Statistics from Altmetric.com
Editor—Congenital deafness occurs in approximately 1 in 1000 live births and 50% of these cases are hereditary. Non-syndromic deafness is classified according to its mode of inheritance as DFN, DFNA, and DFNB (X linked, autosomal dominant, and autosomal recessive, respectively). Non-syndromic recessive deafness accounts for ∼80% of congenital hereditary deafness cases.1 At least 30 DFNB loci have been mapped in the past few years by genetic linkage studies, but the causative gene has been identified for only eight of these loci2-4 (Hereditary Hearing Loss Homepage, http://www.uia.ac.be/dnalab/hhh).
Two of the previously reported loci for non-syndromic recessive deafness are DFNB8 and DFNB10, both located on chromosome 21q22.3 (MIM 601072 and 605316). The DFNB8 locus was originally identified in a large consanguineous Pakistani family, segregating childhood onset deafness,5 while DFNB10 was identified in a large consanguineous Palestinian family, in which deafness was congenital.6 Recently, theTMPRSS3 gene was shown to be mutated in affected subjects of both families.7
TMPRSS3 belongs to a family of transmembrane serine proteases, also including TMPRSS1,8 TMPRSS2,9 and TMPRSS4.10 The TMPRSS3 gene extends over 24 kb and comprises 13 exons. It has four alternative transcripts (TMPRSS3 a,b, c, andd), encoding putative peptides of 454, 327, 327, and 344 amino acids, respectively.7 TMPRSS3a, which contains all 13 exons, is the most abundant transcript and its expression could be detected in various tissues, including fetal cochlea.7 In addition to the serine protease and the transmembrane domains,TMPRSS3 also encodes low density lipoprotein receptor class A (LDLRA) and scavenger receptor cysteine rich (SRCR) domains, which are potentially involved in binding with extracellular molecules and/or the cell surface.7
Material and methods
To identify DFNB8/B10 linked families, we analysed a total of 159 consanguineous Pakistani families that segregate profound congenital deafness and are either large enough to support statistically significant linkage or have at least three affected subjects. Families were ascertained in schools for the deaf in Punjab and Karachi. IRB approval (OH93-DC-016) and informed consent were obtained for all participating family members. Genomic DNA was extracted from venous blood samples according to a standard protocol.11 Linkage to known recessive deafness loci (DFNBs) and refinement of the DFNB8/B10 region in linked families was performed on an ABI-377 sequencer (PE Applied Biosystems) using marker information provided by the Hereditary Hearing Loss Homepage and by Berry et al.12 Linkage analysis was conducted with the FASTLINK version of the LINKAGE program package.13 14 Five of the families showed potential linkage to the DFNB8/B10 locus on chromosome 21q22.3 (fig 1). Medical history and pure tone audiometry testing indicated that all five families segregate congenital, profound, non-syndromic sensorineural deafness.
Results and discussion
To detect mutations in the TMPRSS3 gene in DFNB8/B10 families, we determined the gene sequence in two affected subjects from each family by PCR amplification from genomic DNA of the 13 exons, including intron-exon boundaries, and cycle sequencing of the PCR products using TMPRSS3 specific primers. Primer sequences and PCR conditions are summarised in table 1. PCR products werepurified with the QIAquick PCR purification kit (Qiagen). Sequence analysis using one of the amplification primers or an internal primer was performed with the Big Dye terminator cycle sequencing kit on an ABI-377 sequencer (PE Applied Biosystems). Novel TMPRSS3 mutations were identified in four of the families. The relevantTMPRSS3 exons in all available family members were then amplified by PCR and sequenced. This confirmed that in each family deaf subjects were homozygous for theTMPRSS3 mutation and obligate carrier parents were heterozygous. Family and mutation data, including population study results for each mutation, are summarised in table 2. Several single nucleotide polymorphisms (SNPs) were also identified in both coding and non-coding sequences of theTMPRSS3 gene (table3).
Families PKSN37 and PKSR18b (fig 1A, B) were found to segregate the same TMPRSS3 missense mutation, C407R (table 2). This mutation is a T to C transition at position 1420 (1219 from the first ATG) (GenBank accession number AB038157), located in exon 12, and leads to a cysteine to arginine substitution. C407 is located in the serine protease domain of TMPRSS3, only a few amino acids from the active site residue S401 within the substrate pocket, and thus may be important in substrate specificity. Although C407 is not highly conserved, three out of the other four known TMPRSS proteins (mouse and human TMPRSS2 and human TMPRSS1) have either a cysteine or a threonine at this position, both of which are polar, uncharged amino acids, while the mutation changes the amino acid at this position to the positively charged arginine. All affected subjects were homozygous for the C407R mutation and obligate carriers were heterozygous (fig 1A, B and 2A). C407R was also found in one of 200 normal control Muslim Indian chromosomes. The haplotype for markers flanking theTMPRSS3 gene (834A1.CA78 and 994G8.CA82, located approximately 300 kb and 100 kb from the mutation, respectively) in both families and in the Muslim Indian subject harbouring the C407R mutation was identical (fig 1A, B and data not shown).
Another missense mutation, R109W (table 2), was found in family PKSR51a (fig 1C). This mutation is a C to T transition at position 526 (325 from the first ATG) (GenBank accession number AB038157), located in exon 5, and leads to an arginine to tryptophan substitution. Three out of the other four known TMPRSS proteins have either an arginine or the similar positively charged histidine at this position (mouse and human TMPRSS2 and human TMPRSS4). R109 is located within the SRCR domain of TMPRSS3. The SRCR is an adhesive extracellular domain (PROSITE Database of Protein Families and Domains, http://www.expasy.ch/prosite, accession number PDOC00348), which is potentially involved in binding of TMPRSS3 with extracellular molecules and/or the cell surface.7 All affected subjects belonging to the main branch of the family were homozygous for the R109W mutation and obligate carriers were heterozygous (fig 1C and 2B). Yet, subject I.2, who is deaf, was found to be homozygous for the normal allele. This finding is not surprising, since our initial genotyping data indicated that I.2, who belongs to a remote branch of the family, has a different haplotype at the DFNB8/B10 region than the haplotype shared by the other deaf subjects in the family (fig 1C). Thus, her deafness might be the result of a mutation in a different gene or the result of non-genetic factors.
Family PKB16 (fig 1D) was found to segregate a third missense mutation, C194F (table 2). This mutation is a G to T transversion at position 782 (581 from the first ATG) (GenBank accession numberAB038157), located in exon 7, which encodes part of the SRCR domain, and leads to the substitution of the cysteine at position 194, which is highly conserved among all TMPRSS proteins, to phenylalanine. All affected subjects were homozygous for the C194F mutation and obligate carriers were heterozygous (fig 1D and 2C).
Family PKSR7 (fig 1E) supports a simulated maximal lod score of 3.8 (FAST SLINK15), and deaf subjects are homozygous for markers spanning the DFNB8/B10 region (fig 1E). The region of homozygosity is shared by all affected subjects, but is more restricted in one of the sibships (fig 1E, II.9-13). However, none of the markers is fully informative, resulting in a maximal calculated lod score of 2.9. The disease allele frequency was set at 0.0011 (upper limit for recessive deafness based on estimates from the Indian population16) and the disease was coded as fully penetrant and recessive with a 1/1000 phenocopy rate. Allele frequencies of 0.1 and 0.2 for each allele of markers 834A1.CA78 and 994G8.CA82, respectively, were assumed based on observations in the other analysed Pakistani DFNB8/B10 families. The lod score was not significantly changed by omitting subjects II.9-13 (2.7), nor by increasing 834A1.CA78 and 994G8.CA82 allele frequencies to 0.4 each (2.3). Since linkage analysis for this family did not obtain a significant lod score, it is possible that deafness in family PKSR7 is not actually linked to the DFNB8/B10 locus, and the homozygosity observed in this region is incidental. However, it was previously estimated that 60% of all Pakistani marriages are consanguineous.17 It is unknown whether I.15 is related to family PKSR7. Recalculation of the lod score under the assumption that I.14 and I.15 are first cousins resulted in a maximum two point lod score of 4.2. Thus, this family may have a mutation in a regulatory element of theTMPRSS3 gene, or alternatively it might carry a mutation in a different gene located in the same region.
R109W and C194F were not found in any of the normal control Pakistani or Muslim Indian chromosomes tested, while the C407R mutation was found in 1 of 200 Muslim Indian control chromosomes. In addition, C407R was found in two of our DFNB8/B10 families. Taken together, these findings imply that the carrier frequency for C407R in the Muslim Indian-Pakistani population is higher than for the otherTMPRSS3 mutations described in this manuscript. Moreover, finding of pathogenicTMPRSS3 mutations in four out of a total of 159 Pakistani families segregating profound congenital recessive deafness indicates that TMPRSS3 mutations contribute to approximately 2.5% of the recessively inherited deafness cases in the Pakistani population (95% confidence interval 0.7-6.3). This is a significant contribution, considering the high level of genetic heterogeneity of recessively inherited deafness in general, and in this population in particular.
To estimate the contribution of TMPRSS3mutations to genetic deafness in North America, we sequenced allTMPRSS3 coding exons (exons 2-13) in a panel of 64 deaf North American subjects. DNA samples of North American deaf subjects were obtained from the National DNA Repository for Research on Deafness (NDRRD) based at the Virginia Commonwealth University. None of the subjects included in the screened panel had any obvious syndromic or environmental cause for their deafness based on their medical history. Subjects with identifiable mutations inGJB2 (Cx26) (based on complete sequencing of exon 2 of the GJB2gene) or with known mitochondrial deafness related mutations were also excluded from this panel. The panel includes subjects from both multiplex (25) and simplex (38) families, with the following ethnic origins: 54 whites, two African-Americans, two Hispanic, three Asians, one Indian, and two of unknown origin. We identified several common SNPs in both coding and non-coding sequences of theTMPRSS3 gene (table 3). Interestingly, noTMPRSS3 mutations were detected. Comparison of this finding to another deafness related gene,GJB2 (Cx26), which accounts for approximately 20% of non-syndromic, recessive hearing loss,4 indicates thatTMPRSS3 is not a major contributor to genetic deafness in North America. Direct comparison between our findings in the North American panel and in the Pakistani population is difficult, since some of the subjects included in the panel are sporadic cases. This is unlike the Pakistani families we analysed, which are all consanguineous families with multiple affected subjects. Thus it is possible that some of the North American deaf analysed do not actually have a genetic cause for their deafness, while deafness in the Pakistani families is most probably genetic.
The five TMPRSS3 deafness related mutations identified to date include missense (current report), splice site, and insertion mutations,7 which affect various domains of the TMPRSS3 protein (fig 3). Finding ofTMPRSS3 mutations in several families segregating autosomal recessive deafness DFNB8/B10 indicates that TMPRSS3 is essential for hearing. It remains to be determined whether TMPRSS3 has serine protease activity, as suggested by its conserved serine protease domain. Further work is needed to identify its substrates and the role it plays in the hearing process.
The first three authors contributed equally to this work. We are grateful for the participation of many families in Pakistan in this study. We thank Tenesha Smith, Nikki Liburd, Colette Rossier, and Nobuyoshi Shimizu for their contribution, and Suzan Sullivan and Dennis Drayna for critical reading of the manuscript. Work at the University of Geneva Medical School was supported by grants from the Swiss FNRS 31.57149.99 and 4038-52845.1, the OFES/EU 98-3039, and funds from the University and Cantonal Hospital of Geneva. Work in Pakistan was supported by University Grants Commission (UGC). Work at the Keio University School of Medicine was supported in part by a Grant in Aid for Scientific Research and a Fund for “Research for the Future” Program from the Japan Society for the Promotion of Science (JSPS). This work was supported by intramural funds from the National Institute on Deafness and Other Communication Disorders, National Institutes of Health (1 Z01 DC 00035-04 and 1 Z01 DC 00039-04) to ERW and TBF.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.