New Technologies for Glutamate Interaction: Neurons and Glia (Neuromethods, 207) [2024 ed.] 107163741X, 9781071637418

Citation preview

Neuromethods 2780

Maria Kukley  Editor

New Technologies for Glutamate Interaction Neurons and Glia

NEUROMETHODS

Series Editor Wolfgang Walz University of Saskatchewan Saskatoon, SK, Canada

For further volumes: http://www.springer.com/series/7657

Neuromethods publishes cutting-edge methods and protocols in all areas of neuroscience as well as translational neurological and mental research. Each volume in the series offers tested laboratory protocols, step-by-step methods for reproducible lab experiments and addresses methodological controversies and pitfalls in order to aid neuroscientists in experimentation. Neuromethods focuses on traditional and emerging topics with wide-ranging implications to brain function, such as electrophysiology, neuroimaging, behavioral analysis, genomics, neurodegeneration, translational research and clinical trials. Neuromethods provides investigators and trainees with highly useful compendiums of key strategies and approaches for successful research in animal and human brain function including translational “bench to bedside” approaches to mental and neurological diseases.

New Technologies for Glutamate Interaction Neurons and Glia

Edited by

Maria Kukley Science Park of the UPV/EHU, Achucarro Basque Center for Neuroscience, Leioa, Spain

Editor Maria Kukley Science Park of the UPV/EHU Achucarro Basque Center for Neuroscience Leioa, Spain

ISSN 0893-2336 ISSN 1940-6045 (electronic) Neuromethods ISBN 978-1-0716-3741-8 ISBN 978-1-0716-3742-5 (eBook) https://doi.org/10.1007/978-1-0716-3742-5 © 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.

Preface to the Series Experimental life sciences have two basic foundations: concepts and tools. The Neuromethods series focuses on the tools and techniques unique to the investigation of the nervous system and excitable cells. It will not, however, shortchange the concept side of things as care has been taken to integrate these tools within the context of the concepts and questions under investigation. In this way, the series is unique in that it not only collects protocols but also includes theoretical background information and critiques which led to the methods and their development. Thus, it gives the reader a better understanding of the origin of the techniques and their potential future development. The Neuromethods publishing program strikes a balance between recent and exciting developments like those concerning new animal models of disease, imaging, in vivo methods, and more established techniques, including, for example, immunocytochemistry and electrophysiological technologies. New trainees in neurosciences still need a sound footing in these older methods in order to apply a critical approach to their results. Under the guidance of its founders, Alan Boulton and Glen Baker, the Neuromethods series has been a success since its first volume published through Humana Press in 1985. The series continues to flourish through many changes over the years. It is now published under the umbrella of Springer Protocols. While methods involving brain research have changed a lot since the series started, the publishing environment and technology have changed even more radically. Neuromethods has the distinct layout and style of the Springer Protocols program, designed specifically for readability and ease of reference in a laboratory setting. The careful application of methods is potentially the most important step in the process of scientific inquiry. In the past, new methodologies led the way in developing new disciplines in the biological and medical sciences. For example, physiology emerged out of anatomy in the nineteenth century by harnessing new methods based on the newly discovered phenomenon of electricity. Nowadays, the relationships between disciplines and methods are more complex. Methods are now widely shared between disciplines and research areas. New developments in electronic publishing make it possible for scientists that encounter new methods to quickly find sources of information electronically. The design of individual volumes and chapters in this series takes this new access technology into account. Springer Protocols makes it possible to download single protocols separately. In addition, Springer makes its print-on-demand technology available globally. A print copy can therefore be acquired quickly and for a competitive price anywhere in the world. Saskatoon, SK, Canada

Wolfgang Walz

v

Preface Anyone who stops learning is old, whether at twenty or eighty. Anyone who keeps learning stays young. The greatest thing in life is to keep your mind young. Henry Ford

In the central nervous system, synaptic communication between neurons determines information flow within neuronal networks. In mammals, the amino acid glutamate is the major excitatory neurotransmitter at synapses. Glutamate is released from presynaptic boutons and can be detected by fast ionotropic glutamate receptors of the AMPA, NMDA, or kainate type on postsynaptic neurons, triggering fast excitatory postsynaptic currents (EPSCs). Glial cells are indispensable components of neuronal networks, and they are integrated into these networks in multiple ways. We know very well that astroglial cells extend their tiny processes carrying glutamate transporters to neuronal synapses and perform fast removal of synaptically released glutamate, shaping EPSCs. Astroglial cells also express metabotropic glutamate receptors which trigger intracellular calcium transients, or even calcium waves in glial networks, upon binding of glutamate. Astrocytic calcium signaling is implicated in several important brain functions including regulation of breathing, sleep, and sensory activity. Less attention has been given to the phenomenon of neuron-glia synapses and the functional role of these synapses in neuronal and neuron-glia networks. In 2000, Dwight Bergles and colleagues pioneered the discovery of true synaptic connections between neurons and oligodendrocyte progenitor cells in the hippocampus. At these neuron-glia synapses, fast vesicular release of glutamate occurs from a presynaptic (neuronal) site and is detected by postsynaptic (glial) AMPA receptors, triggering neuron-glia EPSCs with the kinetic properties that are very similar to neuronal EPSCs. Subsequently, neuron-glia synapses have been described in many regions of the central nervous system including the spinal cord and even the white matter of animals and humans. Furthermore, later studies discovered that synaptic communication occurs even between neurons and glioma cells. Despite these amazing discoveries, the awareness of the neuroscience community about neuron-glia synapses and the drive to study them remains relatively low. The major idea of putting together this volume of the Neuromethods series is to emphasize the co-existence of neuronal and neuron-glia synapses in the brain. The volume focuses on major advances in both fields, and consists of three parts. The first part addresses recent advances in the technical approaches for studying glutamatergic synapses between neurons. The second part highlights state-of-the art approaches to study functional role of astrocytes at neuronal synapses. The third part focuses on fast signaling at neuron-glia synapses, and highlights relevant methods with particular focus on slice electrophysiology and in vivo gene delivery techniques.

vii

viii

Preface

I hope that this collection of articles will inspire both neuronal and glial physiologists and foster interactions to combine their efforts in developing new methods and approaches for understanding neuronal and neuron-glia synaptic transmission, which are among the most fascinating phenomena in the central nervous system. I thank all the authors and all the reviewers who helped me to assemble this book. Leioa, Spain

Maria Kukley

Contents Preface to the Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

PART I

v vii xi

NEW APPROACHES TO STUDY GLUTAMATERGIC SYNAPSES BETWEEN NEURONS

1 Visualization of Glutamatergic Neurotransmission in Diverse Model Organisms with Genetically Encoded Indicators. . . . . . . . . . . . . . . . . . . . . . 3 Abhi Aggarwal, Joyce Chan, Amelia K. Waring, Adrian Negrean, Jonathan S. Marvin, Kaspar Podgorski, and Loren L. Looger 2 Surface Glutamate Receptor Nanoscale Organization with Super-Resolution Microscopy (dSTORM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Joana Ferreira and Laurent Groc 3 Ligand-Directed Chemical Labeling for Visualizing and Analyzing AMPA Receptors in Neurons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Shigeki Kiyonaka, Kyohei Soga, Kento Ojima, Hiroshi Nonaka, and Itaru Hamachi 4 Live FRET-FLIM Imaging to Study Metabotropic Signaling via the NMDA Receptor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Mehreen Manikkoth and Kim Dore 5 Quantitative Analysis of Single Glutamatergic Vesicles in the Brain . . . . . . . . . . . . 91 Yuanmo Wang, Ajay Pradhan, Pankaj Gupta, Hanna Karlsson-Fernberg, and Ann-Sofie Cans 6 Imaging Microtubule Network in Rodent Giant Glutamatergic Presynaptic Terminals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 Laurent Guillaud 7 Photooxidation of Genetically Encoded MiniSOG-Fused VGLUT2 for Identification of Glutamatergic Synapses by Transmission and 3D Electron Microscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 Andrew J. Flores, Anna Nilsson, Junru Hu, Mason R. Mackey, Mark H. Ellisman, Thomas S. Hnasko, and Daniela Boassa

PART II

NEW APPROACHES TO STUDY ROLE OF ASTROCYTES AT NEURONAL SYNAPSES

8 Chemogenetic Approaches to Study Astrocytes at Glutamatergic Synapses . . . . . 155 Liam Nestor, Yana Van Den Herrewegen, Zuner A. Bortolotto, Dimitri De Bundel, and Ilse Smolders

ix

x

Contents

9 Studying the Role of Astrocytes at Synapses Using Single-Cell Transcriptomics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 Francisco Pestana, T. Grant Belgard, Thierry Voet, and Matthew G. Holt 10 Analysis of Synaptic Glutamate Clearance as a Possible Indicator of Synaptic Health in the Degenerating Rodent Brain . . . . . . . . . . . . . . . . . . . . . . . 207 Anton Dvorzhak and Rosemarie Grantyn 11 Computational Models of Astrocyte Function at Glutamatergic Synapses . . . . . . 229 Kerstin Lenk, Audrey Denizot, Barbara Genocchi, Ippa Sepp€ a l€ a, Marsa Taheri, and Suhita Nadkarni

PART III TECHNICAL APPROACHES TO STUDY FAST SIGNALING NEURON-GLIA SYNAPSES

AT

12

13 14

15

16

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Coram Guevara, Rodrigo Varas, Marı´a Cecilia Angulo, and Fernando C. Ortiz Synaptic Integration at Neuron-OPC Synapses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wenjing Sun Mapping Synaptic Inputs to Oligodendroglial Cells Using In Vivo Monosynaptic Viral Tracing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Belgin Yalc¸ın and Michelle Monje In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA Receptors in Oligodendrocyte Lineage Cells . . . . . . . . . . . . . . . . . . . . . . Ting-Jiun Chen, Bartosz Kula, and Maria Kukley Studying Synaptic Integration of Glioma Cells into Neural Circuits . . . . . . . . . . . Kiarash Shamardani, Kathryn R. Taylor, Tara Barron, and Michelle Monje

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

267

283

301

313 345

369

Contributors ABHI AGGARWAL • Allen Institute for Neural Dynamics, Seattle, WA, USA; Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, VA, USA; Cumming School of Medicine, University of Calgary, Calgary, AB, Canada MARI´A CECILIA ANGULO • Universite´ Paris Cite´, Institute of Psychiatry and Neuroscience of Paris (IPNP), INSERM U1266, Paris, France; GHU PARIS Psychiatrie et Neurosciences, Paris, France TARA BARRON • Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, USA; Howard Hughes Medical Institute, Stanford University, Palo Alto, CA, USA T. GRANT BELGARD • The Bioinformatics CRO, Orlando, FL, USA DANIELA BOASSA • Department of Neurosciences, University of California San Diego School of Medicine, La Jolla, CA, USA; National Center for Microscopy and Imaging Research, University of California San Diego, La Jolla, CA, USA ZUNER A. BORTOLOTTO • School of Physiology, Pharmacology and Neuroscience, University of Bristol, Bristol, UK ANN-SOFIE CANS • Department of Chemistry and Chemical Engineering, Chalmers University of Technology, Gothenburg, Sweden JOYCE CHAN • Department of Neurosciences, University of California, San Diego, San Diego, CA, USA TING-JIUN CHEN • Department of Psychiatry, Icahn School of Medicine at Mount Sinai, New York, NY, USA DIMITRI DE BUNDEL • Department of Pharmaceutical Chemistry, Drug Analysis and Drug Information, Research Group Experimental Pharmacology, Center for Neurosciences (C4N), Vrije Universiteit Brussel (VUB), Brussels, Belgium AUDREY DENIZOT • Okinawa Institute of Science and Technology, Computational Neuroscience Unit, Onna-Son, Japan; AIstroSight, Inria, Hospices Civils de Lyon, Universite´ Claude Bernard Lyon 1, Villeurbanne, France; University of Lyon, LIRIS UMR5205, Villeurbanne, France KIM DORE • Department of Neurosciences, Center for Neural Circuits and Behavior, University of California San Diego, La Jolla, CA, USA ANTON DVORZHAK • Neuroscience Research Center, Charite´ – University Medicine Berlin, Berlin, Germany MARK H. ELLISMAN • Department of Neurosciences, University of California San Diego School of Medicine, La Jolla, CA, USA; National Center for Microscopy and Imaging Research, University of California San Diego, La Jolla, CA, USA JOANA FERREIRA • Interdisciplinary Institute for Neuroscience – IINS, CNRS UMR 5297, University of Bordeaux, Bordeaux, France; CNC-UC – Center for Neuroscience and Cell Biology, CIBB – Centre for Innovative Biomedicine and Biotechnology, University of Coimbra, Coimbra, Portugal ANDREW J. FLORES • Department of Neurosciences, University of California San Diego School of Medicine, La Jolla, CA, USA BARBARA GENOCCHI • Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland

xi

xii

Contributors

ROSEMARIE GRANTYN • Neuroscience Research Center, Charite´ – University Medicine Berlin, Berlin, Germany LAURENT GROC • Interdisciplinary Institute for Neuroscience – IINS, CNRS UMR 5297, University of Bordeaux, Bordeaux, France CORAM GUEVARA • Instituto de Ciencias Biome´dicas, Facultad de Ciencias de Salud, Universidad Autonoma de Chile, Santiago, Chile; Mechanisms of Myelin Formation and Repair Laboratory, Departamento de Biologı´a, Facultad de Quı´mica y Biologı´a, Universidad de Santiago de Chile, Estacion Central, Santiago de Chile, Chile LAURENT GUILLAUD • Molecular Neuroscience Unit, Okinawa Institute of Science and Technology Graduate University, Okinawa, Japan PANKAJ GUPTA • Department of Chemistry and Chemical Engineering, Chalmers University of Technology, Gothenburg, Sweden ITARU HAMACHI • Department of Synthetic Chemistry and Biological Chemistry, Graduate School of Engineering, Kyoto University, Kyoto, Japan THOMAS S. HNASKO • Department of Neurosciences, University of California San Diego School of Medicine, La Jolla, CA, USA; Department of Veterans Affairs San Diego Healthcare System, Research Service, La Jolla, CA, USA MATTHEW G. HOLT • Instituto de Investigac¸a˜o e Inovac¸a˜o em Sau´de (i3S), Porto, Portugal JUNRU HU • Department of Neurosciences, University of California San Diego School of Medicine, La Jolla, CA, USA; National Center for Microscopy and Imaging Research, University of California San Diego, La Jolla, CA, USA HANNA KARLSSON-FERNBERG • Department of Chemistry and Chemical Engineering, Chalmers University of Technology, Gothenburg, Sweden SHIGEKI KIYONAKA • Department of Biomolecular Engineering, Graduate School of Engineering, Nagoya University, Nagoya, Japan; Institute of Nano-Life-Systems, Institutes of Innovation for Future Society, Nagoya University, Nagoya, Japan MARIA KUKLEY • Achucarro Basque Center for Neuroscience, Leioa, Spain; IKERBASQUE Basque Foundation for Science, Bilbao, Spain BARTOSZ KULA • Del Monte Institute for Neuroscience, Department of Neuroscience, University of Rochester, School of Medicine and Dentistry , Rochester, NY, USA KERSTIN LENK • Institute of Neural Engineering, Graz University of Technology, Graz, Austria; BioTechMed, Graz, Austria; Theoretical Sciences Visiting Program, Okinawa Institute of Science and Technology Graduate University, Onna, Japan LOREN L. LOOGER • Howard Hughes Medical Institute, Department of Neurosciences, University of California, San Diego, San Diego, CA, USA MASON R. MACKEY • Department of Neurosciences, University of California San Diego School of Medicine, La Jolla, CA, USA; National Center for Microscopy and Imaging Research, University of California San Diego, La Jolla, CA, USA MEHREEN MANIKKOTH • Department of Neurosciences, Center for Neural Circuits and Behavior, University of California San Diego, La Jolla, CA, USA JONATHAN S. MARVIN • Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, VA, USA MICHELLE MONJE • Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, USA; Howard Hughes Medical Institute, Stanford University, Palo Alto, CA, USA SUHITA NADKARNI • Department of Biology, Indian Institute of Science Education and Research Pune, Pune, Maharashtra, India ADRIAN NEGREAN • Allen Institute for Neural Dynamics, Seattle, WA, USA

Contributors

xiii

LIAM NESTOR • Department of Pharmaceutical Chemistry, Drug Analysis and Drug Information, Research Group Experimental Pharmacology, Center for Neurosciences (C4N), Vrije Universiteit Brussel (VUB), Brussels, Belgium ANNA NILSSON • Department of Neurosciences, University of California San Diego School of Medicine, La Jolla, CA, USA; National Center for Microscopy and Imaging Research, University of California San Diego, La Jolla, CA, USA HIROSHI NONAKA • Department of Synthetic Chemistry and Biological Chemistry, Graduate School of Engineering, Kyoto University, Kyoto, Japan KENTO OJIMA • Department of Biomolecular Engineering, Graduate School of Engineering, Nagoya University, Nagoya, Japan; Department of Synthetic Chemistry and Biological Chemistry, Graduate School of Engineering, Kyoto University, Kyoto, Japan FERNANDO C. ORTIZ • Mechanisms of Myelin Formation and Repair Laboratory, Departamento de Biologı´a, Facultad de Quı´mica y Biologı´a, Universidad de Santiago de Chile, Estacion Central, Santiago de Chile, Chile FRANCISCO PESTANA • VIB Center for Brain & Disease Research, Leuven, Belgium; Department of Human Genetics, KU Leuven, Leuven, Belgium; KU Leuven Institute for Single Cell Omics (LISCO), Leuven, Belgium KASPAR PODGORSKI • Allen Institute for Neural Dynamics, Seattle, WA, USA; Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, VA, USA AJAY PRADHAN • Department of Psychiatry and Neurochemistry, Institute of Neuroscience & Physiology, the Sahlgrenska Academy, University of Gothenburg, Mo¨lndal, Sweden IPPA SEPPA€ LA€ • Department of Psychology and Speech-Language Pathology, University of Turku, Turku, Finland KIARASH SHAMARDANI • Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, USA; Howard Hughes Medical Institute, Stanford University, Palo Alto, CA, USA ILSE SMOLDERS • Department of Pharmaceutical Chemistry, Drug Analysis and Drug Information, Research Group Experimental Pharmacology, Center for Neurosciences (C4N), Vrije Universiteit Brussel (VUB), Brussels, Belgium KYOHEI SOGA • Department of Biomolecular Engineering, Graduate School of Engineering, Nagoya University, Nagoya, Japan WENJING SUN • Department of Neuroscience, Wexner Medical Center, The Ohio State University, Columbus, OH, USA MARSA TAHERI • Department of Neurobiology, University of California Los Angeles, Los Angeles, CA, USA KATHRYN R. TAYLOR • Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, USA; Howard Hughes Medical Institute, Stanford University, Palo Alto, CA, USA YANA VAN DEN HERREWEGEN • Department of Pharmaceutical Chemistry, Drug Analysis and Drug Information, Research Group Experimental Pharmacology, Center for Neurosciences (C4N), Vrije Universiteit Brussel (VUB), Brussels, Belgium RODRIGO VARAS • Instituto de Ciencias Biome´dicas, Facultad de Ciencias de Salud, Universidad Autonoma de Chile, Santiago, Chile; Mechanisms of Myelin Formation and Repair Laboratory, Departamento de Biologı´a, Facultad de Quı´mica y Biologı´a, Universidad de Santiago de Chile, Estacion Central, Santiago de Chile, Chile THIERRY VOET • Department of Human Genetics, KU Leuven, Leuven, Belgium; KU Leuven Institute for Single Cell Omics (LISCO), Leuven, Belgium

xiv

Contributors

YUANMO WANG • Department of Chemistry and Chemical Engineering, Chalmers University of Technology, Gothenburg, Sweden AMELIA K. WARING • Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, VA, USA BELGIN YALC¸IN • Department of Neurology and Neurological Sciences, Stanford University, Stanford, CA, USA; Howard Hughes Medical Institute, Stanford University, Palo Alto, CA, USA

Part I New Approaches to Study Glutamatergic Synapses Between Neurons

Chapter 1 Visualization of Glutamatergic Neurotransmission in Diverse Model Organisms with Genetically Encoded Indicators Abhi Aggarwal, Joyce Chan, Amelia K. Waring, Adrian Negrean, Jonathan S. Marvin, Kaspar Podgorski, and Loren L. Looger Abstract Glutamate is the principal excitatory neurotransmitter, and occasionally subserves inhibitory roles, in the vertebrate nervous system. Glutamatergic synapses are dense in the vertebrate brain, at ~1/μm3. Glutamate is released from and onto diverse components of the nervous system, including neurons, glia, and other cells. Methods for glutamate detection are critically important for understanding the function of synapses and neural circuits in normal physiology, development, and disease. Here we describe the development, optimization, and deployment of genetically encoded fluorescent glutamate indicators. We review the theoretical considerations governing glutamate sensor properties from first principles of synapse biology, microscopy, and protein structure-function relationships. We provide case studies of the state-of-the-art iGluSnFR glutamate sensor, encompassing design and optimization, mechanism of action, in vivo imaging, data analysis, and future directions. We include detailed protocols for iGluSnFR imaging in common preparations (bacteria, cell culture, and brain slices) and model organisms (worm, fly, fish, rodent). Key words Glutamate, iGluSnFR, Biosensors, Imaging, Synapses

1 Introduction Glutamate is the predominant excitatory neurotransmitter in vertebrates and many invertebrates and subserves inhibitory roles in both clades. It functions in both the central and peripheral nervous systems and in diverse organ systems. Glutamatergic synapses form the bulk of contacts in the brain. Glutamate interacts with AMPA, NMDA, and kainate ionotropic receptors as well as diverse metabotropic receptors, with each receptor type serving distinct roles in neural signalling. More is known about glutamatergic signaling

Abhi Aggarwal and Joyce Chan contributed equally. Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5_1, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

3

4

Abhi Aggarwal et al.

than any other transmitter; however, a great number of aspects remain mysterious. In this chapter, we review recent advances in directly visualizing glutamatergic neurotransmission using engineered genetically encoded biosensors alongside advances in (sub)cellular-resolution imaging. We focus on key aspects of glutamatergic signaling that are best revealed through optical imaging and in particular on detailed methods for state-of-the-art glutamate imaging in a range of common model organisms and preparations. We also point the reader to technical discussions of various aspects of glutamate imaging.

2

Discovery of Glutamate as a Neurotransmitter The detection of glutamatergic neurotransmission has had a long history. Even the identification of glutamate as a neurotransmitter was itself a multi-decade process. Glutamate is abundant in all cells, as it is a common amino acid in proteins and a central player in multiple metabolic pathways including glycolysis, the TCA cycle, and pyrimidine metabolism. That the concentration of glutamate in the brain (~5–10 mM) is several times higher than any other amino acid, despite other amino acids being more abundant in proteins, is insufficient evidence that glutamate plays a role as a signaling molecule. Injection of glutamate into the brains of dogs produced convulsions [1], hinting that the molecule could signal, but these experiments did not confirm any physiological role. About this time, though, the development of electrophysiological methods such as patch-clamping and current amplification allowed direct testing of the effects of compounds such as glutamate on diverse preparations. Jeffery Watkins set about in early 1958 to identify hitherto unknown neurotransmitters using this approach [2, 3]. Importantly, GABA had already been identified as an inhibitory neurotransmitter and was used as a control. Watkins et al. injected electrical current into the dorsal roots of isolated spinal cords of cane toads and recorded from downstream ventral root neurons [4]. Bath application of compounds activated various receptors, modulating the baseline response (clear depolarization). GABA decreased the peak of the evoked signal, consistent with its inhibitory role. Intriguingly, application of low millimolar glutamate increased the response, but higher concentrations (>10 mM) decreased it again [4]. This further cemented the notion of glutamate as a transmitter molecule, but direct interpretation was confounded by the opposing activities of glutamate at different points in the spinal cord circuit. Other agonists were soon discovered. Notably, the excitatory effect of aspartate was similar to that of glutamate, and (confusingly) both the D- and L-enantiomers of both amino acids were active. The excitatory effect of these chemicals was confounded by

Visualization of Glutamatergic Neurotransmission in Diverse Model. . .

5

the fact that Renshaw interneurons, which mediate the contraction and relaxation of complementary muscles, were also activated by glutamate, giving rise to the idea that glutamate might be “nonspecific.” Throw in the fact that homocysteine and N-methyl-Daspartate (NMDA) appeared to be even more potent than glutamate or aspartate, and the specificity of glutamate as a neurotransmitter remained an unresolved question until 1977. In a crucial experiment with individual Renshaw cells, with excitation by microelectrophoretic administration of glutamate or acetylcholine, stimulation of dorsal or ventral roots, and application of the cholinergic antagonist dihydroxydibutylether (DHbE) or the glutamatergic antagonist D-α-aminoadipate, it was shown that L-glutamate was responsible for excitation of Renshaw cells from incoming dorsal root cells [5]. Soon the various receptors of glutamate were discovered and cloned, showing that fast synaptic responses arose from ionotropic glutamate channels/receptors (iGluRs) and slower signals arose from metabotropic, G-protein-coupled glutamate receptors (mGluRs). This discovery led to the use of patch-clamping to measure excitatory postsynaptic currents (EPSCs) arising from inputs to glutamatergic synapses. However, such recordings are technically challenging and very invasive, are limited to single cells at a time, and integrate the inputs into all synapses, reflecting only bulk activity. Investigation of the activities of single glutamatergic synapses would require the development of other techniques.

3

Key Aspects of Glutamatergic Signaling Glutamatergic synapses are ubiquitous in the brain. In the vertebrate brain, most glutamatergic synapses are characterized by the presence of an obvious presynaptic terminal and a small protrusion from a postsynaptic dendrite known as a spine. Spines typically vary between ~0.1 and 2 microns (some are larger) and can vary in length, diameter, and shape across brain regions, cell types, and synaptic strength. Spines consist of a bulbous head containing key postsynaptic proteins such as glutamate receptors and a narrow neck connecting the spine to the dendritic shaft (Fig. 1a). Glutamate is released from the presynaptic terminal through exocytosis of synaptic vesicles, each containing ~500 molecules. Every step of glutamate transmission is quite rapid, with the whole process typically lasting 13%) should also be removed. As SMART-Seq2 generates full-length data, reads usually align across the gene body. However, some libraries may have a higher proportion of reads aligning to the 5′ or 3′-ends of the transcripts, which is indicative of transcript coverage bias. Picard can be used to exclude such libraries by visualizing the distribution of reads across genes (see Note 23). For the remaining libraries, STAR will count the number of reads that match each unique reference gene (e.g., the number of reads that uniquely map to the Gfap gene). However, this step does not factor in the possibility of alternative gene splicing, which may result in several unique transcripts (isoforms) being produced from a single unique gene (e.g., Gfap-201, Gfap-202, and Gfap-203, which are all transcribed from the Gfap gene). This is important as unique transcripts may have specific functions [61, 62]. For users interested in splice analysis, tools as RSEM can estimate transcript abundance by aligning reads to a reference transcriptome, instead of the reference genome alignment performed by STAR [63] (see Note 24). 3.4 Cell-Type Identification Methods

A source of unwanted variation in the sequenced libraries can result, for example, from poor library preparation, differences in sequencing depth, and RNA composition, which complicate the comparison of transcriptomes across cells. Therefore, normalization and scaling are applied to correct for this variation. Most single-cell transcriptome methods, including SMART-Seq2, only detect approximately 30% of transcripts in a given cell. As a consequence, many genes will show zero or near-zero expression [64]. This large number of “dropout” genes results in high-dimensional noisy data. This means that considerable time and effort has gone into generating computational tools to reduce this complexity, allowing comparison of cells based on the similarity of their gene expression profiles, a process known as clustering. For example, Seurat, a widely used analysis tool for single-cell data, uses a standard Louvain clustering algorithm [65]. Briefly in this section, we will provide details on how single cells are pre-processed: cells that do not pass specific quality thresholds are removed from the expression matrices, the expression profiles of the remaining cells are normalized, and cells are then clustered to visually identify astrocytes and distinct astrocyte subtypes (Fig. 5 and below). For this section, knowledge of R is required, although similar tools, such as Scanpy, have been developed for Python users [66]. An example of the following workflow has been implemented in R and deposited on GitHub (https://github.com/fpestana-git/Springer_astrocyte_ synapse/tree/main/Section_3.4).

Astrocyte-Neuron Interactions Revealed by Single Cell Sequencing

Total counts per gene Total reads per cell Doublet/multiplet rate % mitochondrial counts

PCA

Scaling

Dimensionality reduction

0.5

Expressed genes

Normalization Log(x+1)

STD

Quality control

PC

Density

Input reads (106)

20

Clustering Run UMAP

Cluster 1 Cluster 2 Cluster 3

Expression level

3200 1000

191

C1

C2

C3

% mitochondrial counts

Fig. 5 An example workflow for the processing of expression matrices, using Seurat. Individual single-cell libraries are filtered based on multiple indicators, such as the number of expressed genes detected in each cell, or the percentage of mitochondria mRNA counts. The resulting dataset is then normalized and scaled, before dimensionality reduction using principal component analysis (PCA). The number of PCA components considered in further analysis steps is chosen based on the inflex point of a scree (elbow) plot. STD, standard deviation. Cells are then clustered into distinct cell types and subtypes, based on the differential expression of marker genes 3.4.1

Quality Control

Successful detection of a large number of genes depends mainly on the quality of library preparation and the chosen sequencing depth. Unfortunately, dissociation generally leads to cellular stress, which can lead to apoptosis. This leads to a loss of structural integrity in cell membranes, resulting in a loss of bona fide cytoplasmatic mRNA transcripts and increased detection of transcripts otherwise retained in mitochondria. Low-sequencing depth may also result in poor detection of lowly expressed genes. Specific cutoffs can be defined from observing the distribution of sequenced reads and detected genes across all cells. Only cells with a total mitochondrial content 10 kb. 14. Read alignment using STAR often requires access to a dedicated home-built workstation or to a supercomputer network. If access to these resources is not possible, less computationally

200

Francisco Pestana et al.

intensive methods, such as HISAT2, may be an option [91]. On the other hand, HISAT2 does not handle mismatches as well as STAR and generates a higher fraction of unmapped reads [92]. 15. SMART-Seq2 was not developed to use unique molecular identifiers (UMIs) for molecular counting. Instead, we use ERCC spike-ins. ERCCs are a set of synthetic RNA molecules developed by the External RNA Controls Consortium, which are used as spike-in controls to evaluate technical variability and identify potential amplification biases in the data [93]. Generally speaking, only 1–5% of total sequencing reads should come from the ERCCs to allow efficient normalization. Typically, we use ERCCs in the reverse transcription mix at a final dilution of 1:160 × 106. 16. Transcript coverage bias is the technical term for an uneven distribution of sequencing reads along the transcript length. 17. The installation of software packages can cause conflicts with other programs installed on the user workstation. To prevent disruption of analysis pipelines, we recommend installing packages in enclosed environments or containers (e.g., using Conda, Docker, Singularity). 18. The Snakemake workflow we suggest implementing allows the user to run workflows which take full advantage of multicore workstations to process multiple files with high reproducibility. Alternative options, including Nextflow and Galaxy, aim to provide a similar experience. The choice of workflow management system depends largely on user preference. 19. It is recommended to make sure that the most up-to-date reference genome is being used. For mouse, this is currently the Ensembl genome release 109 (same annotation as the GENCODE release M32). The latest reference genomes can be found at https://www.gencodegenes.org. 20. As the sequenced libraries will also include spike-in ERCC reads, the respective spike-in sequences (in Fasta format) should be added at the end of the reference genome annotation file (GTF format), prior to alignment. When building the reference genome, the “sjdbOverhang” parameter should be calculated as the length of the sequencing read minus one base pair (e.g., for 75 bp reads, the value should be 74). 21. i5/i7 adapter sequences and bases with high probability of being incorrectly identified should be trimmed to improve alignment efficiency. If splice analysis is not the primary goal, it is recommended to remove alignments that include noncanonical unannotated junctions and introns, as these can represent technical artifacts [94]. If splice analysis and transcript

Astrocyte-Neuron Interactions Revealed by Single Cell Sequencing

201

analysis is the primary goal, the “quantMode” should be set to “TranscriptomeSAM”, which is necessary for transcriptome reconstruction using RSEM. 22. FastQC generates QC metrics on individual libraries, including base quality, sequence length distribution, and GC content. Tools, such as MultiQC, can aggregate the quality metrics produced by FastQC, allowing comparison of multiple libraries in a single report [95]. 23. In addition to generating coverage bias plots, Picard can also manipulate the alignment files to explore alignment quality, e.g., by looking at the quality of read alignments to specific genome regions, such as exons, introns, or repetitive elements. This function also allows Picard to be used to assess transcript coverage bias. The position and distribution of reads along the lengths of the various transcripts are determined for each library. Then, a curve is fitted across the read distributions of all libraries, to determine the average read distribution and standard deviation. Libraries are removed from further analysis if their read distribution deviates more than three times from the standard deviation of the fitted curve. 24. Although RSEM gives information at the transcript level, it requires realignment of reads to estimate transcript expression. Like STAR, RSEM is computationally intensive. 25. Confounding factors, resulting from biological variability, can be regressed out directly by setting “vars.regress” in the “ScaleData” function of Seurat. Technical batch effects, however, should be regressed out using tools such as Harmony [74]. 26. Choosing the number of PCA components, based on visual inspection of a scree plot, can be somewhat subjective. The number of PCA components can also be determined more quantitatively using a computational approach (see example code at https://github.com/fpestana-git/Springer_astrocyte_ synapse/Section_3.4/tree/main/calculatePC.R). 27. Cluster robustness can be assessed using a resampling approach. A cluster is considered robust if cluster identity is maintained when clustering is rerun using a subset of cells randomly selected from each cluster. 28. Identification of subtypes is largely dependent on the identification of a set of differentially expressed genes at statistically significant levels. 29. Differential gene expression analysis may return a short list if cell populations are very similar. In this case, the threshold values (e.g., log2 fold-change and adjusted p-value) may be loosened. However, we do not recommend using methods

202

Francisco Pestana et al.

that return (by default) a higher number of DEGs, since the number of false positives may be significantly higher compared to the Wilcoxon method. 30. DAVID is one of many tools available that allows for analysis of multiple pathways and biological processes using a single software package. In the case of GO term analysis, DAVID does not currently use the latest GO term database and uses the Panther database from 2019. For GO term analysis we recommend using the Gene Ontology and Panther databases (http://geneontology.org and http://pantherdb.org), which are more frequently updated.

Acknowledgments FP holds a Fundac¸˜ao para a Cieˆncia e a Tecnologia (FCT) scholarship (2020.08750.BD). MGH is the ERA Chair (NCBio) at i3S Porto, funded by the European Commission (H2020-WIDESPREAD-2018-2020-6; NCBio; 951923). The single-cell sequencing protocol and pathway analysis described in this chapter were implemented when he was based at KU Leuven and were supported by grants from the European Research Council (ERC) (Starting Grant 281961; Astrofunc), Fonds Wetenschappelijk Onderzoek (FWO) (Grants G066715N, 1523014N and I001818N), and the Stichting Alzheimer Onderzoek (SAO) (Grant S#16025). The authors thank Dr. C ¸ ansu Akkaya (Lab of Glia Biology, KU Leuven) for providing Fig. 1. References 1. Su¨dhof TC (2018) Towards an understanding of synapse formation. Neuron 100:276–293. https://doi.org/10.1016/j.neuron.2018. 09.040 2. Allen NJ, Eroglu C (2017) Cell biology of astrocyte-synapse interactions. Neuron 96: 697–708. https://doi.org/10.1016/j.neu ron.2017.09.056 3. Hillen AEJ, Burbach JPH, Hol EM (2018) Cell adhesion and matricellular support by astrocytes of the tripartite synapse. Prog Neurobiol 165–167:66–86. https://doi.org/10. 1016/j.pneurobio.2018.02.002 4. Kucukdereli H, Allen NJ, Lee AT et al (2011) Control of excitatory CNS synaptogenesis by astrocyte-secreted proteins Hevin and SPARC. Proc Natl Acad Sci 108:E440–E449. https:// doi.org/10.1073/pnas.1104977108 5. Baldwin KT, Eroglu C (2017) Molecular mechanisms of astrocyte-induced synaptogenesis. Curr Opin Neurobiol 45:113–120.

https://doi.org/10.1016/j.conb.2017. 05.006 6. Barres BA (2008) The mystery and magic of glia: a perspective on their roles in health and disease. Neuron 60:430–440. https:// doi.org/10.1016/j.neuron.2008.10.013 7. Perez-Alvarez A, Navarrete M, Covelo A et al (2014) Structural and functional plasticity of astrocyte processes and dendritic spine interactions. J Neurosci 34:12738–12744. https:// doi.org/10.1523/JNEUROSCI.2401-14. 2014 8. Araque A, Carmignoto G, Haydon PG et al (2014) Gliotransmitters travel in time and space. Neuron 81:728–739. https://doi.org/ 10.1016/j.neuron.2014.02.007 9. Fiacco TA, McCarthy KD (2018) Multiple lines of evidence indicate that gliotransmission does not occur under physiological conditions. J Neurosci 38:3–13. https://doi.org/10. 1523/JNEUROSCI.0016-17.2017

Astrocyte-Neuron Interactions Revealed by Single Cell Sequencing ˜ oz-Manchado AB, Codeluppi S 10. Zeisel A, Mun et al (2015) Cell types in the mouse cortex and hippocampus revealed by single-cell RNA-seq. Science 347:1138–1142. https://doi.org/10. 1126/science.aaa1934 11. Batiuk MY, Martirosyan A, Wahis J et al (2020) Identification of region-specific astrocyte subtypes at single cell resolution. Nat Commun 11:1220. https://doi.org/10.1038/s41467019-14198-8 12. Bayraktar OA, Bartels T, Holmqvist S et al (2020) Astrocyte layers in the mammalian cerebral cortex revealed by a single-cell in situ transcriptomic map. Nat Neurosci 23:500–509. https://doi.org/10.1038/s41593-0200602-1 13. Booher J, Sensenbrenner M (1972) Growth and cultivation of dissociated neurons and glial cells from embryonic chick, rat and human brain in flask cultures. Neurobiology 2:97–105 14. McCarthy KD, de Vellis J (1980) Preparation of separate astroglial and oligodendroglial cell cultures from rat cerebral tissue. J Cell Biol 85: 890–902. https://doi.org/10.1083/jcb.85. 3.890 15. Foo LC, Allen NJ, Bushong EA et al (2011) Development of a method for the purification and culture of rodent astrocytes. Neuron 71: 799–811. https://doi.org/10.1016/j.neuron. 2011.07.022 16. Zamanian JL, Xu L, Foo LC et al (2012) Genomic analysis of reactive astrogliosis. J Neurosci 32:6391–6410. https://doi.org/10.1523/ JNEUROSCI.6221-11.2012 17. Batiuk MY, de Vin F, Duque´ SI et al (2017) An immunoaffinity-based method for isolating ultrapure adult astrocytes based on ATP1B2 targeting by the ACSA-2 antibody. J Biol Chem 292:8874–8891. https://doi.org/10. 1074/jbc.M116.765313 18. Lovatt D, Sonnewald U, Waagepetersen HS et al (2007) The transcriptome and metabolic gene signature of protoplasmic astrocytes in the adult murine cortex. J Neurosci 27: 12255–12266. https://doi.org/10.1523/ JNEUROSCI.3404-07.2007 19. Cahoy JD, Emery B, Kaushal A et al (2008) A transcriptome database for astrocytes, neurons, and oligodendrocytes: a new resource for understanding brain development and function. J Neurosci 28:264–278. https://doi. org/10.1523/JNEUROSCI.4178-07.2008 20. Regan MR, Huang YH, Kim YS et al (2007) Variations in promoter activity reveal a differential expression and physiology of glutamate transporters by glia in the developing and

203

mature CNS. J Neurosci 27:6607–6619. https://doi.org/10.1523/JNEUROSCI. 0790-07.2007 21. Yang Y, Vidensky S, Jin L et al (2011) Molecular comparison of GLT1+ and ALDH1L1+ astrocytes in vivo in astroglial reporter mice. Glia 59:200–207. https://doi.org/10.1002/ glia.21089 22. Heiman M, Schaefer A, Gong S et al (2008) A translational profiling approach for the molecular characterization of CNS cell types. Cell 135:738–748. https://doi.org/10.1016/j. cell.2008.10.028 23. Doyle JP, Dougherty JD, Heiman M et al (2008) Application of a translational profiling approach for the comparative analysis of CNS cell types. Cell 135:749–762. https://doi.org/ 10.1016/j.cell.2008.10.029 24. Lanjakornsiripan D, Pior B-J, Kawaguchi D et al (2018) Layer-specific morphological and molecular differences in neocortical astrocytes and their dependence on neuronal layers. Nat Commun 9:1623. https://doi.org/10.1038/ s41467-018-03940-3 25. Sakers K, Lake AM, Khazanchi R et al (2017) Astrocytes locally translate transcripts in their peripheral processes. Proc Natl Acad Sci U S A 114:E3830–E3838. https://doi.org/10. 1073/pnas.1617782114 26. Mazare´ N, Oudart M, Moulard J et al (2020) Local translation in perisynaptic astrocytic processes is specific and changes after fear conditioning. Cell Rep 32:108076. https://doi. org/10.1016/j.celrep.2020.108076 27. Boulay A-C, Mazare´ N, Saubame´a B, CohenSalmon M (2019) Preparing the astrocyte perivascular endfeet transcriptome to investigate astrocyte molecular regulations at the brainvascular interface. Method Mol Biol (Clifton, NJ) 1938:105–116. https://doi.org/10. 1007/978-1-4939-9068-9_8 28. Boulay A-C, Saubame´a B, Adam N et al (2017) Translation in astrocyte distal processes sets molecular heterogeneity at the gliovascular interface. Cell Discov 3:17005. https://doi. org/10.1038/celldisc.2017.5 29. Bachoo RM, Kim RS, Ligon KL et al (2004) Molecular diversity of astrocytes with implications for neurological disorders. Proc Natl Acad Sci U S A 101:8384–8389. https://doi. org/10.1073/pnas.0402140101 30. Zhang Y, Chen K, Sloan SA et al (2014) An RNA-sequencing transcriptome and splicing database of glia, neurons, and vascular cells of the cerebral cortex. J Neurosci 34:11929– 1 1 9 4 7 . h t t p s : // d o i . o r g / 1 0 . 1 5 2 3 / JNEUROSCI.1860-14.2014

204

Francisco Pestana et al.

31. Zhang Y, Sloan SA, Clarke LE et al (2016) Purification and characterization of progenitor and mature human astrocytes reveals transcriptional and functional differences with mouse. Neuron 89:37–53. https://doi.org/10.1016/ j.neuron.2015.11.013 32. Stogsdill JA, Ramirez J, Liu D et al (2017) Astrocytic neuroligins control astrocyte morphogenesis and synaptogenesis. Nature 551: 1 9 2 – 1 9 7 . h t t p s : // d o i . o r g / 1 0 . 1 0 3 8 / nature24638 33. Paukert M, Agarwal A, Cha J et al (2014) Norepinephrine controls astroglial responsiveness to local circuit activity. Neuron 82:1263– 1270. https://doi.org/10.1016/j.neuron. 2014.04.038 34. Murphy-Royal C, Johnston AD, Boyce AKJ et al (2020) Stress gates an astrocytic energy reservoir to impair synaptic plasticity. Nat Commun 11:2014. https://doi.org/10. 1038/s41467-020-15778-9 35. Wallingford J, Scott AL, Rodrigues K, Doering LC (2017) Altered developmental expression of the astrocyte-secreted factors Hevin and SPARC in the fragile X mouse model. Front Mol Neurosci 10:268. https://doi.org/10. 3389/fnmol.2017.00268 36. Barbar L, Jain T, Zimmer M et al (2020) CD49f is a novel marker of functional and reactive human iPSC-derived astrocytes. Neuron 107:436–453.e12. https://doi.org/10. 1016/j.neuron.2020.05.014 37. Hiller BM, Marmion DJ, Thompson CA et al (2022) Optimizing maturity and dose of iPSCderived dopamine progenitor cell therapy for Parkinson’s disease. NPJ Regen Med 7:24. https://doi.org/10.1038/s41536-02200221-y 38. Chai H, Diaz-Castro B, Shigetomi E et al (2017) Neural circuit-specialized astrocytes: transcriptomic, proteomic, morphological, and functional evidence. Neuron 95:531–549. e9. https://doi.org/10.1016/j.neuron.2017. 06.029 39. Morel L, Chiang MSR, Higashimori H et al (2017) Molecular and functional properties of regional astrocytes in the adult brain. J Neurosci 37:8706–8717. https://doi.org/10. 1523/JNEUROSCI.3956-16.2017 40. Boisvert MM, Erikson GA, Shokhirev MN, Allen NJ (2018) The aging astrocyte transcriptome from multiple regions of the mouse brain. Cell Rep 22:269–285. https://doi.org/10. 1016/j.celrep.2017.12.039 41. Ben Haim L, Rowitch DH (2017) Functional diversity of astrocytes in neural circuit

regulation. Nat Rev Neurosci 18:31–41. https://doi.org/10.1038/nrn.2016.159 42. Pestana F, Edwards-Faret G, Belgard TG et al (2020) No longer underappreciated: the emerging concept of astrocyte heterogeneity in neuroscience. Brain Sci 10:E168. https:// doi.org/10.3390/brainsci10030168 43. Nagai J, Yu X, Papouin T et al (2021) Behaviorally consequential astrocytic regulation of neural circuits. Neuron 109:576–596. https://doi.org/10.1016/j.neuron.2020. 12.008 44. Zeisel A, Hochgerner H, Lo¨nnerberg P et al (2018) Molecular architecture of the mouse nervous system. Cell 174:999–1014.e22. https://doi.org/10.1016/j.cell.2018.06.021 45. Saunders A, Macosko EZ, Wysoker A et al (2018) Molecular diversity and specializations among the cells of the adult mouse brain. Cell 174:1015–1030.e16. https://doi.org/10. 1016/j.cell.2018.07.028 46. Picelli S, Faridani OR, Bjo¨rklund ÅK et al (2014) Full-length RNA-seq from single cells using SMART-Seq2. Nat Protoc 9:171–181. https://doi.org/10.1038/nprot.2014.006 47. Lee H-G, Wheeler MA, Quintana FJ (2022) Function and therapeutic value of astrocytes in neurological diseases. Nat Rev Drug Discov 21:339–358. https://doi.org/10.1038/ s41573-022-00390-x 48. Ahmed S, Holt M, Riedel D, Jahn R (2013) Small-scale isolation of synaptic vesicles from mammalian brain. Nat Protoc 8:998–1009. https://doi.org/10.1038/nprot.2013.053 49. Schmid KT, Ho¨llbacher B, Cruceanu C et al (2021) scPower accelerates and optimizes the design of multi-sample single cell transcriptomic studies. Nat Commun 12:6625. https://doi.org/10.1038/s41467-02126779-7 50. Brewer GJ, Torricelli JR (2007) Isolation and culture of adult neurons and neurospheres. Nat Protoc 2:1490–1498. https://doi.org/10. 1038/nprot.2007.207 51. Bordt EA, Block CL, Petrozziello T et al (2020) Isolation of microglia from mouse or human tissue. STAR Protoc 1:100035. https://doi.org/10.1016/j.xpro.2020. 100035 52. Takele Assefa A, Vandesompele J, Thas O (2020) On the utility of RNA sample pooling to optimize cost and statistical power in RNA sequencing experiments. BMC Genomics 21: 312. https://doi.org/10.1186/s12864-0206721-y

Astrocyte-Neuron Interactions Revealed by Single Cell Sequencing 53. Pal S, Gupta R, Kim H et al (2011) Alternative transcription exceeds alternative splicing in generating the transcriptome diversity of cerebellar development. Genome Res 21:1260– 1 2 7 2 . h t t p s : // d o i . o r g / 1 0 . 1 1 0 1 / g r. 120535.111 54. Tian B, Manley JL (2017) Alternative polyadenylation of mRNA precursors. Nat Rev Mol Cell Biol 18:18–30. https://doi.org/10. 1038/nrm.2016.116 55. Picelli S (2019) Full-length single-cell RNA sequencing with SMART-Seq2. In: Proserpio V (ed) Single cell methods. Humana, New York, pp 25–44 56. Trapnell C, Williams BA, Pertea G et al (2010) Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nat Biotechnol 28:511–515. https://doi. org/10.1038/nbt.1621 57. Rich-Griffin C, Stechemesser A, Finch J et al (2020) Single-cell transcriptomics: a highresolution avenue for plant functional genomics. Trends Plant Sci 25:186–197. https:// doi.org/10.1016/j.tplants.2019.10.008 58. Bru¨ning RS, Tombor L, Schulz MH et al (2022) Comparative analysis of common alignment tools for single-cell RNA sequencing. GigaScience 11:giac001. https://doi.org/10. 1093/gigascience/giac001 59. Dobin A, Davis CA, Schlesinger F et al (2013) STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29:15–21. https://doi. org/10.1093/bioinformatics/bts635 60. Benjamini Y, Speed TP (2012) Summarizing and correcting the GC content bias in highthroughput sequencing. Nucleic Acids Res 40:e72. https://doi.org/10.1093/nar/ gks001 61. Booeshaghi AS, Yao Z, van Velthoven C et al (2021) Isoform cell-type specificity in the mouse primary motor cortex. Nature 598: 195–199. https://doi.org/10.1038/s41586021-03969-3 62. Gupta I, Collier PG, Haase B et al (2018) Single-cell isoform RNA sequencing characterizes isoforms in thousands of cerebellar cells. Nat Biotechnol 36:1197–1202. https:// doi.org/10.1038/nbt.4259 63. Li B, Dewey CN (2011) RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinf 12:323. https://doi.org/10.1186/14712105-12-323 64. Wu AR, Neff NF, Kalisky T et al (2014) Quantitative assessment of single-cell RNA-sequencing methods. Nat Methods 11:41–46. https://doi.org/10.1038/nmeth.2694

205

65. Satija R, Farrell JA, Gennert D et al (2015) Spatial reconstruction of single-cell gene expression data. Nat Biotechnol 33:495–502. https://doi.org/10.1038/nbt.3192 66. Wolf FA, Angerer P, Theis FJ (2018) SCANPY: large-scale single-cell gene expression data analysis. Genome Biol 19:15. https://doi. org/10.1186/s13059-017-1382-0 67. McGinnis CS, Murrow LM, Gartner ZJ (2019) DoubletFinder: doublet detection in single-cell RNA sequencing data using artificial nearest neighbors. Cell Syst 8:329–337.e4. https:// doi.org/10.1016/j.cels.2019.03.003 68. Wolock SL, Lopez R, Klein AM (2019) Scrublet: computational identification of cell doublets in single-cell transcriptomic data. Cell Syst 8:281–291.e9. https://doi.org/10.1016/j. cels.2018.11.005 69. Blondel VD, Guillaume J-L, Lambiotte R, Lefebvre E (2008) Fast unfolding of communities in large networks. J Stat Mech Theory Exp 2008:P10008. https://doi.org/10.1088/ 1742-5468/2008/10/P10008 70. McInnes L, Healy J, Saul N, Großberger L (2018) UMAP: uniform manifold approximation and projection. J Open Source Softw 3: 861. https://doi.org/10.21105/joss.00861 71. Soneson C, Robinson MD (2018) Bias, robustness and scalability in single-cell differential expression analysis. Nat Methods 15: 255–261. https://doi.org/10.1038/nmeth. 4612 72. Hosack DA, Dennis G, Sherman BT et al (2003) Identifying biological themes within lists of genes with EASE. Genome Biol 4: R70. https://doi.org/10.1186/gb-2003-410-r70 73. Stuart T, Butler A, Hoffman P et al (2019) Comprehensive integration of single-cell data. Cell 177:1888–1902.e21. https://doi.org/ 10.1016/j.cell.2019.05.031 74. Korsunsky I, Millard N, Fan J et al (2019) Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods 16:1289– 1296. https://doi.org/10.1038/s41592019-0619-0 75. Welch JD, Kozareva V, Ferreira A et al (2019) Single-cell multi-omic integration compares and contrasts features of brain cell identity. Cell 177:1873–1887.e17. https://doi.org/ 10.1016/j.cell.2019.05.006 76. Tran HTN, Ang KS, Chevrier M et al (2020) A benchmark of batch-effect correction methods for single-cell RNA sequencing data. Genome Biol 21:12. https://doi.org/10.1186/ s13059-019-1850-9 77. Yao Z, van Velthoven CTJ, Nguyen TN et al (2021) A taxonomy of transcriptomic cell types

206

Francisco Pestana et al.

across the isocortex and hippocampal formation. Cell 184:3222–3241.e26. https://doi. org/10.1016/j.cell.2021.04.021 78. Raj B, Blencowe BJ (2015) Alternative splicing in the mammalian nervous system: recent insights into mechanisms and functional roles. Neuron 87:14–27. https://doi.org/10.1016/ j.neuron.2015.05.004 79. Li D, McIntosh CS, Mastaglia FL et al (2021) Neurodegenerative diseases: a hotbed for splicing defects and the potential therapies. Transl Neurodegener 10:16. https://doi.org/10. 1186/s40035-021-00240-7 80. Hagemann-Jensen M, Ziegenhain C, Chen P et al (2020) Single-cell RNA counting at allele and isoform resolution using SMART-Seq3. Nat Biotechnol 38:708–714. https://doi. org/10.1038/s41587-020-0497-0 81. Hagemann-Jensen M, Ziegenhain C, Sandberg R (2022) Scalable single-cell RNA sequencing from full transcripts with SMART-Seq3xpress. Nat Biotechnol 38:1452–1457. https://doi. org/10.1038/s41587-022-01311-4 82. Alon S, Goodwin DR, Sinha A et al (2021) Expansion sequencing: spatially precise in situ transcriptomics in intact biological systems. Science 371:eaax2656. https://doi.org/10. 1126/science.aax2656 83. Krishnaswami SR, Grindberg RV, Novotny M et al (2016) Using single nuclei for RNA-seq to capture the transcriptome of postmortem neurons. Nat Protoc 11:499–524. https://doi. org/10.1038/nprot.2016.015 84. Eraslan G, Drokhlyansky E, Anand S et al (2022) Single-nucleus cross-tissue molecular reference maps toward understanding disease gene function. Science 376:eabl4290. https://doi.org/10.1126/science.abl4290 85. Takeichi M, Okada T (1972) Roles of magnesium and calcium ions in cell-to-substrate adhesion. Exp Cell Res 74:51–60. https:// doi.org/10.1016/0014-4827(72)90480-6 86. Machado L, Geara P, Camps J et al (2021) Tissue damage induces a conserved stress response that initiates quiescent muscle stem cell activation. Cell Stem Cell 28:1125–1135. e7. https://doi.org/10.1016/j.stem.2021. 01.017

87. Liu L, Besson-Girard S, Ji H et al (2021) Dissociation of microdissected mouse brain tissue for artifact free single-cell RNA sequencing. STAR Protoc 2:100590. https://doi.org/10. 1016/j.xpro.2021.100590 88. Reichard A, Asosingh K (2019) Best practices for preparing a single cell suspension from solid tissues for flow cytometry. Cytometry A 95: 219–226. https://doi.org/10.1002/cyto.a. 23690 89. Jungblut M, Tiveron MC, Barral S et al (2012) Isolation and characterization of living primary astroglial cells using the new GLAST-specific monoclonal antibody ACSA-1. Glia 60:894– 907. https://doi.org/10.1002/glia.22322 90. Ohlig S, Clavreul S, Thorwirth M et al (2021) Molecular diversity of diencephalic astrocytes reveals adult astrogenesis regulated by Smad4. EMBO J 40:e107532. https://doi.org/10. 15252/embj.2020107532 91. Baruzzo G, Hayer KE, Kim EJ et al (2017) Simulation-based comprehensive benchmarking of RNA-seq aligners. Nat Methods 14: 135–139. https://doi.org/10.1038/nmeth. 4106 92. Sahraeian SME, Mohiyuddin M, Sebra R et al (2017) Gaining comprehensive biological insight into the transcriptome by performing a broad-spectrum RNA-seq analysis. Nat Commun 8:59. https://doi.org/10.1038/s41467017-00050-4 93. The External RNA Controls Consortium (2005) The external RNA controls consortium: a progress report. Nat Methods 2:731– 7 3 4 . h t t p s : // d o i . o r g / 1 0 . 1 0 3 8 / nmeth1005-731 94. Houseley J, Tollervey D (2010) Apparent non-canonical trans-splicing is generated by reverse transcriptase in vitro. PLoS One 5: e12271. https://doi.org/10.1371/journal. pone.0012271 95. Ewels P, Magnusson M, Lundin S, K€aller M (2016) MultiQC: summarize analysis results for multiple tools and samples in a single report. Bioinformatics 32:3047–3048. https://doi.org/10.1093/bioinformatics/ btw354

Chapter 10 Analysis of Synaptic Glutamate Clearance as a Possible Indicator of Synaptic Health in the Degenerating Rodent Brain Anton Dvorzhak and Rosemarie Grantyn Abstract Neurodegeneration may present multiple challenges when one tries to quantify the astrocytic contribution to synaptic reorganization in a diseased animal. It is therefore desirable to apply tests that would detect alterations occurring in vivo but at sufficiently elementary level of defined cellular interactions. With reliable tools to identify a given synapse with respect to the pathway of origin, one could examine the specific impact of experimentally induced modifications in adjacent astrocytes. Ideally, one might be able study the properties of individual synapses in contact with a predefined type of individual astrocyte. In our lab, respective techniques for single synapse imaging and analysis of synapse-astrocyte relationships were established in the course of a preclinical research project on Huntington’s disease (HD). We aimed at exploring new possibilities of functional rescue by expression of a modified version of the glutamate transporter EAAT2, the type 2 excitatory amino acid transporter. Our focus was on improvement of glutamate clearance, since it has repeatedly been suggested that there might be an excitotoxic component in the pathogenesis of HD. From the very outset, we therefore sought to establish and to verify indicators suitable for independent evaluation of glutamate uptake as opposed to glutamate release. HD is a fatal neurodegenerative disease of monogenic origin. Expression of mutant huntingtin (mHTT) damages the afflicted cells (both neurons and astrocytes) in multiple ways. The most vulnerable brain area would be the dorsal striatum (caudate nucleus in humans), and the most afflicted input is the corticostriatal pathway, The latter is known to control the initiation of movements. Its damage has been associated with symptoms of hypokinesia, i.e., less frequent and slowed spontaneous movement activity, loss of neuropil, increased neuronal excitability, and reduced capacity of glutamate uptake. In the murine striatum, most of the synaptic glutamate uptake is carried out by the excitatory amino acid transporter type 2, EAAT2, but overexpression of native EAAT2 in HD mice provided little evidence for functional rescue. We therefore explored the effects of genetically modified forms of EAAT2 and were successful with a C-terminal-truncated version of EAAT2 (EAAT2-S506X) expressed in striatal astrocytes under the control of a GFAP-promoter. This intervention not only restored the glutamate uptake of transduced astrocytes but also ameliorated the symptoms of hypokinesia in treated HD mice. We therefore wanted to know to what extent the function of individual corticostriatal synapses is shaped by their astroglial environment. The following description of respective experiments is divided in two parts. Part 1 addresses some basic requirements to be considered for the successful implementation of single-synapse glutamate imaging as such. Part 2 will show how single-synapse imaging can be combined with single-astrocyte imaging to

Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5_10, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

207

208

Anton Dvorzhak and Rosemarie Grantyn

characterize glutamatergic synaptic transmission in its dependence on astrocyte activity in HD and the disease-changing effects of astrocytic EAAT2 modifications. It will be demonstrated that the expression of a modified EAAT2 transgene can change the glutamate clearance characteristics of the contacted synapses. Although the outlined applications of single-synapse imaging to preclinical research in HD will require a certain level of experimental skills and good organization of animal supply, they have the advantage that they can be performed with a relatively low-cost equipment. This should make our approach attractive for forthcoming projects on other types of neurodegeneration. Key words Excitatory amino acid transporter type 2 (EAAT2), Glutamate clearance, Tripartite synapse synapse-astrocyte interaction, Single-synapse glutamate imaging, Corticostriatal pathway, Hypokinesia, Huntington’s disease, Adeno-associated virus-(AAV)-mediated gene transfer

1

Introduction The mammalian brain contains four types of high-affinity electrogenic glutamate transporters for the removal of neurotransmitter from extracellular space [1–3]. EAAT2 accounts for the major part of glutamate homeostasis in the striatum, neocortex and hippocampus [4]. It is preferentially expressed in astrocytes forming clusters on the perisynaptic astrocyte processes (PAPs) next to the presynaptic active zone as the site of calcium-dependent glutamate release [5, 6]. Compared to other synaptically enriched proteins, including the AMPA receptors, EAAT2 is very abundant [5, 7, 8]. It has therefore been expected and is now generally assumed that tripartite synapses can cope with a wide range of glutamate concentrations [9, 10]. But the situation may become different when brains undergo neurodegeneration [11–13, 13] and the levels of SLC1A2 (the gene encoding EAAT2) decrease (see for instance [14]). It has already been demonstrated that the presence of mHTT in mixed astrocyte-neuronal cultures may impede the overall glutamate uptake [15–18]. Synaptosomal preparations from HD mice consistently showed a decrease in the stimulated synaptosomal uptake of exogenous [3H]-labeled L-glutamate or D-aspartate [15, 17, 19– 21]. When glutamate uncaging on astrocytes simulates synaptic glutamate release in striatal slices, individual astrocytes decreased their EAAT2-mediated co-transport of sodium (a commonly used measure for glutamate transport) by 20–40% from WT level [22]. This value could represent a rather direct estimate of the HD-related transport deficit, but some doubts remained with regard to the presumed pathogenic role of impaired glutamate clearance in HD, since glutamate imaging experiments with iGluSnFR provided no evidence for EAAT2-related uptake deficits [21, 23]. The latter approach seemed to support a previous notion that this transporter contributes little to the progression of mHTTrelated neurodegeneration [24, 25].

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . .

209

A plausible explanation for these contradictory findings could be that the observed reduction of striatal SLC1A2 expression and glutamate uptake in HD were just part of an adaptive response where glutamate clearance decreases in proportion with glutamate uptake due to the loss of glutamatergic synapses [26–28]. On the other hand, one could not exclude the possibility that EAAT2related changes passed undetected because they were confined to synaptic sites but the applied methods lacked the necessary resolution. The molecular mechanisms controlling the availability of EAAT2 in the tripartite synapse are still not fully understood. Heterologous expression studies with recombinant chimeric EAAT2 variants have shed some light on the importance of intracellular interaction partners and their possible role in the balance between cytoplasmic retention and membrane insertion of the transporter protein complex [29, 30]. It is also known that proteins facilitating internalization of the transporter from the plasma membrane interact with EAAT2 at four lysine sites in the C-terminus of EAAT2 [31]. The C-terminal transmembrane EAAT2 domain is also relevant for the interaction with the transport substrate [32], and it was reported that sumoylated toxic C-terminal fragments may inhibit the Slc1a2 transcription in the astrocyte nucleus [33– 35]. A recent study from our lab investigated the protein binding partners of a wild-type and truncated EAAT2 transgene after its expression in striatal astrocytes [36]. These experiments rendered significant mHTT-related differences in the EAAT2 binding patterns. It has also been shown that the presence of a truncated EAAT2 transgene can prevent pathological binding, which underlines the involvement of astrocytic as opposed to neuronal mechanisms of glutamate uptake. Together, these experiments supported the hypothesis that the presence of mHTT does affect the availability and function of EAAT2 at synaptic sites. It therefore seemed promising to invest more effort into methods increasing the resolution of functional tests at tripartite synapses in situ. Striatal slice preparations have already been used to successfully study single glutamatergic synapses and their specific relationships with diverse astrocytes [37, 38]. The striatum is well suited for this type of research because in itself it lacks glutamatergic neurons but receives strong glutamatergic input from the cerebral cortex and medial thalamus (see [28, 39–41] for more). The cortical control of striatal activity deteriorates in HD [39, 42, 43]. On the presynaptic side of the corticostriatal connection, there is a tendency for a downregulation of vGlut1 immunofluorescence [26], a decrease in vGluT1+ terminal numbers [27, 28], and an impairment of synaptic glutamate release at individual corticostriatal terminals [44]. On the postsynaptic side, corticostriatal synaptic transmission

210

Anton Dvorzhak and Rosemarie Grantyn

is jeopardized by insufficient supply with brain-derived neurotrophic factor, BDNF [45], and reduced signal transfer along the distal dendrites of D2-expressing striatal projection neurons [46]. The duration of the NMDA receptor (NMDAR) component of corticostriatal EPSCs could be prolonged [44], which might be attributed to the reduced glutamate uptake capacity seen in individual striatal astrocytes [22, 47]. Live single-synapse imaging in striatal slices has now added new details on the mechanisms of corticostriatal signal transmission [36, 44]. This approach could also provide a better basis for the testing of astrocyte-directed treatment options. To this end, one would need experimental settings allowing to discriminate alterations (i) in the glutamate release as opposed to clearance, (ii) at one versus the other class of synapses, and (iii) under the influence of one versus another type of genetically modified astrocyte.

2

Methods and Experimental Paradigms

2.1 Analysis of Individual Synapses: Tests That Can be Performed in Acute Brain Slices from Adult Mice to Identify Specific Types of Synapses in a Mixed Population of Glutamatergic Contacts

Our method of single-synapse imaging has already been described in some detail [36, 48]. We would, therefore, like to dedicate more space to particular applications which may help to deal with critical issues in preclinical research.

2.1.1 The Necessity to Investigate the Synapses In Situ and at an Appropriate Age

Figure 1a, b shows one of very many structural variants of the perisynaptic astrocyte process (PAP) around the two neuronal elements of a tripartite synapse. If present, this PAP, would contain a rather defined set of specializations, including the clusters of EAAT2. How effective the latter can operate may depend on the functional state of the connected astrocyte network. In accordance with their general housekeeping function, astroglial cells are extremely sensitive to the conditions of operation. Although still little is known on the extra- and intracellular pathways regulating astrocytic EAAT2, one should be prepared to find this regulation influenced by brain dissociation, inter- and intraregional diversity of astrocyte phenotypes, and the animal’s sex, age, and disease. Therefore, any effort should be made to study synaptic transmission in vivo, or at least in situ. The use of acute parasagittal or horizontal brain slices from adult mice could be a reasonable compromise for

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . .

211

Fig. 1 Analysis of single-synapse glutamate transients at tripartite synapses of the corticostriatal pathway. (a) The concept of tripartite synapse illustrated by an electron microscopic image showing a perisynaptic astrocyte process (PAP) in red. (Modified from [56]). (b) Major components of a tripartite synapse as shown

212

Anton Dvorzhak and Rosemarie Grantyn

the murine striatum. Nevertheless, a set of accompanying electrophysiological or biochemical indicators should ensure a reproducible standard in the overall performance of a given pathway under conditions, for instance, recording of local field potentials or measurement of tissue lactate or extracellular [K+] levels. We typically work with mice at the age of 12–18 months. At this age, heterozygote Q175-KI HD mice are already afflicted by pronounced loss of brain and body weight as well as slight-tomoderate symptoms of hypokinesia. The severity of symptoms correlates with the number of glutamate repeats and, of course, age. However, even siblings with the same number of CAG repeats could vary in the individual expression of HD symptoms. It, therefore, proved to be quite worth the effort to dedicate one or two additional hours to perform a neurological express examination before sacrificing a mouse for subsequent synapse testing. A video-recorded step-over latency test (SOLT) from our lab proved to be both quick and reliable [36]. In combination with open-field imaging of spontaneous motor behavior, the former provided us with criteria to include/exclude a given data set into the sample for inter-animal comparison. Functional embedding of elementary synapse data into a more complete functional system is not only attractive but also required to control the admissibility of data sets from different animals. 2.1.2 Selection of the Appropriate Sensor for the Given Type of Synapse

The striatum contains glutamatergic terminals from few and well identifiable distant sources. The terminals formed by corticostriatal axons either originate in the layer V (PT afferents) or in layers II/III (IT afferents) (Fig. 1c, d). This provides us with a sufficient number of options for synapse identification, including labeling with a fluorescent glutamate sensor. This sensor must match the kinetic characteristics of the synapse of interest. The “ultrafast”

ä Fig. 1 (continued) in (a). Diffusion restriction and active glutamate uptake by the surrounding astrocytes limit transmitter access to pre- and extrasynaptic glutamate receptors, including NMDA receptors (NMDARs) on either side. The AMPA receptor-mediated postsynaptic response is largely unaffected by the efficacy of astrocytic glutamate clearance. At corticostriatal synapses, loading of glutamate into the presynaptic vesicles is carried out by the vesicular glutamate transporter type 1 (vGluT1), while the plasma membrane transport of glutamate is mostly carried out by EAAT2. (c) Sagittal section through the mouse brain illustrating the origin and orientation of corticostriatal axons. The boxed area indicates the preferred area of recording in the dorsal striatum. The corticostriatal pathway can be visualized by using the iGluu expression vector shown in green. The fluorescence in the area of termination was used for the imaging of electrically evoked synaptic glutamate transients. (d) Simplified scheme of the corticostriatal circuitry illustrating the concept of preferential projection of pyramidal tract (PT) neurons to indirect pathway striatal projection neurons (iSPNs) and intratelencephalic (IT) neurons to direct pathway SPNs (dSPNs), with size differences between the IT and PT terminals. (e) Resting iGluu fluorescence merged to the respective 63× DIC image of an axon with three adjacent varicosities and a stimulation pipette at the central bouton. (e–g) Basic information on the requirements for single-synapse imaging

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . .

213

glutamate sensor iGluu used here was developed in the Lab of Katalin To¨ro¨k at St. Georges, University of London [49], and represents a modification of the earlier introduced high-affinity but slow glutamate sensor iGluSnFR [50]. The dissociation time constant of iGluu amounts to 2.1 ms at 20 °C and to 0.68 ms when extrapolated to a temperature of 34 °C [49]. In organotypical hippocampal cultures, glutamate transients of Schaffer collateral terminals rendered decay time constants of about 2.7 ms. In the case of PT terminals, the time constant of glutamate decay was 3.4 ms at 30 °C (compared to the average time constant of decay of AMPA receptor-mediated unitary EPSCs: 4.1 ms). So, from the viewpoint of its dissociation time constant, iGluu seems to be well suited for experiments with PT corticostriatal synapses. The injections and expression times of the iGluu vector should be adjusted to obtain the best possible discrimination of single boutons not only during activity but also at rest. Therefore, the label should be strong but scarce. In our case, this was achieved with multiple (4×) injections of 0.3 μL of AAV9-CamKII.iGluu. WPRE-hGH at a concentration of 7.5 * 1013 gc/ml at the following coordinates with respect to bregma (mm): anterior 1.5, lateral 1.56, 1.8, 2.04, 2.28 and ventral 1.7. With such microinjections into the motor cortex and survival times of >6 weeks, one can obtain satisfying expression patterns, notably if using a CaMKII promoter to prevent labeling of striatal neurons by virus leaking out of the motor cortex. Figure 1e provides an example with three suitable varicosities. 2.1.3 Requirements for Sequential Testing of Different Synapses Within One Slice

As we aimed at comparing synapse pools of similar origin but in contact with different types of astrocytes in the same slice, we needed to have sufficient flexibility for synapse scanning and electrode positioning without bleaching the neighboring sites. Therefore, slices had to be prepared and maintained in the dark (with red light), and the stimulation pipette had to be placed with minimal focal illumination. A Zeiss microscope for one-photon wide-field imaging (AxioExaminer A1) with a 63× /NA 1.0 water immersion objective and a UGA-40 laser positioning system from Rapp OptoElectronic served this purpose. In most cases, glutamate transients could be isolated by applying, in the presence of tetrodotoxin (TTX), depolarizing electrical pulses of 0.63 μm could be used to define a class of “large” synapses (presumably PT type) as opposed to “small” (presumably IT type).

216

Anton Dvorzhak and Rosemarie Grantyn

Fig. 3 The characteristics of IT as opposed to PT terminals and tests for saturation of glutamate clearance. (a) Bimodal distribution of bouton diameters as determined by the suprathreshold resting fluorescence before stimulation. Boutons with diameter ≥ 0.63 μm were defined as “large” and assumed to be issued by PT axons. (b, c) Differential dynamic behavior of IT (b) as opposed to PT (c) synapses. The former typically displayed paired pulse depression, and the latter—paired pulse potentiation. (d, e) At high-frequency

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . .

217

In view of the rather distinct differences in “bouton size,” we asked whether or not transmitter depletion or exhaustion of the glutamate buffering capacity would be more prominent at one or the other kind of terminal. It was found that the first responses elicited with terminals that were classified as “large” also exhibited larger values of “maximal amplitude,” “peak amplitude,” and “spread” (see [44] for details). The differences between “small” and “large” were even more obvious at the end of a series of high-frequency (100 Hz) responses (Fig. 3b, c). These experiments suggest that the pool of “large” must cope with larger total amount of glutamate. Would such synapses be more prone to exhaustion of glutamate clearance? Interestingly, this was not the case. Despite evidence for an accumulation of residual glutamate and higher integral values of IGluu (Fig. 3d, e), we did not find significant differences between the mean “TauD” values (Fig. 3f). Only few (4/17) of the large PT synapses exhibited a prolongation of the glutamate decay when challenged with 100 Hz stimulation. One could imagine that such cases could become more frequent if PT synapses were facing an unfavorable astrocytic environment. 2.1.6 Any Evidence for Saturation of Astrocytic Glutamate Uptake in Normal Mice?

The results illustrated in Fig. 3 shed new light on the much disputed issue of whether glutamate uptake is ever saturated under physiological activation conditions. In agreement with previous notions from other preparations [9, 10], we would suggest that saturation of uptake is unlikely to occur in either subcomponent of the corticostriatal pathway of healthy mice. Concerning the PT synapses challenged by high-frequency activation, one has to consider that in the intact mouse brain, PT units rarely reach frequencies as high as 100 Hz. Moreover, our tests were performed at reduced temperature (about 30 °C) meaning that at normal body temperature, all transporters would work even more effectively. The situation may, however, be different in mice with impaired astrocyte function/deficient EAAT2. To obtain more direct evidence for a possible regulation of PT synapses by astrocytic glutamate uptake, we simulated deficient glutamate clearance by incubating the slices with the competitive inhibitor of glutamate uptake TFP-TBOA. Figure 3g, h presents the outcome of these experiments in WT. The findings demonstrate that the parameter “TauD” is indeed sensitive to a reduction of the tissue clearance

ä Fig. 3 (continued) activation, IT synapses exhibited depletion of release but no sign of clearance deficit. The larger PT synapses maintain stable release with accumulation of increasing amounts of residual glutamate and respective buildup of the integral iGluu intensity values. (f) This leads to slightly wider “spread” of the iGluu signal (not illustrated) but little change in “TauD” of the last as compared to 1st response in the sequence (Fig. 3f). (g, h) In PT terminals glutamate, deterioration of glutamate clearance could be provoked by slice incubation in the blocking substrate TFB-TBOA (100 nM). After such treatment, normal WT mice showed a highly significant increase in the “TauD” values

218

Anton Dvorzhak and Rosemarie Grantyn

capacity. As in the striatum most clearance is carried out by astrocytic EAAT2, this finding can also be regarded as evidence for an involvement of astrocytes in the regulation of PT signaling. 2.2 Single-Synapse Analysis in Transgenic Mice

With the above tools at hand, one can also address more complex issues, including the estimation of disease- and treatment-induced alterations of synaptic glutamate release and clearance in the corticostriatal pathway. However, the quantification of such alterations in transgenic animals would typically require pooling of singlesynapse data for comparison of different animal groups. In the case of thee mhtt-knock-in mice, one has to expect life-long compensatory adjustments in a large variety of variables other than glutamate uptake. Nevertheless, it proved possible (1) to reveal mhtt-related alterations in individual astrocytes, (2) to demonstrate functional recovery after expression of a modified EAAT2 transgene, and (3) to evaluate the properties of single PT-type synapses in their dependence on the type of transgene expressed in the contacted astrocytes.

2.2.1 mHTT-Related Alterations in the State of Individual Astrocytes, as Revealed by Sodium Imaging of Glutamate Uptake with SBFI

Previous studies on a variety of transgenic mouse models of HD consistently showed lower levels of SLC1A2 mRNA and the encoded protein EAAT2 in the dorsal striatum. It was, therefore, suggested that impaired clearance of synaptic glutamate in the striatum could contribute to the HD phenotype. Even though radioactive uptake assays seemed to confirm this idea, the initial results of experiments with glutamate imaging rendered contradictory results, mostly due to insufficient specificity of release activation and low spatial and temporal resolution of the glutamate fluorescence signal in the investigated slices (see Subheading 1 for more). Considering the usually protracted time course of neurodegeneration in most animal models of HD, we deemed it possible that the resulting alterations would appear in form of local hotspots rather than even patterns [44]. We, therefore, need an experimental scheme where results on synaptic glutamate clearance could be set in relation to estimates characterizing the local astroglial glutamate uptake capacity. As glutamate uptake is associated with co-transport of sodium, one can use sodium imaging to evaluate the transport of individual astrocytes. The transport substrate can either be applied by focal uncaging [22] or by local superfusion [36]. In the former case, glutamate concentrations could be adjusted to those expected in the synaptic cleft (1–3 mM). In the following, we provide some details on the SBFI method as used in our lab to record the transport activity of individual astrocytes. Briefly, slices were for 20–30 min incubated in oxygenated artificial cerebrospinal fluid (ACSF) containing 222 μM of the membrane-permeable form of the fluorescent sodium sensor SBFI-AM (Thermo Fisher Scientific). A wide-field fluorescence

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . .

219

imaging was performed using a digital live acquisition system (such as Zyla 4.2 plus from Andor) attached to an upright Zeiss microscope (Axio Examiner A1). Images were collected with a Zeiss 63× NA 1.0 water immersion objective selecting the cells according to their fluorescent tag (mRuby) and the bulk-loaded SBFI. Ultraviolet or visible light was applied by an UVICO system (Rapp OptoElectronic), combined with suitable filter sets from Omega Optical attached to a FW 1000 filter wheel (Applied Scientific Instrumentation). Single-wavelength sodium imaging was performed by excitation of SBFI at 380 nm (sodium-sensitive wavelength). SBFI emission was collected at >510 nm. ROIs with a size of 3.2 μm × 3.2 μm were defined on the cell body. The spatial resolution was then 0.8 μm/pixel and the binning was 8 × 8. The exposure times was set to 150 ms in all experiments. Images were acquired every 3 s. A custom-written software was used to control image acquisition and the valves operating the superfusion system. In the case of experiments with bath application (Fig. 4), exposure to L-aspartate (1 mM) was preceded by 1 min bathing in ACSF and followed by a 3 min washout time. Subsequently, the same routine was repeated in the presence of the glutamate transport blocker TFB-TBOA 2 μM (Tocris). For both traces, the fluorescence change was calculated from the average of the baseline fluorescence intensities for each ROI as ΔF/F = (FSBFI - FSBFI(baseline)/FSBFI(baseline). The difference between the control and TBOA traces represents the Laspartate-induced sodium transient that is mediated by all available glutamate transporters. The aspartate-induced uptake capacity varied from cell to cell. Nevertheless, it was possible to obtain significant genotype-related differences between the animal groups. In comparison with mRubylabeled WT astrocytes, mRuby-labeled HD astrocytes exhibited an average uptake deficit of 40%. A similar deficit was found with focal glutamate uncaging on SR101β-labeled astrocytes [22]. 2.2.2 Experiments to Characterize the HD Phenotype of Glutamate Uptake

The design of the experiments laid out in Fig. 4a pursued a dual analytical purpose (apart from or main aim of exploring new therapeutic possibilities in HD): First, to allow for two independent controls to verify the assumption that there indeed is a distinct HD phenotype and, second, to define a distinct pool of astrocytes, with experimentally altered glutamate uptake, for subsequent single-synapse testing. For the first purpose, we compared three animal groups (Fig. 4a). The “HD:CTRL” reference group consisted of aged (>1 year) Z-Q175-KI mice, a widely used mouse model of HD with hypokinetic motor phenotype [51]. This group is to be compared with matched samples of “WT:CTRL” and “HD: TREATED.” The viral vector constructs used for each group are listed in Fig. 4b.

220

Anton Dvorzhak and Rosemarie Grantyn

Fig. 4 The state of glutamate uptake in individual astrocytes (SBFI imaging). (a) Design of experiments for inter-group comparison. (b) Overview of the vectors used for two-directional testing of mHTT-related changes in perisynaptic EAAT2 protein. The pooled data from HD:CTRL mice were compared to WT:CTRL or HD-TREATED. (c, d) The vectors were injected into the dorsal striatum of carefully matched WT and HD mice (Q175-KI). A single injection of 1 * 10 viral particles is sufficient to label about 66% of the S100β-stained astrocytes. (e, f) Quantification of results from the three test groups. Results from mRuby-positive astrocytes only. Recordings of SBFI fluorescence, in the presence of CBX (100 μM), DNQX (10 μM) and MK801 (1 μM).

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . .

221

In the past, a similar study was performed with full-length EAAT2 as “experimental transgene” [52]. However, these experiments failed to provide evidence for a distinct effect of transgenic EAAT2 in HD mice and, therefore, were inconclusive. We then regarded the possibility of using a genetically modified version of EAAT2, and this served our purposes much better, especially, the C-terminal-truncated variant EAAT2-S506X [36]. For the combination with SBFI or iGluu we selected the red fluorochrome mRuby for astrocyte labeling in all three groups. The bilateral intrastriatal injections of the virus (Fig. 4c) were be carried out under ketamine/xylazine anesthesia (100/10 mg/kg) and with the aid of a stereotaxic frame to target glass micropipettes to the following coordinates with respect to bregma (mm): anterior 0.75, lateral 2.0, ventral 2.5. When using PHP.B as viral shuttle, one injection of 1.0 * 109 gc has been sufficient to label astrocytes throughout the entire striatum (Fig. 1d) and to obtain transduction of about 66% of S100β-positive astrocytes [36]. After about 3 weeks, the animals were ready for the express check of motor behavior, before being sacrificed for the experiments with SBFI. Astrocytic glutamate was evaluated independently for each cell, animal, and animal group. Figure 4e, f illustrates the final outcome of this study by showing the averaged traces from all astrocytes of a group. The differences were significant when comparing HD:CTRL (red trace) with WT:CTRL (grey), on one side, and HD:CTRL (red) with HD:TREATED (green), on the other side. A comparison of HD:TREATED with WT:CTRL provides a measure of recovery. Comparison of pooled data from distinct animal groups is a very common approach in preclinical research. In general, the quality of conclusions is very much dependent on how well defined the sample is at any level variance analysis. One should remember that the evaluation of data from genetically engineered mice may require multilevel nested ANOVA to consider the variance of synapses, cells, or animal genotypes separately [53]. In the experiments illustrated in Fig. 4e, f, data acquisition was restricted to labeled astrocytes. Moreover, as the label had been expressed under the promoter for glial fibrillary acidic protein, ä Fig. 4 (continued) The traces represent an average from all synapses in the group. The amplitude of the signal is expressed as ΔF/F during the last 15 s of L-Asp application. F is the mean fluorescence at rest, before drug application. Please notice color coding. Measurement of glutamate uptake by sodium imaging with SBFI. The differences between WT:CTRL and HD:CTRL and HD:CTRL and HD:TREATED were significant according to two-level (“nested”) statistics (animal level, cell level, and posthoc comparison between groups). (g) Viewfield with three astrocytes, where one is transduced and two are not. (h) Comparison of results obtained from mRuby+ vs. mRuby- astrocytes

222

Anton Dvorzhak and Rosemarie Grantyn

GFAP, we may have preferentially dealt with the so-called “reactive astrocytes”. These astrocytes express GFAP at higher level [52]. Did this have any importance for the detection of HD-related differences? Possibly not, because the uptake deficit obtained in the present samples was about the same as the previously determined deficit in cells labeled with SR101β+, a marker staining all astrocytes [54]. The restriction of the sample to GFAP+ astrocytes could, however, be relevant for the detection of treatment effects. A comparison of transduced as opposed to not transduced astrocytes in one and the same slice indeed indicates that the treatment effect solely occurs in astrocytes with the EAAT2-S506X transgene (Fig. 4g, h). As the presence of the EAAT2-S506X changed the state of the astrocytes almost back to the WT:CTRL level, we now have the option of double control when studying the impact of astrocyte uptake on glutamate clearance. 2.2.3 Imaging of Glutamate Clearance in Synapses Contacting Astrocytes with Genetically Altered Glutamate Uptake

A set of experiments performed with single synapses on genetically altered astrocytes is illustrated in Fig. 5a–c. One can see that the representative specimen from the WT, HD, and treated HD group differ with regard to the time course of an extension of the glutamate elevation but not it’s amplitude. This impression is confirmed by comparing the respective intensity traces (Fig. 5d–f) and sample means (Fig. 5g–i). There is a clear group-related difference in the pooled values of “TauD” and “Spread,” i.e., the indicators of synaptic clearance, whereas “Peak amplitude” as an indicator of glutamate release remained unaffected. We can conclude that the uptake characteristics of the contacted astrocyte influence the dynamics of glutamate clearance in the associated synapses.

2.2.4 Can SingleSynapse Data be Used to Estimate the DiseaseModifying Effect of Altered Gene Expression in Astrocytes?

Considering that neurodegeneration might occur in a discontinuous manner and in dependence on the local astroglial environment, it is particularly important to take a closer look at the distribution of the collected single-synapse variables. The scatterplots of Fig. 5j, k indeed show a remarkable widening in the range of TauD values from HD mice, which suggests that with disease progression the resilience of individual synapses might get increasingly diverse. Apparently healthy synapses seem to coexist with dysfunctional ones. However, TauD values larger than 15 ms were never found in WT. Using this cutoff criterion, one could thus obtain a quick estimate of the TauD-derived “Fraction of dysfunctional synapses” as another phenotypic marker in HD (Fig. 5l). In hypokinetic Q175-KI heterozygotes (our current mouse model of HD), this fraction amounted to about 30%. Most remarkably, treatment with EAAT2-S506X decreased the “Fraction of dysfunctional synapses” back to WT level, i.e., to zero (Fig. 5l).

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . .

223

Fig. 5 Imaging of glutamate clearance in synapses contacting astrocytes with genetically altered level of glutamate uptake activity. (a–c) Representative examples of iGluu imaging of single P-type corticostriatal synaptic terminals in slices from WT, HET, and treated HET. The selected images were acquired at the

224

3

Anton Dvorzhak and Rosemarie Grantyn

Conclusions Thus, the “single synapse–single astrocyte” design of pathway analysis (Fig. 5m) might render very satisfying results. We agree with [55] that “a new era of quantifying glutamate clearance in health and disease” may have started due to the availability of improved genetic tools for glutamate sensors, cell-specific labeling/modification, and optical discrimination. Last but not the least, our experimental approach may bring the astrocytic glutamate transport back to the list of molecular targets for therapeutic interventions in HD.

Acknowledgments This work was supported by CHDI (A-12467), the German Research Foundation, under Germany’s Excellence Strategy (Exc 2049 – 390688087) and intramural Charite´ Research Funds to R.G. References 1. Danbolt NC (2001) Glutamate uptake. Prog Neurobiol 65:1–105 2. Vandenberg RJ, Ryan RM (2013) Mechanisms of glutamate transport. Physiol Rev 93:1621– 1657 3. Jensen AA, Fahlke C, Bjorn-Yoshimoto WE, Bunch L (2015) Excitatory amino acid transporters: recent insights into molecular mechanisms, novel modes of modulation and new therapeutic possibilities. Curr Opin Pharmacol 20:116–123 4. Rothstein JD, Dykes-Hoberg M, Pardo CA, Bristol LA, Jin L, Kuncl RW, Kanai Y, Hediger

MA, Wang Y, Schielke JP, Welty DF (1996) Knockout of glutamate transporters reveals a major role for astroglial transport in excitotoxicity and clearance of glutamate. Neuron 16: 675–686 5. Lehre KP, Danbolt NC (1998) The number of glutamate transporter subtype molecules at glutamatergic synapses: chemical and stereological quantification in young adult rat brain. J Neurosci 18:8751–8757 6. Melone M, Bellesi M, Ducati A, Iacoangeli M, Conti F (2011) Cellular and synaptic localization of EAAT2a in human cerebral cortex. Front Neuroanat 4:151

ä Fig. 5 (continued) response peak to illustrate the spatial extension of elevated glutamate (boxed area). Pink pixels within the black boundaries are pixels where stimulation of the bouton elicited a fluorescence increase (ΔF) to values larger than the resting level prior to stimulation (F). (d–f) iGluu transients for the examples shown in a–c. Recordings before and after electrical stimulation of the given bouton in the presence of TTX. The traces represent the mean fluorescence intensity calculated from all suprathreshold pixels in the ROI. Red curves: Monoexponential functions fitted to the iGluu traces. In black—respective time constants of decay (TauD). (g–i) Evaluation of “TauD,” “Spread,” and “Peak amplitude” values for comparison of the three test groups (nested ANOVA statistics). See Part 1 of this chapter for definitions. (j, k) Scatterplots of results obtained from the individual synapses. Please note the HD-related shift in the range of “TauD” values. Results obtained from the pixel with the highest elevation of iGluu fluorescence. (l) Cumulative probability plot to illustrate the absence of “TauD” larger than 15 ms in WT and in EAAT2-S506X-treated HD. TauD values >15 ms identify pathological synapses in Q175-KI. (m) Only synapses in the immediate vicinity of transduced astrocytes were included, as confirmed by the location of the tested iGluu-expressing varicosity on the territory of mRuby-positive astrocytes (arrowhead)

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . . 7. Cahoy JD, Emery B, Kaushal A, Foo LC, Zamanian JL, Christopherson KS, Xing Y, Lubischer JL, Krieg PA, Krupenko SA, Thompson WJ, Barres BA (2008) A transcriptome database for astrocytes, neurons, and oligodendrocytes: a new resource for understanding brain development and function. J Neurosci 28:264–278 8. Marcaggi P, Attwell D (2004) Role of glial amino acid transporters in synaptic transmission and brain energetics. Glia 47:217–225 9. Diamond JS, Jahr CE (2000) Synaptically released glutamate does not overwhelm transporters on hippocampal astrocytes during high-frequency stimulation. J Neurophysiol 83:2835–2843 10. Tzingounis AV, Wadiche JI (2007) Glutamate transporters: confining runaway excitation by shaping synaptic transmission. Nat Rev Neurosci 8:935–947 11. Pekny M, Pekna M, Messing A, Steinhauser C, Lee JM, Parpura V, Hol EM, Sofroniew MV, Verkhratsky A (2016) Astrocytes: a central element in neurological diseases. Acta Neuropathol 131:323–345 12. Verkhratsky A, Zorec R, Rodriguez JJ, Parpura V (2016) Astroglia dynamics in ageing and Alzheimer’s disease. Curr Opin Pharmacol 26: 74–79 13. Khakh BS, Beaumont V, Cachope R, MunozSanjuan I, Goldman SA, Grantyn R (2017) Unravelling and exploiting astrocyte dysfunction in Huntington’s disease. Trends Neurosci 40:422–437 14. Langfelder P, Cantle JP, Chatzopoulou D, Wang N, Gao F, Al Ramahi I, Lu XH, Ramos EM, El Zein K, Zhao Y, Deverasetty S, Tebbe A, Schaab C, Lavery DJ, Howland D, Kwak S, Botas J, Aaronson JS, Rosinski J, Coppola G, Horvath S, Yang XW (2016) Integrated genomics and proteomics define huntingtin CAG length-dependent networks in mice. Nat Neurosci 19:623–633 15. Bradford J, Shin JY, Roberts M, Wang CE, Li XJ, Li S (2009) Expression of mutant huntingtin in mouse brain astrocytes causes age-dependent neurological symptoms. Proc Natl Acad Sci U S A 106:22480–22485 16. Meunier C, Merienne N, Jolle C, Deglon N, Pellerin L (2016) Astrocytes are key but indirect contributors to the development of the symptomatology and pathophysiology of Huntington’s disease. Glia 64:1841–1856 17. Faideau M, Kim J, Cormier K, Gilmore R, Welch M, Auregan G, Dufour N, Guillermier M, Brouillet E, Hantraye P, Deglon N, Ferrante RJ, Bonvento G (2010)

225

In vivo expression of polyglutamine-expanded huntingtin by mouse striatal astrocytes impairs glutamate transport: a correlation with Huntington’s disease subjects. Hum Mol Genet 19: 3053–3067 18. Shin JY, Fang ZH, Yu ZX, Wang CE, Li SH, Li XJ (2005) Expression of mutant huntingtin in glial cells contributes to neuronal excitotoxicity. J Cell Biol 171:1001–1012 19. Lievens JC, Woodman B, Mahal A, SpasicBoscovic O, Samuel D, Kerkerian-Le Goff L, Bates GP (2001) Impaired glutamate uptake in the R6 Huntington’s disease transgenic mice. Neurobiol Dis 8:807–821 20. Huang K, Kang MH, Askew C, Kang R, Sanders SS, Wan J, Davis NG, Hayden MR (2010) Palmitoylation and function of glial glutamate transporter-1 is reduced in the YAC128 mouse model of Huntington disease. Neurobiol Dis 40:207–215 21. Parsons MP, Vanni MP, Woodard CL, Kang R, Murphy TH, Raymond LA (2016) Real-time imaging of glutamate clearance reveals normal striatal uptake in Huntington disease mouse models. Nat Commun 7:11251 22. Dvorzhak A, Vagner T, Grantyn R (2016) Functional indicators of glutamate transport in single striatal astrocytes and the influence of Kir4.1 in normal and Huntington mice. J Neurosci 16:4959–4975 23. Parievsky A, Moore C, Kamdjou T, Cepeda C, Meshul CK, Levine MS (2017) Differential electrophysiological and morphological alterations of thalamostriatal and corticostriatal projections in the R6/2 mouse model of Huntington’s disease. Neurobiol Dis 108:29– 44 24. Petr GT, Schultheis LA, Hussey KC, Sun Y, Dubinsky JM, Aoki C, Rosenberg PA (2013) Decreased expression of GLT-1 in the R6/2 model of Huntington’s disease does not worsen disease progression. Eur J Neurosci 38:2477–2490 25. Petr GT, Sun Y, Frederick NM, Zhou Y, Dhamne SC, Hameed MQ, Miranda C, Bedoya EA, Fischer KD, Armsen W, Wang J, Danbolt NC, Rotenberg A, Aoki CJ, Rosenberg PA (2015) Conditional deletion of the glutamate transporter GLT-1 reveals that astrocytic GLT-1 protects against fatal epilepsy while neuronal GLT-1 contributes significantly to glutamate uptake into synaptosomes. J Neurosci 35:5187–5201 26. Rothe T, Deliano M, Wojtowicz AM, Dvorzhak A, Harnack D, Paul S, Vagner T, Melnick I, Stark H, Grantyn R (2015) Pathological gamma oscillations, impaired dopamine release, synapse loss and reduced dynamic

226

Anton Dvorzhak and Rosemarie Grantyn

range of unitary glutamatergic synaptic transmission in the striatum of hypokinetic Q175 Huntington mice. Neuroscience 311:519–538 27. Deng YP, Wong T, Bricker-Anthony C, Deng B, Reiner A (2013) Loss of corticostriatal and thalamostriatal synaptic terminals precedes striatal projection neuron pathology in heterozygous Q140 Huntington’s disease mice. Neurobiol Dis 60:89–107 28. Reiner A, Deng YP (2018) Disrupted striatal neuron inputs and outputs in Huntington’s disease. CNS Neurosci Ther 24:250–280 29. Kalandadze A, Wu Y, Robinson MB (2002) Protein kinase C activation decreases cell surface expression of the GLT-1 subtype of glutamate transporter. Requirement of a carboxylterminal domain and partial dependence on serine 486. J Biol Chem 277:45741–45750 30. Underhill SM, Wheeler DS, Amara SG (2015) Differential regulation of two isoforms of the glial glutamate transporter EAAT2 by DLG1 and CaMKII. J Neurosci 35:5260–5270 31. Gonzalez-Gonzalez IM, Garcia-Tardon N, Gimenez C, Zafra F (2008) PKC-dependent endocytosis of the GLT1 glutamate transporter depends on ubiquitylation of lysines located in a C-terminal cluster. Glia 56:963–974 32. Leinenweber A, Machtens JP, Begemann B, Fahlke C (2011) Regulation of glial glutamate transporters by C-terminal domains. J Biol Chem 286:1927–1937 33. Gibb SL, Boston-Howes W, Lavina ZS, Gustincich S, Brown RH Jr, Pasinelli P, Trotti D (2007) A caspase-3-cleaved fragment of the glial glutamate transporter EAAT2 is sumoylated and targeted to promyelocytic leukemia nuclear bodies in mutant SOD1-linked amyotrophic lateral sclerosis. J Biol Chem 282: 32480–32490 34. Foran E, Bogush A, Goffredo M, Roncaglia P, Gustincich S, Pasinelli P, Trotti D (2011) Motor neuron impairment mediated by a sumoylated fragment of the glial glutamate transporter EAAT2. Glia 59:1719–1731 35. Rosenblum LT, Shamamandri-Markandaiah S, Ghosh B, Foran E, Lepore AC, Pasinelli P, Trotti D (2017) Mutation of the caspase-3 cleavage site in the astroglial glutamate transporter EAAT2 delays disease progression and extends lifespan in the SOD1-G93A mouse model of ALS. Exp Neurol 292:145–153 36. Hirschberg S, Dvorzhak A, Rasooli-Nejad SMA, Angelov S, Kirchner M, Mertins P, L€attig-Tu¨nnemann G, Harms C, Schmitz D, Grantyn R (2022) Uncoupling the excitatory amino acid transporter 2 from its C-terminal interactome restores synaptic glutamate

clearance at corticostriatal synapses and alleviates mutant huntingtin-induced hypokinesia. Front Cell Neurosci 15:792652 37. Lines J, Covelo A, Gomez R, Liu L, Araque A (2017) Synapse-specific regulation revealed at single synapses is concealed when recording multiple synapses. Front Cell Neurosci 11:367 38. Octeau JC, Chai H, Jiang R, Bonanno SL, Martin KC, Khakh BS (2018) An optical neuron-astrocyte proximity assay at synaptic distance scales. Neuron 98:49–66 39. Plotkin JL, Surmeier DJ (2015) Corticostriatal synaptic adaptations in Huntington’s disease. Curr Opin Neurobiol 33C:53–62 40. Villalba RM, Smith Y (2018) Loss and remodeling of striatal dendritic spines in Parkinson’s disease: from homeostasis to maladaptive plasticity? J Neural Transm (Vienna) 125:431–447 41. Huerta-Ocampo I, Mena-Segovia J, Bolam JP (2014) Convergence of cortical and thalamic input to direct and indirect pathway medium spiny neurons in the striatum. Brain Struct Funct 219:1787–1800 42. Rebec GV (2018) Corticostriatal network dysfunction in Huntington’s disease: deficits in neural processing, glutamate transport, and ascorbate release. CNS Neurosci Ther 24: 281–291 43. Cepeda C, Levine MS (2022) Synaptic dysfunction in Huntington’s disease: lessons from genetic animal models. Neuroscientist 28:20–40 44. Dvorzhak A, Helassa N, Torok K, Schmitz D, Grantyn R (2019) Single synapse indicators of impaired glutamate clearance derived from fast iGluu imaging of cortical afferents in the striatum of normal and Huntington (Q175) mice. J Neurosci 39:3970–3982 45. Plotkin JL, Day M, Peterson JD, Xie Z, Kress GJ, Rafalovich I, Kondapalli J, Gertler TS, Flajolet M, Greengard P, Stavarache M, Kaplitt MG, Rosinski J, Chan CS, Surmeier DJ (2014) Impaired TrkB receptor signaling underlies corticostriatal dysfunction in Huntington’s disease. Neuron 83:178–188 46. Carrillo-Reid L, Day M, Xie Z, Melendez AE, Kondapalli J, Plotkin JL, Wokosin DL, Chen Y, Kress GJ, Kaplitt M, Ilijic E, Guzman JN, Chan CS, Surmeier DJ (2019) Mutant huntingtin enhances activation of dendritic Kv4 K(+) channels in striatal spiny projection neurons. elife 8:40818 47. Tong X, Ao Y, Faas GC, Nwaobi SE, Xu J, Haustein MD, Anderson MA, Mody I, Olsen ML, Sofroniew MV, Khakh BS (2014) Astrocyte Kir4.1 ion channel deficits contribute to

Analysis of Synaptic Glutamate Clearance as a Possible Indicator of. . . neuronal dysfunction in Huntington’s disease model mice. Nat Neurosci 17:694–703 48. Dvorzhak A, Grantyn R (2020) Single synapse indicators of glutamate release and uptake in acute brain slices from normal and Huntington mice. J Vis Exp e60113 49. Helassa N, Durst CD, Coates C, Kerruth S, Arif U, Schulze C, Wiegert JS, Geeves M, Oertner TG, Torok K (2018) Ultrafast glutamate sensors resolve high-frequency release at Schaffer collateral synapses. Proc Natl Acad Sci U S A 115:5594–5599 50. Marvin JS, Borghuis BG, Tian L, Cichon J, Harnett MT, Akerboom J, Gordus A, Renninger SL, Chen TW, Bargmann CI, Orger MB, Schreiter ER, Demb JB, Gan WB, Hires SA, Looger LL (2013) An optimized fluorescent probe for visualizing glutamate neurotransmission. Nat Methods 10:162–170 51. Menalled LB, Kudwa AE, Miller S, Fitzpatrick J, Watson-Johnson J, Keating N, Ruiz M, Mushlin R, Alosio W, McConnell K, Connor D, Murphy C, Oakeshott S, Kwan M, Beltran J, Ghavami A, Brunner D, Park LC, Ramboz S, Howland D (2012) Comprehensive behavioral and molecular characterization

227

of a new knock-in mouse model of Huntington’s disease: zQ175. PLoS One 7:e49838 52. Vagner T, Dvorzhak A, Wojtowicz AM, Harms C, Grantyn R (2016) Systemic application of AAV vectors targeting GFAPexpressing astrocytes in Z-Q175-KI Huntington’s disease mice. Mol Cell Neurosci 77:76– 86 53. Aarts E, Verhage M, Veenvliet JV, Dolan CV, van der Sluis S (2014) A solution to dependency: using multilevel analysis to accommodate nested data. Nat Neurosci 17:491–496 54. Nimmerjahn A, Kirchhoff F, Kerr JN, Helmchen F (2004) Sulforhodamine 101 as a specific marker of astroglia in the neocortex in vivo. Nat Methods 1:31–37 55. Brymer KJ, Barnes JR, Parsons MP (2021) Entering a new era of quantifying glutamate clearance in health and disease. J Neurosci Res 99:1598–1617 56. Halassa MM, Fellin T, Takano H, Dong JH, Haydon PG (2007) Synaptic islands defined by the territory of a single astrocyte. J Neurosci 27:6473–6477

Chapter 11 Computational Models of Astrocyte Function at Glutamatergic Synapses Kerstin Lenk, Audrey Denizot, Barbara Genocchi, Ippa Sepp€al€a, Marsa Taheri, and Suhita Nadkarni Abstract At tripartite synapses, astrocytes are in close contact with neurons and contribute to various functions, from synaptic transmission, maintenance of ion homeostasis, and glutamate uptake to metabolism. However, disentangling the precise contribution of astrocytes to those phenomena and the underlying biochemical mechanisms is remarkably challenging. This notably results from their highly ramified morphology, the nanoscopic size of the majority of astrocyte processes, and the poorly understood information encoded by their spatiotemporally diverse calcium signals. This book chapter presents selected computational models of the involvement of astrocytes in glutamatergic transmission. The goal of this chapter is to present representative models of astrocyte function in conjunction with the biological questions they can investigate. Key words Computational neuroscience, Simulation, Glutamatergic transmission, Calcium signaling, Tripartite synapses

1

Introduction In this chapter, we present diverse computational models of astrocytes on a wide range of spatiotemporal scales. Our goal is to equip the reader with a concise overview of the available models so that they can use them to investigate research questions of interest. Astrocytes are glial cells that are essential to numerous functions of the central nervous system, from brain development, metabolism, and homeostasis to brain injury repair. They interact with numerous cell types simultaneously. Notably, astrocytes communicate with both blood vessels at specialized subcellular compartments—endfeet—and with neurons at tripartite synapses, where the astrocyte is in apposition to presynaptic and postsynaptic elements [1]. This intercellular communication is believed to be

Kerstin Lenk and Audrey Denizot contributed equally to this work. Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5_11, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

229

230

Kerstin Lenk et al.

mediated by astrocyte Ca2+ signals [2]. One of the largest challenges in the field is to unravel the biochemical reactions that underlie astrocyte function [3]. Because of the complex nanoscopic morphology of astrocytes and the various spatiotemporal properties of astrocyte signals, characterizing their involvement in the brain (patho-)physiology is hindered by technical difficulties, such as the resolution of live microscopy [2, 4]. Computational models of astrocytes have been implemented to overcome those limitations and have provided valuable insights into the involvement of astrocytes in synaptic function [5–7]. A computational model is a simplification of a system that describes its elements, their states, and their interactions. Computational models can guide experimentalists toward the most relevant experiments by forming a theoretical framework to characterize and predict the function of the system of interest [5, 8, 9]. Those models can be used to run studies that would be unfeasible or time-consuming experimentally, in a fully controllable manner. Computational models, thus, make it possible to go beyond correlational observations and to propose causal relationships that govern the dynamics of the system of interest. When data is rare or difficult to get experimentally, models can be used to generate vast amounts of synthetic realistic datasets. Biophysical models [10] describe detailed chemical reactions based on experimental data of a biological phenomenon. The advantage of this type of modeling paradigm is that it can provide quantitative, testable predictions. The potential disadvantage of this approach can be its high computational cost, i.e., a large amount of time or memory required to run the simulations. Another type of approach is phenomenological modeling (the FitzHugh-Nagumo model [11] is a good example), which aims at mimicking the phenomenon of interest without describing its biophysical details. Those models are less computationally expensive, allowing simulations of larger systems and/or reduced simulation time. For more details on the insights gained by experimentalists from computational models and vice versa, the reader can refer to [12] and to dedicated books [13, 14]. Several reviews of computational astrocyte models have been published recently that have contributed to a better understanding of astrocyte function in the brain [5–7], and new ones emerge at a fast pace. The review paper by Oschmann et al. [5] summarizes astrocyte models from the subcellular to the network level. The book chapter from Denizot et al. [6] presents the recent developments in astrocyte modeling approaches at the cellular and subcellular levels that accompanied advances in experimental techniques. Manninen et al. [7] provide a historical overview of computational astrocyte models. Lastly, the book “Computational Glioscience” [15] gives detailed insights into various computational models of

Computational Models of Astrocyte-Neuron Interactions

231

Table 1 Websites with the code of available models Section Model

Website

2.1.

De Pitta` et al.

3.1.

Rǎdulescu et al. https://github.com/scimemia/Glutamate-transporters-estimates

4.1.

Østby et al.

https://models.physiomeproject.org/exposure/d9de93b128da322a4d50 f24589980ea1/ostby_oyehaug_einevoll_nagelhus_plahte_zeuthen_ voipio_lloyd_ottersen_omholt_2008.cellml/view

4.3.

Oschmann et al.

(partially available) https://github.com/FranziOschi/AstroMultiComp

6.1.

Savtchenko et al.

https://senselab.med.yale.edu/ModelDB/ShowModel?model=24350 8#tabs-1

6.2.

Cresswell-Clay et al.

https://github.com/FSUcilab/Compartmental_model_astrocytes

6.3.

Denizot et al.

https://senselab.med.yale.edu/ModelDB/ShowModel?model=247694 #tabs-1

7.1.

Verisokin et al.

https://zenodo.org/record/4552726#.Yrr28C8RppR

7.2.

Lenk et al.

https://github.com/kerstinlenk/INEXA_FrontCompNeurosci2020

https://github.com/mdepitta/comp-glia-book/tree/master/Ch5.DePitta

neuron-glia interactions. We believe that robust and insightful collaborative work between theoreticians and experimentalists relies on the implementation of models that are accessible, reusable (FAIR principles: Findable, Accessible, Interoperable, Reusable) [16], as well as reproducible and replicable [8, 17, 18]. In this line, we provide links to the code of each publicly available model (Table 1) and describe the validation of the model against experimental data whenever the latter was performed. In this chapter, rather than presenting an exhaustive list of astrocyte models, we present a few selected models that are most representative of the diversity of the existing models and highlight the type of biological questions they can investigate. The chapter is organized as follows: We present models of glutamatergic transmission at the tripartite synapse (see Subheading 2), the involvement of astrocytes in glutamate uptake (see Subheading 3), ion homeostasis (see Subheading 4), and metabolism (see Subheading 5). Then, we introduce models that study astrocyte structure-function coupling (see Subheading 6) and astrocyte networks (see Subheading 7). We end the chapter with concluding remarks (see Subheading 8) and a list of resources (see Subheading 9).

232

2

Kerstin Lenk et al.

Signal Transmission at Tripartite Synapses Astrocytes can contact pre- and postsynaptic neurons, forming so-called tripartite synapses [1]. The astrocyte subcompartments that communicate with neurons at synapses are often referred to as perisynaptic astrocyte processes (PAPs). Glutamate released by active glutamatergic presynaptic neurons binds to G-protein-coupled receptors at the astrocyte membrane, which triggers a series of chemical reactions that allow the formation of Ca2+ signals in the astrocyte cytosol. Those Ca2+ signals are essential to various brain functions, from synaptogenesis to memory consolidation [19, 20]. They are believed to have a variety of downstream effects that can modulate neurotransmission and synaptic plasticity, such as the release of gliotransmitters (e.g., adenosine triphosphate (ATP), D-Serine, and glutamate) by the astrocyte into the synaptic cleft [21] or changes in extracellular ion concentrations (more on this in Subheading 4). Gliotransmission has long been debated [22, 23] and a growing body of literature supports the existence of such neuron-astrocyte communication [24, 25]. The influence of astrocyte Ca2+ signals and astrocyte-neuron communication on synaptic transmission and plasticity are still poorly understood. Computational models have been developed to study those interactions and to predict the coupling between neuronal and astrocyte activity. In this section, we present three models of signal transmission at tripartite synapses [26–28], highlighting the different approaches and strategies that have been used as well as the different insights that can be gained from those models (Fig. 1). As glutamate uptake and ionic homeostasis are covered in Subheadings 3 and 4, we here focus on models of neuronal activity-induced astrocyte Ca2+ dynamics and gliotransmission. Those models can be used to predict how astrocyte activity can modulate neurotransmission in various physiological conditions.

2.1 Model of Glutamate-Induced IP3 and Ca2+ Oscillations in Astrocytes

In 2009, De Pitta` et al. [27] introduced a model in which the astrocyte receives neuronal glutamate input. Glutamate then binds to the metabotropic glutamate receptors (mGluRs) at the astrocyte membrane. This binding results in an increase in inositol trisphosphate (IP3) concentration in the astrocyte cytosol. Subsequently, Ca2+ is released from the endoplasmic reticulum (ER) through the opening of IP3 receptor channels (IP3Rs), which leads to IP3-dependent Ca2+-induced Ca2+ release (CICR) in the cytosol, whereby an increased Ca2+ concentration enhances the release of Ca2+ from the ER. At high Ca2+ concentrations, the IP3Rs are inactivated and Ca2+ is pumped back into the ER by the sarco-endoplasmic reticulum Ca2+-ATPase (SERCA) pumps. A long-lasting glutamate stimulus leads to high cytosolic IP3 concentrations, allowing for the alternation of activation and inactivation

Computational Models of Astrocyte-Neuron Interactions

233

Tripartite synapse models Glutamate-induced Ca2+ activity Gliotransmission ex: Nadkarni et al. 2008; De Pittà et al. 2009, 2016

Presynaptic neuron Astrocyte Postsynaptic neuron

Action potential [Ca2+ ]pre,st [Ca2+]α

[Ca2+]pre,ap

α mGluR

Glutamate

Ionotropic glutamate receptors

Whole astrocyte model Signal propagation Signal integration ex: Gordleeva et al. 2018

Nadkarni et al. 2008

Ca2+ diffusion IP3 diffusion

Glutamate

mGluR

PLCβ

soma

PLCδ

Ca2+ IP3 IP3R

cytosol ER SERCA

De Pittà et al. 2009, 2016 Fig. 1 Models of signal transmission at tripartite synapses. Neuron-astrocyte communication at synapses has been modeled using different strategies: simulating neuronal activity-induced Ca2+ activity in astrocytes with (e.g., Gordleeva et al. model [26]) or without (e.g., De Pitta` et al. model [27] and Nadkarni et al. model [28]) taking into account the diffusion of molecules between subcellular compartments

of Ca2+ channels and thus Ca2+ oscillations [29]. The kinetics of IP3Rs are described with the Li-Rinzel model [30] (reviews of IP3R models can be found in [31, 32]). Besides the CICR mechanism, IP3 production by the phospholipase C (PLC) isoenzymes PLCβ and PLCδ and IP3 degradation are described by the model. IP3 degradation can occur in two ways, through dephosphorylation by inositol polyphosphate 5-phosphatase and Ca2+-dependent phosphorylation by IP3 3-kinase. A simplified representation of the modeled chemical reactions can be found in Fig. 1. In this paper, the authors measured the variation of the amplitude and frequency of Ca2+ signals depending on the level of synaptic activity, which is modeled as alterations of the glutamate concentration in the extracellular space (ECS).

234

Kerstin Lenk et al.

In summary, the model describes IP3R-dependent Ca2+ signals in the astrocyte that result from neurotransmission at glutamatergic synapses. It has been used in various other models to describe astrocyte activity, e.g., [26, 33–35]. 2.2 Model of Ca2+ Activity in an Astrocyte Connected to a Neuronal Network

To model astrocyte-neuron communication at tripartite synapses, Gordleeva et al. [26] simulated the activity of a single astrocyte in a network of 36 neurons. Their goal was to investigate the effect of the spatial distribution of Ca2+ signals within the astrocyte on gliotransmitter release and the associated modulation of neuronal activity. To do so, the authors combined their compartmental model of an astrocyte [36] with their model of a neuronal network, which accounts for gliotransmitter release by astrocytes [37]. Astrocyte morphology is described as the assembly of cellular subcompartments, with a cylindrical shape, coupled by IP3 and Ca2+ diffusion (deterministic spatially extended approach, see Subheading 6 for more details). The astrocyte contains 14 processes, each connected to a different synapse amongst the 36 neurons of the network. Astrocyte Ca2+ activity in each process is modeled using the IP3R-mediated Ca2+ signaling model from De Pitta` et al. [27] (see Subheading 2.1). In the model from Gordleeva and colleagues, gliotransmitter release is a function of cytosolic Ca2+ concentration in distal processes and only occurs when local Ca2+ concentration exceeds a given threshold. The neurons are modeled using a conductance-based mathematical formalism, the Hodgkin-Huxley model [38]. The connectivity of the neural network is randomly chosen, with a 20% probability of connectivity for each pair of neurons. Each spike in a presynaptic neuron results in the release of glutamate at synapses, modeled by a Poisson process of fixed frequency. This model allows one to simultaneously monitor Ca2+ signals in different astrocyte compartments, gliotransmitter release at synapses, and postsynaptic neuronal firing. The modeled geometry and reactions are summarized in Fig. 1. To investigate the connectivity between active neurons and astrocytes, Gordleeva et al. [26] have calculated the crosscorrelation between neuronal firing rate and the so-called astrocyte firing rate, corresponding to the frequency of the number of Ca2+ signals in the whole astrocyte. They found a synchronization of the activity of presynaptic neurons and astrocytes with a delay of roughly 2 s. They further analyzed the integration of Ca2+ signals within the astrocyte and predicted that distal processes were the most active subcompartments of astrocytes, occasionally allowing somatic events to occur, the latter backpropagating to all processes. The model was able to predict various effects of astrocyte gliotransmitter release on neuronal activity: glutamate-mediated presynaptic potentiation, inhibition of presynaptic release, and D-serinemediated increase in the postsynaptic current. The authors also investigated the effect of local Ca2+ signals in a PAP contacting an

Computational Models of Astrocyte-Neuron Interactions

235

active synapse on neighboring processes. They found that Ca2+ and IP3 diffusion within the astrocyte can activate the release of gliotransmitters in neighboring processes, resulting in the potentiation or depression of nearby inactive synapses. Similarly, whole-cell Ca2+ events favored the release of gliotransmitters from various processes and thus coordinated the activity of the neural circuits connected to the active astrocyte. Overall, the model from Gordleeva et al. [26] accounts for the spatial morphology of a single astrocyte and its connectivity to various neurons. It is thus well-suited to study the modulation of neuronal network activity mediated by the spatiotemporal integration of Ca2+ signals within a single astrocyte. Please refer to Subheading 7 for network models involving numerous astrocytes. 2.3 Model of Signal Transmission at a Glutamatergic Tripartite Synapse

Nadkarni et al. [28] modeled the closed-loop modulation of synaptic transmission at a tripartite synapse in the hippocampus (Fig. 1). The model is based on the Bertram et al. phenomenological model to describe action potential-driven vesicle release from the presynaptic neuron [39]. Both asynchronous and action potential-independent neurotransmitter release rates are derived from experimental data [40]. This release is followed by a refractory period [41]. The vesicle recycling rates and short-term plasticity are determined by the Tsodyks et al. model [42]. Thus, the Nadkarni et al. model considers both the availability and recovery of neurotransmitter resources. Furthermore, the model describes the Ca2+ activity of an astrocyte, the resulting release of gliotransmitters (here glutamate), and its effect on synaptic transmission. If a vesicle release event takes place, the glutamate in the cleft binds to the postsynaptic receptors and the mGluRs on the membrane of the astrocyte process. The former causes an excitatory postsynaptic current, while the latter results in the production of IP3, causing the release of Ca2+ from the astrocyte ER (modeled as per Nadkarni et al. [43]). The time course of Ca2+ signals in the astrocyte in the model is in qualitative agreement with experimental data from rat hippocampal and visual cortex slices [44]. If the cytosolic Ca2+ levels in the astrocyte are above a threshold, glutamate is released from the astrocyte [45, 46], which leads to the potentiation of synaptic transmission [40, 47, 48]. Since the precise mechanisms of gliotransmitter release by the astrocyte were unknown, the authors modeled gliotransmission with a slow decay that mirrors the timescale of potentiation mediated by astrocytes [48]. In their study, Nadkarni et al. propose that Ca2+ release from the astrocyte leads to a potentiation of neurotransmitter release that can last for minutes. In support of the proposed mechanism, the time course of the ER-mediated Ca2+ signal correlates well with the observed time course of changes in synaptic transmission. Since these Ca2+ fluxes are not temporarily correlated to the action potential-mediated Ca2+ activity, it causes an increase in

236

Kerstin Lenk et al.

asynchronous release, i.e., in action potential-independent neurotransmitter release, which depletes the vesicle resource. The strength of the coupling between the astrocyte Ca2+ and the presynaptic Ca2+ was investigated as an open parameter, “α”. An increase in the value of “α” thus leads to an enhanced vesicle release rate. The authors found a value of “α” that resulted in a neurotransmitter release rate that was concordant with experimental data [47]. Interestingly, it is the value of “α” and the corresponding increase in release probability that seemed to maximize synaptic transmission. The model was robust with respect to a wide range of stimulus frequencies, the number of active zones, and basal levels of vesicular release probability. In summary, the model of Nadkarni et al. describes neuronastrocyte coupling at tripartite synapses and can be used to investigate the complex relationship between astrocyte activity, presynaptic vesicular release rate, and vesicle depletion. It can be used to predict the modulation of neurotransmitter release rate by astrocytes under a range of stimulus protocols and can be extended to calculate downstream changes in plasticity. 2.4

3

Discussion

Numerous models have been implemented to study the roles of astrocytes in synaptic transmission. In those models, astrocytes mainly respond to neuronal activity through the mGluR-dependent activation of IP3R channels that influence Ca2+ dynamics. De Pitta` et al. [27] modeled astrocyte IP3 and Ca2+ oscillations mediated by glutamatergic transmission; Nadkarni et al. [28] described the coupling of neuronal and astrocyte activity in a single tripartite synapse; while Gordleeva et al. [26] depicted the spatiotemporal integration of astrocyte Ca2+ signals in a whole cell, thus predicting its influence on synaptic transmission at various synapses. All three models (Fig. 1) describe IP3R channel dynamics using early computational models [30, 49] that are based on electrophysiological data such as Bezprozvanny et al. [50].

Glutamate Uptake As much as glutamate release is essential for excitatory transmission in the central nervous system (CNS), its rapid removal from the ECS is critical for normal brain function. Glutamate molecules that linger on and diffuse away from the synaptic cleft can compromise the specificity of synaptic signaling [51], a key component of information processing in the brain. A prolonged lifespan of glutamate can also cause neuronal cell death through a phenomenon referred to as excitotoxicity [52]. About 90% of all the released glutamate is taken up by astrocytes, making them the primary cells responsible for glutamate clearance [53–55]. This phenomenon is mediated by its uptake by

Computational Models of Astrocyte-Neuron Interactions

237

glutamate transporters, which are expressed in all cell types in the CNS, with the highest density found in astrocytes [56]. Glutamate transporters can be classified as Na+-independent and Na+-dependent transporters [56, 57]. Even though the affinity of Na+-independent transporters is similar to that of Na+-dependent ones, they contribute to less than 5% of the total glutamate uptake [56]. Na+dependent glutamate transporters are also referred to as excitatory amino acid transporters (EAATs) and consist of several isoforms, e.g., EAAT-1 (also known as glutamate-aspartate transporter or GLAST in rodents) and EAAT-2 (known as glutamate transporter-1 or GLT-1 in rodents). The time course of glutamate uptake can be calculated based on a deconvolution analysis of astrocyte transporter currents such as done by Scimemi and Diamond [58]. In this section, we present three models of astrocyte glutamate uptake [59–61] to study its effect on astrocyte Na+ and Ca2+ dynamics and postsynaptic α-amino-3-hydroxy-5-methyl-4isoxazole propionic acid receptors (AMPAR) and NMDAR activation (Fig. 2). 3.1 Model of Glutamate Uptake in Relation to Glutamate Transporter Density

A recent model proposed by Rǎdulescu et al. [59] investigates the effect of the density of glutamate transporters on the membrane of PAPs on the opening probability of AMPARs and NMDARs at the active (Po-AMPAperi and Po-NMDAperi) and neighboring synapses (Po-AMPAextra and Po-NMDAextra). To do so, they performed Monte Carlo simulations (particle-based approach, see Subheading 6 for more details), in which glutamate molecules were injected at the center of a synaptic cleft and the glutamate concentration profile was monitored at the stimulated synapse as well as at six neighboring synapses (Fig. 2). Glutamate transporter density in astrocytes at different ages in mice (2 weeks to 21 months) was estimated based on dot blot experiments. Combining those values with 3D axial STEM tomography reconstructions allowed the authors to evaluate the variability of glutamate transporter density in different astrocyte subcellular compartments. Simulations with the model allowed the authors to infer the effect of the variability of glutamate transporter density on synaptic transmission depending on mouse age and the subcellular compartment in contact with the synapse. Results indicate that the density of glutamate transporters on PAPs as well as the location of the site of glutamate release play an important role in shaping glutamate receptor activity at local (Po-AMPAperi and Po-NMDAperi) and distant (Po-AMPAextra and Po-NMDAextra) synapses. More specifically, the higher the glutamate transporter density around a synapse, the lower the extra-synaptic activation of glutamate transporters.

238

Kerstin Lenk et al. Tripartite synapse models Glutamate uptake and gliotransmission ex: Flanagan et al. (2018)

Glutamate uptake and volume transmission ex: Rǎdulescu et al. (2022)

Postsynaptic neuron

Presynaptic neuron

Glutamate transporter

NMDAR Glutamate

AMPAR Po - NMDA extra

Glutamate

Po - AMPA extra K+ Glutamate

Na+ Na + /K+ ATPase

mGluR K+

IP3 ER

H+

Ca 2+

K+ EAAT2

Glu

Na+ Ca 2+

Na+

Presynaptic neuron

NMDAR

AMPAR

NCX

Astrocyte

Astrocyte PAP model Glutamate uptake and microdomain activity ex: Héja & Kardos (2020)

Glutamate diffusion Glutamate EAAT mGluR

Postsynaptic neuron

NCX

Ca 2+ diffusion

Bouton

Na+ diffusion Spine

Astrocyte process

Tightly vs. loosely wrapped synapse

Fig. 2 Models of glutamate uptake at tripartite synapses. Flanagan et al. [60] and Rǎdulescu et al. [59] model the interplay between glutamate uptake by astrocytes at tripartite synapses and postsynaptic activity. Additionally, Flanagan et al. describe how those interactions can affect gliotransmission. Rǎdulescu et al. instead simulate the influence of glutamate uptake on extra-synaptic volume transmission. The He´ja and Kardos model [61] describes how glutamate uptake alters astrocyte microdomain activity at the single PAP level

In summary, this model allowed the authors to test the effect of the differential expression levels of glutamate transporters that they observed in different astrocyte compartments on the local and extra-synaptic activation of NMDARs and AMPARs. Their study also highlights the importance of accurate estimates of molecular expression levels to fully grasp the spatiotemporal dynamics of glutamatergic signaling. This model can be used to gain insights into glutamate concentration profiles in the ECS at tripartite synapses, glutamate receptor activity, and glutamate spillover under a range of stimulation protocols.

Computational Models of Astrocyte-Neuron Interactions

239

3.2 Model of Glutamate Uptake by Astrocytes and Its Effects on Postsynaptic Neuronal Excitability

Flanagan et al. [60] extended the tripartite synapse model by De Pitta` and Brunel [62] by adding a biophysical model of EAAT2 transporter activity at the astrocyte membrane (Fig. 2). Thus, they combined in their model glutamate uptake and gliotransmission. Additionally, the model also accounts for Na+ (as EAAT-2 is a Na+-dependent glutamate uptake transporter) and K+ dynamics. The model includes five compartments: the presynaptic and postsynaptic terminals, the soma and process of an astrocyte, and the ECS. The phenomenological presynaptic terminal releases glutamate at a given rate, and spikes trigger a K+ efflux from the presynaptic neuron into the synaptic cleft. The postsynaptic terminal is populated with AMPARs and NMDARs. The binding of glutamate to mGluRs on the membrane of astrocytes results in the activation of IP3Rs on the ER and leads to Ca2+ release into the cytosol (see Subheading 2). This Ca2+ signal can cause a release of gliotransmitter from the perisynaptic compartment of the astrocyte. Moreover, glutamate is taken up by EAAT-2, which is accompanied by a Na+ influx into the astrocyte cytosol. The Na+/K+-ATPase pump is primarily responsible for maintaining Na+ and K+ gradients across the cell membrane; while the Na+-Ca2+ exchanger (NCX) removes Ca2+ from the cell and allows Na+ influx into the cell. Flanagan et al. simulated three basal glutamate concentrations in the astrocyte. The highest of those concentrations (10 mM), nonphysiological and chosen to simulate pathological conditions, led to a delayed removal of glutamate by EAAT-2 from the synaptic cleft. They hypothesize that this slow removal also occurs in the epileptic brain. Due to the prolonged glutamate uptake, the activation of mGluRs was higher, which led to larger IP3-mediated Ca2+ oscillations, allowing intracellular Ca2+ concentration to rise above the threshold for gliotransmission. This increased glutamate release by the astrocyte triggered a slow inward current, which resulted in high-frequency neuronal activity. Interestingly, such a high intracellular glutamate concentration in the astrocyte reduced the minimum value of the neuronal firing rate that could trigger gliotransmitter release events. In summary, the model allows inferring the influence of glutamate uptake and astrocyte intracellular glutamate concentration on gliotransmission and the subsequent alterations of postsynaptic firing rates. It is well suited to study the interplay between astrocyte activity and the excessive glutamate concentrations measured in the ECS in pathological conditions, such as epilepsy.

3.3 A Spatial Model of Glutamate Uptake and Ca2+ and Na+ Signaling in PAP Microdomains

Recently, He´ja and Kardos [61] have developed a model at the nanoscale level to investigate how glutamate uptake as well as Na+ and Ca2+ activity in the astrocyte cytosol are affected by the coverage of the synapse by the astrocyte leaflet. Due to the small size of PAPs, the model is both stochastic and spatially extended (see Subheading 6 for more details). Briefly, the model describes

240

Kerstin Lenk et al.

glutamate diffusion within the synaptic cleft as well as the diffusion of Na+ and Ca2+ within the astrocyte cytosol. The geometry of the synapse is simplified: the presynaptic bouton and the postsynaptic spine are cylinder-shaped while the astrocyte process consists of a hollow cylinder. Astrocyte process geometries of different sizes are used, enabling to test the effect of the astrocyte coverage of the synapse—from loose to tight—on glutamate uptake (see Fig. 2). Neuronal activity is modeled as a punctual infusion of 5000 glutamate molecules at the center of the presynaptic bouton. The model describes the glutamate uptake by EAATs at the plasma membrane of the astrocyte process, which is accompanied by an influx of Na+ within the astrocyte cytosol. EAATs can interact with any glutamate molecule in their vicinity, here described as a 50 × 50 × 50 nm3 interaction space. The model also takes into account the activity of the NCX, which is the only source of Ca2+ influx and efflux in this model. The model by He´ja and Kardos has contributed to a better understanding of the dynamics of glutamate uptake at individual synapses. They distinguished a subpopulation of EAATs on the membrane of astrocytes in the tightly wrapped tripartite synapse configuration that was exposed to high glutamate concentrations and was responsible for the majority of glutamate uptake. Interestingly, this subpopulation was absent in the loosely wrapped synapse configuration. They were further able to characterize the fluctuations of Ca2+ concentration resulting from Ca2+ binding and unbinding to and from NCX and showed that those fluctuations were remarkably stable, displaying little variability upon changes of Na+ and Ca2+ concentrations, neuronal firing rate, NCX activity, or membrane potential. Overall, the model from He´ja and Kardos is an example of a model focusing on astrocyte-neuron communication at the singleastrocyte process level. Such a model is well suited to investigate the effect of spatial properties, such as the shape of the astrocyte leaflet (see also Subheading 6), on astrocyte Ca2+ microdomain activity, and glutamate uptake. 3.4

Discussion

Over the last few decades, experimental and computational studies have characterized the biophysical properties and expression levels of astrocyte glutamate transporters in various (patho-)physiological conditions, in the mature and developing central nervous system. The computational models presented in this section can be used to further our understanding of how astrocyte transporters corral glutamatergic transmission (Flanagan et al. [60] and He´ja and Kardos [61] models) and limit glutamate spillover (Rǎdulescu et al. model [59]).

Computational Models of Astrocyte-Neuron Interactions

4

241

Ion Homeostasis Astrocytes express numerous ion transporters, pumps, exchangers, channels, and receptors that regulate ion homeostasis and the ECS volume [3]. Here, we describe three computational models of astrocyte ion homeostasis that provide key insights into the complex interactions between ion fluxes, ECS shrinkage, glutamate uptake, gliotransmission, and/or astrocyte Ca2+ signaling (Fig. 3). The most common ion pumps and transporters described in these models [63, 64, 66] are (i) Na+/K+-ATPase pumps, which actively exchange Na+ (outward) and K+ (inward) across the plasma membrane; (ii) NCX exchangers, which exchange Na+ (outward) and Ca2+ (inward)—NCX can switch to reverse mode with high Østby et al. (2009) → Regulation of ionic and water fluxes in astrocytes

Presynaptic neuron

ECS

Astrocyte NaK-ATPase

[Na+ ]o [K+ ]o [Cl -]

Astrocyte

o

[HCO3-]o

Postsynaptic neuron

[Na+ ]i [K+ ]i [Cl -]i [HCO3-]i

Voltage-gated ion channels AQP4

NKCC1

KCC1

NBC

Breslin et al. (2018) and Wade et al. (2019) → Interplay between ECS and astrocytic ionic fluxes at tripartite synapses

Oschmann et al. (2017) → Glutamate transporter-dependend ionic fluxes and geometry-dependent ER ratio

Global ECS

Endoplasmic reticulum

[Ca2+ ]

IP3R SERCA CERleak mGluR

INaECSL

Synapse NaK-ATPase KNEU

ER

IKECSL Intracellular space

[Ca2+ ]I [K+ ]I [Na+ ]I [IP3 ]I

NaB KB NaNEU

Glut. transp. NCX NaK-ATPase Kleak Naleak

Perisynaptic ECS [Na+ ]o +

[K ]o [Ca2+ ]o

Astrocyte NaK-ATPase KB [Na+]i KKIR [K+] i EAAT1/2 NaB NCX

[Ca2+ ]i

Perisynaptic cradle

INaPF IKPF ICaPF

Process

* Wade et al. (2019) extensions

mGluR-dependent pathway Extracellular space

Glutamate transporter-dependent pathway glu

[Ca2+ ]

O

[K+ ]

O

[Na+ ]

I : currents : flux direction

O

Fig. 3 Models of ion homeostasis at tripartite synapses. Østby et al. [63] described the ECS shrinkage resulting from astrocyte ionic fluxes. The Breslin et al. model [64] comprises a synapse (describing both pre- and postsynaptic neurons) encapsulated by an astrocyte process. Wade et al. [65] extended the model by adding Ca2+ and Na+ dynamics. The Oschmann et al. model [66] describes the interplay between mGluR-dependent and glutamate transporter-dependent Ca2+ signaling pathways in the astrocyte. The colored arrowheads describe the direction of Na+ (blue), K+ (light red), Ca2+ (dark red), Cl- (green), and HCO3- (orange) fluxes

242

Kerstin Lenk et al.

intracellular Na+ concentrations; and (iii) EAAT-1/2, which exchange Na+ and glutamate (both inward) with K+ (outward). Each astrocyte subcompartment, from the soma to PAPs and endfeet, displays different levels of expression of transporters, pumps, exchangers, channels, and receptors [3, 59]. Hence, the molecular processes underlying ion homeostasis might vary across the astrocyte. This is particularly difficult to examine experimentally as astrocyte fine processes, which account for about 75% of the astrocyte volume [67], are not resolved by diffraction-limited light microscopy. 4.1 Model of Astrocyte Ion FluxesMediated ECS Shrinkage

Østby et al. [63] developed a computational model to study the interplay between K+ buffering and water transport mechanisms. Ionic transport across the membrane change the relative amount of positive and negative charges in the intracellular space (ICS) and the ECS. These fluxes of charges modulate the intracellular osmolarity that drives water fluxes at the plasma membrane. The model accounts for the variations in Na+, K+, and bicarbonate ion (HCO3-) concentrations in the astrocyte and the ECS. Those ion concentration changes are described, together with the volumetric changes of the astrocyte and the ECS, using ordinary differential equations (ODEs), while algebraic equations depict Cl- dynamics and astrocyte membrane potential. The authors implemented the model to investigate stimulation-induced shrinkage of the ECS in the gray and white matter. When active, neurons release K+ and uptake Na+, while astrocytes uptake K+, Na+, and Cl-, which results in a water influx from the ECS into the cell by osmolarity. Figure 3 displays the modeled reactions. The simulation results indicated that volume changes are controlled by the combined action of several processes. The ECS shrinkage seemed to be enhanced by the cotransporters (i.e., the Na+-bicarbonate cotransporter, NBC, and the Na+-K+-Cl- transporter, NKCC1). The rise in the extracellular K+ concentration following neuronal activation causes an astrocyte membrane depolarization, which is sensed by NKCC1 and causes an increase in the influx of both Na+ and HCO3-. Their results further suggested that the activity of the Na+/K+-ATPase limited the ECS shrinkage by keeping the intracellular Na+ concentration low, notably in the presence of an activity-induced increase of Na+ influx. The low intracellular Na+ concentration prevents intracellular osmolarity from reaching high levels, which in turn limits water influx from the ECS into the astrocyte. In summary, the Østby et al. model describes ionic fluxes in the astrocyte and how they can impact the ECS volume during glutamatergic transmission. In several diseases, such as cortical spreading depression and epilepsy, the ECS and astrocyte volume is altered [68]. This model is best suited to study the impact of ionic fluxes in astrocytes on water and ionic homeostasis at synapses.

Computational Models of Astrocyte-Neuron Interactions

4.2 Model of Potassium and Sodium Microdomains in Astrocytes

243

Experimental studies showed the presence of ionic microdomains in thin astrocyte processes [65], which correspond to small portions of the plasma membrane with inhomogeneous distributions of Na+ channels and cotransporters, forming clusters. This spatial organization could result from spatially restricted areas with negatively charged membrane lipids [69]. Breslin et al. [64] hypothesized that this localized negative charge might result in a slow diffusion of cations (positively charged ions) along the astrocyte processes. These localized negative charges create potential wells, characterized by a local minimum of potential energy. Potential wells restrict cation conduction, attracting and trapping the positive charges in the wells created by the negative charge. To investigate their hypothesis, the authors proposed a multicompartmental model of a synapse enwrapped by an astrocyte to explore the ion homeostasis in thin astrocyte processes and the interplay between the astrocyte and neuronal compartments (Fig. 3). With the present model, the authors showed that the cation flow restriction forms a K+ microdomain at the PAP, referred to as the perisynaptic cradle (PsC) in this study. Moreover, they showed that K+ microdomains decrease the electrochemical gradient of K+ and reduce the influx of K+ through inward-rectifier K+ channels (Kir), facilitating the return to basal concentrations of K+ in the perisynaptic ECS. Similar microdomains were observed for Na+. To further investigate the effect of such microdomains, Wade et al. [65] extended the Breslin et al. model by adding Ca2+ channels onto the plasma membrane (Fig. 3). Note that intracellular sources of Ca2+ were omitted. The authors tested the hypothesis that Ca2+ microdomains can be formed in the PsC and that those depend on Na+ microdomains. Na+ microdomains reversed the NCX, instigating an influx of Ca2+ into the astrocyte. Ca2+ microdomains in this case were not formed by potential wells but by the reverse mode of the NCX. Since Na+ influx through EAAT-2channels depends on presynaptic glutamate release, the Wade et al. model allows studying the effects of sustained neuronal activity on the intra- and extracellular ionic concentrations. The formation of Na+ and Ca2+ microdomains was itself sufficient to produce Ca2+ transients, even in the absence of intracellular Ca2+ stores. In summary, Breslin et al. and Wade et al. were the first to hypothesize and simulate the formation of Na+ and K+ microdomains in PAPs and to test their effect on Ca2+ microdomain activity. These models could help to study the effect of the modulation of the volume of astrocyte subcompartments on ionic microdomain formation and local ionic fluxes.

244

Kerstin Lenk et al.

4.3 Model of Ca2+ Dynamics Mediated by Two Different Spatially Segregated Pathways

While many computational models of astrocytes focus on one Ca2+ pathway, Oschmann et al. [66] examined the interactions between (i) glutamate-induced Ca2+ signals and (ii) glutamate uptake. In this model, it is assumed that the activity of glutamate transporters indirectly activates intracellular Ca2+ influx through the activity of NCX, while Na2+ and K+ concentration gradients across the plasma membrane are maintained by the activity of the Na+/K+-ATPase. The model consists of a system of ODEs describing a single compartment of either an astrocyte process or soma divided into three compartments: a cylindrical ICS, a cylindrical ER (internal Ca2+ store) within the ICS cylinder, and a cylindrical ECS, which has the same volume as the ICS (Fig. 3). The model describes the dynamics of the astrocyte membrane potential, intracellular and extracellular ion concentrations (Ca2+, Na+, and K+), and intracellular IP3 concentration. Diffusion is not described as the model is not spatialized (see Subheading 6). On one hand, this model assumes, based on previous experimental studies [70, 71], that somatic Ca2+ signals mostly result from mGluR activity-dependent Ca2+ influx. The soma is characterized by a low-surface-volume ratio and a high-ER-ICS-volume ratio (ratioER). On the other hand, Ca2+ signals resulting from the activity of glutamate transporters are assumed to mostly occur near synapses, in PAPs, whose surface-volume ratio is high and ratioER low. The authors used this model to investigate whether the activity of NCX and glutamate transporters can trigger Ca2+ signals in PAPs. Intracellular Ca2+ concentration in the PAPs only increased when the parameter value for the maximal pump rate of the NCX was increased. Blocking glutamate uptake by the astrocyte prevented Ca2+ influx through the NCX. Ziemens et al. [72] used the equation describing the NCX current from the Oschmann et al. model to predict that the increased Na+ activity in PAPs measured experimentally upon NMDA application triggers NCX-dependent Ca2+ influx (reverse mode) in PAPs. In summary, the novelty of the Oschmann et al. model lies in the spatial separation within the astrocyte of mGluR- and glutamate transporter-dependent Ca2+ signaling pathways. Therefore, the model is best suited for computational studies investigating the distinct Ca2+ activity in the soma and PAPs.

4.4

The maintenance of ion homeostasis is critical to ensure the propagation of action potentials in neurons and to prevent excitotoxicity. Several computational models have been developed to study ion homeostasis at tripartite glutamatergic synapses. For example, the model from Østby et al. can be used to study the interplay between ECS shrinkage, ion uptake, and water transport [63]. The Breslin et al. [64] and Oschmann et al. models [66] allow for studying the involvement of astrocytes in ion homeostasis and glutamate uptake.

Discussion

Computational Models of Astrocyte-Neuron Interactions

245

The Oschmann et al. model describes the interplay between two different Ca2+ signaling pathways, while the Breslin et al. model allows for studying ionic microdomains in astrocyte leaflets and their effect on synaptic homeostasis. Altogether, these models can provide novel insights into the mechanisms by which astrocytes contribute to the regulation of ion homeostasis in the brain.

5

Metabolism The idea that metabolic interactions occur between astrocytes and neurons has existed for more than a century now [3]. It is based on the observation that astrocyte processes are intimately juxtaposed to brain capillaries as well as neuronal synapses [73]. Astrocytes are involved in the uptake of glucose—the main energy source of the brain—from the blood and distribute it to other brain cells [73]. However, the detailed involvement of astrocytes in the metabolic processing of glucose remain unclear and controversial [74]. The major impediment in achieving a clear understanding of this phenomenon has been the subcellular resolution required to monitor metabolic fluxes during neuronal activity in the brain. This has been partially overcome by using more accessible experimental model systems like the retina, co-cultures of neurons and glial cells, as well as mathematical modeling [75]. Despite technological advances, the exact role of astrocytes and lactate in brain energy metabolism is still unresolved. Lactate is an alternative energy source to glucose. In the brain, lactate is produced by both astrocytes and neurons, which convert glucose into lactate through a process called aerobic glycolysis. Most investigations agree that lactate is transferred between astrocytes and neurons but disagree on the direction of this transfer: (1) astrocyte-to-neuron (ANLS) [76] or (2) neuron-to-astrocyte lactate shuttle (NALS) [77]. Most of the existing computational models of astrocyte metabolism are based on the biophysical models proposed either by Aubert et al. [78] or Simpson et al. [79]. Both papers model the transport and processing of metabolites as a series of coupled differential equations that aim to explain experimental data. The second-generation models based on either of these models include recent insights and refined parameters (Fig. 4) [77, 80]. Both models predict accurate glucose and lactate transients. However, despite a similar framework, they predict the opposite outcomes in the direction of lactate transfer. The differences seem to arise primarily from the way the models describe fluxes that are associated with (i) the uptake of glucose by the astrocyte compartment from the basal lamina, (ii) the uptake of glucose by the astrocyte compartment from the interstitium, and (iii) the uptake of glucose by the neuronal compartment.

246

Kerstin Lenk et al.

glucose glucose NADH

pyruvate ADP

GLUT3

Neuron

Astrocyte NAD+

NAD+ LDH1

lactate

MCT2

MCT1,4

lactate

LDH5

pyruvate

glucose

ADP ATP

CO2

glutamate

Capillary

GLUT1 GLUT1

NADH

O2

ATP

glucose

GLUT1

GLS

glutamine

glutamine GS

glutamate

glutamate

3Na+

GluR

Na+/K+ ATPase 2K+

3Na+

Fig. 4 Pathways described by models of astrocyte-neuron metabolic interactions. Released glutamate from the presynaptic bouton can activate glutamate receptors (GluRs) at the astrocyte and postsynaptic membranes, which is an energetically expensive process. Following the uptake of glutamate by astrocytes, glutamate is converted into glutamine and transported to neurons, where it is converted back into glutamate by glutaminases (GLSs). In neurons, lactate can be used as an energy substrate following its conversion to pyruvate by the lactate dehydrogenase-1 (LDH1). Astrocytes and neurons take up glucose via GLUT1 (glucose transporter 1) and GLUT3 (glucose transporter 3), respectively 5.1 Top-Down Model of the Compartmentalization of Metabolic Pathways at Tripartite Synapses

Previous models that have attempted to resolve the ANLS versus NALS debate have relied on a bottom-up approach, wherein the energy needs of each of the biophysical processes involved in signaling were accounted for to calculate the total energy needs [79]. In contrast, the approach by Jolivet et al. [76] is a top-down approach that focuses on the energy that is available rather than required to understand the compartmentalization of different biochemical reactions involved in metabolic activity in neurons versus astrocytes. To that end, they used published datasets that describe the average tissue glucose and oxygen utilization at resting and active brain states [81]. They measured the linear relationship between the total cycling of neurotransmitters and the neuronal oxidative glucose utilization [82]. This allowed them to calculate the average tissue ATP production at rest and in the active state. Jolivet et al. then described the compartmentalization of neuronal and astrocyte oxygen and glucose metabolism (Fig. 4) and used this information to investigate whether glucose is completely oxidized by these cells (based on the calculations by Gjedde et al. [81]). This method allows for a quantification of the energy budget of the brain

Computational Models of Astrocyte-Neuron Interactions

247

constrained by in vitro experimental data and does not have to make any significant assumptions on parameter values. Their results suggest that a larger majority of glucose is taken up by astrocytes, while oxygen is mostly consumed by the neurons, and this consumption is correlated with neuronal activity. Although the model did not include glycogen, it was able to predict a wide range of in vivo data from the human brain. Their key finding that addresses the ANLS/NALS controversy is that astrocytes only oxidize a small portion of the glucose while neurons oxidize glucose-derived metabolites, which strongly supports the ANLS hypothesis since this results in the transfer of glucose-derived metabolites from astrocytes to neurons. The amplitude of this transfer goes up with increased neuronal activity. In summary, the model quantifies the partitioning of the distribution of energy utilization, notably oxygen and glucose, by neurons and astrocytes. It provides a mathematical description of the neurovascular coupling at different spatial and temporal scales, describing the metabolic activity of neurons and astrocytes as well as the BOLD signal. The model can be used to predict the temporal dynamics of the consumption of lactate, glucose, and oxygen by the brain tissue. 5.2 Model of Lactate and Glucose Levels in Neurons and Astrocytes During Visual Stimulation

The study by Mangia et al. [77] describes a mathematical model of NALS based on the Simpson et al. model [79] to gain insights into the compartmentalization of the metabolic activity of different brain cells. The model was implemented using in vivo data from the human brain, notably magnetic resonance spectroscopy, which quantified temporal changes in metabolite concentration during neuronal activity. The model simulates brain glucose and lactate levels in astrocytes and neurons (Fig. 4), based on concentrations and kinetic rates measured experimentally. Parameters that govern the utilization of glucose and lactate by astrocytes and neurons were investigated over a wide range of values. Their results suggest that physiological parameter values predict NALS. Mangia et al. further demonstrate that ANLS is only possible under unrealistic conditions, where astrocytes display a twelve-times increased capacity for glucose transport and neurons do not respond to activation with increased glycolysis. In summary, the main difference between Jolivet et al. [76] and Mangia et al. [77] as well as the rest of the ANLS and NALS models, seems to stem from the parameter values used to describe the amount of glucose that is entering the astrocyte compared to neurons. The NALS models keep the proportion of astrocyte glucose transport at around 20% [77], whereas the value of this parameter is more than 50% in ANLS models [83, 84]. Predictions from the Jolivet et al. and Mangia et al. models may need to be further tested in models of metabolic disorders or ischemic stroke to resolve the debate.

248

5.3

6

Kerstin Lenk et al.

Discussion

Both ANLS and NALS models agree that glucose is partially transported into astrocytes from the blood [85]. The debate is about the proportion of this astrocyte glucose transport. The supporters of the ANLS models promote the idea that there is a shift in glucose utilization from neurons to astrocytes during glutamatergic activity. However, the modeling studies that support the NALS mechanism [77, 86] suggest that glucose and not lactate is the main energy substrate in the brain. This conclusion is based on the theoretical prediction that glucose transport capacity is larger in neurons than in astrocytes. The supporters of ANLS argue that this thesis is not consistent with the absence of a pathological phenotype in transgenic mice with decreased expression of the neuron-specific glucose transporter GLUT3 [87]. On the contrary, a decrease in the expression of GLUT1, the astrocyte-specific glucose transporter, leads to pathological conditions [88]. The strongest argument against the NALS model is the observation that the glucose utilization rate in neurons does not seem to vary depending on activity levels [89]. The NALS versus ANLS model debate thus remains unresolved to date.

Structure-Function Coupling Astrocytes display a very complex nanoscopic morphology. Around 75% of the total astrocyte volume consists of a meshwork of fine processes that are below the diffraction limit, thus unresolved by diffraction-limited light microscopy [90]. Such a complex cellular nano-architecture has been shown to greatly impact the function of various cell types. For example, the shape of dendritic spines controls the local sequestration of signals and thus strongly shapes synaptic function [91–93]. As modifying cell morphology without altering cell physiology is unfeasible experimentally, mathematical and computational modeling approaches are essential to investigate geometrical effects on cell signaling. Consequently, an increasing amount of astrocyte models describe and account for cell morphology and spatial effects. In this section, we present three different computational approaches and three models that can be used to study the effect of cell shape on astrocyte physiology at the singlecell level (Fig. 5). Please refer to Subheading 7 for models taking into account the topology of astrocyte networks. Before describing the three models, we provide a brief overview of modeling approaches taking into account the effects of cell geometry, referred to as spatially extended approaches: • Deterministic spatially extended approaches describe the average behavior of populations of molecules within compartments. They are often referred to as compartmental models. Reactions within each compartment are described by ordinary differential equations.

Computational Models of Astrocyte-Neuron Interactions

249

5 μm

ASTROCYTE COMPARTMENT Major branch

Unresolved fine process

Cell body

Compartmental models ex: Savtchenko et al. 2018 Cresswell-Clay et al. 2018 Neuronal stim.-induced Ca 2+ flux

Stochastic spatial models ex: Héja et al. 2020 Denizot et al. 2019 Cytosol diffusion

200 nm

ATPase Ca2+

soma IP3R

Astrocytic Ca 2+ signals model from Cresswell-Clay et al. 2018

INSIGHTS (examples)

Inter-compartment coupling Signal propagation

PLC Ca2+

cytosol ER SERCA

Cytosol ER

IP3

Ø

cytosol ER IP3R

ER dffusion

Astrocytic process Ca 2+ model from Denizot et al. 2019

Effect of nanoscale morphology Microdomain activity

Fig. 5 Modeling strategies to investigate the effects of geometry on astrocyte activity depending on the compartment under study. This schematic illustrates the main cellular subcompartments that characterize astrocyte morphology as well as examples of modeling techniques used depending on the compartment being modeled. Compartmental models such as implemented by Cresswell-Clay et al. [94] and Savtchenko et al. [95] are best suited to model Ca2+ signal propagation and coupling within a whole astrocyte, while spatial models of fine processes such as developed by He´ja et al. [61] and Denizot et al. [96] make it possible to study the effect of spatial factors such as cell morphology and Ca2+ channels distribution on local microdomain activity

• Stochastic spatially extended approaches consider that molecular interactions are probabilistic events. There are two main stochastic approaches, described below. For more details, see [97]. 1. Particle-based approaches, also referred to as particletracking or microscopic models, describe the position and state of all the molecules being modeled. Diffusing molecules are then tracked individually during simulation time. 2. Voxel-based approaches, also referred to as populationbased or mesoscopic, divide the modeled space into small compartments: voxels, most often cubes, or tetrahedra. Each compartment is considered well-mixed and diffusion events are described as modifications of the number of molecules in two adjacent compartments.

250

Kerstin Lenk et al.

• Hybrid approaches divide the system of interest into subcompartments, each describe by a different spatially extended approach. For more details on spatially extended modeling techniques and tools, please refer to [97–100]. 6.1 ASTRO: A Tool to Simulate Astrocyte Activity in Realistic Astrocyte Ultrastructures at the Whole-Cell Level

ASTRO [95] is a computational tool for simulating astrocyte activity and is based on the NEURON software framework [101]. In addition to developing ASTRO, Savtchenko et al. [95] have developed an algorithm that creates geometries of single astrocytes based on experimental 3D reconstructions. The resulting geometries are adapted to be compatible with NEURON software, thus consisting of trees of 1D compartments. By adding processes randomly to those geometries, the ASTRO software allows the user to vary the tissue volume fraction occupied by processes as well as their surfaceto-volume ratios. Reactions in each compartment are described by ODEs (deterministic spatially extended approach, see above) and compartments are coupled by diffusion (Fig. 5). The software has been extensively validated against experimental data performed on hippocampal astrocytes from the CA1 region, obtained with various approaches such as patch-clamp recordings, electrophysiology, two-photon excitation imaging, spot-uncaging, fluorescence recovery after photobleaching, in vivo Ca2+ imaging, and quantitative correlational electron microscopy. Computational models of astrocytes have similar characteristics to those studied experimentally, such as their intracellular diffusional connectivity and their passive electrical properties. ASTRO extends NEURON by adding features that are relevant for modeling astrocytes, such as the description of surface-volume ratios, sites of glutamate application, as well as the number and location of endfeet and gap junctions. Simulations using ASTRO have provided new insights into astrocyte activity, predicting various mechanisms controlling astrocyte physiology, such as the decrease of Ca2+ wave speed and amplitude caused by increased Ca2+ buffering. They further predicted that local K+ efflux can efficiently prevent the spatial spread of elevations of intracellular K+ concentration resulting from K+ uptake. Finally, the detailed compartmentalization of the model allows for changing local characteristics of astrocyte activity, such as the local Ca2+ channel cluster size. This allowed Savtchenko et al. to illustrate the complex interplay between the inter-Ca2+ channel cluster distance, the associated Ca2+ activity, and its fluorescence readout, mediated by Ca2+ indicators. The ASTRO tool can be used to test the effect of diverse characteristics of subcellular astrocyte subcompartments on cellular dynamics at the whole-cell level, such as membrane voltage spread, input resistance, and the generation of Ca2+ waves.

Computational Models of Astrocyte-Neuron Interactions

251

6.2 A Multicompartmental Model of Ca2+ Activity in an Astrocyte

Cresswell-Clay et al. [94] have developed a model that divides the astrocyte into different major compartments. Reactions involved in Ca2+ signaling differ depending on the location within the cell and Ca2+ diffuses between compartments. Ca2+ can enter PAPs following neuronal stimulation. This reaction depends on Ca2+ influx through Ca2+ channels at the plasma membrane such as the NCX. Larger processes contain some ER and are characterized by IP3R-dependent Ca2+ signaling, including Ca2+-induced-Ca2+ release. Ca2+ removal results from the activity of ATPases at the plasma membrane and the membrane of the ER (SERCA pumps). Larger compartments, i.e., the soma and five major branches, are nonspatial (single-point models), connected to the rest of the astrocyte subcompartments by Ca2+ diffusion in the cytosol and the ER (see Fig. 5). Cresswell-Clay et al. have used this model to study the influence of neuronal input properties, such as its amplitude or frequency, and diffusive properties, such as Ca2+ diffusion coefficient in the cytosol or ER, on Ca2+ spikes in the soma of the astrocyte. They found that concentrating neuronal inputs onto fewer astrocyte processes and increased synchrony of Ca2+ signals in processes facilitated the emergence of somatic Ca2+ spikes. Their results further suggested that cell morphology influenced Ca2+ activity. In particular, an increased somatic volume was associated with a decreased somatic spike probability. Further, they found that an increased Ca2+ diffusion coefficient in the cytosol facilitated the emergence of somatic spikes so that fewer process spikes were needed to result in a somatic event. Finally, Ca2+ diffusion within the ER led to a non-monotonic variation of Ca2+ somatic spikes with the neuronal input intensity, caused by Ca2+ depletion in the ER for high neuronal input frequencies. In summary, the Cresswell-Clay et al. model has improved our understanding of the integration of neuronal inputs by single astrocytes by varying spatial factors, such as the distribution of neuronal inputs over the astrocyte and diffusional properties in the cytosol and the ER, which cannot be performed experimentally. This model is best suited for studying the interactions between astrocyte compartments of different sizes in response to neuronal activity, notably the integration and propagation of Ca2+ signals at the whole-cell level.

6.3 A Spatial Model of Ca2+ Activity in a Perisynaptic Astrocyte Process

Most astrocyte-neuron communication occurs in fine PAPs that contain a very low number of molecules and ions so that the kinetics of the associated reactions are highly stochastic. Stochastic spatially extended approaches are best suited to model astrocyte physiology at this spatial scale. The model from Denizot et al. [96] corresponds to a model of IP3R-dependent Ca2+ signals in fine processes (Fig. 5), implemented both in 2D, with a custom-made particle-based simulator, and in 3D, using the voxel-based STEPS

252

Kerstin Lenk et al.

software [102]. The latter allows running simulations at the nanoscale in both simplified 3D shapes of thin processes and more realistic ultrastructures reconstructed from electron microscopy, for example. Simulations of a nonspatial, non-stochastic implementation of the model highlighted that stochasticity was necessary for spontaneous Ca2+ signals to be triggered in fine processes. The 2D implementation allowed the authors to explore the range of dynamical behaviors that the model displays, suggesting that Ca2+ peak frequency increases when Ca2+ channels are organized into spatial clusters. Simulations of the 3D model implementation were performed in a simplified astrocyte process morphology in 3D, consisting of a 1 μm-long, 100 nm in radius cylinder, which displayed a similar Ca2+ microdomain activity than recorded in organotypic cultures of hippocampal astrocytes. Simulations quantified the alteration of Ca2+ signals by Ca2+ indicators, which are necessary to perform Ca2+ recordings experimentally. Increased concentrations of Ca2+ indicators were notably resulting in a decrease in Ca2+ peak amplitude and frequency. The model from Denizot et al. has recently been used to investigate the effect of remodeling the astrocyte nano-architecture observed in pathological hypo-osmotic conditions [103] on local astrocyte Ca2+ activity at tripartite synapses [104]. Simulation results suggest that the nanoscale reticular morphology of astrocyte processes observed in healthy tissue [105] enhances local Ca2+ activity and that this effect is hindered in pathological conditions, which was confirmed by Ca2+ imaging experiments. More recently, simulations of this model in realistic 3D geometries of PAPs reconstructed from electron microscopy gave new insights into the complex interplay between ER shape and distribution, the clustering of Ca2+ channels, and Ca2+ buffering mechanisms in regulating microdomain Ca2+ activity at tripartite synapses [106]. The high spatial resolution of this model comes at a high computational cost, and simulations of hundred seconds of chemical reactions in a fine process take several days to compute, which is much slower than the compartmental models of Savtchenko et al. [95] and Cresswell-Clay et al. [94], despite simulating smaller subcellular compartments. For that reason, the model from Denizot et al. is best suited to study astrocyte physiology in fine processes and can be used to test the effect of spatial factors, such as cell morphology and the distribution of Ca2+ channels, on astrocyte microdomain Ca2+ activity. 6.4

Discussion

In this section, we have presented some of the recent models that take into account the complex morphology of astrocytes to investigate the effect of spatial properties of astrocytes on their activity. The presented models describe different signaling pathways and cell shapes, using different spatial resolutions and accuracy. Such spatially extended models are useful tools to test the effect of factors

Computational Models of Astrocyte-Neuron Interactions

253

that might be crucial to astrocyte physiology, such as the location and density of gap junctions, the distribution and size of Ca2+ channel clusters, and the local variability of astrocyte morphology. As those parameters vary drastically in pathological conditions and are often inaccessible experimentally, those models offer valuable opportunities to better understand the biochemical processes that underlie astrocyte activity and astrocyte-neuron communication in health and disease.

7

Astrocyte Networks Astrocytes establish complex networks with the numerous cells they are contacting. Notably, in the human brain, a single protoplasmic astrocyte could contact up to two million synapses residing in its territorial domain [107], forming numerous tripartite synapses (see Subheading 2) [108]. Astrocyte and neuronal networks are, thus, tightly interwoven. Astrocyte activity is correlated to neuronal synchronization [109, 110]. Yet, the mechanisms by which neuron-astrocyte communication shapes network activity in health and disease remain poorly understood. Astrocytes are characterized by nonoverlapping spatial domains [111] and are connected to neighboring astrocytes through gap junction channels. This coupling allows for the flow of small molecules (e.g., IP3) and ions from one cell to another [112]. Intercellular Ca2+ waves can spread from one astrocyte to up to 70 neighboring cells in rodent cultures [113]. Such a spatial spread of astrocyte Ca2+ signals can influence the activity of numerous neuronal circuits simultaneously. Studying the interplay between astrocyte and neuronal activity at the network level is thus crucial to expanding our understanding of brain physiology. In this section, we present three computational network models that describe astrocyte-neuron networks that take into account the network topology and include hundreds of astrocytes (Fig. 6).

7.1 A Topologically Realistic Model of Astrocyte Networks

Verisokin et al. [114] have implemented a morphologically detailed network model of astrocytes focusing on the spatial spread of Ca2+ signals (Fig. 6). Each astrocyte shape was generated by randomly transforming confocal microscope images. In the model, astrocytes were randomly placed onto a grid of 0.275 μm/px for the singleastrocyte simulations and 0.55 μm/px for the network simulations while ensuring that astrocyte territories did not overlap. In the model, each astrocyte is subdivided into (i) the soma with thick branches, (ii) the thin astrocyte processes, and (iii) the surrounding ECS. In the model, the astrocytes are stimulated by glutamate released by connected presynaptic neurons, modeled with a stochastic Poissonian distribution. In other words, the neurons are not

254

Kerstin Lenk et al. Neuron-astrocyte network models Glutamate and GABA effects on the network activity ex: Lenk et al. (2020); Li et al. (2020) Lenk et al. (2020) Presynaptic neuron

Postsynaptic neuron Glutamate

Astrocytes

mGluR / NMDAR Adenosine receptor

Presynaptic neuron

Local Ca2+ dynamics

Postsynaptic neuron

Astrocyte network models Ca2+ wave diffusion in the astrocyte network ex: Verisokin et al. (2021) Glutamate

J glu

Astrocyte

Li et al. (2020) Presynaptic neuron

Postsynaptic neuron Glutamate

[Ca 2+]

AMPAR NMDAR

AMPAR GABA BR NMDAR

Iglu

Global Ca2+dynamics

Global Ca2+dynamics Astrocyte

LIF

J in

Adenosine release

mGluR

Glutamate

AMPAR NMDAR

J out ICa

[IP3]

Iun

[Ca C 2++] cyt [Ca2+]ER

mGluR

GABAex

Jpump Jleak JIP P3

GABA B

[Ca 2+]

[IP P3 ] ER Astrocyte

Fig. 6 Astrocyte and neuron-astrocyte network models. Verisokin et al. [114] modeled the propagation of Ca2+ waves in astrocyte networks. Both Lenk et al. [34] and Li et al. [115] models describe the propagation of Ca2+ waves in astrocyte networks coupled with the excitatory and inhibitory transmission in neighboring neuronal networks

explicitly modeled. Astrocyte activity was described using the mGluR-dependent Ca2+ signaling model by Ullah et al. [116]. The contribution of the different Ca2+ pathways (i.e., ERor plasma membrane-mediated) varied depending on the surfacevolume ratio of the subcellular compartment. More precisely, ER-mediated Ca2+ signals prevail in the soma and thick branches, while signals in leaflets are mediated by channels at the plasma membrane. Ca2+ and IP3 diffusion are described both within and between astrocytes. The model was able to reproduce Ca2+ activity measured experimentally in terms of duration and spatial spread at the single-cell level as well as in terms of spatial spread in the network [117]. Simulation results indicated that even though all cells were described similarly, the network presented a pacemaker-like behavior, i.e., the spread of signals originated from a specific pool of cells. This

Computational Models of Astrocyte-Neuron Interactions

255

resulted from differences in cell morphology as well as in the astrocyte-to-astrocyte contacts in the network and drove the activation of multicellular Ca2+ waves, which often displayed similar spatiotemporal properties. The novelty of the Verisokin et al. model is that it simulates the astrocyte Ca 2+ activity within realistic cell shapes. This feature makes the model most suitable to study the effect of cell morphology on Ca2+ dynamics at the whole cell as well as at the astrocyte network levels. 7.2 A Topologically Realistic Model of Neuron-Astrocyte Networks

Lenk et al. [34] introduced a neuron-astrocyte network model with a biologically plausible network topology. The simulations aimed to reproduce neuronal spiking recorded from rodent co-cultures plated on in vitro microelectrode arrays. In the model, the network includes 250 neurons and varying ratios of astrocytes, which are distributed over a 750 × 750 μm2 2D space. Both cell types are modeled as points in space, i.e., they do not have a morphology. The astrocytes are randomly placed in the 2D space and are connected via gap junctions if the inter-soma distance between two cells is lower than 100 μm. The neurons, thereof 80% excitatory and 20% inhibitory, are randomly distributed in the 2D space, and long-distance connections (up to 500 μm) are allowed. Of note is that the model currently only connects excitatory neurons with astrocytes due to the limited information on interactions between inhibitory neurons and astrocytes at the time. Upon incoming spikes, the excitatory neurons release glutamate into the synaptic cleft, which activates glutamate receptors at the membrane of the postsynaptic neuron and the perisynaptic astrocyte. The astrocyte activity is modeled by Lenk et al. using the model from De Pitta` et al. [27] (see Subheading 2.1), to which they added the release of the gliotransmitters glutamate and ATP by the astrocyte into the synaptic cleft (Fig. 6). In the first set of in silico experiments, each excitatory presynaptic neuron was connected to an astrocyte, while astrocytes were not coupled. This network topology resulted in increased spike and burst rates to pathological levels. Then, simulations were performed in the complete neuron-astrocyte network model with 10%, 20%, or 30% astrocytes, which were connected by gap junctions. This topology led to a reduction of the neuronal spiking and bursting rates to healthy ranges. Increasing the number of astrocytes shaped neuronal network activity by preventing overexcitation. In Fritschi et al. [118], the Lenk et al. model was used to investigate four hypotheses on the pathological mechanisms involving astrocytes in schizophrenia: (i) The number of neurons or astrocytes in the network is reduced. (ii) There is an effect of astrocyte ATP on postsynaptic activity. (iii) The release of glutamate from the presynapse and the uptake of glutamate by the astrocyte is

256

Kerstin Lenk et al.

altered in schizophrenia. (iv) The excitatory and/or inhibitory synaptic strength, i.e., the coupling between neurons is stronger in this disease. In summary, Lenk et al. modeled the communication between neurons and astrocytes in networks with a high number of cells. The network topology is highly controllable by the model user, which makes the model useful to study the effect of network topology on neuronal and astrocyte activity in 2D and 3D [119– 121]. 7.3 A Network Model of GABA-Evoked Neuron-Astrocyte Communication

Li et al. [115] developed a model of extracellular GABA (γ-aminobutyric acid) activation of astrocytes, resulting in IP3mediated Ca2+ signals (Fig. 6). The neurons are modeled using the leaky integrate-and-fire (LIF) formalism [122], which describes a neuron as an electrical circuit composed of a capacitor (C) in parallel with a resistor (R), such as in an RC circuit. When a current is injected into the model, the LIF neuron acts as a resistor, and, in the absence of an input current, the membrane potential discharges exponentially to its resting value. Excitatory and inhibitory neurons differ in the model only by their initial excitatory and inhibitory conductances. The presynaptic neurons express glutamate receptors (e.g., NMDARs and mGluRs) and GABA receptors (e.g., GABABRs). The model further describes the activity of NMDA and AMPA receptors in the postsynaptic neurons. The astrocytes express mGluRs at their plasma membrane, whose activation leads to IP3-evoked Ca2+ signaling (see Subheading 2). The novelty of this model is to incorporate GABABRs at the astrocyte plasma membrane, whose activation also results in IP3 synthesis. The 2D network model comprises 500 neurons (400 excitatory and 100 inhibitory), with a 20% connection probability, and 400 astrocytes. The cells are uniformly distributed onto a 10 × 10 mm2 planar grid. The astrocytes in the network are on average connected to 100 neighboring excitatory synapses and four astrocytes. The model simulates the response of the network to an injection of exogenous GABA in the ECS. At the synaptic level, the model describes the activation of GABABRs in the astrocyte and the presynaptic neuron. In the presynaptic neuron, the activation of GABABRs decreases glutamate release probability, which counteracts the increased glutamate release from the astrocyte. Changes in GABA concentrations are based on the presynaptic release and exogenous input and then decay exponentially. The results of this work suggest how elevated extracellular GABA concentrations can increase the duration and amplitude of astrocyte Ca2+ signals in a concentration-dependent manner. Without external GABAergic stimuli, the astrocyte Ca2+ oscillations were slower and more similar to those measured in healthy conditions.

Computational Models of Astrocyte-Neuron Interactions

257

Overall, the Li et al. model describes the effects of GABA release on glutamatergic synaptic transmission and is thus suitable for studying the interplay between excitatory and inhibitory signaling in neuron-astrocyte networks. 7.4

8

Discussion

Astrocytes and neurons in the brain form interconnected networks. Verisokin et al. [114] modeled the influence of cell morphology on Ca2+ activity in astrocyte networks. The model framework is similar to Savtchenko et al. [95] (see Subheading 6.1) but with fewer biochemical details, thus facilitating the simulation of astrocyte activity at the network level. Li et al. [115] and Lenk et al. [34] models describe the interactions between neurons and astrocytes. These models do not describe cell morphology but rather concentrate on neuron-astrocyte communication through neuro- and gliotransmission at the network level. This section illustrates how computational network models can be used to test different hypotheses related to gliotransmission and its effect on neuronal activity (see [22, 23] for reviews on the current debates).

Concluding Remarks One of the biggest unresolved questions in neuroscience lies in understanding the physiological roles played by glial cells, the “other half of the brain,” in different anatomical regions and brain states. By now, it is well established that astrocytes, the most abundant glial cell type, display a rich repertoire of functions that operate over diverse spatiotemporal scales. Since most of the “currency” of these cells (i.e., glutamate, ATP, Ca2+ signals, etc.) is common to that of neurons, disambiguating their precise contribution to brain function in health and disease has been challenging. The history of neuroscience tells us that groundbreaking discoveries have often materialized through a synergy between experimental insights and mathematical and computational models [123– 127]. In the last two decades, we have witnessed a deluge of experimental investigations targeting astrocytes, which has helped deepen our understanding of their contribution to brain function. However, making sense of the resulting high-dimensional data is a major challenge, so that a modeling framework equivalent to that of neurons is critical to fill in the missing gaps. A model is an abstraction with an immediate goal to reduce the dimensionality of the problem under consideration. Thus, this minimal representation of the system carries information about the components that are critically involved in a function of interest. The iterative and trialand-error process of building minimal representations (or models) is typically based on a key observation from experiments and a modeling intuition (a good example is the Hodgkin-Huxley model of action potential propagation [38]). In this scheme,

258

Kerstin Lenk et al.

variables, timescales, and parameters can be systematically explored; what does not fit is thrown out and new components are brought in. Computational models of astrocytes range widely in scale, from the nanoscopic interactions of individual molecules to intercellular processes at the network level. Besides the spatial scale, the temporal scales of these models also vary largely (from milliseconds to seconds). We envision the reader selecting sections and models that are relevant to the experiments they are running and the data obtained. We further aim to provide a concise overview of the types of models that are available, together with a glimpse window into their usage. Our aim with this book chapter is to highlight how computational models complement experiments in the quest for unraveling neuron-astrocyte communication at glutamatergic synapses to foster collaboration between neuroscience disciplines.

9

List of Resources ModelDB (https://senselab.med.yale.edu/ModelDB/) and CellML (https://www.cellml.org/) are the main open-access databases that host numerous models of neurons and astrocytes. Table 1 provides the links to the models described in this chapter that are available online.

Acknowledgments K.L.s research was partially conducted while visiting the Okinawa Institute of Science and Technology (OIST) through the Theoretical Sciences Visiting Program (TSVP). A. D.s work was funded by the Okinawa Institute of Science and Technology Graduate University and by JSPS (Japan Society for the Promotion of Science) Postdoctoral Fellowship for Research in Japan (Standard, P21733). S.N. would like to acknowledge Shweta Shrotri, IISER Pune, for her help with Figs. 1 and 4. References 1. Araque A, Parpura V, Sanzgiri RP et al (1999) Tripartite synapses: glia, the unacknowledged partner. Trends Neurosci 22:208–215 2. Semyanov A, Henneberger C, Agarwal A (2020) Making sense of astrocytic calcium signals — from acquisition to interpretation. Nat Rev Neurosci 21:551–564 3. Verkhratsky A, Nedergaard M (2018) Physiology of astroglia. Physiol Rev 98:151

4. Arizono M, N€agerl UV (2022) Deciphering the functional nano-anatomy of the tripartite synapse using stimulated emission depletion microscopy. Glia 70:607–618 5. Oschmann F, Berry H, Obermayer K et al (2018) From in silico astrocyte cell models to neuron-astrocyte network models: a review. Brain Res Bull 136:76–84

Computational Models of Astrocyte-Neuron Interactions 6. Denizot A, Berry H, Venugopal S (2020) Intracellular calcium signals in astrocytes, computational modeling of. In: Jaeger D, Jung R (eds) Encyclopedia of computational neuroscience. Springer New York, New York, pp 1–12 7. Manninen T, Havela R, Linne M-L (2018) Computational models for calcium-mediated astrocyte functions. Front Comput Neurosci 12:14 8. Manninen T, Havela R, Linne M-L (2017) Reproducibility and comparability of computational models for astrocyte calcium excitability. Front Neuroinform 11:11 9. Gonza´lez J, Pinzo´n A, Angarita-Rodrı´guez A et al (2020) Advances in astrocyte computational models: from metabolic reconstructions to multi-omic approaches. Front Neuroinform 14:35 10. Sejnowski TJ, Koch C, Churchland PS (1988) Computational neuroscience. Science 241: 1299–1306 11. FitzHugh R (1955) Mathematical models of threshold phenomena in the nerve membrane. Bull Math Biophys 17:257–278 12. Covelo A, Badoual A, Denizot A (2022) Reinforcing interdisciplinary collaborations to unravel the astrocyte “calcium code”. J Mol Neurosci 72:1443–1455 13. Giugliano M, Negrello M, Linaro D (eds) (2022) Computational modelling of the brain: modelling approaches to cells, circuits and networks. Springer International Publishing, Cham 14. De Schutter E (ed) (2009) Computational modeling methods for neuroscientists. The MIT Press, Cambridge 15. De Pitta` M, Berry H (eds) (2019) Computational Glioscience. Springer International Publishing, Cham 16. Eriksson O, Bhalla US, Blackwell KT et al (2022) Combining hypothesis- and datadriven neuroscience modeling in FAIR workflows. elife 11:e69013 17. Crook S (2013) Model reproducibility: overview. In: Jaeger D, Jung R (eds) Encyclopedia of computational neuroscience. Springer, New York, pp 1–3 18. Benureau FCY, Rougier NP (2018) Re-run, repeat, reproduce, reuse, replicate: transforming code into scientific contributions. Front Neuroinform 11:69 19. Fellin T (2009) Communication between neurons and astrocytes: relevance to the modulation of synaptic and network activity. J Neurochem 108:533–544

259

20. Adamsky A, Goshen I (2018) Astrocytes in memory function: pioneering findings and future directions. Neuroscience 370:14–26 21. Araque A, Carmignoto G, Haydon PG et al (2014) Gliotransmitters travel in time and space. Neuron 81:728–739 22. Savtchouk I, Volterra A (2018) Gliotransmission: beyond black-and-white. J Neurosci 38: 14–25 23. Fiacco TA, McCarthy KD (2018) Multiple lines of evidence indicate that gliotransmission does not occur under physiological conditions. J Neurosci 38:3–13 24. Kofuji P, Araque A (2021) G-protein-coupled receptors in astrocyte–neuron communication. Neuroscience 456:71–84 25. Gonza´lez-Arias C, Perea G (2019) Gliotransmission at tripartite synapses. In: De Pitta` M, Berry H (eds) Computational glioscience. Springer International Publishing, Cham, pp 213–226 26. Gordleeva SY, Ermolaeva AV, Kastalskiy IA et al (2019) Astrocyte as spatiotemporal integrating detector of neuronal activity. Front Physiol 10:294 27. De Pitta` M, Goldberg M, Volman V et al (2009) Glutamate regulation of calcium and IP3 oscillating and pulsating dynamics in astrocytes. J Biol Phys 35:383–411 28. Nadkarni S, Jung P, Levine H (2008) Astrocytes optimize the synaptic transmission of information. PLoS Comput Biol 4:e1000088 29. Dupont G, Combettes L, Bird GS et al (2011) Calcium oscillations. Cold Spring Harb Perspect Biol 3:a004226 30. Li Y-X, Rinzel J (1994) Equations for InsP3 [Ca2+]i oscillations receptor-mediated derived from a detailed kinetic model: a Hodgkin-Huxley like formalism. J Theor Biol 166:461–473 31. Dupont G, Sneyd J (2017) Recent developments in models of calcium signalling. Curr Opin Syst Biol 3:15–22 32. Siekmann I, Cao P, Sneyd J et al (2019) Datadriven modelling of the inositol trisphosphate receptor (IP3R) and its role in calciuminduced calcium release (CICR). In: De Pitta` M, Berry H (eds) Computational glioscience. Springer International Publishing, Cham, pp 39–68 33. Vuillaume R, Lorenzo J, Binczak S et al (2021) A computational study on synaptic plasticity regulation and information processing in neuron-astrocyte networks. Neural Comput 33:1970–1992

260

Kerstin Lenk et al.

34. Lenk K, Satuvuori E, Lallouette J et al (2020) A computational model of interactions between neuronal and astrocytic networks: the role of astrocytes in the stability of the neuronal firing rate. Front Comput Neurosci 13:92 35. Liu J, McDaid L, Araque A et al (2019) GABA regulation of burst firing in hippocampal astrocyte neural circuit: a biophysical model. Front Cell Neurosci 13:335 36. Gordleeva SY, Lebedev SA, Rumyantseva MA et al (2018) Astrocyte as a detector of synchronous events of a neural network. JETP Lett 107:440–445 37. Gordleeva SY, Stasenko SV, Semyanov AV et al (2012) Bi-directional astrocytic regulation of neuronal activity within a network. Front Comput Neurosci 6:92 38. Hodgkin AL, Huxley AF (1952) A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiol 117:500–544 39. Bertram R, Sherman A, Stanley EF (1996) Single-domain/bound calcium hypothesis of transmitter release and facilitation. J Neurophysiol 75:1919–1931 40. Kang J, Jiang L, Goldman SA et al (1998) Astrocyte-mediated potentiation of inhibitory synaptic transmission. Nat Neurosci 1:683– 692 41. Dobrunz LE, Huang EP, Stevens CF (1997) Very short-term plasticity in hippocampal synapses. Proc Natl Acad Sci U S A 94: 14843–14847 42. Tsodyks MV, Markram H (1997) The neural code between neocortical pyramidal neurons depends on neurotransmitter release probability. Proc Natl Acad Sci U S A 94:719–723 43. Nadkarni S, Jung P (2007) Modeling synaptic transmission of the tripartite synapse. Phys Biol 4:1–9 44. Pasti L, Volterra A, Pozzan T et al (1997) Intracellular calcium oscillations in astrocytes: a highly plastic, bidirectional form of communication between neurons and astrocytes in situ. J Neurosci 17:7817–7830 45. Pasti L, Zonta M, Pozzan T et al (2001) Cytosolic calcium oscillations in astrocytes may regulate exocytotic release of glutamate. J Neurosci 21:477–484 46. Parpura V, Haydon PG (2000) Physiological astrocytic calcium levels stimulate glutamate release to modulate adjacent neurons. Proc Natl Acad Sci U S A 97:8629–8634 47. Perea G, Araque A (2007) Astrocytes potentiate transmitter release at single hippocampal synapses. Science 317:1083–1086

48. Fiacco TA, McCarthy KD (2004) Intracellular astrocyte calcium waves in situ increase the frequency of spontaneous AMPA receptor currents in CA1 pyramidal neurons. J Neurosci 24:722–732 49. Young GWD, Keizer J (1992) A single-pool inositol 1,4,5-trisphosphate-receptor-based model for agonist-stimulated oscillations in Ca 2+ concentration. Proc Natl Acad Sci U S A 89:9895–9899 50. Bezprozvanny L, Watras J, Ehrlich BE (1991) Bell-shaped calcium-response curves of lns (l,4,5)P3- and calcium-gated channels from endoplasmic reticulum of cerebellum. Nature 351:751–754 51. Rusakov DA, Stewart MG (2021) Synaptic environment and extrasynaptic glutamate signals: the quest continues. Neuropharmacology 195:108688 52. Ding S, Fellin T, Zhu Y et al (2007) Enhanced astrocytic Ca2+ signals contribute to neuronal excitotoxicity after status epilepticus. J Neurosci 27:10674–10684 53. Danbolt NC (2001) Glutamate uptake. Prog Neurobiol 65:1–105 54. Eulenburg V, Gomeza J (2010) Neurotransmitter transporters expressed in glial cells as regulators of synapse function. Brain Res Rev 63:103–112 55. Bergles DE, Diamond JS, Jahr CE (1999) Clearance of glutamate inside the synapse and beyond. Curr Opin Neurobiol 9:293– 298 56. Anderson CM, Swanson RA (2000) Astrocyte glutamate transport: review of properties, regulation, and physiological functions. Glia 32: 1–14 57. Rose CR, Ziemens D, Untiet V et al (2018) Molecular and cellular physiology of sodiumdependent glutamate transporters. Brain Res Bull 136:3–16 58. Scimemi A, Diamond JS (2013) Deriving the time course of glutamate clearance with a deconvolution analysis of astrocytic transporter currents. J Vis Exp (78):50708 59. Rǎdulescu AR, Todd GC, Williams CL et al (2022) Estimating the glutamate transporter surface density in distinct sub-cellular compartments of mouse hippocampal astrocytes. PLoS Comput Biol 18:e1009845 60. Flanagan B, McDaid L, Wade J et al (2018) A computational study of astrocytic glutamate influence on post-synaptic neuronal excitability. PLoS Comput Biol 14:e1006040 61. He´ja L, Kardos J (2020) NCX activity generates spontaneous Ca2+ oscillations in the

Computational Models of Astrocyte-Neuron Interactions astrocytic leaflet microdomain. Cell Calcium 86:102137 62. De Pitta` M, Brunel N (2016) Modulation of synaptic plasticity by glutamatergic gliotransmission: a modeling study. Neural Plast 2016: 1–30 63. Østby I, Øyehaug L, Einevoll GT et al (2009) Astrocytic mechanisms explaining neuralactivity-induced shrinkage of extraneuronal space. PLoS Comput Biol 5:e1000272 64. Breslin K, Wade JJ, Wong-Lin K et al (2018) Potassium and sodium microdomains in thin astroglial processes: a computational model study. PLoS Comput Biol 14:e1006151 65. Wade JJ, Breslin K, Wong-Lin K et al (2019) Calcium microdomain formation at the perisynaptic cradle due to NCX reversal: a computational study. Front Cell Neurosci 13:185 66. Oschmann F, Mergenthaler K, Jungnickel E et al (2017) Spatial separation of two different pathways accounting for the generation of calcium signals in astrocytes. PLoS Comput Biol 13:e1005377 67. Bindocci E, Savtchouk I, Liaudet N et al (2017) Three-dimensional Ca 2+ imaging advances understanding of astrocyte biology. Science 356:eaai8185 68. Haydon PG, Parpura V (eds) (2009) Astrocytes in (patho)physiology of the nervous system. Springer US, Boston 69. Elul R (1967) Fixed charge in the cell membrane. J Physiol 189:351–365 70. Srinivasan R, Huang BS, Venugopal S et al (2015) Ca2+ signaling in astrocytes from Ip3r2-/- mice in brain slices and during startle responses in vivo. Nat Neurosci 18: 708–717 71. Patrushev I, Gavrilov N, Turlapov V et al (2013) Subcellular location of astrocytic calcium stores favors extrasynaptic neuron– astrocyte communication. Cell Calcium 54: 343–349 72. Ziemens D, Oschmann F, Gerkau NJ et al (2019) Heterogeneity of activity-induced sodium transients between astrocytes of the mouse hippocampus and neocortex: mechanisms and consequences. J Neurosci 39:2620– 2634 73. Tsacopoulos M, Magistretti PJ (1996) Metabolic coupling between glia and neurons. J Neurosci 16:877–885 74. Dienel GA (2012) Brain lactate metabolism: the discoveries and the controversies. J Cereb Blood Flow Metab 32:1107–1138 75. Magistretti PJ, Allaman I (2015) A cellular perspective on brain energy metabolism and functional imaging. Neuron 86:883–901

261

76. Jolivet R (2009) Deciphering neuron-glia compartmentalization in cortical energy metabolism. Front Neuroenerg 1:4 77. Mangia S, Simpson IA, Vannucci SJ et al (2009) The in vivo neuron-to-astrocyte lactate shuttle in human brain: evidence from modeling of measured lactate levels during visual stimulation. J Neurochem 109:55–62 78. Aubert A, Pellerin L, Magistretti PJ et al (2007) A coherent neurobiological framework for functional neuroimaging provided by a model integrating compartmentalized energy metabolism. Proc Natl Acad Sci U S A 104:4188–4193 79. Simpson IA, Carruthers A, Vannucci SJ (2007) Supply and demand in cerebral energy metabolism: the role of nutrient transporters. J Cereb Blood Flow Metab 27:1766–1791 80. Cloutier M, Bolger FB, Lowry JP et al (2009) An integrative dynamic model of brain energy metabolism using in vivo neurochemical measurements. J Comput Neurosci 27:391–414 81. Gjedde A (2007) 4.5 coupling of brain function to metabolism: evaluation of energy requirements. In: Lajtha A, Gibson GE, Dienel GA (eds) Handbook of neurochemistry and molecular neurobiology: brain energetics. Integration of molecular and cellular processes. Springer US, Boston, pp 343–400 82. Hyder F, Patel AB, Gjedde A et al (2006) Neuronal–glial glucose oxidation and glutamatergic–GABAergic function. J Cereb Blood Flow Metab 26:865–877 83. Barros LF, Courjaret R, Jakoby P et al (2009) Preferential transport and metabolism of glucose in Bergmann glia over Purkinje cells: a multiphoton study of cerebellar slices. Glia 57:962–970 84. Nehlig A, Wittendorp-Rechenmann E, Dao Lam C (2004) Selective uptake of [14C]2deoxyglucose by neurons and astrocytes: high-resolution microautoradiographic imaging by cellular 14C-trajectography combined with immunohistochemistry. J Cereb Blood Flow Metab 24:1004–1014 85. Mason S (2017) Lactate shuttles in neuroenergetics—homeostasis, allostasis and beyond. Front Neurosci 11:43 86. Patsatzis DG, Tingas E-A, Goussis DA et al (2019) Computational singular perturbation analysis of brain lactate metabolism. PLoS One 14:e0226094 87. Tadi M, Allaman I, Lengacher S et al (2015) Learning-induced gene expression in the hippocampus reveals a role of neuron -astrocyte metabolic coupling in long term memory. PLoS One 10:e0141568

262

Kerstin Lenk et al.

88. Wang D, Pascual JM, Yang H et al (2006) A mouse model for Glut-1 haploinsufficiency. Hum Mol Genet 15:1169–1179 89. Herrero-Mendez A, Almeida A, Ferna´ndez E et al (2009) The bioenergetic and antioxidant status of neurons is controlled by continuous degradation of a key glycolytic enzyme by APC/C–Cdh1. Nat Cell Biol 11:747–752 90. Rusakov DA (2015) Disentangling calciumdriven astrocyte physiology. Nat Rev Neurosci 16:226–233 91. Tønnesen J, Katona G, Ro´zsa B et al (2014) Spine neck plasticity regulates compartmentalization of synapses. Nat Neurosci 17:678– 685 92. Holcman D, Korkotian E, Segal M (2005) Calcium dynamics in dendritic spines, modeling and experiments. Cell Calcium 37:467– 475 93. Obashi K, Taraska JW, Okabe S (2021) The role of molecular diffusion within dendritic spines in synaptic function. J Gen Physiol 153:e202012814 94. Cresswell-Clay E, Crock N, Tabak J et al (2018) A compartmental model to investigate local and global Ca2+ dynamics in astrocytes. Front Comput Neurosci 12:94 95. Savtchenko LP, Bard L, Jensen TP et al (2018) Disentangling astroglial physiology with a realistic cell model in silico. Nat Commun 9:3554 96. Denizot A, Arizono M, N€agerl UV et al (2019) Simulation of calcium signaling in fine astrocytic processes: effect of spatial properties on spontaneous activity. PLoS Comput Biol 15:e1006795 97. Burrage K, Burrage PM, Leier A et al (2011) Stochastic simulation for spatial modelling of dynamic processes in a living cell. In: Koeppl H, Setti G, di Bernardo M et al (eds) Design and analysis of biomolecular circuits: engineering approaches to systems and synthetic biology. Springer, New York, pp 43–62 98. Smith S, Grima R (2019) Spatial stochastic intracellular kinetics: a review of modelling approaches. Bull Math Biol 81:2960–3009 99. Andrews SS (2018) Particle-based stochastic simulators. In: Jaeger D, Jung R (eds) Encyclopedia of computational neuroscience. Springer, New York, pp 1–5 100. Blackwell KT (2013) Approaches and tools for modeling signaling pathways and calcium dynamics in neurons. J Neurosci Methods 220:131–140 101. Hines M, Carnevale T, McDougal RA (2019) NEURON simulation environment. In: Jaeger D, Jung R (eds) Encyclopedia of

computational neuroscience. Springer, New York, pp 1–7 102. Hepburn I, Chen W, Wils S et al (2012) STEPS: efficient simulation of stochastic reaction–diffusion models in realistic morphologies. BMC Syst Biol 6:36 103. Arizono M, Inavalli VVGK, Bancelin S et al (2021) Super-resolution shadow imaging reveals local remodeling of astrocytic microstructures and brain extracellular space after osmotic challenge. Glia 69:1605–1613 104. Denizot A, Arizono M, N€agerl UV et al (2022) Control of Ca 2+ signals by astrocyte nanoscale morphology at tripartite synapses. Glia 70:2378–2391 105. Arizono M, Inavalli VVGK, Panatier A et al (2020) Structural basis of astrocytic Ca2+ signals at tripartite synapses. Nat Commun 11:1906 106. Denizot A, Castillo MFV, Puchenkov P et al (2022) The endoplasmic reticulum in perisynaptic astrocytic processes: shape, distribution and effect on calcium activity. https://www. biorxiv.org/content/10.1101/2022.02.2 8.482292v1 107. Oberheim NA, Wang X, Goldman S et al (2006) Astrocytic complexity distinguishes the human brain. Trends Neurosci 29:547– 553 108. Eroglu C, Barres BA (2010) Regulation of synaptic connectivity by glia. Nature 468: 223–231 109. Pirttimaki TM, Sims RE, Saunders G et al (2017) Astrocyte-mediated neuronal synchronization properties revealed by false gliotransmitter release. J Neurosci 37:9859–9870 110. Szabo´ Z, He´ja L, Szalay G et al (2017) Extensive astrocyte synchronization advances neuronal coupling in slow wave activity in vivo. Sci Rep 7:6018 111. Bushong EA, Martone ME, Jones YZ et al (2002) Protoplasmic astrocytes in CA1 stratum radiatum occupy separate anatomical domains. J Neurosci 22:183–192 112. Evans WH, Martin PEM (2002) Gap junctions: structure and function (Review). Mol Membr Biol 19:121–136 113. Giaume C, Venance L (1998) Intercellular calcium signaling and gap junctional communication in astrocytes. Glia 24:50–64 114. Verisokin AY, Verveyko DV, Postnov DE et al (2021) Modeling of astrocyte networks: toward realistic topology and dynamics. Front Cell Neurosci 15:645068 115. Li L, Zhou J, Sun H et al (2020) A computational model to investigate GABA-activated astrocyte modulation of neuronal excitation.

Computational Models of Astrocyte-Neuron Interactions Comput Math Methods Med 2020: e8750167 116. Ullah G, Jung P, Cornellbell A (2006) Antiphase calcium oscillations in astrocytes via inositol (1, 4, 5)-trisphosphate regeneration. Cell Calcium 39:197–208 117. Wu Y-W, Tang X, Arizono M et al (2014) Spatiotemporal calcium dynamics in single astrocytes and its modulation by neuronal activity. Cell Calcium 55:119–129 118. Fritschi L, Lenk K (2021) Parameter inference for an astrocyte model using machine learning approaches. http://biorxiv.org/ lookup/doi/10.1101/2023.05.16.540982 119. Genocchi B, Ahtiainen A, Barros MT et al (2021) Astrocytic control in in vitro and simulated neuron-astrocyte networks. In: Proceedings of the eight annual ACM international conference on nanoscale computing and communication. ACM, Virtual Event Italy, pp 1–7 120. Lenk K, Genocchi B, Barros MT et al (2021) Larger connection radius increases hub astrocyte number in a 3-D neuron–astrocyte network model. IEEE Trans Mol Biol MultiScale Commun 7:83–88 121. Genocchi B, Lenk K, Hyttinen J (2020) Influence of astrocytic gap junction coupling on in silico neuronal network activity. In:

263

Henriques J, Neves N, de Carvalho P (eds) XV Mediterranean conference on medical and biological engineering and computing – MEDICON 2019. Springer International Publishing, Cham, pp 480–487 122. Sacerdote L, Giraudo MT (2013) Stochastic integrate and fire models: a review on mathematical methods and their applications. In: Bachar M, Batzel J, Ditlevsen S (eds) Stochastic biomathematical models: with applications to neuronal modeling. Springer, Berlin, Heidelberg, pp 99–148 123. Blohm G, Kording KP, Schrater PR (2020) A how-to-model guide for neuroscience. eNeuro 7. https://doi.org/10.1523/ ENEURO.0352-19.2019 124. Churchland PS, Sejnowski TJ (2016) Blending computational and experimental neuroscience. Nat Rev Neurosci 17:667–668 125. Levenstein D, Alvarez VA, Amarasingham A et al (2023) On the role of theory and modeling in neuroscience. J Neurosci 43:1074– 1088 126. Wang X-J, Hu H, Huang C et al (2020) Computational neuroscience: a frontier of the 21st century. Natl Sci Rev 7:1418–1422 127. Gerstner W, Sprekeler H, Deco G (2012) Theory and simulation in neuroscience. Science 338:60–65

Part III Technical Approaches to Study Fast Signaling at Neuron-Glia Synapses

Chapter 12 Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices Coram Guevara, Rodrigo Varas, Marı´a Cecilia Angulo, and Fernando C. Ortiz Abstract Communication between neurons and oligodendrocyte lineage cells has attracted a great interest since multiple discoveries revealed its important roles in brain function under physiological and pathological conditions. Oligodendroglia responds to neuronal activity through the activation of a plethora of ion channels and receptors whose expression changes depending on the maturation state and whose characterization helps defining their interactions with neurons. Here, we describe in detail the methodology for carrying out electrophysiological patch-clamp recordings of oligodendroglial cells in acute brain slices of adult mice, with an emphasis on tailor-made solutions to make this experimental approach successfully. Additionally, we describe a protocol for combining photostimulation of neurons with patch-clamp recordings of oligodendroglia. Key words Oligodendrocyte precursor cells, Oligodendrocytes, Whole-cell patch-clamp, Neuron-glia communication, Optogenetics, Electrophysiology

1

Introduction Neuron-glia interactions in health and disease have been a central topic in neuroscience in the last two decades [1–3]. In particular, neuron-oligodendroglia chemical communication has been very intriguing since the discovery of true synaptic contacts between neurons and oligodendrocyte precursor cells (OPCs) more than 20 years ago [4]. Indeed, OPCs express different types of glutamatergic and GABAergic receptors that allow these cells to respond to the synaptic release of glutamate and GABA, the two major neurotransmitters of the CNS [5]. Although the characteristics of these synapses have been well described, less is known about their

Supplementary Information The online version contains supplementary material available at https://doi.org/ 10.1007/978-1-0716-3742-5_12. Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5_12, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

267

268

Coram Guevara et al.

function [4, 6]. Furthermore, although electrophysiological analyses have allowed for the extensive characterization of oligodendroglia membrane properties in neonatal rodent animals (postnatal day (PN) 3 to PN15), less studies have been performed in the adult model (>PN45). Cumulative evidence has shown that neuronal activity can modulate oligodendroglia proliferation, differentiation, and survival, directly through synaptic receptors such as AMPARs, NMDARs, or GABARs [7, 8] or indirectly by, for instance, modulating surrounding astrocytes [9]. However, the mechanisms underlying OPC development, myelin formation, and repair as well as the roles of oligodendroglia in health and disease are still not totally understood. Therefore, oligodendroglia represents an important field of study that contributes to our understanding of brain physiology and pathophysiology. In line with this, state-of-the-art experimental technologies such as optogenetics, super-resolution microscopy, and optical techniques combined with electrophysiological recordings represent powerful tools to analyze oligodendrocyte (OL) lineage cells (i.e. oligodendroglia). In this chapter, we will describe in detail the methodology to carry out whole-cell patch-clamp recordings of oligodendroglial cells in acute brain slices of adult mice. In addition, we will add a short section dedicated to the description of a protocol to combine optogenetic stimulation of neurons with whole-cell patch-clamp recordings of oligodendroglia.

2 2.1

Materials Animals

A suitable way to identify the different developmental stages of oligodendroglia in the adult (>PN45) is the use of transgenic mice expressing fluorescent reporters under the control of lineage stage-specific promoters. Among the existing mouse lines, CNPase-GFP or Sox10-Venus mouse lines, for instance, can be used for the identification of cells of the entire OL lineage [10, 11], PDGFRα-GFP for OPCs [12, 13] and the MBP-GFP mouse line [14] for mature OLs.

2.2 Buffer and Intracellular Solutions

It is recommended to prepare 10 times (10×) concentrated buffer solution stocks containing most of the inorganic salts. Store them at 4 °C and freshly prepare the final 1× solution the day of the experiment.

2.2.1 N-Methyl-DGlucamine (NMDG)-Based Solution

Prepare the 10× stock considering the following composition of the final (1×) solution: 2.5 mM KCl, 1.2 mM NaH2PO4, 30 mM NaHCO3, 20 mM HEPES, and 1.2 mM thiourea. After dilution of this 10× stock to 1×, add 93 mM NMDG, 25 mM glucose, 5 mM Na-ascorbate, and 3 mM Na-pyruvate. This solution has a basic pH (around 10) that must be carefully adjusted to a pH 7.4 by

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices

269

adding 10 M HCl (drop by drop). After bubbled the solution for at least 10 min with a 95% O2/5% CO2 gas mix, add 10 mM MgCl2 and 0.5 mM CaCl2 from the previously prepared 1 M stock solutions. This solution will be used for cutting and recovery of the slices. In the case of using old animals, viable brain slices from older-than-P120 mice are obtained by a step gradient change of extracellular sodium (please see details in [15]). 2.2.2 Artificial Cerebrospinal Fluid (aCSF) Solution

Prepare the 10× stock considering the following composition of the final (1×) solution: 126 mM NaCl, 2.5 mM KCl, 25.9 mM NaHCO3, and 1.3 mM NaH2PO4. After diluting this stock to 1×, add 20 mM glucose and 5 mM Na-pyruvate. After bubbled the solution for at least 10 min with a 95%O2/5%CO2 gas mix, add 1 mM MgCl2 and 2 mM CaCl2 from previously prepared 1 M stock solutions. This solution will be used for the storage and recordings of brain slices.

2.2.3 Intracellular Solutions

To unmask the characteristic Na+-inward current of OPCs, a Cs-gluconate intracellular based solution is recommended (see [16]). This solution contains 130 mM Cs-gluconate, 10 mM 4-aminopyridine, 5 mM tetraethylammonium chloride, 5 mM EGTA, 0.5 mM CaCl2, 2 mM MgCl2, 10 mM HEPES, 2 mM Na2-ATP, 0.2 mM Na-GTP, and 10 mM Na2-phosphocreatine (pH ≈ 7.4), 296 mOsm/L. Alternatively, the following intracellular solution can be used: 130 mM K-gluconate, 0.1 mM EGTA, 0.5 mM CaCl2, 2 mM MgCl2, 2 mM Na2- ATP, 0.5 mM Na-GTP, 10 mM HEPES, 10 mM phosphocreatine. Dilute Cs- (or K-) gluconate and EGTA in 30 to 40 mL of milliQ water and adjust the pH to 7.3 with a solution of CsOH (or KOH, accordingly). Prepare this solution on ice. Add the required volumes of CaCl2 and MgCl2 solutions and dissolve the HEPES. Adjust the pH to 7.3 while mixing. Dissolve Na2-ATP, Na-GTP, and phosphocreatine and adjust again the pH to 7.3. Complete the volume to 50 mL using a volumetric flask. It is recommended to store this solution at -20 °C in 1 mL aliquots.

2.3 Material for Brain Slice Storage and Recordings

Before starting the slice preparation, prepare a chamber for their storage. The “storage” chambers can be obtained commercially or manufactured in the laboratory (see Note 1). A conventional chamber is a table-shaped plastic structure with a nylon net on top, which is secured by a plastic hoop and must be tight to receive the slices (Fig. 1a). The chamber is placed inside a 250 mL glass beaker that contains either a NMDG-based or an aCSF solution (Fig. 1b). Although brain slices must be kept oxygenated inside the “storage” chamber, it is very important to avoid bringing the bubbling into direct contact of the tissue (see Note 2).

270

Coram Guevara et al.

Fig. 1 Materials for brain slice preparation and recording. Picture of the storage chamber before (a) and after (b) placing it in a 250 mL beaker with either aCSF or NMDG-based solutions (b, see the text for details). Note the protective barrier to prevent direct bubbling the tissue into the storage chamber (arrow in a, see Note 1 for details). A custom-made recording chamber (c) is shown. The recording chamber is mounted on an upright epifluorescence microscope (d) (note the needles of the perfusion system visible in the picture, white arrowheads). The slice-anchor avoids the movement of the slice during patch-clamp recordings (e). Scale bar: 1 cm

Once brain slices are obtained, one of them can be transferred to the recording chamber and placed under the microscope, where patch-clamp recordings are performed (Fig. 1c, d). Again, this chamber can be obtained commercially or custom-made in the laboratory. After placing the brain slice in this recording chamber, it is necessary to use a “slice-anchor” (or “harp”) to immobilize it, preventing the slice from floating in the perfused extracellular

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices

271

solution during patch-clamp recordings (Fig. 1e, see Note 3). Although fixing the slice with a harp could cause some tissue damage, it is by far the most common maneuver to immobilize the slice in the chamber without altering significantly the tissue properties. 2.4 Whole-Cell Patch-Clamp Recordings

Borosilicate glass capillaries containing an inner filament (1.5 mm outer diameter, 0.86 mm inner diameter; GC150F-10, Harvard Apparatus) are the most convenient choice to make the recording microelectrodes. Prepare your micropipettes (see Note 4) by pulling these capillaries in a conventional puller (for instance, a P-87, Sutter Instruments, or a P-1000 Microelectrode Puller, WPI). Use a silver wire coated with a thin layer of silver chloride (AgCl) (see Note 5) on your microelectrode holder to mount the micropipette containing an intracellular solution, and test for the pipette resistance using the membrane test pulse of your software (for instance, PClamp or AxoScope series, Axon Instruments) when the pipette is in the solution. A pipette resistance of 5 MΩ is appropriate for patching oligodendroglia.

2.5 Photostimulation During Whole-Cell Recordings

To combine electrophysiological recordings of oligodendroglia with photostimulation of neurons, a transgenic animal expressing a photosensitive ion channel in neurons (i.e., channelrodopsin-2) or, alternatively, the delivery of a genetic construct containing the photosensitive ion channel (i.e., via a viral-driven expression) [17, 18] is suitable to obtain the responsive cells in the tissue. 1. A fully equipped patch-clamp setup. 2. A light source for photostimulation: ultrahigh-power LED-based or laser-based system capable of delivering the desired wavelength (i.e., for channel rodhodpsin-2 λ ~470 nm, [17]). 3. Current controller to power the light source. 4. Patch-fiber cord to stimulate the sample (the characteristics are described in Subheading 3.4). 5. Pulse generator device to create programmable TTL pulses from a software.

3 3.1

Methods Brain Slices

1. Before preparing the slices, the NMDG-based cutting solution—almost frozen—must be bubbled with 95% CO2/ 5% O2 mix for at least 10 min and maintained on ice. The solution must be at the freezing point during the slicing (see Note 6). For the brain extraction, it is recommended to have three beakers or wells with ice-cold NMDG-based solution (Fig. 2): (i) one beaker to clean the head, (ii) another to handle the head during brain extraction (a low well is recommended in

272

Coram Guevara et al.

Fig. 2 Step-by-step summary of the procedures to perform electrophysiological recordings in brain slices (please refers to Subheadings 3.1 and 3.2 for a full detailed description). (This figure was made using BioRender)

this step), and (iii) a third one to clean the brain once it has been extracted from the skull. 2. Place the slicing chamber on the vibratome according to the manufacturer’s instructions. 3. Fill a beaker containing a storage chamber with oxygenated NMDG-based solution (see Note 6). Fill another beaker (containing another storage chamber) with aCSF. In both cases, remove all bubbles adhering to the nylon net using a Pasteur pipette. Ensure that the slice is continuously oxygenated until the end of the experiment, and remember that bubbles should not touch the slices. Both solutions should be kept at 34 °C in a water bath during the slice cutting. 4. Anesthetize the adult mouse by inhalation of isoflurane 1% (we recommend volatile anesthetic since they are quick and painless and requires no particular technical skills; however, the selection of a particular anesthetic agent must take into account possible effects upon the ionic currents of interest). After checking lack of paw pinch reflex, perform the beheading with a sharp guillotine, a sharp-blunt or a blunt-blunt scissor (i.e., 14001-12 scissors, FST instruments). Please follow all guidelines of your local Animal Care Committee. Quickly immerse it in the first glass beaker containing ice-cold oxygenated NMDG-based solution (see Note 7).

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices

273

5. Dissect the brain out by cutting the skull with a fine straight scissors (i.e., 14060-09 scissors, FST instruments) between the eyes. Continue with lateral cuts close to the base of the skull, in both sides of the head from posterior to anterior. It is recommended that all these steps be made inside the well to keep the tissue constantly wet and cold (Fig. 2). 6. Carefully remove the excised skull with fine forceps ideally angled (45 degrees, i.e., 00109-11 forceps, FST instruments) without damaging the brain, and immediately remove and place the brain in the third clean glass beaker containing ice-cold NMDG-based solution. This procedure removes the excess of blood (Fig. 2). 7. In a wet surface (for instance, a filter paper soaked in NMDGbased solution placed on a petri dish cover), dissect the brain area of interest, paying attention to the angle in order to obtain a flat surface. 8. Stick the brain with glue to the slicing platform, taking into consideration the correct brain orientation for the desired slices (i.e., sagittal, coronal, transverse). Fill the slicing platform with ice-cold oxygenated NMDG-based solution (Fig. 2). From beheading to step 8 should take around 1 min. 9. Cut brain slices of 300 μm thick with the optimal vibration amplitude and frequency for your vibratome (see Note 8). 10. As a slice is made, transfer it to the storage chamber filled with oxygenated NMDG-based solution at 34 °C using a Pasteur pipette whose tip has been cut and smoothed by heating. Each slice must remain in this solution for no more than 8 min. Then, it must be transferred to the aCSF-containing storage chamber—kept also at 34 °C—resting there for 15 min. Before performing the electrophysiological recordings, the slices should rest for another 30 min in the same aCSF-containing solution this time at room temperature (RT) to allow them to restore the intracellular osmolarity and adapt to the new temperature. 3.2 Whole-Cell Recording

1. Unfreeze and filter one aliquot of the desired internal solution and store it on ice for the day. 2. Turn on all the electronic equipment on the electrophysiological rig and oxygenate the aCSF solution with the gas mix (95% O2/5% CO2) for a few minutes before use. 3. Before placing the brain slice on the recording chamber, verify that the perfusion system of the electrophysiological rig works properly. Continuously perfuse the recording chamber with the oxygenated aCSF solution by using either a gravity perfusion system for the influx coupled to a peristaltic pump connected to a suction tubing (outflow) or connect both influx

274

Coram Guevara et al.

and suction tubing to the same peristaltic pump. Be careful in the setting of the inflow and outflow in order to keep the recording chamber volume constant. The optimal flow rate of the perfusion should be around 2.5–3.0 mL per minute. The recording chamber is mounted on the stage of an upright fluorescence microscope (Fig. 1c, d) carrying at least one low-magnification objective (i.e., 4×, 5×, or 10×) and one high-magnification water immersion objective (i.e., 40× or 63×). 4. Transfer a brain slice from the storage chamber to the recording chamber of the setup using the cut Pasteur pipette. To stabilize the slice inside the liquid, place the C-shaped sliceanchor on the top of the slice. Make sure that the nylon threads of the slice-anchor do not interfere with the region of interest. 5. Visualize the slice first with a low-magnification objective to identify the region under study. Once in the area of interest, switch to the 40× objective. 6. In order to identify the cell of interest and note its location on the screen, you can alternate fluorescence (for instance, to identify an OPC in case that you are using transgenic mice; Fig. 3a) and bright-field visualization. This helps to approach the target cell with the pipette (Fig. 3b). 7. Once the best cell to record has been identified, fill a glass micropipette with the internal solution and mount it on the microelectrode holder that is attached to the headstage. Check that the intracellular solution is in contact with the AgClcoated silver wire and that there are no air bubbles at the tip of the pipette. Add some positive pressure into the micropipette before entering to the bath solution. 8. Lower the microelectrode until the surface of the slice under visual control. Be careful; do not touch the slice before checking the pipette resistance inside the bath solution. Use a pipette resistance around 4.5–5.5 MΩ. A higher resistance can be used but makes it difficult to achieve a whole-cell configuration, and a lower resistance makes it difficult to seal the cell. 9. After checking the resistance, adjust the pipette current offset to zero on the amplifier. This offset arises from the voltage difference between the operational amplifier and the recording electrode, normally originated from multiple sources: the intrinsic electrode properties such as resistance and capacitance, small drifts in the electrode position over time, and the small amount of current flowing into the input terminals of the amplifier. The latter can create a small current bias inside the amplifier, contributing to the initially detected offset that must be cancelled (see Note 9).

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices

275

Fig. 3 OL lineage cell identification for whole-cell patch-clamp recordings. Identification of fluorescent oligodendroglial cells in transgenic mice (a). In this example, a coronal brain slice from an adult NG2-CreERT2+/-; tdTomatolox/lox mouse. Note the presence of TdTomato+ pericytes clearly located in vessel walls (white arrowheads) compared with oligodendroglia presented as a rounded-shaped sparse cell population (red arrows). Images were acquired at λ emission = 580 nm by using an excitation laser beam source settled at λ 550 nm wavelength (b). After identification, we increased the transmitted light potency (at 40×) allowing for visualization of the pipette during the final approach to patch the target cell (note the pipette’s shade above the cell indicated by a discontinuous yellow line). (c) To illustrate, here we show a representative example of an OPC, this time pseudo-colored in red after image treatment. Note the ovoid shape with almost no processes of an identified fluorescent OPC. (d) Currents elicited in an identified OPC held at -80 mV (Cs-gluconate intracellular based solution) by voltage steps from -120 to +40 mV. Note the presence of a

276

Coram Guevara et al.

10. Once in the bath solution, continue applying a positive pressure to the micropipette by blowing in steadily and very gently before to enter to the tissue. This prevents the pipette from becoming clogged before reaching the cell of interest. 11. Change from positive to negative pressure (gently sucking) when the pipette is already on the surface of the cell membrane in order to get a gigaseal. A change in the holding potential from 0 mV to a negative potential of -60 mV (or up to 80 mV) while contacting the cell surface might also help to get the gigaseal. The formation of a proper gigaseal is crucial to obtain a good recording (see Note 10). 12. When the seal has been stabilized, cancel the pipette capacitance using the amplifier. 13. To obtain a whole-cell configuration, apply brief pulses of suction while monitoring the seal resistance. This pressure is applied to break the membrane at the tip of the pipette. It is recommended to wait a few seconds between each suction to avoid damaging the cell. The whole-cell configuration is reached when the value of the series resistance dramatically drops (from GΩ to hundreds of MΩ). 14. Since OPCs and immature and mature OLs have specific macroscopic current profiles (see ref. 19), it is recommended to apply a depolarizing voltage-step protocol (in whole-cell configuration, see details in Subheading 3.3.2) to obtain the cell macroscopic currents. This procedure will allow for a further confirmation and identification of the cell (for details, see Subheading 3.3.2). 15. When the OL cell type is identified and the recording is stable, apply your experimental protocols (i.e., electrical or pharmacological stimulation). In our hands, a signal acquisition filtered at 4 kHz and digitized at 20 kHz are recommended to obtain proper whole-cell recordings of OPCs and OLs [19]. 3.3 OL Lineage Identification Hallmarks 3.3.1

Morphology

OL lineage cells are not easily visualized in a brain slice under brightfield or DIC microscopy due to their small size (compared to neurons), and the high refraction index observed in myelinated bundles (i.e. white matter tracts) [20]. Thus, as mentioned, the use of transgenic models provides the easiest method for OL lineage cell identification prior to patch-clamp recordings. For instance,

ä Fig. 3 (continued) characteristic fast transient inward sodium current that becomes evident at voltage step values of -40 mV (inset). (e). OL macroscopic currents elicited by the same voltage steps. Note the lack of the inward sodium current and the difference of current amplitudes (scale bars) compared to those obtained from OPC, as well as the classical linear phenotype of a mature oligodendroglia [19, 24]. (f) Spontaneous excitatory postsynaptic currents (EPSC, *) from the same OPC are shown (magnified examples in the inset)

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices

277

OPCs can be identified by their small, hexagonal, or ellipsoidal cell body and the presence of two major branches (see Fig. 3c). A critical aspect to obtain a proper whole-cell recording is that the target cell has a well-preserved cellular membrane. Under DIC microscopy, healthy cells normally present a smooth and bright cellular outline; a swollen cell with a round and dark outline is not a recommended target. 3.3.2 Inward Sodium Currents

OPCs exhibit a characteristic transient inward voltage-dependent sodium current in response to depolarizing voltage steps (Fig. 3d, Note 11). Thus, once the whole-cell recording is stable, apply a voltage-step stimulation protocol starting from a holding potential of -80 mV to steps from +40 to -120 mV in 10 mV steps while recording the cell macroscopic currents (Fig. 3d). The functional expression of this current decreases in amplitude with oligodendroglia maturation, being smaller in pre-oligodendrocytes and completely absent in differentiated oligodendrocytes (Fig. 3e) [17, 20]. Therefore, studying this sodium current allows for further characterization of OL lineage cells. Additionally, OPCs express spontaneous excitatory postsynaptic currents (EPSCs) in different regions of gray and white matter [4, 5, 21, 22]. Then, after identification by their morphology and macroscopic sodium currents, these cells can also be characterized by checking the presence of EPSCs under voltage-clamp configuration (Fig. 3f).

3.4 Photostimulation of Neurons During OPC Whole-Cell Recordings

This procedure is suitable to study neuron-OPC communication properties. Before starting, ensure that the neurons of interest respond properly to photostimulation (please find a comprehensive detailed methodology in our previous publications on this subject [17, 18]). 1. A LED-based (or laser-based) light source and a patch-cord optic fiber(s) are needed to carry on these experiments (see Note 12). This equipment must be settled combined with the electrophysiology workstation in a way that it allows for the manipulation of all the elements necessary for the achievement of both patch-clamp recordings and light delivery pulses (i.e., let free room for maneuver micromanipulators, micropipettes, optic fibers, etc.). 2. Connect the patch-cord optic fiber to the light source. Minimize the length of the patch-cord to reduce the power loss. In order to photostimulate a broad area, it is recommended the use of an optic fiber with a high numerical aperture (i.e., 0.8 NA). Other parameters of the optic fiber such as its length and diameter should be adjusted according to the setup characteristics and experimental aims (Fig. 4).

278

Coram Guevara et al.

Fig. 4 OPC whole-cell patch-clamp recordings during photostimulation. Setting an experimental station for OL lineage cell recordings during photostimulation (a) requires an additional equipment able to deliver light pulses in a controlled manner. This equipment comprises a current controller, a light source, and a fiber patch cord. (b) In the example, light stimulation is delivered upon neurons expressing the channel-rhodopsin 2 (ChR2) during the recording of an OPC in whole-cell configuration. Note the light-evoked AMPAR-dependent excitatory postsynaptic current in response to photostimulation (blue line). Note that these evoked currents are blocked by the AMPAR antagonist NBQX (red) (b)

3. Measure the power at the tip of the optic fiber outside of the recording chamber. The current controller parameters should be adjusted to obtain the light power required for the stimulation (Fig. 4a, see Note 13). 4. Set the equipment to trigger your stimulation pattern. Usually, this step is achieved by connecting an external TTL-pulse generator to the light-delivery system, but it can also be incorporated as an asset of the delivery system (depending on the light-source distributor). 5. Trigger the stimulation pattern and check whether the light source is delivered the light properly (i.e., following the established pattern) outside of the recording chamber. 6. Before starting the experiment, carefully put the optic fiber into the recording chamber and check for photostimulation artifacts. One way to do it is by delivering light pulses over the tip of the microelectrode placed in the bath solution with no patched cell (see Note 14).

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices

279

7. Repeat the procedure to patch a cell in the region of interest (see Subheading 3.2). 8. Bring the tip of the optic fiber just above the recorded cell and deliver the light pulses with the established parameters. The cell response can be recorded simultaneously (Fig. 4b) (For an extended and detailed methodology on optogenetics procedures, see ref. [17]).

4

Notes 1. A file containing the design of the chamber, to be printed in any conventional 3D-plastic printer, can be download in the online version of the chapter. 2. To prevent oxygen bubbles from contacting the tissue, you can build a diffuser. For this, you can cut the bottom of a 1 mL Eppendorf tube and attach it to the chamber inside the beaker; then, put the bubble tubing inside the cut Eppendorf tube. Alternatively, a protective barrier between the bubbling and the tissue can be made by cutting a small plastic sheet or the lid of a petri dish and sticking it to the edge of the chamber (Fig. 1a). 3. To make a slice-anchor, you can use a thick-platinum wire bended in a C-shape and then stick thin nylon threads to this structure crossing from one side to another (Fig. 1e). To make the nylon threads of the slice-anchor, you can use a pantyhose. Stretch one end of a few nylon threads and glue them to the C-shaped wire by using a strong liquid glue. Then, cut off the excess of nylon threads, and make sure you have enough rows to keep the tissue immobile and enough space between each thread to allow the entry of the recording micropipette into the tissue. 4. Before pulling the micropipettes, flame the two ends of the capillaries to make them smooth. This prevents damaging the AgCl-coated silver wire when placing the pipette into the microelectrode holder. Also, clean the center of the capillary with an alcohol solution (70% v/v) before placing it in the puller apparatus. This ensures no impurities on the tip of the micropipette. 5. The “ground” wire should be carefully chlorinated. It is recommended to check the condition of the ground wire (or pellet) the day before the experiment. When not properly chlorinated (i.e., it does not show a uniform grayish color), it can be a source of electrical noise or instability (i.e., a high junction potential). It is recommended to chlorinate the ground by electrochemical deposition of chloride, for instance, by using a 1 M HCl solution in an electrolysis device.

280

Coram Guevara et al.

6. When possible, set the slicing chamber of the vibratome at – 20 °C before starting the procedure. Similarly, the NMDGbased solution could be stored at -20 °C for 20 or 30 min before starting the procedure. Once it starts, you can maintain the NMDG-based solution on ice. You also should prepare a beaker with a storage chamber containing the NMDG-based solution at 34 °C for its use after obtaining the slices. 7. Oxygenated solutions are necessary to improve the longevity of the slices and maintain their suitable pH. Then, oxygenate the solution before starting the procedure but avoid direct contact of gas bubbles with the tissue during the dissection of the brain. 8. In our hands, a vibration amplitude/frequency of 1–1.2 mm/ 50–60 Hz are suitable parameters for cutting in a HM 650V Vibrating-Blade Microtome (Thermo Scientific). 9. There are different voltage offsets associated with patch-clamp recordings. Another common source arises from the generation of a liquid junction potential (LJP) originated when two electrolyte solutions with different ion concentrations get in contact. It occurs due to the unequal rate of diffusion of ions from one solution to the other, resulting in a potential difference between the two solutions. If the LJP is significantly high, it can interfere with the electrochemical-based measurements of the amplifier. Then, in addition to the initial current offset cancelation, a possible LJP must be considered when analyzing the final recordings. For instance, in our hands, by using intracellular solutions based on K-salts, the LJP was negligible; however, Cs-based solutions present a LJP of around -8 to 10 mV. For a further detailed discussion of this subject, please check ref. 23. 10. The corpus callosum (CC) is a common region to perform OPC whole-cell recordings in the neonatal mouse. However, in the adult, almost all oligodendroglia present in the CC correspond to mature OLs, and the tissue exhibits dense myelinated axon bundles, making difficult to perform patch-clamp recordings. A recommendation to work in this region is to enter to the tissue with a high positive pressure at the pipette tip while approaching the cell. A possible pitfall of this procedure is that the target cell can be displaced by this high positive pressure. To overcome this problem, once the target cell is identified, the micropipette tip should be quickly (few seconds) placed over its cell body just before releasing the pressure. This procedure generally produced a “fishing-like” effect where the cell is attracted to the tip of the pipette due to the sudden negative pressure generated by releasing the positive pressure (in occasions this is enough to produce the gigaseal).

Electrophysiological Recordings of Oligodendroglia in Adult Mouse Brain Slices

281

11. In occasions the characteristic sodium current of OPCs might not be observed directly during the recording due to a low signal-to -noise ratio (i.e., cell culture [16]). Nevertheless, in these cases, it is recommended to perform an offline analysis of the amplitude of this conductance after applying a leak subtraction protocol [16]. 12. Prizmatix , Artifex Engineering and Thorlabs are suitable companies to provide both light sources and path chords to perform photostimulation. 13. Most common optogenetic protocols that deliver light intensities are in a range of 0.5–10 mW/mm2 at the tip of the fiber (see refs. 17, 18). Power at the tip can be tested by using fiber optic power meters with Internal Sensor (Thorlabs or Artifex companies). 14. Possible artifact sources originate in photoelectric or photothermic effects on the electrode. Commonly, they can be reduced or eliminated by simply reducing the power of the light source (i.e., light intensity).

Acknowledgments FCO and MCA were supported by the ECOS180013 grant. FCO was supported by Fondo Nacional de Desarrollo Cientı´fico y Tecnolo´gico (FONDECYT) 1210940. M.C.A was supported by grants Fondation pour la Recherche Me´dicale (FRM, EQU202103012626), ANR under the frame of the European Joint Programme on Rare Diseases (EJP RD, project no. ANR-19-RAR4-008-03), and ANR CoLD (ANR, ANR-20CE16–0001-01). M.C.A. is a CNRS (Centre National de la Recherche Scientifique) investigator. We specially thank to Sebastian Vejar for his work designing the chambers to storage the brain slices. References 1. Henn RE, Noureldein MH, Elzinga SE et al (2022) Glial-neuron crosstalk in health and disease: a focus on metabolism, obesity, and cognitive impairment. Neurobiol Dis 170: 105766. https://doi.org/10.1016/J.NBD. 2022.105766 ˜ oz M, Va´zquez2. Mata-Martı´nez E, Dı´az-Mun Cuevas FG (2022) Glial cells and brain diseases: inflammasomes as relevant pathological entities. Front Cell Neurosci 0:314. https:// doi.org/10.3389/FNCEL.2022.929529 3. Patel DC, Tewari BP, Chaunsali L, Sontheimer H (2019) Neuron–glia interactions in the

pathophysiology of epilepsy. Nat Rev Neurosci 205(20):282–297. https://doi.org/10.1038/ s41583-019-0126-4 4. Bergles DE, Jabs R, Steinh€auser C (2010) Neuron-glia synapses in the brain. Brain Res Rev 63:130–137. https://doi.org/10.1016/ J.BRAINRESREV.2009.12.003 5. Habermacher C, Angulo MC, Benamer N (2019) Glutamate versus GABA in neuron–oligodendroglia communication. Glia 67:2092– 2106. https://doi.org/10.1002/GLIA. 23618

282

Coram Guevara et al.

6. Maldonado PP, Angulo MC (2015) Multiple modes of communication between neurons and oligodendrocyte precursor cells. Neuroscientist 21:266–276. https://doi.org/10. 1177/1073858414530784 7. Fro¨hlich N, Nagy B, Hovhannisyan A, Kukley M (2011) Fate of neuron–glia synapses during proliferation and differentiation of NG2 cells. J Anat 219:18–32. https://doi.org/10.1111/J. 1469-7580.2011.01392.X 8. Yang QK, Xiong JX, Yao ZX (2013) NeuronNG2 cell synapses: novel functions for regulating NG2 cell proliferation and differentiation. Biomed Res Int 2013. https://doi.org/10. 1155/2013/402843 9. Ishibashi T, Dakin KA, Stevens B et al (2006) Astrocytes promote myelination in response to electrical impulses. Neuron 49:823–832. https://doi.org/10.1016/J.NEURON.2006. 02.006 10. Yuan X, Chittajallu R, Belachew S et al (2002) Expression of the green fluorescent protein in the oligodendrocyte lineage: a transgenic mouse for developmental and physiological studies. J Neurosci Res 70:529–545. https:// doi.org/10.1002/JNR.10368 11. Shibata S, Yasuda A, Renault-Mihara F et al (2010) Sox10-Venus mice: a new tool for real-time labeling of neural crest lineage cells and oligodendrocytes. Mol Brain 3:1–14. https://doi.org/10.1186/1756-6606-3-31/ TABLES/1 12. Ziskin JL, Nishiyama A, Rubio M et al (2007) Vesicular release of glutamate from unmyelinated axons in white matter. Nat Neurosci 10: 321–330. https://doi.org/10.1038/NN1854 13. Hamilton TG, Klinghoffer RA, Corrin PD, Soriano P (2003) Evolutionary divergence of platelet-derived growth factor alpha receptor signaling mechanisms. Mol Cell Biol 23: 4013–4025 14. Ou-Yang MH, Xu F, Liao MC et al (2015) The N-terminal region of myelin basic protein reduces fibrillar amyloid-β deposition in Tg-5xFAD mice. Neurobiol Aging 36:801– 8 1 1 . h t t p s : // d o i . o r g / 1 0 . 1 0 1 6 / J . NEUROBIOLAGING.2014.10.006 15. Ting JT, Lee BR, Chong P et al (2018) Preparation of acute brain slices using an optimized

N-methyl-D-glucamine protective recovery method. J Vis Exp 2018. https://doi.org/10. 3791/53825 16. Wake H, Ortiz FC, Ho Woo D et al (2015) ARTICLE Nonsynaptic junctions on myelinating glia promote preferential myelination of electrically active axons. Nat Commun. https://doi.org/10.1038/ncomms8844 17. Habermacher C, Manot-Saillet B, Ortolani D et al (2021) Optogenetics to interrogate neuron-glia interactions in pups and adults. Methods Mol Biol 2191:135–149. https:// doi.org/10.1007/978-1-0716-0830-2_9 18. Ortolani D, Manot-Saillet B, Orduz D et al (2018) In vivo optogenetic approach to study neuron-oligodendroglia interactions in mouse pups. Front Cell Neurosci 12:477. https://doi. org/10.3389/FNCEL.2018.00477/BIBTEX 19. Sahel A, Ortiz FC, Kerninon C et al (2015) Alteration of synaptic connectivity of oligodendrocyte precursor cells following demyelination. Front Cell Neurosci 9:1–12. https:// doi.org/10.3389/fncel.2015.00077 20. Schmitt FO, Bear RS (1936) The optics of nerve myelin. J Opt Soc Am 26(5):206–212. https://doi.org/10.1364/JOSA.26.000206 21. Lin SC, Bergles DE (2004) Synaptic signaling between GABAergic interneurons and oligodendrocyte precursor cells in the hippocampus. Nat Neurosci 7:24–32. https://doi.org/10. 1038/NN1162 22. Chittajallu R, Aguirre A, Gallo V (2004) NG2-positive cells in the mouse white and grey matter display distinct physiological properties. J Physiol 561:109–122. https://doi. org/10.1113/JPHYSIOL.2004.074252 23. Neher E (1995) Voltage offsets in patch-clamp experiments. In: Sakmann B, Neher E (eds) Single-channel recording. Springer, Boston. https://doi.org/10.1007/978-1-44191229-9_6 24. Kukley M, Nishiyama A, Dietrich D (2010) The fate of synaptic input to NG2 glial cells: neurons specifically downregulate transmitter release onto differentiating oligodendroglial cells. J Neurosci 30:8320–8331. https://doi. org/10.1523/JNEUROSCI.0854-10.2010

Chapter 13 Synaptic Integration at Neuron-OPC Synapses Wenjing Sun Abstract The synapses between neurons and oligodendrocyte precursor cells (OPCs) in the central nervous system (CNS) have similar ultrastructures as neuron-neuron synapses and follow the quantal transmission rule. Since OPCs have highly ramified processes, each single OPC can receive up to 100 synaptic contacts with neighboring axons. Therefore, addressing how OPCs integrate synaptic inputs from different axons will significantly advance our understanding of the functional implication of these synapses. This chapter describes experimental and computational methodologies to investigate synaptic integration in OPCs. As neuronal activity-induced downstream signaling often occurs locally within the thin processes of OPCs, we will emphasize the methods to explore how OPCs integrate synaptic inputs within local compartments of the thin processes. Key words OPC, Synaptic integration, Voltage-gated ion channels, Glutamate uncaging, Ca2+ imaging, Computational simulation

1

Introduction Oligodendrocyte precursor cells (OPCs) are the only glial cells forming direct synaptic contact with neurons. The physiological properties of neuron-OPC synaptic transmission are similar to classic neuron-neuron synaptic transmission. Taking glutamatergic neuron-OPC synapses as an example, the firing of presynaptic neurons releases glutamate via synaptic vesicle release and then evokes AMPA receptor-mediated inward current in the postsynaptic OPCs. The postsynaptic current displays fast rising and decaying kinetics, similar to the kinetics of neuronal glutamatergic synaptic currents. The release of glutamate to OPCs also follows the quantal release rule [1–3]. Neuron-OPC synaptic communication is believed to provide the information required for downstream signaling cascades to initiate activity-dependent myelination. Like neurons, OPCs express various voltage-gated ion channels (VGCs), particularly Na+ channels [4–7]. Although OPCs generally do not fire action potentials [1, 6, 8–10] (but see [5]), it is

Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5_13, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

283

284

Wenjing Sun

intriguing to investigate how postsynaptic OPCs integrate neuronal synaptic inputs into their membrane potential and how various VGCs shape the postsynaptic potential. Moreover, it is crucial to understand the functional output or downstream signaling pathways of synaptic integration in OPCs. This chapter will discuss available experimental and computational tools to study synaptic integration in OPCs.

2 Electrophysiological and Pharmacological Approaches to Study Somatic Integration OPCs can have up to ~100 synaptic contacts with neurons [2]. Simultaneous firing of multiple presynaptic boutons may produce substantial postsynaptic depolarization in OPCs. Indeed, previous studies showed that maximal synaptic stimulation intensity could evoke up to 250 pA excitatory postsynaptic current (EPSC) [2] and substantially depolarize the postsynaptic OPC membrane to a range that is sufficient to activate various VGCs [11]. Therefore, it is natural first to ask whether the summation of synaptic inputs at the soma will activate VGCs and how VGCs will shape the amplitude and kinetics of somatic excitatory postsynaptic potential (EPSP). To trigger the synaptic vesicle release, one conventional method is to stimulate the presynaptic axons with an electrode. The stimulation-evoked action potential will arrive at the presynaptic boutons and activate voltage-gated Ca2+ channels. The fast Ca2+ entry will then initiate a Ca2+-dependent vesicle fusion, leading to the release of vesicle-contained neurotransmitters into the synaptic cleft. Unfortunately, action potential-evoked vesicle fusion is not a guaranteed event, and the success rate (release probability) is less than 30% in some brain regions [2, 12], which means the same strength of electrical stimuli cannot result in the vesicle release from the exact same group of presynaptic terminals all the time. Moreover, repetitive stimulation might deplete the presynaptic vesicle pool and reduce the release probability. Therefore, this technique can neither precisely control the number of released vesicles nor evoke vesicle release from the same synapses repetitively and is not suited to dissect the contribution of different VGCs to shape EPSP quantitatively. To precisely control the magnitude and timing of postsynaptic depolarization, we can instead inject current in the EPSC waveform to mimic the current passing through the postsynaptic glutamate receptors and record the response potential in OPCs. Since this response resembles how the postsynaptic membrane potential alters upon glutamate receptor activation, it can be considered a “mock EPSP.” Here, we will describe using this methodology to study how voltage-gated Na+ and K+ channels shape somatic EPSP in OPCs.

Synaptic Integration at Neuron-OPC Synapses

285

2.1 Generating EPSC Waveform Template

The EPSC waveform can be derived from miniature EPSC recordings in OPCs. To do so, we patch OPCs in the voltage-clamp mode and hold the membrane potential at -85 mV (close to the resting membrane potential of OPCs). Miniature EPSCs can be recorded with 1 μM tetrodotoxin (TTX) in the bath solution to block the action potential-driven synaptic vesicle release. To increase the frequency of spontaneous vesicle release, we usually add 100 μM of ruthenium red to the bath solution [2, 3]. Since the miniature EPSC represents the postsynaptic conductance change evoked by a single glutamate-filled vesicle release, the waveform will be considered a quantal EPSC waveform template. To create a quantal EPSC template without any baseline recording noise, we average 10–20 detected miniature EPSC events and fit the average event with a double exponential function defining both the rise and decay time constants. The amplitude of the single quantal release can be set to match the miniature EPSC recording. When we scale up the EPSC waveform amplitude by multiplying the single quantal template with an integral number, Q, the scaled EPSC current can then represent the number Q of vesicles released by the stimulus.

2.2 Investigating the Recruitment of Voltage-Gated Ion Channels in Mock EPSP

For mock EPSP recordings, OPCs should be patched in currentclamp mode, and the membrane potential should be held at the same level (e.g., -85 mV) across different cells. Currents in EPSC waveform with increasing amplitudes are injected through the patch pipette, and somatic mock EPSPs are recorded. To explore whether somatic EPSP with fast kinetics readily recruits VGCs in OPCs, we need to compare the recorded mock EPSP to a passive response without activating any VGCs. Although such a passive response can be recorded by pharmacologically blocking all known VGCs expressed by OPCs, the washing-in procedure often takes more than just several minutes. The physiological properties of the recorded cell, such as the cell capacitance and input resistance, could have changed during the washing-in period. Therefore, we recommend generating a real-time theoretical passive response for each recording. In each recording, we can inject a small inverted EPSC waveform (e.g., Q = -5 or -10) to produce a passive hyperpolarizing voltage response (typically with an amplitude of ~5 mV). This passive response can then be scaled up to the same Q number as the recorded mock EPSP to create the corresponding passive EPSP (Fig. 1a, b). In our previous study, we found that when we gradually increased the injected EPSC waveform amplitude, we observed an increasing shortening of mock EPSP duration compared to the theoretical passive EPSP, clearly indicating the activation of VGCs [11] (Fig. 1b). To dissect the individual contribution of VGCs to the shortening of mock PSPs, we can bath apply different VGC blockers and execute the same EPSC waveform current injection protocols (e.g., 1 μM TTX to block Na+ currents, 10 mM TEA to block delayed-rectifier K+ currents, and 4 mM 4-AP to block the A-type K+ currents).

286

Wenjing Sun

Fig. 1 Dissecting the contribution of VGCs in somatic synaptic integration. (a) The scheme shows the experimental paradigm for mock EPSP recording. We patch OPCs in current-clamp mode and inject EPSC waveform (grey line in the inset box) to record somatic mock EPSP. Note that a small inverted EPSC waveform injection is included in each recording trial to obtain the passive response. (b) An example recording shows that larger mock EPSPs recruit VGCs, which is indicated by the difference between the recorded mock EPSP (black line) and the scaled-up passive response (dashed blue line) 2.3 Analyzing Mock EPSP Data across Different OPCs

The cell membrane acts as both a resistor (R) and a capacitor (C). At the steady state, the amplitude of membrane potential response (Vm) follows Ohm’s law and is proportional to the injected current (I) strength (Vm = I * R). However, in the phases of onset and offset of current injection, the capacitance of the cell determines how rapidly the cell membrane reaches the steady state. Both the rising and decaying of EPSCs are very fast. Therefore, both the membrane resistance and capacitance are essential in determining EPSP size. When analyzing the mock EPSP data across different OPCs/groups, we need to remember that the membrane resistance (Rin), but not the cell capacitance, varies greatly among OPCs [4, 7, 11]. The same strength of EPSC waveform stimuli will result in variable mock EPSP response sizes, mainly due to the variability in Rin. Therefore, the intensity of EPSC stimuli is better to be

Synaptic Integration at Neuron-OPC Synapses

287

normalized as the product of Q * Rin instead of Q itself. For instance, injecting 100 Q into an OPC with an input resistance of 250 MΩ yields a similar response to injecting 50 Q into an OPC with 500 MΩ input resistance. 2.4 Consideration of the Driving Force Alteration

When neurotransmitters bind to the postsynaptic receptors, the size of resulting postsynaptic currents not only depends on the conductance of activated receptors (g) but also on the driving force, which is the difference between the membrane potential of the postsynaptic cell (V) and the reversal potential of the ions passing through the receptors (Erev): I = g * (V - Erev). Therefore, for inward currents, when the membrane potential is depolarized, the driving force (V - Erev) becomes smaller, and the required stimuli strength to reach the same magnitude of depolarization becomes larger. In the case of AMPA receptors, the reversal potential is approximately 0 mV. As the resting/holding membrane potential of OPCs is ~ -85 mV, we omit the relatively small change in Q due to the driving force alteration. For GABAergic synaptic inputs, due to high intracellular Cl- concentration, the activation of GABA-A receptors will depolarize the postsynaptic OPC membrane instead of hyperpolarizing. The reversal potential for GABAA receptors in OPCs is around -40 mV [13, 14], which means that the current stimuli needed to depolarize the cell membrane will drastically increase once the membrane potential is approaching -40 mV. The methodology described above using conventional current-clamp recording will not be accurate for quantitatively analyzing the relationship between synaptic stimuli strength and GABAergic PSPs. Instead, dynamic clamp recording that evaluates and simulates the synaptic currents for injection in real time during recording is highly recommended. To record under the dynamic clamp mode, we still use the same current-clamp amplifier. However, we will set up a separate data acquisition board and the connected computer to collect the real-time membrane potential and the intended GABAergic stimuli conductance waveform (g). The computer will compute the driving force (V - Erev) and simulate the resulting stimulus current waveform in a real-time fashion (I = g * (V - Erev)) and inject the simulated current stimuli back into the recorded cell. It is obvious that the sampling rate of the data acquisition board and the computation rate for the computer need to be sufficiently rapid to simulate the conductance change. ITC-18 data acquisition interface or National Instrument DAQ PCI boards generally have sufficient sampling rates. Several software packages are available for dynamic clamp recording (reviewed by [15]). Igor Pro-based recording package mafPC [16] also contains a dynamic clamp interface.

288

Wenjing Sun

3 Utilizing Two-Photon Glutamate Uncaging to Investigate Local Integration of Synaptic Inputs VGCs are not only expressed at the OPC soma but also possibly along the processes [11]. Meanwhile, emerging evidence suggests that OPCs likely receive and integrate neuronal inputs and then trigger downstream signaling cascades locally within the processes. Understanding how locally expressed VGCs contribute to synaptic integration within a spatially restricted compartment (process segments) is essential to deciphering how neuronal activity instructs myelin formation. The two-photon uncaging glutamate technique has been widely utilized in the neuronal dendritic integration field. This technique can generate a point-source release of glutamate within the two-photon excitation volume through brief photolysis of caged-glutamate, thereby effectively mimicking the vesicular release of glutamate (Fig. 2a, b, [17]). Although OPCs do not have spine structures indicating synapses or clusters of neurotransmitter receptors, they express clusters of AMPA receptors along processes (Fig. 2b, [11]). Here, we will describe how to employ two-photon glutamate uncaging techniques in OPCs to investigate the local integration of synaptic inputs. 3.1 CagedGlutamate Compound and the Application to Acute Brain Slices

4-Methoxy-7-nitroindolinyl-glutamate (MNI-glutamate) is the first and most widely used caged-glutamate compound for two-photon glutamate uncaging (Fig. 2a, [17]). MNI-glutamate is stable in physiological solutions and has a high two-photon crosssection at 730 nm (0.06 GM) with relatively low phototoxicity (reviewed by [18]). It is commercially available through Tocris (Catalog # 1490). Due to the high cost of MNI-glutamate ($ 914 for 50 mg) and the relatively high concentration needed for two-photon uncaging (mM range), most labs cannot afford to continuously bath apply MNI-glutamate with a running peristaltic

Fig. 2 The illustration of two-photon glutamate uncaging technique. (a) MNI-caged-glutamate is photolyzed by the near coincident absorption of two IR photons at 720 nm and then converted to glutamate. (b) When placing the uncaging spot along a process of OPC, a brief uncaging pulse (0.6–1 ms, 720 nm) can release a small amount of free glutamate within the two-photon excitation volume and activate the AMPA receptors expressed along OPC processes. The recorded uncaging voltage response (uEPSP) is similar to a synaptic event response size

Synaptic Integration at Neuron-OPC Synapses

289

pump. Several different strategies have been used for MNI-glutamate application. Our previous studies measured the volume of artificial cerebrospinal fluid (aCSF) within the recording chamber to be approximately 1.5 mL. For the uncaging experiments, we stopped the aCSF perfusion and slowly added 300 μL of 30 mM MNI-glutamate stock solution to the recording chamber using a P200 pipette to reach the final concentration of 5 mM [11, 19]. We usually wait 5–8 min after adding the stock MNI-glutamate to ensure an equal final concentration across the whole chamber. With bath application, we assume that the MNI-glutamate concentration at any depth within the brain slice equals 5 mM. However, the photolysis efficiency will vary due to laser power attenuation at various depths (see solution below). In our hands, the recording condition remains stable for up to 30 min with no aCSF perfusion. Alternatively, one can apply MNI-glutamate locally through puff application by placing a puffing pipette (2 h. 3. Stereotaxic injection of rabies virus is performed 3 days after the induction of TVA expression. This allows the initial TVA labeling of starter OPCs (see Note 3). 3.2 Injection of Rabies Virus into the Brain

1. Anesthetize the mouse using isoflurane (in oxygen: induce 3–4% and maintain 1–2%). Place ophthalmic ointment on both eyes. Administer 2 mg/kg Bupivacaine and 5 mg/kg Carprofen subcutaneously. 2. Remove hair from the top of the head. Aseptically prepare the shaved skin by disinfecting the surgical area by alternating between betadine and 70% ethanol for three cycles. 3. Place mouse on the stereotaxic apparatus atop a clean absorbent surface and thermometer monitored heat source to prevent hypothermia. Keep monitoring the mouse for heart rate and absence of a withdrawal reflex during the procedure. 4. Before making an incision, check anesthetic depth. Using aseptic technique, make a midline incision on top of the skull with small surgical scissors. 5. Separate the subcutaneous and muscle tissue, and gently retract the skin to visualize the structures below. 6. Gently scrape clean the bregma and lambda areas using a small bone scraper; keep the skull moist with sterile saline applied with a sterile cotton swab. Adjust the incisor screw to make the head level so that the bregma and lambda are equal; level the head horizontally in the caudal-to-rostral direction. 7. Go to the desired stereotaxic coordinates and perform craniotomy over the target injection site using a handheld drill and a 1–2 mm sterile burr with slight downward pressure. Stop when bone is thin, and the blood vessels and dura become clearly visible (see Note 4). This will generate a small hole (1–2 mm diameter) in the skull to conveniently insert the Hamilton syringe.

Mapping Synaptic Inputs to Oligodendroglial Cells Using In Vivo. . .

307

8. Stabilize a Hamilton syringe on the surgical rig injector, and withdraw ~1 μL of virus. Squirt a small amount of virus to make sure the injector is working and there are no bubbles. Using the Hamilton neuro syringe, deliver 500 nL of the modified rabies virus at the rate of 0.1 μL/min. 9. At the completion of infusion, allow the Hamilton needle to remain in place for a minimum of 5 min, then manually withdrawn slowly to minimize backflow. 10. Following successful injection of the virus, close the incision by sealing the scalp with sutures. Add Neo-Predef over the incision. 11. Remove the mouse from the stereotaxic frame. Postoperatively, hydrate the animal by a 1 cc subcutaneous injection of lactated ringers. Allow to recover on a temperature controlled (37 °C) heating pad, and monitor the animal for heart rate and respiration until it achieves an ambulatory recovery. 12. Once the animal achieves ambulatory recovery, return to the home cage. About 24 h after the operation, administer 5 mg/ kg Carprofen subcutaneously and every 24 h as needed by the pain assessment. Monitor for signs of pain and distress such as difficulty in breathing, hunching, ruffled coat hair, dehydration, loss of weight, or infection at surgery site. Carprofen will not be necessary any more if the animal does not seem in pain. Add Neo-Predef over the incision until the skin heals to prevent infection. 3.3

Tissue Analysis

1. Euthanize the mice by transcardial perfusion with 10 mL of ice-cold PBS, followed by 10 mL of ice cold 4% paraformaldehyde (see Note 5). 2. Remove the brain and store in 4% paraformaldehyde overnight at 4 °C. 3. Switch the brain into 30% sucrose solution in PBS. 4. The brain sinks to the bottom of sucrose solution when dehydrated and is ready for sectioning. 5. Section the embedded brain at 40 μm on a microtome, and collect floating sections (see Note 6). 6. Rabies infected starter OPCs will be GFP+ in the brain region injected (Fig. 2a, b). For identification of this starter OPC population, anti-GFP and anti-Pdgfra co-labelling is used (Fig. 2b). 7. The first-order input neurons are also labelled with GFP and nuclear stain DAPI (Fig. 2a, c).

308

Belgin Yalc¸ın and Michelle Monje

Fig. 2 (a) Rabies tracing identifies synaptic connectivity of corpus callosum OPCs. Neuronal inputs arise from functionally connected cortical and thalamic areas. (b) An example of EGFP+ starter OPC identified by oligodendroglial marker Olig2 and OPC specific marker Pdgfra; scale bar = 10 μm. (c) EGFP+ input neurons arising from medial prefrontal cortex (mPFC); scale bar = 20 μm. (d) Neuronal input/starter OPC ratio shows linear relationship (R2 = 0.8732; slope = 23.46 ± 2.8 standard error). (Adapted from Mount et al. [10])

Perform immunostaining for anti-GFP to achieve a stable fluorescence during the microscopy imaging of the brain sections. To identify the neural subtypes of the input neurons, co-labelling with neuronal markers should be performed. 3.4 Mapping the Synaptic Network

1. A stereomicroscope or a lower magnification objective on a confocal microscope is used to acquire tiled images of each brain section. A 10× objective with image acquisition of 1024 × 1024 pixels works well for tiling a section and allows identifying cell types for analysis. A 1-in-6 sampling rate, one section per 240 μm, for identifying input neurons provides a good overview (see Note 7).

Mapping Synaptic Inputs to Oligodendroglial Cells Using In Vivo. . .

309

2. For identification of brain regions to aid neuronal quantification, acquired images are manually registered to the closest available section from the Allen Brain Mouse Reference Atlas using DAPI fluorescence of the section outline and major neuroanatomical structures to guide fitting. 3. Cell counting for GFP+ neurons on the identified brain regions provide the input neuron number for each brain region. In our experiments, summing all inputs identified to the white matter OPCs, viral input/starter ratios in this context were approximately 23 (slope of linear regression 23.46 ± 2.8 standard error, Fig. 2d), with neuronal input cells clearly identifiable by morphology and GFP expression (Fig. 2c). 4. As the sampling done in 1-in-6 40 μm-thick brain sections, total cell count estimates are derived by multiplying the number of counted cells by 6.

4

Notes 1. The size of the starter OPC population can be adjusted by changing the tamoxifen injections or the age injections are performed. For labelling more starter OPCs, tamoxifen can be injected for up to 5 days, and younger mice can result in more OPC labelling as the cells are proliferating more frequently. 2. Cre-negative mice, TVA-negative mice, or a group of gp4-TVA; Pdgfra-CreERT mice with no tamoxifen injections can be used as control group for testing the specificity of the virus. Alternatively, immunohistochemistry for different cell types can be employed with the experimental groups for validation of the infected starter cell type. 3. Time interval between the tamoxifen administration and injection of rabies should be determined depending on the age of mice. Younger mice exhibit higher OPC proliferation rate, so rabies tracing should be performed before the starter cells start to differentiate. Increasing this interval in younger mice might in turn cause a mismatch between the starter cell to input neuron ratio as those starter OPCs might progress along the lineage towards oligodendrocytes. For older mice, this time interval can be arranged longer because of slower cell turn over. 4. For mapping inputs into OPCs in other brain regions, these coordinates should be modified. Depending on the OPC density in the brain region of interest, tamoxifen injections can be increased to label more or fewer starter OPCs.

310

Belgin Yalc¸ın and Michelle Monje

5. Time for virus spread and euthanasia should be determined depending on the distance of expected afferents. Usually 5- to 7-day incubation works optimal, because during longer spread times, rabies virus become toxic to infected cells and might skew the results. 6. Generally, 40 nm sections provide nice coverage for many brain regions for OPC connectivity tracing. However, the brain section thickness can be adjusted depending on both the mapped region and the distance of potential inputs. The brain sections should encompass both anterior and posterior portions of the injection coordinates to include all the starter cells and local inputs. The needle site can be apparent on the brain surface, which might also help as a guide. Sectioning the whole brain for identifying unexpected long distance input neurons is recommended. 7. Alternative light microscopy techniques could be used such as 3D light sheet microscopy for analysis.

5

Troubleshooting No labelling present: Confirm the TVA expression in the starter cells. Increasing tamoxifen injections is also an option to expand the TVA-expressing starter cell population. Cre-negative control shows extensive labeling: Rabies virus can be leaky causing unspecific labelling. Contact the virus core for strain confirmation and try another batch with confirmed EnvA modification. No long-range input identification: Consider increasing the time for spread of the virus after the intracranial injections. Backflow of virus during injections: Rabies virus mixture can be viscous; try injecting over longer time. Starter cells are limited at a single point; not enough inputs identified: This could be an experimental preference for tracing of a very spatially specific cell population. If spatial expansion is needed for starter cell labelling, viral stock can be diluted and more volume of virus can be injected. This will achieve a bigger injection site without changing the total amount of virus injected.

Mapping Synaptic Inputs to Oligodendroglial Cells Using In Vivo. . .

6

311

Conclusions Monosynaptic rabies tracing is a powerful tool for identifying both the local and long-range synaptic connectivity of defined cell populations. Even though the labelling strategy can be up- or downscaled, this is a population level analysis tool for mapping networks. Therefore, it does not identify connections of a single OPC, but rather it allows mapping afferents into a spatially and genetically restricted OPC populations. Moreover, in combination with other neuroscience tools, these inputs can be specifically accessed and leveraged for optogenetics or further gene expression studies. Further development of monosynaptic rabies tracing with additional tools to target input neurons will allow elucidating the neuron-toOPC synaptic connectivity in physiological and pathological conditions and its potential role in regulating myelin plasticity.

Acknowledgments The authors gratefully acknowledge support from the National Institute of Neurological Disorders and Stroke (R01NS092597 to M.M.), NIH Director’s Pioneer Award (DP1NS111132 to M. M.), National Cancer Institute (P50CA165962, R01CA258384, U19CA264504), Robert J. Kleberg, Jr. and Helen C. Kleberg Foundation (to M.M.), Cancer Research UK (to M.M.). References 1. Bergles DE, Roberts JDB, Somogyl P, Jahr CE (2000) Glutamatergic synapses on oligodendrocyte precursor cells in the hippocampus. Nature 405:187–191 2. Lin SC, Bergles DE (2004) Synaptic signaling between GABAergic interneurons and oligodendrocyte precursor cells in the hippocampus. Nat Neurosci 7:24–32 3. Ziskin JL, Nishiyama A, Rubio M, Fukaya M, Bergles DE (2007) Vesicular release of glutamate from unmyelinated axons in white matter. Nat Neurosci 10:321–330. https://doi.org/ 10.1038/nn1854 4. De Biase LM, Nishiyama A, Bergles DE (2010) Cellular/molecular excitability and synaptic communication within the oligodendrocyte lineage. J Neurosci 30:3600. https://doi.org/ 10.1523/JNEUROSCI.6000-09.2010 5. Kukley M, Capetillo-Zarate E, Dietrich D (2007) Vesicular glutamate release from axons in white matter. Nat Neurosci 10:311–320 6. Lundgaard I, Luzhynskaya A, Stockley JH, Wang Z, Evans KA, Swire M, Volbracht K,

Gautier HOB, Franklin RJM, FfrenchConstant C, Attwell D, Ka´rado´ttir RT (2013) Neuregulin and BDNF induce a switch to NMDA receptor-dependent myelination by oligodendrocytes. PLoS Biol 11:e1001743. https://doi.org/10.1371/journal.pbio. 1001743 7. Mangin JM, Kunze A, Chittajallu R, Gallo V (2008) Satellite NG2 progenitor cells share common glutamatergic inputs with associated interneurons in the mouse dentate gyrus. J Neurosci 28:7610–7623 8. Lin SC, Huck JHJ, Roberts JDB, Macklin WB, Somogyi P, Bergles DE (2005) Climbing fiber innervation of NG2-expressing glia in the mammalian cerebellum. Neuron 46:773–785 9. Mu¨ller J, Reyes-Haro D, Pivneva T, Nolte C, Schaette R, Lu¨bke J, Kettenmann H (2009) The principal neurons of the medial nucleus of the trapezoid body and NG2+ glial cells receive coordinated excitatory synaptic input. J Gen Physiol 134:115–127

312

Belgin Yalc¸ın and Michelle Monje

10. Mount CW, Yalc¸ın B, Cunliffe-Koehler K, Sundaresh S, Monje M (2019) Monosynaptic tracing maps brain-wide afferent oligodendrocyte precursor cell connectivity. eLife 8:1–17 11. Gibson EM, Purger D, Mount CW, Goldstein AK, Lin GL, Wood LS, Inema I, Miller SE, Bieri G, Zuchero JB, Barres BA, Woo PJ, Vogel H, Monje M (2014) Neuronal activity promotes oligodendrogenesis and adaptive myelination in the mammalian brain. Science (80-. ) 344:1252304 12. McKenzie IA, Ohayon D, Li H, Paes de Faria J, Emery B, Tohyama K, Richardson WD (2014) Motor skill learning requires active central myelination. Science (80-. ) 346:318–322 13. Geraghty AC, Gibson EM, Ghanem RA, Greene JJ, Ocampo A, Goldstein AK, Ni L, Yang T, Marton RM, Pas¸ca SP, Greenberg ME, Longo FM, Monje M (2019) Loss of adaptive myelination contributes to methotrexate chemotherapy-related cognitive impairment. Neuron 103:250–265.e8 14. Pan S, Mayoral SR, Choi HS, Chan JR, Kheirbek MA (2020) Preservation of a remote fear memory requires new myelin formation. Nat Neurosci 23:487. https://doi.org/10.1038/ s41593-019-0582-1 15. Steadman PE, Xia F, Ahmed M, Mocle AJ, Penning ARA, Geraghty AC, Steenland HW, Monje M, Josselyn SA, Frankland PW (2020) Disruption of oligodendrogenesis impairs memory consolidation in adult mice. Neuron 105:150–164.e6 16. Mitew S, Gobius I, Fenlon LR, McDougall SJ, Hawkes D, Xing YL, Bujalka H, Gundlach AL, Richards LJ, Kilpatrick TJ, Merson TD, Emery B (2018) Pharmacogenetic stimulation of neuronal activity increases myelination in an axonspecific manner. Nat Commun 9:306 17. Mensch S, Baraban M, Almeida R, Czopka T, Ausborn J, El Manira A, Lyons DA (2015) Synaptic vesicle release regulates myelin sheath number of individual oligodendrocytes in vivo. Nat Neurosci 18:628–630 18. Hines JH, Ravanelli AM, Schwindt R, Scott EK, Appel B (2015) Neuronal activity biases axon selection for myelination in vivo. Nat Neurosci 18:683–689

19. Saleeba C, Dempsey B, Le S, Goodchild A, McMullan S (2019) A student’s guide to neural circuit tracing. Front Neurosci 13:1–20 20. Callaway XEM, Luo L (2015) Monosynaptic circuit tracing with glycoprotein-deleted rabies viruses. J Neurosci 35:8979. https://doi.org/ 10.1523/JNEUROSCI.0409-15.2015 21. Wickersham IR, Lyon DC, Barnard RJO, Mori T, Finke S, Conzelmann K-K, Young JAT, Callaway EM (2007) Monosynaptic restriction of transsynaptic tracing from single, genetically targeted neurons. Neuron 53:639– 647 22. Reardon TR, Murray AJ, Turi GF, Schnell MJ, Jessell TM, Losonczy A, Wirblich C, Croce KR (2016) Rabies virus CVS-N2c strain enhances retrograde synaptic transfer and neuronal viability. Neuron 89:711–724. https://doi.org/ 10.1016/j.neuron.2016.01.004 23. Beier KT, Kim CK, Hoerbelt P, Hung LW, Heifets BD, DeLoach KE, Mosca TJ, Neuner S, Deisseroth K, Luo L, Malenka RC (2017) Rabies screen reveals GPe control of cocaine-triggered plasticity. Nature 549:345– 350 24. Ciabatti E, Gonza´lez-Rueda A, Mariotti L, Morgese F, Tripodi M (2017) Life-long genetic and functional access to neural circuits using self-inactivating rabies virus. Cell 170: 382–392.e14 25. Osakada F, Mori T, Cetin AH, Marshel JH, Virgen B, Callaway EM (2011) NeuroResource new rabies virus variants for monitoring and manipulating activity and gene expression in defined neural circuits. Neuron 71:617–631 26. Chatterjee S, Sullivan HA, MacLennan BJ, Xu R, Hou Y, Lavin TK, Lea NE, Michalski JE, Babcock KR, Dietrich S, Matthews GA, Beyeler A, Calhoon GG, Glober G, Whitesell JD, Yao S, Cetin A, Harris JA, Zeng H, Tye KM, Clay Reid R, Wickersham IR (2018) Nontoxic, double-deletion-mutant rabies viral vectors for retrograde targeting of projection neurons. Nat Neurosci 21(4):638–646. https://doi.org/10.1038/s41593-0180091-7

Chapter 15 In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA Receptors in Oligodendrocyte Lineage Cells Ting-Jiun Chen, Bartosz Kula, and Maria Kukley Abstract Delivery or deletion of genes of interest in the nervous system of animals in vivo is a complex and difficult task. While genetic manipulations can be achieved through multiple techniques and utilizing various tools, producing transgenic animals is time-consuming, inflexible, and sometimes inefficient. Employing a viral gene delivery approach provides a good alternative strategy for targeting the gene of interest in vivo. Retroviruses infect only dividing cells because they enter the nucleus during mitotic breakdown of the nuclear envelope. Hence, a retroviral approach is particularly efficient for the delivery of genes in vivo to the cells, which display high proliferation rates, such as oligodendrocyte precursor cells. This chapter provides experimental details for viral gene delivery to oligodendrocyte precursor cells in vivo and discusses advantages and limitations of the technique. Key words Retrovirus, Oligodendrocyte precursor cells, Oligodendrocytes, Premyelinating oligodendrocytes, Proliferation, Differentiation, Stereotaxic injection

1

Introduction Oligodendrocyte precursor cells (OPCs), also known as NG2 cells, are widespread in the grey and white matter regions throughout the entire central nervous system (CNS) and constitute 2–9% of the total cells in the adult rodent brain [1]. One of the most interesting features of OPCs is that they receive direct fast synaptic input from neurons. Neuronal synapses on OPCs have been demonstrated using several different approaches. Electron microscopic (EM) and/or immunofluorescent analysis showed that axonal membranes are located in close proximity to OPC processes, synaptic vesicles accumulate at the axonal membrane facing the OPC membrane, and type 1 vesicular glutamate transporter (VGLUT1) is present inside the grey and white matter axons at the sites opposing the OPC processes [2–7]. Furthermore, several studies, including

Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5_15, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

313

314

Ting-Jiun Chen et al.

ours, used electrophysiological recordings to demonstrate that neuronal synapses on OPCs in different brain regions are functional, and the excitatory synaptic input is mediated by α-amino3-hydroxy-5-methyl-4-isoxazole propionic acid receptors (AMPARs) expressed in OPCs [2, 5, 7–13], but see [14]. Kinetics and pharmacological properties of AMPAR-mediated currents in OPCs are very similar to those of neuronal excitatory postsynaptic currents (EPSCs). Hence, OPCs represent the second type of glial cells in the CNS (besides the Bergmann glia in the cerebellum), which receive direct synaptic input from neurons. In neurons, AMPARs are tetramers composed of four subunits (GluA1, GluA2, GluA3, and GluA4, also called GluR1–4 or GluRA-D, encoded by four genes gria1–4) [15], and any two GluA dimers can be assembled together in different combinations to form a functional receptor tetramer. Messenger ribonucleic acid (mRNA) and protein for all AMPARs subunits have been detected also in OPCs [16–21]. Of the AMPAR subunits, GluA2 is the most unique, as it can undergo posttranscriptional RNA editing at two sites: the R/G site in the ligand-binding domain 2 (at the position 764, from arginine (R) to glycine (G)), and at the Q/R site located in the second membrane domain (at the position 607, from glutamine (Q) to arginine (R)). Editing determines many important properties of the fully assembled receptor, including kinetic of the channel, single-channel conductance, and Ca2+ permeability [22]. The Q/R site is of special importance as the editing replaces the charge-neutral glutamine (Gln, Q) with positively charged arginine (Arg, R) at a critical position inside the receptor’s channel pore. The positive charge of Arg blocks larger cations from passing through the channel pore; therefore, AMPARs containing the Q/R edited GluA2 subunit are impermeable to Ca2+, while AMPARs lacking the GluA2, or containing the unedited GluA2, are permeable to Ca2+ and Zn2+ [23–26]. As the GluA2 subunit plays a critical role in determining the properties of AMPARs, several studies aimed to define the functional role of AMPARs for the development and behavior of OPCs, using various approaches to manipulate the GluA2 subunit specifically in OPCs. In vivo, ablation of the GluA2 subunit in OPCs in the germline gria3-null mice (gria3 gene encodes for the GluA3 subunit of AMPARs) did not affect proliferation of OPCs but resulted in an acutely diminished density of oligodendrocytes (OLs) at postnatal days 7 and 14 (P7 and P14), which returned to control values later in development [18]. Overexpression of GluA2 subunit specifically in OPCs using a Cre-dependent transgenic mouse line had no effect on the differentiation and proliferation of OPCs at early age but resulted in the increased proliferation of OPCs in adults (postnatal day 75) [18]. Although transgenic mice represent a powerful tool for studying the functional role of proteins and receptors, generating a transgenic mouse line is timeconsuming and expensive. An alternative strategy for studying the

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

315

functional role of a protein is to target it using viral gene delivery approach. This approach has been used to study AMPARs trafficking and potentiation in neurons [27–29]. We have recently used a retroviral gene delivery approach in vivo to introduce acute alterations of AMPARs properties in OPCs [8, 30]. Retroviruses represent a good tool for this purpose, because they exclusively infect actively dividing cells, and OPCs continue to proliferate throughout life [1], while the majority of other cells in the healthy adult brain do not proliferate. In this chapter, we describe this tool for modifying the GluA2 subunit of AMPARs in OPCs: We discuss production of retroviruses, approaches to verify that the desired modifications of the GluA2 subunit have occurred in OPCs, and immunohistochemical procedures for examining the effects of the GluA2 modifications on differentiation and proliferation of OPCs (Fig. 1).

2

Generation of Retrovirus Carrying Modified GluA2 Subunit of AMPARs For generating the recombinant retroviral vector of our interest, we use three plasmid DNAs: (1) a recombinant retroviral vector pRetro-CMV-GFP which expresses green fluorescent protein (GFP) under the CAG promoter (Fig. 2a1). We subclone a modified GluA2 subunit into this retroviral vector (Fig. 2a4; see Subheading 2.1 below); (2) a plasmid containing gag and pol (pCMVgagpol), which expresses the retroviral structure proteins (Fig. 2a2); and (3) a plasmid encoding the viral envelope protein (pCMV-VSVG) (Fig. 2a3).

2.1 Amplification of Plasmids for Retroviral System

If the plasmids are received on the filter paper, the first step is to recover them. The procedure of recovery is the same for all three plasmids. Cut out the circles containing the DNA with a pair of clean scissors and elute the DNA with Tris-EDTA (TE) buffer to get the final concentration of 10 ng/μL of DNA. After waiting for at least 10 min, the eluted DNA can be stored at -20 °C or transformed into the competent cells. Transformation is a method in which a foreign DNA is taken up into bacteria. This process is important for replication of plasmids and further storage. In order to amplify the plasmids for further usage and long-term storage, transform each of the plasmids (50 ng) into SURE 2 Supercompetent Cells separately (Agilent, which is now VWR-Avantor) using heat-shock method according to the manufacturer instructions. SURE 2 competent cells are suitable for cloning unstable DNA with secondary structures which viral genomes often have. They lack components of the pathways that catalyze the rearrangement and deletion of nonstandard secondary and tertiary structures. For further information about these cells, the reader is advised to visit the website

316

Ting-Jiun Chen et al.

Fig. 1 Flow diagram summarizing each step of the procedure for in vivo viral gene delivery to manipulate functional properties of AMPARs in oligodendrocyte lineage cells. The approximate timeline for completing each step is indicated. Note that the timeline could be varied by the experimenter because, in some steps, the sample can be stored in the freezer and the experiment can be continued later

https://us.vwr.com/store/product/24107296/sure-competentcells-agilent-technologies. Next, plate two volumes (20 and 200 μL, ensure single colony can be picked up) of transformation mixtures on the Luria broth (LB) agar plates containing 100 μL/mL ampicillin and incubate the plates at 37 °C for at least 17 h. After that, carefully check the colonies on the LB agar plates, and pick up at least two individual colonies from each plate into 3 mL of LB medium containing 200 μL/mL ampicillin for inoculation (215 revolutions per minute (rpm)), at 37 °C for 12–18 h. To isolate small amounts of plasmid DNA from bacteria, minipreps are carried out by using QIAprep Spin Miniprep Kit (Qiagen)

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

317

Fig. 2 Schematic diagrams illustrating the structure of the retroviral vector, and the subcloning procedure. (a1) The retroviral vector regulates the transcription of the viral genome by using cytomegalovirus (CMV) promoter. The compound promoter chicken-beta actin (CAG) controls the expression of GFP. Various diagnostic restriction enzymes (shown in blue) are used for plasmid validation. Black dash line indicates the size of digested fragment from BamHI and Pmel. Magenta dash line indicates the size of digested fragment from BamHI and NotI. LTR, long terminal repeat; Ψ, viral packing signal; WPRE, Woodchuck hepatitis virus posttranscriptional regulatory element; AmpR, ampicillin resistance. (a2) pCMV-galpol vector expresses the retroviral structure proteins (galpol). Restriction enzymes of EcoRI and Xhol (blue) are used for plasmid validation. Black dash line indicates the size of digested fragment from Xhol. (a3) The pCMV-VSVG encodes viral envelope protein (VSVG). Restriction enzymes of HindIII and EcoRI (blue) are selected for plasmid validation. Black dash line indicates the size of digested fragment from EcoRI. Magenta dash line indicates the size of digested fragment from HindIII and EcoRI. (a4) Newly generated retroviral vector contains the GluA2-insert (Ca2+-permeable GluA2 subunit). The restriction enzymes which verify the orientation of the GluA2-insert are indicated. The size of retroviral vector, GluA2-insert, and WPRE are 6.8 kbp, 3.841 kbp (black dash line), and 0.596 kbp, respectively. Magenta dash line denotes the size of the fragment digested by PvuIII and ClaI, indicating that the GluA2-insert ligates into the recipient vector in the correct orientation. (b1–b4) Schematic drawing shows the procedure of subcloning. (b1) Add restriction enzyme sites to the flank of GluA2-insert from the donor vector by PCR. (b2) Digest the GluA2-insert from the donor vector, and digest the recipient vector, with the restriction enzymes Agel and Pmel. (b3) Ligate the digested GluA2-insert into the recipient vector. (b4) Transform the new vector into the competent cells

318

Ting-Jiun Chen et al.

following the manufacturer instructions and DNA is eluted with nuclease-free water. After isolation of the three plasmids, it is necessary to validate the plasmids by diagnostic restriction digestion before beginning further experiments. For pRetro-CMV-GFP (7.544 kilobase pairs; kbp), we chose the restriction enzymes BamHI, BamHl and PmeI, and BamHI and NotI (Fig. 2a1). They yield linearized plasmid DNA of 7.544 kbp band size, 6.84 and 0.7 kbp band size, and 3.1 and 4.44 kbp band size in agarose gel electrophoresis, respectively. For pCMV-galpol (9.293 kbp), we choose the restriction enzymes EcoRI and XhoI (Fig. 2a2). The EcoRI yields linearized plasmid DNA of 9.293 kbp band size, while the Xhol yields 7.479 and 1.814 kbp band sizes in agarose gel electrophoresis. For pCMV-VSVG (7.58 kbp), we choose the restriction enzymes HindIII and EcoRI (Fig. 2a3). The HindIII yields linearized plasmid DNA of 7.58 kbp band size, while HindIII and EcoRI yield the 1.6, 5.4, and 0.58 kbp band sizes in agarose gel electrophoresis. If the results of the restriction digestion come out as expected, the bacterial glycerol stock solutions for long-term storage can be produced. To get a larger quantity of plasmid DNA for further subcloning and retroviral production (see below), inoculate the bacteria containing plasmid by adding 1 mL miniprep culture into 150 (or 200) mL LB medium containing 200 μL/mL ampicillin (215 rpm at 37 °C for 12–18 h). To isolate larger amounts of plasmid DNA from bacteria, maxipreps are carried out by using QIAfilter Plasmid Maxi Kit (Qiagen) following the manufacturer instructions. Use nuclease-free water to elute the DNA. Validate the plasmids again by diagnostic restriction digest as described above. The plasmid DNA can be stored at -20 °C. Notes and Troubleshooting: • Before the experiment, it is important to check which antibiotic resistance gene is present in the plasmid and to add the specific antibiotic into the LB agar plate and into the LB culture medium (see Fig. 2a). Antibiotic resistance gene on the plasmid is used to ensure that the transformed bacteria containing the plasmid would survive and propagate. Three plasmids described here (pRetro-CMV-GFP, pCMV-gagpol, and pCMV-VSVG) contain the ampicillin gene. Therefore, we included ampicillin into LB agar plate or LB medium. • To make sure that the transformation procedure is working, it is advisable to use the plasmid provided with competent cells as a positive control in the step of plating the cells on the agar plates. Positive control experiment is particularly important if no colonies are growing on the agar plates after transformation.

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

2.2 Subcloning the Gene of Interest

319

Subcloning is a procedure of transferring a DNA sequence from one vector to another vector. Here, we transfer the DNA sequence encoding the Ca2+-permeable GluA2 subunit (called “GluA2insert” in the following text) from the donor vector pCI-CMVEGFPGluA2(R583Q) (gift from Malinow Lab) into the recombinant retroviral recipient vector pRetro-CMV-GFP, to obtain the final product which we call pRetro-CMV-EGFPGluA2(R583Q) (Fig. 2a4). Before starting the subcloning, it is necessary to check which restriction enzyme sites are present at the multiple cloning sites (MCS) where DNA is inserted (see below: (3) Ligation) in the recipient vector. Our recipient vector pRetro-CMV-GFP contains multiple restriction enzyme sites at the MCS, but we choose the restriction enzymes AgeI at the 5′ end and PmeI at the 3′ end on the recipient vector because neither cuts the sequence of donor vector. However, the restriction enzyme sites of AgeI and PmeI are not present in the MCS of our donor plasmid pCI-CMV-EGFPGluA2(R583Q), which makes it important to ligate them into the recipient vector. Therefore, our strategy starts from adding these restriction sites to the flank of GluA2-insert with polymerase chain reaction (PCR). The steps of subcloning are as follows: (1) Amplify GluA2insert and add restriction enzyme sites to the flanks of GluA2insert by PCR. (2) Digest GluA2-insert and recipient vector. (3) Ligate the GluA2-insert into the digested recipient vector and obtain pRetro-CMV-EGPFGluA2(R583Q). (4) Transform the ligated vector pRetro-CMV-EGPFGluA2(R583Q) into the competent bacteria cells (Fig. 2b1–b4). 1. Amplify GluA2-insert and add restriction enzyme sites to flank of GluA2-insert and isolate GluA2-insert by PCR We use the following PCR primers to add the restriction enzyme sites: forward (from 5′ to 3′) (AgeI): TTACCGGTACGACTCACTATAGGCTAGAACTAG; reverse (from 5′ to 3′) (PmeI): TGTTTAAACCCAAGGCCTGCATGCACTGCTTTG (the underlined nucleotides represent the sequence of the enzyme, while the remaining nucleotides represent the complementary sequence of the GluA2-insert). It is critical to have primers long enough that the restriction enzymes can cut the blunt ends (Pmel for example) without problems. The PCR reagents (New England Biolabs) are mixed with 20 ng of donor plasmid pCI-CMV-EGFPGluA2(R583Q), the mixture is placed into the thermocycler, and the following thermal cycle parameters are used: 30 s at 98 °C for initial denaturation, 30 cycles of 10 s at 98 °C for denaturation, 30 s at 70 °C for annealing, 2 min at 72 °C for extension, and 10 min at 72 °C for final extension. The expected size of the DNA fragment of the GluA2-inset flanked by the restriction enzyme sequence after PCR is 3.855 kbp. The size of the DNA fragment should be verified by gel electrophoresis (0.7–1% agarose gel).

320

Ting-Jiun Chen et al.

We cut and extract the 3.855 kbp band, corresponding to the DNA fragment of GluA2-inset flanked by the sequence of restriction enzymes sequence, from the gel using the QIAquick Gel Extraction Kit (QIAGEN) following the manufacturer instructions. Subsequently, the purified DNA fragments can be digested by the restriction enzymes directly (see below) or stored at -20 °C. 2. Digest GluA2-insert and recipient vector After extraction of the GluA2-insert from the gel, the GluA2insert (10 μg) and the recipient recombinant retroviral vector pRetro-CMV-GFP (10 μg) are separately digested with both restriction enzymes at 37 °C for 3 h. The purpose of digestion is to generate the “complementary ends” in the GluA2-insert and in the recombinant retroviral vector for the subsequent ligation. After 3 h of enzyme digestion, the digestion reaction is stopped at 65 °C for 20 min. Next, the GluA2-insert is purified by the QIAquick PCR Purification Kit (QIAGEN) according to the manufacturer instructions. The digested retroviral vector, which is now in the linearized form, is incubated with 1 μL of calf intestinal alkaline phosphatase (CIP; New England Biolabs) at 37 °C for 1 h in order to remove the 5′-phosphate groups from DNA and prevent selfligation. Subsequently, electrophoresis is used to verify that the retroviral vector remains in the linearized form with the expected size of 6.8 kbp. If the size is correct, the retroviral vector can be extracted from the gel using the QIAquick PCR Purification Kit (QIAGEN) according to the manufacturer instructions. 3. Ligate the GluA2-insert into the digested recipient vector Finally, we ligate the digested GluA2-insert into the recipient recombinant retroviral vector. The molar ratio of the vector to insert mixture is critical for success, especially for large and complicated plasmids. For the GluA2-insert, we start with the 1:3 and 1:5 vector to insert molar ratio. The mixture of ligation reactions includes the retroviral vector, the GluA2-insert, the 2X Quick Ligation Reaction Buffer, and the Quick T4 DNA Ligase (Quick Ligation Kit, New England Biolabs). The reaction is incubated at room temperature for 10 min followed by chilling on ice to stop the reaction. As a result, we expect that the GluA2-insert is attached to the recombinant retroviral backbone, generating the complete plasmid (10.6 kbp) ready for transformation (see below). We perform the ligation and the transformation on the same day. 4. Transformation: transfer the ligated vector pRetro-CMV-EGFPGluA2(R583Q) into the competent bacteria cells At the last step of subcloning, we transform the newly generated retroviral vector containing GluA2-insert (pRetro-CMVEGFPGluA2(R583Q)), into the Stbl3 competent bacteria cells (Invitrogen). Retroviral vector contains long terminal repeats

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

321

(LTRs) (Fig. 2a1, a4), encoding a reverse transcriptase and an integrase which mediates the integration of plasmids DNA into the host genome. Stbl3 competent cells are designed for the vectors in which DNA contains LTRs. Those competent cells will reduce the frequency of recombination of LTRs; therefore, we choose Stbl3 for plasmid amplification. This procedure is performed immediately after the ligation step, according to manufacturer instructions. We plate two volumes (200 and 20 μL) of the transformation mixture onto the LB agar plates containing ampicillin (100 μL/mL) and incubate the plates at 37 °C for at least 17 h. Next day, we observe the colonies on the LB agar plates and carefully transfer five to ten individual colonies from any plates into 3 mL LB medium containing ampicillin (200 μL/mL) for inoculation (215 rpm at 37 °C for 12–18 h). We then isolate the plasmid DNA with minipreps following the instructions from the manufacturer. Purified pRetro-CMV-EGFPGluA2(R583Q) plasmid DNA is verified by a diagnostic restriction digestion and then by running the digest on the agarose gel. The goal here is to check whether (1) the size of the plasmid is correct, (2) the plasmid is cut into pieces of the expected size, and (3) the orientation of the insert is correct. We utilize PmeI to verify that the whole size is 10.6 kbp, as in a linearized form. We use the same restriction enzymes AgeI and Pmel to cut the plasmid into two pieces of 6.8 and 3.8 kbp. Last, we use PvuII and ClaI to test the GluA2 insert orientation (Fig. 2a4). If the insert orientation is correct, the expected fragments are 8.27 and 2.35 kbp; otherwise, if the fragments are 7.96 and 2.66 kbp, it indicates incorrect orientation of the insert (Fig. 2a4). We also include undigested plasmid on agarose electrophoresis as an additional control for these reasons. Uncut plasmid is a supercoiled DNA and it appears as three bands of different sizes on the agarose gel because of different conformation. It is useful to check the background of the plasmid on the agarose gel and avoid the false-positive result from the digested fragments. In addition to diagnostic restriction digest, we sequence pRetroCMV-EGFPGluA2(R583Q) by using the sequencing primers: forward (from 5′ to 3′, 3600–3618 on pCI-CMV-EGFPGluA2 (R583Q)): CTGACATTGCAATTGCTCC and reverse (from 5′ to 3′, 4560–4544 on pCI-CMV-EGFPGluA2(R583Q)) AC GTTGCTCAGACTGAG. The primers are designed to sequence the pore region on pRetro-CMV-EGFPGluA2(R583Q) to ensure the Gln (Q) remains at the RNA editing Q/R site. If the results from the diagnostic restriction digestion and the sequencing are as expected, we create bacterial glycerol stocks for long-term storage. For further retroviral production, we may need a higher concentration of the plasmid. We inoculate the bacteria containing the plasmid by adding 1 mL miniprep culture into 150 (or 200) mL LB

322

Ting-Jiun Chen et al.

medium containing 200 μL/mL ampicillin (215 rpm at 37 °C for 12–18 h). To isolate and purify the plasmid DNA from bacteria, maxipreps are carried out by using QIAfilter Plasmid Maxi Kit (Qiagen) following the manufacturer instructions. We verify the plasmids again by diagnostic restriction digest and sequencing as described above. Notes and Troubleshooting: • In case that the picked colonies do not contain the correct plasmid (after examination with diagnostic restriction digestion), pick up and screen more colonies if they are available. Otherwise restart the experiment from ligation to transformation. • We use nuclease-free water to elute DNA and store it at -20 °C, unless stated otherwise. Using nuclease-free water for elution provides greater flexibility during subsequent reactions. • All the restriction enzymes used in our experiments are from New England Biolabs. • Another option of plasmid purification after larger culture is to use “endotoxin-free” plasmid Maxiprep kit. Endotoxins, also known as lipopolysaccharides, are released from bacteria during the lysis step and may influence the transfection efficiency. • We measure the concentration of the plasmid after each miniprep or maxiprep by NanoDrop, which is necessary for further viral production (see Subheading 2.3). 2.3 Retroviral Production with Lipofectamine 2000

When we have all the necessary plasmids ready, we start with production of the retrovirus which we will use for in vivo gene delivery. The viral production includes the following steps: growing of the appropriate cell line, transfection of the cells using the plasmids, and harvesting the virus from the cells. In our experiments, we package the transgene (the Ca2+-permeable GluA2 fused with GFP, or GFP alone) into the retrovirus, using HEK 293T/17 (ACTT) cells. HEK 293T/17 cells are cultured with complete growth medium (Dulbecco’s Modified Eagle’s Medium (DMEM) containing 4500 mg glucose/L, 110 mg sodium pyruvate/L and L-glutamate (PAA and Sigma), 10% fetal bovine serum (FCS, Life technologies), and 1% penicillin/streptomycin (PAA)). TrypsinEDTA (0.25%, Sigma) is used to detach the cells during subculturing. Lipofectamine 2000 (Invitrogen) is used as the transfection reagent. For retroviral production, we follow the procedure described previously ([30] with few modifications [8]) (also see below). The procedure takes 5 days in total. On day 1, we plate the 5 × 106 HEK 293T/17 cells onto six, 10 cm diameter plates using the DMEM containing FCS and antibiotics. Six plates are a good number for easy handling, but

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

323

they yield less virus compared to the 12 plates used by Tahiro et al. [30]. On day 2, we transfect three plasmids (pRetro-CMV-GFP or pRetro-CMV-EGFPGluA2(R583Q), pCMV-galpol, and pCMVVSVG) into the HEK 293T/17 cells. Before starting this procedure, it is necessary to check whether the cells are healthy, and the cell confluency is roughly 40–60%. We gently wash the cells with OptiMEM without detaching them and replace the medium containing antibiotics to 10 mL OptiMEM per plate. We then prepare four 15 mL Falcon tubes with 2.4 mL OptiMEM per tube. Into two tubes, we add into each 45 μg DNA mixture consisting of 22.5 μg pRetro-CMV-GFP (control) or pRetro-CMV-EGFPGluA2(R583Q), 15 μg pCMV-galpol, and 7.5 μg pCMV-VSVG. Into another two tubes, we add 150 μL lipofectamine 2000. Subsequently, we mix the content of a tube with DNA and the content of a tube with lipofectamine 2000. We now have two tubes with the DNA mixtures together with lipofectamine (control or pRetroCMV-EGFPGluA2(R583Q), pCMV-galpol, and pCMV-VSVG), which we incubate for 25–30 min at room temperature. We then transfer 1.6 mL of either control or pRetro-CMV-EGFPGluA2 (R583Q) mixture onto the plates with HEK 293T/17 cells. This procedure should be performed gently in order to avoid the detachment of the cells. We incubate the transfected plates for 6.5–7 h for pRetro-CMV-GFP or 10–12 h for pRetro-CMVEGFPGluA2(R583Q) at 37 °C in the incubator with 5% CO2. After the waiting time, we replace the supernatant with DMEM containing FCS and antibiotics and place the plates back into the 5% CO2 incubator at 37 °C. On day 4, we collect the supernatant into 50 mL Falcon tubes, add 10 mL growth medium with FCS and antibiotics into each plate, and leave those plates at 37 °C in the incubator with 5% CO2. We centrifuge the collected supernatant at 1000 g for 2–3 min to remove cell debris and filter through a pre-wet 0.45 μm filter (Merck) with DMEM containing FCS and antibiotics. Subsequently, we evenly distribute the supernatant to two 36 mL ultracentrifuge tubes (Thermo Scientific). We add PBS till ~0.5 cm from the top of the tubes, balance each tube to the same weight, and centrifuge with the rotor AH-629 (Thermo Scientific) for 2 h at 65,000 g, 4 °C. After centrifugation, viral particles are precipitated to the bottom of the centrifuge tube, the supernatant is carefully discarded. Avoid touching the bottom of the tube because the pellet may not be well visible in this step. Add 400 μL of PBS into each centrifuge tube (two tubes in total) containing viral particles, and store at 4 °C overnight. On day 5 (i.e., 72 h after transfection), we collect the supernatant again and repeat the same procedure as described for day 4. After 2 h of centrifugation, we carefully discard the supernatant in each tube (total of two tubes). Avoid touching the bottom of the

324

Ting-Jiun Chen et al.

tube even if no obvious pellet is visible at this stage. We resuspend the virus in each tube with 700 μL PBS by pipetting it up and down for 20–30 times. Critical step here is to avoid generating any air bubbles during the pipetting. Pool the resuspended virus into one 4 mL ultracentrifuge tube (Thermo Scientific) and fill PBS till 0.5 cm from the top of the tube. Centrifuge using the rotor TH-660 (Thermo Scientific), at 65,000 g, 4 °C, for 2 h. Remove the supernatant as much as possible. A small pellet is precipitated at the bottom of the tube. Add 40 μL PBS into 4 mL ultracentrifuge tube which contains the small pellet, and resuspend the final pellet by vortexing for 30 s and then pipet carefully and gently but avoid generating any air bubbles. Transfer the sample into one 0.5 mL tube. Wash the 4 mL ultracentrifuge tube once with 5–10 μL PBS (depending on the size of the final pellet), and transfer it back to the same 0.5 mL tube. Shake the tube containing the viruses on ice, on a shaker at 4 °C for 1–3 h, as an optional step. Prepare aliquots (2.0–2.5 μL) of each virus in the 0.2 mL tubes, put them into a 50 mL Falcon tube, and store them at -80 °C. Usually this virus production procedure allows for the preparation of roughly 20 aliquots of each virus. We aliquot the virus into 2.0–2.5 μL aliquotes to avoid re-freezing virus after thawing. One tube is enough for bilateral stereotaxic brain injection per mouse. Frozen retrovirus can be used for up to a year, and the infection efficiency diminishes after that time. Notes and Troubleshooting: • Retroviral vectors are classified as Biosafety Level 2. Follow the guidelines of Biosafety Level 2 during viral preparation. We autoclave all the solid and liquid waste before disposal. All the cells and media before transfection are considered as Biosafety Level 1; however, we perform all procedures following the guidelines of Biosafety Level 2. • If the density of HEK293T/17 cell per plate is too high (70–90% per 10 cm plate) or cells look unhealthy on day 2, this may impair the titer of the virus. Do not conduct the experiment further but repeat the procedure from day 1. • The viral titer is higher when we use OptiMEM instead of complete DMEM during lipofectamine transfection. • The incubation time of transfected cells is critical: it should be long enough for protein expression but not too long, else the cells may start dying. • If the cells look unhealthy or the color of OptiMEM turns yellow during transfection (day 2), replace the OptiMEM to the DMEM containing FCS and antibiotics, even before the recommended incubation time for transfection is over.

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

325

• We usually collect and concentrate the virus from the supernatants 48 h (day 4) and 72 h (day 5) after transfection, in order to obtain a higher amount of the virus. However, if many HEK293T/17 cells appear dying during the process, collect the virus as soon as possible to avoid the low titer of viral production. 2.4 Estimation of the Viral Titer

Once the stock solution of a retrovirus is prepared, we assess its potency in vitro using a viral titer assay. This assay evaluates the capability of the virus to infect and replicate within target cells. The typical procedure involves three steps: (1) preparation of the serial dilutions from the virus stock solution, (2) infecting the target cells with the diluted virus, (3) plating the mixtures of diluted virus and HEK293T/17 cells, and (4) quantifying the viral titer by estimating the count of infectious particle (Fig. 3).

Fig. 3 Overview of the experimental workflow for the estimation of the viral titer. (a) Serial dilutions of the viral stock solution: Different concentrations of the virus are prepared by making a series of dilutions in DMEM, in the range from 1:50 to 1:5 × 107. Black numbers under the Eppendorf tubes indicate the dilution factors. (b) Cell density adjustment: Following the dilution of the virus, 100 μL of the HEK293T/17 cell suspensions are added to each of the Eppendorf tubes containing the diluted virus. (c) Plating and incubation: 100 μL of the cell-virus mixture is plated into the wells of a 48-well plate containing 25 μL of culturing medium. Afterward, the plate is incubated for 3 days. (d) Quantification of the fluorescent clusters: After the incubation period, the number of fluorescent clusters in each well is counted. The counting is only reliable if the clusters of the infected cells are falling into the range between 20 and 200. Created with BioRender.com

326

Ting-Jiun Chen et al.

1. Preparation of the serial dilutions from the virus stock solution We initially dilute 2 μL of the viral stock solution in 98 μL DMEM with FCS and antibiotics (1:50) and mix well. Then we make a series of subsequent dilutions as follows: (1) Prepare 6 Eppendorf tubes containing 90 μL medium each. (2) Add 10 μL of the virus solution from the initial dilution into the first Eppendorf tube and mix well. The dilution factor now is 1:500. (3) Add 10 μL of the 1:500 dilution into the second Eppendorf tube and mix well. The dilution factor now is 1:5000. (4) Proceed in a similar way using the remaining Eppendorf tubes, until the dilution of 1:5 × 107 (Fig. 3a). 2. Infecting the target cells with the diluted virus We use HEK 293T/17 cell line as the target cells in this assay. We adjust HEK 293T/17 cells to the density 106 cells/mL, add 100 μL of the cell suspension to each viral dilution, and mix well. The cell density in all Eppendorf tubes is now 5 × 105 cells/mL, while the viral dilution factor in the tubes is 1:103, 1:104, etc., down to 1:108 (Fig. 3b). 3. Plating the mixtures of diluted virus and HEK293T/17 cells Next, we take a 48-well plate, add 250 μL medium per well (6 wells in total), and plate 100 μL of each mixture with diluted virus and HEK293T/17 cells directly into the wells. The viral dilution factor now is from 3.5 × 103 down to 3.5 × 108 due to the 3.5 times dilutions from adding 250 μL medium to each well (Fig. 3c). It is important to mix each tube containing the HEK 293T/17 cells and virus very well before plating. We place the 48-well plate into a 5% CO2 incubator at 37 °C. After growing the cells in the 48-well plate for 3 days, we estimate the viral titer by counting infectious particles. 4. Quantifying the viral titer by estimating the count of infectious particles To quantify the viral titer, we count the fluorescent clusters that originated from the cells infected either with retrovirus containing GFP alone or with the retrovirus containing the Ca2+-permeable GluA2 subunit fused with GFP. We fix the cells with 4% PFA for 20 min at room temperature. In order to obtain a brighter GFP fluorescence, and thus a better visualization of the infected cells for counting, we perform anti-GFP staining using chicken anti-GFP primary antibody (chicken anti-GFP (1:500, Abcam), apply at 4 °C overnight) followed by application of anti-chicken FITC secondary antibody (donkey anti-chicken FITC (1:1000, Dianova), and apply 3 h in room temperature or overnight at 4 °C). We then count the number of fluorescent clusters in each well and take the number of counts from the well containing 20–200 clusters. This ensures that the viral titer falls within a certain range,

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

327

such that the virus is present in sufficient quantities (Fig. 3d). We calculate the viral titer (unit is presented as colony-forming unit (c.f.u.) per mL) as an example shown below: In the 1:107 sample, we count 50 clones. Therefore, the viral titer is 50 × 3.5 × 107 = 1.75 × 109 c.f.u/μL. To express the titer per mL, 1.75 × 109 × 103 = 1.75 × 1012 c.f.u./mL. Typically, we obtain a titer of 108–109/mL.

3

Injection of Retrovirus into the Mouse Corpus Callosum

3.1 Required Equipment, Materials, and Experimental Animals

Animals All animal experiments are performed in accordance with relevant authorities’ guidelines and approved by the relevant authorities. The whole procedure should be adjusted according to their guidelines and regulations. NG2DsRedBAC transgenic mice are at P10–12. The NG2DsRedBAC mice express red fluorescent protein DsRed under control of the NG2 (Cspg4) promoter. These mice allow for easy identification of NG2+ OPCs by their red fluorescence in living or fixed slices. Equipment The following equipment is required: Anesthesia chamber (Harvard Apparatus, USA), computer-controlled stereotaxic frame (Stoelting, USA), pressure-induced drug application system PDES-01D-4 (NPI Electronic, Germany), and P-97 (Sutter Instruments, USA). Materials, Drugs, and Chemicals Metacam (Boehringer Ingelheim), Xylocaine (AstraZeneca), surgical instruments, silk thread (Ethicon, USA), 1.05 ID (inner diameter), and 1.5 OD (outer diameter) borosilicate glass capillaries (GB150T-8P, Science Products, Germany) for viral injection pipettes.

3.2 Experimental Procedures

Mice are anesthetized with a mixture of oxygen and isoflurane (1–3% v/v) and fixed in the stereotaxic frame (Stoelting, USA). For analgesia, Metacam (1 mg/kg body weight, Boehringer Ingelheim), or an alternative compound, is injected subcutaneously before the surgery. The skull of a P10–12 mouse is soft; therefore, it is advisable to put rubber tips onto the ear bars of the stereotaxic frame when fixing the mouse head in the frame. The skin above the skull is disinfected and an anesthetic (e.g., 2% Xylocaine, AstraZeneca) is applied locally. Then a small (~0.5 cm) cut of the skin of the skull is made at the location of the future viral injection. Before

328

Ting-Jiun Chen et al.

making a hole in the skull for injection, a glass micropipette containing a dye is attached to the stereotaxic frame and connected to a fast pressure application system. We use PDES-01D-4 (NPI Electronic, Germany), but an alternative injection system can be used instead (i.e., nanoliter injector). The dye is used to precisely mark the position of the future virus injection site on the skull. We break the tip of the glass micropipette containing the dye and mark each of the two injection sites with a small drop of the dye. To infect as many callosal OPCs as possible in mouse corpus callosum, we use bilateral injections with the following coordinates (in mm from bregma): anteroposterior 0.23, mediolateral ±0.23–0.25, dorsoventral 1.77. However, the sites/coordinates of the injection depend on the target brain area of interest, animal species, and animal age; they should be adjusted as necessary in each experimental series. At the sites marked by the dye, we make two small injection holes using a thin sewing needle. We do not use a dental drill for this purpose because the skull of a young mouse is thin and soft. However, if the experimental goal is to deliver the virus into the brain of an adult mouse, a dental drill should be used for making the injection hole(s). We prepare glass micropipette for virus injection using the puller from Sutter Instrument (or equivalent). A thin tip and long taper of glass micropipette should be used to get a better outcome of stereotaxic injection. Glass micropipette containing ~2.5 μL of viral stock solution (titer of each virus is between 108–109/mL, backfilled with Eppendorf microloader), is mounted onto the pipette holder of pressure-induced drug application system PDES-01D-4. We carefully break the glass micropipette from the most distal tip by using a fine tweezer and apply pressure until a viral droplet comes out of the tip. The critical step here is to maintain a longer tip of micropipette but appropriate tip size for the viral droplet. It ensures the adequate amount of virus injected into the brain and minimizes the tissue damage from injection. We then lower the micropipette containing viral stock into the injection hole (~1.77 mm deep into the brain from the surface) and inject the viral solution. With our system, we use the following injection parameters: pressure 16–20 psi, application duration 60–90 ms, but the best parameters depend on the application system and should be found and tested for each system in a preliminary series of experiments. The total volume of the virus for bilateral intracranial injections into one mouse is between 1.5 and 2 μL. After the injection is completed, the injection pipette is lifted from the brain and removed, the holes in the skull are not filled, and the wound in the skin is sutured with the silk thread (Ethicon, USA). The whole procedure including the surgery and the viral injection usually takes 30–40 min. Subsequently, the mouse is disconnected from the stereotaxic frame and placed under the red light of the photobiomodulation lamp to stimulate wound healing and relieve

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

329

pain [31]. After the surgery, mice usually recover from anesthesia within 20–30 min. Upon recovery, they are returned to their home cages with parents. It is important to observe all operated mice during the following days after surgery in order to ensure that the wounds remain properly closed and do not get infected. If the wound opens, it should be closed with a suture under anesthesia.

Notes and Troubleshooting: • Keep in mind that when the rubber-tip ear bars are used for the fixation of the mouse head in the stereotaxic frame, the stability may be lower compared to the sharp-ended ear bars used for adult animals. Hence, the mouse should be handled as gently as possible for all the surgical procedures. • Before lowering the injection pipette into the brain, it is important to check that the pipette is not clogged. To do this, we apply a few pressure pulses, when the pipette is still above the brain. During this procedure, small droplets are expected to come out from the tip of the injection pipette. If this does not happen, the tip of the injection pipette should be broken with fine forceps a little bit further, and the pressure pulses should be applied again. • If any indication appears that the injection pipette might have been clogged during the injection, it is advisable to break the existing pipette more and repeat the injection. However, if the opening of the pipette is huge, the volume of the injection would be too large, potentially causing tissue damage; therefore, it is better to change to a new pipette. • The volume of the injection should be monitored (at the pipette) to ensure that comparable viral volumes are injected between animals.

4 Electrophysiology to Evaluate Virus-Induced Changes to Functional Properties of AMPARs in OPC Upon introducing an AMPAR subunit into the callosal OPCs using retroviral gene delivery approach, it is important to verify that OPCs express it. Electrophysiology is among the best approaches to do this because it allows testing for functional changes in AMPAR Ca2+-permeability or conductance in individual cells. To verify the presence of the unedited GluA2 delivered as pRetro-CMV-EGFPGluA2(R583Q), we prepare brain slices to record AMPAR-mediated currents in OPCs at different holding potentials (Vhold). The amplitude of the current (IAMPA) is then

330

Ting-Jiun Chen et al.

Fig. 4 Electrophysiological differences between naive and virally expressed modified GluA2 subunit in the callosal OPCs. (a) Top: Schematic of the naive AMPAR GluA2 subunit with arginine substitution at the Q/R editing site and the ions passing through the channel. Q, glutamine; R, arginine. Bottom: Examples of AMPARmediated EPSCs evoked in an OPC infected with retrovirus expressing only GFP. The OPC is held at five different Vh (-90, -40, 0, +20, +40 mV), the currents are recorded, and the current-voltage (I-V) relationship is plotted. Note that even though spermine is present in the patch pipette, it does not block the AMPAR channels and the I-V relationship is linear, indicating that in the control conditions AMPARs in OPCs are not permeable for Ca2+ ions. (b) Top: Schematic of the modified, virally delivered AMPAR GluA2 subunit, with glutamine (Q) at the Q/R editing site, and the ions passing through the channel. Bottom: Examples of AMPARmediated EPSCs evoked in an OPC infected with retrovirus expressing Ca2+-permeable GluA2 subunit. The OPC is held at five different Vh (-90, -40, 0, +20, +40 mV), the currents are recorded, and the currentvoltage (I-V) relationship is plotted. Note that now spermine present in the patch pipette blocks the AMPAR channels at positive voltages, and the I-V relationship appears inwardly rectifying. (c) Top: Formula for calculation of the rectification index (RI). Bottom: Summary of the rectification indexes for AMPAR-mediated EPSCs recorded in OPCs infected with retrovirus expressing only GFP (n = 12 cells; N = 9 mice) and OPCs infected with retrovirus expressing Ca2+-permeable GluA2 subunit (n = 8 cells from 7 mice). Black diamonds represent group mean ± SEM. The dashed line indicates the theoretical rectification index for a perfectly linear I-V relationship (RI = 0.44). (Adapted from Chen et al. [8])

plotted vs. the Vhold. If the resulting current-voltage (I-V) relationship shows inward rectification (i.e., the IAMPA is larger at negative Vhold than at the corresponding positive Vhold), this indicates that unedited GluA2 subunit is indeed present within the AMPAR complex and is functional (Fig. 4b). The inward rectification is attributed to the intracellular polyamines which block the ionic

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

331

flow through the channels at positive potentials [32–35]. On the contrary, AMPARs containing the edited GluA2 subunit limit the passage of divalent ions through the channels, rendering the receptors Ca2+-impermeable; the I-V relationship in this case is linear (Fig. 4a). 4.1 Required Equipment, Materials, and Experimental Animals

Animals NG2DsRedBAC transgenic mice, 3–5 days after the injection of retrovirus containing the Ca2+-permeable GluA2 subunit fused with GFP, or GFP alone. Equipment Anesthesia chamber (Harvard Apparatus, USA), vibratome (Leica VT1200S), customized interface chamber or equivalent (Haas type) upright patch-clamp microscope (FN-1, Nikon, Japan), stimulator (A-M Systems, Model 2100, Science Products, Germany), EPC-8 amplifier (HEKA, Germany), multichannel data acquisition interface (ITC-18, HEKA Instruments Inc., USA), and PC-10 pipette puller (Narishige, Japan). Alternative equipment available may be used instead. Materials, Drugs, and Chemicals Spermine (Sigma), NMDA-receptor antagonist (RS)-3-(2-Carboxypiperazin-4-yl)-propyl-1-phosphonic acid (CPP, Tocris), GABAA receptor antagonist (RS)-3-(2-Carboxypiperazin-4-yl)-propyl-1phosphonic acid (gabazine, Sigma), tetrodotoxin citrate (TTX, Abcam), 6-Cyano-7-nitroquinoxaline-2,3-dione (CNQX, Abcam), and borosilicate glass capillaries (Science Products, Germany) for pulling the pipettes. In addition, solutions for preparation, maintaining, and perfusion of brain slices (e.g., Ringer solution or artificial cerebrospinal fluid) and intracellular solution for patch-clamp recordings are required. Our solutions are described below.

4.2 Experimental Procedures 4.2.1 Slice Preparation for Electrophysiology

Mice are anesthetized with a mixture of oxygen and isoflurane (3% v/v) and decapitated. The brains are dissected in ice-cold N-methyl-D-glucamine (NMDG)-based solution containing (in mM): 135 NMDG, 1 KCl, 1.2 KH2PO4, 20 choline bicarbonate, 10 glucose, 1.5 MgCl2, and 0.5 CaCl2 (pH 7.4, 305 ± 5 mOsm), gassed with carbogen (95% O2, 5% CO2). Coronal brain slices, 270–300 μm thick, are cut in the same solution using Leica VT1200S vibratome. The slices are transferred to a Haas-type interface incubation chamber warmed to 32 °C where they are perfused with Ringer solution containing (in mM): 124 NaCl, 3 KCl, 1.25 NaH2PO4*H2O, 2 MgCl2, 2 CaCl2, 26 NaHCO3, and 10 glucose; 300 mOsm/kg; 7.4 pH; gassed with carbogen. After all slices (usually 6–8) are placed into the

332

Ting-Jiun Chen et al.

interface chamber, the temperature controller is switched off and the chamber is gradually cooled down to room temperature. The slices are allowed to recover in the chamber for 1 h (in total). 4.2.2 Patch-Clamp Recordings

One hour after the preparation, individual slices are transferred to a submerged recording chamber mounted on the stage of an upright microscope (in our experiments, we use FN-1, Nikon, Japan) equipped with infrared differential interference contrast (IR-DIC) filters and a fluorescence light source. The slices are kept at room temperature and superfused continuously (~2 mL/min) with carbogenated Ringer solution. Callosal OPCs are selected for recordings based on the double red-green fluorescence (red because mice express DsRed under the NG2 promoter, and green because OPCs targeted with retrovirus express GFP). As bath solution, we use a Ringer solution containing (in mM): 119 NaCl, 2.5 KCl, 1 NaH2PO4*H2O, 1.3 MgCl2, 2.5 CaCl2, 26.2 NaHCO3, and 11 glucose; 300 mOsm/kg; 7.4 pH; gassed with carbogen during the recording. As pipette solution, we use Cs-based solution containing (in mM): 100 CsCH3SO3H (CsMeS), 20 tetraethylammonium (TEA) chloride, 20 HEPES, 10 EGTA, 2 Na2ATP, and 0.2 NaGTP; 280–290 mOsm/kg; titrated to pH 7.3 with CsOH. We add spermine (Sigma, 100 μM) into the pipette solution in all recordings of the evoked axon-glia currents. Spermine is a polyamine that helps to identify the presence of Ca2+-permeable AMPARs in the recorded cell: It blocks Ca2+-permeable AMPARs leading to an inwardly rectifying I-V relationship of AMPARmediated currents. Spermine may be present in the cells endogenously but may be strongly diluted (“washed out”) during wholecell patch-clamp recordings. Therefore, adding exogenous spermine into the patch pipette has been utilized extensively in the studies aiming to test for the presence of Ca2+-permeable AMPARs. We correct the Vhold for the liquid junction potential (-7 mV) before seal formation. Upon establishing the whole-cell configuration, we first apply a series of depolarizing voltage pulses with an increment of +10 mV from the Vhold of -80 mV and record current responses. The current profile provides an additional, electrophysiological verification that the patched cell is an OPC. To evaluate whether these OPCs contain Ca2+-permeable AMPARs, we hold the cell at several Vhold (i.e., -90, -40, -20, 0, +20, +40 mV) and stimulate callosal axons with isolated pulse stimulator using monopolar glass electrode (resistance 5–6 MO) filled with Ringer solution in order to evoke AMPAR-mediated currents. The stimulation electrode is usually placed 50–150 μm from the recorded cell. Paired (40 ms interpulse interval) monophasic rectangular pulses of 100–250 μs duration are applied every 15 s. Whole-cell currents in response to voltage steps are low-pass filtered at 10 kHz and digitized with a sampling frequency of 20 kHz. All recordings of synaptic currents are low-pass filtered at 1 kHz and digitized with a

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

333

sampling frequency of 10 kHz. To isolate AMPAR-mediated currents, all recordings of evoked synaptic currents are performed in the presence of NMDA-receptor antagonist CPP (10 μM, Tocris) and GABAA receptor antagonist gabazine (5 μM, Sigma). In some experiments, we applied CNQX (10 μM, Abcam) at the end of the recording to ensure that the evoked currents are completely blocked meaning that they are indeed mediated by AMPARs. All drugs are dissolved in Ringer solution and applied via the bath. All patch-clamp recordings are performed at room temperature. Analysis of evoked EPSCs is performed using custom-written macros in IgorPro. To calculate the rectification index, the average value of the EPSC amplitudes at +40 mV is divided by the average value of the EPSC amplitudes at -90 mV (Fig. 4c). Notes and Troubleshooting: • The key step for the successful patch-clamp recordings is the quality of the brain slices. During preparation of the slices, make sure that both NMDG and Ringer/ACSF solutions are carbogenated at least 30–60 min before the beginning of the preparation, stay carbogenated all the time, and the pH is around 7.4. • Keep the procedure of brain isolation and slice cutting as short as possible. Ideally, the brain should be extracted within 45 s of decapitation. Transfer all freshly cut slices to the recovery chamber immediately. • Pre-chill the NMDG solution, preparation tools, Leica cutting chamber, specimen disc, and the cutting tray (preferably metal) on ice ahead of time. Ideally, they should have a temperature of 0–4 °C at the time of the prep. • Recovery of the slices in Ringer/ACSF with gradual decrease of temperature is critical. Ideally, the slices should reach room temperature 30 min into the recovery, at the earliest. • Maintaining ~15–20 mOsm difference between external and internal solutions helps with seal formation and cell stability. • If the experimenters are struggling with seal formation and cell stability, pipettes with slightly smaller tips (6–7 MΩ resistance) can be used at the cost of higher access resistance (Ra). Yet, we recommend that the Ra remains below 40 MΩ in all whole-cell recordings and the change of Ra is not over 30% during the whole recording. • To validate the viability of a newly patched OPC, resting membrane potential (RMP, Vrest) and input resistance (Rin) have to be carefully monitored. At P10 callosal OPCs typically have Vrest ~ -60 to -80 mV and Rin ~ 1000 to 2500 MΩ. Highly depolarized Vrest and very low Rin usually indicate an unhealthy cell.

334

Ting-Jiun Chen et al.

5 Evaluating the Changes in Proliferation and Differentiation of OPCs with Modified AMPARs To examine whether expression of unedited GluA2 in the callosal OPCs results in their altered proliferation and/or differentiation, immunohistochemistry with the markers specific for OPCs and OLs may be used. It is also advisable to perform the anti-GFP labeling in order to amplify the fluorescence of the transfected cells (Figs. 5 and 6). During development of the OL lineage, OPCs first gradually differentiate into premyelinating OLs, which then further mature into OLs (Fig. 5). In mouse brain slices, each of these three developmental stages can be distinguished using different immunohistochemical markers. OPCs are usually distinguished using labeling for neuron-glial antigen 2 (NG2, chondroitin sulfate proteoglycan 4, CSPG4 gene) and platelet-derived growth factor receptor α (PDGFR α) [36]. When OPCs start undergoing differentiation, NG2 and PDGFRα start getting downregulated. The expression of these markers is lost when the cells become OLs. Subsequently, other markers such as proteolipid protein (PLP)/DM20, immature OL antigen O4, galactocerebroside O1, and APC/CC1 (adenomatous polyposis coli, CC1 clone) appear in the premyelinating OLs and immature OLs [37] (Fig. 5) (Note that PLP/DM20 is an early form of PLP, expressed early during the transition process from OPCs to OL. Hence, the marker PLP/DM20 can be used to identify immature OLs). In the final stage of OLs maturation, myelinating OLs express myelin basic protein (MBP), myelinassociated glycoprotein (MAG), and myelin oligodendrocyte glycoprotein (MOG) [37–39]. For quantification of OLs, antibodies targeting antigens expressed in the cell body are usually used, e.g., anti-APC/CC1. To investigate whether GluA2-insert in OPCs affects their proliferation, we use 5-ethynyl-2′-deoxyuridine (EdU). EdU is an alternative to widely used 5-bromo-2-deoxyuridine (BrdU) and, similarly to BrdU, incorporates into the newly synthesized DNA during the S-phase of the cell cycle (Fig. 6) [40]. 5.1 Required Equipment, Materials, and Experimental Animals

Animals NG2DsRedBAC transgenic mice, 3–5 days after the injection of retrovirus containing the Ca2+-permeable GluA2 subunit fused with GFP, or GFP alone. Equipment Anesthesia chamber (Harvard Apparatus), vibratome (Leica VT1200S), microtome (HM 650 V, Thermo Scientific), and epi-fluorescence microscope (Axio Imager Z1m, Zeiss, Germany). We use this equipment in our experiments, but alternative equipment may be used instead.

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

335

Fig. 5 Manipulations of the Ca2+-permeability of AMPARs in callosal OPCs of juvenile mice affect oligodendrocyte lineage progression. (a) Schematic of the oligodendrocyte lineage progression, with markers expressed by the cells at three stages of differentiation. (b) Maximum intensity projections (three successive planes, four color channels) of a confocal scan showing a representative callosal OPC transduced with

336

Ting-Jiun Chen et al.

Materials, Drugs, and Chemicals 5-ethynyl-2′-deoxyuridine (EdU, Thermo Fisher) and solution for preparation of the slices, primary and secondary antibodies, paraformaldehyde (PFA), and other reagents for immunohistochemistry are all described below. 5.2 Experimental Procedures 5.2.1 In Vivo EdU Treatment

5.2.2 Immunohistochemistry

In order to study OPC proliferation, mice are administered with EdU (Thermo Fisher) [41] intraperitoneally at a dose of 25 mg/kg body weight. We prepare EdU stock solution (2.5 mg/mL in DMSO) following the instructions from the manufacturer and store at -20 °C. EdU is administered three times, i.e., on the third, fourth, and fifth day after the viral injection, at 24 h interval. Mice are sacrificed on the fifth day, i.e., 6–7 h after the third EdU injection and immunolabeled for GFP, NG2, and EdU (Fig. 6). Mice are anesthetized with a mixture of oxygen and isoflurane (3% v/v) and decapitated. The brain is isolated and dissected on a cold metal plate. Coronal brain slices (350–400 μm thick) are cut using the Leica VT1200S vibratome in the solution of the following composition (in mM): 87 NaCl, 2.5 KCl, 1.25 NaH2PO4*H2O, 7 MgCl2, 0.5 CaCl2, 25 NaHCO3, 25 glucose, and 75 sucrose. The slices are fixed overnight at 4 °C in 4% PFA in 10 mM PBS. Subsequently, the slices are embedded into 5% agar in PBS and re-sectioned in PBS to 30 μm thick slices using a HM 650 V microtome (Thermo Scientific, USA). All 30 μm thick slices are inspected visually for quality and for the GFP-expressing (green) cells using an Axio Imager Z1m epi-fluorescence microscope (Zeiss, Germany), and slices with no or few transfected cells and visible post-injection debris or tissue damage are removed. From the remaining pool, 4–12 slices per mouse are selected and used for immunohistochemistry and cell counting.

ä Fig. 5 (continued) retrovirus GFP alone. DAPI is shown in blue, GFP in green, CC1 in red, and NG2 in white; the corresponding composite image is shown at the bottom of each column of images. Arrows mark the cell soma. Scale bar: 10 μm. (c) As in (b), but for a premyelinating oligodendrocyte. The stainings and labels are as in (b). Note that the expression of NG2 in a premyelinating oligodendrocyte is weaker than in an OPC. (d) As in (b), but for a mature oligodendrocyte. The stainings and labels are identical to (b). (e) As in (b), but a confocal scan showing a representative callosal OPC transduced with Ca2+-permeable GluA2 subunit. The stainings and labels are identical to (b). (f) As in (e) but for a premyelinating oligodendrocyte. The stainings and labels are identical to (e). (g) As in (e), but for a mature oligodendrocyte. The stainings and labels are identical to (e). (h) The ratio of OPCs (GFP+NG2+CC1-) among all GFP+ cells, compared between the animals that received the injection of retrovirus containing only GFP+ (control; n = 8) and the animals that received the injection of retrovirus containing Ca2+-permeable GluA2 subunit (n = 7). (i) The ratio of premyelinating oligodendrocytes (GFP+NG2+CC1+), compared between the two groups as in (h). (j) The ratio of mature oligodendrocytes (GFP+NG2-CC1+), compared between the two groups as in (h). Black diamonds represent group mean ± SEM. (Adapted from Chen et al. [8])

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

337

Fig. 6 Tracking actively proliferating callosal OPCs upon manipulations of their AMPAR Ca2+-permeability in juvenile mice. (a) Timeline of the EdU labeling, from the moment of virus injection until the immunohistochemical labeling in brain slices: The administration of EdU starts on the third day after the virus injection and continues for 3 days. On day 5 after the viral injection, the animals are sacrificed, brain slices are prepared, and immunohistochemicaly labeled. (b) Schematic of the EdU incorporation into the DNA and the subsequent visualization of EdU+ cells using an EdU Kit. EdU becomes fluorescent after the reaction with Cu2+ ions. (c) Maximum intensity projection (two successive planes, four color channels) of a confocal scan showing representative examples of OPCs labeled with DAPI (blue), GFP (green), EdU (white), NG2 (red), and their composite image, in animals expressing GFP (left) or Ca2+-permeable GluA2 subunit (right). White arrowheads mark GFP+NG2+EdU+ cells. Note that the GFP is fused to the Ca2+-permeable GluA2 subunit, and therefore, the cell soma is not labeled. Scale bar: 10 μm. (d) The ratio of actively proliferating OPCs (GFP+NG2+EdU+) among all GFP+ cells, compared between the animals transduced with GFP (control; n = 8) and the animals transduced with Ca2+-permeable GluA2 subunit (n = 7). (Adapted from Chen et al. [8])

There are two widely used methods for tissue staining. One method is to mount and dry the slices on glass slides before the immunohistochemical procedure. Afterward, the antibody solution is carefully applied onto each slice. In another approach, called freefloating, the slices are floating in the solution during the staining procedure, usually coupled with gentle shaking. A major advantage

338

Ting-Jiun Chen et al.

of the former approach is that it is cost-efficient as it allows using smaller volumes of antibody per slice (~20–25 μL may be sufficient). However, free-floating approach creates more uniform staining, allows antibody penetration from both sides, and is more suitable for staining of thicker slices. We perform all stainings using the free-floating slices (30 μm thick) placed into multiwell plates. For antigen retrieval, slices are incubated in 10 mM citric acid (pH = 6.0) at 37 °C for 1 h. After washing, a blocking solution containing 0.1 M Tris-buffer saline (TBS), 3–5% Albumin Fraction V (MilliporeSigma, USA), and 0.2–0.5% Triton-X (MilliporeSigma, USA) is applied at 37 °C for 1 h. Subsequently, the slices are incubated overnight with the primary antibody applied in the blocking solution. The following primary antibodies are used: rabbit anti-NG2 (1:500, gift from Bill Stallcup, Burnham Institute, La Jolla, USA) to label OPCs, mouse anti-APC/CC1 (1:250, Ab-7, Calbiochem) to label cell soma of OLs for counting, and chicken anti-GFP (1:500, Abcam) to amplify the fluorescence in all cells targeted with retrovirus. Detection is performed using the following secondary antibodies: goat anti-rabbit Cy5 (1:500, Dianova), goat anti-mouse Alexa Fluor 555 (1:500, Invitrogen), and donkey anti-chicken FITC (1:1000, Dianova). Secondary antibodies are applied for 3 h at 37 °C. For EdU visualization, we follow the protocol recommended by Thermo Fisher Scientific. For counterstaining of the nuclei, Diamidino-2-phenylindole dihydrochloride (DAPI, 0.2 μg, Sigma) is used. 5.2.3

Image Acquisition

Confocal microscopy provides high resolution due to thin optical sections and is preferred over fluorescent microscopy in judging co-localization of markers. A confocal laser scanning microscope is used for image acquisition (we used LSM 710, Zeiss, Germany, equipped with a 40× oil-immersion objective (NA = 1.3)). Images containing corpus callosum are acquired and saved as z-stacks with 16 bit pixel depth. We capture the images as tile scans, each consisting of 3 × 7 or 2 × 7 (vertical × horizontal) tile images. Each tile image is 512 × 512 pixels, with pixel size of 0.415 × 0.415 μm. Along the z-axis, the distance between the images within a stack is 1 μm, and we acquire 6–18 z-sections. Confocal microscopy allows separation of signals from different cellular markers provided that those markers are coupled with fluorochromes with unique emission spectra. In our experiments, we use four markers matched to the fluorochromes emitting blue, green, red, and infrared wavelengths, which allows us to record four separate color channels per tile scan, each channel representing a unique cellular marker. The following excitation laser lines and emission detection ranges are used: for DAPI excitation 405 nm, emission 414–490 nm; for

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

339

FITC excitation 488 nm, emission 497–556 nm; for Alexa-555, Alexa-568, or Cy3 excitation 561 nm, emission 569–633 nm; for Alexa-633 or Alexa-647 excitation 633 nm, emission 650–740 nm. The beam splitters for each fluorochrome matched the excitation laser lines. When co-localization of signals from different markers is necessary, it is critical that each of the signals comes from the same optical section (thickness) of the tissue. To maintain the same optical section thickness of 1.2 μm for all channels the pinhole size has to be adjusted individually for each wavelength, within the range between 1.07 and 1.42 airy units (AU). Pinholes for emission wavelengths close to blue light are set at 1.07 AU and were progressively opened for green/red light, finally reaching 1.42 AU opening for infrared light. 5.2.4

Cell Counting

We perform counting of GFP-labeled cells (infected with retrovirus) stained for various molecular markers, in the z-stacks of images, using ImageJ Cell Counter plugin (NIH, USA). Even the highest quality antibodies and the optimized immunohistochemistry protocols may produce differences in the staining intensity, usually due to the inherent, non-uniform marker expression between regions of the labeled structure, tissue viability, or differences between individual animals. Such processes are usually beyond the experimenter’s control, but they can be corrected digitally after the images are acquired. One of the most common methods for such correction is to subtract the background signal from the actual marker signal, therefore reducing the overall variability between samples. In ImageJ, the background signal intensity can be measured by creating an “intensity profile” (plot profile), drawing by hand a curve passing through the non-labeled areas of the image, and then subtracting the mean/median value from each pixel of the original image. The adjustments of brightness or contrast by hand are not advised as they tend to rely on the experimenter and introduce bias, unless automated. After subtracting the background, the next step is to identify the borders of the region of interest (ROI). In case of the corpus callosum, the mediolateral passing of axons and callosal postnatal development enforce specific orientation of the callosal cells which align with the passing fibers and elongate alongside them. For this reason, a simple staining for cell nuclei (such as DAPI) or a staining which highlights the cell soma (such as NG2 or APC/CC1) are sufficient to identify callosal ventral and dorsal borders. To define the lateral borders in each coronal slice, we determined the midline of the brain slice and outlined the area of the corpus callosum approximately 500–700 μm to each side from the midline. Thus, the ROI spans 1–1.4 mm along the mediolateral axis of the corpus callosum and avoids the rapidly dividing neuronal precursor cells in

340

Ting-Jiun Chen et al.

the vicinity of lateral ventricles. For the cells located at this predefined border of the ROI, only those cells are included in the analysis whose nucleus is ≥50% within the border. As transfection rates may show large variations, which may be due to the amount of diving OPCs when retrovirus is injected stereotaxically, the volume of the injection, or the slices we select for the staining, it is advisable to present the results of the cell counting as a ratio between the counted specific subpopulation and the whole population of the transfected cells. In our experiments, cells are counted in 4–12 slices from each mouse and the counts within one animal are summed, resulting in 29–486 GFP+ cells per animal. Individual slices can vary greatly in the number of transfected cells and/or their post-transfection lineage progression. Those discrepancies add to the considerable variability among animals, potentially substantially skewing the estimate of the mean and variance of the sample, resulting in pseudo-replication errors. The selection of the markers to identify the cells of interest is of critical importance. It is widely accepted that at least two distinct markers are necessary to reach ≥90% certainty that a cell belongs to a specific subtype. In case of cells differentiating as rapidly as OPCs, a third marker serving as a validation of the cell’s developmental stage is necessary. In our study, we identify OPCs as GFP+NG2+CC1- cells (Fig. 5b, e, h), premyelinating oligodendrocytes as GFP+NG2+CC1+ cells (Fig. 5c, f, i), and myelinating oligodendrocytes as GFP+NG2-CC1+ cells (Fig. 5d, g, j). NG2 is a marker typical for OPCs, while APC/CC1 is a marker of myelinating cells. However, APC/CC1 starts to be expressed when the cells are committed to the oligodendroglial fate but do not necessarily form myelin sheaths. Because NG2 does not get downregulated instantaneously, NG2+CC1+ cells can be defined as premyelinating oligodendrocytes (the labeling for NG2 is weaker than in OPCs, and the labeling for CC1 is weaker than in OLs). Cells targeted with retrovirus express GFP in our experiments. Thus, combining the labeling for NG2, CC1, and GFP allows for the identification of three distinct developmental stages of the oligodendrocyte lineage cells. For the detection of OPCs proliferation, we count GFP-expressing cells positive for EdU or NG2EdU (shown as an example, Fig. 6d). To avoid an individual bias in cell counting, it is advisable to repeat the counting by two investigators “blinded” with respect to the experimental group of animals. If the resulting differences in counts between the two investigators are minor, the counts can be accepted. If the counts appear strongly different, it is advisable to re-evaluate the data in order to understand and resolve the reasons for the discrepancies.

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . .

341

Notes and Troubleshooting: • Many groups performing immunohistochemistry in brain slices perfuse animals with 4% PFA before preparation of the slices. We try this procedure in our preliminary experiments but find that NG2 staining is much weaker in slices from animals perfused with PFA vs. animals sacrificed without perfusion, even after long antigen recovery. Therefore, we perform all experiments in slices from non-perfused mice. • Dying cells are weakly autofluorescent and can bind antibodies nonspecifically; therefore, a good signal-to-noise ratio is critical during cell counting. Therefore, we only count signals with intensity of at least two times the intensity of the background. • The expression of NG2 is also found in pericytes in the brain, which are localized at the blood vessels. Pericytes are morphologically different from OPCs, as they are embedded in the basement membranes and extend their processes along and around the microvasculature. This typical morphology of pericytes allows their easy distinction from OPCs, which have complex morphology.

6

Conclusion We described an approach for modifying the AMPARs in OPCs in juvenile mice in vivo, using a retroviral gene delivery approach. Unlike the transgenic mouse lines, this retroviral gene delivery approach allows for local modification of AMPARs in proliferating OPCs, within different time windows, and within short time periods. The advantage of this approach lies in the fast expression of the genes of interest. Therefore, the retroviral gene delivery approach is a good strategy to target OPCs in the developing nervous system. However, it may be challenging for targeting OPCs in the adult brain where they proliferate slower or for targeting other cell populations.

Acknowledgments The work of T.-J.C. is supported by NIH/NEI R01: EY031009, and the Knights Templar Eye Foundation. The work of B.K. is supported by National Science Foundation Grant NSF1926781, and National Institutes of Health Grant K01NS110981. The work of M.K. is supported by IKERBASQUE (Basque Foundation for Science), the Spanish Ministry of Science and Innovation (grant PID2019-110195RB-I00), MCIN/AEI/10.13039/ 501100011033/FEDER Unamanera de hacer Europa (Proyecto PID2022-140726NB-I00), and the Basque Government PIBA Project (PIBA2020_1_0030).

342

Ting-Jiun Chen et al.

References 1. Dawson MR et al (2003) NG2-expressing glial progenitor cells: an abundant and widespread population of cycling cells in the adult rat CNS. Mol Cell Neurosci 24(2):476–488 2. Bergles DE et al (2000) Glutamatergic synapses on oligodendrocyte precursor cells in the hippocampus. Nature 405(6783):187–191 3. Buchanan J et al (2022) Oligodendrocyte precursor cells ingest axons in the mouse neocortex. Proc Natl Acad Sci U S A 119(48): e2202580119 4. Haberlandt C et al (2011) Gray matter NG2 cells display multiple Ca2+-signaling pathways and highly motile processes. PLoS One 6(3): e17575 5. Kukley M, Capetillo-Zarate E, Dietrich D (2007) Vesicular glutamate release from axons in white matter. Nat Neurosci 10(3):311–320 6. Sahel A et al (2015) Alteration of synaptic connectivity of oligodendrocyte precursor cells following demyelination. Front Cell Neurosci 9: 77 7. Ziskin JL et al (2007) Vesicular release of glutamate from unmyelinated axons in white matter. Nat Neurosci 10(3):321–330 8. Chen TJ et al (2018) In vivo regulation of oligodendrocyte precursor cell proliferation and differentiation by the AMPA-receptor subunit GluA2. Cell Rep 25(4):852–861 e7 9. Lin SC et al (2005) Climbing fiber innervation of NG2-expressing glia in the mammalian cerebellum. Neuron 46(5):773–785 10. Mangin JM et al (2012) Experience-dependent regulation of NG2 progenitors in the developing barrel cortex. Nat Neurosci 15(9): 1192–1194 11. Muller J et al (2009) The principal neurons of the medial nucleus of the trapezoid body and NG2(+) glial cells receive coordinated excitatory synaptic input. J Gen Physiol 134(2): 115–127 12. Nagy B et al (2017) Different patterns of neuronal activity trigger distinct responses of oligodendrocyte precursor cells in the corpus callosum. PLoS Biol 15(8):e2001993 13. Parri HR, Gould TM, Crunelli V (2010) Sensory and cortical activation of distinct glial cell subtypes in the somatosensory thalamus of young rats. Eur J Neurosci 32(1):29–40 14. Wake H et al (2015) Nonsynaptic junctions on myelinating glia promote preferential myelination of electrically active axons. Nat Commun 6:7844

15. Dingledine R et al (1999) The glutamate receptor ion channels. Pharmacol Rev 51(1): 7–61 16. Cahoy JD et al (2008) A transcriptome database for astrocytes, neurons, and oligodendrocytes: a new resource for understanding brain development and function. J Neurosci 28(1): 264–278 17. Chew LJ et al (1997) Growth factor-induced transcription of GluR1 increases functional AMPA receptor density in glial progenitor cells. J Neurosci 17(1):227–240 18. Kougioumtzidou E et al (2017) Signalling through AMPA receptors on oligodendrocyte precursors promotes myelination by enhancing oligodendrocyte survival. eLife 6:e28080 19. Patneau DK et al (1994) Glial cells of the oligodendrocyte lineage express both kainateand AMPA-preferring subtypes of glutamate receptor. Neuron 12(2):357–371 20. Yoshioka A et al (1995) Alpha-amino-3hydroxy-5-methyl-4-isoxazolepropionate (AMPA) receptors mediate excitotoxicity in the oligodendroglial lineage. J Neurochem 64(6): 2442–2448 21. Zhang Y et al (2014) An RNA-sequencing transcriptome and splicing database of glia, neurons, and vascular cells of the cerebral cortex. J Neurosci 34(36):11929–11947 22. Isaac JT, Ashby MC, McBain CJ (2007) The role of the GluR2 subunit in AMPA receptor function and synaptic plasticity. Neuron 54(6): 859–871 23. Burnashev N et al (1992) Divalent ion permeability of AMPA receptor channels is dominated by the edited form of a single subunit. Neuron 8(1):189–198 24. Geiger JR et al (1995) Relative abundance of subunit mRNAs determines gating and Ca2+ permeability of AMPA receptors in principal neurons and interneurons in rat CNS. Neuron 15(1):193–204 25. Hollmann M, Hartley M, Heinemann S (1991) Ca2+ permeability of KA-AMPA--gated glutamate receptor channels depends on subunit composition. Science 252(5007):851–853 26. Verdoorn TA et al (1991) Structural determinants of ion flow through recombinant glutamate receptor channels. Science 252(5013): 1715–1718 27. Ehrlich I, Malinow R (2004) Postsynaptic density 95 controls AMPA receptor incorporation during long-term potentiation and experiencedriven synaptic plasticity. J Neurosci 24(4): 916–927

In Vivo Viral Gene Delivery to Manipulate Functional Properties of AMPA. . . 28. Poncer JC, Esteban JA, Malinow R (2002) Multiple mechanisms for the potentiation of AMPA receptor-mediated transmission by alpha-Ca2+/calmodulin-dependent protein kinase II. J Neurosci 22(11):4406–4411 29. Zhu JJ et al (2000) Postnatal synaptic potentiation: delivery of GluR4-containing AMPA receptors by spontaneous activity. Nat Neurosci 3(11):1098–1106 30. Tashiro A, Zhao C, Gage FH (2006) Retrovirus-mediated single-cell gene knockout technique in adult newborn neurons in vivo. Nat Protoc 1(6):3049–3055 31. Hamblin MR (2017) Mechanisms and applications of the anti-inflammatory effects of photobiomodulation. AIMS Biophys 4(3):337–361 32. Bowie D, Mayer ML (1995) Inward rectification of both AMPA and kainate subtype glutamate receptors generated by polyaminemediated ion channel block. Neuron 15(2): 453–462 33. Donevan SD, Rogawski MA (1995) Intracellular polyamines mediate inward rectification of Ca(2+)-permeable alpha-amino-3-hydroxy-5methyl-4-isoxazolepropionic acid receptors. Proc Natl Acad Sci U S A 92(20):9298–9302 34. Kamboj SK, Swanson GT, Cull-Candy SG (1995) Intracellular spermine confers rectification on rat calcium-permeable AMPA and kainate receptors. J Physiol 486(Pt 2):297–303

343

35. Koh DS, Burnashev N, Jonas P (1995) Block of native Ca(2+)-permeable AMPA receptors in rat brain by intracellular polyamines generates double rectification. J Physiol 486(Pt 2): 305–312 36. Nishiyama A et al (1996) Co-localization of NG2 proteoglycan and PDGF alpha-receptor on O2A progenitor cells in the developing rat brain. J Neurosci Res 43(3):299–314 37. Zhang SC (2001) Defining glial cells during CNS development. Nat Rev Neurosci 2(11): 840–843 38. Reynolds R, Wilkin GP (1988) Development of macroglial cells in rat cerebellum. II. An in situ immunohistochemical study of oligodendroglial lineage from precursor to mature myelinating cell. Development 102(2):409–425 39. Scolding NJ et al (1989) Myelinoligodendrocyte glycoprotein (MOG) is a surface marker of oligodendrocyte maturation. J Neuroimmunol 22(3):169–176 40. Buck SB et al (2008) Detection of S-phase cell cycle progression using 5-ethynyl-2′-deoxyuridine incorporation with click chemistry, an alternative to using 5-bromo2′-deoxyuridine antibodies. BioTechniques 44(7):927–929 41. Young KM et al (2013) Oligodendrocyte dynamics in the healthy adult CNS: evidence for myelin remodeling. Neuron 77(5):873–885

Chapter 16 Studying Synaptic Integration of Glioma Cells into Neural Circuits Kiarash Shamardani, Kathryn R. Taylor, Tara Barron, and Michelle Monje Abstract Nervous system activity regulates homeostasis, development, and plasticity of the brain. Certain glial cell populations regulate neuronal activity by controlling neurotransmitter availability at the synapse or regulating extracellular ion concentrations. Glial precursor cells can give rise to gliomas, which are the leading cause of brain cancer-related death. Recent studies have demonstrated the influence of neuronal activity on the progression of a range of high-grade gliomas. One process through which neurons influence gliomas is the formation of functional bona fide synapses between presynaptic neurons and postsynaptic glioma cells. Here, we present multiple techniques that can be utilized to study these neuron-to-glioma synapses. We present a coculture method for in vitro studies and layout the process for generating patient-derived xenograft models in mice for in vivo and in situ studies. We describe the use of electron microscopy to observe the structural characteristics of synapses and electrophysiological studies to investigate the electrical properties of such synapses. We also outline two-photon calcium imaging as a powerful tool to study network-level consequences of activity-depended currents in glioma cells. While similar questions can be answered using electrophysiology and calcium imaging, electrophysiology is useful for directly probing synaptic responses and membrane potential changes while calcium imaging is useful for studying networklevel changes. Key words Cancer neuroscience, High-grade glioma, Neuron-glioma interactions, Synaptic signaling, Neuron-glioma coculture, Patient-derived xenograft, Electron microscopy, Electrophysiology, Calcium imaging

1

Introduction Nervous system activity regulates homeostasis, development, plasticity, and regeneration, not only in the brain but also in various tissue types throughout the body [1]. As cancer tends to recapitulate and hijack mechanisms of normal growth and development used by its cell type of origin [2], multiple studies have revealed mechanistic parallels between how the nervous system regulates normal and neoplastic cellular function across a wide range of tissues [1]. High-grade gliomas (HGG) are highly aggressive

Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5_16, © The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

345

346

Kiarash Shamardani et al.

primary brain cancers that are almost universally lethal, making them the leading cause of brain cancer-related death in both children and adults. HGG occur in specific neuroanatomical locations and at specific ages: pontine and thalamic gliomas occurring in mid-childhood, cortical gliomas of childhood occurring in adolescents and young adults, and high-grade gliomas of later adulthood occurring chiefly in the frontotemporal regions [3–5]. This spatiotemporal pattern of glioma formation hints at its origins in neurodevelopment and ongoing neuroplasticity in the brain as well as the critical importance of the tumor microenvironment of the brain. Electrical activity of neurons and activity-associated mechanisms such as neurotransmitters, neurotrophins, and calcium-mediated signaling are major regulators of neural development and plasticity [6–8]. Neuron-glioma interactions are bidirectional: neuronal activity promotes glioma, and gliomas increase neuronal excitability. Studies on patient-derived xenograft slices have shown that gliomas induce hyperexcitability in peritumoral neurons and result in seizures through nonsynaptic glutamate secretion [9, 10], as well as secretion of synaptogenic factors [11, 12] and in some cases loss of inhibitory interneurons [13]. Recent studies have demonstrated the influence of neuronal activity on malignant neural precursorlike cells in a wide range of HGG [14, 15]. The activity of glutamatergic neurons robustly promotes high-grade glioma proliferation and critically regulates glioma growth through secreted paracrine factors such as the synaptic molecule neuroligin-3, which promotes glioma proliferation by activating oncogenic signaling pathways [14]. Though activity-regulated release of growth factors promotes glioma growth, this alone is insufficient to explain the profound effect that neuronal activity exerts on glioma progression. Direct synaptic communication between neurons and glioma cells is a key mechanism by which neuronal activity influences glioma growth [16, 17]. Whole-cell patch-clamp electrophysiological recordings from xenografted patient-derived glioma cells revealed that AMPA-receptor-mediated excitatory neurotransmission exists between presynaptic glutamatergic neurons and postsynaptic glioma cells [16, 17]. Millisecond timescale excitatory postsynaptic currents that are depolarizing occur in a subpopulation of glioma cells and are associated with activity-induced glioma cell calcium transients [16, 17]. These excitatory axon-to-glioma synapses are similar to the axon-glial synapses that form between neurons and normal oligodendrocyte progenitor cells [18]. Glutamatergic AMPA-receptor-dependent neurotransmission through synapses between microenvironmental neurons and malignant glioma cells results in depolarizing currents, which promote glioma proliferation and regulate growth. It has also been shown that gliomas are composed of distinct subpopulations that differentially

Studying Synaptic Integration of Glioma Cells into Neural Circuits

347

support synaptogenesis and tumor pathophysiology [11]. The formation of these synaptogenic subpopulations during tumor progression is correlated with tumor invasion [17], increased aberrant network hyperexcitability, and the onset of seizures [11, 12]. In this chapter, we describe various approaches that can be used to study neuron-to-glioma synapses. We start with neuron-glioma cocultures as an in vitro tool to study neuron-glioma interactions. We then discuss generating patient-derived xenograft models in immunodeficient (NOD-SCID-IL2gamma chain, NSG) mice [19] and utilizing this model to further investigate neuron-glioma synapses. The xenograft models can be investigated using transmission electron microscopy to observe the structural characteristics of synapses, in acute slice electrophysiology to study the electrical properties of glioma cells and neurons, and two-photon calcium imaging to study network-level consequences of these synapses.

2

Neuron-Glioma Coculture Neuron-to-glioma signaling plays a key role in tumor pathophysiology, influencing growth, invasion, initiation, and metastasis. The advantage of culturing two distinct cellular populations, such as neurons and glioma cells, is the investigation of their direct interaction without additional influences from other cell types within the environment. Compared to in vivo modeling, neuron-glioma cocultures offer a relatively high throughput assay to perturb these cellular interactions in a reproducible system. Neuronal activity directly promotes glioma growth via paracrine signaling and direct electrical communication. In a coculture system, the proliferation rate of tumors can be examined in combination with genetic, or pharmacological, interventions of the malignant cells response to activity [16]. In respect to the identification of malignant synapses, immunofluorescent imaging of proteins, which are known to play a key role at pre- and postsynaptic sites, identifies the potential for direct electrical communication between neurons and tumor cells. The coculture system offers an additional level of clarity over tissue preparations for identifying malignant synaptic puncta; however, these structures do not confirm functional synaptic responses nor identify tumor membrane potential changes in response to neuronal activity. It may be feasible to perform electrophysiological recordings of tumor cells in culture; however, it would be more beneficial to measure the tumor cells response while integrated into a functional neuronal network using acute slices of xenografted cells. Glioma cells have a profound effect on neuronal activity by promoting neuronal excitability. Cocultures offer the potential to measure overall electrical activity using electrode arrays [20], which can be confirmed using in situ field potential recordings of glioma xenografts.

348

Kiarash Shamardani et al.

The advantages of using a culture system are also its largest drawback, where the complexity of the tumor microenvironment is lost, both in regard to cellular diversity, as well as tissue structure. Therefore, these experiments serve to support the investigation of factors affecting neuron-to-glioma signaling yet should always be confirmed with orthogonal in vivo neuroscience techniques. 2.1

Materials

2.1.1

Culture Media

2.1.2

Cultureware

1. Buffer: 0.5% BSA in D-PBS with calcium and magnesium (FisherScientific, Cat#14-040-182). 2. Neuron media (50 mL): BrainPhys Neuronal medium (StemCell Technologies); GlutaMAX-I Supplement (Cat#35050061, 500 μL); penicillin/streptomycin (Invitrogen, 500 μL); B-27 Supplement (Invitrogen, Cat#17504-044, 1 mL); BDNF (10 ng/mL, Shenandoah); GDNF (5 ng/mL, Shenandoah), TRO19622 (5 μm, Tocris); β-mecaptoethanol (Gibco, 50 μL) (see Note 1). 1. 24-well culture plate (Falcon). 2. Coverglass (12 mm, Electron Microscopy Services, Cat# 72196-12). 3. Laminin (5 μg/mL, Thermo Fisher Scientific). 4. Poly-l-lysine 0.01% (Sigma, Cat# P4707).

2.1.3 Dissociation/ Isolation Kits

1. Neural Tissue Dissociation Kit (Miltenyi Biotec, Cat#130094-802). 2. Neuron Isolation Kit, mouse (Miltenyi Biotec, Cat#130-115389). 3. MACS SmartStrainers (70 μm, Miltenyi Biotec, Cat#130-098458). 4. Manual separation: LS Columns, MACS Cell Separation (Miltenyi Biotec, Cat#130-042-401), MACS Multistand (Miltenyi Biotec, Cat#130-042-303), QuadroMACS Separator (Miltenyi Biotec, Cat#130-091-051). 5. Automated separation: autoMACS Pro Separator (Miltenyi Biotec, Cat#130-092-545).

2.1.4

Mice

1. P0-P2 mouse pups (see Note 2).

2.1.5

Equipment

1. Temperature controlled centrifuge. 2. Incubator oven with tube rocker/rotator. 3. Tissue culture hood. 4. Sterile forceps, scalpel, dissecting scissors. 5. Wet ice.

Studying Synaptic Integration of Glioma Cells into Neural Circuits

2.2 2.2.1

Methods Preparation

349

Set up the coated coverglass prior to beginning neuron isolation. 1. Using sterile forceps, place one cover glass slip per well. 2. Coat with poly-l-lysine 0.01% and incubate at 37 °C for 1 h. Remove coating and allow to air dry for 5–10 min. 3. Coat with laminin (5 μg/mL) and incubate at 37 °C for a minimum of 1 h. 4. Remove laminin immediately prior to plating the cells.

2.2.2

Neuron Isolation

The Miltenyi Biotech dissociation kit and isolation kit were used to isolate neurons for studying neuron-glioma interactions (see Note 3). The following protocol is briefly outlined from the manufacturers guide and serves as a basis to include additional notes regarding glioma (tumor) culturing. Please refer to the comprehensive manufacturers protocol for detailed instructions. 1. Using sharp sterile surgical scissors, decapitate the pup. Cut the skin along the midline and fold back, make a midline cut in the skull above the brainstem, and remove the top of the skull using forceps. Carefully extract the brain and place into a Petri dish containing buffer. 2. Remove the olfactory bulb and hindbrain using a scalpel. Transfer the cortical region to a 1 mL drop of buffer to a new petri dish. Begin mincing the brain tissue by cutting it repeatedly with a sterile scalpel. 3. Transfer the minced tissue to a 15 mL tube containing 5 mL of buffer using a 1 mL pipette (use sterile scissors to remove approximately 2 mm off the end of the pipette tip). 4. Let the tissue settle (approx. 1–2 min). 5. In a separate tube, prepare the neural dissociation kit enzymes as instructed in the manufacturers guide. 6. To perform tissue dissociation, remove supernatant and add the dissociation enzymes. Agitate the tissue to ensure suspension and incubate as instructed. 7. Repeat step 6 with second dissociation enzyme mix and incubate as instructed. 8. Manually dissociate tissue as instructed and strain the dissociated tissue through a MACS Smart Strainer to collect the single-cell suspension below. 9. Centrifuge at 300 × g for 10 min at 4 °C. 10. Resuspend the cell pellet in buffer. 11. Count the cells (approximately seven million cells to be expected per one pup brain) and proceed to the Neural Isolation Kit (see Note 4). 12. Centrifuge at 300 × g for 10 min at 4 °C.

350

Kiarash Shamardani et al.

13. To the cell pellet, add buffer and the Non-Neuronal Cell Biotin-Antibody Cocktail as instructed. The antibody cocktail will bind all non-neuronal cells, astrocytes, oligodendrocytes, and microglia. 14. Swirl gently to ensure a uniform mix. Incubate and add buffer to wash cells and centrifuge as instructed. 15. Remove supernatant and add buffer and Anti-Biotin MicroBeads as instructed. The microbeads are conjugated to anti-biotin antibodies and will adhere to the non-neuronal biotin conjugated antibody labeled cells. 16. Swirl gently to ensure a uniform mix. Incubate as instructed. 17. Add buffer and proceed to magnetic separation. 18. Magnetic separation can be performed manually using columns or automated as per manufacturer’s instructions. 19. For column separation, the flow through will be collected; for automated separation, the negative fraction will be collected. 20. Centrifuge the neuron isolation. 21. Any red blood cells present can be removed by a 1 min incubation in Red Blood Cell Lysis buffer. 22. Resuspend the cell pellet in Neuronal Medium and count the cells (approximately one to two million cells are expected per pup brain). 23. Remove the laminin from the coated coverglass slips. 24. Plate the neurons at 300,000–400,000 cells/mL/well in Neuronal Medium (see Note 5). 25. Incubate in a cell culture incubator at 37 °C, with 5% CO2. 2.2.3

Neuron Culture

1. After 1 day in vitro (DIV), exchange half of Neuronal Medium for fresh medium. 2. Repeat medium change at DIV 3.

2.2.4 Neuron-Glioma Coculture

1. Dissociate glioma cultures to a single-cell suspension and resuspend in Neuronal Medium. 2. Plate 40,000–70,000 patient-derived glioma cultured cells per well (see Note 6). 3. Incubate in a cell culture incubator at 37 °C, with 5% CO2.

2.2.5 Experimental Paradigms

Neuron-to-Glioma Synaptic Connections Visualize the colocalization of glioma postsynaptic PSD95 with neuronal presynaptic synapsin (Fig. 1): glioma cells expressing RFP-tagged postsynaptic density protein 95 (PSD95) enables the specific staining of glioma postsynaptic puncta. PSD95-pTagRFP (Addgene, plasmid# 52671) [16].

Studying Synaptic Integration of Glioma Cells into Neural Circuits

351

Fig. 1 Representative confocal image of neurons cocultured with PSD95–RFPlabeled glioma cells. White box and arrowheads highlight region of synaptic puncta colocalization; magnified view is shown to the right. Green denotes neurofilament (axon); white denotes nestin staining (glioma cell processes); blue denotes synapsin (presynaptic puncta); and red denotes PSD95–RFP staining (postsynaptic puncta). Scale bars, 10 μm (left) and 2 μm (right) [16]

Glioma Proliferation Rate Measure the proliferation rate of glioma in the presence of neurons: 48 h after plating the tumor cells, EdU is added to each culture well, and this can be done simultaneously with agonists/antagonists to measure proliferation changes over a 24 h period. 2.2.6 Coculture Fixation/ Staining

1. Three days after plating the tumor cells (neuron DIV 8), remove the culture medium, and fix with 4% PFA for 20 min at room temperature. 2. Replace fixative with D-PBS and proceed with staining. 3. Stain with appropriate markers to identify tumor cells, neurons, proliferation, and synaptic puncta, according to the experimental paradigm. Below is an example of antibodies used previously for neuron coculture preparations [16]: Neurons: anti-Neurofilament-H (#NFH, Aves labs, AB_2313552), anti-Neurofilament-M (#NFM, Aves labs, AB_2313554), and anti-MAP 2 (#AB5622, EMD Millipore, AB_91939). Glioma cells: anti-GFP (#ab13970, Abcam, AB_300798) and antihuman Nestin (’#ab6320, Abcam, AB_308832). EdU: Alexa594 ClickIT EdU Cell Proliferation kit (#C10339, Thermo Fisher Scientific). Synaptic puncta: anti-RFP (#600-401-379, Rockland, AB_2209751) and Synapsin (#106-004, Synaptic Systems, AB_1106784).

352

2.3

Kiarash Shamardani et al.

Notes

1. Culture media can be supplemented with 2% fetal bovine serum to increase neuron density and differentiation, and this should be gradually removed and replenished at DIV 1 and 3 with serum-free media if your tumor cultures are grown in serumfree conditions. If there is a contamination of glial cells from your preparation, then 1 μM UFDU (1:1 mix of Urdine (10 mM, #U3750, Sigma) and 5-Fluoro-2′-deoxyuridine (10 mM, #F0503, Sigma)) can be added at DIV 1 and 3 to remove proliferating cells, the media should be fully changed on DIV 5 before plating tumor cells. A 1:1 mix of DMEM/F-12 (#11330-032, 250 mL) and Neurobasal-A Medium (10888-022, 250 mL) can be used in substitute of BrainPhys Neuronal medium. TRO1962 can be removed if specifically studying tumor cell death or mitochondrial function. To avoid the potential for culture contamination, pen/strep is added to the culture media. Antibiotic supplements have been shown to affect electrophysiological properties of neuronal excitability [21]; therefore, it would be beneficial to remove antibiotic supplements when possible. 2. P0 mouse pups produce the healthiest cultures after sorting. CD1 and NOD-SCID-IL2gamma chain (NSG) mice have been used to study neuron-glioma interactions. Other strains have not been determined, but likely suitable. Approximately, one pup brain should produce a sufficient number of neurons to plate 12 wells; this can increased/ decreased depending on pup age and quality of neuron prep. The number of brains collected can be scaled up, but it is important to scale up the reagent volumes and use the appropriate number of strainers for tissue quantity, accordingly. In general, the younger the pup, the better the quality of neuron isolation. The experimental setup can be performed with 48-well plates for higher throughput assays. 3. There are multiple protocols outlining neuron isolation. Here, one example has been given; however, any protocol that isolates purely neuronal cells will be effective. When examining neuron-to-glioma interactions in this context, it is important to not have any glial contamination. The neuron-glioma cocultures offer the advantage of purely neuron-to-glioma interactions. 4. The Neuron Isolation Kit is designed to collect neurons from the brain by antibody depletion of all non-neuronal cells. If a mixture of neurons and other glial cells is of interest, then these should be harvested from the dissociated tissue prior to using this kit. Isolation of certain glial cells will be optimal using older pups for preparation.

Studying Synaptic Integration of Glioma Cells into Neural Circuits

353

5. The density of neurons should be optimized; healthy neurons exhibit small processes by DIV 1 that continue to extend during culture. The neurons should exhibit electrical properties by approximately DIV 5. It is vital that a healthy neuronal preparation is obtained and cultured, and characterization of electrical activity should be confirmed using electrophysiological recordings. 6. The seeding density of different cancer cell cultures will vary based on their proliferation rate. Account for a potential proliferation rate increase upon coculture with neurons, as has been shown previously. The seeding density should also be optimized for application; for example, experiments examining synaptic puncta will require a lower seeding density to enable single-cell imaging at higher objectives (e.g., 63×) (see Subheading 2.2.5). For proliferation rate, a slightly higher seeding density is suitable for imaging multiple cells in one field of view at lower magnification (e.g., 20×).

3

Orthotopic Xenografting of Glioma Cells Although multiple types of animal models are currently available for the investigation of brain tumors in vivo, patient-derived xenografts have been shown to have an increased reliability when reproducing the heterogeneity of the human disease, which may better reflect the therapy response in patients than genetically engineered mouse models (GEMMs) and established human cancer cell lines that have adapted to growth under artificial culture conditions, and are generally considered less relevant for clinical translation due to a more homogeneous, undifferentiated histology. Patient-derived xenografts serve as faithful disease models for preclinical studies as they replicate the human tumor as closely as possible. These models recapitulate histopathology, DNA methylation signatures, mutations, and gene expression patterns of the patient tumors from which they were derived [3, 22]. This allows researchers to study the biology of the tumor in a more clinically relevant context and to identify new therapeutic targets and potential biomarkers. Patient-derived cell culture models of diffuse intrinsic pontine glioma (DIPG) can be generated using a published protocol for the rapid processing of postmortem autopsy tissue samples [23]. For all human tissue studies, informed consent must be obtained, and tissue must be used in accordance with protocols approved by the institutional review board (IRB). In brief, tissue was obtained from high-grade glioma (WHO grade III or IV) tumors at the time of biopsy or from early postmortem donations. Tissue was dissociated both mechanically and enzymatically and grown in a defined,

354

Kiarash Shamardani et al.

serum-free medium designated “tumor stem media” (TSM), consisting of neurobasal(-A) (Invitrogen, Cat# 10888-022), DMEM/ F-12 (Invitrogen, Cat# 11330-032), B27(-A) (Invitrogen, Cat# 12587-010), human bFGF (20 ng/mL; Shenandoah), human EGF (20 ng/mL; Shenandoah), human PDGF-AA (10 ng/mL; Shenandoah) and PDGF-BB (10 ng/mL; Shenandoah), and heparin (2 ng/mL; Stem Cell Technologies, Cat# 07980). 3.1

Materials

3.1.1 Solutions for Cell Culture

1. TrypLE™ Express Enzyme (1X), no phenol red (ThermoFisher, Cat# 12604013). 2. HBSS (Corning, Cat# 21022CV). 3. Deoxyribonuclease I—200X (Worthington Biochemical, Cat# LS002007).

3.1.2 Consumables for Cell Culture

1. Cell Strainer Sterile—40 μm.

3.1.3 Equipment for Cell Culture

1. Nutator.

3.1.4 Consumables for Surgery

1. Unify nylon surgical sutures 5-0.

2. Hemocytometer.

2. Artificial tears lubricant ophthalmic ointment (Akorn Animal Health). 3. Cotton tip applicator—sterile. 4. Neo-Predef with Tetracaine Powder. 5. Lactated Ringer’s solution. 6. Saline solution. 7. 70% Ethanol. 8. Betadine.

3.1.5 Equipment for Surgery

1. Mouse stereotaxic instrument. 2. Tabletop rodent anesthesia machine. 3. 26-gauge, Small Hub RN Needle (Hamilton Company). 4. Hamilton 1700 Series Gastight Syringes: RN termination (Hamilton Company). 5. Microinjection syringe pump. 6. Micromotor high-speed drill. 7. Burrs for micro drill—Tungsten carbide burrs 0.05 mm. 8. Fine Scissors—Sharp. 9. Olsen-Hegar needle holders with suture cutters. 10. Graefe forceps. 11. Pet heated pad.

Studying Synaptic Integration of Glioma Cells into Neural Circuits 3.1.6

3.2

Mice

Methods

3.2.1 Preparing Glioma Cells in a Biological Safety Cabinet

355

1. NOD-SCID-IL2R gamma chain-deficient (The Jackson Laboratory). 1. Using a 25 mL pipette, decant all the medium from the tissue culture flask into a 50 mL conical tube. 2. To the empty culture flask, add 12 mL of TrypLE Express and 120 μL of DNase. Incubate flask at 37 °C for 5 min. 3. Meanwhile, centrifuge the conical tube at 300 × g for 5 min. 4. After centrifuge, aspirate the supernatant. 5. Remove flask from the incubator. Triturate multiple times while holding the flask slightly slanted to get as many cells as possible. 6. Transfer the TrypLE Express mixture to the conical tube and triturate a few times to resuspend the cells. 7. Gently rotate the tube in a 37 °C nutator (pre-warmed) for 5–10 min (see Note 1). 8. Remove the tube from the nutator and triturate the cells 2–3 times with a 10 mL pipette to dissociate the cells. 9. Add 24 mL of pre-warmed HBSS to the cell mixture. Mix the solution with a 25 mL pipette. 10. Filter the mixture through a 40 μm membrane and centrifuge at 300 × g for 7 min. 11. Aspirate the supernatant as much as possible. 12. Resuspend the cells in 1 mL of HBSS and count them with a hemocytometer. 13. Pipette the volume of the cell mixture corresponding to the number of desired cells needed for xenografting into an Eppendorf tube (see Note 2). 14. Centrifuge at 300 × g for 3 min. 15. Remove the supernatant and add desired amount of HBSS to have 200,000 cells/μL. Keep tube on ice.

3.2.2 Stereotactic Injection of Glioma Cells into Mouse Brain

All in vivo experiments must be conducted in accordance with protocols approved by your Institutional Animal Care and Use Committee (IACUC) and performed in accordance with institutional guidelines. 1. Anesthetize NSG mice at postnatal day 28–30 using isoflurane (in oxygen: induce 3–4% and maintain 1–2%). Place ophthalmic ointment in both eyes. 2. Remove hair from the top of the head. Aseptically prepare the shaved skin by disinfecting the surgical area by alternating between betadine and 70% ethanol for three cycles.

356

Kiarash Shamardani et al.

3. Place mouse on the stereotaxis apparatus atop a clean absorbent surface and heat source to prevent hypothermia. 4. Before making an incision, check anesthetic depth. Using aseptic technique, make a midline incision on top of the skull with small surgical scissors. 5. Separate the subcutaneous and muscle tissue, and gently retract the skin to visualize the structures below. 6. Gently scrape clean the bregma and lambda areas using a small bone scraper; keep the skull moist with sterile saline applied with a sterile cotton swab. Adjust the incisor screw to make the head level so that the bregma and lambda are equal; level the head horizontally in the caudal-to-rostral direction. 7. Go to the desired stereotactic coordinates and perform craniotomy over the target injection site using a handheld drill and a sterile burr with slight downward pressure. Stop when bone is thin, and the blood vessels and dura become clearly visible (see Note 3). 8. Using a Hamilton syringe, deliver volumes of 1–3 μL of the cell mixture at the rate of 0.4 μL/min. 9. At the completion of infusion, allow the syringe needle to remain in place for a minimum of 2 min, then manually withdraw at a rate of 0.875 mm/min to minimize backflow of the injected cell suspension. 10. Following successful injection of the tumor cells, close the incision by sealing the scalp with sutures. Add Neo-Predef over the incision. 11. Remove the mouse from the stereotactic frame. Postoperatively, hydrate the animal by a 1 mL subcutaneous injection of lactated ringers. Allow to recover on a temperature controlled (37 °C) heating pad. 12. Once the animal achieves ambulatory recovery, return to original cage. It is recommended to house immunocompromised NSG mice in a strict barrier animal facility. 3.3

Notes

1. The optimal amount of time for enzymatic dissociation of cells should be empirically tested to balance the amount of cell death caused by this step with dissociation efficiency. Some cells might require more than 10 min to dissociate, while in others after 5 min, a significant amount of cell death is observed. 2. Take 10% extra to account for lost volume during the xenografting process. 3. Stereotactic coordinates for CA1 region of the hippocampus: 1.5 mm lateral to midline, 1.8 mm posterior to bregma, 1.4 mm deep to cranial surface. Stereotactic coordinates for the

Studying Synaptic Integration of Glioma Cells into Neural Circuits

357

Fig. 2 (A) Representative confocal images of mouse brain (coronal sections) from SU-DIPG-XIII-FL xenografts expressing GFP in the cortex; white denotes GFP (scale bar = 2000 μm). (B) Representative image of mouse brain (sagittal section) from SU-DIPG-XIII-P* xenografted to the pons; white denotes HNA (tumor cells); DAPI nuclei are shown in blue (scale bar = 2000 μm). (C) Confocal image of infiltrating SU-pcGBM2 cells expressing human nuclear antigen (HNA, red), proliferation marker Ki67 (green) in premotor cortical deep layers and subjacent corpus callosum (MBP, white) (scale bar = 100 μm). (D) Representative confocal micrographs illustrating proliferation of SU-DIPG-VI-ChR2 xenografts. Red denotes human nuclei staining by HNA; white denotes Ki67 (scale bar = 50 μm) [14, 16, 24].

premotor cortex: 0.5 mm lateral to midline, 1.0 mm anterior to bregma, -1.75 mm deep to cranial surface. Stereotactic coordinates for the pons: 1.0 mm lateral to midline, -0.8 mm posterior to lambda, -5.0 mm deep to cranial surface (Fig. 2).

4

Acute Slice Preparation of Hippocampal Xenografts Acute slices of xenografts are useful for both calcium imaging and electrophysiology experiments. These experiments can also be performed in glioma monocultures or cocultures with neurons. However, the benefit of using xenografts is that the glioma cells integrate with the neuronal circuit in a more complex microenvironment more similar to that from which they originated. Xenografting glioma cells into the hippocampus allows for glioma integration into a well-defined circuit, including both glutamatergic projection neurons and local GABAergic interneurons. In acute slices, these neurons can be electrically stimulated with a bipolar stimulator to elicit postsynaptic currents in the xenografted glioma cells. Recording from glioma cells xenografted into the orthotopic location of the tumor may not be feasible; for example, heavy myelination of the brainstem makes patch-clamping of pontine tumor cells xenografted into the pons difficult. While similar questions can be answered using electrophysiology and calcium imaging techniques, electrophysiological recordings are useful for directly probing synaptic responses and membrane potential changes in glioma cells in

358

Kiarash Shamardani et al.

response to neuronal activity. Combined calcium imaging and electrophysiology may be useful to answer questions about how neuron-to-glioma synapses impact calcium transients on a millisecond timescale. Paired recordings of neurons and glioma cells may not be feasible due to the low number of glioma cells that take synaptic input from neurons. A major caveat to consider when performing electrophysiology experiments on glioma cells is that these cells are highly heterogeneous and demonstrate varying electrophysiological properties and responses to neuronal activity [16]. Despite this, electrophysiological recordings of xenografted glioma cells are important to understanding their relationship with neurons in their microenvironment. 4.1

Materials

4.1.1

Solutions

1. Slicing artificial cerebrospinal fluid (ACSF): 125 mM NaCl, 2.5 mM KCl, 25 glucose, 25 mM NaHCO3, 1.25 mM NaH2PO4, 3 mM MgCl2, and 0.1 mM CaCl2, pH 7.4, frozen to form a slush. 2. Recovery ACSF: 100 mM NaCl, 2.5 mM KCl, 25 mM glucose, 25 mM NaHCO3, 1.25 mM NaH2PO4, 30 mM sucrose, 2 mM MgCl2, and 1 mM CaCl2, pH 7.4, warmed to 30 °C.

4.1.2

Equipment

1. Dissection tools (forceps, scalpel, dissecting scissors). 2. Vibratome (such as VT1200, Leica). 3. Brain slice incubator (such as BSK1, AutoMate Scientific).

4.2

Methods

Acute coronal hippocampal slices from mice 6–12 weeks after xenografting can be used for electrophysiology and calcium imaging experiments. To prepare the slices: 1. Anesthetize mouse with isoflurane, euthanize mouse via rapid decapitation using sharp sterile scissors, and carefully remove brain from the skull and immerse in ice-cold slicing ACSF (see Note 1). 2. Remove brainstem and cerebellum with scalpel and mount the cerebrum, cut side down, into a vibratome for coronal sections (see Note 2). 3. Set vibratome to section the tissue into 250 μm-thick slices at 0.16 speed. Collect the slices that contain the hippocampus. 4. Incubate the hippocampal slices in a brain slice incubator for 30 min in warm (30 °C) oxygenated (95% O2, 5% CO2) recovery ACSF to preserve the tissue before allowing to equilibrate at room temperature for an additional 30 min. Slices can be kept in a brain slice incubator at room temperature in oxygenated recovery ACSF for up to 5–7 h depending on tissue health, covered with foil to protect the fluorescence of the cells.

Studying Synaptic Integration of Glioma Cells into Neural Circuits

4.3

Notes

359

1. Alternatively, NMDG- or sucrose-based slicing solution can be used to better preserve the tissue. 2. Alternatively, horizontal or transverse hippocampal slices can be used. Both coronal and transverse slices can be used to successfully record synaptic currents in xenografted glioma cells.

5

Electrophysiological Recording of Postsynaptic Responses in Glioma Cells

5.1

Materials

5.1.1

Solutions

1. Recording ACSF: 125 mM NaCl, 2.5 mM KCl, 25 mM glucose, 25 mM NaHCO3, 1.25 mM NaH2PO4, 1 mM MgCl2, and 2 mM CaCl2, pH 7.4. 2. CsMe-based internal solution: 135 mM CsMeSO4, 12 mM HEPES, 8 mM NaCl, 0.25 mM EGTA, 2 mM MgCl2, 1 mM Mg2ATP, 0.3 mM Na3GTP, and 5 mM phosphocreatine, pH 7.4 and osmolarity of 292–296 mOsM (see Note 1).

5.1.2 Equipment and Software

1. Microscope with DIC optics (such as Olympus BX50WI). 2. Illumination system (such as pE-4000, CoolLED). 3. Stimulation electrode (such as a platinum/iridium metal microelectrode, Harvard Apparatus). 4. Stimulus isolator (such as Iso-flex, A.M.P.I.). 5. Amplifier (such as MultiClamp 700B, Molecular Devices). 6. Data acquisition 8+8, HEKA).

device

(such

as

InstruTECH

LIH

7. Data acquisition software (such as AxoGraph X, AxoGraph Scientific). 8. Platinum/iridium metal microelectrodes. 5.2

Methods

1. Transfer an acute hippocampal slice prepared as above (see Subheading 4.2) to a recording chamber perfused with oxygenated (95% O2, 5% CO2), warmed (28–30 °C) recording ACSF. 2. Use a microscope equipped with DIC optics (Olympus BX50WI) and a pE-4000 illumination system (CoolLED) to identify xenografted GFP-labeled glioma cells. 3. Place bipolar stimulating electrode connected to an Iso-flex stimulus isolator (A.M.P.I.) near the glioma cells to evoke synaptic responses (see Note 2). 4. Using 3–5 MΩ recording pipettes filled with internal solution, whole-cell patch-clamp glioma cells (voltage clamp at -70 mV).

360

Kiarash Shamardani et al.

Fig. 3 Electrophysiological recordings of postsynaptic responses in glioma cells. (a) Postsynaptic current in xenografted glioma cell in response to electrical stimulation before (black) and after (red) inhibition of neuronal activity with 0.5 μM tetrodotoxin (TTX). Blue bars demonstrate a few parameters that can be measured from a postsynaptic current recording: amplitude and rise and decay times. (b) Left, representative trace of a glutamate-mediated postsynaptic current in xenografted glioma cell in response to electrical stimulation before (black) and after (red) application of 10 μM NBQX to block AMPA receptors. Right, time course of NBQX effect on current amplitude as a percent of control (red bar denotes duration). (c) Representative trace of a GABA-mediated postsynaptic current in xenografted glioma cell in response to electrical stimulation before (black) and after (red) application of 50 μM picrotoxin (PTX) to block GABAA receptors. (d) Representative trace of a GABA-mediated postsynaptic current in xenografted glioma cell in response to electrical stimulation before (black) and after (red) application of 10 μM lorazepam (LZP) an allosteric modulator of GABAA receptors [16, 24, 25]

5. Use data acquisition software to set stimulation duration, frequency, and intensity. 6. Acquire signals with amplifier and digitize using a data acquisition device at 10 kHz. 7. Record responses to electrical stimulation. Parameters that can be analyzed include those such as amplitude and rise and decay times. These parameters can be measured before and after application of pharmacological agents (Fig. 3). Other useful experiments may include measuring paired-pulse ratio and spontaneous synaptic currents [16]. 5.3

Notes

1. Cs-based internal solution for voltage clamp recordings provides a better signal, as the glioma cells are highly branched and highly express K+ channels. As an alternative to CsMe, CsCl internal solution can be used to record larger GABA-mediated currents. K-based internal solution should be used for current clamp recordings. Optionally, add Alexa 568 dye to the internal solution to visualize the cell through dye-filling during wholecell recordings.

Studying Synaptic Integration of Glioma Cells into Neural Circuits

361

2. Placement in the strata radiatum or strata oriens will maximize the chance of activating a synaptic input onto the cells. Placing the stimulation electrode before patch-clamping the cell helps to prevent loss of the patch due to tissue movement during stimulation electrode placement.

6

Two-Photon In Situ Calcium Imaging of Glioma Cells In vivo calcium imaging provides temporal resolution at the millisecond range and spatial resolution at micrometer range, allowing the study of neuronal and glial activity [26]. The technique offers several advantages over traditional electrophysiological methods, such as the ability to monitor multiple cells simultaneously and to obtain temporal and spatial resolution. However, there are some limitations to calcium imaging, such as potential phototoxicity and the need for specialized equipment and data analysis techniques. Despite these challenges, calcium imaging remains a powerful tool for investigating the mechanisms underlying neuron-glioma interactions. Genetically encoded calcium indicators (GECIs) can be expressed in specific cell types or subcellular compartments to assess neuronal activity over time. GCaMP, a GFP-based GECI, is the most widely used GECI in biological systems [27]. Recent advances have improved GCaMP through several rounds of mutagenesis, resulting in GCaMP6, which can detect single-action potentials in vivo. GCaMP6 has three variations with slow, medium, or fast fluorescence changes upon calcium binding. GCaMP6S is the most sensitive GECI with slow kinetics, while GCaMP6F is one of the fastest GECIs and has a sensitivity comparable to synthetic calcium indicator OGB1-AM [28].

6.1

Materials

6.1.1

Solutions

6.1.2

Equipment

1. Artificial cerebrospinal fluid (ACSF): 125 mM NaCl, 2.5 mM KCl, 25 mM NaHCO3, 1.25 mM NaH2PO4, 25 mM glucose, 1 mM MgCl2, and 2 mM CaCl2, pH 7.4. 1. Prairie Ultima XY Two-Photon Slice Rig, including the following: Olympus BX-61W upright microscope, 10× Olympus U Plan Fl N lens, 40× Olympus LUM Plan Fl W/IR-2 lens. 2. Iso-flex stimulus isolator (A.M.P.I.). 3. Slice anchor (Warner Instruments).

6.1.3

Software

1. EZcalcium (https://github.com/porteralab/EZcalcium). 2. Detection of peaks in data (https://nbviewer.org/github/ demotu/BMC/blob/master/notebooks/DetectPeaks. ipynb). 3. scikit-image: Image processing in Python [29].

362

6.2

Kiarash Shamardani et al.

Methods

For calcium imaging, genetically encoded calcium indicator GCaMP6s was lentivirally transduced into glioma cells (pLV-ef1GCaMP6s-P2A-nls-tdTomato). In this case, glioma cells containing the GCaMP6s reporter can be identified using the tdTomato nuclear tag. 1. Prepare acute slices of the brain as described in Subheading 4.2. 2. Set the temperature of the ACSF perfusion media to 30 °C, oxygenate with carbogen (95% O2, 5% CO2), and perfuse through the chamber at rate of 2 mL/min. 3. With plastic transfer pipette (cut the tip to widen the opening), transfer a slice from the incubation bath to the imaging chamber. Fix the slice in the chamber by inserting a slice anchor (harp) over the tissue. 4. Set the excitation light wavelength to 920 nm on a tunable Ti: Sapphire laser to allow for excitation of both tdTomato and GCaMP6s. 5. Using the 10× objective and transmitted light locate the area of interest for imaging. Then, using the fluorescent reflected light locate the cells expressing tdTomato and GCaMP6s (see Note 1). 6. After locating the imaging region of interest, move to the 40× objective and focus on the tissue. 7. Turn on live imaging via the software interface and set the laser power, PMT voltages, dwell time, and Pockels cell to obtain an optimal image (see Note 2). 8. Record at ~1.5 Hz (0.65 frames/s) for 30 min to record spontaneous activity and for 10 min to record response to periodic electrical stimulation (Fig. 4, left). 9. For neuronal stimulation experiments, place a bipolar stimulating electrode connected to an Iso-flex stimulus isolator near the cells of interest (similar to the electrophysiology paradigm). Deliver approximately 20 μA over 200 μs. 10. Analyze the two-photon live image sequences with a software package of choice, for instance, EZcalcium. Perform motion correction, ROI detection, and compute ΔF/F and deconvolved traces. To find the local maxima of calcium signals, first smooth the data using a 20-frame window and then find the peaks. Synchronicity between cells can be measured by determining the synchronous cells, the number of synchronous communications, and the time point of the synchronous firing using a window of six frames around each peak using the scikitimage library (Fig. 4, right).

Studying Synaptic Integration of Glioma Cells into Neural Circuits

363

Fig. 4 (Left) Two-photon in situ calcium imaging of hippocampal slice xenografted with GCaMP6s-expressing glioma (SU-DIPG-XIII-FL); 30 min (m) time course without axonal stimulation. Red denotes glioma tdTomato nuclear tag; green denotes glioma GCaMP6s. Scale bar, 50 μm. (Right) Phase-locked traces of GCaMP6s intensity over time in four synchronous glioma cells with axonal stimulation (red line) [16]

6.3

Notes

1. Even though cells are not firing, background fluorescence of GCaMP6s should be visible as very dim green cells in the transfected cells. 2. For our two-photon setup, optimal settings were: Pockels cell set at 10, and PMTs set at 800 for each channel. For these settings, power at back aperture of the objective was approximately 30 mW at 920 nm. The wavelength ranges for the emission filters were PMT1: 607 nm center wavelength with 45 nm bandpass (full-width at half-maximum) and PMT2: 525 nm center wavelength with 70 nm bandpass (full width at half maximum). Adjust the parameters so you can clearly see the fluorescence of GCaMP6s without suturing pixels in your field of view.

7

Immunoelectron Microscopy Neuron-to-glioma signaling plays a key role in tumor pathophysiology, influencing growth, invasion, initiation, and metastasis. The advantage of culturing two distinct cellular populations, such as neurons and glioma cells, is the investigation of their direct interaction without additional influences from other cell types within the environment. Compared to in vivo modeling, neuron-glioma cocultures offer a relatively high throughput assay to perturb these cellular interactions in a reproducible system.

364

Kiarash Shamardani et al.

Immunoelectron microscopy (Immuno-EM) is a powerful technique that combines the specificity of immunological methods with the high-resolution imaging capabilities of electron microscopy. It involves labeling specific molecules or structures of interest with antibodies that are conjugated to electron-dense markers, such as gold particles, which can then be visualized using an electron microscope. It allows the visualization of specific molecules or structures of neurons and glioma cells at high resolution. It enables us to study the interaction between neurons and glioma cells such as visualizing the ultrastructural changes like synapse formation that occur during this interaction [16]. Immuno-EM can also be used to localize and identify specific proteins or other molecules within a cell or tissue sample, providing insights into their subcellular distribution and functional roles. 7.1

Materials

7.1.1

Solutions

1. Karnovsky’s fixative: 2% glutaraldehyde (Electron Microscopy Services, #16000) and 4% paraformaldehyde (Electron Microscopy Services, #15700) in 0.1 M sodium cacodylate (Electron Microscopy Services, #12300), pH 7.4. 2. 21% osmium tetroxide (Electron Microscopy Services, #19100). 3. Ethanol (50%, 75%, 95%, and 100%). 4. Acetonitrile. 5. EMbed-812 resin (Electron Microscopy Services, #14120). 6. 10% periodic acid. 7. 10% sodium metaperiodate in H2O. 8. 0.5 M glycine. 9. Blocking solution: 0.5% BSA, 0.5% ovalbumin in phosphatebuffered saline solution (PBST). 10. PBST. 11. 3.5% uranyl acetate in 50% acetone. 12. 0.2% lead citrate.

7.1.2

Equipment

1. TAAB capsules. 2. Leica Ultracut S (Leica) ultramicrotome. 3. 100-mesh Ni grids (EMS FCF100-Ni). 4. Parafilm. 5. JEOL JEM-1400 (or similar).

7.1.3

Antibodies

Transmission

Electron

Microscope

1. Primary antibody: anti-GFP (1:300 in blocking solution; MBL International) (See Note 1). 2. Secondary antibody: 10 nm gold-conjugated IgG (15,732, TED Pella).

Studying Synaptic Integration of Glioma Cells into Neural Circuits

7.2

Methods

7.2.1 Tissue Collection and Fixation

365

Allow tumors to establish for several weeks after orthotopic xenografting (approximately 8–12 weeks for pediatric glioma, see Note 2), as described in Subheading 3.2. Harvest tissue in a fume hood: 1. Euthanize the mice by transcardial perfusion with 10 mL of ice-cold PBS, followed by 10 mL of ice-cold Karnovsky’s fixative. 2. Remove the brain and store in Karnovsky’s fixative at 4 °C.

7.2.2 Sample Preparation

1. Microdissect the brain region of interest containing the tumor and control sample (see Note 3). 2. Post-fix the sample in 1% osmium tetroxide for 1 h at room temperature. 3. Wash three times with ultrafiltered water. 4. En bloc stain the sample for 2 h at room temperature. 5. Dehydrate the sample in graded ethanol (50%, 75%, 95%) for 15 min each at 4 °C. 6. Equilibrate the sample to room temperature. 7. Rinse the sample in 100% ethanol twice. 8. Incubate the sample in acetonitrile for 15 min at room temperature. 9. Combine Embed-812 resin in a 1:1 ratio with acetonitrile. Incubate the sample in the resin: acetonitrile mix for 1 h. 10. Remove and incubate the sample in a 2:1 Embed-812 resin: acetonitrile mix for 2 h. 11. Remove and incubate in Embed-812 resin for 2 h. 12. Place sample into a TAAB capsule with fresh resin and incubate at 65 °C overnight. 13. Section the embedded sample at 40 or 60 nm on a ultramicrotome. 14. Mount tissue slices on 100-mesh Ni grids.

7.2.3 Immunohistochemistry

All steps are performed at room temperature, unless otherwise stated. 1. Microetch the sample in 10% periodic acid for 15 min. 2. Wash sample three times in ultrapure water for 5 min each. 3. Elute osmium by incubating in 10% sodium metapriodate for 15 min. 4. Wash sample three times in ultrapure water for 5 min each. 5. Quench with 0.5 M glycine for 10 min. 6. Wash twice with PBS for 2 min each. 7. Incubate in blocking solution (0.5% BSA, 0.5% ovalbumin) for 20 min.

366

Kiarash Shamardani et al.

8. Incubate in primary antibody overnight at 4 °C. 9. Wash twice in blocking solution for 10 min each. 10. Wash twice in PBS for 10 min each. 11. Incubate in secondary antibody for 1 h at room temperature. 12. Wash three times in PBS for 5 min each. 13. Rinse twice in ultrapure water. 14. Contrast stain in 3.5% uranyl acetate in 50% acetone for 30 s, followed by 0.2% lead citrate for 90 s. 15. Dry the grids on a filter paper and store them in a grid box until imaging. 7.2.4 Imaging and Analysis of Neuron-Glioma Synaptic Structures

7.3

Notes

The prepared and stained samples are imaged using the Transmission Electron Microscope at 120 kV. Glioma cells are identified by immunogold particle labeling (presence of 3–5+ particles). Structures are confirmed by meeting the following criteria: (1) presence of presynaptic vesicles, (2) visually clear synaptic cleft, (3) a clear postsynaptic density present in the immunogold labeled glioma cell (Fig. 5). 1. The primary antibody used to detect the tumor cells can be directed at a marker introduced into the cells, such as GFP, or an endogenous marker distinct to the tumor cells (e.g., human nestin). 2. Tumors generated from different glioma cultures should be characterized by IF/IHC to confirm diffuse tumor burden and appropriate timepoint post-xenograft for tissue collection and analysis.

Fig. 5 Immunoelectron microscopy of patient-derived DIPG cells SU-DIPG-VI (left) and SU-DIPG-XIII-FL (right; FL denotes frontal lobe tumor) xenografted into the mouse hippocampus. Arrowheads denote immuno-gold particle labeling of GFP. Postsynaptic density in GFP+ tumor cells (pseudo-colored green), synaptic cleft, and clustered synaptic vesicles in opposing presynaptic neuron (pseudocolored magenta) identify synapses. Scale bars, 2 μm [16]

Studying Synaptic Integration of Glioma Cells into Neural Circuits

367

3. For control samples, collect additional slices with no tumorcells, and stain sample with GFP-containing cells with secondary antibody only to assess the background signal. This will help to identify the number of immunogold particles that indicate positive staining.

8

Conclusion The emerging field of cancer neuroscience is focused on elucidating the central role of the nervous system in the pathophysiology of cancer, including primary brain tumors and brain metastases. A more complete understanding of neuron-glioma interactions requires an interdisciplinary approach bringing together methods used in neuroscience and cancer biology. Here, we described various approaches that can be used to study the complex neuronglioma interaction and discussed their advantages and disadvantages.

Acknowledgments The authors gratefully acknowledge support from the National Institute of Neurological Disorders and Stroke (R01NS092597 to M.M.), NIH Director’s Pioneer Award (DP1NS111132 to M. M.), National Cancer Institute (P50CA165962, R01CA258384, U19CA264504), Robert J. Kleberg, Jr., and Helen C. Kleberg Foundation (to M.M.), Cancer Research UK (to M.M.). References 1. Venkatesh H, Monje M (2017) Neuronal activity in ontogeny and oncology. Trends Cancer 3:89–112. https://doi.org/10.1016/j.trecan. 2016.12.008 2. Hanahan D, Weinberg RA (2011) Hallmarks of cancer: the next generation. Cell 144:646– 674. https://doi.org/10.1016/j.cell.2011. 02.013 3. Monje M, Mitra SS, Freret ME et al (2011) Hedgehog-responsive candidate cell of origin for diffuse intrinsic pontine glioma. Proc Natl Acad Sci U S A 108:4453–4458. https://doi. org/10.1073/pnas.1101657108 4. Schwartzentruber J, Korshunov A, Liu X-Y et al (2012) Driver mutations in histone H3.3 and chromatin remodelling genes in paediatric glioblastoma. Nature 482:226–231. https:// doi.org/10.1038/nature10833 5. Sturm D, Witt H, Hovestadt V et al (2012) Hotspot mutations in H3F3A and IDH1

define distinct epigenetic and biological subgroups of glioblastoma. Cancer Cell 22:425– 437. https://doi.org/10.1016/j.ccr.2012. 08.024 6. Flavell SW, Greenberg ME (2008) Signaling mechanisms linking neuronal activity to gene expression and plasticity of the nervous system. Annu Rev Neurosci 31:563–590. https://doi. org/10.1146/annurev.neuro.31.060407. 125631 7. Hasel P, Dando O, Jiwaji Z et al (2017) Neurons and neuronal activity control gene expression in astrocytes to regulate their development and metabolism. Nat Commun 8:15132. https://doi.org/10.1038/ncomms15132 8. West AE, Greenberg ME (2011) Neuronal activity–regulated gene transcription in synapse development and cognitive function. Cold Spring Harb Perspect Biol 3:a005744.

368

Kiarash Shamardani et al.

https://doi.org/10.1101/cshperspect. a005744 9. Campbell SL, Buckingham SC, Sontheimer H (2012) Human glioma cells induce hyperexcitability in cortical networks. Epilepsia 53:1360– 1 3 7 0 . h t t p s : // d o i . o r g / 1 0 . 1 1 1 1 / j . 1528-1167.2012.03557.x 10. Buckingham SC, Campbell SL, Haas BR et al (2011) Glutamate release by primary brain tumors induces epileptic activity. Nat Med 17: 1269–1274. https://doi.org/10.1038/nm. 2453 11. John Lin C-C, Yu K, Hatcher A et al (2017) Identification of diverse astrocyte populations and their malignant analogs. Nat Neurosci 20: 396–405. https://doi.org/10.1038/nn.4493 12. Yu K, Lin C-CJ, Hatcher A et al (2020) PIK3CA variants selectively initiate brain hyperactivity during gliomagenesis. Nature 578:166–171. https://doi.org/10.1038/ s41586-020-1952-2 13. Robert SM, Buckingham SC, Campbell SL et al (2015) SLC7A11 expression is associated with seizures and predicts poor survival in patients with malignant glioma. Sci Transl Med 7:289ra86. https://doi.org/10.1126/ scitranslmed.aaa8103 14. Venkatesh HS, Johung TB, Caretti V et al (2015) Neuronal activity promotes glioma growth through neuroligin-3 secretion. Cell 161:803–816. https://doi.org/10.1016/j. cell.2015.04.012 15. Venkatesh HS, Tam LT, Woo PJ et al (2017) Targeting neuronal activity-regulated neuroligin-3 dependency in high-grade glioma. Nature 549:533–537. https://doi.org/10. 1038/nature24014 16. Venkatesh HS, Morishita W, Geraghty AC et al (2019) Electrical and synaptic integration of glioma into neural circuits. Nature 573:539– 545. https://doi.org/10.1038/s41586-0191563-y 17. Venkataramani V, Tanev DI, Strahle C et al (2019) Glutamatergic synaptic input to glioma cells drives brain tumour progression. Nature 573:532–538. https://doi.org/10.1038/ s41586-019-1564-x 18. Ka´rado´ttir R, Cavelier P, Bergersen LH, Attwell D (2005) NMDA receptors are expressed in oligodendrocytes and activated in ischaemia. Nature 438:1162–1166. https://doi.org/10. 1038/nature04302 19. Shultz LD, Lyons BL, Burzenski LM et al (2005) Human lymphoid and myeloid cell

development in NOD/LtSz-scid IL2Rγnull mice engrafted with mobilized human Hemopoietic stem cells. J Immunol 174:6477–6489. https://doi.org/10.4049/jimmunol.174.10. 6477 20. Fuchs Q, Batut A, Gleyzes M et al (2021) Co-culture of glutamatergic neurons and pediatric high-grade glioma cells into microfluidic devices to assess electrical interactions. J Vis Exp (177). https://doi.org/10.3791/62748 21. Bahrami F, Janahmadi M (2013) Antibiotic supplements affect electrophysiological properties and excitability of rat hippocampal pyramidal neurons in primary culture. Iran Biomed J 17:101–106. https://doi.org/10.6091/ibj. 11242.2013 22. He C, Xu K, Zhu X et al (2021) Patientderived models recapitulate heterogeneity of molecular signatures and drug response in pediatric high-grade glioma. Nat Commun 12:4089. https://doi.org/10.1038/s41467021-24168-8 23. Lin GL, Monje M (2017) A protocol for rapid post-mortem cell culture of diffuse intrinsic pontine glioma (DIPG). J Vis Exp (121): e55360. https://doi.org/10.3791/55360 24. Taylor K, Barron T, Hui A et al (2023) Glioma synapses recruit mechanisms of adaptive plasticity. Nature 623:366. https://doi.org/10. 1038/s41586-023-06678-1 25. Barron T, Yalc¸ın B, Mochizuki A et al (2022) GABAergic neuron-to-glioma synapses in diffuse midline gliomas. bioRxiv. https://doi. org/10.1101/2022.11.08.515720 26. de Melo Reis RA, Freitas HR, de Mello FG (2020) Cell calcium imaging as a reliable method to study neuron–glial circuits. Front Neurosci 14:569361. https://doi.org/10. 3389/fnins.2020.569361 27. Cichon J, Magrane´ J, Shtridler E et al (2020) Imaging neuronal activity in the central and peripheral nervous systems using new Thy1.2GCaMP6 transgenic mouse lines. J Neurosci Methods 334:108535. https://doi.org/10. 1016/j.jneumeth.2019.108535 28. Chen T-W, Wardill TJ, Sun Y et al (2013) Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499:295–300. https://doi.org/10.1038/nature12354 29. van der Walt S, Scho¨nberger JL, NunezIglesias J et al (2014) scikit-image: image processing in Python. PeerJ 2:e453. https://doi. org/10.7717/peerj.453

INDEX A Acute tissue slice ........................................ 210–218, 301, 333, 347, 357, 362 α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor......................................... 35, 36, 53–69, 86, 208, 212, 213, 256, 283, 287, 288, 291, 292, 313–341, 346, 360 Amperometry ....................................................92, 93, 98, 106, 108–110 Astrocyte .......................................... 5, 96, 123, 156, 173, 208, 229, 268, 350 Astrocyte-neuron interactions .....................174–176, 194 Axon-glial synapses ....................................................... 346

B Biosensors ....................................4, 9, 17, 19, 93, 95, 98, 101, 105, 106, 108, 109, 111, 115, 117, 169

C Calcium/calmodulin-dependent protein kinase II (CaMKII).......................... 17, 72, 78–80, 86, 213 Calcium (Ca2+) imaging ..................................... 347, 357, 358, 361–363 Calcium signaling................................................. 157, 293 Cancer neuroscience ..................................................... 367 Carbon fiber microelectrode (CFE).............................. 93, 96, 101–106, 113, 114 Chemical label .................................................. 53–69, 337 Chemogenetic modulation................................. 155–157, 167, 169, 170 Computational neuroscience........................................ 257 Computational simulation ................................... 295–299 Confocal microscopy .............................................. 22, 23, 165, 166, 337 Corticostriatal pathway ................................................211, 212, 215, 218

D Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) ................................ 156–165, 167–170

Differentiation......................................36, 268, 302, 314, 315, 334–341, 352 Direct stochastic optical reconstruction microscopy (dSTORM) ...................................................36–38, 41, 43, 45–48 Dissociated culture..............................122, 124–128, 131

E Electron microscopy (EM) ...............................17, 36, 38, 122, 123, 135–143, 145–151, 250, 252, 313, 347, 348, 364 Electrophysiology..............................................72, 73, 92, 117, 165, 250, 277, 293, 295, 328–333, 347, 357, 358, 362 Excitatory Amino Acid Transporter type 2 (EAAT2) ......................................... 208–210, 212, 215, 218, 220, 221 Exocytosis .................................. 5, 53, 54, 62, 86, 91–94, 96, 98, 99, 108–112, 114–117

F Fluorescence lifetime imaging (FLIM).................................72–76, 82–84, 86, 87 Fluorescence microscopy ....................................... 17, 144 Fluorescent visualization............................. 54, 60, 65, 69 Fusion pore dynamics ................................. 92, 93, 96, 99

G G-deleted rabies racing ........................................ 303, 304 Genetically encoded indicators................................... 3–32 Glioma .................................................................. 345–367 Glutamate ...............................................3, 35, 53, 72, 95, 137, 160, 208, 231, 267, 283, 313, 346 Glutamate clearance ................................... 208, 209, 212, 214–218, 221–224, 236 Glutamate oxidase (GluOx) .......................................... 93, 96, 97, 101, 105 Glutamate receptors (GluRs) ................................. 5, 6, 9, 13, 31, 35, 36, 38, 39, 41, 43, 44, 46–48, 50, 53, 212, 232, 237, 238, 246, 254, 256, 284 Glutamatergic synapse ......................................3, 5, 6, 24, 31, 96, 150, 168, 169, 194, 209, 234, 244, 258

Maria Kukley (ed.), New Technologies for Glutamate Interaction: Neurons and Glia, Neuromethods, vol. 2780, https://doi.org/10.1007/978-1-0716-3742-5, © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Science+Business Media, LLC, part of Springer Nature 2024

369

NEW TECHNOLOGIES FOR GLUTAMATE INTERACTION: NEURONS AND GLIA

370 Index

Glutamatergic transmission .........................................231, 236, 240, 242 Glutamatergic vesicles....................................................... 6 Glutamate sensor...........................................6–10, 30, 93, 96, 97, 101–109, 111, 112, 115, 212, 213, 224 Glutamate uncaging....................................................... 77, 208, 219, 288–295 Gold nanoparticles (AuNPs) .................................. 96, 97, 101, 102, 105, 113, 114

H High-grade glioma (HGG) ........................ 345, 346, 353 Huntington’s disease (HD)................................ 208, 209, 212, 218–221, 224 Hypokinesia ................................................................... 212

I IGluSnFR ............................ 8, 10, 12–32, 169, 208, 213 Image acquisition ......................... 47, 127, 219, 308, 337 Imaging...................................................4, 36, 56, 72, 92, 122, 138, 196, 208, 250, 289, 308, 347 In vivo gene delivery ..................................................... 320 Ion-flux independent ............................72, 73, 77, 79, 83 Ionotropic receptors ............................................ 3, 71, 86

L Ligand-directed chemistry........................................53–69 Live fluorescence imaging ........................................72, 82 Long-term depression (LTD) ....................................... 54, 71–73, 77–79, 83

M Metabotropic receptors (mGluRs)..............................3, 5, 35, 37, 232, 235, 239, 244, 256 Microscopy .......................................9, 12, 17, 20, 22–24, 54, 63, 66, 72, 122, 128, 129, 132, 166, 230, 242, 248, 274, 277, 289, 308, 310, 337, 364, 366 Microtubules .......................................121, 122, 124–132 Mini singlet oxygen generator (miniSOG) ................136, 137, 141, 142, 144, 148–150 Monosynaptic tracing .......................................... 301–311

N Nanoclusters .................................................................... 48 Nanodomains ..................................................... 36–38, 48 Nanoscale organization ..................................... 36–38, 47 Neural network mapping..................................... 233, 302 Neuron-glia communication ............................... 230, 267 Neuron-glia synapses .......................................................vii Neuron-glioma coculture .................................... 347–353 Neuron-glioma interactions ........................................346, 347, 349, 352, 361, 367

O Oligodendrocyte precursor cells (OPCs) ..................................................... 267–269, 274–278, 280, 281, 283–288, 290–299, 301–304, 306–311, 313–315, 326, 328–341 Oligodendrocytes ....................................... 185, 193, 267, 268, 277, 283, 302, 304, 309, 313, 346, 350 OPC synapses ...............................................283–297, 302 Optogenetics ................................ 28, 268, 279, 281, 311

P Patch-clamp ..........................................98, 106, 108, 112, 116, 250, 270, 271, 274, 277, 280, 331–333 Patient-derived xenograft ........................... 346, 347, 353 Photooxidation ................................................... 136–140, 142–145, 148–150 Premyelinating oligodendrocytes........................ 336, 340 Proliferation........................................................... 15, 126, 268, 302, 309, 314, 315, 334–341, 346, 347, 351, 353 Protein phosphatase 1 (PP1)....................................72, 78 Protein-protein interactions ....................... 72, 76, 85, 86

R Rabies tracing ...................................................... 302–304, 306, 308, 309, 311 Retrovirus ............................................................ 315–328, 330–332, 334–336, 338–340

S Serial block-face scanning EM (SBEM)......................135, 144–146, 149, 150 Simulation .................................................. 230, 237, 242, 249, 250, 252, 257, 295, 296, 298, 299 Single-molecule localization microscopy (SMLM)........................................... 36, 44, 47, 48 STED microscopy ...............................122, 124, 128–131 Stereotaxic injection............................................. 306, 328 Subsynaptic domains....................................................... 48 Super-resolution microscopy ......................................... 36, 43, 122, 129, 268 Synapse ......................................................... 3, 36, 77, 91, 122, 141, 168, 173, 208, 229, 267, 283, 301, 313, 346 Synaptic connectivity ............................17, 304, 308, 311 Synaptic integration .....................................................284, 286, 288, 292, 295–299 Synaptic network mapping ........................................... 308 Synaptic plasticity ..............................................31, 38, 72, 77, 79, 86, 93, 165, 167, 232 Synaptic signaling.......................................................... 236

NEW TECHNOLOGIES

FOR

GLUTAMATE INTERACTION: NEURONS

GLIA Index 371

AND

T

V

Transcriptomics .................................................... 176, 196 Transgenic mice.............................................98, 218–221, 248, 268, 274, 275, 304, 314, 326, 331, 334, 341 Transmission EM (TEM) ............................................136, 140, 144, 145, 147, 149, 150 Tripartite synapses............................................... 208–211, 229, 231–236, 238–241, 246, 252, 253 Tubulin .......................................126, 127, 129, 131, 132 Two-photon live microscopy......................................... 11, 23, 24, 80, 250, 361–363

Vesicle quantal size.........................................94, 111–112 Vesicular glutamate transporter (VGLUT2) .............................................. 136–138, 142, 144, 149, 150, 313 Voltage-gated ion channels............................................. 12

W White matter...............................242, 277, 302, 309, 313 Whole-cell patch-clamp ...................................... 268, 270, 275, 278, 332, 346, 359