Establishing the involvement of the novel gene AGBL5 in retinitis pigmentosa by whole genome sequencing

Abstract

While more than 250 genes are known to cause inherited retinal degenerations (IRD), nearly 40-50% of families have the genetic basis for their disease unknown. In this study we sought to identify the underlying cause of IRD in a family by whole genome sequence (WGS) analysis. Clinical characterization including standard ophthalmic examination, fundus photography, visual field testing, electroretinography, and review of medical and family history was performed. WGS was performed on affected and unaffected family members using Illumina HiSeq X10. Sequence reads were aligned to hg19 using BWA-MEM and variant calling was performed with Genome Analysis Toolkit. The called variants were annotated with SnpEff v4.11, PolyPhen v2.2.2, and CADD v1.3. Copy number variations were called using Genome STRiP (svtoolkit 2.00.1611) and SpeedSeq software. Variants were filtered to detect rare potentially deleterious variants segregating with disease. Candidate variants were validated by dideoxy sequencing. Clinical evaluation revealed typical adolescent-onset recessive retinitis pigmentosa (arRP) in affected members. WGS identified about 4 million variants in each individual. Two rare and potentially deleterious compound heterozygous variants p.Arg281Cys and p.Arg487* were identified in the gene ATP/GTP binding protein like 5 (AGBL5) as likely causal variants. No additional variants in IRD genes that segregated with disease were identified. Mutation analysis confirmed the segregation of these variants with the IRD in the pedigree. Homology models indicated destabilization of AGBL5 due to the p.Arg281Cys change. Our findings establish the involvement of mutations in AGBL5 in RP and validate the WGS variant filtering pipeline we designed.

Keywords: ophthalmology; whole genome sequencing.

Fig. 1.

A: pedigree of family RA:RF:KR93 showing segregation of mutations in AGBL5. B: electropherograms showing the sequence encompassing the mutations. C: fundus photos of II:3 at age 24 yr showing bone spicules in the equatorial region with relatively preserved posterior pole. D: Goldman visual fields of II:3 at age 17 yr (I, OD; ii, OS) and at age 24 yr (iii, OD; iv: OS). OD, oculus dexter; OS, oculus sinister.

Fig. 2.

Conservation of p.R281 and the flanking sequence in orthologs of AGBL5.

Fig. 3.

Homology models of sections of the AGBL5 protein product, derived from either 3l2n or 4b6z templates, in which the cartoon backbone of variants is colored blue. Active site representations, including the zinc coordinating residues (His252, Glu255, and His434) and the nucleophile (Arg303), are shown for reference. His434 is only present in the model derived from the 4b6z template. Possible polar contacts are indicated by dashed yellow lines in each model, though these predictions are of low confidence, especially in B and C. The previously reported Asp295Asn variant is shown in A, while the presently identified variant Arg281Cys is shown in B and C using the model derived from the 3l2n template and the 4b6z template, respectively. The models in B and C are aligned to one another to depict the same perspective. A: the Asn variant can replace the Asp without additional steric hindrance; however, the replacement of a side-chain oxygen with nitrogen may have consequences for polar contacts, potentially with the backbone atoms of residues Leu200 and Asp201 and/or with the side chain hydroxyl of Ser199. B and C: the model quality is too poor for high-confidence assertions about specific atomic or residue level interactions; however, the Cys side chain can be accommodated in similar rotameric states to the Arg, suggesting that loss of polar contacts and/or the loss of packing interactions, rather than a steric clash, could explain functional consequences.