Patch-Clamp Applications and Protocols

NEUROMETHODS 0 26

Patch-Clamp Applications and Protocols

NEUROMETHODS 0 26

Patch-Clamp Applications

and Protocols Edited by

Alan A. Boulton Uniuersity of Saskatchewan, Saskatoon, Canada

Glen B. Baker Unioersity of Alberta, Edmonton, Canada

and

Wolfgang Walz University of Saskatchewan, Saskatoon, Canada

Humana Press * Totowa, New Jersey

0 1995 The Humana Press Inc. 999 Riverview Drive, Suite 208 Totowa, New Jersey 07512

All rights reserved. No part of this book may be reproduced, stored m a retrieval system, or transmitted in any form or by any means, electromc, mechanical, photocopying, ml- crofilming, recording, or otherwise without written permisslon from the Pubhsher.

All authored papers, comments, opmions, conclusions, or recommendations are those of the author(s) and do not necessarily reflect the views of the publtsher

This publication is prmted on acid-free paper. B ANSI 239.48-1984 (American National Standards Institute) Permanence of Paper for Prmted Library Materials.

Photocopy Authorization Policy: Authorization to photocopy items for internal orpersonaluse, or the internal or personal use of specific clients, is granted by Humana Press Inc., provided that the base fee of US $4.00 per copy, plus US $00.20 per page, 1s paid directly to the Copyright Clearance Center at 222 Rosewood Drive, Danvers, MA 01923. For those organizations that have been granted a photocopy license from the CCC, a separate system of payment has been arranged and is acceptable to Humana Press Inc. The fee code for users of the Transactional Reportmg Service is: [O-89603-311-2/95 $4.00 + $00.20]. All rights reserved.

Printed in the United States of America,

ISBN O-89603-311-2 ISSN 0893-2336

Preface to the Series

When the President of Humana Press first suggested that a series on methods in the neurosciences might be useful, one of us (AAB) was quite skeptical; only after discussions with GBB and some searching both of memory and library shelves did it seem that perhaps the publisher was right. Although some excellent methods books had recently ap- peared, notably in neuroanatomy, it is a fact that there was a dearth in this particular field, a fact attested to by the alac- rity and enthusiasm with which most of the contributors to this series accepted our invitations and suggested additional topics and areas. After a somewhat hesitant start, essentially in the neurochemistry section, the series has grown and will encompass neurochemistry, neuropsychiatry, neurology, neuropathology, neurogenetics, neuroethology, molecular neurobiology, animal models of nervous disease, and no doubt many more “neuros.“Although we have tried to include adequate methodological detail and in many cases detailed protocols, we have also tried to include wherever possible a short introductory review of the methods and/or related substances, comparisons with other methods, and the relationship of the substances being analyzed to neurologi- cal and psychiatric disorders. Recognizing our own limitations, we have invited a guest editor to join with us on most volumes in order to ensure complete coverage of the field. These editors will add their specialized knowledge and com- petencies. We anticipate that this series will fill a gap; we can only hope that it will be filled appropriately and with the right amount of expertise with respect to each method, substance or group of substances, and area treated.

Alan A. Boulton Glen B. Baker

V

Preface

E. Neher and B. Sakman were the first to monitor the opening and closing of single ion channels and membranes by conductance measurements. In 1976, they used firepolished micropipets with a tip diameter of 3-5 pm to record currents from a small patch of the membranbe of skeletal muscles, thereby decreasing background membrane noise. In order to reduce the dominant source of background noise-the leakage shunt under the pipet rim between membrane and glass- the muscle membrane had to be treated enzymatically. Despite these early limitations, a new technique was born -the patch-clamp technique. The final breakthrough came in 1981 when the same authors, in collaboration with 0. P. Hamill, A. Marty, and F. J. Sigworth, developed the gigaohm seal. Not only did this improve the quality of recordings, it was now possible to gently pull the membrane patch with the attached pipet off the cell and study its trapped ion channels in isolation. Another offshoot of the gigaohm seal technique was the whole-cell patch-clamp technique, in which the patch is ruptured without breaking the seal. This technique is really a sophisticated voltage-clamp technique and also allows for the altering of cytoplasmic constituents if the experimenter so wishes.

The first part of Patch-Clamp Applications and Protocols presents modern developments associated with the technology of patch-clamp electrodes, of cell-free ion channel recording, and of the whole-cell patch-clamp technique. Chapters on recent offsprings of the patch-clamp method, such as the concentration clamp technique, the pressure clamp method, and the perfusion of patch-clamp electrodes, were written for this Neuromethods volume by authors who were intimately involved in their development. Two increasingly widely used methods are the loose-patch-clamp technique and the perforated patch-clamp technique. An important addition is the

vii

Preface

patch clamp technique in brain slices, developed by A. Konnerth et al., who wrote a dedicated chapter for the present book. Finally, molecular biological aspects of the patch-clamp technique are covered by two additional chapters, one on Xenopus oocyte microinjection and ion channel expression, and one on patch-clamp recording and RT-PCR on single cells.

The patch-clamp method is certain to be refined further in the future, as new applications involving the manipulation of cellular constituents, molecular biological techniques, and the various imaging techniques emerge.

Wolfgang Walz

Contents

Preface to the Series .................................................................................... V

Preface ........................................................................................................... Vii

List of Contributors ................................................................................ xvii

Technology of Patch-Clamp Electrodes Richard A. Leuis and James L. Rae

1. Introduction .............................................................................................. 1 2. General Properties of Pipet Glass ........................................................ 3 3. Whole-Cell Pipet Properties: Practical Aspects ................................. 5

3.1. Choice of Glass ............................................................................... 5 3.2. Pulling Whole-Cell Electrodes .................................................... 7 3.3. Elastomer Coating Whole-Cell Electrodes ................................ 7 3.4. Firepolishing Whole-Cell Electrodes ......................................... 8

4. Patch Electrode Fabrication for Single-Channel Recording.. ........ 10 4.1. Choice of Glass ............................................................................. 10 4.2. Pulling Single-Channel Electrodes ........................................... 11 4.3. Coating Single-Channel Pipets with Elastomers ................... 11

4.4. Firepolishing Single-Channel Pipets ........................................ 13 4.5. Fabrication Methods Specific to Quartz .................................. 13 4.6. Low-Noise Recording ................................................................. 15

5. Noise Properties of Patch Pipets ........................................................ 17 5.1. Noise Contribution of the Pipet ................................................ 17 5.2. Thin-Film Noise ........................................................................... 19 5.3. Distributed RC Noise .................................................................. 21 5.4. Dielectric Noise ............................................................................ 24

5.5. R -C Noise .................................................................................... 28

5.6. Sialhoise ...................................................................................... 29

5.7. Summary of Plpet Noise Sources ............................................. 30 5.8. Noise Sources for Whole-Cell Voltage Clamping ................ .32 References ............................................................................................... 36

Whole-Cell Patch-Clamp Recordings Harald Son theimer

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..*..*..*.....**..*.*.*.........* 37

2. Principles (Why Voltage Clamp?) .,..,.,.,,..,,.................,......*............... 38

3. Procedure and Techniques ,,......,,............*.... ..*.*......,..............*.*.......... 39

3.1. Pipets . . ..**.................*......*....... .,...,....,.................,.,..,,..................... 39 3.2, Electronic Components of a Setup ,.*..........*..............*....*....*.... 40

ix

Con tents

3.3. Recording Configuration ........................................................... 42 3.4. Experimental Procedure ............................................................. 44

4. Data Evaluation and Analysis ............................................................ 46 4.1. Data Filtering/Conditioning, Acquisition, and Storage .... ..4 6 4.2. Leak Subtraction ....................................................................... ...4 9 4.3. Determination of Cell Capacitance ......................................... .52 4.4. Dissecting Current Components .............................................. 52 4.5. I-V Curves ..................................................................................... 57 4.6. Fitting of Time-Constants ......................................................... .63 4.7. Data Presentation.. ...................................................................... .64

5. Limitations, Pitfalls, and Errors .......................................................... 65 5.1. Series Resistance and Its Consequences .................................. 65 5.2. Voltage Clamp Errors ................................................................. 68

6. Special Applications ............................................................................. 70 7. Conclusions ............................................................................................ 72

Acknowledgments ................................................................................ 72 Recommended Readings ..................................................................... 72 References ............................................................................................... 73

Pressure/Patch-Clamp Methods Owen P. Hamill and Don W. McBride, Jr.

1. Introduction ............................................................................................ 75 2. General Cell-Attached Patch Recording Procedures.. ................... .76 3. Methods of Applying Suction ............................................................. 77

3.1. Steady-State Methods ................................................................. 77 3.2. Perturbation Methods ................................................................. 78

4. Properties of the Pressure Clamp ....................................................... 79 4.1. Stimulation Protocols ................................................................. .79 4.2. Speed of the Pressure Clamp.. .................................................. .81 4.3. Sensitivity and Noise of the Pressure Clamp ........................ .82 4.4. Range of the Pressure Clamp .................................................... 82

5. Applications of Pressure/Patch-Clamp Methods .......................... .83 5.1. Sealing Protocols and Determination of Functional

Membrane-Cytoskeleton Interactions .................................... .83 5.2. Membrane Viscoelastic and Mechanical Properties ............. 84 5.3. Characterization of Mechano-Gated Channels ..................... .84

6. Conclusion .............................................................................................. 85 Acknowledgments ................................................................................ 85 References ............................................................................................... 85

Cell-Free Ion-Channel Recording C. G. Nichols, M. B. Cannell, and A. N. Lopatin

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..~............................. 89 2. Making an Inside-Out Membrane Patch:

The Problem of Vesicle Formation . ..*..*........*.....*..*.........*...*....*... 90

Con tents xi

2.1. How to Tell When You Have a Vesicle ................................... 90 2.2. How to Deal with the Problem ................................................ 94

3. Analysis of Ion Channels in Cell-Free Patches: Dealing with the Problem of Channel “Rundown” ........................................ 95 3.1. Mechanisms of Rundown .......................................................... 97 3.2. Accounting for Rundown: Statistical Approaches

fo I Analyzing Current Records ................................................. 98 4. Methods for Rapid Change of the Solution Bathing Cell-Free

Membrane Patches .............................................................................. 101 4.1. M’ethods for Rapid Bulk Application of Solution ................ 102 4.2. Laminar Flow Methods of Separating Parallel Solutions . .103 4.3. Separation of Solutions Using an “Oil-Gate” ...................... .103 4.4. Separation of Solutions Using an “Air Gate” ..................... ..lO 6

5. Analysis of Responses to Rapid Concentration Changes.. ......... .107 5.1. Modeling the Pipet Geometry ................................................. 107 5.2. Time Course of Solution Change: The Effects

of Pipet Geometry ...................................................................... 108 5.3. Experimental Measurement of Diffusion Delays ............... .109 5.4. Correcting for Diffusion Delays in Analysis

of Concentration Jump Experiments ...................................... 112 5.5. Advantages and Disadvantages of the Analysis ................ -113

6. Twenty Eight Hints and Tips for Successful Cell-Free Ion-Channel Recording’ ...................................................................... 114

7. Concluding Remarks .......................................................................... 118 References ............................................................................................. 119

Perfusion of Patch Pipets John M. Tang, F. N. Quandt, and R. S. Eisenberg

1. Introduction .......................................................................................... 123 2. Methods ................................................................................................. 124

2.1. Patch-Clamp of Neuroblastoma Cells ................................... 124 2.2. In ternal Perfusion Technology ................................................ 125

3. Results ................................................................................................... 129 3.1. Time Course of Exchange of Internal Solution ................... .129 3.2. Efficiency of Exchange of Internal Solution ........................ .129 3.3. Selectivity of K Channels Measured

by Reversal Potentials ............................................................... 131 3.4. Pharmacology of 4-Aminopyridine ....................................... 133 3.5. Parameters Controlling the Rate of Exchange

of Solution ................................................................................... 134 4. Discussion ............................................................................................. 135

4.1. Applicability of the Technique.. ............................................. .135 4.2. Possible Problems ...................................................................... 137 4.3. Improvements ............................................................................ 138 References ............................................................................................. 139

xii Contents

Concentration Clamp Techniques Norio A kaike

1. Introduction .......................................................................................... 141 2. Setup of Rapid Solution Change ...................................................... 142 3. Preparations ......................................................................................... 144

3.1. Whole-Cell Recording Mode ................................................... 144 3.2. Excised Cell Membranes .......................................................... 146

4. Kinetic Studies Using Concentration Clamp Technique ............ .146 4.1. Receptor-Mediated Ionic Currents ......................................... 146 4.2. Voltage-Dependent Ionic Currents ........................................ 147 4.3. Rapid Change of Physical Conditions ................................... 148 4.4. G-Protein Mediated Response ................................................. 150 4.5. Measurement of Ca2+ Release from Intracellular

Ca2+ Store Sites ............................................................................ 151 5. Limitations ............................................................................................ 151

References ............................................................................................. 151

Perforated Patch-Clamp Techniques Wolfgang Walz

1. Introduction .......................................................................................... 155 2. Dialysis of Cytoplasm by the Patch Micropipet Filling

Solution ................................................................................................. 155 3. Strategies Used to Prevent Dialysis ................................................. 158

3.1. Increase of Pipet Resistance ..................................................... 158 3.2. Addition of a Cytosolic Extract to the Pipet Solution ......... 159 3.3. Use of ATP .................................................................................. 160 3.4. Use of Polyene Antibiotics ....................................................... 160

4. Use of Nystatin .................................................................................... 160 4.1. Properties of Nystatin Pores .................................................... 160 4.2. Perforating the Patch Membrane with Nystatin ................. .162 4.3. Intrapipet Dialysis of Nystatin ............................................... .163 4.4. Composition of Pipet Solutions .............................................. 164 4.5. Detailed Protocol for Use of Nystatin .................................... 165 4.6. Use of a Nystatin-Fluoroscein Mixture ................................. 166

5. Use of Amphotericin B ....................................................................... 167 6. Special Application: The Perforated Vesicle .................................. 168 7. Conclusions .......................................................................................... 169

Acknowledgment ................................................................................ 170 References ............................................................................................. 170

The Loose Patch Voltage Clamp Technique J. H. Calduel and R. L. Milton

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 2. Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174

2.1. Amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174

Contents *.a

Xl11

2.2. Pipets ............................................................................................ 177 2.3. Equipment.. ................................................................................ .179 2.4. Procedure for Performing an Experiment ............................. 180 2.5. Possible Sources of Error .......................................................... 182

3. Variations of the Method ................................................................... 186 3.1. Concentric Electrodes ............................................................... 186 3.2. Current Collector ....................................................................... 187 3.3. Ionophoresis with Loose Patch ............................................... 188

4. Conclusion ............................................................................................ 190 Acknowledgments ............................................................................. ,190 References ............................................................................................. 190

Patch-Clamp Recording and RT-PCR on Single Cells Bertrand Lambolez, Etienne Audinat, Pascal Bochet, and Jean Rossier

1. Introduction .......................................................................................... 193 2. Materials and Methods ...................................................................... 195

2.1. Labware, Reagents ..................................................................... 195 2.2. Solutions ...................................................................................... 196 2.3. Design of the Oligos .................................................................. 198 2.4. Thermocycle PCR Programs .................................................... 199 2.5. Test of the Sensitivity of the PCR ........................................... 200 2.6. Contamination ........................................................................... ,201 2.7. Electrophysiology and Cellular RNA Harvesting...............20 2 2.8 RT-PCR on Single-Cell Step by Step ...................................... 204

3. AMPA Receptor Subunits in Purkinje Cells: * GFAI? in Glial Cells ............................................................................ .206 3.1. Experimental Procedures ......................................................... 206 3.2. Specificity of the PCR and Quantification ............................ 215 3.3. Proportional Amplification of the Fragments ..................... ,216 3.4. Discussion .................................................................................. ,217

4. AMPA Receptor Subunits and GAD in Hippocampal Cells.......21 8 4.1. Experimental Procedures ......................................................... 219 4.2. Results .......................................................................................... 222 References ............................................................................................ ,229

Patch-Clamp Technique in Brain Slices T. D. Plant, J. Eilers, and A. Konnerth

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

2.1. Brain Slices for Patch-Clamp Studies ..*..*......*....*.*............*.... 234 2.2. Patch-Clamp Recording in Brain Slices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 2.3. Combinations of the Patch-Clamp Techniques

in Slices with Other Methods ..a . . . . . . . . t.......... . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..*...................................................... 256

xiv Con tents

Xenopus Oocyte Microiqjection and Ion-Channel Expression T. G. Smart and B. J. Krishek

1. Introduction .......................................................................................... 259 1.1. History of the Xenopus laevis Oocyte as an Expression

System .......................................................................................... 259 1.2. Application to Ion Channel Expression Studies .................. 260

2. Husbandry of Xenopus laevis ............................................................. 261 2.1. Source and Identification of Xenopus laevis .......................... 261 2.2. Housing and Environment ..................................................... ,263 2.3. Feeding ........................................................................................ 264 2.4. Diseases and Parasites ............................................................. .265

3. Removal of Ovary Tissue ................................................................... 267 3.1. Anesthesia of Xenopus laevis .................................................... 267 3.2. Removal of Oocytes .................................................................. .268 3.3. Preparation of Oocytes for Injection ...................................... 272 3.4. Removal of the Vitelline Membrane ..................................... ,274

4. Selection of Oocytes ........................................................................... ,275 4.1. Stages of Oocyte Development ............................................... 276 4.2. Separation of Stage V/VI Oocytes.. ........................................ 276

5. Microinjection of Xenopus Oocytes .................................................. 277 5.1. Inlection Equipment ................................................................. ,277 5.2. Preparation of Glassware and RNA/DNA Solutions ....... ,278 5.3. Fabrication of Injection Micropipets ..................................... ,279 5.4. Cytoplasmic RNA Injection ..................................................... 281 5.5. Nuclear DNA Injection ............................................................. 282 5.6. Optimization of Receptor/Ion-Channel Expression.. ....... ..28 4

6. Culture/Incubation of Injected Oocytes ......................................... 285 6.1. Optimal Culture Conditions for Protein Expression.. ...... ..28 6

7. Electrophysiological Recording from Xenopus Oocytes.. ............ ,287 7.1. Two-Electrode Voltage Clamp ................................................ 287 7.2. Patch-Clamp Recording.. ......................................................... .292

8. Comparison of Xenopus Oocytes with Alternative Expression Systems ................................................................................................. ,296 Appendix 1: Composition of Physiological Solutions .................. 297 Acknowledgments .............................................................................. 300 References ............................................................................................. 301

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307

Contributors

NORIO AKAIKE l Department of Physiology, Faculty of Medicine, Kyushu University, Fukuoka, Japan

ETIENNE AUDINAT l Laborutoire de Neurobiofogie et de Neuropharmacologie du De’veloppement, Centre National de la Recherche Scientifique, Orsay, France

PASCAL BOCHET l Institut Alfred Fessard, Centre National de la Recherche Scientifque, Gifsur-Yvette, France

J. H. CALDWELL l Departments of Cellular and Structural BioZogy and Physiology, and the Neuroscience Program, University of Colorado School of Medicine, Denver, CO

M. B. CANNELL l Department of Pharmacology and Clinical Pharmacology, St. George’s Hospital Medical School, London, UK

J. EILERS l I. Physiologisches Institut, Universitizt des Saarlandes, Homburg, Germany

R. S. EISENBERG l Department of Molecular Biophysics and Physiology, Rush Medical College, Chicago, IL

OWEN P. HAMILL l Department of Physiology and Biophysics, The University of Texas Medical Branch, Galveston, TX

A. KONNERTH l I. Physiologisches Institut, Universitat des Saarlandes, Homburg, Germany

B. J. KRISHEK * Department of Pharmacology, The School of Pharmacy, London, UK

BERTRAND LAMBOLEZ l Znstitut Atfred Fessard, Centre National de la Recherche Scienttfique, Gif-sur-Yvette, France

RICHARD A. LEVIS l Department of Physiology, Rush Medical College, Chicago, IL

A. N. LOPATIN l Department of Cell Biology and Physiology, Washington University School of Medicine, St. Louis, MO

DON W. MCBRIDE, JR. l Department of Physiology and Biophysics, The University of Texas Medical Branch, Galveston, TX

R. L. MILTON l Indiana University School of Medicine, Muncie Centerfor Medical Education, Bull State University, Muncie, IN

C. G. NICHOLS l Department of Cell Biology and Physiology, Washington University School of Medicine, St. Louis, MO

xv

XVi Contributors

T. D. PLANT l I. Physiologisches Institut, Universitiit des Saarlandes, Horn burg, Germany

F. N. QUANDT l Department of Molecular Biophysics and Physiology, Rush Medical College, Chicago, IL

JAMES L. RAE l Departments of Physiology, Biophysics, and Ophthalmology, Mayo Foundation, Rochester, MN

JEAN ROSSIER l lnstitut Alfred Fessard, Centre National de Za Recherche Scientifique, Gif-sur-Yvette, France

T. G. SMART l Department of Pharmacology, The School of Pharmacy, London, UK

HARALD SONTHEIMER l Yale University School of Medicine, New Haven, CT

J. M. TANG l Department of Molecular Biophysics and Physiology, Rush Medical College, Chicago, IL

WOLFGANG WALZ l Department of Physiology, College of Medicine, University of Saskatchewan, Saskatoon, Canada

Technology of Patch-Clamp Electrodes

Richard A. Leois and James L. Rae

1, Introduction The extracellular patch voltage clamp technique has

allowed the currents through single ionic channels to be studied from a wide variety of cells. In its early form (Neher and Sakmann, 1976), the resolution of this technique was limited by the relatively low (-50 MR) resistances that isolated the interior of the pipet from the bath. The high resolution that presently can be achieved with the patch-clamp technique originated with the discovery (Neher, 1981) that very high- resistance (tens or even hundreds of GS2) seals can form between the cell membrane and the tip of a clean pipet when gentle suction is applied to the pipet interior. Although the precise mechanisms involved in this membrane-to-glass seal are still not fully understood, the importance of the GS2 seal is obvious. The high resistance of the seal ensures that almost all of the current from the membrane patch flows into the pipet and to the input of the current-sensitive headstage preamplifier. It also allows the small patch of membrane to be voltage-clamped rapidly and accurately via the pipet, and the mechanical stability of the seal is vital to the whole-cell voltage clamp technique. Of equal importance, the high resistance of the seal greatly reduces the noise it contributes to single-channel measurements. Although the seal can often represent only a small fraction of total patch-clamp noise (particularly as the bandwidth of recording increases), its importance should never be minimized. Without such high

From- Neuromethods, Vol. 26. Patch-Clamp Applications and Protocols Eds. A. Boulton, G. Baker, and W. Walz Q 1995 Humana Press Inc

1

2 Levis and Rae

resistance seals, most of the steady progress to reduce background noise levels would not have been possible.

Of course, the patch pipet is not simply a tool in the formation of GQ seals. The pipet serves as a fluid bridge that connects the current-sensitive headstage amplifier input to the surface or interior of the cell. The insulating properties (both resistive and, more importantly, capacitive) of the glass that forms the wall of the pipet are also crucial to the ability to measure current originating in the patch and to the background noise levels that can be achieved.

For any patch-clamp measurement, several steps are required to construct a proper glass electrode. First, a glass that has optimal properties is selected. The required properties differ substantially for single-channel recordings and whole-cell current recordings. For single-channel measurements, low noise is the most important electrical parameter, whereas for whole-cell measurements dynamic performance is more important than the contribution of the electrode to the background noise. This is simply because the background noise in a whole-cell recording is dominated by the noise from the electrode resistance (actually, the access resistance) in series with the capacitance of the entire cell. The dynamic bandwidth of a whole-cell recording also depends on the same factors. Therefore, the goal in constructing an electrode for whole-cell recording is simply to make it as blunt and as low in resistance as is compatible with sealing it to the cell. In single-channel recordings, the pipet is a major contribu- tor to the background notice and so requires many subtle considerations to produce an electrode optimal for recording single-channel currents.

As a second step in pipet construction, the electrode glass stock is pulled into a pipet with a tip of optimal geometry. This geometry differs for whole-cell and single-channel recordings. In a third step, the outside wall of the pipet is coated with a hydrophobic elastomer possessing good electrical properties. This procedure is essential for low noise single-channel recordings, but can be done much less carefully for whole-cell recordings. Fourth, the tip is firepolished

Patch-Clamp Electrode Technology 3

to round it and clean its surface of any thin film of elastomer coating. This step can also be used to adjust the final tip diameter. Firepolishing promotes seal formation but often is not required. After all these procedures, the electrode can be filled and used.

Several general properties of glasses must be considered when trying to construct optimal electrodes for patch-clamping (see Table 1). Thermal properties determine the ease with which desired tip shapes can be produced and they determine how easily the tips can be heat polished. Optical properties often result in a distinct visual endpoint so that tips can be firepolished the same way each time. Electrical properties are important determinants of the noise the glass produces in a recording situation and determine the size and number of components in the capacity transient following a change of potential across the pipet wall. Glasses are complex substances composed of many compounds and most of their properties are determined to a first order by the composition of the glass used. Glass composition may also influence how easily a glass seals to membranes and whether or not the final electrode will contain compounds leached from the glass into the pipet filling solution, which can activate, inhibit, or block channel currents.

2. General Properties of Pipet Glass

Before proceeding to the details of electrode fabrication, it is useful to consider in more detail glass properties that are important for patch-clamp pipet construct ion. We will begin with thermal properties. It is important that glasses soften at a temperature that is easily and reliably achieved. This formerly was a stringent constraint, since glasses like aluminosilicates, which melt at a temperature in excess of 9OO”C, would shorten the lifetime of a puller heating filament so much that their use was unattractive. Quartz, which melts above 16OO”C, could not even be pulled in commercially available pullers and so was not used at all. Today, at least one puller exists that will do these jobs easily (P-2000,

4 Levis and Rae

Table 1 Glass Properties

Glass

L%o Loss volume Dielectric Softening factor resistivity constant temp., Co Description

7940 SO038 11.8 3.8 1724 .0066 13.8 6.6 7070 -25 11.2 4.1 8161 .50 12.0 8.3 Sylgard .58 13.0 2.9 7059 .584 13.1 5.8 7760 .79 9.4 4.5 EG-6 .80 9.6 7.0 0120 .80 10.1 6.7 EG-16 .90 11.3 9.6 7040 1.00 9.6 4.8 KG-12 1.00 9.9 6.7 1723 1.00 13.5 6.3 0010 1.07 8.9 6.7 7052 1.30 9.2 4.9 EN-l 1.30 9.0 5.1 7720 1.30 8.8 4.7

1580 926 -

604 -

844 780 625 630 580 700 632 910 625 710 716 755

7056 1.50 10.2 5.7 720 3320 1.50 8.6 4.9 780

7050 1.60 8.8 4.9 705 KG-33 2.20 7.9 4.6 827 7740 2.60 8.1 5.1 820 1720 2.70 11.4 7.2 915 N-51A 3.70 7.2 5.9 785 R-6 5.10 6.6 7.3 700 0080 6.50 6.4 7.2 695

Quartz (fused silica) Aluminosilicate Low loss borosilicate High lead #184 Coating cmpd. Barium-borosilicate Borosilicate High lead High lead High lead Kovar seal borosilicate High lead Aluminosilicate High lead Kovar seal borosilicate Kovar seal borosilicate Tungsten seal

borosilicate Kovar seal borosilicate Tungsten seal

borosilicate Series seal borosihcate Kimax borosilicate Pyrex borosilicate Alummosilicate Borosilicate Soda lime Soda lime

Sutter Instruments, Navato, CA) and so virtually any kind of glass can be used routinely. It is generally true that the lower the melting temperature of the glass, the more easily it can be firepolished. Low-melting-temperature glasses, such as those with high lead content, can be pulled to have tip diameters in excess of 100 pm and still be firepolished to a small enough tip diameter that the pipet can be sealed to a


7-10 pm diameter cell. With such glasses, one has greater control over the final shape of the tip than is possible with higher melting temperature borosilicate glasses. Quartz pipets cannot be firepolished with a usual firepolishing apparatus, although with care they can be firepolished in a temperature-controlled flame.

Electrical properties are most important for providing low noise as well as low amplitude, simple time-course capacity transients. As will be discussed later, it is not possible to achieve low background noise without an elastomer coating the outside of the pipet. In general, glasses with the lowest dissipation factors have minimal dielectric loss and produce the lowest noise. There is a wide variety of glasses to choose from that will produce acceptable single channel recordings, although quartz is clearly the best material to date. Good electrical glasses are also necessary for whole- cell recordings, not because of noise properties, but because they result in the simplest and most voltage- and time-stable capacity transients.

Major chemical constituents in glass are important since they determine the overall properties of the glass and because they are potential candidates to leach from the glass into the pipet filling solution where they can interact with the channels being studied. No glass can be deemed to be chemically inert, since even tiny amounts of materials leached in the vicinity of the channels may produce sufficient local concentrations to interact with channels and other cellular processes. Again, quartz would be expected to have fewer chemical impurities than other glasses, but every kind of glass should be suspected of having an effect on the channels being measured.

3. Whole-Cell Pipet Properties: Practical Aspects

3.1. Choice of Glass

Modern computerized pipet pullers are capable of pulling glass with almost any thermal properties (with the exception of quartz) into the proper blunt-tipped geometry

6 Levis and Rae

that is ideal for whole-cell recording. Therefore, almost any glass can be used to form whole-cell pipets. Nevertheless, we feel that some types of glass should usually be avoided, whereas others have some particularly useful properties for this application.

Soda lime glasses, such as Kimble R-6 and Corning 0080, generally should not be used because of their high dielectric loss. When a voltage step is applied across a patch pipet fabricated from one of these glasses, there will be a large slow component in the resulting capacity transient (Rae and Levis, 1992a). For a 2-mm depth of immersion with a moderate coating of Sylgard 184 to within -200 pm of the tip, we have found following a 200 mV voltage stop that is a slow component for a soda lime pipet can be as large as 50 pA 1 ms after the beginning of the step. The slow tail of capacity current can still be as much as 10 pA 10 ms after the step and may require as much as 200 ms to decay to below 1 pA. The time-course of this slow tail is not exponential, but more closely approaches a logarithmic function of time. In addition, we have observed that for soda lime pipets the magnitude of the slow component of capacity current is not always constant during a series of pulses that occur at rates faster than about 1-2/s. Instead, the magnitude of this component is sometimes observed to decrease with successive pulses. Because of these characteristics, these capacitive currents can possibly be mistaken for whole-cell currents. Heavy Sylgard coating can reduce the amplitude of the slow component of capacity current for soda lime glasses, but it is generally better (and certainly more convenient) simply to use glasses with lower loss factors (see Rae and Levis, 1992a, for further discussion).

High-lead glasses, such as 8161, EG-6, EG-16,0010,0120, and KG-12, possess much lower loss factors than soda lime glasses and are particularly useful because of their low melting point. This property allows the construction of initially very large-tipped pipets that subsequently can be firepolished to blunt bullet-shaped tips offering the lowest possible access resistance. This, of course, minimizes series resistance. In


addition, pipets of this shape also draw in the largest surface area patch of membrane when suction is applied. This is useful in perforated patch recordings, since the larger area of membrane available for partitioning by amphotericin or nystatin results in the maximum incorporation of perforation channels and thus the lowest access resistance. KG-12 (Friedrich and Dimmock, Millville, NJ) is a good choice for glasses of this class, since it seals well, has good electrical properties, and is readily available.

Pipets for whole-cell recording can be thin-walled by comparison to those for single-channel recording. In whole- cell measurements, other sources of noise far outweigh the contribution from the pipet per se (see Section 5.8). In terms of total background noise, the major consideration in pipet fabrication is simply achieving the lowest possible resistance. Glass with an OD/ID ratio of 1.2-1.4 will have lower resistance for a given outside tip diameter than will thicker-walled glass, and is therefore useful for whole-cell recording. Some precautions are necessary, however, since if the walls become too thin the pipet will more easily penetrate the cell during the attempt to form a seal.

Other glasses that have been successfully used by many laboratories for whole-cell recording include Pyrex (Corn- ing [Corning, NY] #7740), Kimble’s Kimax, and Corning 7052. Although we usually prefer the high-lead glasses described earlier, these glasses have produced perfectly acceptable results. Note, however, that Corning no longer makes 7052 and so existing supplies will be depleted within a few years.

3.2. Pulling Whole-Ceil Electrodes

This can be done on any commercially available electrode puller. Here one simply strives for as blunt a taper and as large a tip diameter as is compatible with sealing of the electrode to the cell.

3.3. Elastomer Coating Whole-Cell Electrodes

Elastomer coating of electrodes reduces electrode noise in single-channel recordings. In whole-cell recordings, the

8 Levis and Rae

noise associated with electrode glass is usually insignificant in comparison to other noise sources and so elastomer coating is not required for noise reduction. Elastomer coating also reduces electrode capacitance. Commercial patch- clamp amplifiers have the ability to compensate about 10 pF of electrode capacitance. For pipets made from glasses with high dielectric constants (e.g., soda lime and high-lead glasses) immersed deeply into a tissue bathing solution, the electrode (and holder) capacitance may exceed the compensation range of the electronics. Elastomer coating will help to keep the total electrode capacitance within the compensation range. For whole-cell recordings, it is not usually necessary to paint the elastomer close to the tip. Coat- ing that extends from the top of the shank to 1 mm from the tip is sufficient for whole-cell recordings. Many investigators do not use elastomer coating for whole-cell recordings.

3.4. Firepolishing Whole-Cell Electrodes

Finally, to promote GS2 seals and to reduce the possibility of tip penetration into the cell during seal formation, electrode tips should be firepolished. In some cells, firepolishing has proven unnecessary, but we have found that sealing is generally promoted by firepolishing the electrode tip, particularly for cells where seal formation is difficult. Whole-cell and single-channel electrodes are firepolished with the same basic apparatus. Firepolishing can be done either using an upright or an inverted microscope. In fact, many investigators have chosen to coat their pipets and firepolish them using an inverted microscope with a 40x or so long working distance objective.

Another very useful approach is to utilize a standard upright microscope converted to the 210-mm tube length that is standard for metallurgical microscopes. Several microscope companies, but particularly Nikon (Garden City, NY), make extra long working distance and super long working distance high magnification metallurgical objectives. Most noteworthy are the 100x ELWD or 100x SLWD


that have l- mm and 2-mm working distances, respectively. With these objectives and 15x eyepieces and with the electrode mounted on a slide held in the mechanical stage of the microscope, it is possible to move the electrode tip into the optical field and directly visualize the electrode tip at 1500x magnification. At such high magnifications, it is possible to firepolish the tip to a very distinct optical endpoint under direct visualization. This approach ensures very repeatable results from one electrode to the next. The firepolishing itself is accomplished by connecting to a micromanipulator a rod of inert material to which has been fas- tened a short loop of platinum iridium wire. The ends of this wire must be soldered to two other pieces of wire that can be connected to a voltage or current source to allow current to be passed through the platinum wire. The platinum loop generally is bent into a very fine hairpin so that it can be brought to within a few millimeters of the electrode tip under direct observation. Because of early reports that platinum can be sputtered from the wire onto the electrode tip and prevent sealing, the platinum wire is generally coated with a glass like Pyrex (Corning #7740) or Corning #7052 to prevent such sputtering. This is done by overheat- ing the platinum wire and pushing against it a piece of electrode glass that has been pulled into an electrode tip. At high temperatures, the glass melts and flows over the platinum wire ends up thoroughly coating it and forming a distinct bead of glass. If the elastomer has been coated too near the tip, firepolishing causes the tip to droop downward at the juncture where the coating ends. If one desires to paint elastomer extremely close to the tip, it may be necessary to do the majority of the firepolishing before coating and then firepolish lightly again afterward. As a general rule, firepolishing with the electrode tip close to the heating wire at low temperature produces a tip whose inner walls are parallel and relatively close together. With a hotter heating element and the tip farther away, the tip tends to round more and end up quite blunt.

10 Levis and Rae

4. Patch Electrode Fabrication for Single-Channel Recording

4.1. Choice of Glass A limited number of glasses are available for single-chan-

nel patch-clamping. Perhaps the most important feature to consider is the amount of noise in the recording that is owing to the pipet itself. This subject is sufficiently important that we include an entire section dealing with noise sources in pipets in the hope that readers will be able to use the principles to make optimal pipets for their own recording situation. There is no longer any question, however, that quartz is the best glass if noise performance is important. Quartz itself is quite expensive and requires an expensive laser-based puller, and so probably is not the glass for routine studies. Therefore, we consider other glasses here as well. Garner Glass (Claremont, CA) has been particularly helpful in the development of specialty glasses for patch-clamping, although they are no longer able to provide any of the high- lead glasses we find so useful. Any glass tubing selected for the fabrication of patch electrodes should have walls of substantial thickness. Wall thickness results in decreased electrical noise and increased bluntness at the tip, which prevents penetrating the cell during seal formation. Glass tubing with an OD/ID of 2.0-3.0 is easily obtainable and is expected to yield the lowest background noise levels. Gen- erally, the outside diameter chosen is 1.5-1.7 mm. For single- channel recordings, only the glasses with the best electrical properties should be used if optimal noise performance is desired. Corning glasses #8161 and #7760 are particularly good in this regard, but again Corning no longer makes them and the existing supplies are extremely limited. Corning #7052 is also quite acceptable but also will not be available for much longer. Sadly, most of the options for particularly low-noise glasses are running out, and so quartz is expected to become increasingly more attractive even given its cost. Readily available glasses, like Corning 7740 or Kimble’s Kimax, are not particularly quiet glasses. High-lead glasses


like Kimble’s KG-12 give better signal-to-noise ratios than the Pyrex-type glasses, but are substantially worse than the best glasses mentioned earlier.

In our experience, it is usually unnecessary to clean electrode glasses prior to pulling. On occasion, however, normally quiet pipet glasses are found to be noisy in use, and it is imperative to clean the glass for best noise performance. Sonicating the glass in 100% ethanol or methanol in an ultra- sonic cleaner is effective for this purpose. Following any cleaning procedure, it is a good idea to place the glass in an oven at around 200°C for lo-30 min to achieve complete drying. Heat treatment of this sort has also proven necessary if low-noise recordings are required in environments where the humidity is exceptionally high.

4.2. Pulling Single-Channel Electrodes

Single-channel pipets made from glasses other than quartz can be pulled on any commercially available patch electrode puller. Here the tips can be less blunt and higher in resistance. The electrode resistance in series with the patch capacitance is a potential noise source (see Section 5.5). How- ever, as will be seen, this source of noise actually may be minimized by using high-resistance pipets insofar as such high resistance correlates with a small patch area. In addition, sharper tips taper, often leading to higher resistance seals to the membrane. Thus, for best noise performance for single-channel recording it is better not to use the blunt electrode tips that are good for whole-cell situations.

4.3. Coating Single-Channel Pipets with Elastomers

For the lowest noise recordings, electrodes must be coated with a hydrophobic elastomer to within 100 pm or less of their tip. The closer it can be painted to the tip the better. This coating prevents bathing solution from forming a thin fluid film along the outer surface of the electrode. This thin film of bathing solution would be a substantial noise source. A commonly used compound is Sylgard #184 (Dow Corning, Midland, MI). Sylgard also has exceptional electri-

12 Levis and Rae

cal properties (see Table 1) and so improves the electrical properties of most glasses when a thick coat covers the glass surface. Sylgard, meticulously mixed, can be stored at -20°C in small capped centrifuge tubes. The thorough mixing is required to prevent pockets of the compound not adequately exposed to polymerizer. This unpolymerized elastomer can flow to the electrode tip (even against gravity) and render the tips difficult to seal. At freezer temperatures, the mixed Sylgard can be stored for several weeks. A tube of this freezer- stored Sylgard, when brought to room temperature for use in painting electrodes, will last for several hours before it begins to polymerize. Care must be taken not to open the tube until the contents have reached room temperature to prevent water condensation. Condensed water can degrade the electrical properties of the elastomer and increase noise. The Sylgard is applied to the electrode tip with a small uten- sil, such as a piece of capillary tubing pulled to a reasonably fine tip in a flame. Sylgard is applied using dissecting microscopes at magnifications of 10-30x. It is useful, but not required, to modify the dissecting microscope to work in a dark field. This can be done inexpensively with a fiberoptic ring illuminator connected to a fiberoptic light source. The ring illuminator is placed under the stage of the microscope. Three to four inches above the ring light, dark-field illumination is achieved and the walls of the electrode glass show up as bright lines of light against a dark background. Both the Sylgard coat and the tip of the electrode are easily seen with this dark-field illumination. The Sylgard must be directed away from the tip by gravity at all times during the painting procedure or the Sylgard may flow over the tip to make firepolishing and/or sealing impossible. The Sylgard can be cured by holding the tip for 5-10 s in the hot air stream ema- nating from a standard heat gun like those used in electronics to heat shrink tubing. Again, the Sylgard must be gravita- tionally directed away from the tip during this curing process.

Although Sylgard is the most commonly used elastomer, there are a number of other elastomers available that are as good as Sylgard in most respects and better in others.


RTV615A from General Electric has properties nearly identical to Sylgard and can be used in exactly the same way as is Sylgard. Dow Corning Medical Silastic MDX-4 has dielectric properties slightly better than Sylgard #184 but polymerizes more rapidly at freezer temperatures. To date, it has not offered any obvious improvement in noise on a day-to-day basis, but several of the lowest noise measurements done with quartz electrodes utilized this elastomer. It is considerably more expensive than Sylgard #184. Dow Corning #R-6101 is another excellent elastomer, which costs more to buy, but probably not to use, than Sylgard. R-6101 is useful because it does not polymerize appreciably at room temperature and so can be used for up to 2-3 mo without freezing. Its noise properties are as good (should be a little better) as Sylgard #184 and it does result in low noise when used with quartz or some other very good electrode glass. Teflon AF (DuPont, Wilmington, DE) is a Teflon-based coating material with dielectric properties claimed to better than Sylgard. Its solvent must be obtained from 3-M and both the compound and its solvent are expensive. However, it offers some potential to improve electrode noise when procedures are worked out to use it optimally.

4.4. Firepolishing Single-Channel Pipets

The same principles apply here as in the firepolishing of whole-cell electrodes. The same apparatus is used for both. In general, patch electrodes are firepolished with the tip close to the heating filament with the goal of thickening the glass near the tip in addition to rounding it. For high resistance seals, it may be useful to firepolish so that the internal walls of the tip become parallel for several microns. This mode of firepolishing will increase the tip resistance a few Ma but will often result in lower noise because of higher resistance seals (see also Section 5.5).

4.5. Fabrication Methods Specific to Quartz

Quartz softens at about 16OO”C, and so no platinum or nichrome wire-based heat source will melt it because both of

14 Levis and Rae

these materials disintegrate long before 1600°C is reached. Quartz can be pulled in a flame, but the tip geometry is unreliable with such fabrication techniques. The new laser- based P-2000 electrode puller from Sutter Instruments generates enough heat to pull quartz fairly easily. It begins to have trouble when the glass OD exceeds 1.5 mm. It has no difficulty pulling quartz tubing with an OD/ID = 3 so long as the OD does not exceed 1.5 mm. Since the major reason to use quartz patch pipets is for the reduction of single-channel background noise currents, it is best to use quartz with as thick a wall as possible. A 1.5-mm OD with 0.5 mm ID produces about the smallest bore that is practical. Even at 0.5 mm ID, there is some difficulty with the internal Ag-AgCl electrode since it must be made of such flimsy silver wire that it is often damaged (bent) or denuded of silver chloride as the electrode is placed into the small bore. IDS of 0.6-0.75 mm make the pipets much easier to use.

Quartz cannot be firepolished easily with any presently available commercial apparatus. Those that firepolish other glasses, including aluminosilicate, do not generate enough heat to firepolish quartz. It is possible to firepolish it in a carefully controlled Bunsen burner, but that approach is sufficiently unreliable that it is best to try to pull tips whose geometry is good enough to allow sealing without firepolishing. That places an additional constraint on the puller, since most other glass pullers need only to produce electrode tips that are approximately correct since the final tip geometry can be customized while firepolishing. With quartz, the tips must be good enough for use immediately after pulling.

Because of the noise produced by a thin film of bathing solution creeping up the outer surface of an electrode, quartz must be elastomer-coated like any other glass. This bathing solution film is such a large noise source that if an elastomer coating is not used to reduce it, there is absolutely no reason to use quartz electrodes for patch-clamping. It will not perform appreciably better than poor glasses if this noise source is not eliminated or minimized. Because quartz must be elas-


tomer coated, it must also be subjected to the heat polisher. Although the polisher cannot smooth or round the quartz tip as it does with other glasses, it can burn off any residual elastomer and so should be used with quartz electrodes just before filling.

4.6. Low-Noise Recording

Low-noise recording requires meticulous attention to detail. Even with an electrode optimally pulled, coated, and firepolished, there are still many ways in which excess noise can creep in. It is important that the electrodes be filled only to just above the shank. Fluid in the back of the electrode can cause internal noise-generating films and allow fluid into the holder. It is important for low-noise recordings that a suction line with a syringe needle the correct size to fit into the bore of the pipet be maintained near the experimental setup. This suction line can be used to vacuum fluid from the pipet and ensure none gets into the holder or coats the majority of the back of the electrode. Alternatively, silicone fluid or mineral oil can be used to fill the electrode for a short distance in back of its filling solution. These “oils” are somewhat messy and not really required if a proper suction line is used. The internal electrode should be adjusted in length until its tip just comfortably is immersed in the filling solution. In general, the shorter the length of the internal electrode (and of the pipet), the lower the noise will be. Therefore, it is best to use the shortest possible holder and electrode that is practical.

During experiments where low noise is required, it is best to test the noise at intermediate stages. Most modern patch-clamp amplifiers have a root mean square noise meter that can be checked to determine the noise levels at any time. This meter should be checked immediately after inserting the electrode into the holder and placing the electrode tip over the bath but before actually immersing the tip in the bath. Poorly filled electrodes, fluid in the holder, a generally dirty holder, and pickup from the environment will show up as elevated noise. What the actual level of the noise will be depends on the noise of your patch-clamp, the kind of

16 Levis and Rae

holder and electrode glass you are using, and on how well you have shielded against pickup of electrical interference. Specific examples appear in Levis and Rae (1993). As a general rule, however, total noise in this situation should not be more than -lo-20% above that of the open circuit headstage. If you see excess noise, you can remove the electrode, dry the internal electrode, and then test the noise with only the headstage and holder placed above the bath. If this is elevated above what is normal for your setup, either your holder is dirty or you are experiencing pickup from the environment. Environmental pickup often can be seen as noise spikes at discrete frequencies, whereas a dirty holder contributes noise across a broad range of frequencies. You can try to dry the holder by blowing dry, clean air through it, but it is possible that you will have to clean the holder before the noise will go down. This can be done by disassembling it, sonicating it in ethanol, and drying it for several hours in an oven at 60-70°C. Because of the time involved in cleaning the holder, it is wise to have two or more holders available when attempting very low-noise recordings.

The noise of your electronics, holder, electrode glass, and elastomer can be determined by making a thin pad of Sylgard and placing it in the bottom of your chamber. Then seal your electrode to it much as you would sealing to a cell. No suction, however, is required to make the seal. Simply push the tip against the Sylgard and a seal forms. The seal should be 200 GQ or more if you have done it correctly. Under these circumstances, the seal noise is essentially negligible and you are able to quantify the remaining composite noise sources. This noise will depend on how deep the bathing solution is: The deeper the bathing solution, the greater the noise. For most purposes, the bath depth need not be more than 1-3 mm. This simple procedure will let you know what is routinely possible with your setup and give you a baseline for comparing the noise you actually get in experiments. A good seal to a cell will often produce noise that is about the same as the noise you get sealed to Sylgard.

Note, however, that as soon as the electrode tip is placed in the bath, the noise will be enormous since you are now


measuring at best the thermal noise of a l-10 MSZ resistance tip. The readings on the noise meter will not be meaningful until you have obtained a GQ seal. If the seal resistance is ~20 GC2, the majority of the noise will be owing to the seal and really low-noise recordings cannot be achieved.

5. Noise Properties of Patch Pipets

5.1. Noise Contribution of the Pipet

The earliest patch pipets were fabricated from “soft” soda lime glasses. Such glasses were easy to pull and heat polish to any desired tip geometry, primarily because they soften at relatively low temperatures. Unfortunately, such pipets introduced relatively large amounts of noise into patch-clamp measurements. It was soon found that “hard” borosilicate glasses produced less noise, but, owing to their softening at higher temperatures, were somewhat more difficult to pull and heat polish. Probably as a result of these early findings, it has sometimes been assumed that “hard” high-melting- temperature glasses necessarily have better electrical properties than “soft” low-melting-temperature glasses. However, there is no obligatory relationship between the thermal and electrical properties of glass. For example, several low-melting-temperature high-lead glasses (e.g., 8161, EG-6) have been shown to produce less noise than a variety of high-melting- temperature borosilicate and aluminosilicate glasses (e.g., 7740,172O). The reason for these findings becomes clear when the electrical properties of the glasses are considered.

The electrical properties of glass that are important to its noise performance are its dielectric constant and its dissipation factor; the bulk resistivity of a glass might also be important, but is usually sufficiently high to be ignored. The dielectric constant of a substance is the ratio of its permittivity to the permittivity of a vacuum. Thus, for pipets of equivalent geometry and depth of immersion, the higher the dielectric constant of the glass, the higher the pipet capacitance. The dielectric constants for glasses commonly used for patch pipet fabrication range from 3.8 for quartz to more than 9 for some high-lead glasses. The dielectric constant of boro-

18 Levis and Rae

silicates is typically 4.5-6, whereas that of soda lime glasses is near 7. The pipet capacitance generates noise by several mechanisms that will be described later. The dissipation factor is a measure of the lossiness of a dielectric material. Ideal capacitors display no dielectric loss and do not generate thermal noise. However, all real dielectrics are lossy and do produce thermal noise; we refer to this as dielectric noise. Glasses with the lowest dissipation factors are the least lossy and generate the least dielectric noise. Quartz is among the least lossy of all practical dielectrics; its dissipation factor, which is in the range of lo”-lo”, is far lower than that of other glasses used for patch pipets. Several high-lead glasses have dissipation factors of -10”. The dissipation factor of boro- silicates that have been used successfully to fabricate patch pipets varies from about 0.002-0.005. Soda lime glasses have the highest dissipation factor (-O.Ol), which is the principal reason for their high noise.

The best glasses for patch pipet fabrication are those with the best electrical properties, i.e., low dissipation factor and low dielectric constant. However, understanding pipet noise requires more than simply understanding the electrical properties of glass. A variety of other factors also influence the noise performance of the patch pipet, e.g., pipet geometry, depth of immersion, and the type and extent of elastomer coating. Here we will summarize our present understanding of all major pipet noise sources; more detailed discussions can be found elsewhere (Levis and Rae, 1992,1993; Rae and Levis, 1992a,b).

Attaching the electrode holder to the headstage input will slightly increase noise above its minimum level associated with an open circuit input. The mechanisms involved in generating this noise are discussed elsewhere (Levis and Rae, 1993). Here we only note that the contribution of the holder by itself to total patch-clamp noise should be very small. Holder noise is minimized by constructing the holder from low-loss dielectric materials, minimizing its size, and always keeping it clean. Shielded holders will produce more noise than unshielded holders.


Simply adding the pipet to the holder (attached to the headstage input) slightly increases the capacitance at the amplifier input. After the pipet has been immersed into the bath and a GQ seal has been formed, the capacitance at the headstage input is further increased. As will be seen, the capacitance of the immersed portion of the pipet is a consideration in several sources of noise. Here, however, we begin by noting that all of this capacitance will at minimum produce noise because it is in series with the input voltage noise, en, of the headstage amplifier. The current noise produced has a power spectral density (PSD, Amp2/Hz) with rises asp at frequencies above roughly 1 kHz. Of course, this noise is correlated with noise arising from e,, in series with other capacitance (amplifier input capacitance, stray capacitance, capacitance of the electrode holder). The total amount of capacitance associated with an immersed pipet can vary from a fraction of a pF up to 5 pF or more. Low capacitance is associated with heavy elastomer coating and shallow depths of immersion. Obviously, the amount of noise arising from this mechanism increases as the capacitance associated with the pipet increases. However, regardless of the value of the pipet capacitance, the noise it contributes in conjunction with e,, will be small in comparison with other pipet noise sources described later. For low-noise patch-clamp measurements, it is imperative that the pipet capacitance be minimized. The reason for this will become more clear as other noise sources associated with this capacitance are described.

In addition to the mechanism just described, and to noise arising from the membrane to glass seal (which will be discussed separately), the pipet contributes noise by at least four mechanisms. Each mechanism will be described later, followed by a summary of pipet noise sources. Our emphasis is on the minimization of each noise, rather than simply its description.

5.2. Thin-Film Noise

Thin films of solution are capable of creeping up the outer surface of the pipet from the bath (Fig. 1A). The noise associated with such films has previously been shown to

Levis and Rae

Fig. 1. Simplified circuit representations of the major noise mechanisms of the patch pipet. (A) Thin-solution film on the exterior surface of an uncoated patch pipet; noise arises from the thermal voltage noise of the distributed resistance of this film in series with the capacitance of the pipet wall. In (B-D), the pipet is shown coated with a suitable elastomer. (B) Distributed RC noise arising from the thermal voltage noise of the distributed resistance of the pipet filling solution in series with the distributed capacitance of the immersed portion of the pipet wall and its elastomer coating. (C) Dielectric noise of the series combination of the pipet (r,, C,, where y, = WC,D,) and the elastomer coating (r,, C,, where ‘yz = wC,D,). In the region immersed in the bath, the glass wall of the pipet and its elastomer coating are represented by ideal lumped capacitances C, and C,, respectively in parallel with loss conductances x = Z#C,D, and y2= 27&D,. The thermal noise (dielectric noise) of the coated pipet is then 4kT multiplied by the real part of the admittance of the series combination of dielectrics. (D) Re-CP noise arising from the thermal voltage noise of the entire (lumped) resistance, Re of the patch pipet in series with the patch capacitance, C,,. See text for further details.

Patch-Clamp Efectrode Technology 21

be very significant (Hamill et al., 1981). Such a film will have a relatively high distributed resistance, and the thermal voltage noise of this resistance is in series with the distributed capacitance of the pipet wall. It is expected that the PSD of this noise will rise at low to moderate frequencies and then level out at frequencies in the range of several kHz to several tens of kHz. We have estimated with uncoated pipets made from several types of glass that the noise associated with such a film of solution is usually in the range of 100-300 pA rms in a bandwidth of 5 kHz. Evidence for such films has been found in pipets fabricated from all glasses we have tested when elastomer coating has been omitted. However, pipets pulled from GE quartz produce significantly less noise without elastomer coating than any other type of glass. Appar- ently the surface of this glass is less subject to the formation of such thin films.

Coating the pipet with Sylgard 184 or other suitable elastomers can essentially eliminate the formation of external films of solution and eliminate the otherwise large amounts of noise they produce. These elastomers have a hydrophobic surface that prevents the formation of such films. Sylgard 184 is so effective in this regard that we have been unable to detect any thin-film noise in properly coated pipets.

Thin films of solution may also be able to form on the interior surface of the pipet and inside the holder. To avoid the formation of such films, it is possible after filling the pipet with the desired amount of ionic solution to layer a few millimeters of paraffin oil or silicone fluid on top of the filling solution. However, we have found that this is usually unnecessary (and it can get messy) if excess solution is carefully suctioned from the back of the pipet as described earlier.

5.3. Distributed RC Noise

Noise will also arise from the thermal voltage noise of the resistance of the pipet filling solution in series with the capacitance of the immersed portion of the pipet (Fig. 1B).

22 Levis and Rae

Most of the resistance of the pipet resides at or near its tip. However, significant resistance is distributed along the shank distal to the tip. This resistance (and its thermal voltage noise) are in series with the capacitance of the pipet wall distributed along the portion that is immersed in the bath. We refer to noise that results as distributed RC noise. In the frequency range of greatest interest to patch-clamping (DC to 100 kHz or more), the PSD of this noise is expected to rise as 7. Our theoretical predictions of the noise arising from this mechanism (e.g., Levis and Rae, 1992) have relied on idealizations of the pipet geometry. More complicated real-world geometries and factors such as nonuniform thinning of the pipet wall that often occurs during pulling are expected to make such predictions rather imprecise. Because of this, we chose to study distributed RC noise directly. These experiments used quartz pipets pulled from OD/ID = 2.0 tubing that were coated with Sylgard 184 only to the point where the electrode entered the bath (i.e., most or all of the immersed portion of the pipet was uncoated); immersion depth was -1.8 mm, and the pipets were sealed to Sylgard (seal resistance ~200 Go). Our strategy was to vary the ionic strength of the internal filling solution. Changing the ionic strength of the filling solution will change the pipet resistance, but it will have no effect on the pipet capacitance. Because of this, it is expected that for pipets of equivalent geometry and with the same depth of immersion into the bath, the PSD of distributed RC noise will vary as l/M, where M is the ionic concentration of the filling solution. The rms noise in any particular bandwidth is expected to vary as l/M% In our study of this noise, we used NaCl solutions with concentrations from 1.5 mM to 1.5M to fill the pipet. As expected, the noise increased as the ionic strength of the filling solution decreased. When the noise component attributable to distributed RC noise was parsed from total noise (and it was the dominant noise source for ionic strength of 15 mM or less), the predicted behavior was reasonably well confirmed. Also, as expected, the PSD of this noise component increased approximately asfZ as frequency increases.


On the basis of these experiments, we concluded that for uncoated quartz pipets that were pulled from OD/ID = 2 tubing and immersed to a depth of about 1.8 mm and filled with 150 mM NaCl (i.e., the ionic strength typical of most experiments), the PSD of distributed RC noise was approximated by 2.5 x 1O-38p amp2/Hz. The rms noise contribution in a bandwidth B is then (8 x 1O-39 c,B~)~~ amps rms, where c, is a coefficient that depends on the type of filter used (c3 = 1.9 for an &pole Bessel filter). This equation predicts a noise component of -44 pA rms for a 5 kHz bandwidth (-3 dB, B-pole Bessel filter), or about 123 fA rms in a lo-kHz bandwidth. It must be remembered, however, that these results were for relatively thick-walled pipets fabricated from quartz, which has a low dielectric constant of 3.8. It must also be remembered that the pipets were not coated with Sylgard (or other suitable elastomer) in the region immersed in the bath. The capacitance of the wall of the pipet is expected to vary directly with the dielectric constant of the glass (for pipets of the same geometry) and vary inversely roughly in proportion to the log of the OD/ID ratio. The PSD of distributed RC noise should vary in proportion to the pipet capacitance (C,) squared; rms noise in a given bandwidth will therefore vary linearly with C,, Thus, for an uncoated pipet fabricated from OD/ID = 1.4 tubing from a glass with a dielectric constant of 7.6 (twice that of quartz), the numbers given above would be expected to increase by a factor of about 4. On the other hand, coating the immersed portion of a pipet with a suitable elastomer will thicken its walls and therefore reduce C,. Thus, very heavy coating of the pipet with an elastomer, such as Sylgard 184, can dramatically reduce distributed RC noise, and, with such coating, the amount of this noise will become almost independent of the type of glass used. In the experiments described earlier, we measured C, to be in the range of 1.4-1.8 pF. We have found that using the tip-dip elastomer coating method (Levis and Rae, 1993) to build up a heavy coat of Sylgard all the way to the tip of the pipet, we can obtain values of Ce as low as -0.35 pF for a comparable depth of immersion. This should reduce distributed RC

24 Levis and Rae

noise to ~10 fA rms in a 5-kHz bandwidth. Of course, shallow depths of immersion can also reduce distributed RC noise.

From the preceding discussion, it should be clear that the reduction of distributed RC noise is one of the major benefits of coating the immersed portion of the pipet with a low dielectric constant elastomer such as Sylgard 184. This noise component can also be minimized by using thick- walled tubing of glasses with low dielectric constants and by shallow depths of immersion of the pipet into the bath. Distributed RC noise is also expected to depend on pipet geometry, and should be minimized by shapes that reduce the distributed resistance distal to the pipet tip.

5.4. Dielectric Noise

Dielectric noise (Fig. 1C) will also arise from the capacitance of the pipet wall over the region that is immersed in the bathing solution. For pipets fabricated from glasses other than quartz, dielectric noise is likely to be the dominant source of noise arising from the pipet. For a single dielectric with a capacitance C, and a dissipation factor D, the PSD of dielectric noise is given by:

Sd2 = 4kTDC,(2nJ) Amp2/Hz (1)

The rms noise in a bandwidth B is given by:

I, = (4kTDCdc2nB2)vz Amp rms (2)

where k is Boltzman’s constant and T is absolute temperature (OK). c, Is a coefficient that depends on the type of filter used; for an &pole Bessel filter with B as the -3-dB bandwidth, c, = 1.3. It is important to note that the PSD of dielectric noise rises linearly with increasing frequency and that the rms value of this noise is proportional to filter bandwidth. This is quite unlike the other noise sources discussed, and is very useful in experimentally parsing dielectric noise from other types of noise generated by the pipet.


For an uncoated pipet, these equations can be applied simply by noting that Cd is the capacitance of the immersed portion of the pipet (denoted by C, above), and that D is the dissipation factor of the glass. It is instructive to consider two uncoated pipets with the same geometry both pulled from OD/ID = 1.4 tubing and both immersed to a depth of about 2 mm. One pipet is fabricated from quartz (D = 0.0001, dielectric constant = 3.8) and the other pipet is fabricated from a borosilicate with D = 0.005 and a dielectric constant of 5.0. The capacitance (C, or CJ of the quartz pipet should be about 1.5 pF, whereas that of the borosilicate pipet will be about 2 pF because of its higher dielectric constant. Using these numbers, it can be estimated that the uncoated quartz pipet will produce about 16 fA rms dielectric noise in a 5-kHz bandwidth (-3 dB, B-pole Bessel filter), whereas the borosilicate pipet would produce 128 fA rms dielectric noise in the same bandwidth. The superiority of quartz is clear in this case.

Of course, the importance of coating the pipet with a suitable elastomer has already been demonstrated, regardless of the type of glass used. Therefore, it is necessary to consider the dielectric noise in this more complicated situation We have presented a more detailed analysis of the dielectric noise in this case elsewhere (Levis and Rae, 1993). Here, we will summarize our most important conclusions. When the pipet is coated with an elastomer, it is necessary to derive equations that describe the dielectric noise of the series combination of two different dielectrics with capacitances C, and C, and dissipation factors 0, and D, (see Fig. 1C and its legend). For D,, D, << 1, the dielectric noise PSD of the elastomer coated pipet is well approximated by:

~mqj[(W,~ + D2c2c:)/(c1+ c2j21 Amp2m (3)

and the rms noise in a bandwidth B is approximated by:

(4kTc2np2[(QC$l$ +D$$C~)/(CI +C#l}Y2Amps rm (4)

In these equations C, and D, are the capacitance and dissipation factor of the glass wall of the pipet and C, and D,

Leuis and Rae

are the capacitance and dissipation factor of the elastomer coating. The capacitance C, depends on the depth of immersion, the thickness of the pipet wall, and the dielectric constant of the glass. For a 2-mm depth of immersion, C, can vary from as little as about 1 pF for very thick-walled quartz pipets, to more than 6 pF for thin-walled pipets made from glassses with high dielectric constants (e.g., soda lime and high-lead glasses). Of course, the capacitance C, of the elastomer coating also depends on the depth of immersion, the dielectric constant of the elastomer, and the thickness of the elastomer coating. Obviously, heavy elastomer coating will lead to the smallest values of C2. However, it is important to realize that the thickness of the elastomer coating will not be uniform. In particular, it is hard to achieve very thick elastomer coatings near the tip of the pipet. The dip method of elastomer coating (Levis and Rae, 1993) has proved to be useful in building up relatively heavy coats of elastomer all the way to the tip of the electrode, but even with this method the thickness of the coat is still not uniform. Because of this, it is difficult to predict the value of C,. However, we have measured the value of C, (see Levis and Rae, 1993) to be as little as 0.4-0.5 pF for a 2-mm immersion depth when heavy coatings of Sylgard were applied with the dip method. With lighter coating, the value of C, can easily be much higher (2 pF or more).

The dissipation factor D, of the glasses used to fabricate patch pipets have already been discussed; reported values range from as little as 10-5-10-4 for quartz to as much as 0.01 for soda lime glasses. The dissipation factor D, of the elastomer is also very important. Sylgard 184 has a dissipation factor of about 0.002, which is lower than that of most glasses, with the notable exception of quartz. Because of this, coating pipets fabricated from glasses other than quartz will significantly reduce their dielectric noise and the relative reduction will be greatest for the poorest (most lossy) glasses. However, the dissipation factor of Sylgard 184 is a factor of 20 or more higher than that of quartz, and predictions based on Eqs. (3) and (4) indicate that for all realistic values of C,


coating a quartz pipet with Sylgard will actually increase its dielectric noise relative to that which would have been produced by the pipet alone. This is true despite the reduction in overall capacitance produced by the Sylgard coating. Thus, for a quartz pipet with C, = 1.5 pF and D, = 0.0001 and a Sylgard coating with C2 = 0.5 pF and D, = 0.002, Eq. (4) predicts 30 fA rms of dielectric noise in a 5-kHz bandwidth (-3 dB, &pole Bessel filter), whereas as described earlier, the same pipet without the Sylgard coating would have produced only about 17 fA rms dielectric noise in this bandwidth. Estimates of the dielectric noise of quartz pipets coated with Sylgard and several similar elastomers and sealed to Sylgard have produced values that are in good agreement with the predictions of Eqs. (3) and (4).

It is apparent from the earlier discussion that coating a quartz patch Pipet with Sylgard 184 is not desirable in terms of dielectric noise. Nevertheless, coating with Sylgard or some other suitable elastomer is necessary to eliminate thin-film noise and to minimize distributed RC noise. If fact, very heavily Sylgard-coated quartz pipets display the least noise of all pipets, so the small increment in dielectric noise resulting from such coating is more than offset by the benefits in terms of reduction of other types of noise. It is also important to realize that even though Sylgard-coating a quartz pipet will increase its dielectric noise, the final dielectric noise of such a pipet still remains significantly below that of Sylgard- coated pipets fabricated from any other type of glass we have tested. If elastomers with dissipation factors significantly less than that of Sylgard 184 can be found that are otherwise suitable for coating pipets, they could be effective in lowering the dielectric noise of quartz pipets. It can be appreciated from examination of Eqs. (3) and (4) that the dissipation factor of such an elastomer need not be less than that of quartz to lower total dielectric noise of a heavily coated pipet. Such elastomers (if found) should also be very beneficial for other types of glass. Dow Corning R-6101 and Ql-4939 both are reported by the manufacturer to have dissipation factors of 0.00025. However, our preliminary measurements of pipets

28 Levis and Rae

coated with R-6101 have failed to demonstrate any significant advantage over pipets coated with similar thicknesses of Sylgard 184. Although we are unable to account for this finding, it certainly seems possible that the true dissipation factor of this elastomer exceeded the value in the manufacturer’s data sheet.

Because of the volume of material presented regarding dielectric noise, it is probably worthwhile to summarize our conclusions. For thick-walled quartz pipets with heavy Sylgard coating to the tip and seal to Sylgard at an immersion depth of -2 mm, our estimates of dielectric noise has generally been in the range of 20-35 fA rms in a 5-kHz bandwidth. On occasion, with actual excised patches and heavily Sylgard-coated patch pipets, our estimates of dielectric noise have been ~15 fA rms in a 5-kHz bandwidth when the electrode tip has been withdrawn close to the surface of the bath. For other types of glasses, dielectric noise is significantly higher. Our previous measurements of the noise arising from light to moderately Sylgard-coated pipets made from more than 20 different glasses (Rae and Levis, 1984,1992a), indicated that in a 5-kHz bandwidth and with a -2-mm depth of immersion dielectric noise varied from about 100-200 fA rms. The lowest noise was associated with glasses with the smallest loss factor (i.e., dissipation factor multiplied by dielectric constant), whereas the highest noise arose from the very lossy soda lime glasses. Recently, we have measured a few pipets made from Corning 7052 and 7760 (tubing OD/ID = 1.4) that were heavily coated with Sylgard 184 to the tip by the dip method described earlier. These measurements indicated that dielectric noise could be as low as -70 fA rms in a 5-kHz bandwidth for these glasses (with heavy Sylgard coating) at a 2-mm immersion depth. This is somewhat more noise than would be predicted from Eqs. (3) and (4), but less than we have estimated previously.

5.5. R,-Cp Noise

The last pipet noise mechanism that we will consider is the noise that is expected to arise from the thermal voltage


noise of the entire lumped pipet resistance, Re, in series with the capacitance of the patch membrane, CP; we refer to this noise source as Re-CP noise (see Fig. 1D). This noise is expected to have a PSD that increases asf2 up to frequencies of about 1 /2xRBCP (which is usually several hundred kHz); at frequencies below this the PSD is expected to be:

SeP2 = 47c2e:CP2y Amp2/Hz (5)

where ee2 = $kI’R%, i.e., the thermal voltage noise PSD for the pipet. The rms noise attributable to this mechanism in a bandwidth B is then given by:

IeP = (1 .33n2c,ee2Cp2B3)% Amps rms (6)

where c, is a coefficient that again depends on the type of filter used to establish the bandwidth; for an B-pole Bessel filter with a -3-dB bandwidth of B Hz, c, = 1.9.

For single-channel measurements, patch capacitance typically ranges from approx 0.01-0.25 pF for pipet resistances in the range l-10 MQ (Sakmann and Neher, 1983). As expected, higher values of patch capacitance are associated with lower resistance pipets. Because of the inverse relationship between Re and Cr, Eqs. (5) and (6) predict that the smallest amount of Re-Cp noise will arise from the smallest patches, even though such patches are obtained with higher resistance pipets. For example, with Re = 10 MQ and CP = 0.01 pF, Eq. (6) predicts a noise contribution of only about 6 fA rms in a 5-kHz bandwidth (-3 dB, B-pole Bessel filter). On the other hand, with Re = 2 MQ and C = 0.25 pF, the predicted noise is more than 60 fA rms in thePsame bandwidth. This latter amount of noise can exceed the total of all other pipet noise sources for quartz pipets, and remains significant even for pipets fabricated from other glasses. Obviously, however, this noise source itself does not depend on the type of glass, but rather on the geometry of the pipet (and, to some extent, on luck).

5.6. Seal Noise

The noise associated with the membrane-glass seal is less easily predicted. It is expected that the PSD of this noise for

30 Levis and Rae

zero applied voltage will be given by 4kT Re{YJ, where Re{YSh} is the real part of the seal admittance. The minimum value of Re{YSh} is l/R,, where R, is the DC seal resistance, and this leads to a minimum estimate of the seal noise in a bandwidth B of (4kTB/R,)% This is just the thermal current noise of the seal resistance and can be very small for high resistance seals. For example, for a 200 GQ seal and a bandwidth of 5 kHz, (4kTB/R,)% = 20 fA rms. Our measurements from several patches with seal resistances in the range 40- 100 GQ have shown that the noise attributable to the seal is often indistinguishable from the predicted thermal current noise of R, (Rae and Levis, 1992b). Nevertheless, it is certainly possible that seal noise may sometimes exceed this minimum prediction. As anyone who has spent much time trying to achieve low noise with the patch-clamp technique knows, there is a great deal of variability in the noise achieved even when all of the precautions we have described have been followed and when very high seal resistances have been obtained. It is certainly tempting to blame some of this variability on the noise associated with the seal.

5.7. Summary of Pipet Noise Sources

It is important to realize that the noise sources described above (with the exception of noise arising from various capacitances in series with the amplifier’s input voltage noise e,,) are all uncorrelated. Uncorrelated noise sources add in an rms fashion. For example, if four uncorrelated noise sources have rms values denoted by E,, E,, E,, and E,, then the total rms noise resulting from the summation of these sources is given by (E,2 + E,2 + E,2 + E42& Because of this, the largest individual source of noise will tend to dominate total noise.

Of the noise sources described above, only thin-film noise can be completely eliminated (or in any case reduced to negligible levels). Distributed RC noise, dielectric noise, and Re-CP noise can never be eliminated, but they can be minimized. In many cases, precautions taken to reduce one noise source will also be beneficial in reducing other sources of noise. Thus, with any type of glass, thick-walled pipets will,


all else being equal, have less capacitance, and therefore display less distributed RC noise and dielectric noise. Similarly, shallow depths of immersion will also reduce pipet capacitance and simultaneously reduce distributed RC and dielectric noise. Coating pipets with a heavy layer of a low-loss elastomer such as Sylgard 184 will also reduce the pipet’s capacitance and reduce distributed RC noise for all types of glass. For all types of glass other than quartz, a heavy coat of Sylgard 184 extending as close to the tip as possible will also significantly reduce dielectric noise. In the case of quartz pipets, elastomers with dissipation factors comparable to that of Sylgard 184 will actually somewhat increase dielectric noise. However, within the range of realistic thicknesses of the coating, even for quartz, heavy coatings will generally lead to the least dielectric noise; this is compatible with the requirements for minimizing distributed RC noise in quartz pipets. It is also important to recall that even though Sylgard coating somewhat increases the dielectric noise of quartz pipets, the final dielectric noise of a heavily Sylgard-coated quartz pipet is still much less than that of pipets made from any other type of glass. The major distinction in terms of noise between pipets fabricated from quartz and other types of glasses is, in fact, the much lower dielectric noise of quartz. R,-C, noise often has been ignored in the past, and in many situations it is sufficiently small to still be ignored. However, when all other sources of noise successfully have been reduced to the lowest limits presently achievable, it can become significant, and even dominant at very wide bandwidths (Levis and Rae, 1993). X,Cr noise is minimized by forming the smallest patch areas that are consistent with the goals of the experiment being undertaken. Although the data are widely scattered, patch area (or patch capacitance) decreases as pipet resistance increases. The net result is that it is predicted that higher resistance patch pipets with small tips will tend to produce the least amount of R,-Cp noise. The geometry of such electrodes is not necessarily the best selection for minimizing distributed RC noise, but this can be overcome by heavy elastomer coating. Although we have not

32 Levis and Rae

systematically studied the relationship between pipet resistance (tip diameter) and noise, it is our experience that the lowest noise patches are usually obtained from small-tipped high resistance pipets.

It is difficult to assign values to what can be expected as “typical” or “best-case” noise from pipets fabricated from different glasses in different situations. Nevertheless, some rough estimates can be provided. For low-loss borosilicate, aluminosilicate, or high-lead glasses with moderate Sylgard coating extending to within -100 pm of the tip, it is reasonable to expect that in a 5-kHz bandwidth total pipet noise (excluding seal noise) as low as loo-120 fA rms can be achieved with a -2-mm depth of immersion. With very heavy Sylgard coating all the way to the tip, this value should fall to somewhat less than 100 fA rms in this bandwidth. With quartz pipets that are heavily Sylgard-coated to the tip we have been routinely able to keep total pipet noise to -4O-fA rms in a 5-kHz bandwidth for a 2-mm immersion depth. With the quartz pipet raised to the surface of the bath with an excised membrane patch, occasionally we have achieved a total noise of the pipet plus seal as low as 30-35 fA rms in a 5-kHz bandwidth; subtracting the thermal current noise of the seal (as judged from its measured resistance) yields an estimate of lo-20 fA rms for total pipet noise in these cases.

5.8. IYoise Sources for Whole-Cell Voltage Clamping Whereas all of the pipet noise mechanisms described

earlier, with the exception of Re-Cp noise, are present in whole- cell voltage clamping, their relative importance is very much less than is the case for patch voltage clamp measurements. Of course, this is not because these pipet noise sources have become less in the whole-cell situation, but rather because other noise sources have become much higher. In the first place, most whole-cell voltage clamp measurements are made with a patch-clamp headstage amplifier configured with a 500-MQ feedback resistor. In a 5-kHz bandwidth, this resistor alone will produce 400-fA rms noise, which is more than


even soda lime pipets will produce provided they are reasonably Sylgard-coated. Under most situations, however, the dominant source of noise in a whole-cell voltage clamp will be the thermal current noise of the pipet resistance Re in series with the cell membrane capacitance Cm.

As just noted, the whole-cell voltage clamp lacks RB-Cr noise. The reason for this is simply that the patch membrane has been disrupted, or shorted out, as is the case for perforated patch measurements. However, in the whole-cell situation, the entire cell membrane is in series with the pipet resistance and with the thermal voltage noise of this resistance. The noise produced by this has precisely the same mechanism that underlies Re-Cp noise, but, since Cm >> Cp, it is of far greater magnitude. It might also be recalled that the time constant R,Cp will typically be 1 ps or less and so usually can be neglected. However, the time constant ReCm is much larger and its effects can not be ignored, either in terms of noise or dynamic performance.

Of course, the electrode resistance Re is the series resistance in the whole-cell variant of the patch voltage clamp, and many of its effects are well known and need no further comment here. But it seems that some of its effects can never be emphasized often enough. One of these is the filtering effect that uncompensated series resistance has on the measured current. In the absence of series resistance compensation, this filtering effect (equivalent to a simple RC low-pass filter) limits the actual bandwidth of current measurement to 1/2xR,Cm. For example, with Re = 10 MS2 and C, = 50 pF, this is -320 Hz, and it should be remembered that Re, after patch disruption or perforation, usually is higher than the pipet resistance that was measured in the bath. With series resistance compensation, this bandwidth limit is increased. We will define as the fraction of the series resistance compensated (0 < a < l), and p = 1 - a. With series resistance compensation, the uppermost usable bandwidth is extended to 1/2$A,C,. So in the previous example, 90% series resistance compensation (p = 0.1) will extend the actual bandwidth limit to about 3.2 kHz. It will also greatly increase the

34 Levis and Rae

noise at this bandwidth. The PSD, Sem2, of noise arising from the thermal voltage noise of Re in series with Cm is given by:

S,,2 = (4n2e~C,‘f)/(l + 4n2P2R:Cm2f) (7)

where e 2 - - 4k7’Re is the thermal voltage noise PSD of Re. Note that thi’s expression takes into account the effects of series resistance compensation. For 100% series resistance compensation (a = 1, B = 0), Eq. (7) reduces to 4n*e,‘C,*f”, which has exactly the same form as Eq. (5).

From Eq. (7) it can be seen that the PSD of the noise arising from Re and Cm rises with increasing frequency asp until it reachesf = 1/27@R,C,.

Thereafter, this noise plateaus to a value of 4kT/P*R,, which, of course, is many times larger than the thermal current noise of the feedback resistor. This plateau level of the PSD will be maintained until a frequency is reached where it is rolled off by an external filter (or the inherent bandwidth limit of the electronics). As an example of the magnitude of the noise introduced by this mechanism, consider a favor- able example for whole-cell voltage clamping with Re = 5 MSZ and Cm = 30 pF. Without series resistance compensation, the “corner frequency” at which the noise PSD plateaus (and the limit of actual bandwidth of current measurement) is about 1060 Hz. For a -3-dB bandwidth @-pole Bessel filter) of current measurement only 500 Hz, the noise arising from Re and Cm would already be nearly 0.5 pA rms, which is more than a very bad electrode would produce in a bandwidth of 5 kHz. By a bandwidth of 1 kHz, the noise would have increased to about 1.3 pA rms. Increasing the bandwidth of current measurement much beyond 1 kHz without series resistance compensation is not justified, since the measured current will still be effectively filtered at 1.06 kHz (-3-dB bandwidth of the l-pole low-pass filter arising from Re and Cm). This does not mean, however, that setting the external filter to a bandwidth higher than 1 kHz will not add more noise. Increasing the bandwidth of the external filter to 5 kHz will increase the noise to more than 3 pA rms, but it will provide very little signal information that was not contained when the data


was filtered at 1 kHz. Series (pipet) resistance compensation can extend the usable bandwidth, but, of course, it will significantly increase the noise at eternal filter bandwidths higher than 1/2xReCm. Thus, with 90% series resistance compensation, the maximum usable bandwidth of current measurement is extended to 10.6 kHz. In this case, with an external filter @-pole Bessel) with a -3-dB bandwidth of 5 kHz, the noise is increased to almost 15 pA rms. For a lo- kHz bandwidth the noise will increase to about 40 pA rms. In noises of this magnitude, the pipet noise mechanisms previously discussed become quite insignificant. It can therefore be concluded that many of the characteristics of the pipet that were important to patch-clamping are not important to a whole-cell voltage clamp situation.

The noise arising from Re and Cm in whole-cell voltage clamping can only be minimized by minimizing Re and/or C,. Of course, minimizing C, means selecting small cells and often this is not possible. In addition, it should also be noted that if you are studying a particular type of channel in a population of cells of various sizes but the channel density is the same in all cases, there is no clear advantage in terms of signal-to-noise ratio of selecting smaller cells. For a constant value of Re it is simple to show that at a given bandwidth (below l/ 27$3R,C,) the rms noise will decrease linearly as Cm decreases, but, since the number of channels is also proportional to Cm, the signal will also decrease linearly with decreasing C,: Signal-to-noise ratio will be constant. In this case, signal-to- noise ratio only depends on Re and it will improve as l/Revz. So the most practical way to minimize this source of noise is to use the lowest resistance pipets that are capable of sealing to your cells and make every effort to minimize the increase in access resistance that often occurs when the patch is disrupted.

Finally, it is worth emphasizing that another important way of minimizing this noise is to not make the mistake of using a bandwidth of the external filter that is not justified by the situation. Increasing the external bandwidth significantly beyond 1/2@ReC,, essentially adds no information about the signal, but it will add additional noise.

36 Levis and Rae

References

Hamill 0. P., Marty A., Neher E., Sakman B., and S&worth F. J. (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pfliigers Arch. 391,85-100.

Levis R. A. and Rae J. L. (1992) Constructmg a patch-clamp setup. Met/z. Enzymol. 207,18-66.

Levis R. A. and Rae J. L. (1993) The use of quartz patch pipettes for low noise single channel recording. Biophys. ].65,1666-1677.

Neher E. (1981) Unit conductance studies in biological membranes. Tech- niques in CeIIuZar Physiology (Baker, I?. F., ed.), Elsevier, North Hol- land, Amsterdam.

Neher E. and Sakman B. (1976) Single channel currents recorded from membrane of denervated frog muscle fibers. Nature (Land.) 260, 799-802.

Rae J. L. and Levis R. A. (1984) Patch voltage clamp of lens epithelial cells: theory and practice. Mol. Physiol. 6,115-162.

Rae J. L. and Levis R. A. (1992a) Glass technology for patch electrodes. Meth. Enzymol. 207,66-92.

Rae J. L. and Levis R. A. (1992b) A method for exceptionally low noise single channel recordings. Pflrigers Arch. 42,618-620.

Sakmann B. and Neher E. (1983) Geometric parameters of pipettes and membrane patches, in Single-Channel Recording (Sakmann B. and Neher E., eds.), Plenum, New York, pp. 37-51.

Whole-Cell Patch-Clamp Recordings

Harald Sontheimer

1. Introduction

The patch-clamp recording technique, which measures ionic currents under voltage clamp, was designed to study small patches of membranes in which near perfect control of the transmembrane voltage can be achieved readily. The recent application of patch-clamp methodology to the analysis of whole-cell current actually defies many of the original design requirements. Nevertheless, whole-cell recordings are used routinely in electrophysiology laboratories to study electrical currents carried by ions through ion channels, neurotransmitter receptors, and electrogenic transporters in cell types of virtually any origin. Since the introduction of the patch-clamp technique in 1981 (Hamill et al., 1981) and the subsequent rapid development of commercial amplifiers, this method of intracellular recording has nearly replaced sharp electrode recordings, particularly in the study of cultured cells.

This procedure and its applications have been the topics of numerous excellent reviews and book chapters (see Rec- ommended Readings). In addition to summarizing some basic concepts, this chapter specifically emphasizes hands- on procedures and protocols. It is hoped that the chapter will effectively complement previous accounts of the patch-clamp technique.

From Neuromethods, Vol 26 Patch-Clamp Applications and Protocols Eds A. Boulton, G. Baker, and W. Walz 0 1995 Humana Press Inc

37

38 Son theimer

2. Principles (Why Voltage Clamp?)

Electrophysiologists are especially interested in the activity of membrane proteins that provide conductive pathways through biological membranes: ion channels, transmitter receptors, or electrogenic ion carriers. Channel activity, whether through voltage-dependent or ligand-gated ion channels, results in changes of membrane conductance that can be most conveniently evaluated by recording membrane currents at a constant membrane voltage. Under such “voltage-clamped” conditions, current is directly proportional to the conductance of interest.

A two electrode voltage-clamp design was first introduced in the seminal studies of Hodgkin et al. (1952) for the study of ionic conductances of the squid giant axon. In this application, one of the electrodes serves as voltage sensor, whereas the second functions as a current source, both interconnected through a feedback amplifier. Any change in voltage detected at the voltage electrode results in current injection of the proper polarity and magnitude to maintain the voltage signal at a constant level. The resulting current flow through the current electrode is assumed to flow exclusively across the cell membrane and, as such, is proportional to the membrane conductance. The major disadvantage of this technique, however, is its requirement for double impalement of the cell, which restricts its application to large cells (~20 pm) and prevents study of cells embedded in tissue.

Single-electrode switching amplifiers were developed that allowed the use of one electrode to serve double duty as voltage and current electrode in an attempt to solve this problem. For short periods of time, the amplifier connects its voltage-sensing input to the electrode, takes a reading, and subsequently connects the current source output to the same electrode to deliver current to the cell. This approach, however, is limited in its time-resolution by the switching frequency between the two modes, which must be set based on the cell’s RC time constant. Both single electrode switch clamp and double electrode voltage clamp allow direct mea-


surement of the cell’s voltage and avoid the introduction of unknown or unstable voltage drops across the series resistance of the current passing electrode.

The whole-cell patch-clamp technique similarly uses only one electrode. However, by contrast to above techniques, it uses the electrode continuously for voltage recording and passage of current. Consequently, the recording arrangement contains an unknown and potentially varying series resistance in the form of the electrode and its access to the cell. For the technique to deliver satisfactory results, it is essential that this series resistance be small relative to the resistance of the cell. Numerous measures are taken to satisfy these requirements (see Section 5.) including the use of blunt, low- resistance electrodes, small cells with high impedance, and electronic compensation for the series resistance error.

When effectively utilized, the whole-cell technique can yield current recordings of equal or superior quality to those obtained with double- or single-electrode voltage clamp recordings.

3. Procedure and Techniques

3.1. Pipets

In contrast to sharp electrode recordings that utilize pipets with resistances of >50 Ma, comparatively blunt low resistance (l-5 Ma) recording pipets are used for whole-cell recordings. This is done for at least two reasons: Series resistance should ideally be two orders of magnitude below the cell’s resistance, and blunt electrodes (l-2 pm) are required to achieve and maintain mechanically stable electrode-membrane seals.

As described in Chapter 1, electrodes can be manufac- tured from a variety of glass types. Although it has been reported frequently that glass selection has a significant influence on the quality of seal or the frequency at which good seals are obtained, little scientific evidence supports this notion. In this laboratory, it was found to be more critical that glass pipets be thoroughly cleaned by soaking in

40 Sontheimer

acetone for l/2 h followed by drying at 180°C in a lab oven. To achieve the shape ideal for sealing membrane patches, electrodes are pulled from capillary glass pipets in a two- or multistage process using commercially available pullers, such as those of Narashige, Brown-Flaming, and others. The additional step of firepolishing can significantly improve the likelihood of seal formation. Cell surfaces also can be enzymatically cleaned during isolation and culturing procedures, or just prior to each experiment. Once an appropriate set of variables, i.e., cell preparation, glass type, electrode resistance, and shape, is identified, the success rate for stable electrode-membrane seals should be between 50 and 90%.

3.2. Electronic Components of a Setup

The electronic components of a patch-clamp setup are comparatively few: A patch-clamp amplifier, oscilloscope, stimulator, computer, optional external signal filter, and VCR recorder. High-quality patch-clamp amplifiers are available from a number of manufacturers. At present, most laboratories utilize either a List EPC 7 or an Axon Axopatch Series 1 amplifier; both manufacturers have recently introduced new models (EPC 9 and Axopatch 200). These amplifiers are overall similar in design and are both equipped with multistage Bessel filters, variable gain settings, and at least two feedback resistor settings for whole-cell and single-channel recordings.

The single most important electronic component of a patch-clamp amplifier is the current-to-voltage converter, which is contained in the headstage. Its characteristics are described in detail elsewhere (Sigworth, 1983; Fig. 1). Current flow through the electrode (I$ across a resistor of high impedance (R) causes a voltage drop that is proportional to the measured pipet current (IJ. An operational amplifier (OpAmp) is used to automatically adjust the voltage source (Vs) to maintain a constant pipet potential (VP) at the desired reference potential (Vr,J. Because the response of the OpAmp is fast, it can be assumed that for all practical purposes VP = Vrer.

Whole-Cell Patch-Clamp Recordings 41

-L J Vref

Fig. 1. Scheme of current-to-voltage converter (for details, see Sigworth, 1983). Abbreviations used: pipet current, Ip; feedback resistance, R; operational amplifier, OpAmp; reference potential, V,,; voltage source, Vs.

In their whole-cell mode, patch-clamp amplifiers use a 500 Ma feedback resistor allowing measurement of currents of up to 20 nA. For the Axopatch 1D a low gain 50 Ma headstage is available that allows currents of up to 200 nA to pass; however, its use sacrifices the use of capacitance and series resistance compensation.

Although some amplifiers, such as the Axopatch lD, have built-in stimulators, most electrophysiologists prefer the use of an external stimulator that offers greater versatility. Low cost microcomputers can serve as both digital stimulator and on-line recorder. Although the use of a microcomputer is not essential, the growing number of affordable high-quality hard- and software products have made computers and D/A (digital-analog)-A/D converters standard laboratory equipment. Data can be collected and digitized on-line at up to 330 kHz and can be stored to the hard disk of a computer. All necessary components can be purchased at a price well below that of an external stimulator alone.

Most patch-clamp amplifiers are equipped with 4-pole Bessel filters that are of sufficient quality to filter data. However, the use of an external 8-pole Bessel filter (such as a Frequency Devices, Series 920) allows the selection of a wider range of cut-

42 Son theimer

off frequencies and has better frequency responses, specifically a sharper roll-off (see Section 4.1.1.).

Additional equipment is recommended for specific data collection needs. If data are to be collected for extended periods of time, and if the sampling rates are roughly lo-40 kHz, a VCR tape recorder is a useful interim storage medium. Data analysis subsequently can be done by playing data back off- line to a computer equipped with an A/D converter. Even in the presence of a computer-based data acquisition system, oscilloscopes are a convenient means for monitoring data collected by either microcomputer or VCR, and are essential for the “debugging” of environmental electrical noise from a recording setup.

3.3. Recording Configuration

The whole-cell patch-clamp recording setup closely resembles that used for sharp electrode intracellular recordings (Fig. 2A). An electrically grounded microscope on an isolation table serves as the foundation of the recording setup. A recording chamber is mounted to the stage of the microscope (Fig. 2B). Alternatively, if constant perfusion is not desired, a 35-mm Petri dish can be used as a recording chamber, in which case cells may be grown directly on the 35-mm dish. To use Normarski optics, we prefer the use of a Plexiglas recording chamber that has a glass coverslip as the base (Fig. 2B). Various types of recording chambers are commercially available, all of which serve well for most purposes. Elec- trodes typically are placed under visual control (400x) onto a cell by use of a high-quality, low-drift micromanipulator. Numerous hydraulic, piezoelectric, and mechanical designs are commercially available, each offering unique benefits. Hydraulic manipulators, such as the Narishige model MO- 203, combine precise movement with large travel, are very versatile, and are easy to use. Piezoelectric manipulators, such as the Burleigh PCS 250, only allow 70-200 Frn of travel, and as such are only useful when used in combination with a coarse positioning manipulator. Piezoelectric systems provide excellent stability and are the instruments of choice for


Fig. 2. Whole-cell patch-clamp setup (A) and recording chamber (B). Photomicrographs of a typical recording setup based on a Nikon Diaphot microscope. Patch-clamp headstage with electrode holder mounted on a swivel PVC clamp and attached to three axis manipulator constructed from three series 420 microtranslation stages (Newport Inst.). (B) Closeup view of flow-through recording chamber. Chamber was machined from Plexiglas and uses 24 x 50 mm glass coverslip as bottom.

excising patches for single-cell recordings. Stable, low-cost mechanical manipulators (such as Newport instruments series 421) can be assembled from single axis translation stages.

44 Son theimer

Bath B On-cell (no camp.)

c On-cell (cp camp.) D Whole-cell

2 ms

Fig. 3.Oscllloscope traces before and during establishment of whole- cell recording. (A) Electrode in bath (V = 0 mV). (B) On cell after formation of giga-seal (V = 0 mV). (C) As in (B) after Cp compensation. (D) After rupturing patch, whole-cell configuration but prior to cell capacl- tance and series resistance compensation (V = -80 mV).

Their modular design, combined with micrometer screws and DC motors (e.g., model 860A) make them extremely versatile. The arrangement shown in Fig. 2A includes three 421 stages of which the X and Y axes are controlled manually by micrometer screws, whereas the Z axis for electrode placement uses an 860A DC motor controlled by a hand-held battery-operated manipulator (model 861). This arrangement has proven to be stable and relatively inexpensive.

3.4. Experimental Procedure

During electrode placement, electrode resistance is monitored continuously by applying a small voltage pulse (l-5 mV, 2-10 ms) to the electrode (Fig. 3A). Once contact is made with the cell, electrode resistance spontaneously increases

Whole-Ceil Patch-Clamp Recordings 45

by lo-SO%. Application of gentle suction to the electrode by mouth or a small syringe quickly results in the formation of a gigaseal (Fig. 3B). At this point, seal quality can be improved by applying a negative holding potential to the pipet. In this cell-attached configuration, pipet capacitance transients (Cp) are reduced using the fast compensation adjustment at the amplifier (Fig. 3C). This compensation of pipet capacitance is essential for proper series-resistance compensation. Should compensation be incomplete, coating of future electrodes with Sylgard (Dow Corning) or lowering the bath perfusion level is recommended to reduce the residual transients and improve Cp compensation. Following pipet capacitance cancellation, a brief pulse of suction will rupture the membrane patch under the electrode, providing low resistance access to the cell. This also results in large capacity transient arising from the added membrane capacitance (Fig. 3D). Immediately after rupturing the membrane, a reading of the cell’s potential should be obtained (at I = 0), since this access potential is as close to the actual resting potential reading as can be obtained. Within minutes of establishing a whole-cell configuration, the pipet contents will equilibrate with the cell’s cytoplasm and will impose an artificial ionic potential across the membrane. Next, by adjusting the capacitance and series resistance (Rs) compensation and gradually increasing the percent of compensation, effective Rs compensation should be possible under most circumstances. Ideally, access resistance should be 40 MSZ prior to activating Rs compensation. Under these conditions, 80% compensation results in a <2 MQ residual uncompensated series resistance. Series resistance and capacitance compensation result in a change of the step waveform applied without actually changing access resistance or the cell capacitance per se (see later). The procedure for establishing whole-cell configuration and all necessary compensations are nicely illustrated in the manuals of both the List EPC 7 and Axon Axopatch 1D amplifiers.

Because of the intracellular perfusion of cytoplasm with pipet solution, it is advisable to wait several minutes prior to

46 Sontheimer

obtaining recordings to assure that this dialysis has reached a steady-state. Diffusion rates depend on molecular weight and charge of the diffusing particle, such that longer diffusion times are expected for relatively large, uncharged molecules as compared to small ions. For substances of mol wt 23-156,000, diffusion rates have been determined in adrenal chromaffin cells, and these suggest that complete dialysis of these small cells occurs on the order of tens to hundreds of seconds (Pusch and Neher, 1988). If dialysis of the cell is incompatible with the experimental design, e.g., when ionic currents are studied that are under control of second messengers, the perforated-patch method should be used instead (see Chapter 7).

4. Data Evaluation and Analysis

4.1, Data Filtering/Conditioning, Acquisition, and Storage

4. I. I. Filtering Data are hardly ever acquired and stored without fur-

ther modifications. The analog output of the amplifier is typically amplified to make effective use of the dynamic range of the acquisition device. In the case of an A/D converter, this typically translates to an amplification range of -10 to 10 V. At the same time, signals are filtered, often using the built- in signal filter.

Filtering of data is both essential and inevitable. Because of the RC components of the cell membrane-series resistance combination, the cell and electrode are essentially a single- stage RC filter. This is important to bear in mind because it significantly affects the true time-resolution of a recording. Assuming that Rs << membrane resistance (Rm), uncompensated Rs will filter any current flow recorded with a -3 dB cutoff frequency described by F = 1/(2~~ . Rs + Cm). Assum- ing, for example, a cell capacitance of 20 pF and a series resistance Rs of 10 Ma, values typical of small cell recordings, currents across the membrane will be filtered with an effective F of -800 Hz.


Irrespective of the intrinsic filter properties of the analyzed cell, data filtering is essential to reduce signal components that are outside the bandwidth of interest. Filtering “condenses” the time domain of the signal to the domain of interest. The fastest signals recorded under whole-cell conditions are on the order of 200-500 ps (2-5 kHz). Note that this is of the same order of magnitude as the membrane-as filter time-constant above. To eliminate high frequency noise, a low-pass filter is used. An ideal filter has a steep roll-off, and does not greatly distort signals. Bessel filters (4- or B- pole) have excellent characteristics for filtering whole-cell currents. Comparison of the filter characteristics of a 4- and &pole Bessel filter are demonstrated by comparing the onset response of a square pulse before and after filtering at various cutoff frequencies (Fig. 4). In these examples, an B-pole Bessel filter clearly provides excellent signal filtering with least distortions (Fig. 4). However, at cutoff frequencies above 2 kHz, 4-, and B-pole filters do not differ significantly in the onset or settling time.

4.1.2. Sampling Rate and Dynamic Range of Signal When using an A/D converter to digitize signals, it is

important to select the appropriate filter and sampling rates to represent accurately the analog signal of interest. It is inevitable that A/D conversion will reduce the “infinite” dynamic range of the analog signal to a well-defined step- like range of the digital signal. Commonly used 12-bit converters (for example, the Axon models TL-40,125,330) divide the amplitude range into 4096 discrete steps, which at a -10 to 10 V signal range will yield steps of 4.88 mV. In theory, signals can be sampled at the highest possible rate supported by the A/D converter. However, practical limitations exist. Most affordable A/D boards sample at loo-330 kHz on a single channel, thus allowing sampling in 3-10 ps intervals. Depending on the duration of the signal, sustained sampling at 10 ys will generate very large amounts of data, requiring significant disk space. A vast majority of this data will not contain necessary information. The Nynquist Sampling Theo- rem states that the minimum sampling rate (Nynquist fre-

48 Sontheimer

g 60-

; 40-

3 20-

b-4

0

00 02 04 06 08

, o. 2 kHz (-3dB)

50

z 60

3 40

$ 20

0:o 0'2 0.4 0.6 0.8

00 02 04 06 08

Fig. 4. Frequency response of 4- and &pole Bessel filter to square pulse. A 2-ms square pulse was apphed to a Frequency Device model 902 &pole Bessel filter as compared to the built-in Cpole Bessel filter of the Axopatch 1D amplifier. The graphs illustrate differences in response characteristics at three commonly used 3 dB cutoff frequencies (1,2, and 5 kHz).

quency) required to represent accurately an analog waveform is twice the signal bandwidth. As a consequence, if the analog filter is set at a -3 dB cutoff frequency of 3 kHz, a minimum sampling frequency of 6 kHz or 167 ps intervals is required. Although these minimum requirements will allow the reconstruction of data with little error under most circumstances, a sampling frequency five times the -3 dB frequency commonly is recommended for actual recordings.

4.1.3. Aliasing

The Nynquist Sampling Theorem only applies when sampling data digitally at frequencies between 0 and the Nynquist frequency. If frequencies higher than the Nynquist frequency are sampled, they are “folded back” into the low

Whole-Ceil Patch-Clamp Recordings

frequency domain, a process called “alias@.” Essentially, these high frequency signals will affect and distort those signals within the appropriate frequency domain. Aliasing can be prevented if signals above the Nynquist frequency are cut off by a low-pass filter. A proper matching of filter frequency and sampling rate is thus important to accurately reproduce analog waveforms. In practical terms this requires that the cutoff frequency of a low-pass filter be set to no higher than half the sampling frequency. In the previous example, the low-pass filter was set at 3 kHz.

4.1.4. Data Storage On-line digitization has the advantage that data can be

stored directly in computer memory or hard disk. This mode of data storage is preferred since it gives convenient and fast access to the data for future evaluation. Using a 12 bit A/D converter, each data sample uses two bytes of information. Thus a 2048 sample trace requires -4 kbytes of memory or disk space. A continuous sampling of neuronal discharge at a frequency of 50 kHz generates 100 kbyte of data every second. A 5-min recording would thus require 300 x 100 kbyte or 30 Mbyte of disk space. It thus becomes apparent that practical limitations exist as to the on-line digitization of data. For prolonged recordings at high frequencies (~40 kHz), a VCR recorder (such as the Neurocorder) may be used as an interim storage for data. Segments subsequently can be played back to the A/D converter for data analysis. For brief signals, on-line storage is not an issue, because hard disk space has become affordable (<$l/Mbyte). Rewritable optical computer disks with removable 1 Gbyte cartridges represent a practical way to archive data.

4.2. Leak Subtraction Currents across a cell membrane consist of two compo-

nents: ionic current flowing through ion channels of interest and capacitive current that charges the membrane. Capaci- tive current contains useful information pertaining to the cell size, since an approximation of cell size (or more precisely, membrane area of the recorded cell) can be derived from the

50 Sontheimer

capacitive current. However, in the study of ionic currents, capacitive currents are of relatively little interest. Since, ideally, capacitive currents are linear and not voltage-dependent, they can be subtracted from the signal of interest, through a process called “leak subtraction.” This subtraction can be done either on-line or off-line. Two protocols for leak subtraction typically are used:

1.

2.

If the nature of the experiments permit, currents are recorded sequentially in the absence and presence of specific ion-channel blockers (e.g., TEA, TTX, 4-AI?). Because specific blockers will eliminate ionic current but should not alter the capacitive or leakage current, subtraction of the two traces/or set of traces should result in the removal of capacitive and leakage currents. If this approach is not possible, P/N leak subtraction as first proposed by Bezanilla and Armstrong (1977) can be obtained.

In this subtraction scheme, each “test” voltage step is preceded by a series of N (typically 4) “leak” voltage steps of l/N (l/4 or -l/4, depending on polarity) amplitude of the test pulse activated from a potential at which no voltage- activated currents are activated. In a P/4 protocol, these 4 + l/4 amplitude traces are summed together (Fig. 5B) and are subtracted from the actual current trace (Fig. 5A) and will isolate the ionic current of interest (Fig. 5C). The example demonstrated in Fig. 5 actually used a P/-4 protocol, in which four hyperpolarizing pulses of -l/4 amplitude were summed and added to the current trace of interest. It is important to obtain the “leakage” current at potentials at which no voltage-activated currents occur. Most often this can be achieved by stepping to potentials negative of the resting potential (as illustrated). However, some cells express inwardly rectifying (or anomalous rectifying) currents that are active at the resting potential and negative thereof. Under these circumstances leak currents must be recorded at potentials at which the I-V curve is linear and no voltage-dependent currents are activated. Note that subtraction of capacitive and leak-


C 2ms

-145 mv

Fig. 5. Capacitive and leakage current subtraction using P/-4 method. (A) Whole-cell current recorded in response to voltage step from -120 to -20 mV. (B) Summed response of 4 . -l/4 amplitude voltage steps as indicated at bottom of (C). (C) Added response of (A) and (B) eliminating capacitive current.

52 Son theimer

age currents is purely cosmetic. It does not actually improve the signal recorded. However, it may allow revelation of small current components that would otherwise be hard to identify. If capacitive currents are not of interest, it is recommended that P/4 leak subtraction be performed on-line by the data acquisition program (such as pClamp, Axon Instr.) since it significantly reduces the amount of data stored.

4.3. Determination of Cell Capacitance

Biological membranes are lipid bilayers in which membrane proteins (e.g., ion channels and transporters) are contained. The specific capacitance of biological membranes seems to be fairly constant. It is relatively independent of cell type and a value of 1 pF/cm* is typical. The capacitance cancellation circuits of patch-clamp amplifiers are normally cali- brated in pF, and allow direct determination of the cells capacitance by adjusting and minimizing the capacity transients in response to a voltage step, as discussed previously. The dial reading provides at least a rough determination of membrane capacitance on which base estimates can be made as to the membrane area, since 1 pF capacitance represents 100 pm* of membrane.

Capacitance can also be derived from the capacity transient at any time during the recording. In response to a voltage step, capacitance is proportional to the integral of the charging transient, thus it can be derived by determining the area under the transient of a current’s trace. Neher and Marty (1982) have developed a very sensitive approach of measuring changes in membrane capacitance using a phase lock amplifier, which measures currents in and out of phase with a sinusoidal voltage change. This approach can resolve capacitance changes of 10 fS and has been used to resolve the fusion of synaptic vesicles by a step increase in capacitance.

4.4. Dissecting Current Components

Whole-cell recordings integrate the response of a large number of potentially heterogeneous ion channels. Separa- tion of these ionic current components is a critical step dur-

Whole-Cell Pa tch-Ciamp Recordings 53

ing or following current recordings. Four methods of current isolation commonly used to isolate voltage-activated ionic currents are outlined below.

4.4.1. Kinetically Certain types of ion channels activate and inactivate

much faster than others. For example, Na+ currents typically activate within 200-300 ps and inactivate completely within 2-5 ms. In contrast, K+ currents may take several ms to activate, and often inactivate slowly, if at all. Simple current isolation can thus be accomplished by studying whole-cell currents at different time points following stimulation, e.g., determining Na+ current amplitudes at 300-500 ps, and determining K+ current amplitudes after tens or hundreds of ms.

4.4.2. Current Subtraction via Stimulus Protocols Voltage-dependence of the steady-state activation and

inactivation of currents often allows selective activation of subpopulations of ion channels. For example, low or high threshold-Ca2+ currents can be activated separately by voltage steps originating from different holding potentials. Simi- larly, depolarizing voltage steps applied from very negative holding potentials (e.g., -110 mV) can activate both transient “A” type (K,) and delayed-rectifying (K,) K+ currents (Fig. 6A). Voltage steps applied from a more positive holding potential (e.g., -50 mV) will completely inactivate all K, channels while not affecting K, activity (Fig. 6B), such that subtraction of currents recorded with these two protocols effectively isolates K, currents (Fig. 6C).

4.4.3. Through “Isolation Solutions” (Ion Dependence) It is common practice to specifically design the compo-

sition of ionic solutions to favor movements of desired ions. As mentioned previously, dialysis of cytoplasm with patch pipet contents occurs rapidly after whole-cell configuration is achieved, thus allowing for manipulation of internal ionic concentrations. This access can be used to block most K+ channel activity by replacing pipet KC1 with impermeant Cs+ or N-methyl-o-glucoronate (NmDG), thus allowing isolation of

54 Son theimer

I m 70 mV

-50 mV u

-110 mV Kd only

A 70 mV

-50 mV -50 mV

A-B

4 ms

Fig. 6. Isolation of Ka current by subtraction. Current recordings from a spmal cord astrocyte expressing both transient (Ka) and delayed (Kd)- like rectifier K+ currents. (A) Currents activated from holding potential of -110 mV (step protocol, see inset). (B) Same cell and same voltage step protocol but steps originated from holding potential of -50 mV. Subtraction of (A) - (B) yielded Ka currents in isolation.


Table 1 Commonly Used Ion-Channel Blockers

Channel type Compound/reagent

K+ channels Delayed rectifier (Kd) TEA, Ba*+, capsaicin, 4-AP Inward rectifier (Kir) TEA, Cs+, Rb’, Na+, Ba*+ A TEA, AI’, dendrotoxin Wd

Big K Small K

K(ATP) Na+ channels CaZ+ channels

L-type T-type N-type P-type

Cl-/anion channels

Charybdotoxin Apamin TEA, Cs+, Ba*+ TTX, STX, CNQX, agatoxin, scorpion toxin

Nifedlpine, verapamil, Bay K8644, Cd*+, La3+ NP La3+ w-cdnotoxin, La3+ FTX funnel spider toxm Chlorotoxm, avermectin B

Na+ currents. In a similar fashion, acetate, glucoronate, or isothionate can each be substituted for Cl- ions, and tetramethylammonium chloride (TMA-Cl) can be substituted for NaCl. It has even been reported that the contribution of one ion channel population can be determined by replacement of all but the desired ion with glucose or sucrose.

4.4.4. Current Isolation uia Pharmacology Numerous natural toxins and synthetic pharmacologi-

cal agents exist that can be used to reduce or eliminate specific voltage-activated ion channel activity (for an excellent review, see TINS Suppl. Volume, 1994). A list of some of the more commonly used agents is shown in Table 1. Many of these agents are effective against one particular type of ion channel, such as tetrodotoxin and dendrotoxin, which target Na+ and K+ channels, respectively. The use of these compounds, either alone or in combination, allows the isolation of specific current(s).

Experimentally, currents are best identified pharmacologically by recording a family of current traces in both the

56 Son theimer

G2-X control

B -110 mv 4-AP

-O-control -O-control

-60 -60 -40 -40 -20 -20 0 0 20 20 40 40 60 60 60 60

Vm [mV]

Fig. 7. Pharmacologrcal isolation of 4-AI?-sensitive K’ current. (A) Family of current traces recorded from a spinal cord astrocyte using step protocol indrcated in inset. 03) Recording in the same cell using the same stimulus protocol 2 min after application of 2 mM 4-Al’. (C) 4- AP-sensitive current isolated by subtraction of (A) - (B) * (D) Current amplitudes determined 8 ms after onset of voltage steps plotted as a function of applied potential for current traces in (A-C).

absence and presence of drugs. This is illustrated in Fig. 7, in which the block of spinal cord astrocyte K+ currents by 4- amino pyridine (4-Al?) is shown. In the control and treated current traces (Fig. 7A,B), the inward sodium current is unaltered. By subtracting the 4-A&treated current traces from those of control, one can isolate the current component that is 4-A&sensitive. As discussed previously, a side benefit to such current subtraction is the elimination of capacitive and leakage currents.

Neurotransmitter-activated currents are perhaps easier to identify and isolate than their voltage-dependent coun- terparts, because these currents are induced in a time-depen-


dent manner based on the application of exogenous ligands. If need be, ligand-gated current can be isolated from background noise by subtracting recorded currents in the absence and presence of a given ligand.

4.5. 1-V Curves

Current-voltage (I-V) relationships are perhaps the most effective way to summarize the behavior of voltage- and ligand-activated ion channels. A number of important and useful parameters that cannot be accessed easily from the raw data can be readily derived from these plots, including: reversal potential, ionic dependence/selectivity, voltage- dependence (rectification), activation threshold, slope and cord conductance, as well as overall quality of voltage-clamp. I-V curves can be determined in various ways, examples of which are discussed later.

The factors that determine current flow through an open channel are conductance and driving force. Whereas conductance is proportional to the number of open channels, driving force is defined as the difference between actual voltage and the equilibrium potential for the ion(s) permeating the channel, also known as reversal potential (V,,). Thus current can be described as

I = G (V,,, - V,,,>

A plot of I vs Vm is commonly used to derive G (slope) or Vrey (X-intercept). The current evoked at a given Vm can be measured using a variety of protocols, including those listed below.

4.5.1. Peak and Steady-State I-V Curves

Peak currents are measured as the largest current activated by the applied voltage (Fig. 7C). Thus, if a current has a transient peak current amplitude, such as the K, current in Fig. 7A, analysis software can easily determine maximal current and plot these values against voltage (Fig. 7C). If the current to be studied does not have this defined peak, steady-

58 Sontheimer

state values may be used, typically as recorded at the end of a voltage step. This type of analysis was applied to the 4-A& treated current traces of Fig. 7B, with the resulting amplitudes similarly plotted in Fig. 7C.

4.5.2. Continuous “Quasi-Steady-State” I-V Curves A convenient way to establish I-V curves is by alteration

of the membrane potential in a continuous way through a voltage ramp. Because this ramp can be applied very slowly, it allows acquisition of a “quasi-steady-state” I-V relationship. This procedure has proven very useful in determining I-V relationships for transmitter responses. Note, however, that this approach assumes that currents do not inactivate during the length of the voltage ramp. An example of a voltage ramp used to determine the reversal potential of GABA- induced currents is illustrated in Fig. 8. A 200 mV, 400 ms voltage ramp as indicated in the inset to Fig. 8A was applied twice, once prior to application of GABA (control) and once during transmitter application (GABA). By subtracting the two responses one obtains the transmitter induced current in isolation. Since the time axis represents a constant change in the voltage applied, it can be instantaneously replotted as the I-V relationship of the transmitter-induced current (Fig. 8B). Using this approach, I-V curves of transmitter responses can be easily created throughout an experiment. As mentioned earlier, alteration of ionic composition allows shifting of the reversal potential and thus allows determination of ion specificity and relative permeabilities.

For those situations where currents inactivate rapidly, some indication as to the reversal potential can be obtained using a single-step protocol from which the reversal potential can be extrapolated. An example of this approach is illustrated in Fig. 8C. Here, a larger concentration of GABA was bath-applied, which resulted in characteristic receptor desensitization. The cell was maintained at -80 mV and a single 80 mV step (50 ms) was applied prior to and at the peak of the transmitter response. Current levels at the two potentials, -80 mV and 0 mV, were then subtracted and plotted


0 50 100 150 200 Imel -300 - D

1 InAl

0

-80 -60 -40 -20 0 20 ,’ Vm [mV]

-1

I---- -2 I

Fig. 8. Stimulus protocols to define reversal potential of GABA- induced currents. (A) A voltage ramp (inset) was applied prior to (Con- trol) and at the peak of (GABA) GABA-induced currents. The difference of these two current ramps represent the GABA-induced current in isolation, This current was plotted as a function of applied potential in (B) to yield the I-V curve of GABA-induced currents. (C) A single 50 ms, 80 mV voltage step was applied once prior to and once in the peak of the GABA response. The differences of current amplitudes were plotted as a function of applied potential (D) to yield two points of a I-V curve fro GABA-induced currents. The dotted line extrapolates the current reversal potential.

as a function of applied potential (Fig. 8D). The line through the two data points clearly does not reflect the true I-V relationship of the response; however, it permits determination with fair accuracy the reversal potential of the response. To obtain a more complete I-V relationship, the two voltage steps above can be substituted by trains of voltage steps.

I-V plots may also be used to determine the quality of the voltage clamp achieved if the reversal potential (equilibrium potential) is known, as is the case when isolation solu-

60 Son theimer

&I -0 2

P ,E -04

- -06

A 80 mV

-70 mV

,110 mV

Fig. 9. I-V curves used to judge quality of point clamp. (A-C) Repre- sentation of families of current recordings from three different B104 cells (neuronal cell line) using the same step protocol indicated in inset. Peak Na+ currents at each potential were plotted in (Dl. Arrows in (A-C) point to current trace in response to a -40 mV voltage step. Only the recording in (A) is under appropriate point clamp. (B) and (C) are distorted by a delay in current activation at threshold (arrows) and by a shift in I-V relationship (D) overestimatmg the true current reversal potential. ENiI was 40 mV.

tions limit ionic movements to only one ion, or when it is established clearly, as in the case of GABA, receptors, that currents are mediated predominantly by one ion. Under the imposed ionic conditions, currents must reverse close to the theoretical equilibrium potential for the permeable ion. In poorly voltage-clamped cells, reversal potential is either not achieved or at potentials more positive than the equilibrium potential (see Fig. 9D and next section for further discussion).


4.5.3. Conductance-Voltage Curves If the reversal potential is known, a conductance volt-

age (C-V) curve can be readily calculated by dividing current at each potential by the driving force (V-V,,,). These curves are typically sigmoidal and can be fitted by a multistage Boltzmann function. Provided that current through a single channel is linear, conductance is proportional to the number of open channels. Thus the G-V curve resembles an activation curve.

4.5.4. Steady-State Inactivation Curves Voltage dependence of current inactivation permits

determination of the fraction of channels available for activation as a function of voltage. Currents are activated by a step to potentials at which the largest conductance is achieved. This voltage step is preceded by variable prepulse potentials at which the membrane is maintained for 200-1000 ms (Fig. 10A). Current amplitudes at each potential are normalized to the largest current recorded and plotted as a function of prepulse potential (Fig. 10B). These curves can be fitted to a multistage Boltzmann function. In the experiment illustrated, a potential of -40 mV yielded about 50% of Na+ channels available for activation (dashed line).

4.5.5. Deactivation i-V Curves Because voltage-dependent currents are often also

time-dependent, and may activate and/or inactivate in a time-dependent manner, I-V curves as described earlier cannot distinguish between time- and voltage-dependence. An elegant way to determine conductance independent of its time-dependence is provided by analyzing the deactivation process (Fig. lOC,D). After termination of the voltage step, deactivation, which is the reversal of activation, results in tail currents (Fig. lOC, dotted line). Immediately after termi- nating the voltage step, for a brief period of time current continues to flow through open channels and only subsequently terminates with time-dependent channel closure (relaxation of tail currents). Thus, current amplitudes measured at the

62 Son theimer

prep&e

C lo 7 [nAl

10

06

2 06

g04

L

-.----- ._______ _ ____

02

00 -140 -120 -100 -80 -60 -40

prepulse [mV]

[n_Al

1 40 mV

80 mV -12U mV

Fig. 10. Steady-state inactivation and tail-current analysis. Current recordings from two different spinal cord astrocytes. (A) To study steady- state current inactivation, inward Na+ currents were activated by stepping the membrane to -20 mV for 8 ms. Voltage step was preceded by varying prepulse potential ranging from -130 to -30 mV (step protocol, see inset). (B) Peak current amplitudes m (A) were normalized to the largest current amplitude and plotted as a function of prepulse potential. The data were fitted to a two-stage Boltzmann equation (solid line) to yield steady-state inactivation (h,) curve. (C) Tail current analysis. Out- ward currents were activated by a 15 ms voltage step from -80 to 80 mV. This step was followed by a second step to varying test potentials ranging from -30 to -120 mV (see inset), resulting in “tail” currents. (D) Current amplitude of tail currents was measured 500 l.~s after stepping potential to second step potential (dotted line) and plotted as a function of applied potential to yield tail current I-V curve.

peak of these tails resembles time-independent current amplitudes, and these typically yield linear I-V curves (Fig. 10D).


4.6. Fitting of Time-Constants

63

Current activation and inactivation kinetics have been well described by mathematical models. If the model is known, the data can be fitted to the model and will allow derivation of important kinetic properties, such as time constants for activation (2,) and inactivation (z,), respectively (Hodgkin and Huxley, 1952). Often the models used are an oversimplification of the true biology. This is particularly true for fitting of whole-cell data for two reasons. First, whole- cell current may be mediated by the combined activation of numerous channel types, and second, even if one can be reasonably sure that currents are mediated by a single channel population, the biophysics of this channel type, e.g., the number of open and closed states, may be unknown.

Fitting routines are an integral part of numerous data acquisition or data analysis packages. Most commonly these use either a Simplex or a Levenberg-Marquard algorithm to minimize the least squared error. Both algorithms are capable of producing excellent and fast fitting to small data sets. Examples of Levenberg-Marquard fits are demonstrated for two examples in Fig. 11. These were obtained using the script interpreter of Origin (Mica Cal) by fitting to user-defined functions. The examples illustrated fitted multiple parameters simultaneously. Thus, in Fig. 11A transient K+ current activation and inactivation was fitted to a n4g model of the form:

f(t) = A0 + Al . (1 - EXP[- (t - tO)/y,,]}4 . {EXP[- (t - tO)/z,]}

as used by Connor and Stevens (1971) to describe kinetics of A-currents; in Fig. 11B Na+ current activation and inactivation was fitted to the Hodgkin-Huxley equation:

f(t) = A0 + Al . (1 - EXP[- (t - tO)/z,]}3 s {EXP[- (t - to)&])

It is important to keep in mind that data fits are not sufficient to formulate a model, but rather assume that the model is known and used to derive variables such as z, and zh contained in the model.

64 Sontheimer

A 70 mV

40 mV

10 mV

-10 mV

-20 mV

150

200 A I I 2 4 6 @=I

Fig. 11. Fitting of current traces using least squared fit. (A) Transient outward K’ and (B) inward Na+ currents were fitted to established kinetic models using a Levenberg-Marquard algorithm to minimize the least squared error. Fitted curves were superimposed on data. The models used were: (A) n4h describing transient K+ current according to Connor and Stevens (1971): f(t) = A0 + Al . (1 - EXP[- (t - tO)/z,]}’ . {EXP[- (t - tO)/z,,]). (B) The Hodgkin-Huxley (1952) equation described Na+ current kinetics: f(t) = A0 + Al . (1 - EXP[- (t - tO/rJ3 * (EXP[- (t -tO)/z,]).

4.7. Data Presentation

Presentation of patch-clamp data has become significantly easier with the advent of microcomputer-based data acquisition, virtually eliminating the use of scissors and glue.


Digitized data can be easily exported in ASClI format and read into numerous powerful spreadsheet programs (Excel, Lotus, Quattro, SigmaPlot, Origin, Plotlt). More recently, two scientific spreadsheet programs (Origin, Plotlt) have included import modules that permit importation of Axon binary data. Thus, without prior conversion, axon data files can be imported for plotting, data analysis, and graphing. The author’s laboratory is currently using Origin, which is based on a scientific scripting language, LabTalk, for which a script interpreter is part of the program. This allows simple programming of frequently used commands or sequences of commands into “macros” that can be assigned to visual but- tons on the screen. This allows the computer literate (non- programmer) to design custom data analysis and graphing schemes. Since most of the graphing programs are Windows- based, merging of graphs into word processors or other drawing programs is easy. Finally, high-quality, affordable laser printers have made plotters obsolete. Most of these printers, such as the HP Laserjet 4, will readily accept 80 g glossy paper, providing a quality printout that is indistinguishable from glossy photographs.

5. Limitations, Pitfalls, and Errors

5.1. Series Resistance and Its Consequences

As mentioned previously, the major limitations of the whole-cell patch-clamp recording technique lie in its design as a continuous single electrode voltage clamp. The continuous use of one electrode for current passage as well as voltage sensor makes true membrane voltage determination impossible. The technique assumes that pipet voltage equals membrane voltage, because voltage commands are imposed on the pipet, not on the cell. However, recording pipet and access resistance (owing to potential clogging at the electrode tip) are in series with the current recording and the voltage command. This series resistor in conjunction with the membrane resistance acts as voltage divider to all imposed voltages. Consequently, only in cases where the membrane

66 Sontheimer

resistance greatly exceeds the series resistance is an adequate voltage clamp (point clamp) assured. Under experimental circumstances, series resistance accounts for at least 5, more typically lo-15 M!L In order to keep the voltage error below l%, membrane resistance has to be two orders larger than the series resistance, in our example -1 GQ. This is hardly the case, and certainly does not hold true during activation of ionic currents. In order to assure best possible recording conditions, the following steps are absolutely necessary:

1.

2.

Electrode resistance has to be minimized as much as possible, depending on the size of the cells to be studied. In our experience, and dependent on the solutions used, cells from 840 pm in size can be successfully patched with electrodes 1.5-3 M&2 range. However, once the whole-cell recording configuration has been achieved, it is important to frequently check for adequate compensation. If all compensation mechanisms are turned off, one may see a dramatic decrease in the magnitude of initial capacitance transients observed. The most likely explanation for such a change is the clogging of the electrode tip with cell membrane, which directly interferes with clear access to the cell’s interior. This phenomenon of membrane “healing” around the electrode tip can be prevented by buffering [Ca*+], using high concentrations of EGTA or BAPTA. In an acute situation, slight positive or negative pressure can reverse electrode clogging. In our experience, it is possible to achieve access resistances of 5 M&2 prior to series resistance compensation. Series resistance (RS) needs to be compensated for. Most patch-clamp amplifiers provide a positive feedback series resistance compensation circuit, in which a signal proportional to the measured current is added to the command potential. RS is determined by adjusting Rs and Cp controls to square out a command voltage. Sub- sequently, Rs compensation is activated. Although theoretically near 100% compensation is possible, real experiments hardly allow gain setting of more than 50-


3.

80%. Rs compensation scales the command input to account for the voltage loss across Rs. Rs compensation is very sensitive to changes in the fast (pipet) capacitance compensation. It is essential to adjust this compensation properly to achieve maximum percent compensation settings. Note that continuous bath perfusion may result in some oscillations of the bath fluid level. As a consequence, this would also change the effective capacitance of the pipet and would make the fast capacitance compensation unstable and thereby Rs compensation prone to ringing. Effective compensation under those circumstances thus requires stable bath perfusion level. The problem can be reduced by the use of heavily Sylgarded electrodes that have a much reduced capacitance.

In an ideal case, with Rs of 10 MQ, a voltage step of 100 mV results in a current flow of 1 nA, and an apparent input resistance Rcell + Rs of 100 MR. A 1 nA current flow across Rs generates a 10 mV voltage drop across Rs and thus a 10% error. Using Rs compensation, assuming an 80% compensation, the error is reduced to 2 mV or 2% of command voltage, a tolerable error. However, suppose activation of voltage-dependent channels gives rise to a 5 nA current. During the peak of this response, the input resistance falls to 20 Ma and, in the uncompensated situation, the membrane experiences only a 50 mV voltage drop. Even with 80% compensation, a 10 mV (10%) error still remains. Cells with low input resistances are almost impossible to record from. Should the membrane impedance be cl00 MSZ, it is advisable to increase it by inclusion of ion- channel blockers to block conductances that are not of immediate interest. Thus, K+ channel blockers could be included and Cl- replaced by acetate to allow resolution of small Na+ currents.

Uncompensated Rs will have two additional detrimental effects on current recordings. It will affect the time response to a voltage change, and it will result in increased signal noise. The transmembrane voltage resulting from a

68 Son theimer

step change in voltage is described by Vm = Vc [ 1 - exp (- z/ (Rs + Cm)] with an effective time constant of z = RC (if Rs << Rm). Assuming an uncompensated Rs of 10 Ma and Cm of 100 pF, charging of the membrane will be slowed with an effective time constant (z) of 1 ms. Unfortunately, Rs will also filter any current flow recorded with this arrangement. In the absence of compensation, this results in a single pole RC filter with a -3 dB frequency described by F = 1/(2n . Rs . Cm) resulting in a cutoff frequency (F) of 159 Hz for the earlier example.

5.2. Voltage Clamp Errors

Voltage clamp is prone to error, because it makes numerous assumptions that may not be valid under the given experimental conditions. It assumes that the cell is isopotenial and that the voltage measured at any one point across the membrane is the true membrane voltage. Analogously, current injection, which imposes change to the membrane voltage, is thought to be uniformly realized in all parts of the membrane, including distant processes. This, however, is not the case. As a result, two sources of error, namely voltage (point) and space clamp errors, exist. As illustrated later, whole-cell patch-clamp recordings are even more susceptible to error than classical two electrode recordings, and as such, the experimenter needs to be constantly aware of possible sources of this error.

52.1. Space Clamp Space clamp limitations are intrinsic to voltage clamp

and do not differ in their principles between different voltage clamp techniques. Current injected into the cell to maintain or establish a change in membrane voltage will spread radially from the injection site, and decay across distance with the space (length) constant h. In small diameter spherical cells, this is a minimal concern. However, in a process-bearing cell, the current signal may have distorted by the time it reaches distant processes hundreds of urn away from the injection site. In the best of circumstances, the sig-


nal will be attenuated and the voltage changes imposed will be smaller. In the worst case, distant membranes may not experience any voltage change at all.

Double electrode voltage clamp methods are somewhat advantageous in that they allow detection of space clamp problems more readily. In this case, the voltage electrode can be inserted at a distance from the current electrode, and closer to the site of interest. Whole-cell recordings clamp the voltage of the electrode tip, and thus provide no means to establish any true recording at a site distant from the electrode. Some investigators have chosen to insert a second patch- clamp or sharp microelectrode into cells to monitor the true voltage changes observed. Space clamp problems can only be reduced by recording from small cells with simple morphology, ideally spherical cells. Nonspherical cells can sometimes be “rounded-up” by exposure to serum or treatment with dBcAMP. However, often the most interesting cells bear extensive arborized processes. A second way to assure that injected current can travel further is to increase the cell’s impedance. Thus it is often possible to block parts of the cell’s conductance (e.g., K+ conductance) pharmacologically to effectively increase the length constant (h) of the cell.

5.2.2. Voltage (Point) Clamp The whole-cell recordings technique effectively voltage

clamps the electrode, and, by assuming that its resistance is small relative to the cell’s resistance, the cell’s voltage is assumed to be clamped. As described earlier, this is only the case if series resistance is small and well compensated for. Unlike space clamp errors, point clamp errors can be readily detected and often eliminated. After canceling series-resistance error (at least partially, see earlier) the experimenter can calculate the voltage error that now is a linear function of current flow. Point clamp errors produce primarily two distortions to the recorded signal: slowing of current kinetics, and apparent attenuation of true current amplitudes caused by uncompensated voltage error. Errors caused by slow activation can be identified readily from the current traces.

Sontheimer

5.2.3. Determining Quality of Clamp from 1-V Plots I-V plots are a very sensitive way to evaluate point clamp

errors. Figure 9 demonstrates Na+ current recordings from three different cells of a neuronal cell line (B104). Current recordings were obtained under appropriate and poor voltage control (Fig. 9A-C), and peak current values were plotted as a function of membrane potential and superimposed in Fig. 9D. Under the imposed ionic gradients of the recordings, the theoretical equilibrium potential for Na+ was -40 mV. The current traces in Fig. 9A yield an I-V plot that reverses close to EsNa indicative of proper voltage control, whereas the traces m Fig. 9B indicate a 20 mV more positive reversal potential, and in Fig. 9C the currents do not reverse at all. Thus the I-V plot directly indicates the severity of the voltage errors in Fig. 9B,C. Although the recording in Fig. 9B still yielded the same peak in the I-V relationship (at -20 mV) as the recording in Fig. 9A, the voltage step to 40 mV (arrows in Fig. 9A-C) was significantly delayed. In Fig. 9C, voltage control is lost at the threshold of current activation (-50 mV), and currents do not activate in a graded fashion but rather “escape” to reach near peak amplitude with a severe delay. Similar I-V plots of transmitter-induced currents for which the major ion carrying the response is known (such as GABA) permit utilization of the reversal potential as an indicator for proper voltage control.

We have found that by minimizing Rs and utilizing I-V curves, currents of up to 10 nA can be properly voltage clamped over a wide voltage range. However, under all these circumstances potential space camp errors remain, and most likely currents activated in remote processes are not recorded at all.

6. Special Applications A number of specialized applications utilize the whole-

cell patch-clamp recording technique. Three of these, namely perforated patch, patch slice, and single-cell PCR, are described in detail in other chapters of this book. One application worth mentioning is the use of patch electrodes for dye loading.


Since whole-cell recordings allow low-resistance access to the cell’s cytoplasm, it provides a convenient means to load cells with biological markers, such as fluorescence indicators, horseradish peroxidase, or biotin. Loading of cells can be used to address a number of questions:

1. Since fluorescence markers may diffuse through gap junctions, dye-diffusion can be used to search for the existence of gap junctions between cells. Although in principle, numerous low-mol-wt compounds can be and have been utilized for this purpose, Lucifer Yellow (LY) has been a longtime favorite, because it is readily retained in cells even after fixation and is among the fluorescence compounds with the highest quantum yields. We have used LY as a pipet solution constituent in numerous recordings and have not detected any interference with our ability to resolve ionic currents. The use of LY to study gap junction coupling is described in detail elsewhere (Ransom and Sontheimer, 1992).

2. LY may also be used as a cell marker allowing the localization of a cell from which recordings have been obtained. We frequently fill cells with 0.2% LY (potassium or Li+ salt) during recordings and after fixation utilize cell-specific antibodies to antigenically identify the cell. LY is believed to form covalent bonds with elements of the cell’s cytoskeleton, and is retained in cells even after formaldehyde fixation and membrane permeabilization. In some preparations, such as brain slices, LY fills are the only way to resolve the complex arborization of cell processes.

3. Although membrane-permeable AM esters exist for most ratiometric fluorescent indicator dyes, their characteristics can differ depending on the method of cell loading (Almers and Neher, 1985). Dyes can be readily loaded through a patch pipet through which electrophysiological recordings can be obtained while imaging an ion of interest ratiometrically.

72 Sontheimer

7. Conclusions

What started out as a spinoff from single channel patch recordings has resulted in a technique more useful than its inventors had anticipated (Sigworth, 1986). Its ease of use makes whole-cell recordings the most widely used intracellular recording technique to date and in many instances has replaced sharp microelectrode recordings. However, whole- cell patch-clamp recordings are notoriously prone to error and may not always generate accurate recordings. It is thus important to understand the limitations of the technique. If proper care is taken, whole-cell patch-clamp allows the study of almost any small cell of interest, and has opened the field of single-cell electrophysiology. Modifications (perforated patch) and applications of the technique to more intact preparations (patch slice) have provided invaluable insight into nervous system function.

Acknowledgments

The author wishes to thank Mary-Louise Roy and Joseph Santos-Sacchi for critical comments on the manuscript. Dur- ing preparation of this manuscript the author was supported by grants IBN-9310277 from the National Science Founda- tion, and ROl-NS31234 from the National Institutes of Health.

Recommended Readings

For a more in-depth description of the patch-clamp technique, the following books are highly recommended (in chro- nological order):

Single-Channel Recording (1983) (Sakman B. and Neher E., eds.), Plenum, New York.

The Biophysical Basis of Excitability (1985) (Ferreira H. G. and Marshall M. W., eds.), Cambridge University Press, London.

Whole-Cell and Microelectrode Voltage Clump, S. Jones (1990), in Neuromethods, vol. 14 (Boulton A. A., Baker G. B., and Vanderwolf C, H., eds.), Humana, Clifton, NJ.

Ionic Channels of Excitable Membranes (1992) (Hille B., ed.) Sinauer, Sunderland, MA.


Ion Channels, Methods in Enzymology (1992) (Rudy B. and Iverson L. E., eds.), Academic, San Diego.

The Axon Guidefor Electrophysiology 6 Biophysics Laboratory Techniques (1993) (Sherman-Gold R., ed.), Axon Instruments, Inc., Foster City, CA.

References

Almers W. and Neher E. (1985) The Ca signal from fura- loaded mast cells depends strongly on the method of dye-loading. FEBS Lett. 192,13-l&

Bezanilla F. and Armstrong C. M. (1977) Inactivation of the sodium channel: I. Sodium current experiments. I. Gen. Physiol. 70,549-566.

Connor J. A. and Stevens C. F. (1971) Voltage clamp studies of a transient outward membrane current in gastropod neural somata. 1. Physzol, (Land.) 213,21-30.

Hamill 0. P., Marty A., Neher E., Sakmann B., and Sigworth F. J. (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pfltigers Arch. 391,85-100.

Hodgkin A. L. and Huxley A. F. (1952) A quantitative description of membrane current and its application to conduction and excitation in nerve. 1. Physiol. (Land.) 117,500-544.

Hodgkin A. L., Huxley A. F., and Katz B. (1952) Measurement of current-voltage relations in the membrane of the giant axon of lohgo. 1. Physiol. (Land.) 116,424448.

Neher E. (1982) Unit conductance studies m biological membranes. Tech. Cell. Physiol. P121,1-16.

Neher E. and Marty A. (1982) Discrete changes of cell membrane capacitance observed under conditions of enhanced secretion in bovine adrenal chromaffin cells. Proc. N&l. Acad. Sci. USA 79, 6712-6716.

Pusch M. and Neher E. (1988) Rates of diffusional exchange between small cells and a measurmg patch pipette. P’iigers Arch. 411, 204-211.

Ransom B. R. and Sonthelmer H. (1992) Cell-cell couplmg demonstrated by mtracellular injection of the fluorescent dye Lucifer yellow, m Electrophysiologzcal Methods for In Vitro Studies In Vertebrate Neuro- biology (Kettenmann H. and Grantyn R., eds.), Wiley-Liss, New York, pp. 336-342.

Sigworth F. J. (1983) Electronic design of the patch-clamp, m SingZe-Chan- nel Recording (Sakmann B. and Neher E., eds.), Plenum, New York, pp. 3-35.

Sigworth F. J. (1986) The patch-clamp is more useful than anyone had expected. [Review]. Fed. Proc. 45,2673-2677.

Pressure/Patch-Clamp Methods

Owen I? Hamill and Don W. McBride, Jr.

1. Introduction

It is approaching 20 years since the introduction of the single-channel patch-clamp recording technique (Neher and Sakmann, 1976), and over the last two decades its refinements and diverse applications have served to maintain it as the dominant technique in membrane physiology (Neher, 1992; Sakmann, 1992). Historical accounts of the development of the technique have been given (Sigworth, 1986), and an extensive literature exists detailing and updating various aspects of the method (Sakmann and Neher, 1983,1995; and this volume). In this chapter we focus on a critical yet somewhat neglected aspect of the method, namely the magnitude and time course of the suction/pressure applied to the patch and its consequent effects on membrane and channel properties. Although suction is most often used in obtaining the tight seal, it has also been shown that excessive suction alters the properties of specific membrane ion channels (Hamill and McBride, 1992). In particular we describe here recent development of pressure clamp techniques that allow the application of precise and rapid suction/pressure steps to membrane patches and whole cells (McBride and Hamill, 1992, 1993, 1995). The combination of the pressure clamp/patch-clamp techniques provides a means to study the dynamic properties of membrane mechanics and membrane ion channels in response to mechanical stimulation.

From Neuromethods, Vol 26: Patch-Clamp Applrcabons and Protocols Eds A. Boulton, G Baker, and W. Walz 0 1995 Humana Press Inc

75

Hamill and McBride

2. General Cell-Attached Patch Recording Procedures

There are a number basic “tricks” required in obtaining routine tight seal (l-20 GQ) formation (Hamill et al., 1981). The specific details may vary from preparation to preparation and, indeed, when starting a new preparation some time may be required to modify them to suit the new preparation. Nevertheless, these steps include mechanical and/or enzymatic treatment of the cell to ensure the plasma membrane is exposed to direct contact with the patch pipet (this may involve collagenase, protease, low Ca2+, trituration, microdissection, micro-water blasting,” and so on), use of “fresh” pipets (i.e., pulled on the day of recording and used only once; firepolishing of the pipet tip may or may not be necessary), use of ultrafiltered pipet solution, application of positive pressure to the pipet to maintain the internal walls of the pipet tip region as clean as possible by continual efflux of solution as it approaches the cell, diluting the pipet solution to approx 90% of the osmolarity of the bath solution, and finally, applying suction to the pipet after initial contact with the membrane.

The generally accepted concept of seal formation and morphology is shown in Fig. lA, in which a membrane bleb is pulled into the pipet increasing the surface area of interactions between membrane and glass. The exact nature of the interactions and forces underlying the glass-membrane seal (cf, Sokabe and Sachs, 1990; Opsahl and Webb, 1994) and the integrity of membrane-cytoskeleton interactions (Milton and Caldwell, 1990; Ruknudin et al., 1990) remain unknown. In the vast majority of cases the application of suction is the final and critical step that forms the tight seal. However, spontaneous seals have been observed without application of suction (Hamill, 1983; Sakmann and Trube, 1984). We find that such seals form infrequently, are of sporadic nature, and are not mechanically stable (i.e., they do not allow cell-free patch or whole-cell recording configurations). These features may reflect heterogeneities in the cell membrane or variability in

fFessufe/fatch-Clamp Methods 77

Suction seal Spontaneous seal

Fig. 1. Schematic representation of patch morphology for suction- induced (A) vs spontaneous (8) tight seals. Electrically, both seals may have sufficiently high resistance (i.e., >1 G&2) to permit single-channel current recordings. However, the main difference involves the contact/ adhesion area (hatched regions) between the membrane and pipet, which presumably determines the mechanical stability of the patch and may greatly increase the seal resistance (>lO GR).

the ease of sealing among various cell types. Whatever the case, they result in reduced contact area between the membrane and the pipet as represented in Fig. 1B and consequently less stable seals. Nevertheless, it may be that the membrane patch in these seals is the least disturbed in terms of disruption of membrane-cytoskeleton interactions.

3. Methods of Applying Suction

3.1. Steady-State Methods

A variety of different methods has been used to apply suction/pressure to the suction port of the patch pipet holder. The simplest and what still remains the most common method is the use of mouth or a syringe. The suction/pressure can be monitored with a water or mercury filled manometer or an electronic pressure transducer. The latter offers the advantage of being able to monitor and record the

Hamill and McBride

pressure continuously. Although we have generally built our own pressure monitors from discrete parts (McBride and Hamill, 1992) there are also stand-alone commercially available electronic manometers (e.g., SenSym, Malpitas, CA; Omega, Stamford, CT). Although both mouth and syringe- applied suction can also be used to mechanically stimulate the patch and study mechano-gated (MG) channels, they lack precision and the ability to apply rapid, repeatable suction/ pressure steps.

3.2. Perturbation Methods

The basic principle of relaxation techniques is that an appropriate perturbation (e.g., voltage, temperature, or pressure) is applied to a reaction system and the rate at which it relaxes to its new equilibrium is measured (Eigen and DeMaeyer, 1963). The most straightforward way to apply a perturbation is as a step function. Over the last few years a variety of methods has been developed that allows the application of controlled pulses of suction/pressure to cell- attached patches. Fred Sachs and colleagues developed an air- based syringe system driven by a linear stepper motor as well as a servo-controlled oil-based pressure clamp. Unfortu- nately, details concerning these methods are scarce (see cita- tions in Sachs, 1987; Sokabe and Sachs, 1990). Another method involving a microprocessor-controlled piston was used to provide reproducible suction/pressure waveforms (Lane et al., 1991). However, the first detailed description and application of a pressure clamp system was provided by McBride and Hamill (1992). Figure 2 is a schematic illustrating the mechanical arrangement of the pressure/patch-clamp arrangement. The system, by incorporating a three-way valve, allows the flexibility of mouth or pressure clamp- applied suction. This pressure clamp is based on a balancing of positive and negative (suction) pressures to achieve the desired pressure. Critical to this strategy is a proportional piezoelectric valve that controls the influx of N, into a mixing chamber that is under constant suction. In the original form a Maxtek valve (Torrance, CA) was utilized. This valve

Pressure/Patch-Clamp Methods

Needle Valve

Fig. 2. Schematic of the mechanical arrangement of the pressure clamp also showing the principle of operation of the piezoelectric valve. The three-way valve is included to allow convenient switching between pressure clamp operation and mouth/syringe-applied suction. The pressure clamp incorporates two transducers. One is involved in control and feedback necessary for the pressure clamp, and the other measures the pipet pressure independent of the setting of the three-way valve. (For a schematic of the electronic controller, see McBride and Hamill, 1992.)

has also been utilized in a water microjet method used to stimulate hair cells (Denk and Webb, 1992). Figure 3 shows two photographs of the actual mechanical arrangement of the pressure clamp. Figure 3A illustrates the overall layout. Notice the size and placement of the valves and mixing chamber with respect to the patch pipet holder. Figure 3B is a closeup view of the regions surrounding the piezo valve. For further details concerning the individual components and electronic circuitry, see McBride and Hamill (1992).

4. Properties of the Pressure Clamp 4.1. Stimulation Protocols

The pressure clamp was designed with two command sources to control the pressure. An internal source is con- trollable through a potentiometer and is meant to adjust the

Hamill and McBride

Fig. 3. Photographs of the pressure/patch-clamp rrg. (A) An overall view of the pressure clamp in relation to the manipulator, microscope stage, and microscope (with nose piece removed). 1 IS the patch pipet holder connected to the head stage. A tube is connected to the suction port of the pipet holder and runs to the branch point (8 in [B]) between the three-way valve, pipet holder, and monitor transducer (3). 2 1s the pressure transducer involved in control and feedback. 4 is the needle

Pressure/Patch-Clamp Methods 81

background pressure. It is most often used during seal formation and steady-state characterization of MG channel properties. Additionally, an external voltage can be used to define the pressure by supplying an input from a function generator or a computer. Diverse pressure waveforms can therefore be applied, including ramps and sinusoids as well as a variety of simple or complex step protocols (see McBride and Hamill, 1992,1993). Although application of the ramp and sinusoidal waveforms can be useful in some instances, step perturbations are most useful for relaxation analysis.

4.2. Speed of the Pressure Clamp

Physical factors influencing the speed of the clamp have been previously discussed (McBride and Hamill, 1992). Cur- rently, the transition time for a suction/pressure step is -10 ms. This is adequate for the characterization of adaptation kinetics in Xenopus oocytes, which have typical decay constants at resting potential of -100 ms (Hamill and McBride, 1992). However, we have observed that MG channels can turn on with latencies <2 ms and with rise times of <l ms. Thus, the 10 ms rise time may be a limitation in proper characterization of the activation kinetics of MG channels. There are several factors that limit the speed of the clamp. Ones that can be dealt with concern the volume of the system and the response time of the piezo valve. Current efforts to increase the speed of the clamp to submillisecond transition times are centered on miniaturization of the system, in particular the mixing chamber volume, and finding faster valves (McBride and Hamill, 1995).

Fig. 3 (continued) valve for adjusting the vacuum flow. (B) Closeup view of elements surrounding the piezoelectric valve (5). 6 is the mixing-chamber: The port is facing the viewer going to the feedback transducer, the back port barely visible is the vacuum line, and the port to the left goes to the three-way valve (7). 8 is the branch point with equivalent arms to the pipet holder and the pressure monitor transducer. 9 is the tube connecting to the N, tank.

Hamill and McBride

4.3. Sensitioitg and Noise of the Pressure Clamp

There is a reciprocal relationship between sensitivity (i.e., the minimal distinguishable step size) and the speed of the clamp with regard to changing the volume of the system. On the one hand, larger volumes decrease the speed (i.e., for a given flux it takes longer to change the pressure in a larger volume). Yet, on the other hand, a larger volume stabilizes the pressure of the system, making it less sensitive to fluctuations in input or output flux. This decreases the pressure noise of the system, and, also, because it requires a larger change in flux for a given change in pressure, the control of the pressure can be more sensitive, albeit slower. Our preference has been to optimize the time response of the clamp since oocyte and muscle MG channels are activated by moderate pressures (-10-20 mmHg). However, some MG channels have been reported to have half-saturation pressure as low as 1-2 mmHg (Sackin, 1989; Kim, 1993). For these more sensitive MG channels, the minimal pressure increments could be reduced by increasing the mixing chamber volume.

4.4. Range of the Pressure Clamp

To begin with, the maximum suction that can be applied by the pressure clamp is only as good as the vacuum source that, at best, would give a suction of 760 mmHg. On the other hand, the maximum pressure is determined by the feed pressure applied to the piezo valve, and for the Maxtek valve is approx 3000 mmHg. Obviously, both of these limits exceed the rupture pressure of the patch (70- 100 mmHg when using patch pipets with tip diameters of -2 pm) and therefore do not present a practical limitation. However, another consideration is the range of the pressure transducer. Here there is a general tradeoff between the dynamic range of the transducer and its sensitivity. For example, for a transducer with high sensitivity the range is necessarily sacrificed.


5. Applications of Pressure/Patch-Clamp Methods

5.1. Sealing Protocols and Determination of Functional Membrane-Cgtoskeleton Interactions

Common to all patch-clamp experiments is the initial tight seal formation that is typically achieved by mouth- applied suction. Although this procedure is adequate in achieving the tight seal, it lacks a certain precision and reproducibility in terms of the mechanical stresses applied to the patch. There have been mixed reports concerning the integrity of the membrane-cytoskeleton complex in sealed patches (Milton and Caldwell, 1990; Sokabe and Sachs, 1990; Ruknudin et al., 1991). However, in the case of voltage- and ligand-gated channels, there has been general agreement (with a few exceptions) between results obtained using whole-cell, single-channel patch-clamp and conventional intracellular voltage clamp recording techniques. In contrast, with MG channels there are several reports that indicate specific changes, such as the loss of adaptation and sensitivity in MG channel properties, caused by mechanical stresses associated with either sealing and/or mechanical stimulation of the patch (Hamill and McBride, 1992). We have found that using the pressure clamp to apply low (cl-2 mmHg) and reproducible pressure/suction protocols for tight seal formation leads to more consistent MG channel behavior in terms of adaptation and sensitivity.

A relatively unexplored aspect of the biophysics of membrane ion channels is the contribution that membrane- cytoskeleton interactions play in specific channel properties. In the past the major approach to investigating this has been to use pharmacological agents targeted toward disrupting (or stabilizing) particular cytoskeletal proteins. However, a disadvantage of this approach is that no agent is completely specific and it may be that observed changes are the consequence of drug effects on the channel or regulatory enzymes

Hamill and McBride

rather than owing to cytoskeletal effects. Recent studies have implicated the cytoskeleton in specific properties of non-MG channels (see Kimitsuki et al., 1990; Johnson and Byerly, 1993; Rosenmund and Westbrook, 1993). Perhaps the pressure clamp can be used in conjunction with pharmacological agents to alter and control the state of membrane-cytoskeleton interactions while monitoring channel activity. Ideally, one would also like to be able to visualize changes in these interactions with microscopy techniques. For example, the application of fluorescence and confocal microscopy techniques could allow for better resolution of the interactions and resolve the relative contributions of the membrane and cytoskeleton to channel behavior.

5.2. Membrane Viscoelastic and Mechanical Properties

Basic properties of the cell membrane, such as its fragility or membrane strength, can be measured with pressure/ patch-clamp technique in terms of the patch rupture pressure. For example, Cooper and Hamill (1989) used patch rupture pressure as a measure of the relative strength of the sarcolemma of control and dystrophic muscle. Since membrane tension (T) depends on both pressure (P) and patch diameter (d) (T = Pd/4, Laplace’s law), it is critical that patch pipets with identical tip diameters are used in comparative measurements. Although patch rupture pressure may give comparative information regarding membrane-cytoskeleton strength, it is a measurement of an irreversible process. A more physiological measurement of the elastic properties of the membrane can be made with small mechanical perturbations. For such studies, membrane morphology and movements can be monitored by high resolution video techniques and membrane area changes by membrane capacitance changes while applying pressure clamp steps (Sokabe et al., 1991).

5.3. Characterization of Mechano-Gated Channels

Perhaps the most significant use of the pressure/patch- clamp has been to reveal the dynamic properties of single


MG channels. Using this technique, a number of novel properties of MG channels have been revealed. These include rapid and complete adaptation to maintained suction, short activation latencies, and voltage-dependent tail currents on rapid termination of suction/pressure. Furthermore, the ability to apply incrementing steps of suction/pressure allow the measurement of stimulus-response relations of these dynamic properties (Hamill and McBride, 1992; McBride and Hamill, 1992).

6. Conclusion

The patch-clamp technique continues to be the most popular method for investigating the biophysics of membrane ion channels. However, in general the vast majority of patch-clamp studies have ignored for one reason or another the potential role of membrane-cytoskeleton interactions in affecting specific properties of membrane channels. There is growing evidence that indeed the cytoskeleton has a functional role in certain aspects of channel behavior. The pressure/patch-clamp technique in combination with visualization techniques provides the opportunity to study in further detail these interactions.

Acknowledgments

We thank the Muscular Dystrophy Association, NIH, and NSF.

References

Cooper B. J. and Hamill 0. P. (1989) Patch-clamp measurements of viscoelastic propertres and mechanoelectric transductron in dystrophic muscle membrane. Sot. Neurosci. Abstr. 15,412.5.

Denk W. and Webb W. W. (1992) Forward and reverse transduction at the limit of sensitivity studied by correlating electrical and mechanical fluctuations in frog saccular hair cells. Hearing Res. 60,89-102.

Eigen M. and DeMaeyer L. (1963) Relaxation methods. Technique of Organic Chemistry 8,895-964.

Hamill 0. P. (1983) Potassium and chloride channels in red blood cells, in Single Channel Recording (Sakmann B. and Neher E., eds.). Ple- num, New York, pp. 451-471.

86 Harnill and McBride

Hamill 0. I’., Marty A., Neher E., Sakmann B., and Sigworth F. J, (1981) Improved patch-clamp techniques for high-resolution recording from cells and cell-free membrane patches. Pflzigers Arch. 391,85-100.

Hamill 0. P. and McBride D. W. Jr, (1992) Rapid adaptation of the mechanosensitive channel in Xenopus oocytes. Proc. N&Z. Acad. Sci. USA 89,7462-7466.

Hamill 0. P. and McBride D. W. Jr. (1994) Molecular mechanisms of mechanoreceptor adaptation. News Physlol. Scz. 9,53-59.

Johnson B. D. and Byerly L. (1993) A cytoskeletal mechanism for Cat+ channel metabolic dependence and inactivation by intracellular Ca++. Neuron 10,797-804.

Kim D. (1993) Novel cation-selective mechanosensitive ion channel in the atria1 cell membrane. Circulation Res. 72,225-231.

Kimitsuki T., Mitsuiye T., and Noma A. (1990) Negative shift of cardiac Na+ channels kinetics in cell-attached patch recordings. Am. J, Physiol. 258, H247-H254.

Lane J. W., McBride D. W. Jr., and Hamill 0. I’. (1991) Amiloride block of the mechanosensitive cation channel m Xenopus oocytes. I, PhysioZ. Land. 441, 347-366.

McBride D. W. Jr. and Hamill 0. P. (1992) Pressure-clamp: a method for rapid step perturbation of mechanosensitive channels. Pfiigers Arch. 421,606-61'2.

McBride D. W. Jr. and Hamill 0. I’. (1993) Pressure-clamp techmque for measurement of the relaxation kinetics of mechanosensitive channels. Trends Neurosci. 16,341-345.

McBride D. W. Jr. and Hamill 0. I’. (1995) Fast pressure clamp for studying mechano-gated channels, in Single Channel Recording (Sakmann B. and Neher E., eds.), 2nd ed. Plenum, New York, pp. 329-340.

Milton R. L. and Caldwell J. H. (1990) How do patch-clamp seals form? A lipid bleb model. Pftigers Arch. 416,758-765.

Neher E. (1992) Ion channels for communication between and within cells. Scrence 256,498-502.

Neher E. and Sakmann B. (1976) Smgle-channel currents recorded from membrane of denervated frog muscle fibers. Nature (Land.) 260, 779-802.

Opsahl L. and Webb W. W. (1994) Transduction of membrane tension by the ion channel alamethicin. Biophys. J. 66,71-74.

Rosenmund C. and Westbrook G. L. (1993) Calcium-induced actin depolymerization reduces NMDA channel activity. Neuron 10, 805-814.

Ruknudin A., Song M. J., and Sachs F. (1991) The ultrastructure of patch- clamped membranes: a study using high voltage electron microscopy. 1. Cell Biol. 112,125-134.

Sachs F. (1987) Baroreceptor mechanisms at the cellular level. Fed. Proc. 46,12-16.


Sackin H. (1989) A stretch-activated K’ channel sensitive to cell volume. Proc. Natl. Acad. Sci. USA 86,1731-1735.

Sakmann B. (1992) Elementary steps in synaptic transmission revealed by currents through single ion channels. Science 256,503-512.

Sakmann B. and Neher E. (eds.) (1983) Single Channel Recording. Plenum, New York.

Sakmann 8. and Neher E. (eds.) (1995) Single Channel Recording. 2nd ed,, Plenum, New York.

Sakmann B. and Trube G. (1984) Conductance properties of single inwardly rectifying potassium channels in ventricular cells from guinea-pig heart. J. Physiol. (Land.) 347,641-657.

Sigworth F. J. (1986) The patch-clamp is more useful than anyone had expected. Fed.Proc. 45,2673-2677.

Sokabe M. and Sachs F. (1990) The structure and dynamics of patch- clamped membranes: a study using differential interference contrast light microscopy. J. Cell Biol. 111,599-606.

Sokabe M., Sachs F., and Jing Z. (1991) Quantitative video microscopy of patch clamped membranes stress, strain, capacitance, and stretch channel activation. Biophys. J. 59,722-728.

Cell-Free Ion-Channel Recording C. G. Nichols, M. B. Cannell,

and A. N. Lopatin

1. Introduction When membrane patches are isolated from cells, one

gains the ability to regulate precisely the composition of the solution bathing both surfaces of the membrane and to change the composition rapidly. However, in detaching the membrane from the underlying cytoskeleton, one irrever- sibly changes the integrated system of which the channel protein is a part, and begins an inexorable loss of channel function. In this chapter we will deal primarily with the techniques available for measuring the response of patch currents to changes in bathing solution and the practical problems to be minimized, or overcome, in implementing such techniques and analyzing the results obtained. Other chapters will deal with patch-clamping in general. Chapter 6 deals specifically with methods of changing the concentration of the solution bathing the intrapipet face of the membrane, and we will concentrate on consideration of the extrapipet solution changes, i.e., changes at the intracellular membrane surface in inside-out patches, or the external face in outside-out membrane patches.

The purpose of this chapter is to describe practical approaches to methods and problems encountered in performing and analyzing experiments with inside-out membrane patches. Four sections follow, dealing with:

1. Problems of vesicle formation when isolating an inside- out membrane patch;

From: Neuromethods, Vol. 26: Patch-C/amp Applications and Protocols Eds. A Boulton, G. Baker, and W Walz Q 1995 Humana Press Inc.

89

90

2.

3.

4.

Nichols, Cannell, and Lopatin

Problems of channel rundown, and how to deal with them statistically; Methods of rapidly changing the solution bathing the excised patch; and Methods for analyzing the kinetics of patch currents after concentration changes, that account for the problems of diffusion to the membrane in the patch.

A final section will provide a potpourri of tips and hints - _. that we have found useful in our own experiments.

2. Making an Inside-Out Membrane Patch: The Problem of Vesicle Formation

Following the formation of a gigaseal, an inside-out patch can be obtained by removing the pipet tip from the surface of the cell. However, it is possible to remove the tip and maintain a gigaseal without getting an inside-out patch, and instead forming a “vesicle,” or “bleb.” Figure 1 explains what a vesicle is, how it forms, and how it may affect single- channel measurements. An inside-out membrane patch requires not only a high electrical seal resistance but also mechanical stability of the seal. Despite years of practical patch-clamping, many details of the physicochemical basis of gigaseal formation remain unclear. During withdrawal of the pipet tip from the cell surface, the sealed membrane patch within the pipet remains intact while the appearance of a cytoplasmic bridge can be observed in many cases (Fig. 1; Milton and Caldwell, 1990). Depending on experimental conditions, there are only two outcomes for the cytoplasmic bridge: disruption or resealing. It appears that resealing to form a vesicle can be stimulated by mechanical vibration, so that high quality micromanipulators and good vibration isolation would be recommended for this reason.

2.1. How to Tell When You Have a Vesicle

Using giant pipets one may directly observe the patch membrane and determine directly whether or not a vesicle is present. In this instance, the formation of an inside-out

Cell-Free lot-t-Channel Recording 91

I I I I , , I , I I I /

I , , ,/a I , ,,,l I , /I? Xl l,/

~~

I ,,l

I , , / , ,l I ,

, 11,

I , / I I ,

extracellul~ , ,, matrix / II - ,

membrane

cytoskeleton

Fig. 1. Lipid bleb model of inside-out patch and vesicle formation. Normally, the membrane is attached to the cytoskeleton and extracellular matrix. The negative pressure (suction) applied to the pipet produces fast fluid flow close to the pipet rim that in turn causes an additional pressure drop owing to the Bernoulli effect. It is presumed that tiny bulges first occur under the pipet rim where membrane attachments to the cytoskeleton and extracellular matrix are weak. When a critical pressure is reached, the bulges fuse to form an invagination, producing a gigaseal in a small pipet. Following withdrawal of the pipet from the cell surface, a cytoplasmic bridge extendmg several microns can be observed. Further withdrawal causes the cytoplasmic bridge to rupture, producing an inside-out patch (A) or a vesicle (B) depending to some extent on the state of the membrane and composition of the bath solution. In most cases, the cell membrane heals and additional patches can be excised from the same cell.

92 Nichols, CannelI, and Lopatin

patch is seen to be followed by the cytoplasmic contents streaming out of the patch. Most of the time, electrical indicators or other controls must be used to get an idea of what is going on in the tip. When channel activity can be recorded in cell-attached configuration, but single-channel current amplitude apparently declines after removal of the patch from the cell, or single-channel current declines within each opening as though the record was being high-pass filtered (AC coupled), it is probable that a vesicle has formed. It is easy to understand qualitatively how the distortions occur (Fig. 2). The surface area and resistance of the outer membrane may be different depending on the shape of the pipet, stability of the membrane, and many other factors. If surface area (So) is small and membrane resistance (RO) is relatively low, single-channel current may simply be reduced compared to expected. On the other hand, if So is large and resistance R, is high, the currents may become disordered, as shown in Fig. 28. It is easy to detect completely disordered currents or currents flowing in the direction opposite to predicted (resulting from a marked vesicle potential), but single-channel currents may be simply reduced in amplitude, which is harder to recognize. This may be the case when the surface area of the outer membrane S0 is relatively small and resistance R0 is relatively low (damaged outer membrane). The most awkward case would be the forming of a vesicle with the outer membrane having semipermeability. In this case, the ionic conductance of the outer membrane is high

Fig. 2. (omosite page) Electrical indications of vesicle formation. Cur- rents flowing through ion channels can be affected by combinations of RC filters arising from the inner and outer membranes of the vesicle (A). Each membrane forms an RC filter and battery (Eo and E,) since both may contain ion selective channels. (B) Depending on the values of R, R, C, C, and E, E, currents may be disordered (l), reduced (2), or even completely “blocked.” The shape of single-channel current responses may appear to have either slow rising (3) or relaxation (4) phases. Sometimes, however, fast currents apparently may be unchanged (5). The vesicle may also have its own resting potential, E,, reversing currents (6) that should flow in a predicted opposite direction.

Cell-Free lon-Channel Recording

so

B 1

2

membrane

93

94 Nichols, Cannell, and Lopatin

enough to allow measurement of undistorted patch (inner membrane) currents, but the macroscopic permeability of the outer membrane may be so low as to not allow access of bath applied molecules to the (inner) patch membrane.

2.2. How to Deal with the Problem

1.

2.

3.

There are ways to avoid vesicle formation:

The possibility of vesicle formation may be reduced if patch excision occurs in a Ca2+-free solution (Horn and Patlak, 1980). Better results may also be obtained in a solution containing Ca*+ chelators, such as EGTA, EDTA, or fluoride ions. Hilgemann (1991) suggests that moving the patch pipet from side to side during excision rather than simply withdrawing from the cell appears to be useful, especially with giant patches on oocytes, which has also been our experience. It may be better to release negative pressure in the pipet and even apply small positive pressure before patch isolation. It is our impression that this is more obviously useful with giant pipets.

If, after all precautions, vesicles are still formed, the outer part of the vesicle membrane may be disrupted using the following methods.

1.

2.

3.

The pipet tip may be passed briefly through the water- air interface of the bath. The number of passages and their durations will depend on vesicle stability. How- ever, especially with large pipets, this procedure may lead to loss of the gigaseal (patch disruption). The pipet tip may be passed through a water-mineral oil interface, such as described later. Empirically, we have found that this approach is less likely to destroy the gigaseal than passage through an air-water interface. The pipet tip may be passed into a Sylgard ball attached within the bath.

Cell-Free lon-Channel Recording 95

4. With giant patches, we have been able to disrupt the outer membrane of the vesicle using mechanical pierc- ing with another micropipet tip.

5. Theoretically, the outer membrane of the vesicle may be destroyed using detergents (e.g., saponin) or obviated using other substances that will make the outer membrane ion permeable, such as nystatin (Horn and Marty, 1988), but we have no practical experience of such approaches.

3. Analysis of Ion Channels in Cell-Free Patches: Dealing with the Problem of Channel “Rundown” A major problem in measuring channel currents in

inside-out membrane patches is the phenomenon of “rundown,” whereby channel activity decreases during the time after patch isolation, significantly complicating the analysis of measured currents. This phenomenon is seen with almost all ion channels, perhaps not surprisingly since many, if not all, proteins are rather unstable when isolated from their natural environment. Various approaches have been described to partially overcome problems of channel rundown, but, unfortunately, at the present time, rundown cannot be completely avoided or reversed. In electrophysiological practice, channel rundown can be seen as a decrease in mean current with time (under stationary conditions of voltage and ion composition), and results from a decrease in the number of active ion channels in the patch, or a decrease in the channel open probability. We are aware of no experimental evidence that single-channel current (conductance) during rundown is changing. Experimentally, in some cases one may see decreasing apparent single-channel current after removal of a membrane patch from the cell, but this phenomenon can be explained by resealing of the outer membrane and formation of vesicle (see earliev).

In patches containing a large number of similar channels, decrease of integral current amplitude after patch excision occurs smoothly with time and is typically described by


A

n 1 min

B

li

Fig. 3. Rundown of KATp channel activity in cardiac membrane patches. (A) Membrane current in a membrane patch following isolation (at arrow). Initially, 7 channels were active. The activity declined quasi- exponentially over several minutes. (B) Membrane current in a patch containing only 1 channel following isolation from the cell (above) and the time integral of membrane current, the slope of which is a reflection of cumulative open time (below). The patch was isolated at the first arrow. The second arrow marks a step change in channel open probability.

an exponential decay (Fig. 3A). However, in single-channel patches, rundown may occur as an abrupt loss of channel activity, up to which time the channel activity may be stationary (e.g., Nichols et al., 1991). Figure 3B shows rundown of a single KATp channel in an inside-out membrane patch from a rat ventricular cell. It is apparent that channel activity (open channel probability) may change suddenly to a new steady state level at some time, before disappearing completely.

Cell-Free Ion-Channel Recording 97

The time course of channel rundown can be very variable from one preparation to another, and the time constant may range from a few seconds to tens of minutes, considerably complicating experiments that require time-consuming measurements, such as channel activity, in different experimental conditions, or averaging of many current records.

3.1. Mechanisms of Rundown

The mechanism of channel rundown in inside-out patches seems to be different in different channels. Rundown probably results from combinations of alterations in the lipid membrane, the channel itself, or a regulatory protein, or the cytoskeleton. These structures are affected by patch excision because the membrane is isolated from internal metabolism (metabolites), second messengers, and other soluble or insol- uble materials. In many cases, ion channels or their regulatory structures become dephosphorylated and lose activity. Add- ing exogenous reagents, such as ATP or corresponding cofactors and enzymes, to the internal solution can restore activity (in general temporarily) or delay channel rundown (see, e.g., Takano et al., 1990). Phosphorylation-dephosphorylation mechanisms of rundown do not discriminate what structures are subjected to regulation (i.e., the channel itself or a related structure). Specific substances that break down microtubules (colchicine) or microfilaments (cytochalasin B) can speed up channel rundown, whereas agents that support these structures (tax01 and phalloidin, respectively) can prevent it or slow it down (Matsumoto and Sakai, 1979a,b; Fukuda et al, 1981; Matsumoto et al., 1984a,b; Johnson and Byerly, 1993; Rosenmund and Westbrook., 1993). In many cases, Ca2+ at the inner surface of the membrane speeds up the process of channel rundown (e.g., Findlay, 1987), presumably by activating Ca2+- dependent proteases or phos- phatases. In this regard, some success has been achieved at avoiding rundown of Ca channels using the phosphatase inhibitor okadaic acid (Ono and Fozzard, 1992).

Rundown can rarely be completely avoided, but in specific cases, unexpected solutions have been found. Many stud-

98 Nichols, Cannell, and Lopa tin

ies on K,, channels have shown that MgATP can prevent rundown or subsequently restore activity (Ashcroft, 1988). In some cases, the addition of exogenous protein kinase A catalytic subunit (Takano et al., 1990; Ono and Fozzard, 1992) further enhances this effect. Treatment of the inner surface of the membrane with trypsin works even better-channels do not run down spontaneously, and rundown is no longer stimulated by internal Ca*+ (Proks and Ashcroft, 1993). How- ever, trypsin treatment also changes some properties of the channel: ATP sensitivity and sensitivity to sulfonylurea drugs are decreased (e.g., Nichols and Lopatin, 1993), so it appears that trypsin is acting directly on the channel itself or a closely associated protein. In delayed rectifier potassium channels (DRKl) expressed in oocytes, we have found that as the current runs down, the amplitude can be restored to the control value just by changing the holding potential to more negative potentials. The activation kinetics of DRKl are not affected during rundown, suggesting that only steady-state inactivation is shifted to more negative values. Unfortunately, continuous changes of holding potential to keep current amplitude at a constant level cannot be continued indefi- nitely-at very negative potentials, the membrane becomes unstable and “noisy,” and large negative holding potentials eventually destroy the gigaseal.

3.2. Accounting for Rundown: Statistical Approaches for Analyzing Current Records

Suppose, by way of example, that one is measuring the effect of some blocking substance X on a current and the only desired information is the fractional reduction of current amplitude (Fig. 4). Even in the case of severe rundown (33%), the blocking effect can be estimated easily:

Fractional current = l/2 [(aZ/aZ) + (a3/a4)] (1)

Noise analysis is a frequently applied tool for determination of microscopic kinetics and single-channel amplitudes from macroscopic records (see, for example, Sigworth, 1980). Certain general assumptions underlie this method:

Cell-Free ion-Channel Recording 99

Before -

After

Substance X

Fig. 4. During rundown, the blocking effect of substances can be estimated by measuring instantaneous effects on application and washout and averaging them. The errors are less than rundown during the time of measurements. (A) “Voltage-activated currents” before, during, and after washout of blocking drug X. (B) Changes in current amplitude with time (see text).

1. Channels are independent; 2. Channels are homogeneous; and 3. Channels exhibit only one conductance.

With these assumptions, single-channel current (i) and number of channels (IV) in the patch can be estimated (see, for example, Sigworth, 1980) from the following relationship:


d - 30%

Fig. 5. Errors in estimation of current variance may be several times greater than rundown during the time of measurements. The noiseless curve represents the average of all current records. Current records above and below the mean current are individual records at the beginning and end of the measurement period. The small rectangles represent estimation of true noise (-10%) associated with each measurement and the large rectangle represents the noise (-30%) that would be estimated by comparing individual records to the overall average. Estimated noise is proportional to 6. Therefore, without taking local means, variance, which is equal to o*, would be overestimated by factor R = (30/10%)2= 9, for rundown of only 30%.

ow = iw - [W2/N] (2)

d(f) = l/(n - 1) x [I(f) - i(f)]2 (3)

where I is mean current, CT is variance, n is the number of observations, and [I(t) - i(t)] is the deviation of each current sample [i(t)]from its mean value [I(t)]. Care must be used when applying noise analysis to currents that are running down. In contrast to the simple estimation of amplitude of blocking effect of substance X described earlier, the estimation of mean current and variance during channel rundown is not so straightforward (Fig. 5). In practice, it is typically necessary to average up to 50-100 current responses to the same voltage steps to get noiseless mean current and reasonable values for 02. Depending on the channel being investigated, it may take many minutes, sufficient time for considerable


rundown in inside-out patches. Direct averaging of all current records would give mean current somewhere between the maximal value in the beginning of measurements and minimum value at the end. In Fig. 5, thin rectangles show estimation of real signal variation, and the thick rectangle at the right reflects the signal variability that would be estimated using total mean current. It is clear that if channel rundown (0 total -30% for deviation from mean current) exceeds the true value of current fluctuations (otiue -10%) then estimation of variance (02) will be in error and will differ from the real value by a factor R = (crtota,/otruJ2 -9, and single-channel current and other parameters will be incorrectly estimated. Fortunately, this problem can be surmounted using “local means” to calculate current variance. Current records are divided into successive groups containing two or more records. The number of records in a group is based on the estimation of rundown within these records, which should be less than the variability between records. For each group, mean current and variance can be calculated and then these values averaged over all groups. Using local means leads to loss of information from the noise in the current records and thus more records are needed to obtain acceptable scatter in the estimated parameters. In practice, groups containing as few as two to four records are necessary, which means that the total number of records should be increased by 25-50% and that the total time of measurements should be increased to the same extent. Obviously, using a longer total measurement time will mean more rundown, thus reducing the accuracy of estimations. However, with currents that run down, using local means reduces the error of estimation by orders of magnitude compared to the tradeoff increase in recording time.

4. Methods for Rapid Change of the Solution Bathing Cell-Free Membrane Patches

The most straightforward way to change the solution bathing a membrane patch is to change the inflow to the chamber in which the patch is placed. With time constants


for solution change >lOO ms for the fastest rates of flow, this may be suitable for measuring steady-state responses of channel current to changes in ligand or permeant ion concentration, but is too slow for measurements of transient responses to channel agonists and antagonists, which may have time constants of milliseconds or less. Even for “steady-state” measurements, the problem of channel rundown (above) frequently means that one would want to change the solution as quickly as possible. Exposure of cell-free patches to air frequently disrupts the membrane and destroys the patch. It is, therefore, generally impractical to change the solution bathing the patch by physically moving the patch from one solution to another through air. It is also a cumbersome procedure, requiring lifting, lateral movement, and lowering of the pipet. Thus, methods for rapid solution change generally employ lateral movement of the patch or the chamber relative to one another. The new solution is separated from the original solution either by bulk application of the new solution, by having parallel laminar flows, or, as originally invented by Qin and Noma (1988), by separating the two solutions by an oil-filled well.

Below we will outline various methods and their limitations, together with references to original papers.

4.1. Methods for Rapid Bulk Application of Solution

The most frequently used method of obtaining rapid changes in the solution bathing the patch is the so-called “sewer-pipe” method in which the tip of the patch electrode is placed in the opening at the end of a capillary tube down which a continuous stream of solution is flowing. By joining several “sewer-pipes” together, Yellen (1982) described rapidly exchanging the solution at the tip of the patch electrode. Various automations of the movement of the sewer pipes have been described (e.g., Akaike et al., 1986; Mery et al., 1992), involving stepper motors or loudspeaker-driven systems, and these allow the solution change to be timed more precisely and synchronized with other events. In some systems (e.g., Konnerth et al., 1987, Dilger and Liu, 1992) the


multiple barrels are connected to a common tip, and which solution flows to the common tip is controlled by solenoid pinch valves. The dead volume within the tip of such systems can be small enough to allow solution exchange with time constants of tens of milliseconds. However, the mixing of solutions cannot be completely avoided.

4.2. Laminar Flow Methods of Separating Parallel Solutions

Brett et al. (1986) provide a detailed description of a parallel solution method of switching solutions; this method involves a tube containing the test solution placed within a larger chamber that contains the first solution. The patch pipet tip enters the test solution through one of two means. In the first approach, the pipet tip is lowered into the test solution through a small hole placed in the top surface of the tube. A similar method is described by Akaike et al. (1986). In the second variant, the pipet tip is placed into a “liquid-filament” -the stream directly at the open end of the tube. In this case, an outflow siphon from the main chamber is placed immediately downstream of the tube end. With this system, Brett et al. demonstrated very rapid change of solution at the tip of the electrode (tau m 1 ms). In a further development of this approach, Franke et al. (1987) automated the movement of the “filament” relative to the pipet tip by fix- ing the tube discharging the “filament” to a piezo crystal held by a manipulator and its position controlled by application of a voltage pulse (Fig. 6).

4.3. Separation of Solutions Using an “Oil-Gate”

Qin and Noma (1988) first described the construction and use of an ‘oil-gate’ for passing the patch tip from one solution to another without passing through air or mixing of solutions at the interface. The oil-gate consisted of a slit filled with paraffin oil in a partition wall separating the two solutions. The walls of the slit were painted with silicone rubber to make them hydrophobic and retain the oil within the slit. The major consideration in our experience has been the need


fixed electrode

shifting liquid filament

Fig. 6. The liquid filament solution switch. The tube discharging the liquid filament is fixed to a piezo crystal held by a micromanipulator. The patch electrode is held in a fixed position. The prezo crystal shrinks or extends, depending on the applied voltage (1 kV), sufficient to move the tube by 10 pm. This moves the liquid filament so that the electrode tip is either in the filament or out of it (redrawn from Franke et al., 1987).

to keep the slit diameter as small as possible consistent with passage of the tip through it (co.5 mm). Figure 7 shows our version of the chamber (Lederer and Nichols, 1989), which consists of four “channels” that run into the same end-pool. The channels are separated by Perspex partitions, in each of which is placed an “oil-gate.” We simplified the fabrication of the oil-gate by drilling a small well (of a diameter just less than the width of the partition) into the partition, and then cutting out the edge of the well. The well is filled with paraffin oil and it is then possible to move the electrode tip from one chamber, through the oil-filled well (oil-gate), into the next chamber without exposing the tip to air. On moving from one channel to the next, no mixing of aqueous solutions takes place, apart from the small amount of solution carried in the tip between the membrane and the oil. The bath ground electrode is placed in the end-pool. A float, connected to a tension transducer, senses the solution level in the end-pool, and produces a signal that is used to control the solution level (Cannel1 and Lederer, 1986) by varying the rate of outflow from the end-pool.

Cell-Free fan-Channel Recording

Fig. 7. The oil-gate chamber, Solutions flow through four channels (C), supplied through different inflow lines (I), into a common end-pool (l?), and are pumped out through a single outflow (0). A float (F) m the end-pool senses the solution level that is controlled by varying the outflow rate. Cells are placed in the first channel, where a gigaseal is formed. The electrode is then lifted (possibly with the cell attached) and the tip moved through the oil-filled gate (G) into the second channel. The cell is pulled off the electrode tip at the solution-oil interface, formmg an inside-out patch. Subsequent changes of solution bathing the exposed surface of the inside-out patch are made by moving the electrode tip from one channel to another through the oil-gates (G) (redrawn from Lederer and Nichols, 1989).

Micropipets are “sealed” onto cells (placed in the first chamber) by applying light suction to the rear of the pipet. Inside-out patches are then obtained by lifting the electrode and passing the electrode tip through the oil-gate. Only rarely are patches ruptured on passing through the oil-solution interface. Koh and Vogel (1993) have recently described a variant in which a pipet tip can access a microcapsule within a larger chamber by passing through an oil-gate at the surface of the microcapsule as a means to perform oil-gate experiments with a very small volume (10 pL) of test solution.

106 Nichols, Canneii, and Lop&in

Flow m

Bath solution

Fig. 8. The air-gate chamber. A glass capillary with a hole (H) is connected to a polyethylene tube such that suction can be applied to one end, thereby introducmg a solution flow through the tube. The tube contains two reference electrodes, connected together, each comprising an Ag/AgCl pellet connected to the solution by an agar bridge. Two small air bubbles are used to separate different solutions. The pipet tip containing an excised patch is inserted into the hole and Sylgard coatings are used to provide a seal between the tube and the pipet. The dragram illustrates the situation prior to draining the bath (redrawn from Kakei and Ashcroft, 1987).

4.4. Separation of Solutions Using an “Air Gate” Kakei and Ashcroft (1987) have described an ingenious

method of switching solutions via an “air gate” (Fig. 8). This method should be suitable for use with very small quantities of solution. The method involves inserting the patch at the tip of the electrode into a small tube that is perfused independently of the main bath. Solutions flow through the small tube, separated by an air bubble. As the bubble passes the tip of the electrode, the tip is exposed to air for <250 ms, and seals can be maintained through such a procedure. How- ever, it seems that some mixing of the second solution with the thin film of the first that must remain on the tube inner surface is likely to occur. As described, this method also suf- fers from the fact that the referencing electrode(s) placed in the tube (Fig. 8) must be physically separated from the


electrode tip, and the recording circuit will be broken while the air bubble passes between the reference electrode and the pipet tip, unless, as described, two reference electrodes are used, in which case current transients will occur on entering the new solution.

5. Analysis of Responses to Rapid Concentration Changes

No matter how the solution change is achieved, or how rapidly, there is an inevitable problem that arises in analyzing the responses of inside-out membrane patches to concentration “jumps.” This results from the fact that the patch of membrane in the tip of the pipet is recessed some distance from the orifice of the pipet (Sokabe and Sachs, 1990) so that the diffusion-limited time course of the solution change within the pipet itself will contribute to the overall response time of the channels.

5.1. Modeling the Pipet Geometry

In order to develop a means of accounting for the delays associated with diffusion in the response of isolated patches to concentration “jumps,” we have analyzed the effects of pipet geometry on the time course of solution change at the patch of membrane within the pipet (Cannel1 and Nichols, 1991). The approach was to model the geometry of the pipet and then compare the model to observed diffusion-limited patch responses. The pipet geometry was modeled as a right cone with the apex removed (Fig. 9). The patch of membrane within the pipet was considered to be a flattened disk (Sokabe and Sachs, 1990) presenting a reflective barrier for diffusion at a distance I,,,, from the pipet tip whose radius was R,,,. The length of the pipet between the tip and the membrane was split into n elements of thickness dx. The flux of solute (F) across the boundary between two elements is then given by

F = D . 8. {R,,, + [(LPI,- x,>TdN~* (C, - C, + ,VdxW (4)


-1‘ R TIP

Fig. 9. A model of the membrane patch. The geometry of the pipet is modeled as a right cone (half angle e) with the apex removed. The membrane patch within the pipet forms a reflective barrier for diffusion at a distance I.,, from the pipet tip of radius R,,. The length of the pipet between the tip and the membrane is split into n elements of thickness dx.

where C is the concentration of solute in the ith element, x, the pos&on of that element from the patch, and D the diffusion coefficient.

5.2. Time Course of Solution Change: The Effects of Pipet Geometry

After a step change of solution at the tip of the pipet, the simulated time course of solution change at the membrane is reasonably approximated by an exponential decline after a delay. This “delay” increases with increasing LPIP, and is owing to a nonzero pipet cone angle (0) coupled with a finite tip radius. However, the time course of solution change becomes approximately exponential once the concentration has reached 50% of its final value. Figure lOA,B shows the effect of varying LPIp, 0, and RTIP on the time taken for 50 and 90% of the solution change to occur, respectively (for D = 1O-5 cm2/s), for pipet dimensions that are likely to be obtained


1oc

1c

2clm 1 0.1 1 3

Tip Radius &m)

A

3000

1000

1 8 J-TlP

3 1oclm “a 100

!I

ii0

2 P) 10

8

2 0:1 i 3

Tip Radius (pm)

B

Fig. 10. The effect of L,,,,,, I&, and 0 on diffusion time. (A,B) The effect of varying LpIp, 8, and q,, on the time taken for 50% (A) and 90% (B) of solution change to occur (for D = RI5 cmz/s, similar to the diffusion coefficient of calcium).

with standard micropipet pullers from borosilicate glass (Sakmann and Neher, 1983). It is apparent that the time taken for the solution change to occur is highly dependent on the geometry of the pipet, and for realistic ranges of L,, and R,, can vary by more than two orders of magnitude. It should be noted that I.,,, may be larger than 10 pm in some experiments (Sokabe and Sachs, 1990), which will result in solution changes taking several seconds to attain 90% completion. However, the time taken for the solution change will be minimized by using pipets with large R,,, and small 8.

5.3. Experimental Measurement of Diffusion Delays

Figure 11A illustrates the time course of change in current through ATP-sensitive potassium channels when the


I- ([K+] - 4?*

7x103

0 20 40 60 80 100 120 140 [K+l mM

_ _ _-_--e-e e-m--.--- ,

0 200 400

TIME mm2

Fig. 11. The time course of diffusion mto the pipet tip. (A) The time course of change in current through ATP-sensitive potassium channels (m zero ATP) when the electrode was moved from 140 mM K+ to 4 mM K+ (Na+-substituted), through an oil-gate (oil). (B) Measured steady-state [K+]-dependence of patch current (I- normalized to maximum current observed in 140 mM [K+]). Squares show means of 3-4 experiments. Solid line is fit to empirical relationship shown.


electrode was moved from a solution containing 140 mM K+ to a solution containing 4 mM K+ (Na+-substituted), through an “oil-gate.” On entering the oil, patch current falls to zero because of the high resistance of the oil. The emergence of the electrode tip into the 4 mM K+ solution results in an electrical artifact that provides a time-mark for exit from the oil. The patch current decreases owing to the change in electrochemical driving force but decreases approximately exponentially and does not instantaneously attain the final level. Similarly, on switching back to high K+ solution, patch current increases approximately exponentially to the final level (not shown). These changes in current reflect the time course of change in the electrochemical driving force across the membrane patch (Em - EK) and the [K+]-dependence of single-channel conductance. Knowing the dependence of channel current on [I(+] (Fig. 11B) permits the time course of change of [K’] to be inferred from the time course of change of patch current. Fig- ure 11C shows the calculated [K+] at the patch as a function of time. This diffusion-limited change in patch current was then simulated with the model described earlier. The pipet tip radius (R& and cone angle (6) were measured optically with a microscope and eyepiece graticule after the experiment. With the constraints imposed by measurement of R,, and 0, the distance between the pipet tip and the membrane was the only free parameter. As shown by the smooth curve in Fig. llC, the model accurately reproduced the observed time course of change in K+, assuming L,, = 9.8 pm. In similar experiments, the best estimate of L, ranged between 3.2 and 23.0 Frn, in good agreement with the range of values measured optically (Sakmann and Neher, 1983; Sokabe and Sachs, 1990). Refer-

Fig. 11. (contznued) (C) Estimated [K’] at the patch (dots), as a functron of time using empirical relationship derived in (B). Pipet tip radius (R,,,) and cone angle (8) were measured optically with a mrcroscope and eyepiece graticule after the experiment. The distance between the pipet tip and the membrane was the only free parameter and was varied in the model (Fig. 9). The model (smooth curve) accurately reproduced the observed time course of change in [K’] (dots), when L,,, was set to 9.8 urn.


ring back to Fig. 10 shows just how large a delay would then be expected in diffusion times of ligands (tens to thousands of milliseconds).

5.4. Correcting for Diffusion Delays in Analysis of Concentration Jump Experiments

Diffusion delays of much longer than 1 ms will occur for most substances, even when the patch of membrane is formed only a few microns from the pipet tip. A further complication is the fact that the solution to the diffusion equation is not a simple exponential (except at late times) for typical pipet geometries. However, since the time course of solution change at the membrane can be estimated from jumps of permeant ion concentrations, the time course of concentration change of any other modulator of channel activity can be predicted by multiplying the time axis of observed activity changes by the factor Dpermeant JD,,,odulator (since the rate of diffusion is directly proportional to the diffusion coefficient).

Since the actual measured time course of change in channel activity in response to a step change of a modulator at the tip of the pipet arises from the convolution of the kinetics of modulator-channel interaction with the time course of the change in modulator concentration at the membrane (which can be calculated as shown earlier), it is possible to deconvolve the observed channel response to modulators and derive the kinetics of modulator-channel interaction. Given a model for the interaction of a channel with a diffusing modulator, it is straightforward to add the relevant equations to those describing the diffusion of the modulator and obtain the time course of channel response. The rate constants describing the interaction of the channel with the modulator can then be altered to minimize the difference between the observed and simulated channel response and thereby obtain best estimates of the model rate constants (Cannel1 and Nichols, 1991). By simulating the results obtained with a limited number of channels, it is clear that good estimates of rate constants for channel modulators can be obtained until


the modulator off-rate constant becomes orders of magnitude faster than the diffusion rate. As might be expected, an increase in the number of channel records averaged results in a reduction in the error of the estimate of the unblock rate constant (Cannel1 and Nichols, 1991).

5.5. Advantages and Disadvantages of the AnaQsis

It is clear that diffusion limitations will seriously interfere with simple measurements of channel responses to step changes of (modulator) in inside-out patch-clamp experiments, although in outside-out configurations, the membrane does not appear to be recessed into the electrode, and diffusion delays should not then be significant. The simulation method described above permits rate constants for the interaction of a modulator with any ion channel to be estimated by deconvolving the observed time course of channel activity and the estimated diffusion-limited time course of [modulator] following a step change of [modulator] at the tip of the pipet. It allows one to obtain the microscopic rate constants from macroscopic (i.e., multichannel) data. In principle, one could measure these rate constants from analysis of single channel open-closed times, but, as discussed earlier, one rarely has sufficient single-channel data, and for some channels it is very difficult to obtain a single-channel patch. The approach is not as numerically intensive as the method described by Magleby and Weiss (1990), which performs maximum likelihood fits of individual single-channel records. Such a method could also be used to fit kinetic models to changes in modulator concentration, although it is not clear whether the precision of the fitting procedure will be improved by using simulated single-channel records rather than the mass action response of the channel. The rate constants associated with ion-channel activity can be estimated by measuring the activity of single channels and fitting exponentials to dwell time distributions (Colquhoun and Hawkes, 1983). This approach requires quite large numbers of transitions and long periods of steady-state recording. Thus, useful data for interpreting in terms of anything but


the simplest kinetic schemes is unlikely to be obtained from excised patch data. With the simulation method presented here, estimates of rate constants may be obtained in a few seconds.

By analogy with the utility of pulse protocols in voltage clamp experiments, the measurement of the responses of ion channels to rapid change of (modulator) may allow determination of the rate constants for both activation and inactivation In addition, the oil-gate/simulation method may help to circumvent the problems of channel rundown or slow irreversible modification of channel behavior that may obvi- ate long recordings necessary for dwell-time histogram analysis. For use of this approach in practice, the reader is referred to our determination of suitable kinetic schemes and the necessary rate constants for the interaction of KATP channels with ATP (Nichols et al., 1991).

6. Twenty-Eight Hints and Tips for Successful Cell-Free Ion-Channel Recording! We conclude the chapter with a selection of isolated tips

and hints that we consider potentially useful with regard to isolated patch recording.

1.

2.

3.

Inside-out patches can be sufficiently stable to allow “restuffing” of the excised patch back to the cell interior (to see the effect of internal metabolites and second messengers, and so on), the patch surviving the physical interaction with the rest of the cell. This approach has been used successfully with oocytes because of their size. Although rundown is an omnipresent problem with inside-out patch recording, we have also found that certain potassium channels (chimeric DRKl) become inac- tive (rundown), even in cell-attached configuration. It does not appear that the problem is associated with vesicle formation. Maintaining constant solution heights within each channel is important for using an oil-gate chamber. The construction method that we have used, with a common outflow guaranteeing a constant height across all chan-


4.

5.

6. 7.

8.

9.

10.

nels, is probably a useful feature. An electronic feedback control of solution flow (Cannel1 and Lederer, 1986), which is fairly simply constructed, is able to control the solution level accurately. We have also found that a constant vacuum outflow to control the solution level works well, so long as the siphon (say a 22-gage needle) diameter is sufficiently narrow and the speed of air flow is much greater than the required speed of solution flow. Reducing the electrical noise in patch pipets has been achieved by painting the tip with Sylgard. We have found that similar noise reduction can be very simply achieved using a 1:l parafilm:mineral oil mixture kept as a liquid emulsion in an oven at 80-100°C. Immediately before filling, the electrode is dipped into the hot solution, with constant application of positive air pressure to the rear end using a hand held syringe. The positive pressure is maintained for several seconds after dipping, as the parafilm:mineral oil mixture dries on the pipet. The pipet can then be filled by reversing the pressure and sucking solution up into the tip. Very small electrodes (>lO MSJ) form seals more easily than large electrodes but are also more likely to form a vesicle, which is hard to break. Patches in large electrodes (~2 MQ) break easily. Do not use electrodes prepared yesterday. “Fresh” electrodes give seals more easily than old ones. High divalent ion concentrations in the electrode can greatly improve the ease with which seals can be formed (e.g., 100 mM Ba*+ in cardiac cells), but the stability of such seals may be low. Always use gentle suction when forming a seal to prevent the membrane being drawn up too far into the electrode, breaking the patch seal, or causing a premature break-in to the whole-cell configuration. Always maintain positive pressure on the electrode before putting the electrode into the bathing solution, and, if possible, maintain a slight positive pressure until the cell is actually touched by the electrode.


11. Try to allow the electrode resistance to increase by at least 30% before switching to negative pressure. This will greatly improve the chance of seal formation.

12. We have found that certain types of borosilicate and aluminosilicate glass capillaries seem to work better on some cell types than others, and there is no magic glass guaranteed to work. If a given type of glass does not form seals, try another.

13. Do not try to use Cl- free solutions inside the electrode, because the junction potential will become very unstable.

14. When firepolishing electrodes, make sure that the polishing filament cannot evaporate onto the glass. This may be avoidable by premelting a small quantity of glass onto the polishing filament, near the point of approach of the electrode.

15. To fill electrodes, take a 1-mL polyethylene tuberculin syringe and draw to a fine point. To draw the syringe, hold about 4 cm above a 2-cm Bunsen burner flame, rotating the syringe until it becomes transparent, and then pull. With a little practice, such syringes can be pulled routinely to a thin diameter to allow filling of the electrode from the shoulder. Use of plastic (rather than, say, a 26-gage needle for backfilling) is essential to completely avoid heavy metal contamination of the filling solution, since stainless steel filling needles can leach Cr, Ni, Fe, and so on.

16. For single-channel recording, routinely check the noise in the system by burying the tip of the electrode into a Sylgard ball placed in the bath. Excess noise (>250 fA) often arises from filling solution having tracked up inside the pipet holder, and in such cases it will be necessary to disassemble and wash the components thoroughly in deionized water.

17. Avoid exposing Perspex (acrylate) holders to ethanol, because machining stresses will develop into cracks and eventually destroy the holder.

18. The simplest way, in our experience, of chloridizing the electrode silver wire is to dip it into bleach (Clorox).


Cold distiIled water in tiout

T-shape plastic connectors

Insulator Metal needles w

15 cm

Fig. 12. A very simple device for cooling the solutions flowmg to the chamber. Cooling may be useful if trying to avoid the problem of rundown.

19. The next simplest way is to connect the electrode wire to the anode of a 1.5 V battery and dip the electrode into 20-100 mM HCl, with an additional silver wire connected to the cathode and also immersed to complete the circuit. Alternating the polarity while chloridizing does not seem to help.

20. Pipet firepolishing is not necessary in most cases. We have generally found that seals form as easily with unpolished electrodes as with firepolished pipets whereas stability of the inside-out patches is unchanged.

21. In some cases lowering bath temperature may be useful to slow down channel rundown. It is not necessary to rebuild the chamber; a very simple cooling element can be constructed from a couple of plastic tubing T-pieces, some metal needles, and Parafilm (Fig. 12). Such a device can reduce bath temperature up to 10°C (real temperature depends on chamber volume, speed of flow, and so forth) when iced water is used as a coolant. Metal needles are good heat transducers but also make an electrical connection between bath solution and coolant, so, in

118 IYichols, Cannell, and Lopa tin

order to avoid electrical noise coming from coolant, one has to use distilled or deionized water (or other coolant with high electrical resistance). The speed of coolant flow should be high enough to keep temperature within the cooler as low as possible. Insulating material (e.g., tightly wrapped Parafilm) around the cooler may be useful.

22. When working with relatively small cells, such as lym- phocytes, cardiocytes, or neurons, lifting the pipet often detaches these cells from the bottom of the chamber, preventing formation of an inside-out patch. To avoid this detachment, try (a) washing the chamber bottom glass thoroughly with soap, or (b) cleaning the solution containing cells of any debris or contaminant particles before putting into the chamber.

23. We have found that when working with giant patches, that less conical pipet tips give seals more easily. In this case, however, care should be taken about series resistance compensation when currents are large (or channels density is high).

24. With some cells, seal formation may be facilitated by holding the pipet potential at negative or positive values during suction.

25. Do not forget to release negative pressure in the pipet after seal formation since some cells (a good example is Xenopus oocytes) have stretch-activated channels, that will be continuously open, simulating leakage current, while the gigaseal may already be formed.

26. Usually, giant inside-out patches do not “survive.” 27. Not only internal Ca*+, but also Mg*+, may cause chan-

nel rundown in inside-out patches. 28. If bath and pipet solutions are different, then along with

a junction potential, an additional problem can be sucking bath solution into the pipet during seal formation. This problem is best avoided by following point 10.

7. Concluding Remarks

In this chapter, we have attempted to consider practical and statistical approaches to solving problems that are

Cell-Free Ion-Channel Recording

peculiar to inside-out patch-clamp analyses, namely vesicle formation, channel rundown, and the kinetic limitations of the recessed membrane patch. In many places, particularly the final section, which might be the most useful, we have included descriptions of our own personal experiences, with which other practitioners may beg to differ! We can only suggest that the interested reader try these approaches if con- founded by the same problems. We welcome suggestions for improvement on the described approaches, or evidence con- flicting with our claims. We will be happy to provide any readers with our original publications and more details on methods, if possible.

References Akaike N., Inoue M. and Krishtal 0. A. (1986) Concentration-clamp study

of y-aminobutyric acid induced chloride current kinetics in frog sensory neurones. I. PhysioI. (Land.) 379,171-185.

Ashcroft F. M. (1988) Adenosme 5’-triphosphate-sensitive potassium channels. Ann. Rev. Neuroscz. 11,97-118.

Brett R. S., Dilger J. I’., Adams P. R., and Lancaster B. (1986) A method for the rapid exchange of solutions bathing excised membrane patches. Biophys. I. 50,987-992.

Cannel1 M. B. and Lederer W. J. (1986) An experimental chamber for single-cell voltage-clamp and patch-clamp experiments with temperature and flow control and low electrical noise. PfliQers Arch. 406,536-539.

Cannel1 M. B. and Nichols C. G. (1991) The effects of pipette geometry on the time course of solution change in patch-clamp experiments. Biophys. I. 60, H1156-H1163.

Colquhoun D. and Hawkes A. G. (1983) The principles of the stochastic interpretation of ion-channel mechanisms, in Single Channel Record- ing. (Sakmann B. and Neher E., eds.) Plenum, New York, pp. 135-175.

Dilger J. P. and Liu Y. (1992) Desensitization of acetylcholine receptors in BC3H-1 cells. P’iigers Arch. 420,479-485.

Findlay I. (1987) AT&sensitive K+ channels in rat ventricular myocytes are blocked and inactivated by internal divalent cations. Pftigers Arch. 410,313-320.

Franke C. H., Hatt H., and Dude1 J. (1987) Liquid filament switch for ultra-fast exchanges of solutions at excised patches of synaptic membrane of crayfish muscle. Neurosci. Left. 77, 199-204.

Fukuda J., Kameyama M., and Yamaguchi K. (1981) Breakdown of cytoskeletal filaments selectively reduces Na and Ca spikes in cultured neurones. Nature (Land.) 294,82-85.

120 Mchols, Cannell, and Lopatin

Hilgemman D. W. (1989) Giant excised cardiac sarcolemmal membrane patches: sodium and sodium-calcium exchange currents. P’zigers Arch. 4X5,247-249.

Horn R. and Marty A. (1989) Muscarinic activation of ionic currents measured by a new whole-cell recording method. 1. Gen. Physiol. 92,145-159.

Horn R. and Patlak J. (1980) Single channel currents from excised patches of muscle membrane. Proc. Nutl. Acad. Sci. USA 77,6930-6934.

Johnson B. D. and Byerly L. (1993) A cytoskeletal mechanism for Ca*+ channel metabolic dependence and inactivation by intracellular Ca*+. Neuron 10,797-804.

Kakei M. and Ashcroft F. M. (1987) A microflow superfusion system for use with excised membrane patches. Pflugers Arch. 409,337-341.

Koh D.-S. and Vogel W. (1993) A method for rapid exchange of solutions at membrane patches using a 10 ul microcapsule. Pfliqers Arch. 422,609-613.

Konnerth A., Lux H. D., and Morad M. (1987) Proton-induced transformation of calcium channels in chick dorsal root ganglion cells. J. Physiol. (Land.) 386,603-633.

Lederer W. J. and Nichols C. G. (1989) Nucleotide modulation of the activity of rat heart K,, channels in membrane patches. J. Physiol. (Land.) 419,193-211.

Magleby K. L. and Weiss D. S. (1990) Estimating kinetic parameters for single channels with simulation: a general method that resolves the missed event problem and accounts for noise. Bzophys. J. 58, 1411-1425.

Matsumoto G. and Sakai H. (1979a) Microtubules inside the plasma membrane of squid giant axons and their possible physiological function. J. Membrane Biol. 50,1-14.

Matsumoto G. and Sakai H. (197913) Restoration of membrane excitability of squid giant axons by reagents activating tyrosine-tubuline ligase. J Membrane Biol. 50,15-22.

Matsumoto G., Ichikawa M., Tasaki A., Mirofushi H., and Sakai H. (1984a) Axonal microtubules necessary for generation of sodium current in squid giant axons: I. pharmacological study on sodium current and restoration of sodium current by microtubule proteins and 260K protein. J. Membrane Biol. 77,77-91.

Matsumoto G., Ichikawa M., and Tasaki A. (1984b) Axonal microtubules necessary for generation of sodium current in squid giant axons: II. effect of colchicine upon asymmetrical displacement current. 1. Membrane Biol. 77,93-99.

Mery P.-F., Lechene P., and Fischmeister R. (1992) A loudspeaker-driven system for rapid and multiple solution exchanges in patch-clamp experiments. Pfliigers Arch. 420,529-535.

Milton R. L. and Caldwell J. H. (1990) How do patch-clamp seals form? Pftigers Arch. 416,758-765.


Nichols C. G., Lederer W. J., and Cannel1 M. B. (1991) The ATP-dependence of K,, channel kinetics in isolated membrane patches from rat ventricle. Biophys. J. 60,1164-1177.

Nichols C. G. and Lopatin A. N. (1993) Trypsin and a-chymotrypsin treatment abolishes glibenclamide sensitivity of K,, channels in rat ventricular myocytes. P’ugers Arch. 422,617-619.

Ono K. and Fozzard H. A. (1992) Phosphorylation restores activity of L-type calcium channels after rundown in inside-out patches from rabbit cardiac cells. J. PhysioI. (Lond.) 454,673-688.

Proks P. and Ashcroft F. M. (1993) Modification of K-ATP channels in pancreatic beta-cells by trypsin. Pjiigers Arch. 424,63-72.

Qin D. and Noma A. (1988) A new oil-gate concentration jump technique applied to inside-out patch-clamp recording. Am. J. Physiol. 255, H980-H984.

Rosenmund C. and Westbrook G. L. (1993) Calcium-induced actin depolymerization reduces NMDA channel activity. Neuron 10, 805-814.

Sakmann B. and Neher E. (1983) Geometric parameters of pipettes and membrane patches, in Single Channel Recording. (Sakmann B. and Neher E., eds.) Plenum, New York.

Sigworth F. J. (1980) The variance of sodium current fluctuations at the node of Ranvier. J. PhysioI. (Lo&) 307,97-129.

Sokabe M. and Sachs F. (1990) The structure and dynamics of patch- clamped membranes: a study using differential interference contrast light microscopy. J. Cell Biol. 111,599-606.

Takano M., Qin D., and Noma A. (1990) ATP-dependent decay and recovery of K+ channels in guinea-pig cardiac myocytes. Am. J. Physiol. 258, H45-H50.

Yellen G. (1982) Single CaZ+- activated nonselective cation channels in neuroblastoma. Nature (Lo&.) 296,357-359.

Perfusion of Patch Pipets

John M. Tang, E N. GZuandt, and R. S. Eisenberg

1. Introduction

The patch-clamp technique allows the measurement of current through a wide variety of channels under reasonably realistic conditions, while controlling (“voltage clamping”) one component of the driving force for current, the electrical potential. The other component of the driving force is set by the concentrations of permeant ions on both sides of the membrane, and those need to be controlled as well if the function and mechanism of channels are to be studied. In natural biological settings, current through channels is determined as much by chemical messengers, metabolites, modulators, and drugs as by driving force, and these must be applied to one side of the membrane or another if their action is be understood.

In the patch-clamp, solutions on one side of the membrane easily can be changed because that side is an easily accessible bath. However, solutions on the other side of the membrane are difficult to change. They are in the patch pipet and must be isolated by many gigaohms of resistance from (and coupled by very little capacitance to) surrounding solutions and earth if the voltage clamp is to function at all, let alone with reasonably low noise. Changing solution in a compartment isolated by gigaohms and picofarads is not easy, particularly if the solution-changing apparatus is not to interfere with the other necessities of experimentation. The apparatus must not add too much complexity, inconvenience,

From. Neuromethods, Vol. 26: Patch-C/amp Applicsbons and Protocols Eds: A. Boulton, G. Baker, and W. Walz Q 1995 Humana Press Inc.

123

124 Tang, Quandt, and Eisenberg

or cost to the setup. In particular, it must not interfere with the easy changing of pipets. We have developed an apparatus and procedure to change solutions in the pipet that adds little complexity, cost, or noise to the setup for recording single channel currents (Tang et al., 1990, 1992). Here, we modify the apparatus to allow control of the pipet solution during voltage clamp of the whole cell.

The control of pipet solutions permits many kinds of experiments. For example, in nerve cells many types of ionic currents must be identified and separated because multiple voltage-gated channels are activated by depolarization: The current measured in whole cell experiments is the sum of current from many types of channels and of the nonlinear capacity current, called gating current. Isolation of gating current or any one component of membrane current requires the removal of all other components. A pharmacological agent can be applied to selectively remove one type of channel or permeant ions can be removed and replaced by impermeants. For example, substitution of K+ with Cs+ eliminates many K+ currents because many K+ channels are impermeable to Cs’. Often impermeants or blockers must be applied to the internal surface of the channel; e.g., in squid axon, tetraethylammonium ion blocks K+ channels when applied to the internal, but not external side of the membrane (Armstrong and Binstock 1965). Our perfusion apparatus is useful in these cases.

Perfusion of the pipet also helps in studies of the selectivity of ion channels. The permeability ratio of the channel for two ions is usually estimated from measurements of the reversal potential (the potential at which zero current flows through the open channel), if the concentration gradient of the ions is known. Perfusion allows control of the concentration gradient.

2. Methods 2.1. Patch-Clamp of Neuroblastoma Cells

Neuroblastoma cells are grown in tissue culture and are differentiated prior to use in electrophysiological experi-

Perfusion of Patch Pipets 125

ments. Following differentiation, the cells have large voltage-gated K currents. Details of the culture conditions have been published previously (Quandt, 1994). Cells are grown in Dulbecco’s modified Eagle’s medium (DMEM) with 5% fetal bovine serum. To induce differentiation, cells are grown for at least 3 d in DMEM with reduced serum (2.5%) and 1.5% dimethylsulfoxide.

Techniques used to patch-clamp neuroblastoma cells in the whole cell configuration are similar to those published for other preparations (Hamill et al., 1981). Patch pipets have an opening at the tip of 2-3 pm. Typically the seal resistance is >lO Go.

Cells were typically bathed in a normal saline composed of (in mM): 125 NaCl, 5.5 KCl, 3.0 CaCl,, 0.8 M&l,, 25 N- 2-hydroxyethylpiperazine-n’-2-ethanesulfonic acid (HEPES), 25 dextrose. K internal solution consisted of (in mM): 150 KCl, 1 NaHEPES, 5 HEPES, 5 ethylene glycol-bis p-aminoethylether N,N,N’,N-tetra-acetic acid (EGTA). The pH was adjusted to 7.25 with the addition of KOH. Cs internal solution was identical to the K+ internal solution, except 150 n-&I CsCl replaced the KCl, and the pH was adjusted with CsOH. The internal solutions were filtered to minimize clogging of the perfusion system with particulate matter. Experiments were performed at room temperature.

2.2. Internal Perfusion Technology

Internal perfusion requires modifications in the standard patch-clamp apparatus. The electrode holder in the patch- clamp apparatus was modified to make separate ports for inflow and outflow in addition to the usual port for suction. The configuration of the perfusion setup is shown in Fig. 1A. Capsules are used to hold the perfusion solution. For the inflow, one end of a short length (8-10 cm) of Tygon tubing (PElO) is connected to the quartz perfusion capillary within the patch pipet. The tubing leaves the electrode holder through a gasket and pressure fitting and its other end is placed in any one of the several capsules at hand.

Tang, Quandt, and Eisenberg 126

quartz cap&il

patch

b

edle

g h

capsules

resewor

B

Fig. 1. Configuration of the pipet perfusion system for whole-cell patch-clamp. (A) Diagram of the pipet perfusion configuration. Com- ponents are: a, patch pipet; b, cell; c, electrode holder; d, silver wire; e, outflow for suction used to obtain a gigaohm seal; f, outflow reservoir g, h, capsules for the inflow solution; i, vacuum line to pressure generator; j, perfusate outflow line; k, perfusion inflow he. See text for explanation. (B) The patch pipet and the quartz perfusion capillary are shown retouched to outline the perfusron caprllary. The marker is 50 pm.


Internal perfusion is much faster if the pipet has a steep final taper to its opening, and so the shape of the pipet must be carefully controlled in perfusion experiments, even though it is not very important in standard setups. We use a Flam- ing/Brown programmable micropipet puller (model P-SO/ PC, Sutter Instrument Co., San Rafael, CA) to pull the pipet in multiple steps, with progressively increasing heat. It is rather difficult to pull 3-pm steeply tapered pipets: The glass is liable to break unevenly on the final pull, leaving an opening that is rather large and jagged. Pipets are firepolished to reduce the final opening of the tip to about 3 pm. A typical patch pipet used in this study is shown in Fig. 1B.

A second piece of tubing is used for the outflow of the perfusion solution. This tubing (PE50) is connected to a 27- 30-gage needle placed near the back of the pipet. The tubing exits the electrode holder and connects to a reservoir, which collects the perfusate. A vacuum is applied to a port on the reservoir to suck solution from the capsule through the perfusion capillary, then into the patch pipet, out of the pipet, and finally into the reservoir.

Current is collected by a Ag-AgCl, wire and led out of the holder through the standard connector. Suction is applied to a separate port on the electrode holder to produce gigaohm seals between the glass and cell membrane, or to rupture the membrane under the pipet sealed to its tip.

The quartz perfusion capillary (Polymicro Technologies, Phoenix, AZ) is made as described in Tang et al. (1990,1992). Briefly, quartz tubing is softened by heat and then drawn out. The capillary is cleaned of debris and the drawn out tubing is cut (near its tip) to the desired opening diameter. The position of the capillary within the pipet is critical. If positioned as shown in Fig. lB, close to the pipet tip, reasonably rapid exchange of solutions is possible.

Our perfusion procedure starts by flushing the inflow line-the Tygon tubing connected to the quartz perfusion capillary-with the standard (K) internal solution to remove air bubbles and solution left from the previous experiment. The patch pipet is then partially filled with control solution


and mounted in the electrode holder. The quartz perfusion capill%y is next positioned close to the tip (of the patch pipet) under a stereomicroscope and the suction and outflow lines are connected. The inflow line is inserted into the capsule containing the standard K+ internal solution. A valve is used to seal the suction line from atmospheric pressure and a vacuum (typically 60 mm Hg)” is applied to the reservoir on the outflow side to initiate the flow of solution.** The perfusion is then stopped by closing off the line connecting the vacuum generator to the reservoir and the suction line (e in Fig. 1) is opened. A gigaseal can then be made between the pipet and cell membrane, using the standard suction procedure of patch-clamp experiments. At this stage the membrane under the pipet must be broken to allow “whole cell” recording. We increased the suction (or applied a large voltage) to break down the membrane and gain diffusion access to the cell interior. Perfusion can be restarted at any time by closing the suction line e in the Figure and applying a vacuum to the outflow line (i in the Figure). To change the perfusion solution, the vacuum is turned off, the inflow line is carefully moved to the new capsule containing the selected solution, and vacuum is reapplied, monitored, and adjusted if it drifts, presumably because of leakage.

It is important to initiate perfusion with a control solution and to switch to a test solution only after access is gained to the cell interior, so there is a clear start-time of perfusion. Significant flow occurs while establishing a gigaseal, or breaking down the membrane. Flow can also occur in the absence of suction owing to capillary action. Because of this flow the state of the cell and its channels are easier to interpret if control solution is in the perfusion capillary during those pre- parative procedures.

*Generated by a regulated source, such as the Bio-Tek pneumatic transducer, model DPM-1B (Bio-Tek Instruments, Winooski, VT).

**If the meniscus in the patch pipet is seen to rise, fluid is flowmg and perfusron has been establrshed. If the meniscus does not rise, the chances are a perfusion line is blocked or there is a leak somewhere in the vacuum system.


3. Results

3.1. Time Course of Exchange of Internal Solution

The effectiveness of internal perfusion-in particular, the time course of exchange of solutions-can be evaluated by replacing internal K+ with Cs+, an ion that does not permeate many voltage-gated K+ channels. Substitution should completely eliminate outward K current through these channels. Results from a typical experiment are shown in Fig. 2. Every 6 s, the potential was stepped to a more positive value, and the current resulting from this depolarization was recorded. The maximum amplitude (normalized and shown as a fraction of its maximum value) is shown in Fig. 2A. The filled circles show the change of the maximum current (resulting from a 70-mV depolarization) after the solution perfusing the pipet was changed from K+ internal solution to Cs internal solution at time zero. Following a latency of about 150 s, the current rapidly declined, and reached a much lower steady state (note that the baseline toward the right of the figure is made of filled circles). Figure 2A (filled triangles) also shows the effect of reversal, of changing Cs back to K (note that the baseline toward the left of the figure is made of filled triangles). The maximum current was restored to its original value, following a latency similar to that for the onset, showing that the change of current is produced by the switch in ions, not the perfusion itself or some other artifact.

3.2. Efficiency of Exchange of Internal Solution

It is not clear that all the current in the pipet and cell can be changed by perfusion: A residual K current can often be measured following substitution of K+ with Cs+. Although the measured outward current is normally dominated by flux through K+ channels, in the blocked situation other components, e.g., nonlinear leakage, may show themselves. Residual current through K+ channels (following substitution of K+ with Cs+) would be expected to be blocked by external tetraethylammonium (TEA) (Quandt and Im, 1992). Figure 3 shows the effects of Cs+ plus TEA (external). A current volt-


A -I

IOO- -w#992-**. -a--- K to Cs

! -A--CstoKe

z 80- \ cu . JR

2 60- I rA

E

E 40-

it

\, i

3 20- 0

B

0 1 I I I I I I I

0 100 200 300 400 500

Time (set)

& t 20 ms

Fig. 2. Time course of exchange of the internal solution. (A) The amplitude of outward current in response to a depolarizatron to 70 mV is plotted, normalized to the maximum. The circles plot the current at 6-s intervals, followmg a change m the pipet perfusron solution from K internal to Cs internal solution at zero time. The trrangles plot the restoration of current following a change back to K mternal solution at zero time. The record is continuous: Zero tune for the triangles begins immediately after the last time sample given by the circles. Note that the reduction in the outward current is reversible following the reintroduction of K. (B) The membrane current traces are shown superimposed and recorded during the period plotted by the circles in (A). The onset of the depolarization is marked by the arrow. The time (following the change from K to Cs internal solution) is given to the right of each trace.

age curve was first measured with K internal solution in the pipet. The residual outward current was next measured. Finally, 20 mM TEA was added to the solution perfusing the


35

l K 30 A cs

T Cs + TEA-... 25

,

00 ..AAAAA A A I v v t v . .

v . .

7 I I I I I I -20 0 20 40 60 80

Membrane Potential (mV)

Fig. 3. Measurement of residual current through K channels following substitutio+n of K with Cs. The current-voltage curve was measured with K in the pipet (after substitution of Cs for K by perfusion) and during Cs substitution with 20 m.M TEA added to the external solution, TEA was found to block a small amount of outward current after perfusion with Cs solution, revealing residual current through the K channel.

outside of the cell and the current-voltage curve was measured again. No significant current flowed in the doubly blocked preparation and one can thus conclude that current through K channels was reduced by 95% following perfusion of the pipet and cell interior by Cs.

3.3. Selectivity of K Channels Measured by Reversal Potentials

The residual current (through K+ channels after exchange of K+ with Cs+) might result from Cs flowing through the K+ channels. The relative permeability of Cs+ and K+ in blocked channels an be gaged by the shift in the reversal potential for K channel “tail current” following a change in the K+ concentration. A typical experiment is shown in Fig. 4. The


n 150mMK

-04 ! I t 1 I -120 -100 -60 -60

Membrane Potential (mV)

Fig. 4. Measurement of the reversal potential of tail current following Cs replacement. The membrane was depolarized to 60 mV to activate the K channels and repolarized to various potentials to measure the reversal potential. The maximum amplitude of this tail current, measured immediately after repolarization, is plotted. The tail currents were measured first during patch pipet perfusion with K internal solution, and second during perfusion with 50% K and 50% Cs internal solution. Note that the reversal potential shifted to a more depolarized potential in the solution containmg Cs. The K inward current was less than that predicted in the absence of rectification (dashed line).

membrane was depolarized to activate K channels, and then repolarized to a variety of potentials (Vj). The amplitude and polarity of the tail current is a function of the potential Vj to which the membrane is repolarized. The initial amplitude of the tail current (immediately following the repolarization to Vj) is plotted in the figure as a function of Vj. This relationship was measured during perfusion of K+ internal solution (filled squares) and following a change to 50% K+ and 50% Cs internal solution (filled circles). The reversal potential (Vj) at which current is zero) was found to be -78 mV for K+ internal solution and -64 mV for the mixed K+/Cs+ solution:


The reversal potential shifted 14 mV to a more depolarized potential following the change to the solution with reduced K+. The amplitude of this shift is the same as that which is predicted by the Nernst equation for a channel permeable only to K+, suggesting that the channel is impermeable to Cs+. Although Cs+ permeability was not measurable, it may be greater at a large depolarization (see Section 4.).

There may be a fraction of the internal K+ that is not subject to exchange with the patch pipet, for example, owing to an unstirred layer. This unexchanged fraction could produce the residual current following perfusion with Cs+. We do not have any evidence that this is the case. In the experiment described, if the K+ concentration was greater than that in the pipet after perfusion with reduced K+, the magnitude of the shift would have been smaller.

The current-voltage curve for the tail currents of Fig. 4 shows rectification. The outward current is larger than inward current for an equivalent driving force (the absolute value of the difference between the membrane and reversal potential). The dashed line in the figure gives the relationship for the control K+ solution assuming no outward rectification, plotted using linear extrapolation from the potentials exhibiting outward current. The magnitude of the reduction in inward current increased as the membrane was hyperpo- larized. Although not investigated in the present study, the rectification probably arises from a voltage-dependent block by external divalent cations, including Ca, as seen in other monovalent cation channels and preparations (Yamamoto et al., 1985).

3.4. Pharmacology of 4-Aminopgridine

Internal perfusion can be used to apply pharmacological agents--e.g., a K+ channel blocker 4-aminopyridine (4-AP)- to the inside of a cell. Although 4-AP blocks some species of voltage-gated K+ channels from the outside, because it is permeable to the membrane, it may act from the inside. Figure 5 shows a direct test of this idea. The current in control K internal solution is shown as well as the outward current (super-

Tang, Quandt, and Eisenberg

f7-w

2 nA

t /

20 ms

Control

50 pM 4-AP

Fig. 5. Block of K current by internal 4-ammopyridine. Membrane current was measured in response to a depolarization to 70 mV. Super- imposed records were obtained durmg perfusion of the patch pipet with K internal solution and following the addition of 50 u.M 4-aminopyridine to the pipet perfusion solution. The onset of the depolarization is marked by the arrow.

imposed) in the steady state, following perfusion with solution containing 50 lt.M 4-Al?. The 4-AP blocked the current by 50% and reduced the rate of rise of the current, both typical effects of the drug acting on this preparation (Hirsh and Quandt, 1993). 4-AP clearly can block the K+ current in this preparation from the inside.

3.5. Parameters Controlling the Rate of Exchange of Solution

Figure 2A shows a delay, followed by a relatively rapid change in the current. We imagine that the delay is the time to exchange the solution within the lumen of the patch pipet. In this case, the delay should be dependent on the length of external tubing and the rate of perfusion. Indeed, we found


that varying the rate of perfusion (by changing perfusion pressure) varied the delay.

We investigated the effects of parameters that might alter the time course of the rapid rate of exchange. Figure 6A shows the effect of pipet geometry and cell size on the time course of exchange. K+ internal solution was changed to Cs+ internal solution and the peak outward K+ current in response to a depolarization was measured at 6-s intervals. The dead time of the perfusion system, owing to entry of the new solution into the pipet, is not illustrated. Time zero is the time the current first deviates noticeably from zero. The most rapid exchange, illustrated by two experiments (unfilled squares and unfilled triangles) labeled as “optimal,” was obtained under the conditions previously elaborated. A third experiment plotted in the figure (filled circles) compares the effect of perfusion of a large cell. In this case, although the onset of exchange is similar to the optimal conditions, the final rate of exchange is markedly slowed. The cell used in this experiment is shown in Fig. 6B. Note that the cell has a large cell body and long processes. Most of the current is recorded from the cell body and is reduced following exchange with the pipet. However, the long time of exchange is likely to be the time required for diffusion and exchange in the processes. Figure 6 also shows an experiment to examine the effect of the geometry of the pipet (filled triangles). The pipet used in this experiment (shown in Fig. 6C) has a rather gentle final taper, particularly compared to the standard pipet (Fig. lB), and so the perfusion capillary cannot fit far down the patch pipet but must be placed farther from the pipet opening. Perfu- sion does not show a rapid phase under this condition, probably because the solution at the tip is only slowly displaced by the new solution in this situation.

4. Discussion

4.1. Applicability of the Technique The pipet perfusion method used here to control the

internal solution in whole-cell patch-clamp of neuroblastoma cells should be easily applicable to a wide variety of prepa-


o-

Electrode with long taper

-I -100

3 I I I 100 200 300

Time (set)

Fig. 6. Variables affecting the time course of exchange of internal solution. (A) The amplitude K membrane current in response to a depolarization to 70 mV, measured at 6-s intervals, durmg plpet perfusron IS plotted. Experiments on four separate cells are plotted. Following Cs substitution for K in the patch pipet perfuslon solution, the K current declined. To compare the time course of the rapid decay in current, the time at which the current started to decrease was set to zero time. The time course of exchange of K and Cs was slower for large cells, and in the experiment employing a pipet electrode with a long taper, com-


rations; only simple modifications of the existing procedure likely will be required. Although neuroblastoma cells are relatively large (20 pm in diameter), the pipet perfusion should be no less successful for smaller cells, even cells with an irregu- lar or asymmetric geometry, such as cardiac myocytes. The only modification of the recording configuration is the use of an electrode holder to accommodate the inflow and outflow lines and the only special equipment recommended is a regulated vacuum generator. The procedure does add to the pain and duration of recording, because the perfusion capillary must be repositioned and tested each time the patch pipet is changed, but we find the trouble and time (a few minutes) involved to be bearable.

The control of internal solution should aid investigation of the biophysical properties of channels, the pharmacology of internal receptors, and control of membrane phenomena by internal transmitters. The time course of exchange of solution indicates that the effect being studied needs to last some lo-15 min if it is to be recorded faithfully.

4.2. Possible Problems

Occasionally we found that the perfusion would stop even though a constant perfusion pressure was maintained. The solution entering the reservoir can be monitored to determine whether this situation has developed.

In the present studies, the ionic composition of the solution was not varied dramatically. However, it should be noted that the liquid junction potential across the interface between the pipet solution and the Ag2+ wire can change dramatically when the Cl- is replaced. The liquid junction

pared to the optimal condition using small cells and an electrode with a steep taper. (B) The cell marked with a 1 was used in the experiment marked by the filled circles in (A). The cell marked with a 2 is typical of the smaller cells giving an optimal time course. The marker is 100 pm. (C) An electrode with a long taper, such as the one used in the experiment marked by filled triangles, is shown. The marker represents 50 pm.


potential can be measured prior to an experiment to compensate for this potential change under this condition or the setup can be modified to include a KCl/agar type bridge.

We observed a 5% residual outward current through K+ channels following substitution of K+ with Cs+. Two possibilities may explain this observation. Following the substitution there may be a residual concentration of K+ in the cell that cannot be exchanged even in the steady state. However, the shift of the reversal potential we obtained following a change in K+ concentration is not consistent with this idea. Alternatively, Cs+ may flow through the K+ channels. In the latter case, K+ channels must be permeable to Cs+. The measurement of the reversal potential under this condition did not reveal a substantial Cs+ permeability to the membrane. It should be noted, however, that internal Cs+ may flow through the channel with a large depolarization, but not at a membrane potential closer to the reversal potential. Block of current through K+ channels by Cs+ has been observed to escape under some conditions, such as with a large driving force (e.g., Cecchi et al., 1987). This escape could explain the residual current.

We studied the sensitivity to block of the channels to internal 4-Al?. Previous studies have found that single K+ channels are blocked more completely when 50 @I 4-AP is applied to excised inside-out membranes than in the experiments given here. Two possibilities may cause this discrepancy. Multiple types of K+ channels contribute to the whole-cell current. Some of these types of channels may have a lower sensitivity to 4- Al?, A second possibility is that internal 4-AP may diffuse out of the cell so that the steady-state concentration is less than that added to the internal perfusate. A more complete study employing stop flow experiments would be required to distinguish between these possibilities.

4.3. Improvements The configuration we have used for pipet perfusion can

be improved. A positive pressure could be applied to the inflow solution. This negative pressure at the outflow would


then be reduced. The pressure across the cell would then be negligible, so that there would be little tendency for the cell to enter the pipet during perfusion. The disadvantage to this scheme is that the inflow side of the perfusion system then becomes a closed system, making it harder to change solutions. The method used to change perfusate is primitive and mishaps occasionally occur that ruin the experiment. We anticipate that future improvements would increase mechanical stability and reliability without adding too much complexity.

References

Armstrong C. M. and Binstock L. (1965) Anomalous rectification in the squid axon inlected with tetraethylammonmm chloride. I. Gen. Physiol. 48,859-872.

Cecchi X., Woll D., Alvarez O., and Latorre R. (1987) Mechanisms of Cs+ blockade m a Ca2+- activated K+ channel from smooth muscle. Biophys. 1, 52,707-716.

Ham111 0. I’., Marty A., Neher E., Sakmann B., and Slgworth F. J. (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pfltigers Arch. 391,85-100.

Hush J, K. and Quandt F. N. (1993) Ammopyridine block of potassium channels in mouse neuroblastoma cells. J. Pharmacol. Exp. Ther. 267, 604-611.

Quandt F. N. (1994) Recording sodium and potassium currents from neuroblastoma cells, in Methods in Neurosciences, vol. 19 (Narahashi T., ed.), Academic, New York, pp. 3-20.

Quandt F. N. and Im W. B. (1992) Tetraalkylammonium ion block of potassium currents in mouse neuroblastoma cells. 1. Pharmacol. Exp. Ther. 260,1379-1385.

Tang J. M., Wang J., and Eisenberg R. S. (1992) Perfusing patch pipettes. Meth. Enzymol. 207,176-181.

Tang J. M., Wang J., Quandt F. N., and Eisenberg R. S. (1990) Perfusing pipettes. Pfltigers Arch. 416,347-350.

Yamamoto D., Yeh J. Z., and Narahashi T. (1985) Interactions of permeant cations with sodium channels of squid axon membranes. Biophys. J. 48,361-368.

Concentration Clamp Technique

Norio Akaike

1. Introduction

Since the development of the internal perfusion technique (Akaike et al., 1978; Hamill et al., 1981), the opportunity to analyze, in the voltage clamp mode, the kinetics of voltage-gated Na+, K+, and Ca2+ channels of single cells, such as neurons, heart muscle cells, and smooth muscle cells, has been seized in a number of laboratories with striking success. Kinetic studies of the interaction between neurotrans- mitters and the individual receptors are also indispensable for elucidating the underlying molecular mechanisms. Sev- eral complementary approaches have been used. The first approach involves nerve stimulation and relies on the rapid removal of transmitters from the synaptic cleft by diffusion and inactivation processes (Magleby and Stevens, 1972; Kuba and Nishi, 1979; Segal and Barker, 1984). The second method uses a ligand, bis-Q, that has two conformations (cis and trans) with different affinities to the acetylcholine receptor. Rapid changes to the active trans isomer can be made by brief light flashes of appropriate wavelengths (Lester and Chang, 1977; Weinstock, 1983). However, these two methods have the same disadvantage: The time course of the concentration transient is uncertain, and the study is limited to the case of simpler systems, such as the neuromuscular junction and the single ligand trans-bis-Q. A third method has been used for making rapid changes in the external solution perfusing the tissue, ganglia, or isolated cells. The time-scale of the solution

From* Neuromsthods, Vol. 26. Patch-Clamp Appltcations and Protocols Eds: A. Boulton, G. Baker, and W. Walz 0 1995 Humana Press Inc.

141

A kaike

exchange was a few seconds (Akaike et al., 1976; Yellen, 1982), 1 s (Slater and Carpenter, 1984) and 20-100 ms (Krishtal and Pidoplichko, 1980; Fenwick et al., 1982). But even these techniques were too slow to study receptor-mediated ionic currents that can activate within a few milliseconds and greatly desensitize within a few hundred milliseconds. The rapid application of external solution allows kinetic analysis of the interaction between drug and neurotransmitter receptors. This technique is especially important for studying receptor-mediated currents that activate within tens of milliseconds and desensitize within seconds.

Our laboratory has developed a rapid concentration jump technique (termed “concentration clamp” technique) that combines the solution change technique (Krishtal et al., 1983) with intracellular perfusion either in the whole-cell (Akaike et al., 1978) or the excised membrane mode (Hamill et al., 1981). When this concentration clamp technique (Akaike et al., 1986) was applied to various preparations, such as frog sympathetic ganglion cells (Akaike et al., 1989b), frog sensory neurons (Akaike et al., 1986), rat hypothalamic neurons (Akaike and Kaneda, 1989), rat hippocampal neurons (Kaneda et al., 1989), and Aplysia neurons (Ikemoto et al., 1988), the time lag for changing solutions surrounding these cells depended on the cell diameters. However, in the absence of cells, the solution was replaced within 0.1-0.3 ms (Fig. 1). This technique has enabled us to study the kinetics of receptor-mediated ionic currents.

2. Setup of Rapid Solution Change

1. Peripheral ganglia and CNS neurons from slice preparations, heart muscle cells, and smooth muscle cells are obtained by mechanical and/or enzymatic dissociation.

2. Isolated single cells are transferred into a culture dish and drawn into the opening of a patch pipet (1.5-50 pm in diameter depending on the size of cells, which usually varies between 5 and 700 pm) filled with an artificial intracellular solution. The resistance of electrodes


MV

Fig. 1. Schematic illustration of the concentration clamp technique, which combines a rapid change of the external solution with internal perfusion via a patch-suction pipet. SET, solution exchange polyethylene tube; MV, electromagnetic valve; TT, turntable (Akaike et al., 1986).

3.

4.

ranges from 2 KSZ to 5 Ma, depending on the tip diameter of the patch pipet. Negative pressure is applied to puncture the membrane patch, thereby allowing soluble cell contents to exchange with the pipet solution by diffusion (Akaike et al., 1986,1989b). Patch pipets are pulled on a two-stage puller (Narishige, PB-7) and firepolished on a microforge. Transmembrane currents are recorded with a patch clamp amplifier. Both current and voltage are monitored on a storage oscilloscope and simultaneously stored on tape for off-line computer analysis. The test solution is applied rapidly (Fig. 1). The cell- attached tip of the patch pipet is inserted into a polyethylene tube through a circular hole approx 500-1000 pm in diameter. The lower end of the polyethylene tube is submerged in various external solutions contained in Petri dishes and supported on a turntable. Negative pressure (-3 cmHg) is applied to the upper end of the polyethylene tube, which is controlled by an electromagnetic valve driven by 24 VDC and allows the exchange of solution within the tube. The duration of negative pressure is regulated by a pulse generator.

144 A kaike

3. Preparations

3.1. Whole-Cell Recording Mode

3.1.1. Large Cells Aplysia neurons (100-300 pm in diameter) (Ikemoto et

al., 1988; Ikemoto and Akaike, 1988), snail neurons (50-150 pm in diameter) (Ikemoto et al., 1987), frog dorsal root ganglion cells and sympathetic ganglion cells (15-30 pm in diameter) (Akaike et al., 1986; Sadoshima and Akaike, 1991), and mammalian heart muscle cells (Inomata et al., 1989) are dissociated mechanically and/or enzymatically. Single cells sucked into a tapered patch pipet (approximate diameters 500 pm for Aplysia and snail neurons, loo-150 pm for frog ganglion cells and mammalian heart muscle cells) are transferred into a dish 3 cm in diameter. Single cells are clearly visible under binocular magnification of 80x. A whole-cell mode patch recording technique is used for either current or voltage clamp and intracellular perfusion. A Pyrex glass capillary with a 3-mm outer diameter is pulled to a shank length of 2.5-3 mm. The tip of the pipet is cut at an outer diameter of about 120-150 ym for Aplysia neurons, 70 pm for snail neurons, 40 pm for frog ganglion cells, and 15 pm for rat heart muscle cells. The tip is then firepolished to give an inner diameter of about 30,15,4-7, and 2-3 km for Aplysia, snail, and frog neurons, and heart muscle cells, respectively. Part of an individual cell is aspirated through the patch pipet with negative pressure of about 3 cmHg. The aspirated membrane is ruptured spontaneously or can be broken by applying 5- 20 nA squarewave pulses of depolarizing current (lo-50 ms). The resistance between the patch pipet filled with individual standard internal solution and the reference electrode is about 10 KQ for Aplysia neurons, 100 KL2 for snail neurons, 200- 300 KQ for frog ganglion cells, and 1 Ma for rat heart muscle cells. The membrane potential is controlled by a single-electrode voltage-clamp system switching at a frequency of 10 kHz and passing current for 36% of the cycle (Ishizuka et al., 1984). Clamp currents are measured as the voltage drop across a 10 Mln resistor in the feedback path of a headstage

Concentration Clamp Technique 145

amplifier. In this system, the patch pipet electrode can carry time-averaged currents exceeding 100 nA at a switching frequency of 10 kHz without showing signs of polarization or other artifacts.

In the whole-cell recording mode, the time constants of the external solution exchange vary by 10 ms for Aplysia neurons, 3-4 ms for snail neurons, 2-3 ms for frog sensory neurons, l-2 ms for frog sympathetic ganglion cells, and 3 ms for rat heart muscle cells, but 0.1-0.3 ms without the cells. In larger cells, the time constant increases because of the developed connective tissues surrounding the cell and because of many clefts on the cell membrane surface. Fortu- nately, large cells, such as Aplysia neurons, respond very slowly to chemical substances, in the order of a few hundreds of milliseconds. Therefore, kinetic studies using the large neurons of Aplysia and snail can be performed without difficulty.

3.1.2. Small Cells Mammalian CNS neurons (5-30 pm in diameter) are dis-

persed mechanically after enzyme treatment (Kaneda et al., 1988). The isolated neurons are kept in an external solution and are viable for electrophysiological studies up to 18 h after dissection. The neurons are transferred into the experimental chamber of a culture dish and drawn into the opening of a glass patch pipet (about 1.5 pm in diameter) filled with the internal solution. After obtaining a gigaohm seal between the patch pipet tip and the cell membrane, the patched cell membrane is destroyed by negative pressure, and soluble cell contents are exchanged with the pipet-filling solution by diffusion. Patch pipets for small cells are fabricated from glass capillaries (Narishige, 1.5 mm in outer diameter) on a two- stage puller. The resistance between the patch pipet filled with the internal solution and the reference electrode is 2-4 MQ. Ionic currents and voltages are measured with a patch- clamp amplifier (List-electronic, EPC 7) with capacitance and series-resistance compensation, filtered at 2 kHz, digitized at 5 kHz, and analyzed with a computer. The exchange of external solution surrounding the dissociated rat CNS neu-

146 Akaike

rons an be completed within 2 ms for smaller neurons, such as ventromedial hypothalamic and dentate gyrus neurons (5-8 pm in diameter), and 3-4 ms for relatively large neurons, such as spinal motoneuron and cerebellar Purkinje cells (about 30 pm in diameter).

3.2. Excised Cell Membranes

After a gigaohm seal is formed in the cell-attached mode with a patch pipet containing internal solution, a membrane patch is excised in the inside-out configuration. The tip of the patch pipet is inserted into a plastic tube through a hole 0.8-l cm from the end of the tube. In this case, the exchange of solution at the pipet tip can be completed in 0.25-0.5 ms. As preparations, various cells, including dissociated mammalian CNS neurons and smooth muscle cells dissociated from the aortic media of rat (Sadoshima et al., 1988), are used.

4. Kinetic Studies Using Concentration Clamp Technique

4.1. Receptor-Mediated Ionic Currents

4.1.1. Whole-Cell Recording GABA*, strychnine-sensitive glycine, nicotinic acetylcho-

line (ACh), ionotropic glutamate (iGlu), and 5-hydroxytrypt- amine3 (5-HT3) receptors are known as receptor-channel complexes. These receptor-mediated ionic currents have been studied in various neurons freshly dissociated from snail, Aplysia, frog, and rat. Figure 2A shows a typical kinetic study of GABA* receptor-mediated Cl- currents in frog dorsal root ganglion cell (Akaike et al., 1986).

4.1.2. Excised Cell Membrane Recording Figure 2B shows the NMDA-operated microscopic cur-

rents in the outside-out configuration of a dissociated rat nucleus tractus solitarii neuron and the modulatory effect of glycine on the NMDA responses (Shirasaki et al., 1990).


B A outside -out

GABA NMDA 10.4M .’

+Gly5~16-~M - ., .v y v *

----------____ +Gly W6M

c

I , Scorpion toxin 3x10-6 M

10-5

3x10-5

6x10-5

lo-4

2.5 nA

147

I 5 pA

Fig. 2. (A) Cl current (b) elicited by various GABA concentrations at a holding potential (VH) of -10 mV. The peak current increased sigmoi- dally with increasing GABA concentrations, whereas the steady-state ICI reached a maximum around 10-5M GABA (Akaike et al., 1986). (B) Effect of glycine on NMDA-induced microscoprc ionic currents in outside-out configuration of rat nucleus tractus solitarii neuron (Shirasaki et al., 1990). (Cl Modification of the voltage-dependent Na+ current (1~~) by scorpion toxin. Superimposed current tracings were recorded before and 5 and 30 s after the start of toxin application. The dotted line over the current tracings indicates the level of a Vu of -80 mV. The 1~~ was evoked by a depolarizing step to -10 mV. The arrow between the current tracings shows the direction of change in the inactivation phase of INa after the start of toxin treatment (Kaneda et al., 1989).

4.2. Voltage-Dependent Ionic Currents 42.1. Na+ Channel

The kinetic of the effects of tetrodotoxin (TTX), lidocaine, and scorpion toxin on voltage-dependent Na+ channels of dissociated rat hippocampal neurons have been successfully analyzed (Kaneda et al., 1989). TTX and lidocaine concentration-dependently suppressed the Na+ current in a concentration-dependent manner without affecting the current kinetics, but the time courses of inhibition of TTX and lidocaine occurred in the order of minutes and seconds, respectively. The inactivation phase of Na+ current proceeds

148 Akaike

with two exponential components, fast and slow. Scorpion toxin especially increased the time constant of the slow inactivation component (Fig. 2C).

4 2 2 Ca*+ Channel . . . The inhibitory efects of Ca2+ antagonists and lidocaine

on voltage-dependent Ca2+ currents were investigated kinetically in frog sensory neurons (Oyama et al., 1987) and rat hypothalamic neurons (Akaike et al., 1989a). Both D-600 and lidocaine blocked open channels more quickly or more profoundly than closed ones. Figure 3A shows the experimental separation of the current- and voltage-dependent inactivation of Ca2+ current in the frog sensory neuron (Akaike et al., 1988). The results suggest that the inactivation process of Ca2+ current consists of two components, i.e., a dominant component that is dependent on Ca2+ influx and a smaller one that presumably is voltage-dependent.

Figure 3B shows the effect of Ca2+ influx passing through voltage-dependent Ca2+ channels on the GABAA-mediated Cl- current (Inoue et al., 1986). This was the first evidence that an increase in the intracellular Ca2+ concentration sup- presses the GABA response by decreasing the apparent affinity of the GABAA receptor.

4.3. Rapid Change of Physical Conditions

4.3.1. Temperature In the whole-cell recording, the effect of rapid tempera-

ture changes on GABAA receptor-mediated Cl- current was studied in frog sensory neurons (Maruyama et al., 1990; ffrench-Mullen et al., 1988). We have also studied the effect of temperature on the ionic current of the highly temperature-sensitive neurons dissociated from the rat preoptic and anterior hypothalamus (Kiyohara et al., 1990).

4.3.2. Proton (H+) Proton-gated Na+ current was studied in frog sensory

and parasympathetic neurons (Akaike et al., 1990; Kim et al., 1990) and rat hypothalamic neurons (Ueno et al., 1992).


A Ca

a OmV

J,-50mV

-r-----------

b

ki- 5 nA

10Oms

B GABA lo-5 M

100 me

(ms) / 300

0W

Fig. 3. (Aa) Current trace of the CaZ+ current (Ica) elicited by a depolarization step from -50 to 0 mV. The horizontal column above the voltage trace represents the test solutron. (Ab) Switching external test solutions from Ca2+-free solution containing 5 mM Mg2+ to normal CaZ+ solution during a continuous depolarization from -50 to 0 mV. All current traces were corrected by subtracting the leakage current by adding the current responses to equal but opposite voltage steps using a signal averager (Akaike et al., 1988). (BI Suppressing action of Ica on the GABA-activated 1c1. Actual records of 10e5M GABA-gated 1~1, with and without a preceding Ica elicited by the depolarizing pulse from a VH of -50 to 0 mV for various durations (15,25,100, and 300 ms). The leakage and capacitative currents associated with the ionic currents were subtracted during the experiments by adding the current responses to equal but opposite voltage steps using a signal averager. The suppression of GABA-gated Icr reached a plateau with increasing amounts of b (Inoue et al., 1986).

150 A kaike

A ACh

a 0 10 20 30 40 ms

I----

Fig. 4. (A) Current traces induced by ACh at concentrations between 10ms and 10-5M at room temperature (26°C). Note that the activation phase of ACh-induced K+ current (1~) has a latent period, and that this “latency definitely depends on the ACh concentrations (Inomata et al., 1989). (B) Latency between caffeine ap lication and the onset of 1~ (Icarr latency). Current traces of 10m4M-A P h-induced inward current (IAc~) and 10 mM-caffeine-induced outward current (Icar‘) obtained from the same cell were su erimposed. VH was -50 mV. The time when the drug reached the ce 1 was estimated by the onset of Y IACI,. By comparing the onset of IACh (arrow a), and Icaff (arrow b), the Icaff latency was determined (Sadoshima and Akaike, 1991).

4.4. G-Protein Mediated Response

Before the activation of ACh-induced K+ current in isolated guinea-pig atria1 cells, there was a brief latent period after the application of ACh. As shown in Fig. 4A, the latent


period was shortened considerably by an increase in either the ACh concentration or temperature. The results suggest that the latent period of the ACh response is the time lag needed for the activation of K+ channels using the remote sensor, G-protein (Inomata et al., 1989).

4.5. Measurement of Ca2+ Release from fntracellular Ca2+ Store Sites

Vertebrate sympathetic neurons have well-developed Ca2+ store sites located just beneath the cell membrane. Since Ca2+ released from these organs regulates the membrane excitability through either activation or suppression of Ca2+- dependent ionic conductances, we analyzed the activation and inactivation kinetics of K+ current activated by caffeine, which releases Ca2+ from the Ca2+ storage sites (Sadoshima and Akaike, 1991). Figure 4B shows a typical caffeine-induced Ca2+ activated K+ current and a nicotinic ACh receptor- mediated inward current. In this figure, the difference in onset times between a and b indicates the exact latent period until the onset of the caffeine response, reflecting an increase in the intracellular Ca2+ concentration released from Ca2+ storage sites.

5. Limitations

A possible disadvantage of the concentration clamp technique is that intracellular perfusion (via the patch pipet) may wash out cellular metabolites, such as second messengers, which are important for maintaining the channel activities. However, the use of nystatin in the containing pipet solution could prevent such undesirable effects. Thus, the concentration clamp technique offers unequaled opportunities for studying not only drug-operated channels but also proton-, temperature-, and pressure-activated channels.

References Akaike N., Inoue M., and Krishtal 0. A. (1986) “Concentration-clamp”

study of y-aminobutyric-acid-induced chloride current kinetics in frog sensory neurones. J. Physiol. (Land.) 379,171-X45.

Akaike

Akaike N. and Kaneda M. (1989) Glycine-gated chloride current in acutely isolated rat hypothalamic neurons. 1. Neurophysiol. 62,1400- 1409.

Akaike N., Krishtal 0. A., and Maruyama T. (1990) Proton-induced sodium current in frog isolated dorsal root ganglion cells. J. Neurophysiol. 63,805-813.

Akaike N., Kostyuk I’. G., and Osipchuk Y. V. (1989a) Dihydropyridine- sensitive low-threshold calcium channels m isolated rat hypothalamic neurones. J. Physiol. (Lond.) 412,181-195.

Akaike N., Tokutomi N., and Kijima H. (198913) Kinetic analysis of acetylcholine-induced current in isolated frog sympathetic ganglion cells. J. Neurophysiol. 61,283-290.

Akaike N., Lee K. S., and Brown A, M. (1978) The calcium current of Helix neuron. J. Gen. Physiol. 71,509-531.

Akaike N., Noma A., and Sato M. (1976) Electrical response of frog taste cells to chemical stimuli. J. Physiol. (Land.) 254,87-107.

Akaike N., Tsuda Y., and Oyama Y. (1988) Separation of current- and voltage-dependent inactivation of calcium current in frog sensory neuron. Neurosci. Lett. 84,46-50.

Fenwick E. M., Marty A., and Neher E. (1982) A patch clamp study of bovme chromaffin cells and of their sensitivity to acetylcholine. 1. Physiol. 331,577-597.

ffrench-Mullen J. M. H., Tokutomi N., and Akaike N. (1988) The effect of temperature on the GABA-induced chloride current in isolated sensory neurones of the frog. Br. J Pharmacol. 95,753-762.

Hamill 0. I’., Marty A., Neher E., Sakmann B., and Sigworth F. J. (1981) Improved patch-clamp techniques for high-resolution current recordings from cells and cell-free membrane patches. Pfltigers Arch. 391,85-100.

Ikemoto Y. and Akaike N. (1988) Kinetic analysis of acetylcholine- induced chloride current m isolated Aplysia neurones. Pfftigers Arch. 415240-247.

Ikemoto Y., Akaike N., and Kijima H. (1988) Kinetic and pharmacological properties of the GABA-mediated chloride current in Aplysia neurones: a ‘concentration clamp’ study. Br. 1. Pharmacof. 95,883-895.

Ikemoto Y., Akaike N., and Ono K. (1987) 4-Aminopyridine activates a cholinergic chloride conductance in isolated Helix neurons. Neurosci. Lett. 76,42-46.

Inomata N., Ishihara T., and Akaike N. (1989) Activation kinetics of the acetylcholine-gated potassium current in isolated atria1 cells. Am. J. Physiol. 257, C646-C650.

Inoue M., Omura Y., Yakushiji T., and Akaike N. (1986) Intracellular calcium ions decrease the affinity of the GABA receptor. Nature (Land.) 324,156-158.


Ishizuka S., Hattori K., and Akaike N. (1984) Separation of ionic currents in the somatic membrane of frog sensory neuron. J. Mm. Biol. 78,19-28.

Kaneda M., Nakamura H., and Akaike N. (1988) Mechanical and enzymatic isolation of mammalian CNS neurons. Neurosci. Res. 5,299-315.

Kaneda M., Oyama Y., Ikemoto Y., and Akaike N. (1989) Scorpron toxin prolongs an inactivation phase of the voltage-dependent sodium current in rat isolated single hippocampal neurons. Brain Res. 487, 192-195.

Kim D.-K., Tateishl N., and Akaike N. (1990) Proton-gated sodium current in parasympathetic ganglion cells of frog heart. J. Neurophysiol. 63,1060-1067.

Kiyohara T., Hirata M., Hori T., and Akaike N. (1990) Hypothalamic warm-sensitive neurons possess a tetrodotoxin-sensitive sodium channel with a high Q,,,. Neurosci. Res. 8,48-53.

Krishtal 0. A. and Pidoplichko V. I. (1980) A receptor for protons in the nerve cell membrane. Neuroscience 5,2325-2327.

Krishtal 0. A., Marchenko S. M., and Pidoplichko V. I, (1983)-Receptor for ATP in the membrane of mammalian sensory neurones. Neurosci. Lett. 35,41-K

Kuba K. and Nishi S. (1979) Characteristics of fast excitatory postsynaptic current in bullfrog sympathetic ganglion cells. Pfliqers Arch. 378,205-212.

Lester H. A. and Chang H. W. (1977) Response of acetylcholine receptors to rapid photochemically produced increases in agonist concentration. Nature (Land.) 266,373,374.

Magleby K. L. and Stevens C. F. (1972) The effect of voltage on the time course of end-plate currents. 1. Physiol. (Land.) 223,151-171.

Maruyama T., Ikemoto Y., and Akaike N. (1990) Effect of temperature on the inhibition of the GABA-gated response by intracellular calcium. Brain Res. 507,17-22.

Oyama Y., Hori N., Tokutomi N., and Akaike N. (1987) D-600 blocks open Ca*+ channels more profoundly than closed ones. Brain Res. 417,143-147.

Sadoshima J. and Akaike N. (1991) Kinetic properties of the caffeine- induced transient outward current in bull-frog sympathetic neurones. J. Physiol. (Land.) 433,341355.

Sadoshima J., Akaike N., Kanaide H., and Nakamura M. (1988) Cyclic AMP modulates Ca-activated K channel in cultured smooth muscle cells of rat aortas. Am. J. Physiol. 255, H754-H759.

Segal M. and Barker J. L. (1984) Rat hippocampal neurons in culture: voltage-clamp analysis of inhibitory synaptic connections. J. Neurophysiol. 52,469-487.

Shirasaki T., Nakagawa T., Wakamori M., Tateishi N., Fukuda A., Murase K., and Akaike N. (1990) Glycine-insensitive desensitization of N- methyl-o-aspartate receptors in acutely isolated mammalian central neurons. Neurosci. Lett, 108,93-98.

Akaike

Slater N. T. and Carpenter D. 0. (1984) A study of the cholinolytic actions of strychnine using the technique of concentratron jump relaxation analysis. Cell. Mol. Neurobiol. 4,263-271.

Ueno S., Nakaye T., and Akaike N. (1992) Proton-induced sodium current in freshly dissociated hypothalamrc neurones of the rat. J. Physiol. (Land.) 447,309-327.

Weinstock M. M. (1983) Activation and desensitization of acetylcholine receptors in fish muscle with a photolsomerizable agonist. 1. Physiol. (Land.) 338,423-433.

Yellen G. (1982) Single Ca2+-activated nonselective cation channels in neurobIastoma. Nature (Land.) 296,357-359.

Perforated Patch-Clamp Technique

Wolfgang Walz

1. Introduction Application of the conventional whole-cell patch-clamp

method leads to a replacement of the intracellular fluid with the intracellular pipet solution. The speed of this replacement or dialysis depends on the cell volume and electrode tip diameter. Although this mechanism can be used advanta- geously in many experiments, there are conditions where such a dialysis interferes with the current response to be tested. The response will disappear sometimes within minutes, an event that is usually called “rundown.” Several strategies have been developed in the last 10 yr to overcome this rundown. The most successful use nystatin or amphotericin B in the pipet. These- are ionophores that decrease the resistance of the sealed patch of membrane to selected small ions. A breaking of the patch membrane is not involved. Using such a principle, Lindau and Fernandez (1986) conducted experiments using ATP to permealize the patch membrane. ATP is, however, of limited use because it is dependent on a receptor. A breakthrough was the use of nystatin, first introduced by Horn and Marty (1988); this drug is an ionophore that is receptor-independent. A further improvement was the introduction of amphotericin B by Rae et al. (1991).

2. Dialysis of Cytoplasm by the Patch Micropipet Filing Solution Many electrophysiological responses depend on the

integrity of a variety of cytoplasmic and membrane-bound constituents. An example is the release of Ca2+ from intracellu-

From Neuromethods, Vol 26: Patch-Clamp Applicat/ons and Protocols Eds: A Boulton, G. Baker, and W. Walz 0 1995 Humana Press Inc.

155

156

lar stores in response to an elevation of inositol triphosphate. Responses depending on such intracellular messenger systems diminish between 3-25 min in conventional whole-cell recordings. Immediately after establishment of the whole- cell recording, the content of the pipet begins to diffuse into the cell interior and cytoplasmic cell constituents leave the cell. The internal volume of the patch pipet is many orders of magnitude larger than that of the cell. For this reason it can be safely assumed that the ion concentrations within the pipet do not change significantly and that one has to deal exclusively with a dilution of cytoplasmic constituents (dialysis). The speed of diffusion of the cytoplasmic cell constituents out of the cell into the pipet depends on their size. Ions will be the fastest; larger proteins will diffuse much slower, with some larger organelles, such as mitochondria, probably never leaving the cell. Cytoplasmic constituents, Ca2+ and H+, that are buffered by cell organelles will equilibrate more slowly, but they will eventually equilibrate. For this reason, most intracellular pipet solutions contain Ca2+ and pH buffers adjusted to intracellular values.

The relatively fast exchange of ions between pipet tip and cell interior will give rise to a Donnan potential. Many negatively charged proteins and polyanions will diffuse much more slowly than Na+, K+, and Cl-. This will result in a transient negative junctional potential (cytoplasm relative to pipet solution). Fernandez et al. (1984) calculated a Donnan potential of -15 mV. Although this will be the maximum potential at the start of the whole-cell recording, and it will only be transient, such a potential is bound to interfere with voltage-clamp measurements. Ca2+ currents are more sensitive than Na+ and K+ currents to such a dialysis (Hagiwara and Byerly, 1983). Depending on the cell type, these Ca2+ currents disappear between 5-30 min (see Fig. 2 in Korn and Horn, 1989). The addition of ATP, Mg2+, CAMP, and the catalytic subunit of the CAMP-dependent protein kinase slow considerably the loss of Ca2+ currents (Cota, 1986). In other experiments, this rundown of responses was successfully used to analyze intracellular messenger systems and dephos-

Perforated Patch-Clamp Technique 157

0

.

10 20 30 t (min)

Fig. 1, Recovery time of ATP response in cultured rat mlcroglia recorded with conventional whole-cell patch-clamp. The graph displays the relationship of the time between two subsequent ATP applications (t) and the relative amplitude (1,/I,) with standard error (reprinted from Walz et al., 1993).

phorylation/phosphorylation processes. This was accomplished by adding constituents of these systems to the pipet solution.

The “rundown” of responses caused by dialysis should not be confused with the desensitization of a response. Sev- eral authors never found a restoration of a response to a ligand during recordings lasting up to 1 h, even when the ligand was applied at infrequent intervals (Dufy et al., 1986). Walz et al. (1993), working with cultured microglial cells, found two overlapping processes that attenuated the response to extracellular ATP (Fig. 1). A transient depression of the inward current that was apparent after the first application lasted for about 5 min. This depression was attributed to desensitization. Another irreversible and slower

758 Walz

b 1

8 I

16 2b I

4 12 24

DURATION OF EXPERIMENT (min.)

Fig. 2. Variations of adenosine response over time in mammalian central neurones with conventional whole-cell patch-clamp. Filled circles: low resistance electrodes; open circles: high resistance electrodes (reproduced with permission from Trussell and Jackson, 1987).

developing depression was superimposed; it became apparent after a few minutes and the time course was dependent on the electrode resistance. This depression led to the aboli- tion of the response within 20-30 min. This was thought to be owing to dialysis of intracellular constituents necessary for the ionic response.

3. Strategies Used to Prevent Dialysis

3.1. Increase of Pipet Resistance The use of small-tipped electrodes (high resistance elec-

trodes) will drastically diminish the rate of dialysis of cytoplasmic constituents via diffusion through the microelectrode. Figure 2 demonstrates this effect. It represents the adenosine-activated current during whole-cell patch-clamp


recordings from cultured hippocampal cells from the mouse. The authors (Trussell and Jackson, 1987) used two approaches to investigate the time course of the responses to repeated applications of adenosine: One group of experiments used low resistance electrodes (l-4 Ma) with an access resistance of 3-10 Ma. With this group there was a rundown of the response that was complete within 15-20 min. The second group of experiments was conducted with high resistance electrodes (lo-20 MSZ) with an access resistance of 25-100 MC&. If experiments were undertaken with these high resistance electrodes, there was no significant decrease in the decline of the adenosine-evoked current within 24 min of rupturing the seal. Thus, an effective way to diminish or even prevent dialysis is the use of small-tipped electrodes. How- ever, such a high access resistance will lead to errors in speed and steady-state accuracy of the clamp. It is not possible to use series resistance compensation with such high electrode resistance (Jones, 1990) and that will not be acceptable with most applications. Another problem with such electrodes is that there are often spontaneous increases in the access resistance, because of resealing of the patch.

3.2. Addition of a Cgtosolic Extract to the Pipet Solution

In GH, pituitary cells, the response to thyrotropin releasing hormone (TRH) shows a rundown within several minutes after the rupture of the seal. Dufy et al. (1986) successfully employed an aqueous extract of osmotically lysed GH, cells in the micropipet to maintain the response. The use of the extract doubled the initial amplitude of the response and preserved it for the duration of the whole-cell recording.

1.5 x lo6 cells were allowed to lyse for 15 min in the presence of 1 mL distilled water. The supernatant was collected and filtered through a 0.45~pm Millipore filter. The pipet solution was 140 mM K-gluconate, 2 mM MgCl,, 1 mM EGTA, 5 mM HEPES at pH 7.3. It contained 0.4 mg extract/ml pipet solution. The efficacy of the extract was lost on storage over-

160 Walz

night at 4°C. The same extract produced from C6 glioma cells was not effective in maintaining the response.

3.3. Use of ATP

Mast cells are permeabilized if the extracellular side of the plasma membrane is exposed to ATP4-. Lindau and Fernandez (1986) used this property to reduce the access resistance across the membrane patch without breaking the seal. They added ATP into the pipet solution. ATE’ diffused into the extracellular side of the patch and created a permeabilization of the patched part of the membrane alone, not extending to the cell membrane outside the patch. The permeabilization reduced access resistance until there was an equilibrium between ATP in the pipet and in the membrane patch. The pipet contained 150 mM K-glutamate, 10 mM HEPES, 7 mM MgCl,, 200-400 cln/i Na,-ATP, and 200- 400 w BAPTA and was buffered to pH 7.2 with NaOH. This strategy abolished the rundown of the response. The access resistance was still fairly high, at 200-5000 M&I. This approach can be used only in cells that are permeabilized by extracellular ATP, which excludes most cells of interest for neuro- physiologists, because of the lack of the ATT?--receptor in CNS-derived cells. However, these pioneering studies by Lindau and Fernandez paved the way for experiments that use the same principle, but an ionophore that is not dependent on the presence of a receptor.

3.4. Use of Polyene Antibiotics

Nystatin and amphotericin B were used as ionophores in pipets. They function along the same principles as ATP. However, their permeability is restricted to inorganic monovalent ions. They can be applied to any cell type and are not dependent on the existence of receptors for them.

4. Use of Nystatin

4.1. Properties of Nystatin Pores

Nystatin (Fig. 3) is a polyene antibiotic. This antibiotic is an antifungal agent produced by bacteria of the genus Strep-


H H

APPHOTERICIN B

Fig. 3. Structural formulae for nystatin and amphotericin B.

tomycetes. If applied from both sides of a membrane it forms anion-selective channels (Cass et al., 1970). However, when added to only one side of a lipid bilayer, it induces a cation- selective conductance (Marty and Finkelstein, 1975). The addition of about 5-100 l.tg/mL will induce such a cation- selective conductance. The Stokes-Einstein radius is about 4A (corresponding to an approx mol wt of 200). This means anything larger than the size of glucose is impermeable. There is a slight permeability to glucose (Holz and Finkelstein, 1970). However, any important intracellular messengers, as well as the buffer substances usually employed, do not pass through the nystatin pores. Although these pores are more permeant for cations than for anions, there is a nonnegligible permeability for Cl-. However, the pores are impermeable to divalent ions, such as Ca*+, Mg*+, and SO:- (Korn et al., 1991). Nystatin pores show little voltage dependence. Nystatin is not soluble in water, but is somewhat soluble in methanol

162 Walz

Table 1 Typical Access Resistances

Obtained by Different Methods

Ionophore

ATP Nystatm Nystatin-fluorescem mixture Amphotericin B

Access resistance, ML-2

500-2000 20-50 20-40 3-10

and dimethylsulfoxide (DMSO). However, the solutions begin to lose activity soon after preparation. Thus, nystatin cannot directly dissolve in water solution, and DMSO is used most often as a carrier vehicle. The nystatin solutions have to be made up frequently from stock solutions.

4.2. Perforating the Patch Membrane mith Nystatin Nystatin was introduced for patch-clamp recording

without dialysis by Horn and Marty (1988) and Korn and Horn (1989). The basic mode of action is to have an active concentration of nystatin in the patch pipet. After the establishment of a membrane seal, no attempt is made to rupture the seal. Nystatin molecules incorporate into the patch membrane with some delay. This leads to a decrease of the access resistance until a plateau value is reached (see Table 1). This principle of nystatin use can only be applied if there is no, or minimal, lateral diffusion of nystatin molecules out of the patch membrane area into cell membrane areas outside the patch. This would lead to a perforation of the cell membrane. Nystatin, if only applied from one side of the cell membrane, is not capable of crossing the cell membrane. Thus, it will not reach the cell interior (see Section 4.1.). Horn (1991) investigated the possibility of lateral diffusion of nystatin across a seal. Experiments were undertaken in which a membrane seal was established between a pipet and a cell. The seal was not broken. Nystatin was then placed into the bath


and from there it incorporated rapidly into the cell membrane. However, at no point did nystatin become incorporated into the patched area, lowering the access resistance. Thus, one can safely assume that there was no lateral diffusion of nystatin across sealed membrane borders.

4.3. lntrapipet Dialysis of Nystatin If nystatin is present in the pipet solution at the tip of

the pipet during the establishment of the membrane seal, the likelihood of successful recordings is greatly diminished (Horn and Marty, 1988). This is probably owing to diffusion of nystatin out of the patched membrane area before the seal is established, and the subsequent destabilization of the cell membrane. There are two strategies employed to delay the appearance of nystatin at the tip of the electrode. The first one is to fill the tip of the pipet with a nystatin-free solution by dipping the pipet tip into such a solution. The remainder of the pipet is thereafter filled by backfilling with a nystatin- containing solution. This method is widely used and will give the desired stability of the recording. The diffusion of nystatin into the tip can be very unreliable, however, since the volume of the nystatin-free solution is hard to control. This means variable delays of up to 40 min occur before the access resistance reaches the final value. Rae et al. (1991) investigated how the degree of the filling of pipet tips immersed in solution for different times depended on the type of glass used and the tip diameter. With a brief, ~1 s, dip of the tip into the solution, the tip filled for a distance of 200-550 pm in pipets whose tips were l-2.5 pm in diameter. Another method to keep the tip initially nystatin-free is to have the pipet filled with a completely nystatin-free solution. Then a very fine polyethylene tubing that is filled with nystatin-containing pipet solution is introduced inside the patch pipet (50-100 pg/mL; Horn and Marty, 1988; see Fig. 4). With this method low access resistances can be obtained within several minutes after introducing nystatin. An additional approach to this problem is the use of fluorescein as a vehicle (see later).

764 Walz

Fig. 4. Schematic drawings of different patch-clamp configurations (adapted from Horn and Marty, 1988).

4.4. Composition of Pipet Solutions

One has to keep in mind that nystatin is an ionophore for all monovalent ions (although with somewhat less permeability to Cl- than to K+ and Na+). It is also permeable to Li+ and Cs+. Ca2+, Mg*+, and other multivalent ions do not permeate through the nystatin pores (Korn et al., 1991).

The Cl- permeability of the nystatin pores will potentially lead to a Donnan potential owing to the nonpermeable anions contained inside the cell (see Section 2.). To counter the development of such a Donnan potential between electrode tip and cell, the best strategy is to have in the pipet approx the same Cl- concentration as in the cell. The remainder of the anion deficit can be made up with an impermeable anion, for example, Sod*-. Korn et al. (1991) recommended the following composition for the pipet solution (in n&I): KCl, 55; K,SO,, 75; MgCl,, 8; and HEPES, 10; pH 7.35. Na+ is replaced by K+. In order to block K’ currents, K+ can be replaced by Cs+.

Solubility of nystatin in water is minimal. Nystatin has to be dissolved in either DMSO or methanol. Most authors prefer DMSO as a carrier vehicle. Most authors use approx 100 l.rg nystatin/mL pipet solution as a final concentration (see Korn and Horn, 1989; Sala et al., 1991). This concentration might have to be adjusted for each individual use.

165 Perforated Patch-Clamp Technique

4.5. Detailed Protocol for Use of Nystatin

The following protocol was used for recording whole- cell currents from cultured astrocytes for up to 2 h (Walz et al., 1994). The final nystatin concentration in the pipet was 200 &mL pipet solution.

1.

2.

3.

4.

5.

6.

7.

8.

Micropipets with a resistance of 3-6 MQ were pulled and firepolished. They were used within 5 h of the pulling. The micropipet tip was dipped briefly into an Eppen- dorf microcentrifuge tube filled with pipet solution (for composition, see Section 4.4.). The volume of this nystatin-containing fluid at the tip of the electrode was critical for the time course of the decrease in access resistance. We found values <5 s worked the best, but this differs with the micropipet configuration (consult Rae et al., 1991, for specifics). A stock solution of nystatin in DMSO (50 mg/mL) was kept at -20°C for no longer than 8 d in the dark. Immediately before use, 8 PL of this stock solution were dissolved in 2 mL pipet solution by up and down movements into an Eppendorf pipet until the solution was foamy and yellow. It was kept in a syringe in the dark and on ice. After 3 h this solution was discarded and replaced with a new one. This nystatin-containing solution was backfilled through a 0.2~pm filter into the micropipet, whose tip was previously filled with nystatin-free pipet solution. Air bubbles in the microelectrode were removed by tapping with a finger. To reduce electrode capacitance the pipet was dipped into Sigma (St. Louis, MO) coat. The cell was approached with the micropipet tip the same way as in conventional whole-cell patch-clamp measurements and a seal was established. The microscope light illuminating the preparation was now turned off and the current response to application of a regularly applied depolarizing pulse was observed.

166

min

0

F&=34 ML2

Cm = 15.5 pF

J 300 pA

5ms

Fig. 5. Formation of electrical access to the cell mterior during perforated patch recordings. Traces represent membrane current responses at a 20 ms, -10 mV voltage pulse from a holding potential of -60 mV. Numbers to the left of each trace indicate the time (in minutes) after formation of gigaseal patch. In the bottom trace, the capacitive current has electronically neutralized; series resistance and cell capacitance are shown (reproduced with permission from Korn and Horn, 1989).

9. As the nystatin was diffusing into the patched membrane area and creating pores, the formation of electrical access to the cell jnterior was recorded by the development of the capacitive current (see Fig. 5).

10. As soon as the capacitive current reached a plateau value, the experiment was started.

4.6. Use of a Nystatin-Fluorescein Mixture

Disadvantages of the method described above are the long waiting period (20-30 min) for the stabilization of the

Perforated Patch-Clamp Technique

access resistance, as well as the potential cytotoxicity of the carrier vehicle. An improved method for solubilizing nystatin in the pipet solution that addresses these problems was described by Yawo and Chuhma (1993). This involves the use of fluorescein. A stock solution of 5 mg nystatin and 20 mg fluorescein sodium was dissolved in 1 mL methanol. This stock solution could be kept at 4°C for several days. Imme- diately before use, 50 l,t.L of this stock solution were dried in a polyethylene test tube with a stream of N, gas. Then 1 mL of pipet solution was added and vortexed. The procedures had to be carried out under yellow monocolor fluorescent light to prevent bleaching of fluorescein. The solution was filtered through a 0.45~ym cellulose acetate filter and had to be used within 2 h. The micropipets were filled with that solution. There was no need to keep the tip nystatin-free. With this micropipet it was possible to apply positive pressure to the pipet when approaching cell membranes. Slight negative pressure was used for making a seal. It was very important to turn the transmitting light off after establishing the seal owing to photobleaching of fluorescein. The access resistance reached a stable plateau value after about 20 min (Table 1).

5. Use of Amphotericin B Rae et al. (1991) introduced amphotericin B as an alter-

native to nystatin. The access resistance obtained with this substance is routinely around 3-10 Ma and therefore somewhat smaller than the one for nystatin (Table 1). Ampho- tericin B differs from nystatin in that it contains an extra double bond between the diene and the tetraene (Fig. 3). There are no significant differences between the behavior of nystatin and amphotericin B, when applied from either one side or both sides of a lipid bilayer. Amphotericin B, however, seems to induce its one-sided cation-selective conductance at a lower concentration (Marty and Finkelstein, 1975).

The method is similar to the one used for nystatin. Rae et al, (1991) dissolved 6 mg amphotericin B in 100 /.tL DMSO to obtain a 60 mg/mL stock solution by vortexing for about

168 Walz

5 s. Twenty microliters of this solution was added to 5 mL of pipet solution (final concentration 240 pg/mL). The amphotericin-containing solution was used for backfilling only with a 0.2~pm Nalgene syringe filter. The tip of the pipet was filled by dipping into a container with a pipet solution lacking amphotericin. The optimal height of the amphotericin-free column at the tip of the pipet was found to be around 500 pm (Rae et al., 1991). This could be controlled by varying the tip immersion time. With such electrodes the access resistance decreased quickly and reached a steady level within 5-10 min. Thus, compared with nystatin the equilibration time for the access resistance is faster and the values obtained are smaller with amphotericin B. There are no reported disadvantages of amphotericin B in comparison with nystatin.

6. Special Application: The Perforated Vesicle

Levitan and Kramer (1990) developed a special application of the perforated patch-clamp technique, the perforated vesicle. This technique allows the recording of single channels in the outside-out patch but with the local presence of signal transduction cascades. The authors used pituitary tumor cells and recorded Ca*+ and K+ channels. They were able to obtain long term recordings from functional Ca*+ channels and could modulate them. Loading the cells with a fluorescent dye confirmed that the vesicle contained at the tip of the micropipet included cytoplasm (see Fig. 6).

The authors used micropipets that were filled with 100 pg/mL mystatin except for the tip of the electrode, which was nystatin-free. The pipet solution was either (in mM): KCl, 150; MgCl,, 16; and K+ HEPES, 10; pH 7.1; or K,SO,, 120; KCl, 16; MgSO,, 5; and Na+ HEPES, 10; pH 7.1. After establishing a tight seal, the access resistance started to decrease. When the access resistance was below 50 MQ, the clamp was switched to current clamp mode and the pipet was withdrawn to excise a vesicle.

Rae et al. (1991) were able, using amphotericin B, to record single-channel currents from such vesicles for 30 min or longer.


Fig. 6. Perforated vesicle. A cell-attached patch is formed with a pipet containing nystatin. After nystatin enters the membrane (Top) the pipet is withdrawn to form a perforated vesicle (Bottom). The membrane facing the bath contains receptors and ion channels with an outside- out orientation (reproduced with permission from Levitan and Kramer, 1990).

7. Conclusions

The perforated patch-clamp technique has reached the stage where it can be used reliably to obtain low access resistance to cells without disturbing metabolism. Since many biological questions relating to signal transduction, energy

170 Walz

metabolism, and diseases rely on a functionally undisturbed cell interior, this is the method of choice for whole-cell current recording. The method can also be used to record whole-cell currents with simultaneous measurements of the cytoplasmic Ca2+ activity by loading the cells with a dye (see Walz et al., 1993). The introduction of the perforated vesicle technique by Levitan and Kramer (1990) opened up new possibilities for studying single-channel kinetics. One can expect that the introduction of new ionophores and improvements of the perforated vesicle technique will further increase the usefulness of the perforated patch-clamp method.

Acknowledgment

The author thanks N. H. West for critically reviewing the text.

References

Cass A., Finkelstein A., and Krespi V. (1970) The ion permeability induced in thm lipid membranes by the polyene antibiotics nystatin and amphotericin B. I, Gen. Physiol. 56,100-124.

Cota G. (1986) Calcium channel currents in pars intermedia cells of the rat pituitary gland. J, Gen. Physiol. 88,83-105.

Dufy B., MacDermott A., and Barker J. L. (1986) Rundown of GH, cell K+ conductance response to TRH following patch recording can be obviated with GH, cell extract. Biophys. Biochem. Res. Commun. 137, 288396.

Fernandez J. M., Neher E., and Gomperts B. D. (1984) Capacitance measurements reveal stepwise fusion events in degranulating mast cells. Nature (Land.) 312,453-455.

Hagiwara S. and Byerly L. (1983) The calcium channel. Trends Neurosci. 6,189-193.

Holz R. and Finkelstein A. (1970) The water and electrolyte permeability induced in thin lipid membranes by the polyene antibiotics nystatin and amphotericin B. J. Gen. Physiol. 56,125-145.

Horn R. (1991) Diffusion of nystatin m plasma membrane is inhibited by a glass-membrane seal. Biophys. J. 60,329-333.

Horn R. and Marty A. (1988) Muscarinic activation of ionic currents measured by a new whole-cell recording method. J. Gen. Physiol. 92,145-159.


Jones S. W. (1990) Whole-cell and microelectrode voltage clamp, in Neuromethods, vol. 14 (Boulton A. A., Baker G. B., and Vanderwolf C. H., eds.), Humana, Clifton, NJ, pp. 143-192.

Korn S. J. and Horn R. (1989) Influence of sodium-calcium exchange on calcium current rundown and the duration of calcium-dependent chloride currents in pituitary cells, studied with whole cell and perforated patch recording. J. Gen. Physiol. 94,789-812.

Korn S. J., Marty A., Connor J, A., and Horn R. (1991) Perforated patch recording, in Methods in Neuroscience, vol. 4 (Corm P. M., ed.), Aca- demic, San Diego, pp. 364-373.

Levitan E. S. and Kramer R. H. (1990) Neuropeptide modulation of single calcium and potassium channels detected with a new patch-clamp configuration. Nature (Land.) 348,545-547.

Lindau M. and Fernandez J. M. (1986) IgE-mediated degranulation of mast cells does not require opening of ion channels. Nature (Lo&.) 319,150-153.

Marty A. and Finkelstein A. (1975) Pores formed in lipid bilayer membrane by nystatin. I. Gen. Physiol. 65,515-526.

Rae J., Cooper K., Gates I’., and Watsky M. (1991) Low access resistance perforated patch recordmgs using amphotericm B. J. Neuroscl. Meth. 37,15-26.

Sala S., Parsey R. V., Cohen A. S., and Matteson D. R. (1991) Analysis and use of the perforated patch technique for recording ionic currents in pancreatic B-cells. J. Membrane Biol. 122, 177-187.

Trussell L. 0. and Jackson M. B. (1987) Dependence of an adenosine- activated potassium current on a GTP-binding protein in mammalian central neurons. J. Neurosci. 7,3306-3316.

Walz W., Gimpl G., Ohlemeyer C., and Kettenmann H. (1994) Extracel- lular ATP-induced currents in astrocytes: involvement of a cation channel. J. Neurosa. Res. 38,12-18.

Walz W., Ilschner S., Ohlemeyer C., Banati R., and Kettenmann H. (1993) Extracellular ATP activates a cation conductance and a K+ conductance in cultured microglial cells from mouse brain, J. Neurosci. 13, 4403-4411.

Yawo H. and Chuhma N. (1993) An improved method for perforated patch recordings using nystatin-fluorescein mixture. Japan. J. Physiol. 43,267-273.

The Loose Patch Voltage Clamp Technique

J. H. Caldwelland R. L. Milton

1. Introduction

Extracellular microelectrodes have been used for many years to apply focal electrical stimulation to individual cells (for example, see Pratt and Eisenberger [1919] and Huxley and Taylor [1958]). Strickholm (1961) was the first to use a single extracellular electrode for both voltage control and recording membrane current. Strickholm used this method (developed as part of his PhD thesis) to study muscle membrane impedance and capacitance. Strickholm’s method lay dormant for over 20 yr until it was revived by Sti.ihmer and Almers (1982). Almers, Stiihmer, and their collaborators used the loose patch-clamp to study a wide range of muscle sodium and potassium channel properties, including mobility, inactivation, and spatial distribution (Stiihmer and Almers, 1982; Almers et al., 1983a,b, 1984; Stiihmer et al., 1983; Roberts and Almers, 1984,1992; Weiss et al., 1986; Roberts, 1987).

The discovery of the tight patch-clamp method with gigaohm seals was built on a long history of patch-clamp recordings that could be considered loose patch-clamp recordings (Neher and Lux, 1969; Neher and Sakmann, 1976; Neher et al., 1978). In fact, by using the extracellular electrode solely as a current collector and achieving seals in the range of tens of megaohms, Neher and Sakmann were able to record single-channel currents from acetylcholine receptors. However, loose patch-clamp recording is not merely a

From: Neuromethods, Vol. 26: Patch-Clamp Applications and Protocols Eds: A. Boulton, G. Baker, and W. Walz Q 1995 Humana Press inc.

773

174 Caldwell and Milton

stage in the development of the tight patch-clamp method. For some purposes, such as determining the spatial distribution of ion channels, the loose patch technique has important advantages over tight patch recording.

What is the distinguishing feature of the loose patch- clamp? Although loose patch-clamp recordings are usually made with pipets of large tip diameter (2-20 pm), whereas tight patch pipets are almost always small (l-2 pm), it is mis- leading to think in terms of pipet size when contrasting the two techniques. For example, it is possible to achieve gigaohm seals with large pipets, and small loose patch pipets can be used if channel density is high or if the pipet is used purely for current collection. The key element of the loose patch method is the acceptance of, and compensation for, a large leakage current under the rim of the pipet.

One of the major advantages of the loose patch-clamp is its ability to study macroscopic currents in a restricted patch of membrane. Most cells have nonuniform distributions of ion channels; currents recorded from cells in the two-microelectrode or whole-cell method are summed over the surface and can also be distorted by spatial nonuniformity of the clamp (owing, for example, to complicated cell geometry). Thus, the loose patch voltage clamp method is ideal for map- ping the distribution of ion channels. It has been used on large cells, such as muscle fibers (Almers et al., 1983a,b, 1984; Beam et al., 1985; Caldwell et al., 1986; Roberts, 1987; Milton et al., 1992; Ruff, 1992) and Aplysia neurons (Premack et al., 1989). It has also been applied to channel distributions on sensory receptors (taste receptors [Kinnamon et al., 19881; cochlear hair cells [Roberts et al., 19901; photoreceptors [Karpen et al., 19921) as well as single vertebrate axons (Chiu et al., 1985; Shrager, 1987).

2. Techniques

2.1. Amplifier

The circuit diagram shown in Fig. 1 can be used to construct a loose patch-clamp amplifier (see Stiihmer et al., 1983)

Loose Patch-Clamp Technique 175

--------q r-- - - --- - - --- - - q

I Capacitance 1 I I Comwnsation . I Head Stage

I (Clamps top of pipette at VP and ‘I _ A!! ; I amplifies current by -1000)

L, - - --A

pensation

Fig. 1. Circuit diagram for a loose patch voltage clamp amplifier; V, = command voltage, VP = voltage at the back of the pipet, V, = cytoplasmic voltage with respect to the bath, Vout = output of current to voltage converter, R = resistance of the pipet, R, = resistance of the seal, Rm = resistance of the membrane patch, Cm = capacitance of the membrane patch, IP = total current collected by the pipet, IS = seal current, and Im = current across the patch membrane.

using standard, readily available electronic components (low cost operational amplifiers, such as the LF411 from National Semiconductor are perfectly adequate). The headstage of this circuit clamps the tip of the pipet at the command voltage while reversing and amplifying the pipet current by a factor of one thousand. To reduce errors owing to polarization, it is important to connect the headstage amplifier to the pipet with two Ag/AgCl pellets or wires, one for passing current and the other for measuring voltage. The chassis of this circuit contains the main circuit board and three potentiometers. One of these potentiometers, labeled “pipet compensation” in Fig. 1, alters the voltage at which the back of the pipet will be clamped (depending on the magnitude of the pipet current). This allows the tip of the pipet to remain clamped at the


command voltage in spite of potential drops in the pipet caused by the series resistance of the pipet. The second potentiometer, labeled “seal compensation” in Fig. 1, allows a steady current to be subtracted from the output of the headstage amplifier; this permits analog subtraction of the current passing under the rim of the pipet (seal current). Since this seal current can be two orders of magnitude greater than the currents crossing the patch membrane (see Section 2.5.), this analog subtraction is necessary to allow proper amplification of the membrane current without saturating the final current to voltage converter or any A/D converters connected to the output. This analog subtraction also allows the membrane current to be viewed on an oscilloscope in real time. Following analog subtraction of the seal current, the remaining current is fed through a current to voltage converter such that the final output of the loose patch-clamp amplifier measures the membrane current with a scaling factor of 10 mV/nA.

There is also additional circuitry on the main circuit board, labeled “capacitance compensation” in Fig. 1, whose purpose is to amplify and feed back the command voltage to the negative input of the headstage amplifier through a 1 pF capacitor. Adjustment of the potentiometer controlling the amount of this feedback allows for the analog compensation of transient currents owing to charging and discharging of the membrane and stray capacitances at the beginning and end of a voltage pulse. This analog compensation will also help to prevent possible saturation of the output voltage. It should be noted, however, that the simple circuit of Fig. 1 does not reliably allow the capacitive current transients elicited by a square pulse of voltage to be used as a measure of the membrane area under the pipet tip. This is because the actual seal current has a delay imposed on it by the headstage amplifier, whereas the current shunted to ground by the seal compensation potentiometer does not. Therefore, during the onset of a pulse, there is an initial overcompensation for the seal current. The capacitive current transient caused by charging of the membrane under the lumen of the pipet is


thus seriously distorted, so that integration of this current will not give an accurate measure of membrane area. One way to solve this problem is to add circuitry that imposes a time delay on the seal compensation current matching the time delay imposed on the actual seal current by the headstage amplifier (Milton and Caldwell, 1990b). An alternative method is to turn off the seal compensation and subtract an appropriately scaled seal current from the capacitive transient (see Roberts, 1987).

2.2. Pipets

Loose patch pipets can be pulled from any of the standard glass available for making tight patch pipets. We have had good success with borosilicate glass capillary tubing (5068, Rochester Scientific Co., Rochester, NY). Loose patch pipets can be pulled successfully using any standard tight patch pipet puller (for example, an adequate and relatively inexpensive puller is available from Narishige, Model PB-7). The procedure for pulling loose patch pipets is essentially identical to the two-step process used to pull tight patch pipets. First the pipet is pulled at a relatively high heat for a distance of approx 5-7 mm to produce a narrowing of the pipet to a diameter of 300400 pm. This narrowed portion of the pipet is then centered with respect to the heating element and a second pull performed at a lower heat until the pipet separates into two approximately symmetrical halves of similar tip diameter. This procedure will produce pipets of 7 pm or less with a high degree of success. However, obtaining pipets with larger tip diameters is often more difficult since the second pull must be performed at such a low heat that the tip does not break cleanly when the two halves of the pipet separate. This often produces a tip that contains cracks or from which small slivers of glass may protrude. These defects usually cannot be corrected by firepolishing, and, therefore, the pipet is unusable. Consequently, when large pipets are required, great care must be taken in determining the heat setting for the second pull. We have also found that using pipets constructed from lower-melting-point glass (pot-

Caldwell and Milton

ash soda lead from Dagan Corp., Minneapolis, MN) can increase the success rate when attempting to pull pipets with tip diameters >lO pm.

As with tight patch pipets, loose patch pipets should be firepolished before they are used. Any of the standard micro- forges commonly used for firepolishing tight patch pipets can be used with loose patch pipets (such as the Narishige, Model MF-9), or a firepolishing setup can be constructed using a microscope, a platinum heating element, a micromanipulator, and a transformer. The highest magnification available on the microscope used for firepolishing should be approx 500x; any defects in the rim of the pipet can then be easily seen. The pipets should be firepolished until the rim of the pipet is smooth, without any large bumps or irregu- larities. Over firepolishing of the pipet can result in artifacts owing to currents arising from channels under the rim of the pipet (rim currents, see Section 2.5.) and therefore, should be avoided.

One advantage that loose patch pipets have over tight patch pipets is that a single loose patch pipet can be used for an unlimited number of measurements. In fact, we have found that loose patch pipets, if rinsed with distilled water following an experiment, can be stored and reused for many days (we used one particular pipet daily for over 3 wk before it unfortunately broke). This ability to reuse the same pipet for many experiments enhances the accuracy of comparisons made between experiments since this eliminates any errors that might occur as a result of normalizing the currents by the area of the pipet tip.

Pipets are normally filled with the same solution that bathes the preparation. One should keep in mind that the low electrical seal between the pipet and the membrane means that there will be diffusion between the pipet solution at the tip and the bathing medium. This is in contrast to gigaseal recordings where these compartments are kept separate by the membrane-glass seal. For this reason, solution changes should usually be made in both the bath and pipet. The pipet solution can be changed by simply sucking up the


new extracellular solution (the tip diameter is usually so large that this can be done very quickly). If the solution change is made only in the pipet or the bath, care must be taken to see that diffusion from the pipet does not affect the membrane outside the pipet (this can be prevented by constant, rapid perfusion of the bath) or that the concentration in the pipet does not become diluted.

2.3. Equipment

A vibration isolation table is essential for stable recordings. When voltage steps are applied through the pipet, the current under the rim can be on the order of loo-fold greater than the membrane current. Thus, fluctuations in the seal resistance owing to vibrations will be disastrous. For the same reason it is important that the micromanipulator be rigidly mounted to the microscope.

Any microscope, including a dissecting microscope (Almers et al., 1983a), can be used. If the preparation or tissue is sufficiently thin, our preference is an inverted microscope because the electrode tip can be precisely positioned over any region of the cell. If one wishes to map the distribution along the length of a muscle fiber, for example, a muscle that is more than a few fibers thick is difficult to use with an inverted microscope because it is often not possible to be certain that the electrode is always recording from the same fiber. There are ways to circumvent the problems of tissue thickness; for example, the muscle can be enzymatically dissociated (Beam et al., 1985; Caldwell et al., 1986) or fibers along the edge of the muscle can be used (Milton et al., 1992).

A second advantage of the inverted microscope is that the electrode can be bent such that the tip approaches the cell almost vertically. As the electrode is lowered onto the cell it pushes the cell against the bottom of the chamber, providing a support for the cell as the cell is indented to improve the seal. Almers et al. (1983a) devised a special support to serve the same function when they used a dissecting microscope, and a variety of supports for the tissue have been used for tight patch recording (e.g., Neher et al., 1978).


If the tissue is thick and enzymatic dissociation is nei- ther feasible nor desired, an upright microscope is necessary. However, if the electrode approaches the tissue vertically, the microelectrode distorts the surface of the bath and degrades the optical image. One solution is to use a water immersion lens (however, the working distance of these is short, which creates another difficulty). A better solution is to float a piece of a coverslip on the surface and view the microelectrode tip through the coverslip. Unfortunately, because of the meniscus surrounding the patch pipet, the coverslip will usually not float near enough to the pipet to allow clear visualization of the pipet tip. Under these circumstances, another pipet can be used to hold the coverslip against the loose patch pipet, or silk sutures can be glued to the edges of the coverslip and used to hold it against the loose patch pipet.

2.4. Procedure for Performing an Experiment

The steps enumerated below are meant to serve as a guide for performing a typical loose patch-clamp experiment.

1.

2.

The pipet should be placed in the bath and a test voltage pulse applied. This pulse can also be used during the formation of a seal (see step 3) if it is of a size and polarity that will not elicit significant currents from the cell of interest. While viewing the output of the clamp amplifier on the oscilloscope screen, the seal compensation potentiometer should be adjusted until the current elicited in response to the test pulse is nulled out. The resistance of this potentiometer is now equal to one thousandth of the resistance of the pipet. For example, if a 10 kQ potentiometer is used, full scale would represent a 10 Ma pipet. All of these adjustments should be made with the pipet compensation potentiometer set to zero. The resistance of the pipet compensation potentiometer should now be increased until the output of the loose patch amplifier saturates at either the negative or positive supply voltage. This should occur at a value of about


one thousandth of the pipet resistance and provides a check on the value of the pipet resistance obtained by adjusting the seal compensation potentiometer in step 1. It is impossible to predict whether the output will satu- rate at the negative or positive supply voltage since when the resistance of this potentiometer exceeds one-thousandth of the pipet resistance, the headstage operational amplifier becomes unstable, and any noise in its output will drive it into saturation at either the positive or negative supply voltage, depending on the polarity of the noise. However, regardless of the polarity that the saturated output achieves, this procedure gives the setting of the pipet compensation potentiometer necessary to compensate for potential drops because of the series resistance of the pipet tip. Thus, the voltage at the pipet tip will be clamped at the command voltage in spite of any voltage drops owing to current in the pipet. No further adjustments should be made to this potentiometer during the experiment.

3. The cell of interest should now be approached with the pipet. When the pipet contacts the cell, the output of the loose patch amplifier will no longer be saturated and will now give an accurate representation of the pipet current. A pulse of current owing to the command voltage pulse should now be visible on the oscilloscope screen. The seal compensation potentiometer should be adjusted until this pulse is nulled out. The resistance of this potentiometer now represents one-thousandth of the seal resistance between the pipet rim and the cell membrane. Hence, the seal resistance can be read directly from this potentiometer. Any remaining seal resistance that has not been compensated can be subtracted digitally by scaling a small, control pulse.

4. If the seal resistance is not greater than the resistance of the pipet, accurate measurements of patch currents will be very difficult. To increase this seal resistance the pipet can be pressed further down onto the cell surface or suction can be applied to the back of the pipet. However,


care should be observed when applying suction to the pipet since, as discussed in Section 2.5., the application of suction to cells can sometimes result in the formation of large membrane blebs, which can distort both the magnitude and the kinetics of current records.

5. When an adequate value of seal resistance is obtained, the test pulse can be turned off and the desired experimental pulse protocol applied to the cell and the resulting currents recorded.

2.5. Possible Sources of Error

A disadvantage that the loose patch-clamp technique shares with the tight patch is that the actual intracellular potential is not known and is therefore, of course, not controlled. However, although patch currents in the tight patch cell attached mode are unlikely to be large enough to alter cytoplasmic potentials, loose patch currents can become so large that they significantly change the potential within the cell (see Fig. 2). For large cells, such as oocytes or skeletal muscle fibers, an intracellular two microelectrode voltage clamp can be used in conjunction with the loose patch-clamp to control intracellular potential, and for small cells the intracellular potential can be controlled with a whole-cell tight patch electrode. However, with most cells this is an impractical solution. Therefore, the size of the pipet used in loose patch recording should be chosen with care, so that the elicited currents are not so large as to cause significant changes in the intracellular potential.

A problem unique to the loose patch-clamp is that a significant current flows under the rim of the pipet when applying voltage pulses through the loose patch pipet. This seal current varies inversely with the seal resistance and has several effects. First, some of the membrane current is lost through this path. This problem can be corrected by measuring the pipet and seal resistances and scaling the recorded current (Stiihmer and Almers, 1982) or by using the circuit in Fig. 1 that corrects for loss through the seal. Second, when applying a voltage via the recording electrode, the current


Cytoplasmlc Potential -. I/

20 mV I 1

50nA I 1 ms

Fig. 2. Simultaneous recording of Na current across the patch membrane (measured with a loose patch pipet) and the changes this current causes in the cytoplasmic potential (measured with an intracellular pipet 25 pm from the loose patch pipet). Recordings were made near the endplate (where Na channel density IS high) of a collagenase dissociated flexor digitorum brevis (FDB) muscle fiber from the mouse. Patch membrane was held at -85 mV and then was sequentially stepped to -60, -50, -45, -30, -15, and -5 mV. However, these potentials were only the transmembrane potentials immediately following the onset of the step since the Na current further depolarized the cell. Loose patch pipet tip diameter was 15 pm (pipet resistance 150 k&2).

required for the voltage clamp can be orders of magnitude greater than the membrane current. For example, an electrode with a lo-pm diameter tip might have a resistance of 300 k&2. If the seal resistance is the same magnitude as the pipet resistance and a voltage step of 30 mV is applied, the membrane current (typically a few nA) will be superimposed on a background seal current of 100 nA. This simple illustration emphasizes the importance of analog and/or digital compensation for the leakage current under the rim. This current leak during a voltage pulse limits the membrane currents that can be recorded. A third consequence of the low seal resistance is that channels under the rim of the pipet are

Caldwell and Milton

subjected to a voltage that depends on their position under the rim (a treatment of the voltage decay near the pipet rim was given by Strickholm [1961]). This problem of current from channels under or near the rim of the pipet (rim currents) can be reduced by using minimal firepolishing, but these currents cannot be eliminated without additional modifications, such as the use of concentric electrodes or using separate electrodes for voltage clamp and current collection (described in Section 3.). These rim currents place constraints on the usefulness of the loose patch-clamp for studying the details of channel kinetics and voltage dependence. If the density of channels under the rim is sufficiently large (for example at the neuromuscular junction), the recorded currents resemble two-electrode voltage clamp currents with inadequate spatial control of voltage.

One way to reduce rim currents is to increase the seal resistance by advancing the pipet further onto the cell or by applying suction to the back of the pipet. However, applying too much suction to the pipet may introduce another artifact into the measurements, since we have observed the formation of large membrane blebs (Fig. 3) in the pipet following the application of suction (Milton and Caldwell, 1990a,b, 1994). These blebs arose suddenly from small localized regions of membrane near the inner rim of the pipet and could grow to a diameter of 50 pm or more with continued suction. As the blebs grew in size, they remained attached to the cell by a thin tether approx 2 pm or less in diameter. Since blebs contain ionic channels, bleb formation and growth during loose patch recording can distort both the magnitude and kinetics of the currents. Therefore, the formation of these blebs during loose patch recording is to be avoided. Roberts et al. (1990) devised a simple way of preventing bleb formation when recording from enzymatically dissociated cochlear hair cells. A viscous collagen solution was sucked into the pipet tip and allowed to dry. The collagen plug did not affect the pipet resistance but did prevent bleb formation. It is our experience that the likelihood of bleb formation varied greatly with the type of cell being recorded from. Blebs were very


Fig. 3. Loose patch pipet pressed against the surface of a dissociated flexor digitorum brevis (FDB) muscle fiber from the mouse. The fiber diameter was approx 30 pm and the diameter of the pipet tip was 10 pm. (A) Before bleb formation. (8) After bleb formation. Approximately 20 mm Hg of suction was applied to the lumen of the pipet to induce bleb formation.

likely to form on collagenase dissociated muscle fibers, but were unlikely to form, even with large amounts of suction, on nondissociated fibers. Conversely, bleb formation occurs with low suction when making measurements from cultured chick myotubes or Xenopus myoblasts. We suspect that these differences in the likelihood of bleb formation are related to the extent of the extracellular connective tissue surrounding the cell, which we believe helps to stabilize the membrane and prevent bleb formation. However, whatever the cause of these membrane blebs, their formation introduces errors in loose patch recordings, and therefore, if suction is used to increase the seal resistance, it should be kept to a value low enough so as not to induce bleb formation.

The single electrode method, i.e., applying voltage pulses and recording current through the same electrode, has another artifact that we attribute to nonlinearity in the seal resistance. Current through the seal can be studied by press- ing the electrode against an insulating material, e.g., Sylgard. The seal current for large depolarizations is time dependent and looks very much like a delayed rectifier K+ current. More-

Caldwell and Milton

over, equivalent hyperpolarization does not produce a current that is equal and opposite. Thus, the recording of membrane currents that do not inactivate and that get larger with larger voltage steps (e.g., the delayed rectifier K+ current) can be contaminated by this nonbiological current. It is, therefore, important to confirm that the ionic current is real, e.g., by blocking the current pharmacologically, by achieving high seal resistance to reduce the artifact, or by using pulse protocols that inactivate the membrane current in order to separate the membrane current from the seal current.

3. Variations of the Method

3. I. Concentric Electrodes One of the major limitations of the loose patch-clamp

using the same electrode to voltage clamp and record current is the contribution of channels under the rim (described in Section 2.5.). Several approaches have been taken to eliminate the voltage drop under the rim. One method was to increase the effective seal resistance by creating a sucrose gap with a second electrode, larger than and concentric to the patch pipet (Hencek et al., 1969; Fishman, 1975). Sucrose flows through the outer pipet for electrical insulation of the patch from the bath. Another method is to use the outer electrode as a guard electrode. Hencek et al. (1969) devised an elabo- rate arrangement of multiple concentric rings that combined both a sucrose gap and a guard ring that held one annulus at ground. More recently, Roberts and Almers (Almers et al., 1984; Roberts and Almers, 1984) developed a concentric electrode arrangement and clamped both the inner pipet and the outer annulus (formed by the larger pipet) to the same command potential. In this case there is no potential drop under the rim of the inner pipet, and all channels recorded by the inner pipet are subjected to the same voltage. The use of concentric pipets is technically more difficult and places limits on the size of the cell that can be used but is one way to study voltage dependence and kinetics accurately. Alternative

Loose Patch-Clamp Technique

methods of reducing rim current are the use of suction (but see Section 2.5. for artifacts this may induce) and cleaning the cell surface by treatment with enzymes.

3.2. Current Collector

Another approach to eliminating rim currents is to use the electrode only to record membrane currents. The electrode can then be clamped at the bath potential with no potential drop between the inside and outside of the pipet. This method requires that the membrane potential be controlled by a separate set of electrodes. These electrodes can be a pair of intracellular electrodes; this usually requires that the cell be large, for example, molluscan neurons (Neher and Lux, 1969; Johnson and Thompson, 1989) or muscle cells (Neher et al., 1978; Almers et al., 1983b). If the cells are small, a whole cell patch-clamp electrode can be used to control membrane voltage with a loose patch electrode to record current (for example, in taste receptors, Kinnamon et al., 1988). Although this loose patch method is more demanding since one or two additional electrodes are needed, it has several features that make the extra effort worthwhile. First, the membrane potential is known and accurately controlled. It is worth reiterating that in the cell-attached configuration (both for tight patch and loose patch) the only potential controlled is that outside the membrane. Thus, if the current through the patch pipet is large (see Section 2.5.) or if drugs or hormones are added to the bath, the membrane potential may change owing to changes in the intracellular potential. A second strong argument for utilizing the current collector mode is that currents in the picoampere range can be measured, whereas the limit with the single loose patch electrode is hundreds of picoamperes. For example, the limit of resolution for a 10 pm electrode used on muscle cells is roughly 0.5 nA (Almers et al., 1983a; Caldwell et al., 1986) if the seal resistance is not much larger than the pipet resistance. How- ever, currents smaller than 0.5 nA can be detected with signal averaging and by increasing the seal resistance.

Caldwell and Milton 188

A Mmopin Connectors

Suction Port

er Glue

Retainmg Sleeve

C lonophoretic Current Pulse

Loose Patch-Clamp Technique

3.3. Ionophoresis with Loose Patch

189

One shortcoming of traditional loose patch-clamp measurements is that they can only be used to measure the distributions of voltage-gated ion channels. However, by combining loose patch voltage clamping with ionophoretic drug application the distributions of ligand-gated channels can also be measured with the loose patch-clamp method. The essentials of this combined technique are as follows: First, the loose patch pipet should be filled with solution and placed in a holder similar in design to that illustrated in Fig. 4A. Then the holder and pipet should be clamped on a microscope stage and the ionophoretic pipet filled and threaded inside the loose patch pipet. The ionophoretic pipet can easily be threaded to within several micrometers of the tip (Fig. 4B). This threading procedure is not difficult since even if the ionophoretic pipet hits the wall of the loose patch pipet it will not break if the loose patch pipet is filled with fluid. Finally, a drop of super glue should be applied to the back of the retaining sleeve (which can be a short length pipet of the same diameter as the loose patch pipet). This drop of super glue will flow into the space between the retaining sleeve and the ionophoretic pipet and in a few minutes will harden and hold the ionophoretic pipet in place. Care must be taken to keep the retaining sleeve and the back of the ionophoretic pipet dry so that the super glue will set properly. The loose

Fig. 4. (previous page) Pipet assembly for loose patch recording and ionophoresis. (A) Loose patch ionophoresis plpet holder. Holder was approx 1 cm in diameter and 4 cm in length. (B) Loose patch ionophoretic pipet. Diameter of loose patch pipet tip was 7 um. (C) Superposition of Na current and current through acetylcholme receptors (AChR) at the endplate of a dissociated flexor digitorum brevis (FDB) muscle from the mouse. Na current elicited by depolarizing the patch to 0 mV from a holding potential of -110 mV. AChR current was elicited by ionophoresis of ACh into the tip of the loose patch pipet, whereas the potential at the tip of the pipet was held at ground. Pipet tip diameter was approx 10 urn with the internal ionophoretic pipet about 5 p from the tip of the loose patch pipet.


patch pipet can now be pressed against a cell membrane and a seal established in the usual manner. Current pulses applied to the back of the ionophoretic pipet will ionophorese ligand into the tip of the patch pipet and elicit current from any ligand-gated channels in the patch membrane. We have used this method to measure current through voltage-gated sodium channels and acetylcholine receptors in the same patch of membrane at the endplates of dissociated skeletal muscle fibers (Fig. 4C).

4. Conclusion

The loose patch voltage clamp is the method of choice to determine the spatial distribution of channels or to analyze the biophysical properties of channels in restricted regions of the cell surface. Repeated use of the same electrode to sample different regions of the cell surface and recording under physiologically natural conditions (e.g., without damage to the cell surface or alteration of cellular constituents) are advantages over the tight patch-clamp technique. In fact, the tight patch method may not be appropriate for an accurate measure of channel density owing to the possibility of either bleb formation (Milton and Caldwell, 1990a, 1994) or changes in channel behavior in the cell attached (Fahlke and Riidel, 1992) or excised, inside out patch (Karpen et al., 1992). Thus, the loose patch voltage clamp technique can be a valuable complement to the tight patch-clamp and provide information unattainable with other electrophysiological methods.

Acknowledgments

This work was supported by grants from the National Institutes of Health to R. L. Milton (AR40801) and from the National Science Foundation to J. H. Caldwell (IBN 9213199).

References Almers W., Stanfield P. R., and Sttimer W. (1983a) Lateral distribution

of sodium and potassium channels in frog skeletal muscle: measurement with a patch-clamp technique. 1. Physiol. (Land.) 336, 261-284.


Almers W., Stanfield I’. R., and Stiihmer W. (1983b) Slow changes in currents through sodium channels in frog muscle membrane. I. Physiol. (Land.) 339,253-271.

Almers W., Roberts W. M., and Ruff R. L. (1984) Voltage clamp of rat and human skeletal muscle: measurements with an improved loose- patch technique. 1. Physiol. (Land.) 347,751-768.

Beam K. G., Caldwell J. H., and Campbell D. T. (1985) Na channels in skeletal muscle concentrated near the neuromuscular junction. Nature (Land.) 313,588-590.

Caldwell J. H., Campbell D. T., and Beam K. G. (1986) Na channel distribution in vertebrate skeletal muscle. I. Gen. Physzol. 87,907-932.

Chiu S. Y., Shrager P., and Ritchie M. (1985) Loose patch-clamp recording of ionic currents in demyelmated frog nerve fibers. Brain Res. 359,338-342.

Fahlke C. H. and Rude1 R. (1992) Giga-seal formation alters properties of sodium channels of human myoballs. Pfltigers Arch. 420,248-254.

Fishman H. M. (1975) Patch voltage clamp of squid axon membrane. I. Men&r. Biol. 24‘265-277.

Hencek M., Nonner W., and Stampfli R. (1969) Voltage clamp of a small muscle membrane area by means of a circular sucrose gap arrangement. Pfliigers Arch. 313,71-79.

Huxley A. F. and Taylor R. E. (1958) Local activation of striated muscle fibres. J Physiol. (Land.) 144,426-441.

Johnson J. W. and Thompson S. (1989) Measurement of non-uniform current densities and current kinetics in Aplysia neurons using a large patch method. Biophys. J 55,299-308.

Karpen J. W., Loney D. A., and Baylor D. A. (1992) Cyclic GMP-activated channels of salamander retinal rods: spatial distribution and variation of responsiveness. J. Physiol. (Land.) 448,257-274.

Kinnamon S. C., Dionne V. E., and Beam K. G. (1988) Apical localization of K+ channels in taste cells provides the basis for sour taste transduction. Proc. Natl. Acad. Sci. USA 85,7023-7027.

Milton R. L. and Caldwell J. H. (1990a) How do patch-clamp seals form? A lipid bleb model. Pfriigers Arch. 416,758-762.

Milton R. L. and Caldwell J. H. (1990b) Na current in membrane blebs: implications for channel mobility and patch-clamp recording. 1. Neurosci. 10,885-893.

Milton R. L. and Caldwell J. H. (1994) Membrane blebbing and tight seal formation: are there hidden artifacts in single-channel patch-clamp recordings? Comm. Theoret. Biol. 3,265-284.

Milton R. L., Lupa M. T., and Caldwell J. H. (1992) Fast and slow twitch skeletal muscle fibres differ in their distribution of Na channels near the endplate. Neurosci. Lett. 135,41-44.

Neher E. and Lux H. D. (1969) Voltage clamp on Helix Pomatia neuronal membrane; current measurement over a limited area of the soma surface. Pfltigers Arch. 311,272-277.


Neher E. and Sakmann 8. (1976) Single-channel currents recorded from membrane of denervated frog muscle fibres. Nature (Land.) 260, 799-802.

Neher E., Sakmann B., and Steinbach J. H. (1978) The extracellular patch- clamp: a method for resolving currents through individual open channels in biological membranes. Pflugers Arch. 375,219-228.

Pratt F. H. and Eisenberger J. P. (1919) The quanta1 phenomena in muscle: methods, with further evidence of the all-or-none principle for the skeletal fiber. Am. J. Physiol. 49, l-54.

Premack B. A., Thompson S., and Coombs-Hahn J. (1989) Clustered distribution and variability m kmetics of transient K channels in molluscan neuron cell bodies. 1. Neurosci. 9,4089-4099.

Roberts W. M. (1987) Sodium channels near end-plates and nuclei of snake skeletal muscle. J. Physiol. (Land.) 388,213-232.

Roberts W. M. and Almers W. (1984) An improved loose patch voltage clamp method using concentric pipettes. Pfiigers Arch. 402,190-196.

Roberts W. M. and Almers W. (1992) Patch voltage clamping with low- resistance seals: loose patch-clamp. Met/z. Enzymol. 207,155-176.

Roberts W. M., Jacobs R. A., and Hudspeth A. J. (1990) Colocahzation of ion channels involved in frequency selectivity and synaptic transmission at presynaptic active zones of hair cells. J. Neurosn. 10, 3664-3684.

Ruff R. L. (1992) Na current density at and away from endplates on rat fast- and slow-twitch skeletal muscle fibers. Am. J. Physiol. 262, c229-c234.

Shrager I’. (1987) The distribution of sodium and potassium channels in single demyelinated axons of the frog. J. Physiol. (Land.) 392,587-602.

Strickholm A. (1961) Impedance of a small electrically isolated area of the muscle cell surface. J. Gen. Physiol. 44, 1073-1088.

Stuhmer W. and Almers W. (1982) Photobleaching through glass micropipettes: sodium channels without lateral mobility in the sarcolemma of frog skeletal muscle. Proc. Nutl. Acad. Sci. USA 79, 946-950.

Stuhmer W., Roberts W. M., and Almers W. (1983) The loose patch-clamp, m Single-Channel Recording. (Sakmann B. and Neher E., eds.) Ple- num, New York, pp. 123-132.

Weiss R. E., Roberts W. M., Stuhmer W., and Almers W. (1986) Mobility of voltage-dependent ion channels and lectm receptors u-t the sarcolemma of frog skeletal muscle. 1. Gen. Physiol. 87,955-983.

Patch-Clamp Recording and RT-PCR on Single Cells

Bertrand Lambolez, Etienne Audinat, Pascal Bochet, and Jean Rossier

1. Introduction

The technique described in this chapter provides electrophysiologists using patch-clamp with a convenient method to link electrophysiological data to a molecular analysis of the mRNAs expressed in a single cell. This molecular analysis can be used either to correlate cell responses with their molecular basis or to identify cell types according to the expression of specific markers. The core of the molecular analysis is polymerase chain reaction (PCR), which makes it fast, sensitive, and simple.

Figure 1 outlines the general procedure followed in the experiments. Briefly, after recording of a cell with a patch- clamp electrode, the cell content is aspirated through the tip of the electrode and expelled into a test tube with the whole content of the patch electrode. Reagents are then added to perform first strand cDNA synthesis from the mRNA present in the cell. After completion of the reverse transcription (RT) reaction, further reagents are added to the tube to enable a PCR reaction to amplify the cDNA(s) under investigation. Basically, one tube corresponds to one cell, since no change of tubes and very few biochemical manipulations are required. After the first PCR the amplified DNA is analyzed on agarose gel electrophoresis. In some instances, the DNA product from the first PCR is reamplified through a second

From: Neuromethods, Vol 26: Patch-Clamp Applrcations and Protocols Eds: A Boulton, G Baker, and W Walz 0 1995 Humana Press Inc

193

194 Lambolez et al.

r- ELECTRCWYSICLCXiY (!&ok cell recording)

8 Pl

patch ‘.p<<.y

pipette ‘gyf

. . . .

-2x

: : : . * : : .

. . . . . . . . , . . . . . . . . .

. . . . . . . . . ..a.. . . . . . . . . . .*.*.*.*.*. . . . .

~<~~,ce11::

*I

<

1 I

F(NA WESTIffi

+ REVERSE TRANXRIPTICN

brdwn tip

expel1 6 5 ~1

Agarcee gel electmptmresis Restnct~cm analys7s Southern blot analysts

~enplate IS the punfwd first PCR product

35 Rx cycles

Agarcee gel electrophores?s Restnct1cm analysis Southern blot analysis

Fig. 1. Steps of single-cell RT-PCR.

PCR, either to get enough DNA for a refined analysis using restriction enzymes or to select one species among the population of amplified cDNAs.

This technique was designed to meet requirements on both electrophysiological and biochemical experiments. Since

Single-Cell RT-PCR 195

the conditions used in biochemical experiments are rather constant, the composition of the medium contained in the patch electrode is likely to be changed according to specific electrophysiological needs, as long as the subsequent biochemical steps are not affected.

Besides this flexibility, the time course of the experiments is well adapted to the collaborative work between electrophysiologists and molecular biologists, because of the rapid- ity of the PCR process and analysis. In typical experiments, at the end of a day of recordings (during which RT was carried out), the PCR reaction is run overnight and results are available the next morning. This allows electrophysiological experiments to be designed daily according to molecular biology feedback. Even if a second PCR and restriction analysis are necessary, these can easily be completed within the same day.

The present method has now been successfully used after recording cells (both neurons and glia) from different preparations (organotypic slice cultures, dissociated cell cultures, thin acute slices) to detect different mRNA species.

2. Materials and Methods

2.1. Labware, Reagents

The following list indicates in most instances the supplier from which we purchase reagents for RT-PCR on single cells, because this combination of products works in our hands.

1. 2. 3. 4. 5. 6.

7.

Hematocrit capillaries (Assistent or Kimble). Pipets (Gilson, Paris, France). Microloader (or geloader) tips (Eppendorf). Aerosol Resistant Tips (ART). Do not stand autoclave. 1.5 mL Eppendorf tubes (autoclaved). 500 I,I,L tubes (autoclaved) from Perkin-Elmer/Cetus (Norwalk, CT). The tubes have to match the size of the wells of the PCR machine used. Given the high efficiency required in each step of the reaction, this point has to be stressed. We obtained inconsistent results using tubes not well fitted to the Perkin-Elmer thermocycler. Programmable thermocycler (Perkin-Elmer/Cetus).

196 Lambolez et al.

8. 9.

10. 11. 12. 13. 14. 15. 16. 17.

18.

19.

20. 21.

Distilled autoclaved water. Tris-HCl and Tris base (Sigma, St. Louis, MO). HEPES free acid (Sigma). EGTA free acid (Sigma). CsCl (Appligene, Strasbourg, France). MgCl, (Prolabo, Paris, France). KC1 (Prolabo). Gelatin (Merck, Darmstadt, Germany). Dithiothreitol (Merck). Hexamer random primers (Boehringer Mannheim, Mannheim, Germany). Deoxyribonucleotides (dNTPs) (Pharmacia, Uppsala, Sweden). Oligonucleotide primers for PCR. Crude or purified (Genset and Appligene, France). RNasin (Recombinant RNasin, Promega), 40 U/pL. Moloney Murine Leukemia Virus reverse transcriptase (Gibco-BRL [Gaithersburg, MD] Reference 510-8025SA/B), 200 U&L.

22. Taq DNA polymerase (Stratagene, La Jolla, CA), 5 U/pL. 23. Light white mineral oil in 6-mL small bottles (Sigma).

2.2. Solutions

1. Diluting buffer: Tris-HCl, 10 mM, pH 8, autoclaved solution.

2. EGTA 0.2M stock solution: The pH is adjusted to 8 with KOH and the solution is autoclaved.

3. Patch intracellular solution: 140 mM CsCl, 3 mA4 MgCl,, 5 mM EGTA, 10 mM HEPES (pH 7.2). Eighty milliliters of this solution is prepared as follows. CsCl, MgCl, and HEPES are dissolved in 60 mL water and 2 mL of the EGTA stock solution is added. The pH is then adjusted to 7.2 with KOH. The volume is then adjusted to 80 mL with water and the solution is filtered (to remove any particle that may preclude gigaohm sealing) and autoclaved. Aliquots (1 mL) are then stored at -20°C until use. This solution does not seem to stand up to repetitive freezing and melting as far as gigaohm seal perfor-


mantes are concerned. The same solution but with 140 mM KC1 instead of CsCl has been used in some cases without noticeable difference.

The composition of this intracellular solution may be changed. For instance, the MgCl, concentration can be decreased and MgCl, can be added afterward to the RT reaction if necessary, to reach a final concentration of 2 mM. In theory, any patch pipet solution can be used, provided it does not contain inhibitors of the RT or PCR reaction. Since the RT reaction takes place after the recording, the final concentration of the reagents can be adjusted accordingly.

4. 20X DTT (Dithiothreitol): A 1M stock is prepared in water and filtered (millex). The working solution (20X) is 0.2M in water stored in 50 PL aliquots under nitrogen in 1.5 mL screw-cap tubes at -80°C.

5. 5X RT mix: Hexamer random primers (Boehringer Mannheim) dissolved in Tris (10 m&I, pH 8 at 5 mM). Deoxyribonucleotides (dNTPs) (Pharmacia) are each supplied as a lOO-mM solution. A working RT mix solution (5X) of random primers and dNTPs is prepared in Tris, 10 mM, pH 8, with random primers at 25 w and dNTl?s at 2.5 mM each. This working mix is stored as 30-FL aliquots in 500~PL tubes at -20°C.

6. RNasin and reverse transcriptase (RTase): are both stored at -8O”C, each as 7 PL aliquots under nitrogen in 1.5 mL screw-cap tubes.

7. PCR buffers: Two different buffers are used, depending on the PCR reaction. The 10X Taq buffer supplied by Stratagene (100 mM Tris-HCl, pH 8.3,500 mM KCl, 15 mM MgCl,, 0.1% gelatin, and “other stabilizers”) was found suitable for reactions amplifying only one cDNA (see later). However, it does not permit the adjustment of the MgCl, concentration that we found useful in, for instance, the coamplification of GluRl-4 cDNA described later. For this reason, we also use the following 10X PCR buffer: 200 mM Tris-HCl (pH 8.3 at 25”C), 250 mM KCl, and 1 mg/mL gelatin. This buffer is autoclaved and

198 Lambolez et al.

8.

9.

10.

stored at -20°C in 1 mL aliquots. It is supplemented at the desired MgCl, concentration when setting up the PCR. 15 n&I M&l, solution. For adjusting the M&l, concentration in the PCR. This solution is autoclaved and stored at -20°C in 1-mL aliquots. 100X dNTP PCR solution. This solution contains the four dNTP, each at 5 mM in 10 mM Tris, pH 8. This solution is used for the reamplification of the products from the first PCR or in pilot experiments. 100X PCR oligonucleotide primer. The stock is kept at -2O”C, undiluted. The working solution (100X) is diluted at 10 pmol/pL (for 100X) in Tris, 10 mM, pH 8. This 100X solution is stored in 50-PL aliquots at -20°C.

2.3. Design of the Oligos

Especially important is the choice of the oligos. As a rule, oligos with as few hairpins and primer dimers as possible are required. In most instances, failure in pilot experiments could be related to poor design of a primer pair (with stable hairpin or primer dimer). The choice of another primer pair enabled us to perform single-cell RT-PCR.

We use the oligo program (Rychlik and Rhoads, 1989), version 3. The choice of primer pairs amplifying only one cDNA is quite straightforward with the help of the command “good oligos.” For the design of primer pairs coamplifying several cDNAs, we first align the nucleotide sequences, look- ing for regions of high sequence similarity. Keeping in mind that no mismatch should be allowed close to the 3’ end of the primers, we choose from among these similar regions suitable primer sites.

We have until now selected primers generating cDNA amplified fragments ranging from about 350-750 bp. This general rule is a compromise between the advantages of long and short fragments. The efficiency of both reverse transcriptase and Tuq polymerase in synthesizing is reduced with long fragments; however, the amount of amplified material obtained at the end of the first PCR, in theory proportional to the length of the fragment, is greater and therefore facilitates further analysis.

Single-Cell RT-PCR

In order to avoid amplifying genomic DNA from the nucleus, oligos primers separated by an intronic sequence in the genome are selected when harvesting the cell’s nucleus. If the gene structure is unknown, we proceed to amplifications of 200 ng genomic DNA and compare with the cDNA amplification product. In this way, we can be sure that the amplified product comes from the mRNA expressed in the cell and not from nuclear DNA. However, considering the stringent treatments necessary to amplify genomic DNA from single cells (Li et al., 1988), we doubt that the procedure we use can efficiently release DNA from the nucleus as a suitable PCR template. This is the object of ongoing experiments that indicate that DNA from the nucleus genome is not amplified without the stringent treatment of Li et al. (1988).

2.4. Thermocycle PCR Programs

2.4.1. Program for the First PCR

Our standard procedure involves an initial denaturation of 3 min at 94°C (this time is reduced to 2 min when doing hot start), 40 PCR cycles, and a final 5 min elongation at 72°C. We have never gone beyond 40 cycles for two reasons. With a 100% yield at each step, the resulting amplification factor should be of 240 = 101* which, from a single molecule of 750 bases, results in 0.45 pg of amplified DNA (more than enough!). If the yield achieved at each step decreases to 80%, then the same amplification results in only 75 pg DNA (not enough for agarose gel detection). After more than 40 cycles, nonspecific amplification of the primer dimers becomes more and more pronounced. After more than 40 cycles with unsatisfactory amplification conditions, the risk is the cre- ation of a smear from which the desired cDNA can hardly be analyzed.

The 40 PCR cycles each have three temperature steps (denaturation, annealing, elongation), and, in some cases, a ramp between annealing and elongation. The denaturation is done at 94°C for 30 s. The annealing time is 30 s. The annealing temperature is 5°C below the Tm value given by the oligo program in the case of a perfect match between tem-

200 Lambolez et al.

plate and oligos. For primers designed to amplify several related cDNA species (usually having mismatches), the values given by the oligo program are often too low. In this later case, the annealing temperature has to be determined empirically (see later the AMPA receptor example). The elongation temperature is 72”C, and the time is proportional to the size of the fragment. In our experience, a ratio of 30 s/ 500 bases was sufficient, and further increasing the elongation time did not improve the yield.

2.4.2. Program for the Second PCR

For the reamplification of the first PCR product, we perform 35 cycles identical to those used in the first PCR.

2.5. Test of the Sensitivity of the PCR

When a new primer pair is designed, the test described here allows us to determine the experimental conditions necessary to achieve the sensitivity and specificity required in single-cell RT-PCR.

A cDNA stock is first prepared in a test tube containing 0.5 pg of total RNA in 4.5 yL. After 1 min denaturation at 95”C, the tube is chilled on ice and 2 PL of 5X RT buffer (supplied with the enzyme by Gibco-BRL: According to the manufacturer this buffer contains 0.25M Tris-HCl at pH 8.3,0.375M KCl, 15 rnM MgCl,, and 50 mM dithiothreitol), 2 PL 5X RT mix (containing random primers and dNTPs), 0.5 PL 20X DTT, 0.5 PL RNasin, and 0.5 uL RTase (10 CLr, final) are added. After 1 h incubation at 37”C, dilutions of the resulting cDNAs are prepared in water. The tube is then kept at -80°C.

The PCR reaction is then tested using different dilutions of the cDNA stock. We test the PCR within a range of cDNA amounts corresponding to 20 pg of the original total RNA up to 1 ng. PCR detecting widely expressed mRNA (e.g., AMPA receptor or GFAP or GAD) should yield clearly visible bands from 20 pg since the average cell’s content in total RNA is estimated to be about 10 pg. On the other hand, we always obtained PCR products from 1 ng, even for PCR detecting mRNAs with very restricted expression patterns.


The PCR tubes contain 1 FL diluted cDNA, to which is added 99 PL of a mix containing 10 PL 10X PCR buffer, the desired M&I, concentration, 1 PL of the 100X dNTP solution, 1 FL of the 100X sense primer solution, 1 PL of the 100X antisense primer solution, 0.5 JJL Ta9 polymerase, and water. After addition of two drops of oil, we proceed to PCR. We usually test several M&l, final concentrations (0.5,1,1.5,2, and 2.5 mM) on the smallest amount of cDNA that yields a detectable PCR product. We then choose the concentration that results in both the strongest signal and the least primer dimers.

2.6. Contamination

Contamination is, of course, one major concern when using PCR. In this instance, contamination by either RNAse or DNA or RNA molecules must be carefully avoided. Although the absence of contamination for a given PCR reaction has to be routinely tested (see later and in the example section), we have found that the observance of a few rules efficiently prevents contamination. RNAse contamination does not seem to be a major problem, and using standard procedures (see later), contamination, either from reagents, solutions, or labware, can be avoided.

Contamination by laboratory plasmids is the major problem, given the enormous amount of molecules produced (for instance, 1 PL of a 1 mg/mL solution of a 6 kb plasmid contains lO*l copies of the insert). Our advice is therefore not to set up any of the reactions for single-cell RT-PCR in a room used to manipulate plasmids containing the cDNA to be detected. Even contamination of nondisposable labware by washing is a concern in this case. The same advice is given for in vitro transcripts, although the problem is less critical.

As a rule, we have a set of pipets specially reserved for single-cell RT-PCR and related preliminary tests, With these pipets we never manipulate natural RNA or cDNA solutions containing more than 1 ng/pL. We never use these pipets for plasmids or in vitro transcript solutions containing more than lo3 mol/pL. In addition, these pipets are never used for

202 Lambolez et al.

solutions containing PCR-amplified fragments. Although we worked for 1 yr without aerosol blocking tips and avoided contamination, we now routinely use them.

All solutions are kept as aliquots, rather than in stock. Keep in mind that pH electrodes can be a source of contamination. We always wear gloves.

2.7. Electrophysiology and Cellular RNA Hamesting

For coupling electrophysiology with single-cell RT-PCR, recordings have to be performed using the patch-clamp technique. Although we have only used whole-cell recordings, other configurations of the technique are likely to be used if they allow cytoplasm harvesting. Whole-cell recordings are performed as described by Hamill et al. (1981). The formation of a gigaohm seal between the patch pipet and the cell membrane, together with an efficient harvest of the cytoplasm, are two critical parameters for the success of the subsequent biochemical steps.

Patch pipets are usually prepared from hematocrit capillaries, but harder types of glass can be used. The tubes are washed in one bath of ethanol, three baths of distilled water, dried for 1 h at 2OO”C, and kept in a closed box until they are used to pull the electrodes before each recording.

The tip resistance of the patch pipets used for recordings combined with single-cell RT-PCR experiments must be chosen as low as possible. In addition to the clamp problems arising from high series resistances, a poor access to the cell interior will render difficult the harvest of the cytoplasm into the pipet. The pipets used for experiments on large cells, such as Purkinje cells, usually have a tip resistance of l-3 MSX For smaller cells, such as cerebellar granule cells or small hippocampal interneurons, the pipets have a tip resistance of 3-5 MQ. Our experience has led us to discard pipets having a tip resistance higher than 5 MQ.

The patch pipets are back filled with the appropriate volume of patch intracellular solution (8 PL in our case) using Eppendorf microloader (or geloader) tips. Since this volume

Single-Ceil RT-PCR 203

is relatively small, the length of the pipet and of the silver wire that constitute the pipet electrode have to be adjusted carefully. From the initial 8 PL used to fill the pipet, we determined that roughly 6.5 PL are expelled from the patch pipet into the RT-PCR tube. This volume has to be measured to adjust the final volume of the RT reaction to 10 pL.

As long as the resistance of the seal between the patch pipet and the cell membrane is maintained in the gigaohm range, the duration of the recordings does not appear to be a critical parameter for the success of the single-cell RT-PCR. If, while recording, the seal breaks, the cytoplasm should be aspirated immediately. Nevertheless, the best results are obtained when the harvesting is started while the pipet is still firmly sealed onto the membrane.

The harvesting should be done under visual control to harvest as much of the recorded cell as possible and to avoid collecting neighboring cells. To aspirate the cell’s content, a negative pressure is applied into the pipet. If the soma of the cell is large enough, the flow of the cytoplasm into the tip of the pipet is visible under the microscope. By maintaining the suction, the whole soma can be collected, including the nucleus and the plasma membrane. With low resistance pipets (1-3 M!2) a gentle suction can be sufficient to collect the whole content of the soma. With higher resistance pipets, the cell’s content sometimes gets blocked at the tip of the pipet. If high pressure is not enough to unblock the cell’s content, gently breaking the very tip of the pipet, while maintaining a mild suction, will allow collection of the whole soma. In this case, one has to control carefully that other surrounding cells are not also aspirated. At the end of the harvesting procedure, the content of the cell should not remain stuck at the tip of the pipet but should be aspirated at least to the level of the larger shaft of the pipet.

To expel the content of the pipet, a positive pressure is applied with a syringe attached to the back of the electrode while the tip of the pipet is broken onto the inner wall of the PCR test tube.

204 Lambolez et al.

2.8. RT-PCR on Single-Cell Step by Step 2.8.1. General Protocol for the RTReaction

During the course of the electrophysiological experiments, the aliquots of intracellular solution, 5X RT mix, and 20X DTT are kept on ice and RNasin and RTase kept at -20°C. The patch pipet is filled with 8 PL patch intracellular solution. During the recording, 2 l.tL of 5X RT mix and 0.5 lt,L of 20X DTT are pipeted into the RT-PCR tube. After recording and aspirating the cell, the pipet’s content is expelled (we usually collect 6.5 PL) in this tube. We then add 0.5 PL RNasin and 0.5 PL RTase and the tube is flicked and briefly centri- fuged. The final volume should then be roughly 10 PL with final concentrations of: 0.5 mM each dNTP, 5 pM random primers, 10 mM DTT, and 20 U of RNasin and 100 U of RTase.

The tube is then placed for 1 h at 35-37°C. After this incubation, the tube is kept on dry ice until the PCR reaction. The remains of the aliquots are discarded after each experiment.

2.8.2. General Protocol for the First PCR 1. No hot start option: A solution with 90-PL volume for

each cell tube is prepared on ice. It contains per cell tube: 10 PL 10X PCR buffer, a varying amount of 15 mA4 MgCl, (if using the MgC$-free PCR buffer), 1 PL of the 100X sense primer solutron, 1 PL of the 100X antisense primer solution, 0.5 PL Taq polymerase, and water to 90 pL. Cell tubes are then placed on ice and 90 PL of the mix added to the 10 I..~L RT reaction. Two drops of mineral oil (small Sigma M3516 6-mL bottles are very convenient) are then added to each tube. The tubes are then placed in the PCR machine preheated to 80°C and the PCR program started.

2. Hot start option: Hot start is sometimes required when the choice of primer positions is limited and primer pairs that do not generate dimers (see AMPA receptor example later) cannot be selected. In such cases primers should be added in the preheated RT-PCR tubes since primer dimers are mainly owing to the RTase present in the 10 lt,L RT reaction. We prepare two different solutions. The

Single-Cell RT-PCR

volume of solution 1 is 60 PL x number of cell tubes. It contains per cell tube 6 l.tL 10X PCR buffer, a varying amount of 15 mM M&l, (if using the M&J-free PCR buffer), and water to 60 pL. The volume of solution 2 is 30 PL x number of cell tubes. It contains per cell tube 4 PL 10X PCR buffer, 1 PL of the 100X sense primer solution, 1 PL of the 100X antisense primer solution, 0.5 l.tL Tuq polymerase, and water to 30 PL. Sixty microliters of solution 1 are added to the 10 PL RT reaction in each cell tube and overlaid with 2 drops of mineral oil. The tubes are then placed in the PCR machine preheated to 8O”C, and, after 30 s, 30 l.tL of solution 2 are added to each tube on top of the oil. The PCR program is then started.

In both options, the final aqueous volume is 100 PL with final concentrations of 50 ~.LM each dNTP (from the RT reaction), 10 pmol/lOO ltL of each of the primers, 2.5 U/100 PL of Ta9 polymerase, and MgCl, at the desired concentration,

2.6.3. General Protocol for the Second PCR The separation of the PCR product from primer dimers

is necessary to achieve the second PCR. We now routinely use a procedure different from that previously described (Lambolez et al., 1992), which gives much better results, especially for PCR fragments ~600 bp long. Of the first PCR, lo-15 lt.L are loaded on a 1.5% low-melting-point agarose gel containing 1 pg/mL ethidium bromide. After migration, the agarose band containing the PCR fragment is cut under mild UV illumination (portable lamp) and placed in a 1.5-mL Eppendorf tube. The agarose is melted for 10 min at 65”C, and 0.5-5 PL (depending on the amount of DNA in the band) of the melted agarose containing the template PCR fragment are used directly in the second PCR. To the low melting agarose we add 99.5 to 95 PL of a solution containing 10 PL 10X Tu9 buffer (Stratagene), 1 l.l.L of the 100X dNTP solution, 1 ltL of the 100X sense primer solution, 1 i..tL of the 100X antisense primer solution, 0.5 PL Ta9 polymerase, and water. After addition of two drops of oil, we proceed to PCR.

206 Lambolez et al.

3. AMPA Receptor Subunits in Purkinje Cells: GFAP in Glial Cells

The present method was originally applied to determine, after electrophysiological investigation, the glutamate receptor subunits of the AMPA type expressed by individual Purkinje cells in cerebellar slice cultures (Lambolez et al., 1992). Purkinje cells were chosen because they are large cells that can be easily identified in this preparation and since the pharmacology of their responses to excitatory amino acids is well characterized.

Several subunits of a glutamate receptor-channel have been cloned. These subunits, named GluRl, 2, 3, and 4 (GluRl-4), exist in two versions (flip and flop) generated by an alternative splicing. Functional expression of homomeric or heteromeric combinations of these subunits generates receptors responsive to glutamate, quisqualate (QA), kainate (KA), and a-amino-3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA), and on which quinoxalinediones act as competitive antagonists. This pharmacological profile is characteristic of the glutamate receptor of the AMPA subtype, unlike the N-methyl-D-aspartate (NMDA) and high affinity kainate subtypes (for a review, see Wisden and Seeburg, 1993).

In this section we will focus on the experiments initially performed to test the patch intracellular solution and on the experiments we perform when a new primer pair has to prove suitable for single-cell RT-PCR. The general procedure is described in Fig. 2, together with the characteristics of the oligonucleotide primers and of the amplified GluRl-4 fragments.

3.1. Experimental Procedures

3.1.1. Test of the Sensitivity of the PC!? For the amplification of the GluR1-4 cDNAs we used

cDNAs derived from total RNA isolated by the method of Chomczynski and Sacchi (1987) from rat olfactory bulb. The PCR program was: 3 min initial denaturation at 94”C, followed by 5 cycles (94”C, 30 s; 45”C, 30 s; ramp to 72”C, 1 min

Single-Cell RT-PCR

- 1st PCR AMPLIFICATION up & IO primwo -

Glu M-4 coding ------- S-Cc

ampUfled fragments (750 W

up (sense primer) CCTTTGGCCTATGAGATCTGGATGTG mismatch with Ft2 A. :

R3 * T A R4 c

lo (antlssnsa primer) TCGTACCACCATTTGTTTlTCA mismatch with Rl lb C

207

2”d PCR AMPLIFICATION for RESTRICTION ANALYSIS

up (I lo primers

template* 1st PCR product

Sgl I (1900) -_____;_____-___-- Glu Al

Ssp 1286 I(2099) ___-_-_____~_______ GluR2

EC0 47 I I I (1983) --------*___________ GIU ~3

EcoR I(20351 - - - - - - - _ _ 4 _ _ - - _ - - - - Glu R4

Fig. 2. Molecular analysis of the AMPA receptor of a single cell. First PCR amplification: In the GluR1-4 coding sequences, the four putative transmembrane domains are represented as solid boxes, and the flip/flop region as a hatched box (not to scale). The arrows indicate the positions of the primers and their extension by 7’uq polymerase. All nucleotide sequences are written from 5’ (left) to 3’ (right). The up (upstream or sense, 26-mer) primer positions were 1600 on GluRl, 1621 on GluR2, 1630 on GluR3, and 1623 on GluR4 (position 1 is the first base of the initiation codon). The up primer fully matched with GluRl, but had one mismatch with GluR2 and GluR4 and two mismatches with GluR3. The positions and the natures of these nucleotide substitutions are shown. The lo (downstream or antisense, 22-mer) primer position was 2327 on GluRl, 2348 on GluR2,2363 on GluR3, and 2351 on GluR4. The only mismatch with the GluR1-4 cDNAs was with GluRlflip. The position and nature of this mismatch are indicated. Second PCR amplification-restriction analysis: The positions of the restriction sites on the amplified fragments are indicated by vertical dashes. The lengths of the fragments generated by the restriction enzymes were 300 and 449 bp for GluRl cut by Bg21,478 and 271 bp for GluR2 cut by Bsp12861,359 and 396 bp for GluR3 cut by Eco47111,411 and 338 bp for GluR4 cut by EcoRI, as calculated from the positions of the restriction sites.

10 s; 72”C, 30 s), followed by 35 cycles (94”C, 30 s; 49”C, 30 s; 72”C, 30 s), and 5 min final elongation at 72°C. PCR were run with cDNA amounts corresponding to either 1 ng, 100 pg, or 20 pg of the original RNA. Among the MgCl, final concentrations we tested (0.5, 1, 1.5, 2, and 2.5 mM), we found 0.5 mM to be the best in amplifying from 20 pg, since it resulted in both the strongest signal and the least primer dimers.

208 Lambolez et al.

It must be noted however that under these conditions the reaction suddenly stopped working and the MgCI, concentration had to be modified (see “AMPA receptors in hippocampal neurons” section). Although the reason remains unclear, we suspect that the properties of the Tuq polymerase supplied by Stratagene may have been different.

3.1.2. Test of the Patch Intracellular Solution for Reverse Transcription and P CR The efficiency of the reverse transcription reaction in the

patch intracellular solution was compared to that obtained using the reverse transcription buffer purchased from the manufacturer (Gibco-BRL). For this purpose, a 5X patch intracellular solution was prepared. Two reverse transcription reaction tubes were run in parallel: both contained 0.5 pg of rat olfactory bulb total RNA using the protocol described in the Methods section but with the addition of 0.25 PL [cPl?]dATP (3000 Ci/mmol, Amersham, Arlington Heights, IL) in each tube. The two tubes were identical, except that one contained 2 PL of 5X patch intracellular solution as a reverse transcription buffer, whereas the other contained 2 ILL of 5X RT buffer (Gibco-BRL). After 1 h incubation at 37”C, 1 PL of each tube was TCA-precipitated and the ratio (radioactivity incorporated in cDNA/total radioactivity) was measured. It was found that the ratio obtained in patch intracellular solution was only 4% lower than the ratio obtained in the manufacturer’s buffer. The patch intracellular solution did not affect the rate of the reverse transcription reaction or the PCR. This was further demonstrated by performing two PCRs (as described in the previous section) on cDNA amounts corresponding to 20 pg of the original total RNA from each of the RT reactions. The cDNA produced in patch intracellular solution was diluted in 10 PL 1X patch intracellular solution. The cDNA produced in commercial buffer was diluted in 10 PL of a 3 mM MgCl, solution. The final MgCl, concentration was adjusted to 0.5 mM in both cases. The band corresponding to the GluR1-4 amplified product had indeed a similar intensity in both cases (not shown).

Single-Cell RT-P CR 209

3.1.3. Cell Culture and Efectrophysiological Recording Cell cultures and electrophysiological recordings are

described in Lambolez et al. (1992). All drugs were applied with a U-tube microperfusing system. Recordings were performed using the whole-cell configuration of the patch-clamp technique (Hamill et al., 1981). Pipets had a tip resistance of l-3 Ma and were filled with 8 FL of the patch intracellular solution (see Materials and Methods section).

3.1.4. Cellular RNA Harvest and Reverse Transcription At the end of the recording, a negative pressure was

applied to the pipet and the flow of the cell content was observed under the microscope. The cytoplasm harvest was as complete as possible and the nucleus was sometimes harvested as well. The tip of the pipet was then broken into a test tube in which the pipet content was expelled. To the 6.5 uL usually obtained in the test tube was added 3.5 yL of a solution containing hexamer random primers (Boehringer, final concentration 5 PM), dithiothreitol (final 10 mM), the four deoxyribonucleotides triphosphate (Pharmacia, final 0.5 mM each), 20 U of RNasin (Promega, Madison, WI), and 100 U of Moloney Murine Leukemia Virus reverse transcriptase (Gibco-BRL). The resulting 10 PL solution was incubated 1 h at 35°C for the synthesis of single-stranded cDNA, and then kept on dry ice until PCR amplification.

3.1.5. First Amplification of the GluR1-4 cDNA Fragments

The first PCR reaction amplified a fragment of the GluRl, 2,3, or 4 cDNAs of the flip or flop forms. The up (sense, 26- mer) and lo (antisense, 22-mer) primers described in Fig. 2 were used. The Tm of the up and lo primers, calculated using the program Oligo (Rychlik and Rhoads, 1989), were, respectively, 62.6 and 54.4”C. After correction for the existence of mismatches the T,,, values were 53.3”C for the up primer (two mismatches with GluR3) and 49°C for the lo primer (one mismatch with GluRlflip).

210 Lambolez et al.

The reaction was performed in a final volume of 100 pL containing 10 pmol of each primer; the 10 l.tL reverse transcription reaction (final concentrations: deoxyribonucleotides, 50 pM each, MgCl, adjusted to 0.5 mM final; Tris-HCl, 20 mM, pH 8.3; KCl, 25 mM; gelatin, 100 pg/mL; 2.5 U Taq polymerase [Stratagene]). First, 5 cycles were run at 94”C, 30 s; 45”C, 30 s; ramp to 72”C, 1 min 10 s; 72”C, 30 s, followed by 35 cycles at 94”C, 30 s; 49”C, 30 s; 72”C, 30 s with a programmable thermocycler (Perkin-Elmer/Cetus).

To analyze the amplification material, 10 ~.LL of the reaction was run in parallel with a known amount of a molecular weight marker (4~x174, Hue111 digested) on a 1.5% agarose gel stained with ethidium bromide. The amount of amplified DNA was estimated by comparison with the bands of the molecular weight marker. The sizes of the amplified fragments calculated from the published sequences were 749 bp for GluRl, 2, and 4, and 755 for GluR3 (Fig. 2). Any mixture of these fragments should appear as a single DNA band on agarose gel electrophoresis.

Fifty microliters of the amplification reaction were then passed through a chromaspin P400 gel filtration column (Clontech, Palo Alto, CA) to remove primers and primer dimers. The purified PCR product was used for all subsequent amplification steps.

3.1.6. Second PCR Amplification for Restriction Analysis In order to obtain a sufficient amount of GluRl-4

amplified fragments for a restriction analysis, a second round of amplification was performed (Fig. 2), using the up and lo primers and 1 pg cDNA (5.105 molecules of 750 bp) of the purified first PCR product as a template. It was performed in a final volume of 100 PL containing the cDNA template; 10 pmol of each primer; 50 pM of each deoxyribonucleotide; MgCl,, 0.5 mM final; Tris-HCl, 20 mM, pH 8.3; KCl, 25 mM; gelatin, 100 pg/mL; and Taq polymerase (Stratagene), 2.5 U. Thirty-five cycles (94”C, 30 s; 49”C, 30 s; 72”C, 30 s) were performed and, after a chloroform- isoamyl alcohol extraction, the reaction was ethanol precipitated. The DNA was then resuspended in 20 PL H,O and 2-PL aliquots were digested


by a selected restriction enzyme (Fig. 2). Four enzymes were chosen, &II, Bsp12861, EcoRI (New England Biolabs, Beverly, MA), and Eco47111 (Boehringer), which selectively cut the GluRl, 2,4, or 3 PCR fragment, respectively. The restriction reaction was then analyzed by agarose gel electrophoresis, as described earlier.

3.1.7. PCR Amplification of the GFAP cDNA Fragment

Since the rat GFAP cDNA sequence has not yet been published, the primers used for the amplification of the GFAP cDNA fragment were chosen in two regions of the mouse GFAP cDNA (Lewis et al., 1984) sharing high sequence identity with its human counterpart (Reeves et al., 1989). The positions of the sense and antisense primers were 370 and 982, respectively (position 1 is the first base of the mouse cDNA sequence reported by Lewis et al., 1984), predicting a 632 bp fragment as a result of PCR amplification. Their respective sequences were, from 5’ to 3’, AAGCTCCAAGATGAAAC CAACCTGA and GCGATCTCGATGTCCAGGGC. The PCR amplification, performed on the cDNA obtained from single cells in a final volume of 100 PL contained 10 pmol of each primer; the 10 IJL reverse transcription reaction (final deoxyribonucleotide concentrations, 50 @I each); 2.5 U Taq polymerase (Stratagene) (0.5 pL) and 10 FL 10X Taq buffer supplied by the manufacturer. After 40 cycles at 94”C, 30 s; 55”C, 30 s; 72”C, 30 s, 10 FL of the reaction were then analyzed by agarose gel electrophoresis as described earlier.

3.1.7.1. RESULTS

After completion of the electrophysiological and pharmacological characterization of the excitatory amino acid responses (recording durations ranged 5-30 min), a negative pressure was applied to the patch pipet. The flow of the cell content into the tip of the recording electrode was clearly visible under the microscope. The pipet content was then used for reverse transcription and PCR amplification of the mRNAs encoding the AMPA receptor of each cell. The PCR amplification was performed with the up and lo oligonucleotides, the primer pair designed for the PCR amplification

212 Lambolez et al.

of all of the GluR1-4 cDNAs (see earlier and Fig. 2). A detailed analysis of the GluR1-4 cDNAs amplified was performed on 15 Purkinje cells. For each of these neurons, a single DNA band was detected when the product of the first PCR amplification was analyzed on agarose gel electrophoresis using ethidium bromide fluorescence. The amount of DNA present in the band was estimated from comparison with the intensities of bands of the molecular weight marker. The success of the experiment did not seem to depend on the duration of the electrophysiological recording, but rather on the amount of cytoplasm harvested. Indeed, from three cells recorded during more than 20 min we obtained 80 ng of DNA, which is in the upper range of what we obtained with the present method (lo-100 ng). The molecular weight of the DNA band corresponded to that predicted for the PCR amplification of the GluR1-4 cDNAs using the up and lo primers (750 bp) (see Fig. 38, panel a). The presence or absence of the nucleus in the harvest did not affect the yield of the procedure, and the amplification of nuclear DNA was excluded since, because of the presence of three introns in the GluR1-4 genes between the two primer positions (Sommer et al., 1990), the size of the gene amplification product would have been greater than the size of the cDNA amplification product.

Different species of the mRNAs encoding the AMPA receptor could be present in one cell and therefore amplified during the first PCR. We thus determined which of the GluRl-4 fragments were present within the single DNA band obtained from the first amplification. For this purpose, a second round of amplification was performed, using the up and lo primers and 1 pg DNA (5.105 molecules of 750 bp) from the first PCR. The resulting DNA fragment was then cut with either BglI, Bsp12861, Eco47111, or EcoRI, restriction enzymes, respectively, specific of GluRl, 2, 3, and 4 amplified fragment (see Fig. 2). As shown in Fig. 3, the digestion product of each enzyme, when analyzed by agarose gel electrophoresis, consisted of two bands that had the size predicted for the restriction digest of either GluRl, 2, 3, or 4 amplified fragment (Fig. 2 legend). In each lane, the 750 bp


1 st c R3 R2 Rl PCR

Fig. 3. Characterization of the AMPA receptor of single Purkinje cells. (A) Electrophysiological properties. Upper row: left, whole-cell currents induced in Purkinje cell 5 by the application of kainate (1 mM) at various holding potentials (-60, -4O,O, +40, and 60 mV); right, the corresponding plot as a function of the holding potential. Middle row, complex spikes evoked in Purkinje cell 13. Lower row, in the same neuron, currents induced by quisqualate and kainate. (B,Cl Subunit composition of the AMPA receptor of four different Purkinje cells. In all panels (a-e), right to left lanes correspond to Purkinje cells 5, 7, 13, and 14, respectively. (The electrophysiological recordings from cells 5 and 13 are shown in [A]). (B) Agarose gel electrophoresis of the cDNA fragments obtained after the second round of PCR and cut with BgII (panel b), Bsp12861 (panel c), and Eco47III (panel d) restriction enzymes specific for GluRl, 2, and 3 fragments, respectively. Panel e shows the simultaneous digestion of the fragments by the three enzymes and panel a the electrophoresis of cDNA fragments obtained after the first PCR (no enzymatic digestion). The bands present in the external lanes and between panels c and d are 4X174 HaeIII molecular weight marker. The positions of the 872,603, and 310 bp bands are indicated on the right of the gel. (C) Southern blot of the agarose gel shown specific probes in (B) labeled with the GluRl (upper row), GluR2 (middle row), and GluR3 (lower row).

214 Lambolez et al.

band corresponded to the amplified fragments left uncut by the restriction digest. The relative amounts of the digestion products were estimated by their ethidium bromide fluorescence intensities.

This analysis was performed on 15 Purkinje cells. Examples are given in Fig. 3 for cells 5 and 7 from cerebellar monocultures and for cells 13 and 14 innervated by climbing fibers. Figure 3A shows the electrophysiological recordings on cells 5 (upper row) and 13 (middle and lower row). In Fig. 3B, panel a shows the band obtained after the first amplification for the four different cells. Panels b, c, and d show, respectively, the BglI (cutting GluRl fragment), Bsp12861 (GluR2), and Ecu47111 (GluR3) digests. GluRl, 2, and 3 were present in the four cells. The EcoRI cut (specific of GluR4 fragment) did not generate any visible band (not shown). The simultaneous digest by the three GluRl, 2, and 3 specific enzymes (panel e) shows the superimposition of the restriction patterns obtained for the GluRl, 2, and 3 fragments. In this case, comparison between band intensities shows that the GluR2 fragment was more abundant than GluRl or GluR3 for all cells, although this was less prominent for cell 13. The GluRl/GluR2 proportion was about the same for cells 5,13, and 14, but much smaller for cell 7 because of the low amount of the GluRl fragment in this cell (see panel b). Little variation of the GluR3/GluR2 proportion was observed between the four cells.

The Southern blot analysis (Fig. 3C) of the same agarose gel confirmed the identity of the restriction fragments obtained. The GluRl probe heavily labeled the GluRl bands generated by the BglI digest seen in Fig. 3B, panels b and e. Similarly, the GluR2 and GluR3 probes strongly labeled their corresponding restriction fragments. The nonspecific labeling of other subunits restriction fragments by a given probe was a consequence of the high analogy shared by the four subunits in the amplified region (Boulter et al., 1990; Keinanen et al., 1990; Nakanishi et al., 1990). The rest of the 750-bp band left uncut by the digest by all three enzymes was labeled by the three probes, indicating that it is com-


posed of uncut GluRl, 2, and 3. The GluR4 probe (not shown) did not specifically label any of the bands present. In particular, the 750 bp band uncut by the three enzymes (Fig. 3B, panel e) was not labeled by the GluR4 probe. Essentially similar results were obtained from the analysis of six additional Purkinje cells in cerebellar monocultures and five additional Purkinje cells innervated by climbing fibers in olivocerebellar cocultures.

3.2. Specificity of the PCR and Quantification

Additional experiments were designed to control the absence of contamination during single-cell analysis. To that purpose, we recorded from 14 glial cells, characterized by a flattened morphology and their inability to produce fast action potential, and they did not respond to either kainate or quisqualate. The morphology of these glial cells, their lack of sensitivity to quisqualate and kainate together with the fact that they expressed the glial fibrillary acidic protein (GFAP) mRNA (see later) indicate that these cells were most probably type I astrocytes (Wyllie et al., 1991).

No GluR1-4 amplification product was obtained from single glial cells (n = B), unresponsive to either kainate or quisqualate (see Fig. 4 [AMPA] lane G), whereas each Purkinje cell recorded just before a tested glial cell was found positive (Fig. 4 [AMPA] lane I?). When a second round of amplification (see Section 4.1. and Fig. 2, but template was 10 PL of the first PCR reaction) was performed on glial cells (n = 3), it also gave no amplification product (not shown).

In six additional glial cells, the presence of mRNAs encoding the GFAP, an astrocyte specific marker, was detected by means of the same harvesting and amplification method, but using different primers specific to the GFAP. For these cells the amplification product analyzed on agarose gel electrophoresis consisted of a single DNA band (Fig. 4 [GFAP] lane G). The GFAP-specific amplification was negative on single Purkinje cells recorded during the same experiment (Fig. 4 [GFAP] lane P). The Southern blot analysis of the same agarose gel with a mouse GFAP-specific probe confirmed the

216

AMPA receptor

Lambolez et al.

GFAP

Fig. 4. Specificity of the PCR and quantification. (Left) AMPA receptor: Absence of the GluR1-4 mRNA in glial cells. Agarose gel electrophoresis of the amplification products for three Purkinle cells (lane P) and for three glial cells (lane G) recorded sequentially. Note the absence of DNA bands for the glial cells. (Right) GFAP: Agarose gel electrophoresis of the cDNA from three glial cells (lane G) and three Purkinje cells (lane P), recorded sequentially and amplified with the GFAP-specific primers. Note that only the glial cells were GFAP positive. For each gel, the @X174 HaeIII molecular weight marker is in external lanes.

identity of the DNA fragments (not shown). Altogether, these controls showed that the results of the present study are indeed single-cell specific.

The PCR amplifications using the up and lo common primers were designed to be competitive (Gilliland et al., 1990; Wang and Mark, 1990) in that the coamplification of the different GluR1-4 cDNA species should proceed with the same efficiency for each of them. The hybridization temperature used during the first five amplification cycles was well below the T, calculated for the primers (see Section 4.1.), so that each of the GluR1-4 flip or flop cDNAs were equally suitable templates for the amplification. Furthermore, the efficiency of the PCR should be the same with the different GluR14 fragments, since their sizes are equivalent.

3.3. Proportional Amplification of the Fragments

The extent to which the GluR1-4 proportions were maintained throughout reverse transcription and PCR amplification was tested. In vitro transcripts of the GluRl, 2, or 3 clones


Rl/R3 R2lR3

Fig. 5. Competitiveness of the PCR. Proportional amplification of AMPA subunits. 20 pg of GluRl, 2, and 3 in vitro transcripts were amplified using the up and lo primers after reverse transcription. The initial GluR2/GluR3 proportions were 9 (panel a), 3 (panel b), 1 (panel c), l/3 (panel d), and l/9 (panel e). For each panel the right lane corresponds to the uncut fragments and the left lane to the Bsp12861/Eco47111 double digest. Note in panels a and e that the restriction fragments of, respectively, GluR3 and GluR2 are very faint. Panel f: GluRl/GluR3 proportion of 1; right lane: uncut, left lane: BglI/Eco47III double digest. Each lane contains 5 ltL of the PCR reaction. On the right of the figure are indicated the positions of the two GluR3 restriction fragments that are between either the GluRl or the GluR2 restriction fragments. The molecular weight marker is m the left external lane, the position of the 603 bp band being indicated by an arrow.

(flop forms) were mixed in known proportions (total amount 20 pg), submitted to reverse transcription and PCR amplification using the up and lo primers. Restriction analysis specific of the GluRl, 2, or 3 was then performed on the amplified fragments. Figure 5 shows that the proportions of GluR3/GluR2 amplified fragments corresponded to the initial proportions of their transcripts (panels a-e). An initial GluRl/GluR3 ratio of 1 was also maintained throughout the procedure (Fig. 5, panel f). Similar results were obtained with GluRl-GluR2 combinations (not shown).

3.4. Discussion This study demonstrates the feasibility of using single-

cell RT-PCR to analyze the mRNAs present within a cell after patch-clamp recording.

218 Lambolez et al.

The molecular analysis of the mRNAs contained in single cells was specific to the AMPA receptor since the sizes of the amplified fragments corresponded to those predicted for GluR14 flip or flop cDNAs, the restriction digest generated fragments of the predicted size, and all of these fragments were labeled by their corresponding GluR14-specific probe.

Comparisons between glial cells and Purkinje cells show that the analysis of the mRNAs detected in single cells is cell- type specific. Purkinje cells, responsive to quisqualate and kainate, were found to express AMPA receptor mRNAs but not the GFAP encoding mRNA that is glial specific. In glial cells, which indeed were found to express the GFAP mRNA, the absence of quisqualate and kainate responses was correlated with the absence of detectable GluRl-4 mRNAs. Therefore, the amplified product is not derived from genomic DNA and no contamination by exogenous GluR1-4 mRNA or cDNA occurred during single-cell analysis.

The detection of a given GluR14-amplified fragment thus implies the presence of its corresponding mRNA in the single cell analyzed. Conversely, the absence of a given GluR14 fragment in the amplification product was related to the cell specificity of the AMPA receptor subunit composition rather than to a failure of the amplification reaction. Indeed, whereas the subset of subunits detected in Purkinje cells was rather constant, a completely different subunit composition was found for the AMPA receptor of granule cells (Lambolez et al., 1992).

4. AMPA Receptor Subunits and GAD in Hippocampal Cells

A homogeneous population of small round or ellipsoid neurons has been described in culture from hippocampus. These neurons, described as type II neurons, have glutamate receptors of the AMPA subtype that are highly permeable to calcium and display an inwardly rectifying current-voltage (I/V) curve (Iino et al., 1990; Ozawa et al., 1991; Ozawa and Iino, 1993). Most neurons, including the morphologically


heterogeneous population of type I neurons described below, have AMPA receptors with little permeability to calcium and linear or outwardly rectifying current-voltage relationship. AMPA receptors are known to be composed of subunits taken from a set of four proteins, GluRl, GluR2, GluR3, and GluR4 (GluRl-4). To determine if the different functional properties of AMPA receptors in type I and type II cells could result from different subunit composition, we have used RT-PCR on single cell. The method is the same as described in the previous chapter with two exceptions. The conditions of the first PCR have been changed and a second PCR was added to determine the ratios of flop/flip splice variants of the GluR1-4 mRNAs. In addition, we also investigated the presence of the glutamic acid decarboxylase (GAD) mRNA in hippocampal cells (Bochet et al., 1994).

4.1. Experimental Procedures

Dissociated neurons from rat embryonic hippocampus were grown in culture as described (Iino et al., 1990). Visu- ally identified neurons were recorded using the whole-cell configuration of the patch-clamp technique (Hamill et al., 1981) under voltage clamp conditions at different holding potentials. Pipets had a tip resistance of 5 Ma, and were filled with 8 PL of the patch intracellular solution described above. Kainate was applied iontophoretically, except for fast applications where kainate or AMPA were applied using a theta tubing pipet. The harvesting and reverse transcription procedure were done as described in the preceding sections.

4.1.1. First PCR

As stated in the previous section, we were unable to perform the GluR1-4 PCR in the conditions described earlier. Tests of the sensitivity of PCR indicated that a 1.5 mM MgCl, final concentration was now required instead of 0.5 mM. However, these conditions resulted in a strong primer dimer artifact that could be reduced greatly using the following hot start protocol. We prepared two different solutions. The volume of Solution 1 was 60 PL x number of cell

220 Lambolez et al.

tubes. It contained per cell tube 6 PL 10X PCR buffer (MgCl,- free), 8.67 FL 15 mM MgCl,, and water to 60 PL. The volume of Solution 2 was 30 PL x number of cell tubes. It contained per cell tube 4 PL 10X PCR buffer, 1 uL of the 100X sense primer solution, 1 PL of the 100X antisense primer solution, 0.5 uL Tuq polymerase, and water to 30 pL. Sixty microliters of Solution 1 was added to the 10 PL RT reaction in each cell tube and overlaid with two drops of mineral oil. The tubes were then placed in the PCR machine, preheated to 80°C and after 30 s, 30 PL of Solution 2 was added to each tube on top of the oil. The following PCR program was then started: 3 min initial denaturation at 94”C, followed by 5 cycles (94”C, 30 s; 45”C, 30 s; ramp to 72”C, 1 min 10 s; 72”C, 30 s), followed by 35 cycles (94”C, 30 s; 49”C, 30 s; 72”C, 45 s), and 5 min final elongation at 72°C.

4.1.2. Second PCR for Analysis of Flip/Flop Ratios The purified product of the first amplification was

submitted to a second amplification to determine the ratio of the flip/flop splice variants for each subunit expressed in a given cell. The second PCR amplified selectively one of the GluR1-4 cDNAs for which the flip/flop ratio was subsequently determined using restriction enzymes selective of either the flip or the flop form of the cDNA followed by agarose gel analysis.

Since only GluRl and GluR4 were detected in type II cells, second amplifications for flip/flop analysis were performed with the two following primer pairs: The sense primers were specific for either GluRl (Rl:GGACGAGACCA GACAACCAG at position 1717, position 1 is the first nucleotide of the initiation codon) or GluR4 (R4:GAAGGA CCCAGTGACCAGCC at position 1747); the common antisense primer was that used for the first amplification. The products of this second amplification (632 bp long for GluRl and 630 bp for GluR4) were digested, yielding fragments of the following size: 568 and 64 for GluRl-flip cut by BfaI; 577 and 55 for GluRl-flop cut by MseI; 583 and 47 GluR4-


Alternative splicing

Glu Rl-4 coding sequence

8 IyJ

8

lSt amplif led Glu Rl-4 fragment up #:i 750 ___--- _---

-0-- ___---

--

2nd amplified GluRl fragm! I 632 I

I

flop; Msel i

577 Restriction analysis

‘55; I

fhp j 568

Bfal i I ’ 64

Fig. 6. Flip-flop analysis of GluRl.

flip cut by HguI; and 562 and 68 GluR4-flop cut by HpaI (see Fig. 6 schema of flip-flop analysis for GluRl).

4.1.3. Amplifkation of Glutamic Acid Decarboxylase PCR amplification was carried out with the following

oligonucleotides, taken from the sequence of rat GAD45 (Erlander et al., 1991): TCTTTTCTCCTGGTGGTGCC (sense primer at position 713) and CCCCAAGCAGCATCCACAT (antisense primer at position 1085) yielding a product 391 base pairs in length (position 1 is the first nucleotide of the initiation codon). The PCR program chosen was 3 min initial denaturation at 94”C, followed by 40 cycles (94”C, 30 s; 51°C, 30 s; 72”C, 25 s), and 5 min final elongation at 72°C. The PCR amplification, performed on the cDNA obtained from single cells in a final volume of 100 pL, contained 10 pmol of each primer; the 10 ~J.L reverse transcription reaction (final deoxyribonucleotide concentrations, 50 l.t.M each); 2.5 U Taq polymerase (Stratagene, 0.5 PL); and 10 ~J,L of 10X Taq buffer supplied by the manufacturer. The amplification product was cut by MI, which yielded two products at 233 and 158 basepairs.

222 Lambolez et al.

4.2. Results

4.2.1. Absence of GluR2 Subunits in Type II Cells Whole-cell currents induced by iontophoretic applica-

tion of kainate, a nondesensitizing agonist of the AMPA receptors, were recorded at various holding potentials in cultured hippocampal neurons (Fig. 7). In most neurons the plot of the kainate-evoked current against the membrane potential is outwardly rectifying with a reversal potential (EreY) of -1.3 f 3.1 mV (n = 7) (Fig. 7A). As in previous reports (Iino et al., 1990; Ozawa et al., 1991; Ozawa and Iino, 1993), these cells were classified as type I neurons. In a population of neurons with small round or ellipsoid somata, responses to kainate exhibited a strong inward rectification (Fig. 7B). A rectification index (RI) was defined as follows:

RI = [I+,,/(40 - JQlA140&60 - EJI (1)

where I+40 and Id,, are the peak amplitudes of kainate-evoked currents at +40 and -60 mV, respectively (Ozawa et al., 1991). The recorded neuron was classified as a type II cell only when the rectification index was lower than 0.25. Occasionally cells with a rectification index between 0.25 and 1 were encountered. These neurons were discarded from the present study. The reversal potential of the kainate response in type II neurons was +1.25 f 4.14 mV. In order to determine which of the AMPA receptor subunit mRNAs are present, the cellular content of each neuron was aspirated into the recording pipet after it had been classified as a type I or type II cell according to the rectification index. A reverse transcription was performed on this material, followed by a PCR amplification with primers common to all GluR1-4 cDNAs. The presence of the different subunits of the AMPA receptors in the amplified product was then investigated by restriction analysis with enzymes specific for each GluR1-4 fragment.

The gel electrophoresis of Fig. 7D shows the results of such an analysis for four type II neurons including (left lane on each panel) the cell whose electrophysiological recordings are illustrated in Fig. 7B. For these four cells, as well as


for nine other type II neurons, only the DNA fragments corresponding to the GluRl and GluR4 subunits were detected. The absence of the GluR2 subunit in type II neurons was confirmed on Southern blots on which no DNA corresponding to the digestion product of GluR2 could be detected (Fig. 8). On the contrary, in all the eleven recorded type I cells, GluR2 was expressed at a high level (see Fig. 7C where 4 cells have been displayed). In five of these eleven cells, GluRl was the only other subunit detected together with GluR2. Among the other six cells, two cells contained GluRl, 2, and 3 (cell 322 in Fig. 7C, for example), two cells contained GluRl, 2, and 4, and two cells contained the four subunits. This het- erogeneity of the composition of AMPA receptors in type I cells contrasts with the constant pattern of expression of AMPA subunits observed in type II cells, in which only GluRl and GluR4 were detected.

Expression studies of the cloned GluR1-4 cDNAs have shown that receptors possessing the GluR2 subunit are not permeable to calcium and have either a linear or an outwardly rectifying I/V relationship, whereas receptors lacking GluR2 are permeable to calcium and exhibit a strong inward rectification. Therefore, it can be inferred from our results that the calcium permeability and the inwardly-rectifying I/V curve displayed by the AMPA receptors of type II cells are owing to the absence of GluR2 in these neurons.

The unique features of AMPA receptors in type II neurons have led us to characterize these cells further.

4.2.2. Flop Form in Type II Cells

The diversity of the molecular forms of the AMPA receptor is increased by an alternative splicing, generating two forms, flip and flop, for each of the GluR1-4 subunits (Sommer et al., 1990). In type II neurons, we have reamplified the first PCR product with a set of primers specific for either GluRl or GluR4, the only AMPA receptor subunits detected in these cells. Each PCR product was then digested with restriction enzymes specific for the flip or flop forms (Fig. SC). In five out of eight type II cells, only the flop forms of

0 m

224


GluRl and GluR4 were seen. For the three other neurons, the flip forms were also detected, but in much smaller amounts than the flop forms (Fig. SC).

The flip and flop forms differ by an alternative stretch of 38 amino acids. Expression studies have shown that receptors containing the flop forms desensitize more extensively to glutamate or AMPA application than those with the flip forms (Sommer et al., 1990). Accordingly, fast application of a saturating concentration (1 mM) of AMPA induced in type II neurons an initial peak of current that decayed rapidly (z = 7.5 & 0.5 ms, n = 5) to almost baseline level (Fig. 9A). In contrast, a saturating concentration (1 mM) of kainate evoked a nondesensitizing response in these cells (Fig. 9B). The ratio of AMPA/kainate responses at the steady state was very small (4.9 Z!Z 2.7%, n = 5), showing that AMPA-evoked desensitization was extensive. It seems, therefore, that the predominant expression of the flop forms in type II hippocampal neurons is correlated with the extensive desensitization of their AMPA receptors observed on AMPA application.

Fig. 7. (previous page) Rectification properties and subunit composition of AMPA receptors in cultured hippocampal neurons. (A,B) I/V curves of the kainate mduced responses in one typical type I, cell 298 (A) or type II, cell 335 (B) neuron. The recordings from which the I/V curves were drawn are shown in the inset and correspond to the whole- cell currents evoked by a 100 ms (A) or 50 ms (B) iontophoretic application of kainate at holding potentials of -60, -40, -2O,O, +20, +40 and +60 mV from bottom to top traces. (C,D) Analysis of the cDNAs encoding the AMPA subunits in four type I (C) and four type II (D) neurons. Indi- vidual cell numbers are indicated, and in all panels (NC, Rl-R4, TD) the four lanes correspond to the same four different cells. (NC) Electrophore- sis of the cDNA fragments obtained after the first PCR. The other panels show the results of the second amplification cut by the subunit-specific restriction enzymes. (TD) Simultaneous digestion by the four enzymes. Note the absence of digestion product with the enzyme specific for GluR2 in the four type II cells (D) and the presence of these fragments for the four type I neurons (C). The position of molecular weight marker Hue111 digest of +X174 (Q) is indicated between the two gels.

226 Lambolez et al.

lal

type II

\

Q= ICJ

-872 -603

-310

Fig. 8. Absence of GluR2 from type II cells. (A) Gel analysis of one type I cell and 5 type II cells. Electrophoresis of second PCR products cut by Bsp12861 (specific for GluR2) as described in Fig. 2. The bands corresponding to the digestion products of GluR2 are visible only in the type I cell. (B) Southern blot. The DNA fragments were amplified as described in the text, separated on agarose gel, and transferred onto Hybond N+ (Amersham). The probe was made as described (Lambolez et al., 1992) by random priming labeling of GluR2 flop in the pBluescript cloning vector (kind gift of P. H. Seeburg). The 478 bp band corresponding to the digestion product of GluR2 is strongly labeled in the type I cell, Labeling of the 271 bp band is weaker, because of the lower amount of material and the mismatch of the flip fragment with the flop probe. No labeling is seen at the corresponding positions in type II cells, indicating the total absence of GluR2. The faint labeling of the uncut product seen for type II cells is owing to the crossreactivity of the GluR2 probe with the other AMPA receptor subunits.

4.2.3. Glutamic Acid Decarboxylase in Type I! Cells Type II cells represent a homogeneous population that

accounts for ~5% of the cells in hippocampal cultures. They can


lOOpA

100 Ins

603

Fig. 9. Analysis of the splicing variants of GluRl and GluR4 in type II neurons. (A,B) Response of type II neurons to applications of 1 mM AMPA (A) and 1 mM kainate (B) with a fast perfusion system. Note that AMPA response desensitizes to almost baseline level, whereas the kainate response remains unchanged throughout the application. Agonist was applied rapidly with the use of a theta tubing pipet attached to a piezoelectric device (NEC, Japan). The diameter of the tip of the pipet was about 200 pm and the tip was placed within 100 p of the neuron. The solution perfusing the neuron was completely changed in 10-20 ms, as judged by the rise time of the kainate response. Membrane potential was held at -60 mV. (C) Selective amplification and analysis of GluRl or GluR4 for four other type II neurons. The amplification product of the first amplification was purified and submitted to a second amplification with the primers described in the experimental procedures. The uncut PCR product is shown (nc), together with its digestion with enzymes specific for the flip (i) or the flop (0) forms. The identification number of each individual cell is indicated. A band smaller than the uncut form indicates the presence in the cell of the corresponding splice variant. Note the absence of the flip forms except for cell 317 where a minor proportion of GluR4fhp is found.

be identified visually by their small size (12 pm) and the round or ellipsoidal shape of their soma (Ozawa et al., 1991). Further- more, in five out of five other identified type II cells, the mRNA

228 Lambolez et al.

Fig. 10. Presence of glutamic acid decarboxylase (GAD) mRNA in type II cells. (A) Gel analysis of the PCR product of two type II cells (three other type II cells were analyzed with the same results). After electrophysiological identification, PCR amplification was carried out with the ohgonucleotides described in the experimental procedures yielding a product 391 basepairs in length. The amplification product was cut by MI, which yielded two products at 233 and 158 basepairs. (B) Southern analysis of the same gel with a labeled oligoprobe derived from the sequence of the GAD. Only the larger restriction fragment is labeled by the phosphorylated ohgoprobe corresponding to nucleotides 871-896 of rat GAD65: GCCTTGGGGATCGGAACAGACAGCG.

encoding GAD was detected (Fig. 10 shows two of these cells). This observation identifies type II cells as inhibitory GABAergic interneurons.

42.4. Discussion The present study shows that the presence of messenger

RNAs and their splice variants can be characterized at the single-cell level and correlated with the functional properties of this cell.

A clear correlation is indeed observed between the unusual properties of native AMPA receptors in type II neurons and the subunit composition of this receptor as determined by single-cell RT-PCR. Type II neurons express only GluRl and GluR4 and the unusual calcium permeability and rectification properties of their AMPA receptors are owing to the absence of the GluR2 subunit (Hollmann et al., 1991). They

do not express the nonedited form of GluR2 (Sommer et al., 1991) which would have been an alternative explanation for their properties. Indeed, receptors incorporating the nonedited form of GluR2 have the same calcium permeability and rectification properties as receptors lacking GluR2 (Hume et al., 1991; Burnashev et al., 1992). In addition, the very extensive desensitization of AMPA receptors in type II cells is in agreement with the predominance of the flop splicing variant in these neurons.

Until now, the relationships between the electrophysiological response and the subunit composition have been studied in transfected nonneuronal cells. This study demonstrates that this can be performed on natural receptors in neuronal cultures even when the cells of interest are a minor part of a heterogeneous population.

Finally, the detection of the GAD mRNA in type II neurons provided us with additional information on these cells: They are GABAergic interneurons. Single-cell RT-PCR can, therefore, be used to identify a cell from the expression of a cell- type-specific marker.

References

Bochet P., Audinat E., Lambolez B., Crepe1 F., Rossier J., Iino M., Tsuzuki K., and Ozawa S. (1994) Subunit composition at the single-cell level explains functional properties of a glutamate-gated channel. Neu- ron 8,383-388.

Boulter J., Hollmann M., O’Shea-Greenfield A., Hartley M., Deneris E., Maron C., and Heinemann S. (1990) Molecular cloning and functional expression of glutamate receptor subunit genes. Science 249, 1033-1037.

Burnashev N., Monyer H., Seeburg P. H., and Sakmann B. (1992) Divalent ion permeability of AMPA receptor channels is dominated by the edited form of a single subunit. Neuron 8,189-198.

Chomczynski P. and Sacchi N. (1987) Smgle-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162,X6-159.

Erlander M. G., Tillakaratne N. J., Feldblum S., Pate1 N., and Tobin A. J. (1991) Two genes encode distinct glutamate decarboxylases. Neuron 7,91-100.

Gilliland G., Perrin S., and Bunn F. H. (1990) Competitive PCR for quantification of mRNA, in PCR Protocols (Innis M. A., Gelfand D. H., Sninsky J. J., and White T. G., eds.) Academic, New York, pp. 60-69.

230 Lambolez et al.

Hamill 0. I’., Marty A., Neher E., Sakmann B., and Sigworth F. J. (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pflfigers Arch. 391,85-100.

Hollmann M., Hartley M., and Heinemann S. (1991) Calcium permeability of KA-AMPA-gated glutamate receptor channels: dependence on subunit composition. Science 252,851-853.

Hume R. I., Dingledine R., and Heinemann S. F. (1991) Identification of a site in glutamate receptor subunits that controls calcium permeability. Sczence 253,1028-1031.

Iino M., Ozawa S., and Tsuzuki K. (1990) Permeation of calcium through excitatory amino acid receptor channels in cultured rat hippocampal neurones. J. Physiol. (Land.) 424, 151-165.

Keinanen K., Wisden W., Sommer B., Werner P., Herb A., Verdoorn T. A., Sakmann B., and Seeburg I’. H. (1990) A family of AMPA-selective glutamate receptors. Science 249,556560.

Lambolez B., Audinat E., Bochet I’., Crepe1 F., and Rossier J. (1992) AMPA receptor subunits expressed by smgle Purkinje cells. Neuron 9, 247-258.

Lewis S. A., Balcarek J. M., Krek V., Shelanski M., and Cowan N. J, (1984) Sequence of cDNA clone encoding mouse glial fibrillary acidic protein: structural conservation of intermediate filaments. Proc. NutI. Acad. Sci. USA 81,2743-2746.

Li H., Gyllensten U. B., Cui X., Saiki R. K., Erlich H. A., and Arnheim N. (1988) Amplification and analysis of DNA sequences in single human sperm and diploid cells. Nature Wzd.) 335,414-417.

Nakanishi N., Shneider N. A., and Axe1 R. (1990) A family of glutamate receptor genes: evidence for the formation of heteromultimeric receptors with distinct channel properties. Neuron 5,569-581.

Ozawa S. and Imo M. (1993) Two distinct types of AMPA responses m cultured rat hippocampal neurons. Neurosci. Lett. 157,187-190.

Ozawa S., Iino M., and Tsuzuki K. (1991) Two types of kainate response in cultured rat hippocampal neurons. J. Neurophysiol. 66, l-l 1.

Reeves S. A., Helman L. J., Allison A., and Israel M. A. (1989) Molecular cloning and primary structure of human glial fibrillary acidic protein Proc. Natl. Acad. Sci. USA 86,5178-5182.

Rychllk W. and Rhoads R. E. (1989) A computer program for choosing optimal oligonucleotides for filter hybridization, sequencing and zn vitro amplification of DNA. Nucl. Aced Res. 17,8543-8551.

Sommer B., Keinanen K., Verdoorn T. A., Wisden W., Burnashev N., Herb A., Kohler M., Takagi T., Sakmann B., and Seeburg P. H. (1990) Flip and flop: a cell-specific functional switch in glutamate-operated channels of the CNS. Science 249,1580-1585.

Sommer B., Kohler M., Sprengel R., and Seeburg P. H. (1991) RNA edit- ing in bram controls a determmant of ion flow m glutamate-gated channels. Ce22 67,11-19.


Wang A. M. and Mark D. F. (1990) Quantitative PCR, in PCR Protocols (Innis M. A., Gelfand D. H., Sninsky J. J., and White T. G., eds.) Academic, New York, pp. 70-75.

Wisden W. and Seeburg P. H. (1993) Mammalian ionotropic glutamate receptors. Curr. Opin. Neurobiol. 3,291-298.

Wyllie D. J. A., Mathie A., Symonds C. J., and Cull-Candy C. J. (1991) Activation of glutamate receptors and glutamate uptake in identified macroglial cells in rat cerebellar cultures. J. Physiol. (Land.) 432, 235-258.

Patch-Clamp Technique in Brain Slices

T. D. Plant, J. Eilers, and A. Konnerth

1. Introduction

The technique of patch-clamp recording in brain slices is applicable to a large variety of cell types in slices from nearly all areas of the central nervous system (CNS) in animals at many different stages of development (Blanton et al., 1989; Edwards et al., 1989; Konnerth, 1990). To date, the technique has been successfully applied in a number of areas of the CNS, including the hippocampus, cerebellum, striatum, brain stem, corpus callosum, hypothalamus, frontal and visual cortex, medial septum, olfactory bulb, retina, and spinal cord. In addition, the technique has been applied to other tissues, including the pituitary gland and cardiac muscle.

The main advantages of the technique are the ability to use a high-resolution recording technique, without the use of enzyme treatment, on visually-identified neurons and glial cells in situ with their structure and, for neurons, with most of their synaptic contacts preserved. Slices can be made from animals of defined ages for developmental studies. Furthermore, the technique can be combined with other methods, e.g., fluorometric measurements of intracellular ions, confocal microscopy, and single-cell RT-PCR, to obtain precise information about the properties of the cell type being studied.

From: Neuromsthods, Vol. 26. Patch-C/amp Applications and Protocols Eds A. Boulton, G. Baker, and W. Walz 0 1995 Humana Press Inc.

233

Plant, Eilers, and Konnerth 234

2. Methods

2.1. Brain Slices for Patch-Clamp Studies

2.1.1. Preparation and Maintenance of Slices The procedures for the preparation of brain slices are

described briefly below:

1. 2. 3. 4.

5. 6. 7.

8. 9.

10. 11.

Decapitate animal. Open skull. Dissect out brain area of interest. Place tissue in oxygenated ice-cold saline (cl min from decapitation to this stage). Wait at least 10 min while tissue cools. Trim tissue to correct angle. Glue trimmed tissue to slicer stage using cyanoacrylate glue and cover with cold solution as quickly as possible. Cut slices (60-400 p thick) with a vibrating slicer (~10 n-tin). Move each slice to incubation chamber immediately after slicing (using a cut and firepolished Pasteur pipet). Incubate at 32-37°C for around 30 min before recording. Keep remaining slices at 25°C before use (-10 h).

To reduce damage to the tissue, it is particularly impor- _ _ tant to remove the brain as rapidly as possible and immerse it in ice-cold saline. Following cooling, a piece of tissue that is larger than the required structure is then usually prepared and trimmed for slicing using a scalpel. Surrounding tissue confers mechanical stability on the tissue during slicing. The slicing chamber is partly filled with frozen saline and, after gluing the tissue to the stage, is filled up with ice-cold saline to cover the preparation. It is important that the slicer stage and the tissue surface to be glued are dry, otherwise the preparation may not stick properly and detach during slicing. The angle and vibration frequency (usually near the maximum) of the blade should be adjusted to prevent the tissue being pushed while cutting the slices. A dissecting microscope above the stage allows slicing to be performed under visual control. Following the slicing procedure, the slices are incubated at -37°C for 30 min, then at 25°C until

Patch-Clamp Technique 235

Fig. 1. (A) Incubation chamber for brain slices. The inner chamber is made of two Plexiglas rmgs that, when pressed together, clamp a piece of cotton or polyester mesh that supports the slices. A piece of plastic tubing holds this chamber wedged in a glass beaker. The bubbler that supplies 95% 0, and 5% CO, creates a circulation of oxygenated solution in the beaker, resulting in a flow of solution over the slices from above. With this arrangement, slices remain undisturbed on the mesh during the incubation time of up to 10 h. The glass beaker is placed in a water bath that maintains the temperature at 35°C for 30 min after slicing, thereafter at 25°C. (B) Areas of the rat brain used for preparing sagittal slices from the cerebellar vermis (1) and transverse slices of the neocortex (2).

they are used, in a chamber of the type illustrated in Fig. 1A. This type of chamber (developed by Alisdair Gibb, Univer- sity College London), which maintains a flow of oxygenated solution over the slice and in which the slices remain on the cotton or polyester mesh, allows slices to be kept in good condition for up to 10 h.

2.1.2. Making Slices from Various Parts of the Brain Slices can be made from most areas of the brain and spi-

nal cord. It is important for the survival of neurons for the dendritic tree to be intact. Cells whose dendrites have been “excessively” cut off during slicing have an unhealthy shiny appearance and tend not to allow the formation of a seal. The optimal slice orientation has to be found for the neurons to be studied and the individual experimental goal. For the cerebellum, sagittal slices of the vermis preserve most of the

236 Plant, Eilers, and Konnerth

Purkinje neuron dendrites intact, whereas in transverse and parasagittal slices of the hippocampus and visual cortex respectively, specific synaptic pathways are best preserved. Figure 1B shows the regions of the rat brain and planes of section used for transverse slices of the neocortex and sagittal slices of the cerebellar vermis.

2.1.3. Making Brain Slices from Different Animals at Different Ages, from Newborn to Adult Young animals have a number of advantages for the

preparation of slices. Their skulls are softer and easier to open, allowing the brain to be removed more rapidly. The brain is smaller and cools more rapidly when placed in ice-cold saline. Both of these are critical steps in the preparation (see earlier). Furthermore, the tissue from older animals is tougher and more susceptible to anoxia. The increase in connective tissue and myelination in the brains of older animals may result in more damage to the cells and their processes during slicing.

2.2. Patch-Clamp Recording in Brain Slices

2.2.1. Recording Setup The recording chamber (Fig. 2B) is made of Plexiglas

glued to a round piece of glass, cut from a microscope slide. A square hole at the center of the Plexiglas forms the chamber for the tissue. At the sides, the plastic is cut away at an angle to allow easy access of a number of pipets. Circular holes at the front and back, connected to the main chamber by a passage under the level of the solution, act as an inlet and outlet, respectively, for the extracellular solution. The holes act as bubble traps and reduce the transmission of noise, caused, e.g., by the suction outlet, to the recording pipet. A reference (ground) electrode is placed in the outlet hole. The chamber can grounded directly using a Ag-AgCl wire or pellet, or through an agar bridge filled with KC1 (Neher, 1992).

Slices are transferred from the incubation chamber to the recording chamber using a cut and firepolished Pasteur pipet. The slice is held in position using a “grid” (Fig. 2A), a U- shaped piece of flattened platinum wire (0.5 mm diameter)

Patch-Clamp Technique

A r==- perluslon in

237

x-y-mampulator

Fig. 2. Recording chamber and setup. (A) Recordmg chamber for patch-clamp measurements in brain slices. (B) Recording arrangement for patch-clamping visually-identified neurons in slices.

with a parallel array of fine nylon threads (Konnerth et al., 1987; Edwards et al., 1989). Grids of different sizes can be made, depending on the size of the tissue to be studied. The grid is placed on the slice using forceps under a dissecting microscope. For the cerebellum, the position of the grid is not as critical as for the hippocampus, where it is important that the grid does not cover the layer of interest.

The chamber containing a slice is placed on the stage of the recording setup, as illustrated schematically in Fig. 2B. Slices usually are viewed using an upright fixed-stage microscope with a long working distance water immersion objective of high numerical aperture. However, for some applications, e.g., confocal microscopy (see later), it is neces-


sary to use a microscope with a fixed objective and moveable stage. Objectives with 40x magnification and a working distance of about 2 mm allow individual neurons to be identified and a number pipets to be positioned under the objective. For higher resolution, objectives with 63x magnification and a working distance of 1.45 mm are available. The angle between the pipet and the preparation must be low enough to allow pipets to be positioned under the objective without them touching it even when covered with Sylgard (see later). Angles of around 30 and 20” are close to the maximum that can be used with 40 and 63x objectives respectively. The recording stage can be moved horizontally (X-Y) with respect to the objective using micromanipulators positioned underneath it. Both the stage and the micromanipulators for positioning the pipets are mounted on columns (optical rail) fixed on the table top separate from the microscope (see, e.g., Levis and Rae, 1992). A number of pipets can be present in the chamber simultaneously; for recording, cleaning, and stimulation, and local application by iontophoresis or pressure.

Cells can be viewed using differential-interference-contrast (DIC) or bright-field optics (Edwards et al., 1989; Regehr et al., 1992). With both systems, cell bodies, dendrites, and axons can be resolved. The resolution can be improved by combining DIC optics with infrared illumination and an infrared-sensitive video or CCD camera (MacVicar, 1984; Dodt and Zieglgansberger, 1990; Dodt, 1992; Stuart et al., 1993). Long wavelength infrared light is scattered less than visible light during its passage through the slice, accounting for the improved resolution. Fine structures, such as dendrites and axons, below the surface of the slice can thus be more readily identified.

The recording chamber is normally perfused continuously with gassed (95% 0,, 5% CO,) saline by gravity or using a peristaltic pump. Continuous perfusion is important to maintain the supply of gassed saline and thus prevent anoxia and changes in pH, and also to prevent the accumulation of substances, such as glutamate released from dying cells,


in the slice. For high-resolution recordings, which might be influenced by the flow of bath solution, the perfusion can be shut off for a short time during measurements without noticeable damage to the tissue. Most recordings are made at room temperature (20-25°C) but more physiological temperatures can be attained by heating the stage and objective and by prewarming the solutions.

2.2.2. Intra- and Extracellular Solutions

Our standard extracellular solution contains (in n-t/@: 125 NaCl, 2.5 KCl, 2 CaCl,, 1 MgCl,, 1.25 NaH,PO,, 26 NaHCO,, 20 glucose (pH 7.3 when gassed with 95% O2 and 5% CO,).

Various pipet solutions can be used depending on the problem to be studied. As an example, one of our standard solutions contains (in mM): 140 KCl, 2 MgC!,, 1 CaCl,, 10 EGTA, 2 ATP, 10 HEPES (adjusted to pH 7.3 with KOH). For many applications it is important to reduce the K+ conductance of the cell being studied. K+ can be replaced by Cs+, and in some cases, in addition, in part by tetraethylammonium (TEA+) ions. A more physiological intracellular Cl- concentration can be obtained by replacing Cl- by larger anions, such as gluconate. The calcium-buffering capacity of the internal solution is a further important parameter, particularly when studying Ca2+-dependent processes or when measuring the intracellular calcium concentration ([Ca”],). An appropriate Ca2+ buffer and buffer concentration should be chosen depending on the steady-state free [Ca2+], the buffer capacity, and the speed of buffering required (see Section 2.3.). To prevent swelling or shrinkage of the cells, all solutions should have the same osmolarity (-300 mosmol).

2.2.3. identification of Neurons and Glial Cells in Slices

2.2.3.1. VISUAL IDENTIFICATION

Specific cell types on or close to the surface of slices can be identified by their location, shape, and size when viewed through the oculars of the microscope or on a video monitor. If the tissue has a clear laminar structure, cells may be lim-


ited to certain layers or one layer, e.g., Purkinje neurons, the largest neurons in the cerebellum, are found only in the Purkinje cell layer. The location and shape of the dendritic arborization or the axon may also aid identification. These structures may not always be seen easily if they are not parallel to the surface of the slice. The improved resolution afforded by infrared differential-interference-contrast (IR- DIC) video microscopy greatly increases the visibility of fine structures, e.g., dendrites under the surface of the slice, and aids identification of the cells. It is also easier to see small cells, such as interneurons, with this technique. Figure 3B shows cells in a cerebellar slice from an N-d-old rat recorded with IR-DIC video microscopy. Even with this technique, the resolution deteriorates with distance from the surface of the slice and recording is limited to the upper 40-50 pm. 2.2.3.2. LABELING WITH FLUORESCENT DYES

Another method that also may be used to identify cells, confirm their identity, or obtain more detailed information about their morphology, is staining with vital fluorescent dyes. The dye can enter the cell by diffusion from the patch pipet during patch-clamp recording, or membrane-permeable forms can enter all the cells from the bathing medium. Both of these methods require a microscope equipped for fluorescence and an appropriate set of filters for excitation and emission. With the suitable dyes, e.g., fura-2, identification can be combined with measurements of intracellular ion concentrations. Cells load rapidly with fluorescent dyes from the patch pipet following the establishment of the whole-cell configuration. Dyes, such as Lucifer Yellow and Texas Red, enter the cells rapidly and soon label even distal dendrites. Lucifer Yellow is a relatively small molecule and may pass through gap junctions if cells are electrically coupled. Figure 3A shows an example of cells in a slice from the cerebellum of a young rat (P5) loaded with the membrane-permeable acetoxymethyl fura- ester (fura- AM, Molecular Probes) applied via the bathing medium. For this type of experiment, cells were loaded with fura- AM at room temperature for


Fig. 3. Identification of neurons. (A) Cerebellar slice of a 5-d-old rat loaded from the bathing solution with fura- AM (10 PM). Granule cells are clearly visible in the outer (oGCL) and inner (1GCL) granule cell layers. In the center a Purkinje neuron with its soma m the Purkmje cell layer (XL) and dendrites extending into the molecular layer (MCL) is clearly visible. (B) Cerebellar slice of a 15-d-old rat viewed with infrared differential interference contrast videomicroscopy (IR-DIC) A pipet can be seen on the dendrite of the Purkinje neuron in the center of the image.


l-3 h immediately after slice preparation. The uptake of fura- 2 varies between cells and not all the cells in the slice are visible. At this developmental stage, Purkinje neurons with well-developed dendrites could be detected in the Purkinje cell layer (PCL). Many of the more numerous granule cells could be seen in the outer (oGCL) and inner (iGCL) granule cell layers and migrating through the molecular layer (MCL). Ester loading seems to be limited to the neurons of younger animals.

2.2.3.3. RETROGRADE LABELING

A further method for cell identification is retrograde labeling. This allows the somata of specific nerve terminals to be labeled utilizing the retrograde axoplasmic transport of dyes and other markers. Edwards et al. (1989) used the fluorescent dye Evans Blue injected into the hind leg muscles to label motoneurons in the ventrolateral column of the spinal cord. 2.2.3.4. ANTEROGRADE LABELING

Injections of dyes into a nucleus and uptake by the soma allows axons to be followed to the nerve terminals by labeling via anterograde axoplasmic transport.

2.2.4. Cleaning Procedure For many types of cell, a clean surface is necessary to

obtain the kind of high resistance seal between the patch- clamp pipet and the membrane required for high-resolution recording. This is often achieved using proteolytic enzymes. In slices, some cells on the surface are exposed and already clean enough for recording, whereas others are covered by tissue and can be cleaned mechanically without the use of enzymes. The procedure is illustrated in Fig. 4. Cleaning pipets with a steep taper and a tip diameter of -5-7 pm are pulled using soft (soda) glass (OD 2 mm). Larger pipets cause too much damage to the slice and smaller pipets soon become blocked with tissue. They are filled with normal saline and inserted into a normal patch-pipet holder that allows the application of pressure. Using a micromanipula-


cleaning pipette /9

Fig. 4. Cleaning procedure and parallel fiber stimulation. (A-C) Sche- matic rllustration of the cleaning procedure for a Purkinje neuron of the cerebellum. (A) The cleaning pipet IS positioned close to the surface of the slice above a neuron covered with tissue. (B) Gentle pressure generates a stream of solution out of the pipet that disrupts the surface of the slice. (C) Debris is sucked into the cleaning pipet. (D) Arrangement of pipets for simultaneous patch-clamp recording from the soma and parallel fiber (PF) stimulation; a climbing fiber (CF) is also illustrated.

tor, the cleaning pipet is placed close to the surface of the slice near the neuron of interest. Positive pressure applied to the pipet creates a stream of solution out of the tip, which disrupts the neuropil on the surface of the slice above and


surrounding the neuron, but without causing damage to it. The disrupted tissue can be removed by sucking it into the pipet. Tough fibrous pieces of tissue still attached to the slice can be sucked into the pipet and pulled off the slice by moving the pipet with the micromanipulator. It is important to avoid rough treatment of the cell of interest. Damaged cells rapidly become grainy or “spotty” and no longer allow the formation of seals. Cleaning is important for cells that are to be used for patch-clamp recordings together with fluorometric measurements (see later).

2.2.5. Patch Clamping Without Cleaning: “Blind” Patching

Blanton et al. (1989) described a method for patch-clamp recording in brain slices using a dissecting microscope. They moved the patch pipet with positive pressure applied through the tissue until an increase in resistance was observed. Removing the pressure resulted in the formation of high resistance seals. However, with this method it is impossible to identify the fine structure in the slice from which the recording is made, unless the cell is stained and subsequently identified. For some applications, the ability to record from structures without mechanical cleaning can have distinct advantages. This is clearly the case when recording from fine neuronal processes, such as dendrites, or when recording from small cells that may be damaged by the cleaning procedure described earlier. In addition, synaptic inputs may be disrupted during cleaning. For such applications, the patch-clamp pipet can be moved through the intact tissue under visual control using high resolution optics (e.g., IR-DIC video microscopy). A constant positive pressure applied to the pipet prevents clogging of the tip during movement through the tissue. After making contact with the cell, as indicated by an indentation of the cell membrane, the positive pressure is released and gentle suction applied until a seal is formed. This method may have the disadvantage that the pipet tip is slightly clogged with material from the cell surface, resulting in higher series


resistance values (Edwards et al., 1989). With normal illumination and even with IR-DIC video microscopy, it is sometimes difficult to identify the cell from which the recording is being made, and confirmation by dye injection is necessary. With pipet solutions containing fura- or other fluorescence indicators, the application of positive pressure to the pipet results in expulsion of the indicator from the pipet into the slice during the approach to the cell that may later interfere with the measurement of fluorescence signals from the cell. Cleaning, therefore, is preferred for this application.

2.2.6. Recording and Various Configurations of the Patch-Clamp Technique in Slices

2.2.6.1. CHOICE OF CELL

The choice of a good cell is a matter of experience. Healthy cells have a smooth surface and good contrast. Cells that have a grainy or “spotty” appearance, or that have a very high contrast tend to be damaged, probably during slicing or cleaning, and do not form seals. Other transparent, swollen cells on the surface of the slice with prominent nuclei allow seal formation but not stable whole-cell recordings.

2.2.6.2. PIPETS

The pipets used for patch-clamp recording in slices are no different from those used in other tissues. It is advantageous to have pipets with a longer shank to avoid contact with the objective when the pipets are coated with Sylgard. We use borosilicate glass (OD 2 mm, ID 1.4 mm) to make electrodes with resistances between 1 and 7 ML& The choice of resistance depends on the size of the cell and the type of experiment to be performed. Electrodes with a higher resistance make it more difficult to break through the membrane into the whole-cell mode, result in higher access resistances during whole-cell recording, and limit the exchange of substances between the pipet solution and the cell interior. For higher resistance electrodes, e.g., for dendritic recording or for single-channel recording, we use thicker walled borosilicate glass (OD 2 mm, ID 1.16 mm), which also has the


advantage of lower noise. Owing to the low angle of the pipet and the relatively high level of solution with the water immersion objective, it is important to coat the pipets with Sylgard well up the shank to reduce their capacitance. 2.2.6.3. SEAL FORMATION

The cell of interest is approached while applying a slight positive pressure to the pipet. After touching the cell, as seen by a depression in the surface and an increase in resistance, the pressure is released and a short positive pressure pulse applied to the pipet to clear the tip and the surface of the cell. Slight negative pressure is then applied until a high seal resistance (>1.5 Gn) is obtained. Sealing may be aided by the application of a steady negative potential to the patch pipet. The speed of seal formation is very variable and depends on a number of factors (state of the tissue, pipet solution, pipet size, and so on) and may occur rapidly or take a number of minutes. 2.2.6.4. WHOLE-CELL RECORDING

Whole-cell recording in slices has a large number of advantages over other methods for measurement of membrane potential and ionic currents. Compared to the single microelectrode voltage-clamp technique, the main alternative to whole-cell patch-clamp in this type of tissue, the latter is faster and has lower noise levels. The improved time resolution is of great importance when studying fast voltage-gated or synaptic currents. Lower noise levels have allowed direct quanta1 analysis to be applied to synaptic transmission in the CNS (Edwards et al., 1990). Furthermore, impalement of the cell with a microelectrode creates a leak conductance that acts as a shunt at the site of impalement and decreases the input resistance and consequently the membrane time constant of the cell. Much higher values of input resistance and larger membrane time constants have been obtained for cells in slices with the patch-clamp technique (Edwards et al., 1989; Llano et al., 1991; Spruston and Johnston, 1992; Spruston et al., 1994). The patch-clamp technique allows greater control of the intracellular solution,

Patch-Clamp Technique

because of rapid diffusion from the pipet to the cell interior, but may result in the loss of diffusible factors from the cell. The latter can be reduced by employing the perforated patch technique (Horn and Marty, 1988; see also Chapter 7). In slices, cells may have a very extensive dendritic tree, resulting in problems of space-clamp in voltage-clamp (Armstrong and Gilly, 1992). Therefore, it is important to have low-access resistance values throughout experiments, a factor that is also important for control of the intracellular solution and dye loading. Because of the difficulties of maintaining stable whole-cell recordings for long periods, pipets with resistances of <1 M&J are not used when recording from cerebellar Purkinje neurons or hippocampal pyramidal neurons. Nor- mal values are between 1 and 3 Ma, giving access resistances of between 3 and 10 Ma. The problems of space clamp must always be considered when interpreting data from cells with processes. Not only is the control of membrane potential inadequate, signals arising in distal regions of the cell are also attenuated by the cable properties of the cell (Spruston et al., 1994). In addition, the changes in membrane potential in badly clamped regions of the cell can result in the activation of voltage-gated channels. Neurons of younger animals often have less extended dendrites and allow better voltage control. Furthermore, increasing the membrane resistance by reducing the K+ conductance, e.g., when using Cs+ instead of K+ as the major intracellular cation, improves the space clamp of the cell. 2.2.6.5. SINGLE-CHANNEL RECORDING

The techniques for single-channel recording in slices differ little from those in other preparations. Cell-attached, inside-out, and outside-out patches can be formed from cells in slices and used to study both voltage- and transmitter- gated ion channels. 2.2.6.6. DENDRITIC RECORDING

It is important for the understanding of neuronal function to know more about the properties of fine neuronal structures, such as dendrites. Most synapses are located on


dendrites and there is much evidence to suggest that dendrites are not just passive conductors. Thus, there may be different patterns of expression of voltage- and transmitter-gated ion channels in distinct regions of the cell. Although it is possible to identify dendrites and record from them using the cleaning procedure described earlier (Usowicz et al., 1992), it is difficult to clean them without causing damage and difficult to identify the site of recording using conventional optical methods. Recordings are limited to the wider regions of the dendrite, e.g., in Purkinje neurons the primary dendrite and the sites of dendritic bifurcation. With a combination of patch-clamp recording and IR-DIC video microscopy the resolution is improved sufficiently to allow recordings from dendrites close to the surface of the slice under visual control without cleaning (Stuart et al., 1993; Stuart and Sakmann, 1994). The technique is similar to the modification of “blind” patching described earlier. For this technique, it is important that the processes are not damaged and that the plane of slicing allows both soma and dendrites to be identified. In the neocortex, dendrites could be followed for up to 440 pm from the soma (Stuart et al., 1993). Record- ings from such fine structures necessitate the use of fine electrodes (6-12 Ma) to reduce the amount of damage to the cell. All configurations of the patch-clamp technique are possible. For whole-cell recordings, difficulties arise when trying to break the membrane patch and gain access to the cell interior. High negative pressures are required, which, when maintained following breakthrough, result in a loss of the seal. Short, timed pulses (100-300 ms duration) of negative pressure allow a more controlled access. Whole-cell recordings in voltage-clamp are limited by the problems of space- clamp of an elongated structure with a high access resistance. The membrane potential can be recorded and current pulses applied in the current-clamp mode. Because of the high access resistance combined with the pipet capacitance, filtering and even shunting of the voltage signal may occur. It may thus be advisable to use a conventional microelectrode amplifier with capacity neutralization and a bridge circuit for current


injection (Stuart et al., 1993). The site of recording should be confirmed by dye injection. With dendritic recording, it has been possible to study the distribution of voltage-dependent Ca2+ channels in Purkinje neurons (Usowicz et al., 1992), and to measure voltage- and transmitter-gated channels in dendritic patches from neocortical neurons (Stuart et al., 1993). Furthermore, it has also been used to study the distribution of voltage-dependent Na+ channels and the site of action potential generation in neocortical neurons (Stuart and Sakmann, 1994).

2.2.7. Stimulation of Afferent Synaptic Inputs

One great advantage of the slice preparation is the pres- ervation of the in vivo structure and of many of the synaptic contacts. It is thus possible to study synaptic transmission between anatomically and developmentally defined neurons with functionally intact synaptic contacts. With an extracellular electrode, it is possible to stimulate other cells (e.g., interneurons) or afferent fibers and study inhibitory and excitatory synaptic inputs. Furthermore, the technique of simultaneous whole-cell recording from two cells (Barbour, 1993; Vincent and Marty, 1993) allows the specific stimulation of one identified cell while recording from a second. For extracellular stimulation, a second electrode, often the cleaning pipet (5-10 pm tip diameter) or a Teflon-coated platinum wire, is used as the stimulating electrode. Short (-200 ~1s) square voltage pulses (approx 10 V) can be applied between the stimulating electrode and a reference electrode in the bath from a stimulus isolation unit. In the cerebellum this has been used to stimulate the excitatory inputs to Purkinje neurons the parallel and climbing fibers (Konnerth, 1990). As illustrated in Fig. 4D, each Purkinje neuron is innervated by a single climbing fiber that can be stimulated by placing the stimulating electrode close to the Purkinje neuron in the granule cell layer (GCL) 50-100 pm away from the neuron. The exact location must be found by trial and error. The response to stimulation of the climbing fiber (excitatory postsynaptic potential [EPSP] or current [EPSC]) can be rec-


ognized by its large amplitude and its all-or-none behavior. Parallel fibers can be stimulated in the molecular layer, where they make synaptic contacts with the dendrites of the Purkinje neuron, or in the GCL from which the parallel fibers origi- nate. These responses (see e.g., Fig. 6B) are much smaller and graded in response to variations in the stimulus intensity. Inhibitory synaptic responses (inhibitory postsynaptic potentials [IPSPs] and currents [IPSCs]) can be elicited by placing the stimulating electrode close to inhibitory interneurons, e.g., basket cells, which make contacts with Purkinje neurons.

2.3. Combinations of the Patch-Clamp Techniques in Siices with Other Methods

Further important information about the processes being studied in cells using electrophysiological methods can be obtained by combining patch-clamp measurements with other techniques. These allow, e.g., the estimation of the intracellular concentration of ions and second messengers by the use of fluorescence indicators or the identification of messenger RNA (mRNA) expressed in the cell being studied by the use of the single-cell reverse transcriptase polymerase chain reaction method (single-cell RT-PCR).

2.3.1. Fluorometric Monitoring of Ca2+ and Other Ions Modern optical methods with a high spatial and tempo-

ral resolution together with appropriate indicator probes allow the measurement of the internal levels of various ions (e.g., Ca2+, H+, Na+) in different regions of the cell. The methods used in brain slices are similar to those used in other tissues and have been described in detail recently (Eilers et al., in press). This type of measurement can be performed on the setup described earlier if the microscope is equipped for fluorescence. As in other tissues, measurements of intracellular ions can be made with a high sensitivity but spatially unresolved with a photomultiplier, or of lower sensitivity but with a better spatial resolution using imaging techniques. Imaging (using, e.g., a video or CCD camera) can be used to localize changes in ionic concentration, providing important


additional information to measurements of ionic currents in cells with a very complex morphology.

1.

2.

3.

4.

There are some specific problems that arise in slices.

Cells in slices are often not in one focal plane parallel to the surface of the slice. Thus, all regions of the cell are not in focus simultaneously. The blurring of images by fluorescence coming from regions of the cell that are out of focus will reduce their quality and complicate estimates of ion concentrations. As described earlier, leakage of dye from the pipet into the slice during the approach to the cell complicates measurements by contributing significantly to the background fluorescence. Therefore, the amount of positive pressure that can be applied to the pipet is limited. There is a high level of autofluorescence contributed mostly by dead cells close to the surface of the slice. To reach a sufficiently high dye concentration in the distal dendrites within a reasonable time, it is necessary to load the soma with a high concentration of dye, which may influence intracellular ion homeostasis.

The intracellular calcium concentration [Ca2+], is particularly important for many cellular processes. Measurements of [Ca’+], in slices can provide information about Ca2+ entry through voltage-gated as well as transmitter-gated ion channels and about Ca2+ buffering and the role of intracellular Ca2+ stores. Imaging allows the subcellular localization of the Ca2+ changes and conclusions about the distribution of Ca2+- permeable ion channels and intracellular release sites. Low concentrations of indicator dyes, which disturb Ca2+ buffering within the cell only slightly, can be used to estimate the contribution of various processes to the changes in [Ca2+], occurring physiologically. High concentrations that swamp the endogenous buffer capacity allow the measurement of Ca2+ influx through voltage- (Neher and Augustine, 1992) or transmitter-gated channels (Schneggenburger et al., 1993) and an estimate of the fraction of the membrane current carried by Ca2+. Measurements of this type for transmitter-gated

Plant, Eilers, and Konnerth

channels can be performed under more physiological conditions than estimates of Ca*+ permeability from changes in reversal potential.

A combination of the patch-clamp technique with fluorescence measurements of [Ca*+], has proved important in the study of the Ca*+-dependence of synaptic plasticity in cerebellar Purkinje neurons. Simultaneous stimulation of climbing and parallel fiber inputs to a single Purkinje neuron produces a long-term depression (LTD) of the parallel fiber-mediated EPSCs (Fig. 5A; see Ito, 1989, for review). Furthermore, repetitive activation of climbing fiber synapses leads to a potentiation of the IPSCs (Fig. 5B), a phenomenon named “rebound potentiation” (RI?) (Kano et al., 1992). It could be shown that synaptic stimulation induces an increase in [Ca2+ll that is mostly limited to the dendrites (Fig. 5C) and acts as a trigger for changes in synaptic efficiency (Kano and Konnerth, 1992). Pre- vention of the increase in [Ca*+], by the inclusion of a highly effective Ca*+ buffer (BAPTA) in the pipet solution, blocked both the induction of LTD and RI?. The long-term changes in synaptic efficacy that follow appropriate synaptic stimulation could be mimicked by depolarization of the cell and activation of voltage-gated Ca*+ channels. Thus, changes in dendritic [Ca*+], trigger changes in synaptic strength, probably involving further second messenger systems, which modify the effects of excitatory and inhibitory inputs.

2.3.2. ConfocaI Microscopy Confocal laser scanning microscopy (CLSM) has a num-

ber of advantages over conventional microscopy for imaging. It has a higher lateral and axial spatial resolution and allows thin optical sections to be studied. This greatly reduces the blurring of images caused by structures that are out of focus. Confocal microscopes that are commercially available use either UV or visible light for excitation. Microscopes using visible light are limited, however, to indicators sensitive to long wavelengths, e.g., for Ca*+ to such as fluo-3, Calcium Green-l, or Fura-Red. Flu03 and Calcium Green-l do not show the shifts in excitation or emission spectrum necessary for the ratiometric estimation of [Ca”],. Fura-Red does show


800. iii- =L 600-

o dendrites

-10 -5 0 5 10 15 20

time (mln)

Fig. 5. Measurements combining patch-clamp with fluorometric CaZ+ imaging show that brief dendritic calcium signals initiate long-lasting changes in excitatory and inhibitory synaptic signals in cerebellar Purkinje neurons m slices. (A) Time course of long term depression (LTD) of responses to parallel fiber (PF) stimulation. PF axons were stimulated at 1 Hz throughout the experiment. At time 0 the climbing fiber mput (CF) was stimulated concurrently with the PF input (for eight pairings), producing a gradual decrease in EPSC amplitude (means f SD of 60 successive responses). (B) Time course of rebound potentiation (RI’) of IPSCs in response to CF stimulation (five pulses 0.5 Hz) at time 0. Each point is the mean of 100-200 consecutive IPSCs normalized to the values before conditioning. (C) [Ca*+], in the soma and dendrites measured using fura- (200 l&I) during the experiment shown in (B). For experimental details, see Kano et al. (1992) and Konnerth et al. (1992).

Plant, Eilers, and Konnerth

an excitation shift, but has a relatively low level of fluorescence and shows only small changes on Ca*+ binding. This makes it unsuitable for measurements in small compartments, such as dendrites. An advantage of longer wavelengths is the reduction of autofluorescence of the preparation. The setup used for CLSM and patch-clamp in our laboratory is similar to that described earlier except that the rigid connections between the confocal system and the microscope necessitate the use of a conventional moveable-stage microscope. A number of micromanipulators have to be mounted on the stage, making it more critical to have a mechanically-stable stage and small, light micromanipulators.

The high spatial resolution of the confocal microscope allows the measurement of [Ca*+], changes in fine structures, such as dendrites or dendritic spines. Figure 6A shows the local response of a Purkinje neuron, loaded for 45 min with Calcium Green-l, to repetitive parallel fiber stimulation. The parallel fibers were stimulated three times with a pulse of the same amplitude, giving rise to EPSCs measured by the patch pipet at the soma (Fig. 6B). Note the increase in EPSP amplitude that occurs as a result of paired-pulse facilitation. Depolarization of the imperfectly voltage-clamped distal dendrites leads to the activation of voltage-dependent Ca2+ channels and an increase in [Ca*+], in a restricted region of the dendritic tree of the Purkinje neuron.

2.3.4. Single-Cell PCR Whole-cell patch-clamp in fresh slices can be combined

with the amplification, by means of the polymerase chain reaction (PCR) of mRNA harvested from the single cell being studied, just as in organotypic slice cultures (Lambolez et al., 1992) or cultured neurons (Bochet et al., 1994; see also Chapter 9). This allows a single-cell to be characterized using electrophysiological and molecular techniques. The main advantage of fresh slices is the possibility to use identified cells at defined ages without the changes that can occur through isolation or culture. Of particular interest are the changes in expression that take place during development.


Fig. 6. Local dendritic Ca2+ signaling in a Purkinje neuron m a cerebellar slice in response to parallel fiber stimulation. Measurements were made using confocal imaging in combination with whole-cell patch-clamp recording after 45 mm of loading the cell with 500 uh4 Calcium Green-l via the patch pipet. (A) Grayscale fluorescence ratio image of F/F, (fluorescence [F] divided by the basal fluorescence [F,] before strmulatron). The position of the patch pipet on the soma is shown at the top right of the image, the position of the stimulating pipet m the molecular layer by the broken line at the left. (B) EPSCs evoked by repetitive parallel fiber stimulation (three pulses marked by arrows, 10 Hz).


The pipets should have a tip diameter large enough to allow sufficient cytoplasm to be harvested while still permitting high resistance seals to be obtained. For large Purkinje neurons in the cerebellum, pipets with resistances of l-2 MSJ are suitable; for the smaller cerebellar granule cells, resistances of -3-5 ML? are more appropriate.

Acknowledgments

This work was supported by grants from the Bundes- ministerium fur Forschung und Technologie, the Deutsche Forschungsgemeinschaft, the Human Frontiers Science Program Organization, and the European Community to A. K. We thank F. Tempia for providing unpublished data for Fig. 3A.

References

Armstrong C. M. and Gilly W. F. (1992) Access resistance and space clamp problems associated with whole-cell patch-clamping. Meth. Enzymol. 207,100-122.

Barbour B. (1993) Synaptic currents evoked in Purkinje cells by stimulating individual granule cells. Neuron 11,759-769.

Blanton M. G., Lo T. J., and Kriegstem A. R. (1989) Whole cell recording from neurons m slices of reptilian and mammalian cerebral cortex. J. Neurosci. Meth. 30,203-210.

Bochet P., Audinat E., Lambolez B., Crepe1 F. R. J., Iino M., Tsuzuki K., and Ozawa S. (1994) Subunit composition at the single-cell level explains functional properties of a glutamate-gated channel. Neu- ron 12,383-388.

Dodt H. U. (1992) Infrared video microscopy of living brain slices, in Pructicd Electrophysiologd Methods (Kettenman H. and Grantyn R., eds.) Wiley-Liss, New York, pp. 6-10

Dodt H. U. and Zieglgansberger W. (1990) Visualizmg unstained neurons in living brain slices by infrared DIC-videomicroscopy. Brain Res. 537,333-336.

Edwards F., Konnerth A., and Sakmann B. (1990) Quanta1 analysis of mhibitory synaptic transmission m the dentate gyrus of rat hippocampal slices: a patch-clamp study. 1. Physrol. 430,213-249.

Edwards F., Konnerth A., Sakmann B., and Takahashi T. (1989) A thm slice preparation for patch-clamp recordings from neurones of the mammalian central nervous system. P’tigers Arch. 414,600-612.


Eilers J., Schneggenburger R., and Konnerth A. Patch-Clamp and calcium imaging in brain slices, in Single-Channel Recording, 2nd ed. (Sakmann B. and Neher, E., eds.) Plenum, New York, pp. 213-229.

Horn R. and Marty A. (1988) Muscarinic activation of ionic currents measured by a new whole-cell recording method. J, Gen. Physiol. 92,145-159.

Ito M. (1989) Long-term depression. Ann. Rev. Neurosci. 12,85-102. Kano M. and Konnerth A. (1992) Potentiation of GABA-mediated cur-

rents by CAMP-dependent protein kinase. NeuroReport 3,563-566. Kano M., Rexhausen U., Dreessen J., and Konnerth A. (1992) Synap-

tic excitation produces a long-lasting rebound potentiation of inhibitory synaptic signals in cerebellar Purkinje cells. Nature 356, 601-604.

Konnerth A. (1990) Patch-clamping neurons in slices of mammalian CNS. Trends Neurosci. 13,321-323.

Konnerth A., Dreessen J., and Augustine G. J. (1992) Brief dendritlc calcium signals initiate long-lasting synaptic depression, Proc. N&l. Acad. Sci. USA 89,7051-7055.

Konnerth A., Obaid A. L., and Salzberg B. M. (1987) Optical recording of electrical activity from parallel fibres and other cell types m skate cerebellar slices “m vitro.” I, Physiol. 393,681-702.

Lambolez B., Audinat E., Bochet I’., Crepe1 F., and Rossier J. (1992) AMPA receptor subunits expressed by single Purkinje cells. Neuron 9, 247-258.

Levis R. A. and Rae J. L. (1992) Constructing a patch-clamp setup. Mefh. Enzymol. 207,14-66.

Llano I., Marty A., Armstrong C. M., and Konnerth A. (1991) Synaptic- and agonist-induced excitatory currents of Purkinje cells in rat cerebellar slices. 1. Physiol. 434,183-213.

MacVicar B. A. (1984) Infrared video microscopy to visualize neurons in the in vitro brain slice preparation. I. Neurosci. Met/z. 12,133-139.

Neher E. (1992) Correction for liquid junction potentials in patch-clamp experiments. Meth. Enzymol. 207,123-131.

Neher E. and Augustine G. J. (1992) Calcium gradients and buffers in bovine chromaffin cells. J. Physiol. 450,273-301.

Regehr W., Konnerth A., and Armstrong C. M. (1992) Sodium action- potentials in the dendrites of cerebellar Purkinje cells. Proc. N&l. Acad. Sci. USA 89,5492-5496.

Schneggenburger R., Zhou Z., Konnerth A., and Neher E. (1993) Frac- tional contribution of calcium to the cation current through glutamate receptor channels. Neuron 11,133-143.

Spruston N., Jaffe D. B., and Johnston D. (1994) Dendritic attenuation of synaptic potentials: the role of passive membrane properties. Trends. Neurosci. 17,161-166.


Spruston N. and Johnston D. (1992) Perforated patch-clamp analysis of the passive membrane properties of three classes of hippocampal neurons. I. Neurophysiol. 67,508-529.

Stuart G. J., Dodt H. U., and Sakmann B. (1993) Patch-Clamp recordings from the soma and dendrites of neurons in brain slices using infrared video microscopy. Pfltigers Arch. 423,511-518.

Stuart G. J. and Sakmann B. (1994) Active propagation of somatic action potentials into neocortical pyramidal cell dendrites. Nature 367, 69-72.

Usowicz M. M., Sugimori M., Cherksey B., and Llinas R. (1992) P-type calcium channels in the somata and dendrites of adult cerebellar Purkinje cells. Neuron 9,1185-1199.

Vincent P. and Marty A. (1993) Neighboring cerebellar Purkinje cells communicate via retrograde inhibition of common presynaptic interneurons. Neuron 11,885-893.

Xenopus Oocyte Microinjection and Ion-Channel Expression

T. G. Smart and B. J. Krishelc

1. Introduction

1. I. History of the Xenopus laevis Oocyte as an Expression System

It is now over 20 years since the seminal studies by John Gurdon and colleagues established that Xenopus Zaevis oocytes, when injected with messenger RNA (mRNA), were able after a period of incubation to translate the mRNA and appropriately synthesize the relevant protein (Gurdon et al., 1971; Gurdon 1974). In this study, rabbit reticulocyte 9s mRNA was injected and the oocytes produced globins. This important observation led to an amazing variety of proteins being expressed in Xenopus oocytes following injection with mRNAs extracted from different sources, including, for example, viral (adenovirus, mouse mammary tumor virus) and plant mRNAs (barley and maize), and also invertebrate (locust muscle, honey bee), and vertebrate tissue mRNAs (cat skeletal muscle, mouse kidney, rat spleen, Torpedo electric organ; see Lane, 1983; Colman, 1984; Soreq, 1985 for reviews). The value of the Xenopus oocyte for in vitro translation studies is now recognized by the oocyte’s ability to correctly assemble proteins composed of individual subunits and also to ensure the appropriate posttranslational processing of the protein, i.e., insertion into the cell membrane or secretion of the protein product (Lane, 1983; Soreq, 1985).

From Neuromethods, Vol 26 Patch-Clamp Apphcatlons and Protocols Eds A Boulton, G. Baker, and W. Walz 0 1995 Humana Press Inc

259

260 Smart and Krishek

1.2. Application to Ion-Channel Expression Studies

Following the early studies of protein expression in Xenopus oocytes, it was some years later that the value of this preparation was recognized for expressing nicotinic acetylcholine receptors after injecting oocytes with cat skeletal muscle mRNA (Sumikawa et al., 1981; Barnard et al., 1982). This work was quickly followed by the important demon- stration that mRNA extracted from the central nervous system (CNS) of different species and injected into Xenopus oocytes could also result in the synthesis of nascent neurotransmitter-gated receptors and also voltage-gated ion channels (Miledi et al., 1982; Smart et al., 1983, 1987; Gundersen et al., 1984; Houamed et al., 1984; Sigel, 1990). These early observations forged a pathway for many investigators to use the Xenopus oocyte for the expression of ion- channel proteins, and this has been the subject of detailed reviews (Sumikawa et al., 1986; Dascal, 1987; Smart et al., 1987; Snutch, 1988; Sigel, 1990).

In addition to the expression of ligand and voltage-gated ion channels for study predominantly by electrophysiological methods, the Xenopus oocyte has also been a useful addition to the molecular biologist’s armamentarium for the isolation of cDNAs encoding for additional ion-channel proteins by the method of expression cloning (Snutch, 1988; Frech and Joho, 1992). Notable examples of receptor and ion-channel proteins whose primary DNA sequence has been established by this method include: 5-HTlc receptor (Lubbert et al., 1987), a neuropeptide receptor (Masu et al., 1987), a non- N-methyl-o-aspartate (NMDA) receptor (GluRl; Hollmann et al., 1989), and the first examples of the NMDA receptor family (NRl; Moriyoshi et al., 1991), and metabotropic glutamate receptor family (Masu et al., 1991) in addition to two types of potassium ion channels (Takumi et al., 1988; Frech et al., 1989).

Given the importance of the Xenopus oocyte to the study of recombinant ion-channel proteins, the aim of this review is to provide methodological details that we have accrued in

ion-Channel Expression 261

our laboratory over the last 12 yr for the care and maintenance of Xenopus and the use of the oocytes for the injection of RNA and DNA. We also describe how electrophysiological techniques can be used to study the expressed ion-channel proteins. Many of the methodological details have been assembled empirically from our desire to optimize the expression of receptor proteins. Finally, we try to assess the current position of the Xenopus Zaevis oocyte as a viable transcription/translation system in comparison to other more recent cellular-based in vitro expression systems.

2. Husbandry of Xenopus laeois

2.1. Source and Identification of Xenopus laevis

Xenopus laevis is often colloquially referred to as the South African clawed frog or, occasionally, toad, and is typi- fied by the presence of up to three claws present on each of the hind limbs of mature frogs. These frogs are members of the family Pipidae and are classified as Anurans (Verhoeff- de Fremery and Griffin, 1987). The frogs are endemic in South Africa, Botswana, and South West Zimbabwe. There are many suppliers (e.g., Blades Biological, Edenbridge, Kent, UK) that can readily provide wild-type and laboratory-bred frogs. We prefer to use wild-type females and always specify to the supplier to deliver the largest, most mature frogs, although laboratory-reared animals will also produce viable oocytes for microinjection (Goldin, 1992). The larger frogs (mass 100-200 g) tend to have more oocytes than smaller frogs (~80 g) and of these oocytes a greater proportion will be of developmental stages IV-VI (see Section 4.1.). Xenopus laevis can produce oocytes of differing stages of maturity all year round, although in “new” frogs sometimes the quality of oocytes for expression studies can be poor. To circumvent this problem we allow at least a 6-wk acclimatizing period before the “new” frog will be considered for oocyte extraction. Typically, the natural breeding season for Xenopus is from October to the end of December and recently acquired “wild” frogs will continue this cycle such that the number of


Fig. 1. Magnlfled view of a Xenopus hind limb showing the posltlon and securement of a numbered and colored identification tag on the middle toe.

stage V and VI oocytes can be quite low in the summer months. It is therefore important to plan ahead and bring in new frogs at least 3-6 mo before they are required for use. The “new” frogs are quarantined to guard against disease spread throughout the established colony under conditions of constant environment (see Section 2.2.) to disrupt the natural breeding cycle, thereby ensuring a supply of different stage oocytes throughout the year.

The frogs can live for up to 25 yr; our longest surviving example has attained the age of 8 yr. With this longevity, it is important to be able to identify the frogs. Each Xenopus has a characteristic series of marks on the dark dorsal surface; however, when in a tank of water it is often difficult to use this criterion reliably for identification. We have labeled our frogs by suturing an electrical cable marker around one of the middle toes of a hind limb. These markers are numbered and colored and easily visualized under water (size 1, Cat. No. 666470, Radio Spares, Corby, Northants, UK). The suture forms a loose loop and is made of nylon so that it does not shrink in water and ulcerate the toe (Fig. 1). This method has

Ion-Channel Expression 263

the advantage that, apart from the suture breaking, it is permanent, unlike toe clipping, which needs to be repeated every 2 mo. Skin autografts are another method of permanent labeling. This is a more invasive procedure for the frog and visualization under water is often not clear. A more difficult aspect of identification concerns the difference between male and female frogs. Generally, the female is slightly larger in length compared to the male (9.5 to 7.5 cm snout to vent length; Verhoeff-de Fremery and Griffin, 1987) and the cloa- cal valves are usually more prominent.

2.2. Housing and Environment

Xenopus are aquatic air-breathing animals that will rapidly suffer from dehydration and eventual death if denied access to water. It is desirable to house the frogs in small colonies of approx 7-10 in a water tank fabricated from glass or Perspex (dimensions 60 cm length x 40 cm height x 30 cm width) containing approx 40 L of water. One tank is usually reserved to quarantine newly purchased frogs before mixing with the established colonies. The temperature of the tank water can be maintained by thermostatically controlling the room temperature between 18 and 22°C and a filter pump will keep the water clean from particulate debris. The pump filters must be cleaned daily and the tanks cleaned l-2 times per week to remove particles and excreted urea; however, ensure that not more than two-thirds of the water in each tank is changed at any one time to minimize the risk of causing serious stress to the animals. The rate of water pumping should be sufficiently low to avoid causing the appearance of red leg (see Section 2.4.1.). Some gravel, plastic pond weed, and containers can be installed in each tank to more closely replicate the nocturnal environment of the frog, but this seems to have little effect on the quality and production of oocytes.

Since Xenopus skin is sensitive to chloride ions, the water used to fill the tanks should be taken from stored water contained in large open tanks for at least 24 h, which allows the chlorine to evaporate. Alternatively, if water is rapidly required, aeration of tap water with an airstone will remove


chlorine. It is not necessary to routinely aerate the main tank water. To prevent the escape of Xenopus, the tanks must be enclosed with a weighted wire mesh or a V-shaped aluminum lid that fits onto the top of the tank. The strength of the frogs in escaping should not be underestimated.

2.2.1. Lighting On receipt, the frogs are routinely introduced to an arti-

ficial lighting cycle of 12 h illumination and 12 h darkness that is maintained throughout the year to help break the natural breeding cycle. This is achieved simply by placing a timing switch on the main room lights and ensuring that natural light via the windows is excluded if the frogs are housed in a dedicated room, or alternatively, the tank walls can be painted to prevent the ingress of light. It is not desirable to place the light source above and close to the top of the tank, since this seems to introduce unacceptable stress in the frogs, resulting in the shedding of skin and marked secretion of mucus.

2.2.2. Handling

As a reaction to any type of stress, the frogs will exude a slippery mucus secretion that makes handling difficult, particularly if the handler is wearing standard latex-based laboratory gloves. Therefore, we use a small aquarium net sufficient to capture a large frog swimming in the tank with minimal stress. The captured frog is removed from the tank, keeping one hand over the top of the net. Frogs can easily escape by jumping even from the net. To handle the frog it is best to use a rough cloth or piece of muslin and place the first two fingers of your hand between the hind limbs of the frog with the rest of the animal pointing toward your wrist. In this position, it is possible to keep a firm grip.

2.3. Feeding

We feed our established colonies twice weekly and for new frogs two to three times weekly with an alternate combination of fat-free lamb heart and liver. Healthy Xenopus


have a good appetite and start eating immediately as food is placed into the tank. After 30 min, most of the food should be consumed. Overfeeding should be avoided, and generally we would add approx ZOO-300 g of lamb heart or liver to a tank housing 7-10 frogs on each occasion. Any uneaten food is removed manually with a net or removed by the continuous slow recirculation in the filtration system. If diet sup- plementation should be necessary, for example, following illness to the colony, live crickets (A&eta domestica) can be placed in the tank; these float on the surface and are readily eaten by the frogs coming up to the water surface. Occasion- ally, we have rolled the live crickets in a multivitamin pow- der (Blades Biological) prior to placing them in the tank. This schedule would occur once a week until the frogs regain their normal healthy status.

2.4. Diseases and Parasites

To keep the Xenopus colonies in good condition, hygienic procedures, regular and careful observation, and inspection of the animals and gentle handling are necessary. We have not experienced many problems with diseases, since Xenopus are hardy animals with a relatively low annual death rate (approx 2%). However, when any animal shows symptoms of disease (e.g., whitening of one or both eyes, lack of feeding) the animal is isolated in a separate tank and carefully observed for changes in condition. After removing a dead or diseased frog from a tank, the rest of the colony are temporarily housed while the tank is scrubbed clean with copious amounts of hot water. The colony is reintroduced and then monitored over the next 2 wk. To prevent the spread of disease, all tanks should have separate nets and equipment and must be cleaned separately.

2.4.1. Red Leg The most prevalent ailment is the condition of red leg,

which largely results from poor environmental conditions (dirty tank water, inappropriate and poor diet, widely varying water temperature), allowing multiplication of


opportunistic bacteria. The cardinal signs are subcutaneous hemorrhaging localized to the limbs, the abdomen, and also around the mouth. Any Xenopus demonstrating subcutaneous pooling of blood is isolated immediately and treated by the addition of sulfamenazine (sodium salt) at approx 1 g/L tank water (Verhoeff-de Fremery and Griffin, 1987). An additional treatment, described by Verhoeff-de Fremery and Griffin (1987) is to use the antibiotic, aureomycin, 12 g mixed with minced bovine heart, gelatin, and water (500 mL) to form an amorphous mass. In our experience, the frogs do not avidly consume this mix and we prefer to use the sulfamenazine treatment, which can rescue even quite badly affected frogs.

2.4.2. Leeches Parasite infestation is usually quite rare but can occur

particularly with newly acquired Xenopus. One example is infestation with leeches. These small black creatures are attached to the ventral surface but can occasionally be found on the hind limbs, toes, and main dorsal surface. They attach to the frog skin by suction and cannot be dislodged by holding the frog under running water. The easiest approach is to hold each frog and manually remove individual leeches using Watchmaker’s forceps. This process may have to be repeated after 24-48 h, since some leeches are occasionally missed on the dark-colored dorsal surface, but eventually all the parasites can be removed without damage to the skin. If the parasites are left in place they may puncture the skin, allowing opportunistic bacterial infection.

2.4.3. Cold Water Worm infestation This disease is also quite rare, but can devastate the frog

colony. The nematodes, believed to be Pseudocapillaroides xenopi, are too small to be seen easily and will burrow into the epidermis, causing a parasitic dermatitis manifest as a roughening of the skin with a patchy appearance. The frogs lose weight with skin ulceration and secondary bacterial infection causing septicemia. On postmortem examination the frogs exhibit chronic inflammation in the dermis and epi-


dermis and the presence of small nematodes. The microorganisms isolated are Pseudomonas, Aeromonas species, Citrobacter fieundii, and Klebsiella pneumoniae. Antibacterial treatment is initiated with tetracycline applied orally using a l-mL syringe inserted gently into the mouth (5 mg/30 g body weight twice daily for 30 d; Panmycin Aquadrops, Upjohn Ltd., Crawley, W. Sussex, UK; see Section 2.2.2. for handling). The worm infestation is treated with a subcutaneous injection of the anthelmintic, levamisole hydrochloride (5-10 mg/kg body weight, Nilverm, Coopers Animal Health Ltd., Crewe, Cheshire, UK). This injection is repeated 10 and then 20 d after the initial dose (Cunningham, Sainsbury and Cooper, unpublished observations, Institute of Zoology, London). Frogs suffering with multiple skin lesions usually continue to deteriorate and die, whereas other less affected frogs recover uneventfully. The tanks housing the infected colony are cleaned and disinfected with chlorhexidine gluconate (1 in 200) in water before being washed thoroughly with hot water to remove the disinfectant.

3. Removal of Ovary Tissue

A Xenopus female contains about 30,000 large diameter (>1 mm) oocytes when fully mature. Since a typical electrophysiological experiment rarely requires more than 50-100 oocytes, the following procedures were designed for the limited removal of oocytes under anesthetic.

3.1. Anesthesia of Xenopus laevis

1.

2.

3.

Select a large mature Xenopus female and place the frog in a separate small tank (30 cm length x 13 cm width x 13 cm height) containing tap water. Remove the tap water and replace with approx 1000 mL of 0.2% w/v solution of ethyl-m-aminobenzoate (Tricaine, Sigma, Cat. No. A-5040), ensuring the frog is completely immersed. The level of anesthesia should be checked every l-2 min. This can be determined by either the loss of the withdrawal reflex while pinching the toes of a hind limb or


preferably, the frog will demonstrate loss of the “righting-reflex” when sufficiently anesthetized and turned ventral side up.

Anesthesia is usually obtained within 8.5 + 3 min (n = 50 frogs) after immersion in 0.2% Tricaine. We often find that if the frog has been anesthetized previously, the time period for anesthesia may need to be slightly extended (to 15-20 min).

4. Anesthesia can be maintained by cooling the frog using ice (see Section 3.2.).

3.2. Removal of Oocytes

Throughout the following procedure, distilled water should be sprayed frequently onto the animal to prevent drying of the skin and to keep the site of surgery clean and free from exudates, e.g., blood. Furthermore, all bench surfaces should be swabbed with 70% v/v ethanol and all surgical instruments should be sterilized.

1. Transfer the Xenopus and place dorsal side down on a flat bed of ice in a shallow tray to maintain anesthesia.

2. Illuminate the sight of operation with an optic fiber light source (cold light).

3. With a scalpel make a small transverse incision of 3-5 mm (Swann-Morton Surgical Blades No. 11) through the outer layer of the skin on the lateral ventral surface of the frog. This procedure can be facilitated by placing a finger on either side of the operation site, ensuring the skin is pulled taut. Because the skin of the amphibian is quite tough, “light-scoring” of the skin can facilitate the incision (Fig. 2A).

4. To reach the ovary wall, another incision should be made through the connective tissue and subsequently through the muscle sheet in the same manner. While making this second incision, damage to the frog’s internal organs must be avoided by lifting up the exposed muscle sheet using forceps. The ovary wall and the “dark” oocytes should now be visible (Fig. 2B).


5.

6.

7.

8.

9.

10.

11.

12.

Using a fine pair of forceps (Watchmakers No. 5) a section of the ovary can be teased carefully out onto the surface of the frog (about 1-2 cm in length; Fig. 2C). Separate the small section of ovary from the remaining ovary tissue using a small pair of scissors and transfer to a Petri dish containing a modified Barths solution (MBS; Appendix 1) maintained on ice. Additional portions of tissue can be removed in this manner until sufficient tissue containing the desired number of oocytes has been obtained. After collection, decant the medium and replace with fresh cold MBS. The skin and muscle layer should be sutured separately using sterile polydioxanone (clear) monofilament suture (Ethicon Ltd., Edinburgh, UK), which degrades gradually over a period of several weeks. Typically, the tissues are repaired using up to four to five stitches on the inner muscle layer and four to six stitches on the outer skin layer (Fig. ZD-F). Wash the frog in warm tap water to remove any trace of the anesthetic or exudates. Recovery of the Xenopus should occur under subdued lighting in shallow water on a slope formed from wet tissue paper, which ensures that the head remains above water to prevent drowning while anesthetized. The Xenopus should regain consciousness in normally 42 f 18 min (n = 50), after which the tissue paper can be removed. The rate of recovery depends on the ambient temperature and can be significantly accelerated by using warm tap water (approx 24°C). The animal can be returned to the colony 4-6 h later and should remain under observation over the next 24-48 h. A 6-8 wk recovery period should follow before selection for reuse.

3.3. Preparation of Oocytes for l@ection

To increase the long-term survival of the oocytes, the ovary tissue should be washed thoroughly in MBS shortly


after removal from the frog. This removes any debris and dilutes enzymes (from damaged cells) that may damage the integrity of the oocytes. The oocytes should be maintained at low temperatures in MBS (approx 5OC) throughout all subsequent procedures. A stereomicroscope (x10 magnification unless specified otherwise) was used to facilitate all manual dissections.

3.3.1. Separation of Follicular OocyteS Place a clump of oocytes in a lo-cm sterile dish filled with

MBS. The outer ovarian epithelial layer (Fig. 3A) is carefully removed, exposing the underlying oocytes (Fig. 3B). Using one pair of sterile Watchmaker’s forceps grip a clump of oocytes while carefully stripping off individual oocytes from the clump with another pair of forceps. Each oocyte is contained in a small ovarian sac composed of epithelial cells (inner ovarian epithelium; Fig. 4; Dumont and Brummett, 1978); the technique is to grasp the neck of the sac with the forceps while trying not to touch or damage the oocyte. The success and ease of this method will vary from batch to batch of oocytes. Separated oocytes should be transferred, using a blunt and firepolished Pasteur pipet, to a vial or another dish filled with fresh MBS placed on ice.

3.3.2. Manual Defolliculation Manual defolliculation of the oocyte can be achieved at

the same time the oocyte is separated from the clump of ovarian tissue. Watchmaker’s forceps (No. 5) are used in a manner similar to the method described above except that the forceps should be held nearer to the oocyte surface, whereas the second pair of forceps is used to gently pull the epithelial and follicular layers off from around the oocyte (Fig. 4). This method does not always result in the complete removal of the follicular cell layer. A stereomicroscope at a more powerful magnification (x20) is recommended. 3.3.2.1. DEFOLLICULATION USING POLY-L-LYSINE

An alternative method to remove follicular cells was devised by Woodward and Miledi (1989). Oocytes with their


inner epithelial membrane removed manually are rolled across a glass microscope slide previously coated with poly- L-lysine and immersed in MBS, frog Ringer, or Ca*+-free frog Ringer (Woodward and Miledi, 1989). The follicular cells are shed, leaving defolliculated oocytes retaining the vitelline membrane. To achieve complete defolliculation, oocytes must be rolled across the glass slide several times.

3.3.3. Defolliculation Using Collagenase Defolliculated oocytes can also be obtained by enzymatic

treatment with l-2 mg/mL collagenase (approx 0.2-0.5 U/ mL; type lA, Sigma or collagenase A, Boehringer Mannheim, Mannheim, Germany) in a Ca*+-free Ringer for 60-120 min at room temperature (18-20°C). Removal of the follicle cells is facilitated by gentle shaking during enzyme treatment. Teasing apart large clumps of oocytes will aid the action of collagenase, and it is important to wash the oocytes twice in Ca*+-free Ringer prior to and after enzyme treatment before returning the oocytes to fresh MBS. The washing in Ca*+-free Ringer will encourage greater oocyte viability. This procedure using collagenase has the advantage over manual defolliculation that many oocytes can be treated at the same time, but the procedure can also be detrimental to long-term oocyte survival. A less damaging method is to use collagenase for shorter periods of time to achieve only partial removal of follicular cells that may then be completed manually. This treatment requires collagenase l-2 mg/mL for 15-45 min. If possible, follicular oocytes should be selected preferentially for injection since enzymatic treatment or manual dissection causes the defollicular oocytes to become more fragile and susceptible to damage.

3.4. Remooal of the Vitelline Membrane

The removal of the vitelline membrane from defolliculated oocytes is not necessary for injection of RNA or DNA but it does facilitate studies where the plasma membrane of the oocyte must be exposed and relatively clean (Fig. 4). This situation occurs for patch-clamp studies of expressed ion


channels. Removal of the vitelline membrane can be facilitated by placing oocytes in a hypertonic medium (400- 500 mOsm; solution is also referred to as a “stripping-solution,” Appendix 1; Methfessel et al., 1986). The hypertonic solution causes the oocyte to shrink within the vitelline membrane in approx 5-10 min. The vitelline membrane can then be removed carefully from the plasma membrane using two pairs of fine forceps. The vitelline membrane is translucent and appears as a collection of “glass fibers” as it is removed. The oocyte is then washed gently in MBS. Once the vitelline is removed, the oocyte is extremely fragile and care must now be taken to insure the oocyte is contained in solution at all times and not exposed to an air/solution interface where it will disintegrate. Transfer of devitellinized oocytes is achieved using a blunt Pasteur pipet, ensuring the pipet is kept horizontal to avoid the oocyte reaching an air/solution interface. Once placed in the recording bath, immediately ensure the correct position, since the oocyte will quickly attach itself to any surface and subsequent movement will result in rupture of the plasma membrane with leakage of yolk platelets.

4. Selection of Oocytes It is possible to express exogenous RNA and DNA in all

the different developmental stages of the Xenopus oocyte (I- VI). Invariably, fully grown immature oocytes at stages V and VI (Dumont, 1972; Fig. 3B) are preferred since they are the largest cells, frequently enabling the development of large membrane currents and can be injected with up to 100 nL of solution. However, for electrophysiological experiments other considerations are important, including the membrane capacity of the expression system. The large membrane capacity transients observed in stage V and VI oocytes can take many milliseconds to decay and this can obscure the fast kinetics of activation of some membrane currents, e.g., the voltage-sensitive sodium current. It was for this reason that receptor/ion-channel expression in oocytes of earlier stages (II to III; Fig. 3B) was investigated where the size and


Fig. 2. Close-up view of the lateral ventral abdominal surface of a Xenopus showing the various stages involved in oocyte extraction (A- C) and subsequent repair (F) by suturing both internal (D) and external (E) tissues. Calibration bar 1 mm. Viewed through a Nlkon SMZ-1 microscope.

membrane capacity of the oocyte is reduced, providing faster decays to the capacity transients following brief voltage steps under voltage clamp (Krafte and Lester, 1989). The smaller oocytes may also have less prominent calcium-activated chloride currents (ICICcaJ compared to oocytes at stages V-VI. This would be an advantage when this endogenous current con- taminates recordings of expressed voltage-gated potassium

Figure 6


and calcium currents; however, a prominent Icl(caJ could be a necessary asset if G protein-linked receptors are being expressed, which often results in the liberation of phospho- inositides and activation of Icltca) (Krafte and Lester, 1989). The smaller size of stage II-III oocytes will only allow up to 15-20 nL of solution to be injected.

4.1. Stages of Oocute Deoelopment

To separate the various stages of the oocytes, it is necessary to observe some physical characteristics as detailed by Dumont (1972; Fig. 3B). Oocytes at stage 1(50-300 pm diameter) are translucent with no animal or vegetal pole but with a clear nucleus. Stage II is characterized by a diameter of 300-450 vrn with a white cytoplasm toward the end of this stage of development, but no clear polar differentiation. Stage III oocytes are usually 450-600 pm in diameter and exhibit a blackish-brown patchy coloration; some polarity is just becoming evident. Stage IV shows separation into animal and vegetal poles, with diameters in the range 600- 1000 pm. Stage V oocytes (diameter 1-1.2 mm) exhibit clear differentiation of animal and vegetal poles, with the animal pole appearing lighter in color compared to oocytes at stage IV. At stage VI, the oocytes reach maturity and have

Fig. 3. (previous page) A clump of oocytes before (A) and after (B) the removal of the outer ovarian epithelial layer. Various examples of the different developmental stages (I-VI) of the oocytes with their characteristic physical appearance and size are illustrated. The distinctive hemispheres corresponding to the dark animal poles and much lighter vegetal poles in oocytes at stages IV-VI are clearly visible. (Cl Char- acteristic appearance of damaged or poor oocytes. Note the patchy and dull appearance. Calibration bar is 0.5 mm in (A) and (B) and 0.2 mm in (C). Viewed through a Nikon SMZ-1 microscope.

Fig. 6. (previous page) Stage V oocytes placed in a plastic mesh and shown before (A) and after (B) centrifugation. Note the appearance of the nuclei in (B) which is illustrated for one oocyte magnified in (C). After injection, healthy oocytes should have the appearance of the example in (D). Note the central spot indicating the position of impalement by the injection micropipet. Calibration bar is 0.5 mm in (A) and (B), and 0.2 mm in (C) and (D). Viewed through a Nikon SMZ-1 microscope.


VEGETAL POLE

ANIMAL

Inner ovanan eplthellum

Follicle cell

Oocyte plasma membrane

Theta

Vltelllne envelope

NUCLEUS

Fig. 4. Schematic diagram of a Xenopus laevis oocyte illustrating the various anatomical features including the cell and membrane layers surrounding the oocyte plasma membrane.

diameters of 1.2-1.3 mm. The differentiated animal and vegetal poles are distinctly separated by an unpigmented equatorial band compared to stage V oocytes. At all the different stages of development (except stage I) oocytes will possess a vitelline membrane.

4.2. Separation of Stage Vfll Oocytes

If stage V and VI oocytes are to be chosen for injection with RNA or DNA, then oocytes with diameters >lOOO pm that have a distinct boundary between the hemispheres should be selected. The technique of oocyte selection becomes a routine matter for the experienced operator, who is even able to quickly select oocytes “by eye.” Until this confidence is acquired, the following procedure will be useful.

1. Cut the tip of a Pasteur pipet so that the orifice is approx 3 mm wide. Lightly flamepolish the tip.

2. Draw up about lo-20 oocytes along with 1.5 mL of MBS into the cool Pasteur pipet.

Ion-Channel Expression

3. Hold the pipet vertically and allow the larger, more dense oocytes to fall to the bottom and back into the Petri dish. The less dense, immature stage oocytes will remain in the pipet and should be discarded.

4. This procedure should be repeated twice more and then the remaining large oocytes can be transferred into a fresh lo-cm dish filled with MBS.

5. Finally, examine each oocyte according to the aforemen- tioned criteria, including the approximate diameter; presence of clear animal and vegetal poles; presence of an equatorial band/no signs of damage or patchy gray membranes; yolk platelet leakage; or attached ovarian tissue (Fig. 3C). These oocytes should be suitable for microinjection, and any oocytes not meeting these criteria should be discarded.

5. Microiqjection of Xenopus Oocytes 5.1. IrQection Equipment

It is possible to inject Xenopus oocytes with RNA or DNA using quite simple experimental equipment that will include a stereomicroscope with variable magnification and a good depth of field to facilitate focusing on an oocyte and injection micropipet simultaneously (e.g., Nikon SMZ-1; Nikon, Kingston-upon Thames, London, UK); a coarse micromanipulator (Prior, Micro Instruments, Oxford, UK) permitting the smooth vertical and horizontal movement needed to fill the micropipet and inject oocytes; a Drummond lo-PL digital microdispenser (Laser Lab System Ltd., Southampton, UK) providing an accurate volume delivery in the nL range; sterile glass tubing for fabricating the micropipets (Laser Lab System Ltd.; Drummond Scientific, Bromal, PA); and an optic fiber light source providing high-intensity but cold illumination (Schott KL1500, twin optic-fiber lamp; Nikon) to avoid desiccation of the oocytes during injection. It will be necessary to purchase (Drummond) or manufacture a clamp to firmly hold the pipet to the manipulator of choice. For the microinjection of oocytes, a number of experimental arrange-

The glassware used at any stage during the injection procedure must be sterile, and for the injection of RNA it is important that glassware and other surfaces should be free of RNases that can rapidly degrade RNA and result in an apparent failure of protein expression after injection. The following protocol was designed to avoid these problems:

1.

2. 3.

4.

5.

Prepare a solution of 0.1% v/v diethylpyrocarbonate (DEW, Sigma) in distilled water and sterilize by autoclaving at 121°C for 15 min. This will be referred to as DEPC-treated water. Swab all bench surfaces with 70% v/v ethanol. Prepare a fresh solution of 0.1 v/v DEPC in distilled water. Place the glassware in the DEPC solution for 24 h. This removes all protein traces from the glassware. Discard the DEPC solution and rinse the glassware with DEPC-treated and subsequently just sterilized water to remove all traces of DEPC. Place the individually-wrapped glassware in an oven and sterilize with dry heat at 220°C for at least 4 h to destroy RNases and any remaining DEPC on the surface of the glass.

To avoid the reintroduction of RNases on injecting RNA,


ments have been reported (Gurdon, 1974; Contreras et al., 1981; Burmeister and Soreq, 1984; Colman, 1984; Hitchcock et al., 1987).

5.2. Preparation of Glassware and RNA/DNA Solutions

sterile surgical gloves should be worn throughout all subsequent handling of glassware and aseptic techniques should be observed. Prior to the injection it is important to be assured that all the stock solutions of RNA and DNA are relatively free from detergents and salts since these agents can limit oocyte viability. RNA or DNA solutions should be prepared in sterile water and can be stored at -70°C (for RNA) and -20°C (for DNA).

The concentrations used can vary. For mRNA preparations we have used up to 1 mg/mL; for cRNA, 100 pg/ mL-1 mg/mL; and for cDNA, 30 pg/mL-1 mg/mL.


Fig. 5. Schematlc diagram of a Xenopus laevis oocyte illustrating the various anatomical features including the cell and membrane layers surrounding the oocyte plasma membrane.

5.3. Fabrication of Irljection Micropipets

Using the Drummond microdispenser for microinjec- tions, we use thin-walled borosilicate sterile glass tubing (see Section 5.2.; outer diameter 1.17 mm, internal diameter 0.68 mm, length 20 cm) and a vertical microelectrode puller (Model 730, David Kopf Instruments, Tujunga, CA).

1.

2.

3.

4.

The micropipets are pulled using a single heat setting. Typical values for the Kopf electrode puller are: Heat 12.4-12.8 with the solenoid set at approx 2. Place the tubing asymmetrically in the puller such that one micropipet will be 12-14 cm long. The long micropipet is then bent by 30” in a small Bun- sen flame approx 3-5 cm from the tip, which allows vertical positioning of the pipet above the oocyte (Fig. 5). Fix the micropipet in a sterile Petri dish on plasticine and use blunt forceps to break back the tip, producing an enlarged external tip diameter of approx 20 ym. This procedure is facilitated by heating the tip of the forceps in a Bunsen flame prior to touching the micropipet. A fresh micropipet is fabricated before each injection.


5.3.1. Filling the Micropipet with RNA or DNA 1.

2.

3.

4.

5.

Allow the RNA or DNA solution to thaw at room temperature. Spin down the RNA or DNA in an Eppendorf microfuge for approx 20-60 s to ensure all the sample is collected at the bottom of the Eppendorf tube. Store the Eppendorf vial containing the RNA or DNA on ice until it is required for use. Fill a l-mL sterile disposable syringe with mineral oil (light white oil, RNase and DNase free, Sigma) containing a small amount (sufficient to impart a distinct red color) of Sudan IV (Solvent Red 24, Sigma), an oil-soluble dye. The dye is incorporated to facilitate the visualization of the interface between the RNA or DNA and the mineral oil used in the micropipet. Place a sterile spinal needle (25 x 2.5 mm, Monoject, Sherwood Medical, Balleymoney, Northern Ireland) onto the end of the l-mL syringe and expel the air from the needle.

6.

7.

8.

9.

Insert the syringe needle fully into the wide end of the glass micropipet and fill with approx 4 cm of the oil/ dye solution. Place the micropipet fully into the Drummond microdispenser. It is essential to avoid the introduction of air bubbles during assembly. If air is introduced, then a new pipet must be pulled and the procedure repeated. Ensure the collet is tightened. By placing the pipet onto the microdispenser, the oil solution will be forced down toward the tip. Mount the Drummond microdispenser onto the micromanipulator and secure. The Drummond microdispenser can eject volumes as low as 10 nL and at 50 nL has an accuracy of +5%. Eject any air bubbles at the tip of the micropipet by positive displacement using the microdispenser. It is helpful sometimes for visualization when positioning under the microscope to leave a drop of oil on the tip of the micropipet.


10. Using a sterile disposable pipet tip, place l-2 PL of RNA or DNA onto the middle of a strip of Nescofilm (Nippon Shoji Kaisha Ltd., Osaka, Japan) secured to a sterile microscope slide.

11. While viewing the RNA or DNA sample through the stereomicroscope, position the injection micropipet tip into the center of the RNA or DNA droplet.

12. By negative displacement of the digital microdispenser, take up the RNA/DNA into the micropipet. After backfilling is complete, the sample droplet on the microscope slide should be removed and there should now be a visible distinct boundary between the RNA/DNA solution and the Sudan IV dyed mineral oil.

13. To ensure the injection setup is working, eject approx 10 nL (one graduation of the pipet) of the RNA or DNA.

5.4. Cgtoplasmic RNA fqjection

1.

2.

3.

4.

Place a number of previously selected oocytes (4-8) on a sterile microscope slide wrapped in Nescofilm. Ensure each oocyte is fully immersed in discrete individual drops of MBS to prevent drying during the injection. Manipulate the oocytes using a sterile disposable pipet tip until the equatorial region is vertical and points toward the injection micropipet. Injection into this area of the oocyte allows optimal mobility of the RNA and prevents damage of the nucleus (Colman and Drum- mond, 1986). Push the oocyte with the sterile pipet tip to the side of the droplet so that the oocyte is just touching the meniscus of the MBS droplet. The surface tension of the solution causes the oocyte membranes to stretch. This facilitates the injection and any damage from the injection procedure is minimized as the oocytes membrane “shrinks” on immersion back into the center of the MBS droplet. Hold the position of the oocyte stable with the sterile pipet tip and insert the injection micropipet into the equatorial region of the oocyte using the manipulator.


5.

6.

7.

8.

Successful impalement of the oocyte is signified by the sudden disappearance of the dimpled membrane surface around the micropipet tip. Inject up to 100 nL of RNA (typically 40-50 nL is preferred). Remove the injection micropipet from the oocyte and reposition the oocyte into the middle of the MBS droplet. Expel 10 nL of RNA solution between each injection to ensure that the micropipet tip does not become blocked and to avoid dilution of RNA with intracellular contents from previously injected oocytes. Repeat the procedure (steps 2-6) with other oocytes until eventually the oil-water interface can be seen moving to the tip of the micropipet, The whole procedure can be repeated using a new micropipet and fresh RNA solution as required. Do not reuse the micropipet even with the same aliquot of RNA solution.

9. Incubate oocytes as described in Section 6.

5.5. Nuclear DNA Iw’ection

5.5.1. “Blind” Injection of DNA Although injection of the oocyte nucleus requires a more

refined technique and the use of smaller injection volumes, it is still relatively easy to use the Drummond microdispenser without recourse to pressurized pump injectors.

1. Place a number of previously selected oocytes onto a plastic mesh grid fixed to the bottom of a shallow well contained in a Perspex block (Figs. 5 and 6A [see Fig. 6, p. 2741). This block should be designed to fit into a suitable centrifuge bucket. Ensure that sufficient MBS is present to keep the oocytes wet without allowing them to float over the grid.

2. Manipulate the oocytes using a sterile disposable pipet tip until the dark animal pole of the oocyte is uppermost. We use the same dimensions for our DNA injection micropipets as described previously for the injection of RNA.


3. Using the manipulator, insert the filled injection micropipet into the-middle of the animal pole where there is the highest probability of impaling the nucleus. Inject up to 20 nL of DNA into the selected oocyte (Fig. 5).

4. Expel 10 nL of DNA between each injection. Stop injecting oocytes when the oil-water interface can be seen moving to the tip of the micropipet.

5. The whole procedure (steps 3 and 4) can be repeated using a new micropipet and fresh DNA as required (Brown and Gurdon, 1977).

55.2. Direct Nuclear Injection 1. Place a number of previously selected oocytes onto a

plastic mesh grid and manipulate until the animal poles are uppermost, as described for the “Blind” injection technique.

2. Centrifuge the oocytes in a refrigerated centrifuge (Denley BR 401) at 700-11009 for 8-12 min at lo-15°C. A range of values are quoted, since some oocytes even from the same donor may require differing periods of centrifugation before the nuclei appear. Exceeding the top end of these ranges will compromise oocyte viability. Often it is useful to perform a test spin with a small number of oocytes to ascertain the centrifuge settings. Because of differing densities, the germinal vesicles should rise to the surface of the oocytes and their position will be indicated by a delineated area in the dark animal pole (Figs. 5 and 6B,C). Sometimes this area in the animal pole can appear white. Oocytes that have changed their orientation or have become damaged during centrifugation should be discarded.

3. Insert the microelectrode tip directly into the nucleus of the oocyte using the micromanipulator and inject DNA as described for the “Blind” injection technique (see steps 3 and 4) (Kressmann et al., 1977; Rungger and Turler, 1978; Bertrand et al., 1991) (Fig. 6D).

4. The survival rates of the oocytes, following either the blind or direct nuclear injection techniques, are quite


similar. It should be routine to keep at least 70-80% of injected oocytes alive for at least 7-14 d. Oocyte mortal- ity caused by the injection usually becomes evident in the first 3 d.

5.6. Optimization of Receptor/Ion-Channel Expression

The level of receptor/ion-channel expression can vary considerably among different oocytes from different donor frogs and even within oocytes obtained from the same donor. Over 10 yr we have observed several empirically determined criteria that will influence the ability of oocytes to express receptor/ion-channel proteins. Appropriate animal husbandry (Section 2.) in addition to provision of an adequate diet and ensuring the frogs are of 100-200 g mass before use are vital to obtain good quality oocytes. The level and duration of room lighting seems less critical. In addition to these factors, the quality of RNA and DNA will also be important; however, if expression is poor, it usually signifies that the oocytes used are not of good quality for expression studies rather than problems with the RNA/DNA preparations. Occasionally, when using DNA, expression can be very high, which may cause problems when trying to voltage clamp oocytes with induced membrane currents in excess of 10 fl. Under these circumstances we have reduced the concentration of DNA for subsequent injections (typically, from 1 mg/ mL to 10 pg/mL). However, this produces only a small reduction in expression efficiency. We have had more success mixing the receptor/ion-channel DNA with DNA tran- scribing for another protein product that will not interfere with electrophysiological assays of the expressed receptor/ ion channels. Alternatively, the receptor/ion-channel DNA can be mixed together with the same vector lacking the receptor/ion-channel DNA. The DNAs are mixed in different ratios according to the level of expression required, which can only be determined empirically. For example, expressing y-aminobutyric acid (GABA,) receptors, we have mixed

After injection, the oocytes are maintained in MBS, although other balanced salt solutions, including standard frog Ringer, are also adequate. We have observed that by using a modification of Barths solution that is hypertonic, the oocytes seem to survive for longer and tolerate microinjection better (Appendix 1). The following protocol results in reliable protein expression.

1.

2.

3.

After microinjection, transfer the oocytes into 35-mm sterile dishes containing MBS at approx 5°C and leave for 2 h. Any oocytes acutely damaged by the injection procedure will quickly become evident and should be removed. Healthy injected oocytes should then be transferred into sterile 5-mL glass tubes with screwcaps containing fresh MBS. Each tube will contain approx 5-8 oocytes/5 mL of MBS. Ensure the oocytes are not packed together or stacked in solution. Place the tubes in an incubator maintaining the temperature at 1%20°C for 24-72 h (for vertebrate RNA/DNA) or up to 7 d (for invertebrate RNA). The simplest and most economical incubator is a water bath maintained in a cold room and slowly adjusted to the correct water temperature. It is important to avoid sharp fluctuations in temperature that will affect expression and oocyte viability. Alternatively, and more costly, oocytes can be incubated in a commercial cold incubator, which tends to be quite large. A further alternative we have adopted is to build a small incubator incorporating a Peltier device to accurately maintain the temperature of a small water bath that can then be placed on the laboratory bench and is easily transported (Fig. 7).


receptor DNA (30 kg/mL) with the LacZ gene (1 mg/mL), which transcribes for P-galactosidase and is commonly used as a reporter gene in molecular biology.

6. Culture/Incubation of Iqjected Oocytes


Fig. 7. Equipment used to incubate the injected oocytes on the laboratory bench. This consists of a temperature controller (Peltier device), which controls the water temperature in a small bath utilizing a paddle stirrer to provide a constant cuculation of water. At near maximum capacity, using 5-mL glass vials (containing a maximum of 20 oocytes), this apparatus can incubate a total of 700 injected oocytes.

6.1. Optimal Culture Conditions for Protein Expression

During the period of incubation at 1%20°C for 1-2 d the MBS should be replaced every 24 h with fresh sterile MBS and any damaged or dead oocytes removed. To permit adequate receptor/ion-channel expression, oocytes injected with RNA will require incubation for up to 2-3 d. For DNA- injected oocytes, up to l-2 d is sufficient for expression. During early incubation it is often found that the level of expression, especially for DNA-injected oocytes, can increase in a few hours (monitored by measuring membrane currents through expressed ion-channel proteins), and it is therefore advisable to wait until the level of expression becomes constant before commencing electrophysiological recording.

Once the oocytes express the receptor/ion-channel proteins of interest, they should be stored at 10°C and the MBS


replaced every 3-4 d (any damaged oocytes should still be removed daily). The low temperature prolongs the survival of the oocytes such that they can be maintained in a viable state for up to 3-4 wk. Over a 34wk period, the level of protein expression will be reduced, which will be manifest by smaller induced membrane currents depending on the batch of oocytes used and the RNA/DNA preparation. How long the oocytes can be used for electrophysiology will depend on the type of ion channels being expressed and the peak current amplitudes. If only small whole-cell currents (approx 100 nA) are initially recorded after incubation, it is unlikely that such membrane currents will be resolved in oocytes maintained in culture after some weeks.

7. Electrophysiological Recording from Xenopus Oocytes

7.1. Two-Electrode Voltage Clamp

The large size of the Xenopus oocyte easily facilitates the impalement of two microelectrodes for the standard two-electrode voltage clamp technique. The follicular cell envelope can either be removed (see Section 3.3.) or left in situ for this method of recording. Retaining the follicular cells will make impalement of the electrodes more difficult, but the oocyte viability is usually higher. It is important to be assured that if the follicular cells are retained then the expressed ion channels in the oocyte membrane will not be masked by similar endogenous ion channels present in the follicular cell layer or even in the oocyte plasma membrane (Dascal, 1987).

7.1.1. Recording Equipment There are many types of recording bath that can be used

for Xenopus oocytes. The bath should facilitate continuous superfusion with amphibian/frog Ringer solution (Appen- dix 1) and allow drug application. In addition, easy access is required for drug-filled pipets and the recording electrodes. The oocytes are easily transferred from their incubation container to the bath using a blunt and firepolished Pasteur pipet.


Immobilization of the oocyte in the recording chamber can be achieved by positioning within a trench cut in the base of the bath previously coated with a layer of Sylgard resin (Dow Corning, Seneffe, Belgium). Alternatively, the oocyte can be placed in a circle of shortened insect pins previously positioned in a Sylgard base providing a microholder suspend- ing the oocyte just above the base of the bath. In either case, the oocyte is held firmly in position after impalement by the recording electrodes.

The bath is placed on a heavy recording table (an air table is unnecessary) and the bath viewed with a simple compound microscope (magnification 4-20x). Two coarse micromanipulators are sufficient to hold the microelectrodes and perform satisfactory impalements. The required electrical recording apparatus will comprise a conventional two-electrode voltage clamp amplifier, a timer, and pulse generator. The data will need to be low-pass filtered and can be stored on a chart recorder, a tape recorder, or played directly to a computer for analysis (Fig. 8).

7.1.2. Microelectrodes and Oocy te impalement Glass microelectrodes are routinely fabricated from thin-

walled filamented glass (1.5 mm external and 0.86 mm internal diameters; GC150TF-10; Clark’s electromedical, Pangbourne, Reading, UK). Electrodes are pulled on a David Kopf puller (model 730) to give resistances of 0.5-2 Ma. The low resistances are required to allow the passage of sufficient current to voltage clamp the oocyte during responses to drugs. If the resistance is too low, the oocytes become difficult to impale without causing membrane damage and the electrodes are also “leaky,” allowing the electrolyte to freely enter into the cytoplasm, which may affect agonist response amplitudes. The voltage recording electrode is filled with 3M KC1 solution and the current electrode is filled with 0.6M K,SO,. Both electrodes can be filled with 3M KCl, although this can chloride load the oocyte and inconveniently change the reversal potential for ligand-gated channels selectively permeable to Cl- (see Section 7.1.4.). Connections to the


Fig. 8. Typical recording arrangment for two-electrode voltage clamp experiments on Xenopus laevis oocytes showing two micromanipulators, a chamber, and plastic funnels for bath application of solutions. Note the position of the recording microelectrodes and the intervening aluminum shield. Solution removal is achieved by vacuum suction. With minor modifications, this setup can also be utilized for patch-clamp experiments.

amplifier are made via silver/silver chloride wires. Usually the silver chloride coat is renewed before each experiment.

Impalement of an oocyte will cause a dimpling of the cell surface (more severe if using follicular oocytes). The initial membrane potential can be quite variable from batch to batch of oocytes and typically in the range -10 to -70 mV. The voltage electrode is inserted first and then the current electrode is introduced while ejecting pulses of constant current. Successful impalement of the current electrode will be registered by the appearance of an electrotonic potential recorded by the voltage electrode. Gradual disimpalement of the electrodes, to reduce the membrane surface dimpling, followed by sealing of the electrodes, which can take up to 30 min, will result in a viable oocyte allowing many hours of recording. The gradual improvement in the oocyte after impalement can be followed by an increase in the input resistance (from 0.4 f 0.13 Ma to 1.4 f 0.25 Ma) and hyper-

Smart and Krishek

polarization of the membrane (typically from -20 + 5 mV to -50 f 13 mV; data taken from 25 oocytes).

7.1.3. Voltage Clamping The large size of the oocyte, while making for easy

impalements, also carries a significant disadvantage with the amount of membrane that must be charged to voltage clamp the cell. An oocyte of 1 mm diameter will have a geometric surface area of 3 x lo6 pm2. However, measurement of the capacity transients associated with small voltage command steps yields estimates of the membrane capacity in the range NO-250 nF. Assuming a specific membrane capacity of 1 pF/ cm2 (Methfessel et al., 1986), the likely surface area accounting for surface invaginations and microvilli (Dumont and Brummett, 1978) is approx 4-5.5 times greater at 15 x lo6 pm2 compared to the previous simple geometric estimate (Methfessel et al., 1986). Voltage clamping oocytes expressing voltage-gated ion channels with fast kinetics will require a very fast clamp of the membrane with the ability to compensate for the large capacity transients. The response time of the voltage clamp amplifier following a voltage command step will be a function of R; C,/A, where Re is the resistance of the current passing electrode, C, is the membrane capacity, and A the gain of the amplifier. Thus, by using a low resistance current electrode and a high gain on the amplifier, the time for the clamp to respond will be reduced. An additional possibility is to reduce C,. This can be achieved by using oocytes at an earlier stage of development (stages II and III), which have lower membrane capacity compared to stages V and VI (Krafte and Lester, 1989; see Section 4.). To further increase the gain and speed of the amplifier, an earthed shield should be introduced between the two electrodes to reduce capacitative coupling. We use a small sheet of aluminum foil on a wire frame that can be easily positioned between the electrodes without touching the bathing fluid (Fig. 8). If the speed of the clamp is still insufficient, it may be more appropriate to use an alternative expression system for studying very fast voltage-gated currents (e.g.,


activation of sodium currents, see Section 8.). For l&and-gated currents the speed of the clamp amplifier may not be so critical, and a settling time even of lo-20 ms is tolerable since the limiting factor for the measured responses is likely to be the rate of drug administration. As with all voltage clamp procedures, it is advisable to maintain a constant minimal fluid level in the bath just above the oocyte. This helps with visualization of the preparation under the microscope during the experiment and also reduces electrode capacity.

7.1.4. Disadvantages The oocyte is an efficient expression system that can very

often lead a voltage clamp amplifier to try to clamp up to 10 PA of membrane current induced by an agonist or a voltage command step. There are two strategies that can be employed to ameliorate the high expression levels and consequently large membrane currents. Initially, using another batch of oocytes, a “competing” but unrelated cDNA can be injected together with the ion-channel cDNAs that will act in compe- tition for transcription and lower the expression level (see Section 5.6.). This technique can only be performed with hind- sight and, unfortunately, the next batch of oocytes might not express the protein of interest quite so well as previously. Secondly, with the existing oocytes, the concentration of permeant ions can be reduced to lower the conductance induced by the agonist or voltage command step. This may necessitate removing up to 60% of the permeant ion and replacing with a nonpermeant analog that should not interfere with channel operation. Sucrose replacement can also be employed to remove permeant ions while maintaining osmolarity.

In contrast, a lack of expression producing only small membrane currents can also pose a problem. Currents 40 nA are difficult to detect and a lower range of 1-3 nA is the limit of resolution.

The size of the oocyte does not lend itself readily to rapid drug application. Thus, fast kinetics of ligand-gated ion channels may be lost because of slow drug perfusion rates. Recent

Smart and Krishek

attempts have been made to improve this with variations of the “concentration clamp” technique (Akaike et al., 1986), which enable a 90% exchange of the extracellular fluid around the oocyte to be achieved in lo-20 ms (Arvanov and Usherwood, 1991; Madeja et al., 1991). However, should this rate of drug application still prove too slow for analysis of rapid kinetics, adoption of a new expression system using smaller cells is preferable (Section 8.). A recent variation of the oocyte preparation involves using a grease-gap technique where part of the oocyte membrane is permeabilized or ruptured, allowing solution access to the oocyte interior. This method reduces the capacity transient decay time constant to 20-100 j,~s and resolution of small membrane currents down to 1 nA is possible (Taglialatela et al., 1992).

The use of blunt electrodes can cause two problems. First, leakage of electrolyte into the oocyte may cause a disturbance of ionic concentrations. If this causes the reversal potential of the permeating ion(s) of interest to change (e.g., Cl- and responses to GABA on expressed GABA, receptors), then either higher resistance electrodes should be used to reduce leakage or change the electrolyte in one or both electrodes (e.g., use K,SO, in the current passing electrode). Second, occasionally, the recording electrodes can become blocked with yolk platelets. Switching to current clamp conditions and passing a large current in the opposite direction to that employed in the experiment can clean the electrodes. If this fails and the experiment cannot proceed, switch back to voltage clamp and briefly oscillate the amplifier (for <l s) using the capacity neutralization controls. The oocyte is often sufficiently robust to cope with oscillation of the amplifier to try to clean the electrodes, but note that the cell is not completely indestructible! A last resort is to disimpale the blocked electrode(s) and replace with a new electrode(s). The same oocyte can be reimpaled and, with care, the membrane will reseal to near the original resting input resistance.

7.2. Patch-Clamp Recording

Xenopus oocytes can also be used for patch-clamp experiments involving single-channel recording. Cell-


attached, inside-out, and outside-out patches can all be formed using similar methods to those employed for much smaller cells (Hamill et al., 1981). However, the oocyte is too large to be effectively whole-cell clamped using a patch electrode. To, obtain a good tight seal on the membrane surface the follicular cell layer is first removed using collagenase and the vitelline layer must also be manually removed following immersion of the oocyte in a hypertonic solution to cause cell shrinkage (see Section 3.4.). The oocyte is now very clean but also extremely fragile and will stick readily to glass and plastic containers. Sticking can be avoided by manipulating the oocytes in Petri dishes containing 2% agarose. Transfer of the oocyte into the recording bath can easily incur damage and the cell will disintegrate if brought to an air/water interface (Methfessel et al., 1986).

7.2.1. Recording Equipment and Electrode Fabrication Standard patch-clamp amplifiers from many manufac-

turers will be capable of recording single-channel currents from Xenopus oocytes. Patch pipets are fabricated from thick- walled borosilicate glass (GC150-10, Clarks Electrodmedical) and are pulled to have small pipet opening diameters (lo-20 M&II resistance) and a steep pipet taper to reduce the access resistance (Methfessel et al., 1986). These characteristics appear to reduce the prospect of stretch-activated ion-channel activity in the patches. If the patch pipet is immersed deep into the Ringer solution before reaching the oocyte membrane, it may be necessary to coat the electrode shank to near the tip with Sylgard resin.

7.22. Formation of High Resistance Seals The method of obtaining high resistance seals on oocytes

is generally similar to that for smaller cells. Seal resistance increases on touching the oocyte surface, and gentle suction (lo-20 cm water negative pressure) will often result in the development of a high resistance seal (l-20 GQ). The formation of inside-out patches can be achieved by the slow withdrawal of the pipet from the oocyte, which frequently results in the formation of a vesicle distorting single-channel currents. The vesicle may be broken by a variety of methods,


including brief exposure to air as described by Hamill et al. (1981). Outside-out patches require formation of the whole- cell recording mode before withdrawing the pipet as described previously (Hamill et al., 1981).

7.2.3. Recording from Macropa tches Formation of macropatches requires the use of pipets

with large tip diameters (2-8 pm) and low resistance (0.6-2 Ma; Stuhmer, 1992). A hard aluminosilicate glass is preferred to give pipets with smooth rims. The process of seal formation requires more gentle suction than that required for smaller patches and may take several minutes to form (Stuhmer, 1992). There are very few advantages of using the macropatch method over conventional two-electrode voltage clamp. It enables current flowing through many ion channels to be recorded with better time resolution (50-200 i.~s for macropatch compared to 0.5-2 ms for two-electrode voltage clamp) since the membrane capacity is reduced when patching a relatively small area of membrane compared to the two- electrode voltage clamp of the entire oocyte surface area. Using the macropatch method in an inside-out configuration (Stuhmer, 1992), access is allowed to either face of the membrane; however, formation of the outside-out mode is more difficult. Thus, in conclusion, the macropatch method could be considered as a surrogate “mini-whole-cell recording mode” for the Xenopus oocyte. The resolution of membrane current for the macropatch is approx 5-20 pA, whereas for two-electrode voltage clamp currents below 3 nA become difficult to detect. In comparison, patch-clamp recording from conventional cell-attached or excised patches could resolve temporal events down to 30-50 ps and currents of approx 100 fA.

7.2.4. Disadvantages Although patch-clamp recording will allow resolution

of single ion-channel currents, the Xenopus oocyte does contain endogenous ion channels that can interfere frequently with measurements of the expressed ion channels. Stretch-


activated (or mechanosensitive) ion-channel activity is evident on applying either positive or negative pressure to cell-attached patches. With a normal Ringer solution in the pipet, the conductance of the channel is approx 28-38 pS (Methfessel et al., 1986; Taglietti and Toselli, 1988). Stretch- activated ion channels are cation-selective inward rectifiers and the currents reverse at approx -10 to +7 mV membrane potential (Taglietti and Toselli, 1988; Yang and Sachs, 1990). The use of thick-walled patch pipets with narrow tip diameters seems to reduce stretch-activated channel activity. Also, these channels are less evident in outside-out patches, which is the mode of choice for studying ligand-gated ion channels. However, if the expressed ion-channel density is low in the oocyte membrane, the use of larger tip diameter patch pipets may be necessary.

As an alternative, pharmacological intervention may be used to block the stretch-activated ion channels. Gadolinium ions (Gd3+) will block this channel, with a concentration of 10 @4 sufficient to reversibly abolish channel activity in outside-out patches (Yang and Sachs, 1989). Lanthanum and lutetium at concentrations ,100 @PI will also block stretch- activated channel activity. Less potent compounds, such as amiloride (IC,, 500 u&I) and some structural analogs (IC,,s 30-400 w), will also inhibit stretch-activated channels (Lane et al., 1992); however, amiloride does have many other actions on different ion channels (Inomata et al., 1988; Hamill et al., 1992).

Another endogenous ion-channel population that can interfere with recordings from expressed receptors and channels is the calcium-activated chloride channel (Miledi, 1982; Barish, 1983; Dascal, 1987). This conductance is activated by depolarizations from -70 mV to potentials more positive than -20 mV and is characterized by a slowly developing but transient outward current. This current can be useful as an assay for the expression of a variety of G protein-coupled receptors and also ion channels resulting in increased intracellular calcium levels. If this current interferes with the responses


mediated by the expressed ion channels, then external chloride can be replaced with impermeant anions, e.g., methanesulfonate. Alternatively, the oocytes can be injected with calcium chelat- ing agents to prevent chloride current activation; e.g., 100 mM EGTA or BAPTA (injecting 30-50 nL per oocyte). In addition, this calcium-activated chloride conductance can be inhibited by nonsteroidal anti-inflammatory agents, such as niflumic and flufenamic acids (White and Aylwin, 1990; inhibition constants, 17 and 28 v, respectively). However, specificity is not absolute, and these compounds, including others from the same chemical families, may interact with expressed ligand-gated anion channels, e.g., GABA, receptor/ion channels (White and Aylwin, 1990; Shirasaki et al., 1991).

A different problem may beset the formation of outside- out patches in Xenopus oocytes. When forming the whole-cell recording mode, applied suction that is too strong may allow yolk platelets to enter the electrode tip and cause blockage.

8. Comparison of Xenopus Oocytes with Alternative Expression Systems

The Xenopus oocyte is a very efficient expression system accepting a variety of RNAs or DNAs from different species and producing functional receptor/ion-channel proteins. Although the oocyte has proved an exceedingly popular expression system for functional assays of individual receptor/ion-channel subtypes, there are now alternatives mostly involving secondary cell lines (e.g., Buckley et al., 1990) and insect cells that enable electrophysiological recording of expressed ion-channel proteins (Table 1). The choice of the most appropriate expression system will be based on many criteria, for example, experimental design and also the problems that are to be addressed, including

1. Whether transient or “stable” expression is preferred; 2. The level of protein expression; 3. Ease of manipulation for comparative electrophysiologi-

cal, biochemical, and molecular biological experiments;


4. Reproducibility of expression for comparative pharmacology of expressed proteins;

5. Presence of homologous endogenous receptors in the host cell;

6. Appropriate assembly of heteromultimeric receptor subunits compared to native in vivo receptors; and

7. Ability of expressed ion channels and G protein-linked receptors to couple to single or multiple second messenger transduction pathways.

Overall, no one expression system is likely to prove ideal for all the experiments envisaged for cloned ion-channel proteins. There is also some merit in comparing the properties of an expressed ion-channel population in at least two different types of expression system to ensure that inappropriate posttranslational processing is not unduly altering the physiological and pharmacological properties of the expressed ion channels. Some of the salient features of different expression systems, their advantages and disadvantages, and the methods of RNA and DNA incorporation are detailed in Table 1.

Acknowledgments

We thank and owe much to previous postgraduate assistants in our laboratory (Khaled M. Houamed and Derek Bowie) for their experimental innovations helping to con- tinually develop the Xertopus Zaevis oocyte expression system. We are also grateful to Eric Barnard for introducing us to this preparation in 1981 and for help and advice. We thank Steve Coppard for animal husbandry and we are grateful to Chris Courtice (School of Pharmacy) for design- ing and building the oocyte incubator, timers, and pulse generators. We thank Keith Poulton and Nikon UK for microscopy and generous financial support. Our work is supported by the Medical Research Council (UK), The Wellcome Trust, and The School of Pharmacy, University of London.

Tabl

e 1

Com

paris

on

of t

he P

rope

rties

of

Som

e C

ellu

lar-B

ased

Ex

pres

sion

Syst

ems”

Exp

erim

enta

l us

e in

Expr

essio

n sy

stem

G

ene

trans

fer

Type

of

expr

essi

on

Expr

essio

n of

io

n-ch

anne

l pr

otei

ns

Prop

ertie

s of

ex

pres

sion

sy

stem

elec

troph

ysio

logy

(E

) an

d bi

oche

mis

try

(B)

E B

Ref

s.

Xeno

pus

laeo

rs

oocy

te

RN

A,

cyto

plas

mic

in

lect

ion

DN

A,

nucl

ear

inje

ctio

n

Tran

sien

t

Mam

mah

an

cells

Prim

ary

cells

Se

cond

ary

cell

lines

: e.

g.,

hum

an e

mbr

yoni

c ki

dney

ce

lls;

Chm

ese

ham

ster

ov

ary,

C

os;

L ce

lls

Vira

l in

fect

ion,

D

NA

Mic

roin

ject

ion

RN

A an

d D

NA

Lipo

fect

ion,

D

NA

Tran

sfec

tion,

D

NA

Vira

l in

fect

ion

Wac

cmia

, H

erpe

s sz

mpl

ex),

DN

A

Ele

ctro

pora

non,

D

NA

Tra

nsie

nt

Tra

nsie

nt

- or

“s

tabl

e”

Man

y re

cept

ors

Larg

e ce

lls, e

asy to

an

d io

n ch

anne

ls

man

ipul

ate.

Fai

thfu

l tr

ansl

atio

n of

RN

As.

S

ome r

ecep

tors

not

ex

pres

sed,

po

ssib

le

inco

rrec

t N

-link

ed

glyc

osyl

atio

n. U

p to

10

0% ex

pres

sion

. S

easo

nal va

riatio

n fo

r ex

pres

sion

. M

any

rece

ptor

s S

mall c

ells

, use

ful

and

ion

chan

nels

fo

r pa

tch-

clam

p an

d bi

oche

mis

try.

E

xpre

ssio

n ef

ficie

ncie

s var

y w

idel

y lO

-30%

(t

rans

fect

ion)

to

100%

(hpo

fect

ion)

. E

lect

ropo

ratio

n pa

ram

eter

s (v

olta

ge

and

time)

var

y w

ith

cell

type

. N

o se

aso

nal va

riatio

n in

exp

ress

ion

Intr

acel

lula

r +-

reco

rdin

g (i/

c)

i-i-ii

l-5

Pat

ch-c

lam

p re

cord

ing

++

i/c

+-k

-H

+ 67

Pat

ch-c

lam

p 8,

9 ++

++

lO,Z

l

4,5,

12

13,1

4

Pro

duct

ion

of l

arge

am

ount

s of

pro

tein

. V

irally

in

fect

ed

cells

lab

ile w

ith

time.

Bro

ad r

ange

of

cel

ls i

nfec

ted.

S

mal

l ce

lls,

used

l/C

++

++

25,2

6 fo

r pa

tch-

clam

p +-

Infe

ctio

n is

Iytic

, th

us c

ells

bec

ome

patc

h-cl

amp

elec

tric

ally

++

+ “le

aky“

an

d ar

e use

d

with

in 2

4 h.

Lar

ge

prot

ein

yiel

ds.

Cul

ture

cells

at 2

8°C

. V

ery

sma

ll ce

IIs (7

Iun

i/c

++++

27

-19

dia.

). O

uter

cel

l -

wall

must

be

rem

oved

usi

ng z

ymol

yase

to

patc

h-cl

amp

form

sphero

bla

sts.

+-

Seals

requ

ire m

ore

suct

ion

and

take

lo

nger

to f

orm

. E

xpre

ssed

ion

chan

nel c

urre

nts

not

reco

rded

.

Inse

ct c

ells

Sp

odop

fer

a

frugr

per~

(m

oth;

Sf9

ce

lls)

Yea

st

~Sac

chm

omyc

es

cere

vzsi

ae)

Bac

ulov

irus

Tra

nsie

nt

Pota

ssiu

m

mfe

ctio

n (w

ild-

chan

nels

ty

pe o

r rec

ombi

nant

M

-c

AC

h,

Auto

grap

ha

GA

BA

, an

d $-

ca

lifom

ca)

adre

nerg

ic

rece

ptor

s

Tra

nsfo

rmat

ion

“Sta

ble”

B

iosy

nthe

sis

of

prot

eins

, e

g.,

AC

h re

cept

or

subu

nits

, G

prot

ein-

coup

led

rece

ptor

s, a

nd

pota

ssiu

m

chan

nels

.

““G

ene

tran

sfer

” in

dica

tes

the

main

mod

e of

inc

orpo

ratin

g D

NA

in

to t

he c

ells

and

“ex

perim

enta

l use

” in

dica

tes

the

rela

tive

ease

an

d su

itabi

lity

(+)

for

stud

y usi

ng e

lect

roph

ysio

logi

cal

and

bioc

hem

ical

tech

niqu

es.

bRef

eren

ces:

1. S

mar

t et al

., 19

87; 2

. Les

ter,

198

8; 3.

Sig

el, 1

990;

4. Y

ang

et a

l., 1

991;

5. K

arsc

hin,

199

3; 6.

Cap

pecc

hi, 1

980;

7. Ik

eda

et

al.,

1992

; 8. H

olt

et a

l., 1

990;

9. B

arth

el e

t al

., 19

92; 1

0. C

hen

and

Oka

yam

a, 1

987;

11.

Cla

udio

, 19

92; 1

2. G

elle

r et

al.,

199

1; 13

. Orlo

wsk

i an

d M

ir,

1993

; 14.

Tek

Ie e

t al

., 19

91; 1

5. A

tkin

son

et a

l., 1

992;

16.

Kam

b et

al.,

199

2; 1

7. F

ujita

et

al.,

1986

; 18.

Kin

g et

al.,

199

0; 1

9.

Str

osbe

rg a

nd M

aruI

lo,

1992

.


Appendix 1: Composition of Physiological Solutions

Modified Barth’s Solution

110 mM NaCl 1mMKCl 2.4 mM NaHCO, 7.5 mM Tris-HCI 0.33 mM Ca(NO,), 0.41 mM CaCl, 0.82 mM MgSO, 50 mg/L Gentamycin pH 7.6

Frog-Ringer Solution

110 mM NaCl 2mMKCl 5 mM HEPES 1.8 mM CaCl, pH 7.4

The constituents of both solutions should be added and dissolved in the sequence shown. Modified Barth’s solution should be sterilized by autoclaving the solution without adding the gentamycin and NaHCO,. The antibiotic and NaHCO, can be sterilized by filtration through a 0.22~Frn pore filter (Millipore) and added aseptically to the sterile Modified Barth’s solution.

Hypertonic Stripping Solution

200 mM K+ aspartate 20 mM KC1 1 mM MgCl, 10 mM EGTA 10 mM HEPES pH 7.4

This solution is sterilized by filtration using a 0.22~pm pore filter.


References

Akaike N., Inoue M., and Krishtal 0. A. (1986) ‘Concentration-clamp’ study of y-aminobutyric-acid-induced chloride current kinetics in frog sensory neurones. J. Physiol. (Lund.) 379,171-185.

Arvanov V. L. and Usherwood I’. N. R. (1991) Concentration jump studies of intracellularly dialysed Xenopus oocytes show desensitization of kainate receptors. Neurosci. Left. 129,201-204.

Atkinson A. E., Bermudez I., Darlison M. G., Barnard E. A., Earley F. G. I’., Possee R. D., Beadle D. J., and King L. A. (1992) Assembly of functional GABA, receptors in insect cells using baculovirus expression vectors. Neuroreport 3,597-600.

Barish M. E. (1983) A transient calcium-dependent chloride current in the immature Xenopus oocyte. J. Physiol. (Land.) 342,309-325.

Barnard E. A., Miledi R., and Sumikawa K. (1982) Translation of exogenous messenger RNA coding for nicotinic acetylcholine receptors produces functional receptors in Xenopus oocytes. Proc. Roy Sot. Land. B 215,241-246.

Barthel F., Boutillier A. L., Giraud P., Demeneix B. A., Behr J. P., and Loeffler J. P. (1992) Gene regulation analysis by lipopoly- amine-mediated DNA transfer in primary neurons, in Methods in Neuroscience, vol. 9 (Conn P. M., ed.), Academic, San Diego, pp. 292-312.

Bertrand D., Cooper E., Valera S., Runger D., and Ballivet M. (1991) Elec- trophysiology of neuronal nicotinic acetylcholine receptors expressed in Xenopus oocytes following nuclear injection of genes or cDNAs. Meth. Neurosci. 4,174-193.

Brown D. D. and Gurdon J. B. (1977) High-fidelity transcription of 5s DNA injected into Xenopus oocytes. Proc. Natl. Acad. Sci. LISA 5, 2064-2068.

Buckley N. J., Hulme E. C., and Birdsall N. J. M. (1990) Use of clonal cell lmes in the analysis of neurotransmitter receptor mechanisms and function. Biochim. Biophys. Acta. 1055,43-53.

Burmeister M. and Soreq H. (1984) Production and analysis of radioactivity labelled translation products in microinjected Xenopus oocytes, in Molecular Biology Approach to Neurosciences (Soreq H., ed.), Wiley, London, pp. 195-203.

Capecchi M. R. (1980) High efficiency transformation by direct microinjection of DNA into cultured mammalian cells. Cell 22,479-488.

Chen C. and Okayama H. (1987) High-efficiency transformation of mammalian cells by plasmid DNA. Mol. CeII. Biol. 7,2745-2752.

Claudio T. (1992) Stable expression of heterologous multisubunit protein complexes established by calcium phosphate- or lipid-mediated cotransfection, in Methods in Enzymology, vol. 207 (Rudy B. and Iverson L. E., eds.), Academic, San Diego, pp. 391-408.


Colman A. (1984) Translation of eukaryotic messenger RNA in Xenopus oocytes, in Transcriptzon and Translation: A Practical Approach (Hames B. D. and Higgins S. J., eds.), IRL, Oxford, pp. 271302.

Colman A. and Drummond D. (1986) The stability and movement of mRNA in Xenopus oocytes and embryos. J. Emb yol. Exp. Morphol. 97(Suppl.), 197-209.

Contreras R., Cheroutre H., and Fiers W. (1981) A simple apparatus for mjection of nanolitre quantities into Xenopus oocytes. Anal. Biochem. 113,185-187.

Dascal N. (1987) The use of Xenopus oocytes for the study of ion channels. CRC Crit. Rev. Biochem. 22,317-387.

Dumont J. N. (1972) Oogenesis in Xenopus laevis (Daudin), 1. Stages of oocyte development in laboratory mamtained animals. J. Morphol. 136,153-180.

Dumont J. N. and Brummett A. R. (1978) Oogenesls m Xenopus laevis (Daudin). Relationships between developmg oocytes and their investing follicular tissues. J Morphol. 155,73-98.

Frech G, C., VanDongen A. M. J., Schuster G., Brown A. M., and Joho R. H. (1989) A novel potassium channel with delayed rectifier properties isolated from rat brain by expression cloning. Nature (Land.) 340,642-645.

Frech G. C. and Joho R. H. (1992) Isolation of ion channel genes by expression cloning in Xenopus oocytes, in Methods in Enzymology, vol. 207 (Rudy P. and Iverson L. E., eds.), Academic, San Dlego, pp. 592-604.

Fulita N., Nelson N., Fox T. D., Claudio T., Lindstrom J., Riezman H., and Hess G. P. (1986) Biosynthesis of the Torpedo californica acetylcholine receptor a subunit in yeast. Science 231,1284-1289.

Geller A. I., During M. J., and Neve R. L. (1991) Molecular analysis of neuronal physiology by gene transfer into neurons with herpes simplex virus vectors. Trends Neurosci. 14,428-432.

Goldin A. L. (1992) Maintenance of Xenopus laevis and oocyte injection, in Methods in EnzymoIogy, vol. 207 (Rudy B. and Iverson L. E., eds.), Academic, San Diego, pp. 266-279.

Gundersen C. B., Miledi R., and Parker I. (1984) Messenger RNA from human brain induces drug- and voltage-operated channels in Xenopus oocytes. Nature (Lond.I308,421-424.

Gurdon J. B. (1974) The Control of Gene Expression in Animal Development. Harvard University Press, Cambridge, MA.

Gurdon J. B., Lane C. D., Woodland H. R., and Marbaix G. (1971) Use of frog eggs and oocytes for the study of messenger RNA and its translation in living cells. Nature (Land.) 233,177-182.

Hamill 0. P., Marty A., Neher E., Sakmann B., and Sigworth F. J. (1981) Improved patch-clamp techniques for high resolution current recording from cells and cell-free membrane patches. Pfltigers Arch. 391,85-100.


Hamill 0. I’., Lane J. W., and McBride D. W. Jr. (1992) Amiloride: a molecular probe for mechanosensitive channels. Trends Neurosci. 13,373376.

Hitchcock M. J. M., Ginns E. I., and Macus-Sekura C. J, (1987) Microin- jection into Xenopus oocytes: equipment, in Methods in Enzymology, vol. 152 (Berger S. L. and Kimmel A. R., eds.) Academic, San Diego, pp. 276-288.

Hollmann M., O’Shea-Greenfield A., Rogers S. W., and Heinemann S. (1989) Cloning by functional expression of a member of the glutamate receptor family. Nature (Land.) 342,643-648.

Holt C. E., Garlick N., and Cornel E, (1990) Lipofection of cDNAs in the embryonic vertebrate central nervous system. Neuron 4,203-214.

Houamed K. M., Bilbe G., Smart T. G., Constanti A., Brown D. A., Barnard E. A., and Richards B. M. (1984) Expression of functional GABA, glycine and glutamate receptors in Xenopus oocytes injected with rat brain mRNA. Nature (Land.) 310,318-321.

Ikeda S. R., Soler F., Zuhlke R. D., Joho R. H., and Lewis D. L. (1992) Heterologous expression of the human potassium channel Kv2.1 in clonal mammalian cells by direct cytoplasmic microinjection of cRNA. Pfrligers Arch. 422,201-203.

Inomata N., Ishihara T., and Akaike N. (1988) Effects of diuretics on GABA-gated chloride current in frog isolated sensory neurones. Br. J. Pharmacol. 93,679-683.

Kamb A., Korenbrot J. I., and Kitajewski J. (1992) Expression of ion channels in cultured cells using baculovirus, in Methods in Enzymology, vol. 207 (Rudy B. and Iverson L. E., eds.) Academic, San Diego, pp. 423-431.

Karschin A. (1993) Heterologous expression of the membrane proteins that control cellular excitability, in Comparative Molecular Neurobi- ology (Pichon Y., ed.), Birkhauser Verlag, Basel, pp. 31-47.

King K., Dohlman H. G., Thorner J., Caron M. G,, and Lefkowitz R. J. (1990) Control of yeast mating signal transduction by a mammalian BZadrenergic receptor and Gs a subunit. Science 250,121-123.

Krafte D. S. and Lester H. A. (1989) Expression of functional sodium channels in stage II-III Xenopus oocytes. J. Neurosci. Meth. 26,211-215.

Kressmann A., Clarkson S. G., Telford J. L., and Birnsteil M. L. (1977) Transcription of Xenopus tDNA,“” and sea urchin histone. DNA injected into the Xenopus oocyte nucleus. Cold Spring Harbor Symp. Biol. 42, 1077-1082.

Lane C. D. (1983) The fate of genes, messengers, and proteins introduced into Xenopus oocytes, in Current Topics in DeveIopmenful Biology, vol. 18, Academic, New York, pp. 89-116.

Lane J. W., McBride D .W. Jr., and Hamill 0. l?. (1992) Structure-activity relations of amiloride and its analogues in blocking the mechanosensitive channel in Xenopus oocytes. Br. J. Phurmucol. 106, 283-286.

Smart and Krishek

Lester H. A. (1988) Heterologous expression of excitability proteins: route to more specific drugs? Sczence 241,1057-1063.

Lubbert H., Hoffman B. J., Snutch T. I’., Van Dyke T., Levme A. J., Hart& P. R., Lester H. A., and Davidson N. (1987) cDNA cloning of a serotonin 5-HTlc receptor by electrophysiological assays of mRNA- injected Xenopus oocytes. Proc. Natl. Acad. Sci. USA 84,4332-4336.

Madeja M., Mushoff U., and Speckmann E.-J. (1991) A concentration- clamp system allowing two-electrode voltage-clamp investigations in oocytes of Xenopus laevis. J Neuroscz. Meth. 38,267-269.

Masu Y., Nakayama K., Tamaki H., Harada Y., Kuno M., and Nakanishi S. (1987) cDNA cloning of bovine substance-K receptor through oocyte expression system. Nature (Lond.)329,836-838.

Masu M., Tanabe Y., Tsuchida K., Shigemoto R., and Nakanishi S. (1991) Sequence and expression of a metabotropic glutamate receptor. Nature (Land,) 349,760-765.

Methfessel C., Witzemann V., Takahashi T., Mlshina M., Numa S., and Sakmann B. (1986) Patch-Clamp measurements on Xenopus Zaevis oocytes: currents through endogenous channels and implanted acetylcholine receptor and sodium channels. Pffiigers Arch. 407, 577-588.

Miledi R. (1982) A calcium-dependent transient outward current in Xenopus laevis oocytes. Proc. Roy. Sot. Lond. B 215,491-497.

Miledi R., Parker I., and Sumikawa K. (1982) Synthesis of chick brain GABA receptors by frog oocytes. Proc. Roy. Sot. Lond. B 216,509-515.

Moriyoshi K., Masu M., Ishii T., Shigemoto R., Mizuno N., and Nakanishi S. (1991) Molecular cloning and characterisation of the rat NMDA receptor. Nature (Land.) 354,31-37.

Orlowski S. and Mir L. M. (1993) Cell permeabilization: a new tool for biochemical and pharmacological studies. Biochim. Biophys. Acta. 1154,51-63.

Rungger D. and Turler H. (1978) DNAs of simian virus 40 and polyoma direct the synthesis of viral tumor antigens and capsid proteins in Xenopus oocytes. Proc. Natl. Acad. Sci. USA 75,6073-6077.

Shirasaki T., Harata N., Nakaye T., and Akaike N. (1991) Interaction of various non-steroidal anti-inflammatories and quinolone antinu- crobials on GABA response in rat dissociated hippocampal pyramidal neurons. Brain Res. 562,329-331.

Sigel E. (1990) Use of Xenopus oocytes for the functional expression of plasma membrane proteins. J. Membr. Biol. 117,201-221.

Smart T. G., Constanti A., Bilbe G., Brown D. A., and Barnard E. A. (1983) Synthesis of functional chick brain GABA-benzodlazepine-barbi- turate/receptor complexes in mRNA-injected Xenopus oocytes. Neurosci. Lett. 40,55-59.

Smart T. G., Houamed K. M., Van Renterghem C., and Constanti A. (1987) mRNA-directed synthesis and insertion of functional amino acid receptors in Xenopus laevis oocytes. Btochem. Sot. Trans. 15,117-X22.


Soreq H. (1985) The biosynthesis of biologically active proteins in mRNA- microinjected Xenopus oocytes. CRC Crit. Rev. Biochem. 18,199-238.

Snutch T. P. (1988) The use of Xenopus oocytes to probe synaptic communication. Trends Neurosci. 11,250-256.

Strosberg A. D. and Marullo S. (1992) Functional expression of receptors in microorganisms. Trends Pharmacol. Sci. USA 13,95-98.

Stuhmer W. (1992) Electrophysiological recording from Xenopus oocytes, in Methods in Enzymology, vol. 207 (Rudy B. and Iverson L. E., eds.), Academic, San Diego, CA, pp. 319339.

Sumikawa K., Houghton M., Emtage J. S., Richards B. M., and Barnard E. A. (1981) Active multisubunit ACh receptor assembled by translation of heterologous mRNA in Xenopus oocytes. Nature (Land.) 292,862-864.

Sumikawa K., Parker I., and Miledi R. (1986) Xenopus oocytes as a tool for molecular cloning of the genes coding for neurotransmitter receptors and voltage-operated channels. Fortschritte der Zoologie, Band 33: Membrane Control, Gustav Fischer Verlag, Stuttgart, pp. 127-139.

Taglialatela L., Toro L., and Stefani E. (1992) Novel voltage clamp to record small, fast currents from ion channels expressed in Xenopus oocytes. Bzophys, 1. 61, 78-82.

Taglietti V. and Toselh M. (1988) A study of stretch-activated channels m the membrane of frog oocytes: interactions with Ca2+ ions. J. Physiol. (Land.) 407,311-328.

Takumi T., Ohkubo H., and Nakanishi S. (1988) Cloning of a membrane protein that induces a slow voltage-gated potassium current. Sci- ence 242,1042-1045.

Tekle E., Astumian R. D., and Chock I’. B. (1991) Electroporation by using bipolar oscillating electric field: an improved method for DNA trans- fection of NIH 3T3 cells. Proc. N&l. Acad. Sci. USA 88,4230-4234.

Verhoeff-de Fremery R. and Griffin J. (1987) Anurans (frogs and toads), in The Care and Management of Laboratory Animals, 6th ed. (Poole T., ed.), pp. 773-783.

White M. W. and Aylwm M. (1990) Niflumic and flufenamic acids are potent reversible blockers of Ca2+-activated Cl- channels in Xenopus oocytes. Mol. Pharmacol. 37,720-724.

Woodward R. M. and Miledi R. (1989) Effects of defolliculation on membrane current responses of Xenopus oocytes. J. Physiol. (Land.) 416, 601-621.

Yang X.-C. and Sachs F. (1989) Block of stretch-activated ion channels in Xenopus oocytes by gadolmium and calcium ions. Science 243, 1068-1071,

Yang X.-C. and Sachs F. (1990) Characterization of stretch-activated ion channels in Xenopus oocytes. I. Physiol. (Land.) 431,103-122.

Yang X.-C., Karschin A., Labarca C., Elroy-Stein O., Moss B., Davidson N., and Lester H. A. (1991) Expression of ion channels and receptors in Xenopus oocytes using vaccinia virus. FASEB J. 5,2209-2216.

ACh receptor, 299 Acrylate, 116 A/D converter, 42,46,47,49 Adenosine, 159 Adenovirus, 269 /3-Adrenergic receptors, 299 Aeromonas species, 267 Aerosol-resistant tips (ART),

195 Ag-AgCl (see Silver/silver

chloride) Agarose gel electrophoresis,

194,210-212,216 Agatoxin, 55 Air-gate, 106 Aliasing, 48,9 Aluminosilicate(s), 3, 4, 14,17,

32,116 AM esters, 71 Amiloride, 295 y-Aminobutyric acid (GABA),

58-60,284,292,296, 299

4-Aminopyridine (4-AI’), 50, 55,56,133,134,138

Amino-3-hydroxy-5-methyl-4- isoxazolepropionate (AMPA) receptor, 200, 204,206-208,211,213, 215,217-219,222,223, 225-227,229

Amphotericin B, 7,160,162, 167,168

Amplification, 207,221,222

Index

Amplifiers, 15,19,38,40,41, 45,46,66,288

Annealing, 199 Anterograde labeling, 242 4-AI? (see 4-Aminopyridine) Apamin, 55 ART (see Aerosol-resistant tips) ATP, 97,98,109,119,155-157,

160,162 Aureomycin, 266 Autographa californica, 299 Avermectin B, 55

B104 cells, 60 Ba2+ (see Barium) Baculovirus, 299 BAPTA, 66,160,252,296 Barium (Ba2+), 55, 115 Barley, 259 Barths solution, 285 Bay K8644,55 Bernoulli effect, 91 Bessel filters, 24,47 BfaI, 220 BgZI, 211-214 Biotin, 70 Bleach, 116 Blebs, 90,91 Blind injection (of DNA), 282,

283 Blind patching, 244,248 Borosilicate glass, 4,5, 17,32,

109,116 Brain slices, 233-256

307

308 Index

Patch clamp methods in, 234-256

Brain stem, 233 Bright-field optics, 238 Bsp12861,211-214,217

C6 glioma cells, 160 Ca2+ (see Calcium) Ca2+ binding, 254 Ca2+ channels, 55,168,249,252 Ca2+ imaging, 253 Ca2+ signalling, 255 CaCl,, 239,300 Cadmium (Cd2+), 55 Calcium Green-l, 252,254,255 Calcium (Ca’+), 55,66,76,94,

97,98,118,125,133,155, 156,161,164,168,170, 239,249-255,274,296

CAMP, 156 Capacitance, 19,20,26,27,29,

31,45,49-52 Capacitive currents, 49,50-52 Capsaicin, 55 Cardiac muscle, 233 Cardiac myocytes, 136 Cat skeletal muscle, 260 CCD camera, 238,250 Cd2+ (see Cadmium) cDNA(s), 193,194,197-201,

206,210-212,221,291 Cell bodies, 238 Cell culture, 209 Cell-free membrane patches,

101,102 Cell-free ion-channel record-

ing, 89-119 Channel rundown, 95-101 Rapid change of bathing

solution, 101-107 Responses to rapid

concentration changes, 107-l 14

Tips for successful cell-free channel recording, 114-118

Vesicle formation, 90-95 Cerebellar slices, 240,241,253 Cerebellar vermis, 236 Cerebellum, 233,235,236,

240-243,249,253 Cesium (Cs+), 53,55,124-133,

138,164, 196, 197,239, 247

CF (see Climbingfiber) Channel rundown, 95,97,101 Charybdotoxin, 55 Chloride (Cl-), 55, 67, 116,137,

156,161, 164,239,288, 292

Chloridizing, 116,117 Chlorine, 264 Chlorotoxin, 55 Citrobacterfreundii, 267 Cl- (see ChEoride) Climbing fiber (CF), 243,249 CLSM (see ConfocuI her

scanning microscopy) CNQX, 55 Colchicine, 97 Cold water worm infestation,

266 Collagenase, 76,273,274 Concentration clamp

technique, 141-151 Kinetic studies, 151-156 Limitations, 151 Preparations, 144-146 Setup, 142,143

Concentration jumps, 107,112 Conductance-voltage (C-V)

curves, 61 Confocal laser scanning micro-

scopy (CLSM), 252 Confocal microscopy, 237,

252,254,255

Index 309

w-Conotoxin, 55 Corpus callosum, 233 Cs+ (see Cesium) CSCI, 196,197 Current-to-voltage converter.

40,41 C-V (see Conductance-voltage

curves) Cytochalasin B, 97 Cytoskeleton, 85,89,91 Cytosolic extract, 159

D/A (digital-analog)-A/D converters, 41

Data evaluation and analysis, 46 Data presentation, 64 Data storage, 49 dBcAMP, 69 Dead volume, 103 Defolliculation, 272,273 Delayed rectifier potassium

channels (DRKl), 98, 114

Denaturation, 199 Dendrites, 236, 238,247,248,

254 Dendrotoxin, 65 Deoxyribonucleotides

(dNTPs), 196-198,201, 204,205,210

DEPC (see Diefhylpyrocarbonafe)

Dephosphorylation/ phosphorylation processes, 157

Depression, 158 Desensitization, 157 Dextrose, 125 Dialysis of cytoplasm, 155 DIC (see DiffeerenfiaZ-

interference-contrast) Dielectric constants, 4, 18,23-

25

Dielectric noise, 18, 19,24,27, 28,30,31

Diethylpyrocarbonate (DEW), 278

Differential-interference- contrast (DIC), 238

Diffusion, 111-113, 162, 166, 247

Dimethylsulfoxide (DMSO), 125,162,164,165,167

Direct nuclear injection, 283 Dissecting current

components, 52 Dissecting microscope, 244 Dissipation factor, 18 Distributed RC noise, 21,22,

24,30,31 Dithiothreitol (DTT), 196,197,

204,209 DMEM (see Dulbecco’s

Modified Eagle Medium) DMSO (see Dimefhylsuljoxide) DNA(s), 194, 199,201,210,

212,215,216,220,222, 223,226,261,274,275, 278,280-282,284,285, 287,296-299

DNase, 278 dNTPs (see

Deoxyribonucleofides) Donnan potential, 156,164 Double electrode voltage

clamp methods, 69 DRKl (see Delayed rectifier

potassium channels) DTT (see Difhiofhreiol) Dulbecco’s Modified Eagle

Medium (DMEM), 125 Dyes, 70,240,242,245,252,

254,255

Eco47111,211-214,217 EcoRI, 211,212

310 Index

EDTA, 94 EGTA [see 5 Ethylene gIycol

(bis jkxrzinoethyl ether) N,N,N’,N’-tetraacetic acid1

Elastomer coating, 7,8,18-21, 26

Elastomers, ll-14,25 Electrode pullers, 14-40,279 Electrophoresis, 222,226,228 Elongation, 199 EPSCs (see Excitatory

postsynaptic currents) EPSC amplitude, 253 EPSPs (see Excitatory

postsynaptic potentials) Ethanol, 11,116,202 Ethidium bromide

fluorescence, 212 5-Ethylene glycol-bis (B-

aminoethylether) N,N,N’,N’-tetraacetic acid (EGTA), 66,94, 125,159, 196,239,296, 300

Evans Blue, 242 Excitatory postsynaptic

currents (EPSCs), 249, 252,254,255

Excitatory postsynaptic potentials (EPSPs), 249,254

Fiberoptic ring illuminator, 12 Filaments, 103 Filters, 24,35,40,46,47,93 Firepolishing, 4,8,9,13,76,

116,117,234,236,287 Flip and flop forms, 220-225 Flufenamic acid, 296 Flu03,252 Fluorescence, 70,240,242,

250,255,263

Fluorescent dyes, 70,240,242, 245,252,254,255

Fluorescent indicators, 70 Fluoride ions, 94 Formaldehyde, 71 Frog-Ringer solution, 300 Frontal cortex, 233 FTX funnel, 55 Fura-2,240,245 Fura-red, 252

G proteins, 275,295,297,299 GLZ~ seals (see Gigaohm seals) GABA (see FAminobutyric

acid) GABA, receptors, 284,292,

296,299 GAD (see Glutamic acid

decarboxylase) B-Galactosidase, 285 Gap junctions, 70 GCL (see Granule cell layer) Gel electrophoresis, 222,228 Gelatin, 197 Geloader, 202 Gene transfer, 299 Genomic DNA, 199 Gentamycin, 300 GFAP (see Glialfibrillary acidic

protein) GH, pituitary cells, 159 Gigaohm (GQ) seals, 1,2, 17,

19,30,90,94,98,123, 127,196,202

Gigaseal patch, 166 Glass fibers, 274 Glass types, 39 Glassware, 278 Glial cells, 62, 195,206,216,

218,239 Glial fibrillary acidic protein

(GFAP), 200,206,211, 215,216,218

Index

Globins, 259 GluRl-4 amplified fragments,

210,215 GluRl-4 amplified product,

208 GluRl-4 cDNAs, 209,212,216,

220,222,223 GluRl-4 coding sequence, 207 GluRl-4 mRNAs, 216,218 GluRl-4 subunits, 219,223 Glutamate, 160,206,238 Glutamate receptors, 206 Glutamic acid decarboxylase

(GAD), 200,218,219, 221,226,228

Glycosylation, 298 Granule cells, 241,249,250 Granule cell layer (GCL), 249,250 Grease-gap technique, 292

H+ (see Protons) HaeIII, 213,225 Heat polisher, 15 Hematocrit capillaries, 202 HEPES (see 2-

Hydroxyethylpiperuzine- n’-2-ethanesulfonic acid)

HguI, 221 High-lead glass, 4,6,8,32 Hippocampal cells, 208,218,

219,222,225 HippocamRus,.208,218,219,

222,225,233,236 Honey bee, 259 Horseradish peroxidase, 70 HpaI, 221 5-HT c receptor, 260 2-Hy&oxyethylpiperazine-N’-

2-ethanesulfonic acid (HEPES), 125,159,160, 164,168,196,239,300

Hypertonic stripping solution, 300

311

Hypothalamus, 233

I-V curves, 57-59,61,62,70 Inactivation curves, 61 Incubation chamber, 235 Infrared differential-

interference contrast (IR-DIC) video microscopy, 240,241, 244,245,248

Inhibitory postsynaptic currents (IPSCs), 250,253

Inhibitory postsynaptic potentials (IPSPs), 250

Injection micropipets, 279 Inside-out membrane patches,

89-91,96,107,113,118, 293

Interneurons, 240,250 Intrapipet dialysis, 163 Ion channels, 50,52,53,95,259,

260,284,286,296,298 In cell-free patches, 95

Ionophoresis, 188,189 IPSCs (see Inhibitory

postsynaptic currents) IPSPs (see Inhibitory

postsynaptic potentials) IR-DIC (see Infrared

differential-interference contrus t video microscopy)

K+ (see Potassium) K,,, channel(s), 96, 98 KA (see Kuinute) Kainate (KA), 206,213,215,

222,225,227 Kainate (KA) receptors, 206 KCl/agar type bridge, 138 KCl, 196,197,210,239,288 KG-12,ll Klebsiellu pneumoniue, 267

312 Index

Kimax, 4, 7, 10

La3+ (see Lanthanum) LacZ gene, 285 Laminar flow methods, 103 Lanthanum (La3+), 55,295 Leakage currents, 50,51 Levamisole hydrochloride,

267 Levenberg-Marquard

algorithm, 63 Li+ (see Lithium) Liquid filament solution

switch, 104 Lithium (Li+), 71, 164 Locust muscle, 259 Long-term depression (LTD),

252 Loose patch voltage clamp

technique, 173-190 Techniques, 174-186 Variations of the method,

186-190 Low-noise recording 15 LTD (see Long-term depression) Lucifer Yellow (LY), 70,240 Lutetium, 295 LY (see Lucifer Yellow)

Magnesium (Mg2+) (see also MgClJ, 98, 118, 125, 156,161,164

Maize, 25 Manipulators, 5,42,243,244,281 MBS (see Modified Barths

Solution) Mechano-gated (MG)

channels, 78,81-85 Medial septum, 233 Membrane-cytoskeleton

interactions, 83-85 Meniscus, 128 Methanesulfonate, 296

Methanol, 11,161,164 MG (see Mechano-gated

channels) Mg2+ (see Magnesium) MgATP, 98 MgCl,, 196-198,201,204,205,

208,210,220,239 Microdispenser, 282 Microdissection, 76 Microfilaments, 97 Microloader, 202 Micromanipulator, 42,243,244 Micropipet puller, 126 Microscopes, 43,80,84,111,

238,239,244 Microtubules, 97 Micro water blasting, 76 Mineral oil, 15, 196,280,281 Modified Barths Solution

(MBS), 269,271,272, 274,277,282,285,286, 300

Moloney Murine Leukemia Virus, 196,209

Moth, 299 Mouse kidney, 259 Mouse mammary tumor

virus, 259 mRNA(s), 193, 199,200,211,

215,217,218,227-229, 250,259,260

MseI, 220

N-Methyl-D-aspartate (NMDA) receptors, 206,260

N-Methyl-d-glucoronate (NmDG), 53

Na+ (see Sodium) Na+ channels, 55,249 Na+ currents, 53,64,67,70 Nematodes, 266 Neocortex, 236,248

Index 313

Neocortical neurons, 249 Nescofilm, 281 Neuroblastoma cells, 124,135 Neurons, 195,227,237,239-

241,247,249,250 Neuropil, 243 Ni2+ (see Nickel) Nichrome wire, 13 Nickel (Ni*+), 55 Nifedipine, 55 Niflumic acid, 296 NMDA receptors (see N-

Methyl-D aspartate receptors)

NmDG (see N-Methyl-d- glucoronate)

Noise, 17,30,32,82,100 Normarski optics, 42 Nynquist frequency, 48,49 Nynquist Sampling Theorem,

47,48 Nystatin, 7,160,162-165,169 Nystatin-fluorescein mixture,

162,166

Oil-gate, 103-105,110,111 Okadaic acid, 97 Olfactory bulb, 206,208,233 Oligonucleotide primers, 196,198 Oligos, 198,200 Oocytes (see also Xenopus

oocytes), 81, 82,94,98- 118,259-261,268,271- 277,279,281-294,296, 298

Microinjection and ion- channel expression, 259-300

Oscilloscope, 40, 44 Outside-out patches, 293

P/4 protocol, 50 Parafilm, 117,118

Patch micropipet filling solution, 155

Patch-clamp electrodes, l-35 Noise properties of patch

pipets, 17-35 Patch electrocde

fabrication, lo-17 Pipet glass properties, 3-5 Whole cell pipet properties,

5-9 PCL (see Purkinje cell layer) PCR (see Polymerase chain

reaction) PCR buffers, 197 PCR product, 228 Perforated patch-clamp

technique, 155-169 Amphotericin B use, 167,168 Dialysis of cytoplasm, 155-158 Nystatin use, 160-167 Perforated vesicles, 168, 169 Preventing dialysis, 158-160

Perforated vesicles, 168-170 Perfusion of patch pipets, 123-151

Applicability and problems, 135-139

Methods, 124,128 Resuts with, 129-135

Perspex, 116 Perturbation methods, 78 Phosphorylation-dephosphor-

ylation mechanisms, 97 Piezo crystal, 103 Piezoelectric valve, 79, 81,227 Pipet cone angle (O), 108 Pipet noise sources, 30 Pituitary gland, 233 Pituitary tumor cells, 168 Platinum, 9,13 Platinum iridium wire, 9 Polyene antibiotics, 160 Poly-L-lysine, 273

314

Polymerase chain reaction (PCR) (see also RT- PCR), 70,193-198,200- 202,205,207,208, 210-212,215-217,219, 220,222,227,228,254

Positive pressure, 243,244 Potassium (K+) (see also KCZ,

Potassium channels, and Potassium currents), 110,111,124,127-132, 133,135, 137,138, 156, X4,238,247

Potassium (K+) channels, 55, 67, 114, 129, 131,132, 138,168,260,299

Potassium (K+) currents, 53, 54,56,64,125, 134,164

Power spectral density @‘SD), 22,23,29,34

Pressure clamp, 79,81,82 Pressure/patch-clamp

methods, 75-85 Applications, 83-85 General cell attached

recording procedures, 76,77

Methods of applying suction, 77-79

Properties of the pressure clamp, 79-82

Pro tease, 76 Perspex, 282 Protein kinases, 98,156 Protons (H’), 156,250 PSD (see Power spectral density) Pseudocapillaroides xenopi, 266 Pseudomonas, 267 Purkinje cell layer (PCL), 242 Purkinje cells, 206,213,214,

218,236,240-243,248- 255

Pyrex, 4, 7,9,11

Index

Ql-4939,27 QA receptors (see QuisquaZate

receptors) Quartz, 35,10,13,14,21,23-

27,31 Quisqu;F6te2\I$A) receptors,

I

R-6101,27,28 Rabbit reticulocyte, 259 Rat spleen, 259 Rb+ (see Rubidium) RC noise, 23 RC time constant, 38 Re-CP noise, 28,30,33 Rebound potentiation (RR), 252 Recording chamber, 238 Red leg, 265 Resistance, 7,11,20-22,30,34,

41,45,65-67,116,158, 159,163,209,245,293

Restriction analysis, 194,207,210 Restriction enzymes (see also

the individual enzymes), 211,212,225

Retina, 233 Retrograde labeling, 242 Reversal potentials, 70,131 Reverse transcriptase (RTase),

196-198,200,204,209 Reverse transcription (RT)

reaction (see aZso XT PCR), 193,195,197, 203-205,208,210

Righting reflex, 268 Ringer solution, 274,285,287,

295 RNA harvesting, 202 RNAs (see also mRNAs), 200,

201,208,209,261,274, 275,278,280-282,284, 285,287,296-298

RNase, 201,278

index 315

RNasin, 196,197,200,204,209 RP (see Rebound potentiation) RS compensation, 67 RT (see Reverse transcription

reaction) RTase (see Reverse

transcrip tase) RT-PCR, 193-229,233,250

AMPA receptors, GFAP and GAD, 206-229

Materials and Methods, 195-205

Rubidium (Rb+), 55 Rundown, 96,97,100,114,

155,157

Saccharomyces cerevisiae, 299 Scorpion toxin, 55 Seal formation, 246 Seal noise, 29 Silicone fluid, 15 Silver/silver chloride (Ag/

A&l), 106, 127,137, 236,289

Silver wire, 116, 117, 127 Simplex algorithm, 63 Single-channel pipets, 11,13 Skeletal muscle, 259 Small-tipped electrodes (high

resistance electrodes), 158

SO 2- (see SuZfbte) Soda lime glasses, 6,17,28 Sodium (Na+) (see also Na+

channels and Na+ currents), 55, 125, 156, 164,239,250

Sonica ting, 11 Southern blot analysis, 194,

213,214,228 Space clamp, 68,247 Spinal cord, 233 Spinal cord astrocytes, 62

Spodoptera frugiperda, 299 Spontaneous tight seals, 77 Steromicroscope, 272 Stimulator, 40,41 Stimulus protocols, 53 Stokes-Einstein radius, 161 Streptomycetes, 161 Striatum, 233 Stripping-solution, 274 STX, 55 Suction-induced tight seals,

77 Sudan IV, 81 Sulfate (S042-), 161,164 Sulfonylurea drugs, 98 Sylgard, 12,13,16,21,26-28,

31,32,115

Tail current,131 Tuq buffer, 197 Tuq polymerase, 196,198,201,

204,205,207,210,211, 221

TEA (see Tetraethylammonium) Teflon, 13,249 Tetraethylammonium (TEA),

50,55,129-131,239 Tetrodotoxin (TTX), 50,55 Texas Red, 240 Thermal voltage noise, 28 Thermocycle PCR programs,

199 Thermocycler, 210 Thin-film noise, 19,30 Thyrotropin releasing

hormone (TRH), 159 Time constants, 63 Torpedo electric organ, 259 TRH (see Thyrotropin releasing

hormone) Tricaine, 268 Trituration, 76 Trypsin, 98

316 Index

TTX (see Tetrodotoxin) Tungsten, 4 Two-electrode voltage clamp,

287-289 Tygon, 127 Type II cells, 222,223,226,227

U-Tube microperfusing system, 209

Variances, 100 VCR, 40,42 Verapamil, 55 Vesicles, 89,90-95,169,170 Video monitor, 239 Viscoelastic properties, 84 Visual cortex, 233,236 Vitelline Membrane, 274 Voltage-activated currents, 99 Voltage clamp, 68,69,173-

190,247,254,287-291 Voltage-gated ion channels,

260

Whole-cell patch-clamp recordings, 37-72

$X174,213,225

Xenopus oocytes microinjection and ion-channel expression, 259-300

Comparison with alternative expression systems, 296-299

Culture/incubation of, 285- 287

Electrophysiological recording from, 287-296

History and application, 259,260

Husbandry, 261-267 Microinjection of, 277-285 Physiological solution

composition, 300 Removal of ovary tissue,

267-275 Selection of oocytes, 275-277

Xenopus (see also Xenopus oocytes microinjection and ion-channel expression), 81,118, 259-269,271,273,275, 277,279,287,289,292, 294‘296,298

Yeast, 299

Documents

Patch-Clamp Applications and Protocols