Genetic Analysis of the LOXHD1 Gene in Chinese Patients With Non-Syndromic Hearing Loss

Non-syndromic hearing loss (NSHL) is a common neurosensory disease with an extreme genetic heterogeneity which has been linked to variants in over 120 genes. The LOXHD1 gene (DFNB77), encoding lipoxygenase homology domain 1, is a rare hearing loss gene found in several populations. To evaluate the importance of LOXHD1 variants in Chinese patients with NSHL, we performed genetic analysis on LOXHD1 in 2,901 sporadic Chinese patients to identify the aspect and frequency of LOXHD1 causative variants. Next-generation sequencing using a custom gene panel of HL was conducted on 2,641 unrelated patients and whole-exome sequencing on the remaining 260 patients. A total of 33 likely causative variants were identified in 21 patients, including 20 novel variants and 13 previously reported pathogenic variants. Each of the 20 novel variants was evaluated according to ACMG criteria. These findings showed that causative variants in LOXHD1 were found in about 0.72% (21/2,901) of Chinese NSHL patients. This study is by far the largest number of novel variants identified in this gene expanding the range of pathogenic variants in LOXHD1, and suggests that variants in this gene occur relatively commonly in Chinese NSHL patients. This extensive investigation of LOXHD1 in Chinese NSHL patients proposed six recurrent LOXHD1 variants. These findings may assist in both molecular diagnosis and genetic counseling.


INTRODUCTION
Hearing loss (HL) is common and is known to affect approximately one to three out of 1,000 newborns (Kennedy and Mccann, 2004;Fortnum et al., 2001). HL shows a high degree of clinical and genetic heterogeneity, given different modes of inheritance and the phenotype variabilities associated to the severity, age of onset ranging from congenital and early-onset to late-onset, and large numbers of deafness causing genes. Genetics accounts for over half of the congenital cases. According to the Hereditary Hearing loss Homepage (http://hereditaryhearingloss.org/, accessed 08/ 2021), there is a total of 124 non-syndromic hearing loss genes that has been identified. Autosomal recessive nonsyndromic hearing loss (ARNSHL) accounts for~80% of cases of inherited HL. The two most common genes linked to ARNSHL are GJB2 and SLC26A4, observed in multiple populations. In 2009, biallelic variants in LOXHD1 (lipoxygenase homology domain 1, MIM #613072) was identified as the cause of ARNSHL, DFNB77 (MIM #613079) (Grillet et al., 2009).
Individuals with pathogenic variants in LOXHD1 usually present with heterogeneous hearing phenotypes. Although LOXHD1 pathogenic variants tend to be linked to late-onset HL with an onset age of older than 10 years old (yo) (Maekawa et al., 2019), onset may occur at any age from newborn to adult, with considerable variations in severity varying from mild to profound and progression varying from stable to progressive. Recently, the genotype-phenotype correlations of LOXHD1associated HL have been described, indicating that LOXHD1 variants tend to be associated with a down-sloping audiogram (Maekawa et al., 2019;Kim et al., 2021;Yu et al., 2021).
According to HGMD Professional database (release 2021.02), 107 pathogenic variants of LOXHD1 have been identified to be related with HL. To date, over one hundred pathogenic variants in LOXHD1 have been identified. Although a few sporadic cases of LOXHD1-associated HL have been described in Chinese patients (Hu et al., 2018;Shen et al., 2019;Zhang et al., 2019;Bai et al., 2020;Yu et al., 2021;Jin et al., 2022), little is known of the association between the contribution of LOXHD1 and NSHL in China. Here, our objective was to examine the role of LOXHD1 in Chinese NSHL patients and to provide a genotypical spectrum in this group of patients.

Clinical Data
The DNA of 2,901 sporadic NSHL patients belonging to the Han Chinese population was analyzed. The participants were recruited from the Departments of Otolaryngology Head and Neck Surgery, PLA General Hospital, between June 2015 and August 2021, and ranged in age from 6 months to 28 years. All participants presented with bilateral, congenital or late-onset, mild to profound NSHL (age of onset between 0 and 10 years) with no other obvious systemic disorders. Other explanations for HL, such as infection, malformation, or tumors, were ruled out.
The clinical data acquired at the time of assessment included medical history (by questionnaire), temporal bone computed tomography (CT), otoscopy, pure-tone audiometric examination (if over the age of 6 years), auditory steady-state response (ASSR, if below the age of 6 years), acoustic immittance, auditory brainstem responses, and distortion product otoacoustic emission.
HL was assessed using pure-tone hearing thresholds or ASSR response thresholds (0.5, 1, 2, and 4 kHz for both ears). Hearing was scored using the American Speech-Language-Hearing Association (ASHA) guidelines as normal (threshold ≤25 dB), mild HL (threshold: 25.1-40 dB), moderate HL (threshold: 40.1-55 dB), moderately severe HL (threshold: 55.1-70 dB), severe HL (threshold: 70.1-90 dB), and profound HL (threshold >90 dB, including total deafness). The progression of HL was indicated by a 15 dB increase in the hearing or response threshold in one or both ears from one audiogram to the next increased hearing or response threshold. The hearing frequencies were categorized as low (125-500 Hz), mid (1-2 kHz), or high (4-8 kHz). Prelingual HL was defined as that occurring before the development of speech (usually under the age of 3 years), while postlingual loss occurred after speech development (Adam et al., 1993). Evaluations of vestibular function were conducted using the Romberg, caloric, and tandem gait tests.
The audiograms were described according to shape, specifically, down-sloping (≥15 dB difference between the [better] 250-500 Hz thresholds and the 4,000-8,000 Hz thresholds), flat (≤15 dB difference in thresholds from 125 to 8,000 Hz), U-shaped (≥10 dB difference between the poorest mid-frequency [1-2 kHz] threshold and thresholds at higher and lower frequencies), ascending (≥15 dB difference between the 250-500 Hz and the [better] 4-8 kHz thresholds), residual (limited hearing at only a few frequencies), complete deafness (no hearing at maximum outputs regardless of frequency), and unclassified (none of the above).

Next Generation Sequencing and Bioinformatic Analysis
Genomic DNA was extracted from peripheral blood using a blood DNA extraction kit (TianGen, Beijing, China), as per instructions. Whole-exome capture was applied in 260 patients. A custom capture panel (MyGenostics, Beijing, China) encompassing 3,156 regions of approximately 850 kb target size was constructed to cover the exons and flanking intronic sequences (~50 bp) for 159 deafness genes (Supplementary Table S1) was applied in 2,641 patients. Using these methods, approximately 20,000 genes were analyzed with WES and 159 genes were analyzed with targeted panel. Detailed descriptions of the gene capture, sequencing, and bioinformatics analyses are given in a previous publication (Yao et al., 2017). CNV Analysis along with both WES and targeted panel was conducted by use of Cnvkit software and was described in detail in a previous  publication (Li et al., 2021). For sporadic HL cases, assuming either autosomal dominant (de novo) or autosomal recessive inheritance patterns, only variants de novo in study participants or variants homozygous or compound heterozygous in study participants and meanwhile heterozygous in their parents were picked as candidate.
The variants were comprehensive interpreted using VarSome (Kopanos et al., 2019). Manual variant classification was performed using the American College of Medical Genetics and Genomics (ACMG)/Association for Molecular Pathology (AMP) guidelines for genetic hearing loss (Li et al., 2017). The final screened potential pathogenic variants were verified by Sanger sequencing and validated by parental testing on condition that the DNA samples of parents were available. Some cases in this study have been previously genetically screened as part of the newborn screening of the selected high-risk variants, however, this information is incomplete and was not routinely gathered in this study. Novel variants were defined based on absence from the HGMD professional database, ClinVar database and published literatures. The data, including phenotypes and observed variants, have been submitted to ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) under the accession numbers SCV002011797-SCV002011798.

Variant Analysis
In this study, novel variants were defined based on absence from the HGMD professional database, ClinVar database and published literatures. In this case, twenty-one probands (21/2,901; 0.72%) carrying thirty-three likely causative LOXHD1 variants were identified, including 13 reported pathogenic variants and 20 novel variants. The novel variants included 11 missense variants, 3 nonsense variants, 2 frameshift insertion variants, 1 frameshift deletion variant, and 3 splicing variants. Sixteen novel variants clustered within the PLAT domains, while the remaining four were dispersed in intervals between the PLATs (Figure 1; Table 1). According to ACMG/AMP guidelines, 6 out of 20 novel variants were classified as pathogenic, 1 was classified as likely pathogenic, and 13 were classified as of uncertain significance ( Table 1). All the variants identified in the probands were validated in their parents by Sanger sequencing, except patient No.7 (father's sample was not available) and patient No.17 (parents' samples were not available). It needs to be mentioned that we detected an average of 350 CNVs per sample, but none met the genetic model of these pedigrees and no LOXHD1-associated CNVs were detected.
The candidate variants selected from the sequencing data are shown in Supplementary Table S2. The variants that cosegregated in study participants and parents are shown in bold, and were classified as potentially causative. In patient No.19, a reported variant c.5888delG (p.Gly1963AlafsTer136) was detected in the homozygous state in the patient together with a novel heterozygous variant c.1262G > A (p.Arg421Gln). The two variants were in cis in his father whose hearing was normal. The variant c.5888delG in LOXHD1 leading to a premature termination codon was coded as "PVS1+PS1+PM2+PP3+PP5" and listed as "pathogenic" as mentioned above, while c.1262G > A leading to a single amino acid substitution was coded as  "PM2+BP1+BP4" and listed as of "uncertain significance". Therefore, we presumed c.5888delG was the possible causative variant in patient No.19. Patient No.1 showed two heterozygous variants (c.2641G > A and c.1420G > T) in LOXHD1, in addition to two compound heterozygous variants (c.1256A > T and c.1255A > C) in HARS1. Given that the two HARS1 variants were coded as "c.1256A > T: PM2+PP3" and "c.5557C > G: PM2+BP4" and listed as "Uncertain significance" according to the ACMG criteria with the same population frequency of 0.00109 in East Asia, and a lack of reports of their possible pathogenic significance, we assumed HARS1 was not the causative gene in patient No.1. Similarly,in patient No.8,in addition to the compound heterozygous variants (c.2438T > A and c.2635C > T) in LOXHD1, variants (c.3658G > A and c.5557C > G) in MYO15A were identified, with the parents as carriers. There are no reported pathogenic associations of these MYO15A variants, and the ACMG codes are "c.3658G > A: PP2+BS1+BS2+BP4" and "c.5557C > G: PM1+PM2+PP2+BP4"; thus, they were listed as variants of benign and uncertain significance, respectively. MYO15A was, therefore, not considered to be the causative gene in this family. Nevertheless, as it is possible that these variants may have influenced HL, their relevance cannot be discounted altogether and further investigation should be undertaken. In patient No.7, compound heterozygous variants (c.611-2A > T and c.1846T > C) in LOXHD1 were identified in the affected proband and c.611-2A > T was confirmed as inherited from his mother; the c.1846T > C variant could not be classified as de novo or not as the father's blood sample was not available. A similar situation was seen in patient No.17, where the parents' blood samples were not available.

Recurrent Variant
Of all the variants in LOXHD1 identified in this study, a total of six variants were recurrent, including four reported (c.5888delG, c.611-2A > T, c.2641G > A, and c.4247G > A) and two novel (c.1420G > T and c.6413G > A) variants. Among the reported variants, c.5888delG was detected in three unrelated patients in the heterozygous state (Nos.2, 14, and 17) and in one patient in the homozygous state (No.19), and c.611-2A > T was detected in three unrelated patients in the heterozygous state (Nos.8,14,and 19). Both c.5888delG and c.611-2A > T were defined as "pathogenic" according to the ACMG criteria. The frequency of c.5888delG in the Chinese NSHL population was estimated as 0.0862% (5/5,802), which is four times higher than the value of 0.000183 for the East Asian population recorded in the gnomAD Exomes database. The frequency of c.611-2A > T in the Chinese NSHL population was estimated as 0.0517% (3/5,802) and it was not found in the gnomAD exomes or gnomAD genomes databases. The other four variants, c.2641G > A, c.4247G > A, c.1420G > T, and c.6413G > A, were detected in two unrelated patients separately in the heterozygous state and their frequencies in the Chinese NSHL population were estimated as 0.0345% (2/ 5,802). The variants c.2641G > A, c.4247G > A and c.1420G > T were defined as "likely pathogenic", "pathogenic", and "pathogenic", respectively, according to the ACMG criteria and their frequencies in the East Asian population were not found in either the gnomAD exomes database or the gnomAD genomes database. Although c.6413G > A was classified as a variant of "uncertain significance" according to the ACMG criteria, it is noteworthy that the variant frequency is recorded as 0.033 in the East Asian population in the gnomAD exomes database, which indicated that this variant is likely to be benign. Table 2 lists the clinical features of the 21 unrelated participants with LOXHD1 compound heterozygous/homozygous variants, including age of onset, visiting age, audiogram classification, HL progression, and vestibular symptoms. The age of onset varied from newborn to 10 years, with most participants showing evidence of congenital HL between the ages of 0-3 years (19/21, 90.5%). There were no vestibular symptoms present in any of the participants. By evaluating of the types of HL among the 15 probands with available audiograms, relationships between genotype and phenotype could be determined (Figure 1). The HL in patient No.4 (male/3 yo), No.6 (female/1.5 yo), No.7 (male/28 yo), No.8 (female/7 yo), No.10 (male/4 yo), No.14 (female/0.5 yo), No.15 (female/5 yo), No.17 (female/28 yo), No.19 (male/0.5 yo), and No.20 (female/2 yo) showed symmetric down-sloping audiogram configurations (10/15, 66.7%). The HL in patient No.16 (female/ 1.5 yo) and No.18 (female/1 yo) showed flat audiogram configurations with average thresholds over 90 dBHL on both ears at the age of onset (2/15, 13.3%). Patient No.3 (female/4 yo) and No.11 (male/3 yo) showed asymmetric HL with a downsloping audiogram configuration on one ear and a flat audiogram configuration on the other ear (2/15, 13.3%). Patient No.2 (female/4 yo) had asymmetric HL with a flat audiogram configuration on the left ear and a U-shaped audiogram configuration on the right ear (1/15, 6.7%). Besides these, patients No.9,No.12,No.13,and No.21 who had been diagnosed with profound bilateral NSHL had already received unilateral or bilateral cochlear implantation and the follow-up feedback showed satisfactory speech development after implantation.

Clinical Features of Patients and Relationships Between Genotypes and Phenotypes
Of the 11 patients with follow-up information, four complained of progressive HL while seven gave feedback of stable hearing status. Patient No.4, the only patient with multiple hearing examination results, diagnosed with bilateral symmetric severe hearing loss at 1 yo, showed progressive HL on the left ear and a relatively stable hearing status after 2 years of follow-up ( Figure 1; Table 2).

DISCUSSION
It was observed that pathogenic LOXHD1 variants were associated with~0.72% (21/2,901) of sporadic NSHL cases in a patient cohort without apparent disorders in other systems. All the LOXHD1 variant-positive patients were under 30 years old and 90.5% (19/ 21) had HL of congenital onset. In total, 13 reported pathogenic variants and 20 novel variants of LOXHD1 were identified including six recurrent variants.
The protein LOXHD1 is known to be associated with the auditory system. The mouse Loxhd1 homologous to human LOXHD1 was first cloned by Grillet et al., in 2009 and was found to be expressed in the cochlear and vestibular hair cells of young mice and in hair cells along the length of the stereocilia in adult mice (Liu et al., 2020;Grillet et al., 2009;Trouillet et al., 2021). Recently, Loxhd1 was identified as being required for the FIGURE 2 | The prevalence of LOXHD1-associated HL in NSHL cohorts in different countries. The prevalence of LOXHD1-associated HL in NSHL cohorts ranged rom 0.365 to 3.28% in the seven countries (Egypt, Czech, Netherlands, Italy, China, United States, Arab and Japan).
Frontiers in Genetics | www.frontiersin.org May 2022 | Volume 13 | Article 825082 mechanotransduction process by which hair cells convert physical forces into electrochemical responses (Trouillet et al., 2021).  (Bateman and Sandford, 1999;Lemmon, 2008). Although the functions of LOXHD1 have not been fully elucidated, it can be roughly inferred from the function of the PLAT domain as the protein is made up entirely of PLAT repeats. The variants identified in our study were distributed in 14 of the 15 domains and their intervals, with most variants concentrated in the PLAT7 domain. NSHL typified by a down-sloping audiogram configuration suggests LOXHD1 involvement, and the phenotype and inheritance patterns are similar to those seen with SLC26A4 and TMPRSS3. In our study, most cases showed congenital HL (19/21, 90.5%) and bilateral symmetric down-sloping audiogram configurations (10/15, 66.7%) that were in accordance with previous reports (Kim et al., 2021;Bai et al., 2020). To date, the prevalence of LOXHD1-associated HL in relatively larger NSHL cohorts with more than 50 patients has been reported only in seven countries, ranging from 0.365 to 3.28% (Figure 2). Our results also indicate that LOXHD1 variants account for about 0.72% (21/2,901) of NSHL in Chinese patients, an incidence similar to that reported in the American population (0.71%, 8/1,119) (Bateman and Sandford, 1999;Lemmon, 2008). In order to analyze the correlation between variant location within or out of domain and hearing phenotype, we grouped the 21 patients into "PLAT-domain and PLAT-domain", "PLATdomain and interval", and "interval and interval". There were 15 patients in the group "PLAT-domain and PLAT-domain" with two variants both locating within the PLAT domains, six patients in the group "PLAT-domain and interval" with one variant locating within a PLAT domain and the other variant within an interval, and no patients in the group "interval and interval". There seemed to be no difference in the hearing threshold between groups due to large variability between individuals (Supplementary Figure S1-Type1). Of all 33 variants identified in this study, 12 were non-truncating variations (36.4%) and 21 were truncating variations (63.6%). Similarly, we grouped the 21 patients into "truncating and truncating", "non-truncating and truncating" and "non-truncating and nontruncating" to analyze the relationship between variant severity and hearing phenotype. There were 7 patients in the group "truncating and truncating" carrying two truncating variations, 2 patients in the group "non-truncating and non-truncating" carrying two non-truncating variations, and 12 patients in the group "non-truncating and truncating" carrying both types of variations simultaneously. Again, we noticed no difference between the groups in terms of hearing threshold due to the large variability between individuals (Supplementary Figure S1-Type2). Besides, we counted the number of different types of variants in LOXHD1 to analyze their proportions of occurrence. Of the 124 variants of LOXHD1 worldwide, 63 are missense variants, 22 are nonsense variants, 16 are frameshifting variants, 12 are splice-site variants, 9 are intronic variants and 2 are synonymous variants. Of all these variants, 51 were found in the Chinese population, including 27 missense variants, 7 nonsense variants, 6 frameshifting variants, 5 splice-site variants, and 6 intronic variants. The proportions of different types of variation of LOXHD1 both worldwide and in China were shown in Supplementary Figure S2 and the results suggested that LOXHD1-associated hearing loss is caused firstly by missense variants and secondly by nonsense variants, which were consistent with the previous work that the majority of loss of function (LoF) variants in LOXHD1 are nonsense (Azaiez et al., 2018).
Our findings expand the spectrum of variants seen in LOXHD1. Supplementary Table S3 summarizes the position, origin, and pathogenicity classification of the 124 LOXHD1 variants filtered using the Human Gene Mutation Database (HGMD) Professional version (release 2021.02) and appraised through literature review to date, including the 20 novel variants identified in this study. The findings have also expanded the variant spectrum of LOXHD1 in the Chinese deafness population from 28 to 51. Figure 3 shows the loci of LOXHD1 variants seen in different populations on the schematic diagram of the protein domain structure. Among all the 13 reported variants, c.5888delG (p.G1963Afs*136), c.611-2A > T, c.3268C > T (p.R1090W), c.4247G > A (p.W1416*) and c.4212+5G > A (NM_144612.6) have only been reported in Chinese HL patients (Bai et al., 2020;Liu et al., 2020;Yu et al., 2021), c.1270+4A > C was reported in a Japanese family (Kim et al., 2021), and the other variants were reported mainly in European and American multi-ethnic populations (Sloan-Heggen et al., 2016;Zazo et al., 2017;Safka et al., 2020). Notably, both c.5888delG and c.611-2A > T had a relatively high recurrent frequency as 0.0862% (5/5,802) and 0.0517% (3/5,802) in our study, therefore, we speculate that they may be hot spot variants in the Chinese HL population.
This study has two minor limitations. First, most patients in our cohort did not present with multiple audiograms, leading to difficulties in estimating HL progression. We, therefore, had to rely on parental observation. Second, copy number variations in LOXHD1 were not detected in this study; these should be investigated in further studies.
In conclusion, we evaluated the importance of LOXHD1 variants in 2,901 Chinese patients with NSHL and investigated the clinical and genetic features of 21 patients. Our findings indicate that variants in LOXHD1 represent a comparatively common cause of NSHL in the Chinese population. It is hoped that these findings will assist with diagnosis, management, and genetic counseling of NSHL patients, as well as suggesting directions for therapeutic development.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Chinese PLA General Hospital Research Ethics Committee (Beijing, China). Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
W-QW, XG, Y-YY, and PD conceived of the study and participated in its design and draft the manuscript. S-SH, J-CX, and M-YH participated in the data analysis. KY, XZ, S-YY and D-YK participated in the collection of clinical data and blood samples. All authors read and approved the final manuscript.