Wholde Patch Clamps Recordings

Patch-Clamp Recordings 35

0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

35

From: Neuromethods, Vol. 35: Patch-Clamp Analysis: Advanced TechniquesEdited by: W. Walz, A. A. Boulton, and G. B. Baker @ Humana Press Inc., Totowa, NJ

Whole-Cell Patch-Clamp RecordingsHarHarHarHarHarald Sontheimer and Christopher B. Ransomald Sontheimer and Christopher B. Ransomald Sontheimer and Christopher B. Ransomald Sontheimer and Christopher B. Ransomald Sontheimer and Christopher B. Ransom

1. Introduction

The patch-clamp recording technique, which measures ioniccurrents under voltage-clamp, was design to study small patchesof membrane in which near-perfect control of the transmembranevoltage can be readily achieved. The recent application of patch-clamp methodology to the analysis of whole-cell current actuallydefies many of the original design requirements. Nevertheless, whole-cell recordings are routinely used in electrophysiology laborato-ries to study electrical currents carried by ions through ion channels,neurotransmifter receptors, and electrogenic transporters in celltypes of virtually any origin. Since the introduction of the patch-clamp technique in 1981 (Hamill et al., 1981) and the subsequentrapid development of commercial amplifiers, this method of intra-cellular recording has essentially replaced sharp electrode record-ings, particularly in the study of cultured cells.

This procedure and its applications have been the topics ofnumerous excellent reviews and book chapters (see RecommendedReadings). In addition to summarizing some basic concepts, thischapter specifically emphasizes hands-on procedures and proto-cols. It is hoped that the chapter will effectively complement pre-vious accounts of the patch-clamp technique.

2. Principals (Why Voltage-Clamp?)

Electrophysiologists are especially interested in the activity ofmembrane proteins, which provide conductive pathways throughbiological membranes: ion channels, transmitter receptors, or elec-trogenic ion-carriers. Channel activity, whether through voltage-

2

36 Sontheimer and Ransom

0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

dependent or ligand-gated ion channels, results in changes of mem-brane conductance, which can be most conveniently evaluatedby recording membrane currents at a constant membrane volt-age. Under such “voltage-clamped” conditions, current is directlyproportional to the conductance of interest.

A two-electrode voltage-clamp design was first introduced inthe seminal studies of Hodgkin and Huxley (1952) for the study ofionic conductances of the squid giant axon. In this application, oneof the electrodes serves as voltage sensor, whereas the second func-tions as a current source both interconnected through a feedbackamplifier. Any change in voltage detected at the voltage electroderesults in current injection of the proper polarity and magnitudeto maintain the voltage signal at a constant level. The resultingcurrent flow through the current electrode is assumed to flowexclusively across the cell membrane and as such is proportionalto the membrane conductance (mediated by plasma-membrane ionchannels). The major disadvantage of this technique, however, isits requirement for double impalement of the cell, which restrictsits application to rather large cells (>20 µm) and prevents studyof cells embedded in tissue.

In an attempt to solve this problem, single-electrode switchingamplifiers were developed that allowed the use of one electrodeto serve double duty as voltage and current electrode. For shortperiods of time, the amplifier connects its voltage-sensing inputto the electrode, takes a reading, and subsequently connects thecurrent source output to the same electrode to deliver current tothe cell. This approach, however, is limited in its time-resolutionby the switching frequency between the two modes, which mustbe set based on the cell’s RC time constant (the product of cellinput resistance and capacitance). Both single-electrode switchclamp and double-electrode voltage-clamp allow direct measure-ment of the cells voltage and avoid the introduction of unknownor unstable voltage-drops across the series resistance of the cur-rent passing electrode.

The whole-cell patch-clamp technique similarly uses only oneelectrode. However, in contrast to aforementioned techniques, ituses the electrode continuously for voltage recording and passageof current. Consequently, the recording arrangement contains anunknown and potentially varying series resistance in form of theelectrode and its access to the cell. For the technique to deliver


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

satisfactory results, it is essential that this series resistance be smallrelative to the resistance of the cell. Numerous measures are takento satisfy these requirements (see below) including the use of blunt,low-resistance electrodes, small cells with high impedance, andelectronic compensation for the series resistance error. When effec-tively utilized, the whole-cell technique can yield current record-ings of equal or superior quality to those obtained with double- orsingle-electrode voltage-clamp recordings.

3. Procedure and Techniques

3.1. Pipets3.1. Pipets3.1. Pipets3.1. Pipets3.1. Pipets

In contrast to sharp electrode recordings that utilize pipets withresistances of >50 MΩ, comparatively blunt low-resistance (1–5 MΩ)recording pipets are used for whole-cell patch-clamp recordings.This is done for at least two reasons: 1) series resistance shouldideally be two orders of magnitude below the cell’s resistance,and 2) blunt electrodes (1–2 µm) are required to achieve and main-tain mechanically stable electrode-membrane seals.

As described in Chapter 1 (Levis and Rae), electrodes can be man-ufactured from a variety of glass types. Although it has been fre-quently reported that glass selection has a significant influence onthe quality of seal or the frequency at which good seals are obtained,little scientific evidence supports this notion. In our laboratory, wehave found it to be more critical that glass pipets are thoroughlycleaned by soaking in acetone for 1/2 h followed by drying at 180ºCin a lab oven. To achieve the shape ideal for sealing membranepatches, electrodes are pulled from capillary glass pipets in a two-or multistage process using commercially available pullers, suchas those of Narashige, Brown-Flaming, and others. An additionalstep of firepolishing the tips of pipets can significantly improvethe likelihood of seal formation. As an added benefit, the tips ofpipets with very low resistance (steep, tapering tip) can be fire-polished to reduce the tip diameter to appropriate levels. Cellsurfaces can also be enzymatically cleaned during isolation andculturing procedures, or just prior to each experiment. Once anappropriate set of variables, i.e., cell preparation, glass type, elec-trode resistance, and shape, is identified, the success rate for stableelectrode-membrane seals should be between 50–90%.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

3.2. Electronic Components of a Setup3.2. Electronic Components of a Setup3.2. Electronic Components of a Setup3.2. Electronic Components of a Setup3.2. Electronic Components of a Setup

The electronic components of a patch-clamp setup are compar-atively few: A patch-clamp amplifier, oscilloscope, stimulator, com-puter, an optional external signal filter, and VCR recorder. High-quality patch-clamp amplifiers are available from a number of manu-facturers. At present, most laboratories utilize either a List EPC 9or an Axon Instruments Axopatch 200 amplifier. These amplifiersare overall similar in design and are both equipped with multi-stage Bessel filters, variable gain settings, and at least 2 feedbackresistor 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 con-tained in the headstage. Its characteristics are described in detailelsewhere (Sigworth, 1983; see Fig. 1). Current flow through theelectrode (Ip) across a resistor of high impedance (R) causes avoltage drop that is proportional to the measured pipet current(Ip). An operational amplifier (OpAmp) is used to automaticallyadjust the voltage source (Vs) to maintain a constant pipet poten-tial (Vp) at the desired reference potential (Vref). As the responseof the OpAmp is fast, it can be assumed that for all practicalpurposes Vp = Vref. When current flows across the membranethrough ion channels, Vp is instantaneously displaced from Vref.The OpAmp alters Vs to generate an Ip that will exactly opposethe displacement of Vp from Vref. Thus, the current measuredduring a patch-clamp experiment (membrane current flow) isequal and opposite to Ip. In their whole-cell mode, patch-clamp

Fig. 1. Scheme of current-to-voltage converter (for details, see Sigworth,1983). Abbreviations used: Pipet current lp, feedback resistance R, opera-tional amplifier OpAmp, reference potential Vref, voltage source Vs.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

amplifiers use a 500 MΩ feedback resistor allowing to measurecurrents of up to 20 nA. For some amplifiers a low gain 50 MΩheadstage is available that can pass currents of up to 200 nA, how-ever, its use sacrifices the use of capacitance and series resistancecompensation.

Although some amplifiers, such as the Axopatch 1 D, havebuilt-in stimulators, most electrophysiologists prefer the use ofan external stimulator that offers greater versatility. Low-cost micro-computers serve as both digital stimulator and on-line recorder.While the use of a microcomputer is not essential, the growing num-ber of affordable high-quality hard- and software products havemade computers and D/A(digital-analog)-A/D converters stan-dard laboratory equipment. Data can be collected and digitizedon-line at up to 330 kHz and can be stored to the hard disk of acomputer. All necessary components can be purchased at a pricewell below that of an external stimulator alone.

Most patch-clamp amplifiers are equipped with 4 pole Besselfilters, which are of sufficient quality to filter data. However, theuse of an external 8 pole Bessel filter (such as a Frequency Devices,Series 920) allows the selection of a wider range of cutoff frequen-cies and has better frequency responses, specifically a sharper roll-off (See Subheading 4.1, Filtering).

Additional equipment is recommended for specific data collec-tion needs. If data is to be collected for extended periods of time,and if the sampling rates are roughly 10–40 kHz, a VCR tape recorderis a useful interim storage medium. Data analysis can subsequentlyby done by playing data back off-line to a computer equipped withan A/D converter. The introduction of larger hard drives and con-venient inexpensive means to back up data digitally (JAZ disks, CDburners) have made the use of VCR data storage less popular. Evenin the presence of a computer-based data-aquisition system, oscil-loscopes are a convenient means for monitoring data collected byeither microcomputer or VCR, and are essential for the “debug-ging” of environmental electrical noise from a recording setup.

3.3. Recording Configuration3.3. Recording Configuration3.3. Recording Configuration3.3. Recording Configuration3.3. Recording ConfigurationThe 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 servesas the foundation of the recording setup. A recording chamber ismounted to the stage of the microscope (Fig. 2B). Alternatively, if


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

Fig. 2. Whole-cell patch-clamp setup (A) and recording chamber (B).Photomicrographs of a typical recording setup based on a Nikon Diaphotmicroscope. (A) Patch-clamp headstage with electrode holder mountedon a swivel PVC clamp and attached to 3-axis manipulator constructedfrom 3 series 420 microtranslation stages (Newport lnst.) (B) Close-up viewof flow-through recording chamber. Chamber from Warner Instrumentsmodified after our own design.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

constant perfusion is not desired, a 35 mm petri dish can be usedas recording chamber, in which case, cells may be grown directlyon the 35 mm dish. To use Normarski optics, we prefer the use ofa Plexiglas recording chamber, which has a glass coverslip as thebase (Fig. 2B). Various types of recording chambers are commer-cially available, all of which serve well for most purposes. Elec-trodes are typically placed under visual control (400×) onto a cellby use of a high-quality, low-drift micromanipulator. Numeroushydraulic, piezoelectric, and mechanical designs are commerciallyavailable, each offering unique benefits. Hydraulic manipulators,such as Narishige model MO-203 combine precise movement withlarge travel, are very versatile and easy to use. Piezoelectric manip-ulators, such as the Burleigh PCS 500, only allow 70 µm of travel,and as such are only useful when used in combination with a coarsepositioning manipulator. Piezoelectric systems provide excellentstability and are the instruments of choice for excising patchesfor single-channel recordings (see Chapter 3, this volume). Stable,low-cost mechanical manipulators (such as Soma instrumentsseries 421) can be assembled from single-axis translation stages.Their modular design, combined with micrometer screws and DCmotors- (e.g., model 860A) make them extremely versatile. Thearrangement shown in Fig. 2A includes three 421 stages of whichthe X and Y axes are controlled manually by micrometer screws,whereas the Z axis for electrode placement on the cell uses an 860ADC motor controlled by a hand-held battery operated manipula-tor (model 861). This arrangement has proven to be very stableand relatively inexpensive.

3.4. Experimental Procedure3.4. Experimental Procedure3.4. Experimental Procedure3.4. Experimental Procedure3.4. Experimental Procedure

During electrode placement, electrode resistance is monitoredcontinuously by applying a small voltage pulse (1–5 mV, 2–10 ms)to the electrode (Fig. 3A). Once in contact is made with the cell,electrode resistance spontaneously increases by 10–50%. Applica-tion of gentle suction to the electrode by mouth or a small syringequickly results in the formation of a gigaseal (Fig. 3B). At this point,seal quality can be improved by applying a negative holding poten-tial to the pipet. In this cell-attached configuration, pipet capaci-tance transients (Cp) are reduced using the fast compensationadjustment at the amplifier (Fig. 3C). This compensation of pipetcapacitance is essential for proper series resistance compensation.Should compensation be incomplete, coating of future electrodes


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

with Sylgard (Dow Corning) or lowering the bath perfusion levelis recommended to reduce the residual transients and improveCp compensation. Following pipet capacitance cancellation, a briefpulse of suction will rupture the membrane patch under the elec-trode providing low-resistance access to the cell. This also resultsin large capacity transient arising from the added membranecapacitance (Fig. 3D). Immediately after rupturing the membrane,a reading of the cell’s potential should be obtained (at T = 0) as thisaccess potential is as close to the actual resting potential readingthat can be obtained. Within minutes of establishing a whole-cellconfiguration, the pipet contents will equilibrate with the cellscytoplasm and will impose an artificial ionic potential across themembrane. Next, by adjusting the capacitance and series resis-tance (Rs) compensation and gradually increasing the % compen-

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


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

sation, effective Rs compensation should be possible under mostcircumstances. Ideally, access resistance should be <10 MOhmprior to activating Rs compensation. Under these conditions 80%compensation results in a <2 MOhm residual uncompensatedseries resistance. Series resistance and capacitance compensationresult in a change of the step waveform applied without actuallychanging access resistance or the cell capacitance per se (see below).The procedure for establishing whole-cell configuration and allnecessary compensations are nicely illustrated in the manuals ofboth the List EPC7 and Axon Axopatch 1 D amplifiers.

Due to the intracellular perfusion of cytoplasm with pipet solu-tion, it is advisable to wait several minutes prior to obtaining record-ings to assure that this dialysis has reached a steady-state. Diffusionrates depend on molecular weight and charge of the diffusing par-ticle, such that longer diffusion times are expected for relativelylarge, uncharged molecules as compared to small ions. For substancesof molecular-weight 23–156,000 diffusion rates have been deter-mined in adrenal chromatin cells, and these suggest that completedialysis of these small cells occurs on the order of tens to hundredsof seconds (Pusch and Neher, 1988). If dialysis of the cell is incom-patible with the experimental design, e.g., when ionic currents arestudied that are under control of second messengers, the perfo-rated-patch method should be used instead (see Chapter 6, this vol-ume).

4. Data Evaluation and Analysis

4.1. Data Filtering/Conditioning, Acquisition, and Storage4.1. Data Filtering/Conditioning, Acquisition, and Storage4.1. Data Filtering/Conditioning, Acquisition, and Storage4.1. Data Filtering/Conditioning, Acquisition, and Storage4.1. Data Filtering/Conditioning, Acquisition, and Storage

4.1.1. FilteringData are rarely acquired and stored without further modifica-

tions. The analog output of the amplifier is typically amplified tomake effective use of the dynamic range of the acquisition device.In the case of an A/D converter, this typically translates to an ampli-fication range of −10 V to 10 V. At the same time, signals are fil-tered, often using the built-in signal filter.

Filtering of data is both essential and inevitable. Due to the RCcomponents of the cell membrane-series resistance combination,the cell and electrode are essentially a single-stage RC filter. This isimportant to bear in mind as it can significantly affect the true time-resolution of a recording. Assuming that Rs << membrane resistance


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

(Rm), uncompensated Rs will filter any current flow recorded witha −3dB cut-off frequency described by F = 1/ (2π*Rs*Cm). Assum-ing, for example, a cell capacitance of 20 pF and a series resistanceRs of 10 MOhm, values typical of small-cell recordings, currentsacross 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 areoutside the bandwidth of interest. Filtering “condenses” the timedomain of the signal to the domain of interest. The fastest signalsrecorded under whole-cell conditions are on the order of 200–500µsec (2–5 kHz). Note that this is of the same order of magnitudeas the membrane-Rs filter time-constant above. To eliminate high-frequency noise, a low-pass filter is used. An ideal filter has asteep roll-off, and does not greatly distort signals. Four- or 8-poleBessel filters have excellent characteristics for filtering whole-cellcurrents. Comparison of the filter characteristics of a 4- and 8-poleBessel filter are demonstrated by comparing the onset response ofa square pulse before and after filtering at various cut-off frequen-cies (Fig. 4). In these examples, an 8-pole Bessel filter clearly pro-vides excellent signal filtering with least distortions (Fig. 4A–E).However, at cut-off frequencies above 2kHz, 4- and 8-pole filtersdo not differ significantly in the onset or settling time.

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

tant to select the appropriate filter and sampling rates to accu-rately represent the analog signal of interest. It is inevitable thatA/D conversion will reduce the “infinite” dynamic range of theanalog signal to a well-defined step-like range of the digital sig-nal. Commonly used 12-bit converters (for example the AxonDigidata 1200) divide the amplitude range into 4096 discrete steps,which at a −10 V to 10 V signal range will yield steps of 4.88 mV.In theory, signals can be sampled at the highest possible rate sup-ported by the A/D converter. However, practical limitations exist.Most affordable A/D boards sample at 100–330 KHz on a singlechannel, thus allowing to sample in 3–10 µsec intervals. Depend-ing on the duration of the signal, sustained sampling at 10 µsecwill generate very large amounts of data, requiring significantdisk space. A vast majority of this data will not contain necessaryinformation. The Nynquist Sampling Theorem states that the mini-mum sampling rate (Nynquist frequency) required to accurately


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

represent an analog waveform is twice the signal bandwidth. Asa consequence, if the analog filter is set at a −3dB cutoff frequencyof 3 kHz, a minimum sampling frequency of 6 kHz or 167 µs inter-vals is required. While these minimum requirements will allowthe reconstruction of data with little error under most circumstances,a sampling frequency 5 times the −3dB frequency is commonlyrecommended for actual recordings.

4.1.3. AliasingThe 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-frequency domain, a process

Fig. 4. Frequency response of 4- and 8-pole Bessel filter to square pulse.A 2 ms square pulse was applied to a Frequency Device model 902 8-poleBessel filter as compared to the build in 4-pole Bessel filter of the Axopatch1D amplifier. The graphs illustrate differences in response characteris-tics at three commonly used −3dB cutoff frequencies (1, 2, and 5 kHz).


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

called “aliasing.” Essentially these high-frequency signals willaffect and distort those signals within the appropriate frequencydomain. Aliasing can be prevented if signals above the Nynquistfrequency are cut off by a low-pass filter. A proper matching offilter frequency and sampling rate is thus important to accuratelyreproduce analog waveforms. In practical terms this requires thatthe cut-off frequency of a low-pass filter be set to no higher thanhalf the sampling frequency. In the previous example, the low-pass filter was set at 3 kHz.

4.1.4. Data StorageOn-line digitization has the advantage that data can be directly

stored in computer memory or hard disk. This mode of data stor-age is preferred as it gives convenient and fast access to the datafor future evaluation. Using a 12-bit A/D converter, each data sam-ple uses 2 bytes of information. Thus a 2048 sample trace requiresabout 4 kbytes of memory or disk-space. A continuous sampling ofneuronal discharge at a frequency of 50 kHz generates 100 kbyte ofdata every second. A 5-min recording would thus require 300*100kbyte or 30 Mbyte of disk space. It thus becomes apparent that prac-tical limitations exist as to the on-line digitization of data. For pro-longed recordings at high frequencies (<40 kHz), a VCR recorder(such as the Neurocorder) may be used as an interim storage fordata. Segments can subsequently be played back to the A/D con-verter for data analysis. For brief signals, on-line storage is not anissue, as hard-disk space has become affordable at <$0.1 per Mbyte.Recordable CD-R disks are a practical and cheap way to archivedata.

4.2. Leak/Subtraction4.2. Leak/Subtraction4.2. Leak/Subtraction4.2. Leak/Subtraction4.2. Leak/Subtraction

Currents across a cell membrane consist of two components: ioniccurrent flowing through ion channels of interest, and capacitivecurrent that charges the membrane. Capacitive current containsuseful information pertaining to the cell size, as an approxima-tion of cell size (or more precisely, membrane area of the recordedcell) can be derived from the capacitive current. However, in thestudy of ionic currents, capacitive currents are of relatively littleinterest. As 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


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

be done either on-line or off-line. Two protocols for leak subtrac-tion are typically used: 1) If the nature of the experiments permit,currents are recorded sequentially in the absence and presence ofspecific ion-channel blockers (e.g., TEA, TTX, 4-AP). As specificblockers will eliminate ionic current but should not alter the capac-itive or leakage current, subtraction of the two traces/or set oftraces should result in the removal of capacitive and leakage cur-rents. 2) If this approach is not possible, P/N leak subtraction asfirst proposed by Bezanilla and Armstrong (1977) can be obtained.In this subtraction scheme, each “test” voltage step is precededby a series of N (typically 4) “leak” voltage steps of 1/N (1/4 or-1/4 dependent on polarity) amplitude of the test pulse activatedfrom a potential at which no voltage-activated currents are acti-vated. In a P/4 protocol, these 4 * 1/4 amplitude traces are summedtogether (Fig. 5B) and are subtracted from the actual current traceof interest (Fig. 5A) and will isolate the ionic current of interest(Fig. 5C). The example demonstrated in Fig. 5 actually used a P/-4protocol, in which 4 hyperpolarizing pulses of -1/4 amplitude weresummed and added to the current trace of interest. It is importantto obtain the “leakage” current at potentials at which no voltage-activated currents occur. Most often this can be achieved by step-ping to potentials negative of the resting potential (as illustrated).However, some cells express inwardly rectifying (or anomalousrectifying) currents that are active at the resting potential andnegative thereof. Under these circumstances leak currents mustbe recorded at potentials at which the I-V curve is linear and novoltage-dependent currents are activated. Note that subtractionof capacitive and leakage currents is purely cosmetic. It does notactually improve the signal recorded. However, it may allow reveal-ing small current components that would otherwise be hard to iden-tify. If capacitive currents are not of interest, it is recommendedthat P/4 leak subtraction be performed on-line by the data-acquisi-tion program because it significantly reduces the amount of datastored.

4.3. Determination of Cell Capacitance4.3. Determination of Cell Capacitance4.3. Determination of Cell Capacitance4.3. Determination of Cell Capacitance4.3. Determination of Cell Capacitance

Biological membranes are lipid bilayers in which membraneproteins (e.g., ion channels and transporters) are contained. Thespecific capacitance of biological membranes seems to be fairlyconstant. It is relatively independent of cell type and a value of


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

1 µF/cm2 is typical. The capacitance cancellation circuits of patch-clamp amplifiers are normally calibrated in pF, and allow directdetermination of the cells capacitance by adjusting and minimiz-ing the capacity transients in response to a voltage step as discussedpreviously. The dial reading provides at least a rough determina-tion of membrane capacitance on which bases estimates can be madeas to the membrane area, as 1 pF capacitance represents 100 µm2of membrane.

Capacitance can also be derived from the capacity transient atany time during the recording. In response to a voltage step, capac-

Fig. 5. Capacitive and leakage current subtraction using P/-4 method.(A) Whole-cell current recorded in response to voltage step from −120 mVto −20 mV. (B) Summed response of 4 * −1/4 amplitude voltage steps asindicated at bottom of (C). (C) Added response of (A) and (B) eliminatingcapacitive current.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

itance is proportional to the integral of the charging transient,thus it can be derived by determining the area under the transientof a currents trace. Neher and Marty (1982) have developed a verysensitive approach of measuring changes in membrane capacitanceusing a phase-lock amplifier, which measures currents in and out ofphase with a sinusoidal voltage change. This approach can resolvecapacitance changes of 10 fS and has been used to resolve the fusionof synaptic vesicles by a step increase in capacitance.

4.4. Dissecting Current Components4.4. Dissecting Current Components4.4. Dissecting Current Components4.4. Dissecting Current Components4.4. Dissecting Current ComponentsWhole-cell recordings integrate the response of a large number

of potentially heterogeneous ion channels. Separation of these ioniccurrent components is a critical step during or following currentrecordings. Four methods of current isolation are commonly usedto isolate voltage-activated ionic currents:

1. Kinetically:Certain types of ion channels activate and inactivate much faster

than others. For example, Na+ currents typically activate within200–300 µsec and inactivate completely within 2–5 msec. In contrast,K+ currents may take several msec to activate, and often inacti-vate slowly, if at all. Simple current isolation can thus be accom-plished by studying whole-cell currents at different time pointsfollowing stimulation, e.g., determining Na+ current amplitudesat 300–500 µsec, and determining K+ current amplitudes after tensor hundreds of msec.

2. Current subtraction via stimulus protocols:Voltage-dependence of the steady-state activation and inactiva-

tion of currents often allows selective activation of subpopulationsof ion channels. For example, low- or high-threshold Ca2+ currentscan be activated separately by voltage steps originating from dif-ferent holding potentials. Similarly, depolarizing voltage stepsapplied from very negative holding potentials (e.g., −110 mV) canactivate both transient “A” type (KA) and delayed-rectifying (Kd)K+ currents (Fig. 6A). Voltage steps applied from a more positiveholding potential (e.g., −50 mV) will completely inactivate all KAchannels while not affecting Kd activity (Fig. 6B), such that sub-traction of currents recorded with these two protocols effectivelyisolates KA currents (Fig. 6C).

3. Through “Isolation Solutions” (Ion Dependence):It is common practice to specifically design the composition of

ionic solutions to favor movements of desired ions. As mentioned


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

previously, dialysis of cytoplasm with patch pipet contents occursrapidly after whole-cell configuration is achieved, thus allowingfor manipulation of internal ionic concentrations. This access canbe used to block most K+ channel activity by replacing pipet KCIwith impermeant Cs+ or N-methyl-D-glucoronate (NmDG), thus

Fig. 6. Isolation of Ka current by subtraction. Current recordings froma spinal-cord astrocyte expressing both transient (Ka) and delayed recti-fier (Kd) like K+ currents. (A) Currents activated from holding potentialof −110 mV (step protocol see inset). (B) same cell and same voltage-stepprotocol but steps originated from holding potential of −50 mV. Subtrac-tion of (A) − (B) yielded Ka currents in isolation.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

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

4. Current Isolation Via Pharmacology:Numerous natural toxins and synthetic pharmacological agents

exist that can be used to reduce or eliminate specific voltage-acti-vated ion channel activity (for excellent review, TINS Suppl.,VOLUME, 1994). A list of some of the more commonly used agentsis shown in Table 1. Many of these agents are effective againstone particular type of ion channel, such as Tetrodotoxin and den-drotoxin, which target Na+ and K+ channels, respectively. The useof these compounds, either alone or in combination, allows theisolation of specific current(s).

Experimentally, currents are best identified pharmacologicallyby recording a family of current traces in both the absence and

Table 1Commonly Used Ion-Channel Blockers

Channel Type Compound/Reagent

K+ channelsDelayed rectifier (Kd) TEA, Ba2+, capsaicin, 4-AP, margatoxinInward rectifier (Kir) TEA, Cs+, Rb+, Na+, Ba2+

A TEA, 4-AP, dendrotoxinK(Ca)

Big K Charybdotoxin, iberiotoxinSmall K ApaminK(ATP) TEA, Cs+, Ba2+

Na+ channels TTX, STX, CNQX, agatoxin,Scorpion toxin

Ca2+ channelsL-type Nifedipine, verapamil,

BayK8644, Cd2+, La3+

T-type Ni2+, La3+

N-type ω-conotoxin, La3+

P-type FTX funnel spider toxinCl−/anion channels Chlorotoxin, avermectin B, NPPB,

DIDS, Zn2+


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

presence of drug. This is illustrated in Fig. 7, in which the blockof spinal-cord astrocyte K+ currents by 4-amino pyridine (4-AP)is shown. In the control and treated current traces (Fig. 7A,B), theinward sodium current is unaltered. By subtracting the 4-AP-treatedcurrent traces from those of control, one can isolate the currentcomponent that is 4-AP-sensitive. As discussed previously, a sidebenefit to such current subtraction is the elimination of capacitiveand leakage currents.

Neurotransmitter-activated currents are perhaps easier to iden-tify and isolate than their voltage-dependent counterparts, as thesecurrents are induced in a time-dependent manner based on theapplication of exogenous ligands. If need be, ligand-gated currentcan be isolated from background noise by subtracting recordedcurrents in the absence and presence of a given ligand.

Fig. 7. Pharmacological isolation of 4-AP sensitive K+ current. (A)Family of current traces recorded from a spinal cord astrocyte using step-protocol indicated in inset. (B) Recording in the same cell using the samestimulus protocol 2 min after application of 2 mM 4-AP. (C) 4-AP sensi-tive current isolated by subtraction of (A) − (B). (D) Current amplitudesdetermined 8 msec after onset of voltage steps plotted as a function ofapplied potential for current traces in (A–C).


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

4.5. I-V Curves4.5. I-V Curves4.5. I-V Curves4.5. I-V Curves4.5. I-V Curves

Current-voltage (I-V) relationships are perhaps the most effec-tive way to summarize the behavior of voltage- and ligand-acti-vated ion channels. A number of important and useful parametersthat can not be easily accessed from the raw data can be readilyderived from these plots, including: reversal potential, ionic depen-dence/selectivity, voltage-dependence (rectification), activationthreshold, slope and cord conductance, as well as overall qualityof voltage-clamp. I-V curves can be determined in various ways,examples of which are discussed below.

The factors that determine current flow through an open channelare conductance and driving force. While conductance is propor-tional to the number of open channels, driving force is defined asthe difference between actual voltage and the equilibrium potentialfor the ion(s) permeating the channel, also known as reversal poten-tial (Vrev). Thus current can be described as I = G (Vm - Vrev).

A plot of I vs Vm is commonly used to derive G (slope) or Vrev(X-intercept). The current evoked at a given Vm can be measuredusing a variety of protocols.

4.5.1. Peak and Steady-State IV CurvesPeak currents are measured as the largest current activated by

the applied voltage (Fig. 7C). Thus, if a current has a transientpeak current amplitude, such as the KA current in Fig. 7A, analy-sis software can easily determine maximal current and plot thesevalues against voltage (Fig. 7C). If the current to be studied doesnot have this defined peak, steady-state values may be used, typi-cally as recorded at the end of a voltage step. This type of analy-sis was applied to the 4-AP-treated current traces of Fig. 7B, withthe resulting amplitudes similarly plotted in Fig. 7C.

4.5.2. Continuous “Quasi Steady-State” I-V CurvesA convenient way to establish I-V curves is by alteration of the

membrane potential in a continuous way through a voltage ramp.As this ramp can be applied very slowly, it allows to acquire a“quasi steady-state” I-V relationship. This procedure has provenvery useful in determining I-V relationships for transmitter responses.Note, however, that this approach assumes that currents do notinactivate during the length of the voltage ramp. An example of avoltage ramp used to determine the reversal potential of GABAinduced currents is illustrated in Fig. 8. A 200 mV, 400 ms voltage


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

Fig.

8. S

tim

ulu

s p

roto

cols

to d

efin

e re

vers

al p

oten

tial

of G

AB

A-i

ndu

ced

cu

rren

ts. (

A) A

vol

tage

ram

p (i

nset

)w

as a

pp

lied

pri

or to

(Con

trol

) and

at t

he p

eak

of (G

AB

A) G

AB

A-i

ndu

ced

cu

rren

ts. T

he d

iffe

renc

e of

thes

e tw

ocu

rren

t ra

mp

s re

pre

sent

the

GA

BA

-ind

uce

d c

urr

ent

in i

sola

tion

. T

his

curr

ent

was

plo

tted

as

a fu

ncti

on o

fap

pli

ed p

oten

tial

in (B

) to

yiel

d th

e I-

V c

urv

e of

GA

BA

-ind

uce

d c

urr

ents

. (C

) A s

ingl

e 50

mse

c, 8

0 m

V v

olta

gest

ep w

as a

pp

lied

onc

e p

rior

to

and

one

in t

he p

eak

of t

he G

AB

A r

esp

onse

. The

dif

fere

nces

of

curr

ent

amp

li-

tud

es w

ere

plo

tted

as

a fu

ncti

on o

f ap

pli

ed p

oten

tial

(D) t

o yi

eld

two

poi

nts

of a

I-V

cu

rve

for

GA

BA

-ind

uce

dcu

rren

ts. T

he d

otte

d li

ne e

xtra

pol

ates

the

curr

ent r

ever

sal p

oten

tial

.

54


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

ramp as indicated in the inset to Fig. 8A was applied twice. Onceprior to application of GABA (control) and once during transmit-ter application (GABA). By subtracting the two responses oneobtains the transmitter induced current in isolation. As the timeaxis represents a constant change in the voltage applied, it canbe instantaneously replotted as I-V relationship of the transmitterinduce current (Fig. 8B). Using this approach I-V curves of trans-mitter responses can be easily created throughout an experiment.As mentioned earlier, alteration of ionic composition allows toshift the reversal potential and thus allows to determine ion speci-ficity and relative permeabilities.

For those situations where currents inactivate rapidly, someindication as to the reversal potential can be obtained using a sin-gle-step protocol from which the reversal potential can be extrapo-lated. An example of this approach is illustrated in Fig. 8C. Here,a larger concentration of GABA was bath applied which resultedin characteristic receptor desensitization. The cell was maintainedat −80 mV and a single 80 mV step (50 ms) was applied prior toand at the peak of the transmitter response. Current levels at thetwo potentials, −80 mV and 0 mV were then subtracted and plottedas a function of applied potential (Fig. 8D). The line through thetwo data points clearly does not reflect the true IV relationship ofthe response, however, it allows to determine with fair accuracy thereversal potential of the response. To obtain a more complete I-Vrelationship, the two previous voltage steps can be substitutedby trains of voltage steps.

I-V plots may also be used to determine the quality of the volt-age clamp achieved. If the reversal potential (Equilibrium poten-tial) is known, as is the case when isolation solutions limit ionicmovements to only one ion, or when it is clearly established, as inthe case of GABAa receptors that currents are mediated predomi-nantly by one ion. Under the imposed ionic conditions, currents mustreverse close to the theoretical equilibrium potential for the perme-able ion. In poorly voltage-clamped cells, reversal potential is eithernot achieved, or at potentials more positive then the equilibriumpotential (see next paragraph for further discussion).

4.5.3. Conductance-Voltage CurvesIf the reversal potential is known, a conductance voltage (G-V)

curve can be readily calculated by dividing current at each potentialby the driving force (V-Vrev). These curves are typically sigmoidal


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

and can be fitted by a multistage Boltzmann function. Providedthat current through a single channel is linear, conductance isproportional to the number of open channels. Thus the G-V curveresembles an activation curve.

4.5.3. Steady-State Inactivation CurvesVoltage dependence of current inactivation allows to determine

the fraction of channels available for activation as a function ofvoltage. Currents are activated by a step to potentials at which thelargest conductance is achieved. This voltage step is preceded byvariable prepulse potentials at which the membrane is maintainedfor 200–1000ms (Fig. 9A). Current amplitudes at each potentialare normalized to the largest current recorded and plotted as afunction of prepulse potential (Fig. 9B). These curves can be fit-ted to a multistage Boltzmann function. In the experiment illus-trated, a potential of ~ −80 mV yielded about 50% of Na+ channelsavailable for activation (dashed line).

4.5.5. Deactivation I-V CurvesAs voltage-dependent currents are often also time-dependent,

and may activate and/or inactivate in a time-dependent manner,I-V curves as described above cannot distinguish between time-and voltage dependence. An elegant way to determine conductanceindependent of its time-dependence is provided by analyzing thedeactivation process (Fig. 9C,D). Upon termination of the voltagestep, deactivation, which is the reversal of activation, results in tailcurrents (Fig. 9C, dotted line). Immediately after terminating thevoltage step, for a brief period of time current continuous to flowthrough open channels and only subsequently terminates withtime-dependent channel closure (relaxation of tail currents). Thus,current amplitudes measured at the peak of these tails resemblestime-independent current amplitudes, and these typically yieldlinear I-V curves (Fig. 9D).

4.6. Fitting of Time-Constants4.6. Fitting of Time-Constants4.6. Fitting of Time-Constants4.6. Fitting of Time-Constants4.6. Fitting of Time-Constants

Current activation and inactivation kinetics have been well-described by mathematical models. If the model is known, thedata can be fitted to the model and will allow to derive importantkinetic properties, such as time-constants for activation (τm) andinactivation (τh), respectively (Hodgkin and Huxley, 1952). Often


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

Fig. 9. Steady-state inactivation and tail-current analysis. Current record-ings from two different spinal-cord astrocytes. (A) To study steady-statecurrent inactivation, inward Na+ currents were activated by steppingthe membrane to −20 mV for 8 msec. Voltage step was preceded by vary-ing prepulse potential ranging from −130 mV to −30 mV (step protocolsee inset). (B) Peak current amplitudes in (A) were normalized to thelargest current amplitude and plotted as a function of prepulse potential.The data was fitted to a two-stage Boltzmann equation (solid line) toyield steady-state inactivation (h∞) curve. (C) Tail current analysis. Out-ward currents were activated by a 15 ms voltage step from −80 mV to 80mV. This step was followed by a second step to varying test potentialsranging from −30 mV to −120 mV (see inset), resulting in “tail” currents.(D) Current amplitude of tail currents was measured 500 usec after step-ping potential to second step potential (dotted line) and plotted as afunction of applied potential to yield tail current I-V curve.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

times the models used are an oversimplification of the true biol-ogy. This is particularly true for fitting of whole-cell data for tworeasons. First, whole-cell current may be mediated by the com-bined activation of numerous channels types, and second, even ifone can be reasonably sure that currents are mediated by a single-channel population, the biophysics of this channel type, e.g., thenumber of open and closed states may be unknown.

Fitting routines are an integral part of numerous data-acquisi-tion or data-analysis packages. Most commonly these use either aSimplex or a Levenberg-Marquard algorithm to minimize the leastsquared error. Both algorithms are capable of producing excel-lent and fast fitting to small data sets. Examples of Levenberg-Marquard fits are demonstrated for two examples in Fig. 10. Thesewere obtained using the script interpreter of Origin (Mico Cal) byfitting to user-defined functions. The examples illustrated fittedmultiple parameters simultaneously. Thus, in a) transient K+ cur-rent activation and inactivation was fitted to a n4g model of theform: f(t)=AO+Al*(l-EXP(-(t-tO)/,τm))4 * (EXP(-(t-tO)/τh)) as usedby Connor and Stevens (1971) to describe kinetics of A-currents,in b) Na+ current activation and inactivation was fitted to theHodgkin-Huxley equation: f(t)=AO+A1*(1-EXp(-(t-tO)/τm))3 *(EXP(-(t-tO)/τh)). It is important to keep in mind that data fits arenot sufficient to formulate a model, but rather assume that themodel is known and used to derive variables such τm and τh con-tained in the model.

4.7. Data Presentation4.7. Data Presentation4.7. Data Presentation4.7. Data Presentation4.7. Data PresentationPresentation of patch-clamp data has become significantly easier

with the advent of micro-computer based data acquisition, virtu-ally eliminating the use of scissors and glue. Digitized data can beeasily exported in ASCII format and read into numerous powerfulspreadsheet programs (Excel, Lotus, Quattro, SigmaPlot, Origin,Plotlt). More recently, two scientific spreadsheet programs (Origin,Plotlt) have included import modules that allow to import Axonbinary data. Thus without prior conversion, axon data files can beimported for plotting, data analysis, and graphing. The authors’ labor-atory is currently using Origin, which is based on a scientific script-ing language LabTalk for which a script interpreter is part of theprogram. This allows simple programming of frequently used com-mands or sequences of commands into “macros,” which can beassigned to visual buttons on the screen. This allows the computer


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

Fig. 10. Fitting of current traces using least squared fit. (A) Transientoutward K+ and (B) inward Na+ currents were fitted to established kine-tic models using a Levenberg-Marquard algorithm to minimize the leastsquared error. Fitted curves were superimposed on data. The modelsused were: (A) n4h describing transient K+ current according to Connorand Stevens (1971): f(t) = AO+A1*(1-EXP(-(t-tO)/τm))4 * (EXP(-(t-tO)/τh)).(B) the Hodgkin-Huxley (1952) equation to described Na+ current kine-tics: f(τ) = AO+A1 *(1-EXP(-(t-tO)/,τm))3 * (EXP(-(t-tO)/τh)).


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

literate (nonprogrammer) to design custom data-analysis and graph-ing schemes. As most of the graphing programs are Windows-based, merging of graphs into word processors or other drawingprograms is easy. Finally, high-quality affordable laser printershave made plotters obsolete. Most of these printers, such as theHP Laserjet 4, will readily accept 80g glossy paper, providing aquality printout that is indistinguishable from glossy photographs.

5. Limitations, Pitfalls, and Errors5.1. Series Resistance and its Consequences5.1. Series Resistance and its Consequences5.1. Series Resistance and its Consequences5.1. Series Resistance and its Consequences5.1. Series Resistance and its Consequences

As mentioned previously, the major limitations of the whole-cellpatch-clamp recording technique lies in its design as a continuoussingle-electrode voltage clamp. The continuous use of one elec-trode for current passage as well as voltage sensor makes true mem-brane voltage determination impossible. The technique assumes thatpipet voltage equals membrane voltage, as voltage commands areimposed on the pipet, not on the cell. However, recording pipetand access resistance (due to potential clogging at the electrodetip) are in series with the current recording and the voltage com-mand. This series resistor in conjunction with the membrane resis-tance acts as voltage divider to all imposed voltages. Consequently,only in cases where the membrane resistance greatly exceeds theseries resistance is an adequate voltage clamp (point clamp) assured.Under experimental circumstances, series resistance accounts forat least 5, more typically 10–15 MOhm. In order to keep the volt-age error below 1%, membrane resistance has to be two orderslarger than the series resistance, in our example ~1 GOhm. This ishardly the case, and certainly does not hold true during activa-tion of ionic currents! In order to assure best possible recordingconditions, the following steps are absolutely necessary:

1. Electrode resistance has to be minimized as much as possi-ble, depending on the size of the cells to be studied. In our experi-ence, and dependent upon the solutions used, cells from 8–40 µmin size can be successfully patched with electrodes 1.5–3 MOhmrange. However, once the whole-cell recording configuration hasbeen achieved, it is important to frequently check for adequatecompensation. If all compensation mechanisms are turned off, onemay see a dramatic decrease in the magnitude of initial capaci-tance transients observed. The most likely explanation for such achange is the clogging of the electrode tip with cell membrane,


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

which directly interferes with clear access to the cell’s interior.This phenomenon of membrane “healing” around the electrodetip can be prevented by buffering [Ca2+]i using high concentra-tions of EGTA or BAPTA. In an acute situation, slight positive ornegative pressure can reverse electrode clogging. In our experi-ence, it is possible to achieve access resistances of 5 MOhm priorto series resistance compensation.

2. Series resistance (RS) needs to be compensated for. Most patch-clamp amplifiers provide a positive feedback series resistancecompensation circuit, in which a signal proportional to the mea-sured current is added to the command potential. RS is determinedby adjusting Rs and Cp controls to square out a command volt-age. Subsequently, Rs compensation is activated. Although, theo-retically, near 100% compensation is possible, real experimentshardly allow gain setting of more than 50–80%. Rs compensationscales the command input to account for the voltage loss acrossRs. Rs compensation is very sensitive to changes in the fast (pipet)capacitance compensation. It is essential to adjust this compensa-tion properly to achieve maximum % compensation settings. Notethat continuous bath perfusion may result in some oscillationsof the bath fluid level. As a consequence, this would also changethe effective capacitance of the pipet and would make the fastcapacitance compensation unstable and thereby Rs compensationprone to ringing. Effective compensation under those circum-stances thus requires stable bath perfusion level. The problem canbe reduced by the use of heavily sylgarded electrodes which havea much reduced capacitance.

In an ideal case, with Rs of 10 MOhm, a voltage step of 100 mVresults in a current flow of 1 nA, and an apparent input resistanceRcell+Rs of 100 MOhm. A 1 nA current flow across Rs generates a10 mV voltage drop across Rs and thus a 10% error. Using Rs com-pensation assuming an 80% compensation the error is reduced to2 mV or 2% of command voltage, a tolerable error. However, sup-pose activation of voltage-dependent channels gives rise to a 5 nAcurrent. During the peak of this response the input resistance fallsto 20 MOhm and, in the uncompensated situation, the membraneexperiences only a 50 mV voltage drop! Even with 80% compen-sation, a 10 mV (10%) error still remains.

3. Cells with low input resistances are almost impossible torecord from. Should the membrane impedance be <100 MOhm itis advisable to increase it by inclusion of ion channel blockers to


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

block conductances that are not of immediate interest. Thus, K+

channel blockers could be included and Cl− replaced by acetate toallow resolution of small Na+ currents.

Uncompensated Rs will have two additional detrimental effectson current recordings. It will affect the time response to a voltagechange, and it will result in increased signal noise. The transmem-brane voltage resulting from a step change in voltage is describedby Vm = Vc [1-exp (-τ/(Rs* Cm)] with an effective time-constantof τ = RC (if Rs << Rm). Assuming an uncompensated Rs of 10MOHm and Cm of 100 pF charging of the membrane will be slowedwith an effective time-constant (τ) of 1 msec. Unfortunately, Rswill also filter any current flow recorded with this arrangement.In the absence of compensation, this results in a single-pole RCfilter with a -3dB frequency described by F = 1/(2π*Rs*Cm) result-ing in a cut-off frequency (F) of 159 Hz for previous example.

5.2. 5.2. 5.2. 5.2. 5.2. VVVVVoltage-Clamp Errorsoltage-Clamp Errorsoltage-Clamp Errorsoltage-Clamp Errorsoltage-Clamp Errors

Voltage clamp is prone to error, as it makes numerous assump-tion that may not be valid under the given experimental conditions.It assumes that the cell is isopotential and that the voltage mea-sured at any one point across the membrane is the true membranevoltage. Analogously, current injection, which imposes change tothe membrane voltage is thought to be uniformly realized in allparts of the membrane, including distant processes. This, however,is not the case. As a result, two sources of error exist namely voltage(point) and space clamp errors. As illustrated below, whole-cellpatch-clamp recordings are even more susceptible to error thanclassical two electrode recordings, and as such, the experimenterneeds to be constantly aware of possible sources of this error.

5.2.1. Space ClampSpace-clamp limitations are intrinsic to voltage-clamp and do

not differ in their principals between different voltage-clamp tech-niques. Current injected into the cell to maintain or establish achange in membrane voltage will spread radially from the injec-tion site, and decay across distance with the space (length) con-stant λ. In small-diameter spherical cells, this is a minimal concern.However, in a process-bearing cell, the current signal may havedistorted by the time it reaches distant processes hundreds of µmaway from the injection site. In the best of circumstances, the sig-nal will be attenuated and the voltage-changes imposed will be


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

smaller. In the worst case, distant membranes may not experi-ence any voltage change at all.

Double electrode voltage-clamp methods are some-what advan-tageous in that they allow to detect space-clamp problems morereadily. In this case, the voltage electrode can be inserted at a dis-tance from the current electrode, and closer to the site of interest.Whole-cell recordings clamp the voltage of the electrode tip, andthus provide no means to establish any true recording at a sitedistant from the electrode. Some investigators have chosen toinsert a second patch-clamp or sharp microelectrode into cells tomonitor the true voltage changes observed. Space-clamp prob-lems can only be reduced by recording from small cells with simplemorphology, ideally spherical cells. Nonspherical cell can some-times be “rounded-up” by exposure to serum or treatment withdBcAMP. However, often times the most interesting cells bearextensive arborized processes. A second way to assure that injectedcurrent can travel further is to increase the cells impedance. Thusit is often possible to block parts of the cells conductance (e.g., K+

conductance) pharmacologically to effectively increase the lengthconstant (λ,) of the cell.

5.2.2. Voltage (Point) ClampThe whole-cell recording technique effectively voltage-clamps the

electrode tip, and, by assuming that its resistance is small relativeto 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 smalland well-compensated for. Unlike space clamp errors, point-clamperrors can be readily detected and often eliminated. After cancel-ing series resistance error (at least partially, see above) the expe-rimenter can calculate the voltage error, which now is a linearfunction of current flow. Point-clamp errors produce primarily twodistortions to the recorded signal: 1) Slowing of current kineticsand 2) apparent attenuation of true current amplitudes due touncompensated voltage error. Errors due to slow activation canbe readily identified from the current traces.

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

Figure 11 demonstrates Na+ current recordings from 3 different cellsof a neuronal cell line (BlO4). Current recordings were obtainedunder appropriate and poor voltage control (Fig. 11 A–C), and


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

peak current values were plotted as a function of membrane poten-tial and superimposed in Fig. 11D. Under the imposed ionic gra-dients of the recordings, the theoretical equilibrium potential forNa+ was ~40 mV. The current traces in Fig. 11A yield an I-V plotthat reverses close to ENa indicative of proper voltage-control,whereas the traces in Fig. 11B indicate a 20 mV more positive rever-sal potential, and in Fig. 11C the currents do not reverse at all. Thusthe I-V plot directly indicates the severity of the voltage errors inFig. 11B and C. While the recording in Fig. 11B still yielded thesame peak in the I-V relationship (at −20 mV) as the recording inFig. 11A, the voltage step to −40 mV (arrows in Fig. 11A–C) wassignificantly delayed. In Fig. 11C, voltage-control is lost at the

Fig. 11. I-V curves used to judge quality of point-clamp. (A–C) repre-sent families of current recordings from three different B104 cells (neu-ronal cell line) using the same step protocol indicated in inset. Peak Na+currents at each potential were plotted in (D). Arrows in (A–C) point tocurrent 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 delayin current activation at threshold (arrows) and by a shift in I-V relation-ship (D) overestimating the true current reversal potential. ENa was 40 mV.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

threshold of current activation (−50 mV), and currents do not acti-vate in a graded fashion but rather “escape” to reach near-peakamplitude with a severe delay. Similar I-V plots of transmitter-induced currents for which the major ion carrying the response isknown (such as GABA) allow to utilize the reversal potential asan 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 awide voltage range. However, under all these circumstances poten-tial space-clamp errors remain, and most likely currents activatedin remote processes are not recorded at all.

6. Special Applications

A number of specialized applications utilize the whole-cell patchclamp recording technique. Three of these, namely perforated patch,patch-slice, and single-cell polymerase chain reaction (PCR) aredescribed in detailed in other chapters of this book. One applicationworth mentioning is the use of patch-electrodes for dye loading.

As whole-cell recordings allow low resistance access to the cell’scytoplasm, it provides a convenient means to load cells with bio-logical markers such as fluorescence indicators, horseradish per-oxidase, or biotin. Loading of cells can be used to address a numberof questions:

1. As fluorescence markers may diffuse through gap-junctions,dye-diffusion can be used to search for the existence of gap-junctions between cells. While in principle, numerous lowmolecular-weight compounds can be and have been utilizedfor this purpose, Lucifer Yellow (LY) has been a long-timefavorite as it is readily retained in cells even after fixation andis among the fluorescence compounds with the highest quan-tum yields. We have used LY as a pipet solution constituent innumerous recordings and have not detected any interferencewith our ability to resolve ionic currents. The use of LY to studygap-junction coupling is described in detail elsewhere (Ran-som and Sontheimer, 1992).

2. LY may also be used as a cell marker allowing the localizationof a cell from which recordings have been obtained. We fre-quently fill cells with 0.2% LY (potassium or Li+ salt) duringrecordings and after fixation utilize cell-specific antibodies toantigenically identify the cell. LY is believed to form covalent


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

bonds with elements of the cells cytoskeleton, and is retainedin cells even after formaldehyde fixation and membrane per-meabilization. In some preparations, such as brain slices, LYfills are the only way to resolve the complex arborization ofcell processes.

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

7. Conclusions

What started out as a spin-off from single-channel patch record-ings has resulted in a technique more useful than its inventorshad anticipated (Sigworth, 1986). Its ease-of-use makes whole-cellrecordings the most widely used intracellular recording techniqueto date that in many instances has replaced sharp microelectroderecordings. However, whole-cell patch-clamp recordings are notor-iously prone to error and may not always generate accurate record-ings. It is thus important to understand the limitations of the tech-nique. If proper care is taken whole-cell patch-clamp allows thestudy of almost any small cell of interest, and has opened the fieldof single-cell electrophysiology. Modifications (perforated patch)and applications of the technique to more intact preparations (patch-slice) has provided invaluable insight into nervous system function.

Acknowledgments

During preparation of this manuscript HS was supported bygrants RO1-NS31234 and RO1-NS36692 from the National Insti-tutes of Health.

ReferencesAlmers, W. and Neher, E. (1985) The Ca signal from fura-2 loaded mast cells

depends strongly on the method of dye-loading. FEBS Lett. 192, 13–18.Bezanilla, F. and Armstrong, C. M. (1977) Inactivation of the sodium channel:

I. Sodium current experiments. J. Gen. Physiol. 70, 549–566.Connor, J. A. and Stevens, C. F. (1971) Voltage clamp studies of a transient out-

ward membrane current in gastropod neural somata. J. Physiol. 213, 21–30.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

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

Hodgkin, A. L. and Huxley, A. F. (1952) A quantitative description of membranecurrent and its application to conduction and excitation in nerve. J. Physiol.(London) 117, 500–544.

Hodgkin, A. L., Huxley, A. F., and Katz, B. (1952) Measurement of current-volt-age relations in the membrane of the giant axon of loligo. J. Physiol. (London)116, 424–448.

Neher, E. (1982) Unit conductance studies in biological membranes. Techn. Cell.Physiol. P 121, 1–16.

Pusch, M. and Neher, E. (1988) Rates of diffusional exchange between smallcells and a measuring patch pipette. Pflügers Arch. 411, 204–211.

Ransom, B. R. and Sontheimer, H. (1992) Cell-cell coupling demonstrated byintracellular injection of the fluorescent dye Lucifer yellow, in Electrophysi-ological Methods for In Vitro Studies in Vertebrate Neurobiology (Kettemnann,H. and Grantyn, R. eds.), Wiley-Liss, New York, pp. 336–342.

Sigworth, F. J. (1983) Electronic design of the patch clamp, in Single-ChannelRecording (Sakmann, B. and Neher, E. eds.), Plenum Press, New York, pp.3–35.

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

Recommended ReadingsFor a more in depth description of the patch-clamp technique, the following

books are highly recommended (in chronological order):

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

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

Jones, S. (1990) Whole-cell and microelectrode voltage clamp, in Neuromethods,vol. 14, (Boulton, A. A., Barker, G. B., and Vanderwolf, C. H., eds.), HumanaPress, Totowa, NJ.

Hille, B. (2001) Ionic Channels of Excitable Membranes. Sinauer, Sunderland, MA.Rudy, B. and Iverson, L. E. (eds.) (1992) Ion Channels, Methods in Enzymology.

Academic Press, San Diego, CA.Sherman-Gold, R. (ed.) (1993) The Axon Guide for Electrophysiology & Bio-

physics Laboratory Techniques. Axon Instruments, Inc.


0123456789

10111213141516171819202122232425262728293031323334353637383940414243444546474849

http://www.springer.com/978-1-58829-013-7

Wholde Patch Clamps Recordings

Documents

Transcript of Wholde Patch Clamps Recordings