|Home | About | Journals | Submit | Contact Us | Français|
To use clinical and genetic analyses to determine the mutation causing autosomal recessive non-syndromic hearing loss (ARNSHL) segregating in two consanguineous Iranian families.
Members of each family received otologic and audiometric examination for the type and extent of hearing loss. Linkage mapping using Affymetrix 50K GeneChips and STRP analysis localized the hearing loss in both families to the DFNB3 locus. Direct sequencing of the MYO15A gene was completed on affected members of both families.
Family L-3165 segregated a novel homozygous missense mutation (c.6371G>A) that results in a p.R2124Q amino acid substitution in the myosin XVa protein. While family L-896 segregated a novel homozygous missense (c.6555C>T) mutation resulting in a p.P2073S amino acid change.
These are the first MYO15A mutations reported to cause DFNB3 sensorineural hearing loss in the Iranian population. Like other mutations located in the myosin tail homology 4 (MyTH4) domain, the p.R2124Q and p.P2073S mutations are predicted to disrupt the function of the myosin XVa protein, which is integral to the mechanosensory activity of hair cells in the inner ear.
Sensorineural hearing loss (SNHL) is the most prevalent genetic sensory defect in humans. It is estimated that globally 4 out of every 10,000 children born have profound hearing loss1. Non-syndromic SNHL accounts for ~70% of hereditary hearing loss and 80% of SNHL cases have an autosomal recessive mode of inheritance (ARNSHL). To date, 25 genes and 67 loci have been implicated in ARNSHL2.
The first report of hearing loss at the DFNB3 locus was from an isolated village in Indonesia where 2% of the population was hearing impaired3. Subsequently, it was shown that the causative mutation in this village and two unrelated families resided in the MYO15A gene4. Since these original reports, 24 additional DFNB3-causing mutations have been identified in MYO15A, which spans 71 kb on chromosome 17p11.2 and is comprised of 66 exons that encode a 3,530 amino acid protein in its longest form. (Table 1 and Fig.1)5–8.
The encoded protein is the unconventional myosin XVa. Myosin XVa is unique among unconventional myosins in that it includes a long N-terminal domain (coded by exon 2) that is alternatively spliced to generate distinct class 1 and class 2 protein isoforms10, 11. The N-terminal domain is required for normal hearing, as premature stop mutations that result in loss of this domain cause DFNB3 hearing loss6. Myosin XVa also contains domains that are conserved within the myosin protein family, including the motor domain, IQ motifs (calmodulin/myosin light chain binding), MyTh4 domains (Myosin-Tail like Homology region 4), FERM motifs (4.1 protein, Ezrin, Radixin, and Moesin), SH3 domain (Src Homology 3), and the PDZ ligand domain (Fig.1A)10, 12.
DFNB3-causing mutations have been identified in all domains, with the exception of the IQ domains in the neck region, in families from Pakistan, India, Turkey, Indonesia, Brazil and North America (Table 1). In this study, we have characterized two Iranian ARNSHL families and identified two novel DFNB3-causing mutations in MYO15A. These novel missense mutations are both located in the first MyTh4 domain and are predicted to disrupt normal myosin XVa function.
Family L-3165 is a five-generation consanguineous Iranian family segregating apparent ARNSHL (Fig. 2A). Hearing impaired persons appear in the 4th and 5th generations, consistent with autozygosity by descent. Family L-896 (Fig. 2B) is also a five-generation consanguineous Iranian family segregating apparent ARNSHL. In contrast to family L-3165, the inheritance pattern in this family appears to be pseudo-dominant.
For consenting persons in each family, audiologic testing was completed to document the degree of hearing loss. The test results for family L-3165 are displayed (Fig.3). Examination by an otolaryngologist and clinical geneticist excluded a syndromic presentation. Ten milliliters of whole blood were obtained as a DNA source. Human research institutional review boards at the Welfare Science and Rehabilitation University and the Iran University of Medical Sciences, Tehran, Iran, and the University of Iowa, Iowa City, Iowa, USA approved all procedures.
Genomic DNA from individuals I-1, I-2, II-1, II-4, II-5, II-6, II-7, II-8, II-9 in family L-3165 (Fig. 2) was genotyped for 50,000 SNPs using Affymetrix 50K XBA GeneChips at the Translational Genomics Research Institute (TGEN, Phoenix, Arizona). Genotypes were determined using the BRLMM genotyping algorithm13, 14. Genotyping data were examined with PEDSTATS15 for Mendelian inheritance errors, and MERLIN16 for errors based on inferred double recombination events between tightly linked markers.
An autosomal, genome-wide parametric linkage analysis was performed since males and females appeared equally affected. All linkage analysis was performed with MERLIN. Since the pedigree had 18 bits it was analyzed using exact multipoint linkage analysis16. A subset of 6432 single nucleotide polymorphisms (SNPs) spaced approximately 0.5 cM apart across the genome and with an average heterozygosity of 0.43 was chosen from the 50K XBA set to satisfy the linkage equilibirium requirements of the Lander-Green algorithm for linkage analysis17. The selection and assembly of the data files were performed with an in-house PERL script (linkdatagen.pl, http://bioinf.wehi.edu.au/software). An initial parametric linkage analysis was run assuming a fully penetrant autosomal recessive model with a disease allele frequency of Pr(a)=0.0001 and penetrances of pr(disease|aa)=1=pr(disease|aA), Pr(disease|AA)=0. Haplotypes inferred with Merlin were imported into Haplopainter18.
Short tandem repeat (STRP) analysis was completed using markers developed by the Cooperative Human Linkage Center and made available through Research Genetics Incorporated (Invitrogen, Carlsbad, CA) according to a published methodology1. Markers with heterozygosity > 50% were selected for this analysis.
The MYO15A gene was amplified using gene-specific primers (Table 2). Amplification reactions were cycled using a standard protocol on a GeneMate Genius thermocycler (ISC BioExpress, UT, USA). Sequencing was completed with BigDye™ v3.1 Terminator Cycle Sequencing kit (Applied Biosystems, Foster City, CA), according to the manufacturer's instructions. Sequencing products were read using an ABI 3730s sequencer (Perkin Elmer, Waltham, MA). All sequencing chromatograms were compared to published cDNA sequence; nucleotide changes were detected using Sequencher v4.5 (Gene Code Corporation, Ann Arbor, MI).
Conservation scores for each amino acid in myosin XVa when compared to 50 similar protein sequences was determined using the Conseq program (http://conseq.tau.ac.il/). Any conservation score greater than 1 standard deviation above the mean for the myosin XVa amino acid sequence was regarded as highly conserved.
Linkage analysis of family L-3165 detected one hundred and thirty nine Mendelian (0.24% error) and 465 double recombinant (0.8%) errors in the genome-wide analysis. The highest parametric LOD score of 3.7 was achieved for a region on chromosome 17 (Fig. 4). There were no other regions in the genome where the LOD score exceeded 3. The 1-LOD-drop region spanned approximately 56 cM flanked by markers SNP_A-1671362 and SNP_A-1663708 at chromosomal position 17p13.1-q24.3 (9.9–52 Mb). Haplotype analysis reveals that the critical region coincides with the region identified in the linkage analysis (Fig.5). STRP analysis revealed DFNB3 was also the likely disease locus for family L-896 (data not shown).
The critical region in both families contained the known ARNSHL locus DFNB3. Direct sequencing of the 65 exons of MYO15A identified novel homozygous missense mutations. In family 3165 a homozygous c.6371G>A mutation was identified that substitutes a glutamine for an arginine (p.R2124Q) in the first MyTH4 domain (Fig.2A and and6A).6A). In family L-896 a novel homozygous missense mutation (c.6555C>T) was also detected in the first MyTH4 domain that results in substitution of a proline for a serine (p.P2073S) (Fig. 2B and and6B6B).
The R2124 and P2073 residues are both located in the first MyTH4 domain of the MYO15A protein and are conserved between human and mouse. The substitution of an uncharged, polar glutamine for arginine results in loss of the positive charge on the arginine sidechain. The bulky, non-polar sidechain of proline is replaced with a polar serine sidechain. The c.6371G>A mutation was not observed in 42 (84 chromosomes) Iranian control or 94 (188 chromosomes) CEPH control individuals. The c.6555C>T mutation was not observed in 43 Iranian control (86 chromosomes) or 93 (186 chromosomes) CEPH control individuals.
The unconventional myosin XVa protein is required for normal auditory function as its mutation leads to SNHL at the DFNB3 locus. Two other unconventional myosin proteins, VI and VIIa, are similarly required for normal auditory function, and their mutation also leads to hearing impairment (reviewed in Brown 2008). Myosins are molecular motor proteins that drive the movement of actin filaments via ATP hydrolysis to facilitate muscle contraction, organelle trafficking, cell movement, cytokinesis, and signal transduction12.
Hair bundles located at the apex of sensory hair cells in the cochlea are responsible for the mechano-electrical transduction of sound waves in humans. Each hair bundle consists of up to 300 extracellular projections called stereocilia that contain an actin core and are arranged in rows of increasing height, forming a characteristic staircase architecture19. Extracellular lateral projections (side-links) hold the stereocilia of hair bundles together while tip-links at the apex of the stereocilia facilitate signal transduction19, 25. Myosin XVa immunoreactivity has been observed in stereocilia tips at E18.5 in the mouse, coinciding with formation of the hair bundles of the inner ear. Myosin XVa is located at the stereocilia tips following hair bundle maturation and the amount present is directly proportional to the length of the stereocilia11, 20.
Homozygous Shaker-2 (sh2) mice are deficient in myosin XVa due to a premature stop codon in the motor domain of the protein9 (Fig.1). These mice exhibit profound deafness similar to that observed in human DFNB3 patients, in addition to vestibular defects that are not present in human patients. Hair cell stereocilia in sh2 mice are correctly positioned during development, but are much shorter and do not display the characteristic staircase architecture when compared to those of wild-type mice9. Originally, it was hypothesized that the absence of myosin XVa in hair bundles resulted in a defect in mechanosensation21. However, the abnormally short stereocilia of deaf sh2 mice retain mechanosensory activity suggesting that stereocilia length and the architecture of the hair bundle are crucial in maintaining normal hearing function22.
The whirler (whi) mouse, which contains a mutation in the whirlin gene, shows profound hearing loss similar to that of shaker mice. Homozygous whirler mice have a deficiency in the whirlin protein and their stereocilia appear similar to those in shaker mice: abnormally short with no staircase formation4, 9. The tail of myosin XVa, of which the MyTH4 domains form part, interacts with whirlin, suggesting that these two proteins are partners20, 21. The PDZ ligand domain of myosin XVa and two of the three PDZ binding domains of whirlin are required for this interaction. Myosin XVa is the only myosin in mammals known to contain a PDZ domain21.
Homozygous whirler mice (lacking whirlin) have myosin XVa present at their stereocilia tips, while homozygous shaker mice (lacking myosin XVa) have neither whirlin nor myosin XVa present at their stereocilia tips11,21. Myosin XVa has been observed migrating towards the plus ends of actin filaments at stereocilia tips, confirming myosin XVa activity as a motor protein21. These data support the hypothesis that myosin XVa acts as a transport protein and whirlin is its cargo during development of hair bundles in the cochlea.11, 20, 21 This transport most likely occurs during lengthening and formation of stereocilia staircase morphology, given the localization of myosin XVa during this period20,21.
Mutations in all but one protein domain of myosin XVa have been shown to be causative of DFNB3 hearing loss. Three missense mutations in the first MyTH4 domain of the protein have been described previously: two homozygous mutations in Indian and Balinese families, and one hemizygous mutation in a family affected by Smith-Magenis syndrome from North America (Table 1). Here we report two further mutations in the first MyTH4 domain and the first DFNB3 mutations segregating in Iranian families. The arginine-to-glutamine and proline-to-serine substitutions result in the loss of positively charged and bulky, non-polar sidechains and replacement with residues containing polar sidechains. These alterations are predicted to negatively affect the function of the myosin XVa protein, perhaps by disrupting its interaction with whirlin.
Given the large number of DFNB3-causing MYO15A mutations identified in various populations (Fig.1 and Table 1) and the size of the gene, it would be useful to develop a more targeted screening approach for suspected DFNB3 ARNSHL families. In Table 3 and Figure 1 all known MYO15A mutations and the relative conservation of amino acids in each domain of the protein are depicted. A targeted sequencing strategy would focus primarily on the domains with the highest mutation rate and highest percentage of conserved amino acids. The motor domain is the most highly conserved (84.5%) domain and the site of ten mutations. Conversely, the N-terminal domain, which is unique among myosins, contains the least conserved residues and only three mutations. Therefore, sequencing of the motor domain should be a priority.
When the mutations identified in this study are included, the first MyTH4 domain contains a relatively high number of known mutations (5). This domain also contains a high percentage of conserved amino acids relative to other domains of the protein. We therefore recommend prioritizing screening of exons encoding the motor and first MyTH4 domains in ARNSHL families linked to the DFNB3 locus.
The structure of the MyTH4 domain has not been characterized. In other myosins, the MyTH4 domain has been implicated in microtubule binding as well as actin binding to the plasma membrane24. Some data suggest that the MyTH4/FERM domains in myosin XVa are required for localization of myosin XVa to stereocilia tips, although their specific function has not been elucidated25. The co-localization of myosin XVa and whirlin proteins appears essential to form the transmembrane actin microfilament assembly complex at the stereocilia tips. We speculate that the p.R2124Q and p.P2073S mutations interfere with the interaction between myosin XVa and whirlin, thereby preventing the formation of this complex which appears to be required for normal hearing.
We have identified novel mutations in the MYO15A gene in two Iranian DFNB3 families. These mutations are the first to be described in the Iranian population and both affect the first MyTH4 domain of the myosin XVa protein. The first MyTH4 and motor domains are highly conserved at the protein level and constitute common sites for DFNB3-causing mutations. As such they should be preferentially screened in DFNB3 ARNSHL families. Hearing impairment in a mouse model of DFNB3 has been rescued by supplementation of wild-type MYO15A21. Given advances in genetic therapies for hearing loss27, a similar gene replacement strategy could be developed for DFNB3 families like the ones described in this report.
The authors sincerely thank the family for their participation in this study. RJH Smith is the Sterba Hearing Research Professor, University of Iowa College of Medicine, who supported the project with National Institutes of Health (NIH) grant RO1 DCOO3544. No researchers involved in this study report a conflict of interest.
Funding: National Institutes of Health (NIH) grant RO1 DCOO3544.