Whole-Exome Sequencing Identifies Novel Variants that Co-segregates with Autosomal Recessive Retinal Degeneration in a Pakistani Pedigree

Abstract

Purpose

To identify the molecular basis of inherited retinal degeneration (IRD) in a familial case of Pakistani origin using whole-exome sequencing.

Methods

A thorough ophthalmic examination was completed, and genomic DNA was extracted using standard protocols. Whole exome(s) were captured with Agilent V5 + UTRs probes and sequenced on Illumina HiSeq genome analyzer. The exome-Suite software was used to filter variants, and the candidate causal variants were prioritized, examining their allele frequency and PolyPhen2, SIFT, and MutationTaster predictions. Sanger dideoxy sequencing was performed to confirm the segregation with disease phenotype and absence in ethnicity-matched control chromosomes.

Results

Ophthalmic examination confirmed retinal degeneration in all affected individuals that segregated as an autosomal recessive trait in the family. Whole-exome sequencing identified two homozygous missense variants: c.1304G > A; p.Arg435Gln in ZNF408 (NM_024741) and c.902G > A; p.Gly301 Asp in C1QTNF4 (NM_031909). Both variants segregated with the retinal phenotype in the PKRD320 and were absent in ethnically matched control chromosomes.

Conclusion

Whole-exome sequencing coupled with bioinformatics analysis identified potential novel variants that might be responsible for IRD.

27.1. Introduction

Retinal degenerations (RD) are the most common form of retinal dystrophy that affects 1 in 3000 individuals (Boughman et al. 1980). The disease is characterized by the progressive degeneration of photoreceptors, ultimately leading to blindness (Heckenlively 1988; Inglehearn 1998; Daiger et al. 2013). In the Retinal Information Network, currently, there are approximately 256 retinal disease-causing genes (RetNet; Hartong et al. 2006).

Mutations in these genes have been associated with a wide range of retinal phenotypes including cone or cone-rod dystrophy, leber congenital amaurosis, optic atrophy, retinitis pigmentosa, Usher syndrome, x-linked, and mitochondrial (Carr et al. 1978; Boughman and Fishman 1983; Boughman et al. 1983; Cremers et al. 1998; Morimura et al. 1998; Ayyagari et al. 2000; Nochez et al. 2009; Kenney et al. 2013). More than 187 genes have been reported that are involved with arRD (RetNet). Patients with arRD have a wide variation in the age of onset, the rate of progression, the severity of disease, and clinical symptoms (Heckenlively 1988; Zheng et al. 2015).

Here we describe the identification of the pathogenic candidate variants in a pedigree originating in Pakistan by linkage analysis and whole-exome sequencing.

27.2. Methods

27.2.1. Ethics

All research procedures were approved by the Institutional Review Boards (IRB) of National Centre of Excellence in Molecular Biology in Lahore, Pakistan; National Eye Institute in Bethesda, MD; University of California San Diego, in La Jolla, CA; and Johns Hopkins University in Baltimore, MD. Each participant in this study was informed and provided written consent in accordance with the Declaration of Helsinki.

27.2.2. Subjects and Clinical Examination

A three-generation pedigree with multiple affected family members was recruited from the Punjab province of Pakistan. The ophthalmic evaluation was performed as described previously (Duncan et al. 2007).

27.2.3. Exclusion Analysis

The genomic DNA used in all experiments was extracted from the white blood cells of whole-blood samples using QIAamp DNA blood mini kit (QIAGEN, Valencia, CA 91355). Short tandem repeat (STR) markers spanning known and/or reported IRD loci were selected for exclusion analysis. PCR products were mixed with a loading cocktail containing HD-400 size standards (Applied Biosystems) and resolved in an Applied Biosystems 3100 DNA Analyzer. Genotypes were assigned using the GeneMapper software from Applied Biosystems. Linkage analysis was performed with alleles of PKRD320 obtained through the genome-wide scan using the FASTLINK version of MLINK from the LINKAGE Program Package (Lathrop and Lalouel 1984; Schaffer et al. 1994; Naeem et al. 2015).

27.2.4. Whole-Exome Sequencing and Variant Calling

Genomic DNA from two affected and one unaffected individual was used for whole-exome sequencing. Whole-exome sequencing was performed using Agilent V5 + UTRs probes on the Illumina HiSeq (Illumina, San Diego, CA). The paired end (2 × 100 bases) reads were aligned to human genome hg19 after quality control check (Li and Durbin 2009; Genomes Project et al. 2015). The aligned reads were sorted by using SAMtools and stored in BAM format files. The GATK guidelines were followed to call the variants (DePristo et al. 2011; Tarasov et al. 2015). Variants were filtered using exomeSuite, a software (Maranhao et al. 2014) for a homozygous recessive inheritance pattern. Subsequently, we examined the allele frequency, impact of mutation, and expression profile of promising candidate variants (Table 27.1). Finally, all variants were confirmed by segregation analysis through dideoxy sequencing (MacDonald et al. 2012).

Table 27.1.

The list of databases we used for variant annotation

Category	Database	References
Mutation frequency	dbSNP 138	Sherry et al. (2001)
	ExAC	Lek et al. (2016)
	HGMD	Stenson et al. (2014)
Gene	GENCODE (gene, exon, and transcript)	Harrow et al. (2012)
microRNA	mirBase	Kozomara and Griffiths-Jones (2011)
Expression	UniGene	–
Impact prediction	PolyPhen2	–
	SIFT	–
	MutationTaster	–
Retina	RetNet	RetNet

27.3. Results

27.3.1. Linkage Analysis

The linkage analysis localized the disease interval to chromosome 11q. The haplotype constructed from alleles of additional short tandem repeat (STR) markers in that region further supported localization to chromosome 11q (Fig. 27.1).

Fig. 27.1.

Pedigree drawing of family PKRD320 illustrates segregation of variants identified in ZNF408 and C1QTNF4 with the retinal phenotype. Asterisk indicates the individuals that were chosen for whole-exome sequencing

27.3.2. Identification of Rare and Potentially Damaging Variants

59,064–63,286 SNPs were identified from two affected individuals (III:4 and III:7), whereas 63,996 SNPs were identified in the unaffected individual (III:1). On average, 5900 indels were found in each sample (the range 5295–6490). Analysis based on the pattern of inheritance identified 34,972 homozygous common variants in two affected individuals, which were absent or heterozygous in the unaffected. Among the identified variants, only 18 variants were rare (below 0.5%) and also present in the exonic sequence. Further analysis based on expression profile identified seven candidate variants (Table 27.2). Among these, PolyPhen2 predicted three missense variants to be damaging. Sanger sequencing revealed the novel, homozygous missense variant, c.1304G > A; p.Arg435Gln in the ZNF408 gene (NM_024741) segregating with disease in PKRD320 (Fig. 27.1). This variant was predicted to be probably damaging (PolyPhen2 = 0.983). Likewise, another homozygous missense variant c.902G > A; p.Gly301Asp in C1QTNF4 (NM_031909). a gene that is not yet associated with RD, co-segregated with IRD in PKRD320 (Fig. 27.1). This missense variant was predicted to be damaging (PolyPhen2 = 1.000). The arginine 435 in ZNF408 and glycine 301 in C1QTNF4 are highly conserved through evolution (Fig. 27.2). ZNF408 and C1QTNF4 genes are located on chromosome 11 and only 884,907 base-pair distance apart (Table 27.2). Both these variants are rare (0.0001 for the ZNF408 variant and the C1QTNF4 variant is novel) and located in the disease interval identified by the haplotype (Fig. 27.1).

Table 27.2.

Candidate variants observed in PKRD320

CHR	POS	rsID	ExAC	REF	ALT	GENE	cDNA Change	AA Change	PolyPhen2
chr11	46,726,554	rsl85413257	0.0001	G	A	ZNF408	c.1304G > A	p.R435Q	Probably damaging (0.983)
chr11	47,380,364	rs201791759	0.0034	C	T	SP11	c.524G > A	P.R175Q	Benign
chr11	47,611,461	–		C	T	C1QTNF4	c.902G >A	p.G301D	Probably damaging (1)
chr11	61,730,439	–		C	A	BEST1	c.1633C > A	p.P545T	Benign
chr11	63,525,727	–		A	ACGTCGAC	RTN3	c.509_510insCGTCGAC	p.(N171 fs)	Did not segregate
chr11	71,948,779	rs200679872	0.0016	G	A	INPPL1	c.2765G > A	p.R922Q	Benign
chr13	111,294,837	–		A	C	CARS2	c.1448T > G	p.L483 W	Probably damaging did not segregate

Fig. 27.2.

Illustration of amino acid conservation of Arg435 in ZNF408 and Gly301 in C1QTNF4 in their respective orthologs

27.3.3. Control Analysis

Analysis of 100 ethnicity-matched appropriate controls did not detect the novel causative variants identified in this study, indicating that these variants are rare in the Pakistani population.

27.4. Discussion

We report the use of whole-exome sequencing coupled with bioinformatic analysis to identify potential novel variants segregating with IRD in a large inbred family of Pakistani origin. Our analysis identified two homozygous potentially pathogenic variants segregating with disease. These variants reside in a previously reported Zinc Finger Protein 408 (ZNF408) gene and Tumor Necrosis Factor-Related Protein 4 (C1QTNF4) gene, which has not been associated with IRD.

ZNF408 encodes a transcription factor, and mutations in this gene have been associated with vitreoretinopathy and retinitis pigmentosa (Collin et al. 2013; Avila-Fernandez et al. 2015). ZNF408 is expressed in both cone and rod photoreceptor retinal layer. The znf408 knockdown transgenic zebrafish showed abnormal ventral retinal fusion and absence of blood vessels, leading to severe necrosis for the early stage of eye development (Collin et al. 2013).

C1QTNF4 is a member of the CTRPs family and has an important function on cytokine activity. Interestingly, another member of this family, C1QTNF5, has been associated with autosomal dominant late-onset retinal degeneration (Hayward et al. 2003; Ayyagari et al. 2005). This suggests that the homozygous variant identified in C1QTNF4 might play an important role in development of retinal pathology.

In conclusion, this study identified two homozygous variants in two different genes, i.e., ZNF408 and C1QTNF4, segregating with IRD in a Pakistani family. It is rather challenging to delineate the contribution of each of these variants toward the development of the retinal phenotype and establish the causative role of the C1QTNF4 variant in this pedigree. We speculate that the variant identified in ZNF408, a known IRD gene, is the primary cause of the phenotype in PKDR320. If the C1QTNF4 variant has any effect on the retinal tissue, it may exasperate the phenotype due to the ZNF408 variant. Future studies on the role of these mutant proteins and generation of animal models may reveal the potential role of the C1QTNF4 gene in the pathophysiology of retinal degeneration.