New Genotypes and Phenotypes in Patients with 3 Subtypes of Waardenburg Syndrome Identified by Diagnostic Next-Generation Sequencing

Background Waardenburg syndrome (WS) is one of the most common forms of syndromic deafness with heterogeneity of loci and alleles and variable expressivity of clinical features. Methods The technology of single-nucleotide variants (SNV) and copy number variation (CNV) detection was developed to investigate the genotype spectrum of WS in a Chinese population. Results Ninety WS patients and 24 additional family members were recruited for the study. Fourteen mutations had not been previously reported, including c.808C>G, c.117C>A, c.152T>G, c.803G>T, c.793-3T >G, and c.801delT on PAX3; c.642_650delAAG on MITF; c.122G>T and c.127C>T on SOX10; c.230C>G and c.365C>T on SNAI2; and c.481A>G, c.1018C>G, and c.1015C>T on EDNRB. Three CNVs were de novo and first reported in our study. Five EDNRB variants were associated with WS type 1 in the heterozygous state for the first time, with a detection rate of 22.2%. Freckles occur only in WS type 2. Yellow hair, amblyopia, congenital ptosis, narrow palpebral fissures, and pigmentation spots are rare and unique symptoms in WS patients from China. Conclusions EDNRB should be considered as another prevalent pathogenic gene in WS type 1. Our study expanded the genotype and phenotype spectrum of WS, and diagnostic next-generation sequencing is promising for WS.


Introduction
Waardenburg syndrome (WS) is a rare genetic disorder with a reported frequency estimated to be approximately 1 : 40000 in the general population but is present in approximately 3% of all patients with congenital deafness [1,2].The disorder is characterized by the presence of pigmentation abnormalities, including depigmented patches of the skin and hair, heterochromia iridis, and sensorineural hearing impairment.Dystopia canthorum, musculoskeletal abnormalities of the limbs, or Hirschsprung disease, are used for the clinical classification.Four subtypes of WS have been described based on the clinical manifestations.Six genes are involved in this inherited disorder, including PAX3 (encoding the paired box 3 transcription factor), MITF (microphthalmia-associated transcription factor), SOX10 (encoding the Sry box 10 transcription factor), EDNRB (endothelin receptor type B), EDN3 (endothelin 3), and SNAI2 (SNAIL homolog 2).WS type 1 (OMIM # 193500) was first described by Waardenburg [3].Dystopia canthorum, an outward displacement of the inner canthus of the eyes, is the most penetrant feature of WS type 1. PAX3 mutations account for the majority of WS type 1 cases.Features of WS type 2 (OMIM #) show marked interfamilial and intrafamilial variability.There are 3 genes linked to WS type 2, namely, MITF, SOX10, and EDNRB.However, pathogenic genes cannot be detected in 70% of WS type 2 cases [4].WS type 3 is associated with limb deformities together with the symptoms observed in type 1. PAX3 mutations have also been found in the heterozygous or homozygous state in WS type 3 (OMIM # 148820) [5].WS type 4 (OMIM # 277580), also called the Shah-Waardenburg syndrome, is characterized by the association of deafness, depigmentation, and intestinal aganglionosis (called Hirschsprung disease (HD)).The endothelin pathway (endothelin 3 (EDN3), endothelin receptor type B (EDNRB), and Sry box 10 (SOX10)) was found to be involved in WS type 4 [6,7].
It has been determined that pathogenic mutations were not and still cannot be detected in a considerable number of WS cases.Other types of variations may exist, given the limitations of conventional detection technology.Copy number variation (CNV) is a new topic of increasing interest in genetic research.In addition, CNV has been reported to be associated with WS [8][9][10].Recently, one study addressed the molecular etiology investigation of WS in individuals mostly from southeastern Brazil by sequential Sanger sequencing of all coding exons of the 6 WS-associated genes, followed by CNV detection by multiplex ligation-dependent probe amplification (MLPA) of the PAX3, MITF, and SOX10 genes, and revealed novel pathogenic mutations [11].Traditional sequencing methods are designed for point mutation detection without considering the possibility of CNV.Exon capture sequencing for simultaneous SNV and CNV detection in WS has been developed in our study.
To date, there is no large cohort on the genotype and phenotype spectrum of WS patients from China.In this study, we aimed to investigate the genetic etiology and phenotype differences using target exon capture of 6 known causative genes in WS.

Patients and Family
Members.The patients diagnosed with WS were examined occasionally at the Otology Clinic and training schools for deaf and mute individuals in China from September 2006 to February 2018.There were 114 participants, including sporadic WS cases and 18 families (Figure 1, Table 1).All the WS patients were clinically evaluated by at least one otologist.The clinical signs and symptoms of the 90 patients diagnosed with WS are presented in Table S1.Two hundred randomly selected normal hearing individuals were included in this study.Blood samples (4-6 ml) were extracted from the peripheral veins of all the participants for DNA extraction.The ethics committee of Xiangya Hospital, Central South University, had approved this study, and signed informed consents were obtained from each of the subjects or their guardians.

Clinical Evaluation.
A comprehensive clinical history was collected by questionnaire and telephone inquiry.The audiological examinations consisted of otoscopy, pure-tone audiometry (PTA) or auditory brainstem response (ABR), immittance, and distortion product otoacoustic emission (DPOAE).Special attention was given to the color of the skin, hair, and irises and other developmental defects.Assessment of dystopia canthorum on the basis of ocular measurements was described by Farrer et al. [12].
2.3.Exon Capture Sequencing for Simultaneous SNV and CNV Detection in WS.The primer sequences for 6 WS-related genes were designed for the target regions (Table S2), and the regions of interest were captured and enriched.PCR amplification was divided into two rounds.The first round of PCR amplification was to amplify the 6 relevant WS genes.The amplification region includes the promoter regions (~500 bp), 5 ′ untranslated region (5 ′ UTR), coding regions, splice sites (~8 bp), and 3 ′ untranslated region (3 ′ UTR) (Table 2).The multiplex PCR amplification system can amplify 20-30 gene fragments at the same time.Each sample requires approximately 4 multiplex PCRs to complete the first round of enrichment.The second round of PCR added the 4 multiplex PCR products above into a mixture by amplifying the universal sequence.Indexes were added to distinguish between different samples.After these two rounds of amplification were finished, the WS sequencing library was constructed.Bidirectional sequencing validation of the target segments was performed by 2x 250 bp sequencing with an Illumina MiSeq Sequencer.The average effective sequencing depth for each sample was 300x, with all bases having greater than 20x sequencing depth (Figure 2) The CNV detection technique utilizes ligase to hybridize and ligate the region of interest.Then, different lengths of the ligated products corresponding to the loci were obtained by introducing nonspecific sequences of different lengths to the ends of the ligation probes and performing a ligation reaction.The PCR product was amplified by fluorescently labeled universal primers.The amplified products were separated by fluorescence capillary electrophoresis and analyzed by electrophoresis.The peak height at each site was analyzed 2.4.Bioinformatics Analysis.After mapping WS-related gene sequences to the DNA samples, the results were aligned with the reference genome.Sequencing data quality was assessed through the sequencing depth of each fragment of each sample.Inconsistent sites were detected by comparison with the reference genome, which is called SNV calling.Then, SNVs were functionally annotated and functional candidate sites were determined.

Sanger Sequencing for the Segregation of Candidate
Variants.DNA samples of affected siblings and their available first-degree relatives were collected for segregation analysis of candidate variants.For validation and segregation analyses, the primers were designed to amplify the regions flanking the variant as previously described [4,13].Segregation analysis by Sanger sequencing was also performed in 200 ethnically matched individuals with normal hearing.These included variants annotated as nonsense mutations, splicing mutations disrupting either a splice donor or acceptor site,  NA in gene: the molecular etiology of cases that remained unexplained.NA in protein mutation: protein changes that cannot be described.4 Neural Plasticity recorded in the population database with a minor allele frequency of <1/100000 in the population from the database was considered disease causing and therefore remained.

Results
Most proteins associated with the known WS genes are involved in melanocyte migration and neural crest and inner ear cell development.

Neural Plasticity
Han race is black.Hypoplastic iridis, particularly brilliant blue eyes, was present in 86.7% (78/90) of WS patients.The heterochromia was observed to be complete or segmental.The following prevalent phenotype included freckles on the face (20.0%, 18/90), which seems to be a special clinical sign in the Chinese population, given its high occurrence rate.The rate of hypopigmented skin lesions was more rare than that in other populations [18], with only 3 in 90 cases (3.3%) being found.Pigmentation spots (2.2%, 2/90) on the skin might be a special subtype of skin pigmentation disturbances.Other rare and unreported phenotypes, including amblyopia, congenital ptosis, and narrow palpebral fissures, were recorded to accompany WS, but no clear evidence has been found showing that these phenotypes relate to neural crest embryonic cell defects at present (Figure 3).
The subtypes of WS were defined on the basis of the presence or absence of additional symptoms.WS type 1 was characterized by dystopia canthorum and WS type 2 with no additional features.Type 4 was also called Shah-Waardenburg syndrome, Hirschsprung disease included.Twenty-seven cases were diagnosed as WS type 1, 57 as WS type 2, and 6 as WS type 4.There were no WS type 3 cases recruited.WS types 1 and 2 were more frequent than type 4. The diagnostic next-generation sequencing in all 114 participants revealed 119 variants (Table S3); however, only 90 cases were diagnosed as WS according to the widely accepted diagnostic criteria [19].Of the 90 WS patients, 49 were considered causative in WS and recorded in the Human Gene Mutation Database (HGMD) with a known disease mutation detection rate of 53.3% (49/90).Fourteen unreported mutations were detected and selected for further bioinformatics pathogenicity analysis (Table 3).Three CNVs discovered are listed separately in Table 4. Figure 4 shows the position in the protein domains of the new SNVs detected in our study.WS type 2 is mainly caused by heterozygous pathogenic variants in genes MITF (24/57, 42.1%), PAX3 (10/57, 17.5%), and SOX10 (9/57, 15.8%).Homozygous deletions in SNAI2 were also reported to associate with WS type 2 cases [20].One variant was detected in our research, which indicated that SNAI2 mutations seem to be rare in WS type 2 among Eastern and Western populations.In contrast with the reported findings, EDN3 mutations were also detected in the WS type 2 cases (Subjects 72, 76, and 104) in our study.Even after extensive analysis of the 6 known genes, a percentage of the WS type 1 and type 2 cases, 14.8% (4/27) and 26.3% (15/57), respectively, remained molecularly unexplained (Figure 4).No WS type 3 cases were recruited in our study.The EDNRB, EDN3, and SOX10 mutations [21,22] are related to WS type 4, and close to 80% of the WS type 4 cases were found to be caused by mutations in SOX10 in our study.
Two CNVs and 3 SNVs, including c.110_219del110bp in MITF, the whole-gene deletion in SOX10, c.801delT on PAX3, c.642_650delAAG on MITF, and c.127C>T on SOX 10, resulted in a truncated protein with a premature termination, which were loss-of-function (LoF) mutations and therefore considered to be cause of disease.The last CNV meant a duplicate sequence from promoter 2 to exon 1 on MITF, which might be one of the reasons for a gain-of-function (GoF) mechanism.Therefore, protein function is activated.The expression or degradation of proteins resulted in increased protein dosage.All the remaining missense substitutions were selected for pathogenicity prediction and population frequency, and the forecast results and data are shown in Table 3.

Discussion
WS is a genetic disorder with locus heterogeneity and variable expressivity of clinical features [23].WS type 1 and type 2 are conspicuously differentiated by the presence or absence of dystopia canthorum among populations [18].In this present study, we have identified mutations in PAX3 in cases (Subjects 2, 4, 6, 49, 51, 54, 90, 91, 98, and 104) without dystopia canthorum, typical WS type 2 characteristic; conversely, Subjects 67, 68, and 110 with the clinical feature of WS type 1 (with the presence of dystopia canthorum) were found to carry MITF variants.Thus, it is difficult to establish a link between a genotype and a classical WS phenotype in our cohort.This implies that other factors including interactions between genes, gene-environment interactions, and ethnic background may modulate WS phenotypes.
Overall, 55.5% (15/27) of the cases of WS type 1 are caused by pathogenic variants in PAX3.A recent study suggests that homozygous mutation in EDNRB can cause the WS type 1 phenotype [24], whereas in our study, heterozygous mutations in EDNRB (Subjects 48, 65, 66, 85, 108, and 111) were related to WS type 1 for the first time, with a detection rate of 22.2% (6/27).

De novo
Novel refers to variants that were absent in 200 control subjects; de novo refers to variants that were absent in the parents and 200 control subjects.[27].In this study, we identified two novel SOX10 mutations, c.122G>T and c.127C>T, as associated with WS type 2 in the Chinese population.As described in Table 3, c.122G>T in SOX10 was not considered causative due its frequency, which was >1/10000 in the population database of ExAC_EAS.In addition, c.122G>T in SOX10 can be observed in Figure 4 that this mutation affects an amino acid residue located outside the high-mobility group box domain of SOX10.More specifically, the c.122G>T variant in SOX10 (Subjects 20 and 21, Figure 1) was inherited from the unaffected mother.The c.127C>T variant on SOX10 was regarded as likely causative, because of its in silico predictions of pathogenicity and it is absent in 200 normal controls and with no frequency in population databases.
The SNAIL-related zinc-finger transcription factor SNAI2 (Slug) is a member of the SNAIL family of zincfinger TFs that share an evolutionarily conserved role in mesoderm formation [28].SNAI2 is expressed in migratory neural crest cells (NCCs) and is indispensable for melanoblast survival or migration.We had detected 2 variants (c.230C>G and c.365C>T) in SNAI2 in the cohort of WS participants that were not considered causative due to their frequency (>1/10000) in the population database.Therefore, SNAI2 has a minor involvement in WS in the Chinese population.The endothelins are a group of three peptides (ET1, ET2, and ET3) [29].In vertebrates, Ednrb (encoding the endothelin receptor type B (ETB)) is first expressed at the dorsal tip of the neural tube, then in NCCs in both dorsoventral and dorsolateral pathways [30].The c.469A>G, c.553G>A, c.481A>G, c.1015C>T, and c.1018C>G mutations in EDNRB were detected in patients with WS type 1 in the present study in the Chinese population for the first time.Six cases were found to carry c.49G>A in EDN3 that has been linked to WS type 4 in the HGMD database.
The clinical manifestations of WS vary widely between different populations.In the present study, heterochromia iridis and sensorineural deafness were the most frequent features of both WS type 1 and WS type 2. Freckles were not observed in WS type 1 patients but were present in 31.6%(18/57) of type 2 patients, which might be a preidentified indicator between types 1 and 2 in Chinese populations.The present study and those of Silan et al. [31] and Tamayo et al. [32] were the primary screening programs in the institutionalized deaf populations in China, Turkey, and Colombia.In general, the distribution of WS type 2 is more common than that of type 1, which was in agreement with Silan et al. [31] and Tamayo et al. [32].However, the majority of reports [33] presented more cases of WS type 1.We were looking for WS cases in hospitals or schools for deaf and mute individuals.Hearing loss is more common in WS type 2, which might be the reason resulting in the consequence.A white forelock was reported and estimated to be present in at least one-third of both WS type 1 and 2 cases [18].However, the proportion of this phenotype in our population was much lower, which is different from the deaf populations in Turkey [31] and Colombia [32] and could be considered a difference among different populations.Ethnic migration may be the reason underlining this difference.Hypopigmented and depigmented patches on the skin can also be seen in previous reports from other populations [31,32,34].Other associated symptoms described, such as cleft/lip palate, spina bifida, and musculoskeletal anomalies, were not found in the present study.In contrast, some specific signs recorded in our study are rare and were not reported in other populations, such as yellow hair, amblyopia, congenital ptosis, and narrow palpebral fissures.
Three cases (Subjects 70, 72, and 100) with genotypes linked to WS type 4 have type 2 phenotype characteristics in the present study.The observation of different phenotypes in Family 2 with the same mutations may be due to heterochromia penetrance associated with a SOX10 mutation (Subjects 14 and 15).The presence versus absence of WS features in MITF (Subjects 28 and 29, Family 6) also argues for the influence of the genetic background, which supports the hypothesis that there is an interplay between genetic and environmental factors.Neural Plasticity would be time-consuming and expensive.Captured sequencing, also called diagnostic next-generation sequencing, has specifically targeted regions.Compared with whole-exome sequencing, the cost is greatly reduced and the depth of coverage can reach to 300x.The target region capture system in our study was designed to amplify and detect 6 WS pathogenic genes in one reaction system.The observations in our study have also proved that this technology would be a promising tool to identify the molecular etiology in WS.However, the molecular etiology of a variable fraction of cases remained unexplained [18,35], indicating that noncoding regions should also be included for molecular analysis.Moreover, it is possible that novel disease-causing genes are implicated in WS.Zazo Seco et al. [36] identified mutations in KITLG in WS families.Whole-exome sequencing and even whole-genome sequencing will fully reveal candidate and novel genes in molecularly unexplained cases of WS.

Figure 2 :
Figure 2: Multiple PCR target enrichment and next-generation Sequencing of WS-related genes.

3 :
Summary of the new SNV results of the molecular screening of WS-related genes, including location of mutations, pathogenicity predictions, and population data.
et al.[25] had screened PAX3, MITF, and SOX10 CNVs by MLPA and detected PAX3 and MITF in the search for pathogenic variants.However, other associated genes were not included in their cohort.Bocángel et al.carried out the molecular investigation of WS by sequential Sanger sequencing of all 6 coding exons; then, CNV detection by MLPA of PAX3, MITF, and SOX10 genes in selected cases followed[11].Detection by repeated Sanger sequencing

Figure 4 :
Figure 4: The proportion of genotypes detected in WS types 1, 2, and 4 and the position in the protein domains of the new SNVs detected in our study.

Table 1 :
Figure1: Eighteen WS families with at least 2 DNA samples and clinical information collected in the study.Family ID and Subject ID were added to the individuals with DNA samples.A table with all data for family cases was shown in Table1.All data for family cases.

Table 2 :
WS-associated genes and their PCR target regions.

Table 4 :
Summary of new structure variation (SV) or CNV-detected results of the molecular screening of WS-related genes, including location of mutations and population data.