Expression of hyperpolarization-activated current (I_h) in zonally defined vestibular calyx terminals of the crista

Abstract

Calyx terminals make afferent synapses with type I hair cells in vestibular epithelia and express diverse ionic conductances that influence action potential generation and discharge regularity in vestibular afferent neurons. Here we investigated the expression of hyperpolarization-activated current (I_h) in calyx terminals in central and peripheral zones of mature gerbil crista slices, using whole cell patch-clamp recordings. Slowly activating I_h was present in >80% calyces tested in both zones. Peak I_h and half-activation voltages were not significantly different; however, I_h activated with a faster time course in peripheral compared with central zone calyces. Calyx I_h in both zones was blocked by 4-(N-ethyl-N-phenylamino)-1,2-dimethyl-6-(methylamino) pyrimidinium chloride (ZD7288; 100 µM), and the resting membrane potential became more hyperpolarized. In the presence of dibutyryl-cAMP (dB-cAMP), peak I_h was increased, activation kinetics became faster, and the voltage of half-activation was more depolarized compared with control calyces. In current clamp, calyces from both zones showed three different categories of firing: spontaneous firing, phasic firing where a single action potential was evoked after a hyperpolarizing pulse, or a single evoked action potential followed by membrane potential oscillations. In the absence of I_h, the latency to peak of the action potential increased; I_h produces a small depolarizing current that facilitates firing by driving the membrane potential closer to threshold. Immunostaining showed the expression of HCN2 subunits in calyx terminals. We conclude that I_h is found in calyx terminals across the crista and could influence conventional and novel forms of synaptic transmission at the type I hair cell-calyx synapse.

NEW & NOTEWORTHY Calyx afferent terminals make synapses with vestibular hair cells and express diverse conductances that impact action potential firing in vestibular primary afferents. Conventional and nonconventional synaptic transmission modes are influenced by hyperpolarization-activated current (I_h), but regional differences were previously unexplored. We show that I_h is present in both central and peripheral calyces of the mammalian crista. I_h produces a small depolarizing resting current that facilitates firing by driving the membrane potential closer to threshold.

INTRODUCTION

Vestibular hair cells of the inner ear respond to movement and position of the head and drive reflexes important for balance. Mechanosensory signals originating from type I and type II vestibular hair cells are encoded as action potentials in primary vestibular afferent neurons and are transmitted to the central nervous system. Vestibular afferent neurons are bipolar, and their loosely myelin-wrapped cell bodies reside in the vestibular ganglion at a distance of several hundred micrometers from the hair cells in the neuroepithelium. Afferent fibers can fire action potentials with variable interspike intervals and have been broadly classified into groups that exhibit irregular (phasic) or regular (tonic) firing characteristics (1, 2). Dendritic afferent terminals on hair cells vary in their morphology and form either small bouton endings on type II hair cells or expansive cup-shaped endings called calyces on type I hair cells. Calyx terminals can form calyx-only endings enveloping one or more type I hair cells, or they can be part of more complex dimorphic endings. The majority of afferent fibers in mature rodent vestibular epithelia are dimorphic and are fed by synaptic input from calyx endings on type I hair cells as well as from bouton terminals on type II hair cells. In the gerbil crista, approximately one-third of type I hair cells in the central zone (CZ) of the neuroepithelium are contacted by calyx-only afferents. The remaining type I hair cells in central zones and peripheral zones (PZs) of the cristae are contacted by dimorphic afferents (3). Afferents that fire with irregular characteristics are associated with central zones and receive input from calyx-only and dimorphic terminals. Afferents that fire with more regularity receive input from bouton-only or dimorphic fibers and are located in peripheral zones. The mechanisms that drive these different firing characteristics remain unknown.

Voltage-gated Na⁺ and K⁺ ion channels within afferent terminals influence action potential firing and have been primarily studied in the expansive calyx terminal synaptic endings on type I hair cells (2, 4–6). Previous work has also documented the electrophysiological expression of hyperpolarization-activated current (I_h) in acutely dissociated gerbil calyces (7, 8) and in calyx endings in semi-intact preparations of rodent utricle (9–11). I_h is prevalent in cell bodies of isolated vestibular ganglion cells, where the impact of I_h on action potential firing characteristics has been evaluated (10, 12–15). Different types of intrinsic action potential firing patterns have been observed in response to sustained membrane depolarizations. Phasic ganglion cells show a strong adaptive response and typically fire only one action potential at step onset. Intermediate cells show a moderate adaptation of action potentials, and tonic cells demonstrate tonic firing in response to sustained depolarization (16, 17). Initial work in early postnatal mice suggested that the spike rate becomes elevated and firing more regular in vestibular afferent neurons where I_h is enhanced (10). Rat vestibular ganglion neurons that fired with tonic properties showed a larger current density of I_h compared with phasic neurons and converted to phasic firing when I_h was blocked (15).

Despite the significant body of work focused on afferent soma, the contributions of I_h to the initiation of action potentials and shaping of firing patterns in mature vestibular afferent dendrites are less clear. Further investigations are needed at these sites where hair cell sensory signals provide input to afferent terminals and become neural codes. Recently, two modes of synaptic transmission, termed quantal and nonquantal, have been reported between type I hair cells and their calyx terminals. Currents mediated by hyperpolarization-activated, cyclic nucleotide-gated (HCN) channels may shape both forms of transmission in different ways. Nonquantal transmission in central calyces of the turtle crista may be influenced by K⁺ ion flux in the intercellular synaptic cleft and involve I_h (18, 19), whereas quantally evoked excitatory postsynaptic potentials in calyces and auditory afferents are shortened by I_h (8, 20). However, localization of HCN channels to calyx terminal membranes has not been demonstrated to date. Here we report immunostaining of HCN2 channels in calyceal terminals of the crista.

Given the heterogeneity of I_h observed in vestibular ganglion cell bodies, we hypothesized that afferent terminals within the crista of the semicircular canals might show variations in their hyperpolarization-activated current profiles. Indeed, HCN channel subunits may subserve different functions in afferents from the central and peripheral regions of crista. To address potential differences, we examined the expression of I_h in calyx terminals in different zones of thinly sliced gerbil crista with whole cell patch-clamp techniques. We determined that I_h was expressed in almost all calyx endings located in both central and peripheral zones of the crista. Hyperpolarization-activated currents in calyces from both zones were blocked by 4-(N-ethyl-N-phenylamino)-1,2-dimethyl-6-(methylamino) pyrimidinium chloride (ZD7288), confirming that currents were carried by HCN channels. I_h characteristics did not differ between zones in terms of peak current or half-activation voltage, but activation kinetics were significantly faster in PZ calyces compared with CZ calyces. Activation kinetics became more rapid in response to intracellular application of cAMP, which also produced a rightward shift in half-activation of I_h. Current-clamp investigations revealed different types of firing in calyces (spontaneous, phasic, and oscillatory) that were not zonally specific. In the absence of I_h the latency to peak of the action potential increased.

METHODS

Animals and Crista Extraction

In accordance with protocols approved by the University of Colorado’s Institutional Animal Care and Use Committee and NIH guidelines, Mongolian gerbils (Meriones unguiculatus, 2–8 wk old) of both sexes were obtained from a breeding colony on location and deeply anesthetized with an intraperitoneal injection of ketamine (70 mg/kg) and xylazine (5 mg/kg) mixed in sterile saline. After decapitation, ampullae were carefully removed from the bony labyrinth of the vestibular system with fine forceps and immersed in a solution of Leibovitz’s L-15 medium (L-15; pH 7.40–7.45, adjusted with NaOH; osmolality 300–305 mosmol/kgH₂O) combined with 0.5 mg/mL bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO) for a minimum of 50 min at room temperature (21–24°C).

Slice Preparation

Horizontal cristae, which in gerbils lack a nonsensory central area called the eminentia cruciatum, were used for making slices as described previously (11, 21). A cut was made in the roof of the ampulla, and the flattened crista was embedded in a solution of 4% low-gelling temperature agarose (2-hydroxyethylagarose, type VII; Sigma-Aldrich) in Dulbecco’s phosphate-buffered saline. A block containing the crista was glued to the stage of a vibrating blade microtome (Leica VT 1200S; Leica Biosystems, Deer Park, IL) with superglue and immersed in L-15. The block was sliced in the horizontal plane, yielding five or six slices ∼100 µm in thickness. Slices were maintained in L-15 combined with 0.5 mg/mL BSA for up to 4 h. For electrophysiological recordings, a slice was transferred to a recording chamber (Warner Instruments) containing L-15 and secured on the coverslip base of the chamber by a small weight. Slices were viewed under differential interference contrast (DIC) optics with an Olympus BX51WI microscope with a water immersion objective (×40). Zones were classified as the central zone (CZ) occupying the apical one-third of the crista and the adjacent slopes representing the peripheral zones (PZ) (3, 21). Calyces were morphologically identified as cup-shaped terminals surrounding type I hair cells in slices (see Fig. 1 in Ref. 21) and by the presence of large Na⁺ currents in whole cell recordings.

Isolated Preparation

A small subset of recordings were made from calyces encompassing type I hair cells mechanically dissociated from the peripheral cristae without the use of exogenous enzymes as described previously (22). Dissociated cells were allowed to settle to the bottom of the recording chamber before recording.

Electrophysiological Recordings

Capillary glass tubing (G85150T-3; Warner Instruments, Hamden, CT) was inserted into a micropipette puller (P-97; Sutter Instruments, San Rafael, CA) and pulled to make patch pipettes. Pipette tips were heat polished on a Narishige M830 microforge (Narishige International USA, East Meadow, NY) to give an open tip resistance in the extracellular solution ranging from 2.2 to 4.9 MΩ. Tips were coated with silicone elastomer (Sylgard 184; Dow Corning, Midland, MI) to reduce stray capacitance. The electrode solution for filling pipettes contained (in mM) 115 KF, 10 KCl, 2 MgCl₂, 2 NaCl, 10 HEPES, 3 glucose, 10 ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA), and 0–2.5 Na₂ATP, pH adjusted to 7.4 with KOH to give a final K⁺ concentration of ∼140 mM and osmolality adjusted to 300–305 mosmol/kgH₂O with mannitol.

Whole cell patch clamp was used to record from calyx terminals in voltage and current clamp. Gigaseals were initially formed on the outer face of calyces and whole cell recordings made at room temperature (21–24°C) with a patch amplifier (Axopatch 200B; Molecular Devices, Sunnyvale, CA) connected to a PC through an A/D converter (Digidata 1440A: Molecular Devices) and pCLAMP software (v. 10.7). A variety of different step protocols were applied in voltage clamp from a holding potential of −80 mV to obtain recordings of Na⁺ currents, K⁺ currents, and hyperpolarization-activated currents. For the latter, steps to hyperpolarized potentials were as negative as −150 mV and for different protocols were between 0.75 and 3 s in duration. The amplitude of I_h was measured as the difference current (I_start − I_end) between current measured shortly after the capacitance transient close to the start of the voltage step and current at the end of the 1-s-duration voltage step. Data were low-pass filtered online at 2–5 kHz and sampled at 10–50 kHz. Liquid junction potentials were calculated with the Junction Potential calculator (Clampex 10.7) and corrections applied during data analysis. No leak subtraction was performed. Current-clamp protocols were used to monitor action potential firing and classify firing patterns. To test for spontaneous firing, successive periods of 3 s were recorded around the resting potential with a brief hyperpolarizing step of −200 pA, 25-ms duration administered at the start of each 3-s period. The second protocol comprised a −100-pA, 600-ms pulse followed by a series of depolarizing current steps in 50-pA increments, duration 600 ms. Capacitance was measured in a subset of calyces [ages postnatal day (P)28–P52], yielding values of 9.0 ± 3.0 pF in PZ calyces (n = 6) and 10.7 ± 2.8 pF (n = 11) in CZ calyces (means ± SD; t test showed no significant difference between zones, P = 0.264). Uncompensated series resistance averaged 9.3 ± 7.7 MΩ (n = 17).

Drugs

4-(N-ethyl-N-phenylamino)-1,2-dimethyl-6-(methylamino) pyrimidinium chloride (ZD7288) was obtained from Tocris Bioscience (Bio-Techne Corporation, Minneapolis, MN). A 10 mM stock of ZD7288 was made in dH₂O, stored at −20°C, and diluted in L-15 to a final concentration of 100 μM on the day of the experiment. Slices were perfused with L-15 during the recording session with a Gilson Minipuls 3 peristaltic pump (Gilson, Inc., Middleton, WI) at a rate of 0.5–1 mL min⁻¹. ZD7288 solution was perfused onto slices at the same rate. Dibutyryl-cyclic adenosine monophosphate, sodium salt (dB-cAMP; Tocris Bioscience, Bio-Techne Corporation) was used to investigate modulation of I_h. Stock solutions (10 mM) were stored at −20°C until the day of the experiment, when they were diluted in internal solution to a concentration of 500 µM.

Data Analysis

Data were analyzed with pCLAMP 10.7 (Molecular Devices) and SigmaPlot 11 (Systat Software). Statistical significance was determined with Student’s t test for data that were normally distributed or the nonparametric Mann–Whitney rank-sum test. Results are given as means ± standard deviation (SD) or medians; n represents number of cells.

To determine the half-maximum activation potential of I_h (V_1/2), activation curves were fitted to scatterplots obtained from tail currents with a Boltzmann function of the form

$$I/I_{\max }=1/1\mathrm{+}\exp \left[\left(V\mathrm{-}{V}_{1/2}\right)/S\right]$$ (1)

where V is the voltage conditioning step and S the slope factor.

To determine the time constant of I_h activation (τ), I_h current evoked by a step to −140 mV was fitted with a single-exponential equation of the form

$$I\left(t\right)=A{\left[1\mathrm{-}\exp \left(\mathrm{-}t/\tau \right)\right]}^3\mathrm{+}{A}_0$$ (2)

or a double-exponential equation:

$$I\left(t\right)=A{\left[1\mathrm{-}{B}_0\mathrm{·}\exp \left(\mathrm{-}t/{\tau }_0\right)\mathrm{-}\left(1\mathrm{-}{B}_0\right)\mathrm{·}\exp \left(\mathrm{-}t/{\tau }_1\right)\right]}^3\mathrm{+}{A}_0$$ (3)

where A scales the magnitude of the current, B₀ scales the relative proportion of the fast and slow components of the current, and A₀ defines the leak current. A reflects voltage-dependent activation, conductance, and driving force.

R² values were typically >0.95.

Immunohistochemistry

Ampullae from three gerbils (P39–P43) were placed in L-15 and the nonsensory tissue removed by trimming with microscissors. Cristae were then placed in a chamber filled with 4% paraformaldehyde and secured to a small raised stage with nylon threads to flatten the neuroepithelium. After fixing for 20 min, flattened cristae were rinsed twice in phosphate-buffered saline (PBS) for 2 min each and then stored in PBS at 4°C for up to 72 h. Tissue was permeabilized in 1% Triton X-100 in PBS for 30 min at room temperature (RT; 22–24°C) and then was blocked in the AB medium solution containing 0.1 M phosphate buffer (PB; 50 mM KH₂PO₄, 150 mM Na₂HPO₄), 150 mM NaCl, 3 mM Triton-X, and 1% BSA supplemented with 5% normal goat serum (NGS) at RT for 1 h. Tissue was incubated with primary antibodies in AB medium with 2.5% NGS for 48 h at 4°C, followed by rinses in PBS (3×) for 15 min at RT. Primary antibodies were anti-Tubulinβ3 (TubB3; mouse IgG2a; Biolegend no. 801202; RRID: AB_10063408) and anti-HCN2 (rabbit polyclonal; Alomone Labs no. APC-030; RRID: AB_2313726). Tissue was incubated with secondary antibodies goat anti-rabbit Alexa Fluor 405 (Invitrogen catalog no. A48254) and goat anti-mouse IgG2a Alexa Fluor 594 (Life Technologies catalog no. A21135) diluted in AB medium supplemented with 2.5% NGS for 2 h at RT, followed by 3 × 15 min rinses in PBS at RT. Cristae were mounted on slides with clear mounting medium (Fluoromount-G; SouthernBiotech, 0100-01). Images were obtained with an Olympus FV1000 laser scanning confocal microscope. For high-magnification images, a ×100 oil objective [numerical aperture (NA) 1.4] was used and voxel sizes ranged from xy = 0.062 µm² to 0.121 µm² depending on the digital zoom settings. Z step was set at 0.5 µm. Sequential scans were performed to eliminate potential cross talk. Images are shown as maximum projection; the number of collapsed (max z-projected) sections varies. Nonspecific staining was determined by omitting primary antibodies. Images were processed in ImageJ v.1.52p; Java 1.8.0_172 (64 bit) (Wayne Rasband, NIH; http://imagej.nih.gov/ij; RRID:SCR_003070), and figure panels were aligned in Adobe Illustrator 2021 (RRID:SCR_010279).

RESULTS

Hyperpolarization-Activated Current in Calyx Afferent Terminals across Crista Zones

Whole cell tight-seal recordings were obtained from calyx terminals in regions designated as central zones (CZ) or peripheral zones (PZ) of transversely sliced horizontal gerbil cristae (P8–P52) as described previously (11, 21). In gerbil calyx terminals, we have previously described a profile of outward K⁺ currents and inward Na⁺ currents, some of which vary with zonal position (11, 21, 23). In this study we characterized the properties of hyperpolarization-activated currents in the different crista zones. To test for the presence of hyperpolarization-activated current, long-duration hyperpolarizing voltage steps of up to 3 s were applied from a holding potential of −80 mV in whole cell voltage clamp. Since calyces often did not tolerate long-duration steps, most cells were studied with voltage steps between 0.75 and 1.5 s in length. In response to hyperpolarizations, the majority of calyces showed a small, near instantaneous current followed by a slowly activating inward current that was largest at the most negative voltage steps applied. Representative examples from peripheral and central zone calyces are shown in Fig. 1A in response to 1.5-s-duration steps. Slowly developing currents resembled hyperpolarization-activated currents (I_h mediated by HCN channels), which have been described previously in central calyces of the turtle crista (18), dimorphic afferents of mouse utricle (10), and calyces isolated from gerbil crista (7, 8). In crista slices we observed I_h in 34/42 calyces from the peripheral zone (81%, ages P14–P47) and 24/28 calyces from the central zone (86%, ages P8–P52). At these ages calyx terminals are fully formed, and we saw no clear correlation between peak I_h and postnatal age (not shown). Peak currents for cells from the two zones are shown in Fig. 1B and showed considerable variation in magnitude among cells. In most calyces hyperpolarization-activated currents ranged from −30 to −270 pA at 1 s into the step to −140 mV, but in a few cells currents exceeded −400 pA (Fig. 1B). Data were not normally distributed, and a Mann–Whitney rank-sum test showed that median peak I_h was not significantly different in PZ calyces compared with CZ calyces (PZ: −164 pA, n = 34; CZ: −137 pA, n = 24, P = 0.279). The remaining calyces showed no measurable hyperpolarization-activated current or current amplitudes of less than −30 pA at steps to −140 mV.

Figure 1.

Hyperpolarization-activated currents (I_h) are present in calyx terminals across different crista zones. A: currents in response to voltage steps ranging from −140 mV to −90 mV, from a holding potential of −80 mV, are shown for example peripheral zone [PZ; female, postnatal day (P)20] and central zone (CZ; female, P22) calyces. Voltage protocol shown below the PZ calyx; voltage step duration was 1.5 s. Rapid downward deflections on current traces in this and subsequent figures represent excitatory postsynaptic current as described previously in calyx recordings (22, 24, 25). Double-exponential fits (Eq. 3) to the activation of currents at the −140 mV step are shown in red. Fast and slow time constants (τ) were 87 and 349 ms for the PZ calyx and 118 and 539 ms for the CZ calyx, respectively. I_start, current measured close to start of voltage step; I_end, current at end of voltage step. B: scatterplots show current amplitude for individual PZ and CZ calyces (circles) in crista slices. The amplitude of the hyperpolarization-activated current was measured after 1-s steps to −140 mV. Data were not normally distributed, and a Mann–Whitney rank-sum test showed that median peak I_h was not significantly different in PZ calyces compared with CZ calyces (PZ: −164 pA, n = 34; CZ: −137 pA, n = 24; P = 0.279). Box plots show median and 25th and 75th quartiles, and whiskers extend out to the 10th and 90th percentiles. PZ calyces were from 18 males, 14 females, and 2 gerbils in which sex was not determined. CZ calyces were from 10 males, 13 females, and 1 with sex undetermined. Ages ranged from P14 to P47 in the PZ group and from P8 to P52 in the CZ group. C: activation of currents at the −140 mV step compared. Individual time constants (single-exponential fit) are shown (black circles). Mean τ in CZ calyces was 0.278 ± 0.05 s and was significantly different from mean τ in PZ calyces (0.230 ± 0.05 s; **P = 0.003, t test). Mean current values are indicated by red diamonds.

Activation of Hyperpolarization-Activated Currents

The activation time course of I_h in vestibular afferent cell bodies and calyx terminals is voltage dependent and has been described with kinetic fits using single- or double-exponential functions. We used an approach similar to that used in vestibular ganglion cells, where either a single or double exponential provided a best fit to activation of I_h (12). We fit the slow activation of I_h in central and peripheral zone calyx terminals at the −140 mV step with third-order single- and double-exponential functions (26) to determine time constants of activation (Eqs. 2 and 3 in methods). In general, both methods provided good fits, but a double exponential proved slightly better. It was not possible to fit currents in all cells because of noise such as excitatory postsynaptic currents or membrane instability that occurred in some instances. Examples of double-exponential fits to the currents at the −140 mV step are shown superimposed in red for both calyces in Fig. 1A. For a single exponential (3rd-order function) the mean activation time constant (τ) was 278 ± 50 ms (n = 22) in CZ calyces and 230 ± 50 ms (n = 25) in PZ calyces (Fig. 1C). The difference between time constants was significant (P = 0.003). For double-exponential fits, the mean fast activation time constant was 116.7 ± 71.0 ms (n = 19) in CZ calyces and 134.6 ± 88.7 ms (n = 24) in PZ calyces, whereas the slow activation time constant was 689.4 ± 31.0 ms (n = 19) in CZ calyces and 591.5 ± 30.7 ms (n = 24) in PZ calyces. The weighted time constant median values were 0.390 s (n = 17) and 0.292 s (n = 21) in CZ and PZ calyces, respectively, and were significantly different (signed rank test, P = 0.015). This observation that the current activation is faster in peripheral zone afferents infers there may be zonal differences in underlying HCN channel subunit composition. The accelerated kinetics could act to decrease the lag between excitatory postsynaptic potentials and action potential generation. The activation time course was notably slower than that described for I_h in vestibular hair cells of the early postnatal mouse utricle (27). However, time constants for single-exponential fits were similar to those of ∼200 ms reported previously in isolated calyces (8) and in rat vestibular ganglion cells (12). Horwitz et al. (10) used double-exponential functions to fit activation kinetics in cell bodies and utricular calyces and reported mean fast and slow time constant values comparable to our findings. Values were not directly compared between striola and extrastriola zones of the utricle, but there was no significant difference in time constants between calyces in zones arbitrarily defined as central and peripheral regions (10).

Large inward tail currents were visible following the return to the holding potential after hyperpolarizing voltage steps (Fig. 1A), and tail currents were measured to generate activation curves for I_h. The mean value for half-maximal voltage activation (V_1/2) was −120.4 ± 5.3 mV (n = 7) in PZ cells and −119.0 ± 5.6 mV (n = 7) in CZ cells and was not significantly different between zones (P = 0.629, t test). Slope factors had mean values of 8.1 ± 2.5 (n = 7) and 9.3 ± 2.0 (n = 7) in PZ and CZ calyces, respectively, and were not significantly different (P = 0.34, t test). Mean half-activation values were similar to the mean value of −123 mV reported previously in isolated gerbil calyces (8) but were more negative than the V_1/2 values of between −115 and −100 mV reported for early postnatal calyces in the mouse utricle (10) and cell bodies of the rat and mouse vestibular ganglia (12, 28). These differences could be due to differences in developmental stage, culture conditions, and/or endogenous modulatory factors. V_1/2 values have been shown to shift by several millivolts in response to changing cAMP levels or dialysis with electrode solutions in ruptured-patch recordings (13, 20).

Effect of ZD7288 on Hyperpolarization-Activated Currents

Currents mediated by HCN channels are effectively blocked by the drug ZD7288 (29, 30). To test whether the hyperpolarization-activated currents in different crista zones were mediated by HCN channels, we perfused crista slices with L-15 containing 100 µM ZD7288 during calyx recordings. We waited 5–12 min after perfusion onset for full block to be achieved, since ZD7288 is known to take several minutes to block HCN channels (30). Figure 2A shows examples of the block of the hyperpolarization-activated currents in representative PZ and CZ calyx terminals following ZD7288 perfusion. Figure 2B shows control currents and currents measured after ZD7288 perfusion in calyces located in peripheral (red symbols) and central (black symbols) slice locations. ZD7288 produced an almost complete block of I_h in calyces from both zones. The amount of current blocked was 90.8 ± 9.7% in PZ calyces (n = 6) and 83.0 ± 12.2% in CZ cells (n = 6). The overall reduction in hyperpolarization-activated current was 86.9 ± 11.3% (n = 12 calyces). I_h is carried by cations (K⁺ and Na⁺) and in other cell types can contribute a small depolarizing current at the resting potential (29, 31). We measured the resting (zero current) potential in current clamp in seven calyces before and after application of ZD7288 (100 µM). In control the mean value for resting potential was −58.5 ± 4.1 mV, and in the presence of ZD7288 the mean resting potential was −65.2 ± 2.2 mV (Fig. 2C; n = 5 PZ and 2 CZ calyces, P < 0.003, paired t test). This indicates that I_h can influence the resting membrane potential in calyces by providing a small depolarizing current. It has been reported that ZD7288 blocks endogenous Na⁺ currents in dorsal root ganglion cells and Na⁺ currents mediated by Nav1.4 channels (32). To verify the specificity of the block of ZD7288 for I_h in our experiments, we investigated two other major conductances in calyces (Na⁺ and K⁺) to determine whether they changed in magnitude during perfusion of ZD7288. Peak Na⁺ currents in control cells had a median value of −3,200 pA, whereas median peak Na⁺ current in ZD7288 was −3,171 pA (signed rank test, n = 12, P = 0.424). Median steady-state K⁺ currents in calyces showed a small increase following ZD7288 perfusion, but the difference between control and drug application was not significant [control K⁺ currents: 3,820 pA, K⁺ currents in ZD7288: 4,499 pA, n = 12, P = 0.301 (data not shown)]. We conclude that ZD7288 effectively blocked I_h in calyces in both zones without significantly affecting associated transient Na⁺ currents or outward K⁺ currents in calyces.

Figure 2.

4-(N-ethyl-N-phenylamino)-1,2-dimethyl-6-(methylamino) pyrimidinium chloride (ZD7288) blocks hyperpolarization-activated currents (I_h) in calyces from central (CZ) and peripheral (PZ) zones. A: hyperpolarization-activated currents before (black traces) and after (gray traces) perfusion of 100 µM ZD7288 are shown for a PZ [postnatal day (P)26 male] and a CZ (P48 female) calyx for a voltage step from −80 to −140 mV. B: summary of block by 100 µM ZD7288 in 6 PZ (red symbols) and 6 CZ (black symbols) calyces; current was measured at steps to −140 mV. Mean current values for all cells are indicated by gray diamonds (SD). C: zero current (membrane) potential for 7 calyces in control conditions and in the presence of ZD7288. Control mean resting membrane potential was −58.5 ± 4.1 mV, and in the presence of ZD7288 the mean resting potential was −65.2 ± 2.2 mV (n = 5 PZ and 2 CZ calyces, **P < 0.003, paired t test).

Modulation of I_h by cAMP in Central and Peripheral Zone Calyces

Cyclic nucleotides can influence the activity of HCN channels by binding directly to the COOH-terminal cyclic nucleotide binding domain. In many cells, cAMP increases currents through HCN channels and the voltage dependence of I_h moves to more depolarized potentials (31, 33). Modulation of I_h by cAMP has been characterized in vestibular ganglion neurons (12, 34), where cAMP increased the amplitude of I_h and produced a rightward shift in activation of the current. To test whether cyclic nucleotides could modulate I_h in calyces in both zones of the crista, we applied a nonhydrolyzable form of cAMP via the intracellular solution. In response to standard voltage protocols, we observed large hyperpolarization-activated currents in calyces when 0.5 mM dibutyryl-cAMP (dB-cAMP) was included in the patch electrode solution. Figure 3A shows currents in response to a series of hyperpolarizing steps in representative PZ and CZ calyces. In these experiments I_h was typically seen to increase over the first few seconds after membrane breakthrough into whole cell mode. Peak I_h in PZ cells and CZ cells is shown in Fig. 3B, and the median peak current in the presence of dB-cAMP was −204 pA at the step to −140 mV. Compared with control cells (Fig. 1B), peak I_h was significantly greater than the median amplitude of −161 pA in a combination of PZ and CZ calyces (n = 58; P = 0.024, Mann–Whitney rank-sum test). The activation time course of I_h in the presence of dB-cAMP was fit with exponential functions as shown for the currents in response to voltage steps to −140 mV in Fig. 4A.

Figure 3.

Modulation of hyperpolarization-activated currents (I_h) by internal dibutyryl-cAMP (dB-cAMP). A: hyperpolarization-activated currents in a peripheral zone (PZ) calyx [left; postnatal day (P)24 male] and a central zone (CZ) calyx (right; P28 female) in response to voltage steps in 5-mV increments. Double-exponential fit is shown superimposed on currents for the step to −140 mV (red lines). B: scatterplot showing peak I_h when dB-cAMP was included in the electrode solution (PZ n = 7, red circles; CZ n = 5, black circles). Box plot shows median peak current (−204.2 pA) and quartiles). Age range was P21–P29. In all recordings, 0.5 mM dB-cAMP was included in the patch electrode solution and current amplitude following steps to −140 mV was measured at 1 s between 1 and 11 min into the recording. C: Boltzmann activation curves (Eq. 1) for cells in A with dB-cAMP. CZ calyx (black circles): half-maximum activation potential (V_1/2) −108.9 mV, slope 8.5; PZ calyx (red circles): V_1/2 −120.0 mV, slope 8.7. Mean V_1/2 for dB-cAMP calyces was −113.4 ± 5.1 mV (n = 12).

Figure 4.

Classification of firing patterns in calyces. A: examples of spontaneous firing in peripheral zone (PZ) calyces [left: postnatal day (P)34 calyx; center: P24 calyx] and a central zone (CZ) calyx (right: P26 calyx). A brief current step (−200 pA, 25-ms duration) was given at the start of each recording in current clamp. In all 3 cases the interspike intervals were variable, and the PZ calyx at center showed burstlike activity. B: an example of a phasic response consisting of a single action potential following a brief current step (−200 pA, 25 ms) in a CZ calyx (P52; female). C: oscillatory response to brief current step (−200 pA, 25 ms) in a PZ calyx (P47; female). As shown in this cell, a few calyces demonstrated 2–4 spikelike events of decreasing magnitude in response to brief hyperpolarizing current injection. Dashed lines in A–C indicate −70 mV. D: summary of the different firing behaviors observed in current clamp. Black bars represent CZ calyces, and gray bars indicate PZ calyces. For cells that fired action potentials the majority were phasic [age range P20–P52; all but 1 cell expressed hyperpolarization-activated currents (I_h)]. Fewer calyces showed spontaneous firing or membrane oscillations (age range P22–P47; 1 cell lacked I_h). Several cells did not fire action potentials (no APs; age range P8–P43; all cells expressed I_h).

Tail currents were measured for I_h, and resulting scatterplots were fit with the Boltzmann equation (Eq. 1) for PZ and CZ calyces in the presence of dB-cAMP. Examples for the cells in Fig. 3A are shown in Fig. 3C. Half-activation (V_1/2) for the PZ calyx was −120.0 mV (slope factor 8.7), and V_1/2 for the CZ calyx was −108.9 mV (slope factor 8.5). Mean V_1/2 for all dB-cAMP-containing calyces was −113.4 ± 5.1 mV (n = 12), which was significantly different from the mean V_1/2 value for control calyces (−119.8 ± 5.3 mV, n = 14, P = 0.005, t test). Mean slope factor for dB-cAMP cells was 8.3 ± 2.3 (n = 12), which was not significantly different from control cells (8.7 ± 2.3, n = 14, P = 0.66, t test). Together these data show that I_h in calyces is enhanced in the presence of the cAMP analog and its activation shifts to more positive values.

Role of I_h in Firing Properties in Calyces

Sharp electrode recordings have demonstrated that vestibular afferents can fire action potentials spontaneously and in the absence of hair bundle stimulation. Afferents that innervate central zones of vestibular epithelia often fire action potentials with highly variable interspike intervals (termed irregular), whereas afferents populating peripheral zones can fire in a pronounced regular fashion (1, 2). Therefore, firing regularity varies with location of afferent terminals within the epithelia, but the underlying mechanisms driving differences in firing are unclear. In cell bodies of cultured vestibular ganglion neurons, different types of firing responses have been observed, which may correlate with firing responses in terminals. Firing in ganglion cell bodies is evoked by sustained membrane depolarizations and has been generally categorized as phasic, intermediate, or tonic (12, 15–17, 34, 35). Rat vestibular ganglion neurons that fired with tonic properties showed a larger current density of I_h compared with neurons that fired with phasic properties (15). We investigated whether a similar heterogeneity in firing properties also exists among calyx-bearing afferent terminals in crista slices, using a variety of current-clamp protocols.

We first classified firing patterns in calyces. We observed spontaneous firing in current clamp in a few calyx afferents in our sample (4/24 PZ and 3/22 CZ, P22–P34). All of the cells with spontaneous firing expressed I_h when tested in voltage clamp, except for one PZ cell (P26) that fired spontaneous action potentials with variable interspike intervals but had no measurable I_h. Examples of spontaneous spiking activity in PZ and CZ calyces are shown in Fig. 4A. All three calyces shown expressed I_h and had variable interspike intervals. One PZ cell (P24) showed bursting-type activity (Fig. 4A, center). In response to a short current step (−200 pA, 25 ms), many cells fired a single action potential at the end of the step as shown for the CZ cell calyx in Fig. 4B. This behavior resembled the phasic firing type as described in rat vestibular ganglion neurons (15–17) and calyces (36).

A small number of calyces demonstrated a different and more unusual type of firing as shown for a PZ calyx in Fig. 4C. These calyces demonstrated a large action potential at the step end, which was followed by an additional one to four spikelike events of decreasing magnitude. This firing pattern has also been described in a subset of vestibular ganglion neurons and has been termed intermediate (15) or sustained resonating (17).

The distribution of different firing types is summarized in Fig. 4D for calyces from both zones. Most CZ calyces behaved in a phasic manner and fired one action potential after the brief hyperpolarizing current pulse (Fig. 4B). In central crista areas, 9/22 calyces tested with the protocol were classified as phasic. Two CZ calyces showed a series of oscillations consisting of an action potential followed by up to three spikelike events and resembled the “sustained-C” category described in ganglion cells (17). Three CZ calyces fired spontaneous action potentials. The remainder of CZ calyces showed no evoked or spontaneous action potentials. In the peripheral zone, 12/24 calyces fired action potentials in current clamp. Seven PZ calyces showed phasic firing in response to current injection, and four calyces demonstrated spontaneous firing of action potentials. Two PZ calyces showed oscillatory-type behavior where the initial evoked action potential was followed by up to three smaller events of decreasing magnitude (Fig. 4C).

Next, we tested the response of calyces from both zones to a series of hyperpolarizing and depolarizing current steps of longer duration (Fig. 5). Longer hyperpolarizations were used to allow increased I_h activation. Specifically, the protocol consisted of a 600-ms current injection (−100–200 pA) followed by a series of current steps in 50- to 100-pA increments (600 ms). In cells with I_h, hyperpolarizing current typically evokes a time-dependent voltage “sag” response, consisting of an initial membrane hyperpolarization followed by a depolarization that occurs as HCN channels open further with hyperpolarization (12). In response to this current protocol, most calyces (13/19) showed a characteristic sag response during hyperpolarization and a single action potential with the ensuing depolarization (Fig. 5A, PZ cell). Nine of ten CZ cells tested with this protocol showed a single evoked action potential. The remaining calyx showed oscillatory behavior in response to the protocol (not shown). Ten of thirteen PZ calyces showed phasic firing, and three showed oscillations. Figure 5B, top, shows superimposed action potentials of a calyx demonstrating a sag response (black trace) and a calyx that lacked a sag response (gray trace). Figure 5B, bottom, shows the latency to action potential peak relationship between cells that expressed I_h and demonstrated a sag and action potential latency. Mean latency to peak in calyces with a sag was significantly shorter than latency in calyces without a sag. A similar effect was observed when cells were perfused with ZD7288 to remove I_h. Figure 5C shows action potentials evoked after a 25-ms hyperpolarizing pulse in a PZ calyx (P28) before (black trace) and after (gray trace) perfusion with ZD7288 (100 µM). Latency to peak of evoked action potentials was increased in three PZ cells after ZD7288 perfusion.

Figure 5.

Hyperpolarization-activated current (I_h) generates a sag response during prolonged membrane hyperpolarization and decreases action potential latency to peak. A: action potentials observed in response to current-clamp protocol shown below (−100 pA, 600-ms pulse followed by a single series of 600-ms-duration depolarizing current steps). A peripheral zone (PZ) calyx with a sag (arrow) at step onset (indicating the presence of I_h) showed single evoked action potentials with depolarizing current steps [left; postnatal day (P)35 male]. A PZ calyx without a sag response also demonstrated single action potentials with depolarizing current steps (right; P35 male). B: action potentials evoked by the most depolarized current step from the 2 calyces in A are superimposed (top). Latency to action potential peak is longer in the cell that lacked a sag response. Box plots (bottom) compare latency to peak for calyces with and without a sag. Mean latency to peak in calyces with a sag was 2.51 ms (±0.44 SD, n = 13), which was significantly shorter than latency in calyces without a sag [3.29 ms (±0.81 SD, *P = 0.014, t test, n = 6)]. PZ calyces are shown by black circles; central zone (CZ) calyces are shown by white triangles inside black circles. C, left: action potentials in a PZ calyx (P28) before (black trace) and after (gray trace) perfusion with 4-(N-ethyl-N-phenylamino)-1,2-dimethyl-6-(methylamino) pyrimidinium chloride (ZD7288; 100 µM). Right: latency to peak of evoked action potentials was increased in 3 PZ cells (age range P22–P28) after ZD7288 perfusion. Action potentials for all 3 cells were evoked after a 25-ms hyperpolarizing pulse.

HCN2 Subunit Immunoreactivity in Crista

Our electrophysiological experiments confirm that HCN channels are expressed in calyx terminals in both central and peripheral zones of the crista. The relatively negative activation range, intermediate speed of activation, and sensitivity of I_h to cAMP strongly suggest contributions from HCN2 subunits. The presence of HCN channel subunits has been reported in vestibular ganglion cell bodies by immunohistochemistry (10, 12), but the localization of channels within calyx terminals of the crista has not been confirmed. We therefore tested for the presence of HCN2 channels in afferent terminals by immunolabeling whole mount cristae (P39–P43) with antibodies to HCN2 channels and Tubulinβ3 (TubB3) to label vestibular neurons. Figure 6 shows strong immunostaining for HCN2 channels in the crista neuroepithelium that was associated with TubB3-labeled afferent calyces. HCN2 staining was present in the basal and neck regions of calyx terminals around type I hair cells. HCN2 immunolabelling of afferent fibers was sparser, suggesting greater expression of channels at the synaptic regions between hair cell and afferents.

Figure 6.

Immunolocalization of HCN2 channels in gerbil crista. A: crista whole mount preparation [postnatal day (P)43, female] immunolabeled with antibodies to HCN2 (green, left) and Tubulinβ3 (TubB3; magenta, center). Calyx terminals (side view of roughly oval structures surrounding type I hair cells) as seen from the HCN2-labeled clusters within TubB3-labeled calyces (arrows, right). The HCN2 signal appears to colocalize with afferent fibers. Max z-projection from 8 virtual sections. Scale bar, 20 µm. B: HCN2 label is seen in the neck regions of calyces surrounding type I hair cells. Immunolabeling with antibodies to HCN2 (green, left) and TubB3 (magenta, center). Overlay (right) suggests presence of HCN2 channel subunits in the necks of calyx terminals. Single scan close to the apex of the epithelial surface shows circular structures indicating apical part of calyces that surround the neck regions of type I hair cells (P42 male gerbil). Scale bar, 10 µm.

DISCUSSION

Here we report slowly activating, noninactivating inward currents at hyperpolarized potentials in calyx terminals located in peripheral and central areas of the gerbil crista. We established that these currents were mediated by HCN channels since they were blocked by ZD7288 in calyces from both zones. Peak I_h amplitude, half-activation voltages, and sensitivity to cAMP did not vary between zones, but activation of I_h was found to be significantly slower in CZ calyces.

HCN Channel Expression in Vestibular Afferent Neurons

There are four known members in the Hcn family (Hcn1–4), which encode subunits that can assemble as homomeric or heteromeric channels. Isoforms exhibit differences in current activation kinetics, voltage dependence of their activation range, and sensitivity to the cyclic nucleotides cAMP and cGMP (29, 31, 37). mRNA expression of all four subtypes was reported in cell bodies of the vestibular ganglion, but HCN1, 2, and 4 were found to be more prevalent than HCN3 subunits (10, 12). HCN1 channels have more depolarized activation and the most rapid activation kinetics and lack sensitivity to cAMP compared with other subunits (29, 31). HCN1-mediated currents underlying I_h have been described in vestibular hair cells of mouse utricle and activate with relatively rapid kinetics (27). HCN2 and HCN4 are positively modulated by cAMP, whereas HCN3 channels are reported to be inhibited by cAMP. Additionally, the half-activation voltage is more negative for HCN2 channels compared with HCN1 and HCN4. The relatively negative V_1/2 for I_h, the intermediate speed of current activation, and sensitivity of the current to cAMP in both central and peripheral zone calyces are consistent with expression of HCN2 channels across the crista. However, the difference in activation kinetics of hyperpolarization-activated currents between zones suggests heterogeneity in subunit expression. A possible explanation is that HCN2 channels underlie I_h in PZ calyces whereas both HCN4 and HCN2 subunits contribute to I_h in central zone calyces and produce the slower activation kinetics observed in CZ afferents. In early postnatal mouse utricle, electrophysiological expression of I_h in dimorphic terminals occurred in the absence of HCN1 channels, whereas expression of I_h was greatly reduced in vestibular afferents lacking HCN2. In mice lacking both HCN1 and HCN2 a small, very slow hyperpolarization-activated current persisted in ganglion cells, indicating a likely contribution from HCN4 subunits (10). Taken together, HCN2 channels appear to be the main subunit underlying I_h in calyx afferents of the crista and utricle. Previous studies have documented the presence of ZD7288-sensitive currents in afferent terminals within vestibular epithelia in mouse utricle (10) and calyces in central regions of turtle crista (18). Whether there are subtle differences in HCN subunit composition in afferents innervating utricle, saccule, and canal organs remains to be established.

Immunohistochemical Localization of HCN2 Channels

Since the electrophysiological expression of I_h was most consistent with HCN2-mediated currents, we tested for the presence of HCN2 channels with antibody staining and found immunoreactivity associated with calyx terminals (Fig. 6). Confocal imaging revealed HCN2 staining along the length of the calyx cup with staining in the base and neck regions. We saw no evidence for HCN2 staining within hair cells, which aligns with a previous report showing strong expression of HCN1 but not HCN2 channels in vestibular hair cells of the early postnatal mouse utricle (27). We could not resolve whether staining was associated with the inner aspects, facing the hair cells, or outer surfaces of calyces. Given the slower activation kinetics of I_h found in central zone calyces, it would be interesting to investigate other subunits such as HCN4 in future work. HCN1, HCN2, and HCN4 have been detected in mouse utricles at P8 with RT-PCR as well as in vestibular ganglion cells (10, 27).

Role of I_h in Firing in Mature Vestibular Afferent Calyces and Cell Bodies

We studied I_h in calyces in cristae obtained from gerbils at 2–8 wk of age. We found that I_h was present in the majority of terminals across the crista (>80%), but in this study we did not distinguish calyx-only from dimorph terminals. Peak current was highly variable but did not correlate with postnatal age or crista location. Action potential firing data were obtained from cells at 3–8 wk of age. At these ages, distinct afferent firing patterns have emerged in semicircular canal afferents (38, 39). Membrane properties of hair cells in rodent vestibular epithelia show immature profiles around birth and change dramatically during the first postnatal month as hair cell ion channel complements are refined (40–43). In parallel, vestibular ganglion neurons also demonstrate changes in firing properties during postnatal development and maturation of vestibular-driven reflexes (10, 12, 15, 34, 38). An increase in I_h density along with increased HCN subunit expression was observed in cell bodies dissociated from rat vestibular ganglion neurons, and the number of tonic firing neurons increased from P3 to P14 (15). Almanza et al. (12) compared I_h properties in ganglion neurons at P7–P10 and P25–P28 and found an increase in current density and activation kinetics with developmental age. However, another study in rat ganglion cell bodies between P9 and P22 found no correlation between current density of I_h and postnatal age (13).

In light of prior studies in rodent neuroepithelia, we consider firing properties of calyx afferents to be mature at the ages studied here. Using step protocols to elicit action potentials, we found considerable heterogeneity in firing properties between calyx endings. In response to step current injections, the majority of calyces behaved in a phasic manner, firing one action potential at the start of the current pulse. A greater proportion of CZ calyces demonstrated phasic firing, whereas a greater proportion of PZ calyces showed spontaneous firing. A small percentage of both CZ and PZ calyces were oscillatory, showing a series of spikelike events to depolarizing current injection. HCN channels have been shown to contribute to slow membrane potential oscillations in some central neurons (31), and HCN4 channels mediate slow oscillatory behavior in thalamocortical network neurons (44). HCN channels might also contribute to the oscillatory responses we observed in a subset of afferents.

Spontaneous, but not oscillatory, firing has been described in some early postnatal calyces of the mouse utricle and rat saccule (10, 45) and maturing calyces of gerbil crista [P17–P33 (21)]. Interspike intervals were variable, with some calyces firing at very regular intervals and others firing at highly irregular intervals. Spontaneous firing persisted in the absence of mechanotransduction activity, indicating a role for interactions between voltage-gated ion channels within calyx terminals in action potential generation (10, 21). The role of HCN channels in pacemaking is well established in other cell types (29, 33), and initial studies suggested that depolarizing currents through HCN channels might promote highly regular firing patterns in afferent neurons specifically linked to peripheral zone afferents (10, 15). However, subsequent work demonstrated that HCN channels do not promote regular firing in all vestibular ganglion neurons where coexpression with low-voltage-activated K⁺ channels increased spike rates but not regularity (34). Clearly, I_h expression does not vary dramatically in calyces across crista regions, and its contribution to regular versus irregular firing pattern remains unresolved. Horwitz et al. (10) found no differences in expression across utricle; however, their regional classification did not correspond to striola and extrastriola regions. Small depolarizations induced by activation of I_h could influence activity of other ion channels coexpressed within terminals, such as Na⁺ channels (23, 46) and voltage-activated K⁺ channels (11, 21, 36, 41, 47), to generate diverse firing patterns at axon initial segments where spikes are initiated.

As described in vestibular ganglion cells, we found that prolonged hyperpolarization of calyx terminals in current clamp elicits a sag response when I_h is present (12, 13, 15, 34). Although the time course of activation of I_h is much slower than individual action potential duration, I_h can produce a small constant depolarizing current around the resting membrane potential. Despite highly negative half-activation values in PZ and CZ calyces, the membrane potential became more hyperpolarized when I_h was blocked by ZD7288. Single evoked action potentials in calyces were impacted such that action potential peak occurred faster in the presence of I_h. This is consistent with I_h facilitating action potential firing by driving the calyx membrane potential closer to threshold.

Neuromodulatory Effects and I_h

Hyperpolarization-activated currents are common in sensory neurons where the slow kinetics of HCN channels can influence action potentials through persistent activity at rest (48–50). The type I hair cell and calyx pair together in an unusual extended synapse where communication from hair cell to afferent can occur through conventional quantal and unconventional nonquantal mechanisms (18, 22, 24, 25, 45, 51–54). Hair cells release glutamate at ribbon synapses, and I_h can influence quantal transmission by sharpening of excitatory postsynaptic events in isolated terminals (8). We show here that I_h also speeds up action potential onset, allowing a faster postsynaptic response to sensory input. The contributions of I_h to the nonquantal mechanism mediated by resistive coupling mechanisms are intriguing and less clearly understood (2, 18, 55). Outward flux of K⁺ from the hair cell’s basolateral K⁺ channels likely increases K⁺ levels within the intercellular cleft and could lead to depolarization of both hair cell and calyx afferent terminal membranes (56). Although HCN channels are permeable to cations, they show much greater permeability to K⁺ over Na⁺ (29) and HCN-mediated currents in isolated calyces increased in raised K⁺ conditions (8). Blocking HCN channels abolished nonquantal signals in centrally located calyces of the turtle crista, suggesting that they are required for this unusual form of transmission (18). Work to date has focused on nonquantal transmission in central zone calyces (18, 45), which are important for sensing rapid head motion (57). However, since I_h is prevalent in both central and peripheral zone calyces, nonquantal mechanisms may be important for synaptic transmission in both zones in mammalian species.

The wide range in size of peak I_h in calyces (30–700 pA) suggests that endogenous regulation of HCN channels occurs. Cyclic nucleotides such as cAMP are known to influence HCN-mediated currents, and HCN2 and HCN4 channels are particularly sensitive to modulation. We found that both PZ and CZ calyces showed enhanced I_h in the presence of internal dB-cAMP. Cells treated with dB-cAMP showed significantly larger I_h and more depolarized V_1/2 values, and I_h activation kinetics became faster compared with control cells. Similar effects of cAMP on I_h were described in vestibular ganglion cells (12, 34) and resulted in a larger activation of resting I_h, which could lead to an increase in spike rate (10). Endogenous cyclic nucleotide levels within afferent terminals are unknown, but in vivo levels could be adjusted through neurotransmitter mechanisms. Muscarinic signaling has recently been shown to influence HCN channel activity in vestibular ganglion cell bodies, where a muscarinic acetylcholine receptor agonist produced a small positive shift in the half-activation voltage of I_h in P15 rats (13). The muscarinic effect was greatest in small to medium size ganglion neurons, which may project to afferent terminals in peripheral locations (13). Efferent terminals originating in the brain stem make cholinergic synapses with calyx terminals and may provide tonic input in vivo (58). Therefore, efferent-released acetylcholine could drive changes in HCN currents through cyclic nucleotide signaling cascades, providing an effective mechanism for regulation of spontaneous firing in calyx afferent endings. An inhibitory effect of muscarinic agonist on calyx K⁺ currents mediated by KCNQ channels in central crista regions was recently shown (36), but a direct muscarinic effect on I_h within terminals has yet to be demonstrated.

Finally, we have shown that I_h is prevalent in calyces up to P52. In spiral ganglion neurons of the cochlea, electrophysiological expression of I_h was enhanced in mice aged ∼1 yr compared with younger mice (2–3 mo). This increase in I_h expression was linked to an increase in afferent excitability and synapse degradation, perhaps contributing to age-related hearing loss (59). It remains to be established whether changes in I_h expression and associated modulation of excitability occur in vestibular afferents with age.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

This work was supported by National Institute on Deafness and other Communication Disorders (NIDCD) Grant DC DC018786 to K. J. Rennie and by National Institute on Aging Grant AG073997 (K. J. Rennie and A. Peng).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

F.L.M., T.A.V., and K.J.R. conceived and designed research; F.L.M., T.A.V., A.D., and K.J.R. performed experiments; F.L.M., T.A.V., B.G., T.A.B., A.D., and K.J.R. analyzed data; F.L.M., T.A.V., T.A.B., A.D., and K.J.R. interpreted results of experiments; F.L.M., T.A.V., A.D., and K.J.R. prepared figures; F.L.M., T.A.V., and K.J.R. drafted manuscript; F.L.M., T.A.V., T.A.B., A.D., and K.J.R. edited and revised manuscript; T.A.B., A.D., and K.J.R. approved final version of manuscript.