Tr

JN

a

ARRA

KAPNCCSMC

1

iosedOsced

p

C

0d

Journal of Neuroscience Methods 183 (2009) 158–164

Contents lists available at ScienceDirect

Journal of Neuroscience Methods

journa l homepage: www.e lsev ier .com/ locate / jneumeth

he impact of synaptic conductance on action potential waveform: Evokingealistic action potentials with a simulated synaptic conductance

amie Johnston 1,3, Michael Postlethwaite 2,3, Ian D. Forsythe ∗

eurotoxicity at the Synaptic Interface, MRC Toxicology Unit, Hodgkin Building, University of Leicester, Leicester LE1 9HN, UK

r t i c l e i n f o

rticle history:eceived 9 March 2009eceived in revised form 17 June 2009ccepted 18 June 2009

eywords:ction potentialhysiologyeuronal excitabilityurrent injectionurrent clamp

a b s t r a c t

Most current clamp studies trigger action potentials (APs) by step current injection through the recordingelectrode and assume that the resulting APs are essentially identical to those triggered by orthodromicsynaptic inputs. However this assumption is not always valid, particularly when the synaptic conductanceis of large magnitude and of close proximity to the axon initial segment. We addressed this question ofsimilarity using the Calyx of Held/MNTB synapse; we compared APs evoked by long duration step cur-rent injections, short step current injections and orthodromic synaptic stimuli. Neither injected currentprotocol evoked APs that matched the evoked orthodromic AP waveform, showing differences in APheight, half-width and after-hyperpolarization. We postulated that this ‘error’ could arise from changesin the instantaneous conductance during the combined synaptic and AP waveforms, since the drivingforces for the respective ionic currents are integrating and continually evolving over this time-course.

ynapseNTB

alyx of Held

We demonstrate that a simple Ohm’s law manipulation of the EPSC waveform, which accounts for theevolving driving force on the synaptic conductance during the AP, produces waveforms that closely mimicthose generated by physiological synaptic stimulation. This stimulation paradigm allows supra-thresholdphysiological stimulation (single stimuli or trains) without the variability caused by quantal fluctuationin transmitter release, and can be implemented without a specialised dynamic clamp system. Combinedwith pharmacological tools this method provides a reliable means to assess the physiological roles of

s with

postsynaptic ion channel

. Introduction

The intrinsic properties of a neuron are crucial in determin-ng how it integrates, computes and transmits information tother cells. Although influenced by morphology and passive cabletructure (Mainen and Sejnowski, 1996), the suite of ion channelsxpressed and their location on the cell membrane are a majoreterminant of a neuron’s processing abilities (Dodson et al., 2003;gawa and Rasband, 2008). There are two common methods of

timulation used to approximate physiological inputs under currentlamp: injection of step depolarizing current through the recording

lectrode or electrical stimulation of presynaptic axons (ortho-romic synaptic stimulation).

For cells that require excitatory synaptic input to generate actionotentials (APs), orthodromic synaptic activation is an ideal physio-

∗ Corresponding author. Tel.: +44 0116 252 5580; fax: +44 0116 252 5616.E-mail address: idf@le.ac.uk (I.D. Forsythe).

1 Present address: Department of Biology, University of Victoria, BC, Canada.2 Present address: GU Biology, Pfizer Global R & D, Ramsgate Road, Sandwich, Kent

T13 9NJ, UK.3 Both these authors contributed equally to this work.

165-0270/$ – see front matter © 2009 Published by Elsevier B.V.oi:10.1016/j.jneumeth.2009.06.025

out confounding affects from the presynaptic input.© 2009 Published by Elsevier B.V.

logical stimulus. However, some disadvantages arise when studyingthe physiology of the postsynaptic cell: first, pharmacological toolsmay act at pre- and postsynaptic sites (e.g. tetraethylammoniumTEA, blocks Kv3 channels and broadens the AP at both sites, caus-ing increased transmitter release, Ishikawa et al., 2003). Second,there are many presynaptic changes which may confound interpre-tation (e.g. quantal fluctuation, short-term depression, facilitation,presynaptic modulation) particularly during repetitive stimulation.

Direct somatic current injection avoids these problems, but theresultant APs are not accompanied by the synaptic conductanceand associated changes in resistance. A more physiological stimulusshould take account of the ionic driving forces caused by the voltagechanges during the AP.

This study compares different methods of AP generation andprovides some simple insights into designing a current injectionparadigm that more closely mimics a supra-threshold orthodromicexcitatory input. We have used the calyx of Held preparation, sincethis is a secure synapse with a large safety factor. Our approach

has been to make this method broadly applicable, avoiding useof dynamic clamp methods by using a simple computation of theAP and synaptic current. We show that the synaptic conductancehas a profound influence on the AP waveform, and assert thatthis should be considered when studying physiological roles of

oscience Methods 183 (2009) 158–164 159

coom

2

2

(Ci1(1gslr7fict21br

2

espthW

gca4wrwto

awcp7wPtOflwd(bnv

J. Johnston et al. / Journal of Neur

hannels. This method can be simply applied to evoke more physi-logical AP waveforms without the expense or detailed knowledgef the existing conductances required to implement dynamic clampethods.

. Materials and methods

.1. In vitro preparation

Brain slices were prepared as described previouslyPostlethwaite et al., 2007; Johnston et al., 2008a). Briefly,BA/Ca mice or Lister Hooded rats aged P10–P19 were decapitated

n accordance with the UK animals (Scientific Procedures) Act986 and the brain was removed in a slush of iced artificial CSFaCSF) of composition (in mM) 250 sucrose, 2.5 KCl, 10 glucose,.25 NaH2PO4, 0.5 ascorbic acid, 26 NaHCO3, 4 MgCl2, 0.1 CaCl2,assed with 95% O2/5% CO2 (pH 7.4). The brain was placed ventralide up and the brainstem severed with a vertical cut at theevel of the pons angled at ∼20◦ toward the cerebellum. Theostral surface was then glued on the stage of an Integraslice550 MM (Campden Instruments, Loughborough, UK). Four tove transverse slices 150–250 �m thick were cut from the areaontaining the MNTB. Slices were then transferred to aCSF con-aining (in mM): 125 NaCl, 2.5 KCl, 10 glucose, 1.25 NaH2PO4,

Na-pyruvate, 3 myo-inositol, 0.5 ascorbic acid, 26 NaHCO3,MgCl2, 2 CaCl2, gassed with 95% O2/5% CO2 (pH 7.4), incu-

ated at 37 ◦C for 1 h and then stored at room temperature untilequired.

.2. Electrophysiology

For recording, one slice was transferred to a Peltier controllednvironmental chamber on the stage of an upright E600FN micro-cope with DIC optics (Nikon, Tokyo, Japan) and continuouslyerfused with aCSF (∼1 ml min−1) at the experimental tempera-ure of 35–37 ◦C maintained by feedback control of a Peltier elementeat exchange (University of Leicester, Mechanical and Electronicorkshops).Pipettes (2–3 M�) were pulled from thick walled borosilicate

lass (GC150F-7.5, Harvard Apparatus) and filled with a solutionontaining (in mM): 97.5 K-Gluconate, 32.5 KCl, 10 HEPES, 5 EGTAnd 1 MgCl2 (pH 7.2 with KOH) and had series resistances of–12 M�. All series resistances were compensated by at least 75%ith 10 �s lag and cells were excluded from analysis if the series

esistance changed by more than 2 M�. Current clamped neuronesere held at ∼−70 mV, and the bridge balance circuitry was used

o account for the series resistance. The input resistances measuredver the range −70 to −50 mV was 106 ± 14 M� (n = 18).

A bipolar electrode for stimulating trapezoid axons was placedt the midline and a DS2A isolated stimulator (Digitimer, Wel-yn garden city, UK) delivered 0.2 ms pulses of 4–10 V. Synaptic

onnections were detected using an imaging technique describedreviously (Billups et al., 2002). Briefly, slices were bathed in�M Fura2-AM (Molecular Probes, Invitrogen) for ∼4 min afterhich excess Fura2-AM was washed off. Cells were viewed using a

hotometrics CoolSNAP-fx camera after a single 100 ms exposureo 380 nm (provided by a xenon arc lamp controlled by a Cairnptoscan monochromater; Cairn Instruments, Faversham, UK). Theuorescent image was displayed using Metafluor imaging soft-

are (version 4.01, Molecular Devices) and regions of interest wererawn around visualized neurons. A stimulus train was delivered200 Hz, 200 ms) and synaptically connected cells were identifiedy a decrease in the 380 nm signal brought about by the postsy-aptic rise in calcium concentration. These cells were then locatedisually using DIC optics for recording.

Fig. 1. Measurement of synaptically evoked action potential. An example of anorthodromically stimulated action potential evoked from depolarization from acalyx of Held. The four measured parameters are indicated: action potential (AP)height, AP half-width (width at half-amplitude), ADP and AHP.

2.3. Data acquisition and analysis

In order to avoid artifacts caused by current clamp through apatch-clamp head-stage (Magistretti et al., 1996) we used eitheran Optopatch amplifier (Cairn) or Multiclamp 700B with a Digi-data 1322A interface (Molecular Devices). Command signals weredigitised at 50 kHz and data were acquired using pClamp 9.2 andwere filtered at 5 kHz and digitised at 50 kHz. Junction potentials of−7 mV have been added to stated voltages. The measured parame-ters for an orthdromic AP are shown in Fig. 1. We have stated actualvoltages rather than the relative height since it is the actual voltagethat influences other voltage-gated channels (see Johnston et al.,2008b). The AHP does not undershoot the baseline with an ortho-dromic AP; rather the Kv3 mediated rapid AHP rides on top of theADP. This definition of the AHP was applicable to both the ortho-dromic and SSG waveforms. All APs were evoked from a membranepotential of −67.6 ± 0.3 mV (range −65.9 to −69.4, n = 11).

2.4. Calculation of the simulated synaptic conductance (SSG)

This is a simple Ohm’s law transformation of a synaptic current(EPSC) to conductance: GEPSC = IEPSC(Erev·EPSC − Vm). The EPSC wasmeasured under voltage-clamp with the instantaneous voltage dur-ing the action potential waveform measured in the same cell undercurrent clamp. The synaptic conductance waveform (i.e. the EPSCcorrected for the driving force, as shown in Fig. 2) is multipliedby the AP waveform voltage (using either an excel spreadsheet,Microsoft or Igor pro, wavemetrics). The resultant waveform cor-responds to the current flowing through the ligand-gated GluR(i.e. synaptic conductance) during the changes in voltage drivingforce which are continually evolving over the time-course of theAP, therefore we shall refer to it as the simulated synaptic conduc-tance (SSG). The major difference from a simple current injectionis that at the peak of the AP, the voltage is close to the reversalpotential for glutamate receptors (around 0 mV) hence there is no

driving force (and no current flow) then as the AP repolarizes, thedriving force increases and the synaptic channels can again exertan influence.

Results are stated as means ± SEM. No significant differencewas found for AP parameters between mouse and rat (unpaired t-

1 oscien

tAcuS

3

safibae(taSraoba(i

FdSi(s

60 J. Johnston et al. / Journal of Neur

est), so the data have been pooled. A repeated measures, one-wayNOVA with Dunnett’s post hoc test was used to assess whether theurrent injection paradigms were different from the synaptic stim-lation. Unpaired t-tests were used to assess differences betweenSG AP parameters and synaptic AP parameters.

. Results

Comparison of postsynaptic action potentials generated by threetandard types of depolarizing stimulus clearly illustrates that theyre not equivalent (Fig. 2) with each producing a different wave-orm AP in the same neuron. A long current step is often employedn excitable cells to study intrinsic properties and evoke APs, while arief pulse has the advantage of generating a short latency responsend is commonly used to trigger trains of APs. We compared APsvoked by long duration current steps, short duration current stepspulse) and also APs evoked by orthodromic synaptic stimula-ion in the same MNTB neurones under current clamp (using anmplifier with an appropriate headstage for current clamp, seeection 2). The AP generated by a just supra-threshold step cur-ent injection lasting 200 ms shows a delay to AP triggering, a largefter-hyperpolarization and a maintained depolarization through-

ut the current injection (Fig. 2A, top). A larger magnitude (2 nA)ut a brief supra-threshold current pulse lasting 0.2 ms also evokedsingle ‘instantaneous’ AP without the maintained depolarization

Fig. 2A, middle). Comparison of these two APs evoked by currentnjection with the AP evoked by orthodromic synaptic stimulation

ig. 2. Action potential waveforms are influenced by the method of stimulation when coifferent methods of triggering action potentials in the same neurone. Top: Step currenynaptic stimulation by 200 �s 8 V electrical pulse to the presynaptic axons, this evokes a sndicates the start of the current pulse or synaptic stimulation, scale bars apply to all thrAHP), AP height and AP half-width (width at 1/2 amplitude) for each of the three respeame 11 neurones). Stars show significance at <0.05 and <0.001, respectively.

ce Methods 183 (2009) 158–164

(Fig. 2A, bottom) shows distinct differences: most notably there isa slowly decaying after-depolarization (ADP) on the orthodromicresponse which is absent from responses triggered by current injec-tion.

The mean after-hyperpolarization (AHP), AP height, and AP half-width are compared between the three conditions and plottedin the bar graphs in Fig. 2B. The absolute AHP following synap-tic stimulation was −61.6 ± 1.5 mV, while the step paradigm AHPwas −66.7 ± 1.7 mV (n = 11, synaptic vs. step, p < 0.05), the pulseparadigm AHP also differed, being −73.3 ± 0.9 mV (n = 11, synap-tic vs. pulse, p < 0.001). The grey arrow in Fig. 2A show where thesecond stimulus in a 200 Hz train would occur if this method wereused to evoke trains of APs. The pulse paradigm is still hyperpolar-ized from the resting potential whereas the synaptically evokedparadigm is depolarized from rest. This difference is importantwhen studying AP trains since the voltage around threshold andbetween spikes is crucial for sodium channel recovery from inac-tivation and hence AP output (Jung et al., 1997; Johnston et al.,2008b). A significant difference in action potential peak potentialwas also observed between the pulse (+4.2 ± 4.6 mV, n = 11, abso-lute potential) and the synaptic paradigm (−11.0 ± 2.3 mV, n = 11,p < 0.05), whereas the step paradigm showed no difference to the

synaptic paradigm (Fig. 2B middle). The half-width of the pulse-evoked AP was faster than that evoked by synaptic paradigms(0.31 ± 0.03 and 0.42 ± 0.03 ms, respectively; n = 11, p < 0.001) andthere was no significant difference between the synaptic and stepparadigm (Fig. 2B, bottom).

mpared in the same cell. (A) Example records from a single neuron showing threet injection, 200 pA, 200 ms. Middle: Pulse current injection, 2 nA, 0.2 ms. Bottom:upra-threshold synaptic response and an orthodromic action potential. Black arrowee panels. (B) Comparison of the absolute voltages for the after-hyperpolarizationctive stimulation methods (n = 11, all three stimulation modes were applied to the

oscien

tsmclucbor

iobpedtbnp

FDiessc

J. Johnston et al. / Journal of Neur

Since all three paradigms were recorded from each neuron,hese differences cannot be attributed to differences in the intrin-ic properties of the neurons, rather they must solely reflect theethod used to evoke the action potential. This allows us to

onclude that the depolarization of the membrane potential fol-owing the AHP in the synaptically evoked action potential isnlikely to be a resurgent Na+ current but reflects the decayingonductance of the EPSC. Similarly, the slow decay of the mem-rane potential back to rest must be due to the slow componentf the synaptic response, rather than intrinsic voltage-gated cur-ents.

This result clearly shows that the waveform of the evoked APs influenced by the method used to trigger that AP, so generationf a more physiological stimulus must take account of interactionsetween synaptic conductance and the AP. We thus sought to com-are the synaptic conductance time-course to the AP waveformvoked from the same synaptic stimulus in the same neuron. Thisata required recordings under both current and voltage clamp in

he same cell, so neither tetrodotoxin nor QX314 could be used tolock sodium channels. In order to minimise the risk of contami-ating the EPSC waveform by voltage-gated Na+ currents, holdingotentials were reduced from −77 to −67 or −57 mV while record-

ig. 3. The synaptic conductance and the action potential waveform. (A) Under voltageata is shown from two cells (black traces and grey traces; dV/dt traces above and EPSC

ndicated by the black arrow. These were excluded from analysis. (B) EPSC (lower trace) vvoked action potential (upper trace) recorded in the same cell under current clamp (dahown scaled to the same amplitude as the AP and overlaid (black). Note that the peak of thows expanded view of the conductance contributing to an after-depolarizing potentialonductance at latencies corresponding to the peak conductance, AP peak and at the peak

ce Methods 183 (2009) 158–164 161

ing EPSCs. Contamination was detected and displayed as a notch inthe waveform after differentiation of the EPSC (Fig. 3A, arrow lowertraces). An example current clamp and uncontaminated voltage-clamp trace is shown in Fig. 3B.

At least five EPSCs were averaged and converted to conduc-tance by normalizing for the driving force (using Erev of 0 mV).The conductance waveforms obtained from different voltages wereaveraged to give an accurate estimate of the synaptic conductancewhich was then plotted against the action potential (Fig. 3C, greytrace).

The peak synaptic conductance was 135.0 ± 13.6 nS whichdecayed with a double exponential time-course (Taufast0.27 ± 0.02 ms, Afast 93.3 ± 0.9%, Tauslow 6.15 ± 0.85 ms, n = 13)similar to previous reports for the EPSC (Taschenberger and vonGersdorff, 2000; Postlethwaite et al., 2007; Koike-Tani et al., 2005).At the peak of the action potential the synaptic conductance wasstill 132.7 ± 14.2 nS (Fig. 3D, n = 13). Since the synaptic conductance(mediated by AMPA and NMDA receptors; Barnes-Davis and

Forsythe, 1995) reverses around 0 mV, it will act to shunt currentswith reversal potentials differing from 0 mV, i.e. the Na+ current.Also, the slow decay of the synaptic conductance leaves 6.8 ± 1.1 nSat the peak of the ADP (Fig. 3D).

clamp EPSCs were checked for contamination by Na+ currents by differentiation.below) which illustrate contamination in some records, as in the lower grey tracesoltage-clamped at a holding potential of −67 mV and the supra-threshold synapticshed line indicates zero current). (C) The calculated synaptic conductance (grey) ishe AP occurs after the peak synaptic conductance and during the decay phase. Inset(ADP), due to decay of the EPSC. (D) Mean data showing measurements of the EPSC

of the ADP (n = 13).

1 oscien

uoafTt

FwAtc(ia

62 J. Johnston et al. / Journal of Neur

How can this information improve the physiological stimulusnder current clamp conditions? Fig. 3C highlights an obvi-us but crucial issue: the synaptic and voltage-gated currents

re of similar overlapping time-course, hence the driving forceor the synaptic conductance changes with the AP waveform.hus a more accurate approximation to a physiological supra-hreshold orthodromic stimulus is the product of the synaptic

ig. 4. Constructing and testing the simulated synaptic conductance (SSG). (A) The actionhich gives the simulated synaptic conductance (SSG) displayed with the current conventiP. Note that the current injection has two phases in the SSG; the transient minimum (*) i

he reversal potential for the glutamate receptor conductance (Erev·EPSC) at 0 mV. If the Aurrent flow would briefly reverse. (B) Average data comparing action potential parameteC) Example of a SSG-evoked action potential (black, upper trace) and the same cell aftencreased amplitude of the first AP) in response to the same SSG current injection wavefornd after DTx-I. (D) Concatenated SSG stimuli can be used to build a train of stimuli whic

ce Methods 183 (2009) 158–164

conductance waveform and the AP voltage waveform accordingto Ohm’s Law. This is illustrated in Fig. 4A, the voltage waveformis multiplied by the conductance waveform which results in the

synaptic current (similar to the EPSC but taking account of thechanging voltage during the action potential). Note that as theaction potential peaks, the current drops off because the volt-age approaches the EPSC reversal potential (at around 0 mV), then

potential waveform (voltage) is multiplied by the calculated EPSG (conductance)on for current clamp. This current waveform is injected into the neuron to trigger ans caused by the EPSC driving force declining to near zero as the AP peak approachesP were to overshoot then the voltage would become positive to Erev·EPSC and thers evoked by the SSG (14 neurones) and orthodromic synaptic stimulation (n = 13).r application of 10 nM DTx-I (grey, upper trace) showing increased AP firing (andm (black, lower trace). Inset: The same cell in response to the step paradigm before

h does not decline in magnitude with vesicle depletion and short-term depression.

oscien

ii

t(ScooF+ed

cindsenitctbnlpmhTdSnk2

4

ttfsiiinhitaai

4

AetiI1q

J. Johnston et al. / Journal of Neur

ncreases as the action potential repolarizes and the driving forcencreases.

We then used this SSG waveform to evoke APs (e.g. Fig. 4C), withhe expectation that it would evoke physiologically realistic APsseveral software packages such as pClamp, Patchmaster and CEDignal software provide functions to import waveforms to use asommand signals). We found no significant difference in AHP, ADPr the half-width when we compared the SSG-evoked AP with therthodromic synaptically evoked AP (n of 14 and 13, respectively,ig. 4B). However the AP height was significantly more positive at8.68 ± 2.9 mV for the SSG vs. −8.15 ± 1.6 mV for the synapticallyvoked AP (p < 0.0001) which is consistent with a shunting actionuring the orthodromic synaptic stimulation.

Use of this SSG to trigger APs will improve analysis of normal ionhannel function in the postsynaptic neuron. It will avoid confound-ng effects where compounds act at multiple sites, such as the presy-aptic axon, terminal and/or postsynaptic neuron. For instance den-rotoxin has been used to show the role of Kv1 channels in the post-ynaptic cell (Dodson et al., 2002; Brew and Forsythe, 1995), how-ver the interpretation is confounded by the presence of Kv1 chan-els at the hemi-node of the presynaptic terminal which can result

n multiple action potentials invading the terminal causing aberrantransmitter release (Dodson et al., 2003). Stimulation using the SSGombined with application of dendrotoxin-I (DTx-I, 10 nM) showshat under control conditions Kv1 channels prevent multiple APseing triggered during the ADP from a single stimulus (Fig. 4C grey,= 4). The effect of DTx-I on AP firing during a sustained step depo-

arization is shown in Fig. 4C, inset. DTx-I also increased the actionotential height. Since the AP is generated in the axon initial seg-ent, the effect of Kv1 channels on action potential height would

ave been masked by the synaptic conductance at the cell body.he effect of DTx-I on AP height is consistent with Kv1 activationuring the upstroke of the AP (Klug and Trussell, 2006). Trains ofSGs (Fig. 4D) can be constructed to study the effects of ion chan-els during repetitive stimuli by concatenating single SSGs, therebyeeping the waveform constant, as used previously (Johnston et al.,008b) for studies of high frequency firing in MNTB neurons.

. Discussion

This paper addresses the simple yet fundamental observationhat evoked AP waveforms differ depending on the nature of theriggering stimulus. Step current injections evoked APs with wave-orms which were distinct from those elicited by orthodromicynaptic stimulation. By definition, the orthodromic synaptic inputs the most physiological method of AP generation, but currentnjection is often assumed to be equivalent; we show here that its not. There are several situations where synaptic stimulation mayot be possible, where pharmacological agents are in use whichave pre- and postsynaptic actions, or where it is necessary to elim-

nate underlying quantal fluctuation of the EPSP. We demonstratehat the use of a simulated synaptic conductance can provide a goodpproximation to a supra-threshold orthodromic synaptic inputnd evokes APs which are of similar waveform to the physiologicalnput.

.1. The need for somatic injection

To elucidate the physiology and pharmacology of postsynapticP generation, we must minimise changes at presynaptic sites. Forxample, application of TEA will not only broaden the postsynap-

ic AP, it also broadens the presynaptic AP, resulting in greater Ca2+

nflux and potentiated transmitter release (Ishikawa et al., 2003).ndeed the contribution of other ion channels (e.g. Nav, Wang et al.,998; Song et al., 2005; Leao et al., 2006 or Kv3) to the high fre-uency firing of MNTB neurons are usually assessed by delivering

ce Methods 183 (2009) 158–164 163

trains of current pulses. High frequency orthodromic synaptic stim-ulation causes summation of the ADP and hence a depolarizationplateau upon which APs can ride (Taschenberger and von Gersdorff,2000; Johnston et al., 2008b). During trains of pulse-evoked APs, thesubsequent APs are fired from a more hyperpolarized baseline pro-ducing significant consequences for AP waveform which have oftenbeen overlooked.

4.2. The impact of synaptic conductance on action potentialwaveform

In the MNTB, glutamate receptor currents reverse around 0 mVand are restricted to the postsynaptic cell body, therefore the EPSCgenerated by a synaptic stimulus will act to shunt the AP recordedat the cell body. This is a likely explanation for the smaller AP over-shoots seen with orthodromic synaptic stimuli compared to thosegenerated by pulses (Fig. 2). Interestingly, as the synaptic conduc-tance is restricted to the soma, the AP height in the axon shouldbe less affected by the shunt. It seems unlikely that the ADP prop-agates very far down the axon as Kv1 channels are located in theaxon initial segment (Dodson et al., 2002) and will act to shuntthe ADP as it invades the axon. Kv2 channels are also located inthe axon initial segment, where they act to lower the ADP withincreasing frequency, allowing accelerated recovery of Nav chan-nels from inactivation (Johnston et al., 2008b). Resurgent sodiumcurrents have been reported in the MNTB (Leao et al., 2006), butthey cannot explain the ADP seen with synaptic stimulation, sincethe ADP is absent in the same cell when action potentials are evokedby somatic pulses (Fig. 2). Rather the ADP is likely explained by theTauslow AMPAR conductance, which provide 6.8 ± 1.1 nS during theADP (Fig. 3C and D).

4.3. General applicability and alternative methods

We have demonstrated a simple arithmetic manipulation ofa supra-threshold synaptic current waveform to correct for thechanging driving force during the action potential. This mode ofstimulation allows more accurate assessment of pharmacologicalagents on postsynaptic conductances and allows the generationof trains of synaptic stimulation without the inherent variabilityof quantal fluctuation or short-term facilitation or depression. Theempirical method of generating the SSG presented here will notfully account for voltage-dependent unblock of the NMDA recep-tor, however as the data in Fig. 4B shows, this had no significanteffect on a single action potential, but when generating trains ofSSGs an enhanced NMDA current may cause an additional minorcontribution to the depolarizing plateau current.

Ideally one would seek to generate a synaptic con-ductance using dynamic clamp methods (for review seePrinz et al., 2004). Dynamic clamp measures the mem-brane potential at a limiting frequency, and from a model ofthe conductance it calculates in real time and outputs theappropriate current to be injected. Use of a dynamic clampcould improve on the empirical SSG method presented here pro-vided that the real-time simulation of the MNTB conductances area good fit to the actual conductances and that the simulation can bedone fast enough to drive the dynamic clamp during the very rapidAPs of MNTB neurons, with half-widths of <0.5 ms (for example seeLien and Jonas, 2003). The advantage of this SSG method is that itcan be implemented easily, and gives a good approximation to theorthodromic input, without requiring a precise Hodgkin–Huxley

model of the conductances. This method will be equally applicableto neurons that receive remote dendritic inputs, since the actionpotential initiation is usually in the axon initial segment (i.e. closeto the soma) and it is generally applicable to any neuron wheresupra-threshold synaptic inputs are under study.

1 oscien

A

R

B

B

B

D

D

I

J

J

J

64 J. Johnston et al. / Journal of Neur

cknowledgements

Funded by the MRC.

eferences

arnes-Davis M, Forsythe ID. Pre- and postsynaptic glutamate receptors at a giantexcitatory synapse in rat auditory brainstem slices. J Physiol 1995;488:387–406.

illups B, Wong AY, Forsythe ID. Detecting synaptic connections in themedial nucleus of the trapezoid body using calcium imaging. Pflugers Arch2002;444:663–9.

rew HM, Forsythe ID. Two voltage-dependent K+ conductances with comple-mentary functions in postsynaptic integration at a central auditory synapse. JNeurosci 1995;15:8011–22.

odson PD, Barker MC, Forsythe ID. Two heteromeric Kv1 potassium channels dif-ferentially regulate action potential firing. J Neurosci 2002;22:6953–61.

odson PD, Billups B, Rusznak Z, Szucs G, Barker MC, Forsythe ID. Presynaptic ratKv1.2 channels suppress synaptic terminal hyperexcitability following actionpotential invasion. J Physiol (Lond) 2003;550:27–33.

shikawa T, Nakamura Y, Saitoh N, Li W, Iwasaki S, Takahashi T. Distinct roles of Kv1and Kv3 potassium channels at the calyx of Held presynaptic terminal. J Neurosci2003;23:10445–53.

ohnston J, Griffin SJ, Baker C, Forsythe ID. Kv4 (A-type) potassium currents in themouse medial nucleus of the trapezoid body. Eur J Neurosci 2008a;27:1391–9.

ohnston J, Griffin SJ, Baker C, Skrzypiec A, Chernova T, Forsythe ID. Initial segment

Kv2.2 channels mediate a slow delayed rectifier and maintain high frequencyaction potential firing in medial nucleus of the trapezoid body neurons. J Physiol2008b;586:3493–509.

ung HY, Mickus T, Spruston N. Prolonged sodium channel inactivation contributesto dendritic action potential attenuation in hippocampal pyramidal neurons. JNeurosci 1997;17:6639–46.

ce Methods 183 (2009) 158–164

Klug A, Trussell LO. Activation and deactivation of voltage-dependent K+ chan-nels during synaptically driven action potentials in the MNTB. J Neurophysiol2006;96:1547–55.

Koike-Tani M, Saitoh N, Takahashi T. Mechanisms underlying developmental speed-ing in AMPA-EPSC decay time at the calyx of Held. J Neurosci 2005;25:199–207.

Leao RN, Naves MM, Leao KE, Walmsley B. Altered sodium currents in auditoryneurons of congenitally deaf mice. Eur J Neurosci 2006;24:1137–46.

Lien CC, Jonas P. Kv3 potassium conductance is necessary and kinetically optimizedfor high-frequency action potential generation in hippocampal interneurons. JNeurosci 2003;23:2058–68.

Magistretti J, Mantegazza M, Guatteo E, Wanke E. Action potentials recordedwith patch-clamp amplifiers: are they genuine? Trends Neurosci 1996;19:530–4.

Mainen ZF, Sejnowski TJ. Influence of dendritic structure on firing pattern in modelneocortical neurons. Nature 1996;382:363–6.

Ogawa Y, Rasband MN. The functional organization and assembly of the axon initialsegment. Curr Opin Neurobiol 2008;18:307–13.

Postlethwaite M, Hennig MH, Steinert JR, Graham BP, Forsythe ID. Acceleration ofAMPA receptor kinetics underlies temperature-dependent changes in synapticstrength at the rat calyx of Held. J Physiol 2007;579:69–84.

Prinz AA, Abbott LF, Marder E. The dynamic clamp comes of age. Trends Neurosci2004;27:218–24.

Song P, Yang Y, Barnes-Davies M, Bhattacharjee A, Hamann M, Forsythe ID, et al.Acoustic environment determines phosphorylation state of the Kv3.1 potassiumchannel in auditory neurons. Nat Neurosci 2005;8:1335–42.

Taschenberger H, von Gersdorff H. Fine-tuning an auditory synapse for speed andfidelity: developmental changes in presynaptic waveform, EPSC kinetics, andsynaptic plasticity. J Neurosci 2000;20:9162–73.

Wang L, Gan L, Forsythe ID, Kaczmarek LK. Contribution of the Kv3.1 potassiumchannel to high-frequency firing in mouse auditory neurones. J Physiol (Lond)1998;509:183–94.