Introduction

The receptor potentials of hair cells (HCs) are strongly shaped by large outwardly rectifying K⁺ conductances that are differentially expressed according to HC type. Here we report that a specific voltage-gated K⁺ (K_V) channel subunit participates in very different K_V channels dominating the membrane conductances of type I and type II HCs in amniote vestibular organs.

Type I HCs occur only in amniote vestibular organs. Their most distinctive features are that they are enveloped by a calyceal afferent terminal (Wersäll, 1956; Lysakowski and Goldberg, 2004) and that they express g_K,L (Correia and Lang, 1990; Rennie and Correia, 1994; Rüsch and Eatock, 1996a): a large noninactivating conductance with an activation range from –100 to –60 mV, far more negative than other “low-voltage-activated” K_V channels. In addition to selectively attenuating and speeding up the receptor potentials of type I HCs (Correia et al., 1996; Rüsch and Eatock, 1996b), g_K,L augments non-quantal transmission from type I hair cell to afferent calyx by providing open channels for K⁺ flow into the synaptic cleft (Contini et al., 2012, 2017, 2020; Govindaraju et al., 2023), increasing the speed and linearity of the transmitted signal (Songer and Eatock, 2013).

Type II HCs have compact afferent synaptic contacts (boutons) where the receptor potential drives quantal release of glutamate. They have fast-inactivating (A-type, g_A) and delayed rectifier (g_DR) conductances that are opened by depolarization above resting potential (V_rest).

The unusual properties of g_K,L have long attracted curiosity about its molecular nature. g_K,L has been proposed to include “M-like” K_V channels in the K_V7 and/or erg channel families (Kharkovets et al., 2000; Hurley et al., 2006; Holt et al., 2007). The K_V7.4 subunit was of particular interest because it contributes to the low-voltage-activated conductance, g_K,n, in cochlear outer hair cells, but was eventually eliminated as a g_K,L subunit by experiments on K_V7.4-null mice (Spitzmaul et al., 2013).

Several observations suggested the K_V1.8 (KCNA10) subunit as an alternative candidate for g_K,L. K_V1.8 is highly expressed in vestibular sensory epithelia (Carlisle et al., 2012), particularly hair cells (Lee et al., 2013; Scheffer et al., 2015; McInturff et al., 2018), with slight expression elsewhere (skeletal muscle, Lee et al., 2013; kidney, Yao, 2002). K_V1.8^−/− mice show absent or delayed vestibular-evoked potentials, the synchronized activity of afferent nerve fibers sensitive to fast linear head motions (Lee et al., 2013). Unique among K_V1 channels, K_V1.8 has a cyclic nucleotide binding domain (Lang et al., 2000) with the potential to explain g_K,L’s known cGMP dependence (Behrend et al., 1997; Chen and Eatock, 2000).

Our comparison of whole-cell currents and immunohistochemistry in type I HCs from K_V1.8^−/− and K_V1.8^+/+,+/– mouse utricles confirmed that K_V1.8 expression is necessary for g_K,L. More surprisingly, K_V1.8 expression is also required for A-type and delayed rectifier conductances of type II HCs. In both HC types, eliminating the K_V1.8-dependent major conductances revealed a smaller delayed rectifier conductance involving K_V7 channels. Thus, the distinctive outward rectifiers that produce such different receptor potentials in type I and II HCs both include K_V1.8 and K_V7 channels.

Results

We compared whole-cell voltage-activated K⁺ currents in type I and type II hair cells from homozygous knockout (K_V1.8^−/−) animals and their wildtype (K_V1.8^+/+) or heterozygote (K_V1.8^+/–) littermates. We immunolocalized K_V1.8 subunits in the utricular epithelium and pharmacologically characterized the residual K⁺ currents of K_V1.8^−/− animals. Current clamp experiments demonstrated the impact of K_V1.8-dependent currents on passive membrane properties.

We recorded from three utricular zones: lateral extrastriola (LES), striola, and medial extrastriola (MES) (Fig. 3A.1); striolar and extrastriolar zones have many structural and functional differences and give rise to afferents with different physiology (reviewed in Goldberg, 2000; Eatock and Songer, 2011). Recordings are from 412 type I and II HCs (53% LES, 30% MES, 17% striola) from mice between postnatal day (P) 5 and P370. We recorded from such a wide age range to test for developmental or senescent changes in the impact of the null mutation. Above P18, we did not see substantial changes in K_V channel properties, as reported (González-Garrido et al., 2021).

As reported (Lee et al., 2013), K_V1.8^−/− animals appeared to be healthy and to develop and age normally.

KV1.8 is necessary for g_K,L in type I hair cells

The large low-voltage activated conductance, g_K,L, in K_V1.8^+/+,+/– type I hair cells produces distinctive whole-cell current responses to voltage steps, as highlighted by our standard type I voltage protocol (Fig. 1A). From a holding potential within the g_K,L activation range (here –74 mV), voltage steps to –124 mV, which is negative to E_K and the activation range, producing a large inward current through open g_K,L channels that rapidly decays as the channels deactivate. The large transient inward current is a hallmark of g_K,L. Steady-state activation was measured from tail currents after iterated 500-ms step voltages (Fig. 1A). We detected no difference between the Boltzmann parameters of g_K,L G-V curves from K_V1.8^+/– and K_V1.8^+/+ type I HCs.

Figure 1.

Table 1.

For a similar voltage protocol, K_V1.8^−/− type I HCs (Fig. 1B) produced no inward transient current at the step from –74 mV to –124 mV and much smaller depolarization-activated currents during the iterated steps, even at much more positive potentials. Figure 1C compares the conductance-voltage (G-V, activation) curves fit to tail currents (Eq. 1; see insets in Fig. 1A-B): the maximal conductance (g_max) of the K_V1.8^−/− HC was over 10-fold smaller (Fig. 1C.1), and the curve was positively shifted by >40 mV (Fig. 1C.2). Figure 1D shows the G-V Boltzmann fit parameters for type I HCs from mice >P12, an age at which type I HCs normally express g_K,L (Rüsch et al., 1998).

Suppl. Fig. 1 shows how G-V parameters of outwardly rectifying currents in type I HCs changed from P5 to P360. In K_V1.8^+/+,+,– mice, the parameters transitioned over the first 15-20 postnatal days from values for a conventional delayed rectifier, activating positive to resting potential, to g_K,L values, as previously described (Rüsch et al., 1998; Géléoc et al., 2004; Hurley et al., 2006). Between P5 and P10, we detected no evidence of a non-g_K,L K_V1.8-dependent conductance in immature type I HCs (Suppl. Fig. 1B). In K_V1.8^{− /−} type I HCs, g_K,L was absent and G-V parameters did not change much with age from P5 to P370.

The much smaller residual delayed rectifier activated positive to resting potential, with V_half ∼–40 mV and g_max density of 1.3 nS/pF. We characterize this K_V1.8-independent delayed rectifier later. A much larger non-g_K,L delayed rectifier conductance (“g_DR,I”) was reported in our earlier publication on mouse utricular type I HCs (Rüsch et al., 1998). This current was identified as that remaining after “blocking” g_K,L with 20 mM external Ba²⁺. Our new data suggest that there is no large non-g_K,L conductance, and that instead high Ba²⁺ positively shifted the apparent voltage dependence of g_K,L.

KV1.8 strongly affects type I passive properties and responses to current steps

While the cells of K_V1.8^−/− and K_V1.8^+/– epithelia appeared healthy, type I hair cells had smaller membrane capacitances (C_m) and presumably surface areas: 4-5 pF in K_V1.8^−/− type I HCs, ∼20% smaller than K_V1.8^+/– type I HCs (∼6 pF) and ∼30% smaller than K_V1.8^+/+ type I HCs (6-7 pF; Table 2). The decreased C_m may reflect deletion of a highly expressed trans-membrane protein (see discussion of g_K,L channel density in Chen and Eatock, 2000, and for comparison the large decrease in outer hair cell size in the prestin null mutant (Liberman et al., 2002; Takahashi et al., 2018)).

Table 2.

Basolateral conductances help set the resting potential and passive membrane properties that regulate the time course and gain of voltage responses to small currents. To examine the effect of K_V1.8 on these properties, we switched to current clamp mode and measured resting potential (V_rest), input resistance (R_in, equivalent to voltage gain for small current steps, △V/△I), and membrane time constant (τ_RC). In K_V1.8^−/− type I HCs, V_rest was much less negative (Fig. 2A.1), R_in was greater by ∼20-fold (Fig. 2A.2), and membrane charging times were commensurately longer (Fig. 2A.3).

Figure 2.

Figure 3.

The differences between the voltage responses of K_V1.8^+/+,+/– and K_V1.8^−/− type I HCs are expected from the known impact of g_K,L on V_rest and R_in (Correia and Lang, 1990; Ricci et al., 1996; Rüsch and Eatock, 1996b; Songer and Eatock, 2013). The large K⁺-selective conductance at V_rest holds V_rest close to E_K (K⁺ equilibrium potential) and minimizes gain (△V/△I), such that voltage-gated conductances are negligibly affected by the input current and the cell produces approximately linear, static responses to iterated current steps. For K_V1.8^−/− type I HCs, with their less negative V_rest and larger R_in, positive current steps evoked a fast initial depolarization (Fig. 2B.2), activating residual delayed rectifiers and repolarizing the membrane toward E_K. Negative current steps evoked an initial “sag” (Fig. 2B.2), a hyperpolarization followed by slow repolarization as the HCN1 channels open (Rüsch and Eatock, 1996b).

Overall, comparison of the K_V1.8^+/+,+/– and K_V1.8^−/− responses shows that with K_V1.8 (g_K,L), the voltage response of the type I hair cell is smaller but better reproduces the time course of the input current.

KV1.8 is necessary for both inactivating and non-inactivating K_V currents in type II hair cells

Type II HCs also express K_V1.8 mRNA (McInturff et al., 2018; Orvis et al., 2021). Although their outwardlyrectifying conductances (g_A and g_DR) differ substantially in voltage dependence and size from g_K,L, both conductances were strongly affected by the null mutation: g_A was eliminated and the delayed rectifier was substantially smaller. Below we describe g_A and g_DR in K_V1.8^+/+,+/– type II HCs and the residual outward rectifying current in K_V1.8^−/− type II HCs.

K_V1.8^+/+,+/– type II HCs

Most (81/84) K_V1.8^+/+,+/– type II HCs expressed a rapidly-activating, rapidlyinactivating A-type conductance (g_A). We define A current as the outwardly rectifying current that inactivates by over 30% within 200 ms. g_A was more prominent in extrastriolar zones, as reported (Holt et al. 1999, Weng and Correia 1999).

We compared the activation and inactivation time course and inactivation prominence for 200-ms steps from –124 mV to ∼30 mV. Outward currents fit with Eq. 3 yielded fast inactivation time constants (τ_{Inact, Fast}) of ∼30 ms in LES (Fig. 3A.2). τ_{Inact, Fast} was faster in LES than in MES or striola (Fig. 3A.3) and fast inactivation was a larger fraction of the total inactivation in LES than striola (∼0.5 vs. 0.3, Fig. 3A.4).

Table 3.

To show voltage dependence of activation, we generated G-V curves for peak currents (sum of A-current and delayed rectifier) and steady-state currents measured at 200 ms, after g_A has mostly inactivated (Figure 3D.2). K_V1.8^+/– HCs had smaller currents than K_V1.8^+/+ HCs, reflecting a smaller g_DR (Fig. 3D) and faster fast inactivation (Fig. 3A.3). As discussed later, these effects may relate to effects of the K_V1.8 gene dosage on the relative numbers of different K_V1.8 heteromers.

For K_V1.8^+/+ and K_V1.8^+/– HCs, the voltage dependence as summarized by V_half and slope factor (S) was similar. Relative to g_SS, g_Peak had a more positive V_half (∼–21 vs. ∼–26) and greater S (∼12 vs. ∼9, Fig. 3D, Table 4). Because g_Peak includes channels with and without fast inactivation, the shallower g_Peak-V curve may reflect a more heterogeneous channel population. Only g_Peak showed zonal variation, with more positive V_half in LES than striola (∼–20 mV vs. ∼–24 mV, Fig. 3D, Table 4). We later suggest that variable subunit composition may drive zonal variation in g_Peak.

Table 4.

K_V1.8^−/− type II HCs

from all zones were missing g_A and 30-50% of g_DR (Fig. 3B-D). The residual delayed rectifier (1.3 nS/pF) had a more positive V_half than g_DR in K_V1.8^+/+,+/– HCs (∼–20 mV vs. ∼–26 mV, Fig. 3D.2). We refer to the K_V1.8-dependent delayed rectifier component as g_DR(K_V1.8) and to the residual, K_V1.8-independent delayed rectifier component as g_DR(K_V7) because, as we show later, it includes K_V7 channels. Supplemental Figure 3A shows the development of K_V1.8-dependent and independent K_V currents in type II HCs with age from P5 to over P300. In K_V1.8^+/+,+/– type II HCs, g_A was present at all ages with a higher % inactivation after P18 than at P5-P10 (Suppl. Fig. 3A.4). g_Peak did not change much above P12 except for a compression of conductance density from P13 to P370 (partial correlation coefficient = –0.4, p = 2E-5, Suppl. Fig. 3A.3).

We saw small rapidly inactivating outward currents in a minority of K_V1.8^−/− type II HCs (23%, 7/30), all >P12 and extrastriolar (Suppl. Fig. 4). These currents overlapped with g_A in percent inactivation, inactivation kinetics, and activation voltage dependence but were very small. As discussed later, we suspect that these currents flow through homomers of inactivating K_V subunits that in control hair cells join with K_V1.8 subunits and confer inactivation on the heteromeric conductance.

K_V1.8 affects type II passive properties and responses to current steps

In type II HCs, absence of K_V1.8 did not change V_rest (Fig. 4A.1) because g_A and g_DR both activate positive to rest, but significantly increased R_in and τ_RC (Fig. 4A.2-A.3).

Figure 4.

Table 5.

Table 6.

Positive current steps evoked an initial depolarizing transient in both K_V1.8^+/+ and K_V1.8^−/− type II HCs, but the detailed time course differed (Fig. 4B). Both transient and steady-state responses were larger in K_V1.8^−/−, consistent with their larger R_in values.

Absence of K_V1.8 increased the incidence of sharp electrical resonance in type II HCs. Electrical resonance, which manifests as ringing responses to current steps, can support receptor potential tuning (Ashmore, 1983; Fettiplace, 1987; Hudspeth and Lewis, 1988; Ramanathan and Fuchs, 2002). Larger R_in values made K_V1.8^−/− type II HCs more prone to electrical resonance; Figure 4C.1 shows a striking example. Median resonance quality (Q_e, sharpness of tuning) was greater in K_V1.8^−/− (1.33, n=26) than K_V1.8^+/+ (0.66, n=23) or K_V1.8^+/– (0.59, n=44) type II HCs.

K_V1.8 affected the time course of the initial peak in response to much larger current injections (Fig. 4D-E). Fast activation of g_A in control type II HCs rapidly repolarizes the membrane and then inactivates, allowing the constant input current to progressively depolarize the cell, producing a slow rebound (Fig 4D.2). This behavior has the potential to counter MET adaptation (Vollrath and Eatock, 2003).

K_V1.8 immunolocalized to basolateral membranes of both type I and II HCs

If K_V1.8 is a pore-forming subunit in the K_V1.8-dependent conductances g_K,L, g_A, and g_DR, it should localize to hair cell membranes. Figure 5 compares K_V1.8 immunoreactivity in K_V1.8^+/+ and K_V1.8^−/− utricles, showing specific immunoreactivity along the basolateral membranes of both hair cell types in K_V1.8^+/+ utricles. To identify hair cell type and localize the hair cell membrane, we used antibodies against K_V7.4 (KCNQ4), an ion channel densely expressed in the calyceal “inner-face” membrane next to the synaptic cleft (Hurley et al., 2006; Lysakowski et al., 2011), producing a cup-like stain around type I HCs (Fig. 5A). K_V1.8 immunoreactivity was present in hair cell membrane apposing K_V7.4-stained calyx inner face in K_V1.8^+/+ utricles (Fig. 5A.1, A.2) and not in K_V1.8^−/− utricles (Fig. 5A.3).

Figure 5.

In other experiments, we used antibodies against calretinin (Calb2), a cytosolic calcium binding protein expressed by many type II HCs and also by striolar calyx-only afferents (Desai et al., 2005; Lysakowski et al., 2011) (Fig. 5B). A hair cell is type II if it is calretinin-positive (Fig. 5B.1) or if it lacks a K_V7.4-positive or calretinin-positive calyceal cup (Fig. 5A.2, 5B.3, rightmost cells). Hair cell identification was confirmed with established morphological indicators: for example, type II HCs tend to have basolateral processes (feet) (Pujol et al., 2014) and, in the extrastriola, more apical nuclei than type I HC.

Previously, Carlisle et al. (2012) reported K_V1.8-like immunoreactivity in many cell types in the inner ear. In contrast, Lee et al. (2013) found that gene expression reporters indicated expression only in hair cells and some supporting cells. Here, comparison of control and null tissue showed selective expression of HC membranes, and that some supporting cell staining is non-selective.

K_V1.4 may also contribute to g_A

Results with the K_V1.8 knockout suggest that type II hair cells have an inactivating K_V1 conductance that includes K_V1.8 subunits. K_V1.8, like most K_V1 subunits, is not inactivating as a heterologously expressed homomer (Lang et al., 2000; Ranjan et al., 2019; Dierich et al., 2020), nor are the K_V1.8-dependent channels in type I HCs, as we show, and cochlear inner hair cells (Dierich et al., 2020). K_V1 subunits without intrinsic inactivation can produce rapidly inactivating currents by associating with K_Vβ1 (KCNB1) or K_Vβ3 subunits. K_Vβ1 is present in type II HCs alongside K_Vβ2 (McInturff et al., 2018; Jan et al., 2021; Orvis et al., 2021), which does not confer rapid inactivation (Dwenger et al., 2022).

An alternative possibility is that in type II HCs, K_V1.8 subunits heteromultimerize with K_V1.4 subunits – the only K_V1 subunits which, when expressed as a homomer, have complete N-type (fast) inactivation (Stühmer et al., 1989). Multiple observations support this possibility. K_V1.4 has been linked to g_A in pigeon type II HCs (Correia et al., 2008) and is the second-most abundant K_V1 transcript in mammalian vestibular HCs, after K_V1.8 (Scheffer et al., 2015). K_V1.4 is expressed in type II HCs but not type I HCs (McInturff et al., 2018; Orvis et al., 2021), and is not found in striolar HCs (Jan et al., 2021; Orvis et al., 2021), where even in type II HCs, inactivation is slower and less extensive (Fig. 3A).

In K_V1.8^+/+,+/– type II HCs, the time course (Fig. 3A, τ_Fast,Inac of ∼30 ms +30 mV) and voltage dependence of inactivation of g_A (V_half –41 mV, Fig. 6B.2), are consistent with heterologously expressed heteromers of K_V1.4 with K_V1.x and/or K_Vβ1 (Imbrici et al., 2006; Al-Sabi et al., 2011). In further support, we observed K_V1.4-like immunoreactivity along the basolateral membranes of extrastriolar type II HCs in rat utricles (Fig. 6A).

Figure 6.

K_V7 channels contribute a small delayed rectifier in type I and type II hair cells

In K_V1.8^−/− HCs, absence of I_K,L and I_A revealed smaller delayed rectifier K⁺ currents that, unlike I_K,L, activated positive to resting potential and, unlike I_A, lacked fast inactivation. Candidate channels include members of the K_V7 (KCNQ, M-current) family, which have been identified previously in rodent vestibular HCs (Kharkovets et al., 2000; Rennie et al., 2001; Hurley et al., 2006; Scheffer et al., 2015).

We test for K_V7 contributions in K_V1.8^−/− type I HCs, K_V1.8^−/− type II HCs, and K_V1.8^+/+,+/– type II HCs of multiple ages by applying XE991 at 10 µM (Fig. 7A), a dose selective for K_V7 channels (Brown et al., 2002) and close to the IC₅₀ (Alexander et al., 2019). In K_V1.8^−/− HCs of both types, 10 µM XE991 blocked about half of the residual K_V conductance (Fig. 7B.1), consistent with K_V7 channels conducting most or all of the non-K_V1.8 delayed rectifier current. In all tested HCs (P8-355, median P224), the XE991-sensitive conductance did not inactivate substantially within 200 ms at any voltage, consistent with K_V7.2, 7.3, 7.4, and 7.5 currents (Wang, 1998; Kubisch et al., 1999; Schroeder et al., 2000; Jensen et al., 2007; Xu et al., 2007). We refer to this component as g_DR(K_v7). The voltage dependence and Gmax density (G_max/C_m) of g_DR(K_v7) were comparable across HC types and genotypes (Figure 7B.2-4).

Figure 7.

These results are consistent with similar K_V7 channels contributing a relatively small delayed rectifier in both HC types. In addition, the similarity of XE991-sensitive currents of K_V1.8^+/+ and K_V1.8^−/− type II HCs indicates that knocking out K_V1.8 did not cause general effects on ion channel expression. We did not test XE991 on K_V1.8^+/+,+/– type I HCs because g_K,L runs down in ruptured patch recordings (Rüsch and Eatock, 1996a; Chen and Eatock, 2000; Hurley et al., 2006), which could contaminate the XE991-sensitive conductance obtained by subtraction.

In one striolar K_V1.8^−/− type I HC, XE991 also blocked a small conductance that activated negative to rest (Suppl. Fig. 5A-B). This conductance (V_half ∼= –100 mV, Suppl. Fig. 5C) was detected only in K_V1.8^−/− type I HCs from the striola (5/23 vs. 0/45 extrastriolar). The V_half and τ_deactivation at –124 mV were similar to values reported for K_V7.4 channels in cochlear HCs (Wong et al., 2004; Dierich et al., 2020). This very negatively activating K_V7 conductance coexisted with the larger less negatively activating K_V7 conductance (Suppl. Fig. 5C) and was too small (<0.5 nS/pF) to contribute significantly to g_K,L (∼10-100 nS/pF, Fig. 1D).

Other channels

While our data are consistent with K_V1.8- and K_V7-containing channels carrying most of the outwardrectifying current in mouse utricular hair cells, there is evidence in other preparations for additional channels, including K_V11 (KCNH, Erg) channels in rat utricular type I hair cells (Hurley et al., 2006) and BK (KCNMA1) channels in rat utricle and rat and turtle semicircular canal hair cells (Schweizer et al., 2009; Contini et al., 2020).

BK is expressed in mouse utricular hair cells (McInturff et al., 2018; Jan et al., 2021; Orvis et al., 2021). However, Ca²⁺-dependent currents have not been observed in mouse utricular HCs, and we found little to no effect of the BK-channel blocker iberiotoxin at a dose (100 nM) well beyond the IC₅₀: percent blocked at –30 mV was 2 ± 6% (3 K_V1.8^−/− type I HCs); 1 ± 5% (5 K_V1.8^+/+,+/– type II HCs); 7% and 14% (2 K_V1.8^−/− type II HCs). We also did not see N-shaped I-V curves typical of many Ca²⁺-dependent K⁺ currents. In our ruptured-patch recordings, Ca²⁺-dependent BK currents and erg channels may have been eliminated by wash-out of the hair cells’ small Ca_V currents (Bao et al., 2003) or cytoplasmic second messengers (Hurley et al., 2006).

To check whether the constitutive K_V1.8 knockout has strong non-specific effects on channel trafficking, we examined the summed HCN and inward rectifier currents (I_H and I_Kir) at –124 mV, and found them similar across genotypes (Suppl. Fig. 6). In the process, we noted zonal differences in I_H and I_Kir that have not been reported in hair cells. In type I HCs from both control and null utricles, I_H and I_Kir were less prevalent in striola than extrastriola, and, when present, the combined inward current was smaller (Suppl. Fig. 6B).

Discussion

We have shown that constitutive knockout of K_V1.8 eliminated g_K,L in type I HCs, and g_A and much of g_DR in type II HCs. K_V1.8 immunolocalized specifically to the basolateral membranes of type I and II HCs. We conclude that K_V1.8 is a pore-forming subunit of g_K,L, g_A, and part of g_DR [g_DR(K_V1.8)]. We provide evidence that fast inactivation of g_A may arise from heteromultimerization of non-inactivating K_V1.8 subunits and inactivating K_V1.4 subunits. Finally, we showed that a substantial component of the residual delayed rectifier current in both type I and type II HCs comprises K_V7 channels.

K_V1.8 is expressed in hair cells from mammalian cochlea (Dierich et al., 2020), avian utricle (Scheibinger et al., 2022), and zebrafish (Erickson and Nicolson, 2015). Our work suggests that in anamniotes, which lack type I cells and g_K,L, K_V1.8 contributes to g_A and g_DR, which are widespread in vertebrate HCs (reviewed in Meredith and Rennie, 2016). K_V1.8 expression has not been detected in rodent brain but is reported in the pacemaker nucleus of weakly electric fish (Smith et al., 2018).

K_V1.8 subunits may form homomultimers to produce g_K,L in type I hair cells

Recent single-cell expression studies on mouse utricles (McInturff et al., 2018; Jan et al., 2021; Orvis et al., 2021) have detected just one K_V1 subunit, K_V1.8, in mouse type I HCs. Given that K_V1.8 can only form multimers with K_V1 family members, and given that g_K,L channels are present at very high density (∼150 per μm² in rat type I, Chen and Eatock, 2000), it stands to reason that most or all of the channels are K_V1.8 homomers. Other evidence is consistent with this proposal. g_K,L (Rüsch and Eatock, 1996a) and heterologously expressed K_V1.8 homomers in oocytes (Lang et al., 2000) are non-inactivating and blocked by millimolar Ba²⁺ and 4-aminopyridine and >10 mM TEA. Unlike channels with K_V1.1, K_V1.2, and K_V1.6 subunits, g_K,L is not sensitive to 10 nM α-dendrotoxin (Rüsch and Eatock, 1996a). g_K,L and heterologously expressed K_V1.8 channels have similar single-channel conductances (∼20 pS for g_K,L at positive potentials, Chen and Eatock, 2000; 11 pS in oocytes, Lang et al., 2000). g_K,L is inhibited—or positively voltage-shifted— by cGMP (Behrend et al., 1997; Chen and Eatock, 2000), presumably via the C-terminal cyclic nucleotide binding domain of K_V1.8.

A major novel property of g_K,L is that it activates 30-60 mV negative to type II K_V1.8 conductances and most other low-voltage-activated K_V channels (Ranjan et al., 2019). The very negative activation range is a striking difference between g_K,L and known homomeric K_V1.8 channels. Heterologously expressed homomeric K_V1.8 channels have an activation V_half of –10 to 0 mV (X. laevis oocytes, Lang et al., 2000; CHO cells, Dierich et al., 2020). In cochlear inner HCs, currents attributed to K_V1.8 (by subtraction of other candidates) have a near-zero activation V_half (–4 mV, Dierich et al., 2020).

Possible factors in the unusually negative voltage dependence of g_K,L include:

1. elevation of extracellular K⁺ by the enveloping calyceal terminal, unique to type I HCs (Lim et al., 2011; Contini et al., 2012; Spaiardi et al., 2020; Govindaraju et al., 2023). High K⁺ increases conductance though g_K,L channels (Contini et al., 2020), perhaps through K⁺-mediated relief of C-type inactivation (López-Barneo et al., 1993; Baukrowitz and Yellen, 1995). We note, however, that g_K,L is open at rest even in neonatal mouse utricles cultured without innervation (Rüsch et al., 1998) and persists in dissociated type I HCs (Chen and Eatock, 2000; Hurley et al., 2006).

2. The high density of g_K,L (∼50 nS/pF in striolar K_V1.8^+/+ HCs) implies close packing of channels, possibly represented by particles (12-14 nm) seen in freeze-fracture electron microscopy of the type I HC membrane (Gulley and Bagger-Sjöbäck, 1979; Sousa et al., 2009). Such close channel packing might hyperpolarize in situ voltage dependence of g_K,L, as proposed for K_V7.4 channels in outer hair cells (Perez-Flores et al., 2020).

3. Modulation by accessory subunits. Type I HCs express K_Vβ1 (McInturff et al., 2018; Orvis et al., 2021), an accessory subunit that can confer fast inactivation and hyperpolarize activation V_half by ∼10 mV. K_Vβ1 might interact with K_V1.8 to shift voltage dependence negatively. Arguments against this possibility include that g_K,L lacks fast inactivation (Rüsch and Eatock, 1996a; Hurley et al., 2006; Spaiardi et al., 2017) and that cochlear inner hair cells co-express K_V1.8 and K_Vβ1 (Liu et al., 2018) but their K_V1.8 conductance has a near-0 V_half (Dierich et al., 2020).

K_V1.8 subunits may heteromerize with variable numbers of inactivating K_V1.4 subunits to produce g_A and K_V1.8-dependent g_DR in type II HCs

The K_V1.8-dependent conductances of type II HCs vary in their fast and slow inactivation. In not showing fast inactivation (Lang et al., 2000; Ranjan et al., 2019; Dierich et al., 2020), heterologously expressed K_V1.8 subunits resemble most other K_V1 family subunits, with the exception of K_V1.4 (for comprehensive review, see Ranjan et al., 2019). K_V1.4 is a good candidate to provide fast inactivation based on immunolocalization and voltage dependence (Figs. 4, 6). We suggest that g_A and g_DR(K_V1.8) are K_V1.8-containing channels that vary in K_V1.8:K_V1.4 stoichiometry, with possible additional variation in K_Vβ2 and K_Vβ1 accessory subunits.

K_V1.4-K_V1.8 heteromeric assembly could account for several related observations. The faster τ_Inact,Fast in K_V1.8^+/– relative to K_V1.8^+/+ type II HCs (Fig. 3A.3, Suppl. Fig. 2A.1) could reflect an increased ratio of K_V1.4 to K_V1.8 subunits and therefore more N-terminal inactivation domains per heteromeric channel. Zonal variation in the extent and speed of N-type inactivation (Fig. 3A) might arise from differential expression of K_V1.4. The small fast-inactivating conductance in ∼20% of extrastriolar K_V1.8^−/− type II HCs (Suppl. Fig. 4) might flow through K_V1.4 homomers.

In addition or alternatively, K_Vβ subunits are positioned to contribute to fast inactivation. K_Vβ1 is expressed in type II HCs (McInturff et al., 2018; Jan et al., 2021; Orvis et al., 2021), and, together with K_V1.4, has been linked to g_A in pigeon vestibular HCs (Correia et al., 2008). K_Vβ2, also expressed in type II HCs (McInturff et al., 2018; Orvis et al., 2021), does not confer fast inactivation but hyperpolarizes activation voltage by ∼10 mV and accelerates activation and inactivation kinetics (Heinemann et al., 1996). g_A and g_DR(K_V1.8) are not likely to be different kinetic components of current through homogeneous incompletely inactivating K_V channels. The lack of N-type inactivation of g_DR(K_V1.8) is readily explained by homomeric K_V1.8 channels. The double-exponential decay of g_Peak (comprising g_A and g_DR(K_V1.8)) is consistent with two channel populations.

K_V1.8 relevance for vestibular function

In both type I and type II utricular HCs, K_V1.8-dependent channels strongly shape receptor potentials in ways that promote temporal fidelity rather than electrical tuning (Lewis, 1988), consistent with the utricle’s role in driving reflexes that compensate for head motions as they occur. This effect is especially pronounced for type I HCs, where the current-step evoked voltage response reproduces the input with great speed and linearity (Fig. 2).

g_K,L dominates passive membrane properties in mature K_V1.8^+/+,+/– type I HCs such that K_V1.8^−/− type I HCs are expected to have receptor potentials with higher amplitudes but lower low-pass corner frequencies, closer to those of type II HCs and immature HCs of all types (Correia et al., 1996; Rüsch and Eatock, 1996a; Songer and Eatock, 2013). In K_V1.8^−/− epithelia, we expect the lack of a large basolateral conductance open at rest to reduce the speed and gain of non-quantal transmission, which depends on K⁺ ion efflux from the type I HC to change electrical and K⁺ potentials in the synaptic cleft (Govindaraju et al., 2023). In hair cells, K⁺ enters the mechanosensitive channels of the hair bundle from the K⁺-rich apical endolymph and exits through basolateral potassium conductances into the more conventional low-K+ perilymph. For the type I-calyx synapse, having in the hair cell a large, non-inactivating K⁺ conductance open across the physiological range of potentials avoids channel gating time and allows for instantaneous changes in current into the cleft and fast afferent signaling (Pastras et al., 2023).

In contrast, mature type II HCs face smaller synaptic contacts and have K_V1.8-dependent currents that are not substantially activated at resting potential. They do affect the time course and gain of type II HC responses to input currents, speeding up depolarizing transients, producing a repolarizing rebound during the step, and reducing resonance.

Type I and II vestibular hair cells are closely related, such that adult type II HCs acquire type I-like properties upon deletion of the transcription factor Sox2 (Stone et al., 2021). In normal development of the two cell types, the Kcna10 gene generates importantly different ion channels, presenting a natural experiment in functional differentiation of sensory receptor cells.

Materials and methods

Preparation

All procedures for handling animals followed the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committees of the University of Chicago and the University of Illinois Chicago. Most mice belonged to a transgenic line with a knockout allele of Kcna10 (referred to here as K_V1.8^−/−). Our breeding colony was established with a generous gift of such animals from Sherry M. Jones and Thomas Friedman. These animals are described in their paper (Lee et al., 2013). Briefly, the Texas A&M Institute for Genomic Medicine generated the line on a C57BL/6;129SvEv mixed background by replacing Exon 3 of the Kcna10 gene with an IRES-bGeo/Purocassette. Mice in our colony were raised on a 12:12h light-dark cycle with access to food and water ad libitum.

Semi-intact utricles were prepared from ∼150 male and ∼120 female mice, postnatal days (P) 5-375, for same-day recording. Hair cell K_V channel data were pooled across sexes as most results did not appear to differ by sex; an exception was that g_K,L had a more negative V_half in males (Suppl. Table 1), an effect not clearly related to age, copy number, or other properties of the activation curve.

Preparation, stimulation, and recording methods followed our previously described methods for the mouse utricle (Vollrath and Eatock, 2003). Mice were anesthetized through isoflurane inhalation. After decapitation, each hemisphere was bathed in ice-cold, oxygenated Liebowitz-15 (L15) media. The temporal bone was removed, the labyrinth was cut to isolate the utricle, and the nerve was cut close to the utricle. The utricle was treated with proteinase XXIV (100 μg/mL, ∼10 mins, 22°C) to facilitate removal of the otoconia and attached gel layer and mounted beneath two glass rods affixed at one end to a coverslip.

Electrophysiology

For most recordings, we used the HEKA Multiclamp EPC10 with Patchmaster acquisition software, filtered by the integrated HEKA filters: a 6-pole Bessel filter at 10 kHz and a second 4-pole Bessel filter at 5 kHz, and sampled at 10-100 kHz. Recording electrodes were pulled (PC-100, Narishige) from soda lime glass (King’s Precision Glass R-6) wrapped in paraffin to reduce pipette capacitance. Internal solution contained (in mM) 135 KCl, 0.5 MgCl₂, 3 MgATP, 5 HEPES, 5 EGTA, 0.1 CaCl₂, 0.1 Na-cAMP, 0.1 LiGTP, 5 Na₂CreatinePO₄ adjusted to pH 7.25 and ∼280 mmol/kg by adding ∼30 mM KOH. External solution was Liebowitz-15 media supplemented with 10 mM HEPES (pH 7.40, 310 ± 10 mmol/kg). Recording temperature was 22-25°C. Pipette capacitance and membrane capacitance transients were subtracted during recordings with Patchmaster software. Series resistance (8-12 MΩ) was measured after rupture and compensated 60-80% with the amplifier, to final values of ∼2 MΩ. Potentials are corrected for remaining (uncompensated) series resistance and liquid junction potential of ∼4 mV, calculated with LJPCalc software (Marino et al., 2014).

Type I HCs with g_K,L were transiently hyperpolarized to –90 mV to close g_K,L enough to increase R_input above 100 MΩ, as needed to estimate series resistance and cell capacitance. The average resting potential, V_rest, was –87 mV ± 1 (41), similar to the calculated E_K of –86.1 mV, which is not surprising given the large K⁺ conductance of these cells (Fig. 1). V_rest is likely more positive in vivo, where lower endolymphatic Ca²⁺ increases standing inward current through MET channels.

Voltage protocols to characterize K_V currents differed slightly for type I and II HCs. In standard protocols, the cell is held at a voltage near resting potential (–74 mV in type I and –64 mV in type II), then jumped to –124 mV for 200 ms in type I HCs in order to fully deactivate g_K,L and 50 ms in type II HCs in order to remove baseline inactivation of g_A. The subsequent iterated step depolarizations lasted 500 ms in type I HCs because g_K,L activates slowly (Wong et al., 2004) and 200 ms in type II HCs, where K_V conductances activate faster. The 50-ms tail voltage was near the reversal potential of HCN channels (–44 mV in mouse utricular hair cells, Rüsch et al., 1998) to avoid HCN current contamination.

G-V (activation) parameters for control type I cells may be expected to vary across experiments on semi-intact (as here), organotypically cultured and denervated (Rüsch et al.,1998), or dissociated-cell preparations, reflecting variation in retention of the calyx (Discussion) and voltage step durations (Wong et al., 2004) which elevate K⁺ concentration around the hair cell. Nevertheless, the values we obtained for type I and type II HCs resemble values recorded elsewhere, including experiments in which extra care was taken to avoid extracellular K⁺ accumulation (Spaiardi et al., 2017, 2020). The effects of K⁺ accumulation on g_K,L’s steady-state activation curves are small because the operating range is centered on E_K and can be characterized with relatively small currents (Fig. 1A).

Pharmacology

For most experiments, we locally perfused drug-containing solutions with BASI Bee Hive syringes at a final flow rate of 20 μL/min and a dead time of ∼30 s. Global bath perfusion was paused during drug perfusion and recording, and only one cell was used per utricle. Aliquots of test agents in solution were prepared, stored at –20°C, and thawed and added to external solution on the recording day (see Key Resources Table).

Analysis

Data analysis was performed with software from OriginLab (Northampton, MA) and custom MATLAB scripts using MATLAB fitting algorithms.

Fitting voltage dependence and time course of conductances

G-V curves

Current was converted to conductance (G) by dividing by driving force (V – E_K; E_K calculated from solutions). For type I HCs, tail G-V curves were generated from current 1 ms after the end of the iterated voltage test step. For type II HCs, peak G-V curves were generated from peak current during the step and steady-state G-V curves were generated from current 1 ms before the end of a 200 ms step. Sigmoidal voltage dependence of G-V curves was fit with the first-order Boltzmann equation (Eq. 1).

V_half is the midpoint and S is the slope factor, inversely related to curve steepness near activation threshold.

Activation time course of type II HCs

For type II HCs lacking fast inactivation, outward current activation was fit with Eq. 2.

I_SS is steady-state current, τ_w is activation time constant, n is the state factor related to the number of closed states (typically constrained to 3), and I_o is baseline current.

To measure activation and inactivation time course of g_A, we used Eq. 3 to fit outward K⁺ currents evoked by steps from –125 mV to above –50 mV (Rothman and Manis, 2003b).

Z is total steady-state inactivation (0 ≤ Z < 1 means incomplete inactivation, which allows the equation to fit non-inactivating delayed rectifier currents), f is the fraction of fast inactivation relative to total inactivation, I_max is maximal current, τ_zf and τ_zs are the fast and slow inactivation time constants. We chose to compare fit parameters at 30 ± 2 mV (91), where fast and slow inactivation were consistently separable and g_A was maximized. In most K_V1.8^−/− and some striolar K_V1.8^+/+,+/– cells, where fast inactivation was absent and adjusted R² did not improve on a single-exponential fit by >0.01, we constrained f in Eq. 3 to 0 to avoid overfitting.

For Peak G-V relations, peak conductance was taken from fitted curves (Eqs. 2 and 3). To construct ‘Steady-state’ G-V relations, we used current at 200 ms (6 ± 1 % (94) greater than steady-state estimated from fits to Eq. 3 (Fig. 3C-D)).

Percent inactivation was calculated at 30 mV with Eq. 4 :

I_Peak is maximal current, and I_SS is current at the end of a 200 ms voltage step.

The electrical resonance of type II HCs was quantified by fitting voltage responses to current injection steps (Songer and Eatock, 2013). We fit Eq. 5, a damped sinusoid, to the voltage trace from half-maximum of the initial depolarizing peak until the end of the current step.

V_SS is steady-state voltage, V_p is the voltage of the peak response, τ_e is the decay time constant, f_e is the fundamental frequency, and θ is the phase angle shift.

Quality factor, Q_e, was calculated with Eq. 6 (Crawford and Fettiplace, 1981).

Statistics

We give means ± SEM for normally-distributed data, and otherwise, median and range. Data normality was assessed with the Shapiro-Wilk test for n<50 and the Kolmogorov-Smirnov test for n>50. To assess homogeneity of variance we used Levene’s test. With homogeneous variance, we used two-way ANOVA for genotype and zone with the posthoc Tukey’s test. When variance was non-homogeneous, we used one-way Welch ANOVA with the posthoc Games-Howell test. For data that were not normally distributed, we used the non-parametric one-way Kruskal-Wallis ANOVA (KWA) with posthoc Dunn’s test. Effect size is Hedge’s g (g). For age dependence, we used partial correlation coefficients controlling for genotype and zone. Statistical groups may have different median ages, but all have overlapping age ranges. In figures, asterisks represent p-value ranges as follows: *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001.

Immunohistochemistry

Mice were anesthetized with Nembutal (80 mg/kg), then perfused transcardially with 40mL of physiological saline containing heparin (400 IU), followed by 2 mL/g body weight fixative (4% paraformaldehyde, 1% picric acid, and 5% sucrose in 0.1 M phosphate buffer at pH 7.4, sometimes with 1% acrolein). Vestibular epithelia were dissected in phosphate buffer, and tissues were cryoprotected in 30% sucrose-phosphate buffer overnight at 4°C. Otoconia were dissolved with Cal-Ex (Fisher Scientific) for 10 min. Frozen sections (35 μm) were cut with a sliding microtome. Immunohistochemistry was performed on free-floating sections. Tissues were first permeabilized with 4% Triton X-100 in PBS for 1 h at room temperature, then incubated with 0.5% Triton X-100 in a blocking solution of 0.5% fish gelatin and 1% BSA for 1 h at room temperature. Sections were incubated with 2-3 primary antibodies for 72 h at 4°C and with 2-3 secondary antibodies. Sections were rinsed with PBS between and after incubations and mounted on slides in Mowiol (Calbiochem).

Data Availability

Data used in this study are available on Dryad (https://doi.org/10.5061/dryad.37pvmcvrw).

Abbreviations

• g_A,: A-type (inactivating) K_V conductance in type II HCs

• g_DR,: delayed rectifier K⁺ conductance

• g_K,L,: low-voltage-activated K⁺ conductance in type I HCs

• HC,: hair cell

• K_V,: voltage-gated K⁺ conductance

Acknowledgements

This study was supported by NIH grant R01 DC012347 to RAE and AL and an NSF Graduate Research Fellowship to HRM. We thank Drs. Thomas Friedman and Sherri Jones for the generous gift of the K_V1.8^{− /−} mouse line, and Drs. Zheng-Yi Chen and Deborah I. Scheffer for bringing the expression of this subunit in mouse vestibular hair cells to our attention.

We acknowledge Dr. Vicente Lumbreras for insights from his prior experiments on g_A in mouse utricular hair cells, and thank him for helpful further discussions.

We thank Drs. Rebecca Lim and Ebenezer Yamoah for their critical feedback on the manuscript, and Drs. Rob Raphael and Aravind Chenrayan Govindaraju for feedback and many helpful discussions. We thank Drs. Joe Burns, Gabi Pregernig, and Lars Becker (Decibel Therapeutics, Inc.) for helpful discussions.

Author Contributions

HRM designed and performed experiments, analyzed data, and wrote the paper; RAE helped design experiments, analyze data, and write the paper; AL performed immunohistochemistry experiments.

Supplemental Figures

Supplementary Figure 1.

Supplementary Figure 2.

Supplementary Figure 3.

Supplementary Figure 4.

Supplementary Figure 5.

Supplementary Figure 6.

Supplemental Table 1.

Supplemental Table 2.