The Prevalence and Clinical Characteristics of TECTA-Associated Autosomal Dominant Hearing Loss

TECTA is well known as a causative gene for autosomal dominant mid-frequency hearing loss observed in various populations. In this study, we performed next-generation sequencing analysis of a large Japanese hearing loss cohort, including eight hundred and twelve (812) subjects from unrelated autosomal dominant hearing loss families, to estimate the prevalence and phenotype-genotype correlations in patients with TECTA mutations. The prevalence of TECTA mutations in Japanese autosomal dominant sensorineural hearing loss families was found to be 3.2%. With regard to the type of hearing loss, the patients with mutations in the nidogen-like domain or ZA domain of TECTA showed varied audiograms. However, most of the patients with mutations in the ZP domain showed mid-frequency hearing loss. The rate of hearing deterioration in TECTA-associated hearing loss patients and in the normal hearing Japanese control population were the same and regression lines for each group were parallel. We carried out haplotype analysis for four families which had one recurring missense variant, c.5597C>T (p.Thr1866Met). Our results revealed four different haplotypes, suggesting that this mutation occurred independently in each family. In conclusion, TECTA variants represent the second largest cause of autosomal dominant sensorineural hearing loss in Japan. The hearing loss progression observed in the patients with TECTA mutations might reflect presbycusis. The c.5597C>T mutation occurred in a mutational hot spot and is observed in many ethnic populations.


Introduction
Hearing loss (HL) is one of the most common sensory disorders. Congenital HL, in particular, affects one in 500-600 newborns and it is speculated that about 60% of cases are caused by gene mutations [1]. Currently, it is considered that approximately 120 genes are associated with sensorineural hearing loss (SNHL) [2], with non-syndromic HL accounting for 70% of SNHL. Among them, autosomal recessive (AR) inheritance accounts for approximately 75% and autosomal dominant (AD) inheritance constitutes approximately 15-20% of cases. Autosomal dominant sensorineural hearing loss (ADSNHL) tends to be post-lingual and progressive HL [3]. Some causative genes are characterized by typical HL patterns affecting only specific frequencies. Distinct mutations in one of the causative genes for ADSNHL, TECTA (DFNA8/12), are well known to be associated with mid-frequency HL, while other mutations are involved in high frequency HL. All kinds of mutations are found among various populations. TECTA is also a known genetic cause of autosomal recessive HL (DFNB21); however, most of the TECTA-associated HL is ADSNHL. It comprises 23 exons, and encodes a 2155 amino acid protein, α-tectorin, which is non-collagenous glycoprotein that constitutes a major component of the tectorial membrane in the cochlea [4,5]. The tectorial membrane lies over the cochlear hair cells and is critical for both the mechanical amplification of acoustic stimulation by the outer hair cells and for its transmission to the inner hair cells, which are the genuine sensory cells [6]. Previous studies have reported that TECTA-associated ADSNHL showed characteristic mid-or high-frequency HL depending on the position of the mutations [5,7].
Recently, next-generation sequencing (NGS) has become commonly used for the genetic screening of SNHL. Although hundreds or thousands of variants may be identified by NGS, it is difficult to determine which variants are pathogenic. It is particularly difficult to prioritize potential causal variants within a family with AD inheritance in which only one particular variant leads to the production of the trait. Therefore, we should consider not only NGS analysis results but also the phenotypes of the subjects.
In this study, we (1) elucidated the prevalence of HL caused by TECTA mutations in Japanese ADSNHL patients, (2) examined the genotype-phenotype correlations for each domain, (3) analyzed the rate of HL deterioration for TECTA-associated HL patients, and (4) carried out haplotype analysis for one variant identified in four unrelated ADSNHL families to confirm whether the mutation occurred in a mutational hotspot or whether it was a founder mutation.

Subjects
In this study, we enrolled eight hundred and twelve (812) Japanese subjects (age range: 0-86 years, mean age: 37.1 years) from unrelated ADHL families. This study cohort was taken from 67 otolaryngology clinics across Japan between June 2000 and May 2017. All subjects in this study suffered from non-syndromic HL without any other associated symptoms. A written informed consent was obtained from each proband and their family members prior to participation in this study. This study was approved by the Shinshu University Human Genetic Analysis Ethical Committee, as well as by the ethics committee of each participating institute. The study was conducted in accordance with the Declaration of Helsinki, with the protocol approved by the Ethics Committee of the Shinshu University School of Medicine No. 387, 4 September 2012 and No. 576, 2 May 2017. Clinical information was obtained for each proband and relatives from medical charts. In this study, we collected the following data: (1) pure-tone audiograms, behavioral audiometry, or auditory steady state responses (ASSR); (2) medical history, including onset of HL, progression, and episodes of vertigo; and (3) temporal bone imaging (computed tomography and/or magnetic resonance) if available.

Genetic Analysis and Pathogenic Interpretation
In this study, we used amplicon re-sequencing for 68 genes previously reported as genetic causes of non-syndromic hereditary HL (Table S1). In brief, amplicon libraries were prepared using an Ion AmpliSeq TM Custom Panel (ThermoFisher Scientific, Waltham, MA, USA), and Ion AmpliSeq v2 plus kit (ThermoFisher Scientific) in accordance with the manufacturer's instructions. After amplicon library preparation, emulsion PCR, and next-generation sequencing were performed according to the manufacturer's protocol with Ion 200 sequencing kit (ThermoFisher Scientific) and Ion PGM sequencer (ThermoFisher Scientific) or Ion HiQ chef Kit (ThermoFisher Scientific) and Ion Proton sequencer (ThermoFisher Scientific). The detailed protocol has been described elsewhere [8].
The fastq files obtained were mapped against the human genome (build GRCh37/hg19) using the Torrent Mapping Alignment Program and the variants, including the SNVs, insertions, and deletions, were detected by Torrent Variant Caller plug-in software. The effects of each variant on the protein were analyzed using ANNOVAR software [9]. The variants affecting amino acid sequences (missense, nonsense, insertion/deletion, and splicing variants) were selected from the identified variants. The selected variants were then also filtered as less than 1% of (1) the 1000 genome database [10], (2) the 6500 exome variants [11], (3) the Human Genetic Variation Database (dataset for 1208 Japanese exome variants) [12], and (4) the 333 in-house Japanese normal hearing controls. To estimate the pathogenicity of missense variants, functional prediction software, included in the ANNOVAR software (Sorting Intolerant from Tolerant (SIFT) [13], Polymorphism Phenotyping (PolyPhen2) [14], Likelihood Ratio Test (LRT) [15], Mutation Taster, [16] and Mutation Assessor [17]) were used. Direct sequencing was used to confirm the candidate variants identified through the analysis pipeline. Segregation analysis of family members was also performed by direct sequencing.
The pathogenicity of the identified variants was evaluated in accordance with the American College of Medical Genetics (ACMG) standards and guidelines [18]. This system classified variants into five categories; pathogenic, likely pathogenic, uncertain significance, likely benign, and benign based on the various evidence. In addition, we referred to Inter Var when we evaluated variants [19].
A combined annotation dependent depletion (CADD) was also utilized to prioritize potential causal variants [20].

Haplotype Analysis of the c.5597C>T Variant
We conducted haplotype analysis for the c.5597C>T (p.Thr1866Met) variant identified in four unrelated ADSNHL families. The haplotypes within the 1Mb region surrounding position c.5597 were analyzed using a set of 11 single nucleotide polymorphisms (SNPs) (three sites upstream and eight sites downstream). To select the SNPs for haplotype analysis, Tag SNPs were searched by SNPinfo Web server of the National Institute of Environmental Health Sciences with the Hap Map JPN data set [21]. This analysis was performed using the direct sequencing method.
Data Availability Statement: The sequencing data are available in the DDBJ databank of Japan (Accession number: JGAS00000000201).
We next categorized these variants based on the ACMG criteria and Inter Var [18,19]. Then, we compared the CADD Phred scores for each variant with the larger value of the minor allele frequency (MAF) in the ExAC database [25] or in the ToMMo 3.5KJPN database (Tohoku University Tohoku Medical Megabank Organization, Sendai, JPN) [26] to estimate the pathogenicity, as shown in the scatter plot in Figure S1. We further selected the candidate variants as follows; (1) VUS variants (indicated as blue points in Figure S1) identified in 10 probands were less likely to cause HL, thus we removed these variants from further analysis. (2) We removed the variants with high MAF and/or low CADD scores (under the dotted-line in Figure S1). (3) We removed the one nonsense variant c.4302C>A (p.Tyr1434Ter) as a cause for DFNA8/12. Disease causing mechanism for TECTA -associated ADSNHL was thought to be a dominant-negative effect resulting from missense mutations [22,27]. Therefore, it is unlikely that this nonsense mutation is causative for ADSNHL. In addition, the proband (HL0644) with this mutation also carried another variant in the TECTA gene; c.4955G>C (p.Asp1499His), which was categorized as VUS, suggesting autosomal recessive inheritance (DFNB21). Finally, we defined the remaining 22 variants as likely causative variants ( Table 1). The pedigree and audiograms for each patient are shown in Figure 1 and the variants are listed in Table 1. The in silico prediction software scores for novel missense variants are shown in Table S2.
According to the above results, the prevalence of TECTA mutations in Japanese ADSNHL families was considered to be 3.2% (26/812 probands). Furthermore, when we limited our analysis to 'Likely pathogenic' variants, it was considered to be 1.8% (15/812 probands).
Among the identified variants, c.5597C>T was recurrent and identified in four unrelated ADSNHL families. Interestingly, the c.3995G>A variant identified in this study caused a p.Cys1332Tyr amino acid change. Kim et al. have reported the c.3995G>T variant as pathogenic, which despite having a different nucleotide change, resulted in the same amino acid change [28].

Genotype-Phenotype Correlation
α-tectorin has three major components: (1) a nidogen-like (NIDO) domain; (2) a large zonadhesin (ZA) domain containing three trypsin inhibitor-like (TIL) cysteine-rich domains, a von Willebrand factor type C (vWFC) and four von Willebrand factor type D (vWFD) domains; and (3) a zona pellucida (ZP) domain [29]. According to previous reports, mutations affecting the ZP domain are significantly associated with mid-frequency SNHL, whereas mutations in the ZA domain are associated with high-frequency SNHL [5,7]. Figure 2a shows a schema of the α-tectorin domains, and overlapping pure-tone audiograms of individuals corresponding to the position of each domain.    In the NIDO domain, three variants, c.494C>T, c.589G>A and c.605T>C, were found in one family each. One variant, c.589G>A, led to high-frequency SNHL. The other two variants, c.494C>T and c.605T>C, led to mid-frequency SNHL.
In the ZA domain, the audiograms were varied, with mid-frequency HL observed in 6 subjects, flat-type HL in six subjects and high-frequency HL in eight subjects (Figure 2a).
In the ZP domain, seven of 12 variants were associated with mid-frequency HL, and four variants were associated with flat HL. The severity of HL varied among the patients, however nine out of 11 cases had mild-moderate HL (Figure 2a).

Relationship between Age and Hearing Levels
To estimate the rate of hearing deterioration, we compared the age and hearing levels of 26 probands. As shown in Figure 3, each blue point indicated the average hearing level (0.5-4K Hz) obtained at the time of blood sampling. In contrast, the red points are the average hearing level in the Japanese normal hearing population (age: 35-74 y). As a result, the rate of hearing deterioration in both groups was found to be the same and the regression lines for each group were parallel. Thus, hearing deterioration in TECTA patients was age-related and may not be accelerated by the gene mutation.

Discussion
In this study, we examined the prevalence of TECTA mutations in the Japanese ADSNHL population, and identified 26 TECTA-associated HL patients, which account for 3.2% of ADSNHL families. According to previous reports, TECTA mutations account for 2.9-5% of all ADSNHL patients [24,30]. Our result is consistent with those previous studies. The findings show that TECTA variants represent the second largest cause of ADSNHL in Japan, following KCNQ4 variants (6.6%) [31].
In previous studies, only three mutations within the NIDO domain, c.632T>C (p.Phe211Ser), c.589G>A (p.Asp197Asn), and c.710C>T (p.Thr237Ile), were reported in ADSNHL patients [32]. Furthermore, those cases showed mid-or high-frequency HL (Figure 2b) [32]. In our study, three variants, c.494C>T, c.589G>A, and c.605T>C, were identified in one family each, two of them novel. The patient with known c.589G>A variant showed high-frequency HL, but the other patients (with the novel c.494C>T and c.605T>C variants) showed mid-frequency HL (Figure 2a). The c.589G>A variant was reported previously, but that phenotype was mid-frequency SNHL [32]. Thus, the characteristic phenotype of HL was not observed in the NIDO domain variants. Previously, mutations in the vWFD2-D3 or vWFD4 repeat areas included in the ZA domain have been associated

The c.5597C>T (p.Thr1866Met) Variant in the ZP Domain
One missense variant, c.5597C>T (p.Thr1866Met), in the ZP domain was identified in four unrelated families. The audiograms showed flat or mid-frequency HL, and were similar to the previously reported phenotype for variants located in the ZP domain [7].
We carried out haplotype analysis for these four families to determine whether this variant arose in a mutational hot spot or was a founder mutation. Table 3 shows the haplotype patterns within the 1-Mb region surrounding this variant.
The different haplotypes observed suggest that this mutation occurred independently in each family. Thus, we suggest that c.5597C>T (p.Thr1866Met) arose from a mutational hot spot.  429809  rs752979  T  C  T  C  T  T  T  T  T  T  C  C  T  C  T  C  C 0.46 T  0.54  315340  rs4430518  T  G  T  T  G G  T  T  T  T  T  G  T  T  T  T

Discussion
In this study, we examined the prevalence of TECTA mutations in the Japanese ADSNHL population, and identified 26 TECTA-associated HL patients, which account for 3.2% of ADSNHL families. According to previous reports, TECTA mutations account for 2.9-5% of all ADSNHL patients [24,30]. Our result is consistent with those previous studies. The findings show that TECTA variants represent the second largest cause of ADSNHL in Japan, following KCNQ4 variants (6.6%) [31].
In previous studies, only three mutations within the NIDO domain, c.632T>C (p.Phe211Ser), c.589G>A (p.Asp197Asn), and c.710C>T (p.Thr237Ile), were reported in ADSNHL patients [32]. Furthermore, those cases showed mid-or high-frequency HL (Figure 2b) [32]. In our study, three variants, c.494C>T, c.589G>A, and c.605T>C, were identified in one family each, two of them novel. The patient with known c.589G>A variant showed high-frequency HL, but the other patients (with the novel c.494C>T and c.605T>C variants) showed mid-frequency HL (Figure 2a). The c.589G>A variant was reported previously, but that phenotype was mid-frequency SNHL [32]. Thus, the characteristic phenotype of HL was not observed in the NIDO domain variants. Previously, mutations in the vWFD2-D3 or vWFD4 repeat areas included in the ZA domain have been associated with high-frequency HL (Figure 2b). In this study, we identified 13 variants in the ZA domain region; however, various phenotypes were observed, so we couldn't identify any trend regarding HL in the ZA domain.
We also identified 11 variants in the ZP domain and most cases showed mid-frequency HL. This correlation between mutations in the ZP domain and mid-frequency HL has been reported previously ( Figure 2b) [9,17,19]. This study is in agreement with the findings of those previous studies. In previous studies, some genes were reported as causative for mid-frequency HL, particularly, EYA4 (DFNA10), TECTA (DFNA8/12), COL11A2 (DFNA13), POU4F3 (DFNA15), and CCDC50 (DFNA44). Among those genes, mutations in TECTA are the most frequent cause of mid-frequency HL [1,24,33,34]. Yamamoto et al. reported that pathogenic and possibly pathogenic variants of TECTA were found in 6.0% of mid-frequency SNHL patients [35].
In this study, we focused on TECTA mutations identified in patients with ADSNHL as most of TECTA-associated HL is autosomal dominant and rarely autosomal recessive. The relationship between a domain structure and the inheritance mode of a TECTA mutation remains non conclusive. Regarding the type of mutation, missense mutations were predominantly observed in ADSNHL patients, while most loss of function mutations (nonsense, splicing, and frameshift mutations) in TECTA were identified in autosomal recessive cases. These results suggest that the mechanism of ADSNHL TECTA mutations is presumably dominant negative.
One of the clinical characteristics of the TECTA-associated HL was its non-progressive nature [18]. Byung et al. reported that the average HL deterioration rate didn't exceed 1 dB/year for the NIDO domain variant c.710C>T, p.Thr237Ile [32], and in another study, the p.Arg1890Cys mutation identified in a Dutch ADSNHL family, showed non-progressive, mid-frequency HL [9]. In this study, we analyzed the HL deterioration rate for TECTA-associated HL, and found it to be 0.35 dB/year. This HL deterioration rate is comparable to the hearing deterioration rate in a normal hearing control population. A previous report stated that patients with TECTA-associated mid-frequency HL might be prone to presbycusis as they are theoretically exposed to a lower level of sound energy as a result of cochlear amplification deficiency [36,37]. However, our results indicated that the hearing loss progression rate was the same as in the control group. Based on these results, we suggest that the mutations in the TECTA gene cause functional loss or malformation of the tectorial membrane and inhibit cochlear amplification. As a result of disturbed cochlear amplification, mild to moderate HL is observed. The HL progression (0.3 dB/year) observed in patients with TECTA mutations appears to reflect presbycusis, as the HL deterioration rate was comparable with that of normal hearing controls, suggesting that TECTA mutations do not accelerate HL deterioration.
In this study, one missense variant, c.5597C>T (p.Thr1866Met), in the ZP domain was detected in four unrelated families. This mutation was also reported in different populations including Americans, Spanish, and Koreans [22]. Hildebrand et al. performed haplotype analysis of the mutation found in the American and Spanish populations and found that they carried different haplotypes suggesting a unique founder effect could be identified in each population [22]. In this study, we also performed haplotype analysis for four families with this mutation and showed they carried different haplotypes even though all of them are Japanese. The results of this study and previous reports suggested that the c.5597C>T mutation arose in a mutational hot spot and occurs regardless of population.

Conclusions
The prevalence of TECTA mutations in Japanese ADSNHL families is estimated as 3.2% (26/812 probands). Furthermore, when we limited our analysis to probands with 'Likely pathogenic' or 'Pathogenic' variants, it was found to be 1.8% (15/812 probands). In the NIDO and ZA domain regions, various types of HL were observed, so we couldn't identify any trends. In the ZP domain, most cases showed mid-frequency HL. This result is in agreement with the findings of previous studies. The HL progression observed in the patients with TECTA mutations might reflect presbycusis, as the HL deterioration rate was comparable with that of normal hearing controls. A TECTA mutation itself is considered not to accelerate the HL deterioration. The recurrent c.5597C>T mutation might have arisen in a mutational hot spot and can be observed in many ethnic populations.