Sodium channel diversity in the vestibular ganglion: Na_V1.5, Na_V1.8, and tetrodotoxin-sensitive currents

Abstract

Firing patterns differ between subpopulations of vestibular primary afferent neurons. The role of sodium (Na_V) channels in this diversity has not been investigated because Na_V currents in rodent vestibular ganglion neurons (VGNs) were reported to be homogeneous, with the voltage dependence and tetrodotoxin (TTX) sensitivity of most neuronal Na_V channels. RT-PCR experiments, however, indicated expression of diverse Na_V channel subunits in the vestibular ganglion, motivating a closer look. Whole cell recordings from acutely dissociated postnatal VGNs confirmed that nearly all neurons expressed Na_V currents that are TTX-sensitive and have activation midpoints between −30 and −40 mV. In addition, however, many VGNs expressed one of two other Na_V currents. Some VGNs had a small current with properties consistent with Na_V1.5 channels: low TTX sensitivity, sensitivity to divalent cation block, and a relatively negative voltage range, and some VGNs showed Na_V1.5-like immunoreactivity. Other VGNs had a current with the properties of Na_V1.8 channels: high TTX resistance, slow time course, and a relatively depolarized voltage range. In two Na_V1.8 reporter lines, subsets of VGNs were labeled. VGNs with Na_V1.8-like TTX-resistant current also differed from other VGNs in the voltage dependence of their TTX-sensitive currents and in the voltage threshold for spiking and action potential shape. Regulated expression of Na_V channels in primary afferent neurons is likely to selectively affect firing properties that contribute to the encoding of vestibular stimuli.

the vestibular organs of the inner ear sense head orientation and motion, sending signals to the brain that drive powerful ocular, autonomic, and postural reflexes, contribute to our sense of heading, and support learning of complex tasks such as dancing (reviewed in Angelaki and Cullen 2008; Goldberg et al. 2012). Normally, we are unaware of these functions, but their significance is revealed by the vertigo, blurred vision, and disequilibrium of vestibular disorders.

Signals from vestibular hair cells are conveyed to the brainstem and cerebellum by the processes of bipolar vestibular ganglion neurons (VGNs). VGNs that contact the distinct central and peripheral zones of the vestibular sensory epithelia differ in such key features as regularity of spiking, response dynamics, expression of calcium-binding proteins, axonal diameter and conduction velocity, and central projections (reviewed in Goldberg 2000; Lysakowski and Goldberg 2004; Eatock and Songer 2011). Variation in the mechanosensory hair cells also contributes to afferent diversity: type I and type II vestibular hair cells synapse with, respectively, calyceal and bouton afferent terminals, and vestibular afferents are either calyx-only, bouton-only, or, if they contact both types of hair cell, dimorphic.

Mammalian vestibular afferents are remarkable for their high firing rates and bimodal distribution of spike regularity, either highly regular or highly irregular. Evidence from isolated VGNs suggests that the irregular spike timing of central-zone afferents depends in part on afferent expression of low-voltage-activated K (K_LV) channels (Iwasaki et al. 2008; Kalluri et al. 2010). Voltage-gated calcium (Ca_V) channels and Ca²⁺-gated K channels also vary across VGN populations (Limón et al. 2005). Because neuronal firing patterns are affected by differential expression of voltage-gated sodium channel (Na_V) subunits (Bean 2007), we have examined Na_V channel expression by VGNs. The pore-forming α-subunits of Na_V channels are categorized according to their sensitivity to block by tetrodotoxin (TTX). Most α-subunits are TTX sensitive, with half-blocking doses of ∼10 nM, but Na_V1.5 channels (TTX-insensitive) require over 10-fold larger doses, and Na_V1.8 and Na_V1.9 channels (TTX-resistant) are not blocked even by 1 μM TTX. Previous studies on VGNs have reported only TTX-sensitive Na_V currents (Chabbert et al. 1997; Risner and Holt 2006). Here we show that the rat vestibular ganglion expresses mRNA for TTX-sensitive Na_V α-subunits, as expected, but also Na_V1.5, 1.8, and 1.9 subunits. Na_V1.5 channels are strongly expressed in cardiac muscle (Gellens et al. 1992), where their current drives the upstroke of the cardiac action potential. The Na_V1.8 current was first described as a slow current specific to small peripheral nociceptive neurons of the dorsal root ganglia (DRG) and trigeminal ganglia (Akopian et al. 1996; Sangameswaran et al. 1996). We show that many early postnatal VGNs express Na_V1.5-like or Na_V1.8-like current in addition to TTX-sensitive current and that the TTX-sensitive current also differs between VGNs. The expression of different complements of Na_V channels by VGN subpopulations is likely to contribute to spiking heterogeneity that serves vestibular encoding.

MATERIALS AND METHODS

Animals were handled in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and all procedures were approved by the animal care committee at Baylor College of Medicine (RT-PCR), the Massachusetts Eye and Ear Infirmary (physiology and fixation of tissue from reporter mice), and the University of Illinois at Chicago (immunocytochemistry). Chemicals were obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise specified.

RT-PCR

Long-Evans rats, postnatal day (P) 1 or P21, were anesthetized and decapitated and the temporal bones were placed in our standard dissection and bath solution, “modified L-15” (Table 1): chilled, oxygenated Leibovitz-15 medium supplemented with 10 mM HEPES at pH 7.4, ∼325 mmol/kg. The superior division of the vestibular ganglion, which innervates the utricle, lateral, and anterior semicircular canals and part of the saccule, was dissected out bilaterally and placed in RNase/DNase-free tubes. Excess liquid was removed and the tissue samples were placed on dry ice. The methods are identical to those described for utricular maculae and cristae in Wooltorton et al. (2007). Briefly, we isolated RNA from one ganglion at a time with the RNeasy kit (Qiagen, Valencia, CA). The RNA was reverse-transcribed to complementary (c) DNA with the Advantage RT-for-PCR Kit (BD Biosystems, Palo Alto, CA) with Moloney murine leukemia virus (MMLV) reverse transcriptase and random hexamer primers. We controlled for genomic DNA contamination by designing primers to span intron-exon boundaries; homogenizing the tissue to break up genomic DNA; using RNeasy columns, which contain a silica membrane that eliminates most DNA; and treating the column with RNase-free DNase I (Qiagen) for 30 min to remove residual DNA. As the “negative RT” controls, for each primer set and tissue type we substituted water for the reverse transcriptase. No bands were detected on agarose gels with any primer set tested in these control samples (data not shown).

Table 1.

External and internal solutions

	Solution
	External
	NaCl	NMDG	KCl	CsCl	MgCl2	TEACl	4-AP	CaCl2	HEPES	d-Glucose
Cs+ external	80*	0	0	5.4	2.5	75	0	0	5	10
K+ external	100	0	5.4	0	2.5	50	10	0	5	5
NMDG+ external	0	80	0	5.4	2.5	75	0	0	5	10
Modified L-15	138†	0	5.3	0	1.8	0	0	1.3	5	0

	Internal
	NaCl	Na2-Creatine Phosphate	KCl	K2SO4	CsCl	Mg2+	ATP	LiGTP	Na-cAMP	EGTA	CaCl2	HEPES
Cs+ internal	0	3.5	0	0	148*	3.5	3.5	0.1	0.1	5 or 10	0.8	5
K+ internal	5	5	135	0	0	3.5	3.5	0.1	0.1	5 or 10	0.8	5
Perforated patch	0	0	25	75	0	5	0	0.1	0.1	5	0.1	5

Concentrations in mM. Dissections were done in modified L-15. Most voltage-clamp recordings were done with Cs⁺ internal and Cs⁺ external solutions. Current-clamp recordings of spiking were done with K⁺ internal and modified L-15. Osmolality was ∼325 mol/kgH₂O for external solutions except for NMDG⁺ external (315 mol/kgH₂O) and ∼315 mol/kgH₂O for internal solutions except for the perforated patch solution (270 mol/kgH₂O). pH was adjusted to 7.3–7.4 by adding KOH to K⁺ internal (∼17 mM), CsOH to Cs⁺ internal (∼17 mM), and NaOH to external solutions (∼4 mM), except for external solutions containing 4-AP or NMDG⁺, which were titrated to pH 7.3–7.4 with HCl. The free internal Ca²⁺ concentration for 800 μM Ca²⁺ and 5 mM EGTA was calculated as 20 nM (MaxChelator WEBMAXC software, http://maxchelator.stanford.edu/).

^*In some early experiments on the voltage dependence and tetrodotoxin (TTX) sensitivity of total Na_V current, internal CsCl was 130 mM and external NaCl was 65 mM. Inward Na⁺ currents were correspondingly smaller in those cells but no other differences were noted. These data were not included in estimates of reversal potential.

^†L-15 also contains 1.34 mM Na₂HPO₄ and 5 mM Na-pyruvate. See http://www.lifetechnologies.com/us/en/home/technical-resources/media-formulation.80.html.

PCR was done on a PTC-100 thermocycler (MJ Research, Reno, NV) with the TAQ enzyme (Applied Biosystems, Foster City, CA), the primer sets (IDT, Coralville, IA), and the PCR protocol described in Wooltorton et al. (2007). PCR products were resolved on 1.2% agarose gels and visualized with ethidium bromide.

The PCR product for each primer set was sequenced at least once; sequenced products for Na_V1.1–1.7 and β-subunits were from the utricular macula, as reported previously (Wooltorton et al. 2007). Na_V1.8 and Na_V1.9 products, which were not present in the macula, were sequenced from vestibular ganglia (SeqWright, Houston, TX). Each primer set was tested on ganglion tissue multiple times. For all but three primer sets (Na_V1.7, 1.8, and 1.9), we simultaneously tested for expression in standard tissues (brain, heart, or skeletal muscle), as a control for primer quality. Na_V1.7, 1.8, and 1.9 are weakly expressed if at all in these standard tissues.

Cell Preparation for Physiology

VGNs were isolated from Long-Evans rat pups. The superior compartment of the vestibular ganglion was dissected out of the isolated temporal bone and placed in modified L-15 (Table 1). The ganglion tissue was incubated in 0.25% trypsin and 0.05% collagenase for 30 min at 37°C and then dissociated by gentle trituration onto glass-bottomed culture dishes (MatTek, Ashland, MA). The glass was uncoated for acute preparations and coated with poly-d-lysine for overnight cultures.

Usually, VGNs were dissociated from P1–P8 rat pups, allowed to settle at room temperature (20–25°C), and recorded from between 1 and 7 h posttrituration. In a small number of experiments, dissociated VGNs from P1–P11 rats were incubated in culture medium in 5% CO₂-95% air at 37°C for ∼20 h. The culture medium was serum-free minimal essential medium (MEM) with Glutamax (Invitrogen, ThermoFisher, Waltham, MA), supplemented with 10 mM HEPES and 1% penicillin/streptomycin (Invitrogen). Data from overnight-cultured neurons are excluded from analysis except for the quantitative characterization of their TTX-sensitive Na_V currents.

Recording Solutions

During recording sessions, cells were bathed in preoxygenated L-15 medium. The recorded cell was locally superfused, usually with “Cs⁺ external” solution (Table 1), which minimized K⁺ currents by substituting Cs⁺ for K⁺ and 50–75 mM tetraethylammonium (TEA, a K channel blocker) for equivalent Na⁺. Replacing external Na⁺ with TEA⁺ also reduced total Na_V current and series resistance voltage errors. Ca²⁺ currents were minimized by substituting Mg²⁺, a Ca channel blocker, for Ca²⁺; only trace Ca²⁺ was present.

For current-clamp studies, solutions were more physiological. The external solution was modified L-15 and the internal solution was “K⁺ internal” (Table 1). After recording in current-clamp mode, we switched to voltage-clamp mode to characterize the Na_V currents, changing at the same time from modified L-15 to “K⁺ external” solution containing TEA and 4-AP to block K_V currents.

In several recordings, we used the perforated patch method to preserve the intracellular milieu (Horn and Marty 1988). The pore-forming antifungal drug (amphotericin B, 200 μg/ml) was added to an internal solution containing K₂SO₄ and KCl (Table 1).

Whole Cell Recordings

Cells were visualized at ×600 on an inverted microscope equipped with Nomarski optics (IMT-2; Olympus, Lake Success, NY). We chose round cells with diameters >10 μm (range 14–30 μm, most 18–24 μm) and sometimes partly covered by remnants of the enveloping satellite cells. Almost all of these cells had large total Na_V currents (peak values >2 nA), consistent with their being VGNs. These diameters are within the range reported by Limón et al. (2005) for VGNs from P7–P10 rats, but smaller on average, probably because our recordings were from a younger age range (P1–P8). The mean membrane capacitance was 15.2 ± 0.4 pF (n = 239), similar to the mean values we obtained previously for P1–P7 rat VGNs with either sustained or transient firing patterns (Kalluri et al. 2010). Kalluri et al. (2010) found that in the second postnatal week, VGNs that fired transiently were larger than VGNs that fired tonically. Thus, in our sample, some neurons have not reached their mature sizes.

Data were acquired with a Multiclamp 700B amplifier, Digidata 1440A digitizer, and pClamp 10 software (Axon Instruments, Molecular Devices, Sunnyvale, CA) with low-pass Bessel filtering set at 10 kHz and sampling interval set at 10 μs. Electrodes were pulled from borosilicate glass to resistances of 1–4 MΩ in our solutions and coated with a silicone elastomer (Sylgard 184; Dow Corning, Midland, MI) to reduce electrode capacitance. Voltages were corrected for a liquid junction potential between 5 and 6 mV, calculated for each solution with JPCalc software (Barry 1994) in pClamp10.

To document the incidence of each type of Na_V current, we included all VGNs for which Na_V currents could be clearly identified based on their kinetics and voltage dependence. For analyses of voltage-dependent properties and current-clamp behavior, cells were accepted if one of the following criteria was met: input resistance ≥0.5 GΩ; a giga-ohm initial seal with a smooth rupture; and a resting potential of at least −58 mV. For the VGNs analyzed for voltage-dependent properties and voltage responses in current clamp, seal resistances exceeded 1 GΩ and series resistance, R_S, ranged from 4 to 14 MΩ before electronic compensation. For ruptured-patch recordings, R_S was compensated electronically, usually by 80–90%, and the average residual R_S was 1.6 ± 0.1 MΩ (n = 198). For analysis of the relatively large total Na_V current, we reduced residual R_S to 1 MΩ or less to minimize voltage-clamp errors. The mean clamp rise time, τ_clamp = [R_s × C_m × (1-prediction)], where “prediction” is a fraction set with the amplifier, was 54 ± 2 μs (n = 190). Note also that in external solutions for voltage clamp (Table 1), replacement of one-third to one-half of Na⁺ by the K channel blocker TEA⁺ reduced Na_V current and series resistance voltage errors. The perforated-patch recordings (see above) had, as is typical, higher R_s values (mean: 29 ± 4.2 MΩ, n = 11, compensated by 50–85% to 5–15 MΩ), and were not used to measure kinetics or activation.

The holding potential (V_H) in voltage-clamp mode was −70 mV. Most recordings were at room temperature (range 20–25°C, mean 23.9 ± 0.3°C, n = 130), because fast Na_V currents are difficult to voltage clamp and characterize at body temperature. For some recordings (indicated), we heated the bath to 37°C with a heated platform and temperature controller (TC-344B; Warner Instruments, Hamden, CT). Note that VGNs exhibit similar firing pattern categories (“transient” vs. “sustained”) at room and body temperatures (Kalluri et al. 2010).

Pharmacology

Frozen stock solutions of TTX (2 mM or 200 μM in distilled water) were thawed and added to 10 ml of external solution to make the specified concentrations on the day of the experiment. Drugs were applied via controlled local perfusion (Valvelink 8; AutoMate Scientific, Berkeley, CA). Multiple lines were merged at a manifold that flowed into a single tip (∼3-cm long, 250-μm diameter). This dead volume provided a temporal separation between any transient mechanical artifacts induced by line switching and the onset of the drug effect. Perfusion of control solution before the test solution provided an additional control for flow effects. To visualize flow, we added 2 μl of latex beads (0.46-μm mean particle size) to 10 ml of the control and test solutions. Cells were confirmed to be in the path of bead movement and the flow was adjusted to a gentle but steady rate. Wash-in and wash-out of the drug were tracked by a protocol that applied a test voltage step at 10-s intervals. Recordings were analyzed only for cells with stable R_s values in both control and drug solutions. Many cells also provided stable recordings after wash-out of the drug.

Analysis

Analysis was performed in Matlab (The MathWorks, Natick, MA) and Clampfit (Axon Instruments, Molecular Devices). Figures were prepared with Origin software (versions 8–15; OriginLab, Northampton, MA). Means ± SE are presented. The voltage protocol comprised, in sequence, a 25- to 80-ms hyperpolarizing prepulse (from V_H to −120, −125, or −130 mV) to relieve channel inactivation; a test step of iterated voltage (duration determined by the time course of the current under study, voltage iterated from −125 mV to +30 mV), and a brief tail step at a constant voltage (stepping from the test step to −15, −20, or −30 mV). The voltage dependence of activation and inactivation was analyzed only when R_s voltage error was <5 mV at peak current and τ_clamp was <100 μs. For the comparatively small TTX-insensitive and TTX-resistant currents, R_s voltage errors were typically <3 mV and were not corrected post hoc.

Activation curves of conductance, g, vs. voltage, V, were generated by plotting the peak current (I) against step voltage, fitting the linear upper region of the V- or U-shaped I-V curve (typically between 0 and +30 mV) to find the reversal potential (E_rev) and then dividing the peak current by the driving force (V-E_rev) to obtain g. The activation curves were fit by a Boltzmann function:

$$g(V)=\frac{g_{\max }}{1+e^{(V_{\text{act}}-V)/s}}$$ (1)

where g_max is maximum conductance, V_act is voltage of half-maximal activation, and s is the slope factor. Inactivation curves were generated by plotting the peak current produced by a test pulse (to −30, −20, or −10 mV) against the iterated prepulse voltage and fitting with a Boltzmann function:

$$I(V)=\frac{I_{\max }}{1+e^{(V-V_{inact})\hspace{0.17em}/s}}$$ (2)

with V_inact the voltage of half-maximal inactivation.

We fit drug dose-response curves with the Hill equation with Hill coefficient of 1:

$$y=\frac{[\text{drug}]}{K_{\text{d}}+[\text{drug}]}$$ (3)

where y is the fraction of channels blocked by drug. This equation assumes simple single-site block with an excess of drug relative to sodium channels and no conductance in the blocked state. It is successful at approximating TTX dose-response relations for Na_V channels (e.g., Ritchie 1979; Du et al. 2009). Rearranging the Hill equation allows us to calculate the dissociation constant (K_d) from the block at a single drug dose:

$$K_{\text{d}}=[\text{drug}]\times \frac{(1-y)}{y}$$ (4)

For TTX block of total current in cells with both TTX-sensitive and TTX-insensitive currents, we fit the dose-response relation with:

$$y=\frac{A[\text{drug}]}{K_{\text{d1}}+[\text{drug}]}+\frac{(1-A)[\text{drug}]}{K_{\text{d2}}+[\text{drug}]}$$ (5)

where K_d1 and K_d2 are the dissociation constants for TTX binding of the two components, and A and (1 − A) are the relative fractions of the two components.

To explore the effects of TTX on inactivation curves of cells with a mix of TTX-sensitive and TTX-insensitive current, we performed a simple simulation. Combining Eqs. 2 and 3 for two components of current, with A being the fraction of TTX-insensitive current and (1 − A) the fraction of TTX-sensitive current, yields the following equation for total unblocked conductance at each drug concentration, with the first and second terms representing the TTX-insensitive and TTX-sensitive currents, respectively:

$$g(V,[\text{drug}])=\left(1-\frac{[\text{drug}]}{K_{\text{d1}}+[\text{drug}]}\right)\left(\frac{A}{1+e^{\frac{(V-V_{\text{inact1}})}{s1}}}\right)+\left(1-\frac{[\text{drug}]}{K_{\text{d2}}+[\text{drug}]}\right)\left(\frac{1-A}{1+e^{\frac{(V-V_{\text{inact2}})}{s2}}}\right)$$ (6)

V_inact and s values for each current were taken from their average values in Table 2. A and the K_d values were estimated by fitting dose-response data from VGNs (see results).

Table 2.

Electrophysiological properties of Na_V currents in vestibular ganglion neurons

	Inactivation		Activation			Time Course
Current/VGN Subpopulation	V1/2, mV	s, mV	V1/2, mV	s, mV	G density, nS/pF	tpeak, 0 mV, ms	tpeak, −15 mV, ms	τinact, 0 mV, ms
NaV (total)
VGNs with INaI	−76.9 ± 1.5 (13)	7.5 ± 0.4 (13)	−40.1 ± 2.0 (6)	6.0 ± 0.6 (6)	8.4 ± 0.9 (6)	0.33 ± 0.03 (8)	0.42 ± 0.03 (10)	0.26 ± 0.02 (10)
VGNs without INaI or INaR	−79.5 ± 1.3 (3)	6.6 ± 0.2 (3)	−38.6 (1)	6.2 (1)	7.24 (1)	0.25 ± 0.003 (3)	0.31 ± 0.01 (5)	0.21 ± 0.01 (4)
cultured VGNs	−76.3 ± 0.2 (12)	7.6 ± 0.1 (12)	−36.5 ± 1.6 (11)	5.7 ± 0.4 (11)	8.7 ± 1.0 (11)	0.36 ± 0.01 (8)	0.45 ± 0.01 (8)	0.28 ± 0.01 (8)
NaS
VGNs with INaI	−74.6 ± 1.5 (13)	7.0 ± 0.3 (13)	−39.8 ± 2.5 (6)	5.7 ± 0.8 (6)	6.1 ± 0.4 (6)	0.32 ± 0.03 (7)	0.40 ± 0.04 (8)	0.25 ± 0.02 (7)
VGNs with INaR	−65.8 ± 1.6 (8)	10.5 ± 0.3 (8)	−32.3 ± 0.7 (6)	5.9 ± 0.5 (6)	3.1 ± 0.34 (5)	0.45 ± 0.04 (5)	0.57 ± 0.03 (5)	0.312 ± 0.01 (5)
NaI
VGNs with INaI + INaS	−102.1 ± 1.3 (23)	9.9 ± 0.6 (23)	−47.5 ± 1.6 (9)	8.2 ± 0.4 (9)	1.0 ± 0.2 (9)	0.47 ± 0.03 (9)	0.53 ± 0.03 (11)	0.27 ± 0.03 (8)
NaR
VGNs with INaR + INaS	−31.4 ± 0.6 (23)	5.2 ± 0.2 (23)	−16.5 ± 0.6 (8)	4.5 ± 0.2 (8)	1.5 ± 0.3 (7)	1.4 ± 0.1 (6)	3.1 ± 0.2 (6)	2.4 ± 0.1 (11)

Values are means ± SE (n). Sample sizes were smaller for activation parameters because higher recording quality is required. Room temperature, Cs⁺ internal, and Cs⁺ external solutions (Table 1). VGN, vestibular ganglion neurons; see text for additional definitions.

Na_V1.8 Reporter Lines

We looked for evidence of Na_V1.8 expression in two reporter mice: 1) an Na_V1.8 reporter mouse generated by crossing the BAC transgenic “SNS-Cre” line (Agarwal et al. 2004) with a reporter line Rosa26^LSL-tdTomato (Jackson Laboratory) in which a loxP-flanked STOP cassette prevented transcription of tdTomato, a variant of red fluorescent protein, under the CAG promoter (Madisen et al. 2009); and 2) the Na_V1.8-Cre line, in which targeted knockin of Cre replaces the Na_V1.8 sequence at its endogenous locus (Stirling et al. 2005). The mice used were hemizygous for SNS-Cre or heterozygous for Na_V1.8-Cre and heterozygous for tdTomato. As a positive control for Na_V1.8 expression and confirmation of genotyping, we dissected the trigeminal ganglia and DRG from the same animals; Na_V1.8 is well documented in somatosensory neurons, including trigeminal ganglion cell bodies (Thun et al. 2009).

Reporter mouse tissues were also labeled with antibodies against calretinin, neurofilament 200 (NF200), and β-III tubulin, as described next.

Immunohistochemistry

Long-Evans rats (P3-P21) were deeply anesthetized with gaseous isoflurane or Nembutal (sodium pentobarbital, 80 mg/kg ip) and then either decapitated (younger animals, ages ≤ P6) or transcardially perfused (older animals, ages >P6) with 2 ml/g body wt of an aldehyde fixative [4% paraformaldehyde, 1% acrolein, 1% picric acid, and 5% sucrose in 0.1 M phosphate buffer (PB), pH 7.4]. For young animals, the vestibular ganglia were quickly exposed and then tissue was transferred to 4% paraformaldehyde fixative for 20 min followed by rinses in phosphate-buffered saline (PBS, 0.01 M).

For SNS-Cre reporter mice, vestibular ganglia were dissected in L-15 and then fixed in 4% paraformaldehyde for 2–3 h. For Na_V1.8-Cre reporter mice, similar results were obtained whether the tissue was first dissected in L-15 and transferred to fix or fixed by transcardial perfusion of aldehyde fixative. Background fluorescence was reduced by incubating the tissues in a 1% aqueous solution of sodium borohydride for 10 min.

For Na_V1.5 immunoreactivity and tissue from SNS-Cre mice, frozen sections (35 μm) were cut with a sliding microtome. Na_V1.8-Cre ganglia were studied as whole mounts. Antibodies were from Chemicon (Temecula, CA) unless otherwise specified. Immunocytochemistry was performed on free-floating sections, permeabilized with Triton X-100 in a blocking solution of 0.5% fish gelatin and 1% BSA in PBS. Samples of vestibular tissues were incubated with Triton X-100 at concentrations that depended on age: P1: 0.3% overnight at 4°C; P3: 0.5% for 3 h at room temperature; P8: 2% for 1 h, room temperature; and P21 and adult: 4% for 1 h, room temperature. Samples were then incubated with a cocktail of two primary antibodies: goat anti-calretinin and rabbit anti-Na_V1.5 (1:200 in the blocking solution) for 2 days at 4°C with 0.1, 0.3, and 0.5% Triton X-100 for P1, P3, and P8-adult sections, respectively. Specific labeling was revealed with a cocktail of two secondary antibodies: fluorescein-conjugated donkey anti-goat and rhodamine-conjugated donkey anti-rabbit or Alexa Fluor 594-conjugated donkey anti-mouse (from Molecular Probes, Eugene, OR), all at 1:200 in the blocking solution. Sections were rinsed with PBS between and after incubations and mounted on slides in Mowiol (Calbiochem, Darmstadt, Germany). The sections were examined on a laser scanning confocal microscope (LSM 510 META or LSM 710; Carl Zeiss, Oberköchen, Germany). Final image processing was done with Adobe Photoshop CS4 software (San Jose, CA).

For Na_V1.5 immunostaining, two kinds of control reactions were done: 1) no primary antibody; and 2) primary antibody preincubated with the antigenic peptide (10 μg/1 μg antibody) for 1 h at room temperature. Images comparing labeling in control and test conditions were acquired and digitally processed identically.

In addition, Western blots, performed as described in Wooltorton et al. (2007), yielded a band at the appropriate size (227 kDa) for inner ear and control tissue (heart) prepared from adult rats.

RESULTS

Expression of Na_V Channel Subunits in the Vestibular Ganglion

RT-PCR revealed robust expression of Na_V pore-forming (α) subunits and accessory (β) subunits in the superior compartment of the vestibular ganglion, which contains neuronal cell bodies and their satellite cells (Fig. 1). Data were obtained from the ganglion at P1 and at P21. At P1, all known Na_V α-subunits were detected; at P21 all were detected except Na_V1.4, the skeletal muscle subunit. Most α-subunits (Na_V1.1–1.4, 1.6, and 1.7) carry TTX-sensitive currents. Na_V1.5 carries TTX-insensitive current and is prevalent in cardiac muscle but also reported in neurons, including somatosensory ganglion neurons (Renganathan et al. 2002), and in hair cells of the rat utricular macula (Wooltorton et al. 2007). Na_V1.8 and 1.9 carry TTX-resistant current in somatosensory ganglion neurons. All four β-subunits were expressed.

Fig. 1.

Vestibular ganglia expressed mRNA for multiple Na channel α- and β-subunits at both postnatal day (P) 1 and P21. A: sketch of the mammalian inner ear [adapted from Fig. 7 by M. Brödel in Hardy (1934) with permission from John Wiley and Sons]. The superior division of the vestibular ganglion was taken from P1 and P21 rats for RT-PCR experiments. B: representative agarose gels of PCR products (40 amplification cycles) are shown for individual ganglia at P1 (“G1”) and P21 (“G21”) for each α-subunit (1 to 9) and each β-subunit (1–4). Results from P1 utricular epithelia (“U”) processed in the same experimental series serve as negative controls for positive ganglion expression in some cases (Na_V1.3, 1.8, and 1.9). Each ganglion was processed in an independent experiment. PCR products were detected in most ganglia. Products for α1.1, 1.2, 1.3, 1.5, 1.7, and β1, 3, and 4 were detected in every ganglion (n = 5). Products for α1.8 and 1.9 were detected in every P1 ganglion (n = 4 and 5, respectively). At P21, α1.8 was detected in 4 of 5 ganglia and α1.9 was detected in 3 of 4 ganglia. α1.6 was detected in 4 of 5 ganglia at P1 and β2 was detected in 2 of 3 ganglia at P21. Product for Na_V1.4 was detected in 3 of 5 ganglia at P1 but in 0 of 2 ganglia at P21. Controls: water (no template) negative controls are shown in Wooltorton et al. (2007) (their Fig. 11, A and B). (−)RT: controls, in which reverse transcriptase was left out, were also done for each tissue (not shown). Bands were verified by sequencing. [Some RT-PCR bands were previously published in Fig. 9, A and B, of Wooltorton et al. (2007) on Na_V channels in utricular hair cells: all of the utricular bands (“U”) and the bands in ganglia for Na_V1.7, 1.8, and 1.9 subunits. The ganglion data served in Wooltorton et al. (2007) as positive controls for negative results in the sensory epithelia of utricles and semicircular canals.]

The positive results for Na_V1.5, Na_V1.8, and Na_V1.9 primers were remarkable given previous reports that all Na_V current in VGNs was TTX sensitive. These RT-PCR results on vestibular ganglia were obtained in parallel with RT-PCR results for vestibular (utricular and semicircular canal) epithelia and controls that are reported in Wooltorton et al. (2007). Water (no template) controls and −RT (no reverse transcriptase) controls were negative. In vestibular epithelia, all β-subunits and most α-subunits were detected at P1 and/or P21, but Na_V1.8 and Na_V1.9 subunits were never detected. Correspondingly, no TTX-resistant currents were recorded in utricular hair cells. These clear negative results serve as an additional control for the strong Na_V1.8 and Na_V1.9 expression of the vestibular ganglia (Fig. 1B). Positive support for the Na_V1.5 result in vestibular ganglia is provided by strong Na_V1.5-like immunoreactivity in the distal calyceal terminals (Wooltorton et al. 2007; Lysakowski et al. 2011) of VGNs. In this report, we provide further support for expression of Na_V1.5 and Na_V1.8 subunits in vestibular ganglion cell bodies in the form of whole cell currents, immunostaining of the cell bodies (Na_V1.5), and results with reporter mice (Na_V1.8).

To look for functional expression of Na_V channels in vestibular afferents, we dissociated VGNs from P1–P8 rats and recorded whole cell currents (n = 239 VGNs) within 7 h of dissociation. Dissociated neuronal somata are more accessible than neuronal terminals in situ and allow better space clamp. Recordings were restricted to the early postnatal period because increasing myelination of vestibular ganglion cell bodies (Dechesne et al. 1987) interferes with patching the neuronal membrane at later ages. In some previous reports, VGNs have been studied after overnight culturing, which reduces myelin and satellite cell coverage of the cell bodies. As described later, we found that culturing overnight may affect expression of Na_V channels and have excluded data from overnight-cultured neurons except where indicated.

In all VGNs, fast inactivating inward current was evoked by depolarizing voltage steps following a hyperpolarizing prepulse (Fig. 2). Most recordings were made in solutions designed to minimize K⁺ and Ca²⁺ currents (Cs⁺ internal solution and Cs⁺ external solution with TEA and only trace Ca²⁺; Table 1). Experiments described later confirm that these are Na⁺ currents. We refer to the total voltage-dependent Na⁺ current as I_NaV. Application of the classic Na_V channel blocker TTX revealed that I_NaV comprised currents with three different TTX sensitivities in different combinations. TTX-sensitive current (I_NaS) was present in all VGNs. In 32 of 109 (29%) VGNs, the total inactivating current was fully blocked by 300 nM TTX (Fig. 2A). Currents that were not TTX sensitive fell into two distinct categories. In many VGNs (69/111, 62%), a fast inward TTX-insensitive current (I_NaI) remained in 300 nM TTX (Fig. 2B). I_NaI was a small proportion of total I_NaV at −20 mV (Fig. 2B, left) but a much larger proportion at −58 mV (Fig. 2B, right), indicating that I_NaI had a negatively shifted activation range compared with I_NaS. Other VGNs (47/178, 26%) had I_NaR, a TTX-resistant inward current that was not blocked by micromolar levels of TTX and had slower kinetics and a less negative voltage range than either I_NaS or I_NaI (Fig. 2C). (Sample sizes differ because different conditions were used to detect different Na_V components.) In no case did we detect both I_NaI and I_NaR in a single cell, although small levels of expression cannot be ruled out. Evidence on the identities of the channels carrying I_NaI and I_NaR is provided in sequence in the following sections. In addition, we show that the properties of I_NaS differed depending on whether I_NaR was also present.

Fig. 2.

Pharmacological dissection with tetrodotoxin (TTX) revealed 3 types of Na⁺ current composition in vestibular ganglion neurons (VGNs), as shown by these representative examples. Legend in C applies to A–C. Recording solutions in this and all subsequent figures were Cs⁺ internal and Cs⁺ external (Table 1) unless otherwise specified; channel blockers were added to the external solution. V_H = −70 mV in all figures. Voltage was first stepped to −130 mV to de-inactivate I_NaV, then stepped to a depolarized value. All currents are shown on the same time scale, but amplitude has been scaled for display. Transient capacitive artifact at the voltage step was truncated in this and subsequent figures. A: I_NaV was completely blocked by 300 nM TTX in some VGNs, indicating that it was entirely TTX sensitive (I_NaS). Data are from P3 VGN. B: some VGNs expressed a mixture of I_NaS and a more negatively activating TTX-insensitive current (I_NaI). I_NaI was small relative to total I_NaV for a step to −20 mV (left) but more prominent for a step to −58 mV (right, expanded current scale) because it activated at more negative voltages. Data are from P2 VGN. C: other VGNs had a mixture of I_NaS and a slower, more positively activating TTX-resistant current (I_NaR). Note that the residual inward current in 300 nM TTX (red) was little affected by increasing TTX to 5 μM (overlapping blue trace). Data are from P7 VGN. Gray, difference current between the control current and the current in 300 nM TTX.

I_NaI Is Carried by Na_v1.5 Channels

I_NaI resembled current carried by Na_V1.5 channels in its TTX sensitivity, voltage dependence and kinetics, and sensitivity to block by divalent cations.

Reported K_d values for TTX block are 1–10 nM for TTX-sensitive currents and ∼100-fold higher for Na_v1.5 current (Cribbs et al. 1990). To look for Na_V1.5 current, we applied 300 nM TTX, which would block most TTX-sensitive current while significantly sparing any Na_V1.5 current. The small current remaining in 300 nM TTX (I_NaI) activated and inactivated at substantially more negative voltages than total I_NaV in 0 TTX did (Fig. 2B; Fig. 3A and B), indicating that I_NaI was a distinct current component. I_NaI was blocked by a higher concentration of TTX (5 μM) and eliminated by replacing Na⁺ with the impermeant ion NMDG⁺ (Fig. 3C), confirming that it was a Na⁺ current. Even in VGNs with I_NaI, 300 nM TTX blocked total I_NaV by ∼90% (at −25 to −35 mV, Fig. 3A3), indicating that most of the Na_V current was TTX sensitive. In control conditions, therefore, I_NaI was masked by the much larger I_NaS, such that the voltage dependence of total I_NaV was similar to that of TTX-sensitive channels (Fig. 3, B1 and B2, and Table 2). The masking of I_NaI by I_NaS in inactivation curves in 0 TTX is explored in a later section.

Fig. 3.

Some VGNs expressed TTX-insensitive Na⁺ current (I_NaI) with the properties of Na_V1.5 channels, in addition to I_NaS. A and B: current remaining in 300 nM TTX (I_NaI) had more negative voltage dependence than total Na_V current. Data are from P2 VGN. A1: inactivation: as prepulse voltages (bottom) became less negative, the peak current evoked by the step to −30 mV decreased in both control and 300 nM TTX solutions but with different voltage dependence: a prepulse to −102 mV (red traces) decreased total I_NaV only slightly (top) relative to the peak value (blue traces, prepulse to −130 mV) but inactivated approximately half of I_NaI (bottom). A2: activation: at −58 mV (red traces), I_NaV was one-quarter maximal but I_NaI was nearly half-maximal. A3: current-voltage relation: peak inward current in control and 300 nM TTX solutions (top) for the VGN in A1 and A2 shows a small TTX-insensitive component (red) that activated at relatively negative potentials (y-scale expanded, bottom). B: conductance-voltage (G-V) relations for activation and inactivation. B1: normalized plots of activation (peak current vs. test step, open symbols) and inactivation (peak current vs. prepulse, filled symbols) and Boltzmann function fits (curves); for the VGN in A. Gray triangles, total I_NaV; red circles, I_NaI. B2: population data. Thin curves, all fits for room temperature, ruptured-patch data. Thin gray curves, total I_NaV for VGNs with both I_NaS and I_NaI or just I_NaS. Thin red curves, I_NaI, obtained in 300 nM TTX. Thick black and red curves, curves generated with average fit parameters, Table 2. B3: I_NaI voltage range was still negative when recorded at body temperature (red curves) or at room temperature with the perforated patch method (dark cyan curves; inactivation curves only). Thick gray curves, population averages for I_NaI from B2 (ruptured patch, room temperature). C: I_NaI in 300 nM TTX (black) was strongly blocked by 5 μM TTX (blue) and abolished by replacing external Na⁺ with NMDG⁺ (red, no TTX present), demonstrating that it is carried by Na⁺ ions flowing through Na_V channels (K⁺ internal, K⁺ external, P7). D: voltage range of inactivation recorded in 300 nM TTX did not shift negatively with time, as illustrated by inactivation curves taken at different times since patch rupture. Negative inactivation ranges were recorded immediately post rupture: inverted red open triangles, 300 nM TTX, 10 s after rupture. In 0 TTX, inactivation ranges of total I_NaV did not shift negatively with time: cyan triangles and purple diamonds, 2 VGNs, 7.5 min since rupture; green triangles, a 3rd VGN, 13 min since rupture. Removal of TTX shifted inactivation curves positively despite increasing time since rupture: black squares, in 300 nM TTX; gray circles, the same VGN later, after stopping TTX application. E: I_NaI (isolated in 300 nM TTX) was partially and reversibly blocked by 200 μM external Cd²⁺. The Cd²⁺-blocked (difference) current (E1, P4) had similar time course (E2, normalized E1 data) and voltage dependence (E3, different VGN, P2) to the current in 300 nM TTX, consistent with them being one and the same.

The activation and inactivation midpoints of I_NaI were, respectively, ∼10 and ∼25 mV negative to the midpoints for the total current (Table 2; P = 0.009 relative to I_NaS in cells without I_NaR). Reported V_inact values for Na_V1.5 channels range from −106 to −82 mV (DiFrancesco et al. 1985; Frelin et al. 1986; Sheets and Hanck 1992, 1999; Yamamoto et al. 1993; Wang et al. 1996; Cummins et al. 1998; O'Leary 1998; Zhang et al. 1999; Kuo et al. 2002; Wooltorton et al. 2007), compared with −35 to −78 mV for TTX-sensitive channels (Noda et al. 1986; Sangameswaran et al. 1997; Dietrich et al. 1998; Smith and Goldin 1998; Chen et al. 2000; Moran et al. 2003) and approximately −30 mV for Na_V1.8 channels (discussed later). Thus the V_inact value of I_NaI is consistent with Na_V1.5 and no other Na_V α-subunits.

Since I_NaI activated at significantly more negative voltages than other Na_V currents, it contributed significantly to Na⁺ current evoked by steps from hyperpolarized voltages to near resting potential (Fig. 2B, right). The averaged inactivation and activation curves cross over at −71.5 mV for I_NaI and −56.0 mV for I_NaV, with crossover conductance levels at 6% of peak and correspondingly small window currents (Fig. 3B2). Resting membrane potentials range from −75 to −45 mV in isolated VGNs of the first postnatal week (present study and Kalluri et al. 2010) and could be less negative in vivo under the influence of excitatory synaptic input from hair cells. For resting potentials positive to −70 mV, our results indicate that I_NaI channels would be mostly inactivated. We investigated two factors that might negatively shift the voltage range of Na_V channels in our experiments: temperature and second messenger availability. Increasing temperature from room to 37°C can shift the inactivation range of Na_V channels positively (Ruppersberg et al. 1987; Oliver et al. 1997). For I_NaI in our cells, however, V_inact was similar at 37°C (−103.3 ± 1.5 mV, n = 6) and room temperature (−100.6 ± 1.6 mV, n = 15; P = 0.34; Fig. 3B3).

Alternatively, loss of intracellular signaling molecules during ruptured-patch recording might shift V_inact negatively (e.g., Penniman et al. 2010). To test for such washout effects, we recorded from three VGNs with the perforated-patch method. The inactivation range of I_NaI was still very negative (V_inact = −101.8 ± 4.1 mV; no significant difference from ruptured-patch recordings, P = 0.78; Fig. 3B3). Ruptured-patch data in Fig. 3D further indicate that the negative inactivation range was not a washout or rundown artifact: 1) the negative inactivation range could be seen within seconds of patch rupture (red inverted triangles); 2) in 0 TTX, the inactivation range of total current, I_NaV, did not shift negatively in ∼10 min following patch rupture (purple, cyan, and green curves show data from 3 cells); 3) removing TTX from the external solution moved the inactivation range of total Na_V current positively with time (compare black squares, taken early in TTX, and gray circles, taken later from the same VGN after removing TTX).

Another property of Na_V1.5 currents that distinguishes them from TTX-sensitive currents is a relatively high sensitivity to block by Cd²⁺ and Zn²⁺ (Backx et al. 1992; Chen et al. 1992; Heinemann et al. 1992; Satin et al. 1992). Reported half-blocking concentrations (IC₅₀) of Cd²⁺ for Na_V1.5 current are 50–250 μM (DiFrancesco et al. 1985; Scornik et al. 2006; Wooltorton et al. 2008), in contrast to 5 mM for TTX-sensitive currents (Frelin et al. 1986; Roy and Narahashi 1992). Like Na_V1.5 current, I_NaI was ∼half-blocked by 200 μM Cd²⁺ (Fig. 3E), a dose that barely affects TTX-sensitive current. We first isolated I_NaI by recording in Cs⁺-based external solution with 300 nM TTX and then added 200 μM CdCl₂ to the external solution (Fig. 3E1). In 13 cells, 200 μM Cd²⁺ reversibly blocked I_NaI at −20 mV by 53 ± 4%. Assuming that the block is of a single current, the measured block yields a K_d of 179 μM (Eq. 4), consistent with the reported range for Na_V1.5 channels. (Note that the K_d for Cd²⁺ may be lower in 0 TTX because Cd²⁺ competes with TTX for a common binding site; Doyle et al. 1993; Renganathan et al. 2002.) I_NaS at −20 mV was negligibly blocked by 200 μM Cd²⁺ in three of three cells (block of 0.6 ± 4.7%). The current blocked by 200 μM Cd²⁺ resembled the total current in 300 nM TTX (I_NaI) in its kinetics (Fig. 3E2, normalized currents) and voltage dependence of inactivation (Fig. 3E3), supporting the interpretation that I_NaI is dominated by a Cd²⁺-sensitive inward current with fast kinetics and relatively negative voltage range.

Given that I_NaI was partly blocked by 200 μM Cd²⁺, a known blocker of voltage-gated Ca²⁺ channels, how can we be certain that it was not carried by Ca_V channels? This conclusion is based on multiple arguments: 1) I_NaI was almost entirely blocked at 5–10 μM TTX (n = 7; Fig. 3C). In two cells, we were able to test for reversibility and in both cases the additional block by 5–10 μM TTX could be reversed and re-applied multiple times. The difference current obtained by subtracting current in 5 or 10 μM TTX from current in 300 nM TTX had similar biophysical properties to I_NaI, consistent with I_NaI being primarily a single TTX-blockable current. 2) I_NaI was eliminated by replacing external Na⁺ with the impermeant ion NMDG⁺ (Fig. 3C). This effect was almost fully reversible (>95%, 3 VGNs). 3) The biophysical properties of I_NaI were not consistent with those of inactivating T-type Ca²⁺ currents. I_NaI inactivated with a time constant of ∼300 μs at 0 mV and was half-inactivated at approximately −100 mV (Table 2), in contrast to T-type Ca²⁺ currents, which have inactivation time constants ranging from ten to hundreds of milliseconds (depending on voltage and subunit) and V_inact values from −47 to −86 mV (Coulter et al. 1989; Monteil et al. 2000; Chemin et al. 2001; Gomora et al. 2002; Díaz et al. 2005; Nelson et al. 2005; Vitko et al. 2005; Emerick et al. 2006; Zhong et al. 2006). The fast inactivation of L-type Ca²⁺ current is relatively slow, with time constants of ∼15 ms or more (Beuckelmann et al. 1991; Mewes and Ravens 1994; Magyar et al. 2000) and is Ca²⁺ dependent, with reduced extent and speed of inactivation when Ba²⁺ or Na⁺ (in μM Ca²⁺) is the charge carrier (Brehm and Eckert 1978; Lee et al. 1985; Ferreira et al. 2003; Brunet et al. 2009). In contrast, we observed rapid and near complete inactivation of I_NaI when Na⁺ was the charge carrier and current was eliminated by Na⁺ replacement (Fig. 3C). 4) E_rev for I_NaI was consistent with a Na⁺ current. For our usual recording solutions (Cs⁺ internal and Cs⁺ external, Table 1), we estimated E_rev from I-V curves (see materials and methods) as +49 ± 2 mV (n = 10) for total Na_V current and +50 ± 5 mV (n = 6) for I_NaI. E_Na calculated from the nominal Na⁺ concentrations (Table 2) equaled +62 mV but E_rev is expected to be lower because of nonnegligible Cs⁺ permeability, P_Cs. Ratios of P_Cs to P_Na for mammalian cardiac Na_V currents, which I_NaI closely resembles, range from 0.007 to 0.06, with a mean across studies of 0.02 (reviewed in Kurata et al. 1999). In our solutions, the predicted E_rev for a permeability ratio of 0.02 is +53 mV, close to the value (+50 mV) extrapolated from the I_NaI I-V relations. 5) Our external solutions for recording Na_V currents contained only trace (no added) Ca²⁺. Ca_V channels do become permeable to Na⁺ at concentrations of Ca²⁺ <10 μM (Almers and McCleskey 1984; Hess and Tsien 1984), but to achieve such low concentrations usually requires chelators, which we did not include. Moreover, the voltage dependence of Ca_V channels is similar whether the current is carried by Na⁺ or Ca²⁺ (Fukushima and Hagiwara 1985), and, as already noted, the voltage dependence of I_NaI is very different from that of T-type channels. Our external solution included 3.5 mM Mg²⁺ to help counter the effects of low Ca²⁺ on apparent voltage dependence (Frankenhaeuser and Hodgkin 1957) and to block Ca_V channels (Fukushima and Hagiwara 1985; Kurejová et al. 2007). It is improbable that I_NaI, which was as large as 2 nA (below), was carried by trace Ca²⁺ or by Na⁺ through partly blocked Ca_V channels.

Sizes and Proportions of I_NaV, I_NaS, and I_NaI Current

Total peak current (I_NaV) for cells without I_NaR was −13.1 ± 1.8 nA (n = 6; range −5.7 to −17.0 nA) in K⁺ internal solution and K⁺ external solution (100 mM Na⁺) and −11.7 ± 0.9 nA (n = 21) in Cs⁺ internal solution and Cs⁺ external solution (80 mM Na⁺). [Note that external Na⁺ was reduced from physiological levels (∼145 mM) to reduce current size and associated voltage-clamp errors.] The mean peak current in 300 nM TTX (I_NaI) was −1.2 ± 0.2 nA (n = 6) (range −0.6 to −2.2 nA) for K⁺ internal and K⁺ external solutions and −834 ± 115 pA (n = 24) for Cs⁺ internal and Cs⁺ external solutions. Peak I_NaS, estimated by subtracting the current in 300 nM TTX from the control condition, was −12.3 ± 1.6 nA (n = 14; all solutions). Because I_NaI was partly blocked in 300 nM TTX, these estimates underestimate I_NaI and overestimate I_NaS.

We estimated proportions of I_NaS and I_NaI in VGNs with both currents from the dose-response curve for the total Na_V current at −15 mV. A dose-response curve of 11 normalized data points pooled from five cells (Fig. 4A) is better fit by a two-component model (Eq. 5) than a one-component model (Eq. 3); the deviation of the data from a single-component fit exceeds the variability between the data points. According to the two-component fit, I_NaI was 8.4% of the total current and had an IC₅₀ of 774 nM; I_NaS constituted the remainder with an IC₅₀ of 8 nM. These values are consistent with the range reported for Na_V1.5 currents (150 nM to 2 μM; reviewed in Wooltorton et al. 2008) and the range of 1–10 nM typical of TTX-sensitive currents.

Fig. 4.

TTX-insensitive current was ∼10% of the total Na⁺ current in VGNs with I_NaI and was not readily evident in the absence of TTX. A: TTX dose-response data (11 data points from 5 VGNs) deviated from a single-component fit to the Hill equation (Eq. 3, blue, K_d = 13 nM) but was well fit by a 2-component fit (Eq. 5, red), with K_d1 = 8 nM, K_d2 = 774 nM, and A (fraction of TTX-sensitive component) = 0.92. B: inactivation curves recorded for different TTX doses are consistent with the predictions of the 2-component fit. B1: voltage dependence of inactivation at 5 TTX concentrations, simulated with Eq. 6: 10% I_NaI (V_inact = −96 mV, s = 8 mV, K_d2 = 700 nM) and 90% I_NaS (V_inact = −74 mV, s = 5.6 mV, K_d1 = 5 nM). Parameter values for TTX dependence were close to the results of the 2-component fit (K_d1, K_d2, A). Inactivation parameters were the means obtained for I_NaI and I_NaS (Table 2). B2 and B3 inactivation curves from 2 representative VGNs (P2 and P1, respectively) at different TTX doses resemble the simulations in B1 at corresponding TTX doses. Key colors in B1 apply to B1-B3.

If I_NaI is present within the total Na_V current, why is it not visible as a separate component in the inactivation curves? To explore this question, we used Eq. 6 to simulate, for various TTX concentrations, the inactivation curve for total Na_V current in a cell with 10% I_NaI plus 90% I_NaS and K_ds of 800 and 5 nM, respectively, similar to the IC₅₀s of the two-component fit in Fig. 4A. The inactivation midpoints and slopes of the two inactivation components in Eq. 6 were set to the average values for I_NaI and for I_NaS in VGNs with I_NaI (Table 2). The effect of I_NaI on the shape of the simulated inactivation curve at different TTX doses (Fig. 4B1) reproduces data (Fig. 4, B2 and B3) well and shows that in such a mixture, I_NaI cannot be distinguished as a separate component, especially at 0 TTX where I_NaS dominates. At moderate TTX doses (100–300 nM), V_inact shifted negatively and the slope decreased, consistent with a more balanced contribution of two currents. The mean s value for the inactivation curve was 7.4 ± 0.2 mV (n = 27) at 0 TTX; in contrast, s values were between 12.4 and 13.4 in four of five cells in 100 nM TTX. At high TTX doses (1 μM), I_NaI dominated.

Although I_NaI was just 10% of total Na_V current at −15 mV, its maximum size of ∼200 to ∼2,000 pA in 100 mM external Na⁺ (and larger in physiological Na⁺ concentrations) is well above current thresholds for spiking, particularly in VGNs of the first postnatal week (10–100 pA, Kalluri et al. 2010). The mean I_NaI conductance density was 1.0 ± 0.2 nS/pF (n = 9), 8% of the mean total Na_V conductance density per cell (13.0 ± 2.6 nS/pF, n = 9).

Our results suggest several reasons that Na_V1.5 current has not been reported previously in VGNs. First, even in cells with I_NaI, the total current had similar biophysical properties to I_NaS, as predicted for cells with a ∼10% contribution of I_NaI (Fig. 4B). Second, the very negative inactivation range of I_NaI required a deeply hyperpolarized prepulse (to −130 mV in Fig. 3A1) to relieve inactivation and fully reveal the full conductance. Using less negative prepulses (Chabbert et al. 1997; Risner and Holt 2006) would reduce the pool of Na_V1.5 channels available for activation. Third, the practice of culturing ganglion neurons overnight to reduce satellite cell coverage may suppress I_NaI expression either by the conditions or the passage of time ex vivo: in 14 VGNs that we recorded from after overnight culture, I_NaI was small or undetectable. We also did not observe I_NaR in cultured cells but did not specifically test for it. Fourth, it is not uncommon in Na_V channel studies to add Cd²⁺ to block Ca_V channels, and this would block Na_V1.5 channels.

Finally, VGN maturation may play a factor in whether I_NaI is detected in VGNs. In our sample, the incidence of I_NaI in VGN cell bodies decreased toward the end of the first postnatal week (Fig. 5, dark bars). To estimate the incidence of I_NaI, we counted only cells that had total peak I_NaV larger than 4 nA in 0 TTX; for smaller I_NaV, the voltage dependence of the much smaller TTX-insensitive component was difficult to analyze. I_NaI was identified by its TTX insensitivity or Cd²⁺ sensitivity and its kinetics and voltage range. For VGNs dissociated on P1–P5, 60–85% of the recorded VGNs had detectable I_NaI. By P6–P8, the incidence dropped to 30–40%.

Fig. 5.

The incidence of TTX-insensitive and TTX-resistant currents in our samples of isolated rat VGNs changed over the 1st postnatal week. Total number of VGNs at each age is shown above each column.

Na_V1.5-Like Immunoreactivity in VGN Somata

The calyceal terminals of VGNs on type I hair cells in vestibular epithelia are strongly stained by Na_V1.5 antibody during development (Wooltorton et al. 2007) and in adult (Lysakowski et al. 2011). With the same antibody, we also observed immunoreactivity in cell bodies within the ganglion, examined from ages P3–P21 (Fig. 6). Use of calretinin antibody at P8 and P21 showed that some Na_V1.5-immunoreactive VGNs were also immunoreactive for calretinin, a marker for the large, irregularly-firing “calyx-only” afferents that form calyceal terminals on type I hair cells in the central zones of vestibular epithelia (Desmadryl and Dechesne 1992; Kevetter and Leonard 2002; Leonard and Kevetter 2002; Desai et al. 2005a,b).

Fig. 6.

Na_V1.5-like immunoreactivity (red) in the rat superior vestibular ganglion at different ages. Calretinin antibody (green) was added at P8 (B) and P21 (C) to label the relatively large somata of calyx-only afferents, which selectively innervate type I hair cells in central zones of vestibular epithelia. At P3 (A), no calretinin antibody was applied because vestibular afferents are not calretinin-immunopositive. Arrows, examples of bright Na_V1.5 staining at each age. Arrowheads, dimmer staining. The 20-μm scale bar in C applies to A–C.

I_NaR Is Carried by Na_v1.8 Channels

In approximately one-quarter (47/178) of VGNs, depolarizing voltage steps in 300 nM TTX evoked a relatively slow inactivating inward current, I_NaR, which resembled currents carried by Na_V1.8 (SNS, PN3) channels, first identified in DRG (Akopian et al. 1996; Sangameswaran et al. 1996). Distinctive properties of Na_V1.8 current include TTX resistance, relatively positive voltage dependence, slow time course, and rapid recovery from inactivation. I_NaR was readily detected in the presence of other Na_V currents by its distinctive voltage range and time course (Figs. 2 and 7, A–C). In all cases of I_NaR recorded in 0 TTX (n = 20), I_NaS was also present. In all cases of I_NaR recorded in 300 nM TTX to block most I_NaS, either no fast component (9 cells) or only a very small fast component (12 cells) could be discerned. Where it was possible to analyze that fast component (5 cells), it resembled I_NaS rather than I_NaI. Thus we recorded no examples of I_NaR and I_NaI in the same cell.

Fig. 7.

Some VGNs expressed a TTX-resistant Na_V current (I_NaR) with properties of Na_V1.8 channels. A: in responses to an inactivation protocol with V_H −70 mV, 80-ms prepulse, and test step to −15 mV, I_NaR was distinguished from I_NaS by its less negative voltage dependence and slower time course for P6 VGN. Vertical scale bar applies to A1-A3; A1 has a shorter time scale than A2 and A3. A1 and A2: control data (K⁺ internal, K⁺ external). A1's expanded time scale shows separation of I_NaS (blue dashed line) and I_NaR (red dashed line) with voltage and time, allowing estimation of their amplitudes in 0 TTX (see results). After a prepulse to −40 mV (red traces), I_NaS was almost completely inactivated but I_NaR was minimally inactivated; thus peak current after a −40-mV prepulse provided an estimate of I_NaR. Conversely, the difference current (blue dashed line) between −120- and −40-mV prepulses was almost entirely I_NaS. This method was validated by blocking I_NaS with 300 nM TTX (A.3), which did not affect current after the −40-mV prepulse (red trace, compare to control), confirming that it was I_NaR. B: voltage ranges of activation and inactivation. Solutions: Cs⁺ internal, Cs⁺ external. Top: peak I-V relations in control and 300 nM TTX, for the cell in A. Bottom left: activation data (circles) and inactivation data (triangles) for I_NaS obtained by subtracting traces in 300 nM TTX from control traces (black) and for I_NaR obtained in 300 nM TTX (red), for the cell in A. Data were fitted with Eqs. 1 and 2. I_NaR inactivation curve: V_inact = −32.1 mV, s = 5.4 mV; activation curve: V_act = −14.8 mV, s = 5.4 mV, g_max = 51.1 nS. I_NaS inactivation curve: V_inact = −63.6 mV, s = 11.9 mV; activation curve: V_act = 33.4 mV, s = 4.3 mV, g_max = 106.0 nS. For B, g was normalized to g_max. Bottom right: superimposed activation and inactivation curves from individual fits (thin gray lines, room temperature) and average fits (thick black lines) are shown. Average parameter values are in Table 2. Thin red lines, data at 37°C, not in average. C: time constants of fast inactivation of I_NaR decreased 10-fold to ∼2 ms as voltage was made less negative over the range from approximate V_inact to +15 mV; prepulse was −125 mV, 50 ms. Numbers of values averaged (one per VGN) are indicated above each data point. Inset: much smaller averaged values for I_NaS and I_NaR on an expanded ordinate. D: I_NaR was blocked only slightly more (7%) by 5 μM TTX than by 300 nM TTX. Data are from P7 VGN. E: I_NaR was completely and reversibly eliminated by replacing external Na⁺ with NMDG⁺. Data are from P5 VGN. F: I_NaR recovery at −90 mV from inactivation during 25-ms steps at 0 mV, in 4 VGNs. I_NaR recovered ∼75% within ms (left) but took seconds to recover fully (right). Solutions were Cs⁺ internal and Cs⁺ external except for blue triangles: K⁺ internal, K⁺ external (P7). Open circles, P7 (same VGN at left and right). Red inverted triangles, P2. Black triangles, P5.

For analyses of biophysical properties, we used a subset of 23 VGNs that satisfied two criteria: almost all I_NaS was blocked, and unblocked outward current that overlapped in time and voltage dependence with I_NaR was at least 15-fold smaller than the peak inward I_NaR. The voltage dependence (Fig. 7B) and time course (Fig. 7, A and C) of I_NaR (Table 2) were similar to reported values for Na_V1.8 current in somatosensory afferent neurons.

Inactivation and activation curves of I_NaR were well fit by single Boltzmann functions (Fig. 7B, bottom left), with midpoints of −31 and −16 mV (Table 2), comparable to values between −25 to −35 mV and −15 to −20 mV, respectively, for Na_V1.8 currents in DRG neurons (Akopian et al. 1996; Cummins and Waxman 1997; Sangameswaran et al. 1997; Leffler et al. 2005). Na_V1.9 channels (reviewed in Dib-Hajj et al. 2015) have substantially more negative midpoints of activation and inactivation (−50 to −60 mV) and a larger window current. Relative to I_NaI and I_NaS, the time to peak, t_peak, for I_NaR was three- to fourfold slower at 0 mV and eightfold slower at −15 mV (Table 2) and consistent with t_peak values for Na_V1.8 channels of ∼2 ms at ∼0 mV (Elliott and Elliott 1993; Scholz et al. 1998). Inactivation of I_NaR was well fit by a single exponential with a time constant of ∼2.5 ms at 0 mV and ∼8 ms at −15 mV, slower than I_NaS and I_NaI (Table 2 and Fig. 7C) and close to inactivation time constants of Na_V1.8 currents (3.5 ms at 0 mV, 9 ms at −15 mV; Renganathan et al. 2002).

I_NaR was highly resistant to TTX. In Fig. 7D, increasing the TTX concentration from 300 nM to 5 μM, a dose that blocked most of I_NaI, produced only 7% additional block of I_NaR. Solving Eq. 4 for the effect yields a K_d of 61 μM, within the expected range for either Na_V1.8 channels or Na_V1.9 channels (IC₅₀ values 40 to >100 μM; Akopian et al. 1996; Sangameswaran et al. 1996; Rabert et al. 1998; Cummins et al. 1999).

I_NaR was carried by Na⁺. Substitution of external Na⁺ with NMDG⁺ reversibly eliminated the current (Fig. 7E; eliminated in 13/13 cells; reversed in 7 of 7 cells that were tested for reversibility). E_rev for I_NaR in Cs⁺ external solution was +50 ± 2 mV (n = 9), similar to E_rev estimates for total I_NaV and I_NaI (49−50 mV, above).

Na_V1.8 current recovers from fast inactivation an order of magnitude faster than TTX-sensitive current recovers at the same voltage (Elliott and Elliott 1993), consistent with the general observation that time constants for macroscopic transitions of channel states become faster away from the midpoint of the inactivation range. The fast recovery time constant of I_NaR from inactivation at −90 mV (1.0 ± 0.2 ms, n = 3, Fig. 7F, left) is consistent with recovery time constants for Na_V1.8 currents in small DRG neurons (∼1 ms at −100 mV; Cummins and Waxman 1997) and 4 ms at −67 mV (Elliott and Elliott 1993) and one-quarter the recovery time constants for TTX-sensitive current in small DRG neurons (Everill et al. 2001).

In addition to the inactivation described above, Na_V channels can enter into a distinct slow inactivated state, lasting seconds, which probably involves a different channel conformation (reviewed in Ulbricht 2005). Na_V1.8 is unusual among Na_V subunits for its rapid entry into slow inactivation from which it recovers slowly, even for relatively short depolarizations, contributing to adaptation of responses to physiological stimuli (Blair and Bean 2003). In our protocols probing recovery from inactivation (Fig. 7F, left), only ∼75% of I_NaR recovered with the fast time constant. The remainder of the recovery occurred with a slow time constant (1,060 ± 196 ms, n = 4), even for brief inactivating steps (Fig. 7F, right).

In summary, the properties of I_NaR are consistent with previous reports of current through Na_V1.8 channels. As noted for I_NaI, it is very unlikely that Ca_V channels were involved, for similar reasons: I_NaR is carried by Na⁺, there was only trace Ca²⁺ in the external medium, and the voltage dependence and time course differed substantially from reported values for T-type (Ca_V3.1) Ca²⁺ currents, including T-type Ca²⁺ currents in immature mouse vestibular ganglion cells (Desmadryl et al. 1997; Chambard et al. 1999; Autret et al. 2005; Limón et al. 2005). Furthermore, the fast time constant for recovery from inactivation was much faster for I_NaR than for T-type current (1 ms at −90 mV, Fig. 7F, vs. 25–50 ms at −100 mV; Hering et al. 2004).

Size, Proportion, and Incidence of I_NaR

The average peak I_NaR was −2,962 ± 642 pA (n = 4; range −1,620 to −4,470 pA) or −109 ± 16.6 pA/pF in K⁺ internal and K⁺ external solutions and −1,978 ± 231 pA or −78 ± 9.4 pA/pF (n = 7) in Cs⁺ internal and Cs⁺ external solutions. Peak I_NaR conductance was 32.9 ± 4.0 nS and peak conductance density was 1.47 ± 0.25 nS/pF (n = 7; Cs⁺ internal and Cs⁺ external solutions). Since we excluded cells for which peak I_NaR was <15 times the residual (unblocked) outward current, these values may be larger than the true mean.

In five cells, we could segregate I_NaR and I_NaS pharmacologically by applying high doses of TTX to reveal I_NaR and subtracting to obtain I_NaS. In five other cells, we estimated the ratio of peak I_NaS to peak I_NaR by comparing peak I_NaV at −15 mV after a prepulse to −120 mV, which inactivated neither current, and after a prepulse to −40 mV, which inactivated I_NaS fully but I_NaR only by 15% (method illustrated in Fig. 7A1). To calculate proportions of each current, we corrected for the differences in percent maximal activation at the test voltage of −15 mV (100% for I_NaS vs. 76 ± 4%, n = 13, for I_NaR, Fig. 7B) and for the 15% inactivation of I_NaR at −40 mV. The method was validated by applying it to VGNs for which we also had pharmacological isolation of the currents, which showed that similar isolation was obtained (compare the red curves in Fig. 7, A2 and A3). The ratios of peak I_NaS to peak I_NaR did not differ significantly between data collected with and without TTX and were pooled, yielding a mean value of 3.8 ± 0.6 (range 1.8 to 7.5). Thus peak I_NaS was at least twice as large as I_NaR in a given cell. Peak current ratios may underestimate the relative influence of the TTX-resistant current, which, being slow, prolongs depolarization and its influence over other voltage-gated channels and calcium influx.

The incidence of I_NaR in our sample as a function of postnatal age increased rapidly from zero at P1 to ∼30% by P3–P4 (Fig. 5, hatched bars), in contrast to the decreasing incidence of I_NaI. Thus, by the latter part of the first postnatal week, approximately one-third of cells expressed I_NaI and I_NaS, one-third expressed I_NaR and I_NaS, and one-third expressed only I_NaS.

TTX-Sensitive Currents Differed Between VGNs With and Without TTX-Resistant Current

I_NaS was isolated by subtraction of records in 100–300 nM TTX from control records (0 TTX). I_NaS in VGNs with I_NaR differed from I_NaS in VGNs with I_NaI (Fig. 8 and Table 2) in size (about half) and voltage dependence: less negative (P = 0.005 for V_inact; P = 0.004 for V_act) and broader inactivation range (s_inact, P = 0.0007) and was similarly distinct from I_NaS in cells with neither I_NaI nor I_NaR (Table 2). These differences might arise from the expression of different combinations of TTX-sensitive subunits, for which there are a number of candidates (Fig. 1B).

Fig. 8.

The TTX-sensitive current differed between cells with TTX-resistant current and cells without TTX-resistant current. I_NaS in VGNs with I_NaR (red) had a less negative voltage dependence and broader inactivation range than I_NaS in VGNs with only I_NaS (black), and in VGNs with I_NaI (blue). Thin lines, Boltzmann fits for individual cells; thick lines, mean curves. See Table 2 for Boltzmann parameters, numbers of cells in each group, and means ± SE.

In another difference, I_NaS showed use-dependent decline over the course of our standard voltage protocol in many VGNs with I_NaR but in no others. The ratio of peak I_NaS for the 1st vs. 24th sweeps in the protocol was significantly larger (1.28 ± 0.04) for 6 VGNs with I_NaR than for 15 cells lacking I_NaR (1.07 ± 0.04, P = 0.003). Use-dependent decline was described for TTX-sensitive currents in adult DRG cell bodies (Rush et al. 1998).

Firing Properties of Cells with I_NaR

Eighty VGNs were studied in current-clamp mode with more physiological solutions (K⁺ internal and modified L-15, Table 1) and no ion channel blockers. Current-clamp recordings were made immediately upon rupture to reduce possible effects of washout of endogenous molecules on firing patterns. After recording firing patterns, we switched to voltage-clamp mode and K⁺ external solution with K channel blockers and no added Ca²⁺ (Table 1) to determine whether I_NaR was present. I_NaR was recognized by its distinctive voltage and time dependence.

The most obvious differences in firing were between VGNs that expressed I_NaR (n = 11 VGNs) and VGNs that lacked I_NaR (n = 29 VGNs). Differences in other currents, including the differences in I_NaS described above, may contribute. Nevertheless, certain firing properties of VGNs with I_NaR resemble those in Na_V1.8-expressing DRG neurons. The depolarized voltage dependence and fast recovery from inactivation of Na_V1.8 channels are thought to contribute to tonic firing during sustained depolarizations. The slower time course produces broad spikes with an inflection or hump during the repolarization phase (reviewed in Rush et al. 2007). These features were observed in spikes from VGNs with I_NaR, as shown in Fig. 9.

Fig. 9.

Spike shape and firing pattern differed between VGNs with I_NaR and VGNs without I_NaR. K⁺ internal, modified L-15 external. A: representative examples of voltage responses to 500-μs current steps are shown at firing threshold: 1,150 pA for a VGN with I_NaR and 400 pA for a VGN without I_NaR. VGNs with I_NaR had broader action potentials with less negative voltage thresholds and a shoulder during repolarization (arrow). Both cells: P7. Presence of I_NaR for each VGN was determined in voltage clamp after changing to K⁺ external with K channel blockers. B: firing patterns evoked by depolarizing current steps, 200 ms (B1) or 400 ms (B2, B3) long. VGNs with I_NaR fired either 1–2 spikes (B1) or multiple spikes (B2, B3). For multispiking VGNs, the presence of I_NaR was correlated with constant spike height throughout the train. Current steps of 160 pA (B1), 240 pA (B2), and 130 pA (B3) were selected to yield similar steady-state voltage. Insets: individual first spikes from the same trains. B1: P6. B2: P7. B3: P8. Scale bars in B1, including inset, apply to B2 and B3.

The current pulses we used to evoke action potentials were brief (500 μs) to minimize the influence of injected current on the action potential (Fig. 9A); large currents were required to achieve threshold with such brief pulses. The current amplitude was increased in 100-pA increments until a spike was evoked. The properties of this spike were compared for five VGNs with I_NaR and six VGNs without I_NaR at 23–25°C. The voltage threshold of the action potential (voltage at which dV/dt reached 4 mV/ms, chosen to be clearly above noise) was 10 mV less negative in cells with I_NaR (−44.2 ± 2.5 mV) than in cells without I_NaR (−53.8 ± 1.1 mV, P = 0.0003). The peak dV/dt on the upstroke of the action potential was twice as fast for cells without I_NaR (207 ± 24 mV/ms) than for cells with I_NaR (99 ± 11 mV/ms; P = 0.004). These differences may reflect the smaller size and less negative activation range of I_NaS in cells with I_NaR, combined with a strong contribution from I_NaR once voltage exceeds spike threshold. Spikes in VGNs with I_NaR were of similar height to, but broader than, spikes in VGNs without I_NaR (Fig. 9A). Spike height, measured from the spike peak to the afterhyperpolarization trough, was 105.5 ± 2.1 mV for cells with I_NaR and 102.9 ± 3.1 mV for cells without I_NaR. Spike width at half-height was ∼2–4 ms (2.64 ± 0.31 ms) for cells with I_NaR and ∼1 ms (1.07 ± 0.08 ms, P = 0.006) for cells without I_NaR. The slower kinetics and more depolarized voltage dependence of inactivation of I_NaR may broaden spikes, as reported in DRG and nodose ganglion neurons that have Na_V1.8 current (Ritter and Mendell 1992; Djouhri et al. 2003). In addition to being broader, spikes in every VGN with I_NaR had a shoulder on the repolarization phase (n = 11 VGNs, Fig. 9A, arrow) while VGNs without I_NaR did not (n = 29). Although shoulders are often caused by Ca²⁺ currents, the slow properties of Na_V1.8 current can contribute to shoulders (Ritter and Mendell 1992; Cardenas et al. 1997). In DRG neurons with Na_V1.8 channels, ∼60% of the current during the shoulder flows through Na_V1.8 channels, with the rest mostly through high-voltage Ca_V channels (Blair and Bean 2002).

The current threshold for spiking evoked by longer current steps (Fig. 9B) delivered in 10-pA increments was +30 ± 4 pA in 4 VGNs with I_NaR and +24 ± 5 pA in 24 VGNs without I_NaR (NS, P = 0.32). The firing patterns of VGNs in response to depolarizing current steps have been subdivided into “transient” or “single-spiker” (1–2 spikes at the start of the current step) and “sustained” or “multispiker” (Iwasaki et al. 2008; Kalluri et al. 2010). It is hypothesized that transient patterns are typical of irregularly firing afferents from central epithelial zones and sustained patterns are made by regularly firing afferents from peripheral epithelial zones (Iwasaki et al. 2008; Kalluri et al. 2010). In our limited dataset, we did not see a simple correlation between I_NaR and either the “transient” or “sustained” firing pattern. In VGNs with I_NaR (n = 11), firing patterns ranged from a single spike (Fig. 9B1) to multiple spikes that terminated before the end of the step (Fig. 9, B2 and B3): none showed a fully sustained response that lasted the duration of the step. In all six multispiking VGNs with I_NaR, successive spikes were of similar height until spiking abruptly stopped (Fig. 9B2), in contrast to the decrementing spikes of multispiking VGNs without I_NaR (n = 24) (Fig. 9B3). A possible reason for the abrupt termination is accumulating inactivation of TTX-sensitive currents. Because of their more negative activation range, TTX-sensitive currents may dominate the initial phase of the spike around threshold, but TTX-resistant currents should contribute strongly later in the spike (Blair and Bean 2002). In this way, inactivation of TTX-sensitive currents may increase spike threshold until spiking stops without much affecting spike height.

Despite a significant overlap of activation and inactivation ranges (a “window current”), the activation range of I_NaR was positive relative to most resting potentials. Mean resting potentials were −68.2 ± 2.1 mV (n = 6) for cells with I_NaR and −71.6 ± 0.7 mV (n = 26) for cells without I_NaR (P = 0.19). Thus the transient component of the current played little role in resting potential in these isolated neurons, although persistent current might (Han et al. 2015). We observed only two spontaneously firing VGNs in this study, neither of which had I_NaR.

Na_v1.8 Expression in Reporter Mice

Our electrophysiological data suggest that TTX-resistant currents carried by Na_V1.8 channels are expressed by a VGN subpopulation that also differs from other neurons in the properties of TTX-sensitive current and spikes. To visualize this subpopulation, we examined the vestibular ganglia of two mouse lines in which cells that expressed Na_V1.8 (SNS) also expressed the red fluorescent protein tdTomato (see materials and methods). We refer to the BAC transgenic line as SNS-Cre (Agarwal et al. 2004) and the targeted knockin as Na_V1.8-Cre (Stirling et al. 2005).

Confocal images in Fig. 10 illustrate the tdTomato signal in three vestibular ganglia and one DRG. Figure 10A1 shows an SNS-Cre/Rosa26^LSL-tdTomato vestibular ganglion also immunostained for calretinin, which is selectively expressed by calyx-only afferents, and β-III tubulin, which labels most VGNs (Perry et al. 2003). The staining patterns for tdTomato, calretinin, and β-III tubulin for this section are shown individually in Fig. 10, A2-A4. Figure 10B1 shows a second SNS-Cre vestibular ganglion with tdTomato and immunostaining for the heavy neurofilament NF200; Fig. 10B2 is the same section showing tdTomato alone. NF200-like immunoreactivity is strongest in large VGN somata (Ylikoski et al. 1990; Usami et al. 1993a,b), similar to the pattern in the DRG (Lawson and Waddell 1991). tdTomato expression was largely nonoverlapping with intense immunostaining in large VGNs for calretinin and for NF-200; arrows point to an exemplar large VGN that is positive for calretinin (tissue in Fig. 10A) or NF200 (tissue in Fig. 10B) but not tdTomato.

Fig. 10.

In 2 Na_V1.8 reporter mouse lines, different percentages of vestibular ganglion neurons are labeled. A and B: 35-μm sections from the superior vestibular ganglia (VG) of 4- to 6-wk SNS-Cre reporter mice, showing tdTomato (red) in many VGNs. Scale bars (50 μm) in A4 and B2 apply to A1-A4 and B1-B2. A1: SNS-Cre VG colabeled with antibodies to calretinin (blue) and β-III tubulin (green). A2-A4: color channels of A1 shown separately. Calretinin-positive cells (A3) are relatively large and not tdTomato-positive (e.g., arrow). B1: SNS-Cre section from a different mouse, colabeled with NF200 antibody (green). Intensely NF200-positive cells (e.g., arrow) are often large and not tdTomato-positive. B2: same section showing tdTomato alone. C: tdTomato expression in VG (C1) and DRG (C2) of a 5- to 6-wk Na_V1.8-Cre reporter mouse. Scale bars = 50 μm. C1: maximum intensity projection image of the VG. C2: whole mount DRG from the same animal. Despite strong tdTomato label in the DRG, fewer VGNs are positive in this reporter line than in the SNS-Cre mouse.

In vestibular ganglia of the Na_V1.8-Cre mouse line, far fewer neurons were tdTomato-positive (Fig. 10C1), although DRGs from the same animals (Fig. 10C2) had many tdTomato-positive neurons, as did DRGs in the SNS-Cre line (not shown). Differences between transgenic and knockin reporter lines are not unexpected; which is more representative of Na_V1.8 expression requires more study.

DISCUSSION

Na_V Channel Diversity in the Vestibular Ganglion

As reported previously, rodent vestibular ganglion cell bodies have large TTX-sensitive currents (Chabbert et al. 1997; Risner and Holt 2006). A closer look has uncovered heterogeneity. RT-PCR of whole vestibular ganglia revealed mRNA expression of all known Na_V channel subunits, including TTX-insensitive and TTX-resistant α-subunits. Whole cell recordings then provided evidence for functional expression of multiple Na_V currents.

In acutely dissociated VGNs between P3 and P8, before maturation of the inner ear, we isolated TTX-insensitive and TTX-resistant currents that resembled currents carried by Na_V1.5 and Na_V1.8 α-subunits, respectively, in their kinetics and pharmacology. The expression of these subunits was further supported by Na_V1.5-like immunoreactivity and by labeling in two kinds of Nav1.8 reporter mice. Whole cell recordings suggest that Na_V1.5 current is widely expressed in neonatal afferent cell bodies but declines over the first postnatal week, while Na_V1.8 is upregulated in some VGNs over the same period. We further found that the TTX-sensitive Na⁺ currents are heterogeneous, having smaller amplitude, less negative voltage dependence, and less steep voltage dependence of inactivation in VGNs with TTX-resistant current than in other VGNs. The RT-PCR results suggest multiple candidates for the pore-forming α-subunits that carry TTX-sensitive currents, and it is possible that VGNs with TTX-resistant current express a different complement of TTX-sensitive subunits than VGNs without TTX-resistant current. Differences in β-subunit expression also affect properties of TTX-sensitive currents (Namadurai et al. 2015). A likely contributor to TTX-sensitive current in vestibular afferents is the Na_V1.6 subunit, which is expressed at the afferent heminode adjacent to hair cell synapses (Hossain et al. 2005; Lyakowski et al. 2011).

Expression of Na_v1.8 in Peripheral Sensory Neurons

The properties of I_NaR are highly consistent with Na_V1.8 channels but RT-PCR data also showed robust expression of Na_V1.9 (NaN) mRNA in vestibular ganglia. While we cannot rule out a contribution from Na_V1.9 channels to I_NaR, the voltage and time dependence are better matched by known properties of Na_V1.8 channels, and we have no clear indication of more than one TTX-resistant current. Perhaps in vivo or with further maturation, Na_V1.9 channels in VGNs produce the persistent current for which they are known in nociceptors (Dibb-Hajj et al. 2015), which could support the high sustained firing of vestibular afferents.

Our data may be the first voltage-clamp characterizations of Na_V1.8-like current in a ganglion without classic nociceptors. Na_V1.8 channels were first isolated from rat DRG (Akopian et al. 1996; Sangameswaran et al. 1996) and were originally thought to be selectively expressed by unmyelinated small nociceptive neurons. It is now clear that Na_V1.8 channels are also expressed by other sensory neurons. The SNS-Cre transgenic reporter mouse showed transgene expression in 28% of large-diameter DRG neurons in addition to the expected high percentage (93%) of small-diameter DRG neurons (Agarwal et al. 2004). In the Na_V1.8-Cre reporter mouse line, many nonnociceptive DRG neurons are tdTomato-positive and have Na_V1.8-like currents (Shields et al. 2012), consistent with other reports of TTX-resistant current in large DRG neurons (Renganathan et al. 2000; Ho and O'Leary 2011). Outside the DRG, the TTX-resistant current is found in nonnociceptive neurons of the nodose ganglion (Li and Schild 2007; Kwong et al. 2008). In addition, Na_V1.8-like immunoreactivity has been reported in the retina (O'Brien et al. 2008). The apparently widespread distribution of Na_V1.8 and TTX-resistant currents in sensory neurons argues for a broad sensory function not restricted to nociception. Thus candidate pain therapies targeting Na_V1.8 channels (Liu and Wood 2011) should be tested for diverse sensory side effects, including vestibular symptoms. Potential side effects arising from drug action on Na_V1.8 channels in the vestibular ganglion may be avoided by controlling the permeability of agents through the blood-labyrinth barrier (Soto and Vega 2010).

TTX-resistant current has been proposed to play a role in sensitization in inflammatory hyperalgesia and possibly neuropathic pain (Tanaka et al. 1998; Gold 1999; Cardenas et al. 2001; Coggeshall et al. 2004). If so, it might also contribute to the neuritis and vertigo that can follow infection and inflammation in the vestibular nerve (Strupp and Brandt 2013).

Functional Significance of Na_V Channel Diversity in the Vestibular Ganglion

Afferents from central and peripheral zones of vestibular sensory epithelia show strong differences in spike timing (regular vs. irregular) and response dynamics (frequency-dependent gain and phase) (Goldberg 2000). Action potentials in isolated VGNs with regular and irregular firing have different current and voltage thresholds (Kalluri et al. 2010). The unique voltage-gated properties of both Na_V1.5 and Na_V1.8 channels could play a role in these systematic variations in spiking and stimulus encoding, as well as in development.

Na_V1.5 channels.

Na_V1.5 protein may play roles in both vestibular development and mature vestibular function. In cardiac myocytes, with resting potentials of −80 to −90 mV, Na_V1.5 channels are largely responsible for the upstroke of the action potential. In isolated immature rat VGNs, resting potentials range from −45 to −75 mV (Kalluri et al. 2010), so that most Na_V1.5 channels are inactivated at resting potential. Nevertheless, for those VGNs with the most negative resting potentials and/or in the case of hyperpolarizations from rest, Na_V1.5 channels may be critical for excitability because their activation range is also more negative than that of other Na_V channels. In vivo, vestibular afferents may spend significant time at relatively negative potentials, as follows. First, they have high background firing rates (50–100 spikes/s in mammals) and each spike afterhyperpolarization can be 10 mV negative to resting potential (Schessel et al. 1991; Kalluri et al. 2010). Second, negatively directed hair bundle deflections reduce the excitatory glutamatergic input to afferent terminals. Third, resting potential in mature vestibular afferents in vivo may be more negative than we measured, for two reasons: external K⁺ concentration has been estimated at 3 mM (Anniko and Wroblewski 1986), about half the value in our external solutions, which would make E_K and therefore resting potential more negative; and addition of K_LV channels to VGN cell bodies during the first two postnatal weeks (Kalluri et al. 2010) may drive resting potentials closer to E_K.

Na_V1.5 channels may also play a role in development, either through conducting or nonconducting functions (Kaczmarek 2006). Na_V1.5-null mice die between embryonic day 10 and 11 (Papadatos et al. 2002), and knocking out Na_V1.5 homologs in zebrafish causes severe morphological defects (Chopra et al. 2010). Na_V channel β-subunits are part of the cell-adhesion molecule superfamily and regulate neurite outgrowth and axon pathfinding by mechanisms that depend on Na⁺ current (Davis et al. 2004; Brackenbury et al. 2008, 2010). In our sample, a high incidence of Na_V1.5-like current at birth declined during the first postnatal week (Fig. 5), although Na_V1.5-like immunoreactivity was seen at P21 (Fig. 6C). The incidence of Na_V1.5 current in rodent DRG drops from ∼80% at E15 to ∼20% by P0 and ∼3% by adulthood (Renganathan et al. 2002). The expression of Na_V1.5 currents in hair cells of the rat utricle also declines with postnatal age (Wooltorton et al. 2007).

The decreasing incidence of I_NaI in VGNs over the first postnatal week could, alternatively or additionally, reflect trafficking of the channels to afferent terminals. Unlike central neurons, vestibular afferent neurons are bipolar neurons that initiate spikes near their distal terminals on hair cells, as indicated by nodal markers (Lysakowski et al. 2011). Na_V1.5 mRNA was present in the vestibular ganglion at both P1 and P21 (Fig. 1). Na_V1.5-like immunoreactivity in afferent calyceal terminals increases in intensity during postnatal development (Wooltorton et al. 2007) and is present in adult (Lysakowski et al. 2011).

Na_V1.8 channels.

Varying the proportions and types of TTX-sensitive channels in conjunction with TTX-resistant channels could alter firing threshold and sensitivity while ensuring overshooting action potentials. The depolarized activation range of Na_V1.8 may help reduce depolarization block and maintain action potential height during intense stimulation. As in other peripheral sensory neurons, the spike initiation zones of vestibular afferents lie close to the afferent synapse (Lysakowski et al. 2011) and may therefore be particularly prone to depolarization block from sustained synaptic input. VGNs with TTX-resistant current maintained spike height in the presence of a significant sustained depolarization (Fig. 9B2).

The percentage of DRG neurons expressing Na_V1.8 current increases dramatically around the time of birth and persists (Renganathan et al. 2002). We saw an increase in I_NaR prevalence just after birth (Fig. 5) but do not know whether the current persists into adulthood. The Cre reporter mouse indicates whether a cell expressed Na_V1.8 at some point in the past but does not indicate the level of expression or whether expression continues into adulthood. Fewer VGNs were labeled in the knockin Na_V1.8-Cre reporter mouse than in the transgenic SNS-cre reporter mouse, despite similar fractions of labeled DRG neurons in the two lines (Chiu et al. 2014). Expression may be lower in VGNs than in highly expressing DRG subpopulations; expression of Na_V1.8 may be regulated differently in neurons originating from the neural crest (DRG) and placodes (nodose and vestibular ganglia) (Lu et al. 2015). The transgenic mouse may have a higher copy number, while the knockin has a regulatory context closer to native conditions. In the same Na_V1.8-Cre reporter mouse line, extensive tdTomato staining was reported in “the vestibular-geniculate ganglion mass” of adult mice (Gautron et al. 2011).

Function depends on whether the Na_V channel location is somatic, at the initial heminode, or at nodes of Ranvier. In the immature VGNs that we studied, myelination is not complete and the distribution of ion channels may be immature. Na_V1.8 channels cluster at nodes of Ranvier in a subpopulation of myelinated axons in radicular tooth pulp (Henry et al. 2005), consistent with a role in action potential propagation. Even when ion channels are normally restricted to nodes, the retraction of neurites in the process of dissociation may relocate ion channels to the soma.

Relationship to afferent populations.

Unlike DRG neurons, all vestibular afferents receive input from a single kind of mechanoreceptor, the hair cell, and are myelinated. Vestibular afferents are systematically heterogeneous, however, forming parallel streams with different spike timing, spatial and temporal integration, frequency dependence, sensitivity, and dynamic range (reviewed in Goldberg 2000 and Eatock and Songer 2011). Here we speculate about how Na_V1.5, Na_V1.8, and TTX-sensitive subunits of different voltage dependence might complement the expression of low-voltage-activated (K_LV) channels to permit different vestibular afferent types to spike in different voltage regimes.

In vestibular ganglia from SNS-cre reporter mice, costaining for calretinin suggests that Na_V1.8 is most strongly expressed by calretinin-negative afferents (Fig. 10A), and therefore may include the regular afferents that innervate peripheral epithelial zones. The ability of Na_V1.8 channels, with their depolarized inactivation ranges, to counter depolarization block (Patrick Harty and Waxman 2007; Choi and Waxman 2011) could help regular vestibular afferents sustain high tonic firing. Extrapolating from the association of Na_V1.8 currents with TTX-sensitive currents of relatively depolarized voltage dependence (Fig. 8), suggests a combination that would give (some) regular afferents more depolarized spike thresholds than irregular afferents. Isolated VGNs that fire trains in response to current steps, which may correspond to regular afferents, do have less negative spike thresholds than single-spiking VGNs, which may correspond to irregular afferents (Kalluri et al. 2010). Lower expression of K_LV channels in regular afferents (Kalluri et al. 2010; Lysakowski et al. 2011; Schuth et al. 2014) may produce less negative resting potentials, which would complement the less negative voltage ranges of their Na_V currents. For central-zone irregular afferents, the more negative resting potentials could reduce inactivation of the Na_V1.5 channels on their calyceal terminals (Wooltorton et al. 2007; their Fig. 12). Whether these specific possibilities are borne out, we can add diversity of sodium channels to intrinsic properties of afferents with potential to influence analysis of head motions by the inner ear.

GRANTS

This work was supported by National Institute on Deafness and Other Communication Disorders Grants DC-002290 (to R. A. Eatock) and DC-002058 (to A. Lysakowski) and Training Grant T32-DC-000038 (to X.-P. Liu).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

X.-P.L. and R.A.E. conception and design of research; X.-P.L., J.R.A.W., S.G.-N., F.-C.Y., and A.L. performed experiments; X.-P.L. analyzed data; X.-P.L., A.L., and R.A.E. interpreted results of experiments; X.-P.L., A.L., and R.A.E. prepared figures; X.-P.L. and R.A.E. drafted manuscript; J.R.A.W., S.G.-N., F.-C.Y., A.L., and R.A.E. edited and revised manuscript; R.A.E. approved final version of manuscript.