaneous measurement of integrated optical parameters

698

E. Puchkov

of the cells in the flow and acquisition of their digital fluorescent images. This makes it possible to characterize each cell as a whole by its light scattering and fluorescence, as well as to analyze its morphology and/or spatial distribution of certain optical parameters of the intracellular structures (Barteneva et al. 2012). It is important that the method enables rapid quantitative treatment of large populations of cells by a variety of parameters by special computer analysis algorithms (Hennig et al. 2017). Similar to FC, this approach was initially developed for the study of single human and animal cells, which are significantly larger than the cells of microorganisms (Barteneva et al. 2012). This is due to the low resolution of the images obtained in comparison with conventional microscopy. With the advent of commercially available devices, the method became widely used in biomedical research, including research on the interaction of microorganisms with the cells of a macroorganism. Several approaches based on IFC have been developed to study phagocytosis and internalization of Salmonella enterica in macrophages of the RAW 264.7 cell line (Phanse et al. 2012), Yersinia enterocolitica in mononuclear phagocytes (Drechsler-Hake et al. 2016), Neisseria gonorrhoeae in neutrophils (Smirnov et al. 2017), and Toxoplasma gondii and Mycobacterium tuberculosis in various cell lines (Haridas et al. 2017). Morphological changes of the yeast form of cryptococci associated with the acquisition of pathogenicity factors were revealed in mononuclear lymphocytes (Okagaki et al. 2010); identification, classification, and study of the stages of development of protozoa, Plasmodium falciparum, in the blood were carried out using IFC (Dekel et al. 2017). The use of IFC for quantitative characterization of cell distribution by cell cycle stages in the population of S. cerevisiae (Calvert et al. 2008), as well as for quantitative assessment of phytoplankton communities by such parameters as cell size distribution, viability, and metabolic activity has been shown (Dashkova et al. 2017).

Scanning Cytometry Scanning cytometry (SC) is based on detection of fluorescence and light scattering on a small area of the surface, which is illuminated by a narrow beam of laser radiation during horizontal scanning (Kamentsky and Kamentsky 1991). In contrast to FC, it is possible to study with this technique cell samples not in suspension but on a flat surface, for example, on a slide for microscopy, in a well of a cell culture plate, on a membrane filter, etc. This provides a number of advantages over FC, such as the ability to work with small samples, to study cell morphology, and to conduct examination in situ. SC was used mainly in human and animal cell studies (Pozarowski et al. 2013). There are only few examples of SC application in microbiology. Two techniques have been developed using SC on membrane filters to quantify bacteria and fungi of various natural samples (Vanhee et al. 2010) and bacteria in drinking water (Baudart et al. 2005). There are also similar techniques for quantifying mycobacteria in clinical samples (Pina-Vaz et al. 2004) and of Staphylococcus xylosus in biofilms on glass surfaces (Regina et al. 2012). An SC-based method with microfluorimetry of individual cells of Cryptococcus neoformans yeast

30

Analytical Techniques for Single-Cell Studies in Microbiology

699

Table 1 Selected applications of cytometry in the studies of single microbial cells

Technique Flow cytometry

Structural level of analysis Cellular

Studied microorganisms Bacteria, yeasts

Preparative flow cytometry

Cellular

Bacteria

Flow cytometry with visualization

Cellular

Bacteria

Yeasts

Scanning cytometry

Cellular

Bacteria

Yeasts

Analyzed features Population heterogeneity Viability Physiological reactions Ecology Isolation of bacteria with specific properties Internalization by phagocytes

Internalization by phagocytes Morphometric parameters Morphometric parameters DNA content

References Avery 2006 Emerson et al. 2017 Ambriz-Aviña et al. 2014 Wang et al. 2010b; Emerson et al. 2017 Chen et al. 2011; Wang et al. 2009

Phanse et al. 2012; Drechsler-Hake et al. 2016; Smirnov et al. 2017; Haridas et al. 2017 Okagaki et al. 2010 Calvert et al. 2008 Vanhee et al. 2010; Baudart et al. 2005; Pina-Vaz et al. 2004; Regina et al. 2012 Yamaguchi et al. 2007

was used to assess the content of DNA stained with propidium iodide in the study of the cell cycle. Microfluorometry made it possible to trace the dynamics of DNA synthesis through the stages of yeast development and to estimate the ratio of the cells in the population at different cell cycle stages (Yamaguchi et al. 2007). The applications of cytometry in the studies of single microbial cells described in this section are summed up in Table 1.

Quantitative Optical Microscopy Optical microscopy was the “magic window” through which van Leeuwenhoek first saw the world of microorganisms in the seventeenth century. This method is still one of the key tools of microbiology. However, in its traditional execution, it is predominantly qualitative and subjective, and quantitative methods based on the visual examinations are too labor-consuming. In the second half of the twentieth century, optical microscopy has been upgraded up to an analytical level due to the application of lasers, highly sensitive light detectors, digital photography, and computer technology. New optical microscopy methods enabled objective quantitative studies of spatiotemporal and physicochemical characteristics of microorganisms at cellular

700

E. Puchkov

and even subcellular levels. Thus, optical microscopy became quantitative optical microscopy (QOM), an analytical technique for single-cell studies. It should be noted that, unlike many other analytical techniques, quantitative optical microscopy has the possibility of combined visual and quantitative computer analysis. This allows a researcher to visually choose the objects that are most suitable for quantitative processing and the research logic. Computer digital image processing (CDIP) played a key role in the transformation of optical microscopy from a qualitative to quantitative technique. The basics of CDIP can be found in the reviews (Sbalzarini 2016; Kan 2017; Nketia et al. 2017; Wallace et al. 2018) and references therein. For better understanding various applications of CDIP in quantitative optical microscopy of microorganisms presented in this chapter, the main principles of this methodology could be summarized as follows (Puchkov 2016a). CDIP deals with the images (e.g., photographs) obtained in or converted into the digital form. A digitized image is a set of small elements called pixels in 2D space or voxels in 3D space. Each pixel (voxel) contains the digitally encoded information on its X - Y - (Z) location in a Cartesian coordinate system and optical features at this point of space. The optical information depends on the imaging system employed. The information obtained with monochrome digital cameras consists of the data on the light intensity in gray scale units (often 8-bit 256 gray levels from white to black). The digital images from conventional three-channel color cameras or other filter-based imaging systems, such as multispectral and hyperspectral imaging systems (Hagen and Kudenov 2003; Gao and Smith 2015), can be a set of pixels with the optical information encoded in one of the four “color spaces” RGB, HSV, CIE-Lab, and YCrCb. RGB is the most popular 24-bit combination of red, green, and blue values with 8 bits for each color. Finally, the raw digital data of the images are treated by a computer as variables of a huge mathematical model, constructed of various algorithms. The computer manipulation of digital images for improving their quality and a “more expressive” visual appearance of some components (image deconvolution) can be one outcome of this treatment. “Quantitative phase microscopy” (Phillips et al. 2017) is a good example of the application of this approach for improving sharpness of the images obtained with phase contrast microscopy. The extraction of some quantitative digitally encoded information and its analysis, that is, computer digital image analysis (CDIA), can be another outcome. CDIP and CDIA serve as the main basis of QOM. The use of CDIP and CDIA in optical microscopy of microorganisms simplified, accelerated, and automated many existing quantitative microscopy methods. In addition, it provided previously unavailable quantitative information. This information includes spectral characteristics and their spatial distribution; size and shape of individual components of complex geometry, as well as their mutual arrangement; and the dynamics of processes in a wide range of time scale. The main analytical capabilities of QOM for microbiology can best be demonstrated by numerous examples of its use for solving fundamental and applied problems.

30

Analytical Techniques for Single-Cell Studies in Microbiology

701

Brightfield Microscopy Coupling of brightfield microscopy with computer image analysis made it possible to quantitatively analyze morphological features of single cells of microorganisms, that is, to perform microbial morphometry. This methodology has been applied in several areas of research. A method has been presented for determining the time of the first division of individual rod-like bacterial cells of Escherichia coli, Listeria monocytogenes, and Pseudomonas aeruginosa growing on a solid medium (Niven et al. 2006). The digital images of phase-contrast microscopy of the growing bacteria were captured at intervals. The ratio of the area of the smallest rectangle that can be drawn around the cell, divided by the area of the cell itself, was estimated with CDIA (Image-Pro Plus V4.5 and two in-house written Visual Basic software, called ObjectTracker and Boxer). This ratio was calculated at each time point and found to increase suddenly during growth at a time that correlated with cell division as estimated by visual inspection of the digital images. Since the method allowed automation of the data processing, it was successfully used to generate distributions of lag times of single cells for populations of the studied bacteria. The distributions of the time of the first division parameter for populations of food-borne pathogens serve for predicting shelf life or the possible growth of pathogens in food products. Interesting research possibilities are opened by an original micro-flow device, called “mother cell,” which allows using microscopy with computer image analysis to study the dynamics of division of individual E. coli cells and the relationship of this process with their size. The division rate and the change in the cell size of a few hundred generations derived from a single “mother” cell have been analyzed using this device. It has been shown that the average rate of cell division persisted for hundreds of generations. However, there was a “fluctuation” difference in the duration of the cell cycle of each pair of “mother” and “daughter” cells. Based on these data, it has been assumed that each E. coli cell has a robust mechanism of reproduction, and the loss of viability is not the result of a “stochastic” aging but is due to the accumulation of damaging events (Wang et al. 2010a). The nature of the dependence of an increased average cell length on the growth rate was investigated using the same device and a ΔrecA mutant strain of E. coli. It was found that the average cell length increased at a growth rate above a certain threshold value as a result of stochastic suppression of division in a part of the population, which may be due (at least partially) to restricted RecA recruitment into the nucleoid, which is critical for the repair of stalled replication forks (Gangan and Athale 2017). QOM is a tool that is expected to help in deciphering numerous phenomena associated with appearance of various phenotypes in microbial populations and formation of biofilms. To deal with these problems computer techniques for monitoring and tracking of subpopulations and individual cells in growing bacterial colonies are needed. A number of open source image processing software packages for analyzing bacterial cell movies have been developed (reviewed in Balomenos et al. 2017). The main limitations of these techniques were the lack of generality and automation. A computational pipeline, Bacterial image analysis driven Single Cell

702

E. Puchkov

Analytics (BaSCA), was developed, which overcame these limitations (Balomenos et al. 2017). It combines image processing and machine learning algorithms to achieve precise bacterial colonies and single-cell segmentation, tracking and phenotypic characterization. It also allows the fully automated segmentation and morphology/expression analysis of individual cells in time-lapse cell movies. The pipeline allows analyzing efficiently colonies regardless of their cell density. BaSCA extracts various single-cell properties, which get organized into a database. The authors suggested that, due to its high throughput and ability to manage large data sets, the pipeline can be used in Systems Microbiology to tackle the problems of biofilm formation, persisters’ emergence, etc. Microbial morphometry may be a useful tool to gain ecophysiological insights into microbial adaptations in their ecological niches in situ. With this goal in mind, a software application suite called CMEIAS (Center for Microbial Ecology Image Analysis System) was developed to characterize the morphotypes of bacteria in microbial communities using the data of brightfield and phase-contrast microscopy (reviewed in Dazzo and Niccum 2015). The algorithms of this software package can accurately characterize the following features of a bacterial population at single-cell level: (1) cell length; (2) biovolume relevant to body mass; (3) pattern recognition rules for morphotype classification of diverse microbial communities; (4) spatial patterns of coaggregation that reveal relevant to microbial biofilm ecology; and (5) object segmentation of complex color images to differentiate target microbes. The dynamics of development of microbial populations in natural biofilms was studied with CMEIAS (Dazzo et al. 2017). Images of the microbial biofilms formed on microscope slides and transparent polystyrene submerged in a river were acquired using brightfield and phase-contrast microscopy and digitally segmented as the foreground objects of interest. Various phenotypic features of the cells and colonies were extracted quantitatively and evaluated by discriminating statistical tests. The significant impact of substratum physicochemistry was revealed for biofilms during their early immature stage of development in the river ecosystem. The results indicated that river biofilm architecture exhibits significant geospatial structure in situ, providing many insights on the strong influence that substratum hydrophobicity–wettability exert on biofilm development and ecology, including their productivity and colonization intensity and various morphological and physiological traits. The budding yeast Saccharomyces cerevisiae was broadly used in the studies of the global regulation of morphological characteristics. The main approach was based on the search for the relationship of changes in morphological characteristics with certain mutations. Initially, these studies were performed manually by measuring the size of cells and buds (reviewed in Saito et al. 2004). A morphometric system for the high-throughput analysis of yeast morphology for images of live cells without fixing and staining has been developed (Liu et al. 2011b). A study of four strains of live yeast cell in both exponential and stationary phase was performed with this system. The results have shown that statistically significant morphological differences can be identified between strains, and that these differences vary with growth stage. A comparison of the results for the studied strains with the data for the same strains obtained by subcellular morphometry (see below) based on fluorescence microscopy

30

Analytical Techniques for Single-Cell Studies in Microbiology

703

(Saito et al. 2004) has been made. There was difference between the results of these two studies, which was proposed to be due to the different cell growth conditions, or because the fluorescence microscopy data were obtained with the formaldehyde fixed cells. The system was supposed to be of help in the studies of the complex systems-biology relationships between genotype, environment, and the phenotype of cell morphology. An image processing algorithm to automatically classify the live S. cerevisiae cells according to their development phase has been designed (Yu et al. 2011). The algorithm provided enhancement of the transmission microscopy images of the cells and a robust segmentation to extract geometrical features including compactness, axis ratio, and bud size. The features were classified with various machine-learning classifiers for comparison. The linear support vector machine, distance-based classification, and k-nearest-neighbor algorithm were the classifiers used in the study. The training data labeled manually by a single individual were used. It was shown that this algorithm was more consistent compared to manual classification and was fully automatic. All classifiers showed similar accuracy; k-nearest-neighbor was the most versatile classifier, and support vector machine had the fastest processing time.

Fluorescence Microscopy Almost all analytical tools based on the phenomenon of fluorescence (Lakowicz 2006) can now be applied at the level of single microbial cells. This became possible due to the availability of a new generation of fluorescence microscopes based on various principles of imaging (reviewed in Sanderson et al. 2014), numerous fluorescent dyes with a wide range of spectral features, and computer programs for processing and analyzing digital images. The development and application of fluorescent protein (FP) technology has led to significant progress in many areas of research related to the intracellular structure and functions of live cells, including microbial ones (Tsien 1998; Snapp 2005). With the use of quantitative computer processing of the digital fluorescence images it became possible to study intracellular structures with the resolution of about 0.02– 0.04 μm and even to track the localization and movement of individual macromolecules by methods of so-called super-resolution fluorescence microscopy (SRM) (“nanoscopy”) (Coltharp and Xiao 2012). These methods are based on the use of genetically engineered chimeric FP (Yao and Carballido-López 2014), or so-called “selflabelling” proteins (Stagge et al. 2013). The latter are obtained via two-stage procedures. First, one of the special polypeptide tags, the SNAP-, CLIP-, or HaloTag, is genetically fused to the protein of interest. Second, a fluorescent ligand is specifically covalently bond to the polypeptide tag resulting in a fluorescently labeled target protein. There are two main strategies of SRM (Chozinski et al. 2014; Gahlmann and Moerner 2014). The first enables to monitor the movement of a fluorescent molecule of the protein, which is expressed or activated at a very low concentration, one or two molecules per cell. In the second one, a series of fluorescent images of molecules

704

E. Puchkov

of the same protein labeled with so-called “photoswitchable” fluorophores (Chozinski et al. 2014; Minoshima and Kikuchi 2017) are recorded. The fluorescence of these proteins can be “switched on” or “switched off” by separated time light pulses. Computer processing of the obtained images superimposed on each other enables to reveal the position of adjacent fluorescent proteins in the cell; the distance between them may be of 0.02–0.04 μm. The significant upgrading of fluorescence microscopy made it possible to quantitatively study individual cells of such tiny creatures as yeast and bacteria at a subcellular level. The examples below illustrate some of these possibilities. A CDIP software package (CalMorph) that could automatically extract quantitative data from fluorescence microscopy images of budding yeast cells has been developed (Ohtani et al. 2004). The software operated with the images of cells, which were simultaneously stained with FITC-ConA for cell wall identification, DAPI to localize nuclei, and Rhodamine-Phalloidin to visualize the actin distribution. This software extracted data on cell geometry features such as size, roundness, bud neck position angle, bud growth direction, and fits an ellipse to the cell outline. Also, the positions of nuclei and actin patches relative to the cell wall could be monitored with the software. A set of haploid yeasts with nonessential gene disruption has been used to construct a database of yeast cell morphology called S. cerevisiae Morphological Database (SCMD). It included fluorescence microscopy images and CalMorph software package. It was intended as a resource that complements existing sequence and gene expression databases (Saito et al. 2004). Later, modified software named F-CalMorph for analysis of fission yeast cells was developed (Suzuki et al. 2006). Data mining using SCMD with CalMorph software package has been demonstrated (Ohya et al. 2005). The high-dimensional and quantitative phenotyping of yeast mutants has been performed using morphological variations from the SCMD as quantitative traits. In particular, it was found that individual deletion of nearly half of the nonessential genes in the genome affects cellular morphology. The close relationship between cellular morphology of the mutant strains and the functions of the deleted loci has been revealed. As an example, four of five polarisome mutant strains (polarisome protein complex is known to determine cell polarity) displayed a round bud phenotype indicating that a round bud is a characteristic trait of polarisome perturbation. A similar correlation was observed in 902 cases among a total of 368,808 combinations. Particular morphological phenotypes were enriched in 260 of the 1,452 gene ontology database groups analyzed. This finding implies that, in the case of the genes belonging to these annotated groups, the phenotype of a mutant strain deficient in such a locus can be predicted. The use of morphological phenotyping of yeast mutant strains to assign potential functions to unknown genes was proposed. The high-dimensional and quantitative phenotyping based on morphological parameters (Ohya et al. 2005) was used to investigate the natural genetic diversity of S. cerevisiae cellular morphology (Nogami et al. 2007). To this end, 501 morphological parameters in over 50,000 yeast cells from a cross between two wild-type divergent backgrounds were quantified. Extensive morphological differences were

30

Analytical Techniques for Single-Cell Studies in Microbiology

705

found between these backgrounds. The genetic architecture of the traits was complex; the evidence of both epistasis and transgressive segregation was obtained. Quantitative trait loci for 67 traits have been mapped; 364 correlations between traits segregation and inheritance of gene expression levels were discovered. This study illustrated the natural diversity and complexity of cellular traits among natural yeast strains and provided an ideal framework for a genetically dissection of multiple traits. The subcellular morphology of mitochondria, vacuole, endoplasmic reticulum, Golgi body, endosome, spindle pole body, and septin in S. cerevisiae were evaluated with CalMorph after labeling these organelles and structures with corresponding fluorescent dyes and GFP (Negishi et al. 2009). In this study, in addition to the originally developed 501 parameters for cell wall morphology, nuclear DNA, and actin (Ohtani et al. 2004), additional 610 parameters for the morphology of subcellular components have been proposed. This approach was suggested for more detailed phenotypic studies, which is advantageous in yeast functional genomics. Drug-induced morphological changes caused by antifungal agents of three classes with known mechanism of action were investigated using CalMorph and machine learning (Gebre et al. 2015). The comparison of drug-induced morphological changes as a chemogenomic signature with morphological traits of cells with the deletion of 4718 nonessential genes (SCMD) not only confirmed the mode of action of the drugs but also revealed an unexpected connection among ergosterol, vacuolar proton-transporting V-type ATPase, and cell-wall-targeting drugs. Based on the data obtained, a systematic classifier that sorts a newly discovered compound into a class with a similar mode of action without any mutant information was developed. Using well-characterized agents as target unknown compounds, this method successfully categorized these compounds into their respective classes. It was suggested that morphological profiling can be used to develop novel antifungal drugs. In the studies related to localization of certain FP-labeled (tagged) proteins in different yeast strains or in the same strain under different cultivation conditions, it is necessary to automate the quantitative comparison of large data sets. For this purpose, a special computer image analysis program was designed (Chen et al. 2007). A collection of yeast containing FP-labeled proteins has been created. It is expected that it can be used in the future in conjunction with proteomic analysis to study the activity (functions) of proteins at the subcellular level (Chong et al. 2012). Two methods for viscosity measurements in the yeast vacuoles directly in single live (not fixed) S. cerevisiae cells have been developed (Puchkov 2016b). (General aspects of intracellular viscosity were reviewed in Puchkov 2013.) Specially tailored computer image processing algorithms of the ImageJ software for subcellular microfluorimetry have been used in these methods. The first method was based on the assessment of Brownian motion displacement and size of insoluble polyphosphate complexes stained with DAPI in the vacuoles. Using the obtained data, apparent viscosity inside the vacuoles of four cells has been computed by the Einstein– Smoluchowski equation. It was found to be 2.16
0.60, 2.52
0.63, 3.32
0.9, and 11.3
1.7 cP. To verify these data, the second method has been developed. A fluorescent dye quinacrine was shown to be specifically accumulated within the

706

E. Puchkov

vacuoles of the yeast cells with no detectable binding. The fluorescence anisotropy of quinacrine was measured by image analysis microfluorimetry in the vacuoles of 39 cells using images acquired with a fluorescence microscope equipped with polarizers. Using the Perrin plot as a calibration curve, apparent viscosity values of the vacuolar milieu were calculated for each cell. The population of the cells studied was heterogeneous with regard to vacuolar viscosity, which was in the range 3.5
0.4–14.06
0.64 cP. The cells with viscosity values in the range of 5–6 cP were the most frequent. To test if live yeast cells of S. cerevisiae can be used as a model for locating intracellular sites/targets of the DNA-directed drugs, a so-called “pseudospectral” analysis was developed (Puchkov 2016b). Intracellular distribution of anthracycline anticancer drug doxorubicin (DR), as an example, was investigated along with fluorescent DNA markers DAPI and ethidium (E). Red, green, and blue components of the fluorescence intensity were quantitatively assessed with ImageJ in selected subcellular regions of interest. They roughly corresponded to the real spectra of the dyes in solutions and were called “pseudospectra.” It was established using analysis of these “pseudospectra” of the selected subcellular regions of interest that all three dyes were located in the nuclei and in the mitochondria. An appreciable fraction of DR and E was assumed to be associated with the mitochondrial membranes in contrast to DAPI, which interacted only with nuclear and mitochondrial DNA. It was concluded that this approach may be applied when designing new DNA-targeted drugs at the stage of preliminary assessment of their interaction with eukaryotic cells. The geometry (size and shape) of the S. cerevisiae nuclei in three-dimensional (3D) space with a resolution of 30 nm was assessed (Wang et al. 2016). In this study, a specially developed algorithm named “NucQuant” was used for automated image analysis spinning-disk confocal microscopy data. The 3D positions of FP-labeled nuclear pore complexes in the nuclear envelope were determined and the nuclear geometry was deduced. Modifications of the nuclear morphology were observed when changing carbon source, upon quiescence or in G1-arrested cells. A single molecule fluorescence in situ hybridization (smFISH) technique based on visualization and quantitation of single RNA molecules (Femino et al. 1998) was used in a number of studies on yeasts (reviewed in Chen et al. 2018). At least two recently developed protocols for S. cerevisiae should be mentioned. An optimized protocol to examine the expression of two mRNA isoforms of the NDC80 gene, which encodes a kinetochore protein, during vegetative growth and meiosis, was developed. It was shown that only the short isoform was expressed during vegetative growth while the long isoform was expressed specifically in meiosis. These two types of mRNAs were detected in the cytoplasm, suggesting that both were exported from the nucleus (Chen et al. 2018). Another protocol of smFISH was developed to quantify the kinetics of expression of STL1 and CTT1 mRNAs in single S. cerevisiae cells upon 0.2 and 0.4 M NaCl osmotic stress. A Data Descriptor with RNA smFISH data set containing raw images and processed mRNA counts was presented. It was suggested that these data can be used to develop single cell modeling approaches, to study fundamental processes in transcription regulation and develop single cell image processing approaches (Li and Neuert 2019).

30

Analytical Techniques for Single-Cell Studies in Microbiology

707

Single-molecule tracking (SMT) microscopy allows monitoring the motion of individual biomolecules in live systems, as well as characterization of association/ dissociation kinetics between molecules (reviewed in Liu et al. 2016). Robust procedures for quantitative SMT of transcription factors (TF) with their chromatin response elements in single S. cerevisiae cells have been developed. In this study, Halo-Tag technology (reviewed in Stagge et al. 2013) of TF labeling was used. The developed procedures made it possible to characterize nonspecific binding of Ace1p TF (activator of copper-binding metallothionein gene CUP 1) and two other DNA-binding proteins, Hht1p (histone H3) and Hsf1p (heat shock factor). It was shown that the estimated residence time of the three molecules was essentially of the same very short time. This suggests that short-residence molecules bind to DNA nonspecifically and may have a common underlying mechanism (Ball et al. 2016). Genome architecture in 3D of live S. cerevisiae cells was explored using SMT methodology with millisecond time resolution. Astigmatism super-resolution imaging (Kao and Verkman 1994) of fluorescently labeled Mig1protein (a Zn-finger transcription factor) that binds to target DNA sequences was used to enable extraction of 3D protein position data. The likely Mig1 binding sites in 3D were deduced from the analysis by chromosome conformation capture (3C) and bioinformatics techniques. The developed method made it possible to provide more information toward understanding the functional 3D genome architecture in live cells (Wollman et al. 2020). A protocol has been described for the subnuclear localization of a certain region of a chromosome in live (not fixed) fission yeast Schizosaccharomyces pombe. The approach was based on the FP-technology, confocal microscopy, and CDIA. Relative distances between a gene cluster, named Chr1, nuclear membrane, and Spindle Pole Body, were measured and found to be changed after nitrogen starvation (Bjerling et al. 2012). The algorithms have been developed that allow the use of metal-oxide semiconductor (sCMOS) cameras for single-molecule switching nanoscopy. They made it possible to conduct quantitative high-speed fluorescence photoactivation localization microscopy (FPALM) (Huang et al. 2013). This technology was used in the studies of protein localization and dynamics in intact S. pombe yeast cells with spatial and temporal resolution of 35 nm and 1 s, respectively. The structures and assembly of two types of interphase nodes, precursors of the cytokinetic contractile ring, were studied. First, it was shown that the nodes were discrete structural units with stoichiometric ratios and distinct distributions of constituent proteins. During constriction, the nodes moved bidirectionally and the general node organization persisted in the contractile ring. The dynamics of the actin network formation during cytokinesis was observed (Laplante et al. 2016). Second, the composition, assembling, and disassembling of the nodes during mitosis were investigated. It was shown that the total number of type 1 node proteins was constant even when the nodes disassembled. Type 2 nodes remained intact throughout the cell cycle and were constituents of the contractile ring. They were released from the contractile ring as it disassembled and then associated with type 1 nodes around the equator of the cell during interphase (Akamatsu et al. 2017). FPALM was also used in the study of molecular mechanism of actin filament polymerization in clathrin-mediated endocytosis in S. pombe. It was shown that

708

E. Puchkov

Arp2/3 complex, which assembles two zones of actin filaments, was recruited by two independent pathways defined by the nucleation promoting factors Wsp1p (WASp) and Myo1p (myosin-I). The data on dynamics of assembling of the complexes the authors considered to support their two-zone hypothesis to explain endocytic tubule elongation and vesicle scission in fission yeast (Arasada et al. 2018). Super-resolution structured illumination microscopy with single-particle averaging (Burns et al. 2015) was used in the study of microtubule-organizing centers, known as spindle pole bodies (SPBs) in S. pombe. The relationship of protein components and regulators of SPB and how each protein is assembled into a new structure during SPB duplication were explored. These data enabled to build the first comprehensive molecular model of the S. pombe SPB (Bestul et al. 2017). (Some practical aspects of studying single live fission yeast cells with fluorescence microscopy are summarized in the review of Mulvihill 2017). The methodology of SRM made it possible to observe nanoscale structures in live bacterial cells with unprecedented detail and to conduct single-molecule tracking that provides dynamic information about the motions of labeled proteins. Many subcellular studies of bacteria using this methodology were summarized and discussed in a number of reviews focused on general overview (Endesfelder 2019), technical aspects (Haas et al. 2014), and fundamentally important information on bacterial cytoskeleton and nucleoid structure (Yao and Carballido-López 2014); the relations of protein and mRNA copy number (Taniguchi et al. 2010) and ATP level (Yaginuma et al. 2014); the dynamics of transcription (Stracy and Kapanidis 2017) and translation (Gahlmann and Moerner 2014); the dynamics of proteins interacting with DNA (Uphoff 2016); the organization of secretory systems and intracellular compartmentalization (Schneider and Basler 2016); intracellular signaling networks and gene expression kinetics (Kentner and Sourjik 2010); and the mechanisms of action of antibacterial drugs (Choi et al. 2016). The toolbox of SRM methods is constantly replenished for specific subcellular studies of bacteria. Here are some examples. The study was reported (Saurabh et al. 2016) on the use of a highly photostable fluoromodule containing Malachite Green as a fluorogen to genetically label proteins in Caulobacter crescentus. Imaging of the module relied on the activation of the fluorogen and can be used to label proteins sparsely, enabling single-protein detection in live bacteria without initial bleaching steps. The SRM imaged fusions of the complex to three different proteins in live Caulobacter cells had a four-fold better resolution compared to diffraction-limited imaging. To address the question on how does Vibrio cholerae produce its deadly toxin, plasmon-enhanced emission from intrinsically fluorescent proteins in live bacterial cells was used (Flynn et al. 2016). Membrane-bound single fluorescent protein fusions to the virulence regulator TcpP were coupled to extracellular gold nanotriangle arrays. This provided an enhancement in the rate of emission and in the number of photons detected prior to photobleaching. The authors noted that plasmon-enhanced fluorescence is a biocompatible, generalizable path

30

Analytical Techniques for Single-Cell Studies in Microbiology

709

to directly improve the resolution and trajectory lengths of single molecules in live cells. A novel analytic tool for PhotoActivated Localization Microscopy (one of the varieties of SRM) has been developed to improve the nanoscale localization of protein molecules in bacterial cells (Altinoglu et al. 2019). It integrates precisely drawn cell outlines, of either inner membrane or periplasm, labeled by FP, with molecule data for >10,000 molecules from >100 cells. It was shown that the polar anchor HubP in V. cholerae constitutes a big polar complex which includes multiple proteins involved in chemotaxis and the fagellum. Also, HubP was shown to be slightly skewed toward the inner curvature side of the cell, while its interaction partners showed rather loose polar localization. A strategy for the chemoenzymatic labeling of bacterial proteins with azide-bearing fatty acids in live cells using the eukaryotic enzyme N-myristoyltransferase has been proposed and applied on E. coli (Ho and Tirrell DA 2019). Cell-permeant bicyclononyne-functionalized rhodamine spirolactams were used as photoswitchable fluorescent tags of chemotaxis proteins Tar and CheA and cell division proteins FtsZ and FtsA. SRM revealed helical pattern of Tar and banded patterns of FtsZ dispersed throughout the cell. The precision of radial and axial localization in reconstructed images was close to 15 and 30 nm, respectively. The method was expected to be broadly useful for imaging intracellular bacterial proteins in live cells with nanometer resolution. The applications of QOM in the studies of single microbial cells described in this section are summed up in Table 2.

Scanning Probe Microscopy Scanning probe microscopy is based on the interaction of a miniature mechanical probe (cantilever) with the surface of an investigated object during its scanning. Importantly, the size of the probe is so small that it is possible to characterize the physical and chemical properties of the surface (to obtain “physicochemical images”) with a resolution of the size of atoms. There are four main types of scanning probe microscopy: scanning tunneling microscopy (STM), scanning electrochemical microscopy (SECM), scanning magnetic force microscopy (SMFM), and scanning atomic force microscopy (SAFM). In STM, the so-called tunneling current between the sample surface and the probe is measured. Recording of the electrochemical processes between the probe and the surface under study is carried out in the SECM. The interaction of the atoms of the tip and the studied sample due to magnetic field or van der Waals forces is the basis of the CMFM and CAFM, respectively (Bhushan and Marti 2011).

Scanning Atomic Force Microscopy The use of SAFM in the study of microorganisms began relatively long ago, and the results obtained in this area of research are summarized in numerous reviews.

710

E. Puchkov

Table 2 Selected applications of quantitative optical microscopy in the studies of single microbial cells

Technique Brightfield microscopy

Fluorescence microscopy

Structural level of analysis Cellular

Cellular, subcellular, and molecular

Studied microorganisms Bacteria

Analyzed features Morphometric parameters

Yeasts

Morphometric parameters Metabolic reactions Viability Intracellular protein localization, including dynamic studies Intracellular ATP Proteome and transcriptome Dynamics of transcription Dynamics of translation Morphometric parameters

Bacteria

Yeasts

Viability Intracellular protein localization, including dynamic studies Intravacuolar viscosity

References Dazzo et al. 2017; Dazzo and Niccum 2015; Niven et al. 2006; Balomenos et al. 2017; Wang et al. 2010a; Gangan and Athale 2017 Yu et al. 2011; Liu et al. 2011b Ogawa et al. 2005 Schulze et al. 2011 Yao and CarballidoLópez 2014; Uphoff 2016; Schneider and Basler 2016; Cortesi et al. 2017 Yaginuma et al. 2014 Taniguchi et al. 2010 Stracy and Kapanidis 2017 Gahlmann and Moerner 2014 Negishi et al. 2009; Ohya et al. 2005; Gebre et al. 2015; Han et al. 2011 Puchkov 2014 Chong et al. 2012; Bjerling et al. 2012

Puchkov 2016b

Therefore, the available information is presented with the reference to the main ones. Initially, SAFM was used to obtain images of the the surface structure of native microbial cells with a resolution comparable to that of scanning electron microscopy (SEM). The cells were examined in an aqueous medium that distinguishes SAFM from SEM, which is carried out on dry or frozen preparations under vacuum conditions (Dorobantu et al. 2012). The use of SAFM for obtaining images of the

30

Analytical Techniques for Single-Cell Studies in Microbiology

711

structure of polysaccharides, peptidoglycan, teichoic acids, pili, flagella, and many other components on the surfaces of various microbial cells at physiological conditions has been demonstrated (Dufrêne 2014). The resolution of SAFM makes it possible to study the structural arrangement of single protein molecules in supramolecular complexes. One of the examples of realization of this unique possibility of SAFM is the characterization of spatial self-assembly of protein molecules in the envelope of bacterial micro compartments of Haliangium ochraceum (Sutter et al. 2016). High-resolution SAFM can also be used to record the dynamics of formation of supramolecular protein complexes. As an example, structural features of the assembly and activation of MotPS stator component of the Na +-type flagella motor of Bacillus subtilis have been studied with a special technique of high-speed SAFM (Terahara et al. 2017). SAFM enables quantitative analysis of the forces of interaction of various molecular structures on the surface of microbial cells, between themselves and with the environmental surfaces. For this purpose, modification (functionalization) of the cantilever is made using various chemical groups (e.g., amino group or carboxyl group), small balls (e.g., glass or latex), or even whole cells of bacteria (Lower 2011). Thanks to this, SAFM was transformed into a suite of methods: chemical force microscopy (CFM), single-cell force spectroscopy (SSCF), and single-molecule force spectroscopy (SMFS). They make it possible to characterize the spatial organization of chemical groups and charges on the surface of microbial cells (“chemical imaging”), as well as to measure the forces of interaction of microorganisms with each other, with different ligands, and with abiogenic surfaces (Dorobantu et al. 2012; Ott et al. 2017). SAFM-based approaches proved to be the most popular in the study of microbial biofilms (Angeloni et al. 2016). For example, one of the SAFM techniques enables investigation of the division of single E. coli cells growing on a flat surface (Van Der Hofstadt et al. 2015). Multiparameter imaging with SMFS and SSCF was successfully used to study the molecular mechanisms of bacterial cell adhesion in the formation of biofilms on mineral soil microparticles (Huang et al. 2015), on medical tools (Herman-Bausier et al. 2017), and biotechnological equipment (James et al. 2017).

Scanning Electrochemical Microscopy SECM has been used for the study of biochemical processes on the surface of biofilms of a number of bacteria (Zoski 2016). As an example, measurements of the concentration of hydrogen peroxide on the surface of the biofilms of Vibrio fischeri (Abucayon et al. 2014) and Streptococcus gordonii in the mixture with Aggregatibacter actinomycetemcomitans (Liu et al. 2011a) were carried out with SECM. In the first case, it was shown that the catalase activity of two V. fischeri strains, free-living and living in symbiosis with squid cells, was approximately the same. In the second case, it was found by analyzing the “chemical image” of the hydrogen peroxide content on the biofilm surface that hydrogen peroxide, generated

712

E. Puchkov

on the surface of S. gordonii cells, was then decomposed by catalase on the surface of the A. actinomycetemcomitans biofilm. In the study of intercellular signaling in P. aeruginosa biofilms, the distribution of redox-active pyocyanine signal molecules in the 3D space above the surface of the P. aeruginosa biofilm has been characterized with SECM. The possibility of obtaining a “chemical image” of the redox activity of the studied biofilm in 2D space was also demonstrated (Koley et al. 2011). In another study related to intercellular signaling, the reactions of quorum sensing (QS) in 3D space inside and between artificially created aggregates of P. aeruginosa cells were studied by SECM of pyocyanin. This approach revealed the dependence of QS reactions on the size of aggregates (Connella et al. 2014). No information on the use of scanning tunneling microscopy and scanning magnetic force microscopy for the study of single microbial cells was found in the available literature. The applications of scanning probe microscopy in the studies of single microbial cells described in this section are summed up in Table 3. Table 3 Selected applications of scanning probe microscopy in the studies of single microbial cells

Technique Scanning electrochemical microscopy

Scanning atomic force microscopy

Structural level of analysis Cellular and subcellular

Cellular, subcellular, molecular

Studied microorganisms Bacteria

Bacteria

Isolated supramolecular complexes

Analyzed features Electrochemical properties of the cell surface in biofilms

Structure of polysaccharides, peptidoglycan, teichoic acids, pili, flagella, and other cell surface components Spatial organization of chemical groups and charges on cell surface; forces of cell interaction with each other, ligands, and surfaces Protein spatial organization

References Koley et al. 2011; Liu et al. 2011a; Abucayon et al. 2014; Connella et al. 2014; Zoski 2016 Dorobantu et al. 2012; Dufrêne 2014; Angeloni et al. 2016

Dorobantu et al. 2012; Huang et al. 2015; Ott et al. 2017; HermanBausier et al. 2017

Sutter et al. 2016

30

Analytical Techniques for Single-Cell Studies in Microbiology

713

Nanoscale Secondary Ion Mass Spectrometry Nanoscale secondary ion mass spectrometry (NanoSIMS) enables characterization of the presence of certain chemical compounds in the studied objects in 2D Cartesian coordinates with a resolution within the nanometer range (“chemical imaging”). This is done with unique devices, ion microscopes and ion microprobes, based on the mass spectrometry of secondary ions. In the devices of the first type, the sample is irradiated with primary ions over the entire surface at the same time. In the devices of the second type, scanning is carried out using a very narrow irradiating beam of ions. Cesium (Cs +) and oxygen (O ) ions are used as primary ions. The fragments of the broken molecules of the sample, secondary ions, produced under the impact of primary ions, are separated according to their mass in the mass spectrometer detector and recorded as a mass spectrum. All these data in both types of devices are in strict accordance with the 2D Cartesian coordinates of the sample. If there are spatially ordered structures in the sample such as microbial cells, then after a special computer processing of the data, their “chemical image” is obtained reflecting the specific 2D composition of secondary ions. Moreover, the devices are equipped with a microscope that makes it possible to compare the “chemical image” with the optical image (Gao et al. 2016; Nuñez et al. 2017). The use of NanoSIMS is well presented in numerous studies of microorganisms at the level of single cells especially in combination with FISH (reviewed in Gao et al. 2016) and stable isotope probing (SIP) (reviewed in Pett-Ridge and Weber 2012; Musat et al. 2016). There are also examples of NanoSIMS application combined with scanning transmission (soft) X-ray microscopy (STXM) (Behrens et al. 2012), microautoradiography (MAR), and micro Raman spectroscopy (micro RS) (Musat et al. 2012). All these approaches enabled tracking metabolic activities of single microbial cells by imaging natural isotopic/elemental composition or isotope distribution in situ. NanoSIMS in the abovementioned combinations made it possible to study function to be linked to microbial identity directly in complex microbial communities, as well as metabolic interactions within microbial consortia (Musat et al. 2016). Three examples are presented below to illustrate some of the NanoSIMS possibilities. The first example is the study of the dynamics of multiplication of single Staphylococcus aureus cells directly in the sputum of the patients with cystic fibrosis (Kopf et al. 2016). The approach was based on the labeling with deuterium (2Н2О) of newly synthesized branched saturated C14 and C16 fatty acids with a parallel identification of S. aureus by FISH. Pronounced heterogeneity of the S. aureus cells with regard to generation time was found, and the median of this indicator for the studied population in situ was significantly less than the rate limit for cystic fibrosis pathogens previously measured in vitro. The authors emphasize the importance of the data obtained on single cells in situ for choosing the right chemotherapy strategy for cystic fibrosis. The second example is related to the study of microbial ecology. Physiological characteristics of individual bacterial cells of Chromatium okenii, Lamprocystis purpurea, and Chlorobium clathratiforme in the ecosystems of oligotrophic meromictic lakes were studied (Musat et al. 2008). For this purpose, an approach

714

E. Puchkov

was developed, which was based on combining of NanoSIMS and molecular identification by FISH, similar to the previous example. Metabolic activity was characterized by assimilation of H13CO3 and 15NH4+. Heterogeneity in metabolic activity of the cells of each species of the population was revealed. Another interesting observation was the discovery that C. okenii cells, representing less than 0.3% of the population, provided more than 40% of ammonium consumption and 70% of total carbon consumption in the system. This result is particularly illustrative of the fundamental importance and prospects of the studies in microbial ecophysiology at the level of individual cells, in particular by NanoSIMS methods. The third example illustrates a possibility of identifying and grouping phenotypically similar microbial cells by their chemical and isotopic fingerprint by a taxonomy-independent method using multi-isotope SIP and NanoSIMS (Dawson et al. 2016). This method was applied to study sulfur-cycling biofilm collected from sulfidic intertidal vents. The experiments were performed by SIP with 13C-acetate, 15 N-ammonium, and 33S-sulfate. Using a cluster analysis technique based on fuzzy c-means, the cells were grouped into five groups according to their isotope 13C/12C, 15 N /14N, and 33S/32S and elemental C/CN and S/CN ratio profiles. These isotope phenotype groupings reflected the variation in labeled substrate uptake by cells in a multispecies metabolic network dominated by Gammaproteobacteria and Deltaproteobacteria. Populations independently grouped by isotope phenotype were compared with FISH data, demonstrating a single coherent deltaproteobacterial cluster and multiple gammaproteobacterial groups, highlighting the distinct ecophysiologies of spatially associated microbes within the sulfur-cycling biofilm. The applications of NanoSIMS in the studies of single microbial cells described in this section are summed up in Table 4. Table 4 Selected applications of NanoSIMS in the studies of single microbial cells Structural level of analysis Cellular

Studied microorganisms Bacteria

NanoSIMS +SIP

Cellular

Bacteria

NanoSIMS +SIP+FISH

Cellular

Bacteria

NanoSIMS +STXM NanoSIMS +FISH +MAR + micro RS

Cellular

Bacteria

Cellular

Bacteria

Technique NanoSIMS +FISH

Analyzed features Chemotaxonomic features, metabolic reactions Chemotaxonomic features, metabolic reactions Chemotaxonomic features, metabolic reactions Metabolic reactions Chemotaxonomic features, metabolic reactions

References Gao et al. 2016; Kopf et al. 2016

Pett-Ridge and Weber 2012; Musat et al. 2016; Dawson et al. 2016 Musat et al. 2008; Dawson et al. 2016

Behrens et al. 2012 Musat et al. 2012

30

Analytical Techniques for Single-Cell Studies in Microbiology

715

Rotational-Vibrational Spectroscopy Rotational-vibrational spectroscopy makes it possible to record the spectra of quantum transitions between rotational-vibrational energy levels of molecules after excitation by light. Rotational-vibrational spectra are recorded by infrared spectroscopy (IRS) and Raman spectroscopy (RS) (named after C.V. Raman who first described the phenomenon of inelastic scattering of light). IRS enables to measure absorption spectra of infrared radiation caused by rotational-vibrational energy transitions in the molecules. RS is based on the effect of so-called inelastic light scattering. When reflected from molecules, the bulk of the light is scattered in the course of the so-called elastic (Rayleigh) scattering. The spectrum of reflected light does not change. However, a small fraction of the photons undergoes inelastic reflection due to their interaction with the molecules, which is associated with rotational-vibrational energy transitions, and their spectral properties change. This radiation is recorded as a Raman spectrum. It should be emphasized that IRS and RS carry information about different rotational-vibrational transitions of the molecules. Absorption of infrared light is determined by changes in the electrical moment of the molecule during absorption, while the intensity of Raman scattering depends on how the polarizability of the molecule changes when the transition occurs. The changes of polarizability and electrical moment can be differently expressed at various transitions. Therefore, these changes are not equally represented in the IRS and RS. In particular, the polar groups of organic compounds (C¼O, N-H, O-H) are well expressed in the infrared spectra, and nonpolar groups (C¼C, C-C, S-S) are better visible in the Raman spectra (Ferraro et al. 2003; Stuart 2004). IRS and RS have long been used to characterize multicellular microbial preparations with conventional instruments that provide averaged data on the chemical composition of the main components of microbial cells in the population. An example is the use of these methods for identification at the level of genus, species, and even subspecies, since rotational-vibrational spectra of microorganisms are a kind of their “fingerprints” (Lu et al. 2011). In some studies, IRS and RS were used for investigating bacterial biofilm development, mechanisms of cell damage, susceptibility to antibiotics, and a number of other microbial traits and features (Lu et al. 2011; Santos et al. 2015). Several methods have been developed that can be used to study single microbial cells using RS (micro RS), namely, Surface-Enhanced Raman Spectroscopy (SERS), Tip-Enhanced Raman Spectroscopy (TERS), Coherent Anti-Stokes Raman Spectroscopy (CARS), and Resonance Raman Spectroscopy (RRS) (Li et al. 2012; Harrison and Berry 2017). Given that Raman scattering is a very low-intensity radiation, in all these methods, certain specific means are used to enhance the recorded signals. The main advantage of TERS is its ability to record RS with a spatial resolution below the optical diffraction limit (Domke and Pettinger 2010). For instance, TERS was used to study surface structures, including the dynamics of polysaccharides and peptides in bacteria (Neugebauer et al. 2007), as well as for the identification of bacteria and yeast at the level of single cells (Neugebauer et al. 2007; Harz et al. 2009). Enhancement of the Raman scattering signal can also be achieved by using ultraviolet light irradiation. Nucleic acids and aromatic amino

716

E. Puchkov

acids of the proteins of bacterial cells at different growth phases of culture development were studied by this method (Neugebauer et al. 2007). The so-called Raman images of microorganisms can be recorded using micro Raman spectroscopy (Raman imaging) similar to the “chemical images” obtained by NanoSIMS (see section “Nanoscale Secondary Ion Mass Spectrometry”) under in situ or in vivo conditions (Li et al. 2012; Harrison and Berry 2017). For example, taking into account the taxonomic specificity of Raman spectra (see above), the possibility of identification and localization of Mycobacterium gordonae in macrophages (Silge et al. 2015) and of identification, localization, and even determination of the physiological state of different S. aureus populations in endothelial cells (Große et al. 2015) by the use of RRS and special methods of mathematical statistics were shown. RRS and SERS techniques were used for visualization of a rhizosphere bacterium, Pantoea sp. YR343, on the surface of Arabidopsis thaliana roots (Polisetti et al. 2016). Of particular interest is the application of micro RS in combination with stable isotope probing, Raman-SIP (Raman Stable Isotope Probing) (Wang et al. 2016a). The method is based on the fact that the replacement of an atom by its isotope in a chemical compound leads to a shift in the characteristic bands of the atom in the Raman spectrum. This opens up the way for studies of the metabolism of certain compounds by introducing the substrates labeled with stable isotopes into the system under study. Various substrates have been introduced into the culture medium as metabolites labeled with a stable isotope in the studies of the metabolism of whole microbial cells. To date, the possibility of using the substrates labeled with 13C, 15N, and 2H isotopes has been investigated. The most significant results so far have been obtained mainly with 13C- and 2H-labeled analogs of natural substrates. An original method based on the reverse substitution of 13C metabolites by natural 12C metabolites was developed to solve the problem of the absence or high cost of various 13C substrates (Wang et al. 2016b). This method has been used in the study of the symbiotic relationship between bacteria Acinetobacter baylyi ADP1 and E. coli DH5α-GFP. Initially, the cells were grown with commercially available 13CD-glucose as a substrate with the addition of D2O. After 13C and 2H labeling of both strains, the cells were incubated with 12C-citrate. It was found by micro RS that the cells of E. coli Dh5a-GFP, which do not utilize citrate in monoculture, acquired this ability in the presence of the A. baylyi ADP1 cells. Additional studies using mass spectrometry allowed the speculation that putrescine and phenylalanine excreted by A. baylyi ADP1 were probably stimulants of citrate anabolism in E. coli Dh5a-GFP. Another prospect of the use of micro RS is associated with preparative isolation of individual cells with certain properties, Raman Activated Cell Sorting (RACS). The advantages of this approach are noninvasiveness, the absence of additional chemical treatments such as fluorescent probing in FACS (see section “Cytometry”), as well as the possibility to carry out selection of the cells identified by a certain chemical composition of metabolites (chemical “fingerprint”). However, the development of this methodology is constrained by technical difficulties mainly due to the low intensity of Raman light scattering and the need to record spectra at a rate that allows real-time physical separation of the cells possessing certain characteristics (Pahlow et al. 2015; Song et al. 2016).

30

Analytical Techniques for Single-Cell Studies in Microbiology

717

Two examples of successful application of bacterial cell isolation using RACS can be presented, in which a well-defined spectral band of carotenoids was used as a distinguishing feature for preparative cell sorting. The first example is related to the use of an RACS micro flow system, which was developed for continuous separation of carotenoid-containing bacteria. Cyanobacteria Synechocystis sp. PCC6803 were used as a model object. The possibility of separating the cells cultured with 12C and 13C substrates and containing 12C and 13C carotenoids, respectively, with an efficiency of 96.3% was shown (McIlvenna et al. 2016). In another study, RACS separation of bacteria in the samples of seawater of the Red sea was carried out. In this case, identification was based on carotenoid spectra, too, but cell separation was performed using the so-called cell ejection from the substrate on which the cells were fixed. The subsequent genomic analysis of isolated cells revealed new functional genes, encoding the biosynthesis of carotenoids and isoprenoids, and previously unknown phototrophic microorganisms, including unculturable Cyanobacteria spp. (Song et al. 2017). The applications of micro RS in the studies of single microbial cells described in this section are summed up in Table 5. Table 5 Selected applications of micro Raman spectroscopy in the studies of single microbial cells

Technique Tip-enhanced Raman spectroscopy (TERS)

Structural level of analysis Cellular, subcellular

Studied microorganisms Bacteria

Yeasts

Analyzed features Surface polysaccharides and peptides, chemotaxonomic features, proteins and nucleic acids in dynamics Chemotaxonomic features

References Neugebauer et al. 2007; Harz et al. 2009

Neugebauer et al. 2007; Harz et al. 2009 Silge et al. 2015; Große et al. 2015; Harrison and Berry 2017; Polisetti et al. 2016 Polisetti et al. 2016

Resonance Raman spectroscopy (RRS)

Cellular, subcellular

Bacteria

Chemotaxonomic features

Surfaceenhanced Raman spectroscopy (SERS) Raman stable isotope probing (Raman-SIP) Raman activated cell sorting (RACS)

Cellular, subcellular

Bacteria

Chemotaxonomic features

Cellular

Bacteria

Metabolic features with regard to specific substrates

Wang et al. 2016a, b

Cellular

Bacteria

Isolation of bacteria with Raman-spectra specific properties

McIlvenna et al. 2016; Song et al. 2017

718

E. Puchkov

Conclusion Currently, there is a large toolbox of methods for studies of the physical and chemical properties of single cells of microorganisms available to microbiologists. All these methods are based on different principles and, accordingly, allow for the multilateral characterization of microbial cells. This opens up the prospect of a new approach to the solution of many fundamental and applied problems. The methods of quantitative analysis of single cells are fruitfully applied in the studies of the nature of heterogeneity of microbial populations. These methods have already provided unique information on the structural and functional organization of bacteria and yeasts at the molecular level, including the data obtained in the dynamics of their development. The opportunity for direct in situ study in combination with the methods of genomics of the microorganisms unculturable in the laboratory (“microbial dark matter”) emerged (Solden et al. 2016). Studies are underway with the use of data obtained on single cells of microorganisms in systemic and synthetic biology, as well as for data mining information technology. Quantitative analysis of the interaction of single microbial cells with the cells of a macroorganism will undoubtedly play an important role in the study and treatment of infectious diseases. Further development in this area, including the improvement of its methodological component, will undoubtedly contribute to the deepening of our understanding of the microbial world.

References Abucayon E, Ke N, Cornut R, Patelunas A et al (2014) Investigating catalase activity through hydrogen peroxide decomposition by bacteria biofilms in real time using scanning electrochemical microscopy. Anal Chem 86(1):498–505. https://doi.org/10.1021/ac402475m Akamatsu M, Lin Y, Bewersdorf J, Pollard TD (2017) Analysis of interphase node proteins in fission yeast by quantitative and superresolution fluorescence microscopy. Mol Biol Cell 28(23):3203–3214. https://doi.org/10.1091/mbc.E16-07-0522 Altinoglu I, Merrifield CJ, Yamaichi Y (2019) Single molecule super-resolution imaging of bacterial cell pole proteins with high-throughput quantitative analysis pipeline. Sci Rep 9:6680. https://doi.org/10.1038/s41598-019-43051-7 Ambriz-Aviña V, Contreras-Garduño JA, Pedraza-Reyes M (2014) Applications of flow cytometry to characterize bacterial physiological responses. Biomed Res Int 2014:461941. https://doi.org/ 10.1155/2014/461941 Angeloni L, Passeri D, Reggente M et al (2016) Microbial cells force spectroscopy by atomic force microscopy: a review. Nanosci Nanometrol 2(1):30–40. https://doi.org/10.11648/j.nsnm. 20160201.13 Arasada R, Sayyad WA, Berro J, Pollard TD (2018) High-speed superresolution imaging of the proteins in fission yeast clathrin-mediated endocytic actin patches. Mol Biol Cell 29(3):295– 303. https://doi.org/10.1091/mbc.E17-06-0415 Avery SV (2006) Microbial cell individuality and the underlying sources of heterogeneity. Nat Rev Microbiol 4(8):577–587 Ball DA, Mehta GD, Salomon-Kent R, Mazza D, Morisaki T, Mueller F, McNally JG, Karpova TS (2016) Single molecule tracking of Ace1p in Saccharomyces cerevisiae defines a characteristic residence time for non-specific interactions of transcription factors with chromatin. Nucleic Acids Res 44(21):e160

30

Analytical Techniques for Single-Cell Studies in Microbiology

719

Balomenos AD, Tsakanikas P, Aspridou Z et al (2017) Image analysis driven single-cell analytics for systems microbiology. BMC Syst Biol 11(1):43. https://doi.org/10.1186/ s12918-017-0399-z Barteneva NS, Fasler-Kan E, Vorobjev IA (2012) Imaging flow cytometry: coping with heterogeneity in biological systems. J Histochem Cytochem 60(10):723–733 Baudart J, Olaizola A, Coallier J et al (2005) Assessment of a new technique combining a viability test, whole-cell hybridization and laser-scanning cytometry for the direct counting of viable Enterobacteriaceae cells in drinking water. FEMS Microbiol Lett 243(2):405–409 Behrens S, Kappler A, Obst M (2012) Linking environmental processes to the in situ functioning of microorganisms by high-resolution secondary ion mass spectrometry (NanoSIMS) and scanning transmission X-ray microscopy (STXM). Environ Microbiol 14(11):2851–2869. https://doi.org/ 10.1111/j.1462-2920.2012.02724.x Bestul AJ, Yu Z, Unruh JR, Jaspersen SL (2017) Molecular model of fission yeast centrosome assembly determined by superresolution imaging. J Cell Biol 216(8):2409–2424. https://doi. org/10.1083/jcb.201701041 Bhushan B, Marti O (2011) Scanning probe microscopy – principle of operation, instrumentation and probes. In: Bhushan B (ed) Nanotribology and nanomechanics. Springer, Berlin/Heidelberg, pp 37–110. https://doi.org/10.1007/978-3-642-15283-2_2 Bjerling P, Olsson I, Meng X (2012) Quantitative live cell fluorescence-microscopy analysis of fission yeast. J Vis Exp 59:e3454. https://doi.org/10.3791/3454 Brehm-Stecher BF, Johnson EA (2004) Single-cell microbiology: tools, technologies, and applications. Microbiol Mol Biol Rev 68(3):538–559 Burns S, Avena JS, Unruh JR, Yu Z, Smith SE, Slaughter BD, Winey M, Jaspersen SL (2015) Structured illumination with particle averaging reveals novel roles for yeast centrosome components during duplication.eLife 4:e08586. https://doi.org/10.7554/eLife.08586 Calvert ME, Lannigan JA, Pemberton LF (2008) Optimization of yeast cell cycle analysis and morphological characterization by multispectral imaging flow cytometry. Cytometry A 73(9):825–833. https://doi.org/10.1002/cyto.a.20609 Chen S-C, Zhao T, Gordon GJ et al (2007) Automated image analysis of protein localization in budding yeast. Bioinformatics 23:i66–i71. https://doi.org/10.1093/bioinformatics/btm206 Chen CH, Cho SH, Chiang HI et al (2011) Specific sorting of single bacterial cells with microfabricated fluorescence-activated cell sorting and tyramide signal amplification fluorescence in situ hybridization. Anal Chem 83(19):7269–7275. https://doi.org/10.1021/ac2013465 Chen J, McSwiggen D, Ünal E (2018) Single molecule fluorescence in situ hybridization (smFISH) analysis in budding yeast vegetative growth and meiosis. J Vis Exp 135:e57774. https://doi.org/ 10.3791/57774. Choi H, Rangarajan N, Weisshaar JC (2016) Lights, camera, action! Antimicrobial peptide mechanisms imaged in space and time. Trends Microbiol 24(2):111–122. https://doi.org/10.1016/j. tim.2015.11.004 Chong YT, Cox MJ, Andrews B (2012) Proteome-wide screens in Saccharomyces cerevisiae using the yeast GFP collection. Adv Exp Med Biol 736:169–178. https://doi.org/10.1007/978-1-4419-72101_8 Chozinski TJ, Gagnon LA, Vaughan JC (2014) Twinkle, twinkle little star: photoswitchable fluorophores for super-resolution imaging. FEBS Lett 588(19):3603–3612. https://doi.org/10. 1016/j.febslet.2014.06.043 Coltharp C, Xiao J (2012) Superresolution microscopy for microbiology. Cell Microbiol 14(12):1808–1818 Connella JL, Kimb J, Shearb JB et al (2014) Real-time monitoring of quorum sensing in 3D-printed bacterial aggregates using scanning electrochemical microscopy. Proc Natl Acad Sci U S A 111(551):18255–18260. https://doi.org/10.1073/pnas.1421211111 Cortesi M, Bandiera L, Pasini A et al (2017) Reliable measurement of E. coli single cell fluorescence distribution using a standard microscope set-up. J Biol Eng 11:8. https://doi.org/10.1186/ s13036-017-0050-y

720

E. Puchkov

Dashkova V, Malashenkov D, Poulton N et al (2017) Imaging flow cytometry for phytoplankton analysis. Methods 112:188–200. https://doi.org/10.1016/j.ymeth.2016.05.007 Dawson KS, Scheller S, Dillon JG et al (2016) Stable isotope phenotyping via cluster analysis of NanoSIMS data as a method for characterizing distinct microbial ecophysiologies and sulfurcycling in the environment. Front Microbiol 7:774. https://doi.org/10.3389/fmicb.2016.00774 Dazzo FB, Niccum BC (2015) Use of CMEIAS image analysis software to accurately compute attributes of cell size, morphology, spatial aggregation and color segmentation that signify in situ ecophysiological adaptations in microbial biofilm communities. Computation 3:72–98 Dazzo FB, Sexton R, Jain A et al (2017) Influence of substratum hydrophobicity on the geomicrobiology of river biofilm architecture and ecology analyzed by CMEIAS bioimage informatics. Geosciences 7(56):1–36. https://doi.org/10.3390/geosciences7030056 Dekel E, Rivkin A, Heidenreich M et al (2017) Identification and classification of the malaria parasite blood developmental stages, using imaging flow cytometry. Methods 112:157–166. https://doi.org/10.1016/j.ymeth.2016.06.021 Domke KF, Pettinger B (2010) Studying surface chemistry beyond the diffraction limit: 10 years of TERS. ChemPhysChem 11(7):1365–1373. https://doi.org/10.1002/cphc.200900975 Dorobantu LS, Goss GG, Burrell RE (2012) Atomic force microscopy: a nanoscopic view of microbial cell surfaces. Micron 43(12):1312–1322. https://doi.org/10.1016/j.micron.2012.05. 005 Drechsler-Hake D, Alamir H, Hahn J et al (2016) Mononuclear phagocytes contribute to intestinal invasion and dissemination of Yersinia enterocolitica. Int J Med Microbiol 306(6):357–366. https://doi.org/10.1016/j.ijmm.2016.04.002 Dufrêne YF (2014) Atomic force microscopy in microbiology: new structural and functional insights into the microbial cell surface. MBio 5(4):e01363–e01314. https://doi.org/10.1128/ mBio.01363-14 Emerson JB, Adams RI, Román CMB et al (2017) Schrödinger’s microbes: tools for distinguishing the living from the dead in microbial ecosystems. Microbiome 5(10):86. https://doi.org/10.1186/ s40168-017-0285-3 Endesfelder U (2019) From single bacterial cell imaging towards in vivo single-molecule biochemistry studies. Essays Biochem 63(2):187–196. https://doi.org/10.1042/EBC20190002 Femino AM, Fay FS, Fogarty K, Singer RH (1998) Visualization of single RNA transcripts in situ. Science 280(5363):585–590 Ferraro J, Nakamoto K, Brown CW (2003) Introductory Raman spectroscopy, 2nd edn. Elsevier, San Diego Flynn JD, Haas BL, Biteen JS (2016) Plasmon-enhanced fluorescence from single proteins in living bacteria. J Phys Chem C 120(37):20512–20517. https://doi.org/10.1021/acs.jpcc.5b08049 Gahlmann A, Moerner WE (2014) Exploring bacterial cell biology with single-molecule tracking and super-resolution imaging. Nat Rev Microbiol 12(1):9–22. https://doi.org/10.1038/ nrmicro3154 Gangan MS, Athale CA (2017) Threshold effect of growth rate on population variability of Escherichia coli cell lengths. R Soc Open Sci 4(2):160417. https://doi.org/10.1098/rsos.160417 Gao L, Smith RT (2015) Optical hyperspectral imaging in microscopy and spectroscopy - a review of data acquisition. J Biophotonics 8(6):441–456. https://doi.org/10.1002/jbio.201400051 Gao D, Huang X, Tao Y (2016) A critical review of NanoSIMS in analysis of microbial metabolic activities at single-cell level. Crit Rev Biotechnol 36(5):884–890. https://doi.org/10.3109/ 07388551.2015.1057550 Gebre A, Okada H, Kim C et al (2015) Profiling of the effects of antifungal agents on yeast cells based on morphometric analysis. FEMS Yeast Res 15(5):fov040. https://doi.org/10.1093/ femsyr/fov040 Große C, Bergner N, Dellith J et al (2015) Label-free imaging and spectroscopic analysis of intracellular bacterial infections. Anal Chem 87(4):2137–2142. https://doi.org/10.1021/ ac503316s

30

Analytical Techniques for Single-Cell Studies in Microbiology

721

Haas BL, Matson JS, DiRita VJ, Biteen JS (2014) Imaging live cells at the nanometer-scale with single-molecule microscopy: obstacles and achievements in experiment optimization for microbiology. Molecules 19(8):12116–12149. https://doi.org/10.3390/molecules190812116 Hagen N, Kudenov MW (2003) Review of snapshot spectral imaging technologies. Opt Eng 52(9):090901 Han J, Kamber M, Pei J (2011) Data mining: concepts and techniques, 3rd edn. Elsevier, Amsterdam Haridas V, Ranjbar S, Vorobjev IA et al (2017) Imaging flow cytometry analysis of intracellular pathogens. Methods 112:91–104. https://doi.org/10.1016/j.ymeth.2016.09.007 Harrison JP, Berry D (2017) Vibrational spectroscopy for imaging single microbial cells in complex biological samples. Front Microbiol 8:675. https://doi.org/10.3389/fmicb.2017.00675 Harz M, Rösch P, Popp J (2009) Vibrational spectroscopy – a powerful tool for the rapid identification of microbial cells at the single-cell level. Cytometry A 75(20):104–113. https:// doi.org/10.1002/cyto.a.20682 Hennig H, Rees P, Blasi T et al (2017) An open-source solution for advanced imaging flow cytometry data analysis using machine learning. Methods 112:201–210. https://doi.org/10. 1016/j.ymeth.2016.08.018 Herman-Bausier P, Formosa-Dague C, Feuillie C et al (2017) Forces guiding staphylococcal adhesion. J Struct Biol 197(1):65–69. https://doi.org/10.1016/j.jsb.2015.12.009 Herzenberg LA, Parks D, Sahaf B et al (2002) The history and future of the fluorescence activated cell sorter and flow cytometry: a view from Stanford. Clin Chem 48(10):1819–1827 Ho SH, Tirrell DA (2019) Enzymatic labeling of bacterial proteins for super-resolution imaging in live cells. ACS Central Sci 5(12):1911–1919. https://doi.org/10.1021/acscentsci.9b00617 Huang F, Hartwich TM, Rivera-Molina FE, Lin Y, Duim WC, Long JJ, Uchil PD, Myers JR, Baird MA, Mothes W, Davidson MW, Toomre D, Bewersdorf J (2013) Video-rate nanoscopy using sCMOS camera-specific single-molecule localization algorithms. Nat Methods 10(7):653–658. https://doi.org/10.1038/nmeth.2488 Huang Q, Wu H, Cai P et al (2015) Atomic force microscopy measurements of bacterial adhesion and biofilm formation onto clay-sized particles. Sci Rep 5:16857. https://doi.org/10.1038/ srep16857 Ishii S, Tago K, Senoo K (2010) Single-cell analysis and isolation for microbiology and biotechnology: methods and applications. Appl Microbiol Biotechnol 86(5):1281–1292. https://doi. org/10.1007/s00253-010-2524-4 James SA, Hilal N, Wright CJ (2017) Atomic force microscopy studies of bioprocess engineering surfaces – imaging, interactions and mechanical properties mediating bacterial adhesion. Biotechnol J 12:7. https://doi.org/10.1002/biot.201600698 Kamentsky LA, Kamentsky LD (1991) Microscope-based multiparameter laser scanning cytometer yielding data comparable to flow cytometry data. Cytometry 12(5):381–387 Kan A (2017) Machine learning applications in cell image analysis. Immunol Cell Biol 95(6):525– 530. https://doi.org/10.1038/icb.2017.16 Kao HP, Verkman AS (1994) Tracking of single fluorescent particles in three dimensions: use of cylindrical optics to encode particle position. Biophys J 67(3):1291–1300. https://doi.org/10. 1016/S0006-3495(94)80601-0 Kentner D, Sourjik V (2010) Use of fluorescence microscopy to study intracellular signaling in bacteria. Annu Rev Microbiol 64:373–390. https://doi.org/10.1146/annurev.micro.112408. 134205 Koley D, Ramsey MM, Bard AJ et al (2011) Discovery of a biofilm electrocline using real-time 3D metabolite analysis. Proc Natl Acad Sci U S A 108(50):19996–20001. https://doi.org/10.1073/ pnas.1117298108 Kopf SH, Sessions AL, Cowley ES et al (2016) Trace incorporation of heavy water reveals slow and heterogeneous pathogen growth rates in cystic fibrosis sputum. Proc Natl Acad Sci U S A 113(2):E110–E116. https://doi.org/10.1073/pnas.1512057112

722

E. Puchkov

Lakowicz JR (2006) Principles of fluorescence spectroscopy, 3rd edn. Springer Science and Business Media, Berlin. https://doi.org/10.1007/978-0-387-46312-4 Laplante C, Huang F, Tebbs IR, Bewersdorf J, Pollard TD (2016) Molecular organization of cytokinesis nodes and contractile rings by super-resolution fluorescence microscopy of live fission yeast. Proc Natl Acad Sci U S A 113(40):E5876–E5885 Li G, Neuert G (2019) Multiplex RNA single molecule FISH of inducible mRNAs in single yeast cells. Sci Data 6(1):94. https://doi.org/10.1038/s41597-019-0106-6. Li M, Xu J, Romero-Gonzalez M et al (2012) Single cell Raman spectroscopy for cell sorting and imaging. Curr Opin Biotechnol 23:56–63 Liu X, Ramseyb MM, Chena X et al (2011a) Real-time mapping of a hydrogen peroxide concentration profile across a polymicrobial bacterial biofilm using scanning electrochemical microscopy. Proc Natl Acad Sci U S A 108(7):72668–72673. https://doi.org/10.1073/pnas. 1018391108 Liu Y, Aubrey W, Martin K et al (2011b) The analysis of yeast cell morphology features in exponential and stationary phase. J Biol Syst 19:561–575. https://doi.org/10.1142/ S0218339011003968 Liu C, Liu YL, Perillo EP, Dunn AK, Yeh HC (2016) Single-molecule tracking and its application in biomolecular binding detection. IEEE J Sel Top Quantum Electron 22(4):6804013 Lower SK (2011) Atomic force microscopy to study intermolecular forces and bonds associated with bacteria. Adv Exp Med Biol 715:285–299. https://doi.org/10.1007/978-94-007-0940-9_18 Lu X, Al-Qadiri HM, Lin M et al (2011) Application of mid-infrared and Raman spectroscopy to the study of bacteria. Food Bioprocess Technol 4:919–935. https://doi.org/10.1007/s11947-0110516-8 McIlvenna D, Huang WE, Davison P et al (2016) Continuous cell sorting in a flow based on single cell resonance Raman spectra. Lab Chip 16(8):1420–1429. https://doi.org/10.1039/c6lc00251j Minoshima M, Kikuchi K (2017) Photostable and photoswitching fluorescent dyes for superresolution imaging. J Biol Inorg Chem 22(5):639–652. https://doi.org/10.1007/s00775-0161435-y Mulvihill DP (2017) Live cell imaging in fission yeast. Cold Spring Harb Protoc 2017(10):pdb. top090621. https://doi.org/10.1101/pdb.top090621 Musat N, Halm H, Winterholler B et al (2008) A single-cell view on the ecophysiology of anaerobic phototrophic bacteria. Proc Natl Acad Sci U S A 105(46):17861–17866. https://doi.org/10. 1073/pnas.0809329105 Musat N, Foster R, Vagner T et al (2012) Detecting metabolic activities in single cells, with emphasis on nanoSIMS. FEMS Microbiol Rev 36(2):486–511. https://doi.org/10.1111/j.15746976.2011.00303.x Musat N, Musat F, Weber PK et al (2016) Tracking microbial interactions with NanoSIMS. Curr Opin Biotechnol 4:114–121. https://doi.org/10.1016/j.copbio.2016.06.007 Negishi T, Nogami S, Ohya Y (2009) Multidimensional quantification of subcellular morphology of Saccharomyces cerevisiae using CalMorph, the high-throughput image-processing program. J Biotechnol 141:109–117. https://doi.org/10.1016/j.jbiotec.2009.03.014 Neugebauer U, Schmid U, Baumann K et al (2007) Towards a detailed understanding of bacterial metabolism – spectroscopic characterization of Staphylococcus epidermidis. ChemPhysChem 8(1):124–137 Niven GW, Fuks T, Morton JS et al (2006) A novel method for measuring lag times in division of individual bacterial cells using image analysis. J Microbiol Methods 65:311–317. https://doi. org/10.1016/j.mimet.2005.08.006 Nketia TA, Sailem H, Rohde G, Machiraju R, Rittscher J (2017) Analysis of live cell images: methods, tools and opportunities. Methods 115:65–79. https://doi.org/10.1016/j.ymeth.2017.02. 007 Nogami S, Ohya Y, Yvert G (2007) Genetic complexity and quantitative trait loci mapping of yeast morphological traits. PLoS Genet 3:e31

30

Analytical Techniques for Single-Cell Studies in Microbiology

723

Nuñez J, Renslow R, Cliff JB 3rd et al (2017) Nano SIMS for biological applications: current practices and analyses. Biointerphases 13(3):03B301. https://doi.org/10.1116/1.4993628 Ogawa M, Tani K, Ochiai N et al (2005) Multicolour digital image analysis system for identification of bacteria and concurrent assessment of their respiratory activity. J Appl Microbiol 98:1101– 1106. https://doi.org/10.1111/j.1365-2672.2005.02551.x Ohtani M, Saka A, Sano F et al (2004) Development of image processing program for yeast cell morphology. J Bioinform Computat Biol 1:695–709. https://doi.org/10.1142/ S0219720004000363 Ohya Y, Sese J, Yukawa M et al (2005) High-dimensional and large-scale phenotyping of yeast mutants. Proc Natl Acad Sci U S A 102(52):19015–19020. https://doi.org/10.1073/pnas. 0509436102 Okagaki LH, Strain AK, Nielsen JN et al (2010) Cryptococcal cell morphology affects host cell interactions and pathogenicity. PLoS Pathog 6(6):e1000953. https://doi.org/10.1371/annotation/ 1b59fd9e-9ac9-4ea8-a083-14c413c80b03 Ott W, Jobst MA, Schoeler C et al (2017) Single-molecule force spectroscopy on polyproteins and receptor-ligand complexes: the current toolbox. J Struct Biol 197(1):3–12. https://doi.org/10. 1016/j.jsb.2016.02.011 Pahlow S, Meisel S, Cialla-May D et al (2015) Isolation and identification of bacteria by means of Raman spectroscopy. Adv Drug Deliv Rev 89:105–120. https://doi.org/10.1016/j.addr.2015.04.006 Pett-Ridge J, Weber PK (2012) NanoSIP: NanoSIMS applications for microbial biology. Methods Mol Biol 881:375–408. https://doi.org/10.1007/978-1-61779-827-6_13 Phanse Y, Ramer-Tait AE, Friend SL et al (2012) Analyzing cellular internalization of nanoparticles and bacteria by multi-spectral imaging flow cytometry. J Vis Exp 8(64):e3884. https://doi.org/ 10.3791/3884 Phillips ZF, Chen M, Waller L (2017) Single-shot quantitative phase microscopy with colormultiplexed differential phase contrast (cDPC). PLoS One 12(2):e0171228 Pina-Vaz C, Costa-Oliveira S, Rodrigues AG et al (2004) Novel method using a laser scanning cytometer for detection of mycobacteria in clinical samples. J Clin Microbiol 42(2):906–908 Polisetti S, Bible AN, Morrell-Falvey JL et al (2016) Raman chemical imaging of the rhizosphere bacterium Pantoea sp. YR343 and its co-culture with Arabidopsis thaliana. Analyst 141:2175– 2182. https://doi.org/10.1039/c6an00080k Pozarowski P, Holden E, Darzynkiewicz Z (2013) Laser scanning cytometry: principles and applications-an update. Methods Mol Biol 931:187–212. https://doi.org/10.1007/978-162703-056-4_11 Puchkov EO (2013) Intracellular viscosity: methods of measurement and role in metabolism. Biochemistry (Moscow) Suppl Ser A Membr Cell Biol 7(4):270–279. https://doi.org/10.1134/ s1990747813050140 Puchkov EO (2014) Computer image analysis as a tool for microbial viability assessment: examples of use and prospects. J Biosci Med 2:1–6. https://doi.org/10.4236/jbm.2014.23001 Puchkov E (2016a) Image analysis in microbiology: a review. J Comput Commun 4:8–32 Puchkov E (2016b) Microfluorimetry of single yeast cells by fluorescence microscopy combined with digital photography and computer image analysis. In: Berhardt LV (ed) Advances in medicine and biology, vol 98. Nova Science Publishers, New York, pp 69–90 Regina VR, Poulsen M, Søhoel H et al (2012) Quantification of bacteria on abiotic surfaces by laser scanning cytometry: an automated approach to screen the antifouling properties of new surface coatings. J Lab Autom 17(4):293–301. https://doi.org/10.1177/2211068212450013 Saito TL, Ohtani M, Sawai H et al (2004) SCMD: Saccharomyces cerevisiae morphological database. Nucleic Acid Res 32:D319–D322. https://doi.org/10.1093/nar/gkh113 Sanderson MJ, Smith I, Parker I, Bootman MD (2014) Fluorescence microscopy. Cold Spring Harb Protoc 2014(10):pdb.top071795. https://doi.org/10.1101/pdb.top071795 Santos MI, Gerbino E, Tymczyszyn E et al (2015) Applications of infrared and Raman spectroscopies to probiotic investigation. Foods 4(3):283–305. https://doi.org/10.3390/foods4030283

724

E. Puchkov

Saurabh S, Perez AM, Comerci CJ, Shapiro L, Moerner WE (2016) Super-resolution imaging of live bacteria cells using a genetically directed, highly photostable fluoromodule. J Am Chem Soc 138(33):10398–10401. https://doi.org/10.1021/jacs.6b05943 Sbalzarini IF (2016) Seeing is believing: quantifying is convincing: computational image analysis in biology. Adv Anat Embryol Cell Biol 219:1–39. https://doi.org/10.1007/978-3319-28549-8_1. Schneider JP, Basler M (2016) Shedding light on biology of bacterial cells. Phil Trans R Soc B 371:20150499. https://doi.org/10.1098/rstb.2015.0499 Schulze K, López DA, Tillich UM et al (2011) A simple viability analysis for unicellular cyanobacteria using a new autofluorescence assay, automated microscopy, and image J. BMC Biotechnol 11:118–125. https://doi.org/10.1186/1472-6750-11-118 Shapiro HM (2000) Microbial analysis at the single-cell level: tasks and techniques. J Microbiol Methods 42(1):3–16 Silge A, Abdou E, Schneider K et al (2015) Shedding light on host niches: label-free in situ detection of Mycobacterium gordonae via carotenoids in macrophages by Raman microspectroscopy. Cell Microbiol 17:832–842. https://doi.org/10.1111/cmi.12404 Smirnov A, Solga MD, Lannigan J et al (2017) High-throughput particle uptake analysis by imaging flow cytometry. Curr Protoc Cytom 80:11.22.1–11.22.17. https://doi.org/10.1002/ cpcy.19 Snapp E (2005) Design and use of fluorescent fusion proteins in cell biology. Curr Protocols Cell Biol Unit–21.4. https://doi.org/10.1002/0471143030.cb2104s27 Solden L, Lloyd K, Wrighton K (2016) The bright side of microbial dark matter: lessons learned from the uncultivated majority. Curr Opin Microbiol 31:217–226. https://doi.org/10.1016/j.mib. 2016.04.020 Song Y, Yin H, Huang WE (2016) Raman activated cell sorting. Curr Opin Chem Biol 33:1–8. https://doi.org/10.1016/j.cbpa.2016.04.002 Song Y, Kaster AK, Vollmers J et al (2017) Single-cell genomics based on Raman sorting reveals novel carotenoid-containing bacteria in the Red Sea. Microb Biotechnol 10(1):125–137. https:// doi.org/10.1111/1751-7915.12420 Stagge F, Mitronova GY, Belov VN, Wurm CA, Jakobs S (2013) Snap-, CLIP- and halo-tag labelling of budding yeast cells. PLoS One 8(10):e78745. https://doi.org/10.1371/journal. pone.0078745 Stracy M, Kapanidis AN (2017) Single-molecule and super-resolution imaging of transcription in living bacteria. Methods 120:103–114. https://doi.org/10.1016/j.ymeth.2017.04.001 Stuart BH (2004) Infrared spectroscopy: fundamentals and applications. Wiley, Chichester. https:// doi.org/10.1002/0470011149 Sutter M, Faulkner M, Aussignargues C et al (2016) Visualization of bacterial microcompartment facet assembly using high-speed atomic force microscopy. Nano Lett 16(3):1590–1595. https:// doi.org/10.1021/acs.nanolett.5b04259 Suzuki G, Sawai H, Ohtani M, Nogami S, Sano-Kumagai F, Saka A, Yukawa M, Saito TL, Sese J, Hirata D, Morishita S, Ohya Y (2006) Evaluation of image processing programs for accurate measurement of budding and fission yeast morphology. Curr Genet 49(4):237–247 Taniguchi Y, Choi PJ, Li GW et al (2010) Quantifying E. coli proteome and transcriptome with single-molecule sensitivity in single cells. Science 329(5991):533–538. https://doi.org/10.1126/ science.1188308 Terahara N, Kodera N, Uchihashi T et al (2017) Na+-induced structural transition of MotPS for stator assembly of the Bacillus flagellar motor. Sci Adv 3(11):eaao4119. https://doi.org/10.1126/ sciadv.aao4119 Tsien RY (1998) The green fluorescent protein. Annu Rev Biochem 67:509–544 Uphoff S (2016) Super-resolution microscopy and tracking of DNA-binding proteins in bacterial cells. Methods Mol Biol 1431:221–234. https://doi.org/10.1007/978-1-4939-3631-1_16

30

Analytical Techniques for Single-Cell Studies in Microbiology

725

Van Der Hofstadt M, Hüttener M, Juárez A et al (2015) Nanoscale imaging of the growth and division of bacterial cells on planar substrates with the atomic force microscope. Ultramicroscopy 154:29–36. https://doi.org/10.1016/j.ultramic.2015.02.018 Vanhee LM, D’Haese E, Cools I et al (2010) Detection and quantification of bacteria and fungi using solid-phase cytometry. In: Viola Magni M (ed) Detection of bacteria, viruses, parasites and fungi. NATO science for peace and security series a: chemistry and biology. Springer, Dordrecht, pp 25–41 Wallace CT, Jessup M, Bernas T, Peña KA, Calderon MJ, Loughran PA (2018) Basics of digital microscopy. Curr Protoc Cytom 83:12.2.1–12.2.14. https://doi.org/10.1002/cpcy.31 Wang Y, Hammes F, Boon N et al (2009) Isolation and characterization of low nucleic acid (LNA)content bacteria. ISME J 3(8):889–902. https://doi.org/10.1038/ismej.46 Wang P, Robert L, Pelletier J et al (2010a) Robust growth of Escherichia coli. Curr Biol 20(12):1099–1103. https://doi.org/10.1016/j.cub.2010.04.045 Wang Y, Hammes F, De Roy K et al (2010b) Past, present and future applications of flow cytometry in aquatic microbiology. Trends Biotechnol 28(8):416–424. https://doi.org/10.1016/j.tibtech. 2010.04.006 Wang R, Kamgoue A, Normand C, Léger-Silvestre I, Mangeat T, Gadal O (2016) High resolution microscopy reveals the nuclear shape of budding yeast during cell cycle and in various biological states. J Cell Sci 129(24):4480–4495. https://doi.org/10.1242/jcs.188250 Wang Y, Huang WE, Cui L et al (2016a) Single cell stable isotope probing in microbiology using Raman microspectroscopy. Curr Opin Biotechnol 41:34–42. https://doi.org/10.1016/j.copbio. 2016.04.018 Wang Y, Song Y, Tao Y et al (2016b) Reverse and multiple stable isotope probing to study bacterial metabolism and interactions at the single cell level. Anal Chem 88(19):9443–9450 Wollman A, Hedlund EG, Shashkova S, Leake MC (2020) Towards mapping the 3D genome through high speed single-molecule tracking of functional transcription factors in single living cells. Methods 170:82–89. https://doi.org/10.1016/j.ymeth.2019.06.021 Yaginuma H, Kawai S, Tabata KV et al (2014) Diversity in ATP concentrations in a single bacterial cell population revealed by quantitative single-cell imaging. Sci Rep 4:6522. https://doi.org/10. 1038/srep06522 Yamaguchi M, Ohkusu M, Biswas SK et al (2007) Cytological study of cell cycle of the pathogenic yeast Cryptococcus neoformans. Nippon Ishinkin Gakkai Zasshi 48(4):147–152 Yao Z, Carballido-López R (2014) Fluorescence imaging for bacterial cell biology: from localization to dynamics, from ensembles to single molecules. Annu Rev Microbiol 68:459–476. https:// doi.org/10.1146/annurev-micro-091213-113034 Yu BY, Elbuken C, Ren CL et al (2011) Image processing and classification algorithm for yeast cell morphology in a microfluidic chip. J Biomed Opt 16:066008. https://doi.org/10.1117/1. 3589100 Zoski CG (2016) Review–advances in scanning electrochemical microscopy (SECM). J Electrochem Soc 163(4):H3088–H3100. https://doi.org/10.1149/2.0141604jes

Part VI Single Cell Technologies in Cancer

Single Cell Adhesion in Cancer Progression

31

Privita Edwina Rayappan George Edwin and Saumendra Bajpai

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell Adhesion Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cadherins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Integrins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Selectin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Immunoglobulin Superfamily . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cell-To-Extracellular Matrix (ECM) Adhesion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Role of Cell Adhesion in Cancer Progression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Epithelial-Mesenchymal Transition (EMT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tumor Invasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Angiogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Intravasation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Extravasation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Techniques to Study Cell Adhesion and Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bulk Adhesion Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Centrifugation Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hydrodynamic Shear Assays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wash Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Parallel Plate System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rotating Disk System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Radial Flow System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . High-Throughput Cone and Plate (HT-CAP)-Electric Cell-Substrate Impedance Sensing (ECIS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single Cell Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Micropipette Aspiration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Step-Pressure Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biomembrane Force Probe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

730 731 731 732 733 733 734 734 734 734 735 735 736 736 737 737 737 739 739 741 742 742 743 744 744 744 746

P. E. Rayappan George Edwin · S. Bajpai (*) Applied Mechanics, Biomedical Division, Indian Institute of Technology Madras, Chennai, Tamilnadu, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_8

729

730

P. E. Rayappan George Edwin and S. Bajpai

Micropipette Aspiration (Narrow Sense) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Atomic Force Microscopy (AFM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nanofork and Line-Patterned Substratum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical Tweezer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measuring Cell-to-Extracellular Matrix (ECM) Adhesion Strength . . . . . . . . . . . . . . . . . . . . . . . . Förster Resonance Energy Transfer (FRET) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cellular Traction Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantitative Intravital Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluorescent Models to Study Tumor Heterogeneity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical Imaging Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

746 747 749 751 752 754 755 755 759 760 761 762

Abstract

Quantitative assessment of cell adhesion during pathological development can potentially lead to a better elucidation of the mechanisms behind disease progression and improve prognostic accuracy. Concurrent developments in the fields of biosensors and multimodal imaging techniques, and improved understanding of the biophysical principles driving a disease, have contributed to the development of techniques for quantification of cell adhesion strength. Together, these measurements have underscored the importance of a tightly regulated cell adhesion phenotype, exhibited by tissues under pathological progression. Here we discuss some of the techniques that evaluate cell-to-cell and cell-to-substrate adhesion strength across multiple scales of length and time, taking cancer metastasis as our model system.

Introduction Adhesion plays a central role in cellular communication and is instrumental in the development and maintenance of tissues. It is by means of this adhesiveness cells are able to relay signals that regulate major cellular events such as cell migration, cell division, cell differentiation, etc. (Khalili and Ahmad 2015). Typically expressed on cell-surface as transmembrane proteins, adhesion molecules regulate a variety of cellular functions, including signal transduction, cell growth, differentiation, site-specific gene expression, morphogenesis, immunologic function, cell motility, wound healing, and inflammation. Aberrant expression of cell adhesion molecules or signaling mediated by the adhesion molecules results in pathological anomalies such as cancer (Wai Wong et al. 2012). The progression of cancer is a classic case of an orchestrated, multistep process involving cell adhesion molecules that mediate the development of recurrent, invasive, and distant metastases. In brief, the cancer-metastatic cascade begins with the loss of intercellular adhesion followed by exfoliation of cells from the underlying basal lamina. This allows malignant cells to escape from their site of origin, remodel the extracellular matrix, acquire a more motile and invasive phenotype, and finally, invade and

31

Single Cell Adhesion in Cancer Progression

731

metastasize to distant sites. Numerous researches have reported the role played by the cell adhesion molecules in every step of this cascade. Historically, the characterization of the molecular and pathological changes in tissues undergoing metastasis was the primary focus of earliest studies. With the subsequent development of biomechanical tools for cell adhesion quantification, tissue and cell level characterization became more common in research laboratories, with focus on mechanical perturbation of cells and biomolecules in their physiological states at pico- and nanoscale resolutions. It is expected that these multipronged approaches will offer an alternative and better quantification of the onset and progression of diseases, besides identifying targets for therapeutic interventions (Salant et al. 2006).

Cell Adhesion Molecules Adhesion molecules are divided into four major groups: the cadherins, the integrins, the selectins, and the members of the immunoglobulin superfamily (Figs. 1 and 2).

Cadherins Cadherins are integral transmembrane single-chain glycoproteins, which mediate intercellular cell-cell adhesion through homotypic interactions (between molecules of the same type) in the presence of extracellular calcium. Cadherin subfamily

Fig. 1 Schematic of cell-to-cell and cell-to-matrix adhesion along with their contributing molecules. (Adapted from Molecular Biology of the cell, fourth Edition)

732

P. E. Rayappan George Edwin and S. Bajpai

Fig. 2 Schematic of the major classes of cell adhesion molecules showing their extracellular domain, membrane-spanning domain, and cytosolic domain. (a) Cadherin, (b) Integrin, (c) Selectin, and (d) IgCAMs. (Adapted from Molecular Biology of the cell, fourth Edition)

includes E (epithelial)-, N (neural)-, and P (placental)-cadherin (Makrilia et al. 2009). Cadherin function includes tissue morphogenesis, cell recognition and sorting, boundary formation and maintenance, coordinated cell movements, and the induction and maintenance of structural and functional cell and tissue polarity (Halbleib and Nelson 2006).

Integrins Integrins are cell-surface glycoprotein receptors that are composed of a set of non-covalently associated α and β subunits. There are 18 α and 8 β subunits capable

31

Single Cell Adhesion in Cancer Progression

733

of forming 24 known combinations (Luo et al. 2007; Stupack 2007). Integrin consists of a large extracellular domain that binds extracellular ligands (collagen, laminin, fibrinogen, fibronectin, etc.), a single transmembrane domain, and a relatively small cytoplasmic tail, which associates with the intercellular actin fibers (Luo et al. 2007). Integrins are known to be involved in a bidirectional signaling mechanism: (a) inside-out signaling, where intracellular signals induce alterations in the conformational status of the integrin receptor and these alterations are associated with changes in its ligand-binding properties and (b) outside-in signaling, where extracellular ligands result in downstream signaling and drives the cell toward cell division, migration, etc. (Giancotti and Ruoslahti 1999). Integrins transmit signals by interacting with the cytoskeletal structural proteins such as α-actinin, talin and vinculin, and signaling proteins such as focal adhesion kinase (FAK). Through these multiple pathways and signaling events, integrins regulate gene expression and cell cycle progression, enabling cell survival, and inducing proliferation, migration, and differentiation (Giancotti and Ruoslahti 1999; Stupack and Cheresh 2002).

Selectin Selectins are a family of transmembrane proteins containing an amino-terminal lectin domain for binding to ligands, an epidermal growth factor-like domain, 2–9 regulatory protein repeats, a transmembrane domain, and a short cytoplasmic tail. This class of CAMs consists of three proteins, namely, E-, L-, and P-selectin, which mediate the adhesion between endothelial cells, leukocytes, and platelets, respectively (Makrilia et al. 2009). In particular, E-selectin is expressed on the endothelium and accounts for leukocyte adhesion at the site of inflammation or injury in response to inflammatory cytokines. This particular role of selectin seems to be partly responsible for the capture of circulating tumor cells and the organ-specific migration of tumor (Makrilia et al. 2009).

Immunoglobulin Superfamily The immunoglobulin CAMs (IgCAMs) are a large group of cell surface glycoproteins which contain an immunoglobulin-like extracellular domain. Members of this family comprise NCAM (neural), VCAM (vascular), ICAM (Intercellular), plateletendothelial (PECAM), mucosal addressin (Mad CAM), and carcinoembryonic antigen (CEA). These molecules are capable of both homotypic and heterotypic (between different types of molecules, e.g., VCAM-1 to α4β1 integrin) interactions. IgCAMs play multiple roles in several biological processes, including the development of the nervous system and immune response (Makrilia et al. 2009).

734

P. E. Rayappan George Edwin and S. Bajpai

Cell-To-Extracellular Matrix (ECM) Adhesion Cell adhesion to the extracellular matrix (ECM) is mediated by integrins (Frantz et al. 2010). Integrin along with other accessory proteins form focal adhesions which are primary transducers in cell-ECM interactions (Lu et al. 2011). Focal adhesions are large, elongated structures, approximately 2 μm wide and 3–10 μm long, in which clustered integrins bind extracellular matrix (ECM) fibrils on the outside of the cell and associates with contractile actomyosin stress fiber bundles intracellularly (Schwartz 2010). Therefore, proper assembly of focal adhesion complex is crucial to ensure stable adhesion to matrix and generation of contractile force which is essential for many of the cellular processes (Rape et al. 2011; Burridge and Guilluy 2016). The abovementioned interactions are indispensable for the establishment of normal tissue structure and function and motivate a variety of physiological processes such as morphogenesis, embryogenesis, organogenesis, immunological function, wound healing, and inflammation. A growing body of evidence suggests that alteration in the adhesive properties of cells results in deregulated cellular processes leading to pathological diseases like cancer (Makrilia et al. 2009).

Role of Cell Adhesion in Cancer Progression See Fig. 3.

Epithelial-Mesenchymal Transition (EMT) Epithelial-mesenchymal transition (EMT) is a complex, multistep process that converts epithelial cells into mesenchymal cells. During epithelial-mesenchymal transition (EMT), stable cell-cell junctions are disassembled, the apicobasal polarity is lost, and migratory capabilities are enhanced (Theveneau and Mayor 2013) (Fig. 4).

Tumor Invasion Once theepithelial-mesenchymal transition (EMT) process is complete, cells start to invade the basal lamina progressing into the surrounding stroma. Integrins are the primary receptors that coordinate this invasive behavior. This cell adhesiondependent aspect of integrin function plays a critical role in determining a cell’s ability to break through a defined tumor margin in order to locally invade and ultimately metastasize (Seguin et al. 2016). Increased expression of integrins within the primary tumor is associated with poor prognosis and enhanced metastasis in a variety of cancers (Desgrosellier and Cheresh 2010).

31

Single Cell Adhesion in Cancer Progression

735

Fig. 3 Schematic diagram showing the different stages whereby cancer cells spread from a primary tumor site to a distant site in the body. (Adapted from Robbins pathologic basis of disease, 7th edition, Cotran et al. 1994)

Angiogenesis Angiogenesis is the key step toward tumor metastasis. The formation of functional blood vessels involves a series of coordinated biological processes such as cell proliferation, guided migration, differentiation, and cell-cell communication (Adams and Alitalo 2007). Members of the integrin, cadherin, selectin, and immunoglobulin families contribute toward every step of tumor vascularization, by mediating the cell-cell and cell-matrix interactions and by participating in the signaling events that regulate the extension and the maturation of neoforming vessels (Strömblad and Cheresh 1996; Ramjaun and Hodivala-Dilke 2009).

Intravasation Intravasation is the process by which tumor cells enter the bloodstream and is essential for the formation of distant metastases. Tumor cells interact with the endothelial cells and develop a strong adhesion with them. VE-cadherins are abundant in the adherens junction between the endothelial cells. Tumor cells induce phosphorylation of these VE-cadherins, resulting in the breakdown of the endothelial cell-to-cell junction, thereby gaining access to the bloodstream (Aragon-Sanabria

736

P. E. Rayappan George Edwin and S. Bajpai

Fig. 4 Schematic depicting the epithelial-mesenchymal transition (EMT) process. Shown here is a cell losing its adhesive contacts and moving into the matrix. (Adapted from Theveneau and Mayor 2013)

et al. 2017). There is also evidence of the upregulation of integrins and other adhesion molecules so as to facilitate attachment of tumor cells to endothelial cells (Dua et al. 2018).

Extravasation For extravasation of tumor cell to occur at distant sites, they must first be arrested at and adhere to the capillary beds. Generally, tumor cells might contain adhesion molecules, the ligands of which are expressed preferentially on the endothelial cells of the distant target organs (Ruoslahti 2002). Integrins are essential for attachment of the tumor cell to the new homing site, which then develops into the secondary tumor (Ganguly et al. 2013).

Techniques to Study Cell Adhesion and Migration In this section, we discuss some of the key techniques available to quantitatively evaluate the adhesion forces exhibited by the cells.

31

Single Cell Adhesion in Cancer Progression

737

Bulk Adhesion Measurements Initially, investigators relied on certain elementary techniques to estimate cell-matrix adhesion strength for large cell population. These techniques offer some advantages, as they provide high throughput, do not impose stringent measurement conditions, and do not require sophisticated equipment. Nevertheless, they are less sensitive and subject the cell to severe damage during measurements (Park et al. 2016).

Centrifugation Assays In this method, a substrate containing adherent cells is spun at a specific speed in order to apply a controlled detachment force perpendicular to the cell adhesive area, and the number of cells remaining attached to the substrate before and after spinning is quantified (Fig. 5a) (Keselowsky et al. 2003; Reyes and García 2003). The load applied in the centrifuge assay is the body force exerted on each cell, which is given as (Huang and Ingber 1999) F ¼ ðρcell  ρmedium Þ:V Cell :RCF

(1)

where ρcell is the density of the cell, ρmedium is the density of the medium, Vcell is the cell volume, and RCF is the relative centrifugal force in units of g. Based on the centrifugal assay, Reyes reported the variation in cell adhesion strength as a function of ligand density for a fixed centrifugation speed (Fig. 5b). A nonlinear profile with adherent cell fraction plotted against ligand density can be generated, and the ligand density for 50% adhesion strength can be a sensitive indicator of the overall adhesion strength (Reyes and García 2003). This strategy can be particularly useful for experiments involving multiple conditions, as it offers a significantly high throughput. Nonetheless, centrifugation applies relatively low detachment forces (40 independent parameters, and the information content increases exponentially with the number of parameters acquired, allowing for deep immunophenotypic analysis of the immune response. CyTOF utilizes antibodies labeled with heavy metal isotopic tags or fluorophores to label cells. The labeled cells are then sprayed into a plasma torch to ionize the cell contents to destroy the chemical bonds. SIgN CyTOF can distinguish atomic mass differences at a single atomic mass unit level. Therefore, the amount of information obtained from a single cell using CyTOF and >40 cell parameters is enormous. For example, each cell can be classified as positive or negative for each parameter, and a 40-parameter CyTOF analysis can produce 240 (>1 trillion) possible combinations of conditions. The throughput of millions of cells can be measured from experimental samples and has the opportunity to sort out rare cell populations. The Gonzalez team at Stanford University used MassCytometry (Cytometry by Time-Of-Flight, CyTOF) to identify new cell subtypes in ovarian cancer. CyTOF allows cells to be classified into different classes based on the expression of more than 40 proteins in one experiment. Single-cell CyTOF was used to identify a subset of cells commonly found in high-grade serous ovarian tumors. After analyzing more than 800,000 ovarian cancer cells at the single-cell level, 17 patients were found to have different cancer cell subsets than the other patients. By analyzing intact single cells, data reveals rare cells carrying specific subtypes that are covered in a large amount of homogenate. The researchers isolated 56 cell subtypes from tumor cells

32

Single-Cell Technologies for Cancer Therapy

809

and found a rare cell subtype that is strongly associated with early recurrence. They also found a hybrid cell type with dual characteristics of intratumoral and metastatic features (Gonzalez et al. 2018). In 2015, Xie and Zhang’s group published about an important mitochondrial functional protein IscAI, also known as MagR/MAR, belonging to the iron–sulfur cluster protein (Isc) family, which was found to have magnetoreceptor capabilities. In addition, they proposed a biomagnetic signaling-responsive mechanism. IscA can be a magnetic sensor as a magnetosensor (MagS) complex formed by photoreceptor cryptochromes (Kelly et al. 2015). This means that this new technology has the potential to capture a single cell or sort single cells by magnetic force. For example, a special biomarker is linked to a magnetoreceptor or magnetosensor related structure (Qin et al. 2016). TEM/Cryo-EM/SEM/Confocal 3D Images The use of luminescent or fluorescent markers in the single cell technologies has become a common tool in molecular biology experiments, and it can provide convincing evidence by using real-time fluorescent images. Researchers can use photosensitized transformation to detect changes in metabolites with luminescence, or to design real-time fluorescent probes that measure the concentration of target particles in cells. Scientists have been able to see the immediate action of the DNA-binding protein to turn the gene on and off, and with the electron micrograph, the chromatin chain interactions in the 5–24 nm range can be seen (Ou et al. 2017). In addition, known as the Cryo-electron micrograph, the protein or molecular structure is frozen by cryogenic freezing, and the image is reconstructed into 3D, allowing scientists to capture the first 3-D image of the brain receptor. This technique has successfully obtained the conformational change of the AMPA receptor in the glutamate receptor synaptic complex and stargazin (STZ) when the channel is open, and found that the junction is accompanied by activation and desensitization. In tens of thousands of cells, researchers want to find the cells that are labeled, making it easier to work through optics or imaging. However, the interpretation of clinical images still requires experienced medical staffs and long-term observation. Even so, misjudgment can still occur. Currently in the AI and AR technology trend, AI can be trained to learn from general camera photos, TEM, Cryo-EM, SEM, confocal images, CT, PET, NMR, X-ray film, etc., to find the abnormal part, which is very important in the diagnosis of cancer patients. Recently, Google has introduced AR and AI into cancer testing, which can be used to identify cancer cells. Through the image capture and AI analysis function, the computer can be trained through the neural network, the AI algorithm is executed at a near-instant speed, and the cancer cell range is marked in the microscope image by AR.

Global Projects, Database, and Bioinformatics Comprehensive cancer knowledge and treatment strategies require extensive integration from systems biology, bioinformatics, pharmacology, clinical data, etc., all

810

G.-M. Hu et al.

through various databases. From the human genome project (HGP) to the longevity plan and the integrated high-tech omics-based patient evaluation (HOPE) project, as time progresses, more and more new technologies are involved (Fig. 16). The acquired data is assembled into a large-scale database to form a more complete information network, which allows systems biology, systematic biological information, and clinical data to be integrated and matched, and then promotes drug development and disease treatment progress.

Database and Bioinformatics from Single Cells or a Single Cell Systems biology of diseases or cancer requires effective analysis of the database through bioinformatics and understanding of the interrelationships and interactions between different parts of biological systems. Further, bioinformatics is a crossdisciplinary science that combines biology, computer science, applied mathematics, statistics, etc., and is it increasingly valued for the completion of the Human Genome Project (HGP). With the advancement of biomedical technology, how to analyze and organize following large amounts of data and information becomes an important issue. For example, research on microbial gene function, physiology, metabolism, environment interaction, and host-microbial community is an opportunity to understand how related microorganisms affect our health and disease. By decoding the genome, thousands of bacteria, fungi, and viral genomes have been sequenced to complete the establishment of organisms (human, animals, and plants), virus, and microorganisms fine-maps. Ten thousand microbial genome projects (10 K/thousand microbial genome maps) are designed to sequence fingi, viruses, archaea, bacteria, etc. In addition, the Human Intestinal (MetaHIT) project is looking for potential mechanisms for neurogenic diseases, such as Alzhemase’s disease, dementia, and the cause of cancer. Big Data and Data Science have become well-known vocabularies. The industry is actively using and developing the value of big data. These huge amounts of data also bring huge business opportunities. However, the premise of doing real “big data” is that you have collected enough information to make models more and more accurate. For the analysis of the system biological large-scale database, deep computationalomics is an important milestone in systems biology. Today, deep learning has begun to shine in various sub-domains of systems biology. Scientists use the deep computational -omics of expertise in various fields to adjust the model and observe the characteristics of the model to assess the correctness of the direction and generate new hypotheses. The goal of systems biology is to explore the interaction patterns between various molecular systems in biological processes from a broader perspective. In order to understand this complex interaction, mathematics is ideal. The interaction between genome, RNome-wide, and protein is the basis for clarifying regulatory systems and pathogenic patterns. The training model data for deep computational -omics comes from high-throughput experiments, such as base pair sequencing data, microarray, RNAcompete, chromosome immunoprecipitation sequencing (ChIP/CLIP-seq), and systematic evolution of ligands by exponential enrichment (SELEX).

32

Single-Cell Technologies for Cancer Therapy

811

Fig. 16 A review of transcriptomics and important technologies accompanying global projects from the past. (Reprinted with permission from Macmillan Publishers Ltd.: Cieslik and Chinnaiyan 2018 copyright 2018)

812

G.-M. Hu et al.

SELEX is also referred to as SAAB (selected and amplified binding site), CASTing (cyclic amplification and selection of targets), in vitro selection, or in vitro evolution. It utilizes a very large number of single-stranded nucleotide libraries, such as DNA, RNA, and PNA, which are composed of randomly generated sequences to specifically bind to ligands (as aptamers). Once the synthesized nucleotide or peptide nucleic acid (PNA) oligomer is bound to the ligand, the composition that does not bind to the target ligand can be removed by affinity chromatography or captured by ligand–magnetic beads. Thereafter, the desorption conditions can be altered to find a tight binding sequence that leaves a specific target ligand, and the bound sequence is amplified by PCR to prepare for subsequent rounds of selection. Currently, in the biomedical market, SELEX has been used to develop many aptamers, and many nucleotides with chemically modified sugars and bases have been used. These modified nucleotides allow the selection of aptamers with novel binding properties and increased stability. In addition, since PNA is a synthetic polymer similar to DNA or RNA, it has potential as a drug. In recent years, synthetic peptides or nucleic acid oligomers have been used in biological analysis, diagnosis, and treatment. In deep learning, for example, DeepBind uses the Convolutional Neural Network (CNN), and it uses one-dimensional convolution as a feature of the segment detector to detect binding sites, pools the variety of lengths and sequences (max/average pooling), and then assesses the hub in network. In addition to investigating the network relationships between protein regulation, genome, and RNome, weighting matrices are also used to visualize the influence of each nucleotide site, so that the effect of point mutations can be quantified for exploration of the disease mechanism. In addition, metagenomics, also known as environmental genomics, ecogenomics, or community genomics, is a direct method for obtaining all genetic material in the environment. It can be used to identify and study microorganisms in a given environment. In cancer research, metagenomics reveals the relationship between microbial diversity and disease in humans by analyzing tissue samples and microenvironments of cancer patients. In cancer imaging, the application of deep learning can be extended from metagenomics to phenotypes, with neighborhood effects between points, and even convolution and long short-term memory network (LSTM) can be used to analyze contextual associations. However, non-sequence omics data cannot be used, because macrogenomics, metabolomics, and epigenetics are all quantitative. Fortunately, the data can be mapped into a neighborhood by calculating the distance. For example, operational taxonomic units (OTUs) data is embedded in a phylogenetic tree to generate a matrix. OTUs are clusters with multiple similarities of bacterial multiple genes. Paired OTUs derive paternal relationships to generate phylogenetic trees. If the distance between the levels is consistent, the factors of each OTU can be embedded in the matrix according to the order of the trees to train convolutional neural networks (CNNs). A phylogenetic tree embedded architecture for convolution neural networks (PopPhyCNNs) (Fig. 17), which drives CNNs to find the spatial relationship of the taxonomic annotations for matrix training, performs best in CNN methods. After training, the

32

Single-Cell Technologies for Cancer Therapy

813

Fig. 17 Flowchart of PopPhy-CNN. The taxa annotations and count table are used to create and populate a phylogenetic tree, which is embedded into a matrix format and used to train a CNN model. Features are extracted from the trained model. (Reprinted with permission from Macmillan Publishers Ltd.: Reiman et al. 2018 copyright 2018)

feature map is generated by the first screening to observe the model to determine which OTU is an important biomarker (Reiman et al. 2018).

Global Projects at Single-Cell Resolution Since the launch of the Human Genome Project (HGP) in the 1990s, many large databases, ranging from DNA sequencing to single whole-cell diversity, have been completed or are underway. These vast amounts of data will be available for integration and for research or personalized medicine. Scientists use database analysis to get a lot of valuable information, for example, analysis shows that 98% of the human genome is non-coding, and many non-coding regions of regulatory elements are involved in the regulation of gene expression. A number of non-coding regulatory regions of the human genome have been found to be related to DNase I hypersensitive sites (DHS), histone modification regions, DNA methylation regions, and transcription factor binding sites. In addition, mutations in the coding region can result in mutations in the non-coding region, and most of the somatic mutations in the cancer genomes occur in non-coding regions, so these non-coding regions can be speculated to affect

814

G.-M. Hu et al.

the interior of the cell. The degree of gene expression is related to further canceration of the cells (Khurana et al. 2016). Human Genome Project (HGP) and Human Genome Diversity Project (HGDP) Since the human genome plan began in 1990, the human genome project (HGP) achieved rapid progress in 2003, and has been followed by projects such as the announcement of the encyclopedia of DNA elements (ENCODE) in 2008, the completion of the International Thousand Genome Project (1K and 2K Genomes Project in 2012 and 2016), the release of the Million Genomes Project in 2015, and the completion of the 10 Million Genome Project in 2017. As in the HGP, the HapMap project progresses and with the rapid development of high-throughput biochip technology cells will be more purified and used for genome-wide sequencing by allowing hundreds or thousands of single cells to be tracked. For the genome, it is important to establish a global pattern of changes in single cells. At present, the single-cell sequencing (SCS) technology of malignant tumors has been widely used in the pathogenesis of malignant tumors to explore and analyze the genetic information of individual tumor cells. HGP uses DNA sequencing technology to obtain a full-length human cDNA library for the gene mapping and physics map, with genetic tags to determine the arrangement of genes on the chromosome. Further, the analysis of single nucleotide polymorphism (SNP) is used to identify individual differences and perform functional analysis of genes. In September 1994, the mapping of 3000 genetic maps with a resolution of 1-cM (e.g., 1% recombination rate) was completed. A physical map is generated and determined by a sequence-tagged site; thereby, a relative arrangement of genetic information of certain genes and their relative position on the chromosome forms a linear arrangement. The 24 human chromosomes were completely sequenced in 2006. The subsequent sequencing of messenger RNA (mRNA) in a single cell and single cells by the application of new RNA-seq technologies is accelerated by new sequencing methods, global collaboration and integration of computational techniques, development of related physical technologies, and improvements in bioinformatics methods. Advances in these technologies have expanded to include the model organism’s genome project, HapMap and subsequent 1000 genomes (1KG) project, metagenomic project, Human Genome Diversity Project (HGDP), and other large-scale database construction plans to help clarify the relationship between genes, evolution, and disease. The Haplotype Map (HapMap) The haplotype map (HapMap) project aims to establish an SNP database to identify key information about the genes involved in human health and disease, as well as drug and environmental responses. A haplotype or haploid genotype is a set of DNA variations (or polymorphisms) that can be referred to as a combination of alleles or SNPs found on the same chromosome. At present, millions of SNPs have been

32

Single-Cell Technologies for Cancer Therapy

815

discovered and many genome-wide association studies (GWAS) use this data set for disease association studies. However, in June 2016, The National Center for Biotechnology Information (NCBI) began to withdraw the use of the HapMap database, as the use of the HapMap database has been decreasing and the number of novel variants is constantly increasing. This makes the 1K genome’s data relatively more meaningful. It means that the aims of the HapMap project for the SNP database lose the original effect. Human Genome Diversity Project (HGDP) The Human Genome Diversity Project (HGDP) was promoted by the International Human Genome Organization (HUGO) in 1991. The Illumina Bead Station was used to analyze the genomic DNA of 1043 individuals from around the world, and genotypes in more than 650,000 SNP locus have been identified. The Human Genome Project aims to sequence the “human” genome because all human genomes are slightly different. Therefore, the HGDP hopes to explore a full range of genomic diversity. Today, the genome projects of model organisms, including mice, fruit flies, nematodes, zebrafish, and yeast, as well as metagenomics projects to sequence the genomes of commensal flora in humans, are used for cross-species comparisons and the function research of genes involved in human development and health. The Human Proteome Project and Comparative Serum Proteomics Project Proteomics is a study that focuses on the structure and function of proteins. In particular, the effects of protein modification on function involve the search for biomarkers of specific cancer subtypes. Because an organism produces the proteome, the expression of the protein will change in a cell or an organism with different needs or pressures over time. Many studies have shown that the total amount of RNA translated into a protein depends on the physiological state of the cell, and the proteomics research method can directly measure the type and quantity of the protein. Technically, protein purification and mass spectrometry are often used for proteome analysis. In particular, proteins in eukaryotes generally have post-translational modifications, including phosphorylation, ubiquitination, methylation, acetylation, monosaccharide, nitration, and oxidation. Among these modifications, phosphoproteomics and glycoproteomics research methods play a key role in activation of signal transduction and intercellular immune response in single cells. In addition, protein–protein, protein–DNA or protein–RNA molecules form a complex interaction in the cell. These different object combinations involve the alignment and structural stereolocation of the protein amino acid, in which case the purified protein crystals are X-ray diffracted or nuclear magnetic resonance is used to determine their structure. HGP uses human chromosomes as information, and it has found 20,300 proteins that may be produced. However, in the prior art, there are still about 18% of proteins (missing proteins) in the human proteome and there is no way to confirm their existence. To solve this problem, the Human Proteome Project (HPP) was initiated by the Human Proteome Organization (HUPO) to enhance understanding of human biology at the cellular level through high-stringency data for all proteins encoded

816

G.-M. Hu et al.

from the human genome and to serve as the basis for the development of diagnostic, prognostic, therapeutic, and preventive medical applications. Currently, the PRoteomics IDEntifications database (PRIDE) based on mass spectrometry (MS) proteomics data is provided by the European Bioinformatics Institute (Baker et al. 2017; De Fauw et al. 2018). The goal of the HPP is to use proteomics to make reference data for comprehensive omics and genomics more complete, such as establishing each protein-coding gene, as well as sequence variants, post-translational modifications, and splicing isoforms. In the alignment analysis of the human protein database neXtProt, it is believed that there are 20,399 proteins that can be translated by the human genome. Among these proteins, 17,694 have been confirmed to exist by mass spectrometers or antibodies (https://www.nextprot.org/about/protein-existence). The HPP database currently contains two sources, the Chromosome-based Human Proteome Project (C-HPP) and the Biology/Disease Human Proteome Project (B / D-HPP), to construct protein features and isoforms, and to look for “missing proteins.” These unproven proteins (“missing proteins”) include proteins that cannot be treated with trypsin, highly hydrophobic membrane proteins that are difficult to dissolve, proteins that are not transcribed, or whose expression levels are much lower than the sensitivity of the instrument. It may even be manifested only in early development or under the pressure of infection and inflammation. In addition to HPP, www.missingproteins.org also provides relevant resources. Evidence for the presence of HPP constructing proteins from five groups includes PE1 to PE5: PE1: Evidence at protein level PE2: Evidence at transcript level PE3: Inferred from homology PE4: Predicted PE5: Uncertain Therefore, the so-called “missing protein” is the sum of the number of PE2, PE3, and PE4. At present, there are 1939 proteins in PE2, only the complementary DNA (cDNA), which is reverse transcription by messenger RNA (transcription), but no evidence of protein existence. In PE3, there are 563 proteins defined as homologous or non-homologous proteins, that is, these protein sequences are too close, so that they may be misidentified as the same protein, which may cause actual and predicted quantitative estimates. As for the 77 proteins in PE4, the bioinformatics software predicts that they should be protein, but there is no evidence of gene transcription and protein translation. Overall, a total of 2579 still cannot be proved by mass spectrometers or antibodies. Comparative Serum Proteomics Project

In addition, because of the decline in sequencing costs, there are more than 9000 sequencing projects in progress (G10K and Earth Biogen Genome Project). Many

32

Single-Cell Technologies for Cancer Therapy

817

systems biology-related databases have been established through high-quality genome assembly and gene annotation. Comparative proteomics are used to evaluate different non-model organisms in protein research. Proteomics data requires high-quality protein samples and genomic databases to analyze data from mass spectrometers. Currently, the serum proteomic data is derived from 25 to 50 different mammalian species by genome annotation. Although blood protein components cannot be predicted by mRNA transcript abundance, they can be identified by instrumentation and provide relative quantitative information on 100–500 proteins. These comparative proteomics data are currently available on ProteomeXchange (http://www.proteomexchange.org) and Massive and are used to help search for protein comparisons with different species. A comparative proteomic model can show whether a particular gene is expressed or exists. In particular, it can be applied to search for unknown cancer cell markers (Fig. 22) and reveals how microbes affect cells’ canceration (Baker et al. 2017). Human Longevity Project In 2014, the Human Longevity project was deployed by Craig Venter, Chief Executive Office of Human Longevity, Inc. (HLI), and Peter Diamandis, chairman of the X Prize Foundation, with the goal of building the database on human genotypes and phenotypes, to surmount aging related diseases. Since then HLI has also collaborated with drug companies, Celgene and AstraZeneca, and served customers with full genome sequencing to add analysis of risk factors for diseases, including cancer, diabetes, heart and liver diseases, Alzheimer’s disease, and dementia. Noticeably, the genomic risk indication of cancer and Alzheimer’s disease refers to the source obtained from over 12,000 twins and from Alzheimer’s patients. The phenotype is physical features that may reflect downstream of protein interaction from genomic regulation; therefore, combining both the phenotype and genomic data of not only coding regions but also non-coding regions enables one to mine the variation of interaction effects and find new frontiers. The Cancer Genome Atlas (TCGA) The Cancer Genome Atlas (TCGA) is a large research project that began in 2005. The program collects clinical records, tumor tissues, and corresponding normal tissues of a specific cancer patient on a large scale, and performs bioinformatics analysis on the sequence, then integrates the data and discloses the sequence data and analysis results. TCGA has so far collected more than 11,000 cancer patients to study 33 different tumor types, including ten rare cancers. These cancer data are recorded in seven different data types. The amount of data is more than 2.5 petabytes (Hoadley et al. 2018). Human Cell Atlas (HCA) At present, the Human Developmental Cell Atlas (HDCA) is an important part of the Human Cell Atlas (HCA), in which the sequencing of 250,000 developmental cells began in 2018, and will be used to understand the developmental changes of cells. Cells from various tissues were analyzed by scRNA-seq and transcriptome data was

818

G.-M. Hu et al.

established in conjunction with in situ imaging. HDCA uses powerful single-cell genome analysis tools to create genomic reference maps of all cells important to human development to understand early human development and how it affects health or causes disease, such as the relationship between cancers’ root in development and the developmental pathways. Researchers at the Wellcome Sanger Institute and Newcastle University collected genomic data from more than 250,000 cells from donated developmental human tissues, including liver, skin, kidney, and placenta. These genomic data are expected to show which genes are turned on in each cell. At the same time, other similar projects, such as atlas from tumor, lung, intestine, kidney, and immune cells, are also being promoted. The HCA aims to create an open and accessible human cell type reference map (Behjati et al. 2018). It is widely useful to regenerative biology, disease mechanisms, drug discovery, efficiency, resistance and toxicities, and diagnostics. The Human Cell Atlas (HCA) will have a profound impact on biology and medicine, and it will answer basic questions about biology and help to discover the secrets of human disease. Recent studies have shown that the cell cycle can track the response of immune cells to pathogen components. In the cycling process and transient reaction of single cell, the changes of genes expressed in different stages of time can be linked in time and space, and correlation can be established (Fig. 18b) (Regev et al. 2017). The Genotype-Tissue Expression (GTEx) series meets all of the above goals and objectives. GTEx is the US NIH project launched in 2010 to study how genetic variation affects gene expression in 44 normal human tissues (Aguet and Ardlie 2016). Frozen samples collected from GTEx can be used for snRNA-Seq, while a large number of detailed histological images are available for comprehensive analysis of molecular and spatial characteristics (Fig. 18a). TCR Diversity Database-ImmunoMap and Human Microbiome Project (HMP) In addition, in order to understand the properties of antigens and the T cells that recognize them, a research team at Johns Hopkins University Medical School created the ImmunoMap database for tumor antigen-specific TCR diversity in tumor-bearing and tumor-free mice. ImmunoMap can help develop immunotherapy for cancer treatment and R&D of TCR engineering (Sidhom et al. 2018). Many studies have also shown that the interaction between the original living organisms and the human immune system is also one of the causes of major diseases. The results also show that the patient’s intestinal flora is related to the success of immunotherapy. Intestinal bacteria are the key to determining the cure rate of cancer. It may not be possible for cancer patients who lack intestinal probiotics to cause any effect (Gopalakrishnan et al. 2018). In mouse experiments, A. mimipiphila stimulates immune cells to release a chemical signal called IL-2, which is known to regulate T cells and cause T cells to attack. The results of the experiment showed that mice treated with A. muciniphila could convert unresponsive intestinal bacteria into a reaction (Routy et al. 2018). Currently, to study how the microbiome’s strains, functions, and kinetics affect related diseases, the NIH Human Microbiome Project (HMP) offers over 2355 metagenomics datasets of samples from more than 250 people, with multiple body parts at multiple time points. The HMP database will help to understand microbes

32

Single-Cell Technologies for Cancer Therapy

819

Fig. 18 The analysis of GTEx data in the HCA project (a) Analysis of samples from GTEx by RNA-Seq establishes a specific map of gene expression across a large number of human tissues. (b) Gene signatures (cell cycle in Th cells) or PCA assay (immune response, LPS, pathogenicity) in GTEx data. (a) T-distributed stochastic neighbor embedding (t-SNE) is an unsupervised learning type of algorithm that quickly observes the clustering and classification of data in a less precise but visual way and presents high dimensional numerical data under small sampling. t-SNE belongs to the dimensionality reduction algorithm, which can reduce the data dimension. For example, principal component analysis (PCA) is the easiest to understand intuitively among such algorithms. (Reprinted with permission from Springer Nature Ltd.: (Aguet and Ardlie 2016) copyright 2016). (b) Genetic features in individual mouse cells in a low-dimensional space are shown based on RNA profiles, the cell cycle of mouse hematopoietic stem and progenitor cells, the relationship between lipopolysaccharide (LPS) in mouse immune dendritic cells, and the degree of pathogenicity in mice. (Copyright under terms of CC BY 4.0 from (Regev et al. 2017) and refers to the further copyright of Fig. 4 A, B, and C in “The Human Cell Atlas” Published online 2017 Dec 5. https://doi.org/ 10.7554/eLife.27041 with permissions from Scialdone et al. 2015, Shalek et al. 2014, Gaublomme et al. 2015)

820

G.-M. Hu et al.

inside and outside the body in the clinical application of cell canceration and cancer immunotherapy. High-Tech Omics-Based Patient Evaluation (HOPE) Project Research has provided extensive evidence that oncogenic genes drive cancer cells growth and metastases. Hence, genetic sequencing is a valuable tool for identifying molecular targets and finding new drugs for cancer treatment. Next-generation highthroughput sequencing is allowing clinicians access to a rich amount of information on their patient’s cancer. To effectively employ this great amount of available information, the Shizuoka Cancer Center organized the High-tech Omics-based Patient Evaluation (HOPE) project (Yamaguchi et al. 2014; Routy et al. 2018). The HOPE project consists of the multiomics evaluation of the genomics, transcriptomics, proteomics, and metabolomics of a patient’s cancer. The HOPE experimental design is presented in Fig. 4. Blood cells, tumor tissue samples, and normal tissue are collected from the patient. Blood cells and the tumor tissue whole exome are sequenced with next-generation DNA sequencers. Blood cells sequencing provides information on the inherited risk of cancer and the tumor sequencing indicates the cancer-specific genetic changes. The blood cells genetic sequencing for hereditary cancer risk alerts family members to the possibility of carrying the same genetic mutations. This can lead to preventative medicine and an increase in regular check-ups for early detection and treatment of cancer. Tumor tissue also undergoes a microarray gene expression profiling; this and the wholeexome sequencing can give a prognosis, if there is resistance to any drug, and provide molecular targets for targeted therapies. Normal tissue also undergoes a gene expression profiling for comparative analysis with the tumor expression. When required, tumor proteomics and normal tissue proteomics were analyzed. The employment of all these methods can be useful for a more individualized medicine since it provides rich information on the cancer-specific alterations of the patient. Project HOPE is a close collaboration between clinicians and researchers that encourages individualized medicine. Moreover, clinicians have available knowledge of the evolution of cancer, which in the case of recurrence can be useful for the appropriate targeted therapy. Multiomics is still at a very early stage; however, projects like this one will indicate to researchers and clinicians the proper protocol modifications for a more effective and individualized cancer treatment. Follow-up studies taking into consideration the multiomics of CTCs and the heterogeneity of cancer will certainly provide a more comprehensive understanding of future personalized medicine.

Personal Medicine: Gene Sequencing, Gene Drugs, and Precision Therapy at Single-Cell Resolution Precision therapy is part of precision medicine (PM). Precision medicine is a tailored (personalized) medical decision and treatment for individual patients. Diagnostic tests are often used to select the appropriate and optimal therapy based on the

32

Single-Cell Technologies for Cancer Therapy

821

patient’s genetic predisposition, molecular or cellular analysis background. These will involve the correct matching of molecular targeted therapies to the unique genetic and molecular composition of each patient, such as targeted drugs and biomarkers that can be used as therapeutic decisions for patients. Other tools used in precision medicine can also include molecular diagnostics, imaging, and analysis. Cancer patients often experience severe side effects after chemotherapy, such as neutropenia. Therefore, alternative and systemic anticancer therapies, including chemotherapeutics and targeted therapies and immunotherapy, are needed. It is also clinically suitable for the treatment of advanced cancer. Systemic anticancer therapies usually contain a variety of anticancer drugs and supporting care drugs to reduce side effects and improve the effectiveness of multiple drugs, so there may be interactions between different drugs. Therefore, establishing the Drug–drug interaction database is to avoid systemic interaction of cancer treatment drugs. Single-cells culture or organoid (Czerniecki et al. 2018) for drug interaction testing can be used to test a large number of samples. With the help of single cell’s medical detection technology (Fig. 19) and integrated database, personalized cancer treatment goals can be achieved through gene sequencing, gene drugs, and precision therapy. In addition, biopharmaceutical is a new type of immunotherapy, also known as biological response modifier, which is currently used in cancer. The main purpose is to use biological technology to transform the immune cells in the human body in vitro, and then return them to the patient after amplification, which enhances the immunity to fight disease. It is also a trend of personalized medicine. Cancer cells have a variety of mechanisms to escape the identification and killing of immune cells, which is a modern medical problem. The goal of cancer immunotherapy is to improve the immunogenicity of cancer by molecular biological techniques and cell engineering techniques and to stimulate the body’s anti-tumor immune response. It inhibits the proliferation of cancer cells and cooperates with radiation or surgery to increase the success rate of cancer treatment. Cancer biotherapy has been regarded as the fourth treatment after surgery, radiotherapy, and chemotherapy. For some cancers, the efficiency is as high as 70%. At present, monoclonal antibodies, stem cells, cytokines, tumor vaccines, immune lymphocytes, gene therapy, and other biological treatment fields can become cancer treatment methods.

Anticancer Drug Development and Personal Medicine One of the visions of personalized medicine, to select drugs based on individual genes, has gradually been implemented, using pharmacogenomic testing to avoid dangerous drug side effects due to individual physical characteristics, but it is still difficult to apply this technology clinically. However, drug gene testing has many benefits. Drug-to-personal gene interactions, whether severe or mild, can be avoided by prescribing different doses of medications or alternative therapies. A study published in Nature in October 2015 pointed out that 80 drugs are affected by more than 20 genes (Zickenrott et al. 2016).

822

G.-M. Hu et al.

Fig. 19 Liquid biopsies as a method for clinical decision for cancer immunotherapy by investigation of the molecular heterogeneity and immunologic phenotype of tumors. (Reprinted with permission from Quandt et al. 2017 copyright 2017)

Clinically, traditional cancer treatment drugs can be divided into the following types of mechanisms: intrastrand cross-linking of DNA, blocking the synthesis of DNA or RNA, interference or inhibition of DNA/RNA synthesis, topoisomerase action, microtubule assembly, and spindle formation, inhibition of kinases, growth factor, receptor, and signal transduction. In addition, another direction of cancer drug development is the degradation of target proteins in cells, where drugs can act on the pathogenesis of protein degradation in cells. The development of cancer drugs is also an important direction toward the signal path of the target protein. Empirically, gene overexpression of tyrosine kinase is known to be associated with poor prognosis in cancer. Receptor tyrosine-protein kinase erbB-2, also known as human epidermal growth factor receptor 2 (HER2), Neu, CD340, p185, etc., is related to ERBB2 gene expression. The expression of C-erbB-2 is greater than normal and can be detected in ovarian cancer, colorectal cancer, bladder cancer, pancreatic cancer, ovarian cancer, gastric cancer, small cell lung cancer, prostate cancer, etc., which may lead to cancer cells growing faster and spreading to other parts of the body. Therefore, checking the amount of c-erbB-2 on cancer cells may help to plan treatment. Some examples of anticancer drugs based on the physiology of single cancer cells include inhibition of EGFR, HER, cErbB and other growth factors, inhibition of angiogenesis, inhibition of c-raf, inhibition of BCR-ABL phosphatase, inhibition of the proteasome in apoptotic pathway, and blocking of receptors and enzymes regarding gene expression in the cell nucleus (Harrison’s Principles of Internal Medicine, 18e, Chap. 85).

32

Single-Cell Technologies for Cancer Therapy

823

For example, chromatin regulation is the key to nuclear function and affects the process of transcription. For example, polycomb family proteins associated with histone modifications can remodel chromatin to make epigenetic silencing. Methodologically, the regulation of chromatin in single cancer cells can be studied through molecular tools, such as ubiquitin ligase, fluorescent ligands, inhibitors, and other drugs (Fig. 20) (Cermakova and Hodges 2018). RNAi drugs are also an important research and development direction in the treatment of diseases. Although there are many difficulties in the development of RNAi drugs, such as that unpredictable dangerous side effects occur when RNAi is not delivered to the correct position or it appears there are no side effects in animal experiments but some occurred in clinical trials. At present, a very promising RNAi drug is Alnylam Pharmaceuticals’ ONPATTRO™ (patisiran), which has completed the third phase of clinical trials. Onpattro is delivered to liver cells with a lipid nanoparticle (LNP) complex and reduces the production of TTR protein in the liver via the RNAi pathway to treat hereditary polyneuropathy. Notably, the joint research at Harvard Medical School and the Board Institute published in “Science” discovered that the waste ammonia produced by tumor metabolism will be reused by cancer cells and converted into glutamic acid. Non-essential amino acids such as aspartic acid play roles to help tumor growth and proliferation (Spinelli et al. 2017). In addition, the research results from Wang and her team unveiled the mechanism by which cancer cell exosomes regulate tumor-associated fibroblast metabolic pathways. The significance behind this is that the patient’s prognosis can be analyzed from the level of relevant miRNAs in the exosomes, as well as targeted development of new drugs targeting miRNAs (Yan et al. 2018).

Fig. 20 Possible blocking of chromatin in cancer cells by anticancer drugs. (Reprinted with permission from MDPI Publishers Ltd.: Cermakova and Hodges 2018 copyright 2018)

824

G.-M. Hu et al.

Cancer Vaccine There is increasing evidence that activation of oncogenic pathways in tumor cells may impair induction or local anti-tumor immune responses. Therefore, the goal of the next-generation antibody vaccines should focus on enhancing the interaction between dendritic cells (DCs) and B cells through antigen presentation, as the oncogenic pathway is involved in evading anti-tumor immune responses. Cancer vaccines are divided into two categories, one for cancer prevention, such as the FDA approved hepatitis B vaccine and HPV vaccine; the other is for the treatment of cancer, an emerging immunotherapy to stimulate the body’s immune system, kill cancer cells, and prevent cancer from spreading and recurring. Unlike the customized against targets, Ronald Levy’s team used a non-customized approach called in situ vaccination, combining two drugs to stimulate the immune system by drug injections of the malignant tumor in mice. At the first injection, 90 mice were vaccinated, 87 of which were completely cleared of cancer cells, and the success rate of cancer cell disappearance was 97%. After the second injection of the remaining three cancer-reversed mice, the cancer cell clearance effect was 100%, and the cancer vaccine is effective for breast cancer, colorectal cancer, and melanoma mice. The two components used by the Ronald Levy team for the cancer vaccine are CG-enriched oligodeoxynucleotide (CpG) and anti-OX40 antibody. CpG is a Toll-like receptor 9 ligand (TLR9 ligand) that is part of the innate immune system for recognition of the molecular patterns on pathogens and increase of OX40 expression on CD4+ T cells. The other component, the anti-OX40 antibody, activates T cells and causes T cells to attack cancer cells. Notably, this small amount of local vaccination is not only effective but also relatively inexpensive, and it is less likely to cause side effects compared with chemotherapy or radiotherapy. In the experiment, mice had lymphoma tumors in two locations of the body, and anti-cancer-specific T cells were activated by injecting a small amount of vaccine into the tumors at two different locations. Interestingly, tumors with different locations but with the same target protein were eliminated together. In other words, this cancer vaccine treatment avoids the cumbersome way of searching for the unique immune targets of each tumor, does not require extensive immune activation throughout the body, or separate treatment of each patient’s immune cells. This means that cancer cells that have spread to other parts of the body will be cleared with a single shot. A trial of 15 patients with early stage lymphoma is currently underway for clinical trials. Compared to the side effects of chemotherapy or radiotherapy or expensive immunotherapy, if this in-situ injection of anti-cancer vaccine can generally overcome more types of cancer, it will become a new choice for low-cost and rapid cancer treatment (Sagiv-Barfi et al. 2018). Drug Testing and Organoids from Single Stem Cells Many scientists have discovered that by placing stem cells from different tissue sources in 3D medium and carefully adjusting the culture condition, various human organs, such as eyes, intestines, liver, kidney, pancreas, prostate, lungs, and stomach,

32

Single-Cell Technologies for Cancer Therapy

825

can be manufactured. These miniature artificial organs from the 3D cell mass, called organoids, their structure and function are analogous to real organs. In 2014, a study led by the Sloan-Kettering Cancer Center in New York, USA, established prostate cancer organoids for the first time. Later, between 2014 and 2017, the Nicola Valeri team recruited 71 patients with metastatic colon and gastroesophageal cancer, extracted 110 fresh tumor tissues and cultured them into organoids, and conducted four prospective clinical trials. Fifty-five anticancer drugs that have entered clinical or ongoing clinical trials have been tested. The experiment found that these organoids, regardless of their tissue morphology or gene expression patterns, remained highly consistent with the tumor tissue from which they originated. The results show that if a drug acts on the patient’s organoids, then the drug has an 88% chance of functioning in the patient; conversely, if the drug is not effective, then the drug is 100% not valid for this patient. Currently, researchers at the University of Washington School of Medicine have further developed systems that automate the process of culturing stem cells into organoids. The cell type of the organoid produced by the automated machine has human kidney complexity confirmed by single-cell RNA-seq. This automated machine has significantly expanded the use of organoids in basic research and drug discovery, because the cancer organoids can accurately predict the efficacy of drugs as a new method for cancer drug testing and targeted therapy (Czerniecki et al. 2018). Old drugs can also be used for testing in single cells and organoids – aspirin cures cancer. In addition to soothing headaches and preventing heart attacks, aspirin seems to inhibit the spread of cancer cells. In 2000, scientists discovered that aspirin has a second major role in the body – promoting the production of resolving and helping to stop the inflammatory response. Researchers have also recently begun to clarify the third role of aspirin – interfering with cancer cell metastasis and preventing cancer cells from spreading throughout the body. Interestingly, the anti-inflammatory effect of aspirin does not seem to be the protagonist. Studies have found that platelets that enhance blood clotting and produce growth factors to promote angiogenesis and new blood vessels are an indispensable mechanism for feeding nutrients and oxygen to support new metastatic tumors (Sulciner et al. 2018). Another example of new uses for old drugs is Disulfiram (Antabuse), which was originally used to treat alcohol addicts. By interfering with the metabolism of alcohol, it causes the accumulation of acetaldehyde to be five to ten times higher than normal. Further, the alcohol addicts have a severe discomfort and reduce the desire for alcohol intake. Use of disulfiram to prolong the survival of patients with Non-Small Cell Lung Cancer (NSCLC) has been proven (Najlah et al. 2017). The use of single cells and organoids has the possibility to allow many of the old drugs or poisons to be seen with the potential for cancer treatment. For example, arsenic treatment can cure blood cancer. Anne Dejean, Hugues de Th, and Zhu Chen used the revolutionary treatment to give acute promyelocytic leukemia (Hollis et al. 2017) a chance to be cured and won the Sjoberg Prize in 2018.

826

G.-M. Hu et al.

Antibody Drugs and T Cells Immunotherapy The antigens can be delivered to the lymph nodes by diffusion of the naked antigen from the lymph or by delivery of antigen in the form of immune complexes (IC) by non-homologous B cells or local macrophages. IC activity is regulated by the Fab and Fc domains of the antibody (Ab). Fab drives antigen-specificity, while the Fc region determines which innate immune receptors can bind to ICs, including Fc receptors (FcR) and complement receptors (CR). According to the FcR and CR provided in the IC, the antigen can be rapidly destroyed or transported to the germinal center (Joshi et al.) for antigen presentation. Thus, modifications to the Fc domain, including changes in Ab isoforms/subclasses and glycosylation, have a potential impact on the antitumor activity of vaccine-induced immune responses. Fc glycosylation, a direct regulator of B cell immunity, allows next-generation vaccines to utilize both ends of the Ab to enhance B cell affinity maturation. Abnormal glycosylation on human proteins and cell surfaces has been shown to be closely related to human cancer. Nowadays, many tumor-associated carbohydrate antigens (TACA) have been identified. However, only a few specific glycosylated antigens are expressed on the surface of cancer cells and can be used as a target for cancer immuno-antibody drugs to assist in the development of antibodies. Because the monoclonal antibody (mAb) has superior specificity, it can produce monoclonal antibody drugs for cancer immunotherapy by transgenic mouse technology. In addition to directly attacking cancer cells (Table 2, Fig. 10, and subsections in “Cancer Vaccine,” “Cancer Cells and Immune Response,” nanomedicine and DNA nanorobots, and CAR-T therapy), the antibody can also carry anticancer drugs to reach tumors. According to the structure, antibody drugs can be classified into naked mAb, conjugated mAb, and bispecific mAb, and drugs are administered intravenously to patients. However, the antibodies themselves are proteins, so individual antibody drugs sometimes cause allergic reactions. Naked mAb is the most structurally simple antibody, with no additional drugs or radioactive substances, and it is the most common type of mAb used to treat cancer. In contrast, mAb that links with chemotherapeutic drugs, toxic molecules, cytokines, or radioactive particles is called conjugated mAb. The conjugated mAb has the function of navigation and carrying, which can bring the linked substances into cancer cells instead of normal cells to avoid damage to normal cells. Alternatively, it is designed such that two Fab regions of the antibody bind to a target antigen, and a conjugate or effector cell is called a bispecific mAb. For example, blinatumomab (Blincyto) is a bispecific T-cell engager (BiTE) antibody, a so-called CD19-directed CD3 T-cell engager antibody, with simultaneous recognition of CD19 on the surface of B cells and CD3 on the surface of T cells, which are used to treat certain types of acute lymphoblastic leukemia (ALL). Two different teams in the US and Germany have announced successful clinical trial results for two personalized vaccines for different tumor mutations. The clinical trial results of the US team showed that of the six melanoma patients vaccinated, four of the tumors completely disappeared, and there was no recurrence within 32 months. The other two tumors still existed, and the tumor disappeared completely after

32

Single-Cell Technologies for Cancer Therapy

827

receiving adjuvant therapy. The results of another team reported that of the 13 vaccinated patients, eight tumors completely disappeared and there was no recurrence within 23 months. The other five patients had the tumor spread when vaccinated, and two patients had tumor shrinkage, including one where the tumor completely subsided after receiving adjuvant therapy. Tumor metastasis was significantly reduced in patients after vaccination, and in the case of persistent or recurrent tumors, treatment with PD-1 inhibitors was effective. Hence, the combination of vaccines and PD-1 inhibitors has become a key consideration for researchers in subsequent larger clinical trials. CAR-T therapy (subsection “TCR and CAR-T at Single T Cells Resolution”) has been spotlighted in recent years, but bispecific monoclonal antibodies with the same therapeutic goals are also followed on the market. A bispecific monoclonal antibody is a protein-engineered technique that combines two antigen-binding fragments (Fabs) or single-chain variable fragments (scFv) against different antigens into one antibody. The novel antibody drug allows it to bind to two different antigenic epitopes at the same time with specificity. In addition, bispecific antibodies are easier to produce than CAR-T cells and have the advantage of having a wider range of treatment options. A targeted therapy drug is a drug that blocks the growth of cancer cells by a specific molecule that interferes with cancer or tumor proliferation. Prior to “Imatinib,” a target-drug launched in 2001, surgery, radiation therapy, and chemotherapy were common cancer treatments, all of which have relatively serious side effects. When Dr. Tony Hunter studied the protein kinase of the polymavirus that causes cancer, he found that carcinogenesis is related to the phosphorylation of tyrosine. Later, he further confirmed that the tumorigenic transforming caused by polyomavirus and Rous sarcoma virus is related to tyrosine kinase (Oya et al. 2017) and suggested that the mechanism of uncontrolled tyrosine phosphorylation is the key to initializing cancer. According to the active-site theory, tyrosine kinases are classified into receptor tyrosine kinase (receptor TK) on the cell membrane and tyrosine kinase in cytoplasm. When normal cells overexpress tyrosine kinase or lose function of tyrosine phosphatase (function is opposite to kinase), abnormal proliferation of cells and even tumors will be presented. Thus, finding ways to suppress abnormal tyrosine kinases may be able to control cancer. Imatinib is the first target-drug of tyrosine kinase inhibitor (TKI) for the treatment of chronic myelogenous leukemia (CML). The Abelson murine leukemia (ABL) protein is a tyrosine kinase, and the main cause of CML is the presence of chromosomal translocation in bone marrow hematopoietic cells to form ABL/BCR fusion protein from the chimeric gene. When the switch of tyrosine kinase activity is stuck and cannot be turned off, it will lead to unrestricted proliferation of white blood cells – leukemia. Therefore, it is possible to control leukemia by target-drugs to block the tyrosine kinase active site and to suppress excessive messages of the signal pathway in white blood cells. The molecular mechanism of imatinib is a small molecule to block the active site of cytoplasmic tyrosine kinases. From the same blocking concept, Cetuximab

828

G.-M. Hu et al.

(Erbitux®) is the antibody to block the receptor of epithelial growth factor (EGFR) and receptor tyrosine kinase on the cell membrane. Nowadays, Cetuximab is commonly used in targeted-therapy of colorectal cancer, head and neck cancers, and other cancers.

Molecular Synthesis and Bio-genetical Engineering Drugs to Kill DrugResistant In addition to the use of chemical synthesis as a drug, the use of organisms to produce molecular drugs is also an important way, especially bio-protein drugs, protein surface glycosyl groups or modification of bio-protein drugs is related to the effectiveness of the drug and side effects of immune response. For example, a team at the National Institute of Advanced Industrial Science and Technology (AIST) in Japan has successfully used genetic modification technology to create chickens capable of producing eggs containing the anticancer component beta interferon (IFN-β), an extremely valuable anti-cancer drug component. IBM Labs designed molecules to kill drug-resistant super bacteria. These superbugs are usually obtained from hospitals by patients and can cause septic shock and multiple organ failure due to systemic infections. Hedrick’s team conducted polymer testing on mice infected with difficult-to-treat multi-drug resistant bacteria. To systematically understand the response of bacteria to antibiotics or polymers, bacterial RNA-seq was performed on the 10,073 dish samples using the MDR system. These samples are A.baumannii that have been cultured for 30 generations. Experimentally, the use of imipenem, an antibiotic, inhibits bacterial cell wall synthesis. After systematic comparison of imipenem and pEt_20 response genes, 47 common genes were found. The data indicates that the imipenem and the polymer reactive genes have mutually exclusive functions, and the polymer does not have resistance. This means that polymer based antimicrobials may overcome drug resistance (Chin et al. 2018). Drug–Drug Interaction Database and Drug Release Clinically, systemic anticancer therapies include chemotherapeutic agents as well as targeted therapies and immunotherapies. In drug use, there are potential drug–drug interactions between other drugs and systemic anticancer therapy medicines. Therefore, there is a need to research and build a database of interactions between different drugs, for example, ONCHigh and DrugBank. ONCHigh provides recommendations for drug–drug interactions with highpriority severity, and DrugBank only provides drug–drug interactions without severity levels (Phansalkar et al. 2012). At present, self-constitutive hydrogels can be utilized for single cell technology or for stable release of cancer drug treatment. Joshi et al. published a technique for selfassembly of triglycerol monostearate (TG-18) to form an arthritis-flare reaction hydrogel because of the encapsulation stability and long-term hydrolysis of TG-18 hydrogel in PBS (Joshi et al. 2018). It may be used in conjunction with in situ intratumoral injection of cancer drug release applications.

32

Single-Cell Technologies for Cancer Therapy

829

Bioinfomatics and Analytical Framework for Single-Cell MetaAnalysis The application of bioinformatics and systems biology methods is a future trend in personalized medicine, especially the use of biomedical scientific findings in clinical and therapeutic outcomes. Because of the large amount of information and knowledge required for bioinformatics, the inclusion of data, knowledge, and discovery in cancer, etc., has become crucial (Fig. 21). Nowadays, new analytical techniques can understand the complex relationships between people and individuals through data collection and data mining, and identify important and critical factors to overcome cancer prevention and treatment. There is a variety of tools available to support omics research in systems biology that can influence cancer treatment and prognosis strategies. A complete omics provided by systematic medical analysis sources includes gene and gene expression, protein structure and dynamics, nutrition, microbiology, pharmacology, analytical and statistical mathematics, molecular technical data and diagnostic data, clinical medical imaging, past medical experience, psychology, etc. In particular, pharmacogenomics emphasizes differences in human genomics, pathophysiology, and therapeutic response, and these differences involve differences in individual patient conditions. Therefore, regardless of the general disease or cancer prevention, treatment, or prognosis, to achieve personal medicine, complete omics information is absolutely indispensable. Deep learning, artificial intelligence, and block chain are important tools to aid doctors’ judgment and the safety of medical information sharing. In addition, meta-analysis has become a popular method for summarizing a large number of clinical trials and addressing the differences caused by these trials. Metaanalysis is an important integrated research tool, especially considering the integration of different fields and data, and is applied to machine learning or visual analysis. The purpose of the meta-analysis is to systematically evaluate and improve the estimates of the magnitude of the effects in the statistics and to identify old or new targets that are of interest in different studies. The general procedures for this analysis include formulating questions, determining eligibility, determining research, abstracting data, statistical analysis, and reporting results. Thus, metaanalysis techniques collect quite a few analysis results from individual studies, and then perform statistical analysis for a single report to create a more accurate estimate of the effect. Based on the use of the same statistical methods, meta-analysis is also defined as “the analysis of analyses,” which means that many research results are combined into a general conclusion. However, the heterogeneity between individual studies is a key issue in whether research can be combined. Therefore, integrated analysis has combined tests and effect size, as well as graphical and statistical tools, for assessing heterogeneity to describe the commonly used fixed effect models and random effects in metaanalysis. A meta-analysis survey, for example, systematically assesses treatment outcomes or disease risk factors by using a funnel plot. The funnel plot plots the effect of the treatment versus the size of the study, showing deviations or systematic heterogeneity. A good data set yields a symmetric inverted funnel type, while an

830

G.-M. Hu et al.

Fig. 21 Schematic representation of the framework for analyzing millions of single-cell spatial transcriptomes. The first step in the single-cell sequencing process is to collect brain tissue from the brain region of interest and single cells are sorted and captured by using fluorescence-activated cell sorting (FACS), laser capture microdissection (Routy et al. 2018), and microfluidic devices. The captured single cell will be sequenced using amplified nucleic acid technology. Finally, statistical analysis, such as principal component analysis (PCA) and hierarchical clustering, can be used to distinguish individual cell characteristics into different groups. (Reprinted Reprinted adapted under terms of CC BY 4.0 from Zeng et al. 2018 copyright 2018)

asymmetrical funnel type presents a correlation. Notably, the researchers used simulations to verify that the results of the combined dataset and the results of the network meta-analysis were highly correlated (Winter et al. 2019). The single-cell RNA-seq (scRNA-seq) method has been successfully applied to discover and define cell phenotypes and subpopulations through experiments from conditions, techniques, and species. The scRNA-seq significantly deepens our insight into complex tissues, and the latest technology can process tens of

32

Single-Cell Technologies for Cancer Therapy

831

thousands of cells simultaneously. Therefore, in order to be able to analyze a huge amount of data sets, it is necessary to use appropriate tools, such as the bigSCale system. The bigSCale system provides an extensive analytical framework to analyze millions of cells to solve the challenges associated with large data sets. The analytical framework is used for modules of differential expression analysis, cell clustering, and marker identification. The use of directed convolution allows for the processing of large-scale datasets while retaining transcript information from individual units. It uses a directed down-sampling approach to define cell clusters with improved resolution by indexing single-cell transcriptomes that classify the rare populations. For example, the bigSCale has been used to assess 1.3 million cells from the developing rat brain (Iacono et al. 2018).

Algorithm, Grouping, and Data Visualization of Datasets from Single Cells Statistics and algorithms are tools that make big data analysis meaningful (Cannoodt et al. (2016) and Fig. 22). For example, by using genome-wide polygenic scores or protein clusters to identify cancer-related SNPs or factors. Penn software is a system that helps identify cancer cell metastasis and cancer evolution. It uses nextgeneration sequencing techniques to evaluate, trace, and classify tumor endogenous heterogeneous cells to form phylogenetic trees (Jiang et al. 2016). The biggest challenge in cancer databases is the distinction between available subtypes due to the uncertainty associated with database objects. The diagnostic system of mathematical algorithms plays an important role in the analysis of high-dimensional medical cancer databases. The fuzzy clustering techniques have been used in breast cancer databases to demonstrate the effectiveness of the proposed method by clustering accuracy (Kannan et al. 2018). Single cell omics studies systemic dynamic changes by high-throughput and unbiased methods to understand the spatial and temporal variability of targeted cells. Experimental data is through a dimensionality reduction (similarity, manifold learning, clustering, graph, etc.) method and group-based trajectory modeling (pathfinding, cell ordering, etc.) process to find potentially meaningful changes. More than 50 algorithms are currently available for combined applications (https://github. com/dynverse/dynmethods). The dimensionality reduction is applied to data compression and data visualization, while the trajectory modeling presents the potential spatial and temporal development path of each group. Therefore, the two are organized into a common framework for processing experimental data and will provide deep insight into the connection and gene regulation of cell differentiation, which can serve as a clue for laboratory and clinical disease treatment. Because many cellular dynamic processes are non-linear, including branching processes or cyclical modes (Fig.22a), the TI method samples and analyzes individual cells along the trajectory process to reconstruct the interrelationship between the cellular dynamic processes in different stages. Because the TI method has clearly sequenced the cell population along a continuous path, there is no need to establish

INPUT

- Raw counts - Normalised counts - Default parameters - Start cell(s) - End cell(s) - Cell groups - Timecourse - # of end states - # of branches - Marker genes

Fig. 22 (continued)

direct

b Wrapper type 1:

a

outputs subset of

Wrapper type 2: linear pseudotime

required or optional prior information

TI method

Trajectory inference

Wrapper type 3: cyclical pseudotime

5 6 7

4

1 2 3

Wrap

P(B) Pseudotime

Wrapper type 4: end state probability P(A)

- Branch network - Branch assignment - Branch pseudotime - Global pseudotime - End state probabilities - Cluster assignment - Cluster network - Dimensionality reduction - Cell graph

Output

Wrapper type 5: cluster assignment

Milestones Delayed commitment

Edges

Apply common analyses

Wrapper type 7: cell graph

- Visualise trajectory - Visualise expression - Trajectory comparison - Evaluation - ...

Wrapper type 6: orthogonal projection

Potential cell space

Cell positions

Common probabilistic trajectory model

832 G.-M. Hu et al.

Single-Cell Technologies for Cancer Therapy

Fig. 22 Framework, procedures, and demonstrations of the TI method. (A) Interfaces of the TI method. (a) Raw or normalized counts, parameters, etc., after processing by the TI method, will use several wrapper functions to convert the results from TI processing into trajectory model. Finally, general trajectory model is going to do the analysis. (b) Different wrapper functions will produce different conversion graphs. (B) Some examples infer the trajectory through the TI framework. Multidimensional scaling produces dimensional reductions and dataset “consensus” predictions to infer the trajectory together. (a) Linear trajectory (b) Bifurcating trajectory (c) Four disconnected trajectories (d) Cyclic trajectory. (Reprinted with permission from Macmillan Publishers Ltd.: Saelens et al. 2019 copyright 2019)

32 833

834

G.-M. Hu et al.

additional time series models and the cellular heterogeneity and conditions of each group have been synchronized. Here, we introduce an unsupervised clustering algorithm and minimum span clustering (MSC) to classify G-protein coupled receptors (GPCRs) sequences for the GPCR network based on a data set from 2770 GPCRs and 652 non-GPCRs samples (Fig. 23). High detection accuracy can be achieved with an appropriate data set. The positive and negative selective pressure can be used to group data, such as genes or proteins. Finally, based on distance and feature methods to construct and compare phylogenetic trees, the combination of these methods can obtain comprehensive information (Hu 2017).

Principal Component Analysis (PCA) A key challenge in analyzing high-dimensional data in cancer research is how to reduce the data dimension and how to extract relevant features. Sparse principal component analysis (Sparse PCA) is a powerful statistical tool. A large number of high-dimensional data analyses often involve many variables. It is very difficult to understand the correlation patterns of a large number of variables, and some variables may be highly correlated, resulting in a lot of duplicate information. It can also lead to multicollinearity problems, making regression estimates inaccurate, but PCA is a way to reduce this situation. Therefore, the characteristics of the principal component analysis are that using the information of the original variables, the independent and new variables cannot overlap, and the reduced few variables replace the original multiple variables. In cancer analysis, Hsu et al. published the use of sparse PCA (Sparse PCAs) with variance maximization (VM), reconstruction error minimization (REM), singular value decomposition (SVD), and probabilistic modeling (PM), to analyze the genetic characteristics of lung cancer datasets. The results show that Sparse PCAs have high consistency compared to traditional PCA and can accurately predict true loading coefficient values, and when the number of genes is small, sparse PCA performed better (Hsu et al. 2014). Ingenuity Pathway Analysis The Ingenuity Pathway Analysis (Fernandez et al.) system is probably the world’s most comprehensive software with databases for biomedical analysis and research. Currently, many international pharmaceutical companies and well-known research institutions use IPA as an analysis tool for genomics, proteomics, drug toxicology, clinical trials, metabolism, and regulatory pathways. By using IPA to integrate experimental data from different omics platforms, medical doctors or researchers can gain a deeper understanding of the interactions between biomolecules, cell phenotypes, or disease development processes in experimental systems. IPA can predict downstream effects on biological and disease processes, activation and inhibition of transcription factors, and determine the most relevant signaling and metabolic pathways as well as phenotypes. Its interface covers gene and chemical, function and disease, ingenuity canonical pathways and toxicity lists library, and pathway creations (https://www.qiagenbioinformatics.com).

32

Single-Cell Technologies for Cancer Therapy

Fig. 23 (continued)

835

836

Fig. 23 (continued)

G.-M. Hu et al.

32

Single-Cell Technologies for Cancer Therapy

837

The interface is as follows: • Core Analysis, Tox Analysis, Metabolomics Analysis, Functions Analysis, Pathways, Upstream Analysis, Networks • Core Analysis: Analysis of mRNA, miRNA or protein • Tox Analysis: Toxicologically relevant results after analysis, • Metabolomics Analysis: Analysis of Metabolomics experimental data • Functions analysis: biological function, disease and toxicity results affected by molecular changes • Canonical Pathways: Experimental Impact on Signaling Pathway and Metabolic Pathway • Upstream Analysis: Upstream molecules associated with variable molecules, and whether they are activated or suppressed, predicted according to research literature • Networks: Intramolecular network relationships in experimental data

Business Market: Single Cells Related Technology and Personalized Medicine Avoiding “Path Dependence” is a future trend. For example, the monocyte activation test (MAT) replaces a large number of pyrogen tests using live rabbits and blood from limulidae. Using the “material” of human disease itself or “Information” to develop medicines will be the gold standard for future biomedical research and drug development. Nowadays, more than 17 countries around the world have established 3R (Replacement, Reduction, and Refinement) centers or similar units. Thus, The Joint Research Centre of the European Union (JRC) has also begun to develop biomedical research and toxicology without using animals. The expectation is that organ chips, mini-brains, etc., can replace experimental animals. Combining companies with professional strengths will provide the opportunity to build a more complete system of medical knowledge and services for systems biology and bioinformatics. For example, if Regeneron Pharmaceuticals and BlueBird Bio jointly develop anti-cancer therapy, they can share their technology platforms for the discovery, development, and commercialization of new cancer therapies. In terms of technical advantages, Regeneron Pharmaceuticals owns VelociSuite technology, including VELOCIMMUNE and Veloci-T, which can be used to design fully human monoclonal antibodies and to determine T cell receptors (TCRs) for tumor-specific proteins and peptides. In addition, Bluebird bio’s strengths are gene transfer of lentiviral vector and T cell immunotherapy of chimeric antigen receptor T cell (CAR-T) and T cell receptor (TCR) therapy. Therefore, their collaboration can fully grasp the technologies of antibodies and antigens design, ä Fig. 23 An unsupervised clustering algorithm and minimum span clustering (MSC) to classify G-protein coupled receptors (GPCRs). (Reprinted with permission from Macmillan Publishers Ltd.: Hu et al. 2017 copyright 2017)

838

G.-M. Hu et al.

gene editing, and immunotherapy and will be able to make progress in T cell immunotherapy for cancer.

Single Cell Specific Biomarkers and Service for Immunotherapy BioLegend is located in the Sorrento Valley near the “Golden Triangle” of San Diego, about a 10 min’ drive from UCSD. There, BioLegend has four main business units for flow cytometry, biomarker immunoassay (BMIA), molecular cell immunology, and neurology. In the single cells detection service, BioLegend can use more than a dozen fluorescent-labeled antibodies, especially when the sample is small, which can maximize the experimental performance. Different biomarkers exist between different cells. For example, different kinds of T lymphocytes have different biomarkers (e.g., CD4 or CD8). Fluorescent antibodies and biomarkers can be used to calculate individual T cell values by flow cytometry, which is beneficial for subsequent experimental analysis. Currently, flow cytometry, biological antibodies and diagnostic reagents are the core business of BioLegend, and they are also indispensable technologies for immunomedicine, cancer cell classification, and future development of precision medicine (https://www.biolegend.com/). Human Longevity Human Longevity, Inc. (HLI) aims to build the world’s most comprehensive genomic data and phenotypic data, and then use machine learning to help develop new ways to combat aging-related diseases. HLI has reached an agreement with pharmaceutical companies Celgene and AstraZeneca to collaborate on research and is also providing a health service called “Health Nucleus” which provides customers with a range of medical tests, such as whole genome sequencing(GWS) for cancer and Alzheimer’s disease. HLI Health Nucleus is a clinical research and discovery center that relies on the HLI team’s bioinformatics, machine learning, and other technical expertise. Health Nucleus uses integrated data from genome-wide sequencing (GWS), advanced magnetic resonance imaging (Humphries et al. 2017), and other clinical related materials, and then uses it in four major disease areas: cancer, heart, metabolism, and neurodegenerative/neurovascular (https://www. humanlongevity.com/). AI/Deep Learning and Service in Personal Medicine In the global medical IoT market, there are data showing that the combination of AI and medical services, coupled with the $120 billion precision medical market, is estimated to reach $230 billion in the medical market by 2022. Thereby, Amazon founder Bezos, Microsoft Bill Gates, and Warren Buffett have rushed into the smart medical market. Medical big data helps medical doctors to judge the disease, determine the medical condition, and reduce medical risks, and precision medicine can help make sure medical resources are not wasted. AI with medical big data has great potential to save manpower from dull operation and diagnostic negligence. AI’s deep learning is through various algorithms and mathematical models, such as convolutional neural networks (CNN), capsule networks (CapsNet), generated

32

Single-Cell Technologies for Cancer Therapy

839

confrontation networks (GNS), natural language understanding (NLU), to produce human-like discernment wisdom. In particular, Ian Goodfellow’s GNS breaks through the past one-way passive learning of machines. Deep learning with CNN or CapsNets as the main architecture is being widely used in biology and medicine, such as clinical decision, knowledge engineering, natural language processing in healthcare, data analytics and mining, and medical intelligent devices. In particular, there have been many important advances in image recognition and generation. Practical examples show that it effectively classifies images of macular degeneration and diabetic retinopathy, accurately distinguishing between bacterial and viral pneumonia on chest X-rays (Kermany et al. 2018) and magnetic resonance imaging of brain tumors (Afshar et al. 2018). In the electronic market, the well-known NVIDIA Corporation is actively developing GPU chips, using deep learning and artificial intelligence functions for medical care and complex neural networks in medical research, and putting them into miniaturized devices. Now, deep learning and AI are reshaping the life sciences, medicine, and healthcare industries. AI technology with deep learning ability is being applied to tumor imaging interpretation. In February 2017, Andre Esteva of Stanford University and his team used deep learning convolutional neural networks (CNN) trained with 129,450 images and disease labels to display the classification of skin lesions. After training and testing by 21 dermatologists, the software has been able to identify the most common skin cancer (e.g., keratinocyte carcinoma) and malignant melanoma (Esteva et al. 2017). Another study published in the Journal of Onnals of Oncology showed that an AI skin cancer detection system has surpassed the performance of most dermatologists. The study by Holger Haenssle of the University of Heidelberg, Germany, also showed that the rate of successful diagnosis of skin cancer by the AI system was 95%, which is higher than the 86.6% of 58 dermatologists, and the rate of misdiagnosis of benign moles was also low (Haenssle et al. 2018). AI has also made progress in breast cancer detection. Kheiron Medical Technologies, a London-based company, has shown that the AI diagnostic software developed by the company is surpassing radiology diagnostics in screening for breast cancer (https://www.kheironmed.com/). In addition, AI’s diagnostic applications include eye diseases. DeepMind is a Google subsidiary that develops AI diagnostic software trained by 14,884 3D optical coherence tomography scans (OCT scans) for more than 50 kinds of eye diseases, such as glaucoma, diabetic retinopathy, and macular degeneration caused by aging. In the judgment of the correctness of the eye lesions, the results showed that the AI error rate (5.5%) was lower than the specialist’s error rate (6.7–24.1%). The company is also currently developing mammography, which will be used for image screening of breast cancer in the future (De Fauw et al. 2018). For other examples, the Google team uses machine learning to develop imaging diagnostics that can predict heart disease without blood measurement, and the AI system after training with 284,335 patients’ data can accurately infer patients’ age, gender, blood pressure, whether they are smokers, and other features which are important factors in predicting the risk of cardiovascular disease. This new method

840

G.-M. Hu et al.

predicts that the accuracy of major cardiovascular diseases, such as strokes or heart attacks, can reach as high as 70% within 5 years. Furthermore, Google’s Medical Brian team worked with Stanford University, the University of Chicago, and the University of California, San Francisco to study the AI model for predicting patient mortality. Google AI can predict the probability of a patient being hospitalized again and the risk of death in the short term. Surprisingly, this AI system predicts the probability of human death to be as high as 95%, which is higher than traditional medical predictions (Poplin et al. 2018).

IBM Watson IBM Watson is one of the most competitive off-the-shelf artificial intelligence platforms on the market. The Watson Oncology System is trained to help oncologists with a machine learning-based system that is a commercial service provided by IBM Watson Health. The IBM Watson AI system with DeepQA’s technology to “read” more than 300 medical journals, more than 250 medical textbooks, and the text and medical images of 15 million pages from research papers, provides clinical disease judgment and treatment strategies. Since 2015, IBM Watson has collaborated in hospitals and medical centers around the world. At present, IBM Watson is able to provide medical advice on at least 13 cancers, including lung cancer, breast cancer, cervical cancer, endometrial cancer, ovarian cancer, prostate cancer, prostate cancer, rectal cancer, stomach cancer, etc. In recent years, IBM Watson Health has partnered with the Memorial Sloan Kettering Cancer Center (MSKCC), the world’s oldest and most famous private cancer center, to continuously develop and train the Watson system to improve the accuracy, efficiency, and quality of care advice and treatment options. However, the MD Anderson Cancer Center – IBM cooperation suffered setbacks in 2017 and the cooperation has ended, because the University of Texas believes that IBM Watson’s ability is exaggerated. Since beginning cooperation in June 2012, the system is still under development and has already cost more than 62 million US dollars. Also, a survey by STAT, which is a media company, shows that Watson for Oncology did not meet IBM’s expectations. In a simulation test, Watson prescribed a drug that causes bleeding in cancer patients with bleeding symptoms, which can cause death in severe cases. At the real medical scene, many doctors also found unsafe and incorrect treatment opinions given by Watson (https://www.statnews. com). In 2016, IBM Watson spent approximately $4 billion US dollars to acquire four medical data companies, including Explorys, Phytel, and Merge Healthcare, but IBM Watson Health was reported to lay off 50–70% of the employees after 2019. Although Watson’s database holds knowledge of each disease and its decisions are not easily influenced by doctor bias, it still cannot overcome the hurdles between technology and business because it cannot be connected to different health care systems. In fact, IBM’s medical products have not been successfully commercialized, and IBM’s AI doctors have many gaps with their ideals.

32

Single-Cell Technologies for Cancer Therapy

841

It is extremely difficult to create an AI doctor. So far, the FDA has approved only a few AI-based tools in real hospitals and doctor’s offices. These applications are focused in the field of image diagnostics through computer vision tasks, such as X-ray and retinal scanning. In addition, Watson’s “cognitive” ability through NLP is still far from human beings, and it is impossible to find subtle clues that human doctors will notice. At present, no artificial intelligence system can match the understanding and insight of human doctors, so Watson is more like an AI assistant. Although there may be a medical deficiency, its powerful learning ability is not only being applied to medical treatment but also gradually being extended to various fields such as information security. IBM published an endpoint detection called IBM BigFix Detect at the RSA 2017 conference. It uses Watson AI technology to analyze the behavior of running programs and find the security threats to internal devices. It can detect known and unknown computer viruses or hacking attacks and then determine the attack mode. Almost at the same time, a response and remediate plan will be implemented to protect these devices from cyber-attacks. Therefore, although the AI’s medical services cannot completely replace doctors or experts, it is a must for the future to use its powerful learning ability to assist doctors in cancer treatment. At the present stage, artificial intelligence systems can consider more factors than clinical trials and can divide patients into more categories for “personalized care.”

Conclusions Single cell analysis is an emerging field with the potential to improve cancer early diagnosis, treatment efficacy and monitoring. This has been made possible thanks to the technological innovations on single-cell isolation and decreasing costs of highthroughput sequencing. Moreover, CTCs and cell-free DNA identification by liquid biopsies have facilitated its transition into the clinical practice (Alix-Panabieres and Pantel 2016). Longstanding and emerging biotech companies are pushing forward liquid biopsy and cancer early diagnosis and monitoring technology to decrease costs and provide the tests as a routine diagnostic. Further investigations in single cell analysis and cancer heterogeneity will surely improve the treatment monitoring, recurrence predictions, and survival of cancer in the following years. Although monitoring and identification of cancer progression have improved treatment monitoring, a bigger challenge is the generation and approval of drugs to keep up with the current findings and further advance precision medicine. In particular, advances in AI and computer technology will bring single cell technology to another level, because not only cancer but also more difficult neurodegenerative diseases require a lot of knowledge about cell functioning. The combination of biology, information science, pharmacology, nutrition, microbiota, imaging, clinical medicine and diagnostics, etc., is based on information from single cells or a single cell. Thus, the technologies and services related to single cells are still the future focus of research, biotech business, and personal medicine.

842

G.-M. Hu et al.

Acknowledgements We thank Dr. Fan-Gang Tseng and Dr. Tuhin Subhra Santra for their support on the microfluidic device, Dr. Ming-Hsien Chien at the Institute of Clinical Medicine for the advice on the cancer treatment, Dr. Chien-Wen Chung for nanoparticle delivery instruction, Mr. Ding Wang from Linkou Chang Gung Memorial Hospital for the advice on clinical trial, Mr. Min-Han Lee from the Taiwan Semiconductor Manufacturing Company for the advice on AI/deep leaning and database, Dr. Chao-Lung Chiang from National Synchrotron Radiation Research Center (NSRRC) for materials chemistry and chapter revision, King & Queen Entertainment Inc. for funding, and Academia Sinica and BIOGEN Inc. for the NGS guidance.

References Adams DL, Adams DK, He J, Kalhor N, Zhang M, Xu T, Gao H, Reuben JM, Qiao Y, Komaki R, Liao Z, Edelman MJ, Tang CM, Lin SH (2017) Sequential tracking of PD-L1 expression and RAD50 induction in circulating tumor and stromal cells of lung Cancer patients undergoing radiotherapy. Clin Cancer Res 23(19):5948–5958 Afshar P, Mohammadi A, Plataniotis KN (2018) Brain tumor type classification via capsule networks. https://arxiv.org/abs/1802.10200 Aguado LC, Schmid S, May J, Sabin LR, Panis M, Blanco-Melo D, Shim JV, Sachs D, Cherry S, Simon AE, Levraud JP, tenOever BR (2017) RNase III nucleases from diverse kingdoms serve as antiviral effectors. Nature 547(7661):114–117 Aguet F, Ardlie GK (2016) Tissue specificity of gene expression. Curr Genet Med Rep 4(4):163 Alix-Panabieres C, Pantel K (2016) Clinical applications of circulating tumor cells and circulating tumor DNA as liquid biopsy. Cancer Discov 6(5):479–491 AlJanahi AA, Danielsen M, Dunbar CE (2018) An introduction to the analysis of single-cell RNA-sequencing data. Mol Ther Methods Clin Dev 10:189–196 Azzopardi N, Lecomte T, Ternant D, Boisdron-Celle M, Piller F, Morel A, Gouilleux-Gruart V, Vignault-Desvignes C, Watier H, Gamelin E and Paintaud G (2011) Cetuximab Pharmacokinetics Influences Progression-Free Survival of Metastatic Colorectal Cancer Patients. https:// doi.org/10.1158/1078-0432.CCR-11-1081 Baker MS, Ahn SB, Mohamedali A, Islam MT, Cantor D, Verhaert PD, Fanayan S, Sharma S, Nice, EC, Connor M, Ranganathan S (2017) Accelerating the search for the missing proteins in the human proteome. Nature Commun 8:14271 Baral P, Umans BD, Li L, Wallrapp A, Bist M, Kirschbaum T, Wei Y, Zhou Y, Kuchroo VK, Burkett PR, Yipp BG, Liberles SD, Chiu IM (2018) Nociceptor sensory neurons suppress neutrophil and gammadelta T cell responses in bacterial lung infections and lethal pneumonia. Nat Med 24(4):417–426 Baslan T, Hicks J (2017) Unravelling biology and shifting paradigms in cancer with single-cell sequencing. Nat Rev Cancer 17(9):557–569 Behjati S, Lindsay S, Teichmann SA, Haniffa M (2018) Mapping human development at single-cell resolution. Development 145(3):dev152561 Belic J, Koch M, Ulz P, Auer M, Gerhalter T, Mohan S, Fischereder K, Petru E, Bauernhofer T, Geigl JB, Speicher MR, Heitzer E (2015) Rapid identification of plasma DNA samples with increased ctDNA levels by a modified FAST-SeqS approach. Clin Chem 61(6):838–849 Bystrykh LV, Belderbos ME (2016) Clonal analysis of cells with cellular barcoding: when numbers and sizes matter. Methods Mol Biol 1516:57–89 Cannoodt R, Saelens W, Saeys Y (2016) Computational methods for trajectory inference from single-cell transcriptomics. Eur J Immunol 46(11):2496–2506 Cermakova K, Hodges HC (2018) Next-generation drugs and probes for chromatin biology: from targeted protein degradation to phase separation. Molecules 23(8):1958 Chakraborty C, Sharma AR, Sharma G, Doss CGP, Lee SS (2017) Therapeutic miRNA and siRNA: moving from bench to clinic as next generation medicine. Mol Ther Nucleic Acids 8:132–143

32

Single-Cell Technologies for Cancer Therapy

843

Chen C, Xing D, Tan L, Li H, Zhou G, Huang L, Xie XS (2017) Single-cell whole-genome analyses by linear amplification via transposon insertion (LIANTI). Science 356(6334):189–194 Chen J, Niu N, Zhang J, Qi L, Shen W, Donkena KV, Feng Z, Liu J (2018) Polyploid Giant Cancer Cells (PGCCs): the evil roots of cancer. Curr Cancer Drug Targets 19(5):360–367 Chin W, Zhong G, Pu Q, Yang C, Lou W, De Sessions PF, Periaswamy B, Lee A, Liang ZC, Ding X, Gao S, Chu CW, Bianco S, Bao C, Tong YW, Fan W, Wu M, Hedrick JL, Yang YY (2018) A macromolecular approach to eradicate multidrug resistant bacterial infections while mitigating drug resistance onset. Nat Commun 9(1):917 Chuang JC, Jones PA (2007) Epigenetics and microRNAs. Pediatr Res 61(5 Pt 2):24R–29R Cieslik M, Chinnaiyan AM (2018) Cancer transcriptome profiling at the juncture of clinical translation. Nat Rev Genet 19(2):93–109 Cox DBT, Gootenberg JS, Abudayyeh OO, Franklin B, Kellner MJ, Joung J, Zhang F (2017) RNA editing with CRISPR-Cas13. Science 358(6366):1019–1027 Czerniecki SM, Cruz NM, Harder JL, Menon R, Annis J, Otto EA, Gulieva RE, Islas LV, Kim YK, Tran LM, Martins TJ, Pippin JW, Fu H, Kretzler M, Shankland SJ, Himmelfarb J, Moon RT, Paragas N, Freedman BS (2018) High-throughput screening enhances kidney organoid differentiation from human pluripotent stem cells and enables automated multidimensional phenotyping. Cell Stem Cell 22(6):929–940 e924 De Fauw J, Ledsam JR, Romera-Paredes B, Nikolov S, Tomasev N, Blackwell S, Askham H, Glorot X, O’Donoghue B, Visentin D, van den Driessche G, Lakshminarayanan B, Meyer C, Mackinder F, Bouton S, Ayoub K, Chopra R, King D, Karthikesalingam A, Hughes CO, Raine R, Hughes J, Sim DA, Egan C, Tufail A, Montgomery H, Hassabis D, Rees G, Back T, Khaw PT, Suleyman M, Cornebise J, Keane PA, Ronneberger O (2018) Clinically applicable deep learning for diagnosis and referral in retinal disease. Nat Med 24(9):1342–1350 Dupre TV, Doll MA, Shah PP, Sharp CN, Siow D, Megyesi J, Shayman J, Bielawska A, Bielawski J, Beverly LJ, Hernandez-Corbacho M, Clarke CJ, Snider AJ, Schnellmann RG, Obeid LM, Hannun YA, Siskind LJ (2017) Inhibiting glucosylceramide synthase exacerbates cisplatin-induced acute kidney injury. J Lipid Res 58(7):1439–1452 Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, Thrun S (2017) Dermatologist-level classification of skin cancer with deep neural networks. Nature 542(7639):115–118 Fabbri M, Calin GA (2010) Epigenetics and miRNAs in human cancer. Adv Genet 70:87–99 Feil G, Horres R, Schulte J, Mack AF, Petzoldt S, Arnold C, Meng C, Jost L, Boxleitner J, Kiessling-Wolf N, Serbest E, Helm D, Kuster B, Hartmann I, Korff T, Hahne H (2017) Bacterial cellulose shifts transcriptome and proteome of cultured endothelial cells towards native differentiation. Mol Cell Proteomics 16(9):1563–1577 Fernandez SV, Bingham C, Fittipaldi P, Austin L, Palazzo J, Palmer G, Alpaugh K, Cristofanilli M (2014) TP53 mutations detected in circulating tumor cells present in the blood of metastatic triple negative breast cancer patients. Breast Cancer Res 16(5):445 Fernandez MF, Reina-Perez I, Astorga JM, Rodriguez-Carrillo A, Plaza-Diaz J, Fontana L (2018) Breast Cancer and Its Relationship with the Microbiota. Int J Environ Res Pub Health 15(8):1747 Franz A, Wood W, Martin P (2018) Fat body cells are motile and actively migrate to wounds to drive repair and prevent infection. Dev Cell 44(4):460–470 e463 Garvin T, Aboukhalil R, Kendall J, Baslan T, Atwal GS, Hicks J, Wigler M, Schatz M (2015) Interactive analysis and assessment of single-cell copy-number variations. Nat Methods 12(11):1058–60. Gaublomme JT, Yosef N, Lee Y, Gertner RS, Yang LV, Wu C, Pandolfi PP, Mak T, Satija R, Shalek AK, Kuchroo VK, Park H, Regev A (2015) Single-cell genomics Unveils critical regulators of Th17 cell pathogenicity. Cell 163:1400–1412 Gaudelli NM, Komor AC, Rees HA, Packer MS, Badran AH, Bryson DI, Liu DR (2017) Programmable base editing of AT to GC in genomic DNA without DNA cleavage. Nature 551(7681):464–471 Gerlinger M, Rowan AJ, Horswell S, Math M, Larkin J, Endesfelder D, Gronroos E, Martinez P, Matthews N, Stewart A, Tarpey P, Varela I, Phillimore B, Begum S, McDonald NQ, Butler A,

844

G.-M. Hu et al.

Jones D, Raine K, Latimer C, Santos CR, Nohadani M, Eklund AC, Spencer-Dene B, Clark G, Pickering L, Stamp G, Gore M, Szallasi Z, Downward J, Futreal PA, Swanton C (2012) Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med 366(10):883–892 Gonzalez VD, Samusik N, Chen TJ, Savig ES, Aghaeepour N, Quigley DA, Huang YW, Giangarra V, Borowsky AD, Hubbard NE, Chen SY, Han G, Ashworth A, Kipps TJ, Berek JS, Nolan GP, Fantl WJ (2018) Commonly occurring cell subsets in high-grade serous ovarian tumors identified by single-cell mass cytometry. Cell Rep 22(7):1875–1888 Gopalakrishnan V, Spencer CN, Nezi L, Reuben A, Andrews MC, Karpinets TV, Prieto PA, Vicente D, Hoffman K, Wei SC, Cogdill AP, Zhao L, Hudgens CW, Hutchinson DS, Manzo T, Petaccia de Macedo M, Cotechini T, Kumar T, Chen WS, Reddy SM, Szczepaniak Sloane R, Galloway-Pena J, Jiang H, Chen PL, Shpall EJ, Rezvani K, Alousi AM, Chemaly RF, Shelburne S, Vence LM, Okhuysen PC, Jensen VB, Swennes AG, McAllister F, Marcelo Riquelme Sanchez E, Zhang Y, Le Chatelier E, Zitvogel L, Pons N, Austin-Breneman JL, Haydu LE, Burton EM, Gardner JM, Sirmans E, Hu J, Lazar AJ, Tsujikawa T, Diab A, Tawbi H, Glitza IC, Hwu WJ, Patel SP, Woodman SE, Amaria RN, Davies MA, Gershenwald JE, Hwu P, Lee JE, Zhang J, Coussens LM, Cooper ZA, Futreal PA, Daniel CR, Ajami NJ, Petrosino JF, Tetzlaff MT, Sharma P, Allison JP, Jenq RR, Wargo JA (2018) Gut microbiome modulates response to anti-PD-1 immunotherapy in melanoma patients. Science 359(6371):97–103 Gore L, Locatelli F, Zugmaier G et al. (2018) Survival after blinatumomab treatment in pediatric patients with relapsed/refractory B-cell precursor acute lymphoblastic leukemia. Blood Cancer J 8:80. https://doi.org/10.1038/s41408-018-0117-0 Gracz AD, Williamson IA, Roche KC, Johnston MJ, Wang F, Wang Y, Attayek PJ, Balowski J, Liu XF, Laurenza RJ, Gaynor LT, Sims CE, Galanko JA, Li L, Allbritton NL, Magness ST (2015) A high-throughput platform for stem cell niche co-cultures and downstream gene expression analysis. Nat Cell Biol 17(3):340–349 Griffiths JA, Richard AC, Bach K, Lun ATL, Marioni JC (2018) Detection and removal of barcode swapping in single-cell RNA-seq data. Nat Commun 9(1):2667 Haapaniemi E, Botla S, Persson J, Schmierer B, Taipale J (2018) CRISPR-Cas9 genome editing induces a p53-mediated DNA damage response. Nat Med 24(7):927–930 Haenssle HA, Fink C, Schneiderbauer R, Toberer F, Buhl T, Blum A, Kalloo A, Hassen ABH, Thomas L, Enk A, Uhlmann L (2018) Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Ann Oncol 29(8):1836–1842 Hawk JD, Bookout AL, Poplawski SG, Bridi M, Rao AJ, Sulewski ME, Kroener BT, Manglesdorf DJ, Abel T (2012) NR4A nuclear receptors support memory enhancement by histone deacetylase inhibitors. J Clin Invest 122(10):3593–3602 Heitzer E, Ulz P, Geigl JB (2015) Circulating tumor DNA as a liquid biopsy for cancer. Clin Chem 61(1):112–123 Heppner GH, Shekhar M (2014) Tumor heterogeneity is fundamental to the tumor ecosystem. Oncology (Williston Park) 28(9):780–781 Hoadley KA, Yau C, Hinoue T, Wolf DM, Lazar AJ, Drill E, Shen R, Taylor AM, Cherniack AD, Thorsson V, Akbani R, Bowlby R, Wong CK, Wiznerowicz M, Sanchez-Vega F, Robertson AG, Schneider BG, Lawrence MS, Noushmehr H, Malta TM, N. Cancer Genome Atlas, Stuart JM, Benz CC, Laird PW (2018) Cell-of-origin patterns dominate the molecular classification of 10,000 tumors from 33 types of cancer. Cell 173(2):291–304 e296 Hollis GJ, Keene TM, Ardlie RM, Caldicott DG, Stapleton SG (2017) Prehospital ketamine use by paramedics in the Australian Capital Territory: a 12 month retrospective analysis. Emerg Med Australas 29(1):89–95 Horsthemke B (2018) A critical view on transgenerational epigenetic inheritance in humans. Nat Commun 9(1):2973 Hou JM, Greystoke A, Lancashire L, Cummings J, Ward T, Board R, Amir E, Hughes S, Krebs M, Hughes A, Ranson M, Lorigan P, Dive C, Blackhall FH (2009) Evaluation of circulating tumor

32

Single-Cell Technologies for Cancer Therapy

845

cells and serological cell death biomarkers in small cell lung cancer patients undergoing chemotherapy. Am J Pathol 175(2):808–816 Howard SJ, Woodcock S, Ehrich J, Bokosmaty S (2017) What are standardized literacy and numeracy tests testing? Evidence of the domain-general contributions to students' standardized educational test performance. Br J Educ Psychol 87(1):108–122 Hrvatin S, Hochbaum DR, Nagy MA, Cicconet M, Robertson K, Cheadle L, Zilionis R, Ratner A, Borges-Monroy R, Klein AM, Sabatini BL, Greenberg ME (2018) Single-cell analysis of experience-dependent transcriptomic states in the mouse visual cortex. Nat Neurosci 21(1): 120–129 Hsu YL, Huang PY, Chen DT (2014) Sparse principal component analysis in cancer research. Transl Cancer Res 3(3):182–190 Hu GM, Mai TL, Chen CM (2017) Visualizing the GPCR network: classification and evolution. Sci Rep 7(1):15495 Humphries J, Xiong L, Liu J, Prindle A, Yuan F, Arjes HA, Tsimring L, Suel GM (2017) Speciesindependent attraction to biofilms through electrical signaling. Cell 168(1–2):200–209 e212 Iacono G, Mereu E, Guillaumet-Adkins A, Corominas R, Cusco I, Rodriguez-Esteban G, Gut M, Perez-Jurado LA, Gut I, Heyn H (2018) bigSCale: an analytical framework for big-scale singlecell data. Genome Res 28(6):878–890 Ihry RJ, Worringer KA, Salick MR, Frias E, Ho D, Theriault K, Kommineni S, Chen J, Sondey M, Ye C, Randhawa R, Kulkarni T, Yang Z, McAllister G, Russ C, Reece-Hoyes J, Forrester W, Hoffman GR, Dolmetsch R, Kaykas A (2018) p53 inhibits CRISPR-Cas9 engineering in human pluripotent stem cells. Nat Med 24(7):939–946 Jackowiak P, Hojka-Osinska A, Philips A, Zmienko A, Budzko L, Maillard P, Budkowska A, Figlerowicz M (2017) Small RNA fragments derived from multiple RNA classes – the missing element of multi-omics characteristics of the hepatitis C virus cell culture model. BMC Genomics 18(1):502 Jackson HJ, Rafiq S, Brentjens RJ (2016) Driving CAR T-cells forward. Nat Rev Clin Oncol 13(6):370–383 Jafri MA, Ansari SA, Alqahtani MH, Shay JW (2016) Roles of telomeres and telomerase in cancer, and advances in telomerase-targeted therapies. Genome Med 8(1):69 Janik R, Thomason LAM, Stanisz AM, Forsythe P, Bienenstock J, Stanisz GJ (2016) Magnetic resonance spectroscopy reveals oral Lactobacillus promotion of increases in brain GABA, N-acetyl aspartate and glutamate. NeuroImage 125:988–995 Janku F (2014) Tumor heterogeneity in the clinic: is it a real problem? Ther Adv Med Oncol 6(2):43–51 Jiang Y, Qiu Y, Minn AJ, Zhang NR (2016) Assessing intratumor heterogeneity and tracking longitudinal and spatial clonal evolutionary history by next-generation sequencing. Proc Natl Acad Sci U S A 113(37):E5528–E5537 Joshi N, Yan J, Levy S, Bhagchandani S, Slaughter KV, Sherman NE, Amirault J, Wang Y, Riegel L, He X, Rui TS, Valic M, Vemula PK, Miranda OR, Levy O, Gravallese EM, Aliprantis AO, Ermann J, Karp JM (2018) Towards an arthritis flare-responsive drug delivery system. Nat Commun 9(1):1275 Kalmanti L, Saussele S, Lauseker M et al. (2015) Safety and efficacy of imatinib in CML over a period of 10 years: data from the randomized CML-study IV. Leukemia 29:1123–1132. https:// doi.org/10.1038/leu.2015.36 Kannan SRK, Siva M, Ramathilagam S (2018) Effective kernel-based fuzzy clustering Systems in Analyzing Cancer Database. Data-Enabled Discov Appl 2(1):5 Kelly JR, Kennedy PJ, Cryan JF, Dinan TG, Clarke G, Hyland NP (2015) Breaking down the barriers: the gut microbiome, intestinal permeability and stress-related psychiatric disorders. Front Cell Neurosci 9:392 Kermany DS, Goldbaum M, Cai W, Valentim CCS, Liang H, Baxter SL, McKeown A, Yang G, Wu X, Yan F, Dong J, Prasadha MK, Pei J, Ting M, Zhu J, Li C, Hewett S, Dong J, Ziyar I, Shi A, Zhang R, Zheng L, Hou R, Shi W, Fu X, Duan Y, Huu VAN, Wen C, Zhang ED, Zhang

846

G.-M. Hu et al.

CL, Li O, Wang X, Singer MA, Sun X, Xu J, Tafreshi A, Lewis MA, Xia H, Zhang K (2018) Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell 172(5):1122–1131 e1129 Khurana E, Fu Y, Chakravarty D, Demichelis F, Rubin MA, Gerstein M (2016) Role of non-coding sequence variants in cancer. Nat Rev Genet 17(2):93–108 Kotagiri N, Cooper ML, Rettig M, Egbulefu C, Prior J, Cui G, Karmakar P, Zhou M, Yang X, Sudlow G, Marsala L, Chanswangphuwana C, Lu L, Habimana-Griffin L, Shokeen M, Xu X, Weilbaecher K, Tomasson M, Lanza G, DiPersio JF, Achilefu S (2018) Radionuclides transform chemotherapeutics into phototherapeutics for precise treatment of disseminated cancer. Nat Commun 9(1):275 Kovalevsky A, Aggarwal M, Velazquez H, Cuneo MJ, Blakeley MP, Weiss KL, Smith JC, Fisher SZ, McKenna R (2018) “To be or not to be” protonated: atomic details of human carbonic anhydrase-clinical drug complexes by neutron crystallography and simulation. Structure 26(3): 383–390 e383 Lafzi A, Moutinho C, Picelli S, Heyn H (2018) Tutorial: guidelines for the experimental design of single-cell RNA sequencing studies. Nat Protoc 13(12):2742–2757 Lee K, Conboy M, Park HM, Jiang F, Kim HJ, Dewitt MA, Mackley VA, Chang K, Rao A, Skinner C, Shobha T, Mehdipour M, Liu H, Huang WC, Lan F, Bray NL, Li S, Corn JE, Kataoka K, Doudna JA, Conboy I, Murthy N (2017) Nanoparticle delivery of Cas9 ribonucleoprotein and donor DNA in vivo induces homology-directed DNA repair. Nat Biomed Eng 1:889–901 Li Z, Zheng Z, Ruan J, Li Z, Zhuang X, Tzeng CM (2016) Integrated analysis miRNA and mRNA profiling in patients with severe oligozoospermia reveals miR-34c-3p downregulates PLCXD3 expression. Oncotarget 7(33):52781–52796 Li P, You S, Nguyen C, Wang Y, Kim J, Sirohi D, Ziembiec A, Luthringer D, Lin SC, Daskivich T, Wu J, Freeman MR, Saouaf R, Li D, Kim HL (2018a) Genes involved in prostate cancer progression determine MRI visibility. Theranostics 8(7):1752–1765 Li S, Jiang Q, Liu S, Zhang Y, Tian Y, Song C, Wang J, Zou Y, Anderson GJ, Han JY, Chang Y, Liu Y, Zhang C, Chen L, Zhou G, Nie G, Yan H, Ding B, Zhao Y (2018b) A DNA nanorobot functions as a cancer therapeutic in response to a molecular trigger in vivo. Nat Biotechnol 36(3):258–264 Linnemann JR, Miura H, Meixner LK, Irmler M, Kloos UJ, Hirschi B, Bartsch HS, Sass S, Beckers J, Theis FJ, Gabka C, Sotlar K, Scheel CH (2015) Quantification of regenerative potential in primary human mammary epithelial cells. Development 142(18):3239–3251 Lv Q, Gong L, Zhang T, Ye J, Chai L, Ni C, Mao Y (2016) Prognostic value of circulating tumor cells in metastatic breast cancer: a systemic review and meta-analysis. Clin Transl Oncol 18(3):322–330 Ma H, Marti-Gutierrez N, Park SW, Wu J, Lee Y, Suzuki K, Koski A, Ji D, Hayama T, Ahmed R, Darby H, Van Dyken C, Li Y, Kang E, Park AR, Kim D, Kim ST, Gong J, Gu Y, Xu X, Battaglia D, Krieg SA, Lee DM, Wu DH, Wolf DP, Heitner SB, Belmonte JCI, Amato P, Kim JS, Kaul S, Mitalipov S (2017) Correction of a pathogenic gene mutation in human embryos. Nature 548(7668):413–419 Masuda T, Sankowski R, Staszewski O, Bottcher C, Amann L, Sagar, Scheiwe C, Nessler S, Kunz P, van Loo G, Coenen VA, Reinacher PC, Michel A, Sure U, Gold R, Grun D, Priller J, Stadelmann C, Prinz M (2019) Spatial and temporal heterogeneity of mouse and human microglia at single-cell resolution. Nature 566(7744):388–392 Miyamoto DT, Zheng Y, Wittner BS, Lee RJ, Zhu H, Broderick KT, Desai R, Fox DB, Brannigan BW, Trautwein J, Arora KS, Desai N, Dahl DM, Sequist LV, Smith MR, Kapur R, Wu CL, Shioda T, Ramaswamy S, Ting DT, Toner M, Maheswaran S, Haber DA (2015) RNA-Seq of single prostate CTCs implicates noncanonical Wnt signaling in antiandrogen resistance. Science 349(6254):1351–1356 Moustafa A, Li W, Anderson EL, Wong EHM, Dulai PS, Sandborn WJ, Biggs W, Yooseph S, Jones MB, Venter JC, Nelson KE, Chang JT, Telenti A, Boland BS (2018) Genetic risk, dysbiosis, and

32

Single-Cell Technologies for Cancer Therapy

847

treatment stratification using host genome and gut microbiome in inflammatory bowel disease. Clin Transl Gastroenterol 9(1):e132 Moutinho C, Esteller M (2017) MicroRNAs and epigenetics. Adv Cancer Res 135:189–220 Najlah M, Ahmed Z, Iqbal M, Wang Z, Tawari P, Wang W, McConville C (2017) Development and characterisation of disulfiram-loaded PLGA nanoparticles for the treatment of non-small cell lung cancer. Eur J Pharm Biopharm 112:224–233 Navin N, Kendall J, Troge J, Andrews P, Rodgers L, McIndoo J, Cook K, Stepansky A, Levy D, Esposito D, Muthuswamy L, Krasnitz A, McCombie WR, Hicks J, Wigler M (2011) Tumour evolution inferred by single-cell sequencing. Nature 472(7341):90–94 Nicolazzo C, Raimondi C, Mancini M, Caponnetto S, Gradilone A, Gandini O, Mastromartino M, Del Bene G, Prete A, Longo F, Cortesi E, Gazzaniga P (2016) Monitoring PD-L1 positive circulating tumor cells in non-small cell lung cancer patients treated with the PD-1 inhibitor Nivolumab. Sci Rep 6:31726 Ogishima T, Shiina H, Breault JE, Tabatabai L, Bassett WW, Enokida H, Li LC, Kawakami T, Urakami S, Ribeiro-Filho LA, Terashima M, Fujime M, Igawa M, Dahiya R (2005) Increased heparanase expression is caused by promoter hypomethylation and up-regulation of transcriptional factor early growth response-1 in human prostate cancer. Clin Cancer Res 11(3): 1028–1036 Ortega MA, Poirion O, Zhu X, Huang S, Wolfgruber TK, Sebra R, Garmire LX (2017) Using single-cell multiple omics approaches to resolve tumor heterogeneity. Clin Transl Med 6(1):46 Ou HD, Phan S, Deerinck TJ, Thor A, Ellisman MH, O’Shea CC (2017) ChromEMT: Visualizing 3D chromatin structure and compaction in interphase and mitotic cells. Science (New York, N.Y.), 357(6349), eaag0025 Oya H, Howard MA, Magnotta VA, Kruger A, Griffiths TD, Lemieux L, Carmichael DW, Petkov CI, Kawasaki H, Kovach CK, Sutterer MJ, Adolphs R (2017) Mapping effective connectivity in the human brain with concurrent intracranial electrical stimulation and BOLD-fMRI. J Neurosci Methods 277:101–112 Pantel K, Alix-Panabieres C (2019) Liquid biopsy and minimal residual disease – latest advances and implications for cure. Nat Rev Clin Oncol 16:409–424 Peuker K, Muff S, Wang J, Kunzel S, Bosse E, Zeissig Y, Luzzi G, Basic M, Strigli A, Ulbricht A, Kaser A, Arlt A, Chavakis T, van den Brink GR, Schafmayer C, Egberts JH, Becker T, Bianchi ME, Bleich A, Rocken C, Hampe J, Schreiber S, Baines JF, Blumberg RS, Zeissig S (2016) Epithelial calcineurin controls microbiota-dependent intestinal tumor development. Nat Med 22(5):506–515 Phansalkar S, Desai AA, Bell D, Yoshida E, Doole J, Czochanski M, Middleton B, Bates DW (2012) High-priority drug-drug interactions for use in electronic health records. J Am Med Inform Assoc 19(5):735–743 Plottel CS, Blaser MJ (2011) Microbiome and malignancy. Cell Host Microbe 10(4):324–335 Poplin R, Varadarajan AV, Blumer K, Liu Y, McConnell MV, Corrado GS, Peng L, Webster DR (2018) Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning. Nat Biomed Eng 2:158 Qin S, Yin H, Yang C, Dou Y, Liu Z, Zhang P, Yu H, Huang Y, Feng J, Hao J, Hao J, Deng L, Yan X, Dong X, Zhao Z, Jiang T, Wang HW, Luo SJ, Xie C (2016) A magnetic protein biocompass. Nat Mater 15(2):217–226 Quandt D, Dieter Zucht H, Amann A, Wulf-Goldenberg A, Borrebaeck C, Cannarile M, Lambrechts D, Oberacher H, Garrett J, Nayak T, Kazinski M, Massie C, Schwarzenbach H, Maio M, Prins R, Wendik B, Hockett R, Enderle D, Noerholm M, Hendriks H, Zwierzina H, Seliger B (2017) Implementing liquid biopsies into clinical decision making for cancer immunotherapy. Oncotarget 8(29):48507–48520 Radvanyi LG (2018) Targeting the cancer mutanome of breast cancer. Nat Med 24(6):703–704 Ramassone A, Pagotto S, Veronese A, Visone R (2018) Epigenetics and MicroRNAs in cancer. Int J Mol Sci 19(2):459 Regev A, Teichmann SA, Lander ES, Amit I, Benoist C, Birney E, Bodenmiller B, Campbell P, Carninci P, Clatworthy M, Clevers H, Deplancke B, Dunham I, Eberwine J, Eils R, Enard W,

848

G.-M. Hu et al.

Farmer A, Fugger L, Gottgens B, Hacohen N, Haniffa M, Hemberg M, Kim S, Klenerman P, Kriegstein A, Lein E, Linnarsson S, Lundberg E, Lundeberg J, Majumder P, Marioni JC, Merad M, Mhlanga M, Nawijn M, Netea M, Nolan G, Pe’er D, Phillipakis A, Ponting CP, Quake S, Reik W, Rozenblatt-Rosen O, Sanes J, Satija R, Schumacher TN, Shalek A, Shapiro E, Sharma P, Shin JW, Stegle O, Stratton M, Stubbington MJT, Theis FJ, Uhlen M, van Oudenaarden A, Wagner A, Watt F, Weissman J, Wold B, Xavier R, Yosef N, P. Human Cell Atlas Meeting (2017) The human cell atlas. Elife 6:e27041 Reiman D, Ahmed AM, Yang D (2018) PopPhy-CNN: a phylogenetic tree embedded architecture for convolution neural networks for metagenomic data. https://doi.org/10.1101/257931 Routy B, Le Chatelier E, Derosa L, Duong CPM, Alou MT, Daillere R, Fluckiger A, Messaoudene M, Rauber C, Roberti MP, Fidelle M, Flament C, Poirier-Colame V, Opolon P, Klein C, Iribarren K, Mondragon L, Jacquelot N, Qu B, Ferrere G, Clemenson C, Mezquita L, Masip JR, Naltet C, Brosseau S, Kaderbhai C, Richard C, Rizvi H, Levenez F, Galleron N, Quinquis B, Pons N, Ryffel B, Minard-Colin V, Gonin P, Soria JC, Deutsch E, Loriot Y, Ghiringhelli F, Zalcman G, Goldwasser F, Escudier B, Hellmann MD, Eggermont A, Raoult D, Albiges L, Kroemer G, Zitvogel L (2018) Gut microbiome influences efficacy of PD-1-based immunotherapy against epithelial tumors. Science 359(6371):91–97 Saelens W, Cannoodt R, Todorov H, Saeys Y (2019) A comparison of single-cell trajectory inference methods. Nat Biotechnol 37(5):547–554 Sagiv-Barfi I, Czerwinski DK, Levy S, Alam IS, Mayer AT, Gambhir SS, Levy R (2018) Eradication of spontaneous malignancy by local immunotherapy. Sci Transl Med 10(426):eaan4488 Salvianti F, Pazzagli M, Pinzani P (2016) Single circulating tumor cell sequencing as an advanced tool in cancer management. Expert Rev Mol Diagn 16(1):51–63 Schmidt F, Efferth T (2016) Tumor heterogeneity, single-cell sequencing, and drug resistance. Pharmaceuticals (Basel) 9(2):33 Schorn AJ, Gutbrod MJ, LeBlanc C, Martienssen R (2017) LTR-retrotransposon control by tRNAderived small RNAs. Cell 170(1):61–71 e11 Scialdone A, Natarajan KN, Saraiva LR, Proserpio V, Teichmann SA, Stegle O, Marioni JC, Buettner F (2015) Computational assignment of cell-cycle stage from single-cell transcriptome data. Methods 85:54–61 Shah MY, Ferrajoli A, Sood AK, Lopez-Berestein G, Calin GA (2016) microRNA therapeutics in cancer – an emerging concept. EBioMedicine 12:34–42 Shalek AK, Satija R, Shuga J, Trombetta JJ, Gennert D, Lu D, Chen P, Gertner RS, Gaublomme JT, Yosef N, Schwartz S, Fowler B, Weaver S, Wang J, Wang X, Ding R, Raychowdhury R, Friedman N, Hacohen N, Park H, May AP, Regev A (2014) Single-cell RNA-seq reveals dynamic paracrine control of cellular variation. Nature 510:363–369 Shaw JA, Guttery DS, Hills A, Fernandez-Garcia D, Page K, Rosales BM, Goddard KS, Hastings RK, Luo J, Ogle O, Woodley L, Ali S, Stebbing J, Coombes RC (2017) Mutation analysis of cell-free DNA and single circulating tumor cells in metastatic breast cancer patients with high circulating tumor cell counts. Clin Cancer Res 23(1):88–96 Shen S, Lim G, You Z, Ding W, Huang P, Ran C, Doheny J, Caravan P, Tate S, Hu K, Kim H, McCabe M, Huang B, Xie Z, Kwon D, Chen L, Mao J (2017) Gut microbiota is critical for the induction of chemotherapy-induced pain. Nat Neurosci 20(9):1213–1216 Shibata M, Nishimasu H, Kodera N, Hirano S, Ando T, Uchihashi T, Nureki O (2017) Real-space and real-time dynamics of CRISPR-Cas9 visualized by high-speed atomic force microscopy. Nat Commun 8(1):1430 Sidhom JW, Bessell CA, Havel JJ, Kosmides A, Chan TA, Schneck JP (2018) ImmunoMap: a bioinformatics tool for T-cell repertoire analysis. Cancer Immunol Res 6(2):151–162 Sivan A, Corrales L, Hubert N, Williams JB, Aquino-Michaels K, Earley ZM, Benyamin FW, Lei YM, Jabri B, Alegre ML, Chang EB, Gajewski TF (2015) Commensal Bifidobacterium promotes antitumor immunity and facilitates anti-PD-L1 efficacy. Science 350(6264):1084–1089 Sol N, Wurdinger T (2017) Platelet RNA signatures for the detection of cancer. Cancer Metastasis Rev 36(2):263–272

32

Single-Cell Technologies for Cancer Therapy

849

Spinelli JB, Yoon H, Ringel AE, Jeanfavre S, Clish CB, Haigis MC (2017) Metabolic recycling of ammonia via glutamate dehydrogenase supports breast cancer biomass. Science 358(6365): 941–946 Subramanian S, Gao X, Dann CE 3rd, Kearns DB (2017) MotI (DgrA) acts as a molecular clutch on the flagellar stator protein MotA in Bacillus subtilis. Proc Natl Acad Sci U S A 114(51): 13537–13542 Sulciner ML, Serhan CN, Gilligan MM, Mudge DK, Chang J, Gartung A, Lehner KA, Bielenberg DR, Schmidt B, Dalli J, Greene ER, Gus-Brautbar Y, Piwowarski J, Mammoto T, Zurakowski D, Perretti M, Sukhatme VP, Kaipainen A, Kieran MW, Huang S, Panigrahy D (2018) Resolvins suppress tumor growth and enhance cancer therapy. J Exp Med 215(1): 115–140 Svensson V, Vento-Tormo R, Teichmann SA (2018) Exponential scaling of single-cell RNA-seq in the past decade. Nat Protoc 13(4):599–604 Thyagarajan A, Shaban A, Sahu RP (2018) MicroRNA-directed cancer therapies: implications in melanoma intervention. J Pharmacol Exp Ther 364(1):1–12 Tomasetti C, Vogelstein B, Parmigiani G (2013) Half or more of the somatic mutations in cancers of self-renewing tissues originate prior to tumor initiation. Proc Natl Acad Sci U S A 110(6): 1999–2004 Tramontano M, Andrejev S, Pruteanu M, Klunemann M, Kuhn M, Galardini M, Jouhten P, Zelezniak A, Zeller G, Bork P, Typas A, Patil KR (2018) Nutritional preferences of human gut bacteria reveal their metabolic idiosyncrasies. Nat Microbiol 3(4):514–522 Trejo-Becerril C, Perez-Cardenas E, Taja-Chayeb L, Anker P, Herrera-Goepfert R, MedinaVelazquez LA, Hidalgo-Miranda A, Perez-Montiel D, Chavez-Blanco A, Cruz-Velazquez J, Diaz-Chavez J, Gaxiola M, Duenas-Gonzalez A (2012) Cancer progression mediated by horizontal gene transfer in an in vivo model. PLoS One 7(12):e52754 Tsujiura M, Ichikawa D, Konishi H, Komatsu S, Shiozaki A, Otsuji E (2014) Liquid biopsy of gastric cancer patients: circulating tumor cells and cell-free nucleic acids. World J Gastroenterol 20(12):3265–3286 Tyssowski KM, DeStefino NR, Cho JH, Dunn CJ, Poston RG, Carty CE, Jones RD, Chang SM, Romeo P, Wurzelmann MK, Ward JM, Andermann ML, Saha RN, Dudek SM, Gray JM (2018) Different neuronal activity patterns induce different gene expression programs. Neuron 98(3):530–546 e511 Vander Heiden MG, Cantley LC, Thompson CB (2009) Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324(5930):1029–1033 Veglia F, Tyurin VA, Mohammadyani D, Blasi M, Duperret EK, Donthireddy L, Hashimoto A, Kapralov A, Amoscato A, Angelini R, Patel S, Alicea-Torres K, Weiner D, Murphy ME, KleinSeetharaman J, Celis E, Kagan VE, Gabrilovich DI (2017) Lipid bodies containing oxidatively truncated lipids block antigen cross-presentation by dendritic cells in cancer. Nat Commun 8(1):2122 Vishnoi M, Peddibhotla S, Yin W, T Scamardo A, George GC, Hong DS, Marchetti D (2015) The isolation and characterization of CTC subsets related to breast cancer dormancy. Sci Rep 5:17533 Wagle N, Emery C, Berger MF, Davis MJ, Sawyer A, Pochanard P, Kehoe SM, Johannessen CM, Macconaill LE, Hahn WC, Meyerson M, Garraway LA (2011) Dissecting therapeutic resistance to RAF inhibition in melanoma by tumor genomic profiling. J Clin Oncol 29 (22):3085–3096 Wang L, Dumenil C, Julie C, Giraud V, Dumoulin J, Labrune S, Chinet T, Emile JF, He B, Leprieur EG (2017) Molecular characterization of circulating tumor cells in lung cancer: moving beyond enumeration. Oncotarget 8(65):109818–109835 Wang L, Livak KJ, Wu CJ (2018) High-dimension single-cell analysis applied to cancer. Mol Asp Med 59:70–84 Winter C, Kosch R, Ludlow M, Osterhaus A, Jung K (2019) Network meta-analysis correlates with analysis of merged independent transcriptome expression data. BMC Bioinforma 20(1):144

850

G.-M. Hu et al.

Wu SC, Cao ZS, Chang KM, Juang JL (2017) Intestinal microbial dysbiosis aggravates the progression of Alzheimer’s disease in Drosophila. Nat Commun 8(1):24 Yamaguchi K, Urakami K, Ohshima K, Mochizuki T, Akiyama Y, Uesaka K, Nakajima T, Takahashi M, Tamai S, Kusuhara M (2014) Implementation of individualized medicine for cancer patients by multiomics-based analyses-the project HOPE. Biomed Res 35(6):407–412 Yamanaka YJ, Szeto GL, Gierahn TM, Forcier TL, Benedict KF, Brefo MS, Lauffenburger DA, Irvine DJ, Love JC (2012) Cellular barcodes for efficiently profiling single-cell secretory responses by microengraving. Anal Chem 84(24):10531–10536 Yan W, Wu X, Zhou W, Fong MY, Cao M, Liu J, Liu X, Chen CH, Fadare O, Pizzo DP, Wu J, Liu L, Liu X, Chin AR, Ren X, Chen Y, Locasale JW, Wang SE (2018) Cancer-cell-secreted exosomal miR-105 promotes tumour growth through the MYC-dependent metabolic reprogramming of stromal cells. Nat Cell Biol 20(5):597–609 Yano JM, Yu K, Donaldson GP, Shastri GG, Ann P, Ma L, Nagler CR, Ismagilov RF, Mazmanian SK, Hsiao EY (2015) Indigenous bacteria from the gut microbiota regulate host serotonin biosynthesis. Cell 161(2):264–276 Yip A, Webster RM (2018) The market for chimeric antigen receptor T cell therapies. Nat Rev Drug Discov 17(3):161–162 Zabara A, Chong JTY, Martiel I, Stark L, Cromer BA, Speziale C, Drummond CJ, Mezzenga R (2018) Design of ultra-swollen lipidic mesophases for the crystallization of membrane proteins with large extracellular domains. Nat Commun 9(1):544 Zeng Z, Miao N, Sun T (2018) Revealing cellular and molecular complexity of the central nervous system using single cell sequencing. Stem Cell Res Ther 9 Zhalnina K, Louie KB, Hao Z, Mansoori N, da Rocha UN, Shi S, Cho H, Karaoz U, Loque D, Bowen BP, Firestone MK, Northen TR, Brodie EL (2018) Dynamic root exudate chemistry and microbial substrate preferences drive patterns in rhizosphere microbial community assembly. Nat Microbiol 3(4):470–480 Zhang W, Xia W, Lv Z, Ni C, Xin Y, Yang L (2017) Liquid biopsy for cancer: circulating tumor cells, circulating free DNA or exosomes? Cell Physiol Biochem 41(2):755–768 Zickenrott S, Angarica VE, Upadhyaya BB, del Sol A (2016) Prediction of disease-gene-drug relationships following a differential network analysis. Cell Death Dis 7:e2040 Zou S, Fang L, Lee MH (2018) Dysbiosis of gut microbiota in promoting the development of colorectal cancer. Gastroenterol Rep (Oxf) 6(1):1–12 Zovkic IB, Paulukaitis BS, Day JJ, Etikala DM, Sweatt JD (2014) Histone H2A.Z subunit exchange controls consolidation of recent and remote memory. Nature 515(7528):582–586

Analytical Technology for Single-CancerCell Analysis

33

Ching-Te Kuo and Hsinyu Lee

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conventional Analytical Technology for Cancer Cell Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Two-Dimensional (2D) Cell-Based Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Three-Dimensional (3D) Cell-Based Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analytical Technology for Single-Cancer-Cell Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Single-Cell Sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inherent Traits of Single Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanical Traits of Single Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

852 853 853 854 855 855 858 858 860 861

Abstract

Traditional analysis in cancer initiates an empirical screening on bulk tumors based on standard equipment. Nonetheless, the tumors inherently exhibit distinct characteristics, including heterogeneity/plasticity/morphology of cells, cellmatrix or cell-cell interactions, and in-depth mass transport. Such traditional analysis may not acquire the invaluable yield since the critical cells of interest are in the minority as well as their activities could be altered during preparation. In recent years, emerging techniques in single-cell analysis have opened a new avenue targeting precise cancer medicine, including single-cell genome/transcriptome, next-generation sequencing, tumor spheroid formation targeting cancer stem cells, and microfluidic approach for the physical assay of cell size (mass/ C.-T. Kuo (*) Department of Mechanical and Electro-Mechanical Engineering, National Sun Yat-sen University, Kaohsiung, Taiwan e-mail: [email protected] H. Lee (*) Department of Life Science, National Taiwan University, Taipei, Taiwan e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_33

851

852

C.-T. Kuo and H. Lee

volume/density), deformability, and electrokinetic properties from single cells. In this chapter, an overall view of current techniques and commercial equipment for single-cell analysis in cancer and their potential translation into clinic will be present.

Introduction Development of a living system is likely to be that great oaks from little acorns grow. The biological living functions are originally transcribed and translated by single cells, the sole unit of life. Heterogeneity in cells is typically present at the genomic, transcriptomic, and proteomic levels, due to gene mutations or environmental stresses (Saadatpour et al. 2015). The heterogeneity may contribute the treating failure to cancers and instead leverage the disease recurrence, while a treatment targets only one tumor cell population. Another pitfall resulting from such heterogeneity may promote cells acquiring an invasive and metastatic phenotype that intrinsically presents a 90% death of cancer-associated diseases (Chaffer et al. 2016). This malignant phenotype attenuates the efficacy of both the chemotherapeutic drugs and the selective inhibitors that are employed in cell-cell interaction, epithelial to mesenchymal transition (EMT), and cancer stem cells (CSCs). Conventional analytical techniques targeting either phenotypes or genotypes of cells only profile the average traits from a bulk tumor, lacking the potential to explore complex tumor microenvironments. Recently, genome-scale approaches have been successfully applied for single-cell cancer biology studies (Wills and Mead 2015). The paradigm of single-cell technology is typically used to isolate single candidates from a pool of cells, generating genomic, transcriptomic, and proteomic data from the target cells. Techniques used for capturing single cells incorporate flow cytometry (Moldavan 1934), laser capture microdissection (Emmert-Buck et al. 1996), limiting dilution (Goding 1980), manual cell picking (Fröhlich and König 2000), and microfluidics (Gross et al. 2015; Punjiya et al. 2019). Relevant analytical techniques adopted for the captured cells include single-cell sequencing (Jaitin et al. 2014), high-throughput multiplexed quantitative polymerase chain reaction (qPCR) (Guo et al. 2013), mass cytometrybased proteomics (Bendall et al. 2011), and data computation (Amir et al. 2013). The above single-cell analyses would reveal intra-tumor heterogeneity, molecular subtyping, identification of cell types/subtypes, copy number variation (CNV) profiling, and mutation detection as well as lineage inference of interesting cells (Rantalainen 2018). It thereby not only provides fundamental insights into developmental process of human but also ignites potent applications in cancer treatments. Other techniques for single-cell analysis are based on the examination of inherent or mechanical traits of cells. For examples, one single cell incorporating with traits of EMT can self-grow to a unique tumor spheroid along with properties of CSCs and tumorigenesis (Mani et al. 2008). It can be easily assessed from the suspending culture of single cells in multiple cell culture plates. For the mechanical analysis,

33

Analytical Technology for Single-Cancer-Cell Analysis

853

Fig. 1 Single-cell analytical technology for cancer research

measurements of mass, density, deformability, and induced electrokinetic force of single cells can be assessed by microfluidic devices (Bryan et al. 2014; Gill et al. 2019; Hang et al. 2011). Comparing to the genome-based analysis, the nature or mechanical measurement could provide a facile tool and an auxiliary information to evaluate the differentiate traits of single cells derived from bulk tumors. In this chapter, recent technological and computational advances in single-cell analysis for cancer biology will be discussed (Fig. 1). Potential challenges and future application for cancer treatment will be concluded.

Conventional Analytical Technology for Cancer Cell Analysis Two-Dimensional (2D) Cell-Based Assay Conventional analytical technology for cancer cell analysis is mostly accomplished by 2D cell-based assay. This assay typically uses cell culture dishes or multi-well plates to cultivate cells. Drug screening, cell migration, cell invasion, and DNA/protein assay can be performed in situ and evaluated from conventional instruments. For example, cancer cells can be loaded in 96-well plates following standard protocols, and their viability or proliferation can be determined by a commonly used MTT (3-[4,5-dimethylthiazol-2-y1]-2,5-diphenyl tetrazolium

854

C.-T. Kuo and H. Lee

bromide) assay (Kuo et al. 2017b). Similarly, cell migration or cell invasion can be manually performed following protocols or by commercial devices (e.g., Transwell) (Kramer et al. 2013; Justus et al. 2014). The above assays typically needed abundant cells (>1000 cells) to perform one test, which is laborious and not applicable for rare cell study. Recent improvements for such assays are increasingly appreciated by the incorporation with microfluidics. A nanodroplet cell processing platform has been successfully adopted to perform high-throughput drug screenings, in which a critical number of 100 prostate PC-3 cells in a 200-nl droplet was used. The dramatic reduction of both cell numbers and media helped improve the predicted efficacy of in vivo anticancer treatment compared with a standard 96-well plate assay (Kuo et al. 2019). In addition, a novel micro-gap plate has been demonstrated to enable drug response profiling on rare primary tumor samples (Ma et al. 2015).

Three-Dimensional (3D) Cell-Based Assay Compared with the 2D cell culture, 3D tissue culture presents hallmark characteristics in cellular heterogeneity/plasticity, cell-matrix or cell-cell interaction, and complicate mass transport. Despite the distinction, it helps mimic the in vivo tumor microenvironment more closely. Therefore, 3D cell-based assay has emerged as a useful tool to impact broader medical applications for cell biology, tissue engineering, and drug development. Representative hallmarks of such assay included 3D drug screening assay, tumor dissemination assay, 3D invasion assay, and 3D co-culture assay (Lee et al. 2013; Kuo et al. 2017a). Prior to performing each assay, cells with a deterministic number are prepared and dispensed on the examined substrate, such as a U-shaped low attachment plate or a plain surface for performing hanging droplets. For hanging droplets, the dispensed cell droplet can be achieved either by manually pipetting or by a programmable liquid dispenser. The droplet volume dispensed was typically 10 μl (Tung et al. 2011). The volume dispensed from a robot can be, moreover, saved down to 100 nl or less (Kuo et al. 2019). After reversing the substrate to perform hanging droplets, the sediment and aggregated cells will self-assemble to perform a unique tumor spheroid during a specific period based on cell types. Analysis of 3D cells is typically different from 2D cultures. For example, neuroblastoma SK-N-DZ cells cultured in 3D acquired a strong affinity with MG132, a proteasome inhibitor, as compared to 2D culture (Kuo et al. 2018). For chemotherapeutic drugs (e.g., cisplatin and paclitaxel), in contrast, 3D cultures would present a more drug resistance than 2D cultures, highlighting 3D assays may become a potent indicator in cancer treatment (Tung et al. 2011; Kuo et al. 2014, 2017a). The above conventional analytical techniques (2D- and 3D-based assays) only profile the average traits from a pool of cancer cells or from a bulk tumor, which may lack the potential to discover complex and heterogeneous tumor microenvironments. To address this issue, single-cancer-cell analytical technology has been launched. Relevant techniques will be discussed in the following sections.

33

Analytical Technology for Single-Cancer-Cell Analysis

855

Analytical Technology for Single-Cancer-Cell Analysis Single-Cell Sequencing Single-Cell Isolation Single-cell isolation starts from profile and dissociation of solid tumors into a cell suspension, which is a typical first step for single-cell DNA or RNA sequencing (DNA-seq or RNA-seq). Techniques used for single-cell isolation have been extensively developed before and incorporated fluorescence-activated cell sorting (FACS; DNA- or RNA-seq), laser capture microdissection (LCM; DNA- or RNA-seq), limiting dilution (DNA- or RNA-seq), and manual cell picking (DNA- or RNA-seq) (Saadatpour et al. 2015; Baslan and Hicks 2017; Zhou et al. 2019; Yin et al. 2019; Datlinger et al. 2017; Stubbington et al. 2017; Picelli 2017). Recent improvements by microfluidics (DNA- or RNA-seq), microwell-based method (RNA-seq) and combinational indexing (DNA- or RNA-seq) offered a highthroughput and cost-efficient solution for single-cell isolation and library preparation (Punjiya et al. 2019; Hierahn et al. 2017; Vitak et al. 2017). Volume of reagents and yield of isolation can be effectively saved and enriched, respectively. Most recently, droplet microfluidics impacted high-throughput and ultra-sensitive single-cell profiling for deep and quantitative microRNAs and proteome analyses in human cancers (Zhu et al. 2018; Li et al. 2019a). Single-cell isolation from a biopsy of primary or metastatic tumors inherently encounters obstacles. The invasive surgical process is laborious and timeconsuming. In addition, increased anxiety from patients will dominate the failure of relevant cancer treatments. Recently, circulating tumor cells (CTCs) have been revealed to be a potentially substitute liquid biopsy of tumors, due to the advantage of using a minimally invasive process from a conventional blood sample (Hong et al. 2016). Since CTCs are present in an extremely low frequency 1 of 109 blood cells, several techniques have been developed to address such issue to hit facile capturing methods and efficient enrichments. Cellsearch (Veridex) was the most frequently used commercial instrument for capturing and enumeration of CTCs (Riethdorf et al. 2007). It adopts antibodies against EpCAM and cytokeratins a positive selection and a negative selection against leukocyte antigen CD45 incorporated with a nuclear 40 ,6-diamidino-2-phenylindole dye. CTC-chips, by combining with Cellsearch and microchip technology, have provided a newly effective avenue for precise identification and assay of CTCs in cancer patients (Nagrath et al. 2007). Improved technology to address the isolation of CTCs has been recently developed in numerous manners, for example, flow cytometry, single-cell pipette, microfluidic devices/ disks, Magsweeper, RosetteSep (STEMCELL Technologies Inc.), HD-CTC, MINDEC, VTX-1 (Vortex BioSciences), ClearCell FX1 System (Biolidics), CTC-ichip, and EPIC CTC platform (Rantalainen 2018; Zhang et al. 2016). Detailed information about existing single-cell analytical technology can be referred in Table 1.

856

C.-T. Kuo and H. Lee

Table 1 Existing single-cell technology for cancer research Single-cell isolation FACS Laser capture microdissection Limiting dilution Manual cell picking Microfluidics Droplet microfluidics CTC-chip

Molecular/ cell level DNA-seq RNA-seq DNA-seq RNA-seq DNA-seq RNA-seq DNA-seq RNA-seq DNA-seq RNA-seq microRNA WGA

Commercial product N

Reference Saadatpour et al. (2015)

N

Baslan and Hicks (2017)

N

Zhou et al. (2019)

N

Yin et al. (2019)

N N

Punjiya et al. (2019), Hierahn et al. (2017), and Vitak et al. (2017) Zhu et al. (2018) and Li et al. (2019a)

N

Nagrath et al. (2007)

Single-Cell Sequencing Single-cell genomic sequencing is typically initiated from amplification of genomic DNA or complementary DNA (i.e., RNA-seq) before preparation of sequencing libraries. Its potent contribution in cancer biology facilitates the exploration of clonal architecture of tumors. Theoretically, relatively less than 200 cells were needed to reliably examine 1% tumor mass clones through the detection of very low-level tumor clones (Gawad et al. 2014). The amplification and sequencing of single-cell RNA have been demonstrated to be more reliable than DNA-seq, since abundant RNA molecules inherently exist in cells compared to only two copies of DNA. Compared to present single-cell DNA-seq studies, however, numbers of cells typically from 100s to even 1000s cells will be needed to acquire reliable RNA sequencing data. Currently, whole-genome amplification (WGA) together with next-generation sequencing (NGS) is of minimal bias and sequencing error traits and will dominate the single-cell DNA-seq. There were several methods for WGA, including multiple displacement amplification (MDA), multiple annealing, degenerate oligonucleotide-primed (DOP) PCR, and looping-based amplification cycles (MALBAC) (Telenius et al. 1992; Dean et al. 2002; Borgstrom et al. 2017). These methods can be further applied either with a commercial method, the GenomePlex WGA4 (Sigma-Aldrich), or commercial kits, such as AMPLI1, MALBAC, Repli-G, and PicoPlex. For WGA of CTCs, it has been demonstrated that microfluidic devices can promote clinical investigation of single CTC sequencing, with minimum cell loss/human labor and high accuracy and repeatability (Li et al. 2019b). Single-cell transcriptomics examines the gene expressing levels of single target cells by simultaneously detecting their messenger RNAs (mRNAs). It helps unveil the cell biology mystery of heterogeneous cell population, cellular developmental maps, and cellular transcriptional dynamics, since conventional examination

33

Analytical Technology for Single-Cancer-Cell Analysis

857

methods from bulk tumors cannot easily approach the same standards. Single-cell transcriptomic sequencing can advance cancer studies by using single-cell qPCR. In colorectal cancer (CRC), for example, the pharmaceutical domination of multilayered aberrant transcriptional biocircuits will be contributed by the relevant transcriptional drivers and regulators (Kyrochristos and Roukos 2019). A newly developed sequencing method, in case of cellular indexing of transcriptomes and epitopes (CITE-seq), can perform not only the understanding of complex cell populations but also the examination of phenotypic information such as cell-surface protein levels (Stoechius et al. 2017). Another study using CRISPR-based genetic screening method (CRISPR-seq) enabled screening a pool of CRISPRs with single-cell transcriptome precision, facilitating high-throughput functional assay of heterogeneous cell populations and in-depth regulatory mechanisms (Datlinger et al. 2017). These indicate that the diversity of transcriptional bulk tumor subpopulations can be captured by such single-cell sequencings, in which conventional bio-assays cannot unveil the mystery based on population-level data. Single-cell proteomics, typically using flow cytometry, mass spectrometry, and multiplex fluorescent imaging/fluorescent ELISA, possesses an increased contribution to cancer research. For example, a combined technology with microfluidic nanodroplet platform and tandem mass tag (TMT) isobaric labelling can efficiently leverage assay throughput and single-cell proteome coverage (Dou et al. 2019). The labelling can improve throughput single-cell-sized assays to a depth of 1600 proteins with a suitable coefficient of variation (CV) of 10.9% and a good correlation coefficient of 0.98. A commercial single-cell IsoCode chip (isoplexis) combined with ELISA fluorescence-based assay enabled up to 40 proteins can be simultaneously detected from live single immune cells (Liu et al. 2020). This technology can help unveil functional heterogeneity of cells, evaluating medicine responses of patients to cancer immunotherapies. Other applications of single-cell proteomics were also discussed in review (Milardi et al. 2019).

Single-Cell Analysis and Data Computation Hallmarks of statistical and bioinformatic evaluations for single-cancer-cell studies typically incorporate intra-tumor heterogeneity, classification of rare cell subtypes, mutation detection, molecular subtyping, CNV profiling, and lineage inference of interesting cells. Given the nature distinction between single cells and cell populations, specific models and methods should be adopted to simulate singlecell environments instead of following conventional bulk average profiles (Stegle et al. 2015; Bacher and Kendziorski 2016). Numbers of specific methods for analysis of single-cell data have been demonstrated for DNA-seq (by one method of lineage inference) and RNA-seq (by methods of differential expression, pathway analysis, imputation, heterogeneity, pseudo-time ordering, clustering, dimensionality reduction, modelling of latent factors, quality control, and rare cell detection) (Rantalainen 2018).

858

C.-T. Kuo and H. Lee

Inherent Traits of Single Cells Cells grown within a bulk tumor inherently possess distinct characteristics compared to conventional 2D cell cultures. The distinctions are present to cellular heterogeneity, mass transport, and cell-matrix/cell-cell interaction. Cells cultured in 3D can acquire more closely the in vivo microenvironments, as the cellular mechanotransduction and malignant phenotypes in 2D dramatically diminish (Kuo et al. 2018). The most serious phenotype is present as cancer metastasis that mainly causes the mortality of 90% in human-associated cancers. EMT intervenes the metastasis to force cells reprogramming with traits of CSCs that present self-renewal tumorigenesis as well as drug resistance. These reprogrammed cells will be enriched with cell surface CSC markers, such as CD44, CD133, and ABCG2 (Kuo et al. 2014). For single-cancer-cell study, the tumorigenesis is typically used to investigate the traits of metastasis (Mani et al. 2008). The investigation can follow below. First, single cells resulted from serial dilutions can be loaded in an ultralow attachment 96-well plate, in which one well contains one unique cell. Second, the observation of one tumor spheroid derived from one specific cell can be examined by microscope. Finally, the target tumor spheroid can be applied for further examinations, such as genomic sequencing, detection with protein levels, and fluorescent staining with antibodies against EMT and CSC markers. For breast cancer study, BCL11A, a functional oncogene, has been unveiled to enable breast cancer cells acquiring a superior ability of self-renewal spheroids formation through the Wnt/β-catenin signaling activation (Zhu et al. 2019). BCL11A is also detected in numerous breast cancer samples. For gastric cancer study, the activation with KRAS can enrich cells with EMT and CSCs and increase migration and invasion, which was examined through the spheroid formation assay (Yoon et al. 2019). For liquid biopsy CTC study, the tumor spheroid formation from single cells provided the potent evidence of successful ex vivo culture of circulating breast tumor cells (Yu et al. 2014). Both the enriched CTCs and their passaged cell lines present newly acquired mutations already existing in the estrogen receptor gene (ESR1), fibroblast growth receptor gene (FGFR2), and mutations in the PIK3CA gene. Furthermore, it reveals that a potential surrogate therapeutic target can be achieved through the drug evaluation of CTCs with multiple mutations.

Mechanical Traits of Single Cells Size Cell size is one of the most important indicators for determining fundamental physical properties that typically link to physiological tradeoff, overall health, cell cycle, and metabolic function. Cell size can be determined by either mass or volume, and the ratio of mass to volume is density. Since such determination can present as high as 50% variations between cells, density of cells thereby benefits the tight regulation of cell size. In addition, cell density may frequently be adopted to classify

33

Analytical Technology for Single-Cancer-Cell Analysis

859

different populations of cells, instead of mass and volume. There are few methods and tools available to determine the size information of single cells, including flow cytometry and Coulter electronic counter for volume measurement and quantitative phase microscopy for mass data (Tzur et al. 2011; Hirsch and Gallian 1968; Barer et al. 1953). Recently, a microfluidic method has offered precise single-cell measurements with mass, volume, and density in physiological solutions and applicable environments with minimal disturbance to cells (Bryan et al. 2014). This method is accomplished as single cells flow through a suspended microchannel resonator (SMR) containing a fluid with lower density and then travel reversely through the same SMR but containing another fluid with higher density. Following the detection of buoyant mass of cells, the single-cell mass, volume, and density can thereby be measured. The intrinsic variation of cell density is smaller than the mass and volume variations with an approximately 100-fold. In addition, the density precision detected can be as small as 0.001 g/ml (Grover et al. 2011). The capability of measuring single-cell drug effect can be performed through the detection of density change in leukemia cells treating with staurosporine (STS). Furthermore, a link between single-cell mass and single-cell sequencing (RNA-seq) has been demonstrated from patient-derived glioblastoma cell lines, enabling the characterization of biophysical heterogeneity with and without drug treatment (Kimmerling et al. 2018).

Deformability Cell deformability is of label-free bio-indication for cellular fate in disease and physiological situations, in terms of stem cell differentiation to cancer progression and metastasis. The cell deformability can dominate the crucial role in cancer metastasis and relevant diagnostics. Typical approaches for measuring this indicator included atomic force microscopy, magnetic twisting cytometry, particle-tracking microrheology, parallel-plate rheometry, optical stretching, and flow cytometry, which are discussed in review (Wu et al. 2018). By the examination with multisample deformability cytometry (MS-DC), for example, the single-cell deformability strongly correlated with the metastatic potential either from breast cancer or prostate cancer cells but not with their molecular tags (Ahmmed et al. 2018). Drug-induced disruption of cellular actin networks increases the deformability of cancer cells. However, stabilization of cytoskeletal proteins does not alter the inherent property. Another facile method to detect the cell deformability, the fluidic shear stress (FSS), can commonly be used to examine that deformability of CTCs. For example, breast MDA-MB-231 cancer cells were responsible to deform with a slower rate but with a larger deformation threshold compared to non-metastatic MCF7 cells, through the detection with FSS microfluidic devices (Landwehr et al. 2018). Through the analysis of patient samples by this potent indicator of cell deformability, the exploration of potential anticancer candidates by screening of oncogenes can thereby be achieved.

860

C.-T. Kuo and H. Lee

Electrokinetic Properties Electrophoresis, dielectrophoresis (DEP), and electro-osmosis are a family of different physical effects occurring in heterogeneous fluids, called as the electrokinetic phenomena. Similar to living system that typically contains enriched fluids, the electrokinetic nature can also be used to distinguish physical properties from cell population levels. There were several methods well developed to achieve this issue, including using electric field, electromagnetic field, and DEP (Hang et al. 2011; Filipovic et al. 2014; Antfolk et al. 2017). A direct current electric field (dcEF) was demonstrated to affect the directional cell migration, presented as electrotaxis (Hou et al. 2015). Recent studies have demonstrated that the crucial role of dcEF dominates cell activity during cell differentiation and wound healing as well as embryo development. For lung cancer cell studies, the dcEF can influence adherence and tight junction and transcriptional regulation of telomerase RNA component gene hTerc of CL1–5 cells (Hang et al. 2011). Moreover, the benefit of using electrokinetic measurement combined with microfluidics technology can impact the labelfree cell analysis with either distinct physical properties or single-cell DNA/RNA/ protein analyses (Khamenehfar and Li 2016). All of these single-cell analytical methods have demonstrated to contribute a powerful weapon targeting the cancer diversity and mystery. It can be further applied for the early detection and selection of suitable therapeutic approach to personalized cancer medicine.

Conclusion Single-cancer-cell analytical technology is rapidly progressed in decade and has been applied for the investigation of in-deep clinical data. The data can be collected through profiling patient samples in terms of intra-tumor heterogeneity, drug resistance, and cancer evolution. Typical methods for single-cell analysis include single-cell isolation/ sequencing/computation and examination of tumorigenesis and mechanical characteristics from single cells. Although these methods have reached a maturity level, some pitfalls remain to interrogate such technology. For the purpose of high-throughput examination, for example, an integrated approach for single-cell isolation, analysis, computation, and even cultivation of cells needs to be streamlined without disturbing their properties. It may be daunted if not available. For single-cell analysis, it remains challenging to identifying technical variations from biological equivalent. In addition, it is difficult to recognize them from current technology as conventional clustering algorithms are programmed to identify large subpopulation of cells. Despite the addressed challenging, single-cell analytical technology still sustains extreme potentials at a higher precision with single-cell approaches compared to conventional bulk cell average phenotyping. The next step together with the single-cancer-cell analytical technology is anticipated to ignite studies of clinical outcomes from numerous and invaluable patient samples, pursing the goal of personalized cancer medicine. Acknowledgments Financial support from the Ministry of Science and Technology (MOST), Taiwan, under the grant 107-2622-8-002-018 is gratefully acknowledged.

33

Analytical Technology for Single-Cancer-Cell Analysis

861

References Ahmmed S, Bithi S, Pore A et al (2018) Multi-sample deformability cytometry of cancer cells. APL Bioeng 2:032002 Amir E-A, Davis K, Tadmor M et al (2013) viSNE enables visualization of high dimensional singlecell data and reveals phenotypic heterogeneity of leukemia. Nat Biotechnol 31:545–552 Antfolk M, Kim S, Koizumi S et al (2017) Label-free single-cell separation and imaging of cancer cells using an integrated microfluidic system. Sci Rep 7:46507 Bacher R, Kendziorski C (2016) Design and computational analysis of single-cell RNA-sequencing experiments. Genome Biol 17:63 Barer R, Ross K, Tkaczyk S (1953) Refractometry of living cells. Nature 171:720–724 Baslan T, Hicks J (2017) Unravelling biology and shifting paradigms in cancer with single-cell sequencing. Nat Rev Cancer 17:557–569 Bendall S, Simonds E, Qiu P, Amir E-A, Krutzik P, Finck R, Bruggner R, Melamed R, Trejo A, Ornastsky O, Balderas R et al (2011) Single-cell mass cytometry of differential immune and drug responses across a human hematopoietic continuum. Science 6:687–696 Borgstrom E, Paterlini M, Mold J et al (2017) Comparison of whole genome amplification techniques for human single cell exome sequencing. PLoS One 12:e0171566 Bryan A, Hecht V, Shen W et al (2014) Measuring single cell mass, volume, and density with dual suspended microchannel resonators. Lab Chip 14:569–576 Chaffer C, Juan BS, Lim E et al (2016) EMT, cell plasticity and metastasis. Cancer Metastasis Rev 35:645–654 Datlinger P, Rendeiro A, Schmidl C et al (2017) Pooled CRISPR screening with single-cell transcriptome readout. Nat Methods 14:297–301 Dean F, Hosono S, Fang L et al (2002) Comprehensive human genome amplification using multiple displacement amplification. PNAS 99:5261–5266 Dou M, Clari G, Tsai C et al (2019) High-throughput single cell proteomics enabled by multiplex isobaric labelling in a Nanodroplet sample preparation platform. Anal Chem. https://doi.org/ 10.1021/acs.analchem.9b03349 Emmert-Buck MR, Bonner RF, Smith PD et al (1996) Laser capture microdissection. Science 274:998–1001 Filipovic N, Djukic T, Radovic M et al (2014) Electromagnetic field investigation on different cancer cell lines. Cancer Cell Int 14:84 Fröhlich J, König H (2000) New techniques for isolation of single prokaryotic cells. FEMS Microbiol 24:567–572 Gawad C, Koh W, Quake S (2014) Dissecting the clonal origins of childhood acute lymphoblastic leukemia by single-cell genomics. PNAS 111:17947–17952 Gill N, Nyberg K, Lee L et al (2019) A scalable filtration method for high throughput screening based on cell deformability. Lab Chip 19:343–357 Goding JW (1980) Antibody production by hybridomas. J Immunol Methods 185:285–308 Gross A, Schoendube J, Zimmermann S et al (2015) Techniques for single-cell isolation. Int J Mol Sci 16:16897–16919 Grover W, Bryan A, Diez-Silva M et al (2011) Measuring single-cell density. PNAS 108:10992–10996 Guo G, Luc S, Marco E et al (2013) Mapping cellular hierarchy by single-cell analysis of the cell surface repertoire. Cell Stem Cell 13:492–505 Hang C, Chen H, Yen M et al (2011) Gene expression of human lung Cancer cell line CL1–5 in response to a direct current electric field. PLoS One 6:e25928 Hierahn T, Wadsworth MH, Hughes T et al (2017) Seq-Well: portable, low-cost RNA sequencing of single cells at high throughput. Nat Methods 14:395–398 Hirsch J, Gallian E (1968) Methods for the determination of adipose cell size in man and animals. J Lipid Res 9:110–119 Hong Y, Fang F, Zhang Q (2016) Circulating tumor cell clusters: what we know and what we expect. Int J Oncol 49:2206–2216

862

C.-T. Kuo and H. Lee

Hou H, Chang H, Cheng J (2015) Electrotaxis studies of lung cancer cells using a multichannel dual-electric-field microfluidic chip. J Vis Exp 106:53340 Jaitin D, Kenigsberg E, Keren-Shaul H et al (2014) Massively parallel single cell RNA-Seq for marker-free decomposition of tissues into cell types. Science 343:776–779 Justus C, Leffler N, Ruiz-Echevarria M et al (2014) In vitro cell migration and invasion assays. J Vis Exp 88:51046 Khamenehfar A, Li P (2016) Microfluidic devices for circulating tumor cells isolation and subsequent analysis. Curr Pharm Biotechnol 17:810–821 Kimmerling R, Prakadan S, Gupta A et al (2018) Linking single-cell measurements of mass, growth rate, and gene expression. Genome Biol 19:207 Kramer N, Walzl A, Unger C et al (2013) In vitro cell migration and invasion assays. Mutat Res 752:10–24 Kuo C, Chiang C, Chang C et al (2014) Modeling of cancer metastasis and drug resistance via biomimetic nano-cilia and microfluidics. Biomaterials 35:1562–1571 Kuo C, Wang J, Lin Y et al (2017a) Three-dimensional spheroid culture targeting versatile tissue bioassays using a PDMS-based hanging drop array. Sci Rep 7:4363 Kuo C, Wang J, Wo A et al (2017b) ParaStamp and its applications to cell patterning, drug synergy screening, and rewritable devices for droplet storage. Adv Biosyst 1:1700048 Kuo C, Lu S, Chen W et al (2018) Facilitating tumor spheroid-based bioassays and in vitro blood vessel modeling via bioinspired selfformation microstructure devices. Lab Chip 18:2453–2465 Kuo C, Wang J, Lu S et al (2019) A nanodroplet cell processing platform facilitating drug synergy evaluations for anti-cancer treatments. Sci Rep 9:10120 Kyrochristos I, Roukos D (2019) Comprehensive intra-individual genomic and transcriptional heterogeneity: evidence-based colorectal cancer precision medicine. Cancer Treat Rev 80:101894 Landwehr G, Kristof A, Rahman S et al (2018) Biophysical analysis of fluid shear stress induced cellular deformation in a microfluidic device. Biomicrofluidics 12:054109 Lee J, Mhawech-Fauceglia P, Lee N et al (2013) A three-dimensional microenvironment alters protein expression and chemosensitivity of epithelial ovarian cancer cells in vitro. Lab Investig 93:528–542 Li L, Lu M, Fan Y et al (2019a) High-throughput and ultra-sensitive single-cell profiling of multiple microRNAs and identification of human cancer. Chem Commun (Camb) 55:10404–10407 Li R, Jia F, Zhang W et al (2019b) Device for whole genome sequencing single circulating tumor cells from whole blood. Lab Chip. https://doi.org/10.1039/c9lc00473d Liu D, Paczkowski P, Mackay S et al (2020) Single-cell multiplexed proteomics on the IsoLight resolves cellular functional heterogeneity to reveal clinical responses of cancer patients to immunotherapies. Methods Mol Biol 2055:413–431 Ma W, Hsiung L, Wang C et al (2015) A novel 96well-formatted micro-gap plate enabling drug response profiling on primary tumour samples. Sci Rep 5:9656 Mani S, Guo W, Liao M et al (2008) The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell 133:704–715 Milardi D, Grande G, Vincerzoni F et al (2019) Proteomics for the identification of biomarkers in testicular cancer-review. Front Endocrinol (Lausanne) 10:462 Moldavan A (1934) Photo-electric technique for the counting of microscopial cells. Science 80:188–189 Nagrath S, Sequist L, Maheswaran S et al (2007) Isolation of rare circulating tumour cells in cancer patients by microchip technology. Nature 450:1235–1239 Picelli S (2017) Single-cell RNA-sequencing: the future of genome biology is now. RNA Biol 14:637–660 Punjiya M, Nejad H, Mathews J et al (2019) A flow through device for simultaneous dielectrophoretic cell trapping and AC electroporation. Sci Rep 9:11988 Rantalainen M (2018) Application of single-cell sequencing in human cancer. Brief Funct Genomics 17:273–282

33

Analytical Technology for Single-Cancer-Cell Analysis

863

Riethdorf S, Fritsche H, Muller V et al (2007) Detection of circulating tumor cells in peripheral blood of patients with metastatic breast cancer: a validation study of the cell search system. Clin Cancer Res 13:920–928 Saadatpour A, Lai S, Guo G et al (2015) Single-cell analysis in cancer genomics. Trends Genet 31:576–586 Stegle O, Teichmann S, Marioni J (2015) Computational and analytical challenges in single-cell transcriptomics. Nat Rev Genet 16:133–145 Stoechius M, Hafemeister C, Stephenson W et al (2017) Simultaneous epitope and transcriptome measurement in single cells. Nat Methods 14:865–868 Stubbington M, Rozenblatt-Rosen O, Regev A et al (2017) Single-cell transcriptomics to explore the immune system in health and disease. Science 358:58–63 Telenius H, Carter N, Bebb C et al (1992) Degenerate oligonucleotide-primed PCR: general amplification of target DNA by a single degenerate primer. Genomics 13:718–725 Tung Y, Hsiao A, Allen S et al (2011) High-throughput 3D spheroid culture and drug testing using a 384 hanging drop array. Analyst 136:473–478 Tzur A, Moore J, Jorgensen P et al (2011) Optimizing optical flow cytometry for cell volume-based sorting and analysis. PLoS One 6:e16053 Vitak S, Torkenczy K, Rosenkrantz J et al (2017) Sequencing thousands of single-cell genomes with combinatorial indexing. Nat Methods 14:302–308 Wills QF, Mead AJ (2015) Application of single-cell genomics in cancer: promise and challenges. Hum Mol Genet 24:R74–R84 Wu P, Aroush D, Asnacios A et al (2018) A comparison of methods to assess cell mechanical properties. Nat Methods 15:491–498 Yin Y, Jiang Y, Lam K et al (2019) High-throughput single-cell sequencing with linear amplification. Mol Cell S1097-2765(19):30618–30615 Yoon C, Till J, Cho S et al (2019) KRAS activation in gastric adenocarcinoma stimulates epithelialto-mesenchymal transition to cancer stem-like cells and promotes metastasis. Mol Cancer Res 17:1945–1957 Yu M, Bardia A, Aceto N et al (2014) Ex vivo culture of circulating breast tumor cells for individualized testing of drug susceptibility. Science 345:216–220 Zhang K, Gao M, Chong Z et al (2016) Single-cell isolation by a modular single-cell pipette for RNA-sequencing. Lab Chip 16:4742–4748 Zhou W, Ji Z, Fang W et al (2019) Global prediction of chromatin accessibility using small-cellnumber and single-cell RNA-seq. Nucleic Acids Res 47(19):e121 Zhu Y, Piehowski P, Zhao R et al (2018) Nanodroplet processing platform for deep and quantitative proteome profiling of 10–100 mammalian cells. Nat Commun 9:882 Zhu L, Pan R, Zhou D et al (2019) BCL11A enhances stemness and promotes progression by activating Wnt/β-catenin signaling in breast cancer. Cancer Manag Res 11:2997–3007

Transmembrane Receptor Dynamics as Biophysical Markers for Assessing Cancer Cells

34

Mirae Kim and Yen-Liang Liu

Contents Applications of SPT/SMT in Biology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principle of Single-Particle Tracking (SPT) and Trajectory Analysis . . . . . . . . . . . . . . . . . . . . . . . . . Structure-Property-Function-Disease Paradigm of TReD Assay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quantification of Metastatic Potential of Breast and Prostate Cancer Cells . . . . . . . . . . . . . . . . . . . Biomolecular Mechanisms Guiding the Dynamics of Transmembrane Receptors . . . . . . . . . . . . Deep Learning-Based TReD Assay with Better Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future Direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

869 870 873 874 876 877 879 880

Abstract

As biophysical properties of tissues and cells play essential roles in cellular function, morphogenesis, and disease progression, numerous conventional techniques have been developed to characterize cells and differentiate benign cells from cancer cells. Recently, Transmembrane Receptor Dynamics, a singleparticle tracking based biophysical phenotyping assay, was developed as a contact-free technique that can probe the topography of the plasma membrane and nanostructure of the membrane-associated cytoskeleton with sub-diffractionlimited resolution. Here, we review this state-of-the-art technology by narrating the underlying biophysical principles of single-particle tracking and discussing the interpretation of the dynamics of tyrosine kinase receptors. We further elaborate its potential for gaining insights into biology and being translated as a cancer diagnostic tool on the basis of machine learning models. M. Kim Department of Biomedical Engineering, The University of Texas at Austin, Austin, TX, USA Y.-L. Liu (*) Master Program for Biomedical Engineering, China Medical University, Taichung, Taiwan Research Center for Cancer Biology, China Medical University, Taichung, Taiwan e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_38

865

866

M. Kim and Y.-L. Liu

In recent years, researchers investigate biomedical questions using mechanical concepts and technologies from physics and engineering. These interdisciplinary studies provided unprecedented insights into the mechanisms of cell behavior (Van Helvert et al. 2018), and development (Petridou et al. 2017; Engler et al. 2006), and the mechanical response of tissues (Heisenberg and Bellaïche 2013; Sasai 2013). Across multiple spatial scales, the physical and mechanical properties of cells and tissues play essential roles in determining biological function. At the subcellular scale, the cellular signaling pathways are precisely regulated by the dynamics of transmembrane receptors. Their dynamics and binding kinetics can be affected by molecular factors such as the composition of the plasma membrane (Kusumi et al. 2014), cytoskeleton networks (Liu et al. 2019a), and protein-protein interactions (Kasai and Kusumi 2014). In particular, receptor tyrosine kinases (RTKs) participate in many cell decision-making functions such as propagation, differentiation, and migration. Misregulated RTKs are the leading cause of cancers (Blume-Jensen and Hunter 2001). The compromised spatial distribution and the derailed trafficking of RTKs could be hallmarks of carcinogenesis or even increased tumor progression (Casaletto and McClatchey 2012). At the cellular scale, viscoelasticity regulates cell differentiation (Engler et al. 2006) and migration (Van Helvert et al. 2018), and determining how cells interplay with external mechanical forces. For instance, the level of stemness in limbal stem cells is associated with cell elasticity (Bongiorno et al. 2016). On the other hand, the stiffness of the extracellular microenvironment directs stem cells toward various differentiation lineages (Engler et al. 2006) and regulates cell migration (Ng et al. 2012). At a tissue level, mechanical properties of tissues delicately guide collective migration (Barriga et al. 2018) and morphogenesis (Heisenberg and Bellaïche 2013). The abnormal changes in the physical and mechanical properties affect and are affected by the onset and progression of human cancers (Mohammadi and Sahai 2018). The tumor microenvironment transforms cancer cells into softer but highly invasive metastatic cells. The increased stiffness of endothelial cells also leads to atherosclerosis (Hahn and Schwartz 2009). Traditional phenotyping assays characterize cells on the basis of molecular analyses of genomic, epigenetic, transcriptomic, or proteomic biomarkers. While the inheritance of genes encodes the blueprint of living organisms, it is the cell’s physical properties that directly determine its function and behavior. Therefore, understanding the phenotypic features requires understanding the biophysical and biomechanical properties of cells, tissues, and even organs. Researchers have explored the physical characteristics of cancer cells (e.g., morphology (Kenny et al. 2007), viscoelasticity (Calzado-Martín et al. 2016), shear rheology (Bao and Suresh 2003), and motility (Kramer et al. 2013)) to provide a multifaceted description of cancer cells. Numerous biophysical or biomechanical phenotyping assays were developed to quickly and precisely differentiate malignant cancer subtypes from the benign or less invasive subtypes (Network 2013; Darling and Di Carlo 2015). As the current gold standard in the field of mechanobiology, atomic force microscopy (AFM) measures the quasi-static Young’s modulus (E) using a cantilever that scans through the surfaces of cells (Calzado-Martín et al. 2016; Osmulski et al. 2014; Radmacher et al. 1996; Krieg et al. 2019). Other methods to measuring

34

Transmembrane Receptor Dynamics as Biophysical Markers for Assessing. . .

867

stiffness comprise cytoindentation (Shin and Athanasiou 1999), micropillar deformation (Tan et al. 2003), cytometry (Otto et al. 2015), micropipette aspiration (Hochmuth 2000), magnetic twisting cytometry (Wang et al. 1993; Puig-deMorales-Marinkovic et al. 2007; Wang and Ingber 1995), optical tweezers (Dao et al. 2003; Zhang and Liu 2008), and microdroplet deformation (Serwane et al. 2017). However, these assays have their intrinsic limitations (Fig. 1). While AFM-based techniques, cytoindentation, and magnetic twisting cytometry provide subcellular or even submicron resolution, the need for direct physical contact between the probes and the cell surface reduce the throughput, and the external forces may arouse undesired cellular responses. The passive particle-tracking microrheology (Mason et al. 1997; Guo et al. 2014; Wirtz 2009) does allow the measurement of intracellular viscoelasticity without applied forces. Still, the ballistic injection or micro-injection of foreign nanoparticles into cells might diminish the viability of cells that require a few hours to heal from the intervention. Cell traction force (Tan et al. 2003; Paszek et al. 2005) and shear stress resistance(Bao and Suresh 2003). can be measured by microfabricated post array and shear flow techniques, respectively, to provide a better understanding of metastasis. However, all of these techniques are limited to adherent cells, which would restrict the throughput and might constrain their ability to be integrated into BioMEMS-based circulating tumor cell (CTC) capture devices. Micropipette aspiration (Hochmuth 2000) can be applied

Fig. 1 Summary of a variety of techniques for conducting physical or mechanical phenotyping of cells in terms of principles, physical properties to be measured, throughput, and cassette compatibility. (This figure was adapted with permission from Ref. (Liu et al. 2019a). Copyright 2019 Springer Nature)

868

M. Kim and Y.-L. Liu

to suspended cells, but it still does not avoid cell-tool interactions. Optical tweezers (Dao et al. 2003; Zhang and Liu 2008) and optical stretchers (Guck et al. 2005) avoid mechanical contact with the cells, but these light-based cell-manipulation approaches would not be able to measure the whole-cell deformability due to the relatively small-magnitude force (10
9–10
14 N) that can be applied to cells. The external force is too weak to deform cells sufficiently. Microfluidic assays (Hou et al. 2009; Gossett et al. 2012), such as deformation passage and hydrodynamic stretching, provide high-throughput mechanical phenotyping of whole cells under precisely controlled fluid flow. The microfluidic channels need to be properly designed to avoid inducing mechanical stress on cells. The cell rolling (Network 2013) and the cell migration (Kramer et al. 2013) only provide medium throughput. A recent study compared these techniques to each other and highlighted the intrinsic variability of these techniques (Wu et al. 2018). The Yeh group recently created a biophysical phenotyping technique named TReD (Transmembrane Receptor Dynamics) (Liu et al. 2019a). They demonstrated that the dynamics of epidermal growth factor receptor (EGFR) on the plasma membrane could be used to quantify metastatic potentials of breast (Liu et al. 2019a) and prostate (Liu et al. 2019b) cancer cells. The fluorescently labeled EGFRs were monitored using a technique of single-particle tracking (SPT). The EGFR trajectories provide information about the EGFRs: diffusivity of EGFRs on the plasma membrane, transition states of EGFR dimerization, and the subcellular structure in which the EGFRs are located. Their studies illustrate that EGFR dynamics can be indicators for quantifying the metastatic potentials of breast cell lines and monitor the epithelial-mesenchymal transition of benign cell lines (Liu et al. 2019a). In addition, the EGFR diffusivity is positively correlated with cell propagation, motility, and invasiveness of prostate cancer cells (Liu et al. 2019b). As TReD assay is substantially more accessible than the other methods, TReD can be integrated with a microfluidic CTC capture device, which exhibits the highthroughput isolation of CTCs from the blood (Fig. 2). This technique works with suspended cells and does not require mechanical contact or applied force on cells under examination. Once the cells were trapped in the microfluidic chip, and their EGFRs were fluorescently labeled, EGFR tracking was applied to monitoring EGFR dynamics without external flow. In addition, the acquired trajectories from several subcellular regions in one single cell (5–20 trajectories) would provide a higher spatial resolution than those from whole-cell measurements. While researchers have spent tremendous efforts to develop biophysical phenotyping assay in the last two decades, the first large-scale clinical validation of biophysical biomarkers for breast cancer diagnosis was completed in 2020. ARTIDIS AG, a health-tech company, recently announced that its AFM-based nanotechnology platform for breast cancer diagnosis had met the primary endpoint in the “NANO” clinical study with the enrollment of 520 patients. The ARTIDIS platform was able to detect cancer, including lesions with 1.00

0.06

0.31

0.02~0.50

1.58~2.05 1.47~1.72 2.60  0.42

0.55

500

0.81

55  11 170  20 190  70 230  60 47  4

0.76  0.05 0.93  0.03 1.42  0.09 1.38  0.17 0.83  0.02 0.47  0.03 1.52  0.26

1990 Human T cells and B cells

T cells B cells Spleen cells Lymph node cells Thymocytes

625

0.53

1990 Human red blood cells

1000

0.56

1989 Human erythrocytes

57  5

64  6

52  7

Single-Cell Impedance Flow Cytometry (continued)

Gimsa et al. (1996) Sukhorukov and Zimmermann (1996)

Becker et al. (1995)

Asami and Yamaguchi (1993) Kakutani et al. (1993) Wang et al. (1994)

Gimsa et al. (1991)

Kaler and Jones (1990)

Georgiewa et al. (1989) Donath et al. (1990) Hu et al. (1990)

35 897

Subtype

Solution: 8 ms/m 25 ms/m 150 ms/m >150 ms/m 1998 Lilium longiflorum External thunb generative cells conductivity: 3 ms/m 19 ms/m 480 ms/m Lilium pollen External protoplasts conductivity: 3 ms/m 19 ms/m 166 ms/m 470 ms/m Lilium pollen grains External conductivity: 3 ms/m 19 ms/m 186 ms/m 460 ms/m 1998 Saccharomyces Control cells cerevisiae yeast cells Field (80 μT, 50 Hz) exposed 1999 Hamster kidney fibroblast (BHK (C-13))

1997 Saccharomyces cerevisiae yeast cells 1997 Marine alga Valonia utricularis

Year Cell type

Table 2 (continued)

62 6  1.5

250 200 200 50

50 20 50

1.00 0.90 0.70 1.00

1.00 0.90 0.80

2.00

400 100 40

0.85 0.80 0.85

Specific membrane conductance (S/m2)

100 300 200 170  40 30

Membrane capacitance (pF)

0.80 0.70 1.00 1.20  0.02 0.75

0.76

Membrane relative Specific membrane permittivity capacitance (μF/cm2)

(25  0.8)  10 (25  0.2)  10

8

8

0.30~0.42

0.90 0.90 1.10 0.19  0.02 0.20  0.01

0.75 0.90 0.75 0.90

0.60 0.65 0.75

60  10 60  8 75  12

6 4 3 50 60

3 10 5 8

4 4 7

Archer et al. (1999)

Zhou et al. (1998)

Sukhorukov et al. (1998)

0.60

1

Wang et al. (1997)

Holzel (1997)

Cytoplasm relative Cell permittivity number References

1.50 1.50 0.26~2

0.90~1.20

(5.8~50)  10 7

Cytoplasm conductivity (S/m)

Membrane conductivity (S/m)

898 H. Liang et al.

MCF/neo MCF/HER2-11 MCF/HER2-18

T-lymphocytes B-lymphocytes Monocytes Granulocytes

2002 Human red blood cells Bacteria E. coli Saccharomyces cerevisiae yeast cells 2002 Murine myeloma Sp2/ 0-Ag14 cells 2006 Human Jurkat cells Sorbitol: 100 mosm Sorbitol: 200 mosm Sorbitol: 300 mosm Trehalose: 100 mosm Trehalose: 200 mosm Trehalose: 300 mosm

2001 Giardia intestinalis Cyclospora cayetanensis 2001 Schizosaccharomyces yeast cells 2002 Human MCF-7 sublines

2000 Mouse myeloma cells

1999 Bacterial E. coli 1999 Human leukocyte subpopulations

8 8 8

6  0.5 4  0.5

46  21 98  15 No 35  29 162  45 58  35 82  41

0.73  0.01 0.94  0.05 1.35  0.04 1.42  0.05 1.22  0.04 1.40  0.05

490  70

10~100

0.70  0.01

2.09 1.70 2.56

0.67  0.01

1.5  0.5 1.05  0.31 1.26  0.35 1.53  0.43 1.10  0.32 0.65~0.69 4

(1~3.8)  10 10 6 10 6 6

(2  0.8)  10 6 (5.5  0.3)  10

(25  14)  10

5

0.84  0.07

0.58  0.05

0.40  0.02

1.10  0.10

0.57  0.03

0.28  0.02

0.25  0.03

0.50~0.54 0.50 0.55

60 60

0.80  0.20 0.80  0.10

106  10

104  12

138  5

72  13

50 50 55

133  8

104  25 154  40 127  35 151  39 120~130

0.44  0.10 0.65  0.15 0.73  0.18 0.56  0.10 0.60  0.13 0.25~0.58

9

13

10

12

12

11

60

37 27 25

91 49 43 33

(continued)

Reuss et al. (2002) Kiesel et al. (1999)

Mietchen et al. (2002)

Kriegmaier et al. (2001) Cristofanilli et al. (2002)

Kürschner et al. (2000) Dalton et al. (2001)

Holzel (1999) Yang et al. (1999)

35 Single-Cell Impedance Flow Cytometry 899

50

336  73 1810  14

1.04 1.76  0.09 1.31  0.08 0.91  0.05 0.89  0.14 1.23  0.22 1.10  0.90 0.87  0.17 1.56  0.30 0.92  0.15 0.81  0.17 1.88  0.07 3.17  0.12 2.82  0.09

10.1  1.6 13.9  3.2

12.5  2.2

9.8  1.9

17.7  3.4 10.4  2.6 9.2  2.8

2953  82 450  20 113  18

1377  22

1895  59

100

0.60

2600  600

499

Specific membrane conductance (S/m2)

2010 Human Jurkat E6-1 Tcells Human insulinoma β-cells 2010 Human HL-60 cells Genistein, 0 h Genistein, 2 h Genistein, 4 h 2014 Human MDR K562 No treatment leukemia cells Doxorubic: 0.1 μM Doxorubic: 0.3 μM Doxorubic: 0.5 μM Imatinib: 0.2 μM Imatinib: 0.3 μM Imatinib: 0.5 μM 2014 Human glioblastoma DK-MG multiforme (GBM) GAMG cell lines U87-MG 9.4  0.2 28.8  0.9 15.9  0.3

Membrane capacitance (pF)

0.80  0.10

9.0

1.24  0.18 2.06  0.11 1.24  0.09 1.26  0.07

Membrane relative Specific membrane permittivity capacitance (μF/cm2)

2009 Murine polyomavirus

2007 Human red blood cells

Subtype

MCF-7 MCF-7TaxR MCF-7DoxR MCF-7MDR1

Year Cell type

2007 Human MCF-7 cell lines

Table 2 (continued)

Membrane conductivity (S/m)

0.60  0.10 0.27  0.10 0.78  0.19

0.26  0.04

0.41  0.07

0.32  0.08 0.14  0.01

0.50

1

0.23  0.01 0.14  0.01 0.40  0.02 0.27  0.02 0.42~0.55

Cytoplasm conductivity (S/m)

50

50

50~210

420 300 420

Memmel et al. (2014)

Bahrieh et al. (2014)

Wang et al. (2010)

Sudsiri et al. (2007) Berardi et al. (2009) Sancho et al. (2010)

Coley et al. (2007)

Cytoplasm relative Cell permittivity number References

900 H. Liang et al.

2016 Human white blood cells Human lymphoma cells (U937) Human pancreatic cancer cells (PANC1) Human serum-starved PANC1s Human EMT-induced PANC1s Human pancreatic cancer cells (BxPC3) Human serum-starved BxPC3 cells Human BxGR80c, strongly gemcitabineresistant subclone Human BxGR 360c, strongly gemcitabineresistant subclone

2015 Human Raji cells

Human embryonic kidney HEK293 line Human fibroblast cell line HFIB-1 2014 Human cell lines

HEK Jurkat PC3

U373-MG SNB19

0.66~0.84 0.42~0.53 0.40~0.53 0.44~0.65 0.39~0.50 0.45~0.52 0.46~0.57

0.53~0.61

1.30~1.54 1.83~2.33 1.32~1.81 1.68~2.49 2.13~2.39 1.95~2.34 2.50~2.86

2.45~2.85

0.50  0.10 0.65  0.12 0.90  0.15

0.61~0.89

25.3  0.7 21.6  0.7

0.90~1.10

0.33  0.01 0.24  0.01 0.34  0.01 0.79  0.01~0.94  0.03

2.05  0.12

4.00  0.12 3.71  0.15 1.56  0.10

81~99

81~94

76~94

84~98

72~103

96~163

77~101

95~118

79~122

60 60 60

54

55

39

101

29

53

97

57

49

360 360

Liang et al. (2015) Lannin et al. (2016)

Vaillier et al. (2014)

35 Single-Cell Impedance Flow Cytometry 901

902

H. Liang et al.

2002; Mietchen et al. 2002), only electrical data from hundreds of single cells were reported, which cannot meet the requirements of high-throughput electrical characterization of single cells.

Dielectrophoresis In dielectrophoresis, the numbers of cells attached to the electrodes applied with electrical signals as a function of frequency were first obtained, which were further translated into specific membrane capacitance by curve fitting of the Clausius–Mossotti factor spectra (see Fig. 1c) (Gimsa 2001). As pioneers in this field, Hughes et al. used this technique for the characterization of single-cell electrical properties, reporting (1) specific membrane capacitance of 0.82 versus 0.76 μF/cm2 and cytoplasm conductivity of 0.23 versus 0.50 S/m for K562 cells and their counterparts with multidrug resistance (Labeed et al. 2003); (2) cytoplasm conductivity of 0.23, 014, 0.40, and 0.27 S/m for MCF-7, MCF-7 TaxR, MCF-7 DoxR, and MCF-7 MDR1 cells, respectively. Based on a literature survey, key values of single-cell electrical properties measured by dielectrophoresis are summarized in Table 3. However, since the spectra of dielectrophoresis are derived from cell populations, only averaged electrical properties of cells rather than data of single cells can be obtained.

Impedance Flow Cytometry In an impedance flow cytometry, a cell flows through an orifice rapidly with a limited cross-sectional area and blocks the electric lines within this orifice, leading to impedance variations. By interpreting impedance variations for traveling single cells, cellular electrical properties can be obtained, enabling cell-type classification and cellular status evaluation (Cheung 2010; Chen et al. 2015).

Prototype Demonstration As pioneers in this field, Hoffman et al. used both direct currents and radiofrequencies to detect simultaneously the low- and high-frequency impedance changes produced by biologic cells traveling through a microsensing orifice (see Fig. 2a) (Hoffman and Britt 1979). For nonmembranous particles and plastic microspheres, both the direct current volumes and radiofrequency signals were proportional to particle volumes. Cells having intact plasma membranes produced radiofrequency impedance changes dependent additionally on the electrical properties of plasma membranes and intracellular structures. Thus, biologically different cells having the same direct current volumes can be distinguished if they differ in their electrical properties (see Fig. 2b and c) (Hoffman et al. 1981). In comparison to conventional approaches such as electrorotation and patch clamping, in this impedance flow cytometry, single cells traveled continuously

0.89~1.06 1.08~1.20

2006 Human K562 cells

2007 Human normal keratinocytes cells (UP)

10 25 18 18

0.63~0.88

1.75  0.08 1.76  0.09 1.59  0.07 1.31  0.08 0.91  0.05 0.76~0.88

0.89~1.06

Untreated cells Genistein, 0 h Genistein, 1 h Genistein, 2 h Genistein, 4 h K562 K562 + XR9576 K562AR K562Ar + XR9576 Early stage apoptosis Late stage apoptosis Cell necrosis 5K Hha-3 hns Hha-3 5 K hns

Subtype

2006 Human K562 cells

2006 Bacterial strains E. coli

2005 Human K562 cells

Human K562AR cells

2003 Human K562 cells

Year Cell type 1993 Saccharomyces cerevisiae yeast cells 2002 Human promyelocytic HL-60 cells

Specific Membrane membrane capacitance relative permittivity (μF/cm2) Membrane capacitance (pF)

(25  11)  102 (25  10)  102 (24  11)  102 (25  12)  102 (42  10)  102

Specific membrane conductance (S/m2)

Table 3 A summary of key cellular electrical properties quantified by dielectrophoresis

259  10 441  10 498  10 400  10

6

6

6

6

Membrane conductivity (S/m) 2.5  10 7

0.40~0.50

0.27~0.32

0.003 0.48 0.33 0.48 0.31 0.22~0.24

0.03

0.21~0.24 0.19~0.21 0.48~0.51 0.51~0.58 0.50

Cytoplasm conductivity (S/m) 0.50

40~80

(continued)

Chin et al. (2006) Labeed et al. (2006) Broche et al. (2007)

Castellarnau et al. (2006)

Broche et al. (2005)

Labeed et al. (2003)

Cytoplasm relative permittivity References Huang et al. (1993) Wang et al. (2002)

35 Single-Cell Impedance Flow Cytometry 903

2009 Saccharomyces cerevisiae yeast cells Mammalian cells (BA/F3 cells) 2009 Human T-lymphocytes

Human K562AR cells

2008 Human K562 cells

2007 Human Jurkat T-cells

Cell type Human oral squamous cells carcinoma cells (H357) 2007 Human MDR breast cancer cells

Year

Table 3 (continued)

MCF-7 MCF-7TaxR MCF-7DoxR MCF-7MDR1 No treatment Etoposide exposure: 2 h 4h 6h No treatment Verapamil Quinine NPPB XR9576 No treatment Verapamil Quinine NPPB XR9576

Subtype

1.33  0.18

2.30

1.21  0.34 1.04  0.40

1.24  0.18 2.06  0.11 1.24  0.09 1.26  0.07 1.32  0.29 1.25  0.25

Specific Membrane membrane capacitance relative permittivity (μF/cm2) 1.64~2.15 Membrane capacitance (pF)

Specific membrane conductance (S/m2)

2.5  10

7

Membrane conductivity (S/m)

0.58

0.28 0.22 0.27 0.25 0.28 0.50 0.41 0.34 0.25 0.55 0.50

0.23  0.01 0.14  0.01 0.40  0.02 0.27  0.02 0.70~0.90 0.04

Cytoplasm conductivity (S/m) 0.29~0.33

Vykoukal et al. (2009)

Vahey and Voldman (2009)

Duncan et al. (2008)

Pethig and Talary (2007)

Coley et al. (2007)

Cytoplasm relative permittivity References

904 H. Liang et al.

2014

2014

2012

2011

SC23 SC27 Mouse NSPCs E12 E16 Human oral Primary normal keratinocyte cell lines oral keratinocytes (HOK) Dysplastic (DOK) Malignant (H357) Malignant (H157) Human embryonic H1 stem cells H9 RCM1 RH1 T8 H1-MSC H9-MSC RCM1-trophoblast Human Media: 0.03 S/m mesenchymal stem 0.10 S/m cells Chlorella microalgae

B-lymphocytes Monocytes Neutrophils Eosinophils Basophils 2010 Human T lymphocyte Rat insulinoma β-cells 2011 Human NSPCs

5

2.0 4.1

1.09  0.20 1.51  0.26 1.43  0.45 1.61  0.39 1.42  0.37 1.76  0.67 1.79  0.39 2.00  0.53 4.16  1.10 4.94  1.22 2.83  0.73 2.2 4.5

50

1.04 0.99  0.02 0.76  0.03 0.82  0.05 1.07  0.06 0.69  0.06

100

0.99  0.08 0.42  0.08 0.94  0.04 0.98  0.01 1.12  0.12 0.60

10

8

0.50

0.50 0.50

0.42  0.26 0.26  0.06 0.25  0.10

0.71  0.08

0.50

1

60

60 60

50

50

Single-Cell Impedance Flow Cytometry (continued)

Deng et al. (2014)

Adams et al. (2014)

Velugotla et al. (2012)

Mulhall et al. (2011)

Labeed et al. (2011)

Sancho et al. (2010)

35 905

2015 Human skeletal stem and bone cell populations

2015 Human oral mucosa cells

2014 Human cell lines

2014 Chinese hamster ovary cells

Year Cell type 2014 Human oral cancer cells

Table 3 (continued)

HEK Jurkat PC3 Normal Abnormal MG-63 SAOS-2 STRO-1 positive

Subtype Rapid adherent cells Middle adherent cells Late adherent cells 6.8 0.33  0.01 0.24  0.01 0.34  0.01 0.65  0.26 0.38  0.10 1.60  0.70 1.36  0.33 1.07  0.32

409  74 358  70 392  163

244  29

1.01  0.17

Specific membrane conductance (S/m2) 454  64 586  114

Membrane capacitance (pF)

1.51  0.27

Specific Membrane membrane capacitance relative permittivity (μF/cm2) 1.81  0.33

3  10

3

Membrane conductivity (S/m)

0.23  0.05 0.52  0.11 0.34  0.28

0.50  0.10 0.65  0.12 0.90  0.15

0.44  0.08 0.22~0.52

0.36  0.05

Cytoplasm conductivity (S/m) 0.40  0.07

60 60 60

60

Ismail et al. (2015)

Graham et al. (2015)

Saboktakin Rizi et al. (2014) Vaillier et al. (2014)

Cytoplasm relative permittivity References Liang et al. (2014)

906 H. Liang et al.

2016 Mouse hippocampal neuronal Mouse glial cells 2017 Human Raji cells MCF-7cells HEK293 cells K562 cells

2016 Human NSCLC HCC1833 cells

Human HeLa cells

2016 Human Jurkat cells

2015 Human Jurkat cells

No treatment Staurosporine No treatment DOX-treated No treatment Staurosporine No treatment ABT-263: 6 h 10 h 24 h

1.20 1.11  0.09 1.15  0.08 0.90  0.09 1.02  0.07

0.91 0.80 0.77~0.99 0.66~0.77 3.87 3.54 1.25 1.56 1.60 1.60 1~1.20

782  32 114  28 187  22 879  24

12.5~25 1.3~12.5

1179

0.35

0.07~0.46 0.02~0.12 0.38 0.26 0.80 0.96 0.98 0.98 0.75~0.95 60

60

60 70~80

Liang et al. (2017)

Zhou et al. (2016)

Taruvai Kalyana Kumar et al. (2016)

Henslee et al. (2016)

Mulhall et al. (2015)

35 Single-Cell Impedance Flow Cytometry 907

908

a

H. Liang et al.

ENTRANCE CAP

EXIT CAP

FLOW CHAMBER BODY Pt WIRE

10/32 0.031 “O” RING

0.050

CHO CELLS

50 40

c

MICROSPHERES

50 40

12.5 m

20 9.55 m

TRYPSINIZED

PEPSIN TREATED

50 40 30 20 10 GLUT FIXED

Pt WIRE

PLASTIC MICROSPHERES

V79 CONTROL

12.5 m

30

ETHANOL FIXED

50 40 30

CHANNEL No. (0°, 4.5 MHz r f IMPEDANCE)

CHANNEL No. (0°, 4.5 MHz r f FREQUENCY)

10

10/32

JOINING NUT

30 20

Pt FOIL

ORIFICE

PLATINUM TUBING SAMPLE INLET

b

0.013

“O” RING

0.031

10

50

9.55 m

XRT-5h

ACT-D-5h

X-ACT-D-5h

XRT-6H

ACT-D-6h

X-ACT-D-6h

40 30 20 10

50 40 30 20

20

10

10 10 20 30

10 20 30

COULTER dc VOLUME

10 20 30

10 20 30

10 20 30

COULTER dc VOLUME

Fig. 2 (a) Schematic of the prototype demonstration of the impedance flow system where a single sheath stream was used to constrain the traveling cells to the center of the sensing orifice with impedance values detected by electrodes. (b) Contour plot patterns of impedance data for fresh and treated CHO cells including trypsin treatment, pepsin treatment, glutaraldehyde fixation, and ethanol fixation while the solid line in the upper left panel was a calibration provided by plastic microspheres. (c) Contour plot patterns of impedance data for V79 cells damaged by radiation (x-ray) or chemical treatments (actinomycin-D) or in combination while the solid line in the upper left panel was a calibration provided by plastic microspheres. (Reproduced with permission from Hoffman et al. 1981)

through the detection orifice and thus a much higher number of cells was characterized electrically. However, in this prototype demonstration, the geometries of the micro orifices and the relative positions between the electrodes and the orifices were prone to errors, leading to inconsistent measurement results from device to device, and the compromised functions in quantifying single-cell electrical properties.

35

Single-Cell Impedance Flow Cytometry

909

Microfluidic Impedance Flow Cytometry In order to address the aforementioned problems, microfluidic impedance flow cytometry was then proposed where single cells continuously flowed through microfabricated channels as orifices with fine-tuned geometries where electrodes used for impedance measurement were also based on microfabrication. Renaud et al. are the pioneers in the field of microfluidic impedance flow cytometry (Gawad et al. 2001, 2004; Cheung et al. 2005; Mernier et al. 2011, 2012; Meissner et al. 2012; Shaker et al. 2014). As shown in Fig. 3a, cells were flushed through a microfluidic channel integrated with a differential pair of microelectrodes with the impedance data measured at two given frequencies (Gawad et al. 2001). As a functional demonstration, normal erythrocytes and erythrocyte ghost cells (namely the erythrocytes with cytoplasm replaced with phosphate buffer solution) were characterized and differentiated. The impedance data for these two types of cells were found similar at 1.72 MHz indicating comparable cell sizes whereas, significantly different at 15 MHz suggesting differences in cytoplasm conductivity (see Fig. 3b). Furthermore, Renaud et al. improved the microfluidic impedance flow cytometry (Cheung et al. 2005) by replacing the previously reported coplanar microelectrodes with parallel overlap microelectrodes (see Fig. 3c). In this study, opacity was defined as |Zhigh|/|Zref| to remove the dependence of the impedance data on particle sizes. Figure 3d shows the opacity spectrum (|Zhigh|/|Zref|) of red blood cells with increasing glutaraldehyde concentrations. At the high frequency domain, different opacities were located, due to the crosslinking of glutaraldehyde with cytoplasmic macromolecules, which decreased the cytoplasm conductivity.

Microfluidic Impedance Flow Cytometry Based on Constriction Channels The drawback of the aforementioned microfluidic impedance flow cytometry is the lack of close contacts between cells and electrodes when the cells flowed through the detection areas formed by two electrodes. This issue leads to current leakages where electric signals circumvent the cells under measurement by traveling through solutions surrounding the cells. Thus, intrinsic cellular electrical parameters such as specific membrane capacitance and cytoplasmic conductivity cannot be measured. In order to address this technical issue, Chen et al. proposed a microfluidic impedance flow cytometry based on constriction channels with a cross-sectional area smaller than that of biological cells. As shown in Fig. 4a, when a cell is aspirated to deform through the constriction channel, it effectively seals the constriction channel and blocks electric fields, leading to higher impedance values, which are further translated into intrinsic cellular electrical parameters including specific membrane capacitance and cytoplasm conductivity (Zhao et al. 2013a, b). Based on this approach, single-cell electrical properties derived from multiple cell types were collected and correlated with tumor developments (Zhao et al. 2014; Chiu et al.

910

H. Liang et al.

Fig. 3 (a) The schematic of the microfluidic impedance flow cytometry where a microfluidic channel integrated with a differential pair of coplanar microelectrodes was included to quantify two-frequency impedance data of single cells flushed through the measurement area in a highthroughput manner. (Reproduced with permission from Gawad et al. 2001). (b) Normal erythrocytes and erythrocyte ghost cells (erythrocytes with cytoplasm replaced with phosphate buffer solution) were characterized, with comparable impedance data at 1.72 MHz indicating size comparability and significant differences at 15 MHz suggesting cytoplasm conductivity differences. (Reproduced with permission from Gawad et al. 2001). (c) The schematic of the microfluidic impedance flow cytometry where the parallel overlap microelectrodes were used to replace the previously reported coplanar microelectrodes. (Reproduced with permission from Cheung et al. 2005). (d) Opacity spectrum (fref = 602 kHz) of red blood cells and fixed red blood cells with increasing glutaraldehyde concentrations. At the high frequency domain, opacities were noticed to increase with the increase of glutaraldehyde concentrations since glutaraldehydes crosslink macromolecules within cytoplasm, leading to a decrease in cytoplasm conductivity. (Reproduced with permission from Cheung et al. 2005)

2017), stem cell differentiation (Zhao et al. 2016), and classification of white blood cells (Wang et al. 2017) (see Fig. 4b). However, the aforementioned approach still suffered from limited throughputs by only reporting electrical data from hundreds of single cells due to limitations in geometrical structures of the constriction channels. In order to tackle this technical challenge, the same group presented a new microfluidic impedance cytometry, containing a crossing constriction channel, which enabled the characterization of intrinsic cellular electrical markers from 100,000 cells at a rate larger than 100 cells/s (Zhao et al. 2018). As shown in Fig. 4c, single cells are aspirated continuously

35

Single-Cell Impedance Flow Cytometry

911

Fig. 4 (a) The schematic of the microfluidic impedance flow cytometry based on constriction channels for continuous characterization of specific membrane capacitance and cytoplasm conductivity of single cells based on an equivalent circuit model. (Reproduced with permission from Zhao et al. 2013a). (b) For paired high- and low-metastatic carcinoma strains 95D and 95C

912

H. Liang et al.

through the major constriction channel, effectively sealing the side constriction channel and generating changes in impedance values, which are further translated to specific membrane capacitance and cytoplasm conductivity without the requirement of image processing. The approach was demonstrated to classify different tumor cell lines and locate the effects of epithelial-mesenchymal transitions on tumor cells based on cellular electrical properties (see Fig. 4d).

Conclusion and Outlook In this review, major research progresses in the field of single-cell electrical property characterization have been covered. Although conventional techniques such as patch clamping and electrorotation are capable of characterizing intrinsic cellular parameters, they have limited throughputs and cannot collect data from large numbers of cells. The majority of reported microfluidic impedance flow cytometry can collect cellular electrical properties in a high-throughput manner, which, however, are only capable of reporting preliminary electrical properties (e.g., impedance values at several specific frequencies). Although these parameters can indicate membrane capacitance and cytoplasm resistance to an extent, they are dependent on external experimental conditions (e.g., channel geometries and electrode dimensions) and cannot be effectively used to evaluate cellular status and classify cell types. The recently reported microfluidic impedance cytometry based on constriction channels demonstrated the capability of collecting intrinsic electrical markers of single cells in a high-throughput manner, which, however, cannot provide a comprehensive evaluation of cells electrically. For instance, intrinsic bioelectrical markers such as transmembrane potentials and membrane charges still cannot be collected in a high-throughput manner. Thus, further opportunities in technical innovations may lie in the developments of microfluidic impedance flow cytometry enabling high-throughput characterization of multiple intrinsic cellular electrical markers to conduct comprehensive evaluations of single cells. Acknowledgment The authors would like to acknowledge financial supports from the National Natural Science Foundation of China (Grant No. 61431019, 61671430), Chinese Academy of Sciences Key Project Targeting Cutting-Edge Scientific Problems (QYZDB-SSW-JSC011), Instrument Development Program, Youth Innovation Promotion Association and Interdisciplinary Innovation Team of Chinese Academy of Sciences, and Instrument Development Program of Beijing Municipal Science and Technology Commission (Z181100009518001). ä Fig. 4 (continued) cells, significant differences in both specific membrane capacitance and cytoplasm conductivity were observed. (Reproduced with permission from Zhao et al. 2014). (c) The schematic of the microfluidic impedance flow cytometry based on crossing constriction channels for high-throughput characterization of specific membrane capacitance and cytoplasm conductivity of single cells based on an equivalent circuit model. (Reproduced with permission from Zhao et al. 2018). (d) Scatter plots of specific membrane capacitance versus cytoplasm conductivity of A549 (ncell = ~ 100,000) and A549 EMT (ncell = ~ 230,000) cells with significant differences in specific membrane capacitance and cytplasm conductivity. (Reproduced with permission from Zhao et al. 2018)

35

Single-Cell Impedance Flow Cytometry

913

References Adams TNG, Turner PA, Janorkar AV, Zhao F, Minerick AR (2014) Characterizing the dielectric properties of human mesenchymal stem cells and the effects of charged elastin-like polypeptide copolymer treatment. Biomicrofluidics 8(5):054109 Archer S, Morgan H, Rixon FJ (1999) Electrorotation studies of baby hamster kidney fibroblasts infected with herpes simplex virus type 1. Biophys J 76(5):2833–2842 Arnold WM, Zimmermann U (1982) Rotating-field-induced rotation and measurement of the membrane capacitance of single mesophyll cells of Avena sativa. Zeitschrift Fur Naturforschung C J Biosci 37(10):908–915 Arnold WM, Zimmermann U (1988) Electro-rotation: development of a technique for dielectric measurements on individual cells and particles. J Electrost 21(2–3):151–191 Arnold WM, Wendt B, Zimmermann U, Korenstein R (1985) Rotation of a single swollen thylakoid vesicle in a rotating electric field. Electrical properties of the photosynthetic membrane and their modification by ionophores, lipophilic ions and pH. Biochim Biophys Acta Biomembr 813(1):117–131 Arnold WM et al (1987) Electrorotation of mouse oocytes – single-cell measurements of zone-intact and zone-free cells and of the isolated zona-pellucida. Biochim Biophys Acta 905(2):454–464 Asami K, Yamaguchi T (1993) Dielectric spectroscopy of plant protoplasts. Biophys J 63(6):1493–1499 Bahrieh G, Erdem M, Ozgur E, Gunduz U, Kulah H (2014) Assessment of effects of multi drug resistance on dielectric properties of K562 leukemic cells using electrorotation. RSC Adv 4(85):44879–44887 Becker FF, Wang XB, Huang Y, Pethig R, Vykoukal J, Gascoyne PR (1995) Separation of human breast cancer cells from blood by differential dielectric affinity. Proc Natl Acad Sci U S A 92(3):860–864 Berardi V, Aiello C, Bonincontro A, Risuleo G (2009) Alterations of the plasma membrane caused by murine polyomavirus proliferation: an electrorotation study. J Membr Biol 229(1):19 Broche LM, Labeed FH, Hughes MP (2005) Extraction of dielectric properties of multiple populations from dielectrophoretic collection spectrum data. Phys Med Biol 50(10):2267–2274 Broche LM, Bhadal N, Lewis MP, Porter S, Hughes MP, Labeed FH (2007) Early detection of oral cancer – is dielectrophoresis the answer. Oral Oncol 43(2):199–203 Castellarnau M, Errachid A, Madrid C, Juarez A, Samitier J (2006) Dielectrophoresis as a tool to characterize and differentiate isogenic mutants of Escherichia coli. Biophys J 91(10):3937–3945 Chen P, Gillis KD (2000) The noise of membrane capacitance measurements in the whole-cell recording configuration. Biophys J 79(4):2162–2170 Chen J, Xue C, Zhao Y, Chen D, Wu MH, Wang J (2015) Microfluidic impedance flow cytometry enabling high-throughput single-cell electrical property characterization. Int J Mol Sci 16(5):9804–9830 Cheung KC (2010) Microfluidic impedance-based flow cytometry. Cytometry A 77(7):648–666 Cheung K, Gawad S, Renaud P (2005) Impedance spectroscopy flow cytometry: on-chip label-free cell differentiation. Cytometry A 65A(2):124–132 Chin S, Hughes MP, Coley HM, Labeed FH (2006) Rapid assessment of early biophysical changes in K562 cells during apoptosis determined using dielectrophoresis. Int J Nanomedicine 1(3):333–337 Chiu T-K et al (2017) A low-sample-loss microfluidic system for the quantification of sizeindependent cellular electrical property – its demonstration for the identification and characterization of circulating tumour cells (CTCs). Sensors Actuators B Chem 246(Suppl C):29–37 Cole KS (1928a) Electric impedance of suspensions of spheres. J Gen Physiol 12(1):29–36 Cole KS (1928b) Electric impedance of suspensions of arbacia eggs. J Gen Physiol 12(1):37–54 Cole KS (1932) Electric phase angle of cell membranes. J Gen Physiol 15(6):641–649 Cole KS, Curtis HJ (1938) Electric impedance of single marine eggs. J Gen Physiol 21(5):591 Coley HM, Labeed FH, Thomas H, Hughes MP (2007) Biophysical characterization of MDR breast cancer cell lines reveals the cytoplasm is critical in determining drug sensitivity. Biochim Biophys Acta 1770(4):601–608

914

H. Liang et al.

Cristofanilli M, De GG, Zhang L, Hung MC, Gascoyne PR, Hortobagyi GN (2002) Automated electrorotation to reveal dielectric variations related to HER-2/neu overexpression in MCF-7 sublines. Clin Cancer Res 8(2):615–619 Curtis HJ, Cole KS (1937) Transverse electric impedance of nitella. J Gen Physiol 21(2):189–201 Dalton C, Goater AD, Pethig R, Smith HV (2001) Viability of giardia intestinalis cysts and viability and sporulation state of cyclospora cayetanensis oocysts determined by electrorotation. Appl Environ Microbiol 67(2):586–590 De Gasperis G, Wang XB, Yang J, Becker FF, Gascoyne PRC (1998) Automated electrorotation: dielectric characterization of living cells by real-time motion estimation. Meas Sci Technol 9(3):518–529 Debus K, Hartmann J, Kilic G, Lindau M (1996) Influence of conductance changes on patch clamp capacitance measurements using a lock-in amplifier and limitations of the phase tracking technique. Biophys J 69(6):2808–2822 Deng Y-L, Kuo M-Y, Juang Y-J (2014) Development of flow through dielectrophoresis microfluidic chips for biofuel production: sorting and detection of microalgae with different lipid contents. Biomicrofluidics 8(6):064120 Donath E, Egger M, Pastushenko VP (1990) Dielectric behavior of the anion-exchange protein of human red blood cells: theoretical analysis and comparison to electrorotation data. J Electroanal Chem Interfacial Electrochem 23(3):337–360 Donnelly DF (1994) A novel method for rapid measurement of membrane resistance, capacitance, and access resistance. Biophys J 66(1):873–877 Du E, Ha S, Diez-Silva M, Dao M, Suresh S, Chandrakasan AP (2013) Electric impedance microflow cytometry for characterization of cell disease states. Lab Chip 13(19):3903–3909 Duncan L, Shelmerdine H, Hughes MP, Coley HM, Hubner Y, Labeed FH (2008) Dielectrophoretic analysis of changes in cytoplasmic ion levels due to ion channel blocker action reveals underlying differences between drug-sensitive and multidrug-resistant leukaemic cells. Phys Med Biol 53(2):N1–N7 Egger M, Donath E, Ziemer S, Glaser R (1986) Electrorotation – a new method for investigating membrane events during thrombocyte activation. Influence of drugs and osmotic pressure. Biochim Biophys Acta 861(1):122 Egger M, Donath E, Spangenberg P, Bimmler M, Glaser R, Till U (1988) Human platelet electrorotation change induced by activation: inducer specificity and correlation to serotonin release. Biochim Biophys Acta 972(3):265–276 Fernandez JM, Neher E, Gomperts BD (1984) Capacitance measurements reveal stepwise fusion events in degranulating mast cells. Nature 312(5993):453–455 Franz D, Olsen HL, Klink O, Gimsa J (2017) Automated and manual patch clamp data of human induced pluripotent stem cell-derived dopaminergic neurons. Sci Data 4:170056 Fricke H (1924a) A mathematical treatment of the electrical conductivity of colloids and cell suspensions. J Gen Physiol 6(4):375–384 Fricke H (1924b) A mathematical treatment of the electric conductivity and capacity of disperse systems I. the electric conductivity of a suspension of homogeneous spheroids. Phys Rev 24(5):575–587 Fricke H (1925a) A mathematical treatment of the electric conductivity and capacity of disperse systems II. The capacity of a suspension of conducting spheroids surrounded by a non-conducting membrane for a current of low frequency. Phys Rev 26(5):678–681 Fricke H (1925b) The electric capacity of suspensions of red corpuscles of a dog. Phys Rev 26(5):682–687 Fricke H (1925c) The electric capacity of suspensions with special reference to blood. J Gen Physiol 9(2):137–152 Fricke H, Morse S (1925) The electric resistance and capacity of blood for frequencies between 800 and 4(1/2) million cycles. J Gen Physiol 9(2):153–167 Fuhr G, Kuzmin PI (1986) Behavior of cells in rotating electric fields with account to surface charges and cell structures. Biophys J 50(50):789–795

35

Single-Cell Impedance Flow Cytometry

915

Fuhr G, Glaser R, Hagedorn R (1986) Rotation of dielectrics in a rotating electric high-frequency field – model experiments and theoretical explanation of the rotation effect of living cells. Biophys J 49(2):395–402 Gawad S, Schild L, Renaud P (2001) Micromachined impedance spectroscopy flow cytometer for cell analysis and particle sizing. Lab Chip 1(1):76–82 Gawad S, Cheung K, Seger U, Bertsch A, Renaud P (2004) Dielectric spectroscopy in a micromachined flow cytometer: theoretical and practical considerations. Lab Chip 4(3):241–251 Geier BM, Wendt B, Arnold WM, Zimmermann U (1987) The effect of mercuric-salts on the electrorotation of yeast-cells and comparison with a theoretical-model. Biochim Biophys Acta 900(1):45–55 Gentet LJ, Stuart GJ, Clements JD (2000) Direct measurement of specific membrane capacitance in neurons. Biophys J 79(1):314–320 Georgiewa R, Donath E, Gimsa J, Löwe U, Glaser R (1989) Ac-field-induced KCl leakage from human red cells at low ionic strengths: implications for electrorotation measurements. Bioelectrochem Bioenerg 22(3):255–270 Gimsa J (2001) A comprehensive approach to electro-orientation, electrodeformation, dielectrophoresis, and electrorotation of ellipsoidal particles and biological cells. Bioelectrochemistry 54(1):23–31 Gimsa J, Marszalek P, Loewe U, Tsong TY (1991) Dielectrophoresis and electrorotation of neurospora slime and murine myeloma cells. Biophys J 60(4):749–760 Gimsa J, Müller T, Schnelle T, Fuhr G (1996) Dielectric spectroscopy of single human erythrocytes at physiological ionic strength: dispersion of the cytoplasm. Biophys J 71(1):495 Goater AD, Pethig R (1998) Electrorotation and dielectrophoresis. Parasitology 117(Suppl):S177–S189 Golowasch J (2009) Membrane capacitance measurements revisited: dependence of capacitance value on measurement method in nonisopotential neurons. J Neurophysiol 102(4):2161–2175 Graf J, Rupnik M, Zupancic G, Zorec R (1995) Osmotic swelling of hepatocytes increases membrane conductance but not membrane capacitance. Biophys J 68(4):1359–1363 Graham KA et al (2015) A dielectrophoretic method of discrimination between normal oral epithelium, and oral and oropharyngeal cancer in a clinical setting. Analyst 140 (15):5198–5204 Guofeng Q, Wei W, Wei D, Fan Z, Sinclair AJ, Chatwin CR (2012) Bioimpedance analysis for the characterization of breast cancer cells in suspension. IEEE Trans Biomed Eng 59(8):2321–2329 Han X, van Berkel C, Gwyer J, Capretto L, Morgan H (2012) Microfluidic lysis of human blood for leukocyte analysis using single cell impedance cytometry. Anal Chem 84(2):1070–1075 Henslee EA et al (2016) Accurate quantification of apoptosis progression and toxicity using a dielectrophoretic approach. Analyst 141(23):6408–6415 Hoffman RA, Britt WB (1979) Flow-system measurement of cell impedance properties. J Histochem Cytochem 27(1):234 Hoffman RA, Johnson TS, Britt WB (1981) Flow cytometric electronic direct current volume and radiofrequency impedance measurements of single cells and particles. Cytometry 1(6):377–384 Holmes D, Morgan H (2010) Single cell impedance cytometry for identification and counting of CD4 T-cells in human blood using impedance labels. Anal Chem 82(4):1455–1461 Holmes D et al (2009) Leukocyte analysis and differentiation using high speed microfluidic single cell impedance cytometry. Lab Chip 9(20):2881–2889 Holzapfel C, Vienken J, Zimmermann U (1982) Rotation of cells in an alternating electric field theory and experimental proof. J Membr Biol 67(1):13–26 Holzel R (1997) Electrorotation of single yeast cells at frequencies between 100 Hz and 1.6 GHz. Biophys J 73(2):1103–1109 Holzel R (1999) Non-invasive determination of bacterial single cell properties by electrorotation. BBA-Mol Cell Res 1450(1):53–60 Hu X, Arnold WM, Zimmermann U (1990) Alterations in the electrical properties of T and B lymphocyte membranes induced by mitogenic stimulation. Activation monitored by electrorotation of single cells. Biochim Biophys Acta Biomembr 1021(2):191–200

916

H. Liang et al.

Huang Y, Wang XB, Tame JA, Pethig R (1993) Electrokinetic behaviour of colloidal particles in travelling electric fields: studies using yeast cells. J Phys D Appl Phys 26(9):1528 Hughes MP (1998a) Computer-aided analysis of conditions for optimizing practical electrorotation. Phys Med Biol 43(12):3639–3648 Hughes MP (1998b) Computer-aided analyses of electric fields used in electrorotation studies. J Phys D Appl Phys 27(7):1564 Hughes MP, Wang XB, Becker FF, Gascoyne PRC, Pethig R (1994) Computer-aided analyses of electric-fields used in electrorotation studies. J Phys D Appl Phys 27(7):1564–1570 Ismail A, Hughes MP, Mulhall HJ, Oreffo RO, Labeed FH (2015) Characterization of human skeletal stem and bone cell populations using dielectrophoresis. J Tissue Eng Regen Med 9(2):162–168 Johnson SL, Thomas MV, Kros CJ (2002) Membrane capacitance measurement using patch clamp with integrated self-balancing lock-in amplifier. Pflugers Arch - Eur J Physiol 443 (4):653–663 Joshi C, Fernandez JM (1988) Capacitance measurements. An analysis of the phase detector technique used to study exocytosis and endocytosis. Biophys J 53(6):885–892 Kakutani T, Shibatani S, Senda M (1993) Electrorotation of barley mesophyll protoplasts. Bioelectrochem Bioenerg 31(1):85–97 Kaler KV, Jones TB (1990) Dielectrophoretic spectra of single cells determined by feedbackcontrolled levitation. Biophys J 57(2):173–182 Kiesel M et al (1999) Swelling-activated pathways in human T-lymphocytes studied by cell volumetry and electrorotation. Biophys J 76(5):2833–2842 Kodandaramaiah SB, Franzesi GT, Chow BY, Boyden ES, Forest CR (2012) Automated whole-cell patch-clamp electrophysiology of neurons in vivo. Nat Methods 9(6):585–587 Kriegmaier M, Zimmermann M, Wolf K, Zimmermann U, Sukhorukov VL (2001) Dielectric spectroscopy of schizosaccharomyces pombe using electrorotation and electroorientation. Biochim Biophys Acta Gen Subj 1568(2):135–146 Kürschner M, Nielsen K, von Langen JR, Schenk WA, Zimmermann U, Sukhorukov VL (2000) Effect of fluorine substitution on the interaction of lipophilic ions with the plasma membrane of mammalian cells. Biophys J 79(3):1490 Labeed FH, Coley HM, Thomas H, Hughes MP (2003) Assessment of multidrug resistance reversal using dielectrophoresis and flow cytometry. Biophys J 85(3):2028–2034 Labeed FH, Coley HM, Hughes MP (2006) Differences in the biophysical properties of membrane and cytoplasm of apoptotic cells revealed using dielectrophoresis. Biochim Biophys Acta 1760 (6):922–929 Labeed FH et al (2011) Biophysical characteristics reveal neural stem cell differentiation potential. PLoS One 6(9):e25458 Lannin T et al (2016) Automated electrorotation shows electrokinetic separation of pancreatic cancer cells is robust to acquired chemotherapy resistance, serum starvation, and EMT. Biomicrofluidics 10(6):064109 Liang X, Graham KA, Johannessen AC, Costea DE, Labeed FH (2014) Human oral cancer cells with increasing tumorigenic abilities exhibit higher effective membrane capacitance. Integr Biol 6(5):545–554 Liang W, Zhang K, Yang X, Liu L, Yu H, Zhang W (2015) Distinctive translational and selfrotational motion of lymphoma cells in an optically induced non-rotational alternating current electric field. Biomicrofluidics 9(1):014121 Liang W, Zhao Y, Liu L, Wang Y, Li WJ, Lee GB (2017) Determination of cell membrane capacitance and conductance via optically induced electrokinetics. Biophys J 113(7):1531–1539 Lindau M, Neher E (1988) Patch-clamp technique for time-resolved capacitance measurements in single cells. Pflugers Arch - Eur J Physiol 411(2):137–146 Lovelace RVE, Stout DG, Steponkus PL (1984) Protoplast rotation in a rotating electric field: the influence of cold acclimation. J Membr Biol 82(2):157–166

35

Single-Cell Impedance Flow Cytometry

917

Meissner R, Joris P, Eker B, Bertsch A, Renaud P (2012) A microfluidic-based frequency-multiplexing impedance sensor (FMIS). Lab Chip 12(15):2712–2718 Memmel S et al (2014) Cell surface area and membrane folding in glioblastoma cell lines differing in PTEN and p53 status. PLoS One 9(1):e87052 Mernier G, Piacentini N, Tornay R, Buffi N, Renaud P (2011) Cell viability assessment by flow cytometry using yeast as cell model. Sensors Actuators B Chem 154(2):160–163 Mernier G, Hasenkamp W, Piacentini N, Renaud P (2012) Multiple-frequency impedance measurements in continuous flow for automated evaluation of yeast cell lysis. Sensors Actuators B Chem 170:2–6 Mietchen D, Schnelle T, Mller T, Hagedorn R, Fuhr G (2002) Automated dielectric single cell spectroscopy – temperature dependence of electrorotation. J Phys D Appl Phys 35(35):1258–1270 Morgan H, Sun T, Holmes D, Gawad S, Green NG (2007) Single cell dielectric spectroscopy. J Phys D Appl Phys 40(1):61–70 Mulhall HJ, Labeed FH, Kazmi B, Costea DE, Hughes MP, Lewis MP (2011) Cancer, pre-cancer and normal oral cells distinguished by dielectrophoresis. Anal Bioanal Chem 401(8):2455–2463 Mulhall HJ, Cardnell A, Hoettges KF, Labeed FH, Hughes MP (2015) Apoptosis progression studied using parallel dielectrophoresis electrophysiological analysis and flow cytometry. Integr Biol 7(11):1396–1401 Neef A, Heinemann C, Moser T (2007) Measurements of membrane patch capacitance using a software-based lock-in system. Pflugers Arch - Eur J Physiol 454(2):335–344 Neher E, Marty A (1982) Discrete changes of cell membrane capacitance observed under conditions of enhanced secretion in bovine adrenal chromaffin cells. Proc Natl Acad Sci 79(21):6712–6716 Novak P, Zahradnik I (2006) Q-method for high-resolution, whole-cell patch-clamp impedance measurements using square wave stimulation. Ann Biomed Eng 34(7):1201–1212 O’Shaughnessy TJ, Kim YI (1995) A computer-based system for the measurement of membrane capacitance to monitor exocytosis in secretory cells. J Neurosci Methods 57(1):1–8 Pethig R, Talary MS (2007) Dielectrophoretic detection of membrane morphology changes in Jurkat T-cells undergoing etoposide-induced apoptosis. IET Nanobiotechnol 1(1):2–9 Pilwat G, Zimmermann U (1983) Rotation of a single cell in a discontinuous rotating electric field. Bioelectrochem Bioenerg 10(1):155–162 Reuss OR et al (2002) Interaction of fluorinated lipophilic ions with the plasma membrane of mammalian cells studied by electrorotation and dielectrophoresis. J Electrost 56(4):419–434 Rituper B (2013) High-resolution membrane capacitance measurements for the study of exocytosis and endocytosis. Nat Protoc 8(6):1169–1183 Rohlicek V, Schmid A (1994) Dual-frequency method for synchronous measurement of cell capacitance, membrane conductance and access resistance on single cells. Pflugers Arch - Eur J Physiol 428(1):30–38 Saboktakin Rizi B, Braasch K, Salimi E, Butler M, Bridges GE, Thomson DJ (2014) Monitoring the dielectric response of single cells following mitochondrial adenosine triphosphate synthase inhibition by oligomycin using a dielectrophoretic cytometer. Biomicrofluidics 8(6):064114 Sakaba T, Hazama A, Maruyama Y (2012) Patch-clamp capacitance measurements. In: Patch clamp techniques: from beginning to advanced protocols. Springer, Tokyo, pp 277–286 Sancho M, Martinez G, Munoz S, Sebastian JL, Pethig R (2010) Interaction between cells in dielectrophoresis and electrorotation experiments. Biomicrofluidics 4(2):022802 Schwan HP (1957) Electrical properties of tissue and cell suspensions. Adv Biol Med Phys 5:147–209 Schwan HP (1968) Electrode polarization impedance and measurements in biological materials. Ann N Y Acad Sci 148(1):191–209 Shaker M, Colella L, Caselli F, Bisegna P, Renaud P (2014) An impedance-based flow microcytometer for single cell morphology discrimination. Lab Chip 2014(14):2548–2555 Song H et al (2013) A microfluidic impedance flow cytometer for identification of differentiation state of stem cells. Lab Chip 13(12):2300–2310

918

H. Liang et al.

Sudsiri J, Wachner D, Gimsa J (2007) On the temperature dependence of the dielectric membrane properties of human red blood cells. Bioelectrochemistry 70(1):134–140 Sukhorukov VL, Zimmermann U (1996) Electrorotation of erythrocytes treated with dipicrylamine: mobile charges within the membrane show their “signature” in rotational spectra. J Membr Biol 153(2):161–169 Sukhorukov VL, Benkert R, Obermeyer G, Bentrup F, Zimmermann U (1998) Electrorotation of isolated generative and vegetative cells, and of intact pollen grains of Lilium longiflorum. J Membr Biol 161(1):21 Taruvai Kalyana Kumar R, Liu S, Minna JD, Prasad S (2016) Monitoring drug induced apoptosis and treatment sensitivity in non-small cell lung carcinoma using dielectrophoresis. Biochim Biophys Acta 1860(9):1877–1883 Vahey MD, Voldman J (2009) High-throughput cell and particle characterization using isodielectric separation. Anal Chem 81(7):2446–2455 Vaillier C, Honegger T, Kermarrec F, Gidrol X, Peyrade D (2014) Comprehensive analysis of human cells motion under an irrotational ac electric field in an electro-microfluidic chip. PLoS One 9(4):e95231 Valero A, Braschler T, Renaud P (2010) A unified approach to dielectric single cell analysis: impedance and dielectrophoretic force spectroscopy. Lab Chip 10(17):2216–2225 Velugotla S et al (2012) Dielectrophoresis based discrimination of human embryonic stem cells from differentiating derivatives. Biomicrofluidics 6(4):44113 Vykoukal DM, Gascoyne PR, Vykoukal J (2009) Dielectric characterization of complete mononuclear and polymorphonuclear blood cell subpopulations for label-free discrimination. Integr Biol 1(7):477–484 Wang XB, Huang Y, Gascoyne PR, Becker FF, Holzel R, Pethig R (1994) Changes in Friend murine erythroleukaemia cell membranes during induced differentiation determined by electrorotation. Biochim Biophys Acta 1193(2):330–344 Wang J, Sukhorukov VL, Djuzenova CS, Zimmermann U, Müller T, Fuhr G (1997) Electrorotational spectra of protoplasts generated from the giant marine alga Valonia utricularis. Protoplasma 196(3–4):123–134 Wang X, Becker FF, Gascoyne PR (2002) Membrane dielectric changes indicate induced apoptosis in HL-60 cells more sensitively than surface phosphatidylserine expression or DNA fragmentation. Biochim Biophys Acta 1564(2):412–420 Wang X, Becker FF, Gascoyne PR (2010) The fractal dimension of cell membrane correlates with its capacitance: a new fractal single-shell model. Chaos 20(4):043133 Wang K et al (2017) Membrane capacitance of thousands of single white blood cells. J R Soc Interface 14(137). https://doi.org/10.1098/rsif.2017.0717 Watkins NN et al (2013) Microfluidic CD4+ and CD8+ T lymphocyte counters for point-of-care HIV diagnostics using whole blood. Sci Transl Med 5(214):214ra170 White WE, Hooper SL (2013) Contamination of current-clamp measurement of neuron capacitance by voltage-dependent phenomena. J Neurophysiol 110(1):257 Xu Y, Xie X, Duan Y, Wang L, Cheng Z, Cheng J (2016) A review of impedance measurements of whole cells. Biosens Bioelectron 77:824–836 Yang J, Huang Y, Wang X, Wang XB, Becker FF, Gascoyne PRC (1999) Dielectric properties of human leukocyte subpopulations determined by electrorotation as a cell separation criterion. Biophys J 76(6):3307–3314 Zhang H, Qu A, Luo J, Luo J (2010) Error analysis of Cm measurement under the whole-cell patchclamp recording. J Neurosci Methods 185(2):307–314 Zhao Y et al (2013a) A microfluidic system enabling continuous characterization of specific membrane capacitance and cytoplasm conductivity of single cells in suspension. Biosens Bioelectron 43C:304–307 Zhao Y et al (2013b) A microfluidic system for cell type classification based on cellular sizeindependent electrical properties. Lab Chip 13(12):2272–2277

35

Single-Cell Impedance Flow Cytometry

919

Zhao Y et al (2014) Tumor cell characterization and classification based on cellular specific membrane capacitance and cytoplasm conductivity. Biosens Bioelectron 57:245–253 Zhao Y et al (2016) Electrical property characterization of neural stem cells in differentiation. PLoS One 11(6):e0158044 Zhao Y et al (2018) Development of microfluidic impedance cytometry enabling the quantification of specific membrane capacitance and cytoplasm conductivity from 100,000 single cells. Biosens Bioelectron 111:138–143 Zhou XF, Burt JPH, Pethig R (1998) Automatic cell electrorotation measurements: studies of the biological effects of low-frequency magnetic fields and of heat shock. Phys Med Biol 43(5):1075–1090 Zhou T, Ming Y, Perry SF, Tatic-Lucic S (2016) Estimation of the physical properties of neurons and glial cells using dielectrophoresis crossover frequency. J Biol Phys 42(4):571–586 Ziervogel H, Glaser R, Schadow D, Heymann S (1986) Electrorotation of lymphocytes–the influence of membrane events and nucleus. Biosci Rep 6(11):973–982 Zimmermann U, Vienken J, Pilwat G (1981) Rotation of cells in an alternating electric field: the occurrence of a resonance frequency. Zeitschrift Fur Naturforschung C J Biosci 36(1–2):173–177

Cytometry of Single Cell in Biology and Medicine

36

Shunbo Li

Contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mechanisms of Cytometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of Cytometry in Microbial Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Total Bacterial Cell Count . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bacterial Viability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Specific Microbial Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Multifunction Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of Cytometry in Rare Cell Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phenotype Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Genotype Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of Cytometry in Medicine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fluorescent Staining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Drug Discovery in Immunology and Receptor Pharmacology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

922 923 926 927 927 928 929 930 930 932 934 935 936 936 937 938 939 939

Abstract

It has been a very long time for human to understand the physiological processes related to health, disease, and death. Cell is the basic elementary building block of life. However, no two cells are exactly the same. In order to understand the heterogeneity and complexity of the biological system, statistical analysis has to be conducted on the single-cell level, and the corresponding high-precision instruments have to be used for investigating on the cellular level. Cytometry S. Li (*) College of Optoelectronic Engineering, Chongqing University, Chongqing, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd, part of Springer Nature 2022 T. S. Santra, F.-G. Tseng (eds.), Handbook of Single-Cell Technologies, https://doi.org/10.1007/978-981-10-8953-4_24

921

922

S. Li

including flow cytometry and image cytometry is an advanced technology in sensitive single-cell analysis, and it has the capabilities in detection with high sensitivity, high throughput, and high content. Sorting, sample handling, and even sequencing could also be integrated for further analysis for modern cytometry. Recently, microfluidic-based cytometry with smaller size, higher throughput, and multifunctions starts to find its way in single-cell analysis. Because of these unique advantages, single-cell cytometry has wide applications in basic researches of biology process and drug discovery to understand the cell heterogeneity and complexity. This chapter will give a brief introduction of single-cell cytometry and its applications in biology and medicine. Keywords

Cytometry · Single-cell analysis · High throughput · Multi-parameters · Microfluidics

Introduction Cytometry, invented in the 1950s, is a process in which individual cells or particles are made to pass in a single stream, and they are detected by sensors that could measure the physical and biochemical characteristics. It becomes more and more popular in biological analysis due to its high-sensitivity, high-throughput, and highcontent capabilities. It provides a quantitative platform for the measurement of biological particles with scattering and fluorescence signals. For those that have no fluorescence or are difficult to stain, the newly developed cytometry system could also perform analysis according to the electrical impedance differences. The impedance cytometry has been introduced in the previous chapter. Cytometry has the ability to detect, analyze, and handle biological particles on single-cell level. Thus, it is also called single-cell cytometry to highlight the single-cell ability. The newest developments are mainly multifunctionalization and miniaturization to integrate more channels and detection parameters in the system with smaller scale. Heterogeneity and complexity, existing among biological cells, are very common phenomena in biology. They originate from genome mutations, progression through the cell cycle, and changes in the local environment. The understanding of the heterogeneity of an interesting cell population is significant for human’s full understanding of any given biological process. In addition, owning to the development of new technology, it becomes possible to investigate biological process with resolution down to the single-cell level to deepen our understanding of the underlying disease to better conduct treatments and identify medicine targets for diagnostic purposes. In order to measure and understand the biological heterogeneity, statistical analysis is required, and single-cell cytometry is the ideal platform for this. For example, the adaption of bacteria to environment changes could be analyzed by cytometry in either phenotype or genotype way by measuring their viability or conducting PCR afterward. In biological studies, statistical measurements are extremely important

36

Cytometry of Single Cell in Biology and Medicine

923

due to the diversity of biological entities, coming from the inability to control for all variables in biological processes and the heterogeneity nature of cell populations. It is still true for typical biological cells having a very low frequency in the total population. A simple example will make this scenario clearer. If the frequency of an interested cell is 0.1% in the whole population, it is hardly to obtain the rare cell using traditional plating or sorting method. Don’t even consider the real situation of circulating tumor cells in patient’s blood, which has a much lower frequency. With the help of high-throughput cytometry, one could easily analyze a sample with more than 100,000 cells in minutes. Therefore, cytometry is an ideal platform to study the heterogeneity and complexity of biological cells. Cytometry allows assessment of population heterogeneity on the single-cell level, and its data could reflect different aspects of properties for biological cells such as the size and shape, membrane integrity, membrane potential, cell metabolism, and content of DNA. Therefore, lots of information, such as physiology and structure, can be obtained from individual cell by staining them with specific fluorescent dyes. Usually, combinations of dyes with different colors are used to generate multiparameter data of single cells from multichannel flow cytometry. The main advantages of cytometry for microbiology include high throughput (>1000 cells), high sensitivity (with femto-mole sensitivity), rapid assay time (in minutes), and high content (more than ten channels for multi-parameter analysis). Until now, cytometry has wide applications in lots of areas that closely related to our lives including environmental analysis (De Roy et al. 2012; Herrero and Diaz 2015), food quality test (Gunasekera et al. 2003; Kennedy et al. 2011), fermentation industry (Bühligen et al. 2014), medical diagnostics (Aebisher et al. 2017), pharmaceutical, and industry products (Díaz et al. 2010). This chapter will discuss the important applications of single-cell cytometry in biological and biomedicine perspectives. The applications are mainly focused on the study of microbial, rare cells and drug discovery.

Mechanisms of Cytometry Cytometry is regarded as a mature technology in the past decades, which has dramatic advances in sensitivity and throughput in analysis, counting and sorting biological cells. These unique capabilities ensure its wide applications in analyzing the high-content, diverse, and complex biology systems in an extreme highthroughput manner. Since the invention of cytometry, many companies released their commercial products, for instance, Beckman Coulter, Becton Dickinson Biosciences, Partec, Accuri Cytometers, Thermo Fisher Scientific, etc. However, the basic principle of cytometry is similar – distinguishing cells or particles according to the differences in light scattering, fluorescence, or electrical impedance. Cytometry is usually divided into flow cytometry and image cytometry. The working mechanism of fluorescent flow cytometry is illustrated in Fig. 1. Cell suspensions are introduced in a microchannel, and then they are aligned in a single stream by hydrodynamic focusing with sheath flow after entering the first junction. When the single cell passes through the detection zone, the laser beam will

924

S. Li

Fig. 1 Schematics of flow cytometry

hit on it, and the scattered light could be detected by the optical systems, converted to digital signals and processed by the connected computer. Meanwhile, the laser could also excite the fluorescent light of cell and be detected by detectors. The scattered light (with the same wavelength) and emitted fluorescent light (usually with shorter wavelength) could give the information of cell size, cell shape, internal complexity, and membrane integrity according to the probes or dyes that are used to stain. Therefore, both scattering and fluorescent light can, in a general way, discriminate cells into discrete subpopulations. With integration of lasers with different wavelength, multi-parameters could be obtained, which is also called multichannel flow cytometry. These acquired multi-parameters give the high-content information of cells, and it is then called high-content flow cytometry, which is analogous to fluorescent microscope with multiple fluorescence emission wavelengths for imaging. The latest flow cytometers have throughput of thousands of cells per second and could measure tens of parameters on individual ones, allowing the detection of multiple probes using different fluorescent dyes without overlapping. Therefore, flow cytometry can provide highly quantitative information of individual cells in a high-throughput and high-content manner. After the detection and high-content analysis of cells, sorting based on the pre-set criteria could be achieved in the

36

Cytometry of Single Cell in Biology and Medicine

925

downstream of microchannel. Until now, several methods have been applied to separate cells, for instance, electrostatics, magnetics, acoustics, dielectrophoresis, optical traps, as well as valves. Large numbers of data could be obtained in the single-cell flow cytometry, and separation is critical for further analysis. Image cytometry, similar to flow cytometry in structure, could be used to measure similar parameters as flow cytometry. The only difference is the ability to take images in two dimensions or three dimensions while the cell passes through. Image cytometry is generally composed of automated microscope and highperformance computer to take images and perform analysis, respectively. Inherently, image cytometry could detect more information of single cell and generate much more data than flow cytometry. Image cytometry attracts more and more attentions for its extension of cell information in the morphology and structure. Recent development of image cytometry is focused on imaging with high-speed and super-resolution, as well as analyzing based on machine learning or artificial intelligence. Most of the cytometry are based on light scattering and the fluorescent staining for multi-parameter analysis. However, some of them are hard to stain or not suitable to stain with fluorescent dyes. Therefore, the label-free cytometry, mainly based on the electrical properties, has to be developed. As commonly used label-free markers, cellular electrical properties including membrane capacitance and cytoplasm resistance are promising electrical indicators to distinguish cell types and evaluate the cellular status. The detailed introduction of impedance cytometry has been presented in the previous chapter. As introduced previously, cytometry works as a unique tool for the analysis of single cell with a wide range of applications including molecular assembly, highcontent analysis and imaging, multiplexing of biological targets, sorting, drug screening, and discovery. Most commercially available cytometers are bulky and expensive, which require high maintenance costs and well-trained person for operation. Thus, there is a high demand to develop a low-cost, miniaturized, and customized system that will certainly help to make this powerful research tool more accessible. Over the past decades, several microfluidic platforms have been designed and developed, which enable the single-cell analysis in a unique way that could not be achieved using large-scale equipment. Compared to the traditional bulky instruments, microfluidic-based flow cytometry system could integrate the sampling, cell culture, sorting, and enrichment, along with biochemical or genetic analysis of individual cells with high spatial and temporal resolutions. The microfluidic platforms also provide the ability to acquire multi-parametric and highcontent information of individual cells over time. In addition to the small size character, microfluidic system could operate even faster by integrating multichannels using sheathless focusing like acoustic or inertial to align cells other than the conventional sheath flow. Figure 2 shows the developed microfluidic flow cytometry with 384 microchannels and scanning laser detection system (Mckenna et al. 2011). This system was used to investigate protein localization in a yeast model for human protein misfolding diseases.

926

S. Li

Fig. 2 A schematic of microfluidic flow cytometry which uses 384 parallel microchannels and a laser scanning detection system to provide images of cells during high-content screening (Mckenna et al. 2011). (Reprinted with permission from Macmillan Publishers Ltd, copyright (2011))

Application of Cytometry in Microbial Study Microbes including viruses, bacteria, mycobacteria, fungi, protozoa, and some helminths could be found almost everywhere on Earth, even in extreme conditions such as strong acid, high temperature, and high pressure. They have wide applications in fermentation food production, energy production, medicine production, as well as waste treatment in water and soil. They have positive impact on human life in microbial symbiosis. On the other hand, there are lots of pathogenic microbes (e.g., Bacillus anthracis, E. coli (O157:H7), Salmonella typhi, Staphylococcus aureus, Mycoplasma fermentans, etc.) causing infectious diseases. Microbes are also essential tools in biology as model organisms. Because of the importance of microbiology to human health with positive and negative impacts, lots of methods have been developed to count the number of bacteria, identify types of them, and investigate the physical, chemical, and biological interventions. Traditionally, all microbial detection methods rely on cell replication in the culture media or agar plates. Generally, colonies that grow from single cells are the standard target for research. The method used is plating developed by Koch more than a century ago. However, it is only suitable for culturable bacteria which could grow and replicate in the artificial environment. The non-culturable bacteria cannot be counted accurately due to the lack of symbiotic partners and/or a suitable microenvironment. Therefore, representative enumeration, which could count pathogenic bacteria, can only be achieved by direct optical methods for non-culturable cells. The advance of microbiology study has come to direct single-cell analysis which further broadens our understanding of microbial populations and their heterogeneity and complexity. Cytometry is a powerful technique for single-cell analysis by

36

Cytometry of Single Cell in Biology and Medicine

927

measuring scattering, fluorescence, mechanical and electrical properties, and even the images simultaneously. By providing quantitative, high-sensitive, high-throughput, high-resolution, and multi-parameter analysis at the single-cell level, cytometry has gained increasing interests in microbiological research, water quality control, food safety monitoring, and clinical diagnosis. In biotechnology applications, the multicolor staining approach combined with cytometry offers important physiological information about process efficiency on the individual cell level, which is almost impossible to be obtained in any other way. This chapter will introduce the most common applications of cytometry in microbial study, mainly focusing on bacteria, including total bacterial number count, bacterial viability analysis, specific microbial detection and identification, and multi-parameter analysis.

Total Bacterial Cell Count In order to evaluate the biological contaminations in water and food, the total bacterial count without culture is a straightforward method. A good distinguish of the small size entities between bacterial cells and dust particles is the basic requirement for the accurate counting of bacteria using cytometry. Generally, precise evaluation of bacteria concentration can be achieved either by mixing the sample with standard fluorescent beads of known concentration or by measuring the volumetric flow rate of the sample stream. Hammes et al. developed a quantification method based on flow cytometry to precisely enumerate the total bacterial concentration in drinking water by using SYBR ® Green fluorescent dye to improve the detection sensitivity (Hammes et al. 2008). Comparing the developed cytometry method with traditional ones using the cultivation-based plate counting and the adenosine triphosphate measurement, they confirmed that cytometry was more efficient in measuring total bacterial concentration in terms of simplicity, speed, and sensitivity. SYBR dyes, SYTO dyes, SYTOX dyes, BactoView dyes, RedDot dyes, and PicoGreen thiazole orange and acridine orange are commonly used to stain the DNA of bacterial cells for enumeration and quantification. These dyes emit very weak fluorescent light freely in solution, but when they combined with double-strand DNA, hundreds to thousands fold of fluorescence enhancement can be achieved. Other fluorescent dyes such as SynaptoGreen/SynaptoRed dyes and CellBrite Fix dyes are cell membrane staining dyes used to discriminate bacteria from dust particles for bacterial counting.

Bacterial Viability Analysis Total bacterial cell count is a rough estimation of the bacteria contamination in environments since it can only distinguish bacteria from dusts. Characterizing and distinguishing different physiological states of bacteria at the single-cell level is

928

S. Li

more attractive in microbiology. Employing various fluorescent dyes to stain bacteria with different viability states depends on the selective permeability of the cell membrane. The viability analysis of bacteria is widely applied to test bacterial response to microenvironment and drugs, for instance, the tolerance of bacteria under severe conditions, the survival rates of pathogens in oligotrophic and antimicrobial environment, and the effects of toxic substances on microbial viability. Papadimitriou et al. applied in situ flow cytometry with viability assay capability employing carboxyfluorescein diacetate and propidium iodide to identify acid tolerance of Streptococcus macedonicus (S. macedonicus) (Papadimitriou et al. 2007). The single-cell analysis of S. macedonicus during induction of logarithmic-phase acid tolerance response revealed heterogeneity in the rate of tolerance acquisition within clonal populations. Mustapha et al. applied a flow cytometry assay to monitor the viability of Legionella pneumophila (L. pneumophila) after chlorine dioxide (ClO2) treatment (Mustapha et al. 2015). Different viability responses of cells were found when subjecting to ClO2 with different concentrations. Low concentrations of ClO2 (