Whole exome sequencing revealed 14 variants in NDP, FZD4, LRP5, and TSPAN12 genes for 20 families with familial exudative vitreoretinopathy

Familial exudative vitreoretinopathy (FEVR) is a complex form of blindness-causing retinal degeneration. This study investigated the potential disease-causing variants in 20 Chinese families with FEVR. All available family members underwent detailed ophthalmological examinations, including best-corrected visual acuity and fundus examination. All probands and most family members underwent fluorescein fundus angiography. Twenty probands underwent whole exome sequencing; 16 of them also underwent copy number variant and mitochondrial genome analysis. Bioinformatics analysis and Sanger sequencing of available family members were used to confirm the disease-causing gene variant. Twenty families were diagnosed with FEVR based on clinical symptoms, fundus manifestations, and fundus fluorescein angiography. Whole exome sequencing revealed 14 variants in NDP, FZD4, LRP5, and TSPAN12 genes among the 13 families. These variants were predicted to be damaging or deleterious according to multiple lines of prediction algorithms; they were not frequently found in multiple population databases. Seven variants had not previously been reported to cause FEVR: c.1039T>G p.(Phe347Val) in the FZD4 gene; c.1612C>T p.(Arg538Trp) and c.3237-2A>C in the LRP5 gene; and c.77T>A p.(Ile26Asn), c.170dupT p.(Leu57Phe fsTer60), c.236T>G p.(Met79Arg) and c.550dupA p.(Arg184Lys fsTer16) in the TSPAN12 gene. We did not detect any variants in the remaining seven families. These results expand the spectrum of variants in the NDP, FZD4, LRP5, and TSPAN12 genes and provide insights regarding accurate diagnosis, family genetic counseling, and future gene therapy for FEVR.


Background
Familial exudative vitreoretinopathy (FEVR, OMIM: 133780) is a clinically and genetically heterogeneous inherited ophthalmic disorder [1,2]. It is characterized by incomplete retinal vascular development and pathological neovascularization [3]. Patients usually complain of reduced visual acuity or blindness in early childhood. The fundus can exhibit peripheral retinal avascularization, falciform retinal folds, macular ectopia, retinal exudate, retinal neovascularization, and retinal detachment [4]. However, some patients may not complain of any visual impairment; they may only exhibit peripheral avascularization [5]. The reported prevalence is approximately 0.11% in newborns [6], but, the actual prevalence may be underestimated because some Open Access  [7].
Although increasing numbers of gene variants have been identified using next generation sequencing technology, these reported gene variants are responsible for only 50-60% of FEVR cases. Moreover, some patients may exhibit rapid progression without correct diagnosis and intervention. Thus, it is imperative to ascertain genetic etiology and achieve accurate diagnosis for affected patients, especially patients who are asymptomatic and exhibit peripheral retinal involvement alone. In this study, 20 families were diagnosed with FEVR based on clinical manifestations. We performed whole exome sequencing of probands and Sanger sequencing of available family members to elaborate the underling diseasecausing gene variant.

Clinical examinations
Detailed premature delivery history, oxygen uptake history, family history, and birth weight information were acquired for the probands. Exhaustive ophthalmological examinations were completed, including best-corrected visual acuity, intraocular pressure, slit-lamp microscopy, ophthalmoscopy, fundus photography, and fundus fluorescein angiography (FFA). All participants underwent pupillary dilation with a mixture of 0.5% phenylephrine hydrochloride and 0.5% tropicamide eye drops (Santen Pharmaceutical, Osaka, Japan). Fundus photography was performed with a VISUCAM 200 digital fundus camera (Carl Zeiss Meditec AG, Jena, Thuringia, Germany) or Optos Daytona ultrawide field system (Optos PLC, Dunfermline, United Kingdom). FFA was acquired using SPECTRALIS Engineering systems (Heidelberg Engineering Ltd, Hertfordshire, United Kingdom) or Optos Daytona ultrawide field system (Optos PLC) with 20% fluorescein solution (Guangzhou Pharmaceutical Holdings Limited, Guangzhou, China). FEVR was diagnosed based on previously reported criteria [23].

Whole exome sequencing
Genomic DNA samples were prepared from peripheral blood of all participants using a TIANamp Genomic DNA Kit (TIANGEN Biotech, Beijing, China). Library construction was performed from proband samples using xGen Exome Research Panel (Integrated DNA Technologies, Coralville, Iowa, United States). Samples were sequenced on a HiSeq platform (Illumina, San Diego, California, United States) or MGISEQ-T7 platform (BGI Genomics, Shenzhen, Guangdong, China) using a whole exome sequencing protocol, in accordance with the manufacturer's instructions. Sequence data were analyzed for corresponding ophthalmologic inherited genes, especially inherited retinal disease genes; sequences were aligned using Burrows-Wheeler Aligner (http:// bio-bwa. sourc eforge. net/). Variant calling and nomenclature complied with the recommendations of the Human Genome Variation Society (http:// www. hgvs. org/). Variant annotation was performed in accordance with American College of Medical Genetics (ACMG, https:// www. acmg. net/) guidelines.

In silico analysis
Preferentially selected variants were validated and cosegregated by Sanger sequencing, performed using an 3500xL Dx Genetic Analyser (Applied Biosystems, Foster City, California, United States) with ABI BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems). Primers were designed with Primer3 (http:// prime r3. ut. ee/) to amplify NDP, FZD4, LRP5, and TSPAN12 gene fragments. The primer information is provided in Additional file 1: Table S1. Consensus sequences corresponding to proband sequences were downloaded from national center for biotechnology information (https:// www. ncbi. nlm. nih. gov/). All sequences were analyzed using SeqMan II software in the Lasergene software package (DNASTAR, Madison, Wisconsin, United States). Evolutionary conservation among different species for single nucleotide variants were analyzed using MegAlign software in Lasergene software package (DNASTAR). Genomic and protein structures were schematically represented using IBS 1.0 software (http:// ibs. biocu ckoo. org).

Clinical manifestations
All participants had no history of premature delivery, problems with oxygen uptake, or low birth weight. They had no systematic complaints or extraophthalmic abnormalities that could be identified through conventional examinations (e.g., short stature or microcephaly). All probands complained of reduced visual acuity. Fundus examination of both eyes showed increased numbers of peripheral retinal vessels with willow-like retinal vessels. FFA demonstrated non-perfusion areas, retinal leakage, retinal avascularization, straightened vessels, and increased numbers of vessels. FFA manifestations of probands with potential disease-causing variants are presented in Fig. 1. The ophthalmic features of all probands with potential disease-causing variants are summarized in Table 1. Based on clinical manifestations, they were diagnosed with FEVR.

Sequencing and in silico analysis results
After filtering, the cleaned data of all probands met the quality control criteria. The quality control data for the probands are summarized in Additional file 1: Table S2. In total, 20 probands underwent whole exome sequencing; 16 probands also underwent copy number variation and mitochondrial genome analysis. None of the 16 probands harbored any copy number variant or mitochondrial genome variant. Through bioinformatics analysis and Sanger sequencing, we detected 14 variants in the NDP, FZD4, LRP5, and TSPAN12 genes among the 13 families; seven variants had not previously been reported to cause FEVR. We did not detect any variants in the remaining seven families. Variant c.981G>A in the FZD4 gene did not cosegregate with the disease in family 3, possibly because it was a de novo variant or the proband's biological parents exhibited chimerism. Because peripheral blood samples from family 5 were unavailable, we could not complete the cosegregation analysis. Sanger sequencing of the remaining 11 families revealed that the variants cosegregated with the FEVR phenotypes in these families. Sanger sequencing chromatographs of the variants are shown in Fig. 2. The pedigrees and cosegregation of families are presented in Additional file 1: Figure  S1. Variant information for the NDP, FZD4, LRP5, and TSPAN12 genes is summarized in Table 2. The variants were located in different exons or intron of the NDP, FZD4, LRP5, and TSPAN12 genes; they were predicted to affect the various domains of the NDP, FZD4, LRP5, and TSPAN12 proteins. Schematic representations of the genomic and protein structures of all variants are shown in Additional file 1: Figure S2A and S2B.
These variants were not frequently found in ethnically matched populations in multiple population databases; they were predicted to be damaging or deleterious, using multiple lines of prediction algorithms. Population distribution frequencies and predictive functional effects are summarized in Additional file 1: Table S3. Evolutionary conservation alignment of missense variants showed that they were highly conserved among four different species, except variants c.1612C>T p.(Arg538Trp) and c.4084A>G p.(Ile1362Val) in the LRP5 gene (Additional file 1: Figure S3). Based on Sanger sequencing and bioinformatics analysis, we inferred that these 14 variants in the NDP, FZD4, LRP5,

Discussion
In this study, we enrolled 20 probands with reduced vision who were diagnosed with FEVR based on clinical symptoms, as well as fundus and FFA examinations. We performed whole exome sequencing, Sanger sequencing validation, cosegregation analysis, functional prediction, population distribution analysis, and evolutionary conservation alignment. Our results suggested that 14 variants in the NDP, FZD4, LRP5 and TSPAN12 gene were potential disease-causing variants in 13 probands. Seven variants had not been reported to cause FEVR. The remaining seven families did not harbor any variant. Clinical symptoms and fundus appearances can vary distinctly among patients and genetic backgrounds in patients with FEVR; in some instances, disease presentation can vary between eyes in a single patient [27]. For example, the proband F12 exhibited different appearances between eyes, such that the right eye demonstrated normal vision and mild fundus abnormality, while the left eye demonstrated mild reduced vision and moderate fundus abnormality. Moreover, the progress of disease was asynchronous between eyes: the right eye showed minimal progression, while the left eye showed progression with falciform retinal folds and peripheral retinal exudates at the 2-year follow-up. Distinct fundus findings were also present in proband F6, such that the right eye showed severe retinal detachment, while the left eye showed mild abnormality. Although the two probands has similar disease courses and were of similar age, their symptom severities and fundus appearances were different.
NDP, FZD4, LRP5, and TSPAN12 gene variants can impair the Norrin or Wnt/β-catenin signaling pathways, which are responsible for angiopoiesis during retinal development [28]. In the canonical Wnt/β-catenin pathway, FZD4 and LRP5 form a ternary complex as a coreceptor; Wnt binds to the coreceptor and activates downstream β-catenin signaling [29]. In the Norrin/βcatenin pathway, NDP binds to the coreceptor and activates downstream β-catenin signaling with the TSPAN12 auxiliary component [30]. When these signaling pathways are activated, β-catenin translocates to the nucleus and interacts with the T-cell factor/lymphoid enhancing factor family of transcription factors, thus initiating RNA transcription and elongation [31,32].
The NDP gene encodes the Norrin protein, which contains a signal peptide that directs its localization and a typical motif of six cysteines that forms a cysteine knotlike domain [33]. The variant c.118A>G p.(Met40Val) in the NDP gene is located in the highly conserved amino acid motif; it may affect the high affinity between Norrin and the FZD4 transmembrane protein [34].
The FZD4 gene encodes the FZD4 protein, which contains an extracellular cysteine-rich domain, sevenpass transmembrane domains, and a frizzled domain in the extracellular region [35]. The variants c.  [36].
The LRP5 gene encodes the LRP5 protein, which contains a putative signal peptide, four β-propeller motifs at the amino terminal that alternate with four epidermal growth factor-like repeats, three low-density lipoprotein receptor-like repeats, a single transmembrane domain, and a cytoplasmic domain [37]. Although the exact functions of these domains are unknown thus far, studies of LRP6 (with strong homology and similar function to LRP5) showed that the first and second β-propeller motifs, the third and fourth β-propeller motifs formed tandems to function respectively [38]. Variants located in the β-propeller motif may destroy the stable structure of β-propellers and interrupt their interactions with NDP or FZD4. The c. The TSPAN12 gene encodes the TSPAN12 protein, which contains four-pass transmembrane domains and four cysteines in the second extracellular region, forming two extracellular loops and an intracellular loop. Variants in transmembrane domains and extracellular regions can severely impair function, variants in the C-terminal end can moderately impair function, and variants in the N-terminal end can slightly impair function [36]. The transmembrane domains provide a scaffold for extracellular loops to change conformation and interact with FZD4 for allosteric modulation. The variants c.77T>A p.(Ile26Asn) and c.236T>G p.(Met79Arg) in the TSPAN12 gene are located in the transmembrane domain and may disrupt the domain structure of the TSPAN12 protein, potentially preventing TSPAN12 incorporation into the receptor complex and destabilizing the NDP/FZD4/LRP5 interaction [39]. The frameshift variants c.170dupT p.(Leu57Phe fsTer60) and c.550dupA p.(Arg184Lys fsTer16) in the TSPAN12 gene are predicted to create a premature stop codon and a truncated TSPAN12 protein.
Although we found 14 disease-causing variants, including seven novel variants, in 13 FEVR families, there were some limitations in this study. First, we only speculated that variants were potential disease-causing based on clinical manifestations, whole exome sequencing, and bioinformatics analysis. Second, the study collected a small group of samples and thus cannot expand the overall understanding of the pathogenic mechanisms by which these genes cause FEVR. We plan to validate the pathogenicity of these variants by in vivo and in vitro analyses, and we will attempt to enroll more families in a future study.

Conclusions
In conclusion, through whole exome sequencing and bioinformatics analysis, we identified 14 variants in the NDP, FZD4, LRP5, and TSPAN12 gene in 13 families with FEVR. To our knowledge, this is the first report regarding c.1039T>G p.(Phe347Val) in the FZD4 gene; c.1612C>T p.(Arg538Trp) and c.3237-2A>C in the LRP5 gene; and c.77T>A p.(Ile26Asn), c.170dupT p.(Leu57Phe fsTer60), c.236T>G p.(Met79Arg), and c.550dupA p.(Arg184Lys fsTer16) in the TSPAN12 gene as potential diseasecausing variants in patients with FEVR. These results expand the spectra of variants in the NDP, FZD4, LRP5, and TSPAN12 genes. We presume that these findings will provide insights regarding accurate diagnosis, family genetic counseling, and future gene therapy for FEVR.
Additional file 1. Figure S1. Pedigrees and cosegregation of families. Figure S2. Schematic representations of the genomic and protein structures of all variants. Figure S3. Evolutionary conservation of ten missense variants. Table S1. Primers used in Sanger sequencing. Table S2. Quality control information of probands with potential disease-causing variants. Table S3. Population distribution frequencies and predictive functional effects of variants.