Recessive LOXHD1 Variants Cause a Milder Prelingual Hearing Loss: Genotype-Phenotype Correlation and Three Additional Patients With Novel Variants

Background Biallelic mutations in LOXHD1 have been identied as the cause of DFNB77 (deafness, autosomal recessive 77). It is a novel, progressive, severe-profound, and late-onset non-syndromic hearing loss, and is genetically and phenotypically highly heterogeneous. This study aimed to provide an additional three cases of DFNB77 to analyze this complex disease. Methods We presented three cases of pediatric patients with prelingual milder form of the DFNB77 with residual hearing at low frequencies. Trio whole-exome sequencing (WES) was conducted to identify the pathogenic variants. Additionally, we reviewed the literature to further analyze the relationships between the genotype and audiology phenotype of LOXHD1 worldwide. Results Six novel possible pathogenic LOXHD1 variants in three patients were identied by WES, including three missense, one nonsense, and two splicing variants. The literature review showed that 68.5% of DFNB77 patient onset before ve years old; Most variants (60%) were associated with a milder phenotype, particularly variants in the protein domain of PLAT 7 and PLAT 9. We found that compared with homozygous LOXHD1 variants, individuals with heterozygous compound variants had a signicantly milder phenotype, especially individuals carrying one missense and one splicing or bi-allelic missense variants (P <0.05). Audiometric analysis at different ages showed that the hearing loss degree was aggravated at all frequencies in adulthood and more severe in elderhood. Conclusions We report three children with hearing loss carrying six novel LOXHD1 variants identied by WES. Furthermore, our work indicates that DFNB77 may be milder than previously reported, and recommends considering the genotype combination and mutation location of LOXHD1 and race-specicity in DFNB77 molecular diagnoses and management. phenotype. This study delineates the genetic and clinical characteristics of three Chinese patients with prelingual HL, harboring novel pathogenic variants in LOXHD1. Additionally, we summarized the clinical and genetic characteristics of reported patients with DFNB77 and explored the genotype-phenotype correlation. phenotype in HL progressive at all frequencies during adulthood. LOXHD1 changes in the of the stereociliary HL of HL. follow-up


Background
Hearing loss (HL) is the most common sensory impairment, affecting around 466 million people worldwide, of which 7% are children (World Health Organization (WHO; https://www.who.int/pbd/deafness/estimates/en/), accessed June 2020). The incidence of HL is 1 to 2 per 1000 newborns [1]. At least half of the congenital HL cases account for genetic factors [2]. More than 120 genes have been reported associated with non-syndromic hearing loss (NSHL); about 65% related to autosomal recessive inherited non-syndromic hearing loss (ARNSHL) (Hereditary Hearing Loss Homepage; https://hereditaryhearingloss.org), accessed June 2020). The most common genes associated with ARNSHL, such as GJB2, SLC26A4, MYO15A, and CDH23, have been well published worldwide [3] [4]. Relatively, LOXHD1 was a novel gene described by Grillet et al. in 2009 [5]. Knowledge of the genotype-phenotype correlations is limited; this hampers the diagnosis, genetic counseling, and treatment for ARNSHL patients.
LOXHD1 is the causative gene of autosomal recessive deafness 77 (DFNB77, OMIM #613079). DFNB77 is a rare, progressive type of ARSNHL. LOXHD1 is located on chromosome 18q12-21 and encodes the protein lipoxygenase homology domain 1 (LOXHD1), which is highly conserved. LOXHD1 contains 15 PLAT (polycystin-1, lipoxygenase, alpha-toxin) domains. Studies have shown that PLAT domains help proteins target the cell membrane and mediate the interactions of proteins [6] [7]. Mice with a homozygous missense of Loxhd1 demonstrated that the Loxhd1 is mainly expressed in mature hair cells and stereocilia membrane, which is required in maintaining the normal function of cochlea hair cells [5].
To date, about 100 cases with DFNB77 have been reported [8] [9] [10] [11] [5]. However, the acoustic characteristics are heterogeneous. The onset age of HL varied from neonate to adulthood; the severity of HL varied from mild to profound; the progression of HL varied from stable to progressive. Furthermore, it is presently unclear for the correlation between LOXHD1 mutations and the audiology phenotype.
This study delineates the genetic and clinical characteristics of three Chinese patients with prelingual HL, harboring novel pathogenic variants in LOXHD1. Additionally, we summarized the clinical and genetic characteristics of reported patients with DFNB77 and explored the genotype-phenotype correlation.

Subjects
Three Chinese children (two boys and one girl) with prelingual HL from three unrelated families were recruited from our otolaryngology department of Children's Hospital of Fudan University. We obtained the written informed consent from participants' parents before analysis. The study protocol was approved by the Ethics Committee of Children's Hospital, Fudan University (2019-094 and 2019-214).

Clinical evaluations
Medical records and family history using a questionnaire focusing on the onset age and progression of HL, pedigree, tinnitus, and vertigo was obtained from all subjects. All participants received hearing tests, including otoscopy, tympanometry, distortion product otoacoustic emission (DPOAE), and pure tone average (PTA). Auditory brainstem response (ABR) and steady-state auditory response (ASSR) were performed on individuals under ve years old or could not complete PTA. Further speech audiometry was performed on patients over ve years old using standard Chinese single syllable word chart cards. In patients #1 and #3, a temporal bone computed tomography (CT) and magnetic resonance imaging (MRI) scan was performed.
We recorded PTA with an Interacoustics AC40 clinical audiometer and HAD-280 earphones. It was conducted in an audiometric test room where the ambient noise level is less than 25 dB (sound pressure level), and the patient was instructed on how to play the test. The audiometric con gurations were classi ed into six different types: rising, mid-frequency u-shaped, high-frequency u-shaped, high-frequency gently sloping, high-frequency steeply sloping, and at (Supplementary Table 1). Moreover, the audiograms that did not meet any of the criteria above are classi ed as unspeci ed. The grades of hearing loss were classi ed by the pure tone average at 0.5, 1, 2, and 4 kHz in the better ear according to the WHO classi cation of HL, as follows: 26-40 dB, mild HL; 41-60 dB, moderate HL; 61-80 dB, severe HL; and ≥ 81 dB, profound HL. (WHO; https://www.who.int/pbd/deafness/hearing_impairment_grades/en/), accessed June 2020).

Trio whole exome sequencing and bioinformatics analysis
Genomic DNA of peripheral blood was extracted using standard techniques with the QIAamp DNA Blood Mini Kit (QIAGEN, Germany). Whole exome sequencing (WES) was performed on all patients and their parents following the previously published method [12]. Brie y, DNA fragments were enriched for the target region of the coding sequence exons using the Agilent SureSelect XT Human All Exon 50 Mb kit. Subsequently, the sequencing was performed on Illumina's HiSeq2000 sequencer according to its instructions (San Diego, CA, USA).
The sequence data were mapped to the human reference genome (GRCh37/hg19) with a Burrows-Wheeler Aligner (BWA) software (http://bio-bwa.source forge.net/). For variant calling, single-nucleotide variant and small Indels were detected by GATK best practice (https://gatk.broadinstitute.org/hc/en-us). The variants were annotated by ANNOVAR and VEP software [13], and ltered against 1000Genomes projects (https://www.internationalgenome.org/), gnomAD The pathogenicity of a variant was classi ed based on the ACMG (American College of Medical Genetics and Genomics) standards and guidelines [14]. The candidate variants were con rmed by Sanger sequencing with an ABI 3730 Genetic Analyzer (Applied Biosystems). The sequence variant data have been submitted to the ClinVar database (SUB8294387).

Clinical characteristics of patients with LOXHD1 variants
In this study, three affected children from three non-consanguineous Chinese families were identi ed. All of these families have negative family histories of congenital or early-onset HL. The detailed clinical features of these patients are listed in Table 1.
All patients presented with a prelingual onset hearing loss, and the diagnosed age varied from 2 to 8 years old. They all had bilateral, moderate to severe sensorineural HL (Fig. 1A). DPOAEs of all patients were absent bilaterally. PTA revealed that patient #1 and #2 had severe HL. ABR and ASSR revealed that patients #3 had bilateral moderate HL, and no increases in interpeak latencies were observed. The hearing loss was found to worsen with age in patient #1, with three years' follow-up. Speech audiometry showed that the speech recognition thresholds (SRT) were 80 dB and 85 dB in the left and right ear of patient #1; 65 dB and 100 dB in the left and right ear of patient #2 (Fig. 1B). Moreover, a maximum of 95% and 100% speech recognition was attained at 110 dB sound level in the better ear of patient #1 and patient #2. Speech in the noise test showed that 50% speech recognition was attained at the signal-to-noise of 5 dB with 100 dB for patient #1, while 60% speech recognition was attained at signal-to-noise 5 dB with 100 dB for patient #2.
None of the affected individuals complained of tinnitus and vertigo. Besides, none of the family members complained of eye pain, glare, or blurred vision.
Otoscopic examination and external ear inspection were normal in all patients. The MRI and CT of the brain and temporal bones of the left and right ears for patient #1 and patient #2 demonstrated to be normal (Fig. 1C). No known causes of acquired HL were found in our patients, such as prenatal infection, TORCH (toxoplasmosis, rubella, CMV, HSV) infection, aminoglycoside exposure, otoacoustic trauma, meningitis.
None had delayed motor development or dysmorphology. Besides, the Wechsler Intelligence Scale test for children (WISC-III) was performed on patients #2, and the score was 44 of Verbal IQ, 98 of Performance IQ, 63 of full-scale IQ.

Variation detection and functional analysis
We identi ed six novel LOXHD1 variants. All the variants were detected in a compound heterozygous state (Fig. 1D). The variants consisted of one nonsense, two splicing, and three missense variants. The six variants were all located in a well-conserved site ( Fig. 2A). Besides, all variants are located in PLAT domains of LOXHD1 protein, with one except (c.4814T>C, p.Met1605Thr) (Fig. 3A). Based on the ACMG guidelines, the variants were categorized into two pathogenic, three likely pathogenic, and one uncertain signi cance (Table 2).
Molecular models for the homeodomain of wild-type and variant-containing LOXHD1 protein were developed and adopted a β sandwich fold containing PLAT domains. The two missense variants (c.6038G>A, p.Gly2013Glu; c.1891C>T, p.Arg631Cys) were predicted to change the polarity with surrounding active site residues and consequently destabilize the β sandwich structures (Fig. 2B).
The two splicing variants were evaluated in silico using the web tool HSF. The variant (c.3351-1C>A) was predicted to alter the wide-type acceptor sites and may affect splicing by HSF; variants c.4212+5G>A were predicted to alternate the original splicing donor site and then affected the splicing ( Supplementary   Fig. 1).
The missense variant (c.4814T>C, p.Met1605Thr) was absent in the ExAC, 1000 Genomes, and gnomAD databases and highly conserved across different species. Unfortunately, the Met1605Thr protein template cannot be obtained from Swiss-model servers, and more evidence for its pathogenicity was not obtained. Thus, the pathogenicity of this variant is uncertain.

phenotype
To date, a total of 111 cases with DFNB77 were reported, including the three cases in this study (Supplementary Table 2). We found that most patients' onset age was less than ten years old (68.5% for < 5 years and 14.4% for 5-10 years). There have 16.2% of patients shown severe HL, and 50.5% shown profound HL. 28.8% of patients were described with progressive HL, and 56.2 % of them occurred in adulthood ( Table 3). The complete audiometric data were divided into four groups by age (<10, 10-19, 20-55, and >55 years old), and we found that the main audiogram of DFNB77 among different age groups was a highfrequency down-sloping con guration (Fig. 4). The degree of HL was observed increased in all frequencies in adulthood and more severe in elderhood. We further compared the genotype and phenotype among different populations, and we found that the genotype of LOXHD1, the severity of HL, the type of HL, and the audiogram had signi cant differences (P <0.05) ( Table 4). The profound HL, serious form HL, and gently sloping audiogram HL in the Middle East are obviously common than others.

Genotype-phenotype correlations analysis
About 60% of variants were associated with a milder auditory phenotype, with residual hearing at low frequencies (250 Hz, 500 Hz). Moreover, 19% of the variants that cause serious phenotype were located in the PLAT14 domain, indicating that it is the most frequent domain with a serious phenotype (Fig. 3B). 55% of the variants with milder phenotype were missense variants (Fig. 3C). We further compared the different combinations of LOXHD1 mutations. The results showed that compared with homozygous variants of LOXHD1, individuals with heterozygous compound variants had a signi cantly milder phenotype, especially individuals carrying one missense and one splicing or bi-allelic missense variants (P <0.05) ( Table 5).

Discussion
In this study, we investigated three isolated pediatric patients with six novel variants in LOXHD1, including three missense, one nonsense, and two rare splicing variants. So far, about 100 cases of DFNB77 have been reported in the literature. However, few data have been published with the genotype and audiology phenotype correlation.
Presently, we describe in detail the hearing loss in the three patients. All patients had a moderate-to-severe sensorineural hearing loss, which was milder in the low frequency. No increases in interpeak latencies were observed in the ABR, indicating a probably peripheral hearing loss. Speech audiometry corresponded well with pure tone audiometry, but the two patients (#1 and #2) performed poorly on speech in noise tests. Based on the above ndings: HL with a reduced speech in noise performance, absent DPOAE, and normal ABR waves, we speculate that outer hair cells might be lost or that their functionality is reduced [17]. Moreover, this nding consistent with the previous murine study [5].
The six LOXHD1 variants identi ed in this study were all in a compound heterozygous type. Our review also demonstrates that a compound heterozygous state of LOXHD1 variants is the most common combination in DFNB77, except in the Middle East population. In the Middle East population, all LOXHD1 variants were in a homozygous state, which may be caused by the custom of consanguineous marriage. In this study, 83% (5/6) of the variants located in the PLAT domains supporting our review result that most pathogenic LOXHD1 variants are located in the PLAT domains. It re ects that the PLAT domains are crucial for maintaining the normal function of LOXHD1 protein.
Our review indicates that patients with LOXHD1 variants mainly present a prelingual HL. The onset age of LOXHD1-related HL was initially described at 7-8 years based on the patients' own words [5]. However, recent studies suggested that the affected patients mainly had a congenital or early-onset HL [18][8] [9] [19]. Nevertheless, some DFNB77 patients were reported with a late-onset HL, even onset in adulthood [20] [21]. In this study, all three patients presented with prelingual HL while they had delayed diagnosis. According to the parents' information, it was because of the pass of the neonatal hearing screening (patient #1) or inconclusive results of the screening (patients #2 and #3). Therefore, it is likely that the late-onset HL in these reported late-onset individuals was of delayed diagnosis rather than true late-onset. Besides, for mild-to-moderate HL that occurs in early childhood, the diagnosis may be delayed due to the conversational speech within this range (50-60 dB) [22].
This study suggests that LOXHD1-related HL in childhood is a milder form of HL, milder at low frequencies (250 Hz, 500 Hz). A murine study has demonstrated that the first damaged part of cochlear hair cells was the stereocilia in the basal cochlear turn, which respond to the highest frequencies [5], indicating low frequencies should have relative milder damage. A milder HL was also reported in some patients with LOXHD1 variants [5] [8]. However, studies reported that some LOXHD1 variants are associated with a more severe HL [23] [18] [9] [19]. Our review data re ected that about 60% of variants were associated with a milder auditory phenotype and the HL degree in childhood was a milder HL.
Our study implies that the LOXHD1-related HL is progressive and deteriorated at all frequencies during adulthood. A mice experiment demonstrated that the LOXHD1 mutation could lead to degenerative changes in the hair cells of the inner ear's stereociliary [5], indicating that the LOXHD1-related HL should be progressive. However, to date, 42 of 111 patients have been reported with a non-progressive HL. Nevertheless, these results were questioned because only 8 (19%) of the patients received hearing tests in adulthood, and none was conducted with a long-time follow-up [8] [23][24] [9]. Noteworthily, our review showed that 37.5% of DFNB77 patients were reported with progressive HL during childhood, and 6.3% had progressive HL during adolescence. Besides, the degree of HL increased in all frequencies in adulthood and more severe in elderhood. Therefore, we guess that the LOXHD1-related HL aggravates gently during childhood and adolescence and rapidly during adulthood. Moreover, comparing the HL progression among different LOXHD1 variants, we found it was signi cantly different. Thus, except for age, the LOXHD1 variant combinations should also be considered in evaluating the HL progression for DFNB77 patients.
Previous studies have investigated the genotype-phenotype relationships in DFNB77, but no correlation was found due to the limited cases [8] [9]. We explored the genotype characteristics associated with the severity of LOXHD1-related HL to explain the heterogeneity in phenotype. First, variants in the homozygous state seem to cause a more severe HL than variants in the heterozygous state (Table 5). However, most homozygous variants were reported in patients from the Middle East; and they had more severe HL than patients from East Aisa, America, and Europe, where heterozygous variants are common. Thus, we suspect the difference is caused by a race-speci c effect. Second, the heterozygote combination of one missense and one splicing or bi-allelic missense variants are more likely to lead to a milder HL than other combinations (Table 5). In both situations, a certain amount of wide-type protein may remain expressed [25].
Third, the type of missense and splicing variant is more likely to cause a milder HL than other variant types (Fig. 3C). The homozygous missense and splicing variant also had milder HL than the homozygous frameshift, nonsense, and indel variants (Table 5). Fourth, the variants, which are located in the PLAT 7 and PLAT 9 domain, or may only affect 2 of the 5 LOXHD1 isoforms (transcript variant 2, NM_001145472.3; transcript variant 5, NM_001308013.1), are more likely to cause a milder HL (Fig. 3B) [8]. Moreover, this result ts different populations ( Supplementary Fig. 2). Sixth, variants in the PLAT 14 domain seems more likely to lead to a more severe HL (Fig. 3B). Nevertheless, 71% (5/7) of missense variants in this domain were milder HL, and in East Aisa, it reaches 80% (4/5).
Thus, the HL severity in the PLAT 14 domain is ambiguous.

Conclusions
We identi ed six novel variants in the LOXHD1 gene in three Chinese children with prelingual sensorineural HL. Our review indicates that recessive LOXHD1 variants mainly lead to a prelingual and down-sloping HL, which is milder in childhood and aggravates rapidly in adulthood. Our work extends the spectrum of LOXHD1 pathogenic variants, enriched our knowledge of auditory phenotypes, and provided potential genotype-phenotype correlations.

Ethics approval and consent to participate
This study protocol was approved by the Ethics Committee of Children's Hospital, Fudan University (2019-094 and 2019-214). Informed consents for participation in this study were obtained and could be available by requesting the corresponding author.

Author contributions
Zheng-min Xu. and Sha Yu. designed the study. Huijun Wang. supervised the work and completed the genetic data analysis. Other co-authors collected the clinical data. Sha Yu. and Huijun Wang. drafted the manuscript together. All authors joined in the editing and review of the manuscript. All authors approved the nal manuscript.
Availability of data and materials Some of our data supporting this study's ndings are not publicly available because of privacy or ethical restrictions, but the data are available by requesting the corresponding author.   Note: Detail data of these 111 patients in literature are available in Table S2.
Abbreviations: NA, data were not available; HL, hearing loss.    Table S2.
Data are presented with n (%).
Abbreviations: HL, hearing loss; NA, data were not available. a p-Value was calculated using Fisher's exact test. p-Value <0.05 is statistically signi cant*.
b Milder hearing loss means milder at low frequencies (250 Hz, 500 Hz).
c Serious hearing loss means no report of hearing loss milder at low frequencies.