K+ Accumulation and Clearance in the Calyx Synaptic Cleft of Type I Mouse Vestibular Hair Cells.

Vestibular organs of Amniotes contain two types of sensory cells, named Type I and Type II hair cells. While Type II hair cells are contacted by several small bouton nerve terminals, Type I hair cells receive a giant terminal, called a calyx, which encloses their basolateral membrane almost completely. Both hair cell types release glutamate, which depolarizes the afferent terminal by binding to AMPA post-synaptic receptors. However, there is evidence that non-vesicular signal transmission also occurs at the Type I hair cell-calyx synapse, possibly involving direct depolarization of the calyx by K+ exiting the hair cell. To better investigate this aspect, we performed whole-cell patch-clamp recordings from mouse Type I hair cells or their associated calyx. We found that [K+] in the calyceal synaptic cleft is elevated at rest relative to the interstitial (extracellular) solution and can increase or decrease during hair cell depolarization or repolarization, respectively. The change in [K+] was primarily driven by GK,L, the low-voltage-activated, non-inactivating K+ conductance specifically expressed by Type I hair cells. Simple diffusion of K+ between the cleft and the extracellular compartment appeared substantially restricted by the calyx inner membrane, with the ion channels and active transporters playing a crucial role in regulating intercellular [K+]. Calyx recordings were consistent with K+ leaving the synaptic cleft through postsynaptic voltage-gated K+ channels involving KV1 and KV7 subunits. The above scenario is consistent with direct depolarization and hyperpolarization of the calyx membrane potential by intercellular K+.

Intercellular K⁺ accumulation depolarizes Type I vestibular hair cells and their associated afferent nerve calyx.

Mammalian vestibular organs contain two types of sensory receptors, named Type I and Type II hair cells. While Type II hair cells are contacted by several small afferent nerve terminals, the basolateral surface of Type I hair cells is almost entirely enveloped by a single large afferent nerve terminal, called calyx. Moreover Type I, but not Type II hair cells, express a low-voltage-activated outward K(+) current, I(K,L), which is responsible for their much lower input resistance (Rm) at rest as compared to Type II hair cells. The functional meaning of I(K,L) and associated calyx is still enigmatic. By combining the patch-clamp whole-cell technique with the mouse whole crista preparation, we have recorded the current- and voltage responses of in situ hair cells. Outward K(+) current activation resulted in K(+) accumulation around Type I hair cells, since it induced a rightward shift of the K(+) reversal potential the magnitude of which depended on the amplitude and duration of K(+) current flow. Since this phenomenon was never observed for Type II hair cells, we ascribed it to the presence of a residual calyx limiting K(+) efflux from the synaptic cleft. Intercellular K(+) accumulation added a slow (τ>100ms) depolarizing component to the cell voltage response. In a few cases we were able to record from the calyx and found evidence for intercellular K(+) accumulation as well. The resulting depolarization could trigger a discharge of action potentials in the afferent nerve fiber. Present results support a model where pre- and postsynaptic depolarization produced by intercellular K(+) accumulation cooperates with neurotransmitter exocytosis in sustaining afferent transmission arising from Type I hair cells. While vesicular transmission together with the low Rm of Type I hair cells appears best suited for signaling fast head movements, depolarization produced by intercellular K(+) accumulation could enhance signal transmission during slow head movements.

Analysis of pre- and postsynaptic activity in the frog semicircular canal following ototoxic insult: differential recovery of background and evoked afferent activity.

Frogs were treated with a single dose of gentamicin administered intraotically to produce severe degeneration of posterior semicircular canal hair cells and to evaluate the time course of functional damage and recovery both at pre- and postsynaptic level. In isolated canal preparations the endoampullar potential, which reflects the summed receptor potentials of crista hair cells, was progressively reduced in amplitude and completely abolished 6 days after gentamicin treatment. At this time the crista epithelium was devoid of hair cells. The recovery of the endoampullar potential began around 9 days after the ototoxic insult and its amplitude progressively increased to reach, after 20 days, values close to those observed in control experiments. The endoampullar potential amplitude was related to the degree of hair cell regeneration in the crista epithelium. Consistent with the presynaptic damage, the slow generator potential (representing the summed miniature excitatory postsynaptic potential [mEPSP] activity of all posterior nerve fibres) and the resting and evoked spike discharge recorded from the whole ampullar nerve were abolished 6 days after gentamicin treatment. The recovery of the background and evoked afferent activity showed different behaviours. Background spike activity became detectable around 8 days after the ototoxic insult, but was not modulated by canal stimulation at this time, and no generator potential was detected. Moreover, the resting spike frequency fully recovered and reached control values around 15 days after gentamicin treatment, whereas the evoked activity attained normal values only 20 days after the ototoxic insult. These results were confirmed by intracellular recordings from single afferent fibres of the ampullar nerve in intact labyrinth preparations. Absence of any resting and evoked discharge was the most common pattern observed in the early period from 7 to 8 days after gentamicin treatment. Fifty-five percent of impaled afferents were silent while the others showed low resting frequencies of mEPSPs and spikes, and were unresponsive to canal rotation. In the intermediate period from 14 to 15 days after gentamicin treatment, background mEPSP and spike frequencies approached those evaluated in control experiments, but the frequencies of the evoked mEPSPs and spikes were clearly lower than in controls. In the late period, from 18 to 20 days after the ototoxic insult, the impaled afferents showed normal evoked mEPSP and spike frequencies. The present data indicate that the frog semicircular canal completely recovers its pre- and postsynaptic activity following severe ototoxic insult. During the regeneration process, the cytoneural junction regains function and the resting discharge reappears before recovery of mechanoelectrical transduction.

Gradients of expression of calcium and potassium currents in frog crista ampullaris.

In the present work we studied the intraregional expression of voltage-dependent Ca2+ and K+ currents in hair cells of frog crista ampullaris. The currents were recorded in situ from sensory cells of the peripheral region, the most populated region of the crista, by using the whole-cell variant of the patch-clamp technique. Voltage-clamp recordings revealed that the calcium current (I(Ca)) and the outward potassium currents of I(A), I(K) I(KCa) types and the inward rectifier potassium current of I(K1) type exhibited a significant gradient of density (pA/pF) along the region. I(A) density was maximal in cells located at the beginning of the peripheral region and decreased gradually becoming very small at the opposite end. All the other currents showed an opposite gradient of expression. Current-clamp experiments showed that the voltage behaviour of hair cells changed in relation to cell position. Cells located at the beginning of the peripheral region showed large depolarizations from the resting potential (close to -45 mV) which are consistent with the presence of small I(K) and I(KCa), and an I(A) largely inactivated at rest. These cells also exhibited slowly developing and large hyperpolarizations that approached passive ones, due to the lack of I(K1). In contrast, cells located at the opposite side of the region showed smaller depolarizations and hyperpolarizations from the resting potential (close to -65 mV), due to the presence of large I(K) and I(KCa), and I(K1), respectively. The possible role of the intraregional variation of Ca2+ and K+ currents in both hair cell function and afferent discharge properties is discussed.

Inactivating and non-activating delayed rectifier K+ currents in hair cells of frog crista ampullaris.

The possible presence of different types of delayed rectifier K+ current (I(K)) was studied in vestibular hair cells of frog semicircular canals. Experiments were performed in thin slice preparations of the whole crista ampullaris and recordings were made using the whole-cell patch-clamp technique. We found that an apparent homogeneous I(K), isolated from the other K+ currents, could be pharmacologically separated into two complementary components: a capsaicin-sensitive current (I(Kc)) and a barium-sensitive current (I(K,b)). I(K,c) was recruited at potentials more positive than -60 mV and showed a slow activation having a time constant (tau(a)) ranging on average from 12 ms at 40 mV to 32 ms at -20 mV. This current inactivated slowly with two voltage-independent time constants (ta(d1) and tau(d2) were 300 ms and 4 s respectively) and more than 80% of the channels were in an inactivated state at the cell resting potential. I(K,b) was also recruited at potentials more positive than -60 mV, but in contrast to I(K,c), it activated more rapidly (tau(a) ranged on average from 1 ms at 40 mV to 4.5 ms at -20 mV) and it did not exhibit any inactivation process. Current clamp experiments revealed that I(K,b), at variance with I(K,c), contributes to the cell resting potential and represents the main repolarizing current when sensory cells are depolarized from rest. I(K,c) could have a role in hair cells when they are depolarized after hyperpolarizing stimuli, a condition that removes channel inactivation.

Position-dependent expression of inwardly rectifying K+ currents by hair cells of frog semicircular canals.

The identity and the expression of inwardly rectifying ionic currents were studied using the whole-cell variant of the patch-clamp technique in frog semicircular canal hair cells in situ. The currents were examined in club-, cylindrical- and pear-shaped sensory cells located in three discrete regions of the crista. A unique current of I(K1) type was distinguished based on its K+ selectivity, rapid monoexponential activation, dependence of activation on external K+ and blockade by Ba2+ and Cs+. I(K1) was found in virtually all cylindrical hair cells of the central region and in club-shaped cells located in the halves of the peripheral regions closest to the centre of the crista. Pear-shaped cells of the intermediate regions showed no inward rectification. The I(K1) density (pA/pF) varied along the crista depending on cell position, being maximal in cells located in the middle of the central region and decreased towards its ends. In the peripheral regions, the gradient of I(K1) increased towards the centre of the crista. Current clamp experiments showed that sensory cells having larger I(K1) constantly exhibited more negative resting potentials and required more depolarizing current to elicit an active response than cells having small or no I(K1).

Localization of Ca-ATPase in frog crista ampullaris.

The distribution of Ca-ATPase in frog crista ampullaris was mapped ultracytochemically by using a one-step lead citrate reaction. Electron-dense precipitates, as an expression of Ca-ATPase activity, were observed on the surface of stereocilia and on the apical membrane surrounding the cuticular plate of hair cells. Sensory cells of the isthmus region showed more reactivity than those of the peripheral regions of the crista. No reaction products were detectable on the basolateral membranes and in cytoplasmatic organelles. Supporting cells of the crista showed a quite variable Ca-ATPase reaction on microvilli and on basolateral membranes. The presence of an evident reactivity on the stereocilia is consistent with the existence of an apical calcium microdomain involved in the mechano-transduction process and supports the current view that calcium ions enter the stereocilia during natural stimulation. On the other hand, the lack of an observable reactivity on the basolateral membrane of hair cells suggests that in semicircular canals other mechanisms of active transport of calcium ions across the plasma membrane, such as Na-Ca exchange, may be involved in homeostasis of the ion.

Inactivation of delayed rectifier K+ current in semicircular canal hair cells.

Some properties of the inactivation process of delayed rectifier K+ current (Ik) were investigated in vestibular hair cells of the central region of frog crista ampullaris. These cells were chosen since they exhibited a very large Ik. Experiments were performed on thin slices of sensory epithelium using the whole-cell variant of the patch-clamp technique. Ik showed clear time-dependent inactivation over a period of some seconds, but the current did not completely inactivate even after 30 s depolarizing pulses. Another interesting finding was that inactivation could be well fitted by the sum of two exponentials: at 20 mV depolarization the fast time constant was 291.3 ms and the slow time constant was 2662.3 ms. In addition, an analysis of the steady-state inactivation process of Ik revealed that the inactivation curve was incomplete showing a non-inactivating current at potentials more positive than -50 mV. These results suggested that Ik in hair cells of frog crista ampullaris is composed of more than one component: by at least one inactivating and one non-inactivating component. The possible role of these components in hair cell excitability is discussed.

Potassium currents of pear-shaped hair cells in relation to their location in frog crista ampullaris.

Voltage-dependent K+ currents in pear-shaped hair cells of the frog crista ampullaris were investigated in thin slice preparations using the whole-cell variant of the patch-clamp technique. Microscopy observation revealed that pear-shaped cells are located in intermediate and peripheral regions of the crista, whereas they are absent in the central region. Voltage-clamp recordings in cells from the peripheral regions revealed that the total outward K+ current could be separated pharmacologically into three distinct components: a A-type K+ current (IA); an inactivating calcium-activated K+ current (IK(Ca)) and a delayed rectifier K+ current (IK). IK and IK(Ca) exhibited similar magnitude and accounted for most of the membrane cell conductance. The same experimental protocol applied to cells from the intermediate regions showed the presence of a large and sustained IK(Ca) which represented 95% of the total outward current. In this region IA was absent. The present results demonstrated that pear-shaped hair cells located in two discrete regions of frog crista ampullaris exhibit a different complement of voltage-dependent conductances, suggesting that they can play a different role in processing the natural stimulus.

Morphological features of different regions in frog crista ampullaris (Rana esculenta).

The cellular organization of different regions of the crista epithelium from the frog posterior semicircular canal was studied by light, transmission and scanning microscopy. The sensory epithelium consists of hair cells surrounded by supporting cells and basal cells located close to the basement membrane. Three types of hair cells, namely club-like, cylindrical and pear-like cells differentially distributed along the crista could be recognized on the basis of their shape. Club-like cells are located only in the peripheral regions, cylindrical cells both in the central and in the peripheral regions, and pear-like cells appear segregated into the intermediate regions. Sensory cells of the central region are characterized by a ciliary apparatus consisting of stereocilia usually shorter--and in some cases less numerous--than those of cells of the other regions. The presence of large evaginations of the apical membrane of hair cells and of several vesicles of microexocytosis demonstrates that receptor cells have a considerable secretory activity. This secretory activity is also proven by the presence in the supranuclear region of hair cells of numerous Golgi complexes. Moreover, the presence of two kinds of Golgi complexes, one constituted by dilated cisternae containing a moderately electron-dense material and the other made up of flattened electron-transparent cisternae, suggests a diversified secretion of material by the hair cells. This heterogeneous material may provide substances important for cupula formation and the composition of the endolymph.

Isolation of A-type K+ current in hair cells of the frog crista ampullaris.

Different procedures to isolate the K+ A-type current (IA) from other membrane currents were tested on the complex inactivating outward K+ current generated in hair cells from the peripheral regions of the frog crista ampullaris. Experiments were performed in thin slices of epithelium using the whole-cell configuration of the patch-clamp technique. The conventional conditioning voltage protocol did not allow a satisfactory isolation of IA, due to the presence of other K+ currents showing overlapping steady-state inactivation properties. An attempt to block other K+ currents using calcium-free saline containing 50 mM TEA also failed to provide a satisfactory isolation of IA, due to contamination by a residual sustained current, probably consisting of a slow delayed outward K+ current (IK). Use of the selective A-channel blocker 4-aminopyridine (4-AP) at concentrations < 12 mM was also unsatisfactory because at these concentrations 4-AP produced a voltage-dependent blockade. Conversely, use of 4-AP at concentrations of 15-20 mM allowed a good separation of an uncontaminated IA. These results indicate that IA in hair cells of vestibular epithelium can be isolated most effectively by the 4-AP procedure, provided that sufficiently high concentrations of the A-channel blocker are used.

Glutamate excitatory effects on ampullar receptors of the frog.

The action of glutamate on frog ampullar receptors was investigated to assess the potential role of this excitatory amino acid as an afferent transmitter in the hair cell system. Intracellular recordings from single afferent units in the isolated labyrinth revealed that glutamate and the glutamate receptor agonists, N-methyl-D-aspartic acid, quisqualic acid and kainic acid increase dose-dependently the frequency of the resting afferent discharge of EPSPs and spikes and produce long lasting depolarizations. After blocking synaptic transmission by using 5 mM Co(2+), the same compounds elicited only depolarizations of amplitude comparable to those observed in normal saline. Quisqualic acid and kainic acid were much more potent than N-methyl-D-aspartic acid in increasing the frequency of afferent discharge and in causing axonal depolarizations. The depolarization caused by glutamate was reduced dose-dependently by the competitive non-NMDA receptor antagonist 6-cyano-7-nitroquinaxoline-2,3 dione and disappeared almost completely in Na(+)-free Ringer solution. These results are consistent with the hypothesis that glutamate is the afferent transmitter in vestibular organs and indicate that receptors mainly of the non-NMDA type are present not only at postsynaptic level but also in hair cells. Presynaptic glutamate receptors may function as autoreceptors controlling by a positive feed-back mechanism the release of the afferent transmitter.

Differential expression of potassium currents by hair cells in thin slices of frog crista ampullaris.

1. Electrical responses in hair cells located in the peripheral regions and in the central region of the frog crista ampullaris were investigated in thin slice preparations by using the whole-cell configuration of the patch-clamp technique. 2. Hair cells from the peripheral regions exhibited mostly a club-like shape and had an average resting potential of -46 mV, whereas cells from the central region had mostly a cylindrical shape and a more negative resting potential (-57 mV). 3. Voltage-clamp recordings revealed that ionic conductances differed in the two epithelial regions. Cells from the peripheral regions exhibited a transient K+ current of A-type (IA) in conjunction with a slow rectifier outward K+ current (IK). Cells from the central region showed little or no IA and generated an IK together with an inward rectifier K+ current (IIR). In both regions, hair cells showed a rapidly activating Ca(2+)-dependent outward K+ current (IK(Ca)) that rapidly inactivated to reach a steady-state level during 150-ms test pulses. 4. IA activated close to -60 mV and was inhibited by 12 mM 4-aminopyridine (4-AP). The time course of this current showed time to peak values of 3-4 ms at 0 mV. Inactivation was fast and almost voltage-independent. The decay time constant was approximately 35 ms at 0 mV. 5. IK was recruited close to -60 mV and activated slowly, reaching peak values in approximately 100 ms at 0 mV. It showed no evidence of inactivation during 150-ms test pulses and it was insensitive to 4-AP. 6. IIR activated at membrane potentials more negative than -90 mV and was blocked by exposure to 6 mM Cs+ or to a K(+)-free medium. This current showed an outward relaxation at potentials more negative than -140 mV, an effect that disappeared after exposure to a Na(+)-free medium. 7. IK(Ca) was recruited close to -40 mV and was inhibited by exposure to a Ca(2+)-free external medium or to 0.5 mM Cd2+. The time to peak of this current was approximately 3 ms at 0 mV and inactivation was very fast and almost independent from the membrane potential. The decay time constant was approximately 4 ms at 0 mV. 8. IK and IA were prominent in hair cells from the peripheral regions, whereas IK accounted for most of the membrane conductance in cells from the central region. The contribution of IK(Ca) was comparable in cells from both epithelial regions.(ABSTRACT TRUNCATED AT 400 WORDS)

Non-NMDA receptors mediate glutamate-induced depolarization in frog crista ampullaris.

The effect of glutamate on frog crista ampullaris was investigated in order to assess the potential role of this agent as an afferent transmitter in inner ear organs. Intracellular recordings from single afferent axons in the isolated labyrinth showed that, after blocking synaptic transmission with high concentrations of Mg2+, micro-injections of glutamate elicit a dose-dependent postsynaptic depolarization. The amplitude of depolarization was reduced dose-dependently by the competitive non-NMDA receptor antagonist 6-cyano-7-nitroquinoxaline-2,3-dione. When Na+ concentration in the bath was progressively reduced, depolarization decreased gradually and disappeared almost completely in Na(+)-free Ringer. On the contrary, complete substitution of Ca2+ ions in the bath was without apparent effects. These results indicate that the postsynaptic depolarization induced by glutamate in frog semicircular canals involves the activation of non-NMDA amino acid receptors.

Calcium currents in solitary hair cells isolated from frog crista ampullaris.

Some properties of Ca2+ currents in hair cells isolated from frog semicircular canals by enzymatic or mechanical treatment were studied by using the whole-cell configuration of the patch-clamp technique. After blocking the large outward K+ currents by substituting Cs+ for K+ and adding tetraethylammonium to the pipette filling solution, voltage- and time-dependent inward currents were clearly detectable in the presence of 4 mM Ca2+ in the extracellular solution. Ca2+ current was recruited at test potentials more positive than -60 mV, showed a rapid activation, and exhibited no inactivation during 150-ms depolarizing pulses. The maximal amplitude was attained at about -20 mV, with an average value of about 80 pA. When Ca2+ in the extracellular solution was replaced with Ba2+, the magnitude of inward currents increased about twofold. Ba2+ currents were blocked more effectively by Cd2+ than by Ni2+, were suppressed by 0.5 microM omega-conotoxin, and were virtually unaffected by amiloride. The dihydropyridine Bay K 8644 caused a marked voltage-dependent increase in inward currents. The present data suggest that hair cells from frog crista ampullaris are endowed with a homogeneous population of Ca2+ channels having several properties similar to those described for neuronal L channels. Since these channels are recruited in a range of potentials close to the resting level, it is suggested that they subserve the control of both resting and evoked transmitter release from the basal pole of the hair cells.

[K+ and Ca++ currents in hair cells isolated from the semicircular canals of the frog]. / Correnti di K+ e di Ca++ in cellule ciliate isolate dai canali semicircolari della rana.

Hair cells of the inner ear are endowed with different types of ionic channels. To characterize voltage- and ion-dependent channels in vestibular hair cells, experiments were performed in enzymatically isolated hair cells of frog semicircular canals by using the whole-cell configuration of the patch-clamp technique. A large outward current, identified as a K+ current, was recorded when 132 mM KCl were present in the pipette filling solution. It could be dissected pharmacologically into three different components. The first component, which was transient and selectively blocked by 10 mM external 4AP, is most likely an IA-type current. The second one, sensitive to 20 mM external TEA, might be a delayed rectifier K+ current, while the third component insensitive to TEA and showing faster activation time course has been interpreted as a K+ current of IKCa-type. After blocking the outward current by substituting Cs+ for K+ and adding 20 mM TEA to the internal solution, a sustained inward current, identified as a Ca++ current, could be recorded. This current did not inactivate, and was blocked by Cd++ more effectively than Ni++, thus suggesting the presence of Ca++ channels similar to the neuronal "L" channels. Since both K+ and Ca++ channels were recruited at potentials near the resting level, it is suggested that they are involved in the modulation of the resting as well as the evoked transmitter release from the basal pole of the hair cells.

Intracellular recordings from "recurrent neurons" in the rat superior cervical ganglion.

Intracellular recordings in the isolated superior cervical ganglion of the rat showed that electrical stimulation of the cervical sympathetic trunk elicited in a cluster of neurons localized in the caudal part of the ganglion synaptically driven action potentials, and propagated potentials having the features of typical antidromic spikes. The results demonstrate that these neurons, besides synapsing with common preganglionic fibres, project their axons to the cervical sympathetic trunk. The recurrent neurons showed a very low threshold to direct intracellular stimulation and a high input resistance, suggesting that they have a small size. Almost all recurrent neurons were activated synaptically also by stimulating the postganglionic trunks, indicating that they are innervated by collaterals of preganglionic through-fibres which are known to sustain a direct pathway between pre- and postganglionic nerves. Moreover, some recurrent neurons could also be activated antidromically following stimulation of the external carotid nerve, indicating that their axons divide into collaterals which project not only to the preganglionic trunk but also to a postganglionic nerve. The presence of recurrent neurons in the superior cervical ganglion of the rat provides further evidence for the concept that sympathetic ganglia consist of discrete cell subpopulations which are segregated in different regions and probably subserve different functions.

Pre- and postsynaptic excitatory action of glutamate agonists on frog vestibular receptors.

In order to investigate the localization and the type(s) of excitatory amino acid receptors in the frog vestibular system, the exogenous amino acid agonists Quisqualic acid, Kainic acid and N-methyl-D-aspartic acid were tested on the sensory organ of semicircular canals. Intracellular recordings of the resting discharge from single afferents showed that these agonists exerted a complex excitatory action consisting in a rapid and brief increase in frequency of both EPSPs and spikes, followed by a slower and longer lasting membrane depolarization. The progressive impairment of natural transmitter release achieved by adding Mg2+ or Co2+ in the bath caused a dose-dependent decrease of the agonist-induced afferent discharge, without substantially affecting axonal depolarization. These results suggest that the exogenous amino acid agonists act both pre- and postsynaptically on the vestibular organs. Quisqualic acid and kainic acid were much more potent than N-methyl-D-aspartic acid in inducing excitatory effects, suggesting that the amino acid receptors located on both hair cells and afferent endings are mainly of the non-NMDA type. The present findings, while not excluding that an excitatory amino acid may be the afferent transmitter, highlight its possible function as a presynaptic modulator of the afferent transmission in the frog vestibular system.

[Properties of the postsynaptic depolarization produced by glutamate on the vestibular receptors in frogs]. / Proprietà della depolarizzazione postsinaptica prodotta dal glutammato nei recettori vestibolari della rana.

In order to investigate the possible role of glutamate (Glu) as afferent transmitter in the vestibular system, this agent was tested on sensory organs of frog semicircular canals. Intracellular recordings from single afferent axons in isolated labyrinths showed that, after blocking chemical transmission with high Mg++ (12 mM), micro-injections of Glu (5 mM-15 ul) elicited a long lasting postsynaptic depolarization. The amplitude of this depolarization was reduced dose dependently after addition of the amino acid antagonists Kynurenic acid or gamma-D-glutamylglycine to the bath. When the Na+ concentration in the bath was progressively reduced, the depolarization decreased gradually and disappeared almost completely in Na(+)-free Ringer. The removal of K+ affected the depolarization to a lesser extent: in K(+)-free Ringer depolarization decreased only by 30-40%. On the contrary, the complete substitution of Ca++ ions in the bath was without effect. Our results suggest that in the frog semicircular canals the postsynaptic depolarization induced by Glu involves the activation of non NMDA type of amino acid receptors, probably coupled to channels selective for Na+ and K+ ions. The present findings are consistent with the hypothesis that Glu or a related substance may be the transmitter released at the afferent synapses of the vestibular receptors.