Targeted Next-Generation Sequencing Reveals Novel USH2A Mutations Associated with Diverse Disease Phenotypes: Implications for Clinical and Molecular Diagnosis

USH2A mutations have been implicated in the disease etiology of several inherited diseases, including Usher syndrome type 2 (USH2), nonsyndromic retinitis pigmentosa (RP), and nonsyndromic deafness. The complex genetic and phenotypic spectrums relevant to USH2A defects make it difficult to manage patients with such mutations. In the present study, we aim to determine the genetic etiology and to characterize the correlated clinical phenotypes for three Chinese pedigrees with nonsyndromic RP, one with RP sine pigmento (RPSP), and one with USH2. Family histories and clinical details for all included patients were reviewed. Ophthalmic examinations included best corrected visual acuities, visual field measurements, funduscopy, and electroretinography. Targeted next-generation sequencing (NGS) was applied using two sequence capture arrays to reveal the disease causative mutations for each family. Genotype-phenotype correlations were also annotated. Seven USH2A mutations, including four missense substitutions (p.P2762A, p.G3320C, p.R3719H, and p.G4763R), two splice site variants (c.8223+1G>A and c.8559-2T>C), and a nonsense mutation (p.Y3745*), were identified as disease causative in the five investigated families, of which three reported to have consanguineous marriage. Among all seven mutations, six were novel, and one was recurrent. Two homozygous missense mutations (p.P2762A and p.G3320C) were found in one individual family suggesting a potential double hit effect. Significant phenotypic divergences were revealed among the five families. Three families of the five families were affected with early, moderated, or late onset RP, one with RPSP, and the other one with USH2. Our study expands the genotypic and phenotypic variability relevant to USH2A mutations, which would help with a clear insight into the complex genetic and phenotypic spectrums relevant to USH2A defects, and is complementary for a better management of patients with such mutations. We have also demonstrated that a targeted NGS approach is a valuable tool for the genetic diagnosis of USH2 and RP.


Background
Inherited retinal dystrophies (IRDs) comprise a group of monogenic diseases presenting significant clinical and genetic heterogeneities. Retinitis pigmentosa (RP; MIM: 26800), the most common form of IRDs, possesses a global prevalence of 1 in 3500 to 5000 individuals [1]. In the disease course of RP, rod photoreceptors will initially be affected thus leading to night blindness and visual field (VF) constrictions, while subsequent cone defects will cause color blindness, central vision impairments, and finally total vision loss [2].
Clinically, RP can be divided into nonsyndromic and syndromic forms. Former one can be further classified into typical and atypical RP according to the fundus presentations, while the latter one is usually accompanied by systemic abnormalities [2]. Patients affected with RP may manifest great varieties in the disease course, and RP presentations can overlay with other forms of IRDs, thus making it more difficult to obtain better clinical diagnoses for these patients. Genetically, RP, usually a monogenic disorder, demonstrates all three types of mendelian inheritance modes, including autosomal dominant, autosomal recessive, and X-linked patterns. Additionally, digenic inheritance patterns have also been reported [3]. Hitherto, 71 mapped loci involving 64 genes have been identified as disease-causing for nonsyndromic RP (www.retnet. org). The significant genetic heterogeneity calls for an efficient platform for the molecular diagnosis of RP. Targeted nextgeneration sequencing (NGS) strategies show significant improvements when compared with traditional detection approaches. It is more effective, cost-effective, and provides high sequencing accuracy. Applying NGS approaches to IRDs is now relatively sophisticated. A targeted panel of known IRDs causative genes would help to obtain a potential molecular diagnosis for 50-70% of IRDs cases [4].
USH2A (MIM: 608400), located on chromosome 1 and sublocation 1q41, encodes the usherin, which demonstrates a wide but not ubiquitous tissue distribution [5]. Usherin shows two alternatively spliced isoforms that comprise of short ''isoform a'' and longer ''isoform b'' variants [6,7,8]. Isoform a encompasses 21 exons and generates a secreted and extracellular protein of 1546 residues, while isoform b contains 51 additional exons at the C terminal, and encodes a protein with 5202 amino acids ( Figure 1A). As a basement membrane protein, usherin has been localized to the basement membrane of the retina, cochlea, and a series of other tissues [5,9,10]. Despite its wide expression, defects in usherin have only been linked with Usher syndrome type 2 (USH2; MIM: 276901) [11,12,13,14], nonsyndromic RP [15], and nonsyndromic hearing loss (NSHL) [16], implying its special role in maintaining the function of the capillary and structural basement membranes of the two ciliated sensory neurons, namely photoreceptor cells and cochlear hair cells. As indicated by previous studies, usherin is localized to the apical inner segment recess which wraps around the connecting cilia in photoreceptor cells, and is associated with the hair bundles during postnatal development in cochlear hair cells [17]. Although many studies have been focused on the properties of usherin, the mechanism underlying the great phenotypic variety led by USH2A mutations has not been well illustrated yet.
Herein, using a targeted NGS approach, we determined the disease causative mutations in three Chinese families with typical autosomal recessive RP (ARRP), one with RP sine pigmento (RPSP), and one with USH2. Six novel and one recurrent homozygous USH2A mutations were identified with their genotype-phenotype correlations annotated.

Participants and Clinical Assessments
This study, conformed to the tenets of the Declaration of Helsinki, was approved and prospectively reviewed by the Ethics Committee on Human Research of Nanjing Medical University. Seven patients and seven unaffected family members from five unrelated families with autosomal recessive inheritance pattern of retinal disease were recruited from the Frist Affiliated Hospital of Nanjing Medical University. Written informed consents were signed by all participants before their participation. Consanguineous marriages were reported by three families. Pure tone audiometry testing and vestibular tests were performed on all included patients. Detailed ophthalmic examinations were also conducted, including assessments of best corrected visual acuities (BCVAs), VF measurements, funduscopy, electroretinography (ERG), and fundus fluorescein angiography (FFA). Optical coherence tomography (OCT) examination was selectively conducted on patients with macular involvements. Peripheral blood samples were collected using 5 mL tubes containing ethylenediamantetetraacetic acid (EDTA) from all participants and 150 additional negative controls free of retinal or other ocular diseases. Genomic DNA was isolated from leukocytes with a QIAmp DNA blood kit (Qiagen, Valencia, CA) in accordance to the manufacturer's protocols, and stored at 220uC before used.

Targeted NGS Approach
Seven affected patients from five individual families were included in the present study with their ophthalmic and syndromic evaluations detailed in Figure 2 and Table 1. Targeted NGS approach was selectively conducted on patients ARRP01-IV:4, ARRP02-II:3, ARRP03-IV:3, ARRP04-IV:4, and SU01-II:1, and was shown in Table S2. Briefly, coverage of the targeted region was equal to or over 98.55% for each sample, and the average mean depth of five samples was 98.62-fold. Over 2600 variants were initially detected for each sample. After initial filtration, 5, 7, 5, 1, and 11 variations were retained for patients ARRP01-IV:4, ARRP02-II:3, ARRP03-IV:3, ARRP04-IV:4, and SU01-II:2, respectively ( Table 2). Sanger sequencing further confirmed seven USH2A variants as putative disease-causing mutations for the five included families ( Figure 2). Noteworthy, two homozy-gous missense variations were revealed as RP causative for patient ARRP03-IV:3, implying a putative double hit effect. The seven identified mutations included a nonsense variation, two splice site variants, and four missense mutations ( Table 3), all of which were absent in 150 unrelated controls. The conservations of identified variants were detailed in Figure 1B with predictions from the four online software programs listed in Table 3.

Homozygous Mutations Identified in the Three Families with nonsyndromic RP
Patient ARRP01-IV:3 carried a novel homozygous nonsense mutation, c.11235C.G ( Figure 2). Consanguineous marriage was reported in this family. Patient ARRP01-IV:3 suffered from poor night vision in her early 10 s, which was relatively early when compared with other patients investigated in this study. The disease progressed fast that she had tubular vision at her late 10 s. She was 24 years old at her last visit to our hospital and she complained about her rapidly decreased visual acuities. Macular involvement was also indicated by her fundus appearance ( Figure 3A). A novel homozygous missense mutation, c.14287G.C, was identified in all three patients from family ARRP02 (Figure 2). This alteration would lead to an amino acid change from a neutral glycine to an alkaline arginine at residue 4763 (p.G4763R) of usherin, thus changing the theoretical isoelectric point from 6.40 to 6.42. The alteration of the isoelectric point of the protein might contribute to the change of the protein function. The three patients were manifested with more moderate ophthalmic phenotypes when compared with patient ARRP01-IV:3. Among all three affected siblings, the proband, patient ARRP02-II:3, got the youngest RP onset age and fastest disease progression ( Table 1). She complained about her narrowed VF since her late 20 s, and central vision defects since her early 30 s. Typical RP fundi ( Figure 3B), affected OCT presentations ( Figure 3F-H), and diminished ERG ( Figure 3I) were revealed in all three patients.
The mutation found in patient ARRP04-IV:4 was a recurrent one, c.8559-2T.C (Figure 2). She was first diagnosed with RP at age 45 in a local hospital. She claimed to have nyctalopia since age 40 and constricted VF since age 43. Slit-lamp examination revealed mild cataracts in her both eyes. Consanguinity was reported. No hearing impairments, vestibular functional abnormalities, or other syndromic defects were observed in patients from the three abovementioned families, thus leading to the clinical diagnosis of nonsyndromic RP.

Mutations in Patient ARRP03-IV:3 with RPSP
Another two novel homozygous missense mutations, c.8284C. G and c.9958G.T, were found in patient ARRP03-IV:3 ( Figure 2). The former one leads to an amino acid change from a proline to an alanine at residue 2762 (p.P2762A) of usherin, while the latter one causes the alteration from glycine to cystine at position 3320 (p.G3320C). Patient ARRP03-IV:3 showed similar RP progress to that of patient ARRP01-IV:3. She reported to have nyctalopia at age 12 and constricted VF since age 14. Significantly reduced ERG presentations were also indicated ( Figure 3I). Attenuated vessels, waxy optic disc, and moderate macular atrophy were also revealed by funduscopy, while no pigment deposits were identified ( Figure 3D), thus leading to the clinical diagnosis of RPSP. No systemic defects were identified. Consanguineous marriage was indicated.

Mutations Validated in Patient SU01-II:1 with USH2
Heterozygous novel mutations, c.8223+1G.A and c.11156G. A, were found in patient SU01-II:1 (Figure 2). The former one was a single nucleotide change that might affect splicing of USH2A, while the second mutation was a missense variant, which would lead to the substitution of an arginine to a histidine at residue 3719 of usherin. Patient SU01-II:1 reported to have congenital and stable hearing impairments. No vestibular defect was identified. Typical RP presentations were also revealed, that she complained of night blindness since age 5, youngest among all included patients. Her clinical symptoms supported the clinical diagnosis of USH2.

Discussion
USH2A mutations account for over 70% of USH2 cases in multiple populations [11,12,13,14]. Recent studies have revealed the important role of USH2A mutations in causing nonsydromic RP [15] and NSHL [16], indicating a wide phenotypic variety mediated by USH2A mutations. In this study, we report six novel and one recurrent mutations in the USH2A gene as the diseasecausing mutations for five families. Diverse clinical phenotypes, Table 1. Clinical features of included patients. including early, moderate, and late onset non-syndromic RP, RPSP, and USH2, have been shown, supporting the complex genotype-phenotype correlations for USH2A mutations. Noteworthy, a previously identified founder mutation for USH2 in East Asian populations [13,14,33], c.8559-2T.C, was found correlated to late onset RP in the present study, further suggesting the complexity. Meanwhile, to the best of our knowledge, this is the first time that RPSP is found associated with USH2A mutations, and the two identified homozygous USH2A mutations suggest a potential double hit effect.
Of all the seven identified mutations, four are missense substitutions, while the other three would generate truncated proteins or lead to nonsense mediated mRNA decay (NMD), including c.8223+1G.A found in family SU01, c.8559-2T.C in family ARRP04, and c.11235C.G in family ARRP01. The mutation c.8223+1G.A would probably affect a consensus donor splice site essential for splicing of intron 41. All four online programs predicted that this mutation would abolish the original splice donor site, leading to the insertion of intron 41. However, a premature termination codon (PTC) located at c.8223+10 to  c.8223+12 in intron 41, TGA, would suspend translation, leading to NMD and/or the generation of a truncated protein, p.V2742Ifs*3. The sever impacts of this mutation may partly help to explain for the comparatively earlier onset of her clinically manifestations. The mutation c.8559-2T.C, located at the splice acceptor site in intron 42, is recognized as a frequent mutation for USH2 in Asian populations [13,14,33], but has never been reported in causing RP. This mutation is proved to cause the skipping of USH2A exon 43 by reverse-transcription PCR using patients' hair roots thus leading to the elimination of 43 amino acids in usherin (p.Y2854_R2896del43) [10]. The nonsense mutation, c.11235C.G, would generate a PTC, and further lead to dysfunction or a reduction of usherin through truncation of the generated protein or NMD, respectively. Two different forms of usherin have been revealed. The shorter isoform, isoform a, contains five domains, including an N-terminal signal peptide, laminin G-like jellyroll fold, laminin N-terminal, laminin-type EGF-like modules, and fibronectin type III repeats. The longer isoform, isoform b, contains additional domains, including laminin G domains, a transmembrane region, and a Cterminal PSD-95/Discs Large/Zona Occludens 1 (PDZ)-binding motif, via which this form is attached to the plasma membrane of retinal photoreceptors [17]. Isoform b can therefore be divided into three major parts, namely the extracellular region, the transmembrane domain, and the cytoplasmic part. Isoform b has been proved as the predominant form in the retina, and is crucial for the maintenance of retinal photoreceptor cells and the development of cochlear hair cells [17]. For mutations located within the genome context of isoform a, it can usually affect the function of isoforms a and b. All seven mutations identified in the present study locate within isoform b, among which, p.Y3745* in family ARRP01, p.G4763R in family ARRP02, p.P2762A in family ARRP03, p.Y2854_R2896del43 in family ARRP04, and p.R3719H in family SU01, locate within the fibronectin type-III repeats, while another mutation in family ARRP04, p.G3320C, was found in the interval between the 18 th and the 19 th fibronectin type-III domains ( Figure 1A). However, the expression level of the long isoform is low in the cochlea [7], which could possibly help to explain the absence of hearing and vestibular impairments in some patients with USH2A mutations [15,34,35]. Previous studies have also hypothesized a potential correlation between the nature of the variants and clinical phenotypes, indicating that nonsyndromic RP patients would carry USH2A mutations with less perniciousness, or with deleterious impacts in only one allele [34,36], which may also help to account for the diverse clinical phenotypes as demonstrated in the present study.
Herein, we have identified six novel and one recurrent USH2A mutations using targeted NGS approaches, which further demonstrates that targeted NGS approaches are valuable tools for genetic diagnoses. However, many limitations still exist that prevent the routine application of targeted NGS in the clinic, including cost, infrastructural logistics, strategies for patient selection, and interpretation of genetic datasets [4]. Among all, interpretation of genetic datasets is most important while also most problematic [4]. Bioinformatics predictions are useful, but the results can sometimes be confusing or even incorrect [37]. Therefore, more functional analyses are needed to confirm the pathogenesis of distinct mutations, and also to gain a clearer insight into the mechanism underlying the selectively clinical phenotypes caused by different mutations.
In conclusion, our findings highlight the genotypic multiplicity and phenotypic variability for USH2A mutations. A better understanding of the genotype-phenotype correlations would help with a better management for patients with USH2A mutations, Table 3. Characteristics of identified USH2A mutations. provide thorough viewpoints into the pathogenesis of USH2A defects, and assist with the development of putative therapeutic approaches.

Acknowledgments
We thank all patients and their family members for their participation. We also appreciate all clinicians for collecting of samples and clinical data, and are grateful to Liping Guan, Jingjing Xiao, and Jingjing Jiang from BGI-Shenzhen (Shenzhen, Guangdong, China) for technical support.