Activation of GABA_B receptors results in excitatory modulation of calyx terminals in rat semicircular canal cristae

Animals.

Seventy two Sprague-Dawley rats (Charles River Laboratories), 13–18 days old and of either sex, were used for the experiments. All animals were handled in accordance with animal protocols approved by the Institutional Animal Care and Use Committee at the University at Buffalo and carried out in accordance with NIH guidelines.

Tissue preparation.

Dissection of the end organ and tissue preparation were performed as previously described (Ramakrishna et al. 2020; Sadeghi et al. 2014; Songer and Eatock 2013). Briefly, rats were decapitated after being deeply anesthetized by isoflurane inhalation. The skull was opened in the midline, and the bony labyrinth was removed and placed in extracellular solution. Under the microscope, the ampullae of the horizontal and superior canals were opened. Attachments to bone were carefully dissected, and the membranous labyrinth containing the cristae of the horizontal canal, anterior canal, and the macula of the utricle were taken out of the bone. The membranous labyrinth was opened over the two cristae and the utricle. The preparation was secured on a coverslip under a pin, transferred to the recording chamber, and perfused with extracellular solution at 1.5–3 mL/min.

Electrophysiology recording.

The extracellular solution contained (in mM) 5.8 KCl, 144 NaCl, 0.9 MgCl₂, 1.3 CaCl₂, 0.7 NaH₂PO₄, 5.6 glucose, and 10 HEPES, 300 mosmol/kgH₂O, pH 7.4 (NaOH). The intracellular solution contained (in mM) 20 KCl, 110 K-methanesulfonate, 0.1 CaCl₂, 5 EGTA, 5 HEPES, 5 Na₂-phosphocreatine, 4 MgATP, and 0.3 Tris-GTP, 290 mosmol/kgH₂O, pH 7.2 (KOH). To perform patch-clamp recording, tissue was visualized with a ×40 water-immersion objective and differential interference contrast (DIC) optics (Examiner D1 Zeiss microscope), and viewed on a monitor via a video camera (optiMOS, Qimaging). To make patch-clamp recording pipettes, borosilicate glass (1-mm inner diameter, 1B100F-4; WPI) was pulled with a multistep horizontal puller (P-1000; Sutter), coated with Sylgard (Dow Corning), and fire polished. Pipette resistances were 5–10 MΩ. All recordings were performed at room temperature (~23°C). Application of drug solutions was performed (VC-6 channel valve controller; Warner Instruments, Hamden, CT) using a gravity-driven flow pipette (~100-μm tip diameter) placed at ~45° angle with the top of the crista near the area where a calyx was recorded from. The above solutions result in a liquid junction potential of −9 mV, which was corrected offline. Drugs were dissolved daily in the extracellular solution to their final concentrations from frozen stocks. R-baclofen (100 µM), CGP 35348 (300 µM), iberiotoxin (IBTX; 150 nM), apamin (Apa; 300 nM), cadmium chloride (CdCl₂; 0.5–1 mM), and XE-991 (10 µM) were purchased from Tocris Bioscience. Barium chloride (BaCl₂; 100 µM) was purchased from Sigma.

Experiment protocol.

Whole cell patch-clamp recordings were performed from calyx terminals in the central region of the cristae of anterior or horizontal canals. All measurements were acquired using Multiclamp 700B amplifier (Molecular Devices), digitized at 50 kHz with Digidata 1440A, and filtered at 10 kHz by pCLAMP10.7 software. Calyces innervated type I hair cells with typical morphology. Membrane capacitance and series resistance were electronically compensated during recordings (correction and prediction circuits set to 75–80% with a bandwidth of 10–15 kHz). Cells were held at a holding potential of −70 mV, resulting in a holding current of <200 pA. Leak subtraction was not used. To minimize time-dependent changes (Hurley et al. 2006), all data were collected starting ~10 min after the membrane was ruptured. Since the effect of drugs increased with duration of application for the first ~10 min, to avoid time-dependent changes in drug effects, all data were collected 9–12 min after start of application (Ramakrishna et al. 2020). All drug effects were reversible.

During voltage-clamp recordings, a voltage step protocol was used to study calyx properties (Fig. 1A), which consisted of an initial holding potential of −79 mV (100 ms), a hyperpolarizing step to −129 mV (100 ms), 20-mV steps between −129 and +11 mV (300 ms), and return to holding potential of −79 mV. Depolarizing steps resulted in initial large Na⁺ inward currents and slower voltage-sensitive outward (potassium) currents as previously described (Horwitz et al. 2014; Meredith and Rennie 2018; Ramakrishna et al. 2020; Sadeghi et al. 2014; Songer and Eatock 2013). To study the effect of various drugs on responses to the voltage step protocol, current amplitudes were averaged for the final 100 ms of steps. Peak amplitude of tail currents was measured after the return to −79 mV at the end of steps and was plotted against holding potential at the preceding step (denoted as “activation step” on x-axis in figures).

During current-clamp recordings, we first measured the resting membrane potential (i.e., 0-pA current injection). A current was then injected to hold the cells at −79 mV (for 5–10 s), and then current steps of 100 pA, 200 pA, 300 pA, 400 pA, and 500 pA (2-s duration) were injected with 3–5 s between different steps. None of the recorded calyces in the present study had spontaneous firing. The smallest of the five steps that generated an action potential (AP) was considered as the AP generation threshold. The latency of the first spike was measured as the time from the beginning of the step to the peak of first AP for the threshold step. Finally, the number of APs during the steps was also counted. Results were compared before and after application of drugs.

Data analysis.

Clampfit 10 (Molecular Devices), Prism (GraphPad), and MATLAB (The MathWorks) software were used for analysis of data. Results are reported as means ± SD. Paired t test was used for comparison between two parameters, and repeated measures two-way ANOVA with Bonferroni or Tukey post hoc test was used for comparisons of more than two conditions. Level of statistical significance was set at α = 0.05.

GABA_B activation suppressed voltage-gated currents.

To investigate whether GABA_B receptors are present on the calyx terminal and their possible effect on membrane properties of calyces, we applied R-baclofen, an agonist of GABA_B receptors, during calyx recordings. Baclofen application (100 µM) did not affect the resting membrane potential (control, −67.2 ± 1.4 mV; baclofen, −63.0 ± 8.1 mV; paired t test, P = 0.58), but resulted in a decrease in voltage-dependent currents during depolarizing steps in every recorded calyx (n = 15), suggesting the presence of GABA_B receptors on these afferent terminals (Fig. 1A) (response to the largest depolarizing step of +11 mV: control response, 2,466.3 ± 183.0 pA; baclofen, 1,633.0 ± 305.0 pA after baclofen application). For the population of recorded calyx terminals, application of baclofen resulted in an inhibition of currents for the more depolarized voltage steps greater than −29 mV (repeated measures ANOVA, P < 0.0001; Bonferroni post hoc test, P < 0.001 for −29 mV, −9 mV, and +11 mV). As expected, tail currents were also suppressed during baclofen application following depolarizing steps larger than −29 mV (repeated measures ANOVA, P = 0.0001; Bonferroni post hoc test, P < 0.01 for −29 mV and P < 0.001 for −9 mV and +11 mV).

The above-described changes were completely blocked in the presence of CGP 35348 (300 µM), a GABA_B antagonist. Figure 1C shows the average response for four calyx recordings in which voltage-sensitive currents were inhibited with baclofen application (repeated measures ANOVA, P < 0.001 for −9 mV and +11 mV steps). After baclofen was washed, CGP 35348 was applied, which by itself had no effect on the currents in the same calyces (repeated measures ANOVA, P > 0.05 for all steps). This was then followed by application of a combination of CGP 35348 with baclofen, which showed no effect (repeated measures ANOVA, P > 0.05 for all steps). These results strongly suggest that calyx terminals (at least in the central region of the cristae) have functional GABA_B receptors.

Kir channels are not affected by GABA_B activation in calyx terminals.

Previous studies have shown that GABA_B activation in different neurons and glial cells typically increases inward rectifying potassium (Kir) currents (O’Callaghan et al. 1996, Isomoto et al. 1997), including those mediated by Kir4.1 channels (Takeda et al. 2015). Kir4.1 channels are also expressed in calyx terminals (Udagawa et al. 2012), and it has also been proposed that, similarly to glial cells, they play a role in clearance of K⁺ from the closed synaptic space between type I hair cells and calyx terminals. Despite baclofen having a clear inhibitory effect on potassium currents at depolarized membrane potentials in the above-described experiments, we still tested whether GABA_B has any possible effect on Kir channel activity in the calyx. Voltage steps similar to those shown in Fig. 1 were used before and after baclofen application in the absence and presence of barium chloride (BaCl₂; 100 µM), a voltage-independent blocker of Kir channels (O’Callaghan et al. 1996). Addition of baclofen decreased the currents by ~38% in response to the more depolarized steps with or without BaCl₂ (e.g., for +11 mV, control currents were 1,977 ± 377 pA, decreased to 1,217 ± 247 pA after baclofen application, and stayed at 1,249 ± 344 pA after addition of BaCl₂) (n = 6, repeated measures ANOVA, Bonferroni multiple comparison test; control vs. baclofen: P < 0.01 for −29 mV and P < 0.001for −9 mV and +11 mV; control vs. baclofen + BaCl₂: P < 0.01 for −29 mV and P < 0.001for −9 mV and +11 mV; baclofen vs. baclofen + BaCl₂: P > 0.05 for all steps), suggesting that Kir channels were not affected by GABA_B activation (data not shown).

GABA_B inhibits Ca²⁺-sensitive K⁺ channels.

Two recent studies have shown GABA_B-mediated inhibition of BK (large current) potassium channels in the retinal ganglion cells (Garaycochea and Slaughter 2016) and SK (small current) channels in the substantia nigra (Estep et al. 2016). Since calyx terminals also express SK (Meredith et al. 2011) and probably BK (Limón et al. 2005) channels, we studied whether a similar GABA_B-mediated inhibition was also present in the calyx terminal. We first used a combination of antagonists for SK channel (apamin, 300 nM) and BK channel (iberiotoxin, 150 nM). As expected, the outward voltage-dependent currents (Fig. 2A) and tail currents (not shown) were both inhibited in the presence of apamin and iberiotoxin. For the population of recorded calyces (n = 6), the outward current was decreased by 16–26% (for steps larger than −69 mV) with application of apamin and iberiotoxin. The decrease in voltage-dependent currents was significant for the largest depolarizing steps (Fig. 2B). Addition of baclofen resulted in a nonsignificant decrease in currents by 10–20% (repeated measures ANOVA, P = 0.33). Similar effects were observed for tail currents at the end of voltage steps (not shown). Application of apamin (n = 5) or iberiotoxin (n = 4) alone also resulted in a decrease in voltage-dependent currents in response to the three largest steps (repeated measures ANOVA, P < 0.001). Addition of baclofen resulted in an extra 10–20% reduction [for 11-mV step: apamin + baclofen, 11% (re apamin alone); iberiotoxin + baclofen, 20% reduction (re iberiotoxin alone)]. The latter change was statistically nonsignificant compared with apamin or iberiotoxin alone (repeated measures ANOVA, P > 0.05 for all comparisons) but was significant compared with control conditions for the three largest steps (repeated measures ANOVA, P < 0.001). These results show that both SK and BK channels are expressed by the calyx and that both are involved in mediating the effect of baclofen.

A common pathway that could affect both of these potassium channels is inhibition of voltage-sensitive calcium channels. As a first step, to investigate whether the baclofen effect was due to inhibition of a voltage-sensitive calcium current, we added XE-991 (KCNQ channel blocker) to the cocktail to inhibit all major voltage-activated K⁺ currents in the calyx (response to +11-mV step: control, 2,977 ± 285 pA; apamin and iberiotoxin, 2,276 ± 338 pA; apamin and iberiotoxin and XE-991, 1,914 ± 410 pA). Again, baclofen resulted in a small decrease in the average current (n = 5; largest decrease of 163 ± 115 pA for +11-mV step; Fig. 2C). A similar decrease in currents was observed with addition of the nonspecific Ca²⁺ channel blocker cadmium chloride (CdCl₂; Fig. 2C), suggesting that baclofen inhibits voltage-sensitive Ca²⁺ channels. Furthermore, in the absence of calcium in the external solution (0 Ca²⁺), baclofen did not have any effect on the depolarizing steps and tail currents (Fig. 3A). Note that currents were decreased with 0 Ca²⁺ due to the inhibition of voltage-sensitive Ca²⁺ currents and Ca²⁺-activated potassium currents, but there was no further decrease with baclofen application (for +11 mV step: control, 2,901 ± 310; no calcium, 1,811 ± 25 pA; no calcium and CdCl₂, 1,424 ± 40 pA, addition of baclofen, 1,349 ± 135 pA). For the population of recordings in control external solution, 0-Ca²⁺ external solution_, and subsequent baclofen application (n = 7), voltage-sensitive currents and tail currents decreased in 0 Ca²⁺ compared with control, but there was no further change with baclofen application in 0 Ca²⁺ (Fig. 3B). In a second set of experiments in the control external solution, we applied CdCl₂ to block calcium channels (n = 4), which decreased the amplitude of currents during depolarizing steps as well as the tail currents and occluded the effect of baclofen (Fig. 3C). In three recordings, we used CdCl₂ with 0-Ca²⁺ external solution to completely block all sources of external calcium to the cell. This resulted in inhibition of voltage-activated and tail currents and complete block of the effect of baclofen (Fig. 3D). These results further emphasize the role of Ca²⁺-sensitive SK and BK channels in baclofen-mediated effects.

GABA_B activation has an excitatory effect on calyx terminals.

We next used current clamp to investigate the effect of GABA_B activation on firing of action potentials (AP). For all recordings, we injected currents (if necessary) to have an initial calyx membrane potential at −79 mV. We then injected current steps of 100–500 pA in 100-pA steps. In control condition (i.e., before application of drugs), none of the recorded calyces (n = 15) showed spontaneous firing of APs, and the threshold for generating an AP was reached by injecting 320 ± 147 pA. At threshold and more depolarized membrane potentials, all recorded calyces fired only a single AP at best, as previously reported (Ramakrishna et al. 2020; Songer and Eatock 2013). The latency of this single spike was 6.8 ± 2.1 ms for the threshold current. After baclofen application, calyces fired more than one spike during depolarization (Fig. 4A). On average, calyces generated more than one spike after baclofen application compared with control condition for 200-pA, 300-pA, 400-pA, and 500-pA steps (Fig. 4B). The amplitude of the current step required for spike generation also significantly decreased (control, 320 ± 147 pA; baclofen, 219 ± 122 pA; paired t test, P = 0.009; Fig. 4C), and action potential generation threshold decreased from −50 ± 1.37 mV in control condition to −58.3 ± 1.65 mV after baclofen application (paired t test, P < 0.0009). Furthermore, the responses became faster with GABA_B activation as quantified by the first-spike latency at the beginning of step depolarizations (Fig. 4D). For the threshold current, the latency showed a nonsignificant decrease (~2 ms) from 6.8 ± 2.1 ms to 5.0 ± 0.9 ms (paired t test, P = 0.12). However, the latency of the first spike decreased with increasing amplitude of step currents (for the control condition as well as in the presence of baclofen), and the difference between the two conditions was significant for the 400-pA and 500-pA steps (paired t test, P > 0.05 for 100- to 300-pA steps, P = 0.028 for 400-pA step, and P = 0.003 for 500-pA step).

Previous studies have shown that both SK and BK channels affect neuronal spike-frequency adaptation and action potential firing patterns in different brain areas and animal models (e.g., Deemyad et al. 2011, 2012; Gittis et al. 2010; Smith et al. 2002). Although we expected to see changes in firing patterns, it was surprising to see excitatory effects with GABA_B activation. Furthermore, while none of the recordings showed spontaneous AP firing in control condition, 8 of 15 recorded calyces in current clamp exhibited repeated firing of AP in the presence of baclofen (Fig. 4E). Note that the other two reports of excitatory GABA_B effects in retinal ganglion cells and substantia nigra neurons did not show such repetitive firing in the presence of baclofen (Estep et al. 2016; Garaycochea and Slaughter 2016).

Excitatory GABAergic activity in the vestibular periphery.

Typically, activation of the metabotropic GABA_B receptors results in an inhibitory effect. On the presynaptic side, this is mediated by inhibition of voltage-gated Ca²⁺ channels and a resultant decrease in synaptic vesicular release (reviewed in Kantamneni 2016). On the postsynaptic side, inhibition is achieved through activation of K⁺ channels. Only two previous studies have reported an excitatory GABA_B-mediated effect: 1) in retinal ganglion sensory neurons through inhibition of postsynaptic voltage-sensitive Ca²⁺ channels, resulting in inhibition of BK channels (Garaycochea and Slaughter 2016), and 2) in substantia nigra through inhibition of SK channels through a direct pathway involving adenylyl cyclase and protein kinase A (Estep et al. 2016). Notably, baclofen was most effective at membrane potentials between −29 mV and +11 mV (i.e., highest voltages tested), consistent with the peak activity range of voltage-sensitive calcium channels in the vestibular (Vincent et al. 2014) and cochlear (Goutman and Glowatzki 2007) periphery. Because SK and BK channels are both present in the calyx, it is inevitable that both get affected by lack of Ca²⁺ or inhibition of voltage-sensitive Ca²⁺ channels. Although in the present study we did not investigate the underlying pathway(s) for the GABA_B-mediated effect, either or both of the above-mentioned pathways in the retina and substantia nigra could be present in the calyx terminals. It should be noted that we only studied the effect of GABA_B on the two most likely candidates, SK and BK channels. However, it is possible that other Ca²⁺-dependent K⁺ channels are also expressed by the calyx terminals. For example, intermediate-conductance K⁺ channels (K_Ca3.1) or K_Ca4.1 channels could theoretically be present in the calyx. The above-mentioned possibilities could explain the further (albeit nonsignificant) decrease in K⁺ currents with addition of CdCl₂ after block of SK/BK channels. However, this further decrease could also be due to incomplete block of SK/BK channels by antagonists.

Finally, we did not investigate whether GABA_B activation had any effects on sodium channels. Such effects are particularly likely since calyces might have persistent sodium currents (Liu et al. 2016), and a recent study (Li et al. 2017) has shown that baclofen could inhibit persistent sodium currents and subsequently inhibit a Na⁺-dependent K⁺ current in rat olfactory bulb mitral cells. Such effects could result in an excitatory effect, depending on the balance between inhibition of the sodium and potassium currents.

Source of GABA in the vestibular periphery.

Although previous studies have found GABA_A subunits on afferents in rats, mice, and hamsters (Foster et al. 1995; Kitahara et al. 1994) and our results provide ample evidence for the presence of GABA_B receptors on calyx terminals, a source of GABA in the vestibular periphery has not been unequivocally identified. There is evidence for existence of GABAergic efferent fibers and terminals in the peripheral vestibular neuroepithelium in mouse (Kong et al. 2002a), rats (Kong et al. 1998b; Matsubara et al. 1995), squirrel monkey (Usami et al. 1987), and humans (Kong et al. 1998a, 2002b; Schrott-Fischer et al. 2007). However, the main vestibular efferent cell group in the brain stem (near the genu of the facial nerve) was not labeled with anti-GABA antibodies (Perachio and Kevetter 1989). GABA has also been shown to be present in some of the hair cells in toadfish (Holstein et al. 2004) and guinea pigs (López et al. 1990).

Recent studies have suggested supporting cells as a source of GABA in the vestibular periphery. One study used GAD67-GFP or GAD65-GFP mice that express green fluorescent protein (GFP) along with the two isoforms of the enzyme glutamate decarboxylase (GAD), which transform glutamate to GABA (Tavazzani et al. 2014). The results of this study showed that only GAD67 was expressed in vestibular end organs and only in supporting cells in the peripheral zones of the cristae. Since GAD67 is responsible for generating cytoplasmic GABA (compared with GAD65 for vesicular GABA) (Martin and Rimvall 1993; Soghomonian and Martin 1998), supporting cells could have nonvesicular release of GABA similarly to glial cells (Attwell et al. 1993; Héja et al. 2009; Wu et al. 2007). Interestingly, with the use of anti-GABA antibodies, this study found that GABA was present in the supporting cells in both central and peripheral regions of the cristae. Another recent study used GAD2-tdTomato mice, which expressed red fluorescence with GAD2 and found fluorescence in both hair cells and supporting cells in the peripheral regions of the cristae and maculae (Holman et al. 2019). However, presence of GAD2 does not necessarily indicate generation of GABA in hair cells, and it is unlikely that GABA is an afferent neurotransmitter as AMPA receptor blockers suppress all synaptic events in calyx terminals (Sadeghi et al. 2014). Furthermore, GABA antibodies in the mouse failed to label hair cells (Tavazzani et al. 2014).

From all the above-described previous studies, the most likely sources for GABA in the vestibular periphery seem to be supporting cells, efferent fibers, and, possibly, some of the hair cells.

Function of GABA_B activation in the vestibular periphery.

The firing properties of afferents are decided by presynaptic as well as postsynaptic factors. Typically, vesicular release properties of the hair cell such as number of released vesicles, quantal size, and probability of release are the main deciding presynaptic elements. A previous study has shown an excitatory GABA_B activation effect on type I hair cells in guinea pigs (Lapeyre et al. 1993). Here, we did not set up for proper measurement of such effects and cannot rule out similar excitatory effects in rat hair cells. Furthermore, depolarization of type I hair cells could also affect the nonquantal transmission to the calyx terminals in this unique synapse and result in further depolarization of the calyces (Contini et al. 2017; Lim et al. 2011). In the present study, we did not observe any effect on resting membrane potential with baclofen application, thus ruling out the latter effect. On the postsynaptic side, the number and type of glutamate receptors as well as the calyx membrane properties are the main factors affecting the firing properties of the calyx. Here, we propose that modulation of calcium-dependent potassium channels through activation of GABA_B receptors results in a change in firing properties of calyces as observed in their responses to step depolarizations, including a change in action potential threshold, which will affect calyx responses to presynaptic inputs and could partly explain the repetitive firing observed during depolarizing steps as well as with resting membrane potential.

Presence of different types of potassium (e.g., SK, BK, KCNQ, Kir, etc.) channels in calyx terminals decreases their membrane resistances, making it difficult to depolarize the calyx. We recently showed that muscarinic acetylcholine receptors (mAChR) inhibit KCNQ channels in calyx terminals, resulting in similar excitatory effects (Ramakrishna et al. 2020), supporting the findings of previous studies in turtle (Holt et al. 2017). Calcium currents as well as potassium currents (here, SK and BK channels) are involved in the hyperpolarization phase of action potentials (including the after hyperpolarization) and have been shown to affect the spike-frequency adaptation, firing patterns, and filtering properties of the membrane in other brain areas and animal models (e.g., Beraneck and Straka 2011; Deemyad et al. 2011, 2012; Gittis et al. 2010; Smith et al. 2002). Interestingly, calyces in the central regions of the in vitro preparation of the vestibular neuroepithelium can only fire a single AP in response to a step depolarization. This could be due to the fact that all (e.g., KCNQ, SK, BK) potassium channels expressed by the calyx are fully active in the absence of spontaneous efferent inputs (i.e., absence of or decreased acetylcholine and GABA), which is the case in the in vitro preparation since efferent fibers are cut and separated from their cell bodies in the brain stem. This notion is further supported by our recent in vivo results that showed suppression of efferents results in a decrease or complete shutdown of mainly irregular afferents in mice (Raghu et al. 2019). Note that the low release rates from hair cells onto the calyx most likely also contribute to the lack of spontaneous firing; however, under the same conditions, changing the membrane properties of the calyx resulted in repetitive (spontaneous) firing of action potentials in some of our recordings.

A second function for GABA_B receptors can be a gain control mechanism. Inhibition of SK/BK channels and the resulting increase in the number of APs during step depolarizations is akin to an increase in the response gain. Similar increases in the spontaneous firing and response gain have been observed in the vestibular nuclei following inhibition of BK channels (Nelson et al. 2005, 2017; Smith et al. 2002). In addition to direct gain control, changes in the activity of potassium channels affect membrane filtering properties and its resonance at subthreshold potentials (Beraneck and Straka 2011).

Interaction between cholinergic efferents and GABAergic cells.

Whether and how cholinergic and GABAergic inputs interact in the vestibular periphery remain to be explored. Recently, it was shown that supporting cells that contain GAD2 are activated by acetylcholine in the vestibular periphery (Holman et al. 2019). A recent study has shown that GABA_B activation inhibits cholinergic efferent terminals in contact with inner and outer hair cells (Wedemeyer et al. 2013). It is possible that a similar effect is present in efferents that synapse onto type II hair cells, which are similar to cochlear efferents in that they hyperpolarize type II hair cells through a9-containing receptors (Parks et al. 2017; Poppi et al. 2018; Yu et al. 2020). However, an interaction between the two inputs seems unlikely for the efferent-calyx synapse, since cholinergic efferents have an excitatory effect on calyces (Holt et al. 2015, 2017; Lee et al. 2017; Ramakrishna et al. 2020), and therefore, GABAergic inhibition of these efferents would not have resulted in an excitatory effect on calyces.

Conclusion.

Overall, our results provide evidence for an unusual excitatory effect of GABA_B receptors on calyx terminals, which modulates their resting discharge, sensitivity, and membrane properties. This could potentially affect the response properties of irregular afferents, which play a major role in carrying information about fast head movements. The expression of other GABA receptors by the calyx as well as the presence of any GABA-mediated responses in type II hair cells and bouton afferent terminals remains to be explored. Our findings question the traditional notion that only central vestibular pathways are modulated during adaptation and compensation and suggest that GABAergic inputs along with cholinergic efferent inputs can provide the means for continuous fast adjustments of response properties of the vestibular periphery.