Clinical Features
Among 25 patients clinically diagnosed with RP, 10 eyes of 5 probands from 5 pedigrees with nonsyndromic RP resulting from USH2A mutations were included (Figure 1). There were two males and three females in the identified probands. The age at presentation ranged from 31 to 60 years, the age of onset ranged from 12 to 46 years. The median follow-up was 46 months (range: 40–50 months). In the better eye, two patients (40%) had normal visual acuity or mild visual loss (0.9-0.6), one patient (20%) showed moderate visual impairment (0.5-0.2), and the other two patients (40%) presented low vision (<0.2). The patients’ demographic features are summarized in Table 1.
Fifteen family members from five pedigrees participated in our study (including five probands), and all of them were tested for whole-exome sequencing or verified by Sanger sequencing. All participants denied hearing impairment or deafness or balance problems. Anterior segment examinations were unremarkable. The clinical and genetic details of each pedigree are described in the supplemental materials.
Multimodal Imaging and Disease Progression
In the early stage of RP, the pigment deposits of the RPE are punctate or bone spicule-shaped (Figure 2, pedigree 1) in the ultrawide-field fundus autofluorescence (FAF) and progressively aggravated into a mottled appearance in the mid periphery of retina (Figure 2, pedigree 2) with the course advancing. In the advanced stage, a lobular RPE defect along retinal vessels could be seen in the mid periphery of retina (Figure 2, pedigree 3 and 4). In the late stage, patchy RPE defects confluent into a zonal area of atrophy in the mid and far periphery of the retina with macular sparing (Figure 2, pedigree 5).
Moreover, although the multimodal fundus appearances varied among different patients, they shared something in common. All the probands showed the distinctive pattern of diffuse and homogeneous hyperautofluorescence parafoveal ring with macular sparing, which could be observed in ultrawide-field FAF [11].
Genetic and Bioinformatics Analyses
In total, 17 sequence variants (including ten USH2A variants) were identified in five probands. Among the ten variants of the USH2A gene, seven were missense mutations, two were splicing mutations, and one was an insertion mutation. The c.2802T>G (p.934C>W) variant was recurrent and was found in 2/5 probands (40%), 2/10 USH2A variants (20%) in our study, but almost all of the patients carried private sequence variants. Six of the variants have already been reported in the literature [12–16], while the remaining four were novel (Table 2).
Location of the novel mutations in USH2A: The p.C931Y mutant is located within the laminin EGF-like domain (Lam EGF) of USH2A, and the p.G4489S and p.M4853V mutants are located within the fibronectin type III domain (FN3). Figure 4 (in the supplemental materials) shows the schematic diagram of the reported mutations along the USH2A protein domains (without showing the novel intronic mutation IVS22+3A>G).
Allele frequency: Allele frequency for the nine detected USH2A variants (except for IVS22+3A>G) in the general population ranged from 0% to 0.07825% in total. The allele frequency for the three novel missense variants (p.C931Y, p.G4489S, p.M4853V) was 0.005771%, 0.000000%, and 0.005766%, respectively (Table 3).
Conservative prediction: A total of 127 homologous genes of USH2A taken from 127 species were used to construct the phylogenetic tree (Figure 5A). The human USH2A homolog located in a same sub clade together with that in the primate Bonobo and Chimpanzee, suggesting these USH2A genes share the most evolutionary similarity with each other. The Weblogo software draws the seqlogo diagram of the sequence around the amino acids corresponding to each mutation point. As shown in Figure 5B, the third amino acid in the diagram indicates the amino acid corresponding to each mutation point. The larger the letter, the more conservative the amino acid at this site is. The conservative calculation of amino acids corresponding to each mutation site in the phylogenetic tree can be found in Table 3. Among the nine available mutation sites on USH2A, seven had conservative values greater than 0.5, while the amino acids at the 34 and 4853 sites had conservative values less than 0.5.
Pathogenicity prediction: The three novel missense variants (p.C931Y, p.M4853V, and p.G4489S) could not be found in the 1000 Genomes Database. They were predicted to be deleterious (p.G4489S and p.C931Y), and neutral (p.M4853V) by Provean; to be probably damaging (p.G4489S and p.C931Y) and benign (p.M4853V) by PolyPhen-II; to affect protein function (p.C931Y) and to be tolerated (p.G4489S and p.M4853V) by SIFT; and to be radical (p.C931Y), moderately conservative (p.G4489S) and conservative (p.M4853V) by Grantham scores. The bioinformatics analyses of the USH2A variants are summarized in Table 3.
Additionally, IVS22+3A>G is a novel splicing variant that could be considered pathogenic. The prediction and analysis of the secondary structure of DNA around the intron mutation IVS22+3A>G showed that the mutation position formed a hairpin structure (Figure 6 in the supplemental materials, circled in red). Hairpin loop structures are an important structural motif in nucleic acids that have been shown to play important roles in many biological processes [17].
Protein structure modeling and analysis: Three-dimensional structural modeling of the wild-type and mutant USH2A proteins were cut in four regions according to the principle of covering the mutation point to the greatest extent (Table 4 in the supplemental materials). Region 1 (amino acids 1-1468) covers code-shifting mutations caused by insertions at bases 99-100: p.R34Sfs (c.99_100insT); region 2 (amino acids 747-2239) covers three missense mutations, including p.C931Y (c.2792G>A), p.C934W (c.2802T>G), and p.G1861S (c.5581G>A); region 3 (amino acids 3774-5202) covers four missense mutations, including p.G4489S (c.13465G>A), p.Y4673H (c.14017T>C), p.M4853V (c.14557A>G), and p.R5143C (c.15427C>T); and region 4 (amino acids 1869-3369) covers the splicing mutation in exon 42: IVS22+3A>G. Figure 7 shows the three-dimensional structural modeling of the wild-type and mutant USH2A proteins and their alignments in these four regions.
Protein function prediction in mutation point: The amino acid positions 922 and 931 as well as 934 and 948 were found to form disulfide bonds (Figure 8 in the supplemental materials). The predicted results showed that the reliability of disulfide bond formation is 1. In our study, the amino acids at positions 931 and 934 of the USH2A protein were the sites containing the gene mutations, and it can be assumed that the mutations at amino acids 931 and 934 will lead to disulfide bond breaking and thereby affect the structure of the USH2A protein.