Abstract
Purpose :
The purpose of this study is to better evaluate the contribution of deep intronic splicing mutant alleles to inherited retinal diseases (IRD).
Methods :
Whole-genome sequencing was performed for 741 patients who lack a confident molecular diagnosis from a cohort of approximately 7000 IRD patients. We filtered and annotated genomic alterations with a custom pipeline and predicted the effects of intronic variants on splicing using SpliceAI. The candidate splicing variants were restricted to those found in IRD genes and with a SpliceAI score ≧0.02. The candidates were further filtered based on allele population frequency, the distance from the exon-intron junctions, mode of genetic inheritance, and clinical phenotype. To further validate the in silico predictions, we utilized an in vitro minigene system to test a set of 130 candidate deep intronic splicing mutations with a score range of 0.02 to 1. DNA oligos with which sequences include the candidate mutation, SpliceAI-predicted cryptic exon, and flanking regions were synthesized, PCR-amplified, and cloned into a minigene vector. Each construct was separately transfected into HEK293 cells, from which RNA was extracted and RT-PCR was performed to assess the splicing products.
Results :
Within the 741 unresolved patients, 18182 candidate deep-intronic splicing mutations were identified with a SpliceAI cutoff score of 0.02, but only 279 candidates have a score above 0.2, which is the cutoff commonly used by clinical labs. Among the selected 130 candidate IRD deep-intronic splicing mutations that were tested in vitro, all high-scored candidates (≧0.5) resulted in the generation of aberrant transcripts. However, for the mid-scored candidates (0.1≦ SpliceAI score <0.5), only half of the candidates affected splicing in vitro, while the majority of candidates with a score less than 0.1 failed to be validated to affect splicing.
Conclusions :
Our study is the largest-scale study on deep-intronic splicing variants in IRD diseases. Our findings demonstrated that although there is a good correlation between SpliceAI score and splicing effects, given that many low-scored deep-intronic candidates also generated aberrant transcripts, the contribution of intronic variants, especially those deep within introns, to genetic diseases might be underestimated and further optimization of the prediction tool is needed.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.