Whole genome sequencing delineates novel non-coding variants and candidate genes in inherited retinal diseases

Abstract

Purpose : Whole genome sequencing (WGS) has mainly uncovered deep-intronic splicing variants and structural variants (SVs) in unsolved exome-tested cases with inherited retinal diseases (IRD). The contribution of non-coding variants that regulate gene expression and variants in candidate genes has been generally underinvestigated. Here, we leveraged WGS in a prescreened exome-tested IRD cohort to provide insight into the contribution of non-coding variation and of candidate IRD genes.

Methods : WGS was performed in 82 patients (77 probands) with IRD. Coding and non-coding regions of IRD and candidate genes were analyzed using the in-house Seqplorer tool and Franklin (Genoox). SV analysis was done using ExomeDepth, Manta and Delly. Intronic variants were prioritized using SpliceAI and Alamut Visual. Candidate regulatory variants were assessed in silico using multi-omics datasets such as RegRet or in vitro reporter assays. Single-cell expression of candidate genes was investigated using retinal single-cell datasets such as the IOB Retina Atlas and Spectacle. Segregation and clinical reassessment were performed when possible.

Results : Novel (likely) pathogenic regulatory variants were found in 4% (3/77), one of which is a promotor variant of RPGRIP1 c.-152A>C in a coding monoallelic case, predicted to disrupt an OTX2 binding site. Two 5’UTR variants were identified in a monoallelic case (BBS12 c.-11+3dup) and in a patient with macular disease (ELOVL4 c.-187T>G) respectively, likely solving the diagnosis in both cases. Coding variants in novel candidate genes represent 6.5% (5/77), such as ACACB, encoding a player in lipid metabolism. Variants in genes that were recently implicated in mostly syndromic IRD, such as ALPK1, GRN and ITM2B, facilitated a genotype-driven diagnosis in 9% (7/77). Analysis of non-coding regions in monoallelic cases allowed to pinpoint putative deep-intronic splicing variants in 16% (12/77), illustrated by the first deep-intronic ALMS1 variant that was shown to lead to pseudo-exon inclusion by minigene assays. Finally, variants in well-established IRD genes were found in 12% (9/77), including 2 SVs (3%, 2/77).

Conclusions : We demonstrate that non-coding regulatory variants and deleterious variants in candidate genes contribute to over 10% of the genetic architecture of our prescreened IRD cohort. Overall, WGS analysis could solve missing heritability in up to 46%.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.