Diversity of the Genes Implicated in Algerian Patients Affected by Usher Syndrome

Usher syndrome (USH) is an autosomal recessive disorder characterized by a dual sensory impairment affecting hearing and vision. USH is clinically and genetically heterogeneous. Ten different causal genes have been reported. We studied the molecular bases of the disease in 18 unrelated Algerian patients by targeted-exome sequencing, and identified the causal biallelic mutations in all of them: 16 patients carried the mutations at the homozygous state and 2 at the compound heterozygous state. Nine of the 17 different mutations detected in MYO7A (1 of 5 mutations), CDH23 (4 of 7 mutations), PCDH15 (1 mutation), USH1C (1 mutation), USH1G (1 mutation), and USH2A (1 of 2 mutations), had not been previously reported. The deleterious consequences of a missense mutation of CDH23 (p.Asp1501Asn) and the in-frame single codon deletion in USH1G (p.Ala397del) on the corresponding proteins were predicted from the solved 3D-structures of extracellular cadherin (EC) domains of cadherin-23 and the sterile alpha motif (SAM) domain of USH1G/sans, respectively. In addition, we were able to show that the USH1G mutation is likely to affect the binding interface between the SAM domain and USH1C/harmonin. This should spur the use of 3D-structures, not only of isolated protein domains, but also of protein-protein interaction interfaces, to predict the functional impact of mutations detected in the USH genes.

The spectrum of genes and mutations involved in Algerian USH patients is still poorly defined because only a few Algerian patients have been studied so far [23,24]. We analyzed the molecular bases of USH in 18 unrelated Algerian patients, 16 of them being affected by USH1 and the other two by USH2, by targeted-exome sequencing of the ten identified USH genes.

Ethics statement
This study was approved by the local ethics committee of the CHU Bab El-Oued, in accordance with the "Loi sanitaire n°85-05 du 16 février 1985" applicable in Algeria. This study was conducted according to the principles of the declaration of Helsinki. Patients were anonymized, and the corresponding code was conserved in a confidential file. Written consent for genetic testing was obtained from the adult participants and from the parents of minor participants.

Patients
Eighteen unrelated Algerian USH patients (age ranging from 19 months to 54 years old) from the ENT departments of CHU Blida Hospital (15 patients), CHU Bab El Oued in Algiers (one patient), and CHU Tizi Ouzou (2 patients) were included in this study, together with their clinically unaffected parents. Most patients (14 out of 18) were born to parents known to be consanguineous. Clinical history was taken for each patient. The patients underwent audiograms, ocular fundus autofluorescence imaging, and electroretinogram. All patients had bilateral, moderate to profound deafness, and the 12 patients aged at least five years also had reduced visual field and night vision. All USH1 patients had delayed onset of independent walking (> 18 months) and vestibular dysfunction. The parents of all patients had normal hearing and no retinal degeneration on fundus examination.

Targeted-exome sequencing
Genomic DNA was extracted from peripheral blood using standard procedures. Targetedexome sequencing and bioinformatics analysis were performed as previously described [25].
To amplify and sequence all coding exons of the 10 USH genes identified so far (i.e., a total of 366 exons), specific primers were designed using Primer3 (http://bioinfo.ut.ee/primer3-0.4.0/); the sequences of these oligonucleotides are listed in S1 Table. Sequences were run on ABI 3730 DNA analyzer, and compared with Genbank reference sequences (http://www.ncbi.nlm.nih. gov/genbank/) using ABI Prism Seqscape 2.1.
The segregation of the mutations identified in the patients was studied by Sanger sequencing of the corresponding DNA fragments in the parents and other relatives, whenever available. The Sorting Intolerant from Tolerant (SIFT), Mutation Taster, and PolyPhen-2 software programs were used to predict the impact of all amino acid substitutions on the protein structure and function. The NNSplice, ESEfinder, Max Ent Scan, Gene Splicer, and Human Splicing Finder programs were used to predict abnormal splice sites.

Reference sequences for mutation nomenclature and exon numbering
In this article, the nomenclature of all sequence variants and exon numbering refer to the following genomic and cDNA reference sequences (NG_ and NM_ NCBI accession numbers, respectively): All the mutations identified were deposited in the Leiden Open Variation Database (http:// www.lovd.nl/3.0/home).

Mutation modeling
The structural consequences of the two novel missense mutations in CDH23 and of the inframe single codon deletion in USH1G were analyzed on the 3D-structure models of the cadherin-23 EC1-EC2 & protocadherin-15 EC1-EC2 (PDB 4APX) and sans SAM-PBM & harmonin Nter-PDZ1 (PDB 3K1R) molecular complexes, respectively. The PyMOL software program was used for cartoon representations.

Results and Discussion
In a cohort of 180 Algerian deaf patients, we recruited 18 unrelated USH patients (16 USH1 and 2 USH2 patients) on the basis of associated vestibular dysfunction (including delayed onset of independent walking in all 16 USH1 patients) and/or reduced night vision. The presence of a retinal dysfunction was confirmed by electroretinogram analysis. By targeted-exome sequencing of the known USH genes, we found biallelic mutations in all the patients. Sixteen patients carried mutations at the homozygous state, and two patients at the compound heterozygous state. A total of 17 different pathogenic or presumably pathogenic mutations were identified in five USH1 genes (MYO7A, CDH23, PCDH15, USH1C, and USH1G) and in USH2A: 4 nonsense, 1 frameshift, 3 splice-site, and 6 missense mutations, a synonymous mutation predicted to create a splice acceptor site, an in-frame deletion of one codon, and a large intragenic deletion (Table 1, Fig 1, and S1 Fig).
Nine mutations have previously been reported (see Table 1). The p.Arg212His and p. Gly163Arg missense mutations in MYO7A (see Table 2) have also been found in Belgian, Dutch, and American patients [26] and in Algerian and Turkish patients [27], respectively. The c.2283-1G>T splice site mutation in MYO7A, first described in an Algerian patient [27], seems to be frequent in Maghrebian populations [28]. We found this mutation in three Algerian patients, at the homozygous and compound heterozygous states. The other splice site mutation in MYO7A (c.470+1G>A) has been reported in Tunisian patients [29]. The p. Gln2113 Ã , c.6829+1G>A, and p.Gly291Arg mutations in CDH23 (see Table 2) have been found in an American family [30], a French family [31], and a Spanish family [32], respectively. The p.Thr5006Met missense mutation in USH2A (see Table 2) has been reported in French and Danish patients [33]. Finally, the c.2299delG ancestral mutation in USH2A [16] accounts for 47.5% and 15% of the mutant alleles identified in USH2A patients living in Denmark and in Spain, respectively [34,35], but, to our knowledge, had not yet been reported in patients from North Africa.
The other eight mutations had not been previously reported. Six were present at the homozygous state (6 USH1 patients), and two were present at the compound heterozygous state in an USH1 patient (see Table 1). These mutations were absent in HapMap, 1000 genomes, Exome Variant Server, and Deafness Variation databases, and in the Liveset database (LSDB) for USH genes, and they were not detected in 150 Algerian control individuals. The two missense mutations are located in CDH23 exon 3 (p.Tyr39His) and exon 36 (p.Asp1501Asn), encoding the first and 14 th extracellular cadherin (EC) domains of cadherin-23, respectively. Both mutations were predicted to be pathogenic by PolyPhen-2, Sorting Intolerant from Tolerant (SIFT), and Mutation Taster (see Table 2).
The two most amino-terminal EC repeats (EC1+EC2) of cadherin-23 and protocadherin-15 can interact to form an overlapped, antiparallel heterodimer that is believed to be the structural basis of apical fibrous links joining inner ear hair cells' stereocilia [36]. The Tyr39 residue of cadherin-23 belongs to the loop between the 3 10 helix and A strand of EC1, and is located at the cadherin23-protocadherin15 interaction interface, but the Tyr39His mutation does not appear to perturb the hydrophobic environment of this surface according to the structural model of the interaction (Fig 2A). However, we cannot exclude the possibility that Tyr39 plays an important role in another mode of cadherin-cadherin interaction that might also be involved in stereocilia links, especially since this residue is located at the N-terminal end of the  S2 Fig), and the solved 3D-structure of EC1-EC2 indicates that the equivalent aspartic acid residue in the EC1-2 linker contributes to the Ca 2+ -binding site (Fig 2B). The substitution of the negatively charged Asp1501 by Asn is predicted to disrupt the corresponding Ca 2+ -binding site (Fig 2B). The novel in-frame deletion of three nucleotides identified in USH1G (p.Ala397del) removes an alanine residue from the first α-helix of the sterile alpha motif (SAM) domain. This helix (αA) is part of the SAM domain hydrophobic core formed by a bundle of four αhelices. Modeling of the mutated SAM domain lacking the Ala397 residue predicts a disruption of the hydrophobic core (Fig 3A). In addition, Ala397 contributes to the interface of interaction between sans (USH1G) and harmonin, the PSD95-discs large-ZO1 (PDZ) domain-containing   protein encoded by USH1C [14], [37], and the absence of this residue is predicted to affect the docking of the SAM domain αA helix to the surface groove of the harmonin PDZ1 domain (Fig 3B). Incidentally, among the twelve presumably pathogenic mutations previously reported in USH1G, there were only two (monoallelic) missense mutations affecting residues located in the SAM domain, p.Leu420Val and p.Arg447Trp, none of which is predicted to affect the binding to harmonin [38,39]. The three novel nonsense mutations are located in MYO7A exon 39 (p.Lys1810 Ã ), encoding the second myosin tail homology 4 (MyTH4) domain of myosin VIIa, in USH1C exon 10 (p. Glu260 Ã ), encoding the PDZ2 domain of harmonin (common to all harmonin classes and subclasses), and in CDH23 exon 10 (p.Gln362 Ã ), encoding the 4 th EC domain of cadherin-23. In addition, the novel c.5850T>A (p.Ser1950Ser) synonymous mutation in CDH23 is predicted to create a splice acceptor site in exon 44 (encoding the 18 th EC domain of the protein), which also results in a premature stop codon in the mature transcript. All these mutations are expected to result either in a truncated protein or in no protein at all owing to nonsense mediated mRNA decay [40].
Finally, a biallelic deletion encompassing PCDH15 exons 10 to 14 (p.Glu293_Gln530del) was identified in the USH1F patient, and confirmed by quantitative PCR analysis of these exons (data not shown). This large in-frame deletion is expected to affect protocadherin-15 EC3-EC5. The precise boundaries of the deletion were established by PCR amplification on the patient's genomic DNA with specific primers in introns 9 and 14, followed by Sanger sequencing of the amplicon (Fig 4).
These results highlight the genetic heterogeneity in Algerian USH1 patients, even though MYO7A and CDH23 are predominant causal genes, each being involved in 6 out of 16 patients in our series. It is noteworthy that the proportion of USH2 patients among hearing impaired children is most probably underestimated because clinical diagnosis is usually not made before the onset of the visual loss, owing to the absence of vestibular symptoms in this clinical subtype. USH2 patients were indeed underrepresented in our series of USH patients, most of whom were below 15 years old (see Table 1). An additional study, focused on USH2 Algerian patients, is therefore required to determine the spectrum of mutations in these patients.  Table 1). (TIF)