Abstract
Purpose :
Mutations in the ATP-binding cassette transporter type A4 (ABCA4) gene cause Stargardt disease. Non-canonical splice site (NCSS) and deep-intronic variants constitute a large fraction of alleles identified to date, i.e. ~10.4% and ~8%, respectively. However, defining the functional consequences of them remains a challenge. In this study, we aim to define the splicing effect of several previously reported and unpublished NCSS and deep-intronic variants in ABCA4.
Methods :
We selected 7 NCSS and 10 deep-intronic variants based on 5 splice site prediction and 1 Exonic Splice Enhancer prediction tools integrated in Alamut® version 2.10. We generated mutant from wild-type midigene constructs that were described previously (Sangermano et al. 2018). Mutant and wild-type constructs were transfected in Human Embryonic Kidney (HEK293T) cells. Finally, transcriptional and Sanger sequence analysis was performed to determine the splicing effects of these variants.
Results :
In this study, 10 out of 17 variants showed splicing defects. The 3 deep-intronic and 1 near-exonic variants that create pseudoexons and elongates the upstream exon respectively are classified as severe alleles with less than 20% of remaining wild-type RNA. Furthermore, the 6 NCSS variants with splicing effects caused the partial deletion or full skipping of one or more exons in mRNAs. Except c.161G>A, the moderate allele in which 51% of relative wild-type RNA is present, in others, either no or less than 4% remained. Among these 10 variants, 8 lead to premature stop codons and truncated ABCA4 proteins. We did not observe any splicing defects for the other 7 variants, which showed differences between wild-type and mutant mRNAs in either splice site or ESE prediction tools.
Conclusions :
We have shown the splicing effects of 17 rare NCSS and deep-intronic variants in ABCA4. This study suggests that the enhancers and silencers can play essential role in splicing. For instance, the c.4539+1729G>T variant has no effect, even though 4 splice site programs predicted the creation of strong nearby donor site (human splicing finder: 89.3). Therefore, based on the in silico ESSs predictions, silencers are activated and prevent any splicing defect. These findings suggest that more caution needs to be applied while using the in silico splice site, ESEs, and ESSs prediction tools to predict the splice outcome of NCSS and deep-intronic variants.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.