Divergent membrane properties of mouse cochlear glial cells around hearing onset

Spiral ganglion neurons (SGNs) are the primary afferent neurons of the auditory system, and together with their attendant glia, form the auditory nerve. Within the cochlea, satellite glial cells (SGCs) encapsulate the cell body of SGNs, whereas Schwann cells (SCs) wrap their peripherally‐ and centrally‐directed neurites. Despite their likely importance in auditory nerve function and homeostasis, the physiological properties of auditory glial cells have evaded description. Here, we characterized the voltage‐activated membrane currents of glial cells from the mouse cochlea. We identified a prominent weak inwardly rectifying current in SGCs within cochlear slice preparations (postnatal day P5‐P6), which was also present in presumptive SGCs within dissociated cultures prepared from the cochleae of hearing mice (P14‐P15). Pharmacological block by Ba2+ and desipramine suggested that channels belonging to the Kir4 family mediated the weak inwardly rectifying current, and post hoc immunofluorescence implicated the involvement of Kir4.1 subunits. Additional electrophysiological profiles were identified for glial cells within dissociated cultures, suggesting that glial subtypes may have specific membrane properties to support distinct physiological roles. Immunofluorescence using fixed cochlear sections revealed that although Kir4.1 is restricted to SGCs after the onset of hearing, these channels are more widely distributed within the glial population earlier in postnatal development (i.e., within both SGCs and SCs). The decrease in Kir4.1 immunofluorescence during SC maturation was coincident with a reduction of Sox2 expression and advancing neurite myelination. The data suggest a diversification of glial properties occurs in preparation for sound‐driven activity in the auditory nerve.


| INTRODUC TI ON
Spiral ganglion neurons (SGNs) transmit acoustic information from sensory hair cells in the cochlear periphery to central neurons in the brainstem. Type I SGNs, which represent 90%-95% of the population, have synaptic contacts with inner hair cells while the remainder (type II SGNs) innervate outer hair cells (Berglund & Ryugo, 1987).
SGNs are bipolar, with their peripherally-directed processes and cell bodies residing wholly within the cochlea, while their centrally-directed neurites converge to form the auditory nerve, which crosses the peripheral nervous system-central nervous system (PNS-CNS) transitional zone within the cochlear modiolus (Toesca, 1996), before extending into the brainstem. In the cochlea, three broad classes of peripheral glia are associated with SGNs: (a) satellite glial cells (SGCs) which wrap around the neuronal cell body, plus (b) myelinating and (c) non-myelinating Schwann cells (SCs) which associate with the neurites of type I and type II SGNs, respectively.
In rodents, multiple layers of compact myelin originating from SCs wrap the neurites of type I SGNs, whereas fewer, more loosely assembled myelin layers originating from SGCs surround the neuronal cell body (Rosenbluth, 1962). In humans the cell body is reported to be largely devoid of myelin (Arnold, 1987, Ota & Kimura, 1980. In mice, myelin first appears around SGN neurites at postnatal day 4-5 (P4-P5), with the subsequent appearance around the cell body at P8-P9 (Romand & Romand, 1990). Animal models of dysmyelination have revealed the importance of myelin integrity on signal propagation (El-Badry et al., 2007; and SGN survival . Furthermore, dysmyelination that occurs around nodes of Ranvier in a model of noise-induced hearing loss suggests that such trauma adversely disrupts neuro-glial integrity (Tagoe et al., 2014). In humans, mutations in the MPZ or PMP22 genes that encode specific PNS myelin proteins (Greenfield et al., 1973;Lemke & Axel, 1985;Snipes et al., 1992) are associated with auditory neuropathies (Kabzinska et al., 2007;Kovach et al., 2002;Starr et al., 2003;Verhagen et al., 2005). Together, these observations highlight the importance of myelin formation and its maintenance by SCs and SGCs in the auditory periphery.
Beyond their roles in myelination, little is known about the functions of cochlear glial cells. Glial cells elsewhere in the nervous system perform indispensable homeostatic and metabolic functions, including the spatial buffering of K + (Kofuji & Newman, 2004), clearance of neurotransmitters (Anderson & Swanson, 2000), and the provision of energy metabolites to neurons (Lee et al., 2012).
All these roles depend critically on the functional expression of a precise repertoire of glial ion channels, receptors and transporters. Some of the earliest physiological studies of vertebrate glial cells revealed hyperpolarized resting membrane potentials and high permeability to K + (Hild & Tasaki, 1962;Kuffler et al., 1966). These features have since been ascribed to the weak inwardly rectifying K + (Kir) channel, Kir4.1 (Kofuji et al., 2000;Poopalasundaram et al., 2000;Takumi et al., 1995).
While SGNs have been studied in detail over a number of years, little is known about the membrane physiology of their attendant SGCs and SCs. Here we provide a first description of their electrophysiological properties. Using an intact slice preparation from early postnatal mice, we show that SGCs exhibit weak inwardly rectifying currents which were blocked by Ba 2+ , a broad Kir channel blocker. To examine the properties of more mature glial cells, we used dissoci-

Significance
Glial cells in the cochlea support the normal function of peripheral auditory neurons carrying sound information to the brain. This study documents previously undescribed distinct electrophysiological subtypes of glia derived from the hearing cochlea. In addition, we identify a developmental time window during which changes of potassium channel and myelin protein expression lead to apparent region-specific specializations in glial properties mediating normal auditory nerve homeostasis. The data suggest that cochlear glial subtypes could be differentially affected in certain genetic auditory neuropathies and during agerelated hearing loss. distributed in glial cells during the first postnatal week, but then decreased in SCs only as myelination advanced. These results suggest that the membrane properties of SGCs and SCs undergo a profound diversification during cochlear maturation, which may reflect regional functional specializations contributing to normal hearing.

| Animals
Pre-weaning age C57BL/6 mice of either sex were obtained from an in-house breeding facility. Breeding pairs were housed in individual cages under a 12-hr light-dark cycle with a room temperature of 19-23°C and humidity of 55 ± 10%. Food and water were provided ad libitum. Cages were cleaned every 7 days with minimal handling.
Physical enrichment included a small wooden chew block, cardboard nesting house, clear plastic tunnel and a cotton pad nestlet. Mice were killed by cervical dislocation followed by decapitation. Animal work was approved by the UCL Animal Welfare and Ethical Review Body (license 14,798) and conformed to United Kingdom legislation outlined in the Animals (Scientific Procedures) Act 1986. A total of 44 mice were used for the study: 14 in electrophysiological experiments and 30 for immunofluorescence. Tissues were used without regard to the sex of the animal so we are unable to assess sexdependent effects on Kir4.1 expression.
The outer bone of the otic capsule was gently removed to keep the cochlear modiolus, organ of Corti and lateral wall intact. The tissue was then mounted in 4% low-melting temperature agarose (Sigma-Aldrich) in phosphate-buffered saline (PBS) and sectioned at 300 µm on a vibratome (1000 plus system, Intracel). Two to three slices could be obtained from each cochlea. To aid access to glial cell membranes, sections were incubated in 1 mg/ml collagenase P (Sigma-Aldrich) in artificial perilymph solution (see "Electrophysiology" below) for 30 min at 37°C, followed by several washes in artificial perilymph.
Sections were then transferred to the recording chamber for electrophysiological recordings.

| Cochlear cultures
Dissociated cochlear SGN-glia cultures were prepared from P14-P15 mice. Dissections were performed in ice-cold PBS (w/o Ca 2+ or Mg 2+ ) supplemented with 35 mM glucose. Following removal of the outer bone from the otic capsule, the stria vascularis and organ of Corti were carefully dissected away. The basal portion of the modiolus was discarded in order to limit the contamination by central glia, which, by P14 extend into the central modiolus (Hu, 2013). The remaining cochlear tissue was cut into three to four pieces and incubated in 0.25% trypsin in PBS at 37°C for 35 min. Growth medium, DMEM/F-12 supplemented with 10% fetal calf serum (FCS), 1x N2 supplement (ThermoFisher), 2 mM l-glutamine, 17.5 mM glucose, and 1x Penicillin/Streptomycin, was added to quench the reaction.
The cells were pelleted by gentle centrifugation (6 min at 400× g) and the supernatant was discarded. The cell pellet was re-suspended in a small volume of growth medium and the cells were plated onto coverslips pre-treated with poly-l-lysine (100 µg/ml). The coverslips were incubated for 30 min at 37°C, 5% CO 2 , to allow the cells to adhere, then growth medium supplemented with 8 ng/ml BDNF, 8 ng/ ml NT3, and 50 ng/ml LIF (Sigma-Aldrich) was added to each well.
Cell cultures were maintained at 37°C, 5% CO 2 for up to 3 days in vitro (DIV).

| Electrophysiology
Whole-cell patch clamp recordings were performed using an Axopatch 200B amplifier and a Digidata board (Molecular Devices) under the control of pClamp acquisition software (version 8, Molecular Devices). All recordings were performed at room temperature (20-25°C). A peristaltic pump was used to continuously superfuse cells/slices with artificial perilymph solution, containing (in mM): 145 NaCl, 4 KCl, 1 MgCl 2 , 1.3 CaCl 2 , 10 HEPES and 5 glucose, pH 7.3. Patch pipettes were pulled from borosilicate glass (GC120TF-10, Harvard Apparatus) and had a resistance of 3.5-5 MΩ (dissociated cells) or 6-7.5 MΩ (cochlear slices) when filled with intracellular solution containing (in mM): 130 K-gluconate, 5 KCl, 2 MgATP, 2 Na 2 ATP, 0.3 Na 3 GTP, 10 Na 2 -phosphocreatine, 1 EGTA and 10 HEPES, pH 7.2. Series resistance was compensated online by 70%. The liquid junction potential, measured at −13 mV, was subtracted offline. Resting membrane potential was measured using the I = 0 circuitry of the amplifier. The voltage clamp protocol used to assess the currents is described in the Results section and figure legends. BaCl 2 and desipramine hydrochloride (Sigma-Aldrich) were prepared as 1 M and 100 mM stock solutions in water, respectively, before dilution in artificial perilymph solution to a final concentration of 100 µM. Ba 2+ and desipramine were applied via the bath perfusion system using the peristaltic pump. The n numbers stated for electrophysiological experiments refer to the number of cells.
The semi-circular canals were dissected away and the remaining tissue was mounted in 4% low-melting temperature agarose, before sectioning at 150-200 µm intervals on a vibratome (1000 plus system, Intracel). Since the extent of cochlear maturation can vary along the tonotopic axis at a given developmental age (Mann & Kelley, 2011), the basal turns from comparable mid-modiolar sections were selected for age comparisons.

| Immunofluorescence
All steps in the immunofluorescence protocol were performed at room temperature. Sections were blocked and permeabilized in 10% normal goat serum and 0.2% Triton-X-100 in PBS for 1 hr. For experiments using the goat anti-Sox2 antibody, FCS was substituted for the goat serum. Sections were incubated in primary antibodies di-

| Post hoc immunofluorescence
For a subset of electrophysiological recordings, the intracellular solution was supplemented with 0.1% neurobiotin tracer (Vector Laboratories) to allow post hoc identification of recorded cells (Jagger & Forge, 2006). Following the acquisition of electrophysiological data, the patch pipette was carefully withdrawn from the cell, and the coverslip was removed from the recording chamber and fixed in 4% paraformaldehyde for 20 min. Immunofluorescence was then performed as described above, with the additional inclusion of streptavidin conjugated-Alexa Fluor 555 (Life Technologies) in the secondary antibody incubation step, in order to detect neurobiotin.

| Antibodies
Kir4.1 was detected using one of two primary antibodies. A rabbit polyclonal anti-Kir4.1 antibody (APC-035, Alomone Labs, RRID:AB_2040120; immunizing peptide corresponding to amino acids 356-375 of rat Kir4.1; labels ~40 kDa band on Western blot) was used in co-immunolabeling experiments with a goat polyclonal anti-Sox2 antibody (sc-17320, Santa Cruz Biotechnology, RRID:AB_2286684; proprietary immunizing peptide within carboxyl-terminus of human Sox2). A guinea pig polyclonal anti-Kir4.1 antibody (AGP-012, Alomone Labs, RRID:AB_2340962; same immunizing peptide as rabbit antibody; labels ~40 kDa band on Western blot) was used in co-immunolabeling experiments with the rabbit polyclonal anti-myelin protein zero (MPZ; ab31851, abcam, RRID:AB_2144668; proprietary immunizing peptide within last 50 amino acids from the carboxyl-terminus of rat MPZ). The specificity of the rabbit and guinea pig anti-Kir4.1 antibodies has been previously confirmed in immunohistochemistry experiments using Kir4.1 knock-out mouse tissue (Brasko et al., 2017;Djukic et al., 2007). In addition, we conducted control experiments by pre-incubating the rabbit and guinea pig anti-Kir4.1 antibodies with the peptide antigen for 1h at room temperature, before applying to P6 mouse cochlear sections. This resulted in an absence of Kir4.1-specific immunofluorescence within cochlear sections. We conducted no-primary control experiments (to test for non-specific binding by secondary antibodies) by omitting the primary antibodies from the first incubation step in the immunofluorescence protocol. A mouse monoclo-

| Data and statistical analysis
All current-voltage (I-V) plots show the steady-state currents measured during the last 10 ms of the voltage step. To compare the Ba 2+ and desipramine-sensitive currents, the currents were normalized (I norm ) to the maximum inward current amplitude. The degree of inward rectification of the Ba 2+ -sensitive currents was quantified as a rectification index (RI), calculated by dividing the chord conductance at −43 mV by that at −123 mV, using the following equation adapted from a study of Kir4.1 channels in astrocytes (Seifert et al., 2009): where I is the current amplitude of the Ba 2+ -sensitive current at −43 or −123 mV, and E rev is the extrapolated reversal potential.
To assess the extent of Kir4.1 labeling within SGCs during early development, the percentage of SGNs fully enveloped by a Kir4.1positive SGC was calculated within a 146 × 146 µm region of interest (ROI). To determine how Kir4.1 expression changed in the different regions during cochlear development, relative fluorescence intensity values were calculated. Within the same image, a rectangular ROI (50 × 8 µm) was used to measure the median pixel intensity in the two SC regions, where SCs associate with peripherally-(PN) or centrally-directed neurites (CN), and the SGC region. The relative SC/ SGC intensity value for the different SC regions (PN/SGC or CN/ SGC) was then compared across ages. The relative PN/CN intensity value was also calculated to assess any differences in Kir4.1 expression between the two SC regions.
All group data are reported in the text as mean ± standard error of the mean (SE). Shapiro-Wilk tests were performed to test the datasets for normality. Parametric analyses were performed where data were shown to be normally distributed. A t test was used to test for statistically significant differences between two means and an ANOVA was used for three means or more. A paired-sample t test or a repeated measures ANOVA was used to test for differences when measurements were taken from the same cell (e.g., effects of pharmacological block on membrane potential). For repeated measures ANOVA, if the assumption of sphericity was violated (Mauchly's test), Greenhouse-Geisser correction was performed. A Friedman ANOVA was used when repeated measures data were not normally distributed. A two-way ANOVA was used when data were categorized into two groups (e.g., age and SC location). Pairwise comparisons were performed using a post hoc Bonferroni test (ANOVAs) or a Dunn's test (Friedman ANOVA). A value of p < 0.05 was considered significant. Exact p values are reported to three decimal places, with p < 0.001 used where appropriate.

| Kir channels mediate weak inwardly rectifying currents in SGCs
In order to examine the membrane properties of cochlear glia, we initially employed a cochlear slice preparation which allowed the visual targeting of SGCs and SCs. Whole-cell patch clamp recordings were reliably obtained from SGCs in cochlear slices at P5-P6 ( Figure 1). To assess their voltage-activated membrane currents, SGCs were subjected to a series of 200 ms voltage steps from −153 to +17 mV, in 10 mV increments, from a holding potential of −73 mV. This regimen activated currents that were largely voltage independent, as shown by their linear I-V relationship ( Figure 1a). The mean resting membrane potential of SGCs was −83.1 ± 0.5 mV (n = 15), close to the reversal potential for K + (~−89 mV) in these experiments. This value is consistent with a high membrane permeability to K + , and suggests that these currents were mediated by K + . To test whether weak inwardly rectifying Kir channels underlie these currents, the non-selective Kir channel blocker Ba 2+ was added to the bath solution (Figure 1b, c). 100 µM Ba 2+ caused a substantial block of SGC currents, particularly at hyperpolarized test potentials (Figure 1b left and centre panels, and   1c). Digital subtraction of the current recorded during Ba 2+ application from the control current revealed a Ba 2+ -sensitive component ( Figure 1b right panel), which reversed at −82.8 ± 0.63 mV (n = 7) and displayed weak inward rectification (RI = 0.81 ± 0.03, n = 7). The application of Ba 2+ resulted in a small but significant depolarization of the membrane potential, which recovered following washout (control: −84.0 ± 0.7 mV; Ba 2+ : −76.7 ± 1.3 mV; wash: −81.4 ± 0.9 mV, n = 7; F (2, 12) = 23.59, p < 0.001, repeated measures ANOVA; pairwise comparisons, control vs. Ba 2+ : p < 0.001; Ba 2+ vs. wash: p = 0.003; control vs. wash: p = 0.085), suggesting that Kir channels contribute to the resting membrane potential of SGCs. Disappointingly, gigaohm seals could not be reliably achieved in SCs within slice preparations at any age tested, nor could they be obtained routinely in SGCs beyond P6 in this preparation.
Given the limited accessibility of glial cells within cochlear slices, we switched to a dissociated preparation. This approach enabled us to examine the membrane physiology of glial cells from mice after the onset of hearing, which occurs at around P10 (Mikaelian, 1979;Mikaelian & Ruben, 1965). Intact SGC-SGN pairs were apparent immediately after plating P14-P15 dissociated cochlear preparations, and such pairs often remained together for several hours (Figure 2a inset). A small subset of recordings was F I G U R E 2 Weak inwardly rectifying currents in P14-P15 SGCs retaining association with SGNs. (a) Weak inwardly rectifying currents in an SGC still paired with its SGN, shown in inset. Currents were activated using the same voltage clamp protocol described in Figure 1. Application of 100 µM Ba 2+ blocked a large weak inwardly rectifying current. The digitally subtracted Ba 2+ -sensitive current is shown in the right panel.  (Hibino et al., 1999;Jin et al., 2006;Rozengurt et al., 2003) we next sought a more specific Kir channel blocker to examine the contribution of Kir4.1 channels. The tricyclic anti-depressant desipramine blocks members of the Kir3 and Kir4 subfamilies of Kir channels, but not Kir1.1 or Kir2.1 (Kobayashi et al., 2004;Su et al., 2007). Bath application of 100 µM desipramine resulted in a substantial reduction in the weak inwardly rectifying currents in SGCs (Figure 2c

| Diversity in the membrane properties of cultured glial cells
As the membrane currents of SGCs appeared to survive the initial tissue dissociation, we next examined whether cells displaying these currents could be identified after longer periods in vitro. By culturing the dissociated cochlear cells for 1-3 DIV we could also assess the membrane properties of the whole glial cell population. Sox10 and S100 immunolabeling revealed a prominent glial cell population within the cochlear cultures (Figure 3a, b). In general, glial cells had small nuclei and appeared high contrast under differential interference contrast (DIC) optics. In contrast, presumptive fibroblasts were negative for Sox10 and S100. These cells appeared flatter and had a larger cytoplasmic area, and larger nuclei. These morphological distinctions enabled us to target presumptive glial cells during subsequent electrophysiological recordings.  As observed for SGCs in both slices and acutely dissociated SGC-SGN pairs, bath application of 100 µM Ba 2+ caused a substantial reduction of the large weak inwardly rectifying currents in cultured glial cells (Figure 5a, b). The Ba 2+ -sensitive currents reversed at −82.4 ± 0.8 mV and displayed weak inward rectification (RI = 0.71 ± 0.04). Bath application of 100 µM desipramine also resulted in a substantial reduction in the weak inwardly rectifying currents (Figure 5c, d). Comparison of the Ba 2+ -and desipramine-sensitive currents revealed complete overlay of the I-V plots, except at extreme depolarized potentials (Figure 5e), suggesting that Ba 2+ and desipramine target the same Kir channels. Bath application of Ba 2+ resulted in a significant depolarization of the resting membrane potential, which recovered upon washout (Figure 5f; F I G U R E 5 Effect of Kir channel blockers on weak inwardly rectifying glial currents. (a) Currents from a representative glial cell before (control) and after bath application of 100 µM Ba 2+ . Currents were activated using the same voltage clamp protocol described in Figure 1. Digital subtraction of the current in the presence of Ba 2+ from the control current reveals a weak inwardly rectifying Ba 2+ -sensitive current. (2) = 11.14, n = 7, p = 0.004, Friedman ANOVA; pairwise comparisons, control vs. Ba 2+ : p = 0.004; Ba 2+ vs. wash: p = 0.049; control vs. wash: p = 1, post hoc Dunn's test). Similarly, desipramine application resulted in a reversible depolarization of the resting membrane potential (F (1.06, 4.23) = 51.26, n = 5, p = 0.002, repeated measures ANOVA; pairwise comparisons, control vs. Des.: p < 0.001; Des. vs. wash: p < 0.001; control vs. wash: p = 1, post hoc Bonferroni). These results suggest that Kir4 channels contribute to the hyperpolarized resting membrane potential of this subpopulation of glial cells.
Ba 2+ and desipramine were also applied to glial cells exhibiting both weak inwardly rectifying currents and voltage-dependent outward currents (Figure 6). 100 μM Ba 2+ blocked the weak inwardly rectifying component, leaving only voltage-dependent outward currents (Figure 6a, b). 100 μM desipramine also blocked the weak inwardly rectifying current, although some of the voltage-dependent outward current was additionally blocked (Figure 6c, d). The I-V plots for the Ba 2+ -and desipramine-sensitive currents showed complete overlay except at potentials more positive than −30 mV (Figure 6e).
These plots also highlight a difference in the weak inwardly rectifying current at hyperpolarized potentials (−153 to −133 mV) in this type of cell, in comparison to cells with weak inwardly rectifying currents only (compare Figures 5e and 6e). We are uncertain whether the phenomenon of apparent reduced conductance at extreme hyperpolarized potentials in this cell type has physiological relevance.
Nevertheless, the comparable block by Ba 2+ and desipramine suggest that Kir4 channels also mediate the weak inwardly rectifying F I G U R E 6 Effect of Kir channel blockers on cells with both weak inwardly rectifying currents and voltage-dependent outward currents. (a) Currents from a representative glial cell before (control) and after bath application of 100 µM Ba 2+ . Currents were activated using the same voltage clamp protocol as described in Figure 1 In a final set of patch clamp experiments, we performed wholecell dye injections followed by post hoc immunofluorescence to confirm whether cultured glia with weak inwardly rectifying currents expressed Kir4.1 channels. 6/6 cells with large weak inwardly rectifying currents had prominent Kir4.1 immunofluorescence (Figure 7a right panels). In contrast, Kir4.1 immunofluorescence was absent from a glial cell with an outward current only (Figure 7b).

| Kir4.1 is expressed in SGCs and SCs in the early postnatal period
Prior to the onset of hearing in mice (which occurs ~P10), SGNs exhibit bursts of action potentials in response to spontaneous Ca 2+ spikes occurring in immature hair cells (Tritsch et al., 2010). This sound-independent activity may be important for establishing the circuitry of the central auditory pathway prior to hearing onset (Wang & Bergles, 2015). Given the proposed contributions of Kir4.1mediated weak inwardly rectifying currents to the homeostasis of extracellular K + and to the maintenance of neuronal excitability (Kofuji & Newman, 2004), we next sought to examine Kir4.1 expression in SGCs during this critical period of cochlear development.
An antibody against the transcription factor Sox2 was used to label the nuclei of both broad types of cochlear glia (i.e., SGCs and peripherally-and centrally-directed neurites by P4 (Figure 8o, p).

| Progressive decline of Sox2 and Kir4.1 in SCs during cochlear maturation
The prevalence of Kir4.1 expression in SCs during early postnatal development (P0-P4; Figure 8) was unexpected, particularly given the reported lack of expression in these cells in the mature cochlea (Hibino et al., 1999). This led us to investigate the expression of Kir4.1 at later developmental time points. At P6, strong Kir4.1 immunofluorescence was still present in SCs associated with peripherally-and centrallydirected neurites (Figure 9a-c). However, at this age, the proportion of SCs associated with SGN neurites that expressed Sox2 was clearly

| Declining Kir4.1 expression in SCs is coincident with advancing myelination
In the previous experiments we identified a developmental period during which Kir4.1 expression decreased coincidently with a reduction in Sox2 expression. This observation may be consistent with the role of Sox2 as an inhibitor of myelination by SCs in vivo (Roberts et al., 2017). Consequently, we next investigated the relative timing of Kir4.1 subunit expression compared to the advancement of auditory nerve myelination. Functionally mature SGNs propagate action potentials via regularly spaced nodes of Ranvier located along their neurites, with SCs myelinating the internodal regions. In the mouse cochlea, myelination by SCs begins around P4-P5 (Romand & Romand, 1990), which is close to the onset in the decline of Kir4.1 immunofluorescence in SCs from around P6 (Figure 9). Using an antibody against myelin protein zero (MPZ), the most abundant peripheral myelin protein (Greenfield et al., 1973), we explored the relationship between myelination and Kir4.1 expression in more detail (Figure 10). By P6, MPZ immunofluorescence was detected in the regions where SCs associate with peripherally-and centrally-directed SGN neurites (Figure 10a). In addition, there was some labeling of SCs along the length of neurites running through the spiral ganglion ( Figure 10a). Higher power imaging revealed distinct MPZ labeling in segments along some peripherally-and centrally-directed neurites (Figure 10b, c). By P10, the extent of MPZ labeling had increased in SCs in both regions (Figure 10d-f). As described in the previous section, by P10, Kir4.1 immunofluorescence had declined substantially in SCs associated with peripherally-directed neurites (Figure 10e), and to a lesser extent in SCs associated with centrally-directed neurites (Figure 10f), with short lengths of SC membrane still retaining Kir4.1 expression. By P14, Kir4.1 expression was largely undetectable in SCs, and strong MPZ immunofluorescence was prevalent in SCs associated with peripherally-and centrally-directed neurites, including those running through the spiral ganglion (Figure 10g, i).

| D ISCUSS I ON
SGNs are among the fastest spiking neurons in the mammalian nervous system, with spontaneous rates ranging from 0 to 150 Hz, and evoked rates exceeding 1 kHz at tone onset (Taberner & Liberman, 2005). This suggests that the glial cells in the auditory periphery will be subjected to unusually high metabolic and homeostatic pressures. As in other nervous tissues, auditory glial cells are likely to play indispensable roles in supporting and maintaining neuronal excitability, through processes including myelination, K + homeostasis, neurotransmitter clearance, and provision of metabolites. Many of these roles depend on the expression of particular subtypes of ion channels and transporters on both glial and neuronal cell membranes. Here, we provide the first description of the voltage-activated membrane currents of cochlear glial cells and highlight the prominence of a weak inwardly rectifying current within SGCs.
SGCs had hyperpolarized resting membrane potentials, a property that depended on Ba 2+ -and desipramine-sensitive channels. Other cell types lacked this current though, suggesting that cochlear glia have specific membrane properties that support their distinct physiological roles.

| Weak inwardly rectifying currents in SGCs
The weak inward rectification of SGC currents, and their pharmacological block by Ba 2+ and desipramine suggest that these currents are mediated by Kir4 channels. Further to this, Kir4.1 immunofluorescence in cultured SGCs with large weak inwardly rectifying currents is consistent with an involvement of Kir4.1 subunits. Kir4.1 subunits can form heteromers with Kir5.1 (Tanemoto et al., 2000), but the lack of Kir5.1 immunofluorescence in cochlear SGCs (Hibino et al., 2004), suggests that Kir4.1 homomeric channels may predominate. Kir4.1 channels are expressed in SGCs of other sensory ganglia, including the vestibular ganglion, trigeminal ganglion, and the dorsal root ganglion (Hibino et al., 1999;Tang et al., 2010;Vit et al., 2006). Trigeminal ganglion SGCs from mature mice display a comparable weak inwardly rectifying current, which is also Ba 2+ -and desipramine sensitive (Tang et al., 2010).
Knock-out of Kcnj10 (which encodes Kir4.1) abolished the Ba 2+and desipramine-sensitive currents in trigeminal SGCs, although these experiments were restricted to the P8-P9 developmental period as these mice display severe neurological defects (Tang et al., 2010).

| Physiological roles for weak inwardly rectifying currents in SGCs
Within the mature PNS, Kir4.1 expression is largely restricted to SGCs (Hibino et al., 1999). In the CNS, however, both astrocytes and oligodendrocytes express Kir4.1 (Higashi et al., 2001;Li et al., 2001;Olsen et al., 2006;Poopalasundaram et al., 2000). One of the proposed physiological functions of glial Kir4.1 channels is to regulate K + homeostasis (Olsen & Sontheimer, 2008). Since neuronal K + efflux occurs during the repolarization phase of an action potential, spiking activity is associated with a local rise in extracellular K + . Without clearance by glial cells, the accumulation of K + could result in neuronal depolarization and hyper-excitability. The pharmacological block of Kir4.1 channels, or their genetic deletion from CNS glia results in impaired K + clearance (Bay & Butt, 2012;Chever et al., 2010;Djukic et al., 2007;Kucheryavykh et al., 2007;Larson et al., 2018;Neusch et al., 2006), highlighting the importance of Kir4.1 in K + buffering during neuronal activity.
SGNs are endowed with specialized ion channel mechanisms that enable them to fire action potentials at unusually high frequencies (Browne et al., 2017). Without clearance mechanisms in place during this high-frequency firing, substantial K + accumulation could occur around SGNs. Since SGNs are myelinated, the sites of K + efflux are presumably restricted to their nodes of Ranvier, including the peri-somatic nodes adjacent to the cell body (Hossain et al., 2005;Smith et al., 2015). However, various voltage-gated K + (Kv) channels including Kv1.1 and Kv1.2 subunits are expressed in SGNs (Mo et al., 2002;Wang et al., 2013) where they localize to the plasma membrane of the cell body (Smith et al., 2015), suggesting somatic K + efflux may also occur. The Kir4.1 channels in SGCs are well positioned to contribute to K + clearance in the spiral ganglion. The absence of Kir4.1 in mature SCs, though, suggests that other mechanisms of K + clearance may occur at more distal nodes of Ranvier.
As has been described for other glial cells (Djukic et al., 2007;Olsen et al., 2006;Seifert et al., 2009;Smith et al., 2020), blockade of the weak inwardly rectifying current in spiral ganglion SGCs by Ba 2+ or desipramine led to membrane depolarization, suggesting that this current is important for setting their resting membrane potential. Within astrocytes, the hyperpolarizing effect of Kir4.1 on the resting membrane potential is proposed to maximize synaptic glutamate clearance during neuronal activity (Olsen & Sontheimer, 2008).
SGCs in the spiral ganglion are located a considerable distance from synapses, however, they express the glutamate transporter GLAST (Furness & Lehre, 1997) and glutamate receptors (Safieddine & Wenthold, 1997), suggesting roles for SGCs in extra-synaptic glutamate clearance and/or signaling.

| Implications of Kir4.1 dysfunction in hearing loss
Kir4.1 knock-out mice are profoundly deaf (Rozengurt et al., 2003) and in humans, mutations in KCNJ10, encoding Kir4.1, cause EAST syndrome where sensorineural hearing loss is a consistent feature (Bockenhauer et al., 2009;Scholl et al., 2009). Kir4.1 is expressed in several different cell types within the mature cochlea. In addition to its expression in SGCs, Kir4.1 is also detected in supporting cells of the organ of Corti (Hibino et al., 1999;Taylor et al., 2012), outer sulcus root cells (Jagger et al., 2010), and intermediate cells of stria vascularis in the cochlear lateral wall (Ando & Takeuchi, 1999).
Stria vascularis generates the endocochlear potential which provides a driving force for hair cell mechano-electrical transducer currents, and is essential for normal hearing. The loss of the endocochlear potential is suggested to be a major cause of deafness in Kir4.1 knock-out mice (Rozengurt et al., 2003) and in patients with EAST syndrome (Abdelhadi et al., 2016). However, the expression of Kir4.1 in cochlear epithelial cells and glia raises the possibility that multiple sensorineural deficits contribute to the associated pathologies. A recent study has highlighted progressive changes in Kir4.1 expression in the cochleae of aging mice and humans, including within SGCs, leading to speculation this may in part contribute to the hearing loss associated with aging (Liu et al., 2019).
Again, the involvement of Kir4.1 in the functions of numerous cochlear tissues complicates the interpretation of its role in agerelated hearing loss.

| Heterogeneous membrane properties of cochlear glial cells
In addition to describing glial cells with weak inwardly rectifying currents, cells with voltage-dependent outward currents were identifiable within our cultures from hearing mice. Two apparent subsets of cells displayed voltage-dependent outward currents only. These currents had distinct voltage thresholds of activation. Kv currents are a characteristic feature of cultured mammalian SCs (Chiu et al., 1984;Konishi, 1989aKonishi, , 1989b, and several Kv channel subunits have been detected (Mi et al., 1995;Sobko et al., 1998). It therefore seems likely that the cells with voltagedependent outward currents in our cultures represent the endogenous cochlear SC population. There are at least two subtypes of SCs associated with afferent SGNs in the cochlea: myelinating and non-myelinating SCs. Efferent neurons, whose cell bodies reside within the medial and lateral superior olivary complex in the brainstem, also have axons that project into the peripheral cochlea.
These are associated with myelinating and non-myelinating SCs, respectively (Lopez-Poveda, 2018). Further studies are required to determine whether these different subpopulations of SCs differ in their membrane properties.
Interestingly, studies in rodent sciatic nerve have demonstrated that while embryonic SCs display only Kv currents (Konishi, 1990), there is a transient appearance of Kir currents around the onset of myelination which then disappears over the first postnatal week (Konishi, 1990, Wilson & Chiu, 1990. Although Kir2 channels may mediate currents around nodes of Ranvier in mature SCs (Mi et al., 1996), the molecular identity of the channels mediating the Kir currents in developing sciatic nerve SCs has not been determined.
The finding that Kir4.1 is transiently expressed in cochlear SCs at a similar developmental stage raises the possibility of a comparable role for these channels in SC maturation. A subpopulation of cochlear glial cells displayed both voltage-dependent outward currents and weak inwardly rectifying currents in culture. It has been demonstrated that the culturing of SCs from sciatic nerve results in the loss of Kir currents within 2 DIV (Konishi, 1992). One possibility is that like in sciatic nerve SCs, cultured cochlear SGCs lose Kir4.1 channels in vitro, and then upregulate outwardly rectifying currents.
Although our cultures were limited to 3 DIV, changes in the membrane properties of SGNs within cochlear cultures have been observed over a similar time frame (Cai et al., 2017).

| Developmental expression of Kir4.1 channels as a marker of auditory nerve functional maturation
In rat cochlea, Kir4.1 immunolabeling was first detected in SGCs at P8, reaching a plateau by P14 (Hibino et al., 1999). It was proposed that the timing of Kir4.1 expression may be correlated with the onset and maturation of sound-evoked action potentials in the auditory nerve, which are first detected ~P10 in mouse (Mikaelian & Ruben, 1965, Mikaelian, 1979. In contrast, we detected Kir4.1 immunofluorescence in mouse SGCs as early as P1. Although sound-evoked action potentials do not occur until ~P10 in the mouse, during early postnatal maturation SGNs fire action potential bursts in response to Ca 2+ spikes in inner hair cells which are further propagated to higher auditory centers (Tritsch et al., 2010). In vivo recordings from cats, which have a similar hearing onset to mice, demonstrate spontaneous action potentials as early as P2-P3 (Romand, 1984), which also show a bursting behavior (Jones et al., 2007). Our surprisingly early postnatal detection of Kir4.1 immunofluorescence in mouse SGCs, with most SGNs fully enveloped by Kir4.1 positive SGCs at P4, may reflect the need for peri-neuronal K + buffering to support this early sound-independent activity.
Although action potential bursts occur in rodent SGNs during the early postnatal period, nodes of Ranvier do not form until ~P6-P9 (Kim & Rutherford, 2016). This suggests that early action potentials propagate in a non-saltatory fashion, and the ion channels responsible for this firing may be distributed more evenly along the SGN plasma membrane. This could result in K + efflux occurring along the entire length of SGN neurites, and could account for the transient expression of Kir4.1 we observed in SCs during this early developmental period. A similar transient expression of Kir4.1 occurs in guinea pig SCs, beginning around embryonic day 40 (Jin et al., 2006). The loss of Kir4.1 immunofluorescence in mouse SCs from ~P8 onwards correlates with the timing of node of Ranvier formation, when K + efflux is most likely to be restricted to sites around nodes and thus negates the need for Kir4.1 expression throughout the entire SC membrane.
The transient expression of Kir4.1 in cochlear SCs may signal changes in their maturational state. Kir4.1 expression in SCs first occurs a few days prior to the onset of myelination and gradually declines as myelination progresses along SGN neurites. We show here that the decline in Kir4.1 expression and concomitant upregulation of MPZ coincides with reduced immunofluorescence for the transcription factor Sox2. These findings lead us to hypothesize that Sox2 may also be a negative regulator of myelination in SCs in the auditory nerve, as has been described for the mouse sciatic nerve in vivo (Roberts et al., 2017). The relationship between Kir4.1 and SC-mediated myelination deserves further study, particularly in disease models where extensive neural remodeling is required to repair damaged hearing.
In conclusion, our data show that there are distinct membrane properties between glial subtypes in the auditory nerve in hearing mice. There is a divergence of Kir4.1 expression among cochlear glia following the period of early postnatal maturation, and this may reflect the specialized roles of SGCs and SCs during normal hearing.
Our study has defined conditions under which we can routinely examine the electrophysiological properties of cochlear glia in vitro, an advance that will allow us to increase our understanding of auditory nerve homeostasis in health and disease.

DECL AR ATION OF TR ANS PAREN C Y
The authors, reviewers and editors affirm that in accordance to the policies set by the Journal of Neuroscience Research, this manuscript presents an accurate and transparent account of the study being reported and that all critical details describing the methods and results are present.

ACK N OWLED G M ENTS
This work was supported by the Biotechnology and Biological Sciences Research Council (BBSRC) grant BB/M019322/1 to D.J.J.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/jnr.24744.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.