Skip to main content
NCBI home page
Frontiers in Genetics logoLink to Frontiers in Genetics
. 2026 Mar 18;17:1791840. doi: 10.3389/fgene.2026.1791840

Novel splice-site variants in TMPRSS3 impair hearing via exon skipping and abrogated protease activity

Xiang Wang 1,2,†, Wei-Qian Wang 2,†, Sha-Sha Huang 3,4,5,6,7,†, Xi Wang 2, Li-Min Hu 2, Xing Liu 2, Qiu-Quan Wang 2, Zhen-Dong Wang 2, Jin-Cao Xu 2,*, Yong-Yi Yuan 3,4,5,6,7,*, Xue Gao 1,2,*
PMCID: PMC13038232  PMID: 41923883

Abstract

Background

Hearing loss (HL) is genetically heterozygous, making its genetic diagnosis challenging. Identification of novel HL-associated genes and variants will enhance our understanding of the molecular mechanisms and improve genetic diagnosis. TMPRSS3, encoding a transmembrane serine protease, is implicated in autosomal recessive nonsyndromic hearing loss (ARNSHL), designated as DFNB8 (postlingual onset) or DFNB10 (prelingual onset). Although over 100 pathogenic TMPRSS3 variants have been reported, only seven splice-site variants have been documented to date.

Aims/objectives

This study investigates the molecular etiology of ARNSHL in three Chinese family and functionally characterizes five novel TMPRSS3 variants, including one non-canonical splice-site variant, two canonical splice-site variants, and two missense variants.

Material and methods

Whole-exome sequencing and gene panel sequencing identified candidate variants, followed by validation with Sanger sequencing. Functional analyses included minigene splicing assays to evaluate the variants’ effect on mRNA splicing and a yeast-based functional assay to assess their impacts on protease activity.

Results

Three families carrying compound heterozygous TMPRSS3 variants were identified. The proband with c.205 + 5G>C/c.923T>C (p.Met308Thr) presented with progressive, postlingual, high-frequency–predominant, nonsyndromic sensorineural hearing loss, consistent with DFNB8. Probands with c.572 + 1G>A/c.967G>A (p.Val323Met) and c.1348-2A>G/c.271C>T (p.Arg91Ter) exhibited prelingual, nonsyndromic sensorineural hearing loss. Functional studies revealed that all five novel variants (c.205 + 5G>C, c.923T>C, c.572 + 1G>A, c.967G>A, and c.1348-2A>G) led to reduced protease activity. Notably, the splice-site variants caused exon skipping and resulted in a more pronounced loss of activity compared to the missense variants.

Conclusion and significance

This study expands the spectrum of TMPRSS3 variants and provides mechanistic insights into their pathogenicity. Intriguingly, splice-site variants led to a more severe impairment of enzymatic activity compared to missense variants, providing molecular evidence for the considerable genotype-phenotype heterogeneity observed in TMPRSS3-associated hearing loss.

Keywords: ARNSHL, DFNB8/10, hearing loss, splice-site variant, TMPRSS3

1. Introduction

Hearing loss (HL) is among the most prevalent sensory disorders worldwide, with genetic factors contributing to over 50%–80% of congenital HL cases (Shearer et al., 1993; Choe et al., 2023). The extreme genetic heterogeneity of HL, involving hundreds of genes, poses significant challenges for molecular diagnosis and genetic counseling. Autosomal recessive nonsyndromic hearing loss (ARNSHL) accounts for approximately 70%–80% of hereditary HL cases, with 88 associated genes identified to date (Hereditary Hearing Loss Homepage, accessed, 10/19/25).

TMPRSS3 (MIM *605511), which encodes a transmembrane serine protease, is an important causative gene associated with both prelingual (DFNB10, MIM #605316) and postlingual (DFNB8, MIM #601072) forms of ARNSHL. TMPRSS3-related hearing loss is highly heterogeneous in severity, ranging from mild to profound, and typically follows a progressive course. Pathogenic variants in this gene are among the top ten genetic cause of HL (Sang et al., 2019). The TMPRSS3 protein is expressed in key cochlear cell types, including hair cells, supporting cells, the stria vascularis, and the spiral ganglion neurons (Chen et al., 2022; Guipponi et al., 2002; Arora et al., 2025; Tang et al., 2019).

To date, over 100 pathogenic TMPRSS3 variants have been reported, including missense, nonsense, frameshift, and splice-site alterations (Nisenbaum et al., 2023). Among these, splice-site variants remain exceptionally rare, with only seven documented cases (Gao et al., 2017a; Lechowicz et al., 2017; Imtiaz et al., 2011; Richard et al., 2019). The phenotypic severity of HL is influenced by both the type and genomic location of the variant, with splice-site and missense variants often exerting distinct functional effects (Guipponi et al., 2008).

Despite advances in genetic testing, molecular diagnosis of HL remains challenging due to the large number of candidate genes and the presence of variants of uncertain significance (VUS). Functional validations are therefore essential, particularly for variants suspected to affect splicing or enzymatic activity. Although a few TMPRSS3 splice-site variants have been reported, their precise molecular mechanisms and genotype-phenotype correlations are not fully understood (Boudewyns et al., 2018; Kim et al., 2017).

In this study, we investigated three Chinese families with ARNSHL and identified five novel compound heterozygous TMPRSS3 variants. The variants include one non-canonical splice-site variant (c.205 + 5G>C), two canonical splice-site variants (c.572 + 1G>A, c.1348-2A>G), and two missense variants (c.923T>C, p. Met308Thr, c.967G>A, p. Val323Met). We performed functional analyses—minigene splicing assays and yeast-based protease activity assays—to assess their pathogenicity. Our findings expand the mutational spectrum of TMPRSS3 and provide new mechanistic insights into the genotype-phenotype correlations underlying DFNB8/10-related HL.

2. Materials and methods

2.1. Study subjects

This retrospective study utilized in-house databases of genetic hearing loss from the College of Otolaryngology Head and Neck Surgery at the Chinese PLA General Hospital. The clinical assessment for HL included medical history collection (via questionnaire), otoscopy, pure tone audiometry, acoustic immittance, and stapedius reflexes. HL severity was classified according to the 2021 WHO guidelines, based on the average pure tone audiometry (PTA) threshold at 0.5, 1, 2, and 4 kHz (mild: 26–40 dB HL; moderate: 41–60 dB HL; severe: 61–80 dB HL; profound: ≥81 dB HL). Frequency-specific HL was defined as low (125–500 Hz), mid (1–2 kHz), and high (4–8 kHz). Prelingual (onset before age 3) and postlingual (onset after age 3) HL were distinguished. Vestibular function was evaluated using tandem gait, Romberg, and caloric test, while temporal bone CT were also assessed.

2.2. Whole exome sequencing (WES) and variant analysis

Whole-exome sequencing (WES) and targeted gene panel sequencing were performed for each proband. Sequence reads were aligned to the GRCh38 reference genome, followed by variant calling and annotation. Variant filtering was prioritized on known HL-related genes. Candidate variants were classified based on American College of Medical Genetics and Genomics (ACMG)/Association for Molecular Pathology (AMP) guidelines for genetic hearing loss and validated by Sanger sequencing (Richards et al., 2015).

2.3. Construction of TMPRSS3 minigene plasmids

Three TMPRSS3 splice-site variants (c.205 + 5G>C, c.572 + 1G>A, and c.1348-2A>G) were functionally tested using the pSPL3 exon-trapping vector (Miaolinbio, Wuhan, China) (Westin et al., 2021). Genomic DNA from HEK293T cells served as the template for generating mutant fragments by PCR using gene-specific primers (Supplementary Table S1, TsingkeBio, Beijing, China). For the c.205 + 5G>C mutant, exon 3 along with 338 bp of the upstream intron and 476 bp of the downstream intron were amplified. The first PCR (primers 205-F1 and 205-R1) amplified exon 3 with a 338 bp segment of intron 2, while the second PCR (primers 205-F2 and 205-R2) amplified exon 3 with a 476 bp segment of intron 3. These two fragments were then fused by overlap-extension PCR using primers 205-F1 and 205-R2 to generate the full-length mutant construct. The same strategy was applied for constructing the c.572 + 1G>A mutant fragment (exon 5 with 335 bp of upstream intron and 442 bp of downstream intron) and the c.1348-2A>G mutant fragment (exon 13 with 450 bp of upstream intron and 396 bp of downstream intron).

For cloning, the pSPL3 plasmid was digested with EcoRI and NotI (NEB, US). The ligation ratio was determined using the online tool provided by Vazyme (https://crm.vazyme.com/cetool/restructure.html). The ligation products were transformed into competent E. coli DH5α cells (Takara, Japan) and plated onto ampicillin-containing agar plates, followed by overnight incubation at 37 °C. Single colonies were picked and inoculated into LB medium supplemented with ampicillin, then cultured for 3 h. Colony PCR and Sanger sequencing were performed using primers SD-6 and SA2 to confirm correct insertions. Colonies with correct constructs were expanded in 8 mL of ampicillin-containing LB medium overnight at 37 °C with shaking. Plasmid DNA was subsequently extracted using the EndoFree Plasmid Midi Kit (CWbio, Beijing, China).

2.4. Cell transfection and splicing assay

HEK293T cells were cultured in DMEM (Thermo, US) supplemented with 10% fetal bovine serum (FBS) and penicillin-streptomycin (Beyotime, Shanghai, China) and seeded into 12-well plates. After 24 h, the cells were transfected in triplicate with either wild-type or mutant pSPL3 plasmids (c.205 + 5G>C, c.572 + 1G>A, and c.1348-2A>G) using the jetPRIME® transfection system (Polyplus, Shanghai, China), with 0.8 μg DNA, 1.6 μL jetPRIME® reagent, and 75 μL jetPRIME® buffer per well. Total RNA was extracted 48 h post-transfection using a kit (ABclonal, Beijing, China) and reverse-transcribed into cDNA (ABclonal, Beijing, China).

PCR amplification was performed in a 40 μL reaction volume. The cycling conditions were as follows: 95 °C for 12 min; 35 cycles of denaturation at 95 °C for 30 s, annealing at 60 °C for 30 s, and extension at 72 °C for 30 s; followed by a final extension at 72 °C for 5 min. The PCR products were separated on a 2% agarose gel at 160 V for 15 min and visualized under UV light. The remaining products were subjected to Sanger sequencing to confirm the splicing alterations resulting from the variants.

2.5. Molecular cloning and construction of wild-type (WT) and mutant TMPRSS3 expression plasmids

After confirming that splice-site variants alter mRNA expression, wild-type TMPRSS3 cDNA was obtained from HeLa cell (which express TMPRSS3 protein according to the Human Protein Atlas, www.proteinatlas.org). Mutant TMPRSS3 constructs were subsequently generated by overlap extension PCR using gene-specific primers (see Supplementary Table S1). For functional assays in yeast, wild-type and mutant TMPRSS3 sequences were cloned into the pGBKT7 vector (carrying a TRP1 auxotrophic marker), and the TMPRSS3 substrate epithelial sodium channels (ENaC, Coolaber, Beijing, China) was cloned into the pGADT7 vector (carrying a LEU2 auxotrophic marker). Following transformation and selection, yeast strains of each genotype were plated onto YPD medium, incubated, and colony numbers were quantified using the ImageJ software. The experiments were performed in triplicate, and the collected data are presented as mean ± SD. Statistical significance was analyzed using one-way ANOVA followed by Tukey’s post hoc test.

3. Results

3.1. Clinical findings

Family 1: the proband (II:1, a 30-year-old Han Chinese male, Figure 1A) presented with progressive, postlingual, severe sensorineural HL with onset at age 10 (Figure 1B). Audiograms revealed mid-to-high-frequency HL with absent stapedius reflexes. Tinnitus was reported. Vestibular function tests (tandem gait, Romberg, caloric tests) were unremarkable. There were no structural abnormalities of the outer, middle or inner ear were observed on temporal bone CT (Figure 1C). The patient currently uses bilateral hearing aids for auditory rehabilitation. No family history of HL was reported.

FIGURE 1.

Panel A shows pedigrees of three families with TMPRSS3 gene variants linked to hearing loss, corresponding gene sequencing chromatograms with variant annotations, and legend. Panel B contains audiograms indicating hearing thresholds for affected individuals in each family, with blue for left ear and red for right ear. Panel C displays two grayscale CT scans of the inner ear structure.

Open in a new tab

Pedigree, audiograms, temporal bone CT of families with TMPRSS3 causatives. (A) Family 1: WES identified compound heterozygous TMPRSS3 variants c.205 + 5G>C and c.923T>C (p.Met308Thr); the spouse tested negative for common deafness genes. Family 2: A hearing-loss gene panel detected c.572 + 1G>A and c.967G>A (p.Val323Met). Family 3: WES revealed c.1348-2A>G and c.271C>T. (B) Audiological findings. Family 1 II:1: Hearing loss noted at age 10; audiogram obtained at 24 years. Family 2 II:1: First audiological assessment at 1 year 2 months. Family 3 II:1: Hearing impairment observed after age 1; preoperative thresholds recorded at 2 years 10 months before right cochlear implantation. (C) Temporal-bone CT of the Family 1 proband. CT demonstrated patent external auditory canals, normal tympanic membranes, well-aerated middle ears, and symmetrical internal auditory canals without stenosis or dilation.

Family 2: the proband (II:1, a 6-year-old Han Chinese male) presented with prelingual, severe sensorineural HL. The patient underwent left cochlear implantation at 1 year and 2 months of age, and the audiogram represents his preoperative hearing status (Figure 1B). Following implantation, his auditory and speech-language development demonstrated significant improvement. No family history of HL was reported.

Family 3: the proband (II:1, a 10-year-old Han Chinese male) presented with progressive, prelingual, severe sensorineural HL with onset at age 1. The patient underwent right cochlear implantation at 2 years and 10 months of age (Figure 1B). The presented audiogram, obtained preoperatively, documents the baseline hearing level prior to surgery. His postoperative listening and language abilities have shown marked improvement. No family history of HL was reported.

3.2. Identification of TMPRSS3 variants

Previously reported TMPRSS3 variants were compiled from the Human Gene Mutation Database (HGMD) and cross-referenced with the PubMed literature. These data were then integrated to construct a comprehensive map of known TMPRSS3 variants (Figure 2A). The retrieved variants, detailed in Supplementary Table S2, revealed that splice-site variants are considerably less frequent than missense variants. Among all documented variants, missense variants constituted the majority (68%), followed by nonsense (12%), frameshift (11%), and splice-site (9%) variants (Figure 2B). In this study, we identified five novel variants, comprising three splice-site variants and two missense variants (Figure 2C). WES identified three novel TMPRSS3 variants c.205 + 5G>C, c.923T>C (p.Phe308Leu), and c.1348-2A>G with robust sequencing quality (median depth >100× for c.205 + 5G>C and c.923T>C, >200× for c.1348-2A>G; >98–99% of targeted regions covered at ≥30×). Based on ACMG/AMP guidelines, c.205 + 5G>C was classified as a variant of uncertain significance (PM2, PP3, PP4), c.923T>C as likely pathogenic (PM2, PM3, PP3, PP4), and c.1348-2A>G as pathogenic (PVS1, PM2, PM3, PP3, PP4, PP5). Panel sequencing revealed two additional novel variants with adequate coverage: c.572 + 1G>A (splicing), classified as pathogenic (PVS1, PM2, PM3, PP4), and c.967G>A, classified as a variant of uncertain significance (PM2, PM3, PP4).

FIGURE 2.

Panel A presents a gene schematic with exons, domains, and various variant types color-coded and labeled, including missense, nonsense, frameshift, splicing, and newly found variants. Panel B shows a pie chart quantifying each variant class, with missense variants comprising sixty-eight percent, nonsense twelve percent, frameshift eleven percent, and splicing nine percent. Panel C displays mRNA and protein sequence alignments for three families, comparing wild type with specific point mutations and their effects on encoded amino acids.

Open in a new tab

Summary of reported and novel TMPRSS3 variants. (A) Distribution of known and novel TMPRSS3 variants. Previously reported variants were compiled from HGMD and PubMed and mapped across the gene (see Supplementary Table S2). (B) Proportion of variant types. Incorporating five novel variants from this study, the overall distribution was: missense 68%, nonsense 12%, frameshift 11%, and splice-site 9%. (C) Novel variants identified in this study. Comparative mRNA and protein sequence diagrams illustrate the local molecular changes for each family.

The two missense variants are located in close proximity to each other, the c.923T>C variant results in a substitution of methionine by threonine at position 308 of the protein, and the c.967G>A variant leads to a substitution of valine by methionine at position 323. These variants alter a highly conserved residue within the catalytic serine protease domain. Located proximal to the most common pathogenic variant, c.916G>A, these variants are likely to exert similar deleterious effects (Gao et al., 2017a; Weegerink et al., 2011).

3.3. Splicing analysis

In vitro splicing assays were performed using pSPL3 plasmids carrying the splice site variants c.205 + 5G>C, c.572 + 1G>A, and c.1348-2A>G. To analyze the effect of each variant on splicing, we constructed minigenes containing either the wild-type or mutant sequences, which included the exons flanking the splice sites and portions of the adjacent introns. These constructs were transfected into HEK293T cells to evaluate splicing outcomes.

Sequencing confirmed that the wild-type TMPRSS3 fragments were correctly expressed, with proper splicing of exons 3, 5, and 13 (Figures 3A,B).

FIGURE 3.

Figure containing two sections: Part A shows schematics of gene structure and splicing for exons 3, 5, and 13, including normal and mutant transcripts, with Sanger sequencing traces illustrating exon skipping in the respective splice site variants; Part B presents a gel electrophoresis image comparing PCR products for wild type and mutant alleles of each exon, labeled with a size marker from one hundred to seven hundred fifty base pairs.

Open in a new tab

Analysis of the effects of splice-site variants using minigene assays. (A) Sequencing chromatograms of wild-type and mutant minigene constructs. SAv and SDv correspond to the upstream and downstream exons surrounding the inserted target gene. For c.205 + 5G>C, the minigene (338 bp upstream and 476 bp downstream of exon 3) showed normal splicing in the wild type and exon 3 skipping in the mutant. For c.572 + 1G>A, the construct (335 bp upstream and 442 bp downstream of exon 5) produced mixed transcripts in the mutant, including a 33 bp truncation and complete exon 5 skipping. For c.1348-2A>G, located downstream of the stop codon, the wild-type construct showed no SAv expression, whereas the mutant induced exon 13 skipping and restored SAv expression. (B) PCR results of wild-type and mutant minigenes. Gel electrophoresis showed paired bands for each exon: wild-type and mutant products for exon 3, exon 5, and exon 13, respectively.

The c.205 + 5G>C is a non-canonical donor site variant in intron 2. It is predicted to cause skipping of exon 3, resulting in a 111 bp truncation of the transcript (Figure 3A). This event is predicted to disrupt the LDL receptor class A (LDLRA) domain, affecting protein stability.

The c.572 + 1G>A is a canonical donor site variant in intron 5 within the scavenger receptor cysteine-rich (SRCR) domain. This variant results in two splicing outcomes: complete skipping of the exon 5, similar to c.205 + 5G>C, resulting in a 126 bp truncation of the transcript; or a 33 bp truncation within exon 5 (Figure 3A). Both outcomes are likely to induce misfolding of the highly conserved SRCR domain, leading to loss of enzymatic activity or impaired substrate recognition, thereby inhibiting ENaC activation.

The c.1348-2A>G is a canonical acceptor site variant upstream of exon 13 within the serine protease domain. It led to exon 13 skipping (Figure 3A). Since exon 13 contains the stop codon, its exclusion results in a frameshift and incorporation of downstream intronic sequences, leading to the formation of an aberrant serine protease (SP) domain.

Minigene assays confirmed that the splice-site variants led to aberrant splicing. While the wild-type construct produced the expected normal transcript, the mutant constructs resulted in abnormal splicing patterns. These findings were consistently replicated in three independent experiments in HEK293T cells. Sequencing all bands confirmed break points and splicing events.

3.4. Mutant TMPRSS3 proteins showed aberrant protease activity

Previous studies have demonstrated that TMPRSS3 cleaves ENaC, and the cleavage efficiency can serve as an indicator of TMPRSS3 enzymatic activity (Guipponi et al., 2002). Cleavage efficiency correlates with colony growth on selective medium. Yeast successfully co-transformed with the TMPRSS3 and its substrate plasmids carried the LEU and TRP markers, respectively, and were selected on minimal media containing 2% glucose lacking leucine and tryptophan (Leu−/Trp−). Subsequently, the Leu+/Trp+ transformants were then replica-plated onto YPD medium supplemented with 2% sucrose and 0.5 mg/mL antimycin A. Quantification of colony number and size was performed using the ImageJ software, serve as a proxy for protease activity (Kim et al., 2017; Wong et al., 2020).

To assess the pathogenic potential of the identified TMPRSS3 variants, we evaluated the proteolytic activity of the mutant alleles in comparison to the wild-type protein. The results demonstrated a marked reduction in enzymatic activity across all mutant constructs. When colonies larger than 0.005 mm2 were quantified, those carrying the variants exhibited a significantly slower growth rate than the wild-type. By day 3, colonies harboring splice-site variants were scarcely visible, while those with missense variants were detectable but had a significantly smaller average area. Mutant colonies that remained undetectable by day 3 were further incubated alongside wild-type controls until day 5, at which point colonies corresponding to all three splice-site variants became visible, confirming their viability and excluding the possibility of complete growth failure. Collectively, these functional findings support the pathogenic potential of all five novel variants identified in this study (Figure 4).

FIGURE 4.

Eight petri dish photographs show yeast colony growth under various TMPRSS3 genetic variants at three and five days, revealing different colony densities. Below, two bar graphs quantify average yeast colony area at three and five days, indicating significant reductions for mutant variants compared to wild-type TMPRSS3 with statistical markers.

Open in a new tab

Functional assessment of TMPRSS3 variants using a yeast protease assay. Colony growth was quantified using standard morphological criteria. After 3 days at 30 °C (top row), the assay shows the wild-type control followed by the missense variants c.923T>C and c.967G>A, and the splice-site variants c.572 + 1G>A, c.205 + 5G>C, and c.1348-2A>G. After 5 days (bottom row), the wild-type control and the three exon-skipping mutants (c.205 + 5G>C, c.572 + 1G>A, c.1348-2A>G) are shown. Colony numbers at both time points are presented in the accompanying graphs, demonstrating deleterious effects for all variants, with splice-site variants causing more severe growth impairment than missense variants. All experiments were performed with three independent biological replicates, and data are presented as mean ± SD. Statistical significance was analyzed using one-way ANOVA followed by Tukey’s post hoc test, with p < 0.05 considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001.

While observing a decrease in protein activity, we also predicted the protein structural changes caused by each variant. Point variants did not cause obvious changes in the protein backbone, whereas splice-site alterations had a greater impact. However, predictions for the intramembrane and transmembrane regions differed markedly, and the confidence of the predictions was low (Supplement figure).

Interestingly, the colony growth defect associated with the non-canonical splice-site variant (c.205 + 5G>C) appeared less severe than that caused by the two canonical splice-site variants (c.572 + 1G>A and c.1348-2A>G) (Figure 4). This observation suggests a potential gradation in pathogenicity based on splice-site type, although further evidence is required to support this conclusion.

4. Discussion

In this study, we identified five novel compound heterozygous TMPRSS3 variants (c.205 + 5G>C, c.923T>C, c.967G>A, c.572 + 1G>A, and c.1348-2A>G) in three Chinese individuals with prelingual autosomal recessive non-syndromic hearing loss (ARNSHL). The splice-site variants are particularly noteworthy, as they not only expand the mutational spectrum of TMPRSS3 but also exhibit more severe pathogenic effects than the missense variants.

While over 100 TMPRSS3 variants have been reported, splice-site variants remain exceptionally rare, with only seven had been clearly documented prior to this report. Our identification of three additional splice-site variants contributes significantly to this limited dataset. Functional characterization through minigene splicing assays confirmed that these variants disrupt normal mRNA processing, and protease activity assays demonstrated a marked reduction in enzymatic function. These findings extend beyond prior studies, which have predominantly focused on missense variants that impair protease activity, and provide further mechanistic support for establishing the genotype-phenotype correlations in TMPRSS3-related HL.

Yeast-based protease activity assays have served as an important functional tool for evaluating TMPRSS3 since 2003. Studies have consistently shown that pathogenic variants such as Asp103Gly, Arg109Trp, Cys194Phe, Trp251Cys, Pro404L, Cys407Arg, and Ala426Thr abolish yeast growth on sucrose plates, indicating a severe loss of proteolytic activity, whereas benign polymorphisms like Gly111Ser and Ile253Val do not impair function (Lee et al., 2003). Subsequent studies has corroborated this pattern, with variants like Val116Met and Val291Leu being highly deleterious, and others such as Arg80His, Leu184Ser, Ala418Val, and Trp251Cys also demonstrated severe functional impairment (Kim et al., 2017; Wong et al., 2020).

Previous work has shown that loss-of-function variants in any of the three major domains—LDLRA, SRCR domain, and serine protease—can result in HL (Guipponi et al., 2002; Masmoudi et al., 2001). In line with this, the variants identified in our study, including two missense and three splice-site alterations, all resulted in impaired protease function. Minigene assays specifically demonstrated that the splice-site variants c.205 + 5G>C, c.572 + 1G>A, and c.1348-2A>G caused the skipping of exons 3, 5, and 13, respectively. Notably, these splicing defects were associated with the most severe loss of enzymatic in our functional assays.

To date, 10 TMPRSS3 splice-site variants have been associated with HL of varying severity. A pattern emerges where canonical splice-site variants (e.g., c.1195-1G>C and c.783-1G>A) generally cause prelingual, profound HL, likely due to complete exon skipping and protein truncation. In contrast, non-canonical variants such as c.323-6G>A and c.205 + 5G>C tend to produce postlingual or progressive phenotypes (DFNB8). This clinical distinction may be explained by partial splicing defects that preserve residual protein function, a hypothesis consistent with the possibility of incomplete exon skipping induced by non-canonical variants.

The pathogenic severity of TMPRSS3 variants vary widely, and compound heterozygosity further complicates clinical interpretation. While missense variants have been more extensively studied, splice-site variants remain relatively under-investigated. Our study addressed this gap by functionally validation three novel splice-site variants, confirming they cause aberrant splicing and exon skipping, and demonstrating that all five novel variants impair proteolytic function. This provides direct molecular evidence of their pathogenicity.

The molecular pathogenesis underlying TMPRSS3-related HL have been elucidated through cellular and animal studies. Mice carrying a nonsense variant (Tmprss3 Y206X) develop postnatal degeneration of cochlear hair cells beginning at postnatal day 12 (P12), demonstrating the protein’s essential role in hair cell survival and cochlear homeostasis (Shearer et al., 2025). Mechanistically, TMPRSS3 is known to activate the ENaC, maintains potassium homeostasis, and support normal outward K+ currents in inner hair cells. Taken together, these findings reinforce the critical functional role of TMPRSS3 in the auditory system (Molina et al., 2013).

Globally, TMPRSS3 variants account for approximately 0.36%–12% of genetic HL cases (Walsh et al., 2006; Wattenhofer et al., 2005; Damrongchietanon et al., 2025; Bharadwaj et al., 2025; Li et al., 2024), with notable differences among populations (e.g., only 0.36% in a large Japanese cohort) (Nisenbaum et al., 2023; Colbert et al., 2024; Moon et al., 2021; Gao et al., 2017b). Although not the most common ARNSHL gene, TMPRSS3 remains clinically important due to: (i) its association with both prelingual (DFNB10) and postlingual (DFNB8) HL forms; (ii) phenotypic variability that correlates with variant type and domain; and (iii) its potential as a target for emerging gene therapies.

Recent advances in TMPRSS3-targeted gene therapy have been encouraging. Adeno-associated virus (AAV)-mediated gene replacement has successfully restored hearing in mouse models, paving the way for potential clinical translation (Du et al., 2023; Richard and Delprat, 2024). Our findings demonstrate that splice-site variants lead to adjacent exon skipping and that splice-site and missense variants differentially affect TMPRSS3 function, suggest that TMPRSS3-related HL may be amenable to tailored therapeutic strategies. Specifically, antisense oligonucleotide-based approaches designed to correct splicing defects could represent a viable treatment avenue for patients carrying splice-site variants.

5. Conclusion

In conclusion, our findings advance the understanding of TMPRSS3-related HL by functionally characterizing novel splice-site and missense variants and delineating their distinct impact on protein function. The more severe enzymatic impairment caused by splice-site variants provides a mechanistic explanation for the phenotypic heterogeneity observed in patients. This work underscores the necessity of precision medicine approaches in HL treatment, where therapeutic strategies may need to be tailored based on the specific molecular consequences of each patient’s variants. As gene therapies for HL advance toward clinical application, functional studies like this will be crucial for accurately matching patients with the appropriate therapeutic interventions. Integrating genetic diagnosis with detailed functional validation is a critical step toward achieving personalized medicine for hereditary hearing loss.

Acknowledgements

We sincerely thank all the family members for their participation and cooperation in this study. We thank Professor Wei Xiong of Chinese Institute for Brain Research for his technical assistance and valuable suggestions.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. This research was supported by grants from National Natural Science Foundation of China (82171158, 82271177 and 82271185). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Footnotes

Edited by: Roberto Valli, University of Insubria, Italy

Reviewed by: Sedigheh Delmaghani, Institut de l’Audition, Institut Pasteur, France

Qiong Li, Air Force Medical University, China

Data availability statement

The original contributions presented in the study are included in the article/supplementary material. The study’s WES data are not publicly available due China’s Regulation of Human Genetics Resource. Further inquiries can be directed to the corresponding authors.

Ethics statement

The studies involving humans were approved by the Chinese PLA General Hospital Research Ethics Committee (reference number S2016–120–02). Written informed consent was obtained from all participants or their legal guardians for participation. All in vivo experiments were carried out in accordance with Chinese Association for Laboratory Animal Science (CALAS) guidelines for the care and use of laboratory animals and were approved by the Animal Care and Use Committees of the Chinese PLA General Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin. Written informed consent was obtained from the individual(s), and minor(s)’ legal guardian/next of kin, for the publication of any potentially identifiable images or data included in this article.

Author contributions

XaW: Writing – original draft, Visualization, Conceptualization, Data curation. W-QW: Writing – review and editing, Conceptualization, Visualization. S-SH: Writing – review and editing, Visualization, Data curation, Funding acquisition. XiW: Data curation, Writing – review and editing, Visualization. L-MH: Writing – review and editing, Visualization, Data curation. XL: Writing – review and editing, Visualization, Data curation. Q-QW: Data curation, Visualization, Writing – review and editing. Z-DW: Data curation, Writing – review and editing, Visualization. J-CX: Writing – review and editing, Visualization, Data curation. Y-YY: Visualization, Writing – review and editing, Conceptualization. XG: Writing – review and editing, Funding acquisition, Conceptualization, Project administration.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fgene.2026.1791840/full#supplementary-material

SUPPLEMENTARY FIGURE S1

Functional assessment of TMPRSS3 variants using a yeast protease assay. AlphaFold was used to perform 3D prediction and reconstruction of the protein structure. IPTM is the interface predicted template modeling score, PTM is the predicted template modeling score, and pLDDT is the predicted Local Distance Difference Test. (blue pLDDT > 90, cyan 90 > pLDDT > 70, yellow 70 > pLDDT > 50, and orange pLDDT < 50).

Table2.xlsx (23.3KB, xlsx)
Image1.jpeg (11.6MB, jpeg)
Table1.xlsx (11.2KB, xlsx)

References

  1. Arora R., Pifková L., Deutschmann A. U., Reisinger E. (2025). Tmprss3 is expressed in several cell types of the inner ear including type II but hardly in type I spiral ganglion neurons. Front. Cellular Neuroscience 19, 1690227. 10.3389/fncel.2025.1690227 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bharadwaj T., Acharya A., Manyisa N. R., Aboagye E. T., Peigou Wonkam R., Xhakaza L., et al. (2025). The diverse genetic landscape of hearing impairment in South African families. Clin. Genetics 108 (5), 511–520. 10.1111/cge.14765 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Boudewyns A., van den Ende J., Sommen M., Wuyts W., Peeters N., Van de Heyning P., et al. (2018). Role of targeted next generation sequencing in the etiological Work-Up of congenitally deaf children. Otology and Neurotology Official Publication Am. Otological Soc. Am. Neurotol. Soc. Eur. Acad. Otology Neurotol. 39 (6), 732–738. 10.1097/mao.0000000000001847 [DOI] [PubMed] [Google Scholar]
  4. Chen Y. S., Cabrera E., Tucker B. J., Shin T. J., Moawad J. V., Totten D. J., et al. (2022). TMPRSS3 expression is limited in spiral ganglion neurons: implication for successful cochlear implantation. J. Medical Genetics 59 (12), 1219–1226. 10.1136/jmg-2022-108654 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Choe G., Park S. K., Kim B. J. (2023). Hearing loss in neonates and infants. Clin. Experimental Pediatrics 66 (9), 369–376. 10.3345/cep.2022.01011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Colbert B. M., Lanting C., Smeal M., Blanton S., Dykxhoorn D. M., Tang P. C., et al. (2024). The natural history and genotype-phenotype correlations of TMPRSS3 hearing loss: an international, multi-center, cohort analysis. Hum. Genetics 143 (5), 721–734. 10.1007/s00439-024-02648-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Damrongchietanon T., Wattanasirichaigoon D., Khongkraparn A., Noojarern S., Tiravanitchakul R., Kasemkosin N., et al. (2025). Diagnostic yield of whole exome sequencing with targeted gene analysis in prelingual sensorineural hearing loss in Thailand. Sci. Reports 15 (1), 32784. 10.1038/s41598-025-18038-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Du W., Ergin V., Loeb C., Huang M., Silver S., Armstrong A. M., et al. (2023). Rescue of auditory function by a single administration of AAV-TMPRSS3 gene therapy in aged mice of human recessive deafness DFNB8. Mol. Therapy The Journal Am. Soc. Gene Ther. 31 (9), 2796–2810. 10.1016/j.ymthe.2023.05.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Gao X., Huang S. S., Yuan Y. Y., Xu J. C., Gu P., Bai D., et al. (2017a). Identification of TMPRSS3 as a significant contributor to autosomal recessive hearing loss in the Chinese population. Neural Plast. 2017, 3192090. 10.1155/2017/3192090 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Gao X., Yuan Y. Y., Wang G. J., Xu J. C., Su Y., Lin X., et al. (2017b). Novel mutations and mutation combinations of TMPRSS3 cause various phenotypes in one Chinese family with autosomal recessive hearing impairment. BioMed Research International 2017, 4707315. 10.1155/2017/4707315 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Guipponi M., Vuagniaux G., Wattenhofer M., Shibuya K., Vazquez M., Dougherty L., et al. (2002). The transmembrane serine protease (TMPRSS3) mutated in deafness DFNB8/10 activates the epithelial sodium channel (ENaC) in vitro . Hum. Molecular Genetics 11 (23), 2829–2836. 10.1093/hmg/11.23.2829 [DOI] [PubMed] [Google Scholar]
  12. Guipponi M., Antonarakis S. E., Scott H. S. (2008). TMPRSS3, a type II transmembrane serine protease mutated in non-syndromic autosomal recessive deafness. Front. Bioscience A Journal Virtual Library 13, 1557–1567. 10.2741/2780 [DOI] [PubMed] [Google Scholar]
  13. Imtiaz F., Taibah K., Ramzan K., Bin-Khamis G., Kennedy S., Al-Mubarak B., et al. (2011). A comprehensive introduction to the genetic basis of non-syndromic hearing loss in the Saudi Arabian population. BMC Medical Genetics 12, 91. 10.1186/1471-2350-12-91 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kim A. R., Chung J., Kim N. K. D., Lee C., Park W. Y., Oh D. Y., et al. (2017). The analysis of A frequent TMPRSS3 allele containing P.V116M and P.V291L in A cis configuration among deaf koreans. Int. J. Mol. Sci. 18 (11), 2246. 10.3390/ijms18112246 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Lechowicz U., Gambin T., Pollak A., Podgorska A., Stawinski P., Franke A., et al. (2017). Iterative sequencing and variant screening (ISVS) as a novel pathogenic mutations search strategy - application for TMPRSS3 mutations screen. Sci. Reports 7 (1), 2543. 10.1038/s41598-017-02315-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Lee Y. J., Park D., Kim S. Y., Park W. J. (2003). Pathogenic mutations but not polymorphisms in congenital and childhood onset autosomal recessive deafness disrupt the proteolytic activity of TMPRSS3. J. Medical Genetics 40 (8), 629–631. 10.1136/jmg.40.8.629 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Li J., Zhou S., Pei J., Li W., Cui R., Ren X., et al. (2024). Spectrum of DNA variants for patients with hearing loss in 4 language families of 15 ethnicities from Southwestern China. Heliyon 10 (20), e38802. 10.1016/j.heliyon.2024.e38802 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Masmoudi S., Antonarakis S. E., Schwede T., Ghorbel A. M., Gratri M., Pappasavas M. P., et al. (2001). Novel missense mutations of TMPRSS3 in two consanguineous Tunisian families with non-syndromic autosomal recessive deafness. Hum. Mutation 18 (2), 101–108. 10.1002/humu.1159 [DOI] [PubMed] [Google Scholar]
  19. Molina L., Fasquelle L., Nouvian R., Salvetat N., Scott H. S., Guipponi M., et al. (2013). Tmprss3 loss of function impairs cochlear inner hair cell Kcnma1 channel membrane expression. Hum. Molecular Genetics 22 (7), 1289–1299. 10.1093/hmg/dds532 [DOI] [PubMed] [Google Scholar]
  20. Moon I. S., Grant A. R., Sagi V., Rehm H. L., Stankovic K. M. (2021). TMPRSS3 gene variants with implications for auditory treatment and counseling. Front. Genetics 12, 780874. 10.3389/fgene.2021.780874 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Nisenbaum E., Yan D., Shearer A. E., de Joya E., Thielhelm T., Russell N., et al. (2023). Genotype-phenotype correlations in TMPRSS3 (DFNB10/DFNB8) with emphasis on natural history. Audiology and Neuro-Otology 28 (6), 407–419. 10.1159/000528766 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Richard É. M., Delprat B. (2024). Local administration of an AAV-TMPRSS3 vector in aged Tmprss3 mutant mice with progressive deafness restores hearing. Med. Sciences M/S. 40 (5), 402–404. 10.1051/medsci/2024042 [DOI] [PubMed] [Google Scholar]
  23. Richard E. M., Santos-Cortez R. L. P., Faridi R., Rehman A. U., Lee K., Shahzad M., et al. (2019). Global genetic insight contributed by consanguineous Pakistani families segregating hearing loss. Hum. Mutation 40 (1), 53–72. 10.1002/humu.23666 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Richards S., Aziz N., Bale S., Bick D., Das S., Gastier-Foster J., et al. (2015). Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American college of medical Genetics and Genomics and the association for molecular pathology. Genet. Medicine Official Journal Am. Coll. Med. Genet. 17 (5), 405–424. 10.1038/gim.2015.30 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Sang S., Ling J., Liu X., Mei L., Cai X., Li T., et al. (2019). Proband whole-exome sequencing identified genes responsible for autosomal recessive non-syndromic hearing loss in 33 Chinese nuclear families. Front. Genetics 10, 639. 10.3389/fgene.2019.00639 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Shearer A. E., Hildebrand M. S., Odell A. M., Smith R. J. H. (1993). “Genetic hearing loss overview,” in GeneReviews®[Internet]. Editors Adam M. P., Bick S., Mirzaa G. M., Pagon R. A., Wallace S. E., Amemiya A. (Seattle (WA): University: of Washington, Seattle; ), 1993–2026. [PubMed] [Google Scholar]
  27. Shearer A. E., Chen Y. S., Rouse S. L., Wang X., Marin Fermin J., Booth K. T., et al. (2025). Endocochlear potential contributes to hair cell death in TMPRSS3 hearing loss. J. Clinical Investigation 135 (18), e186395. 10.1172/jci186395 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Tang P. C., Alex A. L., Nie J., Lee J., Roth A. A., Booth K. T., et al. (2019). Defective Tmprss3-Associated hair cell degeneration in inner ear organoids. Stem Cell Reports 13 (1), 147–162. 10.1016/j.stemcr.2019.05.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Walsh T., Abu Rayan A., Abu Sa, Shahin H., Shepshelovich J., Lee M. K., et al. (2006). Genomic analysis of a heterogeneous Mendelian phenotype: multiple novel alleles for inherited hearing loss in the Palestinian population. Hum. Genomics 2 (4), 203–211. 10.1186/1479-7364-2-4-203 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Wattenhofer M., Sahin-Calapoglu N., Andreasen D., Kalay E., Caylan R., Braillard B., et al. (2005). A novel TMPRSS3 missense mutation in a DFNB8/10 family prevents proteolytic activation of the protein. Hum. Genetics 117 (6), 528–535. 10.1007/s00439-005-1332-x [DOI] [PubMed] [Google Scholar]
  31. Weegerink N. J., Schraders M., Oostrik J., Huygen P. L. M., Strom T. M., Granneman S., et al. (2011). Genotype-phenotype correlation in DFNB8/10 families with TMPRSS3 mutations. J. Assoc. Res. Otolaryngology JARO 12 (6), 753–766. 10.1007/s10162-011-0282-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Westin I. M., Jonsson F., Österman L., Holmberg M., Burstedt M., Golovleva I. (2021). EYS mutations and implementation of minigene assay for variant classification in EYS-associated retinitis pigmentosa in northern Sweden. Sci. Reports 11 (1), 7696. 10.1038/s41598-021-87224-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Wong S. H., Yen Y. C., Li S. Y., Yang J. J. (2020). Novel mutations in the TMPRSS3 gene may contribute to Taiwanese patients with nonsyndromic hearing loss. Int. J. Mol. Sci. 21 (7), 2382. 10.3390/ijms21072382 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SUPPLEMENTARY FIGURE S1

Functional assessment of TMPRSS3 variants using a yeast protease assay. AlphaFold was used to perform 3D prediction and reconstruction of the protein structure. IPTM is the interface predicted template modeling score, PTM is the predicted template modeling score, and pLDDT is the predicted Local Distance Difference Test. (blue pLDDT > 90, cyan 90 > pLDDT > 70, yellow 70 > pLDDT > 50, and orange pLDDT < 50).

Table2.xlsx (23.3KB, xlsx)
Image1.jpeg (11.6MB, jpeg)
Table1.xlsx (11.2KB, xlsx)

Data Availability Statement

The original contributions presented in the study are included in the article/supplementary material. The study’s WES data are not publicly available due China’s Regulation of Human Genetics Resource. Further inquiries can be directed to the corresponding authors.

Articles from Frontiers in Genetics are provided here courtesy of Frontiers Media SA

RESOURCES