Three novel variants in SOX10 gene: Waardenburg and PCWH syndromes

Waardenburg syndrome (WS) is a rare genetic disorder characterized by musculoskeletal abnormalities, deafness and hypopigmentation of hair and skin. This article’s aim is to investigate clinical and genetic characteristics of WS in three unrelated Caucasian individuals. The first patient was a 25-year-old female with congenital bilateral hearing loss, bright-blue-eyes, hypopigmentation of hair and skin, megacolon, language retardation, tenosynovitis and neuromas. The second case was an infant symptomatic from birth, with dysphagia, Hirschsprung disease and neurological abnormalities. The third patient was a 14-year-old boy with congenital bilateral hearing loss and ileocolic Hirschsprung disease. In order to identify variants in potentially causal genes of the patients’ phenotype, genetical testing was conducted: targeted clinical exome, targeted exome and trio exome, respectively. We identified three novel variants spread throughout the coding sequence of SOX10. The c.395C>G variant identified de novo in patient 1 was a single nucleotide substitution in exon 2. The c.850G>T variant identified as heterozygous in patient 2 was a loss-of-function variant that generated a premature stop codon. The c.966dupT variant identified in patient 3 was a duplication that generated a premature stop codon. It had been identified in his father, arising a possible germinal mosaicism. According to in silico predictors the variant identified in patient 1 was considered as pathogenic, whereas the other two were classified as likely pathogenic. An exact description of the mutations responsible for WS provides useful information to explain clinical features of WS and contributes to better genetic counselling of WS patients.


Background
Waardenburg syndrome (WS), also known as auditorypigmentary syndrome, is a rare genetic disorder typically characterized by musculoskeletal abnormalities, deafness and hypopigmentation of the skin. Basic clinical symptoms of WS include dystopia of the canthus; abnormal pigmentation of the skin, hair, and eyes; different degrees of unilateral or bilateral sensorineural deafness; and high and wide nasal base [1].
WS is inherited in an autosomal dominant manner and has an incidence of 1 in 20,000. Four distinct clinical subtypes of WS (WS1-4) are defined based on the presence or absence of additional symptoms. Facial dysmorphic features are present in WS1 (MIM#193500) and WS3 (Klein-Waardenburg syndrome; MIM#148820) patients. The absence of additional features characterizes WS2 (MIM#193510), whereas an association with Hirschprung disease (HD, aganglionic megacolon or gastrointestinal atresia) defines WS4 (MIM#277580), also known as Shah-Waardenburg syndrome or Waardenburg-Hirschprung disease [2]. WS is both phenotypically and genetically heterogeneous [3]. Mutations in six genes have been identified to be associated with WS: paired box gene 3 (PAX3), melanocyte inducing transcription factor (MITF), snail family transcriptional repressor 2 (SNAI2), SRY-box 10 (SOX10), endothelin 3 (EDN3) and endothelin receptor type B (EDNRB). De novo mutations in the SOX10 gene are rarely observed, even if heterozygous SOX10 mutations have been described in 15% and 45-55% of WS2 and WS4 patients respectively. Distinct classes of SOX10 mutations are responsible for a more severe phenotype of this condition evolving peripheral and central nervous system known as PCWH syndrome, that comprises peripheral demyelinating neuropathy, central dysmyelinating leukodystrophy, Waardenburg syndrome and Hirschsprung disease [4,5].
This article's aim is to investigate clinical and genetic characteristics (using Sanger and whole-exome sequencing) of WS and PCWH in three unrelated Caucasian individuals. Interpretation of the variants was based on the guidelines of American College of Medical Genetics and Genomics (ACMG) [6].

Case presentation
We identified three unrelated Spanish individuals with an unusual but similar clinical phenotype. They did not have family history of congenital or neurological disorders, and their parents were non-consanguineous. A comprehensive clinical history was gathered and clinical examinations were meticulously performed for each affected individual. Written informed consent was obtained from patients included in this study or their parents.

Patient 1
A 25-year-old female with congenital bilateral hearing loss (Fig. 1A), chronic constipation due to megacolon, gastroesophageal reflux, diabetes and mild language retardation, was referred for asymmetric pain in the lateral aspect of the legs, triggered by exercise and relieved by rest, which was attributed to tenosynovitis.
On examination, bright blue eyes with partial heterochromia iridis, hypopigmentation of hair (white forelock) and skin (café-au-lait spots on the right hand, breasts, right thigh, knees and scalp, Fig. 1B) and pes cavus, were found. Neuromas were present in the left arm, right leg and forehead.
On neurological examination, tendon reflexes were 1/4 in upper limbs and absent in the lower ones. Muscle bulk and power, stance, gait and pain and vibration sense were preserved. Coordination and cranial nerves were normal.
An electrophysiological study revealed decreased sensory (31 m/s) nerve conduction velocity (NCV) in the right median nerve, and absent action potentials in both sural nerves. Motor NCV was decreased in the median, peroneal and tibial nerves, while increased motor latencies were found in the median nerve.
Referred for genetic study by Multiplex Ligationdependent Probe Amplification (MLPA) of the PMP22 gene and obtaining a negative result, she then went through a massive sequencing study (targeted clinical exome, ExoNIM ® , NIMGenetics, Spain) with the aim of identifying variants in 103 selected genes associated with Charcot-Marie Tooth and other HMSN. Four variants of uncertain significance were obtained.
As so, genetic testing was amplified with a massive sequencing of the complete human exome of the patient and his parents (ExoNIM ® Trio Approach, NIMGenetics, Spain). The c.395C>G p.(Ala132Gly) variant was identified de novo in the case sample consisting in a single nucleotide substitution in exon 2 of the SOX10 gene (transcript ID: LRG_271). According to the ACMG variant classification guidelines, this missense variant was classified as pathogenic since PS2, PM1, PM2_supporting, PP2, PP3 and PP4 criteria were met.

Patient 2
Infant symptomatic from birth, with dysphagia, a diagnosis of Hirschprung disease and carrier of gastrostomy and ileostomy. Neurologically, abnormal tendon reflexes, rhythmic and horizontal head jerks, limb tremor and rotating eye movements were observed from day 15 of life. The patient died aged 20 months due to complications derived from intestinal obstruction.
With these clinical features, we were asked to carry out an array comparative genomic hybridization (aCGH, qChip ® Post, qGenomics, Spain), with no relevant findings. Then, we amplified the genetical study by conducting a targeted exome (ExoNIM ® Clinical, NIMGenetics, Spain) in order to identify variants in 6102 selected genes that may present an association with our case's phenotype. The heterozygous c.850G>T p.(Glu284*) variant identified in SOX10 (transcript ID: LRG_271) in the patient's sample was a nonsense variant classified by ACMG as likely pathogenic (PVS1_strong, PM2_supporting and PP4 criteria).

Patient 3
A 14-year-old boy with congenital bilateral hearing loss, required cochlear implants at one year of age. He had no heterochromia iridis or skin hypopigmentation. The patient was diagnosed with ileocolic Hirschprung disease at 6 months, with ileostomy discharge at 15 months. A colectomy was performed to remove the non-functional PVS1_strong, PS2, PM2_supporting and PP4 ACMG criteria.

Discussion
Three novel SOX10 variants were identified including c.850G>T, c.966dupT and c.395C>G (Table 1). The  [4,7,8]. SOX10, in synergy with PAX3, acts as a transcription factor of the MITF gene, playing a key role in the development of melanocytes [9]. Moreover, the hearing loss in WS may relate to abnormal proliferation, survival, differentiation or migration of NCC-derived melanocytes (a type of intermedia cells of stria vascularis) [10]. Lastly, SOX10 may play roles in activating several previously undescribed genes, including genes involved in integrin and focal adhesion kinase signaling [11]. All this may explain the wide range of phenotypic abnormalities caused by reduced SOX10 expression.
In our study, we identified three novel variants spread throughout the entire coding sequence of SOX10 in three patients with WS4 (patient 3)/PCWH (patients 1 and 2). The pathogenicity of the variants was supported by in silico predictors, sequence conservation analysis and cosegregation studies of the variants within families. Two of the identified variants were in exon 4 (c.850G>T and c.966dupT), while the other was found in exon 2 (c.395C>G). Previous studies have suggested that more severe phenotypes, including PCWH, were associated with truncating mutations in the last coding exon (exon 4), because the resulting mRNAs escaped NMD, leading to production of the mutant protein and triggering of a potent dominant-negative process. In contrast, truncating variants with mutations in the first exons (exons 2 and 3) were considered to activate the NMD machinery, resulting in less-severe phenotypes. The fact that missense mutations are not subject to NMD are use in favor or other mechanism at the origin of the phenotype variability. Earlier results suggested that the phenotypes could not actually be predicted based on the genotype alone and that additional information from in vitro studies is essential for making genotype-phenotype predictions [9,12]. In our case, the c.395C>G p.(Ala132Gly) variant identified de novo in the patient 1 sample was a single nucleotide substitution in exon 2 of the SOX10 gene, which changed the amino acid Ala132 for a Gly. This missense variant, without associated population frequency, had not been previously registered in the literature or in the consulted databases associated with a specific phenotype. However, it was located in a hotspot region in which pathogenic missense variants had been described in adjacent codons associated with WS and PCWH [12][13][14]. According to splice AI, HSF, and MaxEnt tools, the c.395C>G variant could possibly create a new donor splice site, which would lead to a frameshift in the spliced mRNA. Based on the results of several in silico prediction tools (16/21), conservation analysis, population frequency, and de novo character of the variant, it was considered a pathogenic variant. The SOX10 c.850G>T p.(Glu284*) variant identified as heterozygous in the patient 2 sample was a loss-offunction variant that generated a premature stop codon, resulting in an early termination of the coding protein sequence at amino acid position 283. This change could have made the mutant messenger RNA (mRNA) escape the nonsense-mediated mRNA decay (NMD) pathway, a cellular surveillance mechanism for mRNA that prevents the expression of truncated or erroneous proteins; or generated a truncated protein with a negative dominant effect. This variant had not been previously registered in the literature or in the consulted databases associated with a specific phenotype. Thus, it was considered a likely pathogenic variant according to all the consulted in silico predictors (9/9), conservation scores and population frequency.
The SOX10 c.966dupT p.(Ala323Cysfs*79) variant identified in the samples from the patient 3 and his asymptomatic father was a duplication that caused a change in the reading pattern, generating a premature stop codon and resulting in a mutant protein consisting of a sequence of 400 amino acids, with a divergent frameshift segment. This change escaped the NMD mechanism as it was located in the last exon of the gene, although variants of loss of function in posterior positions have been previously described as pathogenic in the ClinVar database (Variation ID: 7402, 518422, 805532). Likewise, this variant had been identified in the father with an allelic frequency of 7%, arising a possible germinal mosaicism, previously described in the literature [13,15]. Given the absence of the variant in the control population and its effect on the protein, the obtained in silico predictions and the conservation analysis, pending further studies, it was considered a likely pathogenic variant.

Conclusions
These three cases highlight the importance of detailed clinical phenotyping in cases in which HMSN are suspected, since non-neuropathic phenotypic characteristics warn of possible rare causes and should be taken into account in the performance and interpretation of genetic tests, especially when massive sequencing techniques are used.
Mutation detection in WS is important, as this syndrome is genetically heterogeneous. In addition, SOX10 mutations give a recurrence risk of 50% for offspring. As information on the functional characterization of SOX10 variants is limited, and phenotypes cannot be predicted accurately based on the genotype alone, further investigation is needed. In vitro studies of each variant are necessary to provide a better understanding of the pathogenic mechanisms and better phenotype predictions, while in vivo analysis would further improve our understanding of the phenotypic effects of SOX10 variants. An exact description of the mutations responsible for WS provides useful information to explain clinical features of WS and contributes to better genetic counselling of WS patients.