The generation of resting membrane potentials in an inner ear hair cell system

Usher Syndrome: Genetics and Molecular Links of Hearing Loss and Directions for Therapy

Usher syndrome (USH) is an autosomal recessive (AR) disorder that permanently and severely affects the senses of hearing, vision, and balance. Three clinically distinct types of USH have been identified, decreasing in severity from Type 1 to 3, with symptoms of sensorineural hearing loss (SNHL), retinitis pigmentosa (RP), and vestibular dysfunction. There are currently nine confirmed and two suspected USH-causative genes, and a further three candidate loci have been mapped. The proteins encoded by these genes form complexes that play critical roles in the development and maintenance of cellular structures within the inner ear and retina, which have minimal capacity for repair or regeneration. In the cochlea, stereocilia are located on the apical surface of inner ear hair cells (HC) and are responsible for transducing mechanical stimuli from sound pressure waves into chemical signals. These signals are then detected by the auditory nerve fibers, transmitted to the brain and interpreted as sound. Disease-causing mutations in USH genes can destabilize the tip links that bind the stereocilia to each other, and cause defects in protein trafficking and stereocilia bundle morphology, thereby inhibiting mechanosensory transduction. This review summarizes the current knowledge on Usher syndrome with a particular emphasis on mutations in USH genes, USH protein structures, and functional analyses in animal models. Currently, there is no cure for USH. However, the genetic therapies that are rapidly developing will benefit from this compilation of detailed genetic information to identify the most effective strategies for restoring functional USH proteins.

Sensory transduction at the frog semicircular canal: how hair cell membrane potential controls junctional transmission

At the frog semicircular canals, the afferent fibers display high spontaneous activity (mEPSPs), due to transmitter release from hair cells. mEPSP and spike frequencies are modulated by stimulation that activates the hair cell receptor conductance. The relation between receptor current and transmitter release cannot be studied at the intact semicircular canal. To circumvent the problem, we combined patch-clamp recordings at the isolated hair cell and electrophysiological recordings at the cytoneural junction in the intact preparation. At isolated hair cells, the K channel blocker tetraethylammonium (TEA) is shown to block a fraction of total voltage-dependent K-conductance (IKD) that depends on TEA concentration but not on membrane potential (V_m). Considering the bioelectric properties of the hair cell, as previously characterized by this lab, a fixed fractional block of IKD is shown to induce a relatively fixed shift in V_m, provided it lies in the range −30 to −10 mV. The same concentrations of TEA were applied to the intact labyrinth while recording from single afferent fibers of the posterior canal, at rest and during mechanical stimulation. At the peak of stimulation, TEA produced increases in mEPSP rate that were linearly related to the shifts produced by the same TEA concentrations (0.1–3 mM) in hair cell V_m (0.7–5 mV), with a slope of 29.8 Hz/mV. The membrane potential of the hair cell is not linearly related to receptor conductance, so that the slope of quantal release vs. receptor conductance depends on the prevailing V_m (19.8 Hz/nS at −20 mV; 11 Hz/nS at −10 mV). Changes in mEPSP peak size were negligible at rest as well as during stimulation. Since ample spatial summation of mEPSPs occurs at the afferent terminal and threshold-governed spike firing is intrinsically nonlinear, the observed increases in mEPSP frequency, though not very large, may suffice to trigger afferent spike discharge.

K+ pump: from caterpillar midgut to human cochlea

Deafness is a serious condition that affects millions of people and can also lead to dementia. Moreover, Karet and associates reported in 1999 that mutations in the gene encoding H(+) V-ATPase subunit B(1) lead to deafness. Yet ionic flows that enable humans to hear high-pitched sounds at 20,000 cycles/sec (20 kHz) are not well understood. Sound is transduced to electrical signals by stereocilia of hair cells by influx of Ca(2+) and K(+) as the "transducer channel" opens transiently and reduces the ∼90 mV (endolymph positive) endocochlear potential (EP) by ∼20 mV as the receptor potential. The EP as well as concentrations of Ca(2+), H(+) and K(+) must remain constant to produce reliable signals. Ca(2+) entry is balanced by Ca(2+) exit via a plasma membrane Ca(2+) ATPase (PMCA2a) but the Ca(2+) exit is coupled to H(+) entry. Moreover, K(+) entry is balanced by K(+) exit via a long diffusion route through several channels which is too slow to account for 20 kHz signaling. The problem is solved by a new hypothesis in which an H(+) V-ATPase generates the EP and removes the H(+) while a new K(+)/H(+) antiporter uses the voltage to drive H(+) back in and the K(+) back out. In the new model, Ca(2+), H(+) and K(+) cycle between unstirred layers on the endolymph- and cytoplasmic- borders of the stereocilial membrane through distances of ∼20 nanometers with travel time of ∼10 μs, which is fast enough to account for the 50 μs open/close time for 20 kHz signaling. Central to this model is the hypothesis that a K(+) pump which secretes K(+) into a K(+)-rich compartment is composed of a voltage producing (electrogenic) H(+) V-ATPase that is electrically coupled to a voltage-driven (electrophoretic) K(+)/nH(+) antiporter (KHA). Conversely, for an H(+) V-ATPase to secrete K(+) into a K(+) rich compartment, it must be coupled to a KHA. Richard Keynes reviewed evidence in 1969 that such a K(+) pump, which he called a Type V pump, is present in the stria vascularis of cochlea and the goblet cell apical membrane of caterpillars. Its signature is a large outside positive potential of ∼100 mV, K(+) secretion into a K(+) rich compartment and reversible inhibition by anoxia. The key role of the Type V K(+) pump in generating the EP was recognized by Sellick and Bock in 1974 and others but has disappeared from the hearing literature during the past decades. Its revival here is based on immunolocalization of KHA2 in the stereocilial membrane and Gillespie's generously shared mass spectroscopy evidence that all but one of the V(1) ATPase subunits are detected in isolated chicken stereocilia but V(o) and KHAs are not detected (implying that KHAs must be in the membrane). The new model proposed in the present paper could lead to important changes in our understanding of sensory physiology.

The receptor potential in type I and type II vestibular system hair cells: a model analysis

Several studies have shown that type I hair cells present a large outward rectifying potassium current (g(K,L)) that is substantially activated at the resting potential, greatly reducing cell input resistance and voltage gain. In fact, mechanoelectrical transducer currents seem not to be large enough to depolarize type I hair cells to produce neurotransmitter release. Also, the strongly nonlinear transducer currents and the limited voltage oscillations found in some hair cells did not account for the bidirectionality of response in hair cell systems. We developed a model based in the analysis of nonlinear Goldman-Hodgkin-Katz equations to calculate the hair cell receptor potential and ionic movements produced by transducer current activation. Type I hair cells displaying the large g(K,L) current were found to produce small receptor potentials (3-13.8 mV) in response to mechanoelectrical transducer current input. In contrast, type II cells that lack g(K,L) produced receptor potentials of about 30 mV. Properties of basolateral ionic conductances in type II hair cells will linearize hair bundle displacement to receptor potential relationship. The voltage to obtain the half maximal activation of g(K,L) significantly affects the resting membrane potential, the amplitude, and the linearity of the receptor potential. Electrodiffusion equations were also used to analyze ionic changes in the intercellular space between type I hair cell and calyx endings. Significant K(+) accumulation could take place at the intercellular space depending on calyx structure.

Ca selectivity of the transduction channels in the hair cells of the frog sacculus

The extracellular receptor currents evoked by step displacements of the otolithic membrane of the isolated saccular macula of Rana esculenta were recorded under transepithelial voltage clamp conditions. With the aim to depolarize the hair cells and increase the fractional resistance of the apical membranes, the basal side of the preparation was bathed in saline with an increased K+ concentration (62 mM). This caused a shift in the non-linear receptor current-voltage relation along the voltage axis of -51 mV +/- 10 mV; (mean +/- SD; n = 32) and a reduction in the transepithelial resistance of 10%. Under these conditions the electrical properties of the macula are assumed to be controlled by the apical membranes. The effects of different concentrations of Ca2+ in the apical solution on the receptor current-voltage relation were examined. Change of the apical Ca2+ concentration (range 3 mM to 70 microM) varied the transepithelial voltage at which the receptor current was zero (Vrev). Fitting a modified constant field equation to the relation between the apical Ca2+ concentration and the change in Vrev gave an estimate of PCa/PK of the transduction channels of 212. Furthermore, a high relative permeability of the transduction channels for other divalent cations (Ba2+, Sr2+) was measured, whereas Mn2+ inhibited the receptor current. The receptor current was inhibited by amiloride (IC50 3.2 microM +/- 1.7 microM) and nifedipine (IC50 1.9 microM +/- 0.6 microM). Reduction of the apical Ca2+ concentration to 90 microM in standard apical solution reduced the size of the receptor current to 67% +/- 30% (n = 17) compared to control but did not affect the shape of the receptor current-voltage relation. Subsequent substitution of K+ by Na+ caused a further reduction of the receptor current to 32% +/- 29% (n = 9), changed the receptor current-voltage relation into a linear relation and diminished the adaptation of the receptor current. These results indicate that the mechano-electrical transduction channels of the frog saccular hair cells are highly selective to Ca2+ and that the conductance of the channels may be influenced by the apical monovalent cation species.

Influence of ATP and ATP agonists on the physiology of the isolated semicircular canal of the frog (Rana pipiens)

In the present study, the influence of extracellular ATP and ATP agonists in the physiology of the vestibular organs was examined, using the in vitro model of the isolated semicircular canal of the frog (Rana pipiens). The firing activity of the afferent nerve, the d.c. nerve potential and the transepithelial potential were measured in the absence and presence of mechanical stimulation of the sensory epithelium. Administration of ATP into the perilymphatic compartment, from 10(-12) to 10(-3) M, increased the firing rate of the afferent fibers recorded in the absence of mechanical stimulation. Recordings of the d.c. nerve potential indicated that the afferent fibers were hyperpolarized. The presence of the purine also modified the transepithelial potential. During mechanical stimulation of the sensory epithelium, both the evoked afferent firing and the evoked variation of the d.c. nerve potential were reduced in the presence of ATP. However, ATP did not effect the evoked modulation of the transepithelial potential, evoked by the mechanical stimulation. Administration of the P2x purinoceptor agonists, alpha, beta-methylene-ATP and beta, gamma-methylene-ATP, at concentrations between 10(-12) and 10(-3) M, did not significantly modify the different bioelectrical activities investigated. In contrast, 2-methylthio-ATP, a P2y purinoceptor agonist, more potent and efficacious than ATP in its effect on the spontaneous firing. Concurrently, no modification of the d.c. nerve potential, the transepithelial potential and their variation during mechanical stimulation was observed. In opposition to the ATP effect, the total amplitude of the evoked firing was increased in the presence of 2-methylthio-ATP. These data suggest that extracellular ATP, present in the perilymphatic compartment, may act as a neuromodulator in the vestibular physiology. The effects of the purine appear to be mediated by the activation of a P2y subtype of purinoceptor. The absence of an effect of ATP and 2-methylthio-ATP on the evoked variation of the transepithelial potential suggest that the purine did not affect the processes responsible for the generation of the receptor potential but more likely modified the mechanisms involved in the release of the neurotransmitter from the hair cells and/or acted on the afferent endings.

Ionic selectivity of the mechano-electrical transduction channels in the hair cells of the frog sacculus

In the isolated saccular macula of Rana esculenta extracellular hair cell receptor currents evoked by mechanical stimulation of the otolithic membrane were recorded under transepithelial voltage clamp conditions. The ionic selectivity of the mechano-electrical transduction channels of the hair cells was determined by examining the effects of different concentrations of Ca2+ and K+ in the apical solution on the transepithelial voltage at which the extracellular receptor current was zero (Vrev). Changing the concentration of Ca2+ from 0.26 mM to 0.026 and to 2.6 mM at a constant K+ concentration caused changes in Vrev of -15 +/- 7 mV (mean +/- SD; n = 9) and 20 +/- 6 mV (n = 13), respectively. The relative ionic permeabilities of the transduction channels were estimated from a modified Goldman, Hodgkin and Katz equation, assuming that 80% of the transepithelial resistance is located in the apical membranes of the hair cells. The permeability of the transduction channels for Ca2+ was found to be two orders of magnitude larger than that for K+. The measured effects on Vrev of changing the concentration of K+ at constant ionic strength and at different constant Ca2+ concentrations were well predicted by the same equation. These results indicate that the transduction channels of the frog saccular hair cells are highly selective to Ca2+.

Evidence against a hypothesis of vestibular efferent function

Efferent stimulation and nicotinic agonists can either decrease or increase the frequency of occurrence of EPSPs recorded from VIIIth nerve afferents in the frog. It has been hypothesized that the distribution of hair cell resting membrane potentials overlaps the equilibrium potential dictated by the nicotinic-gated channels on the hair cells. Nicotinic mediated increases in EPSP frequency would then be due to depolarization of hair cells that were more hyperpolarized at rest, while decreases in EPSP frequency would be due to hyperpolarization of hair cells more depolarized at rest. In order to test this hypothesis, while recording from afferents which showed an increase in EPSP frequency due to bath application of the nicotinic agonist DMPP (1,1-dimethyl-4-phenylpiperizinium iodide), hair cells were depolarized with 10 mM K+ in the bath, and then the effects of DMPP on EPSP frequency were assessed. In this situation, DMPP still increased EPSP frequency, suggesting that the equilibrium potential for the nicotinic-gated channel was much more positive than the resting potentials of the hair cells. An alternative hypothesis then seems likely, that the nicotinic receptors on hair cells are able to activate different iontophores that result in either hair cell depolarization or hyperpolarization, dependent upon which iontophore predominates in the hair cells innervating a particular afferent.

Effects of modifications of extracellular and intracellular calcium concentrations on the bioelectrical activity of the isolated frog semicircular canal

In the inner ear, calcium appears to play a major role in different processes including mechanoelectrical transduction, adaptation during prolonged stimulation and electrical resonance. The aim of the present study was to investigate the effect of an increase of the cytosolic calcium content and a reduction of the perilymphatic concentration of calcium, on the bioelectrical activity of the isolated frog semicircular canal. Under resting conditions, the spontaneous activity of the afferent fibers and the difference of potential between the endolymphatic and perilymphatic compartments, called endolymphatic potential, were recorded. When the sensory epithelium was mechanically stimulated three additional parameters were investigated: the variations of the endolymphatic potential (ampullar direct current), the variations of the ampullar nerve potential (nerve direct current) and the frequency of the evoked afferent spikes. Increase of the intracellular calcium concentration by administration of the calcium ionophore A23187 (3 x 10(-6) M, 20 min) into the perilymphatic compartment, caused a biphasic effect on the spontaneous activity of the ampullar nerve which increased rapidly, reaching a maximum within 15 min, and then gradually declined to stabilize at 74% of the control 1 h after withdrawal of A23187. A23187 did not induce any modifications of the endolymphatic potential, the ampullar direct current or the frequency of the evoked afferent spikes. In contrast, A23187 induced a significant reduction of the nerve direct current which decreased by 31% of the control 1 h after withdrawal of the ionophore. Gradual reduction of the perilymphatic concentration of calcium (from 2 to 1 mM) induced a dose-dependent increase of the spontaneous activity of the ampullar nerve and the frequency of the evoked afferent spikes. Reduction of the perilymphatic calcium concentration from 1.6 to 1.2 mM caused a transient increase of the endolymphatic potential, while 1 mM Ca2+ induced a decrease to 88% of the control. The nerve direct current slightly increased for calcium concentrations ranging from 1.8 to 1.4 mM and decreased in the presence of 1.2 mM CaCl2. These data suggest that an increase of calcium into the cytosol induces an alteration of the mechanisms responsible for the spontaneous release of the afferent neurotransmitter and the electrogenic spreading of the postsynaptic potentials. In contrast, an excess of calcium does not impair the mechanisms involved in the generation of the action potentials. Our results also suggest that reduction of the perilymphatic calcium concentration may lead to modifications of the physical and electrical properties of the cell membranes of the labyrinthine epithelium and/or the ampullar afferent fibers.

Effect of lymph composition on an in vitro preparation of the alligator lizard cochlea

The effects of different artificial lymphs on the cochlear duct of the alligator lizard were studied in an in vitro preparation. The duct was dissected and cemented to the glass floor of a chamber that had been filled with an artificial lymph. The vestibular membrane was removed and latex beads (1-5 microns in diameter) were allowed to settle on the endolymphatic surface of the duct. During perfusion with an artificial lymph solution, the positions of beads were measured and video images of the duct were obtained. Artificial lymphs were isosmotic and included artificial endolymph (AE), artificial perilymph (AP), Leibovitz's L-15 culture medium, an AE solution whose calcium concentration was the same as that of AP, and AE and AP solutions in which gluconate was substituted for chloride ions. Results obtained in AE were consistently different from those in other lymphs. The displacements of beads, the projected area of the papilla, the occurrence of blebs, and direct observation of cells in the duct all indicated that the tissue swelled in AE (with or without 2 mmol/l Ca) but showed no consistent shrinking or swelling in any of the other artificial lymphs. Thus for the solutions we used, the presence of both potassium and chloride was required to elicit the swelling response to isosmotic artificial lymphs. There were some regional differences in the swelling response: the swelling of the endolymphatic surface of the tissue in a direction orthogonal to the basilar membrane surface was smaller on the free-standing region of the basilar papilla than either on the tectorial membrane or on the hyaline epithelial cells. The preparation was osmotically stable in AP and in both AE and AP solutions in which gluconate was substituted for chloride ions. After exposure to these solutions for as much as 300 min, the preparation showed no gross signs of deterioration visible with the light microscope, and continued to exhibit a highly specific osmotic response to the composition of the bathing medium.

Is GABA an afferent transmitter in the vestibular system?

This study was undertaken to determine the possible role of GABA as an afferent transmitter in the vestibular system of the axolotl. We studied the effects of GABA, muscimol, bicuculline and picrotoxin on the spontaneous spike discharge of the afferent fibers of the sacculi lagena and anterior semicircular canal. It was found that GABA and muscimol produce a very weak excitatory effect which does not mimic either the temporal course or the amplitude of the response of vestibular afferents to physiological stimuli. The GABA antagonist bicuculline has no significant effect on these fibers, and picrotoxin partially blocks the spontaneous activity in 33% of the fibers studied. These results indicate that GABA is probably not an afferent transmitter in the vestibular system as has previously been proposed.

Differential effect of ischemia on spontaneous and sinusoidal-evoked activity in semicircular canal afferents in the bullfrog

Spontaneous and sinusoidal-evoked nerve activity in semicircular canal afferent fibers of the bullfrog was evaluated prior to and following the production of ischemia of the labyrinthine arterial supply by mechanical occlusion of the vestibular artery. Neuronal spontaneous firing rates were observed to diminish by up to 100% within 10 min following the onset of ischemia. In most neurons there was a substantial increase in firing rate during the first few minutes. The sensitivity of the fibers to natural stimulation as determined by the gain in their responses to sinusoidal motion also diminished by as much as 75% over the same period. No detectable changes in the membrane potentials of the neurons were observed. The changes in excitability were closely correlated with the changes in spontaneous firing rate, but not all the neurons whose responses changed showed changes in spontaneous activity. Likewise, the relative magnitude of change varied from neuron to neuron.

Presynaptic actions of cholinergic agents upon the hair cell-afferent fiber synapse in the vestibular labyrinth of the frog

Spontaneous activity of semicircular canal afferents in the isolated labyrinth of the frog is altered by bath application of cholinergic agonists. Muscarinic agonists can produce an increase in action potential frequency of individual afferents. This increase develops slowly and is prolonged in the time course of its action. Nicotinic agonists can either increase (most cases) or decrease afferent activity. These effects occur rapidly and decay during the period of activation, suggesting desensitization. Muscarinic effects are blocked by prior administration of atropine and nicotinic effects (both increases and decreases in action potential frequency) by curare. Intracellular recordings reveal that the nicotinic effects on afferent action potential frequency are the result of alterations in the frequency of spontaneous synaptic potentials, indicating a presynaptic site of action on the hair cells for these compounds. This conclusion is supported by the fact that in the presence of high Mg2+/low Ca2+, which blocks hair cell release of transmitter, cholinergic agonists do not affect the resting membrane potential of the vestibular afferent. Electrical stimulation of the VIIIth cranial nerve can result in either an increase or a decrease in spontaneous synaptic potential and action potential frequency of an afferent. These effects are blocked by prior administration of curare or of nicotinic agonists. Repetitive or continuous stimulation of the VIIIth nerve results in a reversible reduction of the evoked response, suggesting desensitization. Transection of the VIIIth cranial nerve two weeks prior to recording eliminates these actions of electrical stimulation, but not the responses to cholinergic agonists, indicating that the effects of electrical stimulation are mediated by centrally arising efferents. These findings confirm that acetylcholine is probably the transmitter released from centrally arising vestibular efferents, and, in addition, demonstrate that efferent-mediated effects are predominantly expressed through nicotinic receptors. Studies comparing the effects of isolation of the semicircular canal alone versus the intact labyrinth suggest that the method of isolation may be an important factor in determining whether efferent activity results in a predominant increase or decrease in afferent activity.

Is the Na+,K+-ATPase symmetrically distributed in the neuroepithelium of the vestibular system in the axolotl (Ambyostoma mexicanum)?

This study was undertaken to assess the localization of the Na+,K+-ATPase in the neuroepithelial cells of the macula sacculi. In vitro perilymphatic (basolateral) perfusion with ouabain produced a significant drop in the membrane potential. Endolymphatic (apical) application of ouabain had practically no effect on membrane potentials. This suggests that Na+,K+-ATPase is asymmetrically distributed in the neuroepithelial cells.

Quantitative localization of Na-K pump sites in the frog sacculus

The Na-K pump site distribution within the macula, perimacula, and wall epithelia of the sacculus in the frog inner ear was examined with quantitative [3H]ouabain autoradiography. Excised tissue was incubated for 10-30 min (23 degrees C) in micromolar concentrations of high specific activity [3H]ouabain (14-70 muCi ml-1, 5-15 Ci mmol-1), washed for 30 min (4 degrees C), then rapidly frozen (-175 degrees C) and processed for light and electron microscope autoradiography. Control experiments based on (1) high K+ (50 mM) in the incubation and (2) low specific activity [3H]ouabain (1 mM, 0.013-0.025 Ci mmol-1) indicated negligible nonspecific binding of the [3H]ouabain. Measurable levels of specific [3H]ouabain binding occurred in all saccular regions examined. Binding was localized to the basolateral cell membranes with no detectable binding to the apical membranes. [3H]ouabain binding across the apical-basal axis of the saccule macular epithelium was nonuniform. Binding was low in the apical region, rose to a peak in the middle two-thirds, and then fell again close to the basement membrane. Electron microscope autoradiography suggested that this peak was due to ouabain binding to nerve terminals. Denervation of the sacculus eliminated the peak in [3H]ouabain binding and quantitative grain density analysis revealed that 45% of the Na-K pumps within the saccule macula were located on the nerve terminals. Na-K pump site density per unit volume was estimated by quantitative grain density analysis and the following values were obtained (sites micron-3 X 10(3), means +/- S.E.M.): saccule macula, 1.9 +/- 0.2; saccule perimacula, 1.1 +/- 0.1; saccule wall, 2.3 +/- 0.3. Stereological analysis of conventionally fixed tissue was used to estimate overall plasma membrane surface area per unit volume (Sv). Na-K pump site densities per unit membrane area for the various regions were calculated by combining the autoradiographical and stereological data. The following values were obtained (sites micron-2 +/- 25%): saccule macula, 2500; saccule perimacula, 2500. Values for individual cells within the macula (sites micron-2 +/- 25%) were: hair cells, 3000; nerve terminals, 3000; supporting cells, 1500.

Electrochemical profile for potassium ions across the cochlear hair cell membranes of normal and noise-exposed guinea pigs

The electrochemical driving force for movement of potassium ions across the hair cell membranes was determined in normal and noise-exposed guinea pigs. The measurement of the electrical potential and the potentiometric determination of K+ activity difference across the cell membranes were accomplished with double-barreled K+-selective liquid membrane microelectrodes. Identification of hair cells was based on the sudden increase of the a.c. component of the receptor potential associated with the appearance of the membrane potential and an increase in K+ activity. The results suggest that K+ ions in the hair cell interior and the extracellular space of the organ of Corti are near electrochemical equilibrium. However, the electrochemical gradient for K+ between the hair cell interior and the subtectorial endolymph was very high. These findings imply that the resting potential of hair cells is mainly generated by the diffusion of K+ across the basolateral hair cell membrane and is not affected by contact of the apical cell membrane with K+-rich endolymph. Although cochlear microphonics recorded extracellularly were severely suppressed in guinea pigs exposed to broadband noise at 115 dBA for 7 days, the electrochemical profile for K+ across cell membranes of surviving hair cells did not show marked changes. The ratio of intracellular a.c. receptor potential to extracellular cochlear microphonics was much greater in surviving hair cells of noise-exposed guinea pigs.