Abstract
Amniotes evolved a unique postsynaptic terminal in the inner ear vestibular organs called the calyx that receives both quantal and nonquantal (NQ) synaptic inputs from Type I sensory hair cells. The nonquantal synaptic current includes an ultrafast component that has been hypothesized to underlie the exceptionally high synchronization index (vector strength) of vestibular afferent neurons in response to sound and vibration. Here, we present three lines of evidence supporting the hypothesis that nonquantal transmission is responsible for synchronized vestibular action potentials of short latency in the guinea pig utricle of either sex. First, synchronized vestibular nerve responses are unchanged after administration of the AMPA receptor antagonist CNQX, while auditory nerve responses are completely abolished. Second, stimulus evoked vestibular nerve compound action potentials (vCAP) are shown to occur without measurable synaptic delay and three times shorter than the latency of auditory nerve compound action potentials (cCAP), relative to the generation of extracellular receptor potentials. Third, paired-pulse stimuli designed to deplete the readily releasable pool (RRP) of synaptic vesicles in hair cells reveal forward masking in guinea pig auditory cCAPs, but a complete lack of forward masking in vestibular vCAPs. Results support the conclusion that the fast component of nonquantal transmission at calyceal synapses is indefatigable and responsible for ultrafast responses of vestibular organs evoked by transient stimuli.
SIGNIFICANCE STATEMENT The mammalian vestibular system drives some of the fastest reflex pathways in the nervous system, ensuring stable gaze and postural control for locomotion on land. To achieve this, terrestrial amniotes evolved a large, unique calyx afferent terminal which completely envelopes one or more presynaptic vestibular hair cells, which transmits mechanosensory signals mediated by quantal and nonquantal (NQ) synaptic transmission. We present several lines of evidence in the guinea pig which reveals the most sensitive vestibular afferents are remarkably fast, much faster than their auditory nerve counterparts. Here, we present neurophysiological and pharmacological evidence that demonstrates this vestibular speed advantage arises from ultrafast NQ electrical synaptic transmission from Type I hair cells to their calyx partners.
Introduction
The vestibular system evolved over 400 million years ago in primitive fish (Higuchi et al., 2019), and is the evolutionary precursor to the modern mammalian cochlea (Manley, 2012). The five peripheral vestibular sensory organs detect linear and angular accelerations in three-dimensional space (Goldberg et al., 2012; Curthoys et al., 2017). Fibers from all sense organs travel to the brainstem and cerebellum and terminate in the vestibular nuclei, where secondary neurons project to sensory and motor systems, contributing to several levels of nervous function, including some of the fastest reflexes in biology designed to maintain visual and postural stability during dynamic head and body movements.
Collectively, primary vestibular afferent neurons provide the central nervous system with information spanning a broad frequency bandwidth and dynamic range, with individual neurons specialized to encode specific directional and temporal features of diverse gravito-inertial stimuli. Afferent fibers innervating the utricle, precisely encode the timing of transient inertial stimuli, and generate precisely timed action potentials to sinusoidal stimulation (McCue and Guinan, 1994; Curthoys et al., 2016) up to ∼3 kHz with a temporal fidelity exceeding that of auditory spiral ganglion neurons in the same species (Palmer and Russell, 1986; Curthoys et al., 2019). These sensitive vestibular neurons make calyceal synaptic contacts on Type I hair cells in the striolar region of the sensory epithelium (Curthoys et al., 2006; Lysakowski et al., 2011; Goldberg et al., 2012). The postsynaptic calyceal terminal responsible for this high-fidelity transmission is unique to amniote vestibular organs and supports three distinct forms of neurotransmission: (1) glutamatergic quantal transmission analogous to that responsible for synaptic transmission of sound information between Type I spiral ganglion neurons and cochlear inner hair cells (Matsubara et al., 1999; Ruel et al., 1999; Rennie and Streeter, 2006; Highstein et al., 2015; Rutherford et al., 2021); (2) K+ build up in the synaptic cleft responsible for a slow nonquantal (NQ) synaptic current (Holt et al., 2007; Lim et al., 2011; Highstein et al., 2014; Contini et al., 2017, 2020; Govindaraju et al., 2023); and (3) direct electrical coupling between the hair cell and the calyx terminal responsible for an ultrafast NQ current (Songer and Eatock, 2013; Contini et al., 2020). Here, we examine the hypothesis that the ultrafast electrical NQ current is responsible for synchronized short latency responses to transient stimuli in the guinea pig utricle.
Ultrafast electrical transmission between the Type I hair cell and the calyx terminal has been reported for calyceal synapses in turtle posterior semicircular canal based on paired presynaptic and postsynaptic recordings (Contini et al., 2020), and this is consistent with recordings in immature postnatal (P1–P4) rat saccular calyceal terminals, suggesting the presence of an ultrafast NQ component (Songer and Eatock, 2013). We hypothesized that the same NQ current is present in adult mammals, and underlies short latency neural responses, including vestibular stimulus evoked potentials (VsEPs) commonly used for vestibular phenotypic screening in small amniotes (Jones and Jones, 1999; Jones and Lee, 2021). VsEPs are dominated by vestibular compound action potentials (vCAP) generated by synchronized firing of many vestibular afferent neurons (Goldstein and Kiang, 1958; Brown et al., 2017; Pastras et al., 2023a,b).
We first confirmed the possibility of fast, nonquantal transmission by perfusing glutamate (AMPA) blockers into the inner ear of our guinea pig model and comparing the effects on vestibular and cochlear synaptic transduction. Second, we examined and compared the latency of vCAPs to cCAPs recorded from the guinea pig. Finally, we examined the effects of a forward masking paradigm, in which paired stimuli are presented in short succession with the interval between the pulses controlled [paired pulse interval (PPI)]. Here, we aimed to examine the presence of the classically described phenomenon of forward masking (Goldstein and Kiang, 1958) present in the cochlea, which is thought to arise primarily from the kinetics of depletion/replenishment of the readily releasable pool (RRP) of synaptic vesicles within inner hair cells (Peterson et al., 2014).
Materials and Methods
Animal preparations and surgery
Experiments were performed on 25 adult tri-colored guinea pigs (Cavia porcellus) of either sex, weighing between 300 and 500 g. All procedures were approved by the University of Sydney Animal Ethics Committee (protocol #2019/1533) and were conducted in accordance with the Society for Neuroscience Ethics Policy for Use of Animals and Humans in Neuroscience Research. All animals had a positive Preyer's reflex indicating good hearing. This was further confirmed via inspection of the middle ear cavity under the surgical microscope, and later via auditory cCAP thresholds. Before surgeries, guinea pigs received preanesthetic medications of Atropine sulfate (600 µg in 1 ml, Pfizer), and Buprenorphine HCl (Temgesic, 300 µg in 1 ml, Indivior). Thereafter, guinea pigs were transferred to a Perspex induction chamber and were anaesthetized using 2–4% isoflurane (99.9% isoflurane liquid inhalation; Henry Schein Medical). Once lacking a foot-withdrawal reflex, animals were transported to the surgical table to be tracheotomized and artificially ventilated with a mixture of oxygen and isoflurane (2%). Heart rate and blood oxygenation (SpO2) levels were continuously monitored (Nonin Medical Inc.), with the animal's core temperature regulated via a heating pad (Kent Scientific). Animals were then mounted between custom-made ear bars and were electrically grounded via a Ag/AgCl electrode placed in the neck musculature.
Experimental design
This study tested the hypothesis that vestibular calyx afferents have a different mode of synaptic transmission than cochlear spiral ganglion afferents to transient stimuli. Specifically, we hypothesized that calyx afferents transmit synchronized APs via nonquantal transmission, unlike cochlear spiral ganglion afferents, which transmit compound APs via glutamatergic quantal transmission. This hypothesis was tested using three main experimental paradigms. (1) By examining the effects of pharmacological blockade of glutamate (AMPA) receptors on cochlear versus vestibular neural potentials in the guinea pig inner ear. (2) By quantifying latency differences between hair cell and neural potentials in the cochlea versus vestibular system of the guinea pig. (3) By measuring cochlear versus vestibular nerve responses with changes following a paired pulse interval (PPI) paradigm against refractory characteristics of the nerve. To measure these differences, we used compound field potentials recorded near the cochlear or vestibular afferent neurons and hair cells, known as the compound action potential and microphonic microphonic potential, respectively. These inner ear compound field potential responses were chosen as they are robust allowing for controlled experimental perturbations over long time periods, with relatively low intra-animal (<±10% SD) and interanimal (±10–30% SD) variability. A power analysis using a two-sample t test (power, 1-β = 0.8; α = 0.05) estimated a minimum sample size of five animals for each experimental paradigm. Final numbers exceeded five total observations for each paradigm, respectively.
Stimuli and recordings
All stimuli and responses were generated and recorded using customized LabVIEW programs (National Instruments). Stimuli were produced using an external soundcard (SoundblasterX7, Creative Inc.). Physiologic responses were amplified by 10,000× (80 dB) with a 1-Hz to 10-kHz bandpass filter (IsoDAM, WPI), before being digitized at 40 kHz and 16-bit using an analog-to-digital converter (NI9205, National Instruments). Physiologic and acceleration responses were averaged using 100 stimulus presentations using a running (or moving) average in LabVIEW, which approximately improved signal-to-noise ratio (SNR) in proportion to the square root of the number of measurements.
To record neural responses from the vestibular system, the dorsolateral bulla was opened, and a noninverting Ag/AgCl electrode was inserted into the bony facial nerve canal near the superior branch of the vestibular nerve. The inverting reference electrode was inserted nearby in the neck musculature. Auditory potentials such as the cochlear nerve CAP and cochlear microphonic were recorded from the round window using air-conducted sound (ACS) or bone-conducted vibration (BCV) pulses. Linear acceleration transients were delivered to the skull via BCV pulses using a Brüel & Kjær minishaker (type 4810), which was rigidly coupled to the contralateral ear-bar. A three-axis accelerometer (830M1; flat frequency response >15 kHz; TE Connectivity) was secured to the ear-bar and was used to quantify input stimuli in G (1 G = 9.81 m/s2).
Cochlear ablation and utricular exposure
The cochlea was accessed via a ventral surgical approach, as previously described (Pastras et al., 2017, 2018a,b ,2020, 2021; Fig. 1D). Fine-tip forceps, a hooked-needle, and scalpel were used to surgically ablate the cochlea, starting at the apex, and moving to the base. Tissue wicks were inserted to absorb fluid and blood following ablation. Cochlear tissue and bone were cleared to expose the basal surface of the utricular macula (Figs. 1E). Tissue wicks were implanted adjacent to the macula to control fluid build-up on the epithelium.
Application of 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX)
A 10 mm stock solution of CNQX was first produced by dissolving 50 mg of CNQX disodium salt (molecular weight = 276.12, ab120044, Abcam) in 18 ml of artificial perilymph (130 NaCl, 4 KCl, 1 MgCl2, and 2 CaCl2; pH 7.3; osmolarity 320 mOsmol). Stock solutions were stored as aliquots in tightly sealed vials at −20°C, as per supplier guidelines. Aliquots were used within two weeks of freezing, before the recommended useable period of one month. Before use, and before opening the vial, solutions were allowed to equilibrate to room temperature for ∼1 h. Thereafter, stock solutions were diluted to 100 μm in fresh artificial perilymph. Approximately 0.1 ml of 100 μm CNQX/Artificial perilymph solution was drawn up in a 1-ml syringe with a 27-gauge Luer lock needle. Two to three drops of 100 μm CNQX were placed onto the round window or above the perilymphatic layer of the utricular macula under the visual guidance of an operating microscope (corresponding to the approximate location denoted by the asterisk symbol in Fig. 1D). CAP responses were monitored before, during and after application of the drug.
Electrical stimulation of vestibular neural targets
Central stimulation of the peripheral afferents was chosen over peripheral stimulation, to avoid possible complexities in interpretation of the electrically evoked vestibular compound action potential (evCAP). For example, electrical stimulation of the periphery can result in multiple activation sites of the primary afferent neuron, such as its peripheral axon, soma, or along its central axon, which has been documented in the cochlea (Stypulkowski and Van den Honert, 1984). Moreover, electrodes placed in the periphery can evoke electrophonic responses that mimic fiber activation, which arise from the sensory hair cells (Moxon, 1971; Van den Honert and Stypulkowski, 1984). Third, central vestibular stimulation evoking antidromic electrically evoked vCAPs have been shown to fully cancel the orthodromic BCV vCAP (Pastras et al., 2018b), which is evidence that the antidromic electrically evoked vCAP arises from the same neurons as the BCV vCAP. This can be explained by the electrically evoked antidromic action potential colliding with, and cancelling, the vibration-evoked orthodromic action potential in the same afferents. Hence, by stimulating the vestibular afferents at their proximal ending in the central vestibular system, we were confident that the evCAP arose from the same neural targets that generate the BCV-evoked vCAP.
To electrically stimulate the proximal endings of the primary vestibular afferents at the central vestibular system, surgery was undertaken to expose the floor of the fourth ventricle for placement of bipolar stimulating electrodes. Specifically, a small incision was made between the caudal point of the occipital bone and λ. Thereafter, a posterior craniotomy was performed, and the dura mater was cut. A section of the flocculus and paraflocculus were aspirated to fully expose the floor of the fourth ventricle. Fluid build-up was controlled with tissue wicks. A pair of platinum bipolar electrodes (1.0 MΩ impedance, 400 µm in diameter) coated with a parylene-C layer (Microprobes) were used for stimulation of vestibular neural targets. Stimulating electrodes were positioned lateral of the floor of the fourth ventricle under the guidance of a surgical microscope using a 3-axis clamp mount micromanipulator (MM-33, ADInstuments). Electrode placement was based on guinea pig stereotaxic map coordinates (Rapisarda and Bacchelli, 1977; Voitenko and Marlinsky, 1993) and guinea pig immunohistochemistry studies (Motts et al., 2008). The bipolar electrodes were positioned at the same lateral-medial plane as the crossed olivocochlear bundle, at the midline floor of the fourth ventricle, and positioned lateral of the facial nerve genu beneath the sulcus limitans (Pastras et al., 2018b). Low threshold current twitches indicated stimulation of the facial nerve genu or abducens nucleus. To locate appropriate vestibular afferent targets, subsequent dorsolateral electrode adjustment/positioning was required. Final positioning of the bipolar electrodes dorsolateral of the midline, was determined by the maximal physiological effect on the vCAP with the lowest shock intensity, analogous to previous work (Pastras et al., 2018b). Current shocks (bipolar, 100-μs biphasic pulse width) were delivered via an isolated, biphasic current stimulator (Model DS4, Digitimer Ltd.).
Statistical analysis
Differences in field potential responses were assessed using univariate and multivariate repeated measures ANOVA, using IBM SPSS Statistics 23 software. To further explore statistical effects, post hoc tests were conducted through pair-wise comparisons, including two-sample t tests, assuming equal variances. Reported values are presented as mean ± SD, with statistical significance defined as p-value < 0.05.
Data availability
The datasets used during the current study, as well as the code for data acquisition and analysis are available from the corresponding author on reasonable request.
Results
Experiments were undertaken in adult guinea pigs comparing robust functional cochlear and vestibular responses to examine differences in synaptic transmission modes.
We present evidence that short latency vCAPs evoked by transient vibration in the guinea pig utricle arise from the ultrafast electrical component of NQ synaptic transmission at calyceal synapses. We begin with pharmacological evidence, followed with evidence based on timing, and finally with evidence based on forward temporal masking experiments.
Effect of CNQX on cochlear versus vestibular CAPs
To assess the existence of a nonglutamatergic form of synaptic transmission in the vestibular system, we applied the glutamate (AMPA) antagonist CNQX (100 μm in artificial perilymph) to the utricular macula (Fig. 1D), as well as to the cochlear round window (Fig. 1A). Here, opposite ears of the same animal were used (Fig. 1). While CNQX abolished any detectable cCAP response within 10 min of drug application (Fig. 1B,C), vCAPs persisted even after prolonged CNQX perfusion (Fig. 1E,F). Importantly, auditory hair cell responses (cochlear summating potentials) were unaffected by CNQX perfusion into the inner ear, demonstrating CNQX blocked neurotransmission between the presynaptic and postsynaptic junction, and effects on the cCAP were not because of presynaptic dysfunction of the hair cells (Fig. 1B). vCAP and cCAP input-output growth functions also revealed no relation between stimulus intensity and CNQX mediated effects on the neuroepithelia (Fig. 1B,E).
Effect of 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) on cochlear versus vestibular CAPs in the guinea pig. CNQX (100 μm in artificial perilymph) was applied to the (A) cochlear round window, and (D) the utricle via the macular epithelium (approximate drug delivery locations denoted by asterisks). B, E, Averaged cochlear compound action potential (cCAP) and vestibular compound action potential (vCAP) waveforms during input-output response recording over a 40- and 20-dB stimulus range, respectively, before and after CNQX administration (100 stimulus presentations). CNQX blocks auditory cCAPs (blue) but not the presynaptic hair cell response, the “summating potential” (black). CNQX had no effect on the vCAP across all stimulus intensities delivered. C, F, Time chart of cCAP and vCAP amplitudes normalized as a percentage of baseline. The cCAP was diminished while vCAPs remained unperturbed demonstrating differential effects of CNQX on cochlear versus vestibular afferent function. Abbreviations: RW = round window; FNC = facial nerve canal; hSCC = horizontal semicircular canal; Stapes FP = stapes footplate; SO = saccular otoconia; CM = cochlear microphonic; cCAP = cochlear compound action potential; vCAP = vestibular compound action potential.
Vestibular versus auditory response latency
Next, we compared latencies of the vestibular and cochlear evoked potentials (vCAPs and cCAPs), using the onset of hair cell responses [vestibular microphonic (VM) or cochlear microphonic (CM)] as a temporal reference point (Fig. 2B,C). Given the line of polarity reversal in the utricle, vestibular microphonic field potentials undergo response cancellation, and therefore functional measurements from utricular hair cells require local glass microelectrode recordings from the macula surface (Fig. 2B). Vestibular nerve CAP responses can be recorded simultaneously with VMs, via an electrode sealed into the facial nerve canal, near the superior vestibular nerve branch. By contrast, auditory hair cells do not have a reversal line, and summate to produce large extracellular field potentials, which can be recorded alongside the cochlear nerve CAP from the round window, which are separated by latency and polarity inversion (Fig. 2C). Both responses were evoked by a 0.6-ms condensation (COND +) or rarefaction (RARE −) Gaussian ACS pulse, with the polarity of the stimulus resulting in significant differences in the latency of the vCAP, because of push-pull units either side of the line of reversal (McCue and Guinan, 1994). Results reveal that condensation and rarefaction ACS pulses produced a latency of 0.354 msCOND (±37 µs) and 0.349 msRARE (±61 µs) between the vCAP and VM (Fig. 2D, N = 5), suggesting an average delay of 0.35 ms to peak action potential voltage in the guinea pig utricle. In contrast, the latency of the cCAP relative to the CM for a condensation or rarefaction ACS pulse was 1.347 msCOND (±59 µs) and 1.280 msRARE (±60 µs) respectively (N = 5), which is an overall difference of 0.994 msCOND (±56 µs) and 0.927 msRARE (±48 µs) compared with the latency of the vCAP. Post hoc statistical analysis between vestibular and cochlear response latencies for both COND (+) and RARE (−) stimuli revealed both vCAP-VM and cCAP-CM latencies are significantly different to one another. Results from a two-sample t testCOND (equal variances) are as follows: observations = 10, df = 8, pooled variance = 0.002429, p-value (P(T<=t) one-tail: 5.10044E-10, absolute value t statistic: 31.88, t critical value one-tail: 1.859; t testRARE (equal variances): observations = 10 total, df = 8, pooled variance = 0.00362, p-value (P(T<=t) one-tail: 4.3E-09, absolute value t statistic: 24.362, t critical value one-tail: 2.306.
vCAP and cCAP response latencies. A, Schematic cross-section of the sensory regions of the utricle and the cochlea highlighting relevant sensory generators for latency measurement between hair cell and afferent responses. B, Simultaneous recordings of the VM and vCAP from a representative animal taken from the surface of the utricular macula and the facial nerve canal, respectively. The latencyCOND difference between the onset of the VM (first peak = P1) and the onset of the vCAP (first peak = N1) is 0.35 ms. C, Recordings of gross extracellular cochlear potentials from the same animal, in the opposite ear, with an early latency cochlear microphonic (CM) and a longer latency cCAP response, recorded from the round window. The latencyCOND difference between the CM (onset peak = P1) and cCAP (onset peak = N1) is 1.35 ms. Note: Both vestibular and cochlear responses in B and C were evoked by an identical 0.6-ms air-conducted sound (ACS) pulse, with a 0.3-ms rise-fall time at a stimulus intensity 20 dB above threshold. D, Hair cell to afferent latency comparisons across five animals. Comparison of vestibular and cochlear response latencies revealed statistically significant differences [t testCOND (equal variances): p-value (P(T<=t) one-tail: 5.10044E-10; t testRARE: p-value (P(T<=t) one-tail: 4.3E-09]. Average vCAP-VM latencies were 0.354 msCOND (±37 µs) and 0.349 msRARE (±61 µs), while average cCAP-CM latencies were 1.347 msCOND (±59 µs) and 1.276 msRARE (±60 µs), respectively. cCAP = cochlear compound action potential; CM = cochlear microphonic; COND = condensation; RARE = rarefaction; vCAP = vestibular compound action potential; VM = vestibular microphonic.
Forward masking
We employed a forward masking paradigm to determine whether the short latency vCAP could be reduced by depleting the RRP of synaptic vesicles in hair cells. Experiments followed the approach illustrated in Figure 3A, where brief interaural acceleration stimuli were applied to the temporal bone (minishaker), to simultaneously evoke vCAP and VM responses. vCAPs primarily reflect synchronized firing of many vestibular afferent fibers, while VMs primarily reflect mechano-electrical transduction (MET) currents at the electrode tip near the striola. The stimuli consisted of two pulses separated in time by the paired pulse interval (PPI). In the auditory nerve, which exhibits forward masking, the magnitude of the response to the second pulse is reduced as the PPI becomes small (Goldstein and Kiang, 1958). Example averaged vCAP and VM data are shown in Figure 3B, using a 7-ms PPI. Shorter PPIs result in the averaged vCAP decreasing in amplitude. A plot of vCAP and VM amplitude versus PPI for this animal is shown in Figure 3C, for PPIs ranging from 3.7 ms to 107 ms. The VM amplitude was completely unchanged by the PPI, analogous to results in the cochlea with the CM. In contrast, the vCAP amplitude declined for PPIs <7 ms, as would be expected as the second neural response falls within the relative refractory period of the action potential evoked by the first pulse.
Effects of paired pulse interval (PPI) stimuli on vestibular hair cell versus afferent function. A, Bone-conducted vibration pulses were delivered to the guinea pig temporal bone via a minishaker, simultaneously evoking neural vestibular compound action potentials (vCAPs) and transient hair cell potentials called vestibular microphonics (VMs). Vestibular PPIs were varied between 107 and 3.7 ms. B, Averaged VM and vCAP waveforms to 100 stimulus presentations for PPI stimuli of 7 ms. C, VM and vCAP PPI response curves with amplitudes normalized as a percentage of baseline. Averaged vCAPs were insensitive to changes in PPI between 107 and 10 ms yet decreased with PPIs below ∼8 ms. The VM amplitude was insensitive to changes across all PPIs.
We further examined the refractory period of sensitive calyx bearing afferents using paired electrical stimulation. For this, electrically evoked compound action potentials (evCAPs) were generated using the paired pulse antidromic electrical stimuli (Ramekers et al., 2015) as illustrated in Figure 4A. The antidromic evCAP response includes a saturating nonlinearity as a function of increasing shock intensity (Fig. 4B), consistent with the sensitivity and input-output function of peripheral vestibular afferents to increasing mechanical and electrical stimuli (Pastras et al., 2018b). The relative refractory period of the primary calyx afferents was then characterized by varying the PPI of electrical shocks between 15 and 2 ms (Fig. 4C) and monitoring the amplitude of the second evCAP response. Results demonstrated the second evCAP response amplitude exponentially declined with PPIs <8 ms.
Refractory characteristics of vestibular primary afferents. A, Electrically evoked vestibular compound action potentials (evCAPs) were used to probe refractory properties of vestibular primary afferents. Bipolar current pulses were delivered to the proximal throughputs of the primary vestibular afferents at the central vestibular system, to evoke antidromic evCAPs, recorded near the peripheral vestibular nerve branch. B, The input-output function of the antidromic evCAP displays a saturating nonlinearity, whereas the early-latency electrical artifact is linear with increasing current intensities. C, To determine the relative refractory properties of the evCAP, the interval between successive 100-µs biphasic current pulses or the electrical paired pulse interval was varied between 20 and 2 ms.
A, Images taken during surgery of the guinea pig utricle and cochlea in a representative animal. B, The mammalian vestibular system (utricle) contains Type I VHCs surrounded by unique chalice shaped calyx afferents, which synchronously fire action potentials to the onset of motion, producing vestibular nerve compound action potential (vCAP) responses. The mammalian cochlea contains inner hair cells which innervate multiple Type-I SG afferents, which evoke gross cochlear nerve compound action potential (cCAP) responses to the onset of acoustic stimuli. C, Example paired pulse interval (PPI) paradigms for 100- and 7-ms intervals between stimuli. D, Forward masking of vCAPs versus cCAPs. Normalized amplitude of mechanically evoked paired-pulse vCAPs (red) versus mechanically evoked paired-pulse cCAPs (blue) versus electrically evoked vCAPs (black). E, Example records showing lack of forward masking in vCAPs and significant forward masking in cCAPs for a PPI of 8 ms. F, vCAP input-output functions were measured to inform appropriate stimuli for (G) vCAP PPI intensity curve testing. Mechanically evoked paired pulse vCAPs recorded at four different BCV stimulus levels, and vCAP normalized to max. amplitudes displayed in inset revealed no forward masking associated with increased stimulus levels. H, Corresponding vCAP responses associated with 7.0- and 0.8-mG PPI intensity curves in G. Acceleration units are given in milli-G; where 1 G = 9.81 m·s−2.
We then compared vCAPs and cCAPs evoked by paired-pulse stimuli (Fig. 5; ∼15–20 dB above threshold) to determine whether forward masking shown previously in auditory spiral ganglion neurons is also present in vestibular afferent neurons. Note, PPI response curves in both the cochlea and vestibular system were identical for both BCV (bone-conducted vibration) or air-conducted sound (ACS) pulses. The surgical approach is illustrated Figure 5A, and example vCAP and cCAP signals are shown in Figure 5B. BCV-evoked vestibular (vCAP, red), electrically evoked vestibular (evCAP, black), and ACS-evoked cochlear (cCAP, blue) response amplitudes (N1-P1) are shown in Figure 5D for paired pulse stimuli with PPIs varied from 3.7 to 107 ms. Auditory cCAPs exhibited forward masking for PPIs <80 ms, while vestibular responses, whether evoked electrically or via BCV, only started to decline in magnitude only for PPIs <8–10 ms, which is almost an order of magnitude less. This difference is highlighted in Figure 5E, where vCAP and cCAP response waveforms to 8- and 100-ms PPI stimuli are presented (Fig. 5E).
Finally, previous work in the cochlea has shown that auditory forward masking varies as a function of stimulus intensity, with greater forward masking occurring at higher stimulation levels (Harris and Dallos, 1979; Li et al., 2021), putatively because of increased depletion of the RRP (Eggermont, 2015). To investigate this effect, vCAP PPI response curves were recorded at four different BCV pulse stimulus intensities, corresponding to a broad section of the vCAP input-output growth function (Fig. 5F), just above threshold (0.8 mG) to saturating intensities (7 mG). Results reveal that stimulus intensity had no impact on vCAPs forward masking (Fig. 5G,H).
Discussion
It has been suggested that evolution of the vestibular calyx synapse in amniotes was driven by pressure to increase the speed of locomotor and reflex neural circuits, which depend heavily on the speed of inertial sensation by inner ear vestibular organs (Eatock, 2018). The present report examined the latency between brief transient acceleration stimuli and action potentials evoked in sensitive calyx bearing afferent neurons in the guinea pig utricle. Results demonstrate the most sensitive vestibular afferents are remarkably fast, much faster than their auditory nerve counterparts. Here, we present evidence that the vestibular speed advantage arises from ultrafast NQ electrical synaptic transmission from Type I hair cells to their calyx partners (Contini et al., 2020). Direct electrical synaptic transmission improves speed for two major reasons. First, synaptic delay associated with chemical transmission is completely eliminated by electrical coupling, demonstrated here by the equivalence of electrically and mechanically evoked refractory periods (Fig. 5), and by the very short latency between MET currents and vCAP (VM vs vCAP; Fig. 2). Second, electrical coupling does not suffer from depletion of presynaptic vesicles, evidenced here by the lack of forward masking in vCAPs (indistinguishable from the AP refractory period; Fig. 5). This was demonstrated by comparing paired-pulse mechanical stimuli to paired-pulse electrical stimulation of the nerve bundle, which distinguished presynaptic and postsynaptic mechanisms underlying the decline in vCAP magnitude for short PPIs. Remarkably, evCAPs recorded for paired electrical pulses were identical to vCAPs recorded for paired pulse mechanical stimuli, demonstrating that the reduction in vCAPs with short PPIs results from refractory properties of AP generation and not from hair cell synaptic transmission. Moreover, stimulus intensity had no effect on short latency vCAP forward masking, as would be expected for electrical NQ coupling at the calyx synapse.
The fundamental innovation of the calyx was to eliminate chemical transmission in favor of electrical coupling that is ultrafast and immune to transmitter depletion (Fig. 5). The fact that auditory cCAPs were eliminated by the antagonist CNQX, while vCAPs were completely unchanged, offers additional strong evidence that glutamatergic quantal transmission is not required for short-latency vestibular responses (Fig. 1). Importantly, hair cell microphonics were unaffected by CNQX suggesting that this blockade disrupted neurotransmission between hair cells and afferent neurons, and not mechanosensory hair cell function. This result alone does not prove that neurotransmission underlying vCAPs reported here is nonquantal, but the lack of forward masking combined with significant latency differences strongly argue against quantal transmission underpinning the vCAP response.
Short-latency responses are present only in calyx-bearing afferents contacting Type I hair cells in the striola (Curthoys et al., 2016). Three forms of synaptic transmission are present at these synapses: quantal, slow NQ K+ build up in the cleft, and ultrafast electrical NQ coupling. All three forms play a physiological role (Contini et al., 2022), but only the ultrafast electrical component is consistent with the present short-latency vestibular action potential data. K+ build up in the cleft is simply too slow (Lim et al., 2011; Highstein et al., 2014; Govindaraju et al., 2023) to account for the data. Quantal transmission is also too slow (Palmer and Russell, 1986; Highstein et al., 1996; Li et al., 2021), fatigable, vulnerable to forward masking (Harris and Dallos, 1979), and sensitive to CNQX. Present results rule out K+ build up and quantal transmission as candidate mechanisms but are entirely consistent with the ultrafast electrical NQ hypothesis.
Results provide a new framework for understanding how the mammalian vestibular system generates high acuity/synchronized sensory data signals likely to support ecological functions such as maintenance of postural stability and compensatory reflexes. This has clinical relevance for better understanding alterations in vestibular sensitivity in health and disease, and for understanding the generation of vestibular reflexes at the first afferent synapse. Further exploration of key synaptic mechanisms of the calyx is likely to provide useful information regarding optimal stimulus paradigms for generating sensitive and robust reflex responses for vestibular testing in the clinic, and for understanding how peripheral vestibular function is modulated in various pathologic states. Current evidence suggests the stimulus onset is key for evoking synchronized vestibular nerve potentials at the first synapse (Pastras et al., 2023a,b), and further investigation of key synaptic mechanisms at the periphery will enhance this understanding.
Results raise the possibility that differences between auditory and vestibular responses to transient stimuli might offer a means to more clearly disambiguate auditory versus vestibular evoked potentials. Transient stimuli delivered by BCV and ACS activate the otolith organs and the cochlea at the same time, leading to complex evoked potentials that reflect a mixture of auditory and vestibular responses. The auditory contribution to vestibular responses can be reduced by using acoustic masking (Jones and Lee, 2021), but our results based on latency and forward masking suggest it might be possible to design even more discriminating stimuli.
Finally, despite the supportive evidence that synchronized vCAPs are driven by fast NQ synaptic transmission at the vestibular hair cell–afferent synapse in this study, it does not preclude the possibility of “mixed” quantal and nonquantal transmission in vestibular responses associated with other experimental and/or natural stimuli; especially given the presence of synaptic ribbons in Type I hair cells and postsynaptic glutamate receptors on the calyx.
Footnotes
This work was supported by National Institutes of Health Grants DC 006685 and DC 018919 (to R.D.R.) and the Macquarie University Research Fellowship MQRF0001126 (to C.J.P.).
The authors declare no competing financial interests.
- Correspondence should be addressed to Christopher J. Pastras at christopher.pastras{at}mq.edu.au
