Hearing Research, 56 (1991) 69-78 $3 1991 Elsevier Science Publishers B.V. Ail rights reserved 0378-5955/91/$03.50

69

HEARES 01629

Differential modulation of spontaneous and evoked neurotransmitter release from hair cells: some novel hypotheses

Paul S. Guth lr2v3, Anne Aubert ‘, Anthony J. Ricci ‘,3 and Charles H. Norris 1,2~3 Departments of ’ Orolaryngofogy, 2 Pharmacobgy and 3 The Neuroscience Program, Tulane C’nirersify, School of Medicine, New Orleans,

Louisiana, U.S.A.

(Received 26 December 1990; accepted 28 April 1991)

It has been generally accepted that even in the absence of mechanical stimulation of the transductional elements, a resting depolarizing current exists which is ultimately responsible for the spontaneous release of neurotransmitter. Movement of the transductional elements modutates this resting current and thereby the evoked release of neurotransmitter occurs. Recent data from our laboratory and others have led us to question whether the relationship between spontaneous and evoked neurotransmitter release is as simple as stated, Indeed, a variety of experimental manipulations appear to influence the two modes of release differently. Examination of our results and the results of others has led us to four hypotheses:

1. the two modes of neurotransmitter release are processed differently by the hair cells; 2. cyclic AMP is involved in spontaneous but not evoked neurotransmitter release; 3. there is a positive feedback step involving an excitatory amino acid and its receptor on the hair cell in evoked neurotransmitter release

and; 4. different pools of calcium are involved according to the mode of release. Accordingly, there may be several biochemical steps between the transductional movement of the stereocilia at the apex of the hair cells and

the ultimate release of the neurotransmitter at the base of these cells. Some of these biochemical steps are different depending on whether the mode of release is spontaneous or evoked. These biochemical steps may amplify or at least interact with the biophysical processes previously described in the hair cells.

Hair cell; Neurotransmitter; Octavolateralis; Semicircular canal; Glutamate: Calcium

Introduction

The production of action potentials in the afferent neu- rons depends completely on tr~~~itter release from hair cells

The neurosensory epithelium of the inner ear or- gans contains hair cells which synapse on their basolat- era1 (perilymphatic) sides with afferent and efferent neurons. In the absence of stimulation of the neu- rosensory epithelium the afferent fibers innervating the hair cells of the semicircular canal (XC), and most probably the other acousticolateralis organs, exhibit a spontaneous firing rate (spontaneous activity) which is believed to be entirely dependent on a ‘basal’ release of the neurotransmitter from the hair cells. The evi- dence supporting the assertion that the afferent neu- rons themselves are not spontaneously active but that their firing depends entirely on a basal release of neurotransmitter is as follows:

1. Manipulations designed to inhibit neurotransmit- ter release from hair cells such as replacement of

Correspondence tao: Paul S. Guth, Department of Pharmacology, Tulane University, School of Medicine, 1430 Tulane Avenue, New Orleans LA 70112, U.S.A.

normal perilymph with a solution containing high Mg2+ and low Ca2+ (SCC-Valli et al., 1985; Cochlea-Siegel and Relkin, 1987) or cobalt application (Lateral line- Sewell, 1990) cause the afferent fibers innervating these hair cells to become silent. However, the role of charge screening caused by these divalent cations in the si- lence of the afferents must be considered.

2. Application of excitatory amino acid antagonists (eg. kynurenate) (Annoni et al., 1984; Soto and Vega, 1988) or spider venom (Cousillas et al., 1988) can cause complete cessation of firing of the afferents.

3. In the absence of hair cells, apparently normal cochlear afferents exhibit no spontaneous activity (Durham et al., 1989).

4. Afferent fibers became silent when deprived of hair cell input by acute (Bernard, 1983) or chronic (Norris et al., 1990) aminoglycoside treatment.

5. The frequency of afferent firing in type I neurons goes to zero when the hair cells are transductionally hyperpolarized (Precht, 1976).

6. Passing D.C. currents across the neuroepithelium of the SCC (Fig. 2) in which endolymph is negative with regard to the perilymph can cause the afferents to become silent when the current amplitude is suffi- ciently high (Ricci et al., 1991). These currents are

70

thought to act only on the hair cells for several reasons. First, passage of hyperpolarizing currents affects spon- taneous and mechanically-evoked afferent firing differ- ently. If the current were acting on the afferents them- selves both spontaneous and mechanically-evoked transmission might be equally affected. Second, during the passage of hyperpolarizing current, the afferents may be caused to fire by the application of a solution containing 10 mM potassium.

7. When endolymphatic KC is replaced by Na+ and tetraphenylboron is added to the endolymph, while perilymph is kept unchanged, the afferents become silent (Valli et al., 1985).

8. o-tubocurarine in endolymph (not perilymph) can completely reduce spontaneous and evoked afferent neural activity (Valli et al., 1974) presumably by inter- fering with transduction.

9. Electrical stimulation of the efferents which end primarily, if not totally, on hair cells of the lateral line (Russell and Roberts, 19721 or the frog semicircular canal (Valli et al., 1986) can cause total suppression of spontaneous afferent firing. Therefore, the recording of the afferent neuronal activity provides a dependable means of monitoring the release of neurotransmitter from the hair cells.

Differential modulation of ‘spontaneous’ and ‘euoked’ ~fferent acti~i~

Classically, a resting current across the transduc- tional elements is believed to account (through the release of neurotransmitter) for the generation of the spontaneous afferent activity. Modulation of this cur- rent by the mechanical stimulation of the sensory ep- ithelium is responsible for the modulation of the firing rate which is called evoked activity. This current is mainly carried by the entry of K+ (Valli et al., 1979; Hudspeth, 1983) and is accompanied by the entry of Ca*+ (Ohmori, 1989) from endolymph. Thus move- ment of the stereocilia toward the kinociiium induces a depolarization of the hair cells (transductional depolar- ization) which is accompanied by an increase of the firing rate of the afferent neurons presumably by an increase of the release of the neurotransmitter from the hair cells. In contrast, the decrease in the firing rate recorded when the stereocilia are displaced in the opposite direction is caused by a hyperpolarization of the hair cells (transductional hype~olarization) and apparently a reduction of the hair cell neurotransmit- ter release.

These events involve the interactions of biophysical mechanisms such as entry of Kf through transduction channels and subsequent electrical modulation of volt- age-sensitive Ca2 + channels, calcium-dependent Kf channels and Kt channels. In this paper attention is drawn to the possible interactions of membranal and intracellular biochemical mechanisms with the above-

mentioned biophysical processes. This paper suggests that the modulation of the release of neurotransmitter under spontaneous and mechanically evoked condi- tions is achieved by different although complementary processes and that these processes are largely biochem- ical.

Differences between ‘Lypontaneous’ and ‘euoked’ afferent actiz!i~

A series of observations, by ourselves and others (Bernard, 1980; Drescher and Drescher, 1987; Starr and Sewell, 1990; Gleich et al. 19901, has led to the conclusion that there are differences between sponta- neous and mechanically evoked afferent activity as regards mechanisms responsible for the release of the hair cell neurotransmitter. These observations are that evoked and spontaneous neurotransmitter release can be differentially affected by a variety of chemical and physical manipulations.

The observations given in the Tables I and II illus- trate several general differences between spontaneous and evoked neurotransmitter release.

Note that evoked and spontaneous activities may be independently modulated to cause either an increase or a decrease in the release of the hair cell neurotrans- mitter. Note also that either evoked or spontaneous release of transmitter may be the more readily af- fected. Taken together the observations of Tables I and II strongly suggest that the spontaneous and evoked modes of neurotransmitter release are modulated by different mechanisms and that these mechanisms oper- ate presynaptically in the hair cells. Any change in the afferent dendrite would likely affect both spontaneous and evoked activity similarly. Of particular clarity is the demonstration that a D.C. current flowing across the neuroepithelium of the SCC can, depending on its intensity and duration, cause a complete abolition of evoked neuronal activity but only a partial inhibition of

TABLE I

Evoked neurotransmitter release is affected more than spontaneous.

A.

B.

C.

D.

Glutamate antagonists reduced evoked more than spontaneous firing. 1. Lateral line (Bledsoe et al., 1988) 2. Cochlea (Cousillas et al., 1988) 3. Saccule (Starr and Sewell, 1990) In the cochlea, perfusions of glutamate caused a reduction in tone-evoked activity without a change in spontaneous rate (Gleich et al., 1990). In the semicircular canal electrical D.C. currents across the neuroepithelium, in which the endolymph was negative relative to the perilymph, decreased evoked more than s~ntaneous afferent activity (Fig. 2). When of sufficient amplitude the current can completely inhibit all afferent activity. In the lateral line, evoked activity has been shown to be more affected than spontaneous with decreasing calcium and increas- ing magnesium concentrations (Drescher and Drescher, 1987).

TABLE II

Spontaneous neurotransmitter release is affected more than evoked.

A.

B.

C.

D.

E.

F.

G.

In the semicircular canal an increase of cyclic AMP (CAMP) produced an increase in spontaneous, but not evoked firing rates (Fig. 3). In the semicircular canal Ach application produced an increase in spontaneous, but not evoked firing rates (Fig. 4). In the semicircular canal application of a calcium ionophore caused an increase in spontaneous but not evoked activity. (Au- bert et al., 1991). In the semicircular canal 4-aminopyridine (4-AP) caused a de- crease in spontaneous which was much greater than the decrease in evoked activity (Fig. 51, (Ricci, et al., 1991). In the lateral line, salicylate decreased spontaneous more than evoked activity (Puel et al., 1989). 1. In the semicircular canal, (Fig. 6) and in cochlea (Siegel and Relkin, 1987) substituting low CaZC, high Mg*+ solutions for perilymph caused a marked reduction in spontaneous activity before the subsequent reduction in evoked activity occurred. 2. Likewise, diltiazem or streptomycin (thought to act also on Ca*+ channels) both caused spontaneous activity to fail before evoked activity (A. Ricci-Personal communication). 3. In the lateral line, spontaneous activity is affected more than evoked for 2-4 mM calcium (Drescher and Drescher, 1987). In the semicircular canal, D.C. currents across the neuroepithe- lium in which the endolymph is positive to the perilymph in- creased spontaneous activity much more than evoked (Ricci et al., 1991).

71

spontaneous activity (Fig. 2). In addition, the results of three different experimental manipulations, all of which affect spontaneous and evoked afferent activity differ- ently, have especially captured our attention: 1) the modulation of cyclic AMP levels; 2) the effect of gluta- mate and glutamate antagonists and 3) the modifica- tion of calcium homeostasis.

Methods

A schematic of the experimental preparation is given in Fig. 1. The posterior semicircular canal of the frog, Rana pipiens, was isolated from the labyrinth. This canal was then mounted in a two chambered bath in which the endolymphatic and perilymphatic spaces could be maintained independently. The endolym- phatic solution contained: 75 mM KCl, 50 mM NaCl, 2 mM CaCl,, 5 mM ‘NaHCO,, and 5 mM NaH,PO,. The artificial perilymph solution contained: 2.5 mM KCl, 105 mM NaCl, 2 mM CaCl,, 5 mM NaHCO,, 5 mM Na,HPO,, and 5 mM NaH,PO,. All solutions were bubbled to saturation with a gas mixture of 95% 0, and 5% CO,. The pH of each solution was ad- justed to 7.2 with NaOH (Norris et al., 1988, Valli et

Transepithelial Potential

Efldolymph

J

SuctionElectroCk .

Fig. 1. The two-compartment bath-the experimental arrangement. The isolated posterior, vertical semicircular canal (PA) is shown mounted in a bath of artificial perilymph (right side of partition). Artificial endolymph is placed in the left side of the bath and is in the lumen of the semicircular canal. Test substances may be applied by bath substitution (Rx) or by injection close to the neuroepithelium (not shown). A suction electrode is used to record the multiple unit firing rate of afferent fibers in the cut ampullar nerve. A piezo electric hydraulic drive (MS.) is used to move fluid, displacing the cupula thereby evoking neural activity. A transepithelial potential is recorded between electrodes in either

compartment of the bath. Ge = ground electrode, Cro = oscilloscope, pi= perilymph.

al., 1985): The low calcium-high magnesium solutions contained 0.1 mM calcium and 10 mM magnesium.

Mechanical stimulation of the hair cells was accom- plished with a piezoelectric bimorph coupled, via the plunger of a fluid filled syringe, to the endolymphatic space, so that a movement of the bimorph would elicit a movement of the endolymph (Norris et al., 1988). This movement deflected the cupula, moving the stere- ocilia, resulting in either a depolarization or hyperpo- larization of the hair cells depending on the direction of plunger displacement. The stimulus across the bi- morph was a 5 s ramp (of 0.5 ~1 maximal displacement) in the depolarizing direction followed by a rapid 2 second return to baseline.

The frequency of firing was calibrated using a Hewlett Packard function generator to generate the standard frequencies. The ‘zero’ level was set by re- moving all input to the window discriminator. During the experiment, the window was first set above the electronic noise level and then raised so that the affer- ent nerve firing never exceeded the upper timits of the calibration. This allowed for increases in mechanically- or chemically-stimulated firing to occur, over a range in which the equipment responded linearly. The gain of the mechanical stimulator was set at the beginning of each experiment so that a sub-maximal firing rate of the afferent neurons was achieved, thus allowing for either increases or decreases in firing to be recorded. All afferent firing data were normalized as percent of the control data. No experimental data were included in the analysis if the effects of the drugs applied were not reversible by washing to within 95% of control values. Measurements were made as follows: sponta- neous activity was averaged over one minute prior to any drug application. Mechanically evoked activity was measured as the increase in frequency of firing over spontaneous activity. Five cycles of stimulus were aver- aged once the activity had stabilized. Similar measure- ments were made during a CAMP-altering drug appli- cation once the response had equilibrated. Measure-

-MECHANICAL STIMULUS ,

-electrical stimulus- -lo -20 -30 -40

lUAl

Fig. 2. The preparation depicted in Fig. 1 was used to demonstrate the sensitivities of spontaneous and mechanically evoked afferent firing rates to different DC currents applied across the neuroepithe- hum. Ten &A current steps were applied to illustrate the graded inhibition of mechanically evoked firing rate. It can be seen that the maximal current used (40 PA) completely inhibits evoked activity but that spontaneous activity is much less affected. The frequency is

measured as multiple unit impulses per second.

ments were made again after each drug was washed out. All changes in firing were compared using a two- tailed Student&t test with significance chosen to be at the P = 0.05 level. The transepithelial potential was measured as the potential difference between the en- dolymphatic and perilymphatic solutions. The change in transpithelial potential induced by mechanical stim- ulation is a measure of the electrical state of the hair cell. It was measured differentially between silver/ silver chloride electrodes placed in the endolymph and perilymph respectively. The signal was taken across a Grass amplifier (1000 X 1 and displayed on an oscillo- scope and strip chart recorder. The two parameters of transepithelial potential observed were a change in baseline and a change in magnitude of the response induced by the mechanical stimulus. Only relative changes were recorded in the transepithelial potential.

In all experiments Ach was applied to the neuroep- ithelium of the semicircular canal as an injection of 10 ~1 of a 300 FM solution over 10-s period. A threaded syringe attached to a glass pipette was utilized for the administration of the bolus. Artificial perilymph was administered as the control injection.

AI1 other drugs were applied by bath substitution to the perilymph solution at the indicated concentrations. Forskolin was first dissolved in 95% ethanol and then diluted in artificial perilymph. Controls of ethanol in perilymph were carried out and no effect was noted in the parameters studied at the ethanol concentrations (0.5%) present in artificial perilymph at the final drug dilutions. 3-isobutyl-l-methylxanthine (IBMX) and Dibut~l CAMP were dissoived directly in perilymph. All chemicals were purchased from Sigma Chemical Company.

Results

D. C. electrical stimulation Fig. 2 depicts the differential effect of electrical

stimulation with D.C. currents as mentioned in Tabie IB. Such electrical stimulation completely inhibits the response of the afferent neuron to mechanical stimula- tion, but spontaneous activity remains, albeit at a re- duced level. This result not only demonstrates the primary point of this paper (i.e. differential modulation of spontaneous and evoked transmitter release) but suggests that such D.C. electrical stimulation acts pri- marily on the hair cells and not on the afferent neu- rons. If it acted primarily on the postsynaptic element, the afferent neuron, then both spontaneous and evoked activity might be equally affected or even that sponta- neous activity might be the more affected because there might be less transmitter release under the spon- taneous than under the evoked condition. It is worthy of note that as the current is increased, it is the

73

I -----Control- - _-, ~--------~o~sko~~~------~~---~~sh-___-____~ .FP

5

$E._ M.E.

3 5 400

2 200 m _

Fig. 3. Three drugs known to cause increases in CAMP were applied. This figure illustrates the effect of drugs expected to cause an increase in cyclic AMP concentrations. The responses to all three such drugs-forskolin, IBMX and dibutyryl cyclic AMP were all qualitatively similar. This preparation is as depicted in Fig. 1. The upper trace depicts the transepithelial potential measured across the neuroepithelial layers of the isolated semicircular canal and at least partially reflective of hair cell activity. The lower trace is of multiunit afferent firing rate recorded from the cut ampullar nerve. Spontaneous afferent firing is increased by applying CAMP-elevating drugs. However, the response to intermittent mechanical excitat;o~ (M.E.) (shown in the lower trace as tall reguiar ~uctuations in the afferent firing rate) and the response to a~etylcholine (20 ~1.1 of a 1 mM solution applied by injection close to the neuroepithelium indicated by Ach) are not modified. The frequency is in multiunit

afferent impulses per second. The dashed horizontal line in the lower trace indicates basal spontaneous activity.

depolarizing (upward) response to mechanical stimula- tion that is inhibited first. The h~erpo~a~zin~ (down- ward) response is maintained until the 40 PA current level is reached.

Only the response to forskolin, the stimulator of adenylyl cyclase is shown (Fig. 3). However, a11 the drugs used to produce an elevation of CAMP (dibutyryl CAMP and the CAMP phosphodiesterase inhibitor IBMX) caused similar changes in the transepithelial

potential and in the spontaneous rate of afferent firing. Note that while the transepithetia~ potential is falling and the spontaneous firing rate is increasing, there is no change in the response to Ach or mechanical stimu- lation (as noted in Table IfA).

It can be seen from Fig. 4 that Ach application increased spontaneous afferent activity without in- creasing mechanically evoked activity, It may even be

tACH ,lmin +

t ACH

- ME”-- -M.E.-

Fig, 4. Effect of acetylcholine on spontaneous and mechanically evoked afferent activity. The preparation is as depicted in Fig. 1 and described in Methods. The trace is of multiunit activity recorded from the cut end of the ampullar nerve and is expressed as impulses per second (frequency). Ach indicates the response to an injection of 2.0 ,u1 of a 1 mM solution of a~~~cho~ine close to the neur~pithelium. Evoked indicates the mechanicai stimulation of the semicircular canal. The evoked activity, marked by the dot, points out a cycle of mechanical stimulation of bigher volume than the other cycles. This was done in order to illustrate the ability of the preparation to increase in rate and the electronics to record such increase. The dashed curve emphasizes the response of the spontaneous afferent activity to Ach application during mechanically-evoked afferent activity. This trace demonstrates that the response to Ach increases spontaneous activity but does not affect mechanically evoked activity

(or even decreases it). Frequency is measured in impulses per second.

74

argued that Ach caused a reduction in evoked activity suggesting that the response to Ach and mechanical stimulation somehow compete or are physiological an- tagonists. Ach is thought to act on the hair cell by means of two receptors, one atropine-preferring and the other strychnine-preferring (Norris et al., 1988). The atropine-preferring receptor acts to depolarize the hair cell by reducing the hyperpolarizing calcium-de- pendent potassium current (Housley et al., 1990). The site of this competition might be a pool of calcium required for transmitter release. This competition is also suggested by Fig. 6. That is, in the face of a high Mg2+-low (10 mM) Ca’+ external (0.1 mM) solution, despite the fact that spontaneous activity has already failed, hair cell transmitter can still be released to cause afferent firing either by Ach or mechanical stim- ulation. But, when the effect of either Ach or mechani- cal stimulation is exhausted, the as-yet-unapplied stim- ulus also becomes ineffective. This parallel exhaustion suggests that both stimuli operate on similar intra-hair cell mechanisms, for instance intracellular calcium

4-AP ACTION ON SP~NT~EOUS ACTIWY

LOG CONCENTRATION #dsi

4-AP ACTION ON EVOKED ACTWIlY

too

!

~

i

“t---l---“- 3 4 5

LOG CONCENT~TION phi

Fig. 5. These dose-response curves were obtained from application of 4-aminopyridine (4-AP) into the perilymph of the preparation de- picted to Fig. 1. Data has been normalized as percentages of the control and errors bars represent standard errors. These curves ciearly demonstrate that spontaneous and mechanically evoked firing

can be regulated independently.

t E 00 &oo g 0

t ACH tACH t ACH -

rxi. ME. M.E

Fig. 6. Usage of the procedure previously described (Fig. I), the perilymphatic solution was replaced with a high magnesium, low calcium solution (10 mM, Mg H; 0.1 mM, Ca’” f. These results clearly demonstrate that the response to mechanical stimulation evoked is much less susceptible to changes in extracellular calcium than the spontaneous neurotransmitter release which rapidly goes to zero. Ach at the arrow indicates a 10 ,ILI boius of ace~lcholine (I mM) applied to the perilymphatic side of the ampulla. Similar results to those seen in high magnesium-low calcium were obtained using the calcium channel blockers diltiazem, cadmium, cobalt or strepto- mycin. Frequency is measured as multiple unit afferent impulses per

second.

pools. Other agonists such as glutamate when injected increase spontaneous activity without affecting evoked activity (Carlos Erostegui, personal communication).

~i~erenziu~ effect of 4-AP An entire publication describing the effects of 4-AP

on the SCC is forthcoming (Ricci et al., 1991). These data (Fig. 5) are only included to demonstrate the entire dose-response curves of the differential effects of this agent on spontaneous and evoked activity as in Table IID.

The differential effect of low calcium-high magnesia on spontaneous and evoked afferent activity

There are two important points made in Fig. 6. First, that spontaneous activity was completely sup- pressed by placing the preparation in low Ca*‘-(0.1 mM) high Mg2+ (10 mM) solution at a time when some afferent firing may still be elicited by mechanical stimu- lation (Table IIF). Second, that mechanical stimulation (in the face of low Ca2+-high Mg2+ solution) to ex- haustion of response eliminates the response to Ach. Conversely, when Ach was given first, in the presence of low Ca2+-high Mg 21, then the response to it elimi- nates the subsequent response to mechanical stimula- tion (not depicted). This observation again suggests, as do the results depicted in Fig. 4 that Ach and mechani- cal stimulation may compete for a common step in transmitter release such as a limited (i.e. intracellular) pool of calcium.

Discussion

The role of cyclic adenosine monophosphate (CAMP) The expected increases of CAMP levels caused by

applying dibutyryl CAMP, stimulating adenylyl cyclase

75

with forskolin or by inhibiting CAMP catabolism by mean of the phosphodiesterase inhibitor IBMX (3-iso- butyl-1-methylxanthine) all led to increases in sponta- neous activity with no change or even slight reductions in evoked activity (Fig. 3). This result suggests a role for CAMP in the control of the spontaneous but not the evoked release of the afferent neurotransmitter (Ricci et al., 1991~) The effect of manipulations de- signed to increase CAMP levels is probably exerted on the hair cells. For, if the effect were to take place postsynaptically then both evoked and spontaneous firing would likely be affected.

A putative glutamate autoreceptor on the hair cell Hair cells have been shown to release glutamate in a

calcium-dependent manner (Jenison et al., 1985; Drescher et al., 1987b). Although Drescher et al. (1987)

SPONTANEOUS

found that glutamate release was not inhibited by a supranormal magnesium concentration (10.1 mM). Sev- eral groups have reported that both glutamate antago- nists and spider toxins, known to oppose the action of glutamate, reduced the evoked afferent activity more than the spontaneous (Bledsoe et al., i988) (recording from Xenopus laevk lateral line afferents) Cousillas et al. (1988) (recording from the cochlear afferents; spider toxin study); and Starr and Sewell (1990) (recording from goldfish saccular afferents). Valli et al. (1985) found that glutamate and one of its antagonists, alpha-amino-adipate, exerted presynaptic effects on the release of neurotransmitter from the hair cells of the semicircular canal. Prigioni et al. (1990) using excita- tory amino acid agonists not only confirmed the earlier finding (Valli et al., 1985) of both pre- and postsynaptic effects of excitatory amino acids but suggested that

EVOKED

Ach

Fig. 7. Schematic illustrating the different events involved in the s~ntaneous (left) or evoked (right) release of the neurotransmitter from the hair cells of the frog semicircular canal according to the novel hypotheses presented in this paper. The solid lines and arrows point out generally accepted processes while the dashed lines and arrows correspond to the novel hypotheses reported in this paper. (Lefr) In the absence of mechanical stimulation of the neurosensory epithelium, a resting influx of K+, through the transductional elements (I), is believed to be responsible for the activation of voltage-dependent calcium channels (2), located on the basolateral side of the hair cells. The resulting influx of calcium into the sensory cells is in turn responsible for the spontaneous release of the neurotransmitter (3). The possible role of a calcium influx from the endolymph (4), in the spontaneous mode of neurotransmitter release, has not yet been determined. The spontaneous release of the neurotransmitter is enhanced by an increase in CAMP levels or by a local application of acetylcholine (Ach, 5). These two experimental manipulations could be responsible for an increase of the intracellular concentration of free calcium and thus an increase of the amount of neurotransmitter released at rest. (Right) When the sensory epithelium is mechanically stimulated in the depolarizing direction, the influx of potassium through the transductional elements (1) and thus the entry of calcium across voltage-dependent channels (2) are increased. The influx of calcium from the endolymph has also been shown to increase (4) during the mechanical stimulation of the hair ceils. The consequent increase of the intracellular free calcium also as a result of release from intracellular stores in (3) is in turn responsible for the increase of neurotransmitter released. This evoked release can be enhanced by the release of glutamate (6) which acts on a presynaptic glutamate receptor on the hair cells. Data also suggest that a part of the evoked release of the neurotransmitter can be controlled by a mobii~zation of the

intracellular stores of calcium (7) either by a triggering entry of calcium (4) or by second messengers.

7h

these amino acids may function presynapticaily in a positive feedback manner. Starr and Sewell (1990) sug- gested that glutamate antagonists act presynaptically on the saccular hair cells to reduce the amount of neurotransmitter released by sound stimulation. These data, indicating the existence of a glutamate receptor on the hair cell, suggest that there is normally a gluta- mate-mediated recruitment of the afferent neurotrans- mitter during the evoked release but not during the spontaneous release mode. Guth et al. (1988) have in fact produced evidence that glutamate does not appear to be involved in the spontaneous release of the neuro- transmitter. These authors showed that of two manipu- lations of the SCC leading to a loss in response to applied glutamate neither caused a change in the spon- taneous multiple unit afferent firing rate. These two manipulations were: applications of glutamate in quick succession to cause glutamate desensitization and bath application of the enzyme glutamate decarboxylase to catabolize extracellular glutamate. Likewise, Gleich et al. (1990) found that glutamate application to the point of desensitization did not affect the spontaneous rate in cochlear afferents. In contrast, Rebillard and Bryant (19891 employed the enzyme glutamate dehydrogenase in perfusions of the cochlea to test whether catabolism of glutamate would alter cochlear electrophysiology. Unfortunately, these authors did not examine sponta- neous activity but only sound-evoked compound action potentials. Their finding of a shift to the right of the input-output curve to compound action potential com- ports with the suggestion that glutamate may be in- volved in evoked activity. Thus, when the sensory ep- ithelium is mechanically stimulated, glutamate release by the hair cells may act on a hair cell autoreceptor that potentiates neurotransmitter release and thus greatly increases afferent firing. Such a glutamate- mediated positive feedback is unnecessary and even undesirable in the spontaneous condition.

The observation that glutamate and glutamate an- tagonists may act on a presynaptic receptor is not a new one (Usherwood and Machili, 1966; Florey and Woodcock, 1968; Colton and Freeman, 1975; Thieffry and Bruner, 1978; Cotman et al.? 1986; Forsythe and Clements, 1988). Nevertheless, such a positive feed- back is unusual for transmitters, the majority of them exerting a negative feedback control over their own release (Vizi, 1979). Glaum et al (1990) have reported that glutamate is capable of increasing intracellular free Ca2+ concentrations by affecting both Ca’+ mobi- lization and influx. Both of these processes may be involved in glutamate’s actions on the hair cell. Gluta- mate can also act on postsynaptic receptor as described in the cochlea (Bobbin and Thompson, 1978; Jenison et al., 1986) and SCC (Valli et al., 1985) and such an action by glutamate or other excitatory amino acid cannot be ruled out.

The roles of L~U~U~ c~~~i~rn sources in the spontaneous and evoked modes of transmitter release

The manipulations of Ca*’ concentrations and fluxes listed in Tables I and II suggest not only that CaZt is involved in both spontaneous and evoked re- lease of transmitter but that its involvement in the two modes .of release is different. Calcium is probably in the final common path to transmitter release in both the spontaneous or evoked modes. However, the source of the calcium in the two modes may be different. The most parsimonious explanation for the results seen with calcium channel blockers and with low Ca*+-high Mg*’ (Fig. 3) (i.e. that spontaneous fails before evoked activity) is that spontaneous activity is more dependent on the entry of extracellular calcium through ion chan- nels whereas the evoked mode can cause the mobiliza- tion of intracelIular stores, at least for a time. It is important to emphasize that the two sources of cal- cium, intra- and extracellular, are both parts of a calcium circuit (i.e. Ca2+ flows into the cell through ion channels is bound to intracellular organelles from which it may be mobilized and it may be transported back across the cell membrane to the extracellular fluid). The intracellular pools, being much smaller than the extracellular source, are ultimately dependent on the extracellular source (i.e., if extracellular calcium is removed, intracellular calcium will in time be reduced or depleted). Thus the intra- and extracellular pools are interdependent and a change in one pool may ultimately cause a change in the others. Therefore, experimental manipulations designed to influence one mode of transmitter release may over time, affect the others. The evidence implicating one or another of these pools as proximate sources of Ca”” for sponta- neous or evoked transmitter release in the intact organ is at the moment indirect.

Moreover, the existence of an influx of caIcium from the endolymphatic compartment, previously described by Ohmori (19891, could be the biochemical stimulus responsible for the mobilization of intracellular stores of calcium. This influx of calcium may act directly or via second messengers to release intracellular stores of calcium (as reviewed in Mayer and Miller, 19901. The effect of an increase of intracellular Ca”+ concentra- tion on Ca*+ mobilization from intracellular pools is biphasic. Indeed, as reviewed by Yamamoto and Karaide (1990) low concentrations of Ca” promote and higher concentrations inhibit intracellular cell mo- bilization.

In any case, it seems clear that mechanisms respon- sible for the afferent neurotransmitter release during spontaneous and evoked activities are different and should be the focus of new research activity. A note of caution-the majority of studies cited and illustrated are those carried out in the SCC. It is not yet clear the

77

extent to which these hypotheses may be generalized to other hair cell-bearing organs.

Acknowledgements

The authors wish to thank Drs. Paolo Valli and W.F. Sewell for their very helpful criticisms of the manuscript. This work was supported by grant No. DC-00303 from the National institutes of Health.

References

Annoni, J,-M., Cochran, S.L. and Precht, W. (1984) Pharmacology of the vestibular hair cell-afferent fiber synapse in the frog. J. Neurosci. 4, 2106-2116.

Aubert, A., Bernard. C. and Vaudry, H. (1991) Effects of modifica- tions of extracellular and intracellular calcium concentrations on the bioelectrical activity of the isolated frog semicircular canal. Brain Res. fin press).

Bernard, C. (1980) Electrical activity of isolated semicircular equal in the frog: action of streptomycin. Plugers Arch. 388, 93-99.

Bernard, C. (1983) Action of streptomycin and calcium on the apical membrane of hair cells of the frog isolated semicircular canal. Acta Gto-Laryngol. 96, 21-20.

Bledsoe, SC., Bobbin, R.P. and Puel, J.L. (1988) Neurotransmission in the inner ear. In: A.E. John and J. Santos-Sacci (Eds.), Raven Press, New York, pp. 385-405.

Bobbin, R.P. and Thompson, M.H. (1978) Glutamate stimulates cochlear afferent nerve fibers. Fed. Proc. 37, 613.

Colton, C.K. and Freeman, A.R. (1975) Dual response to lobster muscle fibers to -glutamate. Comp. Biochem. Physiol. 51,275-284.

Cotman, C.W., Flatman, J., Ganong, A.H. and Perkins, M. (1986) Effect of excitatory amino acids in synaptic transmission of the Schaffer collateral-Commisural pathway of the rat hippocampus. J. Physiol. (London) 378, 403-415.

Cousillas, H., Cole, K.S. and Johnstone, B.M, (1988) Effect of spider venom on cochlear nerve activity consistent with glutaminergic transmission at hair cell-afferent dendrite synapse. Hear. Res. 36, 213-220.

Drescher, D.G. and Drescher, M.J. (1987) Calcium and magnesium dependence of spontaneous and evoked afferent neural activity in the lateral line organ of the Xenopus luelts. Comp. Biochem. Physiol. 87A, 305-310.

Drescher, M.J., Drescher, D.G. and Hatfield, J.S. (1987) Potassium- evoked release of endogenous primary amine-containing com- pounds from the trout saccular macula and saccular nerve in vitro. Brain Res. 417, 39-50.

Durham. D.. Rubel, E.W. and Steel, K.P. (1989) Cochlear ablation in deafness mutant mice: 2-deoxyglucose analysis suggests no spon- taneous activity of cochlear origin. Hear. Res. 43, 39-46.

Florey, E. and Woodcock, B. (1986) Presynaptic excitatory actions of glutamate applied to crab nerve-muscle preparations. Comp. Biochem. Physiol. 26, 651-661.

Forsythe, I.D. and Clements, J.D. (1988) Glutamate autoreceptors reduce EPSC’s in cultured hippocampal neurons. Sot. Neurosci. Abs. 14, 791.

Glaum, S.R., Holzworth, J.A. and Miller, R.J. (1990) Glutamate receptors activate Ca*+ mobilization and CaZf influx into astro- cytes. Proc. Nat1 Acad. Sci. 87, 3454-3458.

Gieich, G., Johnstone, B.M. and Robertson, D. (1990) Effects of L-glutamate on auditory afferent activity in view of its proposed

excitatory transmitter role in the mammalian cochlea. Hear. Res. 45, 295-312.

Guth, P.S., Norris, C.H., Barron, S.E. (1988) Three tests of the hypothesis that glutamate is the sensory hair cell transmitter in the frog semicircular canal. Hear. Res. 33, 223-228.

Housley, G.D., Norris, C.H. and Guth, P.S. (1990) Cholinergically-in- duced changes in outward currents in hair cells isolated from the isolated semicircular canal of the frog. Hear. Res. 43, 121-134.

Hudspeth, A.J. (1983) Mechanoelectrical transduction by hair cells in the acoustico-lateralis system. Ann. Rev. Neurosci. 6, 187-215.

Jenison, G.L., Bobbin, R.P. and Thaimann, R. (1985) Pota~ium-in- duced release of endogenous amino acids in the guinea pig cochlea. J. Neurochem. 44, 1845-1853.

Jenison, G.L., Winbety, S. and Bobbin, R.P. (1986) Comparative actions of quisgualate and N-Methyl-o-Aspartate, excitatory amino acids agonists, on guinea pig cochlear potentials. Comp. Biochem. Physiol. 84C, 385-389.

Mayer. M.L. and Miller R.J. (1990) Excitatory amino acid receptors, second messengers and regulation of intracellular Ca*+ in mam- malian neurons. Trends in Pharmacol. Sciences 11, 254-260.

Norris, C.H., Housley, G., Williams, W., Guth, S., and Guth, P. (1988) The acetylcholine receptors of the semicircular canal in the frog. Hear. Res. 32, 197-206.

Ohmori, H. (1989) Mechanoelectrical transduction of the hair ceil. Jap. J. Physioi. 39, 643-657.

Precht, W. (1976) Physiology of the peripheral and central vestibuiar system. In: R. Llinas and and W. Precht (Eds.), Frog Neurobiol- ogy, Springer Vertag, Berlin-Heidelberg, pp. 481-512.

Prigioni, I., Russo, G., Valli, P. and Masetto, S. (1990) Pre and postsynaptic excitatory action of glutamate agonists on frog vestibular receptors. Hear. Res. 46, 253-260.

Puel, J.L., Bledsoe, SC., Bobbin, R.P., Caesar, G. and Fallon, M. (1989) Comparative actions of salicylate on the amphibian lateral line and guinea pig cochlea. Comp. Biochem. Physiol. 93, 73-80.

Rebiilard, G. and Bryant, G.M. (1989) Effects of in vivo perfusion of glutamate dehydrogenase in the guinea pig cochlea on the VIII nerve compound action potential. Brain Res. 494, 379-382.

Ricci, A., Norris, C. and Guth, P.S. (1991a) Effects of 4-Amino Pyridine (4AP) on the semicircular canal of the frog. Submitted for publication.

Ricci, A, Woiverton, S., Aubert, A., Norris, C. and Guth, P. (199Ib) Responses of the isolated semicircular canal to electrical stimula- tion. Abstr. Assoc. Res. Otolaryngol. p. 95.

Ricci, A., Norris, C.H. and Guth, P.S. (1991) Cyclic AMP modulates spontaneous activity in the semicircular canal. Submitted for publication. Ricci, A., Norris, C.H. and Guth, P.S. (1990) A role for cyclic AMP in the semicircular canal. Abstr. Assoc. Res. Otolaryngol. p. 52.

Russell, I.J. and Roberts, B.L. (1972) Inhibition of spontaneous lateral-line activity by efferent nerve stimulation. J. Exp. Bioi. 57, 77-82.

Sewell, W.F. (1990) Synaptic potentials in afferent fibers inne~ating hair ceils of the lateral line organ in Xenopus 1aet:is. Hear. Res. 44, 71-82.

Siegel, J.H. and Relkin, E.M., (1987) Antagonistic effects of periiym- phatic calcium and magnesium on the activity of single cochlear afferent neurons. Hear. Res. 28, 131-147.

Soto, E. and Vega, R. (1988) Actions of excitatory amino acids agonists and antagonists on the primary afferents of the vestibu- lar system of the axolotl. Brain Res. 462, 104-111.

Starr, P.A. and Sewell, W.F. (1990) Effects of amino acid antagonists on auditory nerve synaptic potentials. Sot. Neurosci. A&r. 14, 331.

Thieffry, M. and Bruner, J. (1978) Direct evidence for a presynaptic action of glutamate at a crayfish neuromuscular junction. Brain Res. 156, 402-406.

7x

Usherwood, P.N.R. and Machili, P. (1966) Chemical transmission at

the insect excitatory neuromuscular synapse. Nature (London)

210, 634-636.

Valli, P., Taglietti, V. and Rossi, M.L. (1974) Effects of d-tubo-

curarine on the ampullar receptors of the frog. Acta Oto-Laryn-

gol. 7x. 51-58.

Valli, P., Zucca, G. and Casella, C. (1979) Ionic composition of the

endolymph and sensory transduction in labyrinthine organs. Acta

Oto-Latyngol. 87, 466-471.

Valli, P.. Zucca, G., Prigioni, I., Botta, L., Casella, C. and Guth, P.S.

(1985) The effect of glutamate on the frog semicircular canal. Brain Res. 330. l-9.

Valli. P., Botta, L., Zucca, G. and Casella, C. (1986) Functional

organization of the peripheral efferent vestibular system in the

frog. Brain Res. 361, 92-97.

Vizi. E.S. (1979) Presynaptic modulation of neurochemical transmis-

sion. Prog. Neurobiol. 12, 181-290.

Yamamoto, H. and Kanaide, H. (1990) Release of intracellularly

stored Ca’+ by inositol 1,4,S trisphosphate-an overview. Gen.

Pharmac. 21, 387-393.