INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The vast majority of primary auditory neurons in the mammalian cochlea are excited by the release of neurotransmitter from the inner hair cells (IHCs) (Liberman 1982; Liberman and Oliver 1984; Robertson 1984; Spoendlin 1972). Recordings of the gross sound-evoked potentials from the guinea pig cochlea during perfusion with pharmacological blockers of voltage-gated Ca²⁺ channels have provided evidence that the release of excitatory neurotransmitter from IHCs during acoustic stimulation may be under the control of dihydropyridine-sensitive (L-type) channels (Bobbin et al. 1990; Zhang et al. 1999). Recent experiments in knockout mice lacking the 1-D variant of L-type channels support this notion, showing a concomitant loss of cochlear auditory sensitivity and L-type calcium currents in single IHCs (Platzer et al. 2000). In addition, capacitance measurements from isolated IHCs have provided further evidence to support the notion that the exocytosis of transmitter is at least partially under the control of L-type Ca²⁺ channels (Moser and Beutner 2000).

These previous studies raise a number of questions. Gross sound-evoked neural responses give no information about the spontaneous firing rates of auditory afferents. Hence it is not known whether the background transmitter release from IHCs is also controlled by L-type Ca²⁺ channels. This is an important issue because abnormal levels of spontaneous transmitter release may relate to pathological conditions, such as tinnitus and auditory neuropathy. Similarly, gross potential measurements cannot inform us as to whether the subpopulations of primary afferents that are known to emanate from each IHC (Liberman 1982; Liberman and Oliver 1984; Robertson 1984; Winter et al. 1990) are all controlled by the same Ca²⁺ channel subtype. There is evidence in other hair cell types that N-type Ca²⁺ channels may co-exist with L-type (Su et al. 1995), and hence there is the possibility that the diverse output properties of individual synapses may in some way be related to a corresponding diversity of their presynaptic Ca²⁺ channels.

In this study, we attempt to address these questions by combining microelectrode recording from single primary auditory neurons with intracochlear perfusion with solutions containing various antagonists and agonists of voltage-gated Ca²⁺ channels. The results show clearly that both spontaneous and sound-evoked firing rates of all subpopulations of primary auditory afferents are modified by the pharmacological manipulation of L-type but not N-type channels.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiments were performed on 22 young pigmented guinea pigs (340-440 g) of either sex. All procedures conformed to the Code of Practice of the National Health and Medical Research Council of Australia and were approved by the Animal Experimentation Ethics Committee of The University of Western Australia. Animals were anesthetized according to previously published procedures (see for example Lowe and Robertson 1995). For monitoring cochlear condition during initial surgery, a fine silver wire, insulated except at the tip, was placed on the bony shelf near the round window, and the visual detection threshold of the gross compound action potential response of the auditory nerve fibers (CAP) to repeated brief tone bursts, 10-ms duration, 1-ms rise-fall times and repetition rate of 5/s (Johnstone et al. 1979) was assessed using an on-line averaging program. The visual detection threshold corresponded to a CAP amplitude of approximately 3 µV.

For single-neuron recording, the scala tympani of the first cochlear turn was opened surgically as previously described (Robertson 1984; Robertson et al. 1980), and glass micropipette electrodes filled with 2 M potassium acetate (d.c. resistances ranging from 60 to 100 M) were inserted either into the auditory nerve within the modiolar core (method described by Alder and Johnstone 1978) or the spiral ganglion (Robertson et al. 1980) (Fig. 1).

View larger version (27K): [in this window] [in a new window]   Fig. 1. Sketch showing the recording and perfusion arrangement.

Measurements of frequency-versus-threshold curves (ftc's) were made for each single neuron encountered using custom-written software that automatically controlled the delivered sound frequency and intensity and estimated the single-neuron threshold from statistical measures of action potential firing rate (see Winter et al. 1990 for more detailed description of the threshold algorithm used). From the ftc for each fiber, its most sensitive, or characteristic frequency (CF) was estimated (±0.5 kHz accuracy). When single-neuron recordings were obtained using the modiolar approach, recordings were confined to fibers with CFs in the range of 12-20 kHz, corresponding to the portion of the basal cochlear turn in the immediate vicinity of the perfusion pipette. In the case of the spiral ganglion approach, recordings were made from locations in the ganglion directly radial to the placement of the perfusion pipette (Robertson and Johnstone 1979). These latter fibers also ranged in CF from 12 to 20 kHz.

Because the scala tympani was widely opened in these experiments, no outlet hole was made in the cochlear apex, and the perfusates were simply flushed within the confines of the basal turn. Overflow of perfusate from the scala tympani hole accumulated in the bottom of the tympanic bulla and was removed continuously with fine tissue wicks inserted into the middle ear cavity. Ftc's were measured before, during, and after perfusion. Spontaneous firing rates were estimated with the same software from 10 s samples taken at regular intervals throughout the period of recording from each fiber.

Artificial perilymph was identical in composition to that employed by previous workers (Bobbin et al. 1990; Zhang et al. 1999). It contained (in mM) 137 NaCl, 5 KCl, 2 CaCl₂, 1 MgCl₂, 1 NaH₂PO₄, 12 NaHCO₃, and 10 glucose. The pH was checked immediately prior to all perfusions and was adjusted to 7.4 at 37°C as required. Nimodipine (Alomone Laboratories) was dissolved in DMSO and made up in the artificial perilymph to a final concentration ranging from 3 to 10 µM with a corresponding final DMSO concentration ranging from 0.03 to 0.1%. Conotoxin GVIA (Alomone Laboratories) was made as a stock as instructed by the manufacturer and diluted for each experiment with artificial perilymph to give a final concentration of 300 nM. S() BAY K 8644 (Sigma) was dissolved in artificial perilymph to a final concentration of 20 µM. Control perfusions generally consisted of artificial perilymph, except for the nimodipine experiments when the control perfusion consisted of artificial perilymph with the addition of 0.1% DMSO. Perfusion rates were 2.5 µl/min in all cases.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 80 intracochlear perfusions were carried out while recording the activity of 68 separate auditory afferent neurons. Of the total neuron sample, 54 had spontaneous firing rates before perfusion that were >10 spikes/s (range 10.4-135.3) and 17 had spontaneous firing rates below 10 spikes/s (range 0-10). Of the 80 perfusions, 22 were with control solutions, 39 were with solutions containing nimodipine at concentrations ranging from 3 to 10 µM, 13 were with conotoxin GVIA at a concentration of 300 nM, and 6 were with BAY K8644 at a concentration of 20 µM.

Figure 2 shows two examples of control perfusions. One example (Fig. 2, A and B) is a perfusion with artificial perilymph and the other (Fig. 2, C and D) with artificial perilymph containing DMSO at a concentration of 0.1% (the control perfusion for nimodipine experiments). It can be seen that there are only slight and transient effects on the neurons' spontaneous firing, and there is no detectable change in the ftc's. In the entire sample of 22 control perfusions, there were sometimes small changes in spontaneous firing rates, either increases or decreases, but these were not systematic and they were transient, mainly occurring during or in the first minute after the perfusion. When the data were averaged across all neurons, the mean maximum change in spontaneous discharge rate measured after all control perfusions was a slight increase to 105 ± 26.8% (mean ± SD) of preperfusion values. There was no significant difference between perfusions with artificial perilymph without or containing 0.1% DMSO nor were there any significant effects of any of the control perfusions on threshold sensitivity at any part of the neurons' ftc's.

View larger version (18K): [in this window] [in a new window]   Fig. 2. Absence of effect of control perfusions on single primary auditory neurons. A and C: spontaneous rate. B and D: corresponding frequency-vs.-threshold curves (ftc's). Solid bars in A and D represent period of perfusion. Symbols and arrows in A and C indicate times at which ftc's in B and D were obtained. Data in A and B were from control perfusion with artificial perilymph. Data in C and D were from control perfusion with artificial perilymph containing 0.1% DMSO.

Figures 3 and 4 show typical results obtained with nimodipine perfusions. In dramatic contrast to the results shown in the preceding text for control perfusions, a marked depression of spontaneous firing rates and an approximately uniform elevation of acoustic thresholds across the extent of the ftc were consistently observed. Figures 3 and 4, A and C, illustrate the spontaneous rate finding for six neurons with different initial spontaneous firing rates. In several cases, spontaneous firing was completely suppressed by nimodipine perfusion. There was no significant difference (P > 0.05) in the mean effects on spontaneous rate among 3, 5, and 10 µM concentrations of nimodipine tested (single-factor ANOVA with Tukey's B post hoc comparisons). When the data from all nimodipine perfusions and all neurons were pooled, the mean maximum change in spontaneous firing rate was a reduction to 13.2 ± 14.7% (n = 39) of preperfusion values.

View larger version (16K): [in this window] [in a new window]   Fig. 3. Four examples of effect of 5 µM nimodipine on spontaneous firing rates of single auditory afferents. Note depression of rate in all cases, despite wide variation in initial spontaneous firing rates. , timing of perfusion.

View larger version (23K): [in this window] [in a new window]   Fig. 4. Illustration of effects of nimodipine perfusion (10 µM in both cases) on spontaneous firing rates and ftc's in 2 neurons with different initial spontaneous firing rates. Layout of figure and symbols as in Fig. 2. Note marked elevation of both "tip" and "tail" thresholds of ftc's at same time as depressed spontaneous firing rate. In B and D, the threshold data points at the highest intensities are either equal to or greater than the maximum sound output of the delivery system.

We obtained results from nimodipine perfusions for 8 low spontaneous rate (<10 spikes/s) and 27 high spontaneous rate (>10 spikes/s) neurons, indicating that nimodipine affected them roughly equally. For the low spontaneous rate group, mean spontaneous rates fell to 6.7% of the preperfusion value after nimodipine perfusion and for high spontaneous rate neurons, the mean change was a drop to 14.8% of preperfusion values. A single-factor ANOVA indicated that any apparent difference between these two groups in the effect of nimodipine on spontaneous firing rates had a high probability of being due to sampling error alone (P = 0.166).

In many cases, recovery of spontaneous firing rates from the nimodipine effects was not able to be demonstrated. This was because mechanical limitations to the perfusion arrangement did not readily permit flushing with a second control perilymph while maintaining contact with a single neuron, and the natural flow of perilymph had to be relied on to wash out the nimodipine. Recovery was therefore usually very slow and great stability of the recording was required to remain in contact with the same neuron throughout this procedure. The slow nature of recovery from nimodipine perfusion in these experiments also meant that the yield of single neurons per animal was low. Despite these difficulties however, we demonstrated near complete recovery of spontaneous firing rates in 11 neurons and partial recovery (contact was lost with the neurons prematurely, before full recovery could be demonstrated), in eight cases.

The magnitude of the effects of nimodipine perfusion on the ftc's varied considerably from one experiment to another, but it is clear that in broad terms, decreases in sensitivity to sound accompanied the spontaneous firing rate depression (Fig. 4, B and D). In some cases in which spontaneous firing was severely depressed, the acoustic thresholds were so elevated that no response could be obtained at the highest sound intensities that could be delivered. In other cases, the elevation of thresholds was more moderate, and this correlated with a failure of the nimodopine perfusion to fully depress spontaneous firing. In all cases however, thresholds were elevated both in the region of the sharply tuned "tip" of the ftc and on the low frequency "tail." However, there was generally a significantly greater elevation of thresholds in the vicinity of the sharply tuned tip than on the low frequency tails. Figure 5 shows comparison of tip and tail threshold changes after nimodipine perfusion in 12 neurons. Multiple measurements of the two thresholds were taken immediately after perfusion and during recovery. We observed substantial recovery of ftc's along with spontaneous firing rate recovery in five neurons and incomplete recovery in another seven before contact with the neurons was lost. In a number of cases of recovery, the low-frequency tail thresholds recovered before tip thresholds had returned to normal (see, for example, Fig. 4D).

View larger version (19K): [in this window] [in a new window]   Fig. 5. Pooled data comparing changes in tail and tip thresholds caused by nimodipine. , expected change if only final output of inner hair cell (IHC)-afferent synapse were affected. Repeated measures of tip and tail thresholds were obtained from 12 single neurons immediately after perfusion and during recovery. Tip thresholds measured at the initial characteristic frequency and tail thresholds measured 1 octave below the characteristic frequency. In cases where thresholds exceeded the maximum sound pressure able to be delivered, data points were omitted.

Figure 6 shows typical effects of perfusion with an agonist of L-type Ca²⁺ channels BAY K8644 (Sarmiento et al. 1987). A prolonged and marked elevation of spontaneous firing rates was observed in neurons with widely differing initial spontaneous firing rates. In total six neurons were investigated with spontaneous rates ranging from 2.5 to 140 spikes/s. The mean spontaneous rate after perfusion with BAY K was 395% of the preperfusion value. Ftc's were not investigated during BAY K perfusions.

View larger version (15K): [in this window] [in a new window]   Fig. 6. Examples from 3 different neurons showing the effect of perfusion with 20 µM BAY K8644 on spontaneous firing rate. , period of perfusion.

In contrast to the marked effects of the L-type Ca²⁺ channel antagonist and agonist described in the preceding text, perfusion of the cochlea with solutions containing 300 nM conotoxin GVIA had no significant effects on either spontaneous or driven rates of single primary afferents. Representative examples are shown in Fig. 7, A-D, in which two neurons are illustrated with substantially different spontaneous firing rates. Similar to control perfusions, no marked alterations were caused by this toxin in either spontaneous firing rate or acoustic thresholds over the whole range of the neurons' ftc's. In the pooled sample of neurons from all conotoxin perfusions, the mean maximum change in spontaneous firing rate was an increase to 104 ±13.3% of preperfusion values. There were no significant changes in ftc's. This lack of effect of conotoxin was found for neurons with a wide range of preperfusion spontaneous firing rates.

View larger version (21K): [in this window] [in a new window]   Fig. 7. Examples from 2 different neurons of the lack of effect of perfusion with 300 nM GVIA conotoxin on spontaneous firing rate and frequency vs. threshold curves. Organization of figure and symbols as in Fig. 2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study, by using single neuron recording from the intact mammalian cochlea, adds to our understanding of the role of L-type voltage-gated Ca²⁺ channels in controlling transmitter release from IHCs. Although a number of earlier studies have clearly implicated L-type channels in the release process (Moser and Beutner 2000; Zhang et al. 1999), a critical test of the L-type channel hypothesis requires that a direct effect of channel modulators such as nimodipine and Bay K on the functional neuronal output from the IHC-afferent synapse be clearly demonstrated, as in the present study. The results clearly show that both spontaneous firing rates and acoustically-driven responses of primary auditory neurons are depressed when the L-type blocker nimodipine, is infused into the scala tympani. The involvement of L-type channels in spontaneous release of neurotransmitter is further supported by the finding of increased spontaneous firing rates in the presence of a channel agonist. The fact that, in some cases, nimodipine can reduce spontaneous firing rates to zero and elevate acoustic thresholds to the point where neurons can be become essentially unresponsive strongly suggests that L-type channels are the major, if the not only, Ca²⁺ channels involved in directly regulating transmitter release in this system. Although N-type, as well as L-type channels have been reported in noncochlear hair cells (Su et al. 1995), the lack of effect of conotoxin GVIA in our experiments suggests either that N-type channels are not present or that they have no role in controlling the presynaptic transmitter release in mammalian IHCs.

The finding that the sensitivity to acoustic stimulation is elevated across the ftc, in particular, the fact that thresholds to low "tail" frequencies are affected as well as those in the vicinity of the sharply tuned "tip" of the ftc's, lends considerable strength to the argument that nimodipine acts at the IHC-afferent synapse. Alterations to the function of the outer hair cells would be expected to have a frequency-selective action near the tuning curve tip and little or no effect on the low-frequency tail (for review, see Patuzzi and Robertson 1988; Yates et al. 1992). There was some evidence for a mixed action of nimodipine on both IHCs and outer hair cells in that the tips of ftc's were more elevated than tails and took longer to recover (Figs. 4D and 5). This finding is in agreement with our previous results from gross recordings in which we obtained evidence for a "pure" IHC locus of action of nimodipine only when outer hair cells were removed by kanamycin administration (Zhang et al. 1999).

The present data also demonstrate that nerve fibers with widely differing spontaneous firing rates appear to all be vulnerable to the effects of the same pharmacological blocker, nimodipine. Unfortunately however, the present data suffer from limitations inherent in the experimental preparation that do not allow it to be determined quantitatively whether different neurons are equally susceptible to nimodipine. There was insufficient control over the efficiency of perfusion of the cochlea, and the length of time for which reliable recordings could be obtained from single fibers during perfusion, for us to say whether or not nimodipine can totally eliminate both spontaneous and driven activity in all spontaneous rate categories of primary afferents. Similar limitations may also explain some of the details of the effects of the Ca²⁺ channel agonist BAY K. Unknown diffusion barriers and lack of control over washout rates may contribute to the rather slow onset of increased firing and to the failure to demonstrate reversibility of the effect with this drug (Fig. 6).

If, as the present results suggest, L-type channels are the only ones that control transmitter release in mammalian IHCs, there are several mechanisms that could be invoked to explain the differences in functional properties between different afferents driven by the same IHC. There may be differences between synapses in the number, density, and location of the voltage-gated Ca²⁺ channels relative to the vesicular release sites and to stores and buffers. Thus each of the 20 or so synaptic active zones in each IHC could act as private Ca²⁺ microdomains with the local level of Ca²⁺ varying from synapse to synapse (see for example Issa and Hudspeth 1994; 1996; Roberts et al. 1990; Tucker and Fettiplace 1995).

Another explanation for the variation may lie in some postsynaptic property, such as the threshold of the action potential initiating site on each afferent, or even variations in the type of postsynaptic receptors for the transmitter released from the IHC. Although the dominant postsynaptic receptor appears to be of the AMPA type (see for example Ruel et al. 2000), there is some evidence for a variety of ionotropic and metabotropic postsynaptic receptors for glutamate and other putative transmitters on primary auditory dendrites (Kleinlogel et al. 1999; Salih et al. 1998). Yet another factor that it may be necessary to take into account is the influence of synapses arising from efferent pathways that terminate on afferent dendrites beneath the IHC (Warr and Guinan 1979; Warr et al. 1997). In support of this notion, Liberman (1990) has shown that elimination of all efferent innervation to the cochlea changes the distribution of spontaneous firing rates observed in auditory nerve fibers. Further experiments are needed to differentiate between these various possibilities.