Hearing impairment is defined as the total or partial inability to hear sound. It is the most common sensory defect in humans. Hearing loss has a wide range of biological and environmental causes and hearing tests in deaf people provide limited information regarding the molecular mechanism of pathogenesis. There is a strong genetic component to hearing impairment and analysis of the hereditary forms provides an opportunity to identify genes essential for the development and function of the auditory system. Through linkage and mutation analysis in two large families segregating deafness, we have discovered that mutations in the TMPRSS3 gene, cause autosomal recessive non-syndromic hearing loss (DFNB8/10). TMPRSS3 maps on chromosome 21q22.3 and is made of 13 exons spanning 24 kb of genomic sequence. It encodes for a type II transmembrane serine protease that contains transmembrane, low density lipoprotein receptor A (LDLRA), scavenger receptor cysteine rich (SRCR) and serine protease domains. TMPRSS3 is synthesized as an inactive zymogen of 454 residues that requires activation by cleavage in a conserved activation motif. Sixteen different mutations in the TMPRSS3 gene have been described to date. These pathogenic mutations were found to disrupt the proteolytic activation of TMPRSS3 in vitro, thereby resulting in inactive forms of the protease. We and others have investigated the prevalence of TMPRSS3-related childhood deafness in various populations. The frequency of TMPRSS3 mutations in a large cohort of deaf children originating from different European countries was calculated at 0.4%. However, the frequency was much higher in consanguineous populations. For example, TMPRSS3 mutations accounted for approximately 5% of familial childhood deafness in the Tunisian population. RNA in situ hybridization in mouse inner ear tissues revealed that Tmprss3 is expressed in the spiral ganglion, the organ of Corti and the stria vascularis. Transient expression of wild-type or tagged TMPRSS3 protein in several cell lines showed a primary localization in the endoplasmic reticulum. Using the Xenopus oocyte expression system, we have shown that TMPRSS3 regulates the activity of the sodium channel (ENaC) which is expressed in many sodium-reabsorbing tissues including the stria vascularis. TMPRSS3 could therefore participate in the maintenance of the low sodium concentration in the endolymph, which is indispensable for normal cochlear function. To investigate the physiopathology of TMPRSS3-related deafness, we have generated a mouse carrying a nonsense mutation in Tmprss3 (Y260X). This mutation is predicted to result in a prematurely truncated protein lacking most of the protease domain. Auditory brainstem response analysis revealed that Tmprss3 homozygous mutant (Tmprss3Y260X/260X) mice are completely deaf. Histological examination showed degeneration of cochlear hair cells at the onset of hearing at postnatal day 12, which progressed very rapidly following a base to apex gradient and reached completion within 2-3 days. We also evaluated the vestibular function of Tmprss3Y260X/260X mice and found mild vestibular syndrome due to a slow degeneration of saccular hair cells. Finally, Tmprss3Y260X/260X mice had normal endocochlear potential and stria vascularis structure. Taken together, these data indicate that TMPRSS3 acts as a permissive factor for cochlear hair cell survival at the onset of hearing and is required for saccular hair cell maintenance. In addition, TMPRSS3 is unlikely to play an important role in endolymphatic sodium homeostasis through the regulation of ENaC activity in the stria vascularis. Genes involved in hearing loss can be grouped into functional categories (ion channels, transcription factors, motor molecules, cytoskeletal components). TMPRSS3 does not belong to any of the existing groups and could represent the first member of a new functional category (proteolysis components). To address this hypothesis, we evaluated the 16 known TMPRSS genes for three biological criteria: expression in inner ear tissues, mapping within a genomic locus harbouring a so far unidentified deafness gene, and evaluation of hearing status of Tmprss knockout mice. Four TMPRSS genes (TMPRSS1, 2, 5 and 10) showed strong likelihood of involvement in hearing loss: Tmprss1 knock-out mice, the only mutant line available (at the time of the study) for hearing evaluation, exhibited profound hearing loss. In addition TMPRSS2, 5 and 10 were found to be expressed in inner ear tissues and map within deafness locus PKSR7, DFNB24 and DFNB25, respectively. These 4 candidate genes were then prioritized for mutation analysis in a large cohort of 562 sporadic and familial deaf cases. Numerous novel sequence variations were identified including three potential mutations in TMPRSS5. Interestingly, these three mutants showed no detectable or significantly reduced proteolytic activity in vitro suggesting that TMPRSS5 activity below threshold might cause hearing impairment. As already mentioned Tmprss1-null mice showed profound hearing loss. Interestingly Tmprss1-null cochleae displayed abnormal tectorial membrane development, reduction in nerve fiber compaction and decreased expression of the myelin proteins. As similar cochlear defects have been reported in animal models of hypothyroidism, we examined thyroid hormone levels in Tmprss1-deficient mice and found a significantly reduced free thyroxine levels. Pre- or postnatal administration of thyroxine did not correct deformities of the tectorial membrane nor ameliorate hearing loss, suggesting that Tmprss1 deficiency could affect response to thyroxine in the inner ear. These data showed that hearing loss in Tmprss1-null mice is characterized by a combination of various abnormalities that are likely to affect different cochlear processes. Unlike Tmprss3-null mice, the organ of Corti of Tmprss1 knock-out mice appeared normal as we did not observe any loss of sensory hair cells. The difference in cochlear phenotypes between Tmprss1- and Tmprss3-null mice suggests that both genes have a unique contribution to inner ear function and interact with different signalling pathways. Overall, we have demonstrated that TMPRSS1 and 3 play and TMPRSS5 might play important roles in the function of the inner ear. From the identification and initial characterization of the TMPRSS network implicated in hearing function, we are moving towards the elucidation of its complexity and working mode. The findings reported in this thesis provide a better understanding of the role of this family of serine protease in inner ear function and insights into the pathogenesis of TMPRSS-related deafness.