Flow Cytometry Protocols (Methods in Molecular Biology, 2779) [5 ed.] 1071637371, 9781071637371

Citation preview

Methods in Molecular Biology 2779

Teresa S. Hawley Robert G. Hawley  Editors

Flow Cytometry Protocols Fifth Edition

METHODS

IN

MOLECULAR BIOLOGY

Series Editor John M. Walker School of Life and Medical Sciences University of Hertfordshire Hatfield, Hertfordshire, UK

For further volumes: http://www.springer.com/series/7651

For over 35 years, biological scientists have come to rely on the research protocols and methodologies in the critically acclaimed Methods in Molecular Biology series. The series was the first to introduce the step-by-step protocols approach that has become the standard in all biomedical protocol publishing. Each protocol is provided in readily-reproducible step-bystep fashion, opening with an introductory overview, a list of the materials and reagents needed to complete the experiment, and followed by a detailed procedure that is supported with a helpful notes section offering tips and tricks of the trade as well as troubleshooting advice. These hallmark features were introduced by series editor Dr. John Walker and constitute the key ingredient in each and every volume of the Methods in Molecular Biology series. Tested and trusted, comprehensive and reliable, all protocols from the series are indexed in PubMed.

Flow Cytometry Protocols Fifth Edition

Edited by

Teresa S. Hawley Bethesda, MD, USA

Robert G. Hawley Washington, DC, USA

Editors Teresa S. Hawley Bethesda, MD, USA

Robert G. Hawley Washington, DC, USA

ISSN 1064-3745 ISSN 1940-6029 (electronic) Methods in Molecular Biology ISBN 978-1-0716-3737-1 ISBN 978-1-0716-3738-8 (eBook) https://doi.org/10.1007/978-1-0716-3738-8 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024 This work is subject to copyright. All rights are solely and exclusively licensed 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 Humana imprint is published by the registered company Springer Science+Business Media, LLC, part of Springer Nature. The registered company address is: 1 New York Plaza, New York, NY 10004, U.S.A. Paper in this product is recyclable.

Dedication Dedicated to the memory of Howard Shapiro.

v

Preface The ability of cytometry to simultaneously measure multiple aspects of cellular mechanisms at the single cell level holds promise for a profound understanding of how perturbations to homeostasis are associated with human diseases. Over the years, cytometry has evolved to encompass polychromatic (conventional) flow cytometry, full-spectrum (spectral) flow cytometry, imaging flow cytometry, mass cytometry, and imaging mass cytometry. Cytometry has proved to be a powerful tool in the surveillance of emerging infectious diseases. The prospects and challenges of integrating cytometry instrumentation within highcontainment laboratories are reviewed in one chapter. Reproduction of the highly educational introductory chapter from the last edition, “Flow Cytometry: The Glass Is Half Full”, written by the late Howard Shapiro, is a tribute to the cytometry luminary. A pioneer in the field as well as a historian with an encyclopedic mind, he recounted the landmark discoveries leading to the birth of flow cytometry. His insightful and philosophical account of the journey, both historical and personal, revealed how this enabling technology had made it possible to answer questions that no one even knew how to ask a few decades ago. One of us (Robert) takes pride in presenting it to students taking microscopic anatomy courses where they also learn about Robert Hooke and his classic book Micrographia in which he gave “cells” their name. The current edition aims to present established and emerging methodologies in cytometry. The protocol chapters explain the principles behind the methodologies, present stepby-step procedures, and highlight notes for successful execution. Other chapters provide comprehensive overviews of useful tools and technologies as well as novel applications and troubleshooting guides. Understanding the basic principles of cytometry enhances the appreciation of these meticulously prepared protocols. For flow cytometry, a practical way to achieve that is to learn how to build a simple yet fully functional conventional flow cytometer. Detailed instructions listed in one chapter ensure a fruitful and gratifying outcome. An overview of available lasers ranging in wavelengths from deep ultraviolet to infrared provides guidance in choosing fluorophores for conventional as well as spectral flow cytometry. Spectral overlap among fluorophores results in signal spillover of one fluorophore into detectors other than the detector that captures maximal signal intensity of that fluorophore. Spillover correction (using compensation or unmixing) is employed to identify individual fluorophores in a mixture. However, this process reveals the associated measurement error as data spread that impairs data resolution in the affected detectors. For panel design optimization, it is beneficial to know the expected spreading produced by any combination of fluorophores. One chapter describes metrics that can be used to evaluate spreading. Spectral flow cytometry is a relatively new concept that has generated intense enthusiasm since the launching of commercial spectral flow cytometers in the early 2010s. Newer spectral flow cytometers allow simultaneous detection of 40 markers at the single-cell level. Whereas a conventional flow cytometer uses the same number of detectors as the number of fluorophores and assigns a specific detector (the detector that captures maximal signal intensity) to each fluorophore, a spectral flow cytometer utilizes more detectors than the number of fluorophores without dedicating any detector to a particular fluorophore. The

vii

viii

Preface

array of detectors on a spectral flow cytometer captures the full emission profile of each fluorophore from multiple lasers and identifies each fluorophore by its unique spectral signature. In addition, intrinsic cellular autofluorescence can be incorporated into the analysis. The process of thoughtful panel design and iterative optimization for spectral flow cytometry, which is also generally applicable to conventional flow cytometry, is documented in one chapter. Well-established flow cytometric methodologies presented in this edition include comprehensive immunophenotyping, fluorescence-activated cell sorting, proliferation monitoring, and apoptosis assays. The chapter on comprehensive immunophenotyping presents a panel that includes co-stimulatory molecules and inhibitory checkpoint molecules. The chapter on cell sorting explains the principles of this immensely useful technology and underscores considerations for a successful cell sorting experiment. Updated chapters on proliferation and apoptosis include modifications that incorporate spectral flow cytometry. Simultaneous detection of mRNAs and proteins using branched DNA technology coupled with fluorochrome-conjugated antibodies was documented in the last edition. An alternative methodology known as Cellular Indexing of Transcriptomes and Epitopes by sequencing (CITE-seq) uses oligonucleotide-conjugated antibodies to identify cell surface proteins with sequence-based readout, enabling simultaneous profiling of the RNA transcriptome and surface protein expression in single cells. One chapter describes co-staining of fluorochrome-conjugated and oligonucleotide-conjugated antibodies that allows cell purification by fluorescence-activated cell sorting prior to CITE-seq. Numerous aspects of flow cytometry are instrumental in clinical and biomedical research applications such as SARS-CoV-2 (COVID-19) monitoring and Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) gene editing. The chapter on COVID-19 describes a quantitative and standardized pseudovirus-based assay for the measurement of neutralizing antibodies in serum samples. Such assays are critical to COVID-19 diagnosis, surveillance, and vaccine development. The chapter on CRISPR describes a facile protocol to append a bioluminescent tag to an endogenous oncogenic transcription factor using a co-expressed fluorescent protein to isolate the gene-edited cells, facilitating the search for drugs that selectively target the oncoprotein. Flow cytometry continues to play significant roles in diverse fields, including the discovery of small molecule fluorescent probes (as an alternative to antibodies) capable of identifying specific live immune cell types. One chapter describes the screening for a B cell selective probe. The last edition presented an overview on fluorescence lifetime flow cytometry that measures the timing of fluorescence decay from excited fluorophores on cytometers and imagers. Applications include measurements of exogenous fluorophores/ microspheres, endogenous fluorophores, fluorescent proteins, and Fo¨rster resonance energy transfer (FRET). The updated chapter also includes a discussion on new high-throughput fluorescence lifetime imaging microscopy (ht-FLIM) approaches. Imaging flow cytometry combines phenotyping capability with morphological assessment while assaying single-cell suspensions. Blur-free imaging techniques coupled with realtime intelligent image processors have enabled microfluidic sorters to sort cells, albeit at much lower throughput than conventional fluorescence-activated droplet cell sorters. Recently, high-throughput image-enabled cell sorting has been achieved on a conventional droplet sorter enhanced with innovative camera-free imaging technology and ultra-fast signal processing. In addition, the sorter is capable of full-spectrum fluorescence detection. One chapter introduces this impactful achievement illustrated by a novel application.

Preface

ix

Mass cytometry (also known as Cytometry by Time-of-Flight or CyTOF), unlike flow cytometry that uses a flow cytometer to analyze cells labeled with antibodies attached to fluorescent tags, employs a mass spectrometer to analyze cells labeled with antibodies attached to heavy metal isotopes. As this technology produces minimal signal spillover among detectors, it has been used productively for studies involving simultaneous detection of 40 or more markers. The updated chapter on CyTOF delineates intracellular cytokine assays for the detection of antigen-specific T cells. Integration of imaging capability into the latest model of mass cytometers adds a significant dimension to this platform. Imaging mass cytometry is capable of deciphering cellular interactions by examining spatial localization of various cell types in situ. It can also be performed on archival samples. One chapter describes immune profiling using this methodology as well as modifications that integrate RNAscope to detect RNA transcripts, enabling simultaneous RNA and protein detection. The ability to measure an increasing number of markers simultaneously creates highdimensional data, ushering in a new era of computational analysis. Computational tools originally developed for single-cell RNA sequencing have been adapted for cytometry and are continuously being refined to keep pace with the challenges in visualization and interpretation of large complex datasets. Typically, data are cleaned, compensated or unmixed, transformed, and normalized. Subsequently, algorithms performing dimensionality reduction and clustering are applied to facilitate data visualization and cell population identification. Two chapters demonstrate some promising strides made in the clinical setting for both flow and mass cytometry. We are grateful to our contributors for their commitment as well as their creativity in meeting the submission deadline (even citing different time zones!). Their willingness to impart their extensive knowledge exemplifies the spirit of cooperation that is pervasive in the cytometry community. We would like to thank John Walker for his invitation and expert editorial guidance. We would also like to dedicate this edition to the memory of Howard Shapiro. Bethesda, MD, USA Washington, DC, USA

Teresa S. Hawley Robert G. Hawley

Contents Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v vii xiii

1 Flow Cytometry: The Glass Is Half Full . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Howard M. Shapiro 2 Basic Principles of Flow Cytometer Operation: The Make your Own Flow Cytometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . William G. Telford 3 Laser Sources for Traditional and Spectral Flow Cytometry . . . . . . . . . . . . . . . . . . William G. Telford 4 How to Measure “Spillover Spread”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Debajit Bhowmick and Timothy P. Bushnell 5 Comprehensive Immunophenotyping by Polychromatic Cytometry . . . . . . . . . . . Takuto Nogimori and Takuya Yamamoto 6 Panel Design and Optimization for Full Spectrum Flow Cytometry . . . . . . . . . . . Laura Ferrer-Font, Sam J. Small, Evelyn Hyde, Katherine R. Pilkington, and Kylie M. Price 7 Practicalities of Cell Sorting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mark Cheetham, Derek Davies, Christopher Hall, Charlotte Christie Petersen, Reiner Schulte, and Rachael Walker 8 Image-Enabled Cell Sorting Using the BD CellView Technology . . . . . . . . . . . . . Malte S. Paulsen 9 Monitoring Cell Proliferation by Dye Dilution: Considerations for Panel Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Joseph D. Tario Jr., Kah Teong Soh, Paul K. Wallace, and Katharine A. Muirhead 10 Multiparametric Analysis of Apoptosis by Flow Cytometry . . . . . . . . . . . . . . . . . . . William G. Telford 11 Quantitative and Standardized Pseudovirus Neutralization Assay for COVID-19 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jerilyn R. Izac, Edward J. Kwee, Adolfas Gaigalas, and Lili Wang 12 CRISPR-Cas9-Mediated Bioluminescent Tagging of Endogenous Proteins by Fluorescent Protein-Assisted Cell Sorting . . . . . . . . . . . . . . . . . . . . . . . Robert G. Hawley and Teresa S. Hawley 13 Co-staining with Fluorescent Antibodies and Antibody-Derived Tags for Cell Sorting Prior to CITE-Seq . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiaoshan Shi, Gisele V. Baracho, Woodrow E. Lomas III, Hye-Won Song, Stephanie J. Widmann, and Aaron J. Tyznik

1

xi

11 33 69 85 99

125

145

159

217

259

273

287

xii

14

15

16

17

18 19 20

Contents

Discovery of Live Cell Selective Fluorescent Probes and Elucidation of Their Mechanisms: Case Study of B Cell Selective Probe CDgB. . . . . . . . . . . . Haw-Young Kwon, Young-Tae Chang, and Nam-Young Kang Fluorescence Lifetime Measurements and Analyses: Protocols Using Flow Cytometry and High-Throughput Microscopy . . . . . . . . . . . . . . . . . . Jessica P. Houston, Samantha Valentino, and Aric Bitton Using Artificial Intelligence to Interpret Clinical Flow Cytometry Datasets for Automated Disease Diagnosis and/or Monitoring . . . . . . . . . . . . . . . Yu-Fen Wang, Jeng-Lin Li, Chi-Chun Lee, Paul K. Wallace, and Bor-Sheng Ko Approaching Mass Cytometry Translational Studies by Experimental and Data Curation Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paulina Rybakowska, Marta E. Alarco n-Riquelme, ˜ o n and Concepcion Maran CyTOF Intracellular Cytokine Assays for Antigen-Specific T Cells . . . . . . . . . . . . Dongxia Lin and Holden T. Maecker Imaging Mass Cytometry for In Situ Immune Profiling . . . . . . . . . . . . . . . . . . . . . Kevin Hu, Andrew Harman, and Heeva Baharlou Cytometry in High-Containment Laboratories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Melanie Cohen, Julie Laux, and Iyadh Douagi

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

305

323

353

369

395 407 425 457

Contributors MARTA E. ALARCO´N-RIQUELME • Pfizer-University of Granada-Junta de Andalucı´a Centre for Genomics and Oncological Research (GENYO), Granada, Spain; Institute for Environmental Medicine, Karolinska Institutet, Stockholm, Sweden HEEVA BAHARLOU • Centre for Virus Research, The Westmead Institute for Medical Research, Westmead, NSW, Australia; School of Medical Sciences, Faculty of Medicine and Health, The University of Sydney, Sydney, NSW, Australia GISELE V. BARACHO • BD Biosciences, La Jolla, CA, USA DEBAJIT BHOWMICK • Flow Cytometry Facility, Brody School of Medicine, East Carolina University, Greenville, NC, USA ARIC BITTON • Cytek Biosciences, Fremont, CA, USA TIMOTHY P. BUSHNELL • Center for Advanced Research Technologies and Department of Pediatrics, University of Rochester Medical Center, Rochester, NY, USA YOUNG-TAE CHANG • Department of Chemistry, Pohang University of Science and Technology, Pohang, Republic of Korea; SenPro, C5 building, Pohang University of Science and Technology, Pohang, Gyeongbuk, South Korea MARK CHEETHAM • Cambridge, UK MELANIE COHEN • Flow Cytometry Section, Research Technologies Branch, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD, USA DEREK DAVIES • Francis Crick Institute, London, UK IYADH DOUAGI • Flow Cytometry Section, Research Technologies Branch, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD, USA LAURA FERRER-FONT • Hugh Green Cytometry Centre, Malaghan Institute of Medical Research, Wellington, New Zealand ADOLFAS GAIGALAS • Biosystem and Biomaterials Division, National Institute of Standards and Technology (NIST), Gaithersburg, MD, USA CHRISTOPHER HALL • Flow Cytometry Facility, Babraham Institute, Cambridge, UK ANDREW HARMAN • Centre for Virus Research, The Westmead Institute for Medical Research, Westmead, NSW, Australia; School of Medical Sciences, Faculty of Medicine and Health, The University of Sydney, Sydney, NSW, Australia ROBERT G. HAWLEY • Department of Anatomy and Cell Biology, School of Medicine and Health Sciences, George Washington University, Washington, DC, USA TERESA S. HAWLEY • Bethesda, MD, USA JESSICA P. HOUSTON • Department of Chemical & Materials Engineering, New Mexico State University, Las Cruces, NM, USA KEVIN HU • Centre for Virus Research, The Westmead Institute for Medical Research, Westmead, NSW, Australia; School of Medical Sciences, Faculty of Medicine and Health, The University of Sydney, Sydney, NSW, Australia EVELYN HYDE • Malaghan Institute of Medical Research, Wellington, New Zealand JERILYN R. IZAC • Biosystem and Biomaterials Division, National Institute of Standards and Technology (NIST), Gaithersburg, MD, USA NAM-YOUNG KANG • SenPro, C5 building, Pohang University of Science and Technology, Pohang, Gyeongbuk, South Korea; Department of Creative IT Engineering, Pohang University of Science and Technology, Pohang, Gyeongbuk, South Korea

xiii

xiv

Contributors

BOR-SHENG KO • AHEAD Medicine Corporation, San Jose, CA, USA; AHEAD Intelligence Ltd, Taipei, Taiwan; Department of Hematological Oncology, National Taiwan University Cancer Center, Taipei, Taiwan; Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan; School of Medicine, College of Medicine, National Taiwan University, Taipei, Taiwan EDWARD J. KWEE • Biosystem and Biomaterials Division, National Institute of Standards and Technology (NIST), Gaithersburg, MD, USA HAW-YOUNG KWON • Department of Chemistry, Pohang University of Science and Technology, Pohang, Republic of Korea; SenPro, C5 building, Pohang University of Science and Technology, Pohang, Gyeongbuk, South Korea JULIE LAUX • Flow Cytometry Section, Research Technologies Branch, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD, USA CHI-CHUN LEE • Department of Electrical Engineering, National Tsing Hua University, Hsin-Chu, Taiwan JENG-LIN LI • Department of Electrical Engineering, National Tsing Hua University, Hsin-Chu, Taiwan DONGXIA LIN • Sirona Dx, Portland, OR, USA WOODROW E. LOMAS III • BD Biosciences, San Jose, CA, USA HOLDEN T. MAECKER • Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA, USA; Institute for Immunity, Transplantation, and Infection, Stanford University School of Medicine, Stanford, CA, USA CONCEPCIO´N MARAN˜O´N • Pfizer-University of Granada-Junta de Andalucı´a Centre for Genomics and Oncological Research (GENYO), Granada, Spain KATHARINE A. MUIRHEAD • SciGro, Inc., Middleton, WI, USA TAKUTO NOGIMORI • Laboratory of Precision Immunology, Center for Intractable Diseases and ImmunoGenomics, National Institutes of Biomedical Innovation, Health and Nutrition, Osaka, Japan MALTE S. PAULSEN • Novo Nordisk Foundation Center for Stem Cell Medicine (reNEW), University of Copenhagen, Copenhagen, Denmark CHARLOTTE CHRISTIE PETERSEN • The FACS Core Facility, Aarhus University, Aarhus, Denmark KATHERINE R. PILKINGTON • Cytek Biosciences, Fremont, CA, USA KYLIE M. PRICE • Hugh Green Cytometry Centre, Malaghan Institute of Medical Research, Wellington, New Zealand; Malaghan Institute of Medical Research, Wellington, New Zealand PAULINA RYBAKOWSKA • Pfizer-University of Granada-Junta de Andalucı´a Centre for Genomics and Oncological Research (GENYO), Granada, Spain REINER SCHULTE • Flow Cytometry Facility, Cambridge Institute for Medical Research, University of Cambridge, Cambridge, UK HOWARD M. SHAPIRO • One World Cytometry, Inc., West Newton, MA, USA XIAOSHAN SHI • BD Biosciences, San Jose, CA, USA SAM J. SMALL • Hugh Green Cytometry Centre, Malaghan Institute of Medical Research, Wellington, New Zealand; Malaghan Institute of Medical Research, Wellington, New Zealand KAH TEONG SOH • Roswell Park Comprehensive Cancer Center, Buffalo, NY, USA; Agenus, Inc., Lexington, MA, USA HYE-WON SONG • BD Biosciences, La Jolla, CA, USA JOSEPH D. TARIO JR. • Roswell Park Comprehensive Cancer Center, Buffalo, NY, USA

Contributors

xv

WILLIAM G. TELFORD • Laboratory of Pathology, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA AARON J. TYZNIK • BD Biosciences, La Jolla, CA, USA SAMANTHA VALENTINO • Department of Chemical & Materials Engineering, New Mexico State University, Las Cruces, NM, USA RACHAEL WALKER • Flow Cytometry Facility, Babraham Institute, Cambridge, UK PAUL K. WALLACE • Roswell Park Comprehensive Cancer Center, Buffalo, NY, USA; SciGro, Inc., Sedona, AZ, USA LILI WANG • Biosystem and Biomaterials Division, National Institute of Standards and Technology (NIST), Gaithersburg, MD, USA YU-FEN WANG • AHEAD Medicine Corporation, San Jose, CA, USA; AHEAD Intelligence Ltd, Taipei, Taiwan STEPHANIE J. WIDMANN • BD Biosciences, La Jolla, CA, USA TAKUYA YAMAMOTO • Laboratory of Precision Immunology, Center for Intractable Diseases and ImmunoGenomics, National Institutes of Biomedical Innovation, Health and Nutrition, Osaka, Japan

Chapter 1 Flow Cytometry: The Glass Is Half Full Howard M. Shapiro Abstract Accompanied by a historical perspective of the field of cytometry, this introductory chapter provides a broad view of what flow cytometry can do; hence, the glass is half full. Key words Micrographia, Cells, Blood cells, Dyes, Fluorescence, Microscopy, Hemacytometry, Flow cytometry, Electrostatic sorting, Coulter volume, Poisson statistics

1 Introduction This book presents ample evidence that flow cytometry has provided the means for developing an armamentarium of reagents and measurements that make it possible to answer questions about cells that nobody even knew how to ask when the field got started. The technology now accounts for a multibillion dollar market, with tens of thousands of instruments, most of which cost at least tens of thousands of US dollars, now in use worldwide. Most of the annual expenditure is aimed, directly or indirectly, at improving the overall health of our species, which may require suppressing or eliminating cells from other species and rogue elements from our own. A recent PubMed search on “flow cytometry” returned 180,038 references, dating back to the 1960s; over 75,000 have been added since I wrote a chapter for the previous edition of this compendium in 2010. There are almost certainly not tens of thousands of people who know how to make optimal use of the full range of capabilities of any state-of-the-art flow cytometer; books such as this one are designed to help the users keep up with the apparatus and the

Dr. Shapiro is deceased. This chapter is reproduced as a tribute to him from: Shapiro, H.M. (2018). Flow Cytometry: The Glass Is Half Full. In: Hawley, T., Hawley, R. (eds) Flow Cytometry Protocols. Methods in Molecular Biology, vol 1678. Humana Press, New York, NY. https://doi.org/10.1007/978-1-4939-7346-0_1. Teresa S. Hawley and Robert G. Hawley (eds.), Flow Cytometry Protocols, Methods in Molecular Biology, vol. 2779, https://doi.org/10.1007/978-1-0716-3738-8_1, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

1

2

Howard M. Shapiro

methodology, both of which make demands on the user. This chapter and those that follow, except for the last one, will provide a broad view of what flow can do. At the end of the book, I will focus on what flow cannot do and on what can now be done using alternative methods, both elaborate and simple, in hopes of improving readers’ perspectives. I remember strolling through Glasgow in 2015 with Bob and Teresa Hawley and some other folks, during the CYTO meeting of the International Society for the Advancement of Cytometry (ISAC), an organization with which I would guess most readers of this piece are acquainted. The major lecture of an ISAC meeting is the Robert Hooke lecture, given that year by Carl June, who has been doing flow cytometry since 1986 and who has attracted worldwide attention in recent years by engineering T cells to fight cancer, a project in which flow cytometry has been and is expected to be critical. It was impressive to hear him speak and even more impressive to hear the discussion that followed, with several members of the audience who had also ventured into cancer immunotherapy comparing notes on their respective successful treatment methods. The Hooke lecture was named because Robert Hooke had given cells their name in a book written before he had actually seen what we would today recognize as cells. Hooke’s classic Micrographia: or Some Physiological Descriptions of Minute Bodies Made by Magnifying Glasses with Observations and Inquiries Thereupon, impressively illustrated by the author, was commissioned by the Royal Society of London, for which he was curator of experiments, and appeared in January 1665. In it, Hooke, citing their similarity to cells in a honeycomb, called the spaces visible in thin longitudinal and transverse slices of cork “cells,” with no inkling that they had been formerly occupied by living components of the tree. Lenses (so named because they were lentil-shaped) had been used to start fires since ancient times (the “focus” is where the fire starts) and correct vision since around 1300, but it was not until about 1600 that Italian and Dutch spectacle makers combined them to bring faraway objects closer, thereby inventing the telescope, and to bring objects otherwise too small to see into view, inventing the microscope. Hooke conceived these devices as extending the sense of vision. Although there is no evidence of his having provided a jacket blurb, Samuel Pepys, the famous diarist, noted then that he had sat up until 2 a.m. reading Micrographia and described it as “the most ingenious book that ever I read in my life.” An even better indication of the book’s popularity is given in the writings of Jonathan Swift, born 2 years after Micrographia was published. In the 1726 novel Gulliver’s Travels, Swift’s surgeon protagonist, Gulliver, describes an encounter with giant Brobdingnagian beggars:

Flow Cytometry: The Glass Is Half Full

3

There was a woman with a cancer in her breast, swelled to a monstrous size, full of holes, in two or three of which I could have easily crept, and covered my whole body. There was a fellow with a wen in his neck, larger than five wool-packs... But the most hateful sight of all, was the lice crawling on their clothes. I could see distinctly the limbs of these vermin with my naked eye, much better than those of a European louse through a microscope, and their snouts with which they rooted like swine.

Gulliver’s last two sentences make it clear that Swift had some familiarity with Hooke’s book; the drawings of the louse undoubtedly attracted more attention from the lay audience than did those of the cork slices. Although I didn’t get around to it until after I wrote my cytometry book [1], I’ve been through Micrographia at least a couple of times. When I have asked how many people in audiences listening to Hooke Lectures at ISAC meetings have read it, however, I haven’t seen a lot of hands go up. That may explain why there are so many places on the Internet in which it is erroneously claimed the analogy was made to cells in a monastery or prison. This cytometric urban legend, like the notion that forward scatter measures cell size, is harder to kill than Dracula. Micrographia is fascinating for many reasons, written in understandable English, and available free online; I modestly propose that you read it if you haven’t. Hooke did not actually see living cells until years after Micrographia was published, when the Royal Society asked him to check up on reports from a self-taught Dutch fabric merchant, Antoni van Leeuwenhoek, who used simple microscopes of his own design that provided much higher magnification than was available from the compound microscopes then used by Hooke and others. The property van Leeuwenhoek used to distinguish “animate” particles, now known as cells, from “inanimate” ones was motility, which kept him obsessively interested in bacteria, protozoa, sperm, and other “animalcules” and largely indifferent to yeast and the considerable contributions it had made to humanity over the millennia. Many early microscopists anticipated that improvements in optics would quickly enable them to visualize atoms; they also tended to attribute morphological and biological characteristics of humans and other vertebrates to microorganisms, in which van Leeuwenhoek notably estimated the sizes of livers, kidneys, and other internal organs he expected would eventually be discernible. The expectation that the parts would scale as did the wholes was incorrect. A real Gulliver might have known what a microscope was; he would not have known what a cell was. What we now call “cells” were known by many other names until the mid-1800s, by which time improvements in microscopy including substage condensers and achromats and other lenses that reduced aberrations and increased resolution had made it easier to distinguish biologic structures from artifacts. Both Matthias Schleiden and Theodor

4

Howard M. Shapiro

Schwann, prime movers of but hardly sole contributors to what has been known since that time as the cell theory [1], used and favored the term. Whereas Hooke had used it to describe an empty space, it now referred to the membrane-bounded “elementary particle” of biology. Swift’s 1726 narrative speaks of a breast cancer and a wen (sebaceous cyst) at a time when no one had yet conceived of them as representative of two types of abnormal growth or that an infected and inflamed wen also exemplified bacterial growth and a proliferative response by the patient’s immune cells. By the late 1800s, both the metabolic versatility and pathogenic capability of microorganisms had been revealed by Louis Pasteur, Robert Koch, and others. The cell theory had become accepted, and Rudolf Virchow’s famous quote, “Omnis cellula e cellula,” embodied the efforts of many pathologists to understand disease at the cellular level. Human blood cells had come to microscopists’ attention; they were relatively easy to procure and could provide some information about patients’ overall state of health. Although anemias and leukemias had been described by this time, their causes were unclear; there were no known treatments for either, but their diagnosis and prognosis could be indicated by changes in the numbers of morphologically different cell types in the blood over time. The term “cytometer,” coined around 1880, described a device in which cells within a defined volume of specimen could be counted. “Cytometry” described the process. The cells most often came from blood, giving us the “hemacytometer” and “hemacytometry.” “Flow” and “cytometry” would not be combined until the 1970s; there could be no instrumental alternatives to microscopy until the 1950s. A fascinating account of the development of cell biology from medieval times until the twentieth century is given by the late Sir Henry Harris in The Birth of the Cell [2]. I have written at length on the history, technology, and philosophy of cytometry in my book [1] and, more recently, in a chapter in the previous edition of this compendium [3], a review/overview [4], and two additional book chapters [5, 6]. A detailed retrospective view of the origins of analytical flow cytometry, among other things, was also presented by the late Leonard Ornstein [7]. Because light scattering and absorption by most cells were insufficient to permit visual discrimination of internal details, synthetic dyes began to be used by the 1860s to stain specimens, with Paul Ehrlich providing much leadership. As a medical student in the 1870s, he recognized that different colored organic dyes with different chemical affinities would be bound to different degrees to different parts of different cells. This provided a basis for identifying cells within mixed populations; Ehrlich’s first practical success was in classifying the different types of white blood cells using dye samples provided by manufacturers. By 1880, he had experimented

Flow Cytometry: The Glass Is Half Full

5

with several stains containing mixtures of acidic and basic dyes, the former tending to stain cytoplasm and the latter nuclei. He had also used the blue basic dye methylene blue to stain bacteria. In 1882, Ehrlich joined forces with Koch and developed a stain that identified Mycobacterium tuberculosis (Mtb), newly discovered by Koch, by its ability (“acid fastness”) to retain stains after being washed in strongly acidic alcohol solutions. Slight modifications by others yielded the Ziehl–Neelsen (ZN) stain, which has remained the standard for detecting Mtb by transmitted light microscopy since 1883, with over 50 million slides analyzed annually. Ehrlich’s work also inspired Christian Gram’s initial work on staining bacteria. Ironically, Europe’s burgeoning dye industry had had its beginnings in the 1850s; a British chemistry student’s failed attempt to synthesize quinine, which could be used to treat malaria and was much in demand, serendipitously yielded the dye mauve, made fashionable by Queen Victoria. Quinine itself had been isolated in 1820 from cinchona bark; it had been known as an antimalarial (and one of the only effective drugs against any disease) since the 1600s. In the 1630s, after European invaders and their African slaves brought malaria to the Americas, Jesuits brought cinchona, a native Peruvian folk remedy for chills back to Rome, based on the unscientific but correct suspicion that it might cure malaria, a disease known since ancient times and common enough in Rome to have killed several popes. “Jesuit Powder” became an effective, though scarce and expensive, remedy for the disease, which until the 1950s was a problem in northern as well as southern regions of the world. No similarly effective treatment for any other infectious disease appeared before 1890. The discovery of malaria parasites by Alphonse Laveran in 1880 motivated an intensive search for dyes, which facilitated identification of these organisms in the blood of infected patients, who could then be treated with quinine. Laveran, a French military physician in Algeria, had examined the unstained blood of a malaria patient and found motile particles containing a blackish-brown pigment (now called hemozoin) known to be associated with the disease, but his findings would not be widely accepted until the pathogen’s morphology was better characterized by staining. It took over 20 years to come up with “the new black” for parasites. Gustav Giemsa’s stain, developed in 1904 and containing the red acid dye eosin, methylene blue, and the blue basic dye azure B, quickly became and has remained the “gold standard” for blood smear microscopy. Noting in 1891 that methylene blue by itself stained malaria parasites, Ehrlich had procured a supply from a dye company and successfully treated two malaria patients with it, anticipating his later success in curing syphilis with two of the over 900 compounds

6

Howard M. Shapiro

he tested against that disease. Ehrlich coined the term “chemotherapy,” and his demonstrations of it prompted many dye companies to expand into pharmacologic research. The company from which he obtained methylene blue is still in business and is now known as Hoechst. The year 1891 also saw the emergence of diphtheria antitoxin, the first successful immunotherapy, developed by Emil Behring (von Behring after he won the first Nobel Prize in medicine in 1901), a longtime colleague of Ehrlich. Immunoprophylaxis had been around longer. Variolation, which intentionally infected its subject with what was hoped would be a mild case of smallpox in hopes of preventing infection with the virulent natural form, had been known for centuries but often proved fatal. Beginning around 1800, it was replaced by vaccination, introduced by Edward Jenner, who had noted that milkmaids who had contracted cowpox, which rarely caused serious illness, were thereafter immune to smallpox. Pasteur decided to honor Jenner by calling the immunoprophylaxes for anthrax and rabies he himself later developed “vaccines.” The list of vaccine targets continues to increase. It was not until the 1890s that it was accepted that eukaryotic cells gave rise to new cells only by mitotic division, and the role of the chromosomes (the name of which provides the clue that they were not readily visible without staining) in heredity was not elucidated until the next century. By 1900, the chemistry of proteins was beginning to be understood, but the two types of nucleic acids then recently found to comprise “nuclein” had not yet been named and would not be called DNA and RNA for decades. Although photography permitted a more objective recording of microscope images than did drawings, the enhanced visual sense given to observers only allowed them to describe the sizes, shapes, colors, and textures of cells and their components, and motility and growth in culture offered the only indications of viability. Most microscopists still relied on sunlight as an illumination source; electric light and bright mantle lamps fueled by oil or gas only arrived on the scene late in the nineteenth century. Detection, characterization, and counting of cells were dependent on fallible human observers who, in almost all cases, had no objective means to confirm their findings and no alternative to manual data input and analysis. Advanced dark-field microscopy techniques requiring only sunlight illumination had, by the early 1900s, permitted visual observation of light scattered by particles below the resolution limit; “ultramicroscopes” documented the Brownian motion of large colloid molecules and confirmed Einstein’s predictions. Fluorescence microscopy, introduced around 1915, allowed observation of viruses stained with fluorescent dyes decades before the physicochemical bases of the phenomenon were clarified. By the 1930s, the development of photoelectric sensors and

Flow Cytometry: The Glass Is Half Full

7

electronics allowed spectrophotometers and microspectrophotometers to be built. There were already cytometric problems for them to solve. What we now know as the “rare event” problem arose almost as soon as it became necessary to examine sputum slides looking for tuberculosis (TB) and blood slides looking for malaria. In both diseases, it is likely that a slide taken from a patient with active disease will contain only a few pathogens. Clinicians needed to know how much of a sample needed to be analyzed to detect these. In 1907, William Sealy Gossett, writing under the pseudonym “Student,” had published a paper on the error of counting cells in a hemacytometer. The cells of interest to him were yeast; the pseudonym was necessary because his employers at the Guinness brewery feared their competitors might discover the utility of statistics in brewing. The statistical distribution involved had been discovered by Poisson a century earlier and is now known to be useful in determining sample sizes needed for determining a selected level of precision of counts of just about anything. Gossett’s methodology was adopted by Ronald Ross in 1910 to determine how much blood needed to be analyzed to get accurate estimates of the number of malaria parasites contained in the sample, which led to his introduction of the now standard “thick smear.” Poisson statistics are now widely used in cytometry not only for cell counting but also in the context of photon counting, which determines the precision of fluorescence measurements. By the 1930s, when at least some types of anemia had become treatable, Maxwell Wintrobe defined “red cell indices,” obtained by dividing bulk measurements of cell volume and blood hemoglobin by cell counts. Although these could, in principle, be used to distinguish normal and anemic blood and define types of anemia, visual counting of thousands of red cells to obtain the required precision was impractical. It was suggested in a 1934 article in Science that a photocell could be used to detect and count cells in suspension as they passed through an illuminated region of a capillary tube under a microscope, but the text suggested that the author’s initial experiments had not succeeded. The first working flow cytometer was built at the US Army’s Camp Detrick during World War II and was designed to detect anthrax spores in aerosols using dark-field illumination to measure light scattering. Although it could capture signals from almost 60% of bacteria resembling anthrax, it could not distinguish one species of bacterium from another or a bacterium from an organic or inorganic particle of approximately the same size. It could hardly be expected to do the job for which it was designed on a dusty battlefield. In 1951, an apparatus optically similar to the anthrax machine was demonstrated to be able to detect red blood cells in saline suspension and was eventually produced in England as a

8

Howard M. Shapiro

hematology counter. It could distinguish between the scatter signals from red cells and those from platelets; white cells were counted as red cells, which in normal blood and most abnormal blood would not significantly affect the results. Competitive optical blood cell counters were soon developed by several companies. Wallace Coulter, an American engineer who had attempted to produce a similar instrument, instead found it easier to build a counter, which detected the increase in electrical impedance produced as blood cells in saline flowed through a small orifice. It was soon established that impedance measurements accurately reflected particle volume. Both optical and Coulter counters were on the market by 1960; the latter could use small orifices to detect platelets, and white cell counts could be obtained from both types of counters after red cells were chemically lysed. A German group showed in 1964 that staining whole blood with acridine orange allowed white cells to be discriminated from red cells by a modified scatter-based counter using their nuclear or cytoplasmic fluorescence signals. This paper apparently represents the first instance of both fluorescence measurement and multiparameter analysis in flow cytometry. I had not run across any of the early work described above, by 1964, but I had had some introduction to cytometry. I was a “science brat”; my father was an MD interested in cancer treatment, and my mother, who had first studied microbiology, ran an electron microscope and was working on her PhD thesis in structural biology. Dinner table talk was frequently about science. I would hang out in her lab after school and help her stain slides, at least some of which involved stains for DNA, probably to distinguish developmental stages of the trematode sperm she studied. I did my first cytometry experiment in 1957, as a high school senior. It involved counting mitoses in sections of rapidly proliferating tissues fixed at varying intervals after animals were sacrificed and showed that mitosis could be completed in cells after death. It would be unacceptable and probably illegal to do the same experiment under the same circumstances today. In college, I studied biochemistry, taking a lab course in which we duplicated the experiments of Jacob and Monod et al. as they were published. I also learned about computers and built a mathematical model of metabolic fluxes in Escherichia coli. I went on to medical school at New York University (NYU), where a mainframe computer had been fitted with an analog-to-digital converter to process electrocardiograms (EKGs) recorded on tape; that gave me both computer time for my model-building and reasonable compensation for writing EKG analysis software. By 1965, I was a surgical intern at Bellevue Hospital in New York, where the researchers had Coulter Counters and routine clinical blood cell counting was done by third year medical students using microscopes and the classical Giemsa and Wright’s stains that

Flow Cytometry: The Glass Is Half Full

9

had been around since the early 1900s. I had learned to do differential white blood counts from my mother when I was around 12 years old and initially was sure that hematologists and pathologists must learn some secret handshake that allowed them to identify unusual white cells correctly. As an intern, I should have had no time to do so, but I still managed to sneak a glimpse at Science magazine every week when it came out. I realized that the October report by Louis Kamentsky et al. at IBM, who incorporated Caspersson’s ultraviolet (UV) nucleic acid measurements and a scatter-based size measurement into the first analytical flow cytometer, described something about which I had heard rumors the year before. In November, Science published Mack Fulwyler’s paper on electrostatic sorting of cells based on Coulter volume, which I also found of great interest. There were not a lot of new MDs with computer experience in those days; mine helped me get a job at the National Institutes of Health (NIH). I would be joining the National Cancer Institute (NCI) in July 1967 and working on building an automated microscope to study DNA synthesis in leukemic cells in children using thymidine autoradiography. NCI sent me to a meeting in early June at which Kamentsky described the current version of his instrument, which was measuring four parameters, including fluorescence, had acquired fluidic sorting capability, and was about to be equipped with a minicomputer. Even before my colleagues at NIH and the National Bureau of Standards and I started to build what was for a time the fanciest automated microscope around, I was solidly sold on flow cytometry. By the time I left NIH in 1971, having both helped build and used the minicomputer-controlled microscope and gotten some experience looking at leukemic blood with conventional stains, I knew there were no magic handshakes. I also realized that doing most serious cell-based diagnostic tasks was likely to require measurement of more parameters than could be done by the flow cytometers then in existence. Since there did not seem to be an available opportunity for me to make that happen, I had to shelve my empty glass for a couple of years. By 1976, I had access to a completely computer-controlled, three-laser, eight-parameter flow cytometer, which did not have a sorter because it had been designed for clinical blood cell counting and which would never be used for that because we found, when using it, that a single-laser, four-parameter instrument could do the job, be more reliable, and cost much less. I, at least, thought it would be great for research and be a much better front end for a cell sorter than was available for the two such instruments then available. At the 1976 Engineering Foundation meeting on Automatic Cytology, in Pensacola, Florida, manufacturers of sorters assured me that nobody would ever need all those beams and all those parameters, and would not pick up the project, so we scrounged

10

Howard M. Shapiro

some grants and stumbled on. Also, at that meeting, it was decided that what the participants were doing would thereafter be known as “flow cytometry.” Happy Hour had begun. We can now measure dozens of parameters with huge libraries of specific reagents and generate terabytes of data with a dozen or more lasers, most of which plug into the wall and do not cost more than a few thousand dollars. Detectors have gotten better, smaller, and, in some cases, cheaper. But there are problems. Flow has made it too easy to continue working at the single-cell level, or at the level of aggregates of too few cells, to relate meaningfully to what happens in extremely multicellular organisms like us. Human sperm can be analyzed in flow cytometers and sorted according to their sex chromosome content to minimize the likelihood that an embryo produced by in vitro fertilization (IVF) will be affected by a sex-linked disease. Human oocytes are a bit large for standard flow cytometers, and the 5- or 6-day-old blastocysts implanted in IVF, which have already differentiated to form about 40 trophoectoderm cells and 40 inner cell mass cells, have diameters of over 200 μm, allowing the transfer process to be monitored by nothing fancier than a light microscope. Most cytometry can be done without flow, but it is not as easy as it should be to get started doing it that way. So some people’s glasses are full, and others’ are empty. And on average. . .stay tuned. References 1. Shapiro HM (2003) Practical flow cytometry, 4th edn. Wiley-Liss, Hoboken 2. Harris H (1999) The birth of the cell. Yale University Press, New Haven 3. Shapiro HM (2011) The cytometric future: it ain’t necessarily flow! Methods Mol Biol 699: 471–482. https://doi.org/10.1007/978-161737-950-5_23 4. Shapiro HM, Apte SH, Chojnowski GM, Hanscheid T, Rebelo M, Grimberg BT (2013) Cytometry in malaria--a practical replacement for microscopy? Curr Protoc Cytom

Chapter 11:Unit 11 20. https://doi.org/10. 1002/0471142956.cy1120s65 5. Shapiro HM (2015) Microbial cytometry: what it was, is, and may be. In: Wilkinson MG (ed) Flow cytometry in microbiology: technology and applications. Caister Academic Press, Norfolk, pp 1–16 6. Shapiro HM (2017) Cytometry. In: Rifai N, Horvath AR, Wittwer CT (eds) Tietz textbook of clinical chemistry and molecular diagnostics, 6th edn. Elsevier 7. Ornstein L (1987) Tenuous but contingent connections. Electrophoresis 8:3–13

Chapter 2 Basic Principles of Flow Cytometer Operation: The Make your Own Flow Cytometer William G. Telford Abstract Flow cytometry is a critical technology for biomedical analysis and is an essential component of almost any study of the immune system. Widespread usage and increasing instrument complexity have, however, led to increasing neglect of education in their basic operating principles, a common situation with many technologies. This chapter describes the basics of flow cytometer operation using the Make Your Own Flow Cytometer (https://www.cytometryworks.com), a working cytometer than can be assembled by students into a functional instrument. This project and others like it is seeing widespread usage in biomedical education and can serve as models for like-minded investigators who wish to build their own systems. They also provide a good mechanism to introduce the key operational principles of flow cytometry as illustrated here. Key words Flow cytometry, Laser, Hydrodynamic focusing, Dichroic mirror, Bandpass filter, Photomultipler tube, Photodiode, Preamplifier, Amplifier, Digitizer, Analog-to-digital converter, Digital signal processor

1

Introduction Flow cytometry is an indispensable technology in cell biology and biomedical research, particularly in the analysis of the immune system [1]. Flow cytometers allow the analysis of individual cells in a hydrodynamically focused liquid stream, using lasers and sensitive detectors for light scattering measurement of cell size and optical density and fluorescence detection of cellular proteins, intracellular structures, and physiological status [2]. Modern fluorescence-based flow cytometers can detect more than 40 fluorescent markers simultaneously, thus allowing precision analysis of the complex cell mixtures that make up the immune system [3, 4]. The high-throughput nature of flow cytometry allows millions of individual cells to be rapidly analyzed, permitting the

Teresa S. Hawley and Robert G. Hawley (eds.), Flow Cytometry Protocols, Methods in Molecular Biology, vol. 2779, https://doi.org/10.1007/978-1-0716-3738-8_2, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

11

12

William G. Telford

detection of rare cell populations and identification of new cell subsets. Cell sorters, flow cytometers equipped for physical cell separation, permit isolation of individual cell populations for further analysis. While the technology of flow cytometry has advanced considerably since its inception nearly 50 years ago, the basic optical and fluidic principles behind instrument operation have not changed. Like many advanced technologies, flow cytometers have largely evolved into “black boxes” where students new to the field are unaware of the many basic physical principles behind their operation. In the spirit of many similar programs in the technology “maker” culture, we have designed the Make Your Own Flow Cytometer system, a fully functional instrument that can be assembled from individual parts into a working instrument. This system is based on the original Build Your Own Flow Cytometer system developed by engineers and educators at the National Flow Cytometry Resource (NFCR) located at the Los Alamos National Laboratories (LANL) in the 1990s, one of the birthplaces of the modern cell sorter. The original Build Your Own system continues to see wide usage in flow cytometry education programs. The newer Make Your Own Cytometer is smaller and lighter, allowing it to be transported to flow cytometry educational workshops around the globe. As of 2023, it has been used as an educational tool in flow cytometry workshops held in over 18 countries in all 6 inhabited continents. It heavily relies on open-source, three-dimensional (3D) printed components, giving scientists and educators the inspiration and opportunity to construct their own system. A fully assembled system is shown in Fig. 2. In this chapter, we will use the Make Your Own Flow Cytometer to illustrate the principles of flow cytometer operation. Full details about the system, including free video presentations of its assembly and use, can be found at https://www.cytometryworks.com.

2

Materials 1. A breadboard composed of one of the following: (a) A carbon fiber (CarbonVision GmbH, Unterschleissheim, Germany). (b) FDM (fused deposition manufacturing). (c) 3D printed with aluminum-threaded inserts: acetonitrile butadiene styrene (ABS) for somewhat rigid components, carbon fiber-infused ABS for highly rigid or nonreflective components, or thermoplastic polyurethane (TPU) for flexible components.

Basic Principles of Flow Cytometer Operation

13

2. Laser: A Coherent Bioray cyan 488-nm direct diode laser emitting output attenuated to 5 mW (Coherent Laser, Mountain View, CA, USA). 3. A laser focusing lens (plano-convex 100-mm focal length) and steering prisms (5-mm flats: N-BK7 glass (Thorlabs, Newton, NJ, USA). 4. A flow cell: sources from BD Biosciences (San Jose, CA, USA). 5. Filters and dichroics: sources from Semrock/IDEX Health and Science, LLC (Rochester, NY, USA). 6. A photomultiplier tube (PMT) model H6779 (Hamamatsu Photonics, Shizouka, Japan). 7. Low-pressure air regulators (Dwyer Instrument, Michigan City, IN, USA). 8. Pneumatic switches and metal fluidic fittings (Pneumadyne/ Norgren, Plymouth, MN, USA). 9. Power supplies: sources from Mean Well Enterprises Co. (New Taiwan City, Taiwan). 10. Azurite acquisition electronics (Darkling X, LLC, Los Alamos, NM, USA). 11. Kytos acquisition software (Darkling X, LLC). More information can be found at https://www. cytometryworks.com or by contacting the author.

3

Methods

3.1 Instrument Assembly

While flow cytometers vary in design and complexity, they all have common features (Fig. 1). Flow cytometers require (1) a cell suspension, usually labeled with one or more fluorescent markers; (2) a method to introduce the cells into a narrow liquid stream, usually confined in a cuvette or flow cell; (3) a laser aligned with and focused on the cell stream, allowing “interrogation” of individual cells for both light scatter measurement and excitation of attached fluorescent probes; (4) optics, mirrors, and filters to capture and separate light scatter and fluorescent signals from the cells; (5) sensitive detectors to measure these signals; and (6) electronics, computer systems, and software to process, collect, and display the resulting data. All of these systems are illustrated below in the Make Your Own system. A fully assembled Make Your Own system is shown in Fig. 2a, b. The system uses a baseplate where all the components are assembled (Fig. 2c), containing all the necessary attachment points for each component. Each component illustrates an important element in flow cytometer operation. Assembly of the system provides a good introduction to how flow cytometers operate.

14

William G. Telford

Fig. 1 Basic elements of a flow cytometer, including the characteristics of a cell population suitable for cytometry, generation of a hydrodynamic sample flow, interrogation of cells using a laser, detection of light scattering and fluorescence using collection optics, filters, dichroics and detectors, and processing and display of data 3.1.1 Lasers and Their Optics

Flow cytometers rely on lasers for both light scattering measurement of cells’ physical characteristics and to excite any fluorescent probes incorporated into cells [5]. It is no coincidence that the development of the earliest flow cytometers and the invention of lasers both occurred within the same time period. Early flow cytometers relied on large water-cooled lasers with significant cooling and power requirements. Modern cytometers use solid-state lasers, including direct diodes and diode-pumped modules. These are smaller, easier to operate, and longer-lived than their earlier gas counterparts and are now the dominant laser type among commercial instruments. The Make Your Own system uses a small cyan 488-nm direct diode laser and is shown in Fig. 3a (installed on the cytometer) and Fig. 3b. Historically, the cyan 488-nm laser light (originally generated by argon ion gas lasers) was the primary excitation source for flow cytometers due to its ability to excite common fluorochromes such as fluorescein. This wavelength continues to be used in solid-state form. However, modern cytometers usually employ multiple laser sources, with red, violet, ultraviolet (UV), and yellow being common additions. These arrays of multiple lasers allow many fluorescent probes to be excited

Basic Principles of Flow Cytometer Operation

15

Fig. 2 (a) A fully assembled Make Your Own Flow Cytometer. (b) A CAD schematic of the fully assembled system. (c) A CAD schematic of the empty breadboard prior to construction

simultaneously. Lasers applicable to flow cytometry are usually single mode (with a circular or elliptical beam output), with good output stability and low noise levels. The laser beam needs to be aligned with the center of a flow cell, thus allowing it to intercept the cell stream and excite the target cells. Lasers therefore need to be modified in terms of alignment, focus, and beam shape to optimally intercept the cell stream. Laser beam modification optics used in most flow cytometers are installed on the cytometer, as shown in Fig. 3a. Figure 3c shows a focusing lens used to reduce the beam diameter from >1 mm to 510 nm and reflecting light 0.9) can be used; however, they should be preferably placed on markers that are not co-expressed on the same cell types, or if co-expressed, only where separate populations can still be distinguished from one another, e.g., high-density well-resolved populations. When antigen density and fluorophore brightness are balanced, impacts from SE are minimized [3]. 2. Fluorophore brightness: Best practice is to inversely pair the fluorophore brightness with antigen density levels by combining brighter fluorophores with lower antigen expression levels (tertiary antigens) and dimmer fluorophores with markers expressed in abundance (primary antigens) (see Note 2 for antigen density classification). 3. Spreading error: Co-expressed markers and markers selected for functional readouts should be matched with fluorophores that receive minimal spread from other fluorophores in the panel. 4. Practical considerations: Designing a table similar to that shown in Fig. 1c will facilitate the fluorophore selection process. However, a range of other considerations should also be taken into account for optimal fluorophore selection, these include: (a) Assigning fluorescent reporter genes first, as they have an inflexible fluorescent signature. (b) Assigning very rare markers or those with limited fluorophore options (e.g., CXCR5 to PE). (c) Assigning the most common markers to rare fluorophores to provide greater panel design flexibility (e.g., CD45 to Alexa Fluor™ 532). (d) Assigning the remaining markers to the remaining fluorophore options, considering marker co-expression and fluorophore spread: (i) If markers are not co-expressed, use any bright fluorophore remaining that does not spread into fluorophores already assigned to tertiary markers. (ii) If markers are co-expressed, choose a medium intensity fluorophore that introduces less spreading error. (e) Assign dump markers/channels (see Note 3). Dump markers/channels can be assigned to individual fluorophores or the same fluorophore so that they are removed from the analysis at the same time. If assigning multiple markers to one fluorophore, single stained reference control optimization and unmixing will be easier and more accurate if a non-tandem dye is used.

Panel Design and Optimization for Spectral Flow Cytometry

107

(f) Unless non-viable cells are to be interrogated further, the viability dye can be selected last, regardless of any spreading issues, due to the wide range of available fluorophores (see Note 4). 5. Complete table: Once all the above considerations have been addressed, complete the working version of the table in Fig. 1c including: (a) The fluorophore chosen for each marker. (b) The relative brightness of the fluorophore selected (dim, medium, bright, very bright). 3.1.5 Review the Theoretical Panel Design

The key considerations for final theoretical panel review are as follows: 1. Are the chosen fluorophores distributed evenly across the spectrum? 2. Are the fluorophores introducing spread to other fluorophores assigned to co-expressed antigens? If the answer is affirmative, what is the level of antigen expression? Will marker resolution be affected? 3. Do the readout markers receive a minimal amount of spread? 4. Are the fluorophore-brightness to antigen-density levels well balanced?

3.2 Panel Optimization (Fig. 3) 3.2.1

Antibody Titrations

An essential first step in panel optimization is titrating the antibodies involved in the panel. Using the correct concentration of antibody will improve resolution, minimize non-specific binding, reduce spread, and reduce cost. The optimal concentration of antibody is one that gives rise to the best separation between the positive and the negative population. Each new lot of antibody should be titrated using the tissue and cells of interest. Viability dyes should also be titrated to avoid live cells taking up excess dye, which leads to spread issues and false positive populations. It is important to consider that certain markers or cell types might not be present in control samples or tissues. It is therefore important to perform antibody titrations under the corresponding experimental

Fig. 3 Schematic steps for panel optimization

108

Laura Ferrer-Font et al.

conditions. Step-by-step instructions for titrating antibodies have been described by the authors elsewhere [4]. 3.2.2 Preparation and Evaluation of Optimal Single-Stain (SS) Controls

To determine whether to use compensation beads or cells for individual SS controls, both should be tested in parallel using the same staining conditions as those used for the fully stained sample. The unmixing accuracy can be evaluated upon acquisition of both the bead and cell controls for each antibody used. To optimally unmix the data, when using beads as SS controls, place tight scatter gates around the beads, on a forward scatter vs. side scatter plot. In the case of using cells as SS controls, place the scatter gate on the cells of interest (i.e., if the marker of interest is CD3, place the scatter gate around lymphocytes, but if the marker of interest is CD14, the scatter gate should be placed around monocytes). Make sure that the SS control positive gate is placed on the brightest events. Remember that the negative should always match the autofluorescence of the positive (e.g., cells with cells and beads with beads). The accuracy of beads as SS controls should be tested first and then evaluated per marker by analyzing the single stained cell samples. If unmixing using beads as controls is not accurate, then cells are recommended as SS controls for that specific marker to provide accurate unmixing.

3.2.3

Before analyzing the fully stained (FS) sample for unmixing errors, it must be “cleaned” of false positive events. The FS sample used for unmixing verification should also contain positive events for all markers in the panel. Once this has been confirmed, we can assess if the selected SS controls, either beads or cells (Subheading 3.2.2), optimally unmix the FS sample, following these steps:

Unmixing Accuracy

1. Gate out false positive events by appropriate time gates, gating out doublets, gating out dead cells, and gating out aggregates. 2. Ensure the FS sample contains positive events for all markers in the panel by creating a 1 × N array (Side scatter-A versus each fluorophore). 3. Check how the SS controls selected in step 2 unmix the FS sample. Evaluate the unmixing accuracy using N × N plots for all markers. Ensure the MdnFI of the positive and negative populations of each marker on the N × N plots are well aligned. 4. If there are no major unmixing errors (i.e., the positive and negative populations of each marker in the N × N plots are well aligned), continue to evaluation of marker resolution. If there are detectable unmixing errors, select a different SS control. It might also be necessary to prepare new SS controls that are as bright or brighter than the FS sample using another tissue type or even another marker conjugated to the same fluorophore (this is recommended only for non-tandem dyes). If there are

Panel Design and Optimization for Spectral Flow Cytometry

109

still unmixing issues even once the unmixing has been performed correctly and optimally, it is possible to make small adjustments ( CD45 + leukocytes) to remove the complex AF associated with the tissue type. 10. Ideally the tissue being analyzed would be used to create the RRM, but in some cases the tissue type is not able to be cryopreserved or there is not sufficient sample. In such situations, a surrogate tissue must be selected. The tissue needs to be preservable, and it must represent as many antigens being analyzed within the experimental samples as possible, with these antigens being maintained after freeze/thawing. PBMCs are often selected as an RRM as they are easily isolated and maintain high viability after cryopreservation. Complications arise when the markers of interest in experimental samples cannot be found in such surrogate samples as PBMCs. As further described in Note 12, it may not be possible to normalize expression of these markers. Commercial cellular mimics for the cell type(s) of interest may be available; however, care must be taken if implementing these as the expression pattern of markers on these mimics may not be a close enough representation of that seen on biological samples and thus will not serve as an appropriate RRM. 11. If multiple activation conditions will be included for experimental samples, these conditions should be replicated in the RRM, including a non-activated control. Normalization works best when antigen expression is similar between RRM and experimental samples; therefore, an activated sample will require an activated RRM. 12. The CytoNorm algorithm can only normalize parameters for which there is corresponding expression in the RRM; therefore, if there are markers not represented in the RRM, these should be omitted from normalization. Other algorithms, e.g., CyCombine [21] or iMUBAC [22], do not require an RRM; therefore, if normalization of a marker not present in the RRM is required, utilization of a different normalization algorithm will be necessary. 13. If working with large datasets with batch-to-batch variation presents in viability staining and/or staining of clean-up markers such as CD45, it can be useful to skip these two data cleaning steps so that they can be included in normalization. This may allow easier clean-up post normalization as set gates can be employed without need for individual adjustment. 14. Normalization results can be assessed by examining the stain index of individual markers pre- and post-normalization. The stain index of individual markers should have reduced variation across data collection timepoints after normalization, indicating that the expression profiles have been well aligned. This is a

Panel Design and Optimization for Spectral Flow Cytometry

123

simple way to quantitatively assess the impact of normalization on each marker in addition to the visual assessment done through expert gating or dimensionality reduction techniques. 15. Normalization may require several iterations to reach the optimal outcome. The number of metaclusters in the underlying FlowSOM (if using CytoNorm) may need to be reduced to ensure batch-specific clusters are not being created, for example. Troubleshooting the CytoNorm output usually involves altering metacluster numbers, and/or adjusting the markers being normalized. A protocol about how to use CytoNorm with full spectrum cytometry data has been published elsewhere [16].

Acknowledgements This work was enabled by the Hugh Green Cytometry Centre at the Malaghan Institute of Medical Research, and we wish to thank the Hugh Green Foundation for their funding and support. The authors are also very grateful to Maria Jaimes for her expert opinon and feedback on the manuscript. References 1. Mahnke YD, Roederer M (2007) Optimizing a multicolor immunophenotyping assay. Clin Lab Med 27(3):469–485, v. https://doi.org/ 10.1016/j.cll.2007.05.002 2. Nguyen R, Perfetto S, Mahnke YD, Chattopadhyay P, Roederer M (2013) Quantifying spillover spreading for comparing instrument performance and aiding in multicolor panel design. Cytometry A 83(3):306–315. https://doi.org/10.1002/cyto.a.22251 3. Ashhurst TM, Smith AL, Jonathan N, King C (2017) High-dimensional fluorescence cytometry. Curr Protoc Immunol 119: 5.8.1–5.8.38. https://doi.org/10.1002/ cpim.37 4. Ferrer-Font L, Small SJ, Lewer B, Pilkington KR, Johnston LK, Park LM, Lannigan J, Jaimes MC, Price KM (2021) Panel optimization for high-dimensional immunophenotyping assays using full-spectrum flow cytometry. Curr Protoc 1(9):e222. https://doi.org/10. 1002/cpz1.222 5. Park LM, Lannigan J, Jaimes MC (2020) OMIP-069: forty-color full spectrum flow cytometry panel for deep immunophenotyping of major cell subsets in human peripheral blood. Cytometry A 97(10):1044–1051. https://doi.org/10.1002/cyto.a.24213

6. Jalbert E, Shikuma CM, Ndhlovu LC, Barbour JD (2013) Sequential staining improves detection of CCR2 and CX3CR1 on monocytes when simultaneously evaluating CCR5 by multicolor flow cytometry. Cytometry A 83(3): 280–286. https://doi.org/10.1002/cyto.a. 22257 7. Hally KE, Ferrer-Font L, Pilkington KR, Larsen PD (2022) OMIP 083: a 21-marker 18-color flow cytometry panel for in-depth phenotyping of human peripheral monocytes. Cytometry A 101(5):374–379. https://doi. org/10.1002/cyto.a.24545 8. Van Der Maaten L, Hinton G (2008) Visualizing data using t-SNE. J Mach Learn Res 9: 2579–2605 9. McInnes L, Healy J, Melville J (2018) UMAP: uniform manifold approximation and projection for dimension reduction. arXiv:1802.03426. https://doi.org/10. 48550/arXiv.1802.03426 ˜ ez NG, 10. Brummelman J, Haftmann C, Nu´n Alvisi G, Mazza EMC, Becher B, Lugli E (2019) Development, application and computational analysis of high-dimensional fluorescent antibody panels for single-cell flow cytometry. Nat Protoc 14(7):1946–1969.

124

Laura Ferrer-Font et al.

https://doi.org/10.1038/s41596-0190166-2 11. Schaefer PM, Kalinina S, Rueck A, von Arnim CAF, von Einem B (2019) NADH autofluorescence-a marker on its way to boost bioenergetic research. Cytometry A 95(1): 34–46. https://doi.org/10.1002/cyto.a. 23597 12. Jameson VJ, Luke T, Yan Y, Hind A, Evrard M, Man K, Mackay LK, Kallies A, Villadangos JA, McWilliam HEG, Perez-Gonzalez A (2022) Unlocking autofluorescence in the era of full spectrum analysis: implications for immunophenotype discovery projects. Cytometry A 101(11):922–941. https://doi.org/10.1002/ cyto.a.24555 13. Peixoto MM, Soares-da-Silva F, Schmutz S, Mailhe MP, Novault S, Cumano A, Ait-Mansour C (2022) Identification of fetal liver stroma in spectral cytometry using the parameter autofluorescence. Cytometry A 101(11):960–969. https://doi.org/10.1002/ cyto.a.24567 14. Kharraz Y, Lukesova V, Serrano AL, Davison A, ˜ oz-Ca´noves P (2022) Full spectrum cytoMun metry improves the resolution of highly autofluorescent biological samples: identification of myeloid cells in regenerating skeletal muscles. Cytometry A 101(10):862–876. https://doi. org/10.1002/cyto.a.24568 15. Novo D (2022) A comparison of spectral unmixing to conventional compensation for the calculation of fluorochrome abundances from flow cytometric data. Cytometry A 101(11):885–891. https://doi.org/10.1002/ cyto.a.24669 16. Ferrer-Font L, Kraker G, Hally KE, Price KM (2023) Ensuring full spectrum flow cytometry data quality for high-dimensional data analysis. Curr Protoc 3(2):e657. https://doi.org/10. 1002/cpz1.657

17. Van Gassen S, Gaudilliere B, Angst MS, Saeys Y, Aghaeepour N (2020) CytoNorm: a normalization algorithm for cytometry data. Cytometry A 97(3):268–278. https://doi. org/10.1002/cyto.a.23904 18. den Braanker H, Bongenaar M, Lubberts E (2021) How to prepare spectral flow cytometry datasets for high dimensional data analysis: a practical workflow. Front Immunol 12: 768113. https://doi.org/10.3389/fimmu. 2021.768113 19. Kimball AK, Oko LM, Bullock BL, Nemenoff RA, van Dyk LF, Clambey ET (2018) A beginner’s guide to analyzing and visualizing mass cytometry data. J Immunol 200(1):3–22. h t t p s : // d o i . o r g / 1 0 . 4 0 4 9 / j i m m u n o l . 1701494 20. Ashhurst TM, Marsh-Wakefield F, Putri GH, Spiteri AG, Shinko D, Read MN, Smith AL, King NJC (2022) Integration, exploration, and analysis of high-dimensional single-cell cytometry data using Spectre. Cytometry A 101(3):237–253. https://doi.org/10.1002/ cyto.a.24350 21. Pedersen CB, Dam SH, Barnkob MB, Leipold MD, Purroy N, Rassenti LZ, Kipps TJ, Nguyen J, Lederer JA, Gohil SH, Wu CJ, Olsen LR (2022) cyCombine allows for robust integration of single-cell cytometry datasets within and across technologies. Nat Commun 13(1):1698. https://doi.org/10.1038/ S41467-022-29383-5 22. Ogishi M, Yang R, Gruber C, Zhang P, Pelham SJ, Spaan AN, Rosain J, Chbihi M, Han JE, Rao VK, Kainulainen L, Bustamante J, Boisson B, Bogunovic D, Boisson-Dupuis S, Casanova J-L (2021) Multibatch cytometry data integration for optimal immunophenotyping. J Immunol 206(1):206–213. https://doi. org/10.4049/jimmunol.2000854

Chapter 7 Practicalities of Cell Sorting Mark Cheetham, Derek Davies, Christopher Hall, Charlotte Christie Petersen, Reiner Schulte, and Rachael Walker Abstract Cell sorting is a technique commonly used in academic and biotechnology laboratories in order to separate out cells or particles of interest from heterogeneous populations. Cell sorters use the same principles as flow cytometry analyzers, but instead of cell populations passing to the waste of the instrument, they can be collected for further studies including DNA sequencing as well as other genomic, in vitro and in vivo experiments. This chapter aims to give an overview of cell sorting, the different types of cell sorters, details on how a cell sorter works, as well as protocols that are useful when embarking on a journey with cell sorting. Key words Flow cytometry, Cell sorting, Fluorescence-activated cell sorting (FACS)

1

Introduction Cell sorting is a method of separating particles of interest in a sample from unwanted particles. The history and technical development of cell sorting have been described in numerous publications [1–3]. In brief, Mack Fulwyler, a physicist working in Los Alamos, was looking at the effects of atmospheric nuclear fallout [4]. One of his colleagues, who analyzed red blood cells on a Coulter counter, could see two populations and wanted to know if one of them was the immature red blood cells. Fulwyler disputed this and wanted to isolate the abnormal population to analyze it in isolation on the Coulter counter. In 1964, Richard Sweet patented the use of electrostatic deflection of ink droplets which would eventually lead to the development of the inkjet printer [5]. Fulwyler saw that this might be a way of separating the cell populations. He used the Coulter counter to size the object and eject them through an orifice (nozzle) and then charged the drops containing cells of interest and deposited them into tubes. This led to his patent in 1965 which covered the analysis and sorting of a stream

Teresa S. Hawley and Robert G. Hawley (eds.), Flow Cytometry Protocols, Methods in Molecular Biology, vol. 2779, https://doi.org/10.1007/978-1-0716-3738-8_7, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

125

126

Mark Cheetham et al.

of flowing cells [6]. Leonard and Leonore Herzenberg, two immunologists based in Stanford, wanted to analyze immune cells in a high-throughput manner using quantitative data and fluorescence. Herzenberg looked into options for sorting these cells and approached Mack Fulwyler about his sorter based on Coulter volume and asked if fluorescence could be added. Herzenberg had seized upon a visionary idea to build an instrument capable of rapidly sorting live cells identified by their distinctive surface molecules detected by monoclonal antibodies conjugated to fluorescent tags, and the fluorescence-activated cell sorter (FACS™) was born. Herzenberg worked with Becton Dickinson (BD) to commercialize the cell sorter with a FACSII sorter being released in 1976. This one laser instrument could measure forward scatter and fluorescence at 530 nm, to be able to look at fluorescein. The basic principles of cell sorting are largely unchanged from this time with the major advances being in the miniaturization of components, the speed of the detection electronics, and developments in dye chemistry resulting in a diverse range of fluorochromes. We will briefly look at alternative ways of sorting cells toward the end of this chapter, but for the most part, we will concentrate on electrostatic drop deflection cell sorters which are by far the most common.

2

Cell Analysis To sort cells, they must first be analyzed, essentially in the same way as in an analytical flow cytometer. In a pressurized fluidics system, cells are hydrodynamically focused so that they pass individually through one or more laser beams (Fig. 1). The emitted fluorescence and scattered light are collected and, after passing through appropriate optical filters, the light is quantitated by a detector, for example, a PMT (photomultiplier tube), APD (avalanche photodiode), or siPD (silicon photodiode). The desired populations are defined by a series of user-generated regions in the software. There are two approaches that manufacturers have taken in cell sorters as to where fluorescence is generated. In systems such as the Aria series and the Sony systems, the fluorescence measurement is made in a cuvette prior to stream ejection from the nozzle (cuvette-based systems). In other systems, such as the MoFlo family (XDP, Astrios, and Influx) and Bigfoot, the stream is ejected through the nozzle before the cells pass through the lasers (jet-inair systems). A list of jet-in-air and cuvette-based cell sorters is given in Table 1. Each system has advantages and disadvantages. Jet-in-air instruments are generally faster, while sense-in-channel sorters can give higher sensitivity. This is not due to the frequency of drop formation but rather to the acceleration of the cells within the

Practicalities of Cell Sorting

127

Sheath Flow

Nozzle Body Pressurized Sheath

Flow Channel

Nozzle Tip Nozzle holder Nozzle and O ring

Sheath/Sample Jet

Stream or Jet

100 µm Orifice

Lasers

Fig. 1 Schematic of cell injection into the flow cytometer light collection device

nozzle assembly before they are ejected into air. At a pressure of 70 psi, the sheath and sample fluid are accelerated to around 35 m/ s very rapidly. This means that they pass through the lasers very quickly and it is unlikely that more than one cell will be analyzed simultaneously and therefore the system is making single cell measurements. In the cuvette-based sorters, cells pass more slowly through the laser beams, approximately 6 m/s, so they are interrogated for longer. This means that more photons are produced and this in turn may aid the detection of weak signals. Post interrogation, the sheath and sample streams are accelerated to around 30 m/s prior to their ejection through the nozzle. In a sense-inchannel sorter, the light collection lenses can be gel-coupled to the flow channel, allowing a higher numerical aperture and thus allowing more light to be collected. Cell aggregates must be eliminated from a sort as they are a potential source of reduced purity. By analyzing the pulse shape, a single event can be classified as a doublet and gated out of the analysis and therefore sort criteria. As the sample pressure is increased by the operator to increase the sort speed, the coincident events increase thus reducing the number of cell available to sort and consequently affecting the total yield. As cuvette-based systems slow down the events as they pass through the lasers, cells may arrive at the intersection point before the system has finished analyzing a previous event. The event is aborted thus causing a drop in overall yield; this is termed a hard or electronic, abort. In jet-in-air systems, the events are traveling much faster, so they clear the area of detection much faster. With the jet-in-air systems, the acceleration is not stepwise and this means that sorting is quicker because hard aborts are reduced.

128

Mark Cheetham et al.

Table 1 A snapshot of droplet-based systems available at time of going to press Jet-in-air System name

Manufacturer

Web info

BD Influx

Becton Dickinson

www.bd.com

MoFlo Astrios EQ

Beckman Coulter

www.beckmancoulter.com

Bigfoot

Thermo Fisher

www.thermofisher.com

S3e

BioRad

www.bio-rad.com

Hydris

Cytonome

www.cytonome.com

Viva

Cytonome

www.cytonome.com

Cuvette-based System name

Manufacturer

BD FACSymphony A6

Becton Dickinson

www.bd.com

BD FACSDiscover S8

Becton Dickinson

www.bd.com

BD FACSMelody

Becton Dickinson

www.bd.com

BD FACSAria III

Becton Dickinson

www.bd.com

FX500

SONY

www.sonybiotechnology.com

SH800

SONY

www.sonybiotechnology.com

MA900

SONY

www.sonybiotechnology.com

Aurora CS

Cytek

www.cytekbio.com

Cytoflex SRT

Beckman Coulter

www.beckmancoulter.com

Large particle sorting Instrument name

Manufacturer

COPAS Series

Union Biometrica

www.unionbio.com

Closed microfluidic systems Instrument name

Manufacturer

TyTo

Miltenyi Biosystems

www.miltenyibiotec.com

CGX10 Cell Isolation System

SONY

www.sonybiotechnology.com

Gigasort

Cytonome

www.cytonome.com

Highway1 (beta-testing phase)

Cellular Highways

www.cellularhighways.com

Analytical cell dispensing systems Instrument name

Manufacturer

Wolf Sorter

NanoCellect Biomedical

www.nanocellect.com

F.Sight

Cytena

www.cytena.com

On Chip Sorter

On Chip Biotechnologies

www.on-chipbio.com

Practicalities of Cell Sorting

129

Differences in sensitivity between cuvette-based systems and jet-in-air systems are further exaggerated as jet-in-air systems typically must be optically aligned, either manually or automatically, each time the sheath flow is established or a nozzle is changed due to slight differences in the location of the jet. The chamber of a cuvette-based system is fixed in position and so alignment of the laser beam and core stream is only required to account for any drift in laser beam position. This ease of setup has widened the appeal of cuvette-based systems in many laboratories especially when the sorter is not situated in a core facility.

3

Drop-Based Sorting Once identified, cells of interest need to be separated from other cells in the sample. Drop-based sorters rely on the production of dispersed drops, the aim being that each drop should contain a single cell or no cell at all. Drops containing the required cells are separated, and the rest are sent to a waste tank. Several populations of cells can be collected simultaneously, with a high purity and recovery, and the sort should be completed quickly.

4

Sorting Principle of Drop-Based Sorters The process of cell sorting requires an open stream or jet to be created. This is done by passing a column of liquid through a small orifice (nozzle) where the drops are created by vibrating a piezoelectric device attached to the body of the flow cell. This applies a standing wave of vibration to the column of fluid. The vibration causes drops of equal size and spacing to form in the sheath stream at a given distance from the exit of the nozzle. Both the frequency (number of vibrations per second) and amplitude (amount of vertical movement of the piezo) can be adjusted by the sort operator. If the sheath pressure, vibration frequency, and amplitude remain constant, the break-off point of the drops from the stream remains at a constant and stable distance from the orifice. After a cell has been analyzed, a decision is made by the instrument on whether it meets sort criteria set by the operator. These “sort gates” can be set on any combination of scatter or fluorescence parameters. When a droplet is deemed to contain a target cell and reaches the stream break-off point, the fluid stream is charged during the drop formation. The formed drop retains the electric charge, while the fluid stream receives the opposite charge to be electrically neutral in time for the formation of the next drop. The charged drop passes through an electrostatic field created by a pair

130

Mark Cheetham et al.

Fig. 2 Summary of the electrostatic drop sorter. Shows the important aspects of the cell sorter including how droplets are formed, drop delay, charging of the droplets and deflection of the droplets through electrostatic plates. The images to the left show camera images of the droplets and sorting streams as seen on the cell sorter

of high-voltage deflection plates, and the drop containing the target cell is deflected into a collection container by this electric field (Fig. 2). By varying the charge (positive or negative) and the strength of charge (+, ++, +++) up to six populations can be sorted simultaneously. The sorted populations can be visualized by a camera as side streams at varying distances from the waste (uncharged) stream. For the correct drop to be charged and deflected, it is vital to know how long it takes for a cell to pass from the interrogation point to the point where it is in the last attached drop—this is known as the drop delay or drop charge delay.

5

How Is the Drop Delay Calculated? In the early iterations of cell sorters, drop delay was calculated by measuring the distance from the interrogation point to the breakoff point. However, this does not account for small changes in the stream or sample velocity, so in most current cell sorters, the calculation is performed using beads; a summary of the ways this is achieved is in Table 2. All methods of drop delay calculation aim to maximize cell yield, i.e., ensure as few wanted cells as possible are missed. It is sometimes useful, particularly when sorting a heterogeneous population of cells to calculate a metric “RMax” that allows the efficiency of the drop delay to be established [7].

Practicalities of Cell Sorting

131

Table 2 Different ways of calculating the drop delay, utilizing automated software and reagents supplied by the manufacturer of the instrument Measuring the distance from the laser to the break-off

Monitoring the side streams

Counting beads in the waste or side stream

Monitoring the last attached drop area

Accudrop™ method employed on BD Aria and BD Influx Systems

Pro-drop™ method employed on the Biorad S3, Bigfoot Drop Delay Software

Employed on the SONY Droplet Sorting systems

Employed on the Beckman Coulter MoFlo Astrios and CytoflexSRT

Specific bright beads are loaded onto the instrument as a sample A video image is used to compare waste and side stream brightness The software sorts the beads and runs through a range of drop delays When the side stream is at its brightest and the waste stream is at its dimmest the delay is optimal

Specific bright beads are loaded on to the instrument as a sample A laser-based particle counter is built into the waste catcher (S3) or side stream catcher (Bigfoot) Software runs through a range of drop delays Software sorts the beads at each delay System counts the beads in the waste, when there are zero beads in the waste the delay is optimal (S3). In the case of the Bigfoot the system counts the beads in the side stream

Specific bright beads are loaded onto the instrument as a sample A broad image laser flashes at the last attached drop, in time with the droplet charging Software runs through a range of drop delays Software sorts the beads at each delay Using a video capture image, when all the flashes accumulate in the last attached drop only then the delay is optimal

Does not use any beads for auto delay setting, but beads may be necessary to optimize delay fully A video image is used by the system software to ensure the droplet breakoff is occurring in the required location for a given nozzle type For higher than 90% accuracy a manual drop delay should be carried out by sort beads onto a side and checking the bead count at each delay

6

Considerations for a Successful Cell Sorting Experiment The size of the sorter nozzle should be approximately 4–5 times the size of the largest cell in the sorting sample (not just the cell that is being sorted). In practice, this means that for lymphocytes and splenocytes, a 70 μm nozzle would be used and for larger epithelial cells or cell lines such as HeLa, a 100 or 130 μm nozzle is used. Larger nozzles use more fluid and generate fewer drops per second resulting in a slower sort and larger post sort volume. Using an inappropriately small nozzle may lead to clogging of the orifice, spraying of side streams, and an unstable drop delay (resulting in lower yield). Nozzle size therefore impacts the frequency at which the cell sorter operates which relates directly to the speed of cell sorting. A 100 μm nozzle will typically use frequencies between 30 and 40 kHz, i.e., will produce 30,000–40,000 drops per second. The

132

Mark Cheetham et al.

sorting speed should not exceed one quarter of the frequency resulting in events per second (EPS) rates between 7500 and 10,000. Running at higher speeds will result in a lower yield because the sort mode (see below) often dictates that there should not be cells (whether of interest or not) in adjacent drops. Choosing the correct frequency for the nozzle involves finding the highest frequency, lowest amplitude, and most stable break-off point. As each nozzle will have its own unique characteristics and will have several harmonic frequencies (stable break-off point), this will need to be determined on a per-sort basis. The most important factor is stability although the higher the frequency, the more drops are produced and the faster the sample can be run.

7

Sort Modes Cells do not arrive evenly spaced out at the interrogation point which means that there can often be more than one cell in a drop. As the arrival at the interrogation point follows a Poisson distribution, this is why it is recommended that cells are analyzed at a speed of approximately ¼ of the drop drive frequency, i.e., the number of drops produced per second. For example, if the frequency is 60,000 drops per second, an event rate should be no higher than 15,000 cells per second. In this way, we know that only approximately 2.5% of drops will contain more than 1 cell. Additionally, although cells are interrogated at a known point, their arrival at the break-off point is uncertain and follows a Gaussian distribution, so a cell may be slightly ahead or slightly behind where it is expected to be. To take both these uncertainties into account, sorters will use different sort modes to maximize the chance of sorting a wanted cell. A sorter will consider every single drop and decide whether it can sort it or not and whether it does so will depend on which sort mode has been selected. Broadly, there are three sort modes; although their names may differ depending on manufacturer, all are designed to maximize either purity, recovery (enrich or yield), or count accuracy (single cell). The most common sort mode is Purity, and as the name suggests, this produces as pure a population of interest as possible. In most cases, this mode will sort one drop at a time but if the wanted cell is toward the leading or trailing edge of the drop, the sorter will charge and sort two drops (as long as there are no unwanted cells in that leading or trailing drop as this would compromise purity). In the event of there being two cells in one drop but of conflicting decisions, then the drop would not be sorted— this is referred to as a coincident, or software, abort. In some situations, a maximum recovery is needed, i.e., it is desired to sort all relevant cells at the expense of a reduction in purity. In this case, an enrich mode would be used—any drop that

Practicalities of Cell Sorting

133

Fig. 3 Examples of sort/no sort decisions with different sort modes on a Becton Dickinson FACSAria cell sorter. Each drop is divided into a number of segments (usually 32 but this depends on the sorter). The drop closest to the break-off is the leading drop, the drop with the cell of interest is the target drop, and the drop behind is the trailing drop. With purity mode, if the cell of interest (green) is somewhere in the target drop and an unwanted cell (orange) is in a particular distance away from it, the drop will be sorted. However, if the cell of interest and the unwanted cell are too close, for example, in the same drop, the drop will not be sorted. In enrich mode, if the cell of interest is close to the border and there are no contaminants in the trailing drop, both drops will be sorted. However, if the cell of interest is away from the drop border, the sorter will make a decision to sort just this one drop. In single cell mode, the cell of interest needs to be precisely in the center of the drop to be sorted. Any deviation from the center will cause an abort and a no sort decision. This mode is used for single cell deposition into a multi-well plate as it ensures that only one drop is sorted

contains a cell of interest—or might contain a cell of interest—is sorted so no wanted cells are lost. For single cell sorting or a sort when an accurate count is needed, the single cell mode would be used. In this mode, drops are only sorted when the cell of interest is in the middle of the drop and there are no unwanted cells in the preceding and following drop. This is generally the mode that is used when cells are sorted for downstream genomics applications such as scRNASeq. These sort modes are illustrated in Fig. 3. At the end of a sort, it is important to determine its success. There are three criteria by which this success is judged—purity, cell recovery, and yield. Figure 4 shows how to judge the success of your sort.

134

Mark Cheetham et al.

Fig. 4 How to judge the success of your sort. Purity is defined as the number of desired sorted cells compared to the number of sorted cells expressed as a percentage. In an optimized sort, this should be greater than

Practicalities of Cell Sorting

8

135

How Long Does It Take to Sort Your Target Population Assuming a sort rate of 20,000 cells per second, Table 3 gives an overview of expected sort times using a 1–2 drop purity mode. For example, if the cells of interest contribute 0.1% of the total population and you would like 1,000,000 cells, this will take 13.8 h to sort. This is for guidance only as it will be sample- and cytometerspecific.

9

Bio-containment All sorts should be subject to a risk assessment which will determine the containment level that is needed for a particular cell type or experiment. This risk assessment should be carried out in conjunction with the local health and safety department and will follow both institutional and governmental regulations. Cell sorting technology produces small aerosols which can easily be inhaled. Therefore, particular focus not only needs to be given to aerosol management in normal operation but also when blockages occur resulting in extra high levels of aerosols. For specific discussion of these topics, refer to Cohen et al. (Chapter 22, this book) [8, 9].

Table 3 Sort times Percentage of population # cells requested 1000

0.01% 8.3 min

0.1%

1%

5%

50 s

5s

1s

10%

40%

10,000

83.3 min

8.3 min

50 s

10 s

5s

100,000

13.8 h

83.3 min

8.3 min

100 s

50 s

12.5 s

13.8 h

83.3 min

16.6 min

8.3 min

125 s

1,000,000

5.75 days

ä Fig. 4 (continued) 99%. Normally, this would be assessed by re-running a small sample of the sorted population through the sorter immediately after the sort. Cell recovery is the number of cells actually in the sorted tube divided by the number that the sorter says it has sorted. This takes into account any cell loss (cells missing the sort tube, cell death) during the drop deflection process. Typically, this should be greater than 90%. Importantly, cell counts should be made before any manipulation of the sample, e.g., centrifugation. Yield (sometimes also called recovery) is the proportion of desired cells sorted compared to the total population of desired cells in the sample pre-sort. Some cells will be lost to hardware or software aborts, but a sorter should have greater than 70% yield

136

10

Mark Cheetham et al.

Quality Control (QC) All flow cytometers will need a QC procedure to ensure that lasers are in alignment to the stream and that the sensitivity of detectors is within normal range. Cuvette-based systems do not require alignment on a daily basis, but jet-in-air cell sorters will require alignment of the lasers to the stream and the stream to the detection lenses. In some newer cell sorters, this process is automated (e.g., ThermoFisher Bigfoot), but in some this will require manual alignment (e.g., Moflo XDP and BD Influx). In addition to automated QC procedures or on systems with manual alignment, a simple QC procedure using alignment beads (e.g., Spherotech 8-peak beads) should be used after instrument start-up. Using reference beads allows verification of the laser alignment, signal detection on all scatter and fluorescent parameter, and sensitivity of all detectors. A rigorous QC regime not only helps to identify sudden hardware failures but also helps to identify small changes over time (e.g., detector failures, deteriorating laser power).

11

Cell Preparation and Sort Buffers Sample preparation is perhaps the most crucial aspect to a successful sort. Flow cytometry is a single cell technology, and we want to ensure that we make single-cell measurements. Although hydrodynamic focusing will help with this, it is imperative that a robust sample preparation protocol is in place. This will vary depending on the cell or tissues that are being studied. Cells that grow in suspension or peripheral blood are relatively easy but adherent cells, primary tissue, frozen biopsies, and paraffin embedded material will be much more troublesome. It is beyond the scope of this chapter to discuss all the sample types that can be used but example protocols for peripheral blood mononuclear cells (PBMCs) and adherent cells are given below.

12

Cell Preparation for Cell Sorting

12.1 Protocol: PBMC (Peripheral Blood Mononuclear Cells)

1. To a 10 mL round bottom centrifuge tube, add 3 mL of Histopaque-1077 (Sigma-Aldrich) and bring to room temperature. 2. Prepare one tube for each 3 mL of whole blood. 3. Carefully layer 3 Histopaque-1077.

mL

of

whole

blood

onto

4. Centrifuge at 400 × g, 30 min, 20–22 °C, lowest brake.

the

Practicalities of Cell Sorting

137

5. Carefully transfer the opaque interface (the MNCs) with a blunt needle on a 5 mL syringe into a clean conical 15 mL centrifuge tube. It is important to first harvest the cells in the periphery, and then the rest of the cells. 6. The interface gets clear when all cells are harvested. 7. Transfer cells from two tubes to one conical tube (transfer as little plasma as possible by placing the needle close to the interface). 8. First wash: Add PBS/2% FCS/1 mM EDTA to a total volume of 15 mL. Mix by gentle aspiration. 9. Centrifuge at 300 × g/10 min/room temperature (RT)/ medium brake. (The pellet can be very loose, due to the Histopaque. If the pellet is not solid, centrifuge at 400 × g/ 10 min/RT.) 10. Carefully decant the supernatant and discard. 11. The last (second) wash: Add a small volume of PBS/2% FCS without EDTA to each pellet. 12. Resuspend the pellets and transfer all pellets to one tube. Add PBS/2% FCS without EDTA to this tube to a total volume of 15 mL. 13. Mix by turning the tube by hand twice. Count the cells. 14. Centrifuge at 150 × g/12 min/RT/medium brake (this wash removes thrombocytes). 15. Decant the supernatant and discard. 16. Resuspend the cells in PBS/2% FCS or PBS/1% BSA at the desired concentration. Stain with fluorochrome-conjugated antibodies of interest for 20 min at RT in the dark. 17. Wash twice with 5 mL PBS/1% BSA (centrifuge at 300 × g, 5 min). Discard the supernatant. Resuspend in 300–500 μL PBS/1% BSA and cells are ready to be sorted. Keep cells at 4 °C until the sort starts. 12.2 Protocol: Adherent Cells from Cell Line

1. Loosen your adherent cells according to the cell line (e.g., remove the cell culture medium. Rinse the cells with PBS. Add a solution of, for example, trypsin for 5–10 min at 37 °C or Lidocain for 5–15 min at RT). Keep an eye on the cells under a microscope. 2. When most cells have detached, transfer the cell suspension to a 15 mL tube, add 4 mL non-stick buffer,1 centrifuge at 300–400 × g for 7 min (centrifugation force and time depends on cell type/size).

1

Non-stick buffer: PBS or HBSS supplemented with 1% BSA, 2.5 mM EDTA and 25 mM HEPES buffer.

138

Mark Cheetham et al.

3. Discard the supernatant. Resuspend your cells in 6 mL non-stick buffer and centrifuge again at 300–400 × g for 7 min. 4. Discard the supernatant and resuspend the pellet in a suitable amount of non-stick buffer. If you are working with transfected or transduced cells, you may sort them directly from this step. It is recommended to add a viability dye to your cells (e.g., PI (propidium iodide) or Draq7) to exclude dead cells during your sort. 5. If you need further staining of your cell line, add the antibodies of interest to your cell pellet (200–300 μL). Mix cells and antibodies well. 6. Incubate in the dark at room temperature for 20 min or 30–45 min at 4 °C. 7. Wash twice using 10 mL non-stick buffer (centrifugation at 300–400 × g). 8. Resuspend in a final volume of non-stick buffer (about 10 × 106 cells/mL). 9. Bring the cells for sorting on ice. 10. Add a viability dye immediately before sorting (e.g. PI or Draq7). With all sorting, it is important to complete the sort in as short a period of time as possible to get cells back to a favorable environment or into their downstream assay. Anything that can be done to keep cells happy is important, so deciding on both the sample buffer and collection buffer is critical. Protocols often use an ill-defined “FACS buffer”, but it is important to understand why this matters to your sort. Cells can be in culture medium—sometimes it is best if this is phenol red-free as phenol red is fluorescent and can lead to a reduction in the signal to noise ratio. A small amount of protein will also be beneficial as it helps keep cells apart which will facilitate the single cell measurements and also provides a happy environment for the cells. However, too high a protein concentration can cause issues with light scatter and can also lead to clogging of sample lines. A typical concentration would be about 2% FCS or 1% BSA. If cells tend to stick, or if there is appreciable cell death adding EDTA or DNase to the sort buffer is useful. In addition, using a buffer that is good at maintaining a physiological pH, e.g., HEPES is recommended.

13

Cleaning Before starting a cell sorting experiment, the post-sort application should be considered. For cultivation of cells after a sort, the cell sorter requires to be as aseptic as possible, whereas for most genomic applications, cell sorter sterility is less important. In any case, a

Practicalities of Cell Sorting

139

regular (daily, weekly, and monthly) cleaning and maintenance protocol needs to be implemented for any type of cell sorter. With some cell sorter systems, cleaning protocols are already part of the regular system start-up and shut-down (e.g., Melody, Aria, Bigfoot) but other systems give more flexibility in sorter cleaning procedures. Mostly this will involve cleaning the lines and sterilizing them as well as cleaning the sheath tank and possibly replacing inline sheath filters. If full sorter sterility is needed for an application, only systems with disposable cartridges (e.g., Miltenyi Tyto and SONY CGX10) can be used. No cuvette-based or jet-in-air system currently on the market allows for full sorter sterility. At the end of a sort, the sorter needs to be cleaned and de-contaminated. Any part of the cell sorter that has been in contact with the cell sample needs particular attention. Standard procedure would be to run detergent and bleach through the sample line and cleaning the nozzle or flow cell of the sorter. Many modern cell sorters have a cleaning and shut-down procedure built into the software. Nevertheless, it is good practice to clean the instrument prior to shutting it down. 13.1 A Generic Cell Sorter Cleaning Protocol Would Be

1. Prepare a fresh tube of sodium hypochlorite (the concentration depends on the manufacturer’s recommendation) and load it onto the cell sorter. Run the tube on a high flowrate for 5 min. 2. Prepare a tube with detergent solution (e.g., Decon90 diluted to manufacturers recommended concentration). Unload the tube with sodium hypochlorite and load the detergent tube. Run this on a high flowrate for 5 min. 3. Remove the tube with detergent and prepare a tube with filter H2O. Unload the tube with detergent and load the tube with H2O. Run this for 5 min. 4. After this cleaning protocol, shut the sorter down following the manufacturer’s shut-down protocol.

13.2 Cleaning Before a Sort for Cell Culture (Aseptic Sorting)

1. Load a tube with 70% ethanol and run at highest flowrate for 20 min. Unload the tube.

13.3 Cleaning Before a Sort for RNA Purification

1. Use gloves at all times and clean workspace, pipettes, and sort chamber with RNaseZap RNase decontamination solution (ThermoFischer, cat. no. AM9780).

2. Load a tube with sterile filtered H2O and run at highest flowrate for 10 min. Unload the tube.

2. The temperature control for both sample and collection device must be set to 4 °C. 3. Load a tube with 10% bleach and run at highest flowrate for 5 min. Unload the tube.

140

Mark Cheetham et al.

4. Backflush the sample line for 30 s. 5. Load a tube with sterile filtered H2O and run at highest flowrate for 5 min. Unload the tube. 13.4 Cleaning Between Users or Before Running the Instruments Shutdown Procedure

14

1. Load a tube with a detergent (such as 15% Contrad) and run at highest flowrate for 5 min. Unload the tube. 2. If DNA bindings dyes have been used, load a tube with 10% bleach, and run at highest flowrate for 5 min. Unload the tube. 3. Load a tube with filtered H2O and run at highest flowrate for 5 min. Unload the tube.

Example Protocol for Sorting GFP Transfected Cell Lines 1. Create plots for FSC-A vs SSC-A, FSC-H vs FSC-A, and SSC-A vs SSC-H. A viability plot should be set up for the viability dyes vs SSC-A. Gates should be applied to show single, viable cells. 2. Analysis template for setting sort gates should include a dot plot showing GFP (probably 530/30 488 nm) vs. Autofluorescence (usually 580/40 488 nm or equivalent). A dot plot should be used to allow for autofluorescent cells to be visualized and more accurate gating to be applied to the GFP positive cells (Fig. 5). 3. Care should be taken when sorting negative and positive populations to have an adequate gap between the gates to maximize purity especially with overlapping populations.

15

Evaluation of Performance When Sorting into Plates and PCR Tubes Sorting into 96-well plates is not difficult. Typically, you would test the sort precision on the lid of the 96-well plate and finally sort a single cell into 100–200 μL medium. Compared to the size of the drop, the 96-well is quite wide. When sorting single cells into either 384-well plates (or 1536-well plates) or into PCR tubes, the precision must be very accurate to make sure the sorted drop enters the center of the well or tube and is not deposited on the side. Using an enzymatic test to evaluate the precision is both fast an accurate; the procedure was first described by Rodrigues and Monard [10]. 1. Set up your sorter for single cell sorting. For the drop delay, make sure the precision (CV) is 98% or higher. 2. Add 2 μL TMB (3,3′,5,5′-Tetramethylbenzidine) high sensitivity substrate (BioLegend, cat. no. 421501) into the bottom of all wells/tubes you would like to test.

Practicalities of Cell Sorting

a

b 150

GFP+ 70.0

GFP+ 1.82 Count

Count

400

GFP– 98.2

GFP– 30.0

141

300 200

100

50

100 0

0 0

103

104

105

530_30 488nm-A :: GFP

104

105

530_30 488nm-A :: GFP

c

d 105

105 GFP, Auto subset 80.5

104

585_42 488nm-A :: Auto

585_42 488nm-A :: Auto

103

0

103

102 0

104

103

GFP, Auto subset 1.15

102 0

0

103

104

530_30 488nm-A :: GFP

105

103

0

104

105

530_30 488nm-A :: GFP

Fig. 5 Histograms utilized for sort regions. (a) Shows GFP negative cells on a 530/30 488 nm histogram plot. (b) Cells transfected with GFP showing a heterogenous expression of GFP. (c) Shows a dot plot of the same cells as in (a) but using the dot plot of 530/30–488 nm vs 585/42–488 nm to show autofluorescence. (d) Utilizing the same dot plot, the heterogenous GFP cells in (b), can now clearly gated as GFP negative and GFP positive and gives a much more accurate percentage of GFP positive events (80.5% compared to 70% as shown in the histogram in (b))

3. Use your sorter’s wizard or software tool to calibrate the instrument, e.g., 384-well plate sorting. 4. Place the 384-well plate in the sorter. 5. Use the software to set up the single cell sort. You may choose to sort a single bead in each well, or you may choose different amounts of beads in different columns of the plate. You will be able to see a color difference between sorting 1 and 2 beads.

142

Mark Cheetham et al.

Fig. 6 (a) Test of single bead sorting into 384-well plate. Column 1, 1 bead; column 2, 2 beads; column 3, 5 beads; column 4, 10 beads; column 5, 0 beads and so forth. (b) Test of single sorting into PCR tubes. Tubes 1–3, 1 bead; tube 4, 0 beads. (c) Test of sorting a single bead into each well of a 384-well plate. To each well, 2 μL TMB was added. It can be appreciated that 20 wells are clear (= not blue), meaning the success rate for this sort is expected to be 364/384 = 95%

6. Mix 1 drop of beads (Polybead Carboxylate Microspheres, Polysciences Inc. PA, 6 μm cat. no. 17141-5, or 10 μm cat. no. 18133-2) with 500 μL 100 μg/mL HRP (horse radish peroxidase, ChemCruz, cat. no. sc-280786).

Practicalities of Cell Sorting

143

7. Start acquisition of the beads in HRP solution, place a gate around single beads. Define this as the sort gate. Start the sort. 8. When the sort is done, incubate the plate for 5–10 min in the dark. Look at the plate on a white background. Results of these tests can be seen in Fig. 6 where the color changes indicate the presence or absence of a bead. References 1. Shapiro HM (2003) Practical flow cytometry, 4th edn. Wiley-Liss, New York 2. Melamed MR, Lindmo T, Mendelsohn ML (1990) Flow cytometry and sorting, 2nd edn. Wiley-Liss, New York, p 824 3. Ortolani C (2022) Flow cytometry today: everything you need to know about flow cytometry. Springer, Cham, p 544 4. Herzenberg LA, Parks D, Sahaf B, Perez O, Roederer M, Herzenberg LA (2002) The history and future of the fluorescence activated cell sorter and flow cytometry: a view from Stanford. Clin Chem 48(10):1819–1827 5. Sweet RG (1965) High frequency recording with electrostatically deflected ink jets. Rev Sci Instrum 36(2):131–136 6. Fulwyler MJ (1965) Particle separator. US Patent US3380584A

7. Riddell A, Gardner R, Perez-Gonzalez A, Lopes T, Martinez L (2015) Rmax: a systematic approach to evaluate instrument sort performance using center stream catch. Methods 82:64–73 8. Schmid I, Merlin S, Perfetto SP (2003) Biosafety concerns for shared flow cytometry core facilities. Cytometry A 56(2):113–119 9. Aspland A, Chew C, Douagi I, Galland T, Marvin J, Monts J et al (2021) Risk awareness during operation of analytical flow cytometers and implications throughout the COVID-19 pandemic. Cytometry A 99(1):81–89 10. Rodrigues OR, Monard S (2016) A rapid method to verify single-cell deposition setup for cell sorters. Cytometry A 89(6):594–600

Chapter 8 Image-Enabled Cell Sorting Using the BD CellView Technology Malte S. Paulsen Abstract This chapter is an extension of the original publication by Schraivogel et al. (Science 375:315–320, 2022) which described, for the first time, image-enabled and high-speed cell sorting based on the BD CellView technology. It summarizes the technical aspects of the instrument in an easy-to-digest form and provides example-based guidance toward implementation of the CellView-based image cell sorting technology. As an example, it explains how to use the image-enabled cell sorter to analyze the chemically induced fragmentation of the Golgi apparatus in HeLa cells—an experiment that was alluded to in the original publication but was not included in the manuscript due to space constraints. The chemically induced Golgi fragmentation sort illustrates an elegant example of the utility of image-enabled cell sorting as a significant expansion of the single-cell toolbox. It is such a striking phenotype when analyzed with image cytometry but undetectable when using conventional flow cytometry. Described in a straightforward and concise manner, this experiment serves as a standard system assurance for image-based cell sorters. Key words Image-enabled cell sorting, On the fly image analysis parameters, Chemically induced Golgi fragmentation, High-speed cell sorting, Flow cytometry

1

Introduction

1.1 The Rationale for Using Image-Enabled Cell Sorting

Many critical cellular processes and phenotypes are not regulated by an up-or-down change in protein abundance and hence lack a suitable biomarker enabling traditional fluorescence-activated cell sorters (FACSs) to isolate specific phenotypes from a sample. Such processes could be the fragmentation of a cell organ, the translocation of a transcription factor or signaling cascade member into the nucleus or cytoplasm, or simply the separation of DNA during mitosis. Isolation of such spatially encoded phenotypes has only been possible using powerful yet very slow laser-microdissection capture methods [1] or larger microscopy-microfluidics hybrid systems [2]. Compared to FACS instruments that can isolate cells at speeds of several thousand events per second, microdissection methods deliver around 1–10 cells per minute at best, while some

Teresa S. Hawley and Robert G. Hawley (eds.), Flow Cytometry Protocols, Methods in Molecular Biology, vol. 2779, https://doi.org/10.1007/978-1-0716-3738-8_8, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

145

146

Malte S. Paulsen

microfluidics-based systems are getting closer to stably enabling 500–1000 cells/sec throughput. Modern genetic screening methods based on CRISPR alone and in combination with other conditional factors, such as drugs or targeted biologics, have been hampered in achieving their full technical potential due to the lack of high throughput when sorting millions of target cells is a prerequisite for large-scale screens. A genome-wide CRISPR-library screen, utilizing six gRNAs per gene, should harvest around 70–100 cells per gRNA to ensure robust outputs. This means that a 20,000 gene CRISPR KO screen with a 5% target population for a specific phenotype will require the analysis of 28–40 million cells to sort out 1.4–2.0 million target cells with a desired phenotype. The need for a high-throughput approach is obvious, especially when dealing with unfixed cells that may have a transient response to a conditional stimulus. The image-enabled cell sorting setup published by our team and Becton Dickinson can overcome the throughput gap in imageenabled cell sorting and connect many spatially encoded screening methods with the versatility of state-of-the-art genetic and proteomic analysis spheres [3]. 1.2 BD CellView Image-Enabled Cell Sorting Technology

The technology is rooted in the concept of the FIRE technology originally described by Eric Diebold [4]. A single-frequency laser beam is passed through a Mach–Zehnder interferometer containing opto-acoustic deflectors and frequency combs. The output of a perfectly tuned system is a laser beam split into 120 individual beamlets building a comb array. Within the comb array, each beamlet has its specific frequency in the MHz range—usually increasing between 1 and 100 MHz. The beam array can be optically aligned with other traditional, but non-frequency modified lasers, into a cuvette used within many conventional sorters today. The final spread of the beamlet comb is controlled by an objective with a lower numerical aperture to allow for a broader range of cell sizes to be analyzed. This results in a field of view of about 60 × 60 μm. The speed of the sheath flow is lower than what can be attained in conventional cuvette sorters, about 1 m/s compared to 6 m/s. This slightly reduces the throughput capability compared to nonimage-based FACS instruments but allows for the generation of more fluorescence signals when a cell passes through the beamlet comb. Increased dwell time of the cell in the laser comb results in more fluorescence cycles per dye. This enhances the instrument’s sensitivity—a necessity as each beamlet has only 1/240th or so of the original power of the laser after passing through the beamlet generator. The CellView technology (Fig. 1) enables the creation of an image of the cell as it passes through the beamlet combs. Each beamlet has a specific frequency that can be decoded from the event waveform as the photon deluge is transferred into an electrical

Image-Enabled Cell Sorting Using the BD CellView Technology

147

Fig. 1 Overview of the working principle of the CellView technology. A monofrequency 488 laser is split into two beams that are optoacoustically modified to generate a clean comb of 120 beamlets each carrying a specific MHz-based frequency. The frequency-resolved beamlet comb is aligned in the flow cell of the cuvette. It generates event specific “waveforms” carrying positional and intensity information (x coordinate and its respective intensity) and from scatters and fluorescence in the cell in the frequency and intensity domain. The process is ultrafast as it uses PMTs instead of a camera. The analysis is coupled with onboard FPGA systems that process a set of image analysis parameters on the fly for sorting decisions. The image was kindly provided to the authors with permission to reprint from BD Biosciences (San Jose, CA, USA)

signal via a frequency detection-enabled PMT. This unique feature makes the technology faster and more sensitive than a camera using a similar dwell time. As the spatial information is coded in the frequency domain and the time-of-flight parameter (the cell moving through the thin beamlet comb in the direction of the flow), it is possible to obtain spatially resolved image information using ultra-fast onboard field programmable gate arrays (FPGAs). As a lot of sophisticated coding and signal processing occur in the background, this technological advance is less transparent than the straightforward FACS technology of the last five decades. The outcome is a data package that first calculates a predefined set of imaging analysis parameters on the fly that is fast enough for cell sorting decisions and then reconstructs an image for the different fluorescent channels as well as scattering/light loss parameters while retaining classical intensity data. With the current technical setup installed on the commercial BD FACSDiscover S8 unit (BD Biosciences), the CellView imaging technology has a resolution limit of 1.5 μm per pixel. It is essential to keep this in mind as any signal strong enough to pass the threshold of the parameter mask and is smaller than 1.5 μm × 1.5 μm will result in a 1.5 μm × 1.5 μm pixel whether it is 0.5 μm × 0.5 μm or 1.5 μm × 1.5 μm. The resolution limit is a hard border and guides the usability of the technology to certain areas of biology.

148

Malte S. Paulsen

1.3 Sort-Enabled Image Analysis Parameters

As mentioned above, the CellView technology achieves its analysis and sort speeds by exploiting a preset list of broadly applicable image-analysis parameters hard-coded into the system. Although the list is short, we have found eccentricity, max intensity, size, radial moment, diffusivity, and the correlation of two fluorescent channels very useful when looking at a diverse set of biological questions (Fig. 2). This list of preset image-analysis parameters may not be as elaborate as analysis options on other imaging flow cytometry technologies, but these parameters are entirely sortenabled. In addition to being available for live sorting at speeds exceeding 10,000 events/sec, they are very adaptable to describing biological phenomena such as separations, colocalization, dissolving of structures, cell shapes, or even polarity changes in cells. Combining the parameters increases the descriptive power of the system and enhances the overall functionality. This is relatively easy for midlevel experienced users of flow cytometry as all imaging parameters are described numerically and can be plotted in traditional ways using classic flow cytometry plots and gating strategies while cross-referencing the images of a specific analysis gate. They can be readily combined with the other classic flow cytometry parameters available on the instrument, facilitating the expansion of existing and established protocols with imaging—a valuable design feature for evaluating initial experimental questions. The technology is even more powerful when used in combination with machine learning and unsupervised clustering methods when analyzing training datasets offline. This is currently not possible to do live on the system. Nonetheless, our previous study on cell cycle sorting showed the ease of including the offline clustering analysis into a daily workflow to enhance gating strategies beyond the initial manual gating strategy [3]. The sorting mechanism is based on the long-established BD cuvette sorter. However, in contrast to a traditional cell sorter, high-pressure sorts using a 70 μm nozzle are impossible on the imaging system as the necessary stable sheath speed is not ideal for imaging. Otherwise, sorting with the CellView-enabled technology on the FACSDiscover S8 is very similar to sorting with the FACSAria or FACSymphony S6 cell sorters. It is capable of sorting multiple target populations, or single-cell deposition on slides and multi-well plates. The only difference compared to the Aria and S6 is that the basic cytometry suite of the S8 is based on a fullspectrum detection model aside from the separate imaging laser system. Samples for the image sorter should be of very high quality, i.e., devoid of aggregates. The system is slightly less tolerant to cell clusters and large debris even though doublets and triplets are nicely tolerated. When handling cell culture cells with a size of around 15–20 μm/cell, enormous debris clouds can potentially cause stress on the stream. If the samples are carefully prepared,

Image-Enabled Cell Sorting Using the BD CellView Technology

149

Fig. 2 Overview of the most functional image-analysis parameters available on the BD FACSDiscover S8 for every imaging-enabled channel. The figure explains the underlying logic of the hard-coded eccentricity, max intensity, size, radial moment, correlation, and diffusivity parameters in the BD FACSDiscover S8 image sorter. The individual drawings are slightly exaggerated but serve as an excellent reference to have at hand when setting up and evaluating the image parameters needed to describe a phenotype. Of note, correlation and diffusivity are parameters that use the essential light-loss cell mask as a reference to limit the boundaries of the analysis field

high purity can be achieved with sorts around 5000–15,000 events/sec. This sorting throughput is comparable to conventional cell sorting. 1.4 Example of a Novel Application: Measuring and Sorting Out Chemically Induced Golgi Fragmentation in GalNac-GFP HeLa Cells

1. Cell culture, cell perturbation with Golgi-Plug, and cell suspension generation. (a) HeLa cells with the Golgi stack enzyme N-acetylgalactosaminyltransferase-2 (GalNAc-T2) fused to the green fluorescent protein (GFP) GalNac-T2-GFP were a gift from the Pepperkok team at EMBL that specializes in microscopy methodology and molecular transport biology [5]. The GalNac-T2-GFP specifically labels the Golgi apparatus and, to a lesser extent, the endoplasmic reticulum if the Golgi is unperturbed or the cell is not going through mitosis. The cells were cultured in DMEM high glucose, supplemented with 10% FCS, 1% PenStrep, and 1% sodium pyruvate at 37 °C and 5% CO2. Cells were passaged every second day using TrypLE and re-seeded at a split ratio of 1:10. (b) Chemical treatment with Golgi-Plug: 60% confluent 25 cm flasks of HeLa-GalNac-T2 GFP cells were treated for 0.5 h with Golgi-Plug (BD Biosciences), a brefeldin A-based inhibitor reagent, to induce the dispersion perturbation of the Golgi to be analyzed with the image sorter. At this time point, 10–30% of the cells started to show the dispersion phenotype with a large proportion of

150

Malte S. Paulsen

Fig. 3 Effect of increased positivity concerning background contribution in images. A low-intensity signal, and hence lower separation of positive cells from autofluorescence, has a higher chance of resulting in more background signal contribution to the image data. This effect can be offset if the desired phenotype is a very small and condensed point-like structure with all of the signals concentrated into this area, similar to a geneloci FISH. Otherwise, with a pan-cytosolic distribution for the protein of interest, selecting bright clones for any project aimed at image-enabled cell sorting is advisable

intact Golgi remaining in the sample and many cells displaying an intermediate stage of dispersal with multiple bright signals in a cell. Mock cells were treated with the same concentration of DMSO vehicle. The reporter is very bright, making it easy to use a fluorescence-enabled cell culture microscope to evaluate the dispersion process.

(c) Single-cell suspensions from treated and control cells were generated with TrypLE dissociation for 5–10 min at 37 °C and 5% CO2. Before spinning the cells down at 100xg for 5 min at room temperature and resuspending them in ice-cold PBS supplemented with 0.1 μg/ml DAPI, the

Image-Enabled Cell Sorting Using the BD CellView Technology

151

Fig. 4 Initial gating strategy to identify cells with intact or dispersed Golgi. Top row displays the standard gating strategy for single cells using conventional singlet/doublet gating strategy utilizing H/W on FSC (lightloss) and SSC signal. The CellView image-enabled SSC eccentricity and radial moment parameters interpret the image information allowing sensitive doublet detection beyond the conventional strategy. Middle row focuses on the gating strategy utilizing conventional flow and imaging analysis parameters, distinguishing cells with intact and very condensed Golgi from those with dispersed Golgi. Image-wall examples highlight the gating strategy’s purity using the cells’ images within the final gates. Bottom row illustrates the choice of imaging parameters used for the initial acquisition correlated with the information in Fig. 2

Fig. 5 Setting up thresholds for the imaging channels. This example illustrates the manual process of setting up the threshold for noise on the light-loss channel. Finding the correct balance between a too-high or too-low threshold is essential. The technology will have to improve as this step is user-subjective, and little support is given to correlate the black/white threshold image with the base image. As the threshold affects the calculation of the imaging-analysis parameters, users should invest time in determining the threshold when dealing with new setups

cells were carefully but thoroughly mixed by pipetting with quenching DMEM after the TrypLE digestion. This is important to ensure a very high number of single cells and to reduce the amount of larger cell clusters which might stress the image sorter. Samples were passed through 30 μm mesh filters (blue cap tubes) prior to loading on the image sorter. The cells were agitated at 300 rpm and kept at 4 °C while in the sample injection chamber.

152

Malte S. Paulsen

Fig. 6 Improved gating strategy removes the need to homogenize input levels of GFP intensities for robust Golgi phenotype detection. Top row displays the standard gating strategy for single cells using classical H/A gates. Middle row focuses on improving the Golgi phenotype’s image analysis. Note that the analysis of GalNac-T2-GFP “Size” vs. “diffusivity” utilizes the entire intensity spectrum of the positive reporter cell population. Bottom row shows the final sort gates differentiating between small and condensed, elongated and condensed, and diffused Golgis within all input cells. FlowJo with Lens was utilized to analyze and plot data or to arrange cell-specific images

2. Setting up the image sorter to efficiently separate cells with either an intact or a dispersed Golgi. (a) Prepare the image sorter (either Prototype or FACSDiscover S8) for daily operation by strictly following the basic startup procedure described in the manual, but include an

Image-Enabled Cell Sorting Using the BD CellView Technology

153

Fig. 7 Reanalysis of purified, diffuse Golgi containing cells. Top row: left panel shows the 30 min Golgi-Plug challenged input sample used for image-based sorting of cells with a diffused Golgi; middle panel shows the pre-sort profile; right panel shows the post-sort profile after sorting out 50,000 diffused Golgi-containing cells. Bottom row shows the corresponding images. FlowJo with Lens was utilized to analyze and plot data, or to arrange cell-specific images

extra step of cleaning the cuvette with 2% Helmanex III (Sigma-Aldrich, St. Louis, MO) after the initial rebuffering from ethanol. Extra cleaning of the cuvette is necessary for the imaging quality, and the accompanying simple CST runs do not readily recognize any blurring. Good performance of the imaging laser/cuvette setup becomes apparent when flakes and tiny crystals can be well observed in the stream running BD FACS Clean reagent before any experimental work. Additionally, storing the image sorter in 0.2% Helmanex III or the BD cleaning detergent overnight (or the weekend) helps to ensure the

154

Malte S. Paulsen

performance level. Scrubbing the cuvette with detergent/ air bubbles is less frequently necessary. (b) Setting up the GFP imaging channel and the voltages to observe the Golgi is straightforward but requires focusing on the height parameter of the imaging laser to ensure that the signal is unsaturated—most events should be below channel 5 × 104. The calculated area signals of the imaging channels saturate later due to the calculation of the area under the peak and have a different scaling on the instrument. The separation of negative cells and positive cells on the height parameter is the most critical factor for good resolution using the image-analysis parameter. A half-decade separation from the negative cells gives slightly noisier data, whereas cells with a full log-decade of separation from the negative cells provide clean data for the imaging parameters (Fig. 3). However, very condensed signals, such as p53BP spots, can be seen even with smaller distance between unstained and stained cells. Nonetheless, if the fluorescence is more distributed over a larger area, then better separation is key (Fig. 3). (c) One of the major advantages of image cell sorting is the assurance of single cells using image analysis-based detection of doublets. To achieve this, the light-loss parameter can be exploited using eccentricity and radial moment to separate the doublets hidden in the data after applying traditional singlet gating strategies (Fig. 4). Traditional singlet gating is highly effective, but there are exceptions especially with large or tiny cells; adding image-based singlet gating has been shown to improve the standard in the field so far. The data in Fig. 3 were produced using the unperturbed, vehicle-treated sample as it provides a small number of dispersed Golgi in cells undergoing mitosis. Correctly setting up the detector voltage on the GFP imaging channel (ImgBlue1-A/H/W in CellView) is crucial in avoiding saturated events and allowing all cells to be correctly image analyzed. This is best done by ensuring that the signal height (ImgBlue1-H) of the positive population is set around 2 × 104 (Fig. 4). Placing the desired positive population in this region allows for more dynamic depth to describe the data and improves numeric and displayable resolution on the calculated parameters. It is necessary to set the threshold for each imaging parameter to calculate meaningful image-analysis data from the imaging channels. The system forces users to do this for the main “light-loss” image parameter during the setup of the experiment, but users must take care of the fluorescence-based parameters themselves before

Image-Enabled Cell Sorting Using the BD CellView Technology

155

recording any data and should check accuracy before sorting. The threshold should be set in such a way that it can detect low amounts of signal while robustly excluding background autofluorescence and non-sample associated noise (see Fig. 5 for an illustration based on the light-loss channel). There is only one threshold for each image channel, and the top end is limited by the resolution limit of the PMT that can be taken care of by staying away from the last bin on the respective “height” signal. (d) At this step, things start deviating from the traditional way of interpreting positive signals compared to standard flow cytometry. In order to separate cells with an intact Golgi from those with a dispersed Golgi, it is important to supply the onboard image analysis algorithms with cells having similar global intensities of the GalNac-TC2-GFP reporter. The “max intensity” parameter provides the intensity value of the brightest pixel within the image. Using cells with too broad range of GFP intensities (e.g., a broad selection of GFP positive events) may confuse the analysis algorithm because a bright but diffused Golgi could have a similar max intensity of the GFP signal as a dim condensed Golgi. Therefore, it is advisable to limit any initial gating strategy using a tight gate on the GFP intensity histogram to reduce the general spread of signal intensity. The threshold mask for the imaging channel does not have to be changed if it is based on the dimmest positive cells of the original sample. (e) Limiting the GFP intensity will provide a cleaner image analysis and make the initial gating attempts easier. Plotting GFP “max Intensity” and “diffusivity” of the GalNac-T2-GFP signal enables the direct identification of dispersed Golgi (Fig. 4) and concurrently allows using the cells with the most intense GFP signal for further image analysis to optimally resolve and select the most condensed Golgi specimen with higher resolution. This is done by plotting the GFP “size” and “radial moment” of the “max intensity” gated GFP cells. A low radial moment indicates a roundish character of the signal source, and a small size indicates a very condensed area that the GFP signal originates from, thus providing the means to gate out the desired intact Golgi population with high specificity. 3. Refining the gating strategy for an efficient and productive sort. The initial gating approach above could be directly used to sort out the desired phenotypes, yet it would be very wasteful as only 30–40% of the possible input was used due to the homogenization

156

Malte S. Paulsen

of the GFP input signal. We concluded that a first rough gating strategy was the right approach to initially formulate and record a good-sized data set which was then reanalyzed offline using either FlowJo-Lens or R-based tools to evaluate whether other gating strategies would be even more efficient (see ref. [3]) for the cell cycle gating strategy). Collecting 50,000 cells, including all images, resulted in a data file of about 10 GB. This technology’s data storage requirements must be considered when planning extensive experiments collecting large amounts of cells associated with images. Reanalyzing the data and using the recorded images to formulate the refinement of the gating strategy, we noticed that using the GFP “max intensity” parameter indeed allowed for the most accessible identification of condensed Golgi-containing cells, but at the cost of homogenizing the GFP intensity input. Replacing the GFP “max intensity” with the GFP “size” parameter resulted in a gating strategy independent of the GFP input gate and provided a similarly good description of condensed and dispersed Golgi containing cells. The resolution was also good enough to describe the elongated but intact Golgi phenotypes (Fig. 6). The offline-derived gating strategy was established on the S8 by redrawing the gates that were subsequently utilized to purify dispersed Golgi cells from a short-term Golgi-Plug treated culture. Cells were sorted using a 100 μm nozzle at 2000–3000 events/sec input speed. The sorting mode was set to purity, and the cells were collected in 5 mL FACS tubes containing 200 μL PBS at the bottom and chilled to 4 °C for the duration of the sort. The resulting purity of the sort was satisfying, with a purity of around 96% in the target gate and minor occurrence of events outside the generously drawn gates (Fig. 7). It should be noted that accuracy of the image analysis gates can vary slightly from day to day and that minor shifts of populations can occur during a sort. However, the overall stability of threshold settings and the basic imaging features are remarkably reproducible when used on biological replicates.

2

Conclusion The Golgi phenotype sorting example demonstrates the power of the fusion of conventional and image-enabled cell sorting. The data presented here show unequivocally that it would be impossible to sort out the dispersed Golgi-containing cells from the sample utilizing the traditional total intensity-based FACS approach. The underlying change in the Golgi phenotype is purely spatially coded and not discernable to a functional degree by total intensity changes. It is worth noting that the Golgi example lies within the sweet zone of resolution that the technology provides. Not every spatially encoded scientific question can be addressed with the

Image-Enabled Cell Sorting Using the BD CellView Technology

157

technology, considering the resolution limit and the constraint of imaging round cells in suspension. What we can conclude from our experience with the technology is that it provides a very robust and functional tool for biology to significantly enrich for certain phenotypes, thereby providing cleaner inputs into the remarkably diverse downstream analysis realm. This advance in technology is a very important enhancement in FACS instrumentation which has been one of the main biotechnological isolation methods for the last 50 years [6]. Although fast and powerful in its ability to detect and quantify fluorescence of a single cell, it has remained “blind in one eye” toward the spatial information offered by the cells. The CellView technology is a promising new beginning. Now that this deficiency has been fixed, we are looking forward to what can be done with this technology in the next 50 years. Currently, in summer 2023, there are very few commercially available image cell sorting technologies on the market in parallel to BD CellView. One technology, Deepcell, uses a sophisticated but much slower approach, utilizing brightfield images to establish machine learning-enabled morphometry-based sorting strategies. Deepcell’s focus on the brightfield is rooted in the desire to establish label-free image cell sorting which is of interest to the humandisease market [7]. A second technology in the fluorescent-enabled “image-based” cell sorting market is ThinkCyte’s Ghost cytometry. The technology has been around since its publication in 2018 [8], but ThinkCyte’s system is still in the early stages of market entry. The interesting part of the technology driving ThinkCyte is “machine vision.” The system utilizes structured illumination that creates data pulses not reconstituted into an image but interpreted in its raw form by machine learning. The algorithm’s output is an unbiased clustering map that interprets the different cellular phenotypes identified by the system. Attempting to interface the data description (clustering results) and the ultimate biology at the cellular level is somewhat cumbersome. It requires additional imaging methods to correlate purified clusters identified by machine vision using supplementary instrumentation. Nonetheless, the technology will mature and represents, with its explicit embedding of an unbiased machine learning approach into the sort process, an excellent addition to the burgeoning field. References 1. Hipp JD, Johann DJ, Chen Y, Madabhushi A, Monaco J, Cheng J et al (2018) Computeraided laser dissection: a microdissection workflow leveraging image analysis tools. J Pathol Inform 9:45. https://doi.org/10.4103/jpi.jpi_ 60_18

2. Lee K, Kim S, Nam S, Doh J, Chung WK (2022) Upgraded user-friendly image-activated microfluidic cell sorter using an optimized and fast deep learning algorithm. Micromachines (Basel) 13:2105. https://doi.org/10.3390/ mi13122105

158

Malte S. Paulsen

3. Schraivogel D, Kuhn TM, Rauscher B, Rodrı´guez-Martı´nez M, Paulsen M, Owsley K et al (2022) High-speed fluorescence image-enabled cell sorting. Science 375:315–320. https://doi. org/10.1126/science.abj3013 4. Diebold ED, Buckley BW, Gossett DR, Jalali B (2013) Digitally synthesized beat frequency multiplexing for sub-millisecond fluorescence microscopy. Nat Photonics 7:806–810. https://doi.org/10.1038/nphoton.2013.245 5. Storrie B, White J, Ro¨ttger S, Stelzer EH, Suganuma T, Nilsson T (1998) Recycling of golgi-resident glycosyltransferases through the ER reveals a novel pathway and provides an explanation for nocodazole-induced golgi

scattering. J Cell Biol 143:1505–1521. https:// doi.org/10.1083/jcb.143.6.1505 6. Filby A, Carpenter AE (2022) A new image for cell sorting. N Engl J Med 386:1755–1758. https://doi.org/10.1056/NEJMcibr2200971 7. Salek M, Li N, Chou H, Saini K, Jovic A, Jacobs KB et al (2023) COSMOS: a platform for realtime morphology-based, label-free cell sorting using deep learning. Commun Biol 6:971. h ttps://d oi.org/10.1038/s420 03-02305325-9 8. Ota S, Horisaki R, Kawamura Y, Ugawa M, Sato I, Hashimoto K et al (2018) Ghost cytometry. Science 360:1246–1251. https://doi. org/10.1126/science.aan0096

Chapter 9 Monitoring Cell Proliferation by Dye Dilution: Considerations for Panel Design Joseph D. Tario Jr., Kah Teong Soh, Paul K. Wallace, and Katharine A. Muirhead Abstract High dimensional studies that include proliferation dyes face two inherent challenges in panel design. First, the more rounds of cell division to be monitored based on dye dilution, the greater the starting intensity of the labeled parent cells must be in order to distinguish highly divided daughter cells from background autofluorescence. Second, the greater their starting intensity, the more difficult it becomes to avoid spillover of proliferation dye signal into adjacent spectral channels, with resulting limitations on the use of other fluorochromes and ability to resolve dim signals of interest. In the third and fourth editions of this series, we described the similarities and differences between protein-reactive and membrane-intercalating dyes used for general cell tracking, provided detailed protocols for optimized labeling with each dye type, and summarized characteristics to be tested by the supplier and/or user when validating either dye type for use as a proliferation dye. In this fifth edition, we review: (a) Fundamental assumptions and critical controls for dye dilution proliferation assays; (b) Methods to evaluate the effect of labeling on cell growth rate and test the fidelity with which dye dilution reports cell division; and. (c) Factors that determine how many daughter generations can be accurately included in proliferation modeling. We also provide an expanded section on spectral characterization, using data collected for three proteinreactive dyes (CellTrace™ Violet, CellTrace™ CFSE, and CellTrace™ Far Red) and three membraneintercalating dyes (PKH67, PKH26, and CellVue® Claret) on three different cytometers to illustrate typical decisions and trade-offs required during multicolor panel design. Lastly, we include methods and controls for assessing regulatory T cell potency, a functional assay that incorporates the “know your dye” and “know your cytometer” principles described herein. Key words Cell division, Cell tracking, CellTrace™ dyes, CellVue® dyes, Cytotoxicity, Dye dilution proliferation assay, Flow cytometry, Panel design, PKH dyes, Regulatory CD4pos T cells

Teresa S. Hawley and Robert G. Hawley (eds.), Flow Cytometry Protocols, Methods in Molecular Biology, vol. 2779, https://doi.org/10.1007/978-1-0716-3738-8_9, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

159

160

1

Joseph D. Tario et al.

Introduction Dyes that act as general protein or membrane tags are widely used for cell tracking when the cells of interest are difficult to transfect or where genetic reporters are unable to provide desired information (e.g., division history for individual cells). Although commercially available tracking dyes vary widely in their chemistries and fluorescence properties, the great majority fall into one of two classes based on their mechanism of cell labeling. “Protein dyes,” typified by CFSE [1], are amino-reactive fluorochromes that form stable covalent bonds with random cell proteins. “Membrane dyes,” typified by PKH26 [2], are lipophilic fluorochromes that intercalate stably but noncovalently into the bulk lipid phase of cell membranes via strong hydrophobic interactions. Both classes of dyes come in a wide range of colors, and each class has its own advantages and limitations [3, 4]. The key to successful use is therefore understanding the critical variables for optimal use of each class, particularly in multicolor studies where multiple dyes are used to track different cell types [5, 6]. Due to their stability of cell association, cell tracking dyes of both classes are often used for in vivo studies of cell trafficking, lifetime, and recruitment in contexts such as transplantation [7–9], infection [10, 11], stem cell identification [12, 13], and cancer immunotherapy [14, 15]. Both dye types have also proven valuable for a wide range of in vitro studies including antigen presentation [16–18], trogocytosis [19, 20], mechanism and specificity of cytotoxic effector killing [21–25], and regulatory T cell activity [5, 26] (see Subheading 3.6). Infectious agents [27, 28] and subcellular components (e.g., plasma membrane) [29, 30] can also be tracked, as can the fate and bioactivity of cell-derived vesicles [31–33]. Combining fluorescent cell tracking dyes with stably expressed genetic markers has become increasingly common as the spectral choices available for both probe types have increased. This strategy has been used to monitor extent and/or symmetry of cell division [34–37], something not possible with genetic markers alone, and also to detect active cycling in T cells responding to antigen, both before tracking dye dilution was evident and after daughter cells could no longer be distinguished from unlabeled cells [11]. As used here, the term “proliferation dye” refers to a protein or membrane tracking dye that: (a) exhibits sufficiently good chemical and metabolic stability to partition approximately equally between daughter cells at mitosis and (b) is sufficiently non-perturbing, even at high initial labeling intensities, to allow multiple rounds of cell division to be followed based on dye dilution. Monitoring the proliferative status of stem/progenitor and immune cells is among the most common applications of both classes of cell tracking dyes [5, 26, 34, 38–42], due to the significant limitations

Panel Design Considerations for Proliferation Dyes

161

associated with alternative methods. Using dye dilution to assess extent of cell division avoids the radiation safety and regulatory issues associated with tritiated thymidine incorporation and has several advantages over non-radioactive DNA precursors such as bromodeoxyuridine (BrdU) and ethynyldeoxyuridine (EdU). In particular, only cells actively synthesizing DNA during the pulse become labeled with BrdU and EdU, and viable daughter cells cannot be isolated for functional studies because detection requires fixation and permeabilization. In addition, although BrdU and EdU also dilute out as cells proliferate, division monitoring is typically limited to one or two daughter generations before labeled cells become indistinguishable from labeled ones due to the toxicity associated with high levels of incorporation [43]. As the spectral capabilities of flow and imaging cytometers and the range of choices available for proliferation dyes have expanded, so has our ability to design multiplex, high parameter studies to decipher complex biological systems [7, 11, 18, 29, 32, 37, 42]. The associated challenges for reagent panel design make it even more essential to understand the advantages—and limitations—of different proliferation dye(s) in order to select probe(s) wellmatched to the needs of a given application. Overall, there are more similarities than differences between the two classes of cell proliferation dyes (Table 1). The major differences between them lie in the techniques required for achieving optimal labeling and how soon after labeling cell division monitoring can be started. Key considerations for obtaining bright, homogeneous labeling with tracking dyes differ considerably from those for labeling with antibodies, and also for protein dyes vs. membrane dyes [3–5, 39, 44]. Subheadings 3.1 and 3.2 review the protocols used to label cultured U937 cells and human peripheral blood mononuclear cells (PBMCs) for the work presented here. Before a new cell tracking dye is used for proliferation monitoring, it is important to verify that rate of dilution is linearly correlated with the rate of cell growth in a system where unstained cells are also present and where growth can be independently measured. Subheading 3.3 illustrates two methods for doing this using continuously dividing cultured tumor cell lines. In systems where both responders and nonresponders are present, mathematical modeling [4, 38, 45] is often used to quantify: (a) extent of population expansion in response to a stimulus (typically reported as “Proliferation Index” or “Expansion Index”) and/or (b) proportion of the initial population able to proliferate in response to that stimulus (typically reported as “Precursor Frequency” or “Percent Divided”). In such cases, it is also important to know the maximum number of daughter generations that can be followed before highly divided dye-positive cells begin to overlap with unstained cells. This is illustrated in Subheading 3.4 for lymphocyte cultures proliferating in response to stimulation with anti-CD3 and anti-CD28 antibodies. Methods for assessment

162

Joseph D. Tario et al.

Table 1 Similarities and differences between proliferation Dye classesa Goal

Protein-reactive dyes (e.g., CFSE)

Membrane dyes (e.g., PKH26)

Similarities Reproducible staining Staining is not saturable Must specify and accurately reproduce both dye and cell concentration intensity No effect on cell function(s)

Must be verified for each system (positive and negative biological controls required)

Appropriate instrument setup

High degree of spectral overlap in adjacent channels can complicate compensation Cross-laser excitation can be substantial for some dyes

Data acquisition and gating

Exclude dead or apoptotic cells and contaminating cell types (e.g. phagocytes) Acquire sufficient cells for accurate curve fitting when modeling dye dilution profiles

Data analysis method and metric (s) matched to study goals

(Non-)Proliferative Fraction—negative biological control required Curve Modeling—negative staining and biological controls required —daughter “peaks” not required —manual definition of daughter regions not recommended Differences

Bright, homogeneous, Rapid mixing matters stable staining of Intensity varies with esterase activity parent population and cell size Division-independent dye loss for 6– 24 h (fixed T0 sample may not be appropriate biological or compensation control)

Rapid mixing is even more important than with protein dyes Intensity varies with cell size

Stable to fixation with Moderate intensity decrease for all methanol-free cells; shape of dye dilution profile formaldehyde unaffected

Intensity and shape of dye dilution profile unaffected

Stable intensity immediately poststaining (fixed T0 sample OK as biological or compensation control)

a

Refs. [3–5, 38, 39, 45, 61, 63, 70]

of spectral compatibility with available instrumentation and associated panel design considerations are discussed in Subheading 3.5, using data collected for three protein dyes (CellTrace™ Violet, CellTrace™ CFSE, and CellTrace™ Far Red) and three membrane dyes (PKH67, PKH26, and CellVue® Claret) on four different cytometers to illustrate the impact of laser configuration, laser power, and optical filter choices on the degree of spectral overlap and/or compensation required in other spectral windows. Subheading 3.6 illustrates the final and essential step of verifying that labeling conditions chosen do not alter the functional potency or proliferative behavior of labeled cells relative to unlabeled controls.

Panel Design Considerations for Proliferation Dyes

2

163

Materials

2.1 Cell Isolation and Cell Culture

1. Complete Medium (CM): RPMI 1640 supplemented with 10% heat inactivated fetal bovine serum (FBS), 25 mM HEPES, 0.1 mM non-essential amino acids, 1 mM sodium pyruvate, 2 mM fresh glutamine, 50 μg/mL gentamicin sulfate, and 5 × 10-5 M β-mercaptoethanol. 2. 10% Formaldehyde, methanol free, ultra-pure. Dilute to 2% in PBS (pH 7.4) and store refrigerated. 3. Hanks’ Balanced Salt Solution (HBSS) without phenol red, magnesium, or calcium. Store at room temperature until opened, then at 4–8 °C. 4. Histopaque®-1077. Store at 4–8 °C and use at room temperature. 5. IL-2 (Aldesleukin Proleukin for injection, NDC 53905–99101; Novartis, New York, NY). Dilute stock (2.2 × 106 IU/mL) in sterile HBSS to 1 × 105 IU/mL, aliquot, and store at -80 ° C. Do not refreeze after thawing; store at 4–8 °C and discard thawed product after 7 days. 6. Phosphate Buffered Saline (PBS): Prepare 10× stock containing 1.37 M NaCl, 27 mM KCl, 100 mM Na2HPO4, and 18 mM KH2PO4. Adjust to pH 7.4 with HCl if necessary. Sterilize by 0.2 μm filtration and store at room temperature. Prepare 1× working solution by dilution of one part with nine parts tissue culture grade water. 7. Human peripheral blood mononuclear cells (hPBMCs). Isolate hPBMC from heparinized peripheral blood or from TRIMA filters [5] using the laboratory’s standard density gradient fractionation protocol, with the addition of a final low speed wash (300 ×g) to minimize platelet contamination (see Note 1). 8. K562 Cell Line (see Note 2). Kind gift of Dr. Myron S. Czuczman, Roswell Park Comprehensive Cancer Center, Buffalo, New York; also available for purchase (American Type Culture Collection, Manassas, VA). 9. U937 Cell Line (see Note 2). Generously provided by Paul Guyre, Lebanon, NH; also available for purchase (American Type Culture Collection). 10. 24-well polystyrene plates are useful for plating quadruplicate samples for kinetic studies. 11. 96-well U-bottom polypropylene stripwell plate consisting of 1.1 mL polypropylene tubes in strips of 8, racked in plates and sterile.

164

Joseph D. Tario et al.

2.2 Antibodies and Immunophenotyping Reagents

1. 1.0 mg/mL anti-CD3 (clone OKT3) and 1.0 mg/mL antiCD28 (clone 28.2). Azide free, unconjugated preparations (ThermoFisher Scientific, Waltham, MA). 2. CD45 allophycocyanin (APC, clone 2D1) and CD45 Brilliant Violet 510 (BV510, clone HI30) (BD Biosciences, San Jose, CA). 3. CD4 phycoerythrin cyanine 7 (PE-Cy7, clone SK3), CD25 allophycocyanin (APC, clone 2A3), and CD127 phycoerythrin (PE, clone hIL-7R-M21) (all from BD Biosciences). 4. CD45-Pacific Blue (PacBlue, clone HI30) (BioLegend, San Diego, CA). 5. IgG block. Prepare from Cohn fraction II and III γ-globulins (cat# G-4386, Sigma Aldrich, St. Louis, MO) by reconstituting to 12 mg/mL in RPMI 1640 supplemented with 25 mM HEPES, 20 μg/mL gentamicin sulfate, and 2 mg/mL bovine serum albumin (BSA) (Sigma-Aldrich). Store frozen at -20 °C until use. Once thawed, store at 4–8 °C for no longer than 1 month. 6. ACK lysing buffer (cat. # A1049201, ThermoFisher Scientific).

2.3 Flow Cytometry Reagents

1. FCM Buffer: 1× PBS (pH 7.2) supplemented with 1% BSA, 0.1% sodium azide, and 40 μg/mL tetrasodium ethylenediaminetetraacetic acid. 2. LIVE/DEAD® Fixable Violet (Invitrogen/ThermoFisher Scientific). Reconstitute with DMSO according to the manufacturer’s instructions. Store frozen at -20 °C for no longer than 6 months. Thaw a fresh aliquot daily and dilute 1:50 in PBS. Add 5 μL per test and incubate for 30 min in a buffer free of exogenous protein before washing and fixing in 2% methanolfree formaldehyde for assessment of viability by flow cytometry. 3. 4′,6-diamidino-2-phenylindole (DAPI). Reconstitute powdered solid to 5 mg/mL in deionized water and store at 4–8 ° C. Prepare a working stock by diluting to 5 μg/mL in deionized water. Add 5 μL of working stock to each 100 μL of cells (0.25 μg/mL final) and let stand on ice for 30 min prior to data acquisition. 4. 7-Aminoactinomycin D (7-AAD) (cat #A1310, ThermoFisher Scientific). Reconstitute powdered solid to 1 mg/mL in PBS and store at -20 °C. Prepare a working stock by diluting thawed 1 mg/mL stock to 100 μg/mL in PBS and store at 4–8 °C. Add 4 μL of working stock to each 100 μL of cells (4 μg/mL final) and let stand on ice for 30 min prior to data acquisition.

Panel Design Considerations for Proliferation Dyes

165

5. TO-PRO-3 (cat. # T3605, ThermoFisher Scientific). 6. Rainbow 6-Peak Calibration Particles for instrument setup (Spherotech, Lake Forest, IL). Use for establishing reference intensities in employed fluorescence detectors as described in Subheading 3.3.1. 7. AccuCount Particles for cell counting (Spherotech). Use for single platform cell enumeration as described in Subheading 3.3.1. 8. PKH26 Reference Microbeads (Sigma-Aldrich). Use for single platform cell enumeration as described in Subheading 3.3.1. 2.4 Cell and Proliferation Tracking Dyes

1. CellTrace™ Violet (CTV), CellTrace™ CFSE (CFSE), and CellTrace™ Far Red (CTFR) (ThermoFisher Scientific); CytoTrack™ Yellow (CYY) (Bio-Rad, Hercules, CA). CFSE is also available from other suppliers. Reconstitute lyophilized aliquots for cell labeling according to the manufacturer’s recommended concentrations: 5 mM for CTV, 5 mM for CFSE (see Note 3), 1 mM for CTFR, and 500× for CYY. Labeling chemistries are similar for all four dyes: non-fluorescent precursor compounds freely diffuse across the plasma membrane into the cytoplasm, where their acetate substituents are cleaved by non-specific esterases. This results in trapping of the charged fluorescent product and random protein labeling via covalent bond formation between free amino substituents and the dye’s succinimidyl esters. 2. PKH67, PKH26, and CellVue® Claret (CVC) fluorescent cell linker kits (Sigma-Aldrich). Kits contain 1 mM dye solutions in ethanol and cell labeling diluent for general cell membrane labeling (Diluent C). CVC is also available from another supplier (Molecular Targeting Technologies, Inc., West Chester, PA). These dyes are incorporated into membranes based on hydrophobic forces that drive partitioning from the aqueous phase, in which the dyes are highly insoluble, into cell membranes where they are stably retained due to strong noncovalent interactions between their long alkyl tails and those of membrane lipids. Store tightly capped at room temperature to avoid evaporation of ethanol and associated increases in dye concentration. If any dye solids are visible, sonicate dye stocks to redissolve before use and verify that dye absorbance remains within the range specified on the Certificate of Analysis available for each kit. This can be done by preparing the indicated dilution in 100% ethanol and measuring its absorbance at the indicated wavelength.

166

Joseph D. Tario et al.

2.5 Flow Cytometers, Other Equipment, and Data Analysis Software

For routine data acquisition, any flow cytometer capable of acquiring forward and side scatter, DAPI, BV421, BV510, FITC, PE, and APC would be appropriate. Data in this chapter were collected using four different flow cytometers. For all figures showing flow cytometric data, axis labels follow the convention of Ref. [46]. 1. LSR II (BD Biosciences). The instrument used for studies shown in Figs. 1, 6, and 8 was fitted with 355 nm (100 mW), 405 nm (25 mW), 488 nm (20 mW), 561 nm (50 mW), and 640 nm (40 mW) lasers. From the 488 nm laser, FSC and SSC were measured using 488/10 nm bandpass (BP) filters. From the 355 nm laser, DAPI fluorescence was measured using a 450/50 nm BP filter. From the 405 nm laser, CTV fluorescence was measured using a 450/50 nm BP filter and CD45BV510 was measured using 525/50 nm BP filter. From the 488 nm laser, CFSE and PKH67 fluorescence were measured using a 530/30 nm BP filter and 7-AAD fluorescence was measured using a 695/40 nm BP filter. From the 561 nm laser, PKH26 fluorescence was measured using a 582/15 nm BP filter. From the 640 nm laser, CTFR, CVC, and CD45APC fluorescence were measured using a 660/20 nm BP filter. The instrument used for the studies of Figs. 11 and 12 had no UV laser and was fitted with 407 nm (25 mW), 488 nm (20 mW), and 635 nm (20 mW) lasers. From the 407 nm laser, PacBlue or LIVE/DEAD Fixable Violet was measured using a 450/50 nm BP filter; from the 488 nm laser, CFSE or PKH67, PKH26, 7-AAD, and CD4 PECy7 fluorescence were detected using 530/30 nm, 575/26, 685/35, and 780/60 nm BP filters, respectively. CVC fluorescence was excited using the 633 nm line and collected using a 660/20 nm BP filter. 2. LSR Fortessa (BD Biosciences). Fitted with 355 nm (60 mW), 405 nm (50 mW), 488 nm (50 mW), and 640 nm (40 mW) lasers. From the 488 nm laser, FSC and SSC were measured using 488/10 nm BP filters. From the 355 nm laser, DAPI fluorescence was measured using a 450/50 nm BP filter. From the 405 nm laser, CTV fluorescence was measured using a 450/50 nm BP filter and CD45 BV510 fluorescence was measured using a 525/50 nm BP filter. From the 488 nm laser, CFSE and PKH67 fluorescence were measured using a 530/30 nm BP filter and PKH26 fluorescence was measured using a 575/26 nm BP filter. From the 640 nm laser, CTFR and CVC fluorescence were measured using a 670/14 nm BP filter. 3. MACSQuant Analyzer 10 (Miltenyi Biotec, San Diego, CA). Fitted with 405 nm (40 mW), 488 nm (30 mW), and 635 nm (20 mW) lasers. From the 488 nm laser, FSC and SSC were measured using 488/10 nm BP filters. From the 405 nm laser,

Panel Design Considerations for Proliferation Dyes

167

CTV fluorescence was measured using a 450/50 nm BP filter. From the 488 nm laser, CFSE and PKH67 fluorescence were measured using a 525/50 nm BP filter and PKH26 and CYY fluorescence were measured using a 585/40 nm BP filter. From the 640 nm laser, CTFR and CVC fluorescence were measured using a 655–730 nm spectral window. 4. MACSQuant VYB (Miltenyi Biotec). Fitted with 405 nm (40 mW), 488 nm (50 mW), and 561 nm (100 mW) lasers. From the 561 nm laser, FSC and SSC were measured using 561/10 nm BP filters. From the 405 nm laser, CTV fluorescence was measured using a 450/50 nm BP filter. From the 488 nm laser, CFSE and PKH67 fluorescence were measured using a 525/50 nm BP filter. From the 561 nm laser, PKH26 was measured using a 585/15 nm BP filter and CTFR and CVC fluorescence were measured using a 661/20 nm BP filter. 5. NovoCyte 3000 (Agilent, Santa Clara, CA). Fitted with 405 nm (50 mW), 488 nm (60 mW), and 640 nm (40 mW) lasers. From the 488 nm laser, FSC and SSC were measured using 488/10 nm BP filters. From the 405 nm laser, CTV fluorescence was measured using a 445/45 nm BP filter. From the 488 nm laser, CFSE and PKH67 fluorescence were measured using a 530/30 nm BP filter and PKH26 fluorescence was measured using a 572/28 nm BP filter. From the 640 nm laser, CTFR and CVC fluorescence were measured using a 675/30 nm BP filter. 6. A tube rotator (#13916-822; VWR, West Chester, PA) was used for monocyte depletion of hPBMC and preparation of accessory cells for the studies described in Subheading 3.6. 7. FACS DiVa™ 8.0.1 (BD Biosciences). 8. FCS Express 6.0 (De Novo Software, Glendale, CA). 9. FlowJo™ v10.2 (FlowJo, LLC, Ashland, OR). 10. NovoExpress v1.2.4.1602 (Agilent, Santa Clara, CA). 11. WinList™ v8.0 (current version is v9.0) and ModFit LT™ v4.0 (Verity Software House, Topsham, ME).

3

Methods Virtually any eukaryotic cell can be stained with either class of tracking dye after a single-cell suspension has been obtained (see Notes 4 and 5). The labeling conditions described below have been successfully used to stain hPBMCs and cultured cells used for the proliferation and cytotoxicity assays discussed here but are likely to require modification for other cell types, assay systems, or dye combinations (see Notes 6 and 7). Although CTV, CFSE, and

168

Joseph D. Tario et al.

CTFR are used herein to represent typical protein labeling dyes and PKH67, PKH26, and CVC to represent typical membrane labeling dyes, the principles described here also apply to optimization of staining conditions and flow cytometer choice for use of the many other commercially available tracking dyes. 3.1 Cell Line and hPBMC Labeling with Protein Dyes (CTV, CFSE, CTFR, or CYY)

The method for CellTrace™ labeling described here is a simplification of the protocol described by Quah and Parish [3]. The method for CYY labeling is the manufacturer’s recommended protocol, as adapted for the labeling of cultured U937 cells. 1. Prepare a stock solution in anhydrous DMSO by adding the recommended volume to the provided pre-weighed single use vial, vortexing, and visually inspecting the vial to ensure complete dissolution. The manufacturer’s recommended stock concentrations are: 5 mM for CTV, 5 mM for CFSE (see Note 3), 1 mM for CTFR, and 500× for CYY. 2. Wash cells to be labeled twice in serum-free PBS or HBSS. After resuspension of the cell pellet from the first wash, remove an aliquot for cell counting (see Note 8). After final wash, resuspend cells in serum-free buffer (see Note 9) at a final concentration of 1 × 107 cells/mL (range for hPBMC: 0.5–50 × 106 cells/mL), using a tube that will hold at least six times the volume of the cell suspension. For CYY labeling, pellet cells and carefully aspirate supernatant. 3. Immediately prior to cell labeling, prepare working solutions of CellTrace dyes (50 μM for CTV and CFSE; 10 μM for CTFR) by making a 100-fold dilution of the DMSO stock solution from step 1 in PBS (see Note 10). Prepare a 2× working solution of CYY by making a 250-fold dilution of the DMSO stock solution from step 1 in PBS. 4. For a final staining concentration of 1 μM CellTrace dye, add appropriate amount of working dye solution per mL of cell suspension: 20 μL/mL of cells for CFSE or CTV; 100 μL/mL of cells for CTFR (e.g., to stain 2 mL of hPBMC at a final concentration of 1 × 107 cells/mL and 1 μM CTV, add 40 μL of working dye solution; Notes 7, 11, and 12). For a final staining concentration of 2× CYY, resuspend cell pellet from step 2 in 100 μL of working dye solution per 106 cells (e.g., to stain 5 × 106 U937 cells at a final concentration of 107 cells/ mL and 2× CYY, resuspend cell pellet from step 2 in 500 μL of 2× working solution prepared in step 3). 5. Immediately triturate or vortex tube briefly to disperse dye throughout cell suspension. Incubate at ambient temperature (~21 °C) for 5–15 min, with occasional mixing either manually or on a rotator, protected from light (see Notes 13 and 14).

Panel Design Considerations for Proliferation Dyes

169

6. Stop the reaction by adding a 5× volume of CM or a 1× volume of FBS and mixing well (see Note 15). Centrifuge at 400 ×g for 5 min at ~21 °C and discard the supernatant. 7. Wash the cells twice with 5–10× volumes of CM. After resuspension of the cell pellet from the first wash, remove an aliquot for cell counting. After the final wash, adjust concentration to the desired cell density for functional testing during the final resuspension in CM. 8. Assess recovery, viability, and fluorescence intensity profile of labeled cells immediately post-staining to determine whether to proceed with assay setup (see Note 16 and Ref. [5]). 9. At 24 h post-labeling, verify that labeled cells are well enough resolved from unstained cells for purposes of the assay to be performed and that dye fluorescence can be adequately compensated in spectral windows to be used for measurement of other probes (see Subheading 3.5; Note 17). If samples are to be fixed and analyzed in batch mode, verify that loss of intensity due to fixation does not compromise ability to distinguish desired number of daughter generations (see Note 18 and Ref. [5]). 10. Verify that labeled cells are functionally equivalent to unlabeled cells (see Subheading 3.6; Notes 6 and 19). 3.2 Cell Line and hPBMC Labeling with Membrane Dyes (PKH26, PKH67, or CVC)

The method described here is illustrated in detail in Ref. [4]. 1. Wash cells to be labeled twice in serum-free PBS or HBSS (see Note 9), using a 5 mL Eppendorf conical tube with snap cap (see Note 20) or similar conical polypropylene tube able to hold at least six times the final staining volume in step 5. After resuspension of the cell pellet from the first wash, remove an aliquot for cell counting (see Note 8) and determine the volume needed to prepare a 2× working cell suspension (step 4 below) at a concentration of 1 × 108 cells/mL for hPBMCs (range = 2–100 × 106 cells/mL) or 2 × 107 cells/mL for U937 cells. For example, to stain a total of 5 × 107 hPBMCs at a final concentration of 5 × 107 cells/mL, the volume of 2× cell suspension would be 0.5 mL. 2. Following the second wash in step 1, aspirate the supernatant, taking care to minimize amount of buffer remaining (no more than 15–25 μL) while avoiding aspiration of cells from the pellet (see Note 21). Flick the tip of conical tube once or twice with a finger to disperse the cell pellet in the small amount of fluid remaining, but avoid significant aeration as this reduces cell viability. 3. Prepare 2× working dye solution of PKH67, PKH26, or CVC. To a second conical polypropylene tube (see Note 20), add the same volume of Diluent C staining vehicle (provided with each

170

Joseph D. Tario et al.

membrane dye kit) calculated in step 1 for preparation of the 2× cell suspension. Add the appropriate amount of 1 mM ethanolic dye stock to the Diluent C (e.g., for a 2× working dye solution to give a final dye concentration of 5 μM after admixture with 2× cells in step 5, add 5 μL of dye stock to 0.5 mL of Diluent C). Gently vortex tube to ensure complete dispersion of dye in diluent, avoiding loss of fluid to cap or deposition of droplets on tube wall. Proceed with steps 4 and 5 as rapidly as possible (see Notes 22 and 23). 4. Prepare a 2× cell suspension by adding the volume of Diluent C calculated in step 1 to the partially resuspended cell pellet from step 2. Triturate 3–4 times to ensure complete dispersion of the pellet and proceed immediately to step 5. 5. Rapidly admix the 2× cell suspension from step 4 into the 2× working dye solution from step 3, dispensing the cells directly into the 2× dye and immediately triturating 3–4 times to achieve as nearly instantaneous exposure of all cells to dye as possible (see Notes 24 and 25). 6. After 3 min, stop the labeling by adding a 5× volume of CM (containing 10% FBS) or a 1× volume of FBS or other cellcompatible protein and mixing well (i.e., if 1 mL of cells was combined with 1 mL of dye, then add 10 mL of CM or 2 mL of FBS; Note 26). 7. Centrifuge at 400 ×g for 5 min at ~21 °C and discard the supernatant. Wash the cells twice with 5–10 volumes of CM, removing an aliquot for cell counting after resuspension of the cell pellet from the first wash. After the final wash, count and resuspend the cells in CM at the final desired cell density for functional testing. 8. Assess recovery, viability, and fluorescence intensity profile of labeled cells immediately post-staining to determine whether to proceed with assay setup (see Note 16). 9. Verify that labeled but non-proliferating cells (e.g., unstimulated control) are well enough resolved from unstained cells for purposes of the assay to be performed and that membrane dye fluorescence can be adequately compensated in adjacent spectral windows used for measurement of other probes (see Subheading 3.5; Note 27). 10. Verify that labeled cells are functionally equivalent to unlabeled cells (see Subheading 3.6; Notes 6 and 19). 3.3 Evaluating Linearity of Dye Dilution

Cultured cell lines are useful for testing whether there is a linear correlation between rate of cell growth and rate of dye dilution for a particular cell tracking dye. Such systems allow direct measurement of growth by cell counting, whereas indirect measures such as tritiated thymidine uptake must be used to estimate extent of

Panel Design Considerations for Proliferation Dyes

171

proliferation in more complex systems (e.g., hPBMC, see Table S1 of Ref. [38]). Before evaluating linearity of dilution, however, it is necessary to verify that the staining conditions used do not alter cell growth rate compared with that of unlabeled cells. Subheadings 3.3.1 and 3.3.2 describe methods for testing dye effects on the growth rate using (a) parallel cultures of stained and unstained cells or (b) co-cultures of stained and unstained cells, respectively. Subheading 3.3.3 illustrates how the resulting data can be used to evaluate the linearity of dye dilution and compare dye dilution rates for different proliferation dyes. 3.3.1 Relative Growth Rate of Stained vs. Unstained Cells Using Counting Beads (Nonvolumetric Cytometer)

Beads from different vendors (here, Spherotech or Sigma-Aldrich; Subheading 2.3) may be employed, so long as their fluorescence can be distinguished from that of labeled cells in at least one detection channel. Although it is not necessary to know the absolute concentration of beads/mL, a sufficient number of beads should be present in the initial culture to allow a statistically meaningful number (>100) to be counted when high density cultures are harvested and analyzed by flow cytometry at later time points. 1. Harvest logarithmically growing U937 cells and label with tracking dye of interest as described in Subheading 3.1 or 3.2, leaving an aliquot of harvested cells unlabeled to serve as a dye-naı¨ve control. Adjust labeled and unlabeled cell suspensions to 2.5 × 105 cells/mL in CM. 2. Prepare counting beads for addition to cell cultures by removal of azide and detergents. From the stock solution of fluorescent counting beads, remove an aliquot sufficient to give a concentration of 2.5–10 × 104 beads/mL after admixing with the cell suspension to be used for culture initiation. (e.g., if a bead concentration of 10 × 104/mL is desired after admixing with 50 mL of cell suspension, aliquot a volume of bead stock containing at least 5 × 106 beads.) Centrifuge the bead suspension at 400 ×g for 5 min at ~21 °C, discard the supernatant, and wash twice with 10 mL CM, discarding the supernatant after each centrifugation. After the last wash, re-suspend beads in 1 mL of CM. 3. To the cell suspension from step 1, add washed beads from step 2 at the volume needed to give a final bead concentration of 2.5–10 × 104 beads/mL but do not exceed 1% of the volume of cell suspension. (e.g., if 5 × 106 beads were washed and resuspended in 1 mL CM, adding 200 μL of bead suspension to 40 mL of cell suspension gives a final bead concentration of 2.5 × 104 beads/mL). 4. Mix cell and bead suspensions to homogeneity and separately plate 1 mL of each of suspension into replicate wells of a 24-well flat bottom polystyrene plate for each time point in

172

Joseph D. Tario et al.

the kinetic study (e.g., to evaluate 6 time points in quadruplicate, plate 24 wells). Incubate labeled and unlabeled cell cultures in a humidified 37 °C incubator with 5% CO2 for 7 days. 5. Harvest replicate samples of plated cell cultures for each time point into 12 × 75 mm round bottom tubes by triturating each well to homogeneity with a P1000 pipette fitted with a clean, sterile tip, and place tubes on ice. For the T0 samples, this should be performed immediately after plating the suspensions from step 4, in order to ascertain the starting cell-to-bead ratio. 6. After collecting the samples for Day 3 and Day 5, triturate all remaining wells in the plate as in step 5. Remove and discard 500 μL of mixed cell/bead suspension from each well, and replenish with 500 μL of pre-warmed, CO2-equilibrated CM in order to maintain the cultures in log phase growth. After addition of CM, collect a post-dilution sample and determine the exact dilution factor by comparing its cell:bead ratio to the pre-dilution cell:bead ratio. 7. Add viability dye (50 μL of a 5 μg/mL DAPI solution or 40 μL of a 100 μg/mL 7-AAD solution) to each mL of harvested cells and incubate on ice for 30 min prior to data acquisition. Do not wash or further manipulate harvested cell suspensions, in order to avoid selective losses of either cells or beads. 8. Acquire harvested samples using a flow cytometer, collecting forward and side scatter characteristics (pulse area, height, and width) as well as all relevant fluorescence detection channels (pulse area). Ensure that the acquisition threshold is configured to allow fluorescent beads and cells to be collected using the same instrument settings. Using the gating strategy summarized in Fig. 1a, set a “Stopping Gate” on R4 of 2500 fluorescent bead events. 9. Establish the appropriate detector voltage or gain for each dye’s primary detection channel, placing unlabeled controls fully on-scale in the first decade, and confirm that labeled cells are fully on-scale. If >5% of labeled cells are off-scale high, reduce voltage as needed to bring them on-scale and re-acquire unlabeled controls. 10. After detector settings have been satisfactorily configured for each dye, acquire fluorescent Rainbow 6-Peak Calibration Particles and record mean intensity values from all peaks that are well resolved and fully on-scale in a given detector. To ensure consistent fluorescence intensities and enable direct comparison of data collected on subsequent days, use the recorded intensities as target values when re-establishing detector settings for the remainder of the experiment.

Panel Design Considerations for Proliferation Dyes

173

11. Using appropriate software and the gating strategy illustrated in Fig. 1a, quantify the number of beads acquired during sample analysis. Use serially gated bivariate plots of: Time vs. SSC-A to include only events associated with stable flow rate (R1); and FSC-A vs. SSC-A to discriminate bead events (R3) from cellular events. Gating on “R1&R3,” create a bivariate plot for any two detectors where beads are expected to be brighter than cells (e.g., Y710/50-A and V780/60-A) and establish a “beads-only” region (R4). Use the combined “R1&R3&R4” gate to generate a bivariate beads-only plot of Time vs. FSC-A and establish a rectangular region (R5) to exclude doublet bead events and provide a singlet bead count.

Fig. 1 Relative growth rate determination using counting beads (see Subheading 3.3.1). U937 cells were labeled as in Subheading 3.1 at the concentrations of CTV, CFSE, or CTFR indicated in Panel C and a final concentration of 1 × 107 cells/mL. After addition of 3 × 104 beads/mL to each sample, parallel cultures of labeled and unlabeled cells were cultured for 7 days, with the addition of fresh CM on Days 3 and 5 to maintain logarithmic growth. On each day, quadruplicate wells for each test article were separately triturated to homogeneity, harvested, stained with DAPI (CFSE, CTFR) or 7-AAD (CTV) for viability assessment, and acquired on the LSRII cytometer. Panel A. Data files were analyzed using the gating strategy described in Subheading 3.3.1, steps 11 and 13, with a Stopping Gate (R4) set to 2500 beads. Panel B. Histogram overlays for viable U937 cells (“R1&R2&R7& NOT R6”) present in unstained (filled distribution) and Day 0–Day 7 CTFR-labeled cultures, normalized to the volume of sample associated with a bead count of 250. Inset: Days 0–3 on an expanded scale. Panel C. Cell/bead ratio increased over time at similar rates in all samples, indicating that growth rate was not altered by labeling with any of the three dyes at the concentrations tested. Calculated doubling times ranged from a low of 24.8 h (unstained cells) to 26.3 h (CFSE stained cells)

174

Joseph D. Tario et al.

12. Separately, quantify the number of viable singlet cells acquired during sample analysis (Fig. 1a). Create a FSC-A vs. FSC-H plot to eliminate doublet cells (R2). Use an R1 gated plot of SSC-A vs. DAPI U450/50-A (or 7-AAD B695/40-A for CTV) to identify dead cells and beads (R6). Create an “R1&R2&NOT R6” gated plot of FSC-A vs. SSC-A to distinguish small debris from live U937 cells (R7) and an “R1&R2&R7& NOT R6” gated histogram of dye fluorescence (e.g., CTFR R660/20-A) to provide a singlet cell count and monitor dye dilution over time (Fig. 1b). 13. Calculate cell-to-bead ratio (cells/bead) for each acquired sample by dividing number of events in the “R1&R2&R7&NOT R6” gate (cells) by number of events in the “R1&R3&R4&R5” gate (beads). For the unstained control and each dye labeled population, plot Cells/Bead vs. Time, corrected for the measured dilution factors on days 3 and 5 (Fig. 1c), and use the slope of log(cells/bead) vs. time to calculate and compare population doubling times (see Note 28). 3.3.2 Relative and Absolute Cell Growth Rate Determination in Cocultures (Volumetric Cytometer)

Cytometers with volumetric cell counting capability allow growth rate to be monitored in absolute units (cells/mL) as well as relative units (% labeled cells in co-cultures) without the addition of counting beads. 1. Harvest logarithmically growing U937 cells and label with tracking dye of interest as described in Subheading 3.1 or Subheading 3.2. 2. For co-cultures, add 1 × 105 stained cells and 1 × 105 unstained cells to a sufficient amount of CM to give a final volume of 10 mL and place in a 25 cm2 culture flask. For 100% stained cultures, add 2 × 105 stained cells in a sufficient amount of CM to yield a final volume of 10 mL. To a control flask, add 2 × 105 unstained cells in a sufficient amount of CM to yield a final volume of 10 mL. Incubate in a humidified 37 °C incubator with 5% CO2 for 5–7 days. 3. Immediately after cells are placed in culture flasks, and at approximately 24 h intervals, thereafter, mix well by triturating and withdraw a 1.0 mL aliquot for flow cytometric analysis. On Day 3 after trituration, remove the entire volume of cell suspension and dispense 2.5 mL back into culture flask, reserving 1.0 mL for pre-dilution analysis. Add 7.5 mL fresh pre-warmed, CO2-equilibrated CM to maintain logarithmic cell growth, mix well by triturating, and withdraw a second 1.0 mL (post-dilution) sample to confirm exact dilution factor.

Panel Design Considerations for Proliferation Dyes

175

4. Use the laboratory’s standard reference material(s) to ensure that the flow cytometer is giving reproducible intensity values in all detectors to be used (as described in Subheading 3.3.1). Configure the flow cytometer to acquire data from all fluorescence detection channels and adjust FSC and SSC detector settings to ensure that cells of interest are resolved from electronic noise/small debris and large aggregates. 5. Use unlabeled cells to establish detector settings that place their measured fluorescence distribution fully on-scale within the first decade in the primary detection channel for each tracking dye being evaluated. Using the same instrument settings, confirm that the labeled cell populations are fully on-scale. If more than 5% of the labeled cells fall off-scale high, decrease detector voltage to bring the labeled cells fully on-scale, re-acquire the unlabeled cells at the new setting. 6. Use appropriate software to generate a single-parameter histogram for each tracking dye in its primary detection channel. At a minimum, histograms should be serially gated on: bivariate plots of Time vs. SSC-A to eliminate any events associated with unstable flow rate; FSC-A vs. SSC-A to discriminate cellular events from small debris and large aggregates; and additional viability and/or doublet gates as needed. 7. Using the gated histograms, record the following for each time point at which labeled and unlabeled cell populations remain non-overlapping (see Notes 29 and 30): (a) Number of labeled cells/mL. (b) Number unlabeled cells/mL. (c) Dilution factor from step 3 (pre-dilution cell count divided by post-dilution cell count), if appropriate. (d) % labeled cells. 8. Using Microsoft Excel or similar software, compare growth rates for labeled vs. unlabeled cells by plotting % labeled cells vs. Time (Fig. 2a) and cells/mL vs. Time (Fig. 2b) to identify concentration(s) that do not impair cell growth. 3.3.3 Assessing Linearity of Dye Dilution and Relative Dilution Rates

Once labeling conditions that do not perturb cell growth rate have been established, absolute cell counts obtained from parallel cultures or co-cultures (as described in Subheading 3.3.2) can be used to evaluate the correlation between cell growth rate and dye dilution rate for dyes of interest used as single label (Fig. 3) or in combination (Fig. 4).

176

Joseph D. Tario et al.

Fig. 2 Relative and absolute growth rate determination using volumetric counting. (see Subheading 3.3.2) Adapted with permission from Springer Nature Customer Service Centre GmbH: Springer; Methods Mol Biol. Monitoring Cell Proliferation by Dye Dilution: Considerations for Probe Selection. Tario JD Jr., Conway AN, Muirhead KA, Wallace PK. Copyright 2018. U937 cells were separately stained on Day 0 with each of the indicated tracking dyes as in Subheadings 3.1 and 3.2 (final concentrations: 1 × 107 cells/mL and 10 μM dye) and placed in 1:1 co-culture with unstained cells as in Subheading 3.3.2. An unstained control at the same total density was cultured in parallel. Aliquots withdrawn at each time point were acquired on a MACSQuant 10 Analyzer during the Bowdoin 2014 Annual Course in Flow Cytometry. Baseline separation between stained and unstained populations was maintained through Day 5 for all dyes except CYY. Unstained cells in co-culture with CYY stained cells became indistinguishable from stained cells by Day 1. Panel A. Stable values for % dye positive cells over time indicated that cells labeled with CTV, CFSE, PKH67, PKH26, CTFR, and CVC were growing at rates similar to unstained cells in co-cultures. In contrast, % CYY stained cells increased dramatically from Day 1 to Day 2 due to loss of resolution between stained and “unstained” cells. CYY was the only dye for which unstained cells exhibited such a rapid and extensive right shift, suggesting that the cause was dye-specific and not due to trogocytosis or other cell-type specific transfer mechanisms (see Note 29). Panel B. Plots of absolute cell counts vs. time (Day 3 value = average of pre- and post-dilution counts) confirmed that both stained (“pos”) and unstained (“neg”) populations grew logarithmically during the 5 day co-culture period, with doubling times similar to those of the unstained control culture (see plot legend)

Panel Design Considerations for Proliferation Dyes

177

Fig. 3 Correlation between growth rate and dye dilution rate (see Subheading 3.3.3). Adapted with permission from Springer Nature Customer Service Centre GmbH: Springer; Methods Mol Biol. Monitoring Cell Proliferation by Dye Dilution: Considerations for Probe Selection. Tario JD Jr., Conway AN, Muirhead KA, Wallace PK. Copyright 2018. U937 cells labeled as described in Fig. 2 and unstained controls were placed in separate parallel cultures (see Subheading 3.3.2; 2013 data) or in 1:1 co-cultures (see Subheading 3.3.2; 2014 and 2015 data) for 4–5 days. Aliquots withdrawn at each time point were acquired on 3 different MACSQuant 10 Analyzers during the 2013–2015 Annual Courses in Flow Cytometry and analyzed as described in Fig. 2.

178

Joseph D. Tario et al.

Fig. 4 Relative dye dilution rates for CTV and PKH26 on double labeled U937 cells (see Subheading 3.3.3). Adapted and reprinted by permission from Methods in Cell Biology. Soh KT, Tario JD, Muirhead KA, Wallace PK. Probing Cell Proliferation: Considerations for Dye Selection, in press, Copyright Elsevier (2023). Logarithmically growing U937 cells were stained separately or sequentially with CTV (see Subheading 3.1) and PKH26 (see Subheading 3.2) at a final concentration of 1 × 107 cells/mL and 5 μM dye. After overnight culture to allow for early division-independent loss of CTV (Fig. 3a), triplicate aliquots were harvested daily for each condition from Day 1 to Day 6 and analyzed on the Novocyte 3005 cytometer for cell concentration and fluorescence intensity. On Day 4, the cell culture was split 1:4 with fresh CM to maintain logarithmic growth. Fluorescence intensity was measured in the indicated spectral window(s) for each dye combination. Dye dilution rates were similar in dual stained and single stained cells, indicating that dual labeling did not affect cell growth rate. Fold decrease in CTV intensity was greater than for PKH26 in both single and dual stained cells, indicating that even after an overnight stabilization period CTV exhibits a second slower phase of division-independent intensity loss similar to that reported for CFSE [62] ä Fig. 3 (continued) Panel A. All protein dyes tested exhibited much greater dye dilution between T0 and T1 (~20 h) than could be attributed to cell growth alone, reflecting the early, division-independent intensity loss characteristic of this dye class. Growth rates were similar for all three dyes (1.3–1.6-fold increase in cell counts) but CTV and CTFR showed less division-independent dye loss than CFSE from T0 – T1. During the remainder of the culture period dye dilution was linear for all three dyes and more closely reflected cell growth rate, with slopes slightly greater than theoretical (range: 1.4–1.8). Panel B. All membrane dyes tested exhibited relatively linear dilution from T0 throughout the entire culture period, with slopes similar to or slightly less than theoretical (range: 0.6–1.2)

Panel Design Considerations for Proliferation Dyes

179

1. For each dye (or dye combination) and time point (T) after culture initiation (T0), calculate: (a) Fold Growth = count (T)/count (T0). (b) Fold Dye Dilution = median fluorescence intensity (T0)/ median fluorescence intensity (T). 2. Use Microsoft Excel or similar software to plot Fold Growth vs. Fold Dye Dilution for each dye of interest (Fig. 3; see Note 31). A perfect two-fold decrease in dye intensity at each cell division would give a linear correlation with slope = 1.0. 3.4 Biological Considerations for Cell Proliferation Modeling

Cultured cell systems are useful for characterizing a cell tracking dye in terms of linearity and rate of dye dilution (see Subheading 3.3) and spectral compatibility with the laboratory’s cytometer(s) and other fluorescent probes (see Subheading 3.5 below). However, additional considerations arise when the biology of interest includes both responders and non-responders (e.g., immune cell populations responding differentially to a specific stimulus). In such cases, mathematical modeling is used to quantify the frequency of cells within the starting population that go on to respond to the stimulus and/or overall expansion within the responding subpopulation. Because no currently available tracking dyes of either type give baseline resolution between daughter generations, selection of an appropriate proliferation dye requires knowing the number of daughter generations that can be reliably discriminated from undivided non-responders using a given dye. This is determined by several factors: (a) the highest tolerated labeling concentration, which will dictate the initial cell labeling intensity; (b) extent of division-independent dye loss for a given dye (Fig. 3a); and (c) effect of stimulation on the autofluorescence of dye-naı¨ve cells. In addition, in situations where a distinguishable non-responder peak is not present, it is important to know whether the location of stained but unstimulated controls from the same time point can be used as a surrogate to establish the expected location of non-responders in the stimulated sample. Assessment of these considerations is illustrated here using hPBMC stimulated with anti-CD3 and anti-CD28 to generate a polyclonal T cell response. 1. Generate a preparation of hPBMCs from peripheral blood using the laboratory’s standard density gradient fractionation protocol, with the addition of a final low speed wash (300 ×g) to minimize platelet contamination (see Note 1). 2. Enumerate harvested cells using the laboratory’s preferred methodology.

180

Joseph D. Tario et al.

3. Label hPBMCs with desired tracking dye (see Subheadings 3.1 and 3.2), reserving a sufficient number of unlabeled hPBMCs to employ as autofluorescence assay controls. Generally, an equal number of dye-labeled and unlabeled cells are required. 4. Adjust unlabeled and dye-labeled cells to 2 × 105 cells/mL in CM and divide both labeled and unlabeled hPBMC into two equal volumes. To half of each cell suspension add azide-free anti-CD3 (clone OKT3) to a final concentration of 1 μg/mL plus anti-CD28 (clone 28.2) to a final concentration of 0.5 μg/ mL (stimulated culture). Leave the other half unstimulated. 5. Separately dispense 0.5 mL (1 × 105 total cells) of each test article into quadruplicate wells of a 96-well U-bottom polypropylene stripwell plate for each time point to be evaluated and incubate in a humidified 37 °C incubator with 5% CO2 for 96 h (see Note 32). For example, to evaluate 5 time points in quadruplicate, establish 20 stimulated wells and 20 unstimulated wells for each dye-positive hPBMC test article and, in parallel, establish 20 stimulated wells and 20 unstimulated wells for unlabeled dye-naı¨ve control cells. 6. At each time point, remove the appropriate dye-positive and dye-negative stripwell tubes from the plate, and return the remainder to the humidified 37 °C incubator with 5% CO2. Separate individual tubes from each selected strip, lightly vortex to re-suspend cells and then transfer the stripwell tubes to the bottom of individually labeled 12 × 75 mm tubes, without washing or manipulating the samples. 7. Add viability dye (25 μL of a 5 μg/mL DAPI solution or 20 μL of a 100 μg/mL 7-AAD solution) to each tube containing 0.5 mL of cultured cells and let stand on ice for 30 min prior to data acquisition. 8. Acquire data on the flow cytometer as described in Subheading 3.3.1, steps 9 and 10. 9. Use appropriate software (e.g., WinList, FCS Express, or FlowJo) and the gating strategy shown in Fig. 5a to analyze acquired data. Use serially gated bivariate plots of: Time vs. SSC-A to include only events associated with stable flow rate (R1); FSC-A vs. FSC-H to select single cells (R2) and eliminate doublets; SSC-A vs. DAPI-A (or 7-AAD-A) to exclude dead cells (R3); FSC-A vs. SSC-A to define cellular events (R4); and the relevant fluorescence parameter(s) vs. SSC-A to generate fluorescence histograms for unlabeled or tracking dye-labeled lymphocytes (see Note 33). 10. For every experimental condition assayed, generate overlay histograms of unlabeled and dye-labeled fluorescence distributions from each time point and stimulation condition, as illustrated in Fig. 5b, c. Identify the intensity at which highly

Panel Design Considerations for Proliferation Dyes

181

Fig. 5 Effect of protein dye choice on division-independent dye dilution and ability to resolve highly divided cells from unlabeled cells (see Subheading 3.4). Adapted with permission from Springer Nature Customer Service Centre GmbH: Springer; Methods Mol Biol. Monitoring Cell Proliferation by Dye Dilution: Considerations for Probe Selection. Tario JD Jr., Conway AN, Muirhead KA, Wallace PK. Copyright 2018. hPBMCs were separately stained with the indicated tracking dyes as described in Subheadings 3.1 and 3.2 (final cell concentration: 1 × 107 cells/mL; final dye concentrations as shown) and cultured in quadruplicate without (Panel B) or with (Panel C) anti-CD3 + anti-CD28 stimulation. Data were acquired on an LSR Fortessa flow cytometer; one representative result is shown for each test condition. Panel A. Data files were collected using the gating strategy shown, with a Stopping Gate of 50,000 events in R4. Representative data and gating regions are shown for one of four replicate 96 h cultures. Panel B. Overlays show Day 0–4 histograms for

182

Joseph D. Tario et al.

divided cells can no longer be distinguished from the background fluorescence associated with unstained but stimulated cells. This will determine the maximum number of daughter generations that can accurately be modeled when estimating the fraction of the original population responding to a given stimulus and/or the extent of population expansion during the response (see Note 29 and Refs. [4, 38, 45]). The data gathered in step 8 can also be used in conjunction with peak modeling software (e.g., ModFit LT, FCS Express, or FlowJo) to test whether the dye dilution rate for non-responding cells is the same under stimulated and unstimulated conditions (Fig. 9 in Ref. [44]). 3.5 Spectral Characterization

Cell tracking dyes are commonly combined with each other, with fluorescent antibodies, and/or with genetic markers to enable: (a) in vitro or in vivo tracking of phenotypically defined subsets within heterogeneous populations (e.g., [7, 29]) or (b) identification and characterization of cell populations that do or do not proliferate in response to a given stimulus (e.g., [11, 26, 47]). The multiplicity of spectral detection options available on most digital flow cytometers means that when selecting fluorochrome combinations for such studies, it is important to evaluate the impact of candidate cell tracking dye(s) on the ability to simultaneously detect other common fluorochromes. This must be done in the context of each cytometer’s optical configuration, taking into account relative signal intensities expected from the different probes. Due to the extremely bright staining typically obtained with all cell tracking dyes, both spectral overlap into channels excited by the same laser and signal arising from cross-laser excitation must be taken into account. Where cells of interest are in limited supply, this can conveniently be done using cultured cells

ä Fig. 5 (continued) viable singlet lymphocytes from unstimulated cultures. Since lymphocytes do not proliferate under these conditions, T0–T1 intensity decreases represent division-independent dye loss. As was seen for U937 cells (Fig. 3a), CTV and CTFR showed much less early division-independent dye loss than CFSE. As previously reported for CFSE [62, 63], all three dyes also showed continuing but slower rates of divisionindependent dye loss throughout the study period although this is more difficult to see on the log scale used here than on the linear scale of Fig. 4. Panel C. Overlays show Day 0–4 histograms for viable singlet lymphocytes (“R1&R2&R3&R4” gate) in anti-CD3 + anti-CD28 stimulated cultures. As previously observed (Figs. 1a and 2a in Ref. [39]), stimulated but unstained lymphocytes exhibit measurably increasing autofluorescence over time, particularly in the V450/50 channel. As a result, even though CTV exhibits less division-independent early dye loss than CFSE, the ability to resolve unstained cells from stained but highly divided cells on Days 3 and 4 post-stimulation is actually slightly worse for cells labeled with 1 μM CTV than for those labeled with 1 μM CFSE. In situations where it is important to maximize resolution of unstained vs. highly divided cells but CTV concentration cannot be further increased without impacting cell function (see Subheading 3.6), use of a longer emitting tracking dye (e.g., PKH26, CTFR, CVC) may prove helpful

Panel Design Considerations for Proliferation Dyes

183

as described in step 1, Subheading 3.4. However, fluorochrome choices for immunophenotypic markers must be verified using the cells of interest. In general, it is best to place a bright fluorochrome, or one not present in/on cells of interest, in the spectral channel (s) closest to the proliferation dye. While this reduces the need for excessive compensation, appropriate compensation controls are still needed to verify that the presence of undivided proliferation dye positive cells will not impair ability to distinguish antibody positive and negative cells (see Subheading 3.5.2). 3.5.1 Spectral Overlap and Cross-Laser Excitation

1. Harvest logarithmically growing U937 cells and label with tracking dye(s) of interest as described in Subheading 3.1 or 3.2. 2. Configure the flow cytometer to acquire data from all fluorescence detection channels and adjust FSC and SSC detector settings to ensure that cells of interest are resolved from electronic noise/small debris and large aggregates. 3. Use unlabeled cells to establish detector settings that place the measured fluorescence distribution fully on-scale (preferably at least 95% of events with intensity ≥0). Using the same settings, confirm that each labeled cell population is fully on-scale in the primary detection channel for the tracking dye being evaluated (see Note 34). 4. Use data analysis software to generate single-parameter histograms for each detection channel as shown in Figs. 6 and 7. Histograms should be serially gated on bivariate plots of Time vs. SSC-A to eliminate any events associated with unstable flow rate; FSC-A vs. FSC-H to eliminate doublet events (see Note 35); and FSC-A vs. SSC-A to discriminate cellular events from small debris. In theory, it is desirable to use the highest non-perturbing concentration of proliferation dye in order to maximize the number of daughter generations that can still be distinguished from unstained cells. In practice, both spectral overlap and cross-laser excitation can lead to compensation problems in spectral window(s) adjacent to the dye’s primary emission. Extent of overlap will depend on both the tracking dye and the optical configuration of the cytometer. For example, on the Fortessa: (a) CTV spillover is only substantial in the V525/50-A channel, and cross-laser excitation is noted only in the 355 nm excited, U450/50-A channel (Fig. 6); (b) CFSE (Fig. 6) and PKH67 (Fig. 7) give substantial spillover into all channels associated with the 488 nm laser, and modest cross-laser excitation in all 405 nm-excited channels except V450/50-A;

184

Joseph D. Tario et al.

Fig. 6 Spectral characteristics of selected protein dyes on a BD LSR Fortessa cytometer (see Subheading 3.5.1). Adapted by permission from Springer Nature Customer Service Centre GmbH: Springer; Methods Mol Biol. Monitoring Cell Proliferation by Dye Dilution: Considerations for Probe Selection. Tario JD Jr., Conway AN, Muirhead KA, Wallace PK. Copyright 2018. The goal of the study shown in Figs. 6 and 7 was to compare extent of spectral overlap for U937 cells stained with equimolar concentrations (107 cells/mL, 10 μM dye) of five established proliferation dyes (CTV, CFSE, PKH67, PKH26, and CVC) with the then-new CTFR on two different cytometers. A preliminary titration found that under these conditions CTFR intensities were off-scale high in all detectors associated with 640 nm excitation. U937 cells were therefore labeled as described in Subheading 3.1 at final concentrations of 1 × 107 cells/mL and either 10 μM CFSE, 10 μM CTV, or 1.25 μM CTFR (the highest concentration for which this dye was fully on scale). All samples were analyzed 24 h post-labeling so that CFSE could be run at voltage settings that placed unlabeled U937 cells in the first decade for each fluorescence detector, as would typically be done for a dye dilution study. When evaluating compatibility of a given tracking dye with other fluors, key considerations include starting intensities of dye-labeled cells in the primary channel, spillover signal in non-primary channels, and cross-laser excitation. Compensation settings for the samples shown here and for the same samples run on a BD LSRII are shown in Fig. 9a, b, respectively. Percent compensation cannot be calculated where the peaks are completely off-scale high in their primary channel, as was seen for 10 μM CTFR. Although WinList v8.0 was also able to obtain a compensation value for samples stained with 2.5 μM or 5 μM CTFR, where the median intensity was on-scale, these values were too high because the nominal intensity in the primary channel was underestimated as the peak moved increasingly off-scale. The spillover profile of 10 μM CTFR, which was off-scale high in all detectors associated with the 640 nm laser on both cytometers, is shown in Fig. 10

(c) CTFR (Fig. 6) gives high levels of spillover in all 640 nmexcited channels. Both CTFR and CVC give modest cross-laser excitation in 405 nm-excited channels that are restricted by bandpass filters near their emission range (660 nm – 780 nm). For CVC, cross-laser excitation is also noted in the 355 nm-excited U740/35-A channel (Fig. 7);

Panel Design Considerations for Proliferation Dyes

185

Fig. 7 Spectral characteristics of selected membrane dyes on a BD LSR Fortessa cytometer (see Subheading 3.5.1). Adapted by permission from Springer Nature Customer Service Centre GmbH: Springer; Methods Mol Biol. Monitoring Cell Proliferation by Dye Dilution: Considerations for Probe Selection. Tario JD Jr., Conway AN, Muirhead KA, Wallace PK. Copyright 2018. U937 cells were labeled as described in Subheading 3.2 at final concentrations of 1 × 107 cells/mL and either 10 μM PKH67, 10 μM PKH26, or 10 μM CVC. Since membrane dyes exhibit minimal division-independent dye losses, all samples were analyzed immediately post-labeling at voltage settings that placed unlabeled U937 cells in the first decade for each fluorescence detector. Compensation settings required for PKH67 and PKH26 under these conditions are shown in Fig. 9a (Fortessa) and Fig. 9b (LSR II). The spillover profile of CVC, which was off-scale high in all detectors associated with the 640 nm laser on both cytometers, is shown in Fig. 10

(d) For PKH26 (Fig. 7), substantial spectral overlap is seen for all channels associated with the 488 nm laser except the 530/30-A channel, with negligible cross-laser excitation measured in any other channels. 5. For each cytometer of interest, plot median or geometric mean fluorescence intensity (GeoMFI) as a function of increasing excitation and detection wavelengths (Fig. 8). 6. For samples where dye positives are fully on-scale high at detector settings for which unstained cells are fully on-scale low, generate color-coded heat maps of instrument- and dye-specific compensation values. For each dye, access the compensation matrix generated from the analyses performed in step 4, copy/paste it into spreadsheet software such as Microsoft Excel, and apply “Conditional Formatting” as shown in Fig. 9 to allow comparison among dyes and cytometers. For example, on the Fortessa, >40% compensation is required for CFSE in the B575/26 detector where PE would be measured vs. ~30% for PKH67. On the LSRII, where PE

186

Joseph D. Tario et al.

Fig. 8 Spectral profiles of selected proliferation dyes on BD Fortessa and LSR II cytometers (see Subheading 3.5.1). Geometric mean fluorescence intensities (GeoMFI) were determined for each of the U937 samples analyzed in the study of Figs. 6 and 7 and plotted as a function of increasing laser and detector wavelengths on either a Fortessa (Panel A) or an LSR II (Panel B) cytometer, using a logarithmic intensity scale to allow comparison of indicated dyes with the autofluorescence of unstained cells. Autofluorescence profiles did not differ for cells exposed to Hanks Balanced Salt Solution (HBSS; vehicle used for CTV, CFSE and CTFR staining) and Diluent C (Dil C; vehicle used for PKH67, PKH26, and CVC staining) when using voltage settings that placed

Panel Design Considerations for Proliferation Dyes

187

would be measured in the Y582/15 detector, no significant compensation is required for either of the green proliferation dyes. And although the 10 μM PKH26 sample was fully on-scale in its primary channel (B575/26) on the Fortessa, where it is sub-optimally excited by the 50 mW 488 nm laser (~excitation efficiency @ 561 nm ~20% of maximum), it was completely off-scale high on the LSRII in both the Y582/15 and Y610/20 detectors on the LSR II, where it is much more optimally excited by the 50 mW 561 nm laser (~65% of maximum). 7. For cytometers with wider dynamic ranges, where it is possible to identify detector settings that keep stained and unstained populations fully on scale (e.g., Novocyte 3005), it is preferable to make comparisons of the type shown in Fig. 9. Alternatively, for cytometers with spectral unmixing and autofluorescencecorrection capabilities, spectral profiles or overlays of the type shown in Fig. 11 can aid in choosing appropriate fluorochrome combinations. For example, if violet-emitting antibodies are critical to a particular reagent panel, CVC may be a better choice than CTFR for proliferation monitoring since CVC gives substantially less violet cross-laser excitation (Figs. 10 and 11). 8. When fluorescence intensity distributions are partially or fully off-scale in the primary channel when the unlabeled distribution is placed in the first decade (e.g., CTFR and CVC at 10 μM), accurate values for % compensation cannot be determined. In such cases, raw intensity data can be used to qualitatively determine which detectors will be most affected by spectral overlap or cross-laser excitation from a given dye (Fig. 10). It does not matter which measure of central tendency is used (e.g., median, mean, geometric mean, etc.), provided ä Fig. 8 (continued) the autofluorescence of unstained cells in the lowest decade of each detector. Panel A. On the Fortessa, accurate GeoMFI values could not be determined for cells stained with 2.5, 5, or 10 μM CTFR or with 10 μM CVC because their peaks fell partially or completely off-scale at detector settings that placed autofluorescence fully on scale in the first decade. Cells stained with 10 μM PKH26, which is suboptimally excited at 488 nm, gave a distribution that was fully on scale in that dye’s primary detector (B575/26). Panel B. On the LSR II, accurate GeoMFI values could not be determined for cells stained with 2.5, 5, or 10 μM CTFR or with 10 μM CVC because their peaks fell partially or completely off-scale at detector settings that placed autofluorescence fully on scale in the first decade. Cells stained with 10 μM PKH26, which was more optimally excited by the 561 nm laser on this instrument, gave a distribution that was at the very high end of the scale but not off scale in Y582/15 and partially off scale in the adjacent Y610/20 detector. Comparing results for the same samples analyzed on both cytometers indicates that compensation for CTV spillover into the V525/50 channel will be more problematic on the LSR II than on the Fortessa and that substantial compensation will be required for CFSE or PKH67 spillover into the “PE” channel on the Fortessa (B575/26), while there is minimal spillover from either green dye into the “PE” channel on the LSR II (Y582/15)

188

Joseph D. Tario et al.

Fig. 9 Comparison among dyes and cytometers (see Subheading 3.5.1). (a) LSR Fortessa: percent compensation required as a function of dye (CTV, CFSE, PKH67, and PKH26) and concentration (CTFR). (b) LSRII: percent compensation required as a function of dye (CTV, CFSE, and PKH67) and concentration (CTFR)

Panel Design Considerations for Proliferation Dyes

189

Fig. 10 Geometric mean intensities in non-primary channels for off-scale dyes of Figs. 6 and 7 (see Subheading 3.5.1). Left panel, LSR Fortessa; right panel, LSRII

that all fluorescence measurements employ the same metric. Comparing the Fortessa and LSR II cytometers in this fashion indicates that: (a) While it would be possible to use CTFR or CVC with PE-Cy7 on the Fortessa, where it would be detected in B780/60, compensation is likely to be difficult-to-impossible on the LSRII where PE-Cy7 is detected in Y780/60; (b) Significant yellow cross-laser excitation is observed with both CVC and CTFR; (c) Less violet cross-laser excitation is observed with CVC than with CTFR. 3.5.2 Assessing the Impact of Fluorochrome Choice on Resolution of Antibody Positive and Negative Cells

Proliferation dyes typically give fluorescence intensities several orders of magnitude greater than the antibody reagents used for immunophenotyping. It is therefore critical to verify the tracking dye does not impair the ability to resolve subsets of interest. To ensure that antibody positive and negative populations are sufficiently well resolved and avoid excessive color compensation, good panel design principles include avoiding the use of dim

190

Joseph D. Tario et al.

Fig. 11 Spectral and cross-laser excitation profiles of far red proliferation dyes on 5-Laser Cytek Aurora. (see Subheading 3.5.1) Human PBMCs were labeled as in Subheadings 3.1 and 3.2 with CTFR or CVC (final concentrations: 107 cells/mL, 2 μM dye) and analyzed using the manufacturer’s CytekAssaySettings, in which the gains for all channels are set to provide optimal resolution of human CD8pos and CD4pos lymphocytes labeled with various fluorochromes. Panel A. Raw intensity data indicated that all five lasers gave measurable signal in the far red (650–830 nm; red bars along x-axis), with CTFR (upper plot) giving substantially greater violet cross-excitation than CVC (lower plot) but slightly less UV and yellow-green cross-excitation. Panel B. Greater violet laser cross-excitation for CTFR persisted when normalized, autofluorescence-corrected spectral profiles for each dye were overlaid using the Cytek Full Spectrum Viewer. Greater violet laser cross excitation was also seen for CTFR on the Fortessa and LSR II cytometers (Fig. 10), suggesting that this difference is dye specific rather than instrument specific

Panel Design Considerations for Proliferation Dyes

191

fluorochromes and weakly expressed antigens in adjacent spectral regions. Figure 12 illustrates the impact of optimal vs. sub-optimal panel design during setup for a simple four-color assay of CD4pos T lymphocyte proliferation after stimulation with anti-CD3 and antiCD28 to generate a polyclonal T cell response. 1. Generate a preparation of hPBMCs from peripheral blood using the laboratory’s standard density gradient fractionation protocol, with the addition of a final low speed wash (300 ×g) to minimize platelet contamination (see Note 1 and Ref. [48]). 2. Enumerate harvested cells using the laboratory’s preferred methodology. 3. Select candidate proliferation dye based on spectral compatibility with other fluorochromes to be used in the study and ability to achieve acceptable starting intensity without adverse effects on cell viability or function (see Notes 6 and 19). 4. Label hPBMCs with selected dye (see Subheadings 3.1 and 3.2). 5. Reserve a sufficient number of unlabeled hPBMCs and cells stained only with proliferation dye to serve as assay and compensation controls (Table 3). Generally, an equal number of dye-labeled and unlabeled cells are required. 6. Resuspend hPBMC labeled with proliferation dye in FCM buffer and incubate for 10 min with human IgG block (10 μg/test) to block Fc receptors. 7. Prepare autofluorescence control, single color compensation controls, and the fully stained test samples according to Table 3 and incubate for 20 min with the appropriate mAbs. 8. If necessary, lyse red blood cells with ACK lysing buffer and wash twice in FCM Buffer. 9. Add appropriate viability dye (e.g., 4 μL of 100 μg/mL 7-AAD working stock per 100 μL of cells for a final concentration of 4 ug/mL or 1 μL of 1 μM TOPRO-3 working stock per 100 μL of cells for a final concentration of 10 nM final) (see Notes 36 and 37). 10. Collect cells on the flow cytometer as defined in in steps 2–5 of Subheading 3.5.1. 11. Analyze samples on flow cytometer using the gating strategy shown in Fig. 12 and use the instrument setup controls listed in Table 3 to assess the ability to resolve the antibody positive and negative populations of interest.

192

Joseph D. Tario et al.

Fig. 12 Effect of panel design on ability to immunophenotype lymphocytes labeled with PKH26 on a BD Fortessa cytometer (see Subheading 3.5.2) Adapted from Ref. [4] and used by permission from Tario, J.D., Humphrey, K., Bantly, A.D., Muirhead, K.A., Moore, J.S., Wallace, P.K. Optimized Staining and Proliferation Modeling Methods for Cell Division Monitoring using Cell Tracking Dyes. J. Vis. Exp. (70), e4287, doi:10.3791/ 4287 (2012). Human PBMCs were isolated from 24-hr old blood and replicate aliquots were labeled with PKH26 as described in Subheading 3.2. Post-staining viabilities were similar for all samples (88–92%). Immediately after labeling with PKH26, cells were counterstained with two different combinations of immunophenotypic and viability reagents and analyzed on an LSRFortessa flow cytometer with the following optical configuration: 488 nm laser: FSC-A (488 nm); SSC-A (488/10 BP), FITC-A (530/30 BP); PKH26-A (575/26 BP); 7-AAD-A or PerCP-A (695/40 BP). 640 nm laser: APC-A or TO-PRO-3-A (670/14 BP). Single color controls (Table 3) were used at the time of data acquisition to perform color compensation using BD DiVa software. Data from the PKH26neg control (gray filled histograms; Table 3, Tube 7) are overlaid for reference in panels A(v), B(v), and C(v). Panel A. Cells labeled with PKH26 at a final concentration of 2 μM dye and 1 × 107 cells/mL were counterstained using anti-CD3-FITC, anti-CD4-APC, and 7-AAD. After gating on viable (7-AADneg) CD3pos lymphocytes (i) and exclusion of debris and aggregates based on FSC and SSC (not shown), PKH26 intensity was evaluated in combination with anti-CD4-APC. Whether uncompensated (ii) or compensated (iii), this fluorochrome combination resulted in good resolution between CD4pos T cells and CD4neg T cells, as verified both by a fluorescence minus one (FMO) control stained with all the reagents except antiCD4-APC (iv). Viable PKH26posCD3pos lymphocytes were well resolved from unstained PBMCs (v). Panel B. Using the same fluorochrome combination as in Panel A, but increasing the final PKH26 concentration to 4 μM did not adversely affect the ability to resolve CD4pos T cells from CD4neg T cells. Panel C. A replicate aliquot of cells independently labeled with PKH26 at a final concentration of 2 μM dye and 1 × 107 cells/mL was counterstained using anti-CD3-FITC, anti-CD4-PerCP, and TO-PRO-3. After gating on viable (TO-PRO-3neg) CD3pos lymphocytes (i) and exclusion of debris and aggregates based on FSC and SSC (not shown), PKH26

Panel Design Considerations for Proliferation Dyes

193

3.6 Evaluating the Effect of Tracking Dye Labeling on Cellular Function(s)

Regulatory T cells (Treg) exert potent immunosuppressive effects in autoimmune diseases, transplantation, and graft-versus-host disease [49], inhibiting proliferation of effector T cells (Teff) primarily by downregulating induction of their IL-2 mRNA [50]. Phenotypically, Treg are defined by their co-expression of CD3, CD4, CD25, the transcription factor FoxP3, and dim expression of CD127, along with several other surface markers shared with activated T cells such as GITR and CTLA-4 [49, 51, 52]. As reviewed by Brusko et al. [53], assays that use tracking dyes to monitor Treg suppression of anti-CD3 plus IL-2-induced effector T cell (Teff) proliferation have significant advantages over in vitro suppression assays using 3H-thymidine, a standard measure of Treg activity. In particular, although they require approximately ten-fold more cells, tracking dye-based assays reflect total Teff proliferation throughout the 4 day culture period rather than simply measuring DNA synthesis during the final hours of the response and can be extended to simultaneously monitor low-level Treg proliferation as well [53]. Our experience with a single color in vitro suppression assay has been that it can be difficult to reliably distinguish highly proliferated CFSEdim Teff from unlabeled Treg, since both populations express similar levels of CD4. Use of a second tracking dye has the advantage of not only simplifying discrimination between Treg and CFSEdim Teff but also allowing assessment of whether increasing numbers of Teff in the assay have any effect on Treg proliferation. In the variation described here, isolation of CD4+ Treg and Teff by sorting was combined with CFSE labeling of Teff and CVC labeling of Treg to ascertain the effect of Treg:Teff ratio on the proliferative response of each cell type (Fig. 13; see Notes 38–40). Parallel studies using unstained Treg confirmed that their ability to suppress Teff proliferation was unaltered by labeling with CVC (Fig. 14).

3.6.1 Preparation of Monocyte Depleted Lymphocytes (See Note 41)

1. Prepare TRIMA filtrate by draining TRIMA filter into a 50 mL conical tube, followed immediately by rinsing filter with 40 mL of 10% ACD in PBS to dislodge trapped cells [54] (see Note 5). 2. Isolate hPBMC from the TRIMA filtrate using the laboratory’s standard density gradient fractionation protocol, with addition of a final low speed wash (300 ×g) to minimize platelet contamination (see Note 1).

ä Fig. 12 (continued) intensity was evaluated in combination with anti-CD4-PerCP. Substantial spectral overlap of PKH26 into the PerCP channel is evident in the uncompensated data (ii), and resolution between PKH26pos CD4pos and PKH26posCD4neg events is marginal even after compensation is applied (iii and iv). With this reagent combination, CD4posPKH26pos events could no longer be resolved from CD4negPKH26pos T cells when PKH26 concentration was increased to 4 μM because spillover signal from PKH26 (em. ~575 nm) into the PerCP channel (em. ~ 675 nm) exceeded the signal from anti-CD4-PerCP (not shown)

194

Joseph D. Tario et al.

Fig. 13 Simultaneous analysis of Teff and Treg proliferation during an in vitro suppression assay (see Subheading 3.6). Adapted with permission from Springer Nature Customer Service Centre GmbH: Springer; Methods Mol Biol. Tario JD, Jr., Muirhead KA, Pan D, Munson ME and Wallace PK Tracking immune cell proliferation and cytotoxic potential using flow cytometry. Copyright 2011. After isolation by sorting as described in Ref. [5], Teff cells were stained with CFSE (final concentrations: 5 × 107 cells/mL, 5 μM dye) and co-incubated for 4 days with sorted Treg stained with CVC (final concentrations: 1 × 106 cells/mL, 1 μM

Panel Design Considerations for Proliferation Dyes

195

3. To separate lymphocytes from monocytes via cold aggregation [55, 56] resuspend hPBMC in 50 mL of cold CM and dispense 12.5 mL each into four 15 mL conical polypropylene tubes. Affix tubes onto the fins of tube rotator and rotate along their horizontal axis, parallel to the benchtop, at 18 rpm at 4 °C to induce monocyte aggregation (see Note 42). After 30 to 45 min visible 1–3 mm aggregates will form that contain primarily monocytes. 4. Remove tubes from the rotator and place vertically on ice for 15 min., permitting aggregated cells to precipitate at 1 g to the bottom of each tube. 5. Harvest supernatant containing the monocyte depleted lymphocytes, wash twice with cold HBSS, and use for isolation of Treg, Teff and accessory cells (see step 2 in Subheading 3.6.2 and Notes 43 and 44). 3.6.2 Isolation of Treg, Teff, and Accessory Cells by Flow Cytometry and Sorting (See Note 45)

1. Adjust monocyte depleted lymphocytes from Subheading 3.6.1, step 5 to 5 × 107 cells/mL in HBSS and incubate for 10 min with 600 μg/mL human IgG to block Fc receptor binding. 2. Add a mAb cocktail containing anti-CD127 PE, anti-CD4 PECy7, and anti-CD25 APC to the IgG-blocked lymphocytes and incubate on ice for 30 min (see Note 46). 3. Wash cells twice with HBSS and resuspend at 1.5 × 107 cells/ mL in HBSS. 4. Sort antibody-labeled cells on a fluorescence-activated cell sorter (e.g., FACSAria II or equivalent) into glass tubes containing CM at a rate that provides for purities of 95% or greater (see Note 47). The gating logic used to sort Treg, Teff, and accessory cells is illustrated in Ref. [5].

ä Fig. 13 (continued) dye) in the presence of anti-CD3, anti-CD28 and sorted, irradiated accessory cells (see Subheading 3.6.3 for details). Representative data are shown for one of three triplicate samples at a Treg:Teff ratio of 0.25:1. LIVE/DEAD Fixable Violet was used to exclude dead cells (R1, upper left plot; accessory cells = red-brown, non-viable Teff = gray and non-viable Treg = red) from all other data plots. CVC staining was used to distinguish viable Treg (R4, center right plot) from viable but highly proliferated Teff (R5, center right plot). A single parameter CFSE (530/30) proliferation profile for Teff (lower left plot) was generated by gating on cells that were CFSEpos (R5), CD4pos (R3), viable (not R1), and had lymphocyte scatter properties (R2). A single parameter CVC (660/20) proliferation profile for Treg was generated by gating on cells that were CVCpos (R4), CD4pos (R3), viable (not R1), and had lymphocyte scatter properties (R2). Note the generous lymphocyte region (R2) defined to include lymphocyte blasts. Proliferative fractions, representing the percent of cells that have undergone one or more divisions, were calculated as described in Subheading 3.6.5 (R6 = 78.6% and R7 = 37.6% for Teff and Treg, respectively). Proliferative indices (PI), representing foldexpansion of Teff and Treg populations during the culture period, were calculated as described in Subheading 3.6.5 using Gaussians to model each of the generational peaks (e.g., blue = parental generation, orange = first daughter generation, etc.)

196

Joseph D. Tario et al.

Fig. 14 Inhibition of Teff proliferation by Treg cells (see Subheading 3.6). Adapted with permission from Springer Nature Customer Service Centre GmbH: Springer; Methods Mol Biol. Tario JD, Jr., Muirhead KA, Pan D, Munson ME and Wallace PK Tracking immune cell proliferation and cytotoxic potential using flow cytometry. Copyright 2011. Teff cells were stained with CFSE and incubated with graded ratios of Treg cells in the presence of anti-CD3, anti-CD28, and accessory cells, as described in Fig. 13. The maximum proliferative potential of Teff was assessed in the absence of Treg cells (Treg:Teff ratio of 0:1). As increasing numbers of Tregs were added to the culture system, increasing inhibition of Teff cell proliferation was observed, as expected. Similar results were obtained with both CVC stained (solid line) or unstained (dashed line) Treg, indicating that staining with the CVC tracking dye did not affect the potency of inhibition by Treg cells. Proliferative fraction (%P) and Proliferative Index (PI) were determined as described in Fig. 13 and Subheading 3.6.5. Data points in panels A and B represent the mean ± 1 standard deviation of triplicate samples. Panel A. Effect of Treg:Teff ratio on Proliferative Fraction of Teff. Panel B. Effect of Treg:Teff ratio on Proliferative Index of Teff. The same samples were also used to monitor proliferation of CVC stained Treg cells at different Treg: Teff ratios. Treg are generally anergic and, as expected, did not proliferate when incubated with anti-CD3, anti-CD28 and accessory cells in the absence of Teff cells (Treg:Teff ratio of 1:0). However, as the proportion of Teff in the cultures was increased (i.e., as the Treg:Teff ratio decreased), the extent of Treg proliferation also increased (Fig. 10 of Ref. [5])

Panel Design Considerations for Proliferation Dyes 3.6.3 Proliferation Protocol

197

1. Stain sorted Treg with CVC (final cell concentration of 1 × 106/mL; final dye concentration 1 μM) according to the procedures described in Subheading 3.2. Wash in CM, count, and adjust to 1 × 106 cells/mL. 2. Stain sorted Teff cells with CFSE (final cell concentration 5 × 107/mL; final dye concentration 5 μM) according to the procedures described in Subheading 3.1. Wash in CM, count, and adjust to 5 × 105 cells/mL. 3. In a 96-well round bottom plate, make triplicate serial 1:2 dilutions of the Treg as follows: Pipet 200 μL of the stained Treg suspension into the first well and 100 μL of CM into an adjacent set of 4 wells. Serially transfer 100 μL of Tregs from the first well to the second, then from the second to the third etc. ending with a transfer of 100 μL from the fourth well to the fifth well and removal of 100 μL of cell suspension from the fifth well (see Note 48). 4. Add 100 μL of stained Teff to each well, creating Treg to Teff ratios of 2:1, 1:1, 0.5:1, 0.25:1, and 0.125:1 (see Note 49). 5. Add 50 μL of Treg and 100 μL of Teff cell to the Treg-only and Teff-only wells, respectively (see Notes 50 and 51). 6. Centrifuge sorted accessory cells (400 ×g), pool into a 50 mL conical tube, adjust to 1 × 106 cells/mL with CM, and irradiate with 3000 rad of gamma irradiation to inhibit proliferation. After irradiation, adjust concentration to 5 × 105 cells/mL in CM. 7. To an aliquot of accessory cells commensurate with the size of the experiment, add azide-free anti-CD3 (clone OKT3) to a final concentration of 3 μg/mL and anti-CD28 (clone 28.2) to a final concentration of 1.5 μg/mL. Add 0.1 mL of this preparation to each test well from step 4, yielding a final concentration of 1 μg/mL anti-CD3 and 0.5 μg/mL anti-CD28 in a final volume of 0.3 mL/well. 8. Add CM to bring each well to a final volume of 0.3 mL and incubate in a humidified 37 °C incubator with 5% CO2 for 96 h (see Note 51). 9. After the 96 h incubation, remove the plate from the incubator and harvest cells from each well into individually labeled 12 × 75 mm round bottom tubes compatible with the laboratory’s flow cytometer and place on ice. Rinse each well with 200 μL cold HBSS, adding with the appropriate tube. 10. Centrifuge at 400 ×g and resuspend each pellet in 100 μL of cold FCM buffer, adding 10 μL of human IgG to block Fc receptor binding.

198

Joseph D. Tario et al.

11. Incubate for 10 min on ice and then label with anti-CD4 PECy7 (clone SK3) and 5 μL of LIVE/DEAD® Fixable Violet reagent (see Subheading 2.3, item 4 and Note 52). 12. Incubate for 30 min on ice and then wash 2 times with FCM buffer. Resuspend cells in 300 μL of FCM buffer for flow cytometric analysis. 3.6.4 Flow Cytometric Data Acquisition and Analysis

1. Establish appropriate voltage settings using autofluorescence and single color controls from Table 4 (see Notes 50 and 54). 2. Using the single color controls from Table 4, adjust compensation settings according to the laboratory’s and/or instrument manufacturer’s standard procedures (see Notes 5 and 54). 3. Acquire data on the flow cytometer using the gating strategy shown in Fig. 13 (see Note 50, Table 4).

3.6.5 Calculation of Proliferative Fraction and Proliferative Index

Either Proliferative Fraction (%P), a semi-quantitative estimate of percent proliferating cells, or Proliferative Index (PI), a more quantitative estimate of fold population expansion, may be used to analyze extent of proliferation. In either approach, the starting point is a single parameter tracking dye dilution profile for the appropriate subpopulation of viable lymphocytes (here CFSE for Teff and CVC for Treg), created using the gating strategy described in Fig. 13. 1. Calculation of Proliferative Fraction (%P). To calculate %P, a stained, unstimulated control is used to set the upper boundary for enumeration of daughter cells, selecting an intensity that gives an acceptably low value for dividing cells in the absence of stimulus (e.g., 1–5%, see Ref. [5]). An unstained control is used to define the lower boundary for enumeration of proliferating cells, selecting an intensity that gives an acceptably low value for dividing cells in the absence of proliferation dye. The Proliferative Fraction is then defined as the percentage of proliferating cells with fluorescence intensity less than the stained but unstimulated control and more than the unstained control. 2. Calculation of Proliferative Index (PI). To calculate PI, a specifically designed peak modeling software such as ModFit LT (Verity Software House, Topsham, ME), FCS Express (De Novo Software, Los Angeles, CA), or FlowJo (TreeStar, Ashland, OR) is used to fit the viable, lymphocyte-gated, single parameter CFSE and CVC data. These programs use a nonlinear least squares analysis to iteratively find the best fit to the raw data by changing the position, height, and CV of each generational Gaussian. After loading and gating the histogram, users

Panel Design Considerations for Proliferation Dyes

199

define the location of the parental generation, its spacing and if necessary, its SD. When modeling lipophilic dyes, an equal spacing between generations is assumed, whereas when modeling CFSE, an unequal spacing must be assumed to adjust for observed non-linearities in peak spacing (possibly due to continued slow dye loss even after 24 h; Fig. 4). The area under each Gaussian is taken as a measure of the relative number of cells in that generation and the sum of all Gaussians correspond to the relative number of cells in the total population. These values are then used internally by the software to calculate the PI. 3. Calculation of percent suppression. The degree of suppression observed when Treg cells are co-cultured with Teff cells is calculated as: % Suppression = 1 -

Px TregþTeff Px Teff

× 100%

where Px can be either %P or PI but not simultaneously both.

4

Notes 1. Platelets present in variable amounts act as “hidden” sources of added protein or membrane that can affect labeling efficiency even when hPBMC and dye concentrations are carefully reproduced. Addition of a final low-speed wash step (5 min at 300 ×g) minimizes platelet contamination of hPBMC and improves consistency of staining with both protein and membrane labeling dyes. 2. Maintain cell cultures in logarithmic growth phase in CM using a fully humidified 37 °C incubator with 5% CO2. 3. Commercially available single use vials of CellTrace™ dyes offer the convenience of pre-weighed amounts of dye for dissolution in small volumes (18 μL) of DMSO but cost significantly more per mg than bulk dye. CFSE is available as a bulk powder reagent, and if purchased, it should be accurately weighed out and made up as a 5 mM stock solution (MW 557.47 g/mol) in freshly opened anhydrous DMSO. Aliquots of 5 mM CFSE dye stock in DMSO can be stored in a desiccator at -20 °C for several months. Repeated freezing and thawing of a given aliquot should be avoided since DMSO takes up moisture from the air and reduces labeling efficiency, due to hydrolysis of both the diacetate ester moieties required for entry into cells and the succinimidyl ester moieties required for covalent reaction with amino groups under physiologic conditions. If the entire contents of a bulk dye vial are dissolved

200

Joseph D. Tario et al.

in a calculated volume of DMSO, final dye concentration should be confirmed spectrophotometrically (e.g., by absorption at 490 nm) and adjusted as needed for consistency, since exact weights contained may vary sufficiently from vial to vial to require re-titration of new vs. old dye stocks in order to avoid toxicity. 4. Lymphocytes and monocytes are typically isolated from anticoagulated blood using standard Ficoll Hypaque density centrifugation techniques, but cryopreserved PBMCs, adherent cell lines (harvested using trypsinization), and non-adherent lines are also suitable for staining. Adherent cells may be labeled by flooding the culture dish or flask with dye solution. However, this typically gives considerably more heterogeneous intensity distributions, especially for membrane dyes [57], and makes interpretation of dye dilution profiles more complex. Labeling of single cell suspensions is therefore generally preferred. 5. Labeled cells are typically placed back into culture for in vitro assays or injected into animal models for in vivo functional studies. Standard sterile technique should therefore be followed throughout the labeling protocols described in Subheadings 3.1 and 3.2. 6. Amount of dye required for bright but non-toxic staining will in general increase as total number and/or size of cells to be stained increases. However, exact concentrations resulting in over-labeling and loss of function will vary depending on cell type and class of tracking dye use (e.g., Table S1 in Ref. [38]). Therefore, viability and functionality of labeled cells should always be verified by comparing with unlabeled cells (e.g., Figs. 1, 2A, 14). Similarly, both cell and dye concentrations used for labeling should be reported in any publication. 7. Total number of cells to be stained will depend on the number of replicates and controls required by the experimental protocol. Staining intensities are most easily reproduced when staining is done in volumes ranging from 0.5 to 2.0 mL. Once an approximate cell concentration has been established based on these factors, a preliminary dye titration experiment is recommended to determine or verify the optimal concentration of tracking dye [5, 58, 59]. 8. Obtaining reproducible starting intensities from study-tostudy requires accurately reproducing both dye and cell concentrations. Cell counting using a Coulter counter or other automated cell counter is recommended rather than manual counting using a hemocytometer, since results of replicate hemocytometer counts often vary by as much as 15–20%. 9. Exogenous protein reduces labeling efficiency for both protein and membrane dyes and is therefore normally removed by washing the cells with a protein-free buffer such as PBS or

Panel Design Considerations for Proliferation Dyes

201

HBSS prior to staining. However, when protein dye labeling must be done at relatively low cell concentrations due to limited numbers of cells or other experimental concerns, addition of exogenous protein may aid in protecting against overlabeling and resultant loss of cell viability or functionality [21]. If addition of exogenous protein must be avoided due to other experimental considerations, the working dye stock prepared in Subheading 3.1, step 3 may be further diluted in buffer prior to initiation of cell labeling in Subheading 3.1, step 4. Alternatively, resuspension in serum-free culture medium will also reduce labeling efficiency due to the presence of free amino acids that compete with cellular proteins for dye binding, although extent of reduction varies for different protein dyes [60]. In either case, the time between initial dilution and initiation of cell labeling should be minimized since hydrolysis begins immediately upon dilution of the DMSO stock into aqueous solution and proceeds very rapidly. 10. If bulk CFSE powder has been previously dissolved in DMSO and frozen, ensure that the aliquot to be used is completely thawed prior to preparation of the working stock, but minimize the length of time that the DMSO stock is exposed to ambient conditions to limit uptake of moisture. The CFSE working stock solution should be clear and colorless. If there is any sign of yellowing, it should not be used, since this indicates conversion to the charged fluorescent hydrolysis product, carboxyfluorescein, which will not enter cells. 11. These concentrations were chosen such that following a 24 h stabilization period, the fluorescence intensity of non-dividing lymphocytes should be completely onscale at the upper end of the intensity scale when unstained cells are placed in the first decade (see Notes 18 and 34). 12. For an hPBMC concentration of 1 × 107/mL, final concentrations of up to 1.0 μM for CFSE, CTV, or CTFR are recommended to avoid over-labeling. These dyes label proteins at random sites and unintended modification of critical residues can interfere with signal transduction pathways, proliferation, and other cell functions even when cell viability remains acceptable (see Table S1 in Ref. [38]). More extensive labeling increases the likelihood of altered cell function(s), and the extent of labeling is a function of dye concentration, cell concentration, labeling time, and labeling conditions (temperature, mixing, etc.). Final cell and dye concentrations given here should therefore be taken only as a starting point and verified in each user’s experimental system.

202

Joseph D. Tario et al.

13. Because uptake into cells and reaction with free amino groups occurs rapidly, it is important to disperse the dye solution quickly and evenly throughout the cell suspension immediately after addition. 14. Once formed by hydrolysis, the fluorescent forms of CTV, CFSE, and CTFR are sensitive to photobleaching. Therefore, covering with aluminum foil or placing in a darkened location is recommended to protect tubes or wells containing labeled cells from exposure to high intensity light or prolonged exposure to room light. 15. Inclusion of protein in the stop solution is essential, since it reacts with and inactivates unbound dye. Free amino acids in culture medium further aid in the inactivation. Alternatively, PBS or HBSS containing 1–2% serum albumin may be used as a stop solution. 16. For starting cell numbers of 107 or more, recoveries of at least 85% and viabilities of at least 90% should be obtained for freshly drawn hPBMC (e.g., Fig. 1a and Table 3 in Ref. [61]). However, recoveries typically decrease at lower cell numbers and may also be lower for preparations in which the cells are older or have been subjected to other stresses (e.g., pheresis, elutriation, or cryopreservation and thawing). Staining intensity and CV will vary for different cell types, but a bright, symmetrical fluorescence intensity profile coupled with poor recovery and/or viability usually indicates substantial over-labeling and the need to increase cell concentration, decrease dye concentration, or both. Conversely, heterogeneous and/or dim staining (