Introduction

Mammalian cochlear hair cells (HCs) comprise a single row of inner hair cells (IHCs) and three rows of outer hair cells (OHCs). HCs transduce sound-mediated mechanical force into neural electrical codes for ear-brain intercommunication. The IHCs are the predominant afferent transducers, while the OHCs amplify low-level sound. Mammalian HCs are terminally differentiated by birth, and they are susceptible to damage by ototoxic drugs, noise-overexposure, aging, and environmental insults, resulting in hearing loss, the most common sensory deficit(Neitzel and Fligor, 2019; Rybak and Ramkumar, 2007; Wilson and Tucci, 2021; Wu et al., 2020). Emerging understanding of the mechanisms of transcription factors that induce HC differentiation (Atkinson et al., 2014; Bermingham et al., 1999; Garcia-Anoveros et al., 2022; Zheng and Gao, 2000), and potential induction for regeneration are promising, but none have produced new HCs with sustained functions(Iyer et al., 2022b; Zine et al., 2021). Thus, the current treatment for hearing loss ensuing from HC loss is cochlear implants, despite the potential advantages of HC-replacement therapy (Hinton et al., 2021; Sullivan et al., 2020).

In the mammalian cochlea, each HC is separated from the next by intervening supporting cells (SCs), forming an invariant and alternating mosaic along the cochlea’s length. Cochlear SCs can divide and trans-differentiate into HCs, serving as a potential resource for HC differentiation, using transcription and developmental signaling factors (White et al., 2006). Atoh1, a basic helix-loop-helix factor, induces SCs trans-differentiation into HCs. Upregulation of Atoh1, GFI1, and POU4F3 triggers HC differentiation, but the fledgling HCs invariably degenerate, suggesting pre-maturity (Iyer et al., 2022a; Liu et al., 2012). Additionally, inhibition of the Notch signaling or upregulation of the wnt1 pathway suffices to drive HC formation from SCs (Tang et al., 2023), but the functional features of the newly developed HC are circumpect (Mizutari et al., 2013; Waqas et al., 2016). While transcription factors are potent targets for developmental regulation, a cadre of these essential DNA-binding proteins and the precise timing of expression are required to complete the developmental cascade(Jahan et al., 2015). Thus, the key is identifying HC-developing bands of transcription and signaling factors to treat hearing loss. Recent studies identified the transcription factors INSM1 and IKZF2 as the regulators of OHC fate, while the transcription factor TBX2 specifies and maintains HC and SC fate(Kaiser et al., 2022; Li et al., 2023), advancing understanding of HC-subtype developmental specification mechanisms.

Contact-mediated lateral inhibition is among the final developmental events, where once a cell fate is determined, it inhibits neighboring cells from becoming that cell type. The HC-SC interphase is laced with tight junction proteins (TJPs), which may mediate lateral inhibition mechanisms in nonmammalian vertebrates, though their function is unclear in mammals (Chrysostomou et al., 2012; Goodyear and Richardson, 1997; Petrovic et al., 2014). In addition, TJPs were also found to regulate cell proliferation(Bhat et al., 2020; Díaz-Coránguez et al., 2019). The TJP scaffold cingulin regulates lateral inhibition in HC-SC rearrangements in the avian basilar papilla (Goodyear and Richardson, 1997), and claudin b (cldnb), an ortholog of the cldn4 in humans, upregulation controls cellular patterning during HC regeneration in zebrafish (Montalbano et al., 2021). Moreover, damaged HC extrusion and the breaking of intercellular junctional adhesions may trigger differentiation and regenerative proliferation(Corwin et al., 2007). In mammals, E-cadherin’s junctional expression negatively alters HC regenerative capacity (Burns et al., 2008; Burns et al., 2013), while Cldn9 is a positive regulator of cell proliferation(Hong et al., 2014; Zavala-Zendejas et al., 2011). Although the high expression of Cldn9 in the organ of Corti (OC) is well-established(Kudo et al., 2018; Nakano et al., 2009) (Fig. 1A-C), it is unknown whether regulation of the TJP contributes to sensory cell differentiation.

Figure 1.

To determine the roles in vivo of Cldn9, we generated doxycycline (dox)-tet-OFF-Cldn9 transgenic mice to regulate expression levels of Cldn9. The downregulation of Cldn9 resulted in functional supernumerary (ectopic) putative IHCs along the cochlear contour. Auditory neurons innervated ectopic mechanically transducing IHCs with synaptic features resembling normal IHCs. Analogous additional putative IHCs differentiation was observed when Cldn9 was knocked down using shRNA injection in postnatal (P) days (P1-7) mice, suggesting that regulation of Cldn9 levels coordinates embryonic and postnatal development differentiation of SC into IHCs. Notably, the putative ectopic (PE) IHCs at the apical and middle-frequency contour of the cochlea were preserved for over half the mouse’s life cycle (15 months), making Cldn9 a viable target for generating transformed IHCs.

Results

To control Cldn9 levels in vivo, we generated a mouse model with a site-specific genetic switch that was regulated by using dietary doxycycline (dox) and dox-containing drinking water without interfering with the typical profile of Cldn9 expression. Figures 1D-F show the design constructs and Southern blot analysis to confirm the insertion cassette in the ES cells. PCR of tail tissue samples performed genotyping. Transgene was generated on mixed C57/B6 and backcrossed into a CBA-CaJ (CBA) background after 12 generations to reduce accelerated progressive hearing loss (Peguero and Tempel, 2015; Sergeyenko et al., 2013; Willott and Erway, 1998). Results of RT-PCR from the three groups of animals, including wild-type littermates (Cldn9^+/+), heterozygote (Cldn9^+/T), and Homozygous (Cldn9^T/T) with (1mg/ml) and without dox treatment. Cldn9^T/T and Cldn9^+/Tmice demonstrate ∼55 and a 40-fold increase in Cldn9 mRNA expression in cochlear tissue (Supplement figure 1 (S1)). Treatment of Cldn9^+/T mice with dox (1 mg/ml) resulted in an ∼0.4-0.6-fold decline in mRNA levels compared to Cldn9^+/+cochleae (S1), translating to a marked difference in Cldn9 protein expression (S1). Immunoelectron microscopic analysis showed that Cldn9 levels reduced by ∼8-fold in the Cldn9^+/T cochlea (S2). The Cldn9^T/Tmice had a reduced survival rate (1.5±0.3 months (mos)) relative to the Cldn9^+/T littermates (23±2 mos). Thus, all experiments were restricted to Cldn9^+/T and Cldn9^+/+ littermates fed on dox-water (1 mg/ml). There were no recognizable differences in body weight between Cldn9^+/T and Cldn9^+/+ mice (S1). All animals were in the CBA background. Cldn9 downregulation in the Cldn9^+/T cochlea showed a qualitative decrease in the Cldn6 and an increase in ILDR1 TJP levels but no comparative differences in others(Kitajiri et al., 2004; Kitajiri and Katsuno, 2016) (S3).

Downregulation of Cldn9 induces the production of ectopic cochlear HCs

5-week-old mice Cldn9^+/T cochleae displayed a notable row of ectopic HCs (Fig. 2A-C). The ectopic HCs were observed along the cochlear contour (S4), ranged in abundance from base to apex (Fig. 2), and had contact with innervating neurons, shown in cochlear sections. A distinctive single of "U"-shaped IHC bundles was apparent in scanning electron microscopic (SEM) images (Fig. 2E). In addition to their shape, the ectopic HCs were positively labeled with anti-myosin VIIa antibody and phalloidin-labeled stereocilia bundles, which are features of typical HCs. Moreover, the ectopic HCs were negatively labeled with an anti-prestin antibody (S5), a marker for OHCs, suggesting the new HCs were likely derived from the IHC lineage. Additionally, the ectopic HCs show IHC bundle features (Fig. 2), expressed multiple CtBP2 labeling in contrast to typical OHCs (Figs. 3, 5), and reacted positively to otoferlin antibodies (Pangrsic et al., 2010; Roux et al., 2006; Strenzke et al., 2016) (data not shown). We denote the new HCs as "putative ectopic" (PE) IHCs. On average, HC counts from randomly selected 6-8-week-old cochlea from Cldn9^+/T and Cldn9^+/+ mice showed a ∼1.5-fold increase in IHCs in the Cldn9^+/T mice (Fig. 2). The PE IHCs may subserve function and be a viable alternative to the original IHC. IHC counts at different ages (P2-P21) and the cochlear frequency segments (4-32 kHz) demonstrate that the Cldn9-induced ectopic IHCs were most prominent at the cochlear apex but remained statistically significant at the base (Figs 2E-F). SEM was used for high-resolution analyses to evaluate IHC and their bundle morphology. The original and PE IHCs in Cldn9^+/T mice had normal morphology, including intact hair bundles IHCs (Fig. 2). However, the stereocilia bundle orientation was less orderly when compared to those in Cldn9^+/+ cochlea (Fig. 2D). Viable PE IHCs were identified in 15-mos old Cldn9^+/T mice (Fig. 3). Interestingly in OHCs, the numbers along the cochlear contour, apex, middle, and base, were not significantly different among the two genotypes (S6).

Figure 2.

Figure 3.

Functional features of Cldn9-induced putative ectopic IHCs resemble normal IHCs

Motor responses of both mouse genotypes’ (Cldn9^+/+ and Cldn9^+/T) to auditory stimuli (Preyer’s test) were normal. To evaluate the status of inner hair-cell function, we analyzed auditory brainstem responses to various sound-pressure levels. Cldn9^+/Tmice responses were generally indistinguishable from their Cldn9^+/+ littermate counterparts. They exhibited similar characteristic responses to broadband clicks and pure tones of 8, 16, and 32 kHz stimuli (Fig. 4A), with ∼5-15 dB threshold elevation in the Cldn9^+/T mice. The pattern of hearing threshold remained virtually constant from 2-8 months of monitoring. Moreover, Cldn9^+/+ and Cldn9^+/T mice yielded similar distortion products (Fig. 4B), suggesting normal OHC function. The apparent normal morphology of the PE IHCs led to the hypothesis the IHCs may exhibit functional mechano-electrical transducer (MET) currents. Since active MET channels are partially open at rest, the rapid uptake of the channel permeable lipophilic dye FM1-43 (Gale et al., 2001) was assessed. Results were determined from FM1-43 dye loading of apically-located original and PE IHCs of 4-week-old Cldn9^+/Tmice representing characteristic frequencies (CFs) of ∼4-6 kHz. Local perfusion of 10-μM FM1-43 resulted in intense dye labeling at the hair bundle level of original and PE IHCs. The dye-membrane partitioning and diffusion across the IHCs’ basal aspects occurred within seconds. Z-stacked-time-lapse images were taken below hair bundle level 2-sec post dye exposure and at the basolateral compartment (Fig. 4C). The time constants (τ) of dye loading at the bundle and supra-nuclear basal membrane levels for original IHCs were (data from 4 mice); 19.4±1.1 sec (n=27) and 29.0±3.8 sec (n=27), and for PE IHCs were; 23.1±4.5 sec (n=27) and 47.0±8.4 sec (n=27) (Fig. 4D). These results are consistent with functioning IHCs at rest. We conclude that PE IHC bundles are set at the optimum dynamic range to transduce MET current at rest like the original counterparts. IHC MET current magnitudes and kinetic profiles from the basilar and ectopic rows at the ∼3-4-kHz cochlear-place map in P21 mice were comparable, as summarized in figures 4E-G. The normalized current-displacement relationships were well-fitted with a two-state Boltzmann function, portraying marked similarities between the original and PE IHC MET currents (Fig. 4H). However, the displacement-response relationship for the PE IHCs was right-shifted, indicative of reduced sensitivity.

Figure 4.

Synaptic features of PE IHCs match original IHCs

To determine whether the PE IHCs had additional properties in terms of systemic functions, we examined features such as neuronal innervation in both the original and PE IHCs. We labeled auditory neurons and IHCs with calretinin (Calb2) antibodies (Sun et al., 2018). Results show Calb2-positive-subtype neurites at modiolar aspects of both IHCs (Fig. 5). The synapses between the IHCs and auditory neurons at the apical, middle, and basal cochlear locations from 5-week-old Cldn9^+/+ and Cldn9^+/T mice show substantial differences. The organization of afferent synapses was significantly different, identified as paired presynaptic-CtBP2 (red) and postsynaptic-Homer1 (green) immunopuncta. Contrasting Cldn9^+/+ from Cldn9^+/T cochlear samples, results showed reduced synaptic numbers in the Cldn9^+/T (Fig. 5B-C). Moreover, quantifying the mean number of synapses per IHC among the original and PE IHCs showed variations along the cochlear axis (Fig. 5D). This data suggests that PE IHCs are equipped to serve as afferent receptors capable of transducing and transmitting mechanical displacement into neural codes.

Figure 5.

Postnatal induction of putative ectopic IHCs by shRNA knockdown

Because pregnant mothers were fed on dox-water from gestation, the PE IHCs in Cldn9^+/T cochlea were of embryonic origin. We designed an efficient shRNA construct to knock down Cldn9 postnatally. Viral transfection in vivo, through round window injection into the cochlear tissue, was monitored with a GFP reporter gene (Fig. 6A). We first assessed the efficiency and specificity of the shRNA knockdown of Cldn9 using quantitative RT-PCR (S7) and immunofluorescence microscopy (Fig. 6A). Four days post-injection, there was an ∼20-fold reduction in Cldn9 mRNA relative to nontargeting scrambled shRNA injected cochlea (Fig. 6A, S7). Transfection in vivo of Cldn9 shRNA into P1-7 inner ears yielded cochleae with PE IHCs compared to internal controls, consisting of opposite cochlea injected with nontargeting scrambled shRNA injected, which did not display PE IHCs. By contrast, the P14-21 inner ear transfected with Cldn9-shRNA produced no detectable increase in PE IHCs as counted by three independent blinded observers (Fig. 6B-D). Ultrastructural SEM analysis of Cldn9-shRNA transfected P1-7 inner ears show PE IHC with hair bundles resembling original IHCs (Fig. 6C). The MET currents invoked from the ectopic-IHCs induced by postnatal Cldn9 knockdown were in keeping with functional HCs.

Figure 6.

The endocochlear potential and K⁺ concentrations in Cldn9^+/T mice

The cochlear duct is furnished with cellular syncytia, K⁺ channels, and transporters/pumps that operate to orchestrate a unidirectional (basal to apical) flux of K⁺ at the lateral wall to produce the endocochlear potential (EP, ∼+80 mV), an extracellular potential, subserving the proverbial powerhouse for HC functions (Von Bekesy, 1952). A remnant of K⁺ flux is a high K⁺ endolymph (∼140 mM) restricted from leakage into the basolateral aspects of HCs by TJ between HCs and SCs. At four months of age, the magnitude of the EP and the K⁺ concentration endolymph and perilymph of Cldn9^+/Tmice were variable but insignificant from age-matched littermates. A slight decline in the amplitude of the EP and a substantial rise in perilymph K⁺ was detected in 8-month-old Cldn9^+/T(S7). It is unclear whether the modest changes in the EP and K⁺ concentration of perilymph can account for the threshold increase in the Cldn9^+/Tmice.

Discussions

Fate determination is typically completed by birth in cochlear HCs, the primary receptors for mechanosensory sound detection. Generating functional HCs in the mammalian cochlea, with a proper cellular organization that allows for cochlear sound frequency selectivity, has been a demanding yet unsolved challenge. Consequently, deafness resulting from HC loss, which constitutes a significant portion of SNHL, is incurable. Multiple aspects must be considered in inducing HC differentiation since HCs have distinct functional and structural features along the cochlear axis. For optimal sensing of different sound frequencies, cochlear apical-to-basal HCs have diverse structural configurations and ion channel configurations and densities that sculpt and preserve low-to-high frequency sound processing (Kimitsuki et al., 2001). Additionally, HCs are connected to SCs by TJ proteins, enhancing the sensitivity of cochlear OHC sound amplification and maintaining high K⁺ concentration in the endolymph at the apical surface and low K⁺ concentration in the perilymph milieu surrounding HCs (Cohen-Salmon et al., 2002). This process establishes a unidirectional K⁺ flux in the cochlear duct to generate the EP, which boosts the receptor potential of HC by ∼80 mV (Hudspeth, 2008; Von Bekesy, 1952). Thus, besides overcoming the insurmountable terminal differentiation of HCs, newly differentiated HCs in vivo should be equipped with features that resemble the original primary HCs to integrate into the specialized cochlear environment.

Among the multiple approaches used with limited success in HC replacement are overexpressing transcription factors involved in prosensory cell differentiation and silencing inhibitory factors in the induction of HC fate. These methods include 1) Transfecting proneural genes, such as Atoh1, in embryonic, newborn, and damaged mature cochlear tissue. However, hair-cell-competent cells invariably lose their responsiveness post-birth and in adult animals. In the case of the damaged cochlea, it is challenging to distinguish between transdifferentiated and repaired HCs (Gubbels et al., 2008; Izumikawa et al., 2005; White et al., 2006). 2) Simulating cell division using cell-cycle inhibitors, this strategy activates apoptotic genes, leading to cell death and deafness (Lowenheim et al., 1999). 3) Inhibiting Notch signaling using a γ-secretase inhibitor to stimulate HC differentiation from potential inner ear resident stem cells after noise trauma(Jeon et al., 2011; Mizutari et al., 2013). The scarcity of resident stem cells in the adult cochlea may limit this strategy. Thus, multiple therapeutic armamentariums are required to restore hearing in the translational setting.

Previous studies demonstrated that claudin-9 is essential for hearing function and the maintenance of auditory HCs, using an ethylnitrosourea-induced Cldn9 mutant mouse model (Nakano et al., 2009), which resulted in OHC degeneration. The current results demonstrate that embryonic regulation of Cldn9 levels, but not null deletion using the dox-tet-OFF-Cldn9 transgenic strategy induces functional ectopic IHCs, but not OHCs, along the cochlear contour with increasing numbers from base to apex. These Cldn9 downregulated-induced IHCs mature, acquiring robust MET currents and neural innervation with synaptic structures markedly similar to resident IHCs. Results also revealed that postnatal downregulation of Cldn9 levels in vivo, using shRNA, suffice to coordinate SC differentiation into IHCs. Because the PE IHCs remain viable for a sizable duration of the mouse’s lifespan, the Cldn9 regulatory strategy to induce IHC differentiation subserves a feasible approach to replace lost HCs. The downregulation of Cldn9-mediated selective IHC-increase indicates Cldn9’s role during the latter phase of HC differentiation, perhaps post-OHC-fate determination(Garcia-Anoveros et al., 2022). The findings suggest Cldn9-mediated effects may be upstream of the transcriptional factor-mediated trans-differentiation of HCs since PE ectopic HCs had features of IHCs, in contrast to primordial HCs generated by Atoh1 (Yang et al., 2012). It deepens our understanding of the importance of tapping into later stages of HC differentiation that likely will result in end-organ-specific HCs.

Moreover, a pragmatic strategy requires titrated levels of the TJP to render new HCs without compromising the sensory epithelial cellular syncytial, a decline in the EP, and, significantly, a gradual extracellular K⁺ increase that mediates undue HC depolarization and death (Nakano et al., 2009). The critical period at which alteration of TJP level can induce PE and new IHCs remains unclear, although, in the current report, we have demonstrated that functional and viable mature PE IHCs can be generated by regulating Cldn9 levels.

Our findings that downregulation in the TJP, Cldn9, can regulate IHC differentiation are in conceptual agreement with reports demonstrating that lateral inhibition can affect HC specification(Lanford et al., 1999; Stone and Rubel, 1999). Similar accounts are described where newly formed HCs express delta1-like (DII1) and jagged 2 (Jag2) ligands to mediate Notch1-receptor activation in adjacent antecedent cells, thereby inducing the expression of hairy and enhancer of split (Hes1/5), which suppresses pro-HC transcription factors (Bermingham et al., 1999; Chrysostomou et al., 2012). Consistent with this scheme, interruption of Notch1 signaling during HC development leads to HC overproduction(Brooker et al., 2006; Lanford et al., 1999; Zine et al., 2000). In lateral inhibition, the foremost developing cells adopting HC fate antagonize the neighboring cells from differentiating into HCs through direct cell-to-cell communication (Brown and Groves, 2020; Mizutari et al., 2013). A possible explanation for the current findings is that TJ proteins, mainly Cldn9, are signaling in Notch-mediated lateral induction (Daudet and Lewis, 2005; Lewis, 1998). Canonical Notch signaling is activated when a Notch ligand, such as Delta-like1 (DI1) in an adjacent HC, binds to a receptor in the SC, resulting in the release of the intracellular domain of the Notch receptor (NICD), which translocates to the nucleus to activate the transcription of Notch target genes. The findings also confirm that the Notch signaling pathway is responsible for homeostatic TJP expression in vitro and promotes barrier function in vivo in the RAG1-adoptive transfer model of colitis (Mathern et al., 2014). Indeed, occluding junction depletion disrupts Notch and mitogen-activated protein kinase (MAPK) signaling in intestinal tissue (Fairchild et al., 2016). In the scheme provided in Figure 7, Cldn9 subserves the signaling catalyst to activate NICD cascades that suppress neighboring SCs from trans-differentiation. A limitation of the model is that if Cldn9-induced effects were solely dependent on the Notch signaling, putative ectopic OHCs and IHCs would have ensued. Future studies and emerging findings on HC differentiation will likely address these shortcomings (Kaiser et al., 2022; Li et al., 2023). In cochlear tissue, downregulation of Cldn9 led to concomitant reduced expression of Cldn6 and increased ILDR1. It is unclear whether the induction of PE IHCs resulted from reduced expression of Cldn9 alone or combined TJP alterations. It is conceivable that targeting a different TJP may have similar effects on OHC differentiation, requiring impending studies.

Figure 7.

Materials and Methods

All procedures were performed under research guidelines of the institutional animal care and use committee of the University of Nevada, Reno. Mice of either sex were studied. Doxycycline (dox)-tet-OFF-Cldn9 transgenic mice were generated. In the mouse line, dox concentration can regulate the level of Cldn9 gene expression. The construct consisted of a tetracycline-based genetic switch (tTA cassette) made of three main modules: 1) The tetracycline-controlled transcriptional activator (tTA); 2) The neomycin resistance gene flanked by LoxP sites; and 3) Six copies of the tet operator (tetO) fused to the minimal CMV promoter. The tTA cassette was inserted at the −110 -nucleotide position upstream of the translational start of Cldn9 to generate the targeting vector. The targeting vector was electroporated into B6 mouse embryonic stem cells. Following the selection in G418, DNA samples from the neomycin-resistant ES cell clones were prepared for short-arm PCR/sequencing analysis and Southern blot analysis to confirm the insertion of the tTA cassette into ES cells. Genetically modified ES cells containing one copy of the tTA cassette were injected into healthy albino B6 blastocysts, and the injected blastocysts were transplanted into the uterus of an albino B6 mouse to generate the chimeric mouse. The chimeric mouse was then bred with albino B6 mice to produce the F1 heterozygous mouse, and the germline transmission was confirmed by tail DNA genotyping. Deletion of the selection marker in the tTA cassette by crossing the F1 mouse with the embryonic Cre line (B6.129S4-Meox2^tm1(cre)Sor/J). We backcrossed the B6/129S4 background unto the CBA/CaJ mouse background for 12 generations to prevent the masking of age-related hearing loss effects. 1.0 mg/ml of dox water were fed to Cldn9 breeding pairs from breeding day one, and heterozygote (Cldn9^+/T) and Homozygous (Cldn9^T/T) mice and wild-type littermates (Cldn9^+/+), through the time until for the sample collections. The body weights of mice were recorded. Genotyping was performed by PCR using a set of primers that flank the knockin in the Cldn9 gene: forward primer Cldn9 knockin-F (knockin-sequence): 5’–ATCCACGCTGTTTTGACCTC–3’, Cldn9 R3 (Reverse): 5’– TCTGGACCACACAGGACATC– 3’. PCR fragments were separated with 2% agarose gel for an 800 bp product in wild-type, 365 bp in homozygous mutants, and two in heterozygous littermate mice (Fig. 1).

Auditory brainstem recordings (ABR) and Distortion product otoacoustic emissions (DPOAE) measurements

Cldn9 mice (Cldn9^+/+, Cldn9^+/T, and Cldn9^T/T) littermates were tested at 2-8 months (mos) of age. Mice were anesthetized with ketamine and xylazine by intraperitoneal (IP) injection (25 mg/kg). Body temperature was monitored using a rectal probe and maintained at 36.8±1.0°C using a homeothermic device (Harvard Apparatus). ABR and DPOAE measurements were described previously (Dou et al., 2004). For ABR assays, thresholds were obtained by presenting tone bursts at 4, 8, 16, and 32 kHz and a clicking sound from 0 dB to 90 dB sound pressure levels (SPL) in 5 dB intervals. Tones were 2.5 ms, while click was 0.1 ms in duration, with a repetition rate of 21/s. Electrodes were placed subdermally behind the tested ear (reference), the vertex (active), and the back (ground). Evoked potentials were averaged over 512 repetitions and collected using a Tucker Davis Technology (TDT) RZ6 processor and BioSigRZ software. The threshold was defined as the lowest intensity of stimulation that yielded a repeatable waveform based on an identifiable ABR wave.

DPOAE measurements were performed using the same TDT system with two calibrated MF1 speakers connected to an ER10B+ microphone. Data were collected every 21 ms and averaged 512 times. DPOAEs were recorded using two pure tones with frequencies f1 and f2, using an f2/f1 ratio 1.2. Input/output (I/O) functions were obtained by increasing the primary tone L1 (and corresponding L2) in 5-dB steps from 20 to 80 dB SPL at 8, 16, and 32 kHz frequencies. During DPOAE testing, the probe assembly was placed in the mouse’s left ear canal after visual inspection to ensure no ear infection or inflammation of the tympanic membrane. DPOAE thresholds were defined as the lowest level of f1 required to produce a DPOAE ≥ −5 dB SPL(Chen et al., 2021).

Cochlear mapping and hair cells and synaptic counts

The cochlea was micro-dissected into three to five pieces following the method described (Montgomery and Cox, 2016). Cochlear pieces were measured, and a frequency map was computed based on a 3D reconstruction of the sensory epithelium for HCs and synapse count of associated structures to relevant frequency regions using a custom plug-in to ImageJ (Muniak et al., 2013). Confocal z-stacks of the 4, 8, 16, and 32 kHz areas were collected using a Leica Stellaris8 (Leica) and Nikon A1R laser scanning confocal microscope (Nikon Instruments Inc.). Images were gathered in a 512 x 512 raster using a high-resolution oil immersion objective (60x). IHCs and OHCs at the frequency locations were quantified using myosin-VIIa-positive as an HC-marker within a 70-100-μm field (Chen et al., 2021). Synaptic ribbons could be counted manually using 3D (x-y-z axis) representations of each confocal z-stack with the microscopic image analysis software Imaris (Oxford Instruments, USA).

Inner ear histological analysis

The cochleae were intra-labyrinthine perfused through the oval and round windows with 4% paraformaldehyde (PFA). The samples were decalcified in 10% EDTA up to 72-96 hrs, depending on the age, at 4°C. Microdissected pieces were immunostained with antibodies to the following: (1) mouse anti-C-terminal binding protein 2 (pre-synaptic-marker, BD Biosciences, 1:200, Cat # 612044), (2) rabbit anti-myosin-VIIa (HC-marker, Proteus Biosciences, Inc,1:600, Cat # 25-6790), (3) mouse anti-sox2 (supporting cell (SC)-marker, Santa Cruz Biotechnology, Inc, 1:200, Cat # sc-365823), and (4) rabbit anti-Homer 1 (post-synaptic marker, Synaptic Systems, 1:250, Cat # 160 003), (5) rabbit anti-immunoglobulin like domain containing receptor 1 (ILDR1) (Antibodies-online.com, 1:200, Cat # ABIN1386369), (6) mouse anti-Cldn9 (Santa Cruz, 1:200, Cat # sc-398836), (7) mouse anti-Cldn6 (Santa Cruz, 1:200, Cat # sc-393671), (8) rabbit anti-Sox2 (Abcam, 1:200, Cat # ab97959), (9) rabbit anti-Tuj1 and chicken anti-Tuj1 (Abcam,1:500, Cat # ab18207, ab41489), (10) goat anti-calretinin (Swant Inc., 1:500, code # CG1), (11) mouse anti-calretinin (Millipore Sigma, 1:200, Cat # MAB1568), (12) rabbit anti-Prestin (Santa Cruz, 1:200, Cat # sc-22692), and (13) rabbit anti-calbindin (Cell signaling technology, 1:200, Cat # 13176S) with appropriate secondary antibodies coupled to Alexa-405, −488, −568, and −647 fluorophores.. DAPI labeled the cell nucleus after secondary antibody incubation. Samples were stained with phalloidin and mounted with Fluoro-Gel (Electron Microscopy Sciences). Images were captured under a confocal microscope.

RNA extraction and quantitative RT-PCR of Cochlear tissue

The cochleae were dissected from the mouse and homogenized on ice. Because of limited tissue, we combined 10-15 mice cochleae for the study. Total RNA was isolated using the RNeasy Plus Mini Kit (Qiagen), and cDNA was generated using the RT2 First Strand Kit (Qiagen). cDNA was combined with RT2 SYBR Green Master Mix (Qiagen), specific qRT-PCR primers, and qRT-PCR analysis was run using the ViiaTM 7 Real-Time PCR System (ABI). Primer efficiencies were determined by standard dilution curve analysis. Three separate samples were used from 10 animals for each group. The experiments from each sample were performed in triplicate, and average cycle threshold (Ct) values were normalized to GAPDH expression. ΔΔCt values were determined relative to Cldn9^+/+ cochlear samples. Fold change was defined as 2(−ΔΔCt). Primers used include Gapdh (SA Biosciences) and Cldn9 (ThermoFisher).

Electron Microscopy

Transmission electron microscopy (TEM) and scanning electron microscopy (SEM) of cochlear sensory epithelia were performed as described (Schwander et al., 2007; Senften et al., 2006). Five to eight-week-old Cldn9 mice and littermates were sacrificed for TEM, and the cochleae were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer at 4°C overnight. After several washes with buffer alone, cochleae were fixed in 1% osmium tetroxide at RT for 1-hr. After that, the fixed cochleae were decalcified in 10% EDTA for 3–4 days. Fixed and decalcified cochleae were dehydrated using a graded ethanol series and embedded in epoxy resin. Ultrathin sections were cut with a diamond knife. Specimens were examined using an electron microscope.

For SEM, mice were perfused with 4% PFA in 1x PBS, inner ears were isolated, and the stapes footplate was removed. Ears were flushed and fixed overnight in 4% PFA and 2.5% glutaraldehyde in 1x PBS. After washing in ddH₂O 3X for 1 hour, the samples were post-fixed with 1% osmium tetroxide for approximately 1 hour. Samples were washed before decalcifying for 3-4 days in 0.25 M EDTA at 4°C with daily solution changes. The cochleae were microdissected, the tectorial membranes removed, and gradually dehydrated in 30%, 50%, 70%, 80%, 90%, 100% ethanol, 2:1 ethanol/hexamethyldisilazane (HMDS, Thermo Scientific #A15139.AE)), 1:2 ethanol/HMDS, and finally 100% HMDS. Samples were transferred to an open well plate in HMDS and allowed to air dry overnight in a fume hood. They were then mounted on aluminum stubs (Ted Pella #16111) using double-sided carbon tape (EMS #77817-12) and stored in a specimen mount holder (EMS #76510) sealed in a desiccator until sputter-coated with Au/Pd (Emitech Sputter Coater K550) and viewed. Images were captured utilizing the Hitachi S-4800 SEM. An accelerating voltage of 1kV and 5kV was used. Images were compiled using CorelDRAW X7 graphic suite software.

Measurements of the endocochlear potential (EP) and K⁺ concentration

Cldn9 heterozygote mice and wild-type littermates were anesthetized using ketamine and xylazine (100/25 mg/kg, i.p.) and K⁺ concentration, and the EP was measured using double-barreled microelectrodes. An incision was made along the midline of the neck, and soft tissues were bluntly dissected laterally to expose the trachea and the animal’s left bulla. A tracheostomy was made, and the musculature over the bulla was cut posteriorly to expose the bone. A small hole was made in the cochlear capsule directly over the scala media of the lower basal turn. The EP electrode was filled with 300 mM NaCl, the K⁺-selective barrel was silanized, and the tip was filled with a liquid ion exchanger (Fluka 60398, K⁺ ionophore I-Cocktail B) that was backfilled with 150 mM KCl. A round-window approach made measurements in the basal turn of the cochlea through the basilar membrane of the first turn. The K⁺-selective electrode was calibrated in solutions with known cation (K⁺ and Na⁺) concentrations in situ at 37°C.

shRNA CLDN9 knockdown

siRNAs were designed, using siRNA at WHITEHEAD software, and cloned under U6 promoter in pSilencer5.1-U6 vector to produce hairpin siRNAs (shRNAs; Vector Biolabs, NM_020293). shRNA was packaged into adeno-associated virus constructs: AAV/Anc80L65-GFP-U6-mCLDN9-shRNA(4) at 1.0X10¹² GC/mL titer. shRNA sequence for mouse CLDN9 is 5’-CACC GTGCTTCGGGACTGGATAAGACTCGAGTCTTATCCAGTCCCGAAGCAC TTTTT-3’. The most efficient shRNA 1.0 µl round window injection reduced cochlear RNA by a ∼20-fold expression level compared with scrambled shRNA injection. A round window injection was carried out in mice at P1 – P15. Mice were anesthetized by induced hypothermia and kept on a cold on a cold surface during the injection procedure. After disinfecting the skin with 70% ethanol and Povidone iodine, an incision was made in only the left ear (experiment) and right ear (scramble shRNA injection). Underlying fat and soft tissue were carefully dissected to expose the round window of the cochlea. The glass pipette was pulled with a P-2000 (Sutter Instrument, Novato, CA) and sharpened with a BV-10 Micropipette Beveler (Sutter Instrument, Novato, CA). The shRNA was injected using a nanoinjector (Sutter Instrument, Novato, CA). After all the shRNA was injected, the pipette was left in place for 30 seconds before removal. The muscles and fat tissue were covered, and the skin was closed with a polypropylene suture. Before returning the pups to the home cage, the mice were put on the worm bedding heated by a heating pad for recovery before returning to their mother. Total surgery time did not exceed 15 min.

Voltage-clamp recording of hair cell mechanoelectrical transducer (MET) current

Patch-clamp experiments were performed in the standard whole-cell mode using an Axopatch 200B amplifier (Axon Instruments). Patch electrodes were pulled with a horizontal puller (Sutter Ins. Navato, CA) and had a resistance of 2–3 MΩ when filled with pipette solution consisting of (in mM) 135 CsCl, 10 HEPES, 2.5 EGTA, 0.25 CaCl₂, MgCl₂, 4 MgATP, and 0.4 Na₂GTP (pH adjusted to 7.3 with CsOH). The bath solution consisted of (in mM) 130 NaCl, 3 KCl, 1 MgCl₂, 10 HEPES, 2.5 CaCl₂, and 10 glucose (pH was adjusted to 7.3 using NaOH). Currents were sampled at 20 kHz and filtered at 2 kHz. Voltages were not corrected for a liquid junction potential. No leak current subtraction was performed. Cells were held at −80 mV. All electrophysiological experiments were performed at RT (21-22°C). We used stepwise and sinewave mechanical stimulation of IHC bundles through a piezo-driven fluid-jet stimulator to record IHC mechanoelectrical transducer (MET) currents. We represented the bundle displacement in the form of applied piezo-driven voltage. Bundle displacement was not calibrated for each cell because of variations in stimulating probe positions relative to the stimulated hair bundle.

Data analysis

The ABR and DPOAE data were analyzed using GraphPad Prism 7 (GraphPad Software, San Diego, CA, US) and OriginPro 2020 (OriginLab Corp., Northampton, Mass, US). Two-way ANOVA was used to analyze threshold and amplitude data. One-way ANOVA was used for the pre-synapse count and HC count. Significance was assumed at a p-value of 0.05 in all statistical analyses.

Acknowledgements

We thank members of our laboratory for their comments on this manuscript. This work was supported by grants to E.N.Y. from the National Institutes of Health (DC016099, DC05135, AG051443, DC015252, and AG060504).

Author Contributions

E.N.Y and BF designed the research, Y.C. M.C.P.F., J.H.L, J.L, S.P, M.K, B.P, J.K, J.C, L.L, MAG, and E.N.Y., analyzed data and wrote the manuscript. All authors read and approved the final manuscript.

Author Information

Correspondence should be addressed to E.N.Y., Program in Communication Science, Department of Physiology and Cell Biology, School of Medicine, University of Nevada Reno, NV. Email: enyamoah@gmail.com.