Modifier controls severity of a novel dominant low-frequency MyosinVIIA (MYO7A) auditory mutation
- 1V.M. Bloedel Hearing Research Center, Otolaryngology–HNS Department, University of Washington, Seattle, WA
- 2Otolarynogology–HNS Department, University of Washington, Seattle, WA
- Correspondence to: Dr V Street Box 357923, V.M. Bloedel Hearing Research Center, Otolaryngology – HNS Department, University of Washington, Seattle, WA 98195;
Hearing impairment is a common sensory deficit with both genetic and environmental aetiologies. Pre-lingual hearing loss affects approximately 1 in every 1000 children in the United States with a genetic basis in about 50% of the cases.1 An additional 1 per 1000 individuals experience auditory deficits prior to adulthood.1 Large pedigrees with monogenic non-syndromic hearing impairment have allowed genetic mapping of at least 80 chromosomal locations harbouring auditory-related deafness (DFN) loci with the identification of over 30 DFN genes.2 The DFN inheritance pattern is designated by A dominant, B recessive, and M modifier, with a number following A, B, or M indicating the relative order in which the locus was identified. For example, DFNA1 represents the first dominantly inherited deafness locus mapped in humans.
Several myosin gene products have been implicated in hearing loss. Myosins constitute a family of motor proteins playing roles in diverse biological events such as muscle contraction, cell adhesion, organelle translocation, cytokinesis, and cell movement.3,4 Myosin proteins share a conserved globular head domain with actin- and ATP-binding sites joined to varied amino and carboxy-terminus regions that determine the unique cellular role of each motor protein. For example, the Drosphila phototransduction NINAC (neither inactivation nor afterpotential C) protein couples a myosin head with a N-terminal kinase domain.5 The first myosin discovered, myosin-II, was isolated from muscle and shown to form myosin filaments by association of bipolar tail domains.3 Myosin-II has historically been referred to as a conventional myosin. Unconventional myosins do not form filaments and constitute a large rapidly-growing group of proteins.6 Mutation of one conventional myosin, non-muscle myosin, heavy polypeptide 9 (MYH9),7 and five unconventional myosins, myosin IA (MYO1A),8 myosin IIIA (MYO3A),9 myosin VI (MYO6),10 myosin VIIA (MYO7A),11 and myosin XV (MYO15A)12 are known to cause auditory dysfunction.
Over 80 mutations in MYO7A have been identified,2,13 most leading to a diagnosis of Usher syndrome type 1B (USH1B), a disease characterised by profound, congenital, sensorineural deafness with progressive retinitis pigmentosa leading to visual loss and vestibular areflexia. Four non-syndromic families with MYO7A alterations have also been reported; three with a recessive (DFNB2)14,15 and one with a dominant (DFNA11)16 inheritance pattern. In this article we describe a large American pedigree [referred to as HL2, for hearing loss family 2] experiencing low-frequency hearing loss associated with a novel heterozygous mutation in MYO7A, representing a second DFNA11 family. The severity with which this disease mutation manifests in the auditory system appears to be influenced by a genetic modifier.
MATERIALS AND METHODS
Research subjects and controls
Under a protocol of informed consent approved by the institutional review board (IRB) of the University of Washington, Seattle, 5 ml of blood were obtained by venipuncture for high molecular weight DNA isolation using standard techniques. Control DNA samples were taken from a predominantly Caucasian population. Half of these controls underwent pure-tone air conduction screening at 250, 500, 1000, 2000, 40000, and 8000 Hz (performed by V Street, under the guidance of K Kiemele) to ensure that the control subjects’ hearing levels were within normal ranges for their age.17 The HL2 pedigree is of English decent. Male-to-male transmission is observed confirming autosomal dominant inheritance. Audiologic evaluations were either conducted as part of this study at a clinic near the HL2 family member or previous test results were released to the study. The pure-tone air and bone conduction thresholds for the HL2 family members were established by certified audiologists in sound-attenuated booths. Symmetrical hearing loss was detected in all affected HL2 family members. For clarity, only right ear responses are plotted on the audiograms in fig 1. Audiometry data for a normal hearing individual of similar age to the research subject are included on each audiogram plot.17 Immittance testing evaluated middle ear pressures, ear canal volumes, and tympanic membrane mobility. Each study participant completed a hearing and balance questionnaire to assess their medical and noise-exposure history. For the fundus examination, the pupil was dilated and the retina was examined by a certified ophthalmologist.
Mutations within myosin molecules can lead to syndromic and non-syndromic hearing impairment.
We describe the genetic mapping of progressive sensorineural hearing loss first affecting low-frequency auditory thresholds within a large human pedigree to chromosome 11q13.5. A maximal pairwise LOD score of 7.23 was obtained with marker D11S4207.
We identified a myosin VIIA (MYO7A) G2164C mutation that co-segregates with auditory dysfunction in the pedigree. The mutation results in a predicted G722R substitution at an evolutionarily conserved glycine residue in the MYO7A head domain.
This pedigree represents a second DFNA11 family.
The clinical severity of the G2164C mutation varies between individuals in different family branches with similar medical and noise-exposure histories, indicating involvement of a genetic modifier.
Single nucleotide polymorphism (SNP) analysis in the family suggests that amino acid changes within the opening-reading-frame (ORF) of MYO7A commonly found in the general population and within the GJB2 (connexin 26, Cx26) ORF are not responsible for the marked differences in the clinical manifestation of the G2164C mutation.
Genotyping and linkage analysis
A genome-wide scan (ABI PRISM Linkage Mapping Set Version 2, PE Biosystems) was performed in pedigree HL2 with informative microsatellite markers18 spaced at approximately 10 centimorgan (cM) intervals. Polymerase chain reaction (PCR) conditions were performed according to the manufacturer’s recommendation (PE Biosystems). PCR products were multiplexed and separated by capillary electrophoresis using an ABI PRISM 310 Genetic Analyzer (PE Biosystems). Microsatellite allele data were analysed with GENESCAN version 3.1.2 and GENOTYPER version 2.0 (PE Biosystems). Amplification products generated with the ABI panel sets were sized according to CEPH (Centre d’Etude du Polymorphisme Humain) control DNA (1347-02) and assigned allele numbers consistent with the CEPH designations (http://www.cephb.fr). New alleles not reported in the CEPH database include a 218 basepair (bp) product (designated allele 9 in our study) for marker D11S905, a 252, 261, 270 bp product (designated allele 15, 16, 17) for marker D11S4207, and a 175 bp product (designated allele 15) for marker D11S4175. Under a model of autosomal dominant inheritance, pairwise and maximal LOD scores were calculated using the LINKAGE computer program package19 version 5.1. Equal recombination rates in males and females were assumed and a DFN gene frequency of 0.0005 was used for the calculations. Equal microsatellite marker allele frequencies were employed in the analyses. Haplotypes were constructed based on known marker orders.18 The SimWalk2 program version 2.83 was employed to estimate the multipoint LOD scores.20
Mutation and polymorphism detection
Previously published PCR primers21 were used to amplify the 49 (48 coding) MYO7A exons and immediate flanking genomic intron sequences with exceptions noted in table 1.
The previously published exon 42 primers span a region that in the HL2 family contained a C/T SNP (67 bp 3′ of exon 42 ORF). This SNP prevented the PCR amplification of both MYO7A alleles, therefore a new set of primers were designed for exon 42. PCR incubation mixture and thermocycling parameters were performed as noted previously.22 The GJB2 exon 2 was amplified as described previously.23 PCR amplification products were separated by gel electrophoresis on 2% SeaKem LE agarose gels (BioWhittaker Molecular Applications), purified using a QIAquick Gel Extraction Kit (Qiagen) and sequenced in both directions using the Big Dye Terminator Cycle Sequencing Ready Reaction Kit (PE Biosystems). Sequencing reactions were purified on AutoSeq G50 columns (Amersham Pharmacia Biotech) prior to analysis on an ABI 377 sequencer (PE Biosystems). Electrophoregrams were analysed using the DNASTAR software package.
Protein evolution and mutation modeling
To determine the evolution rate of the G722 residue we used the ConSurf server (http://bioinfo.tau.ac.il).25 The Protein Data Bank (PDB) identification number for MyoIE24 (1lkx) was entered into ConSurf. The server selected 50 myosin homologues from the PSI-BLAST database to calculate the site specific evolution rate. To model the structural impact of the G722R substitution, we utilised the DeepView (Swiss-PdbViewer) program26,27 version 3.7 (http://www.expasy.org/spdbv). The MyoIE24 PDB template was loaded into DeepView, the appropriate glycine residue was mutated to arginine, and the program was allowed to select the best rotamer of arginine. The image was rendered with POV-Ray version 3.5.
Progressive non-syndromic low-frequency hearing loss
The hearing loss onset in family HL2 is noticed generally between 20–30 years of age. Audiologic evaluation of the family members demonstrates normal immitance testing and bone conduction values that equal the air conduction measurements, suggesting sensorineural hearing impairment (data not shown). Audiograms from six affected individuals (50, 63, 65, 66, 68, 69) documenting hearing status at <40 years of age indicate that low-frequency loss is consistently present at these ages (fig 1). Individuals 50, 63, 66, and 69 show similar audiologic profiles with loss of sensitivity at both low- and high-frequency pure tone regions with relative sparing of the mid-frequencies. Family members 65 and 68 show similar S-shaped profiles with greater sensitivity at the higher frequencies compared with the low- and mid-frequency ranges. Serial audiograms from individuals 50, 66, and 69 clearly demonstrate the progressive nature of the HL2 hearing loss (fig 1). Audiograms for family member 50 are shown (right ear only) covering a 21-year span at ages 36 (red-line), 42 (blue-line), 53 (green-line), and 57 (orange-line). Between the ages of 36 and 57 a 30–55 dB worsening in pure tone thresholds is observed across the frequency ranges examined resulting in a flat moderately severe sensorineural hearing loss at 57 years of age. Audiograms for male 66 are shown over a 15-year span at ages 25 (red-line), 37 (blue-line), and 40 (green-line). Between the ages of 25 and 37 a 30-35 dB worsening in pure tone thresholds at 2000 and 4000 Hz is noted resulting in a flat moderately severe sensorineural hearing loss at 37 years of age. Female 69 demonstrates a 20–25 dB worsening of pure tone thresholds at 2000, 4000, and 8000 Hz over a 4-year time span. The documented hearing loss progression in these three individuals can not be accounted for by exposure to environmental noise or ototoxic drugs. Affected individual 60 was exposed to considerable military-related noise exposure for over 15 years and displays profound sensorineural hearing loss across all frequencies (fig 1). He received a unilateral cochlear implant at 69 years of age which has enhanced greatly his communication abilities. Affected individual 59 was exposed to recreational noise and demonstrates an audiogram with a downward sloping configuration.
The HL2 family members do not complain of night-blindness or vestibular problems. Two family members experiencing moderate to severe hearing loss (individuals 48 and 50, ages 66 and 60, respectively) underwent a fundus examination with normal retinal findings (data not shown). Given the absence of detectable ophthalmological dysfunction using these approaches, the HL2 family was considered non-syndromic.
Linkage to chromosome 11q
Prior to performing microsatellite marker analysis, the linkage detection power of the HL2 pedigree was simulated using 400 replicates with the SLINK program28,29 at a recombination frequency θ = 0.00 between the disease and marker loci assuming a frequency of 0.00005 for the disease locus and five equally frequent alleles at the marker locus. Under these conditions, a linked marker could generate a maximal pairwise LOD score (Zmax) of 8.3 and an average LOD score of 5.2 indicating that the HL2 pedigree is powerful enough to detect significant linkage based on the traditional LOD-score-of-3 criterion. First, linkage to two previously identified non-syndromic low-frequency hearing loss DFN loci, DFNA130 and DFNA6/DFNA14,31,32 was excluded (LOD = −2.0) (data not shown). A genome-wide scan to detect linkage was then initiated in the HL2 pedigree. A maximal pairwise LOD score of 7.23 was obtained with marker D11S4207 establishing linkage to chromosome 11 (table 2). Significant pairwise LOD scores Zmax⩾3) were obtained with multiple markers from chromosome 11q13.4–q14.3 (table 2). A maximal multipoint LOD score of 10.2 for marker D11S937 was calculated in the HL2 pedigree. Haplotype analysis indicated that four markers D11S4207, D11S937, D11S901, and D11S4147 were non-recombinant with the auditory trait in family HL2 (fig 1). Individual 53 displayed a crossover between markers D11S987 and D11S4207, defining D11S987 as a proximal flanking marker. Individual 68 displayed a crossover between markers D11S4147 and D11S4175 defining D11S4175 as a distal flanking marker. This haplotype analysis suggests that the HL2 DFN gene lies between markers D11S987 and D11S4175 within a 24 cM interval that contains MYO7A located at 11q13.5.
Novel mutation identified in conserved MYO7A head domain residue
Given the coincident map position of MYO7A and the HL2 gene, the 49 exons of MYO7A were compared between two affected (persons 48, 50) and one unaffected (person 64) HL2 family members. The sequence comparison revealed a heterozygous G-to-C transversion at position 2164 (fig 2A) in exon 17 leading to a G722R substitution within the MYO7A head domain (fig 2B). The heterozygous nucleotide change is consistent with the dominant HL2 hearing loss inheritance pattern. Direct nucleotide sequence analysis demonstrated that the mutation faithfully co-segregated with hearing loss in the HL2 family. The mutation was absent in 200 unrelated control chromosomes, supporting the hypothesis that it represents a causative mutation, not a rare polymorphism. The G722 residue is conserved across eukaryotic myosin head domains (fig 2C). The 3D structure from the recently crystallized unconventional Dictyosteliumdiscordeum myosin-IE (MyoIE) motor domain24 was employed to perform a ConSurf calculation25 to determine the evolution rate of the homologous glycine residue. The ConSurf analysis used 50 myosin homologues to generate an evolution rate score of 9 (on a scale of 1–9) indicating that this glycine residue is evolving slowly and therefore conserved maximally (fig 2D). To our knowledge, mutations at this MYO7A G722 residue have not been described previously in USH1B11,21,33–36 DFNB2,14,15 or DFNA1116 families, nor in Myo7ashaker-1 mouse alleles.37 Therefore, alteration of this glycine residue in the HL2 pedigree represents a novel mutation in MYO7A.
Structural impact of G722R mutation
The 3D structure from the recently crystallized MyoIE motor domain24 served as a structural template for MYO7A comparative modelling using the DeepView program. The MYO7A G722 residue is part of the myosin head converter domain. The converter domain interacts with a distant portion of the myosin head referred to as the relay loop.24 The relay loop communicates conformational changes caused by actin- and nucleotide-binding to the converter domain that moves along the actin filament.24 The relay loop and converter domain are coupled to each other in part by a conserved hydrophobic pocket formed by F727 (MyoIE, F678) from the converter domain with I482 (MyoIE, I431) and Y477 (MyoIE, Y426) from the relay loop (fig 3A). The model indicates that the substituted arginine side chain projects into the conserved hydrophobic pocket creating steric hindrance with neighboring amino acid residue F727 (MyoIE, F678) (fig 3B)
Genetic modifier controlling severity differences
Whereas the HL2 family members exhibit related patterns in their audiologic profiles, some members within different branches of the HL2 pedigree show marked variation in their level of hearing impairment. First cousins 63 and 65 are the two most mildly affected family members. Female 65 clearly demonstrates the affected haplotype and the G2164C mutation. Her audiologic profile demonstrates low-frequency hearing loss greater than expected for her age; however, by definition she does not have abnormal hearing (threshold elevations >25 dB). Figure 4A compares the hearing sensitivity of female 65 (at age 37) with female 68 (at age 31) from a different family branch. Both females display an S-shaped audiogram, and neither reports a significant medical or noise exposure history. Even though female 65 was 6 years older when the hearing test was performed she demonstrates markedly better hearing responses, with differences between the two females ranging from 40–50 dB in the low- and mid-frequencies to 30 dB in the high frequencies. A striking disparity is also seen between male 63 (at age 39) and 66 (at age 37), with differences between the two males ranging from 35–50 dB in the low- and mid-frequencies (fig 4B). Salient variations in the degree of hearing loss between similarly aged G2164C individuals of the same sex with comparable medical and noise exposure histories suggests the presence of a modifier gene. To investigate the single nucleotide polymorphism (SNP) that may be responsible for the modifier affect, we first examined the opposing allele of MYO7A as a nucleotide change acting in trans could impact expression of the mutated G2164C MYO7A allele, especially given that MYO7A likely forms homodimers.14 We analysed the segregation pattern of nine MYO7A SNPs (fig 2B): two intron SNPs flanking intron 17 containing the G2164C mutation; and seven opening-reading-frame (ORF) SNPs leading to amino acid changes commonly found in the general population (table 3). We also compared the GJB2 (gap junction protein, beta 2) gene exon 2 sequence with known allelic heterogeneity between DNAs from mildly versus more severely affected HL2 family members. However, none of these SNPs segregated with the phenotypic variation in the pedigree (table 3).
We have provided evidence that a G2164C mutation, resulting in a predicted G722R substitution in the MYO7A head domain, leads to progressive hearing loss first affecting low-frequency auditory thresholds. The G2164C mutation co-segregates with hearing loss in the HL2 family, is not found in 200 control chromosomes, and leads to a non-conservative G722R amino acid substitution at a slowly evolving residue. The predicted arginine substitution may disrupt the conserved hydrophobic pocket formed by amino acid side chains in the converter domain and relay loop of the myosin head. This evidence strongly supports a causative role of the MYO7A G2164C nucleotide alteration in hearing loss. However, the HL2 DFN gene maps to a large 24 cM region containing a host of genes. Further confirmation of the causative nature of the MYO7A G2164C alteration would be provided by functional studies such as sliding-filament assays indicating altered motility, creation of a G722R MYO7A mouse mutant that recapitulates the HL2 phenotype, or discovery of an unrelated hearing impaired family demonstrating alteration of the same MYO7A glycine residue. Interestingly, mutation of the homologous glycine residue (G674) in MYO1A was found in a person suffering from moderate to severe sensorineural hearing loss.8
Whereas alteration of this preserved glycine negatively impacts the auditory system, the consequences appear to be milder for the visual system. Porter and Montell39 mutagenised the DrosophilaninaC (human homolog is MYO3A) myosin head domain targeting the same glycine residue altered in the HL2 pedigree, in addition to three other highly preserved residues within this 6-amino acid region [NINAC residues 1015–1020; GXTKXF to DXAMXS] (fig 2C). The transformant line expressing the NINAC1015.4 protein containing these four homozygous substitutions displayed temperature-dependent protein degradation. At 18°C the NINAC1015.4 protein showed only a twofold decrease in levels relative to wild-type NINAC. Under these conditions, the NINAC1015.4 transformants displayed a normal ERG and only minor retinal degeneration.39 At 29°C the NINAC1015.4 protein concentration was less than 5%, and the NINAC1015.4 transformants exhibited an abnormal electroretinogram and severe retinal degeneration.39 When present in a stable protein, these four altered residues do not appear to have a significant impact on the function of NINAC in the Drosophila visual system. Consistent with these findings, a fundus examination on two G2164C HL2 family members affected by hearing loss yielded normal observations. The fundus exam does not exclude the possibility of subtle changes in photoreceptor function that may be detected by electroretinography. The DFNA11 family members, with a MYO7A three-amino-acid-residue (A886-K887-K888) deletion in the coiled-coil domain, were examined by electroretinography and found to produce normal photoreceptor responses.40 The opthalmological findings in the DFNA11 and HL2 families suggest that these two different dominant MYO7A mutations do not result in obvious damage to the eye.
Whereas the families share this feature in common, the HL2 and DFNA11 pedigrees manifest distinct audiologic profiles. DFNA11 family members demonstrate a gently downward sloping audiogram indicating that hearing loss is greatest in the high-frequency ranges.40 The HL2 family displays a unique audiogram contour with greater low-frequency loss suggesting that the G2164C mutation first affects hair cells in the cochlear apex. This finding is noteworthy as the vast majority of non-syndromic deafness initiate with high-frequency hearing loss.41
The HL2 audioprofile is similar to the progressive non-syndromic low frequency hearing loss characteristic of DFNA1 and DFNA6/DFNA11 caused by heterozygous mutations in the diaphanous 1 (DIAPH1)42–44 and Wolfram syndrome 1 (WFS1)45–48 genes, respectively. The dominant WFS1 mutations show allelic heterogeneity, but present a uniform clinical entity of low-frequency hearing loss.45 In contrast, the two dominant MYO7A mutations identified in the HL2 and DFNA11 pedigrees lead to dissimilar audioprofiles. The mechanism by which the predicted MYO7A G722R substitution and A886-K887-K888 deletion mutations differentially lead to pathogenesis of the cochlear apex versus base, respectively, remains to be elucidated. In concert with our findings in the HL2 pedigree, the dominant headbanger (Hdb) mouse mutant has been shown recently to demonstrate low-frequency hearing loss associated with a novel heterozygous Myo7a missense mutation in the motor domain (Rhodes et al, personal communication).
The MYO7A G2164C mutation consistently causes low-frequency hearing loss in the HL2 pedigree; however, the clinical severity between HL2 members within two family branches is dissimilar, distinctly suggesting the presence of a genetic modifier. Based on our hearing and balance questionnaire and conversations with the HL2 family members, there is no apparent environmental influence that could result in the striking pure-tone air conduction threshold variations between different G2164C sensorineural hearing loss individuals. Previous human studies provide additional support for the role of modifier genes in the ear: intra-familial phenotypic variation has been noted for mutations in GJB249,50; mitochondrial and nuclear auditory gene interaction has been observed;51–53 and in one large family, a dominant modifier, DFNM1, that suppresses recessively inherited deafness, DFNB26, has been mapped in humans.54 In mice, several genetic modifiers controlling inherited deafness have been reported as reviewed by Riazuddin.55 Interestingly, the severity of the headbanger mouse mutant phenotype varies in different genetic background strains, suggesting interaction of the primary mutation with a modifier (Rhodes et al, personal communication).
Characterisation of genetic modifiers in humans requires exceptionally large pedigrees displaying divergent levels of hearing impairment. The HL2 pedigree is extensive with several marriages resulting in numerous offspring: couple A and B, 10 children; couple 1 and 2, 16 children; couple 3 and 4, eight children; and couple 7 and 8, 15 children. Analysis of additional family members may allow us to determine incidence of the mild auditory phenotype, mode of inheritance, and ultimately identification of the SNP(s) controlling manifestation of the MYO7A G2164C mutation. Whereas the MYO7A and GJB2 SNPs examined in this report did not segregate with the differences in pure-tone air conduction thresholds between G2164C HL2 individuals, it is possible that SNPs within the MYO7A ORF not assayed in this study could partially rescue or exacerbate the mutated allele, or SNPs within the MYO7A promoter may control the amount of normal versus mutant gene transcribed. In addition, SNPs in other gene products known to interact with MYO7A in the ear such as harmonin (USH1C) and cadherin 23 (CDH23)56 should also be considered as candidates. Identification of SNPs regulating the clinical severity of hearing loss will enhance understanding of gene product interactions within the auditory system and may provide predictive value in counselling patients carrying these mutations and genetic variations.
We are grateful to the HL2 family for their cooperation throughout this study. We thank Dr B Tempel and J Goldy for comments on this manuscript. We appreciate the cooperation of audiology clinics across the country that participated in the hearing evaluations. This work was funded by NIH grants DC04945 (VAS), DC02739 (B L T), and P30 DC04661 (V. M. Bloedel Core).