jphysiol 566.1 (2005) pp 49–59 - neuroscience · the journal of physiology online articles are...

DOI: 10.1113/jphysiol.2005.087460

2005;566;49-59; originally published online May 5, 2005; J. Physiol.

Juan Diego Goutman, Paul Albert Fuchs and Elisabeth Glowatzki

Facilitating efferent inhibition of inner hair cells in the cochlea of the neonatal rat

This information is current as of June 15, 2006

publication unless article is open access. This version of the article may not be posted on a public website for 12 months after

http://jp.physoc.org/cgi/content/full/566/1/49

This is the final published version of this article; it is available at:

[email protected]: publication. No part of this article may be reproduced without the permission of Blackwell

articles are free 12 months afterThe Journal of Physiology Online. http://jp.physoc.org/subscriptions/ go to: The Journal of Physiology Onlinepublished continuously since 1878. To subscribe to

is the official journal of The Physiological Society. It has beenThe Journal of Physiology Online

at JOHN HOPKINS UNIVERSITY on June 15, 2006 jp.physoc.orgDownloaded from


http://jp.physoc.org/subscriptions/

mailto:[email protected]

http://jp.physoc.org

J Physiol 566.1 (2005) pp 49–59 49

Facilitating efferent inhibition of inner hair cells in thecochlea of the neonatal rat

Juan Diego Goutman, Paul Albert Fuchs and Elisabeth Glowatzki

The Cochlear Neurotransmission Laboratory, Center for Hearing and Balance, Department of Otolaryngology–Head and Neck Surgery, Johns HopkinsUniversity School of Medicine, Baltimore, MD, USA

Cholinergic brainstem neurones make inhibitory synapses on outer hair cells (OHCs) in themature mammalian cochlea and on inner hair cells (IHCs) prior to the onset of hearing. We usedelectrical stimulation in an excised organ of Corti preparation to examine evoked release of acetyl-choline (ACh) onto neonatal IHCs from these efferent fibres. Whole-cell voltage-clamp recordingrevealed that low frequency (0.25–1 Hz) electrical stimulation produced evoked inhibitory post-synaptic currents (IPSCs) at a relatively high fraction of failures (65%) and with mean amplitudesof about −20 pA at −90 mV, corresponding to a quantum content of ∼1. Evoked IPSCs hadbiphasic waveforms at −60 mV, were blocked reversibly by α-bungarotoxin and strychnineand are most likely mediated by the α9/α10 acetylcholine receptor, with subsequent activationof calcium-dependent potassium (SK2) channels. Paired pulse stimulation with intervals of10–100 ms caused facilitation of 200–300% in the mean IPSC amplitude. A train of 10 pulses withan interpulse interval of 25 ms produced increasingly larger IPSCs with maximum amplitudesgreater than−100 pA due to facilitation and summation throughout the train. Repetitive efferentstimulation at 5 Hz or higher hyperpolarized IHCs by 5–10 mV and could completely preventthe generation of calcium action potentials normally evoked by depolarizing current injection.

(Resubmitted 24 March 2005; accepted after revision 4 May 2005; first published online 5 May 2005)Corresponding author E. Glowatzki: 521 Traylor Building, Johns Hopkins University School of Medicine, 720 RutlandAvenue, Baltimore, MD 21205-2195, USA. Email: [email protected]

Mechanosensory hair cells of the vertebrate ear are sub-ject to feedback regulation by cholinergic efferent neuro-nes originating in the brainstem. Electrical stimulation ofthese axons in the floor of the fourth ventricle causedsuppression of the compound afferent action potential(Galambos, 1956). Similar methods were used to describethe inhibitory effect on single afferent axons (Wiederhold,1970; Wiederhold & Kiang, 1970) where the efficacy ofinhibition was shown to depend on the rate of electricalstimulation, implying some plasticity in efferent synaptictransmission. Synaptic facilitation of efferent inhibitionwas shown directly by intracellular recording from haircells in the turtle’s inner ear (Art et al. 1984); however,equivalent effects have not been demonstrated in themammalian cochlea until now.

Medial olivocochlear efferents innervate outer hair cells(OHCs) in the first postnatal week and maintain thesesynapses through adulthood. In contrast, inner hair cells(IHCs) have contacts with medial efferent fibres beforebirth, but these presumptive synapses disappear after theonset of hearing in the second postnatal week (Simmonset al. 1996; Katz et al. 2004). These transient efferentsynapses can release ACh onto neonatal IHCs (Glowatzki

& Fuchs, 2000; Katz et al. 2004; Marcotti et al. 2004). It hasbeen suggested that this efferent feedback may be involvedin directing IHC maturation, including the formation ofafferent synapses (Simmons, 2002).

Spontaneous IPSCs in OHCs (Oliver et al. 2000;Lioudyno et al. 2004), and in neonatal IHCs (Glowatzki& Fuchs, 2000) are served by α9α10-containingACh receptors, allowing calcium influx that activatescalcium-dependent SK2 potassium channels (Elgoyhenet al. 1994; Dulon et al. 1998; Yuhas & Fuchs, 1999;Elgoyhen et al. 2001; Lustig & Peng, 2002). However,the random timing of these spontaneous events pre-vents any direct assessment of efferent release mechanicsor plasticity. Further study of the efficacy of hair cellinhibition requires the ability explicitly to evoke releasefrom the efferent endings. Here we report the resultsof experiments using electrical stimulation to causeinhibitory postsynaptic currents in IHCs of the neonatalrat cochlea. As previously described for efferent inhibitionin turtle hair cells (Art et al. 1984), the resting probability ofrelease at the IHC efferent synapse was low, but facilitatedmarkedly during repetitive stimulation of the efferentaxons. The resulting large inhibitory postsynaptic currents

C© The Physiological Society 2005 DOI: 10.1113/jphysiol.2005.087460



50 J. D. Goutman and others J Physiol 566.1

(IPSCs) could essentially clamp the IHC membranepotential to the potassium equilibrium potential, andprevent the generation of calcium action potentials inneonatal IHCs.

Methods

The preparation

Procedures for preparing and recording from the post-natal rat organ of Corti were essentially identical to thosepublished previously (Glowatzki & Fuchs, 2000). Animalprotocols were approved by the Johns Hopkins UniversityAnimal Care and Use Committee. Sprague-Dawley ratpups, 7–11 days old, were anaesthetized using pento-barbital and decapitated. The organ of Corti was exposedand the apical turn removed for recording. IHCs inthe apical turn of the organ of Corti were visualizedusing an Axioscope microscope (Zeiss, Oberkochem,Germany) with a 40× water immersion DIC objective, 4×magnification and a NC70 Newvicon camera (Dage MTI,Michigan City, IN, USA) for display. Whole-cell, tight-sealvoltage-clamp recordings were made with Sylgard-coated1 mm borosilicate glass micropipettes (WPI, Sarasota, FL,USA) ranging from 8 to 10 M� resistance. Electrodeswere advanced through the tissue under positive pipettepressure to avoid extensive dissection. In this way it waspossible to maintain the integrity of synaptic contacts onthe IHCs.

The pipette solution was (mm): 150 KCl, 3.5 MgCl2,0.1 CaCl2, 5 EGTA, 5 Hepes, 2.5 Na2ATP; pH 7.2 (KOH).The extracellular solution was (mm): 5.8 KCl, 155 NaCl,1.3 CaCl2, 0.7 NaH2PO4, 5.6 glucose, 10 Hepes; pH 7.4(NaOH). Mg2+ was excluded from the extracellularsolution to maximize current flow through the hair cellACh receptor (Weisstaub et al. 2002; Gomez-Casati et al.2005). In control experiments it was shown that quantumcontent and facilitation were not altered by the absence ofexternal Mg2+, although quantal size was approximately40% smaller in (0.9 mm) Mg2+, as predicted.

Recordings were made at room temperature (22–25◦C).Solutions, including experimental drugs, were applied bygravity flow into the bath chamber (1.5 ml volume) at arate of 2–3 ml min−1. All drugs were purchased from Sigmaand diluted from frozen stocks.

Electrical stimulation and recording

Bipolar electrical stimulation of efferent axons wasprovided by a theta glass micropipette (20–30 µm tipdiameter) positioned about 20 µm modiolar to the baseof an IHC, near the course of the inner spiral bundle.The position of this pipette was adjusted until currentpassage through it caused synaptic currents in the IHCunder study. An electrically isolated constant current

source (model DS3, Digitimer Ltd, Welwyn Garden City,UK) was triggered via the data-acquisition computer togenerate pulses up to ∼50 µA, 50–500 µs long to stimulateefferent axons. With some experience in positioning thestimulation pipette, 50 µs long pulses were sufficient forstimulation. With this method we were able to stimulateIPSCs in every IHC recorded.

Currents and voltages were recorded with pCLAMP9.2software and an Axopatch 200B amplifier (AxonInstruments, Union City, CA, USA), low-pass filtered at2–10 kHz and digitized at 10–50 kHz with a Digidata1322A board.

Indicated holding potentials were not corrected forliquid junction potentials (−4 mV). Series resistancecompensation of 50–70% was used, and the remainingseries resistance error for the recorded synaptic currentswas calculated to be smaller than 5 mV. Synapticcurrents and potentials were analysed with Minianalysis(Synaptosoft, Decatur, GA, USA). IPSCs were identifiedmanually using a search routine for event detection witha threshold of 5 times the RMS noise (7–12 pA). Risetimes (10–90%) and decay time constants were analysedonly in IPSCs that were not compromised in shape by thestimulus artifact or following events. The decay of IPSCswas fitted with a monoexponential. For further analysiswe used Origin 7.5 (OriginLab Corp., Northampton, MA,USA). A two-sample Student’s t test assuming unequalvariances and a one-tailed t test were used for statisticalanalysis. Mean values are presented ± standard deviation(s.d.) unless stated otherwise.

Quantal analysis

The amplitude distributions of spontaneous IPSCs werefitted by a single Gaussian equation. To test whetheramplitude distributions for evoked IPSCs were shaped bydistinct peaks at integral multiples of the first peak, asestablished for the neuromuscular junction (Del Castillo& Katz, 1954), we used the following equation (Sahara &Takahashi, 2001):

N (x) = Nt

k∑

r=1

mr

emr !

1√2πrσ 2

e− (x−rµ)2

2rσ2 (1)

where N(x) was the number of events with x amplitude, N t

was the total number of evoked events, r was the numberof quanta released, m was the quantum content, and µ andσ 2 were the mean and variance of the distribution. We usedthis equation under the assumption that low values for thequantum content were observed, with k, the maximumvesicle number, taking a value of 3 in most cases. Thisfunction uses for each peak a mean and variance given bya Gaussian distribution, whereas the relative amplitude ofeach maximum is estimated by Poisson statistics governingthe probability of release of one vesicle, two vesicles, or

C© The Physiological Society 2005



J Physiol 566.1 Facilitating cochlear inhibition 51

more. This approach allowed us to calculate the quantumcontent ‘m’. The Poisson distribution also could be usedfor a separate estimate of quantum content based on thefraction of failures (del Castillo & Katz, 1954).

Best fits for amplitude distributions were evaluated byfinding the least squared difference from the data usingOrigin 7.5.

Results

Electrically evoked IPSCs in IHCs

To investigate the effects of cholinergic efferent input onneonatal IHC activity, we did whole-cell patch clamprecordings from 43 IHCs in excised apical turns of organof Corti from 7- to 11-day-old rats. In these recordings weelectrically stimulated efferent fibres that directly synapseonto the IHCs. These fibres originate in the brainstem inthe medial olivocochlear system and during developmentreach the IHC area before birth (Simmons, 2002). Medialefferent axons were stimulated electrically with a thetaglass micropipette positioned close to the IHC underinvestigation, about 20 µm modiolar to the IHC’s base(for details see Methods) (Fig. 1A).

When timed current pulses (of ∼50 µA in size,50 µs long) were delivered via the stimulating pipette,transient inward currents (evoked IPSCs) at a holdingpotential of −90 mV were evoked in every IHC tested(Fig. 1B). Application of 1 µm tetrodotoxin to the pre-paration produced a reversible loss of the synapticresponse, confirming that these currents resulted frompropagation of action potentials in presynaptic axons,presumably to release ACh onto IHCs. Consistent withthis interpretation, the IPSCs were reversibly blocked by100 nm strychnine (Fig. 1C) and 300 nm α-bungarotoxin(Fig. 1D), like the α9α10-containing nAChR (Elgoyhenet al. 1994; Verbitsky et al. 2000; Elgoyhen et al. 2001),and spontaneous synaptic currents recorded in IHCs(Glowatzki & Fuchs, 2000).

Evoked IPSCs were similar in shape to spontaneousIPSCs and to IPSCs produced by elevated extracellularpotassium as previously described in both IHCs(Glowatzki & Fuchs, 2000; Katz et al. 2004) and OHCs(Oliver et al. 2000; Lioudyno et al. 2004). The mean10–90% rise time of the evoked IPSCs was 9.6 ± 3.8 ms(mean ± s.d., n = 298 IPSCs from 3 IHC recordings)and the mean time constant of decay was 50.2 ± 17.4 ms(n = 244 IPSCs from 3 IHC recordings) at a holdingpotential of −90 mV.

Like the previously described spontaneous IPSCs,evoked IPSCs reversed from completely inward at −90 mVto a biphasic inward and outward waveform at membranepotentials positive to the potassium equilibrium potential(EK = −82 mV) (Fig. 1E). The initial inward current ofthe biphasic IPSC is thought to result from sodium

and calcium flux through the α9α10-containing nAChR(Elgoyhen et al. 1994; Verbitsky et al. 2000; Elgoyhenet al. 2001; Gomez-Casati et al. 2005). The subsequentactivation of calcium-dependent small conductance SK2potassium channels underlies the outward current (Dulonet al. 1998; Yuhas & Fuchs, 1999). Negative to EK, at−90 mV, both the AChR current and SK current areinward. Between EK and −40 mV the AChR currentis directed inward whereas the SK current is directedoutward, resulting in a biphasic response. Positive to−40 mV the much larger outward current obscures thesmall initial current through the AChRs.

To characterize the effect of efferent stimulation on IHCmembrane potential, we recorded evoked inhibitory post-synaptic potentials (IPSPs) in current clamp (Fig. 1F). Thesynaptic potentials mirrored the synaptic currents, startingwith a short depolarization and followed by a longerand larger hyperpolarization at potentials close to theIHC resting membrane potential at −60 mV. This hyper-polarization makes the effect of efferent synaptic activityon IHCs inhibitory (see also Fig. 5).

Taken together, the pharmacological data and thebiphasic IPSC waveform confirm that with electricalstimulation we were able to exert cholinergic efferentsynaptic inhibition of postnatal IHCs.

Synaptic strength of efferent contacts

As a measure of synaptic strength at the IHC efferentsynapse, we calculated quantum content, ‘m’, the numberof vesicles released per stimulation. We analysed ourdata in the following way. At a stimulation rate of0.25 or 1 Hz we observed a relatively high fraction offailures of 0.65 ± 0.16 (n = 32 IHC recordings) and amean amplitude of ‘successes’ of −22 ± 9 pA (n = 32IHC recordings, 11 900 IPSCs analysed) at a holdingpotential of −90 mV. Figure 2A illustrates a recordingwith three successful stimulations and one failure.Additionally, spontaneous IPSCs occurred during therecordings that were not time-coupled with the stimulus(Fig. 2A). In each recording, evoked IPSC amplitudesfluctuated considerably (Fig. 2B). Examples of amplitudedistributions for evoked and spontaneous IPSCs are shownin Fig. 2C.

We estimated quantum content by four differentapproaches for six IHC recordings that provided enoughdata for both evoked and spontaneous IPSCs; a summaryof the results is provided in Table 1. One approach wasto divide the mean evoked IPSC amplitude by the meanquantal size (q). Assuming that the spontaneous IPSCsare uniquantal events, we estimated the mean quantalsize q from the amplitude distributions of spontaneousIPSCs fitted by a single Gaussian equation and obtained−18 ± 2 pA (Fig. 2C, inset). Dividing this value into





Figure 1. Inhibitory postsynaptic currents (IPSCs) evoked in neonatal rat apical IHCs by electricalstimulation of cholinergic efferent axonsA, for bipolar electrical stimulation of efferent fibres some supporting cells were removed and a theta glassmicropipette, 10–15 µm in diameter, was positioned ∼10–20 µm below the base of the IHC. Scale bar 10 µm.Positions of 2 IHCs and recording pipette outlined. B, evoked IPSCs were eliminated reversibly by 1 µM tetrodotoxin(TTX) (n = 2). C, IPSCs were reversibly blocked by 100 nM strychnine, which blocks α9α10 AChR (n = 2). D, IPSCswere reversibly inhibited by 300 nM α-bungarotoxin, an α9α10 AChR antagonist (n = 2). B–D, traces for eachcontrol, drug application and washout are averages of 50–100 sweeps (stimulation at 1 Hz). Each new solutionwas applied for about 5 min before data collection. Vh = −90 mV. E, IPSC waveform varied with holding potential.IPSCs reversed from completely inward at −90 mV to a biphasic inward (arrowheads) and outward (stars) waveformat membrane potentials positive to the potassium equilibrium potential. F, synaptic potentials mirrored synapticcurrents and varied similarly with membrane potential: an initial depolarization (arrowheads) was followed by ahyperpolarization (stars). Arrow indicates time of electrical shock in E and F.





the mean of evoked IPSCs yielded a mean quantumcontent (m) of 1.0 ± 0.5 vesicles released per stimulus.Alternatively, we estimated an average quantum contentof 0.9 ± 0.5 from the ‘fraction of failures’ based on thePoisson distribution (del Castillo & Katz, 1954). ThePoisson assumptions also underlie the estimate of m asequivalent to the reciprocal of the coefficient of variationsquared (m = 1/CV2) and provided an average value of1 ± 0.6 for the six cells studied. Finally, the overall evokedIPSC amplitude distributions were fit with eqn (1) (seeMethods) (Fig. 2C). The first peak of this fit providedan estimate for the quantal size q of −20 ± 5 pA, quitesimilar to the value provided by the Gaussian fit ofthe spontaneous IPSC distribution. The mean quantumcontent for all IHCs obtained from fitting the evoked IPSCamplitude distributions again was 1 ± 0.4. However, whilethis average value was quite comparable, it should be noted

Figure 2. Amplitude distributions of evoked and spontaneous IPSCsA, electrical stimulation (arrows) of the efferents at 1 Hz caused 3 IPSCs and 1 failure. Some spontaneous IPSCs wereobserved as well. B, evoked IPSCs varied in amplitude. C, amplitude distribution of 114 evoked IPSCs producedby 1 Hz stimulation (300 times), and amplitude distribution of spontaneous IPSCs (inset) from the same cell (thenoise floor is shown by the 3 bars centred around 0 mV). The spontaneous IPSC amplitude distribution was fitwith a single Gaussian providing a mean of −19 ± 5 pA (n = 38 spontaneous IPSCs). Dividing mean evoked IPSCamplitude by the mean spontaneous IPSC amplitude provided a quantum content of 0.64. Quantal parametersalso were obtained from the Poisson distribution derived from the fit of eqn (1) (Methods) to the evoked IPSCamplitude distribution. This gave a mean evoked IPSC amplitude of −22 ± 7 pA for the first peak (n = 114 IPSCs)and a value of 0.75 for the quantum content.

that the estimates of quantum content obtained from thisapproach differed somewhat from those obtained by othermethods (Table 1).

Nonetheless, taken together the different estimates ofquantum content were similar: the efferent IHC synapsehas a low probability of release at low frequency, with arange of 0.4 up to 2 vesicles released per action potential(at 1 Hz).

Two-pulse facilitation of efferent release

Does the efficacy of the IHC efferent synapse change withhigher rates of stimulation? We used paired pulse protocolsto probe facilitation at the IHC efferent synapse. Pulsepairs with intervals ranging from 1 to 500 ms were pre-sented 50 times each, at a rate of 0.25 Hz (Fig. 3A–D).IPSCs or failures could be seen following the first, or the





Table 1. Quantal analysis of cholinergic IPSCs recorded in neonatal IHC (Vh = −90 mV)

Quantal size (q)Quantum content (m)

(a) Peak (b) Firstof sIPSC peak of eIPSC (i) Mean (iv) By

amplitude amplitude eIPSC/ fittingdistribution distribution peak of sIPSC (ii) By (iii) By of eIPSC

Cell no. (pA) n (pA) n amp dist failures 1/CV2 amp dist

1 −15 ± 5 51 −15 ± 5 160 0.62 0.52 0.44 0.862 −20 ± 5 85 −29 ± 11 247 1.78 1.73 1.97 0.923 −19 ± 5 140 −17 ± 5 188 1.16 0.99 0.97 1.864 −19 ± 6 114 −15 ± 4 123 0.40 0.40 0.36 1.035 −19 ± 5 39 −22 ± 7 115 0.64 0.48 0.56 0.756 −15 ± 3 52 −20 ± 6 101 0.92 0.70 0.70 0.85Mean ± S.D. −18 ± 2 −20 ± 5 1.0 ± 0.5 0.9 ± 0.5 1.0 ± 0.6 1.0 ± 0.4

Evoked (e)IPSCs were obtained in protocols with stimulation at 1Hz. Spontaneous (s)IPSCs were not coupled to stimulation. Quantalsize q for every cell was calculated by two methods: (a) finding the peak (mean value) of the sIPSC amplitude distribution by a singleGaussian fit, and (b) finding the first peak of the eIPSC amplitude distribution by fitting with equation 1 (see Methods). Both methodsprovided similar results in comparison. The number of IPSCs in each distribution is included (n). Means ± S.D. are shown. Quantumcontent was estimated in four different ways. (i) Dividing mean eIPSC amplitude by the quantal size (mean sIPSC amplitude), as foundin (a). The mean eIPSC amplitude included failures. (ii) By the rate of failures, with m = ln(N/N0), where N is the number of pulses andN0 the number of failures. (iii) By coefficient of variation CV, with m = 1/CV2. (iv) By fitting the eIPSC amplitude distributions usingeqn (1). Methods (ii), (iii) and (iv) assumed Poisson distribution of the number of quanta released per stimulus.

second (Fig. 3A, insets) or both pulses (not shown). Theresulting mean IPSC amplitudes, which include failures,were calculated as a ‘facilitation ratio’ (S2 − S1)/S1, withS2 representing the response to the second pulse, and S1

representing the response to the first pulse (Mallart &Martin, 1968). The facilitation ratio can be considered ameasure of the percentage mean IPSC amplitude increasedue to facilitation. The 1 ms and 5 ms paired pulse intervalswere too short to measure IPSC amplitudes in response tothe first pulse for S1. Therefore, in addition to the pairedpulse protocol, which provided S2, we used a single pulseprotocol at 0.25 Hz recorded in the same IHC to assess S1.This was possible, as IPSCs evoked at 0.25 Hz rate did notshow facilitation.

Facilitation became evident with paired pulse inter-vals of 10–100 ms (Fig. 3A and E) with mean IPSCamplitudes S2 increasing 200–300% compared to S1. Forexample, for a 25 ms interval, S1 was −11 ± 12 pA and S2

was −24 ± 25 pA (n = 11 IHC recordings). In addition,although we have not quantified this effect, we noted thatpaired pulse protocols with facilitation also showed anincreased number of IPSCs that were not time-coupledto the stimulus.

As presented, the mean IPSC amplitude, Sx , reflects twofactors, the probability of occurrence, Px , and the meanamplitude of the successful IPSCs, Ax ; such that Sx = AxPx .To better understand the mechanism of facilitation, Ax

and Px were analysed separately in the original records(Fig. 3F). As seen there, the principal effect of pairedpulse stimulation was to increase the probability of IPSCoccurrence after the second pulse by 150–250%. That is,

IPSCs occurred much more reliably when two pulses werepresented with a 10–100 ms interval. The effect on themean amplitude of successful IPSCs was also significant,though smaller with a 30% amplitude increase of A2. Forexample, for a double pulse interval of 25 ms, the meanratio of P2/P1 was 1.78 ± 0.73 (n = 11 IHCs, one-tailedt test, P < 0.01) and the mean ratio A2/A1 was 1.34 ± 0.33(n = 11 IHCs, one-tailed t test, P < 0.01). We estimated(both by ‘method of failures’ and by 1/CV2) the quantumcontent for the facilitated IPSC S2 and compared it withthat of S1 (Fig. 3G). For paired pulse intervals of 25 ms,quantum content rose 150–550% (0.09–0.99 for S1 and0.13–2.08 for S2) in 9 of 10 IHCs.

Facilitation and summation during shock trains

Inhibition in the intact cochlea has been most often studiedusing extended trains of electrical shocks to activate theefferent axons (Galambos, 1956; Wiederhold & Kiang,1970). The time course of efferent facilitation in the excisedorgan of Corti was characterized further using trains of 10pulses with an interpulse interval of 25 ms; 30–50 suchtrains were presented at a 20 s interval, with the IHCat a holding potential of −90 mV. Facilitation as wellas summation augmented the average compound IPSCamplitudes to a value of −128 ± 117 pA (n = 14 IHCrecordings), about 6 times the size of the mean IPSCamplitude in response to single pulses. The biggest averagecompound IPSC reached a value of −450 pA (Fig. 4A).We analysed the facilitation ratio in 14 IHC recordingsand compared each mean IPSC amplitude S2 to S10 in the





pulse train to the mean amplitude of the first IPSC S1 inthe train (Fig. 4B). In these instances, the amplitude ofthe facilitated IPSC is slightly underestimated, since thetrue preceding ‘baseline’ was not available. Nonetheless,every mean IPSC amplitude S2 to S10 in the pulse train wassignificantly larger than S1 showing facilitation (one-tailedt test, P < 0.05, n = 14 IHC recordings, 30–50 pulse trainsper IHC averaged). Peak values for an increase in meanIPSC amplitude were found for S4 and S5 (200%), thendeclined to values of 150% for S6 and S7 and stayed at100% increase for S8 through S10. This decline of increase ofmean IPSC amplitudes during the stimulation train couldbe due to saturation of nAChRs, or other pre- and post-synaptic processes, and therefore this phenomenon willneed further examination. In addition to facilitation of

Figure 3. Paired-pulse facilitation of IPSCs in IHCsA, average (n = 50) IPSCs evoked with paired pulses 10 ms apart. Inset shows individual records illustrating thevariability of response: failure occurring after the first, or second shock (x scale: 50 ms, y scale: 20 pA). B–D,average (n = 50) IPSCs evoked with different pulse intervals. E, facilitation of the second IPSC as a function ofpulse interval. The mean IPSC amplitude Sx = PxAx , the probability of occurrence, Px , times the mean amplitudeof successful IPSCs, Ax . The facilitation index was computed as the ratio (S2 – S1)/S1. For intervals between 10and 100 ms IPSCs showed facilitation: S2 was 200–300% bigger than the S1 (∗P < 0.01, †P < 0.05, one-tailedt test). F, IPSC amplitude Ax (failures not included) and IPSC probability Px of success were plotted separately.Facilitation was mostly due to a 150–250% rise in probability Px of transmitter release; however, there was alsoa significant, about 30% increase in IPSC amplitude Ax for 10, 25 and 50 ms pulse intervals (∗ and † same as inE). E–F, facilitation for 50 pulse pairs presented at 0.25 Hz for 5–11 IHCs (mean and standard error of the meanshown). G, quantum content (calculated as 1/CV2) of the second IPSC compared to that of the first with a 25 msinterval. Each data point represents 1 IHC recording. Nine out of 10 IHCs showed facilitation as an increase inquantum content (range 0.09–0.99 for S1 and 0.13–2.08 for S2).

single IPSCs, summation also contributed to the amplitudeof the compound IPSCs.

Following the last pulse, the compound IPSC decayedslowly, over the course of nearly 1 s. This is considerablyslower than the 50 ms time constant of decay for singleIPSCs. Several processes could contribute to this slowdecay. One possibility is suggested by the inset of Fig. 4A,which shows an individual record in which markedfluctuations are observed on the falling phase of thecompound IPSC that were not seen in the averagedrecords. Some of these fluctuations were similar in wave-form to ‘spontaneous’ IPSCs that occasionally followedthe paired pulse protocols. It was already mentioned thatin double pulse protocols that induced facilitation, thenumber of IPSCs increased that were not time-coupled





to the pulse. This effect also may occur after a train ofpulses and could reflect delayed release from the efferentterminal. A second possibility is that there is enhanced anddelayed calcium-induced calcium release from the sub-synaptic cistern of the hair cell (Lioudyno et al. 2004),causing spontaneous transient currents through the post-synaptic SK channels. Like in OHCs (Saito, 1980), a sub-synaptic cistern that could serve as a calcium store hasbeen found postsynaptically facing neonatal IHC efferentsynapses (Ginzberg & Morest, 1984; Sobkowicz, 1992;Bruce et al. 2000). This phenomenon therefore needsfurther investigation.

Figure 4. Repetitive stimulation of IHC efferents resulted inlarge compound IPSCs due to summation and facilitationA, response of an IHC at Vh = −90 mV to a 10 pulse, 25 ms interval,train. The pulse train was presented at a 20 s interval and repeated30–50 times, the average compound IPSC is shown. Inset: anindividual record showing marked fluctuations on the falling phase ofthe compound IPSC. B, facilitation relative to the first response duringa 10 pulse, 25 ms interval, train. Data from 14 IHC recordings areincluded, each with 30–50 repeats of the 10 pulse train. Mean IPSCamplitude was significantly increased throughout the train(mean ± S.E.M., ∗P < 0.01, †P < 0.05, one-tailed t test).

In summary, the experiments of Figs 3 and 4 showthat compound IPSCs can grow much stronger dueto facilitation and summation at the efferent synapsecompared to the effect of a single IPSC.

Inhibition of IHC excitability

Neonatal IHCs generate calcium action potentialsspontaneously or in response to injected current (Kroset al. 1998; Marcotti et al. 2003a,b). We used depolarizingcurrent steps through the recording electrode to causethe activation of repetitive action potentials in rat IHCsduring the second postnatal week. Three levels of currentinjection were used (30–50 pA) to depolarize the IHC from−80 to levels between −70 and −55 mV. The latter twosteps evoked repetitive activity at frequencies between 2and 10 Hz (Fig. 5). When these excitatory stimuli werecombined with efferent pulse trains at frequencies of 5 Hzand greater, action potentials were essentially eliminatedand the steady membrane potential hyperpolarized by5–10 mV (Fig. 5C). Thus, efferent inhibition becomespotent when transmitter release has been facilitated byhigher frequency firing.

Discussion

Cholinergic efferent fibres innervate adult OHCs andneonatal IHCs in the mammalian cochlea. It has beenshown that the cholinergic effect on both these typesof hair cells is essentially identical, involving nicotinicAChRs composed of α9 and α10 subunits (Elgoyhenet al. 1994; Elgoyhen et al. 2001; Lustig & Peng, 2002), andsubsequently activated calcium-dependent SK channels(Evans, 1996; Dulon et al. 1998; Yuhas & Fuchs, 1999).It remains to be seen whether a postsynaptic calcium store(the synaptoplasmic cistern) supports inhibition of IHCs,as has been found for OHCs (Evans et al. 2000; Lioudynoet al. 2004). There is substantial evidence that cholinergicinhibition of hair cells in various end-organs of othervertebrates results from similar, if not identical, synapticmechanisms (Flock & Russell, 1973; Art et al. 1984; Fuchs& Murrow, 1992; Sugai et al. 1992). When functionalmeasures of efferent inhibition have been made, it hasbeen found that relatively high frequency trains of efferentaction potentials (assumed to result from the electricalshocks used for activation) were required to produceconsistent, measurable effects (Wiederhold & Kiang, 1970;Flock & Russell, 1973; Art et al. 1984). Studies in turtle haircells demonstrated that the probability of release to singleaction potentials was significantly less than unity, rangingfrom 0.08 to 0.3, and that transmitter release from efferentendings facilitated strongly during repetitive firing (Artet al. 1984).





The present study demonstrates that efferent axons inthe mammalian cochlea show resting levels of release,and plasticity similar to that found in the turtle. Singleshocks at frequencies less than 2 Hz evoked only smallIPSCs in IHCs with a mean probability of occurrence of0.35. However, with repeated stimulation, IPSCs were bothlarger and more likely. For trains of 10 shocks the summed,facilitated IPSCs reliably reached more than −100 pA inamplitude. Likewise, calcium action potentials in IHCs

Figure 5. Efferent stimulation could prevent calcium action potentials in neonatal IHCsA, current injections of 30–50 pA for 10 s induced repetitive firing of calcium action potentials in IHCs, with largercurrents causing higher frequency firing. B, efferent stimulation (onset indicated by downward arrow) at 2 Hz hadlittle effect on IHC excitability. The downward deflections in the records are stimulus artifacts (examples indicated byarrow heads). C, efferent stimulation at 5 Hz caused a hyperpolarization of 5–10 mV and suppressed the generationof calcium action potentials. A few calcium action potentials still occurred during the hyperpolarization (indicatedby stars). Time scale for all figures: 2 s.

were only inhibited by efferent activity at rates of 5 Hz orgreater. This frequency nearly corresponds to the inter-pulse interval (100 ms) at which facilitation first becomesevident in two-pulse protocols. The implication of thisresult is that efferent inhibition only is effective whenoccurring repetitively, and at sufficiently high frequenciesso that facilitation of transmitter release can occur. In thisway feedback from the central nervous system operatesin a kind of failsafe mode, having consequence only when





strongly driven, and spontaneous, inadvertent activity willnot alter cochlear function significantly. Given the strongconservation of efferent synaptic mechanisms amongvertebrate hair cells, it seems likely that the behaviourdescribed here for neonatal IHCs will also occur in OHCsof the mature mammalian cochlea. The ability to stimulatethe efferents electrically and produce strongly facilitatedinhibition of defined duration, will enable studies ofthe mechanisms of fast and slow inhibition seen in vivo(Sridhar et al. 1997), as well as further investigationof possible longer-term effects on OHC electromotility(Dallos et al. 1997).

The present study also reaffirms the conclusion thatfunctional inhibitory synapses are present on neonatalIHCs prior to the onset of hearing. What purposedo these synapses serve? Clearly they can suppress theexcitability of the IHCs, and presumably prevent evokedand spontaneous transmitter release from ribbon synapsesknown to be functional at this stage (Glowatzki & Fuchs,2002). Previous studies have suggested that this trans-ient efferent innervation may play a role in the ultimatefunctional maturation of cochlear hair cells (Simmons,2002). Most impressively, surgical lesion of the efferentnerve supply caused kittens to fail to develop normalhearing (Walsh et al. 1998). The pattern of loss was muchlike that seen with OHC damage. Also, in transgenicmice lacking the principal voltage-gated calcium channelof cochlear hair cells, the normal maturation of IHCexcitability failed to occur and efferent synapses remainedin place long after the normal period (Brandt et al. 2003).Thus, a neonatal period of synaptic cross-talk betweenefferent and afferent neurones may be necessary to theultimate innervation pattern, and the final functional stateof both IHCs and OHCs.

It is unknown at present what, if any, endogenous firingtakes place among hair cells and efferent neurones beforethe onset of hearing in vivo. A few studies documentedspontaneous, or even rhythmic, activity in afferent nervefibres prior to the onset of hearing (Lippe, 1994; Gummer& Mark, 1994). What remains uncertain is whether thatactivity depends on hair cell transmitter release, or arisesfrom intrinsic excitability of the afferent neuron, as canoccur in vitro (Lin & Chen, 2000). Likewise, several studieshave shown that IHCs are capable of firing spontaneouscalcium action potentials in preparations in vitro (Marcottiet al. 2003a,b); it remains to be determined whether theydo so in vivo. Finally, neonatal medial olivocochlear neuro-nes recorded in a brainstem slice preparation, identified byretrograde labelling from the cochlea, have the ability todischarge tonically, when stimulated with current injection(Fujino et al. 1997). Ultimately it will be necessary toaddress these questions in the intact, developing earto learn if mechanisms of spontaneous activity will beessential to activity-dependent maturation, as shown forthe retina (for review see Shatz, 1996).

References

Art JJ, Fettiplace R & Fuchs PA (1984). Synaptichyperpolarization and inhibition of turtle cochlear hair cells.J Physiol 356, 525–550.

Brandt A, Striessnig J & Moser T (2003). CaV1.3 channels areessential for development and presynaptic activity ofcochlear inner hair cells. J Neurosci 23,10832–10840.

Bruce LL, Christensen MA & Warr B (2000). Postnataldevelopment of efferent synapses in the rat cochlea. J CompNeurol 423, 532–548.

Dallos P, He DZ, Lin X, Sziklai I, Mehta S & Evans BN (1997).Acetylcholine, outer hair cell electromotility, and thecochlear amplifier. J Neurosci 17, 2212–2226.

Del Castillo J & Katz B (1954). Quantal components of theend-plate potential. J Physiol 124, 560–573.

Dulon D, Luo L, Zhang C & Ryan AF (1998). Expression ofsmall-conductance calcium-activated potassium channels(SK) in outer hair cells of the rat cochlea. Eur J Neurosci 10,907–915.

Elgoyhen AB, Johnson DS, Boulter J, Vetter DE & Heinemann S(1994). Alpha 9: an acetylcholine receptor with novelpharmacological properties expressed in rat cochlear haircells. Cell 79, 705–715.

Elgoyhen AB, Vetter DE, Katz E, Rothlin CV, Heinemann SF &Boulter J (2001). alpha10: a determinant of nicotiniccholinergic receptor function in mammalian vestibular andcochlear mechanosensory hair cells. Proc Natl Acad Sci U S A98, 3501–3506.

Evans MG (1996). Acetylcholine activates two currents inguinea-pig outer hair cells. J Physiol 491, 563–578.

Evans MG, Lagostena L, Darbon P & Mammano F (2000).Cholinergic control of membrane conductance andintracellular free Ca2+ in outer hair cells of the guinea pigcochlea. Cell Calcium 28, 195–203.

Flock A & Russell IJ (1973). The post-synaptic action ofefferent fibers in the lateral line organ of the burbot Lota lota.J Physiol 235, 591–605.

Fuchs PA & Murrow BW (1992). Cholinergic inhibition ofshort (outer) hair cells of the chick’s cochlea. J Neurosci 12,800–809.

Fujino K, Koyano K & Ohmori H (1997). Lateral and medialolivocochlear neurons have distinct electrophysiologicalproperities in the rat brain slice. J Neurophysiol 77,2788–2804.

Galambos R (1956). Suppression of auditory nerve activity bystimulation of efferent fibers to cochlea. J Neurophysiol 19,424–437.

Ginzberg RD & Morest DK (1984). Fine structure of cochlearinnervation in the cat. Hear Res 14, 109–127.

Glowatzki E & Fuchs PA (2000). Cholinergic synapticinhibition of inner hair cells in the neonatal mammaliancochlea. Science 288, 2366–2368.

Glowatzki E & Fuchs PA (2002). Transmitter release at the haircell ribbon synapse. Nat Neurosci 5, 147–154.

Gomez-Casati ME, Fuchs PA, Elgoyhen AB & Katz E (2005).Biophysical and pharmacological characterization ofnicotinic cholinergic receptors in rat cochlear inner haircells. J Physiol 566, 103–118.





Gummer AW & Mark RF (1994). Patterned neural activity inbrain stem auditory areas of a prehearing mammal, thetammar wallaby (Macropus eugenii). Neuroreport 5,685–688.

Katz E, Elgoyhen AB, Gomez-Casati ME, Knipper M, VetterDE, Fuchs PA & Glowatzki E (2004). Developmentalregulation of nicotinic synapses on cochlear inner hair cells.J Neurosci 24, 7814–7820.

Kros CJ, Ruppersberg JP & Rusch A (1998). Expression of apotassium current in inner hair cells during development ofhearing in mice. Nature 394, 281–284.

Lin X & Chen S (2000). Endogenously generated spontaneousspiking activities recorded from postnatal spiral ganglionneurons in vitro. Brain Res Dev Brain Res 119, 297–305.

Lioudyno M, Hiel H, Kong JH, Katz E, Waldman E,Parameshwaran-Iyer S, Glowatzki E & Fuchs PA (2004). A‘synaptoplasmic cistern’ mediates rapid inhibition ofcochlear hair cells. J Neurosci 24, 11160–11164.

Lippe WR (1994). Rhythmic spontaneous activity in thedeveloping avian auditory system. J Neurosci 14,1486–1495.

Lustig LR & Peng H (2002). Chromosome location andcharacterization of the human nicotinic acetylcholinereceptor subunit alpha (α) 9 (CHRNA9) gene. CytogenetGenome Res 98, 154–159.

Mallart A & Martin AR (1968). The relation between quantumcontent and facilitation at the neuromuscular junction of thefrog. J Physiol 196, 593–604.

Marcotti W, Johnson SL, Holley MC & Kros CJ (2003a).Developmental changes in the expression of potassiumcurrents of embryonic, neonatal and mature mouse innerhair cells. J Physiol 548, 383–400.

Marcotti W, Johnson SL & Kros CJ (2004). A transientlyexpressed SK current sustains and modulates actionpotential activity in immature mouse inner hair cells.J Physiol 560, 691–708.

Marcotti W, Stuart LJ, Rusch A & Kros CJ (2003b). Sodium andcalcium currents shape action potentials in immature mouseinner hair cells. J Physiol 552, 743–761.

Oliver D, Klocker N, Schuck J, Baukrowitz T, Ruppersberg JP &Fakler B (2000). Gating of Ca2+-activated K+ channelscontrols fast inhibitory synaptic transmission at auditoryouter hair cells. Neuron 26, 595–601.

Sahara Y & Takahashi T (2001). Quantal components of theexcitatory postsynaptic currents at the rat central auditorysynapse. J Physiol 536, 189–197.

Saito K (1980). Fine structure of the sensory epithelium of theguinea pig organ of Corti: afferent and efferent synapses ofhair cells. J Ultrastructural Res 71, 222–232.

Shatz CJ (1996). Emergence of order in visual systemdevelopment. Proc Natl Acad Sci U S A 93, 602–608.

Simmons DD (2002). Development of the inner ear efferentsystem across vertebrate species. J Neurobiol 53, 228–250.

Simmons DD, Mansdorf NB & Kim JH (1996). Olivocochlearinnervation of inner and outer hair cells during postnatalmaturation: evidence for a waiting period. J Comp Neurol370, 551–562.

Sobkowicz HM (1992). The development of innervation in theorgan of Corti. In Development of Auditory and VestibularSystems, ed. Romand R, pp. 59–100. Elsevier, Amsterdam.

Sridhar TS, Brown MC & Sewell WF (1997). Uniquepostsynaptic signaling at the hair cell efferent synapsepermits calcium to evoke changes on two time scales.J Neurosci 17, 428–437.

Sugai T, Yano J, Sugitani M & Ooyama H (1992). Actions ofcholinergic agonists and antagonists on the efferent synapsein the frog sacculus. Hear Res 61, 56–64.

Verbitsky M, Rothlin CV, Katz E & Elgoyhen AB (2000). Mixednicotinic-muscarinic properties of the alpha9 nicotiniccholinergic receptor. Neuropharmacology 39, 2515–2524.

Walsh EJ, McGee J, McFadden SL & Liberman MC (1998).Long-term effects of sectioning the olivocochlear bundle inneonatal cats. J Neurosci 18, 3859–3869.

Weisstaub N, Vetter DE, Elgoyhen AB & Katz E (2002). Thealpha9alpha10 nicotinic acetylcholine receptor is permeableto and is modulated by divalent cations. Hear Res 167,122–135.

Wiederhold ML (1970). Variations in the effects of electricstimulation of the crossed olivocochlear bundle on cat singleauditory-nerve-fiber responses to tone bursts. J Acoust SocAm 48, 966–977.

Wiederhold ML & Kiang NY (1970). Effects of electricstimulation of the crossed olivocochlear bundle on singleauditory-nerve fibers in the cat. J Acoust Soc Am 48, 950–965.

Yuhas WA & Fuchs PA (1999). Apamin-sensitive,small-conductance, calcium-activated potassium channelsmediate cholinergic inhibition of chick auditory hair cells.J Comp Physiol [A] 185, 455–462.

Acknowledgements

This work was supported by research grants from the NationalInstitute for Deafness and Communication Disorders of theNational Institutes for Health R01 DC00276 to PAF and R01DC06476 to EG and a research grant from the Human FrontierScience Program RGY19/2004 to EG.




DOI: 10.1113/jphysiol.2005.087460

2005;566;49-59; originally published online May 5, 2005; J. Physiol.

Juan Diego Goutman, Paul Albert Fuchs and Elisabeth Glowatzki Facilitating efferent inhibition of inner hair cells in the cochlea of the neonatal rat

This information is current as of June 15, 2006

& ServicesUpdated Information

http://jp.physoc.org/cgi/content/full/566/1/49including high-resolution figures, can be found at:

Permissions & Licensing

http://jp.physoc.org/misc/Permissions.shtmlor in its entirety can be found online at: Information about reproducing this article in parts (figures, tables)

Reprints http://jp.physoc.org/misc/reprints.shtml

Information about ordering reprints can be found online:



http://jp.physoc.org/misc/Permissions.shtml

http://jp.physoc.org/misc/reprints.shtml


jphysiol 566.1 (2005) pp 49–59 - neuroscience · the journal of physiology online articles are...

Documents