Non-syndromic, autosomal-recessive deafness

Transcription

1 Clin Genet 2006: 69: Printed in Singapore. All rights reserved Review Non-syndromic, autosomal-recessive deafness # 2006 The Authors Journal compilation # 2006BlackwellMunksgaard CLINICAL GENETICS doi: /j x Petersen MB, Willems PJ. Non-syndromic, autosomal-recessive deafness. Clin Genet 2006: 69: # Blackwell Munksgaard, 2006 Non-syndromic deafness is a paradigm of genetic heterogeneity with 85 loci and 39 nuclear disease genes reported so far. Autosomal-recessive genes are responsible for about 80% of the cases of hereditary non-syndromic deafness of pre-lingual onset with 23 different genes identified to date. In the present article, we review these 23 genes, their function, and their contribution to genetic deafness in different populations. The wide range of functions of these DFNB genes reflects the heterogeneity of the genes involved in hearing and hearing loss. Several of these genes are involved in both recessive and dominant deafness, or in both non-syndromic and syndromic deafness. Mutations in the GJB2 gene encoding connexin 26 are responsible for as much as 50% of pre-lingual, recessive deafness. By contrast, mutations in most of the other DFNB genes have so far been detected in only a small number of families, and their contribution to deafness on a population scale might therefore be limited. Identification of all genes involved in hereditary hearing loss will help in our understanding of the basic mechanisms underlying normal hearing, in early diagnosis and therapy. MB Petersen a and PJ Willems b a Department of Genetics, Institute of Child Health, Athens, Greece, and b GENDIA, Antwerp, Belgium Key words: autosomal recessive deafness genes non-syndromic Corresponding author: Dr Michael B. Petersen, Department of Genetics, Institute of Child Health, Aghia Sophia Children s Hospital, GR Athens, Greece. Tel.: þ ; fax: þ ; inchildh@otenet.gr Received 6 January 2006, revised and accepted for publication 6 March 2006 Approximately one in 1000 children is affected by severe or profound hearing loss at birth or during early childhood, which is defined as pre-lingual deafness (1). More than 50% of pre-lingual deafness in developed countries is attributed to monogenic defects (2). The relative proportion of genetic cases is increasing with time as public health conditions improve, leading to a decrease in the prevalence of hearing loss resulting from non-genetic causes such as infections (2). Genetic deafness is divided into syndromic forms, in which hearing loss is associated with a variety of other anomalies, and non-syndromic forms. The syndromic forms account for 30% of pre-lingual genetic deafness and include several hundred deafness syndromes (3), with the underlying genetic defect being found in about 30 of them (4, 5). In non-syndromic genetic deafness of prelingual onset, autosomal-recessive inheritance predominates (80%), but autosomal-dominant (20%), X-linked (1%), and mitochondrial (<1%) forms have also been described (1). In post-lingual, non-syndromic deafness, autosomalrecessive inheritance is very rare. The autosomalrecessive forms are usually more severe than the other forms and are almost exclusively due to cochlear defects (sensorineural deafness). Nonsyndromic deafness represents extreme genetic heterogeneity, as 85 loci have been mapped and 39 nuclear genes have been identified to date (December 2005), including 23 DFNB genes (Van Camp G, Smith RJH. Hereditary Hearing Loss. Several reviews have been published over the past few years (4 14), but with new genes being discovered at the speed of sound, we present here an updated and comprehensive list of loci and genes and review their contribution to deafness in different populations. We recently presented a review on autosomal-dominant deafness (15) and focus here on autosomalrecessive deafness. The loci and genes for non-syndromic, autosomal-recessive deafness reported to date are presented in Tables 1 and 2. A total of 23 different genes have been identified, and they are described briefly below. 371

2 Petersen and Willems Table 1. Loci and genes for autosomal-recessive, non-syndromic deafness Locus Chromosomal location Gene Reference(s) DFNB1 13q11 q12 GJB2 (16, 20) GJB6 (51) DFNB2 11q13.5 MYO7A (67, 70, 71) DFNB3 17p11.2 MYO15 (79, 82) DFNB4 7q31 SLC26A4 (86, 87) DFNB5 14q12 (213) DFNB6 3p21 TMIE (97, 98) DFNB7 9q13 q21 TMC1 (100, 101) DFNB8 21q22.3 TMPRSS3 (106, 111) DFNB9 2p23.1 OTOF (116, 118) DFNB10 21q22.3 TMPRSS3 (108, 111) DFNB11 9q13 q21 (214) DFNB12 10q21 q22 CDH23 (131, 133) DFNB13 7q34 q36 (215) DFNB14 7q31 (216) DFNB15 3q21.3 q25.2/19p13.3 p13.1 (62) DFNB16 15q15 STRC (142, 144) DFNB17 7q31 (217) DFNB18 11p15.1 USH1C (145, 218) DFNB20 11q25 qter (219) DFNB21 11q23 q25 TECTA (153) DFNB22 16p12.2 OTOA (161) DFNB23 10q21.1 PCDH15 (166) DFNB26 a 4q31 (220) DFNB27 2q23 q31 (221) DFNB29 21q22.1 CLDN14 (167) DFNB30 10p11.1 MYO3A (170) DFNB31 9q32 q34 WHRN (172, 173) DFNB32 1p22.1 p13.3 (222) DFNB33 9q34.3 (223) DFNB35 14q24.1 q24.3 (224) DFNB36 1p36.3 ESPN (179) DFNB37 6q13 MYO6 (181) DFNB38 6q26 q27 (225) DFNB39 7q11.22 q21.12 (226) DFNB40 22q11.21 q12.1 (227) DFNB42 3q13.31 q22.3 (228) DFNB44 7p14.1 q11.22 (229) DFNB46 18p11.32 p11.31 (230) DFNB48 15q23 q25.1 (231) DFNB49 5q12.3 q14.1 (232) DFNB53 6p21.3 COL11A2 (186) DFNB55 4q12 q13.2 (233) NLD b 1p35 p33 GJB3 (191) NLD b 6q21 q23.2 GJA1 (201) NLD b 7q22.1 SLC26A5 (204) a Dominant deafness modifier DFNM1 that suppresses deafness maps to 1q24. b No locus designation. GJB2 (connexin 26) DFNB1 The most important locus for non-syndromic, autosomal-recessive deafness (DFNB1) was originally assigned to chromosome 13q11 by linkage analysis in two large consanguineous Tunisian families with pre-lingual, profound deafness (16). Subsequent linkage studies in New Zealand/Australian (17) and Italian/Spanish deafness families (18) indicated that this locus is a major contributor to pre-lingual deafness in these populations. Mutations in the GJB2 gene encoding the gap junction protein connexin 26, which had been mapped to 13q11 q12 (19), were subsequently identified in three consanguineous Pakistani families with profound deafness genetically linked to 13q11 (20). The GJB2 gene was the first DFNB gene to be identified in The GJB2 gene has a single coding exon and the protein belongs to the large family of connexins having four transmembrane domains, which have been implicated in gap-junctional intercellular communication (21). Six connexin subunits bind together to form a hexamer (connexon) in the plasma membrane, and each connexon 372

3 Table 2. Twenty-three nuclear genes involved in recessive, non-syndromic deafness, categorized by function Protein Gene Non-syndromic HL Additional phenotype Cytoskeletal proteins Myosin IIIA MYO3A DFNB30 Myosin VI MYO6 DFNB37 DFNA22 Myosin VIIA MYO7A DFNB2 DFNA11 Usher 1B Myosin XV MYO15 DFNB3 Structural proteins Stereocilin STRC DFNB16 a-tectorin TECTA DFNB21 DFNA8/12 Otoancorin OTOA DFNB22 Collagen 11a2 COL11A2 DFNB53 OSMED, Stickler Ion transport proteins Connexin 26 GJB2 DFNB1 DFNA3 Vohwinkel, PPD, KID a Connexin 30 GJB6 DFNB1 DFNA3 Clouston Connexin 31 GJB3 DFNA2 EKV b Pendrin SLC26A4 DFNB4 Pendred TMC1 TMC1 DFNB7 DFNA36 Claudin 14 CLDN14 DFNB29 Prestin SLC26A5 Unknown function TMIE TMIE DFNB6 TMPRSS3 TMPRSS3 DFNB8/10 Otoferlin OTOF DFNB9 Otocadherin CDH23 DFNB12 Usher 1D Harmonin USH1C DFNB18 Usher 1C Protocadherin 15 PCDH15 DFNB23 Usher 1F Whirlin WHRN DFNB31 Espin ESPN DFNB36 a PPD, palmoplantar keratoderma; KID, keratitis-ichthyosis deafness. b EKV: erythrokeratodermia variabilis. Autosomal-recessive deafness associates with another connexon in an adjacent cell to form an intercellular channel; and multiple channels, in turn, cluster in a specialized membrane region to form a gap junction. Connexons may be homomeric (composed of identical connexin subunits) or heteromeric (composed of more than one species of connexins) (21). Connexons are important for recycling of potassium ions into the cochlear endolymph through the network of gap junctions that extends from the epithelial supporting cells to the fibrocytes of the spiral ligament and to the epithelial marginal cells of the stria vascularis (22, 23). The ion homeostasis is essential for normal hearing, and mutations in several genes encoding connexins or ion channels lead to hereditary deafness (22 24). Mutations in the GJB2 gene represent a major cause of pre-lingual, non-syndromic, recessive deafness, as they are responsible for as much as 50% of such cases in many populations (24 30). One specific mutation, the 35delG mutation, accounts for the majority of the GJB2 mutations detected in Caucasian populations and represents one of the most frequent disease mutations identified so far (26, 31). The 35delG mutation consists of a deletion of a guanine (G) in a sequence of six Gs extending from position leading to a frameshift and premature stop codon at nucleotide 38 (25, 26). A carrier frequency of the 35delG mutation as high as % has been detected in the Italian and Greek populations, implying that deafness due to homozygosity for this mutation could affect as many as one in 2500 newborns in these populations (31, 32). The carrier frequency of the 35delG mutation in southern Europe and in the Mediterranean region is thus higher than the carrier frequency of the major DF508 mutation of the CFTR gene causing cystic fibrosis (33). A south-north gradient in carrier frequency of the 35delG mutation has been reported among European countries with an average carrier frequency of 2.8% in southern Europe vs 1.3% in central and northern Europe (34). It was originally proposed that the high prevalence of this mutation was due to a mutational hotspot, but haplotype sharing in a very small chromosomal interval in patients homozygous for the 35delG mutation suggested that this mutation was derived from a common ancient founder about 500 generations 10,000 years ago (35). The contribution of the GJB2 gene and of the 35delG mutation in particular to pre-lingual deafness in different populationbased studies is shown in Table 3. The relative 373

4 Petersen and Willems Table 3. Proportion of non-syndromic, pre-lingual, sensorineural deafness due to GJB2 mutations in different populations Country Patients (n) GJB2 (%) a 35delG/35delG (%) Reference Australia (234) Austria (235) Austria (236) Brazil (237) China (238) Czech Republic (239) Denmark (240) Egypt (241) France (242) Germany (243) Germany (244) Germany (245) Ghana (246) Greece (247) India (248) Iran (249) Israel (250) Italy (251) Italy (252) Italy/Spain (31) Italy/Spain NI (253) Japan (254) Japan (37) Japan (255) Jordan (223) Korea (256) Kurdish (257) Lebanon (258) Oman (259) Pakistan (260) Palestinian (261) Sicily (262) Slovakia (263) Slovakia b (264) Spain/Cuba NI (51) Taiwan (265) Thailand (266) Turkey (56) Turkey 235 ND 20.4 (267) Turkey (268) UK (269) USA (270) USA (271) USA (28) NI, not indicated; ND, not determined. a Percentage of unrelated patients with biallelic GJB2 mutations. b Romany (Gypsy) population from Eastern Slovakia. contribution of the GJB2 gene to non-syndromic, pre-lingual deafness varies from 0 to 40% in the populations studied, demonstrating genetic heterogeneity, but some of the studies were based on small numbers of patients, and also ascertainment criteria and mutation screening methods differed between the studies. High frequencies of GJB2 mutations other than 35delG have been reported in other ethnic groups, including the 167delT mutation in Ashkenazi Jewish (36), 235delC in Japanese (37), and R143W in Africans (38), suggesting founder events (36, 39). In total, around 90 different GJB2 mutations have so far been reported to be associated with recessive, non-syndromic hearing loss (Ballana E et al. Connexins and Deafness. A genotype phenotype correlation has been demonstrated and shows that patients with two truncating mutations have significantly more severe hearing impairment than truncating/missense compound heterozygotes and that patients with two missense mutations have even less hearing impairment (40, 41). 374

5 Autosomal-recessive deafness Whereas the GJB2 gene is the major gene responsible for non-syndromic, recessive deafness in many populations, there is some controversy as to the role of GJB2 in dominant deafness (DFNA3) (42). Several heterozygous GJB2 mutations located in a particular domain of the protein (first extracellular domain) have been reported to segregate with autosomal-dominant hearing loss in a small number of families with a different phenotype, consisting of pre-lingual to late-childhood onset, mild to profound, progressive hearing loss (14, 43 45). Mutations in the GJB2 gene are also responsible for syndromic forms of hearing loss, including autosomal-dominant mutilating keratoderma with sensorineural deafness (Vohwinkel syndrome), other forms of autosomal-dominant palmoplantar keratoderma with deafness, and the ectodermal dysplasia keratitis ichtyosis deafness syndrome (46 50). Most of these GJB2 mutations are located in the first extracellular domain (14). GJB2 is therefore an example of a connexin gene (like GJB3 and GJB6) associated with both non-syndromic deafness and a genodermatosis with/without deafness. GJB6 (connexin 30) DFNB1 A study of 422 unrelated subjects from Spain and Cuba with pre-lingual, non-syndromic deafness detected biallelic GJB2 mutations in 129 patients (30.6%), while 44 patients (10.4%) had a mutation in only one GJB2 allele (51). In 22 of the 44 subjects, compound heterozygosity for both a GJB2 mutation and a 342-kb deletion truncating the GJB6 gene was detected (51). A deletion involving the GJB6 gene telomeric to GJB2 was also identified in patients from Ashkenazi Jewish and French families, who were heterozygous for a GJB2 mutation in trans to the GJB6 deletion (52, 53). A multicenter study from nine countries (54) showed that the deletion was present with high frequencies in Spain, France, Israel, and the UK ( % of all DFNB1 alleles). Haplotype analysis showed a clear founder effect for this mutation in Ashkenazi Jews and some countries in Western Europe (54). The deletion has so far not been detected in Turkish, Italian, and Austrian patients (54 56). The GJB6 gene encoding connexin 30 is located next to the connexin 26 gene, and the deletion extends from the 5 0 prime end of the GJB6 gene toward the GJB2 gene. Therefore, it is possible that the GJB6 deletion also deletes control regions of the GJB2 gene. Much more deaf individuals are heterozygous for one GJB2 mutation than hearing individuals, suggesting the existence of one or more unknown GJB2 mutations in the non-coding region of GJB2, possibly involving the same control region as the deletion. Alternatively, digenic inheritance is also possible, as both genes are expressed in the same inner ear structures and share 77% identity in amino acid sequence (57, 58). Furthermore, homozygous Gjb6-deficient mice lack the endocochlear potential and show degeneration of the cochlear sensory epithelium by cell apoptosis (59). Digenic inheritance has also been reported in some other recessive disorders such as retinitis pigmentosa and Bardet Biedl syndrome (60, 61) and has been suggested as a mechanism in a few other deafness families based on linkage studies (62, 63). A second 232-kb deletion in the DFNB1 locus was described recently (64). The deletion was also found in trans with pathogenic GJB2 mutations in affected subjects and arose by unequal homologous recombination (64). Only one dominant GJB6 mutation associated with non-syndromic deafness has been identified in a small number of patients (64). As for several other connexins, a limited number of GJB6 missense mutations cause an inherited autosomaldominant skin disorder, hidrotic ectodermal dysplasia (Clouston syndrome), which is sometimes associated with hearing loss (66). MYO7A (myosin VIIA) DFNB2 The DFNB2 locus was mapped by a genome search to 11q13.5 in a highly consanguineous family from Tunisia segregating non-syndromic, profound deafness (67). The USH1B gene responsible for an autosomal-recessive form of Usher syndrome type 1 had been assigned to the same chromosomal region and the underlying gene shown to be MYO7A encoding myosin VIIA (68). Autosomal-recessive deafness shaker-1 in the mouse (showing hyperactivity, head-tossing and circling due to vestibular dysfunction, together with dysfunction and progressive degeneration of the organ of Corti) was also shown to be due to mutations in the myosin VII gene (69). Finally, in 1997, MYO7A mutations were detected in two of eight Chinese nuclear families with non-syndromic, congenital, profound hearing loss and vestibular dysfunction (70), and in the original DFNB2-affected consanguineous family from Tunisia (67, 71). The MYO7A gene was the second DFNB gene to be associated with recessive, non-syndromic deafness (Table 1). Myosins are a family of actin-based molecular motors that use energy from hydrolysis of ATP 375

6 Petersen and Willems to generate mechanical force. Unconventional myosins have functions that are less well understood but thought to regulate intracellular membrane traffic. They are actin-based motor molecules which transduce chemical energy into the production of a force enabling them to move along actin filaments (72, 73). All myosins share a common structural organization consisting of a conserved NH 2 -terminal motor domain followed by a variable number of light-chain binding (IQ) motifs and a highly divergent tail. The MYO7A gene is a typical unconventional myosin consisting of 48 coding exons (72). Expression was found in several mouse and human tissues, including the retina and cochlea (68, 72). In the inner ear of mouse embryos, only the cochlear and vestibular sensory hair cells expressed the myosin VIIA gene, suggesting that both vestibular dysfunction and deafness might result from a defective morphogenesis of the hair cell stereocilia, the highly specific mechanical properties of which are critical for the mechanotransduction process (72). More than 50 distinct MYO7A mutations have been reported in USH1B, four different mutations have been found in DFNB2, and two in dominant deafness (DFNA11) (74 76). The mutations are dispersed throughout the MYO7A gene (77). MYO15 (myosin XV) DFNB3 In an ethnic isolate on Bali, 2% of the residents have profound, congenital, non-syndromic, sensorineural deafness segregating in an autosomalrecessive pattern (78). By homozygosity mapping in this family, the DFNB3 gene was mapped to chromosome 17p (79), which was confirmed in two Indian families (80). On the basis of conserved chromosomal synteny, the autosomalrecessive mouse deafness mutant shaker-2 was proposed as the homologue of DFNB3 (80). The shaker-2 mouse has a mutation in the Myo15 gene, an unconventional myosin gene (81). The human MYO15 gene was identified and mapped to the DFNB3 critical region, and sequence analysis of MYO15 revealed mutations in the three linked families (82). From a total of more than 100 consanguineous families segregating recessive hearing loss from Pakistan and India, seven families were found to show linkage to DFNB3 (83). Sequencing of the MYO15 gene revealed three novel homozygous mutations segregating in three of the Pakistani families, suggesting that MYO15 mutations are responsible for at least 5% of recessive, profound hearing loss in that population (83). Full-length human myosin XV is encoded by 66 exons (84), and expression studies demonstrated that MYO15 is expressed in a number of tissues in addition to the inner ear (82). In the shaker-2 mouse, the presence of very short stereocilia, and a long abnormal actin-containing structure that projects from the base of auditory hair cells, suggested that myosin XV is necessary for actin organization in hair cells (81, 85). SLC26A4 (pendrin) DFNB4 A genome search in a Druze family from Israel with pre-lingual, severe deafness found linkage to a 5-cM region on human chromosome 7q31 in an interval containing the gene mutated in Pendred syndrome (SLC26A4) (86). Affected members of this family were later found to have goitre and Pendred syndrome, and therefore this family does not represent DFNB (87). A large consanguineous family from India with congenital, profound, non-syndromic deafness showed linkage to the same region (DFNB4) (87, 88). Affected individuals were homozygous for a missense mutation involving a conserved residue in SLC26A4, representing another example where allelic variants are associated with both syndromic and non-syndromic forms of deafness (87). SLC26A4 mutations were detected in a high proportion of individuals with sensorineural hearing loss and temporal bone abnormalities ranging from dilated vestibular aqueduct to Mondini dysplasia (89 92). In a large series of pre-lingually deaf probands (n ¼ 374) and consanguineous deafness families (n ¼ 318) from East and South Asia, SLC26A4 mutations were detected in approximately 5% of the cases, providing an estimate of the prevalence of SLC26A4- associated deafness in these populations (93). SLC26A4 encodes a transmembrane protein pendrin, which functions as a transporter of chloride and iodide and is expressed in the thyroid gland, the inner ear, and the kidney (94). Functional studies showed that mutations associated with Pendred syndrome have complete loss of chloride and iodide transport, while mutant alleles in patients with DFNB4 are able to transport both iodide and chloride, although at a much lower level than wild-type pendrin (90). To explain the associated temporal bone abnormalities, it has been hypothesized that SLC26A4 controls fluid homeostasis in the membranous labyrinth, which in turn affects development of the bony labyrinth (91). Enlarged vestibular aqueduct (EVA) is a frequent symptom in patients with Pendred syndrome 376

7 Autosomal-recessive deafness (and in some other syndromes such as BOR syndrome), but it can also be present as an isolated finding together with sensorineural hearing loss. In the majority of cases, one or two SLC26A4 mutations have been identified (95, 96). TMIE (transmembrane inner ear expressed gene) DFNB6 Homozygosity by descent was demonstrated in a consanguineous nuclear family from southern India with pre-lingual deafness defining the DFNB6 locus (97). Four additional small families (one Indian and three Pakistani) also showed linkage to DFNB6 (98). On the basis of conserved synteny between distal mouse chromosome 9 and human chromosome 3p, the spinner strain of deaf mice was suggested to be the mouse model for human DFNB6 hearing loss (97). Mutations in the novel gene Tmie were found to be responsible for hearing loss and vestibular dysfunction in spinner mice (99), and subsequently homozygous mutations in the human ortholog TMIE were detected in the five DFNB6 families (98). The predicted TMIE protein exhibited no significant similarity to any known protein, and expression was demonstrated in many human tissues (98). The cochlear ultrastructure in spinner mice revealed irregular stereocilia bundles at the apical surfaces of both inner and outer sensory hair cells, suggesting a requirement for Tmie during maturation of hair cells, but the exact cellular location and function of the protein remain to be investigated (99). TMC1 (transmembrane cochlear-expressed gene 1) DFNB7 A locus for pre-lingual, severe to profound hearing impairment was mapped to chromosome 9q13 q21 in two consanguineous families from India defining the DFNB7 locus (100). Approximately 230 consanguineous Indian and Pakistani families segregating recessive, severe to profound, pre-lingual deafness were screened for linkage to the DFNB7 region with ascertainment of 10 more families (101). A novel gene termed TMC1 was identified within the critical interval, and seven different TMC1 mutations were identified in 11 recessive families (one of the original two Indian DFNB7 families and the 10 Indian/ Pakistani families) (101). The mutations included nonsense, frameshift, missense, genomic deletion, and splice-site mutations, all in homozygous state. TMC1 mutations seem a rather common cause of recessive deafness in India and Pakistan but have also been found in Turkish families (102). A heterozygous missense mutation in TMC1 has also been reported in a large North American family segregating autosomal-dominant, post-lingual, rapidly progressing hearing loss (DFNA36) (101). A 1.6-kb deletion encompassing exon 14 of the Tmc1 gene has been detected in the recessive deafness (dn) mouse mutant (101, 103). The TMC1 gene was shown to belong to a family of transmembrane channel-like (TMC) genes with eight paralogs (TMC1 TMC8) predicted to encode proteins with 6 10 transmembrane domains and a novel conserved 120-amino acid sequence termed the TMC domain (104). The TMC6 and TMC8 genes are identical to the EVER1 and EVER2 genes mutated in epidermodysplasia verruciformis, a rare recessive genodermatosis characterized by susceptibility to cutaneous human papilloma virus infections and associated non-melanoma skin cancers (104, 105). Expression analyses of TMC1 detected transcripts in human fetal cochlea and mouse inner and outer cochlear hair cells as well as in neurosensory epithelia of the vestibular end organs (101). TMPRSS3 (transmembrane serine protease) DFNB8/DFNB10 A large consanguineous Pakistani family showing hearing loss with onset in childhood and rapid progression to profound deafness within 4 5 years was found to be linked to chromosome 21q22.3, defining the DFNB8 locus (106, 107). An independent genome search in a large inbred Palestinian family with severe, pre-lingual deafness detected linkage also to the telomeric region of chromosome 21 (DFNB10) (108, 109). Direct sequencing of the TMPRSS3 gene (transmembrane protease, serine 3), a novel gene within the critical region (110), revealed a homozygous splice-site mutation resulting in a frameshift in the Pakistani DFNB8 family (111). The DFNB10 Palestinian family was shown to have a complex rearrangement with deletion of 8 bp and insertion of 18 complete b-satellite repeat monomers in exon 11 of the TMPRSS3 gene. Linkage analysis in 159 consanguineous Pakistani families segregating profound congenital deafness identified five families with potential linkage to the DFNB8/DFNB10 locus (112). Homozygous TMPRSS3 mutations were detected in four of these families (112). Sequencing of TMPRSS3 377

8 Petersen and Willems in 64 deaf North American patients (both familial and sporadic cases) did not reveal any mutation (112). Two homozygous missense mutations in the serine protease domain were detected in two (out of 39 studied) consanguineous Tunisian families with severe to profound, pre-lingual deafness (113). A large study of 448 unrelated Caucasian patients with severe to profound childhood deafness from Greece, Spain, Italy, and Australia identified only four mutations, indicating that TMPRSS3 mutations are not a frequent cause of pre-lingual deafness in the Caucasian population (114). The TMPRSS3 gene has 13 exons encoding transmembrane (TM), low-density-lipoprotein receptor A (LDLRA), scavenger-receptor cysteine-rich (SRCR), and serine protease domains similar to other proteases (111). The mouse ortholog of TMPRSS3 is expressed in the spiral ganglion, the cells supporting the organ of Corti and the stria vascularis, primarily in the endoplasmic reticulum membranes (115). The epithelial sodium channel EnaC, which is involved in the regulation of sodium concentration in the endolymph, was found to have a similar expression as Tmprss3 in rat inner ear. Consequently, it was suggested that EnaC was a substrate for TMPRSS3 (115). Functional expression studies in Xenopus laevis oocytes demonstrated that TMPRSS3 significantly activates ENaC, whereas TMPRSS3 missense mutations causing DFNB8/DFNB10 deafness failed to activate ENaC in this model (115). OTOF (otoferlin) DFNB9 A gene responsible for pre-lingual, profound, sensorineural deafness was mapped by a genomewide search to chromosome 2p23 p22 (DFNB9 locus) in a Sunnite consanguineous family living in an isolated village of Lebanon (116). A consanguineous kindred from Eastern Turkey (117) and three (out of 30) Lebanese families (118) also mapped to the DFNB9 locus. A homozygous mutation in a novel gene (OTOF ) encoding otoferlin was detected in the four Lebanese families (118). Mutation screening in a consanguineous family from India with severe to profound, prelingual, sensorineural hearing loss revealed a homozygous splice-site mutation (119). Other OTOF mutations reported in single families include a splice-site mutation in a consanguineous Druze family from Galilee (120), a nonsense mutation in a consanguineous family from the United Arab Emirates (121), two missense mutations in the consanguineous Turkish family previously mapped to the DFNB9 locus (117, 122), and a premature stop codon in exon 22 (Q829X mutation) in a Spanish family (123). Eleven cases (10 Spanish cases and one Cuban case) of 269 unrelated cases of recessive hearing loss (negative for GJB2 mutations) from Spain and Cuba also had the Q829X mutation (123). Haplotype analysis strongly suggested a common ancestor for the Q829X mutation, which seems responsible for about 3% of all cases of recessive, pre-lingual deafness in the Spanish population (123). OTOF mutations have also been described in non-syndromic, recessive auditory neuropathy, which is characterized by moderate to profound, sensorineural hearing loss with normal otoacoustic emissions (OAEs), indicating preserved outer hair cell function, and lack of any other detectable peripheral neuropathy and with no benefit from hearing aids (124, 125). In a large study of Spanish patients with OTOF-associated deafness, 11 of 21 subjects carrying two mutations in the OTOF gene had preserved OAEs in at least one ear (126). Consequently, genetic diagnosis of patients with profound hearing impairment, but with preserved OAEs, should be directed toward the OTOF gene. Furthermore, these findings are important for newborn screening programs for hearing impairment using OAEs as the first detection test (126). The human OTOF gene was shown to be composed of 48 coding exons predicting a 1997 amino acid protein otoferlin with alternatively spliced transcripts predicting several long isoforms (with six C2 domains) and short isoforms (three C2 domains) (119). Analysis of otoferlin revealed a transmembrane domain at the C-terminus with the rest of the protein predicted to have a cytoplasmic location and three Ca 2þ -binding C2 domains (118). A function in Ca 2þ -triggered synaptic vesicle membrane fusion was hypothesized (118). Homology was found to the Caenorhabditis elegans spermatogenesis factor fer-1 (127), to human dysferlin (encoded by DYSF), a skeletal muscle gene mutated in Miyoshi myopathy and limb-girdle muscular dystrophy type 2B (128, 129), and to human myoferlin (encoded by MYOF) (130). Strong expression in mouse tissues was observed in cochlea, vestibule and brain (118). In situ hybridization of mouse inner ear showed mainly expression in the cochlear inner hair cells and vestibular neuroepithelia. CDH23 (otocadherin) DFNB12 Linkage analysis in a consanguineous Sunni family, living in an isolated village in Syria, 378

9 Autosomal-recessive deafness defined the DFNB12 locus for pre-lingual, profound, sensorineural hearing loss to chromosome 10q21 q22 (131) in an interval overlapping with the region for Usher syndrome type 1D (USH1D) (132). Thirteen consanguineous families (Pakistani, Indian, and Turkish) segregating autosomal-recessive, profound, congenital deafness (n ¼ 7) or USH1 phenotypes (n ¼ 6) demonstrated linkage to 10q21 q22 (133). Sequencing of all 18 positional candidate genes demonstrated mutations in the novel cadherinlike gene CDH23 in nine of the 13 families (five DFNB12 families and four USH1D families) (133). Two mutations were also identified in a large, inbred Cuban USH1 family (134), and CDH23 mutations accounted for about 10% of patients in a study of 52 USH1 cases (135). In a large study including 69 families with Usher syndrome type 1D and 38 families with recessive, non-syndromic deafness from many different countries, a total of 36 different CDH23 mutations were identified in 45 families (136). Nine of 38 recessive, non-syndromic deafness families drawn from an original population of 157 non-syndromic deafness patients had a CDH23 mutation, suggesting that as much as 5% of nonsyndromic deafness is caused by mutations in the CDH23 gene (136). A clear genotype phenotype correlation was established in this study: all Usher patients except three who presented with an atypical Usher syndrome phenotype showed one or two truncating mutations, whereas all DFNB12 patients showed two missense mutations (136). The DFNB12 phenotype demonstrated a large intra- and interfamilial variation, with hearing loss ranging from moderate to profound deafness and age at diagnosis between 3 months and 6 years (136). There were no visual field deficits, but ophthalmologic examinations demonstrated asymptomatic retinitis pigmentosa-like manifestations (subnormal ERG response, subnormal fundus examination) (136). Loss of function mutations in the mouse Cdh23 ortholog cause stereocilia disorganization of the inner and outer hair cells in the waltzer mouse, thereby identifying otocadherin as a critical component for proper hair bundle formation. The CDH23 gene is a very large gene consisting of at least 70 exons encoding 3353 amino acids. Northern blot analysis of CDH23 showed a 9.5- kb transcript expressed primarily in the retina, but cochlear expression was also demonstrated by sequencing of PCR products from a human cochlear cdna library (133). The CDH23 gene belongs to the cadherin superfamily of intercellular adhesion proteins that typically have large extracellular domains (characterized by cadherin repeats that have been demonstrated to provide cell-to-cell adhesion), a membrane-spanning region, and cytoplasmic domains highly divergent among family members (137). Recent mouse data have implicated cadherin 23, harmonin (see DFNB18) and myosin VIIa (see DFNB2) in a single functional network essential to ensure the cohesion of the stereocilia of the hair bundle (138, 139). CDH23 was recently shown to be part of the tip links involved in cross-linking stereocilia (140, 141). STRC (stereocilin) DFNB16 Linkage analysis in 29 Pakistani and 12 Middle Eastern families indicated that three consanguineous families showed linkage to chromosome 15q15 q21 markers, defining the DFNB16 locus for non-syndromic, autosomal-recessive deafness (142), with a large non-consanguineous Spanish family reported later to be linked to the same region (143). With the use of a mouse inner ear cdna library, a novel gene STRC was isolated that was expressed almost exclusively in the inner ear and matched with several human genomic clones from the DFNB16 chromosomal region (144). Mutation screening in the previously described linked families (142, 143) identified two frameshift mutations and a large deletion in two of the families (144). Re-evaluation of the region of homozygosity in the two other DFNB16 families suggested that the DFNB16 locus might contain a second deafness gene (144). The STRC gene contains 29 coding exons and was shown to be tandemly duplicated with a stop codon in exon 20 in the B copy, which might represent a pseudogene (144). The deduced protein stereocilin shows no significant homology to any other known protein (144). Immunofluorescence studies demonstrated that in the mouse inner ear, stereocilin is expressed only in the sensory hair cells, with intense staining along the hair bundle composed of stereocilia (144). USH1C (harmonin) DFNB18 Pre-lingual, profound, non-syndromic, sensorineural deafness segregating in a large, consanguineous Indian family was mapped by a genome-wide search to chromosome 11p15.1 p14 (DFNB18) (145), encompassing the region for Usher syndrome type 1C (USH1C) (146). A splice-site mutation in intron 12 of the USH1C gene was detected in homozygous state in the Indian family (147). Additional mutations were later identified in other families 379

10 Petersen and Willems ( ). Harmonin is only rarely implicated in non-syndromic deafness, as mutation screening of 32 Chinese families with non-syndromic, pre-lingual deafness (150) and 16 Caucasian sib pairs from the UK (151) revealed only one mutation. The USH1C gene was shown to contain 28 exons (147) and encodes a PDZ domain-containing protein, harmonin (from the Greek word armonia, meaning assembling in a correct order ). Immunohistofluorescence detected harmonin in the sensory areas of the inner ear, especially in the cytoplasm and stereocilia of hair cells (147). Eight different transcripts were identified in mouse inner ear (147). Harmonin was shown to bind to otocadherin (see DFNB12) and to interact with myosin VIIA (see DFNB2) suggesting a functional unit underlying the formation of a coherent hair cell bundle (138, 139). Mutations of the mouse ortholog Ush1c gene were described to cause congenital deafness and severe balance deficits (deaf circler) (152). Inner ear pathology of mutant deaf circler mice revealed a progressive loss of hair cells and a secondary degeneration of spiral ganglion cells (152). The hair cell degeneration was preceded by disorganization of stereocilia as visualized by scanning electron microscopy (152). TECTA (a-tectorin) DFNB21 Linkage analysis in a consanguineous Lebanese family with pre-lingual, severe to profound hearing loss identified a novel locus for recessive deafness DFNB21 (153) in a region on chromosome 11q23 q25 encompassing the TECTA gene responsible for DFNA8/DFNA12 deafness ( ). In the DFNB21 family, a splice-site mutation in intron 9 of TECTA was found (153). Homozygous frameshift mutations in the TECTA gene were found in two consanguineous families from Iran and Pakistan (157). Mice homozygous for a targeted deletion in a-tectorin have tectorial membranes that are detached from the cochlear epithelium and lack all non-collagenous matrix (158). Dominant TECTA missense mutations have been reported in Austrian, Belgian, French, and Swedish DFNA8/DFNA12 families (156, 159, 160). These missense mutations probably have a dominant-negative effect that disrupts the structure of the tectorial membrane. TECTA encodes a-tectorin, one of the major non-collagenous extracellular matrix components of the tectorial membrane that bridges the stereocilia bundles of the sensory hair cells (158). OTOA (otoancorin) DFNB22 The mouse otoancorin gene was isolated from a cdna library prepared from the sensory epithelia of the mouse vestibular apparatus and predicted to encode a membrane-anchored protein with inner ear-specific expression (161). No homology with known protein domains was found apart from a weak sequence homology with megakaryocyte potentiating factor (MPF)/ mesothelin precursor, suggesting that otoancorin may act as an adhesion molecule (161, 162). Otoancorin in the mouse inner ear is specifically located on the apical surface of epithelial cells where they contact the overlying acellular gels (161). Otoancorin was suggested to mediate attachment of the tectorial membrane in the cochlea, and the otoconial membranes and cupulae in the vestibule. The corresponding human gene, OTOA, consists of 28 exons and maps to chromosome 16p12.2 (161). One consanguineous Palestinian family with moderate to severe, prelingual, sensorineural, recessive deafness from a collection of 200 deafness families was shown to segregate with an interval on 16p13.1 q11.2 (DFNB22), the chromosomal region containing OTOA (161). A splice-site mutation at the exon 12/intron 12 junction was found to co-segregate with the hearing impairment in this family (161). No other OTOA mutation was detected in 150 probands from families with non-syndromic, recessive deafness (mainly of Caucasian and Chinese origin), and no other family with deafness linked to the DFNB22 locus was found in an additional collection of 150 large families (from Israel and Spain) (161). This indicates that OTOA mutations are not frequent causes of deafness. PCDH15 (protocadherin 15) DFNB23 The PCDH15 gene encoding protocadherin 15 had previously been shown reponsible for Usher syndrome type 1F (163, 164), whereas recessive mutations of Pcdh15 cause deafness in the Ames waltzer (av) mouse (165). The PCDH15 gene was therefore thought to be a good candidate for non-syndromic hereditary deafness. When screening 400 families with autosomal-recessive, pre-lingual hearing loss, two families with nonsyndromic, severe to profound deafness could be linked to the PCDH15 region (DFNB23), and sequence analysis demonstrated homozygous PCDH15 missense mutations in the two families (166). Protocadherin 15 immunoreactivity has been detected in mouse retinal photoreceptors, 380

11 Autosomal-recessive deafness organ of Corti and vestibular hair cells. The immunoreactivity was seen along the length of stereocilia, in the cuticular plate, and diffusely distributed in the cytoplasm of inner and outer hair cells (166). Ames waltzer (av) mouse shows disorganized stereocilia bundles and degeneration of inner ear neuroepithelia (165). CLDN14 (claudin 14) DFNB29 The DFNB29 locus was mapped to chromosome 21q22.1 by linkage analysis in two large consanguineous Pakistani families with pre-lingual, profound deafness (167). The critical interval contained the CLDN14 gene, which was thought a good candidate gene. Sequence analysis of CLDN14 identified a homozygous single nucleotide deletion in transmembrane domain 2 in the two DFNB29 families (167). Among 100 Pakistani recessive deafness families, only these two families showed linkage to CLDN14 (167). No CLDN14 mutation was identified in a study of 60 Turkish families with autosomal-recessive, non-syndromic hearing loss (56). Cldn14 knockout mice, created by targeted deletion of Cldn14, show normal endocochlear potential but deafness due to rapid degeneration of cochlear outer hair cells, followed by slower degeneration of the inner hair cells (168). Claudins comprise a multigene family of integral membrane proteins identified as major cell adhesion molecules working at intercellular tight junctions (169). Immunofluorescence studies in the mouse at postnatal day 4 demonstrated claudin 14 expression in the inner and outer hair cell region of the organ of Corti and in the sensory epithelium of the vestibular organs (167). Between postnatal days 4 and 8, claudin 14 expression diminished in the hair cells and appeared in the supporting cells of the organ of Corti, an expression pattern coinciding with the development of the endocochlear potential (167). It has been hypothesized that the absence of claudin 14 from tight junctions in the organ of Corti leads to altered ionic permeability of the paracellular barrier of the reticular lamina and that prolonged exposure of the basolateral membranes of outer hair cells to high potassium concentrations may be the cause of cell death of hair cells (168). MYO3A (myosin IIIA) DFNB30 A genome search in a three-generation Israeli family with progressive hearing loss suggested linkage to chromosome 10p defining DFNB30 (170). The hearing loss started in the second decade of life, first affecting the high frequencies, and by age 50 was severe in high and middle frequencies and moderate at low frequencies. The MYO3A mapped within the DFNB30 region and seemed to be an excellent candidate (171), as four other myosins had been associated with hearing loss (170). Three different loss-offunction MYO3A mutations were identified to segregate with hearing loss in the Israeli family, seven members being homozygous and 11 compound heterozygous for pairs of mutant alleles. No other families with MYO3A mutations have been described. Myosin IIIA is an actin-dependent motor protein belonging to the class III unconventional myosins with 36% identity to Drosophila NINAC, mutations of which cause retinal degeneration (171). Based on the conservation of domains between NINAC and MYO3A, the actin-binding function was likely to be conserved, but the ligand of the myosin IIIA tail domain remains to be identified (170). As the motor domain of the MYO3A peptide most closely resembles human myosin VIIA, it is probable that MYO3A has a function in the mechanotransduction process, but the exact function in the mammalian ear remains to be investigated. Myosin IIIA expression had previously been demonstrated in human retina (171), and murine expression was shown in cochlea, where it was restricted to the neurosensory epithelium, especially to inner and outer hair cells (170). WHRN (whirlin) DFNB31 Homozygosity mapping in a consanguineous Palestinian family from Jordan mapped a locus (DFNB31) for pre-lingual, profound hearing impairment to chromosome 9q32 q34 (172). The murine region syntenic to the DFNB31 interval on chromosome 4 contains the locus for the recessive deafness mutant whirler (wi), which in the homozygous adult causes the shaker waltzer syndrome: deafness and circling with tossing of the head ( ). Sequence analysis of a novel gene (Whrn encoding whirlin) in this region identified a deletion of 592 bp in wi mice (173), and a homozygous nonsense mutation was detected in the original DFNB31 family (172). Screening of 150 probands from recessive deafness families (mainly of European or Chinese descent) for mutations in WHRN did not detect any anomaly (173), whereas linkage analysis in 63 Tunisian families only identified linkage to DFNB31 in one family (176), indicating that WHRN 381

12 Petersen and Willems mutations do not contribute much to recessive deafness. The human gene was shown to comprise 12 exons with three PDZ domains and one prolinerich domain, the closest related protein being harmonin (see DFNB18), which also contains three PDZ domains (173). Immunofluorescence studies in the mouse showed whirlin expression overlapping with actin staining in stereocilia at the growing ends of actin filaments (173). The findings suggest that whirlin acts by controlling actin polymerization and membrane growth of stereocilia (173, 177). The elongation of stereocilia was found to be defective in wi homozygotes with eventual degeneration of both inner and outer hair cells (178). ESPN (espin) DFNB36 The DFNB36 locus was mapped by a genomewide search in two consanguineous Pakistani families segregating pre-lingual, profound hearing loss and vestibular areflexia (179). The linkage interval on chromosome 1p36.3 contained ESPN (espin), a gene known to cause deafness and vestibular dysfunction in the jerker mouse (180). Two homozygous ESPN frameshift mutations were detected in the two Pakistani families (179). The human ESPN gene consists of 13 exons and was predicted to encode an 854 amino acid protein with eight ankyrin repeats, two prolinerich regions, an actin-binding WH2 domain, and a coiled coil domain important for actin bundling (179). The espins are actin-bundling proteins. In both the cochlea and vestibule of the mouse inner ear, espin was localized mostly to the stereocilia (180). Espin was absent from the stereocilia of jerker mice, eventually leading to complete loss of all sensory hair cells (180). MYO6 (myosin VI) DFNB37 A genome-wide scan in a large consanguineous Pakistani family with profound, sensorineural, pre-lingual hearing loss revealed co-segregation with 6q13 and defined the DFNB37 locus (181). In addition to deafness, vestibular dysfunction and mild facial dysmorphism were present in this family. Two Pakistani families with nonsyndromic deafness were later also linked to the DFNB37 locus (181). The linkage region included the MYO6 gene encoding myosin VI. MYO6 was previously found to be responsible for an autosomal-dominant form of post-lingual progressive deafness in one Italian kindred (DFNA22) (182), whereas mouse Myo6 is responsible for deafness and vestibular dysfunction in the Snell s waltzer (sv) mouse (183). Mutation screening of the MYO6 gene identified homozygous mutations in the three Pakistani families (181). A missense mutation in the highly conserved motor domain of MYO6 has been associated with autosomal-dominant sensorineural hearing loss co-segregating with hypertrophic cardiomyopathy and prolongation of the QT interval in one pedigree (184). The cardiac symptoms were mild or absent in most affected family members and might escape detection in other pedigrees with MYO6 deafness (184). The MYO6 gene is an unconventional myosin highly expressed at the base of the stereocilia of the inner and outer hair cells (185). In the Snell s waltzer mouse, giant stereocilia are seen along with degeneration of hair cells (183). COL11A2 (collagen 11a2) DFNB53 A genome-wide scan carried out in a consanguineous Iranian family identified a novel locus on 6p21.3 (DFNB53) (186). The clinical findings were consistent with non-syndromic, pre-lingual, profound hearing loss. The linked interval contained the COL11A2 gene associated with DFNA13 dominant non-syndromic deafness in an American and a Dutch family (187). Mutation screening of the COL11A2 gene in the Iranian family identified a missense mutation in exon 21 in homozygous state in affected individuals (186). Several other COL11A2 mutations have been reported, including recessive mutations in otospondylomegaepiphyseal dysplasia (OSMED) (188) and dominant mutations in non-ocular Stickler syndrome (189), all syndromic types of hearing loss. A phenotype genotype comparison suggests that mutation type and location are critical determinants in defining the phenotype of COL11A2-associated diseases (186). GJB3 (connexin 31) The GJB3 gene encoding the gap junction protein connexin 31 was cloned and mapped to chromosome 1p35 p33 and thought to be a good candidate for hereditary hearing impairment, as other connexins have also been implicated in deafness (190). Currently, none of the known DFNB loci maps to the 1p35 p33 chromosomal region. Mutation screening in 25 Chinese families with recessive deafness identified two small families 382