Abstract
Purpose :
Missense variants in ABCA4 constitute approximately 50% of causal variants in Stargardt disease (STGD1). Their pathogenicity is attributed to amino acid changes, but their potential role in splicing disruption has not been studied systematically.
Methods :
To select missense and synonymous variants of interest, we used SpliceAI to ascribe defective splice predictions on a dataset of ABCA4 variants from 5,579 bi-allelic STGD1 cases of which 961/1,619 variants have an attributed severity score (PMID 35120629). Variants were selected if SpliceAI predicted an acceptor gain (AG) delta score >0.12 or donor gain (DG) >0.2, and if no previous effect on splicing was reported. Eighteen ABCA4 variants were selected, seven of which predicted to create a new splice acceptor site (SAS) and eleven to create a new splice donor site (SDS). The variants were introduced in midigenes that contained 4 to 12 kb of ABCA4 genomic sequence. The mutant and wild-type (WT) constructs were expressed in HEK293T cells. Transcripts from the midigenes were identitifed by RT-PCR and Sanger sequencing, and semi-quantified using ImageJ. Statistical analysis was performed with two-tailed Student’s t-test and the Pearson correlation factor.
Results :
RT-PCR of RNA of midigene-transfected HEK293T cells revealed predicted splice aberrations for 17 out of 18 analyzed variants. We identified statistically significant differences vs. WT in 15 variants (4/7 predicted SAS variants and 11/11 predicted SDS variants). A correlation was observed between the SpliceAI delta scores and the percentage of expressed WT for the AG variants (R2=0.77, p<0.05), but not for the DG variants (R2=0.01, p=0.7). The latter is likely due to the higher mean DG delta scores (0.83) that led to big mis-splicing effects (as opposed to mean AG delta scores of 0.45). DG variants with delta scores >0.65 contained between 0 and 12.5±1.4% of WT RNA, with the exception of c.2128A>G that contained 86.6±1.2% of WT RNA. Moreover, 4 of 5 analyzed synonymous variants showed statistically significant differences vs. WT with 0 to 47.5±1.8% of WT RNA.
Conclusions :
By combining SpliceAI predictions and midigene splice assays, we report evidence for missense and synonymous ABCA4 variants to be causal by attributing their impact on splicing. These results improve proper classification of missense and synonymous ABCA4 variants and could lead to new treatment strategies for STGD1.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.