Unlocking TMPRSS3 Hearing Loss Mechanism

For the majority of genes associated with hearing loss (HL), the underlying pathologic mechanism is not known. Determining the precise molecular mechanism underlying each form of genetic HL is critical for development of new treatment modalities, which currently are limited to hearing aids and cochlear implants but in the near future may include correction of underlying pathology with molecular or gene therapies. One gene for which the underlying disease mechanism is still elusive is TMPRSS3.

Biallelic pathogenic variants in TMPRSS3 cause 2 forms of sensorineural HL: early-onset severe to profound HL (DFNB10) and postlingual, progressive, high-frequency HL (DFNB8) (1). TMPRSS3 is commonly implicated as a cause of human HL (2). We and others have previously shown that, unlike other forms of genetic HL, TMPRSS3-related HL is associated with variable speech perception outcomes after cochlear implantation in adults and is correlated with duration of deafness, possibly due to reduced spiral ganglion neuron (SGN) function (3–5). In contrast, in children with TMPRSS3-related HL, cochlear implant outcomes are more positive (6). Given its relatively substantial contribution to genetic HL and the variable cochlear implant outcomes associated with TMPRSS3-related HL, determination of the function of TMPRSS3 and how its dysfunction results in HL is critical for a comprehensive understanding of DFNB8/10.

The TMPRSS3 gene encodes the transmembrane serine protease III protein, a widely expressed protein with multiple tissue-specific isoforms (7). To date, there are more than 70 reported pathogenic HL-associated variants in TMPRSS3 (8, 9). In the inner ear, TMPRSS3 is expressed in inner hair cells (IHCs), outer hair cells (OHCs), and in type II, but not type I, SGNs (7, 10, 11). The previously developed Tmprss3-null mouse model (Tmprss3Y260X/Y260X) shows rapid degeneration of cochlear hair cells (HCs) beginning at the onset of hearing at P12, leading to compete loss of HCs by P14 and resulting in profound deafness (7, 11). HCs s develop normally, but they quickly undergo cell death at the onset of hearing, a unique pathology compared with other forms of genetic HL.

The mechanism for HC death at the onset of hearing in Tmprss3Y260X/Y260X mice is not understood. However, due to the coincident timing of normal development of the endocochlear potential (EP) (12) and rapid HC degeneration in Tmprss3Y260X/Y260X mice, we hypothesized that EP may play a role in HC loss and subsequent deafness in Tmprss3Y260X/Y260X mice. The apical surface of cochlear HCs is bathed in endolymph, which is high in extracellular K+ and provides the +90 mV EP. Interestingly, the period at which massive HC death occurs in Tmprss3Y260X/Y260X mice coincides with development of the EP, because there is a rapid increase from +16 mV to +60 mV, reaching the mature +90 mV by P15 (12). The EP is generated by the cells of the stria vascularis and provides a steep electrochemical gradient that drives sensory transduction current into HCs, carried primarily by K+ and Ca2+ (13). One route for K+ efflux from HCs is via big potassium (BK) channels, which are large-conductance, calcium-activated K+ channels. KCNMA1 is the gene that produces the pore-forming α subunit of BK channels, which localize on the neck of mature IHCs (11, 14). Normally, Kcnma1 is first expressed at P12; however, KCNMA1 channels are not detected in IHCs of Tmprss3Y260X/Y260X mice (11, 15). Interestingly, KCNMA1-null mice have normal hearing at 4 weeks of age and develop a mild, progressive, high-frequency HL based on alterations of OHC function (16). Based upon the temporal correlation between rapid HC death with EP maturation, along with alterations in HC K+ channel expression, we hypothesize that alterations of EP contribute to HC death and HL in Tmprss3Y260X/Y260X mice.

Here, we investigated this hypothesis by modifying the intracochlear environment and assessing effects on HC survival in Tmprss3Y260X/Y260X mice both in vitro, using a cochlear explant model, and in vivo, crossing Tmprss3Y260X/Y260X mice with 2 mouse strains that lack development of EP. We also measured EP directly in Tmprss3Y260X/Y260X mice and show that EP aberration is associated with the rapid HC death and HL in these mice. In addition, pharmacologic reduction of the EP with systemic administration of furosemide led to reduction in HC death, providing a new possible avenue for treatment of this common cause of human HL.

HCs show normal early development and function in Tmprss3Y260X/Y260X mice. We evaluated the morphology and function of HCs in early stages of postnatal development in Tmprss3Y260X/Y260X mice prior to the onset of hearing and before the previously identified rapid and complete HC degeneration of IHCs and OHCs from P12 to P14. At P6, the HCs showed typical stereocilia morphology and cellular arrangement via phalloidin immunostaining (Figure 1A). The hair bundle ultrastructure of Tmprss3Y260X/Y260X mice is indistinguishable from that of heterozygous and wild-type littermates. Mechanosensory transduction currents in mouse HCs are detected at P0 at the base and reach mature levels between P5 and P10 in a tonotopic gradient from base to apex (17, 18). We noted typical HC uptake of FM1-43, a styryl dye that specifically labels HCs via uptake through functional sensory transduction channels, at P7 (Figure 1B). Consistent with prior reports, we found that Tmprss3Y260X/Y260X IHCs and OHCs undergo rapid and complete degeneration by P14 (Figure 1A). These data indicate normal early development and function of HCs in Tmprss3Y260X/Y260X mice prior to rapid degeneration.

Tmprss3Y260X/Y260X mice demonstrate normal HC and stereocilia morphology and physiology prior to rapid-onset degeneration. (A) Phalloidin immunostaining and scanning electron microscopy (SEM) show normal architecture of IHCs and OHCs at P6 and P14 in wild-type mice and complete HC degeneration at P14 in Tmprss3Y260X/Y260X mice. Scale bars: 10 μm (P7), 20 μm (P14). (B) FM1-43 uptake by HCs in Tmprss3Y260X/Y260X mice is equivalent to that of wild-type mice at P7. Scale bars: 20 μm.

SGN patterning is normal at early stages in Tmprss3Y260X/Y260X mice. SGNs innervate IHCs and OHCs and consist of type I and type II SGNs. Type I SGNs compose 95% of SGNs, transmit sound information from the IHCs to the central nervous system, and consist of 3 unique subtypes (IA, IB, and IC) based upon single cell transcriptomic profiling (10, 19) and spontaneous firing rates (20). Type II SGNs make up the remaining 5% of the SGN population, synapse on multiple OHCs, and are not essential for sound transmission; instead, they are likely involved in damage perception (21). SGNs undergo major refinement in molecular phenotype in the first few postnatal weeks (19). Proper segregation of type I and type II SGNs and subtype specification of type I SGNs is dependent upon the spontaneous activity of HCs and functional sensory transduction prior to the onset of hearing (19). Genetically engineered mice with defective synaptic transmission or sensory transduction have severe alterations in the type 1 SGN subtyping (19). TMPRSS3 expression is not detected in type I SGNs and is limited to only type II SGNs (10). To test if loss of TMPRSS3 affected SGN function, we profiled the composition of SGNs in Tmprss3Y260X/Y260X mice at early time points (Figure 2). The total number of Tuj1-positive SGNs was no different between Tmprss3Y260X/Y260X and control mice (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/JCI186395DS1). Staining with antibodies specific for type IA (calretinin), type IB (calbindin), type IC (BRN3A), and type II (nerve growth factor receptor [NGFR] SGNs revealed identical subtype composition between the control mice (Tmprss3Y260X/+) and Tmprss3Y260X/Y260X mice at P11 (Figure 2, A and B). Despite HC degeneration by P14 in Tmprss3Y260X/Y260X mice, no significant differences in type I SGNs were observed at P21. The type II SGNs were slightly, though significantly, decreased in Tmprss3Y260X/Y260X mice at P21. These data demonstrate that loss of expression of TMPRSS3 does not affect activity-dependent subtype patterning of type 1 SGNs and provides further support for the indication that loss of TMPRSS3 does not disrupt HC transduction, which is critical to IHC development.

Spiral ganglion subtype composition is normal in Tmprss3Y260X/Y260X mice, aside from type II SGN increase. (A) Representative sections of spiral ganglion taken at P11 stained for neuron-specific class III β-tubulin (TuJ1) and antibodies specific for type IA SGNs (calretinin), type IB SGNs (calbindin), type IC SGNs (POU4F1, BRN3A), and type II SGNs (NGFR). Top row shows TuJ1 channel, middle row shows SGN subtype-specific channel, and the bottom row shows merged channels. Arrows indicate costaining; arrowheads indicate negative costaining. Scale bars: 20 μm. (B) Counted subtype-specific neurons at P11 and P21 for Tmprss3Y260X/Y260X and wild-type mice show no difference aside from increased type II SGNs in wild-type mice at P21. Each point represents cell count from 1 cochlea (n = 3 cochleae per genotype and day).

HC death in Tmprss3Y260X/Y260X mice is mediated by the intracochlear environment. The endolymph within the scala media comprises extracellular fluid containing 157 mM K+ that is dependent on the KCNJ10 channel of the lateral wall of the cochlea (22). The positive EP within the scala media of 80–100 mV is generated by the stria vascularis and reaches peak levels at P14 at the onset of hearing (23).

We had previously shown that inner ear organoids derived from stem cells of Tmprss3Y260X/Y260X mice are morphologically and functionally identical to wild-type inner ear organoids and do not undergo the same rapid degeneration seen in vivo (24). Based on prior observations of sudden IHC and OHC death in Tmprss3Y260X/Y260X mice between P12 and P14 (11, 15), we hypothesized that the extracellular environment — the EP, in particular — may contribute to the early rapid HC death in Tmprss3Y260X/Y260X mice. To test this hypothesis and investigate if altering the extracellular environment could prevent HC death, we performed cochlear explant cultures. Mouse cochleae were microdissected at P7 and the sensory epithelium was placed in explant culture media for an additional 7 days in vitro (DIV). We observed complete preservation of HCs in the Tmprss3Y260X/Y260X mice (Figure 3).

Cochlear explant cultures implicate intracochlear environment in HC death in Tmprss3Y260X/Y260X mice. Representative images of cochlear explant cultures from control (Tmprss3Y260X/+) (A and B) and Tmprss3Y260X/Y260X (E and F) mice. Tmprss3Y260X/Y260X explant at P14 equivalent (P7 explant + 7 days in vitro [DIV]) demonstrates no HC degeneration of either IHCs or OHCs, which is maintained to P30 explant (C and G). Tmprss3Y260X/Y260X mice display near complete HC degeneration at P14 in vivo (D and H). Scale bar: 20 μm. (I and J) Quantification of IHCs and OHCs in explant culture for control and Tmprss3Y260X/Y260X mice demonstrating slowly progressive OHC loss in both. There was no significant difference in cell counts up to 30 DIV (D30) (P > 0.11 in all cases). Each point represents the cell count from 1 culture. Exact Wilcoxon rank-sum test (Mann-Whitney U test) was used to compare 2 groups (Tmprss3Y260X/+ and Tmprss3Y260X/Y260X) in each culture age. Up to 5 cultures were quantified for each day, but only 2 cultures were available for D30 control (see Supporting Data Values for full details).

To determine the length of survival in culture, we cultured P7 Tmprss3Y260X/Y260X explants for up to 30 days (Figure 3, A–I). Although we observed some loss of OHCs with progressively longer culture times, there was no statistically significant difference in HC survival between Tmprss3Y260X/Y260X explants and control Tmprss3Y260X/+ explants (Figure 3J; n = 3–5 cultures per condition). This supported our hypothesis that the extracochlear environment causes massive HC death in Tmprss3Y260X/Y260X mice.

EP measurement in mouse models. Tmprss3 transcript expression is detected in the rare spindle root cells of the lateral wall and not within the major stria vascularis cells such as the marginal, intermediate, or basal cells (25). To determine if TMPRSS3 expression alters EP, we first performed EP measurement on adult mice at P28. We observed no difference in mature EP between Tmprss3Y260X/Y260X and control mice regardless of sex (Figure 4, A and B).

Direct EP measurements show supraphysiologic rise in EP in Tmprss3Y260X/Y260X mice. (A) EP was recorded through the basal turn of left cochleae in live mice ranging from P7 to P24. Representative tracing of EP recorded from the scala media of a live control mouse. (B) Direct EP recordings from P28 mice demonstrating no difference between control (n = 5) (Tmprss3+/–) and Tmprss3Y260X/Y260X (n = 6) mice. Open circles and squares represent female mice. Filled circles and squares represent male mice. nd, no statistical difference determined. (C) Direct EP recordings from mouse models. Each data point represents 1 reading from 1 animal and is plotted as the mean with SD. Mean EP at P12–P15 is significantly different (P < 0.0001) between Tmprss3Y260X/Y260X and control (Tmprss3Y260X/+) mice. Wilcoxon rank-sum test (Mann-Whitney U test) was used to compare 2 groups (control Tmprss3Y260X/+ and experimental Tmprss3Y260X/Y260X) in each postnatal age group (P7–11: n = 12, n = 11; P12–15: n = 15, n = 11; P18–P24: n = 10, n = 9, for control and experimental mice). (D) Schematic demonstrating time frame of HC death in Tmprss3Y260X/Y260X mice in the context of physiologic rapid phase of EP development. Neither Pou3f4delJ nor MitfMi-wh/+ knockout mice generated EP and neither had HC degeneration despite profound HL.

Next, we sought to evaluate EP dynamics in Tmprss3Y260X/Y260X mice during the critical period of EP development from P7 to P24. Similar to our adult measurements, we directly measured EP in live control and Tmprss3Y260X/Y260X mice, using glass micropipettes inserted into the scala media. The EP was recorded as the voltage difference between the micropipette’s stable positions in the scala media and the scala tympani (Figure 4A).

We found that the EP was, on average, 8.9 mV higher at ages P7–P11 in Tmprss3Y260X/Y260X mice compared with Tmprss3Y260X/+ mice (48.9 ± 16.4 mV vs. 40.0 ± 10.0 mV, respectively) (Figure 4C). This difference was not statistically different (P = 0.26), as measured by a Wilcoxon rank-sum test (Mann-Whitney U test). The EP was 30 mV higher in Tmprss3Y260X/Y260X mice than in controls when tested at P12–P15 (88.3 ± 8.9 mV vs. 58.2 ± 12.9 mV, respectively; P < 0.0001) (Figure 4C).

After the onset of hearing, the EP stabilized at approximately 90–95 mV at P18–P21 and was not significantly different between Tmprss3Y260X/Y260X and control mice (89.2 ± 10.9 mV vs. 94.7 ± 5.6 mV, respectively). Thus, the premature rise in EP noted here in the Tmprss3Y260X/Y260X mice prior to P16 occurred concurrently with previously identified complete HC degeneration. The rapid increase in EP from P10 through the onset of hearing at P14 is temporally correlated with the rapid HC degeneration between P12 and P14 seen in Tmprss3Y260X/Y260X mice (Figure 4D). This early increase in EP may be a secondary consequence of an as-yet unidentified role of TMPRSS3 in the lateral wall, the lack of KCNMA1 expression in the Tmprss3Y260X/Y260X HCs, or a consequence of HC degeneration.

Reduction of EP in vivo prevents HC death. To further investigate the role of the EP on TMPRSS3-mediated HC death in vivo, we crossed the Tmprss3Y260X/Y260X line with the 2 different mutant mouse lines: Microphthalmia-White (Mitf+/Mi-wh) and Pou3f4delJ. Mitf+/Mi-wh mice are a model for human deafness-pigmentation syndromes, Waardenburg syndrome type 2a, and Tietz syndrome, caused by mutations in MITF gene and characterized by profound deafness along with melanocyte deficiency. Mitf+/Mi-wh mice carry heterozygous mutations in the Mitf gene and have profound HL and an EP that is less than 20 mV in adult mice (26, 27). Progressive OHC loss is observed in the Mitf+/Mi-wh mice by P28 in a tonotopic gradient from base to apex (26). Pou3f4delJ mice are a model for human X-linked nonsyndromic deafness (DFN3) caused by mutations in the gene POU3F4, which is a transcription factor. Compared with controls, Pou3f4delJ mice have reduced EP (85 mV vs. 38 mV, respectively) (28). POU3F4 is required for generation of the EP but not for generation of cochlear HCs (28). As anticipated, we confirmed a lack of EP in the Tmprss3Y260X/+;Mitf+/Mi-wh mice at 1 month of age (n = 3; EP = 0 mV). Unsurprisingly, due to a complete lack of EP, these mice had profound HL as measured by auditory brainstem response (ABR) at all frequencies tested at P16 (Supplemental Figure 2).

Next, we examined the loss of EP on HC survival and, intriguingly, there was a dramatic preservation of both IHCs and OHCs in the Tmprss3Y260X/Y260X;Mitf+/Mi-wh double-mutant mice (Figure 5, A and B). Although there was robust preservation of IHCs in Tmprss3Y260X/Y260X;Mitf+/Mi-wh mice (Figure 5B), we observed a base-to-apex gradient of OHC loss in Tmprss3Y260X/Y260X;Mitf+/Mi-wh mice that was not significantly different than Tmprss3Y260X/+;Mitf+/Mi-wh mice and is consistent with previously reported HC loss in the mouse line Mitf+/Mi-wh (Figure 5B) (29). We also observed no significant progression in OHC cell death P16 to P21 (Figure 5B).