Clinical and whole-exome sequencing findings in Individuals from Yunnan Province with Familial Exudative Vitreoretinopathy

Background : Familial exudative vitreoretinopathy (FEVR) is a rare inherited retinal disorder characterized by the failure of peripheral retinal vascularization at birth. FEVR can cause further pathological changes, such as neovascularization, exudation, haemorrhage, and retinal detachment, in turn. The molecular diagnosis enables a deeper understanding of this disease, so the differentiation of FEVR is important for the accuracy of genetic counselling. However, to date, only six genes have been identified as being responsible for this disease. Methods : Without a known FEVR gene mutation, six families were enrolled in our study between 2016 and 2017 from the clinical practices of ophthalmologists. The referring physician made a diagnosis for FEVR probands, and clinical data and DNA were collected from each participant. Whole-exome sequencing was used to detect the mutations present in the probands. The raw reads were obtained from Illumina Miseq. Then, an in-house bioinformatics pipeline was performed to detect both single nucleotide variants (SNVs) and small insertions/deletions (InDels). The pathogenic mutations were identified with stringent criteria and were further confirmed by conventional methods and cosegregation in families. Results : Using this approach, bioinformatic predictions showed that six mutations were found in our study: three mutations in the known genes of ZNF408, LRP5 and KIF11 and three mutations in the newly identified genes NR2E3,KRT3 and FOXL2 .To test the hypothesis that cases of FEVR are caused by NR2E3, KRT3 and FOXL2 , probands who were diagnosed as FEVR by a physician using wide-field fluorescein angiography were found to not have any mutations in any of the six known FEVR genes. Compared with previous reports, mutations in NR2E3, KRT3 and FOXL2 are believed to cause a broader spectrum of ocular disease. The NR2E3,KRT3 and FOXL2 genes likely play a role in retinal vascular development. Conclusions : This report is the first to describe FEVR mutations in Yunnan province children with FEVR. This study would provide information on the genetic forms of

FEVR most often presents as visually significant in childhood, although it might progress at any age with sight-threatening manifestations [13]. In 1998, a 5-stage FEVR classification scheme was described by Pendergast and Trese [14]. In Stage 1, the avascular periphery is notable, while Stage 2 occurs when neovascularization develops with the avascular periphery. Stage 3 is defined by macula-sparing retinal detachment and proceeds to Stage 4 with macula-involving retinal detachment. Stage 5 is complete retinal detachment, with the A versus B designations being for the absence versus the presence of exudate [13].
To date, 6 genes have been identified that account for some cases of FEVR. These genes are FZD4 (Frizzled 4 receptor), LRP5 (low-density-lipoprotein receptor-related protein 5), NDP (Norrie disease protein), TSPAN12 (tetraspanin-12), ZNF408 (zinc finger protein 408), and KIF11 (kinesin family member 11) [15][16][17][18][19][20][21][22][23]. Mutations in these gene have been implicated in abnor mal signalling of known biological pathways and are inherited in autosomal dominant (AD), autosomal recessive (AR), and Xlinked recessive fashions [3]. AD is the most common mode of inher itance and is associated with mutations in FZD, LRP5  However, the detailed mutational spectrum of these six genes was unknown in Chinese individuals [29]. Even genetic testing has not been carried out in Yunnan province, and there are few reports regarding mutations in the 25 ethnic minority groups in Yunnan. In the present study, genetic testing was performed on individuals with FEVR in Yunnan province to investigate the genetic causes of FEVR.
WES was used to sequence DNA from affected proband of FEVR pedigrees with no known causative FEVR gene mutations (FZD4, LRP5, NDP, TSPAN12, ZNF408, and KIF11). We identified three pedigrees with a heterozygous ZNF408, LRP5 and KIF11 profile and three pedigrees with a heterozygous NR2E3,KRT3 and FOXL2 mutations. To test the hypothesis that more cases of FEVR are caused by NR2E3, KRT3 and FOXL2 mutations, we screened for mutations in the NR2E3, KRT3 and FOXL2 genes in a cohort of FEVR probands who did not have a mutation in any of the known FEVR genes.

Patient 1
The mutation, ZNF408, p.G290S, identified in a proband, was associated with a FEVR phenotype. The family, which has one girl (the proband) and unaffected parents, was recruited into the FEVR project specifically because of the fundus appearance mimicking FEVR. The proband was a 13-day-old baby at the time of recruitment.

Patient 2
The mutation in LRP5 involved in splicing, identified in a proband was associated with a stage 4-FEVR phenotype. The family, which has two girls (the proband, and an unaffected elder sister of proband) and unaffected parents, was recruited in the FEVR project specifically because of the stage 4-FEVR.
The proband was a 3-year-old girl at the time of recruitment. She had been noted as have retinal abnormalities at age 3 years. She was born full-term with a birth weight of 3.2 kg. The elder sister was examined at age 5 but she is normal.

Patient 3
The mutation in KIF11, p.1423T, identified in proband was associated with a FEVR phenotype. The family, which has two boys (the proband and an unaffected eld brother) and unaffected parents, was recruited into the FEVR project specifically because of stage-1FEVR. The proband was an 8-month-old baby at the time of recruitment. The retinal abnormalities were first noted at age 14 days. He was born full-term with a birth weight of 2.9 kg. There was no family history of health problems compatible with a diagnosis of FEVR, and the mutation was found in his unaffected father and elder brother.

Patient 4
The mutations in NR2E3, p.H361R, identified in a proband were associated with a FEVR phenotype.
The family, which has one boy (the proband) and an unaffected elder sister, as well as unaffected parents, was recruited into the FEVR project specifically because of stage-4 FEVR. The proband was a 3-year-old boy at the time of recruitment.

Patient5
The mutation, KRT3, p. R271H, identified in proband was associated with a FEVR phenotype. The family, which has one boy (the proband) and unaffected parents, was recruited in the FEVR project specifically because of stage-4FEVR. The proband was a 2-month-old boy at the time of recruitment.

Patient 6
The FOXL2 mutation, p.333-337del, was identified in a proband was associated with a FEVR phenotype. The family, which has one boy (the proband) and unaffected parents, was recruited in the FEVR project specifically because of stage-4 FEVR. The proband was an 11-month-old boy at the time of recruitment.

Genetic Analysis
WES was initially performed on an affected probands (pedigree I-Ⅵ) to identify a shared diseasecausing variant. The reads aligned to the human genome, mapped to the target region with a mean coverage between 99.8 and 99.9%, SNPs and indels are shown in Table 1. Between 11,631 and 11,844 were identified as SNPs from six affected individuals, whereas 833-918 indels were identified in the unaffected individuals. Analysis based on the pattern of inheritance identified common homozygous and heterozygous variants in six affected individuals (Table 1). Among the identified variants, the numbers of selected gene were displayed in Table 1 according to our mutational analysis. Further analysis based on expression profiles identified six candidate variants (Table 1).
Among these, the four software packages predicted six of the missense variants to be damaging.
After filtering for suspected sequencing artefacts and with a multiple allele frequency greater than 5%, the shared missense or nonsense variants identified were displayed in Table 2.
In the families (pedigree I-Ⅲ), the remaining heterozygous, ZNF408, LRP5 and KIF11 genes were selected as the candidates because heterozygous mutations in these genes has previously been reported as being causative for FEVR [22]. There were no other siblings (pedigree Ⅲ) and no family history of a condition compatible with a diagnosis of FEVR. The mutation was absent father, mother, and mother in family I,Ⅱ, and Ⅲ, respectively. In families (pedigree Ⅳ-Ⅵ), NR2E3,KRT3 and FOXL2 were selected as the candidate. The unaffected father (pedigree Ⅳ), mother (pedigree Ⅴ) and mother (pedigree Ⅵ) presented with mutations.
Sanger sequencing identified additional novel heterozygous mutations in 6 probands ( Figure 1 family tree).
Gene type, amino acid changes and nucleotide changes of all mutations are shown in Table 3, and the amino acid changes in family2 (LRP5) are in an acceptor splice site.
Additionally, these variants were absent in 60 normal control individuals.

Discussion
In this study, a comprehensive molecular screen for mutations in six families with FEVR was performed with a WES approach. In these six families, the probands were found by fluorescein angiography, which is critical for the diagnosis and manage ment in FEVR [30]. Several bioinformatics tools predicted a strong effect on the amino acid changes inducing exons 5, 6, 11, 8, 25 and 18, in families 1, 2, 3, 4, 5 and 6, respectively. These novel variants found in our patients were neither reported in the literature nor in public databases. To investigate the effects of mutation on the transcript, we analysed ZNF408, LRP5, KIF11, NR2E, KRT3 and FOXL2 DNA obtained from patient's whole blood. The potential regions were amplified by standard PCR. The nucleotide substitutions in these families caused the amino acid changes in exons. As a result, we were able to obtain accurate mutations in FEVR disease gene from these families (pedigree I-III) and identify putative pathogenic mutations for 50% of the patients from genes that are published in the literature. We were also able to find mutation in three genes, NR2E3, KRT3 and FOXL2, in the families (pedigree III-VI).
In these three families (pedigree I-III), there were newly occurring mutations, with most of the alleles identified having never been reported on before. Through previously reported work, it is known that a number of variants discovered in the genes ZNF408, LRP5 and KIF11 correlate with FEVR. Our study allowed us to obtain an accurate estimation of the number of patients carrying mutations in each FEVR gene. In our patient cohort, ZNF408, LRP5, and KIF11, which are the most frequently mutated, were found in families (pedigree I-III). In one Dutch family and a Japanese family, novel mutations in ZNF408 have been reported. Subsequently, it has been shown to play a putative role in retinal vasculogenesis through the use of a knockdown of znf408 in zebrafish [31]. That mutations in LRP5 can lead to abnormal retinal vascular formation, delayed retinal vascular development and subsequent pathological glomeruloid vessels was demonstrated by the use of a Lrp5 knockout (Lrp5−/ −) mouse model of FEVR [32]. The LRP5 gene has been shown to exhibit two inheritance patterns, AD and AR. Due to compound heterozygous genotypes consisting of missense mutations, AR of LRP5 is most common [33]. AR inheritance patterns have been observed in our study. This proband carries compound heterozygous mutation, and LRP5 mutations are found in the unaffected father, while only wild-type variants are present in the unaffected mother. Mutations in KIF11 are associated with microcephaly, chorioretinopathy, or mental retardation [34]. Mutations of this gene have been recently identified in individuals affected with FEVR. Furthermore, there is a phenotypic overlap between the two diseases caused by KIF11 mutations [35][36][37]. With this hypothesis, the mutations of NR2E3, KRT3 and FOXL2 had been found in three patients with FEVR. In 2000, the nuclear receptor subfamily 2, group E, member 3 gene (NR2E3) on chromosome 15q23, which encodes a liganddependent transcription factor, was found to play an important role in photoreceptor development and differentiation and is responsible for the enhanced S-cone syndrome (ESCS) and was identified [38,39]. Subsequently, mutations in NR2E3 were found in patients with Goldmann-Favre syndrome and clumped pigmentary retinal degeneration [40], autosomal recessive retinitis pigmentosa, and dominant retinitis pigmentosa [41]. These mutations of NR2E3 gene expanded the phenotypic variations for people with retinal dystrophy. Keratins are a group of structural proteins in the epithelia of cornea [42]. In 1997, keratin 3 (KRT3) was first linked to mutations in Meesmann's corneal dystrophy [43].The gene family of Fork head box (Fox),which named after the Drosophila melanogaster fork head gene (fkh), is an evolutionarily ancient gene family [44]. The Forkhead Box L2 gene (FOXL2 )which encodes the fork head transcription factor FOXL2 and plays an important role in thedevelopment of the eyelids and ovary, is the primary gene that underlies Blepharophimosis Syndrome [45,46] .We found these mutations in these gened in families (pedigree Ⅳ-Ⅵ),with FEVR.
Further research needs to be carried out on these genes: NR2E, KRT3 and FOXL2.
To the best of our knowledge, the roles of NR2E3, KRT3 and FOXL2 were first reported in this study, and the expression patterns of NR2E3, KRT3 and FOXL2 in different cell types are still not fully defined. Interestingly, traces of NR2E3, KRT3 and FOXL2 transcript in whole blood samples were found to allow for the direct study of the effects of the mutations, making the in vitro model redundant for the mutation assay. Further research will clarify the roles of the NR2E3, KRT3 and FOXL2 genes in the development of retinal vessels.

Conclusion
In summary, the use of the WES method in this study allows us to efficiently identify all of the most likely causative variants in disease-associated genes associated with FEVR.. Mutations in these genes cause a broader spectrum of ocular disease than previously reported. The molecular diagnosis together with the clinical phenotype provides potentially useful information for the diagnosis of the disease, the prognosis for the patient, and guidance for genetic counseling . This information is particularly valuable, given the phenotypic variation found even within the same family.

Participants and Clinical Data Collection
This study was approved by the Ethics Committee of the Children's Hospital of Kunming Medical University, and written informed consent was obtained from either the participants or their guardians.
No stipend was offered.
With the aim of identifying FEVR genes and describing the phenotypic spectrum of the disease, participants were recruited from six nonconsanguineous families by the Children's Hospital of Kunming Medical University (Kunming, China). In these families, children suffered from FEVR; however, the parents and other members of the family were normal. Additionally, our study enrolled 60 individuals of normal control aged between 6 and 35 years old, including 35 males and 25 females without associated hereditary diseases.
Clinical data were collected prospectively by the referring physician. These data included results from eye examinations and, whenever possible, ultrasonography and fundus photography. Parents and siblings at risk were also invited to participate and an eye examination was performed, including bestcorrected visual acuity, ocular alignment, slit lamp examination, dilated fundus examination and, in some cases, intravenous fluorescein angiography (IVFA).
The ethnicity of each participant was recorded by the referring physician who completed the study history questionnaire. This information was used in the analysis of the novel mutation screening.

WES and variant analysis
According to the principles of the Declaration of Helsinki, two-millilitre peripheral blood samples were
The polymerase chain reaction (PCR) duplicates were removed by using the Picard programme [47] [48]. The single nucleotide polymorphisms (SNPs) were determined by using the SOAP programme [49]. The reads were realigned using Burrows-Wheeler Aligner (BWA) software 0.7.15, and the deletions and insertions (indels) were detected with Genome Analysis Toolkit software 3.7. In addition, the identified indel SNPs were annotated with the Exome-assistant programme

Mutation validation
In these families, Sanger sequencing and PCR (polymerase chain reaction) with an ABI3500 sequencer were used to confirm potential causative variants. Regions sequenced included the 5 coding exons of the ZNF408 gene, the 6 coding exons of the LRP5 gene, the 11 exons of the KIF11 gene (including the 5′ untranslated and promoter regions), the 8 coding exons of the NR2E3 gene, the2 coding exons of the KRT3 gene, the 1coding exons of the FOXL2 gene. The sites of variation were identified in order to compare the DNA sequences using the corresponding GenBank (www.ncbi.nlm.nih.gov) reference sequences. Table 1 displays the sequences of forward and reverse primers. Thermocycling conditions were as follows: an initial denaturation of 95˚C for 10 min, 35 cycles of denaturation at 94˚C for 30 sec., annealing at 64˚C for 30 sec., extension at 72˚C for 45 sec. and a final extension of 72˚C for 5 min. The sequences of forward and reverse primers are in Table 4.

Availability Of Data And Materials
The data sets used and analysed during this study are available from the corresponding author on reasonable request.