Nonsyndromic hearing loss DFNA10 and a novel mutation of EYA4: Evidence for correlation of normal cardiac phenotype with truncating mutations of the Eya domain

Dominant, truncating mutations of eyes absent 4 (EYA4) on chromosome 6q23 can cause either nonsyndromic hearing loss DFNA10 or hearing loss with dilated cardiomyopathy (DCM). It has been proposed that truncations of the C‐terminal Eya domain cause DFNA10 whereas upstream truncations of the N‐terminal variable region cause hearing loss with DCM. Here we report an extended family co‐segregating autosomal dominant, postlingual‐onset, progressive, sensorineural hearing loss (SNHL) with a novel frameshift mutation, 1490insAA, of EYA4. The 1490insAA allele is predicted to encode a truncated protein with an intact N‐terminal variable region, but lacking the entire C‐terminal Eya domain. Clinical studies including electrocardiography, echocardiography, and magnetic resonance imaging (MRI) of the heart in nine affected family members revealed no DCM or associated abnormalities and confirmed their nonsyndromic phenotype. These are the first definitive cardiac evaluations of DFNA10 hearing loss to support a correlation of EYA4 mutation position with the presence or absence of DCM. These results will facilitate the counseling of patients with these phenotypes and EYA4 mutations. Published 2007 Wiley‐Liss, Inc.


INTRODUCTION
Nonsyndromic autosomal dominant sensorineural hearing loss (SNHL) is genetically heterogeneous, with more than 40 distinct loci that have been mapped. Many of these loci, termed DFNA loci, have been positionally cloned [Friedman and Griffith, 2003]. For example, mutations in the eyes absent 4 (EYA4) gene cause hearing loss at the DFNA10 locus on chromosome 6q23 . The function of EYA4 is unknown, although it has sequence similarity to the EYA1 gene on chromosome 8q13.3, which has been implicated in transcriptional regulation of inner ear development [Zhang et al., 2004]. Dominant mutations of EYA1 cause ear malformations and hearing loss as part of the branchio-oto-renal (BOR) syndrome [Abdelhak et al., 1997].
The EYA4 gene product has two distinct domains: an amino-terminal variable region followed by a carboxy-terminal Eya domain. Two different truncating EYA4 mutations have been identified in American and Belgian families segregating DFNA10 hearing loss without other reported clinical features . A third truncating mutation of EYA4 co-segregates with hearing loss in a Hungarian family for which the presence or absence of syndromic associations was not mentioned [Pfister et al., 2002]. These three mutations of EYA4 are 1468insAA (American family), 2200C > T (a nonsense mutation in the Belgian family), and 1558insTTTG (Hungarian family), all of which partially or completely delete the Eya domain. The only other reported EYA4 mutation (E193) is a 4,846bp genomic deletion that results in loss of the Eya domain as well as part of the variable region [Schonberger et al., 2005]. E193 co-segregates with hearing loss and dilated cardiomyopathy (DCM) in a single large family, suggesting that truncations affecting only the Eya domain cause SNHL alone whereas truncations affecting the variable region lead to SNHL and DCM [Schonberger et al., 2005]. In vitro studies of mutant EYA4 proteins provide functional evidence for this correlation of mutation position with cardiac phenotype [Schonberger et al., 2005].
However, cardiomyopathy displays age-related penetrance that becomes symptomatic later than SNHL in the family segregating E193 as well as in a second family segregating SNHL and cardiomyopathy associated with a MYO6 mutation [Mohiddin et al., 2004]. Therefore a cardiomyopathy phenotype may have been overlooked in the DFNA10 families [O'Neill et al., 1996;Verhoeven et al., 2000;Pfister et al., 2002] since they were described before the association of EYA4 with DCM had been established [Schonberger et al., 2005]. Here we report a family cosegregating dominant hearing loss and a novel truncating mutation of EYA4 that deletes the Eya domain. Comprehensive clinical evaluations confirm a nonsyndromic DFNA10 phenotype and support the correlation of cardiac phenotype with EYA4 mutation position.

Subjects
The study subjects were 8 male and 11 female members of LMG265, a North American Caucasian family of mixed European ancestry (Fig. 1). This study was approved by an Institutional Review Board at the National Institutes of Health (National Institute of short tandem repeat marker genotypes at the DFNA10 locus on chromosome 6q23. The 19 participating family members are identified with bold Arabic numerals. Individual IV-10 was anamnestically reported to be affected but did not undergo genotype analysis. DFNA10 HEARING LOSS AND EYA4 MUTATIONS of Neurological Disorders and Stroke and the National Institute on Deafness and Other Communication Disorders). Written informed consent was obtained from all subjects and parents of minor subjects.

Genotype Analysis
Genomic DNA was extracted from peripheral blood samples using PureGene (Gentra Systems, Minneapolis, MN). Genotypes of microsatellite markers at known DFNA loci (see the online Supplementary Table I at http://www.interscience. wiley.com/jpages/1552-4825/suppmat/index.html) were analyzed and all 20 coding exons and flanking intronic regions of EYA4 were PCR-amplified for bidirectional nucleotide sequence analysis as described [Bork et al., 2001]. Genomic DNA fragments were PCR-amplified with primer pairs shown in Supplementary Table II (see the online Supplementary Table II at http://www.interscience.wiley. com/jpages/1552-4825/suppmat/index.html), KOD Hot Start DNA Polymerase (Novagen, San Diego, CA), and the following cycling parameters: initial denaturation at 958C for 2 min, 35 step-cycles of denaturation at 958C for 15 sec, annealing at 558C for 30 sec, extension at 688C for 1.5 min, with a final extension at 688C for 5 min. Exon 12 amplification products were subcloned and sequenced in unaffected and affected subjects. Ninety-six genomic DNA samples from unrelated, ethnically matched (Caucasian) controls were obtained from Coriell Cell Repositories (Camden, NJ).

Phenotype Analysis
We defined affected auditory phenotype status as an air-conduction pure-tone (1, 2, and 4 kHz) threshold average greater than 30 dB HL on the subject's most recent audiogram. Nine subjects with the EYA4 mutation, 1490insAA, underwent evaluations at the NIH Clinical Center including general medical and developmental history interviews and physical examinations, interviews and examinations by cardiology consultants, pure-tone and speech audiometry, middle ear immittance testing, videonystagmography including caloric testing, magnetic resonance imaging (MRI) of the inner ears and temporal bones, genetic counseling, electrocardiography, echocardiography, chest X-ray, and cardiac MRI.

Statistical Analysis
We estimated age-related progression of hearing loss as the slope of a simple linear regression of hearing threshold and age using statistiXL version 1.6 (downloaded from http://www.statistixl.com/) with the Windows TM version of Microsoft Excel TM . The same software was used to compare regression slopes and intercepts among genders and families by analysis of variance (ANOVA).

LMG265 Auditory Phenotype
LMG265 is a North American Caucasian family of mixed European ancestry segregating autosomal dominant SNHL (Fig. 1). There are no extra-auditory phenotypes co-segregating with the SNHL. The SNHL was anamnestically reported to start during the second to fourth decade of life, primarily beginning in the middle and high frequencies, and progressing to moderate to severe levels affecting the entire frequency range (Fig. 2). The SNHL at 0.5, 1, 2, and 4 kHz was more severe in males compared to females (P < 0.05) but there were no differences (P > 0.05) in the rates of progression (Fig. 4A). Word recognition scores (not shown) are within expected ranges for the degree of hearing loss when the hearing loss is of cochlear origin [Yellin et al., 1989]. Normal tympanometry and acoustic reflex test results are indicative of normal middle ear function (not shown).
In comparison with the Belgian DFNA10 carriers of 2200C > T , the 1490insAA carriers had slower progression of hearing loss (P < 0.05) (Fig. 4B). We were unable to extract accurate pure-tone threshold average data for the other published families segregating EYA4 mutations [Schonberger et al., 2000;Verstreken et al., 2000;De Leenheer et al., 2001].
We performed videonystagmography on six affected individuals (IV-1, IV-2, IV-3, IV-6, V-1, V-4) who reported episodes of dizziness and one affected individual (VI-1) who denied any history of dizziness. Subjects IV-1, IV-3, and V-1 had normal findings on videonystagmography. Subject IV-2 reported two remote episodes of brief, self-limiting vertigo while supine. He had a positive response to the Dix-Hallpike maneuver with the right ear down, indicative of benign paroxysmal positional vertigo affecting the right ear. Caloric testing of IV-6 elicited a borderline reduced left-sided response to caloric irrigations and a right directional preponderance suggestive of a left-sided peripheral vestibular weakness. Both V-4 and VI-1 had borderline reduced left-sided responses to caloric irrigations suggestive of a potential left peripheral vestibular pathology.
Magnetic resonance images of the temporal bones of four affected individuals (IV-1, IV-3, IV-6, V-4) revealed no structural malformations (not shown).   (Fig. 1) and no other DFNA loci (not shown). LMG265 family member IV-11 was excluded from the initial linkage screen because he had a pure-tone average of 28.3 dB HL at 55 years of age (Fig. 2) and his phenotype status could not confidently be assigned. Since mutations of EYA4 had previously been shown FIG. 4. Comparison of EYA4 mutant auditory phenotypes. A: Pure-tone air conduction thresholds for the better hearing ears of affected male (black squares) and female (gray circles) family members of LMG265 are plotted against age for each stimulus frequency (indicated above each graph). Females had better hearing than males at 0.5, 1, 2, and 4 kHz (P < 0.05). B: Pure-tone (1, 2, and 4 kHz) threshold averages plotted against age for individuals carrying the DFNA10 mutations 1490insAA (this study) and 2200C > T . The 1490insAA carriers had slower progression in comparison with the carriers of 2200C > T (P < 0.05). The male carriers of 1490insAA had more severe SNHL than female carriers (P < 0.05), whereas there was no gender-specific difference in hearing associated with 2200C > T (P > 0.05).
1596 MAKISHIMA ET AL. to cause DFNA10 hearing loss , we PCR-amplified and sequenced all 20 exons and flanking intronic sequences of EYA4 from genomic DNA. We detected a heterozygous insertion of AA at position 1490 (1490insAA) that co-segregated with SNHL in LMG265. We confirmed the mutation by nucleotide sequence analysis of subcloned amplification products of exon 12 from affected subjects (Fig. 3A). We did not detect 1490insAA in unaffected LMG265 members (including IV-11) or 96 ethnically matched control DNA samples. The 1490insAA allele is predicted to encode a truncated EYA4 protein with an intact variable domain and a deleted Eya domain (Fig. 3B).

LMG265 Cardiac Phenotype
Comprehensive cardiac evaluation of nine affected LMG265 members revealed a variety of abnormalities, including hyperlipidemia and hypertension (not shown), but no evidence of DCM (Table I). The abnormalities we detected are also reported to be present in unaffected family members (data not shown) and, given their high prevalence in the general population, are likely to be unrelated to 1490insAA or SNHL.

DISCUSSION
We identified a novel frameshift mutation, 1490insAA, of EYA4 co-segregating with dominant hearing loss at the DFNA10 locus in family LMG265. Our study comprises the first detailed cardiac evaluation of DFNA10 hearing loss to support the proposed correlation of EYA4 mutation position with the presence or absence of DCM [Schonberger et al., 2005]. We cannot rule out the possibility that this is a spurious correlation arising from interfamilial differences in genetic or environmental modifiers of the cardiomyopathy phenotype.
The postlingual-onset, progressive SNHL phenotype segregating in family LMG265 is similar to those which have been reported for DFNA10 in the American (1468insAA) and Belgian (2200C > T) families De Leenheer et al., 2001]. The affected members of the Belgian family did not show (P > 0.05) the gender difference in severity of SNHL that we observed in LMG265. It is possible that our observation was spurious, but we cannot rule out sex-linked genetic or environmental factors modifying the DFNA10 phenotype in LMG265. Similarly, our observation of small but significant differences in hearing loss associated with 1490insAA versus 2200C > T may also be spurious, but could reflect a correlation with EYA4 genotype.
The vestibular findings in LMG265 family members IV-2, IV-6, V-4, and VI-1 may or may not be direct effects of the EYA4 mutation since they were unilateral in those individuals and absent in affected relatives. The significance of the caloric response variations in IV-6, V-4, and VI-1 and their causal relationship to DFNA10 are even less clear.
The results of our study can now be used to guide the molecular diagnosis and genetic counseling of patients with these phenotypes and EYA4 mutations.