Sodium-activated potassium channels shape peripheral auditory function and activity of the primary auditory neurons in mice

Potassium (K+) channels shape the response properties of neurons. Although enormous progress has been made to characterize K+ channels in the primary auditory neurons, the molecular identities of many of these channels and their contributions to hearing in vivo remain unknown. Using a combination of RNA sequencing and single molecule fluorescent in situ hybridization, we localized expression of transcripts encoding the sodium-activated potassium channels KNa1.1 (SLO2.2/Slack) and KNa1.2 (SLO2.1/Slick) to the primary auditory neurons (spiral ganglion neurons, SGNs). To examine the contribution of these channels to function of the SGNs in vivo, we measured auditory brainstem responses in KNa1.1/1.2 double knockout (DKO) mice. Although auditory brainstem response (wave I) thresholds were not altered, the amplitudes of suprathreshold responses were reduced in DKO mice. This reduction in amplitude occurred despite normal numbers and molecular architecture of the SGNs and their synapses with the inner hair cells. Patch clamp electrophysiology of SGNs isolated from DKO mice displayed altered membrane properties, including reduced action potential thresholds and amplitudes. These findings show that KNa1 channel activity is essential for normal cochlear function and suggest that early forms of hearing loss may result from physiological changes in the activity of the primary auditory neurons.

of ion channels that regulate encoding of auditory signals as part of the afferent signalling complex 3 , we used RNAseq to obtain whole transcriptomes from intact preparations of the organ of Corti and SGNs isolated from mice ( Fig. 1). Following the classification of the IUPHAR/BPS Guide to Pharmacology, we mined these transcriptomes to determine expression of subsets of genes encoding for known voltage-gated ion channels (97/146 known genes expressed; Fig. 1A), potassium channels (54/79 genes; Fig. 1C), Ca 2+ -and Na + -activated potassium channels (7/8 genes; Fig. 1D) and, specifically, expression levels of genes encoding the SLO family of ion channels (Fig. 1E). For this group of K + channels, Kcnma1, which encodes K Ca 1.1/SLO1/BK, was most abundantly expressed (35.1 ± 4.7 RPM). Kcnt1, which encodes K Na 1.1/SLO2.2/Slack, was expressed at intermediate levels (6.7 ± 1.5 RPM). Kcnt2, which encodes K Na 1.2/SLO2.1/Slick, was absent from two of the three replicates and expressed at <1 RPM in one replicate. Kcnu1, which encodes K Ca 5.1/SLO3, was expressed at ≈1 RPM (1.1 ± 0.2 RPM). To validate the utility of RNAseq to identify SLO transcripts in the sensorineural structures of the inner ear, we also examined the expression of K Na 1-encoding transcripts in other tissues collected in parallel. RNAseq analyses revealed the following expression levels (in RPM): Kcnt1: 38 ± 1.7 (cerebellum), 0.18 ± 0.09 (heart) and 0 (liver); Kcnt2: 4.5 ± 0.82 (heart), 0 (cerebellum) and 0.060 ± 0.31 (liver). Thus, RNAseq analyses yields results consistent with qPCR detection of Kcnt1 and Kcnt2 in these tissues 19 . All values are expressed as mean ± SEM.
We suspected that K Na 1-encoding transcripts in the sensorineural structures were specifically expressed by the SGNs for two reasons. First, K Na 1 channels have been observed in other primary sensory neurons 16,19,[21][22][23][24] . Second, activation of K Na 1 channels requires rises in intracellular Na + mediated by activation of TTX-sensitive, voltage-gated Na + channels and/or ionotropic AMPA-type glutamate receptors 12 . SGNs express TTX-sensitive (persistent and resurgent) voltage-gated Na + channels 4,25 as well as AMPA-type glutamate receptors 26,27 . In contrast, mature hair cells do not express voltage-gated Na + currents 28,29 or glutamate receptors. To examine cell-specific expression of K Na 1 channels in the organ of Corti and SGNs, we utilized a variety of commercially available antibodies against these channels. Unfortunately, none of these antibodies yielded consistent or reliable results (data not shown). Therefore, as an alternative strategy, we used single molecule fluorescent in situ hybridization (smFISH) to localize expression of K Na 1-encoding transcripts in cochlear sections (Fig. 2). Transcripts encoding for K Ca 1.1, K Na 1.1 and K Na 1.2 (Kcnma1, Kcnt1 and Kcnt2) but not K Ca 5.1 (Kcnu1) were detected (as green particles) in the TUJ1-labeled (red) SGN somas ( Fig. 2A). To quantify relative transcript abundance, the total number of RNA molecules detected per SGN was calculated across independent replicates (Fig. 2B). Kcnma1 (encoding K Ca 1.1) was richly expressed, Kcnt1 and Kcnt2 (encoding K Na 1.1 and K Na 1.2) were expressed at intermediate values, and Kcnu1 (encoding K Ca 5.1) and no probe controls showed little to no expression (Kcnma1: 13 ± 1.8 mRNA/SGN, n = 3 replicates; Kcnt1: 4.9 ± 0.69 mRNA/SGN, n = 4 replicates; Kcnt2: 3.6 ± 0.12 mRNA/ SGN, n = 4 replicates; Kcnu1: 1.1 ± 0.42 mRNA/SGN, n = 3 replicates; no probe: 0.82 ± 0.34 mRNA/SGN, n = 4 replicates). All values are expressed as mean ± SEM. Together, smFISH and RNAseq analyses suggest that of the two K Na 1-encoding transcripts, K Na 1.1-encoding transcripts are more abundantly expressed in the sensorineural structures and specifically SGNs. Differences in the relative expressions of these two transcripts between the two techniques most likely arises from dilution of transcript expression in the intact preparation used for RNAseq. The majority of transcripts encoding voltage-gated ion channels encode potassium channels, with transcripts corresponding to a total of 68% of known potassium channels (54/79 genes) expressed. (D) Of the potassium channels, 88% of the Ca 2+ -and Na + -activated potassium channels (7/8 genes) are expressed. (E) Three of the four members of the subset of Ca 2+ -and Na + -activated potassium channels encoding the SLO family of ion channels are expressed. For this group of ion channels, Kcnma1, which encodes K Ca 1.1/SLO1/BK, is most abundantly expressed. Kcnt1, which encodes K Na 1.1/SLO2.2/Slack, is expressed at intermediate levels. Kcnt2, which encodes K Na 1.2/SLO2.1/Slick, is absent from two of the three replicates and expressed at <1 RPM in one replicate. Kcnu1, which encodes K Ca 5.1/SLO3, is expressed at low levels. Data are plotted to show individual replicates (animals) and mean ± SEM. Values (mean ± SEM) are provided in the Results. . smFISH localizes transcripts encoding the SLO channels K Ca 1.1, K Na 1.1, K Na 1.2 but not K Ca 5.1 to the spiral ganglion neurons (SGNs). Expression of K Na 1-encoding transcripts in the SGNs was examined using smFISH in excised preparations of the OC/SGN isolated from 6-week-old mice. (A) RNA molecules encoding for K Ca 1.1 (Kcnma1), K Na 1.1 (Kcnt1), K Na 1.2 (Kcnt2), K Ca 5.1 (Kcnu1) and no probe controls were detected as fluorescent puncta (green) in TUJ1-positive (red) SGN somas. For easier visualization of fluorescently labelled mRNA molecules, identical views are provided in which detected mRNA molecules are represented instead as spheres (green) and SGNs are shown in grey. (B) The mean number of RNA molecules detected per SGN was calculated as described in the Methods. Kcnma1 was most abundantly expressed, Kcnt1 and Kcnt2 were expressed at intermediate values, and Kcnu1 and no probe controls showed little to no expression. Data are plotted to show individual replicates (animals) and mean ± SEM. Values (mean ± SEM) are provided in the Results.
www.nature.com/scientificreports www.nature.com/scientificreports/ comparisons). In addition to absolute thresholds, we also examined wave I amplitudes and latencies as a function of sound intensity. For both WT and DKO mice, ABR waveform amplitudes increased and latencies decreased as stimulus intensity increased. To compare changes between genotypes and sound stimuli, we calculated I/O linear regression slopes for wave I amplitudes (Fig. 3C) and latencies (Fig. 3D) as a function of stimulus intensity. Wave I amplitude I/O linear regression slopes were significantly reduced in response to click and tone pips at 8 and 16 kHz in DKO compared to WT mice (p-values = 0.0078 for clicks, 0.0061 for 8 kHz, 0.0061 for 16 kHz and 0.8047 for 32 kHz, ordinary one-way ANOVA with Sidek's correction for multiple comparisons). Although mean values for wave I latency I/O linear regression slopes were reduced in response to click and tone pips at 16 kHz in DKO compared to WT mice, these differences were not statistically significant (p-values = 0.4501 for clicks, 0.9998 for 8 kHz, 0.8392 for 16 kHz and 0.7198 for 32 kHz, one-way ANOVA with Sidek's correction for multiple comparisons). Mean values ± SEMs are provided in Table 1. These results indicate that function of the auditory nerve is altered in K Na 1 DKO mice. In fact, DKO mice show a form of hearing loss termed "hidden hearing loss", which is characterized by normal thresholds but reduced suprathreshold wave I responses. Hidden hearing  Cochlear morphology, spiral ganglion cell density, and architecture of the afferent synapses are normal in K Na 1 DKO mice. In both animal models and humans, overt hearing loss is associated with loss of sensorineural structures and "hidden" hearing loss is particularly associated with loss of synapses between the sensory inner hair cells (IHCs) and SGNs [30][31][32][33] . Therefore, we assessed both the morphology of the sensorineural structures and the integrity of the synapses between the IHCs and SGNs in DKO compared to WT mice. Spiral ganglion cell (SGC) density and overall cochlear morphology was examined in mid-modiolar serial sections in cochleae isolated from (6-week-old) WT and DKO mice (Fig. 4). No abnormalities were seen in the cells and tissues of the cochlea, including the inner and outer hair cells, SGCs, stria vascularis, spiral ligament and all supporting structures of the cochlear duct (Fig. 4A). Importantly, both the packing density of the SGCs as well as the density of auditory nerve fibers were visually comparable between WT and DKO mice. When SGC density was quantified, no statistically significant differences were observed between WT and DKO mice in either cochlear apical (WT: 158 ± 3 SGC/mm 2 , n = 5; DKO: 176 ± 11 SGC/mm 2 , n = 4, p value = 0.7827, one-way ANOVA with Sidek's correction for multiple comparisons) or basal segments (WT: 171 ± 14 SGC/mm 2 , n = 5; DKO: 184 ± 9 SGC/mm 2 , n = 4, p value = 0.9278, one-way ANOVA with Sidek's correction for multiple comparisons). Because the SGNs can show delayed loss after much earlier loss of synaptic contacts to the IHCs 30 , we also examined the synaptic connections between the IHCs and SGNs at three tonotopic locations (8,16 and 32 kHz) in intact preparations of the organ of Corti and SGNs isolated from (6-week-old) WT and DKO mice (Fig. 5). There were no observable qualitative differences between WT and DKO mice in the organization of afferent synapses, identified as paired CtBP2 (green) and GluR2/3 (red) immunopuncta (Fig. 5A). Quantification of the mean number of synapses per IHC indicated no statistically significant differences between WT and DKO mice when comparing between tonotopic regions ( Fig. 5B; p-values = 0.9543 for 8 kHz, 0.8315 for 16 kHz and 0.3834 for 32 kHz, ordinary one-way ANOVA with Sidek's correction for multiple comparisons). Mean values ± SEMs are provided in Table 2. Thus, despite differences in the ABR wave I responses between WT and DKO mice, which mimic "hidden" hearing loss in the DKO mice, there were no indications of morphological alterations or synaptopathy in the cochleae of DKO mice.
These findings suggested that the phenotype of hidden hearing loss in the DKO mice may result from changes other than synaptopathy in the SGNs. Therefore, we examined the expression and distribution of various proteins positioned to shape SGN excitability using immunofluorescence in the isolated preparation of the organ of Corti and SGNs from (6-week-old) WT and DKO mice. We particularly examined placement of these proteins at the SGN afferent dendrites, where synapses are made between the SGNs and IHCs and where spike generation occurs. The precise alignment of proteins here is expected to underlie SGN excitability and perhaps also firing synchrony 34 . We examined expression of the Na + , K + -ATPase α3 (ATP1A3, green, Fig. 6A), a transporter expressed in the SGNs 35 and known to regulate neuronal excitability 36 , the patterns of myelination indicated by distribution of myelin basic protein 37 (MBP, red, Fig. 6A), and the expression of various ion channels known to support SGN firing, including the low voltage-activated K V 1.1 [38][39][40] (green, Fig. 6B), the high voltage-activated K V 3.1 41,42 (green, Fig. 6C) and the voltage-gated Na V 1.6 43,44 (red, Fig. 6C). We found no visible evidence of altered ATP1A3 expression or patterns of myelination in the DKO compared to WT mice. Moreover, we found no visible evidence of altered expression or localization of K V 1.1, K V 3.3 or Na V 1.6 compared to previous topographical characterization 34 . In both WT and DKO mice, K V 1.1 was localized to heminodes and nodes and K V 3.1 and Na V 1.6 were colocalized at heminodes and nodes. These data were collected from the mid-cochlear (16 kHz) region. Although not shown, expression patterns of ATP1A3, MBP, K V 1.1, K V 3.3 and Na V 1.6 were similarly expressed at other regions as well as in the somata of the SGNs from WT and DKO mice. These findings indicate that loss of K Na 1 channels does not alter the molecular and cellular architecture of proteins critical for shaping SGN responses in vivo. These findings, in turn, motivated our investigation of possible alterations in the physiology of SGNs lacking K Na 1 channels. spiral ganglion neurons isolated from K Na 1 DKO mice do not have Na + -sensitive outward K + currents and display altered action potential waveforms. To investigate directly the contribution of K Na 1 channels to SGN responses, we performed whole cell patch clamp recordings on SGNs isolated from (6-week-old) WT and DKO mice. The use of genetic models circumvents the lack of pharmacological tools to block K Na 1 channels 12 and/or methodological approaches that require comparison of recordings from separate cells with different internal Na + concentrations 17,19,45 . In the voltage-clamp configuration, whole-cell currents revealed a transient inward current, mediated by voltage-dependent inward (Na + ) current, as well as outward currents in SGNs isolated from both WT (Fig. 7A, control) and DKO (Fig. 7B, control) mice in response to varying depolarization from a holding potential of −80 mV to 60 mV in 10-mV increments. Bath application of 1 µM TTX completely blocked voltage-sensitive inward currents in SGNs isolated from both WT (Fig. 7A, +TTX) and DKO (Fig. 7B, +TTX) mice. The difference-current generated by subtraction of currents recorded in the presence of TTX from those recorded before TTX application (control) revealed the TTX-sensitive currents in SGNs isolated from WT (Fig. 7A, difference) and DKO (Fig. 7B, difference). These difference currents consisted of a fast inward current and a slow outward current. The TTX-sensitive inward current arises from activation of voltage-dependent Na + channels, whereas the outward component is inferred to arise from the Na + -activated K + channels. From a total of 31 basal SGNs isolated from WT mice, 16 expressed sizable TTX-sensitive K + currents, ranging from 10 to 32% of the total outward current. The remaining 15 SGNs expressed substantially less TTX-sensitive K + current (approximately 2-5% of the total outward current). In contrast, SGNs isolated from DKO mice were www.nature.com/scientificreports www.nature.com/scientificreports/ always devoid of TTX-sensitive K + currents. To quantify findings across basal SGNs, the current density-voltage relationship was generated using the steady state K + current amplitude for both WT mice (Fig. 7C, n = 10 cells from which both voltage and current clamp data were collected) and DKO mice (Fig. 7D, n = 13 cells). The total www.nature.com/scientificreports www.nature.com/scientificreports/ outward K + current density at 0 mV ( Fig. 7E) was 72.6 ± 1.7 pA/pF in SGNs from WT mice (n = 10) and not significantly different from the total outward K + current density of 77.3 ± 2.6 pA/pF in SGNs from DKO mice (n = 13; p value = 0.3348, ordinary one-way ANOVA with Sidek's correction for multiple comparisons). In contrast, when present, the mean TTX-sensitive outward K + current density at 0 mV was 17.6 2 ± 1.3 pA/pF in SGNs from WT mice (n = 10) and significantly greater than the mean TTX-sensitive outward K + current density at 0 mV in SGNs from DKO mice (5.0 ± 0.5 pA/pF, n = 13; p value < 0.0001, ordinary one-way ANOVA with Sidek's correction for multiple comparisons).  Table 2. Statistical analyses are provided in the Results.
www.nature.com/scientificreports www.nature.com/scientificreports/ Similar findings were also observed when recording from SGNs isolated from the apical one-third of the cochlea from (6-week-old) WT and DKO mice. For 23 apical SGNs, 14 SGNs had a mean total outward current density at 0 mV of 73.0 ± 10.0 pA/pF. The mean TTX-sensitive K + current at 0 mV was calculated to be 14.7 ± 6.0 pA/pF (or approximately 20% of the total outward current). The remaining 9 neurons expressed substantially less TTX-sensitive K + current (approximately 2-5% of the total outward current). Again, SGNs isolated from DKO mice showed no measurable TTX-sensitive K + currents. Similar findings were also observed when extracellular Na + was replaced by either Li + , which permeates voltage-gated Na + channels but would not be expected to activate K Na 1 channels fully 46 or by NMG + , a bulkier monovalent cation that is less likely to permeate Na + channels and activate Na + -dependent K + currents. Both Li + and NMG + -sensitive K + current densities calculated at 0 mV for SGNs isolated from WT mice (Li + : 8.7 ± 0.6 pA/pF, n = 10; NMG: 4.5 ± 0.5 pA/pF, n = 10) were greater than the equivalent currents calculated for SGNs isolated from DKO mice (Li + : −2.7 ± 0.4 pA/pF, n = 13; NMG: −1.5 ± 0.2 pA/pF, n = 13). Similar subtractive methods and Na + substitution have been used previously to identify K Na 1 currents [45][46][47] .
Together these data indicate the presence of Na + -activated K + currents in approximately half of both apical and basal SGNs isolated from WT mice. The absence of this current in DKO mice is consistent with the current being carried by K Na 1 channels. Furthermore, the reduction of this current in SGNs from WT mice when voltage-gated Na + channels are blocked by TTX or when Na + is replaced by either Li + or NMG, strongly suggest that the K Na 1 channels in SGNs are activated, at least in part, by Na + influx most likely via the TTX-sensitive voltage-dependent Na + channels.
In other neurons, K Na 1 currents contribute to setting the resting membrane potential, shaping the action potential waveform and altering repetitive firing 12,13 . To examine the effects of K Na 1 channels on membrane properties in SGNs, we performed current-clamp recordings on the majority of SGNs also examined by voltage-clamp recordings. In this way we could compare SGNs from DKO mice with SGNs that express K Na 1 currents from WT mice. In general, while resting membrane potentials appeared similar in SGNs from both genotypes, SGNs isolated from WT mice (Fig. 8A) required increased current injection to evoke an action potential compared to SGNs isolated from DKO mice (Fig. 8B). Comparatively, action potentials evoked in SGNs from WT mice were generally slower to initiate and larger in amplitude compared to those evoked in SGNs from DKO mice (Fig. 8C). Quantifying across cells, the resting membrane potentials of SGNs isolated from basal cochlear turns of WT and DKO mice were not significantly different (Fig. 8D, p value = 0.9333, unpaired, two-tailed t test). Nevertheless, active membrane properties did significantly differ between SGNs from DKO compared to WT mice ( Fig. 8E-H). In general, SGNs isolated from DKO compared to WT mice were more excitable, showing significantly reduced action potential latencies (Fig. 8E, p value = 0.0007, unpaired, two-tailed t test) and thresholds (Fig. 8F, p value < 0.0001, unpaired, two-tailed t test). Additionally, action potential amplitudes were significantly decreased in SGNs from DKO compared to WT mice (Fig. 8G, p value = 0.0004, unpaired, two-tailed t test) whereas action potential durations (measured as the width at half-maximal spike amplitude) were significantly increased in SGNs from DKO compared to WT mice and durations (Fig. 8H, p value < 0.0001, unpaired, two-tailed t test). Similar trends were observed in SGNs isolated from the apical one-third of the cochlea, with the exception that apical SGNs from DKO mice also showed more negative resting membrane potentials (RMPs) compared to SGNs from WT mice. Values are provided in Tables 3 and 4. These data suggest, that across frequencies, the presence of K Na 1 channels in SGNs normally serves to delay action potential generation, increase the threshold of action potential generation, and increase action potential amplitude.

Discussion
To accelerate the discovery of K + channels that regulate encoding of auditory signals, we used RNA sequencing to obtain transcriptomes from the intact sensorineural structures, including the organ of Corti and SGNs, isolated from adult (6-week-old) mice. Transcriptomic analyses revealed a vast repertoire of K + channels positioned to regulate encoding of auditory signals. In fact, of the tissues investigated by transcriptomic analyses, K + channel diversity was greatest in the organs of Corti and SGNs (with 68% or 54/80 genes known to encode K + channels expressed) compared to cerebellum (65% or 49/80 genes), heart (59% or 47/80 genes) and liver (35% or 28/80 genes). In prioritizing K + channels for further investigation, we were especially interested in the expression of the SLO family of ion channels and particularly the sodium-activated K + channels, K Na 1.1 and K Na 1.2. These channels are unique in their regulation by intracellular [Na + ], and recent work has identified their central role in regulating neuronal function, development and plasticity 12 .
In this work, expression analysis indeed identified K Na 1-encoding transcripts in the sensorineural structures of the organ of Corti and SGNs (by RNAseq) and specifically localized transcript expression to the SGNs (by smFISH). Multiple independent observations corroborate SGN-specific expression of K Na 1-encoding transcripts and K Na 1 channels. First, K Na 1-encoding transcripts (Kcnt1 and Kcnt2) have been identified in earlier microarray expression analyses of FACS-sorted mature (P15) SGNs 48 and absence of both Kcnt1 and Kcnt2 in FACS-sorted mature (P16) hair cells 49 . Although a previous microarray study examining isolated mature (P25-30) hair cells    www.nature.com/scientificreports www.nature.com/scientificreports/ detected low expression of Kcnt1 and Kcnt2 in both the inner and outer hair cells 50 , a subsequent more sensitive RNAseq study found no expression of either gene in the mature hair cells 51 . Second, K Na 1 currents and K Na 1-encoding transcripts have been previously detected in rat vestibular ganglion neurons using a combination of electrophysiology and reverse-transcription PCR 47 . Immunofluorescence indicated the presence of K Na 1 in the somata and dendrites of the vestibular ganglion neurons and the absence of K Na 1 in the vestibular hair cells. Third, the expression of K Na 1-encoding transcripts and currents specifically in the SGNs and not hair cells is consistent with the neuronal distribution of K Na 1 channels in the central nervous system 12 and lack of sources for Na + influx in the mature hair cells 28,29,52 . Finally, previous microarray datasets indicate increasing expression of the K Na 1-encoding transcripts Kcnt1 and Kcnt2 in FACS-sorted SGNs from E12 to P15 48 . Thus, K Na 1 currents could have been missed in earlier investigations, which more routinely examine immature SGNs 53 .
To investigate the contribution of K Na 1 channels to peripheral auditory function, we took advantage of K Na 1 DKO mice. We observed normal absolute thresholds but reduced suprathreshold wave I amplitudes in K Na 1 DKO compared to WT mice. This phenotype mimics a recently identified form of hearing loss coined hidden hearing loss that has been attributed to loss of the synapses between the sensory inner hair cells and SGNs 30 . However, our anatomical characterization indicates that DKO mice have normal numbers of SGNs and normal numbers of synapses between the SGNs and IHCs. Moreover, the molecular and cellular architecture of proteins critical for shaping SGN excitability also appears normal in SGNs from DKO mice. These findings suggest that physiological Figure 7. Na + -sensitive outward K + currents are absent in spiral ganglion neurons (SGNs) isolated from K Na 1 DKO mice. Whole cell patch clamp recordings were performed on SGNs isolated from the basal one-third of the cochlea from 6-week-old WT and DKO mice. (A) Whole-cell currents revealed a transient inward current and also outward currents in SGNs isolated from both WT and DKO mice in response to varying depolarization from a holding potential of −80 mV to 60 mV in 10-mV increments (control). Bath application of 1 µM TTX completely blocked voltage-sensitive inward currents in SGNs isolated from both WT and DKO mice (+TTX). The difference current generated by subtraction of currents recorded in the presence of TTX from those recorded before TTX application (control) revealed the TTX-sensitive outward currents in SGNs isolated from WT and DKO mice (difference, TTX-sensitive). (B) To compare findings across SGNs, the current densityvoltage relationship was generated using the steady state K + current amplitude for both WT and DKO mice. (C) The mean TTX-sensitive outward K + current density calculated at 0 mV was significantly greater in SGNs from WT compared to DKO mice. Data are plotted to show individual replicates (animals) and mean ± SEM. Values (mean ± SEM) and statistical analyses are provided in the Results. (2019) 9:2573 | https://doi.org/10.1038/s41598-019-39119-z www.nature.com/scientificreports www.nature.com/scientificreports/ rather than anatomical properties of the SGNs underly the phenotype of hidden hearing loss in these mice. For these reasons, we examined excitability of isolated SGNs using patch clamp electrophysiology.
Electrophysiological characterization of the SGNs revealed three consistent findings in SGNs isolated from both apical and basal cochlear turns. First, we observed expression of K Na 1 currents in some but not all SGNs Figure 8. Properties of evoked action potentials (APs) are altered in spiral ganglion neurons (SGNs) isolated from K Na 1 DKO mice. To examine the effects of K Na 1 on membrane properties in SGNs, current-clamp recordings were performed on SGNs isolated from the basal one-third of the cochlea from 6-week-old WT and DKO mice. (A) SGNs isolated from WT mice required greater current injection to evoke action potentials (APs) compared to SGNs isolated from DKO mice. (No DC current was injected to set the resting potential for different cells to the same value.) (B) Comparatively, APs evoked in SGNs from WT mice (black traces) were generally slower to initiate and larger in amplitude compared to those evoked in SGNs from DKO mice (grey traces). Dotted blue lines indicate −60 mV and 0 mV. (C-G) Across SGNs, the resting membrane potential (RMP) was not significantly different between WT and DKO mice (C). In contrast, significant differences in the AP latency, (D), membrane potential threshold for AP generation (E), AP amplitude (F) and AP duration (G) were observed between SGNs from WT and DKO mice. Data are plotted to show individual replicates (animals) and mean ± SEM. Values (mean ± SEM) from basal and apical SGNs are provided in Tables 3 and 4. For these data, significant differences between genotypes were determined using unpaired two-tailed t test.  Table 3. Membrane properties of SGNs isolated from the basal one-third of the cochlea from 6-week-old WT and K Na 1 DKO mice.   Table 4. Membrane properties of SGNs isolated from the apical one-third of the cochlea from 6-week-old WT and K Na 1 DKO mice.
www.nature.com/scientificreports www.nature.com/scientificreports/ isolated from WT mice. Second, we found reduced thresholds for action potential firing in SGNs isolated from K Na 1 DKO mice compared to SGNs expressing K Na 1 currents from WT mice. Third, action potential amplitudes were reduced in SGNs isolated from DKO mice compared to SGNs expressing K Na 1 currents from WT mice. Together, these three findings suggest that K Na 1 currents are normally expressed in a subset of SGNs, where they serve to increase action potential thresholds and amplitudes of those SGNs.
Although extrapolation of results obtained from SGNs in vitro to auditory nerve responses in vivo must be done cautiously, changes in the response properties of SGNs isolated from K Na 1 DKO mice may contribute to the phenotype of hidden hearing loss observed in these mice. Specifically, differences in threshold sensitivity among SGNs have been documented in a variety of mammals [54][55][56][57] . These variations in threshold sensitivity enable encoding of sound intensities over a large dynamic range: SGNs with lower thresholds contribute to auditory nerve (wave I) responses to lower sound intensities (near threshold), whereas SGNs with higher thresholds contribute to responses to higher (suprathreshold) sound intensities 31 . Thus, the reduced growth of auditory nerve (wave I) responses to suprathreshold sound intensities observed in K Na 1 DKO mice may result from the diminished contribution of a subpopulation of SGNs with higher action potential thresholds. Consistent with this reasoning, noise exposure that causes hidden hearing loss, the phenotype mirrored by K Na 1 DKO mice, appears to result specifically in loss of auditory nerve fibers with high thresholds 58,59 . Moreover, the reduced growth of auditory nerve (wave I) responses to suprathreshold sound intensities observed in K Na 1 DKO mice may also result from the reduced action potential amplitudes of a subpopulation of SGNs. Finally, we found normal expression and distribution of ion channels essential for action potential generation in the SGNs from K Na 1 DKO mice. Moreover, no other K + channels could replicate the Na + -activation of K Na 1 channels. Nevertheless, we cannot rule out the possibility that there may be quantitative differences in expression and/or functional changes in the properties of these or other ion channels that lead to the altered responses we observed in vivo and/or in vitro in response to genetic deletion of K Na 1.
This work also raises questions for future consideration about the regulation of K Na 1 channels in both SGNs and neurons more broadly. First, future studies should take advantage of SGNs to identify the sources of intracellular Na + ions required for activation of K Na 1 currents. During the voltage steps used in our experiments (<100 ms), the TTX-sensitive outward K + currents in SGNs showed little degradation. Na + influx through transient Na + channels can produce sustained activation of K Na 1 currents if the two channels are coupled intimately and local Na + concentration remains elevated. Alternatively or additionally, K Na 1 current in SGNs may be activated by Na + influx via persistent Na + channels that have been identified previously in mouse SGNs 25 . In either case, our findings suggest close coupling of the voltage-gated Na + channels and K Na 1 channels, which would allow Na + influx during the upstroke of the action potential to activate K Na 1 currents and modify properties of action potential threshold and latency in ways we observed in this report. Indeed, activation of K Na 1 currents during action potential initiation has been proposed by others 19 . Examination of the interaction of Na + influx via voltage-gated Na + channels and K Na 1 channel activation may also provide insight into the mechanisms underlying the reduced action potential amplitudes we unexpectedly observed in SGNs isolated from K Na 1 DKO mice.
Second, future experiments should utilize SGNs to examine mechanisms that regulate K Na 1 channel expression and activity. In this study, electrophysiological examination revealed K Na 1 currents in a subset of SGNs. In contrast, smFISH analyses indicates expression of K Na 1-encoding transcripts in all SGNs. Similarly, recent work using single-cell RNA sequencing to define three subtypes of (type I) SGNs, reported expression of Kcnt1 in all three SGN subtypes 60 . These findings suggest that expression of K Na 1 channels in subpopulations of SGNs is regulated post-transcriptionally. Indeed, a variety of intracellular signalling cascades and neuromodulators are known to modulate K Na 1 channel activity in other cell types, including neurons of the dorsal root ganglion 16,24,[61][62][63][64][65] . These mechanisms may give rise to the differences in K Na 1 channel activity between subpopulations of SGNs within a specific cochlear region, between apical and basal SGNs, and between SGNs and other neurons.
Third, our work highlights the need for further investigation of K Na 1 in SGNs to deepen our understanding of the molecular mechanisms that contribute to heterogeneous responses of the auditory neurons and correlated susceptibilities to excytotoxic damage. More specifically, our observation of a phenotype that mimics hidden hearing loss without synapse loss in the K Na 1 DKO mice suggests that the anatomical synaptopathy underlying hidden hearing loss may, in some cases, be preceded by physiological changes in SGN excitability. If so, identification of the molecular cascades by which the absence of K Na 1 channels exert these changes in SGN excitability and, in turn, their possible link to synapse loss and hearing loss, will inform the development of treatments that intervene in the earliest stages in the progression of hearing loss.
In conclusion, our results show that K Na 1 channels are part of a much larger repertoire of K + channels positioned to regulate the primary auditory neurons. This work provides an efficient strategy to identify, prioritize and characterize the contributions of these K + channels to function of the peripheral auditory system both in vitro and in vivo.

Methods
Animals. All experimental protocols were approved and carried out in accordance with the relevant guidelines and regulations in place at the University Medical Center Groningen (UMCG) and the University of Nevada Reno. K Na 1 double knockout (DKO) mice were bred onto a C57BL/6 background for 12 generations 19 . Because utilization of wildtype (WT) littermates were not feasible, age and gender matched C57BL/6 mice were obtained from either the C57BL/6 stock maintained at the UMCG Central Animal Facility or The Jackson Laboratory. No comparative differences between WT C57BL/6 from these two sources were observed. Nevertheless, whole genome scanning was performed to confirm strain identity and assess genetic quality between stocks. The C57BL/6 J sub-strain was confirmed via single nucleotide polymorphism (SNP)-based genome scanning (performed by Jackson Laboratories). 100% of the 150 SNP markers evenly spaced over the 19 autosomes and the X chromosome were identical in the C57BL/6 J colony maintained at the UMCG Central Animal Facility compared to the sub-strain maintained by Jackson Laboratories (data not shown).
www.nature.com/scientificreports www.nature.com/scientificreports/ RNA isolation and sequence analysis. Micro-dissection of cochlear tissue. Mice were anaesthetized with isoflurane before being sacrificed by decapitation. All mice were male and sacrificed at the same time of day to avoid hormonal and circadian variations in transcript expression between replicates. Cochleae were isolated from the temporal bones in ice-cold phosphate buffered solution (PBS). Cochlear tissue was micro-dissected, with the organ of Corti with SGNs saved separately from the lateral wall tissue (the stria vascularis with the spiral ligament). Micro-dissection was performed without decalcification or other pre-treatment (including harsh mechanical or chemical lysis), and care was taken to remove the overlying bone and associated vasculature as much as possible, including the red blood cell niche at the apex of the cochlea. Micro-dissected tissues were immediately transferred to ice-cold TRIzol reagent and processed for RNA isolation. For comparison, cerebellum, heart and liver were simultaneously collected and identically prepared.
RNA isolation. Micro-dissected tissues were homogenized in TRIzol reagent using a rotor-stator homogenizer. RNA extraction was performed using the ARCTURUS PicoPure RNA Isolation Kit with the addition of a DNase treatment. RNA quality and quantity were verified with a ThermoFisher Nanodrop. Samples with highest RNA quantity were checked for RNA quality by capillary electrophoresis using a Perkin Elmer LabChip GX. Samples with distinct 18S and 28S peaks were chosen for RNA sequencing.
RNA sequencing. RNA sequencing (RNAseq) and quality control (QC) were performed by the Genome Analysis Facility (GAF) at the UMCG. Illumina TrueSeq RNA sample preparation kits were used to generate sequence libraries using the Perkin Elmer Sciclone NGS Liquid Handler. cDNA fragment libraries were sequenced on an Illumina HiSeq2500 (single reads 1 × 50 bp) in pools of multiple samples. A total of 3 independent replicates (from 3 mice) were analyzed. The Mus musculus GRCm38 Ensembl Release 82 reference genome was used to align trimmed fastQ files with hisat. Sorting of aligned reads was performed using SAMtools. Gene level quantification was performed by HTSeq and Ensembl version 82 was used as gene annotation database. FastQC was used for QC measurements of raw sequencing data. Picard-tools calculated QC metrics for aligned reads. Sequence counts were standardized against total number of high quality reads for each sample. Because only one fragment was sequenced per transcript, length normalization was not necessary. For each gene, the mean values were generated from three replicate standardized values. Transcripts were considered present only if 2 of the 3 reads were greater than 0 RPM.

Single molecule fluorescence in situ hybridization (smFISH) with RNAscope. Preparation of cochlear sections.
Mice were anesthetized with an intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg) and then transcardially perfused with diethyl pyrocarbonate (DEPC)-treated phosphate-buffered saline (PBS) and 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer. The cochleae were harvested and immersed in a 4% paraformaldehyde (PFA) solution (DEPC-treated) overnight on a shaker at 4 °C. The cochleae were washed with PBS and decalcified in 0.35 M ethylenediaminetetraacetic acid (EDTA) for 5 days on a shaker at 4 °C and washed with PBS. Samples were cryoprotected by sequential immersion in 10%, 20%, and 30% sucrose solution at 4 °C for 1 h, 2 h, and overnight, respectively. Samples were transferred into optimal cutting temperature (OCT) compound for a minimum of 1 hr at 4 °C and then snap frozen, using a dry ice-ethanol mixture. Samples were cryo-sectioned to a thickness of 12 μm, placed onto Superfrost slides and stored at −80 °C until further use.
Probe hybridization and subsequent immunofluorescent staining. Probe hybridization closely followed the manufacturer's instructions (Advanced Cell Diagnostics). Sections were immersed in pre-chilled 4% PFA for 15 min at 4 °C. They were then dehydrated at room temperature (RT) in 50%, 70% and 100% ethanol (2X) for 5 min each and allowed to dry for 1-2 min. Fixation and dehydration was followed by protease digestion, using Protease 4 for 30 min at RT. Sections were then incubated at 40 °C with the following solutions: 1) target probe in hybridization buffer A for 3 hours; 2) preamplifier in hybridization buffer B for 30 minutes; 3) amplifier in hybridization buffer B at 40 °C for 15 minutes; and 4) label probe in hybridization buffer C for 15 minutes. After each hybridization step, slides were washed with wash buffer three times at RT. For fluorescent detection, the label probe was conjugated to Alexa Fluor 488. Probes for K + channels and a blank negative control were obtained from Advanced Cell Diagnostics. Sequences of the target probes (for the specified K + channels), preamplifier, amplifier, and label probe are proprietary. Detailed information about the probe sequences can be obtained by signing a nondisclosure agreement provided by the manufacturer.
For subsequent immunofluorescent staining, slides were treated with 10% blocking solution for 10 min at RT, incubated with anti-Tubulin β3 (TUJ1, BioLegend, 1:300 dilution), overnight at 4 °C, washed with PBS three times for 5 min each, incubated with the appropriate Alexa Fluor secondary antibody (ThermoFisher) diluted 1:500 for 2 hours at RT, and again washed with PBS three times for 5 min each. Incubation in Hoechst 33342 solution for 15 s at RT was performed to label cell nuclei. Slides were then mounted in Fluoromount-G and sealed under a coverslip.
Imaging and image analysis. Confocal micrographs were obtained as described below. Individually fluorescently labelled mRNA transcripts appeared as puncta. To quantify the number of mRNA transcripts per SGN, individually fluorescently labelled mRNAs within a given field of view (FOV) were detected using the spots function in Imaris 6.4 software (Bitplane). mRNA counts were normalized to the number of TUJ1-labeled SGNs marked manually in the same FOV.

Measurement of auditory brainstem responses.
Mice were anesthetized with an intraperitoneal injection of 75 mg/kg ketamine and 1 mg/kg dexmedetomidine and placed in an acoustic chamber. ABRs were recorded in response to both click and pure tone pips (8, 16, 32 kHz) stimuli produced with an open field speaker (2019) 9:2573 | https://doi.org/10.1038/s41598-019-39119-z www.nature.com/scientificreports www.nature.com/scientificreports/ as described previously 66 . Responses were averaged over 512 recordings. P1, N1, P2 and N2 were detected manually blind to genotype and frequency and used to calculate wave I and II amplitudes and latencies. Input-output (I/O) function slopes of the amplitude and latency growth function curves (that is, amplitude and latency as a function of stimulus intensity) were calculated as described previously [67][68][69][70] over stimulus intensities ranging from 40-90 dB SPL. I/O function slopes were only calculated when distinct positive and negative peaks could be unambiguously identified.
Histological assessment of the cochlear morphology. Mice were anesthetized with an intraperitoneal injection of 300 μg/g Avertin, (2,2,2-tribromethanol) and transcardially perfused with PBS followed by a fixative solution, containing 4% PFA and 2% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4). The cochlea was isolated, perilymphatically perfused with and then immersed in the fixative overnight at room temperature. The cochleae were post-fixed with 1% osmium tetroxide, decalcified with 120 mM EDTA at 23 °C for 48 h, and then dehydrated and embedded in an epoxy resin. Semi-thick (1 µm) sections of the cochleae were cut in the mid-modiolar plane and stained with toluidine blue for examination by light microscopy. Images (20×) were captured using a Nikon Eclipse 80i microscope. Spiral ganglion cells from the lower basal and apical segments were quantified from four to five images spaced 100 µm apart. Final figures were assembled using Adobe PhotoShop and Illustrator software (Adobe Systems).

Immunofluorescence, confocal microscopy and image analysis of isolated auditory sensory epithelia.
Mice were anaesthetized with isoflurane before being sacrificed by decapitation. Cochleae were isolated from the temporal bones in ice-cold phosphate buffered solution (PBS) and then fixed for 1 to 3 hours in a fixative solution containing 4% PFA. Auditory sensory epithelia were isolated and immunostained as described previously 35 . The primary antibodies used in this study included anti-C-terminal-binding protein 2 (CtBP2, mouse IgG1, BD Biosciences 612044), anti-glutamate A-receptor 2/3, (GluR2/3, rabbit polyclonal, Millipore AB1506), anti Na,K-ATPase α3 (ATP1A3, mouse IgG1, ThermoFisher MA3-915), anti-myelin basic protein (MBP, mouse IgG2b, Covance SMI-99), anti K V 1.1 (rabbit polyclonal, Alomone Labs APC-009), anti-tubulin J (TUJ, mouse IgG2a, Covance TUJ1), anti K V 3.1 (rabbit polyclonal, Alomone Labs APC-014), and anti-Na V 1.6 (mouse IgG1, NeuroMab 75-026) and were used diluted 1:300. The appropriate Alexa Fluor secondary antibodies (ThermoFisher) were used diluted 1:500. To determine frequency regions in isolated organs of Corti, low magnification micrographs of organs of Corti were obtained using a Leica DM4000B fluorescent microscope. If necessary, the Stitching plugin in ImageJ was used to create a single montage image. Tonotopic maps were then overlaid on the image using a specially developed plugin in Image J (https://www.masseyeandear.org/research/otolaryngology/investigators/laboratories/eaton-peabody-laboratories/epl-histology-resources/ imagej-plugin-for-cochlear-frequency-mapping-in-whole-mounts) and the previously determined place-frequency map of the mouse cochlea 71 . High magnification confocal micrographs were collected using a Leica SP8 confocal microscope with a 63× oil immersion lens under the control of the LAS X software. Z-stacks of the entire inner hair cell (IHC) synaptic pole from the 8, 16 and 32 kHz region were collected at a scan speed of 200 Hz and zoom of 1. The step size (optical section thickness) was determined by stepping at half the distance of the theoretical z-axis resolution (the Nyquist sampling frequency). Images were acquired in a 1024 × 1024 raster (x = y = 184.52 µm × 184.52 µm) at sub-saturating laser intensities for each channel. Images are presented as z-projections through the collected optical stack. All quantitative image analysis was performed on the raw image stacks, without deconvolution, filtering, or gamma correction. The number of synaptic elements per IHC were determined from 3D reconstructions generated using Imaris 6.4 software (Bitplane) as described previously 72 . Final figures were assembled using Adobe PhotoShop and Illustrator software (Adobe Systems). patch clamp electrophysiology of isolated spiral ganglion neurons. Isolation of spiral ganglion neurons (SGNs). SGNs were isolated from male and female WT and K Na 1 DKO mice as described in detail previously 53,[73][74][75] . Mice were anaesthetized and the temporal bones were removed in a solution containing Minimum Essential Medium with Hank's salt (Invitrogen), 0.2 g/L kynurenic acid, 10 mM MgCl 2 , 2% fetal bovine serum (FBS; v/v), and 6 g/L glucose. The SGN tissue was dissected and split into three equal segments: apical, middle and basal segments across the modiolar axis. Apical and basal thirds were used to obtain viable neuronal yield for experiments. Additionally, tissue was pooled from three mice into each SGN culture. The apical and basal tissues were digested separately in an enzyme mixture containing collagenase type I (1 mg/mL) and DNase (1 mg/mL) at 37 °C for 20 min. After a series of gentle trituration and centrifugation in 0.45 M sucrose, the cell pellets were reconstituted in 900 mL culture media (Neurobasal-A, supplemented with 2% B27 (v/v), 0.5 mM L-glutamine, 100 U/mL penicillin) and filtered through a 40-µm cell strainer for cell culture. SGNs were cultured for 24 to 48 h to allow detachment of Schwann cells from neuronal membrane surfaces. All electrophysiological experiments were performed at RT (21-22 °C).