K_v3 channels are high-voltage-activating type of K⁺ channels that are highly expressed in neurons with high frequency firing capability [1, 2]. One important feature distinguishing K_v3 from other voltage activating K⁺ channels is the rapid activation and deactivation in response to voltage changes. Such kinetic property of K_v3 enables rapid action potential (AP) repolarization while keeping the relative refractory period minimal. There are 4 members in K_v3 family: K_v3.1, K_v3.2, K_v3.3, and K_v3.4. Each member, apart from being regulated by different sets of kinases and signaling molecules, possess different inactivation properties. K_v3.1 and K_v3.2 exhibit little inactivation upon prolonged depolarization. In contrast, K_v3.3 and K_v3.4 show partial or rapid and near complete inactivation at membrane potential more positive to +10 mV, respectively.

Auditory neurons are well known for their capability of firing brief APs with high frequency and thus, able to faithfully follow the temporal pattern of high frequency inputs [3-5]. Not surprisingly, many auditory neurons express high level of K_v3 channels [6, 7]. Genetic or pharmacological manipulation of K_v3 channels in these neurons have significant impact on the firing pattern and ultimately the auditory function. In vitro studies have demonstrated that K⁺ channel inhibitor such as TEA reduced the MNTB neuron’s ability to follow high frequency neural input [8] and disturbed spike-timing in dorsal cochlear nucleus neurons [9]. Similarly, MNTB neurons from K_v3.1-null mice failed to fire high frequency APs [10]. K_v3.1-null mice exhibited attenuated acoustic startle response [11].

The widespread expression and in vitro impact of K_v3 channels in various auditory neurons has lead many to consider K_v3 as a potential therapeutic target for hearing deficit. In drug- or noise-induced hearing deficit animals, central auditory neurons often exhibited increased spontaneous firing rate [12-14] and K_v3 modulator suppressed the hearing deficit-associated hyperactivity. AUT00063, a K_v3 modulator causing leftward shift in voltage dependence and slowing deactivation speed, suppressed the hyperactivity in inferior colliculus neurons [15] and dorsal cochlear nucleus neurons [16] in noise-induced tinnitus animal. In addition, K_v3 modulator alleviated temporal processing deficit in animal with deafferented cochlea [17] and hyperacusis in Fragile X mouse [18].

Then, the next question is if K_v3 modulators exert any unwanted effect on other K_v3 expressing neurons including the ones in the auditory end organ. Multiple K_v3 subtypes have been found in the cochlear nerve fibers [19, 20] and therefore, it is likely that K_v3 modulators affect the neuronal excitability in the cochlea. There are 4 major groups of nerve fibers in the sound sensing area of the cochlea. Type 1 afferent, which comprise ~90~95% of cochlear afferent nerve fibers, typically make single synaptic contact with single inner hair cell (IHC) [21]. Type 2 afferent, on the other hand, receives synaptic inputs from multiple outer hair cells (OHCs) [21, 22]. Efferent nerve fibers originated from the lateral olivary complex of the brainstem (LOC efferent) contact the most distal segment of the type 1 afferents and are thought to modulate the IHC-type 1 afferent synaptic transmission [23, 24]. Another group of efferent nerve fiber contacting the OHCs are originated from the medial olivary complex (MOC efferent) and regulate the sensitivity of sound detection system [23]. Until now, most studies provided only limited amount of information on K_v3 subtypes in cochlear afferents [19, 20, 25, 26] and efferent neuronal somas in the brainstem [6, 7, 27]. The subtype and subcellular localization of K_v3 channels in each neuronal component and their precise role at the hair cell synaptic transmission are yet to be discovered. The ultimate goal is to identify the precise localization of all K_v3 family K⁺ channels within the cochlea and thereby, help assessing a therapeutic potential and adverse effect of K_v3 modulators. Here, we will focus on a subset of K_v3, namely K_v3.1b, K_v3.3, and K_v3.4, since their possible role in cochlear synaptic transmission has been anticipated from electrophysiological studies [26, 28-31].

Sprague Dawley rats (4 to 9 weeks old, either sex) were used. Ultimate aim of this study was to identify K⁺ channel subunit expression in cochlear nerve fibers of fully developed rat. At first, cochleas of 8~9-week-old rats were examined. Then, we increased our dataset by using cochlear tissues from 4~5-week-old rats. Our decision to do so can be justified because cochlear hair cell innervation pattern has already reached the adult form by the 4th postnatal week in rats and mice [32, 33]. Indeed, we did not notice any difference in the expression of various neural markers and K⁺ channel subunits between 4~5-week vs 8~9-week-old rats. In cochleas of young rats and mice the fundamental properties of cochlear hair cell synaptic transmission did not exhibit any significant sex difference [34, 35]. Again, we did not notice any qualitative sex difference in a few cases where the sex of experimental subject was identified before the euthanasia. Therefore, we pooled the results from all ages and sex groups together. All procedures were performed in accordance with the animal protocol approved by the Mokpo National University Animal Care and Use Committee. Animals were euthanized by sevoflurane overdose and cochlear tissues were isolated for experiments. Tissue fixation and subsequent immunolabeling procedures were adapted from the methods previously described [36, 37]. The cochlea was perfused with 4% ice cold paraformaldehyde prepared in PBS (pH 7.4) through oval or round window and then fixed for 1 hour at 4°C. After 3 brief rinses with PBS the cochlear coil was carefully excised and immersed into a blocking/permeabilizing buffer (PBS supplemented with 5% donkey serum and 0.25% Triton-X 100) for 1 hour at room temperature. Then, the sample was incubated with primary antibodies diluted in the blocking/permeabilizing buffer at 4°C for 20 hour. Next day, the tissue was washed 3 times (each for 20 minutes) with the blocking/permeabilizing buffer. Fluorescence-labeled secondary antibodies diluted in blocking/ permeabilizing buffer were incubated with the sample for 1 hour at room temperature. Then, the sample was rinsed once with blocking buffer for 20 minutes and twice with PBS for 10 minutes each. The tissue was mounted on a slide glass using FluorSave^® mounting medium (Calbiochem, 345789). High-resolution images were obtained using a confocal laser scanning microscope (ZEISS, LMS 710). Confocal z-stacks were collected at 0.3~0.99 μm interval. Image analyses and reconstructions were carried out using image viewing software provided by the microscope manufacturers (Zeiss Zen), Imaris (version 7.3.0, Bitplane, Switzerland) and ImageJ (NIH). No labeling was observed when the primary antibodies were omitted. The list of primary antibodies is provided in Table 1. Donkey anti-chicken secondary antibody conjugated with Alexa Fluor 647 was purchased from Millipore (AP194SA6, 1:1,000). All other secondary antibodies (conjugated with Alexa Fluor 488, 555 or 633, generated in donkey) were purchased from Molecular Probes/Invitrogen and used at 1:1,000.

K_v3.1b

K_v3.1 has 2 different splice variants, K_v3.1a and K_v3.1b. K_v3.1b appeared to be the primary isoform found in adult auditory neurons [38]. Expression of K_v3.1b isoform in the soma of cochlear afferent nerve fibers (also known as spiral ganglion neurons (SGN)) and the contribution to neuronal firing is also well established [28]. However, it is unclear whether K_v3.1b also modulates the excitatory postsynaptic potentials (EPSPs) at the IHC-type 1 afferent synaptic junction or participate only in the transmittance of action potential (AP) once large EPSP evokes AP at the heminode, the AP generation zone of type 1 afferent (corresponding to the fiber region penetrating habenular perforata in Fig. 1 diagram). Electrophysiological recording from the dendritic terminal of type 1 afferent suggested an existence of high-voltage-activating K⁺ current at the IHC-type 1 afferent synaptic junction [36]. In contrast, previous immunolabeling studies reported K_v3.1b signal at the heminode and the node of Ranvier [25, 26, 39] but not at the synaptic junction. To resolve this discrepancy, we closely examined the unmyelinated dendritic segments in addition to the cell body and the myelinated segment of the type 1 afferent. First, we compared K_v3.1b signal with that of Na⁺, K⁺ ATPase α3 (NKA), a cellular marker for the type 1 afferent and MOC efferent nerve fibers [40]. Fig. 1A shows K_v3.1b signal at the NKA-positive SGNs and nerve endings near IHCs. Higher magnification images of cell bodies (Fig.1B) and the distal segments (Fig. 1C) of the type 1 afferent fibers indicate that K_v3.1b is present at the cell body as well as the most distal segment, the presumed heminode and node of Ranvier of NKA-positive fiber.

Co-labeling of K_v3.1b and calretinin further confirmed that K_v3.1b is expressed at the IHC-type 1 afferent synaptic junction. Calretinin is a cytosolic Ca²⁺ binding protein found in the hair cells and a subset of type 1 afferent fibers [37]. K_v3.1b signal is found at the calretinin-positive dendritic terminals (Fig. 2A, yellow arrow) where they contact the calretinin-positive IHC. We also found some calretinin-negative but K_v3.1b-positive dendritic endings (Fig. 2A, white arrow) and cell bodies (Fig. 2B white arrow). In fact, type 1 afferent fibers, regardless of their calretinin content or the anatomical location of their synaptic junctions, appeared to express K_v3.1b at their endings. Previous findings by us as well as by others consistently proposed that the absence of calretinin is one of the important cellular markers differentiating cochlear afferent fibers with low spontaneous firing rate, high-threshold to sound stimuli, and higher vulnerability to trauma [37, 41, 42]. Here, we hoped to find out whether there is a similar subgroup-specificity in the expression of K_v3.1b, a K⁺ channel well known for regulating firing property of a neuron. However, K_v3.1b signal was present in all NKA-positive SGNs, suggesting that K_v3.1b essential component for type 1 cochlear afferents regardless of their sensitivity to sound or traumatic noise. Yet, one can speculate that the calretinin content might affect K_v3.1b activity to some degree via intracellular Ca²⁺-involved signaling cascade. For example, the physiologic characteristics of K_v3.1b are known to be modulated by phosphorylation at a Ser residue of its cytoplasmic C-terminal domain [2]. Lower Ca²⁺ buffering capacity in calretinin-negative neurons might result in longer intracellular Ca²⁺ transient during neuronal activation, and higher chance of Ca²⁺-dependent isoform of PKC activation, leading to K_v3.1b phosphorylation and ultimately suppression of K_v3.1b conductance.

Triple immunolabeling of K_v3.1b, NKA and Caspr-2 (the marker for node of Ranvier) demonstrated the K_v3.1b-hotspots at the heminode (Fig. 2C) and nodes of Ranvier (Fig. 2D) of NKA-positive fibers. K_v3.1b is a high-voltage-activating K⁺ channel with rapid activation and deactivation kinetics. Such kinetic property of K_v3.1b would significantly speed up the AP repolarization phase while keeping the relative refractory period minimal. Therefore, we speculate that the K_v3.1b at the heminodes and nodes of Ranvier of cochlear afferents would contribute to generating and transmitting very brief APs at high rate along the cochlear afferent nerve fibers once suprathreshold synaptic potentials are evoked at the dendritic segment. Together, these results suggest that K_v3.1b modulate IHC-driven synaptic potentials as well as subsequent AP transmittance along the type 1 afferent fibers.

K_v3.3

K_v3.3 is a high-voltage-activating K⁺ channel with N-type inactivation mechanism [2]. Expression of K_v3.3 in mouse SGN soma has been previously demonstrated by immunostaining and mRNA analysis [19]. In many auditory brainstem neurons K_v3.1b and K_v3.3 are often co-expressed and co-contribute to high frequency AP firing with minimal latency [6, 27]. Thus, we investigated whether K_v3.3 exhibited similar expression pattern as K_v3.1b or was present in a distinctive group of nerve fibers. Double-immunolabeling of K_v3.3 and NKA revealed a weak K_v3.3 signal at the unmyelinated dendritic segments (Fig. 3A) and cell bodies of NKA-positive type 1 afferent fibers. Unlike K_v3.1b, we did not detect any K_v3.3 signal at the heminode or node of Ranvier of type 1 afferent. K_v3.3 signal at the spiral ganglion area largely overlapped with parvalbumin and NFH signals (Fig. 3B), indicating K_v3.3 expression in the soma of most type 1 afferent. In contrast, none of the peripherin-positive neurons showed K_v3.3 signal (Fig. 3C). Therefore, we concluded that K_v3.3 is expressed at the unmyelinated dendritic segment and cell body of type 1 but not type 2 afferent.

K_v3.4

K_v3.4, the last member of K_v3 family, has a high-voltage-activating and N-type inactivating property [2]. The presence of K_v3.4 immunoreactivity has been reported in guinea pig SGNs [20]. However, a more thorough examination was necessary because electrophysiological evidence from rat SGN suggested K_v3.4 expression in type 2 but not type 1 afferent fibers [30]. Initial immunolabeling data with anti-K_v3.4 revealed 3 different types of nerve fiber-like structures in the organ of Corti (Fig. 4); radial fibers projecting towards OHCs, spiral fibers near IHC, and spiral fibers near OHC. The shape and trajectory of the first and the second groups of K_v3.4-positive fibers are reminiscent of MOC and LOC efferent fibers, respectively. Because both MOC and LOC efferent fibers are predominantly cholinergic [23] we tested whether the K_v3.4-positive structures expressed choline acetyltransferase (ChAT). At the same time, we compared K_v3.4 signal with NKA-positive type 1 afferent and MOC efferent fibers [40]. The K_v3.4-positive radial fibers near OHCs had bouton-like endings contacting the OHC base, and expressed both NKA and ChAT (Fig. 5A), indicating that the first group of fibers are indeed MOC efferent. In the area near IHCs, the K_v3.4-positive spiral fibers only crossed the NKA-positive fibers (Fig. 5B) and exhibited ChAT-signal (Fig. 5C), suggesting that this group of K_v3.4-positive fibers are LOC efferent.

The third group of K_v3.4-positive fibers expressed neither NKA nor ChAT. Based on their spiral trajectory near OHCs, we speculated them to be type 2 afferent. Because cell bodies and the peripheral processes of type 2 afferent fibers are known to express peripherin [43, 44] co-labeling of K_v3.4 with peripherin was performed. As expected, K_v3.4-positive neurons in spiral ganglion were NKA-negative (Fig. 6A) but peripherin-positive (Fig. 6B). Also, the K_v3.4 and peripherin signals co-localized in the spiral fibers near OHCs (Fig. 6C), indicating that this group of fibers are type 2 afferent.

K_v3.1b and K_v3.3 in auditory neurons

Coexpression of K_v3.1b and K_v3.3 had been observed in several central auditory neurons [6, 45]. Similarly, gene analysis studies by others as well as our immunolabeling data consistently show that type 1 afferent in the cochlea, the first auditory neuron in the ascending pathway, co-express K_v3.1b and K_v3.3. Quantitative RT-PCR of cultured mouse SGNs found mRNAs for K_v3.1 and K_v3.3 [19]. More recent studies using single cell RNA seq or RNAscope techniques reported K_v3.1 and K_v3.3 genes in type 1 but not in type 2 mouse SGN [41, 46]. Here, we found K_v3.1b and K_v3.3-immunoreactivities in type 1 but not type 2 SGNs. K_v3.1b and K_v3.3 are both high-voltage-activating K⁺ channels but have different inactivation properties. K_v3.1b shows only slight inactivation at positive potentials while K_v3.3 inactivates upon prolonged depolarization to +10 mV or more positive [2]. Electrophysiological recordings of auditory brainstem neurons indicated that both K_v3.1b and K_v3.3 facilitate high frequency repetitive firing [8, 27] and different K_v3 subunits in same neuron appeared to exert compensatory role although the relative contribution in different sets of neurons might vary. For example, K_v3.3 has dominant role in LSO neurons while both K_v3.1b and K_v3.3 contribute in MNTB neurons [27]. In type 1 afferents, it is still unclear whether K_v3.1b and K_v3.3 contribute equally to the high frequency firing or not. Patch clamp recordings of cultured mouse SGN have reported that dendrotoxin (selective K_v1 inhibitor; [47])-resistant high-voltage-activating K⁺ current was non-inactivating type [26, 31], supporting dominant role of K_v3.1b over K_v3.3. However, the amount of K_v3.3 mRNA was higher than that of K_v3.1 in freshly isolated mouse SGN [19]. Also, high-voltage-activating K⁺ current recorded from freshly isolated neonatal rat SGNs revealed inactivating component [29]. Therefore, further investigation using freshly isolated mature SGN is necessary to determine the relative contribution of K_v3.1b and K_v3.3 to the high-voltage-activating K⁺ current in type 1 afferents.

Previous immunolabeling studies reported K_v3.1b signal at the heminode and the node of Ranvier [25, 26, 39] but not at the unmyelinated dendritic segment of type 1 afferent. In contrast, we consistently found K_v3.1b and K_v3.3 at the unmyelinated dendritic segments of type 1 afferent (Fig. 2A). Interestingly, studies on Na_v1.6 in type 1 afferent reported similar contradictory findings. Hossain and colleagues reported Na_v1.6 expression at unmyelinated dendritic segment as well as the heminode, nodes of Ranvier of type 1 afferent in mouse cochlea [48] while others reported Na_v1.6 signal at the heminode, nodes of Ranvier but not at the dendritic segment [25, 39]. Considering the observation of occasional stronger K_v3.1b signal at the heminode than at the dendritic segment in some cochlear tissues (Fig. 2A vs. 2C), we suspect that the discrepancy might be due to an underestimation of dendritic signal juxtaposing the relatively strong signal at the heminode.

Then, what is the physiological role of the high-voltage-activating ion channels at the dendrite, away from the AP initiation zone of type 1 afferent? The IHC-type 1 afferent synaptic junction is <50 μm away from the AP initiation zone [40, 49] and therefore, the membrane potential at the synaptic junction could easily reach the voltage range of K_v3 activation during synaptic events. Intracellular recording from the dendritic terminal of type 1 afferent confirmed that significant portion of IHC-driven EPSPs were large enough to generate APs (in P7-14 rat ~18% [36], in P17-19 rat 81~97% [49]). Furthermore, the voltage dependence and other biophysical properties of K_v3 are known to be modulated by multiple mechanisms [2]. For example, dephosphorylated K_v3.1b, compared to their phosphorylated counterpart, showed significantly more negative activation voltage ranges (V_1/2max -3.92 mV vs. 16.9 mV) and larger amplitude [50]. It implies that the impact of K_v3 on IHC-type 1 afferent synaptic transmission could get even more extensive depending on the strength of various modulatory mechanisms.

K_v3.4 in type 2 afferent, MOC and LOC efferent

A presence of K_v3.4 in type 2 afferents had been initially predicted in a simultaneous patch clamp recording and dye-filling experiment on neonatal rat SGNs [30]. The OHC-innervating type 2 SGNs, compared to the IHC-contacting type 1 afferents, exhibited much more rapidly and prominently inactivating A type K⁺ current. From the kinetic properties and voltage dependence, K_v3.4 has been proposed as one of the channels responsible for the A current in type 2 SGNs. A later immunolabeling study suggested K_v3.4 expression in both type 1 and type 2 SGNs because 86~91% of guinea pig SGNs were K_v3.4-positive [20]. Recent gene analysis on mouse SGNs indicated K_v3.4 in type 2 but not in type 1 SGNs [41, 46]. In our immunolabeling result, only a small fraction of 4~9-week-old rat SGNs were K_v3.4-positive. Furthermore, K_v3.4 signal was found in peripherin-positive SGNs and nerve fibers near OHC but not in NKA-positive SGNs. Therefore, we conclude that K_v3.4 is specific for type 2 afferent at least in rats and mice.

Additionally, we found K_v3.4 in MOC and LOC efferent fiber axons in the organ of Corti. Both branches of efferent systems are known to provide cochlear protection upon noise exposure. MOC efferent fibers, by regulating electromotility of OHCs, modulate the sensitivity of ‘cochlear amplifier’ and prevent hair cell damage upon noise exposure [51-53]. Similarly, LOC efferent fibers, at least in part by modulating glutamate release from IHC, protect type 1 afferents from excitotoxic damage [24, 54]. Not surprisingly, neurotransmitter release from cochlear efferent fiber axons are modulated by multiple mechanisms [55, 56]. Metabotropic glutamate receptor, voltage gated Ca²⁺ channels and Ca²⁺-activated BK K⁺ channel at the presynaptic efferent fiber axons had been reported to contribute to the efferent synaptic transmission [56-58]. Here, based on our finding of K_v3.4 at the MOC and LOC efferent fiber axons, we propose to add K_v3.4 to the list of proteins modulating cochlear efferent neurotransmitter release. Currently available electrophysiological evidence from MOC and LOC neurons also do not reject this possibility. Patch clamp recordings from LOC neurons in mouse brain slices exhibited prominent A type current [59]. In another study MOC and LOC neurons were recorded after retrograde labeling by dye injection to the rat cochlea and both labeled MOC and LOC neurons displayed A type K⁺ current although the kinetic properties, voltage dependence, and relative sensitivity to a K⁺ channel inhibitor 4-AP differed somewhat [60]. Indeed, contribution of K_v3.4 to synaptic release has been demonstrated in other peripheral and central neurons. For example, in dorsal root ganglion nociceptors, K_v3.4 has been shown to participate in AP repolarization at the presynaptic axon terminal and consequently to modulate pain synaptic transmission [61-63]. In cerebellar stellate cells, the availability and state of K_v3.4 at synaptic boutons could be rapidly altered by the prior activity history of the neuron, which enabled rapid modulation of AP waveform at the synaptic boutons and ultimately provided flexibility to the synaptic transmission [64]. Beside the rapid voltage-dependent inactivation, K_v3.4 can also be switched from rapidly inactivating A type to non-inactivating type by modification on the N-terminal inactivation domain [2, 61]. It suggests that the prior activity history of each K_v3.4-expressing nerve fiber might modulate the status of K_v3.4 differently and therefore, provide another layer of fine tuning to neurotransmitter release.

To summarize, we have demonstrated that K_v3.1b and K_v3.3 are expressed in type 1 afferent and K_v3.4 in type 2 and efferent cochlear nerve fibers. These findings suggest that different K_v3 subtypes are likely assigned for different auditory tasks, for example, K_v3.1b and K_v3.3 in ascending signaling while K_v3.4 in protective feedback upon loud noise exposure. The data also indicates that K_v3 modulators, depending on the spectrum of action, might affect the neuronal excitability of different sets of the cochlear nerve fibers and modify different aspects of auditory processing in the cochlea.

Authors appreciate late Dr. Young-Woo Seo of Korea Basic Science Institute Gwangju Center for his technical support and critical comments during the early stage of this work.

WBK, K-WK, and KS performed the experiments and analyzed the data. WBK and EY wrote the manuscript. EY supervised the project.

This work was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2019R1I1A3A01063625 to EY).

Primary antibodies used in this study