rs45598239 — TMPRSS3

TMPRSS3 — The Inner Ear Protease Essential for Cochlear Hair Cell Survival

The TMPRSS3 gene on chromosome 21q22.3¹¹ chromosome 21q22.3 encodes a type II transmembrane serine protease indispensable for the survival and maturation of cochlear hair cells — the sensory cells that translate mechanical sound vibrations into electrical nerve signals. When TMPRSS3 function is lost or severely reduced, hair cells begin degenerating at the precise moment hearing first activates²² hair cells begin degenerating at the precise moment hearing first activates
In mouse models, hair cell loss begins at postnatal day 12 — the exact onset of hearing — starting in the high-frequency basal cochlear turn and sweeping toward the apex within 48 hours. Bi-allelic loss of TMPRSS3 causes DFNB8 (postlingual progressive deafness) and DFNB10 (congenital profound deafness).

This missense variant at the TMPRSS3 locus tags population-level variation in hearing loss risk. Large-scale genome-wide association studies³³ genome-wide association studies
Population-scale GWAS meta-analysis with 125,749 cases and 469,497 controls across five cohorts, Communications Biology 2022 have identified common coding variants at this locus — including the Ala90Thr change (MAF=0.06) — as contributors to adult hearing loss susceptibility, bridging the gap between the rare Mendelian forms of TMPRSS3 deafness and common age-related hearing decline.

The Mechanism

TMPRSS3 is expressed in inner hair cells, outer hair cells, spiral ganglion neurons, and the stria vascularis⁴⁴ inner hair cells, outer hair cells, spiral ganglion neurons, and the stria vascularis
The stria vascularis maintains the endocochlear potential, the +80 mV electrochemical gradient essential for hair cell mechanotransduction of the developing and mature cochlea. The protein's serine protease catalytic domain — containing the canonical His-Asp-Ser catalytic triad — is required for a still-incompletely understood signaling cascade that prevents apoptosis of newly activated hair cells. Current evidence suggests TMPRSS3 may regulate ion channel activity (particularly ENaC, the epithelial sodium channel) and cochlear fluid homeostasis, though ENaC alone cannot fully explain the phenotype⁵⁵ ENaC alone cannot fully explain the phenotype
Patients with pseudohypoaldosteronism and ENaC loss-of-function have normal hearing, arguing against ENaC as the sole target.

Pathogenic variants in TMPRSS3 fall into two severity classes: loss-of-function alleles (frameshift, nonsense, canonical splice) that completely abolish protease activity, and missense alleles that reduce — but do not eliminate — enzyme function. The hearing phenotype is determined by the combination: two loss-of-function alleles produce prelingual profound deafness (DFNB10), while a loss-of-function paired with a missense allele typically produces postlingual progressive high-frequency hearing loss (DFNB8) with a characteristic ski-slope audiogram⁶⁶ ski-slope audiogram
Down-sloping configuration with disproportionate loss at 2,000 Hz and above, progressing at approximately 0.3 dB/year.

The Evidence

TMPRSS3 accounts for a substantial proportion of autosomal recessive nonsyndromic hearing loss (ARNSHL) in many populations: 0.7% in Japanese, 3% in Pakistani, 4.6% in Chinese, 5–6% in Tunisian, 5.9% in Korean, and up to 11% in Turkish⁷⁷ 0.7% in Japanese, 3% in Pakistani, 4.6% in Chinese, 5–6% in Tunisian, 5.9% in Korean, and up to 11% in Turkish ARNSHL cases. In contrast, it accounts for less than 1% of ARNSHL in Caucasian populations, suggesting population-specific founder variants.

The most documented pathogenic variants include p.Ala306Thr (c.916G>A), identified in German, Dutch, Korean, and Chinese families, and p.His70Thrfs*19 (c.208delC), a frameshift causing a premature stop codon that is particularly prevalent in Slovenian populations. In East Asian populations, p.Ala306Thr allele frequency in gnomAD is approximately 6 per 10,000 alleles.

A landmark international multi-center cohort study of 127 individuals⁸⁸ international multi-center cohort study of 127 individuals
Colbert et al. Human Genetics 2024 established that cochlear implantation produces a mean word recognition score of 76% in TMPRSS3-related hearing loss — excellent outcomes that support early implantation. Crucially, age at implantation — not genotype — was the strongest predictor of speech recognition outcome, declining approximately 0.3% per year of delay.

At the population level, large-scale GWAS analyses have demonstrated that TMPRSS3 locus variants contribute to common adult hearing loss⁹⁹ TMPRSS3 locus variants contribute to common adult hearing loss
53 loci identified affecting hearing loss risk; TMPRSS3 Ala90Thr top variant at one locus; MAF=0.06 in the studied population.

Practical Implications

For heterozygous carriers (one copy of the risk variant), hearing is typically normal in childhood and early adulthood based on both human carrier studies and mouse heterozygote data.

For any individual with confirmed TMPRSS3-related hearing loss (biallelic variants), cochlear implantation should be pursued promptly. The evidence is clear and consistent across multiple populations: outcomes are excellent, and delay worsens speech recognition scores.