Purchase this article with an account.
Maureen Neitz, Jay Neitz, Jessica Rowlan, Candice Davidoff; Observed versus predicted exon 3 skipping in OPN1LW using a hexamer-linear-additive model (HAL). Invest. Ophthalmol. Vis. Sci. 2016;57(12):3186.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
In modern humans, recombination has intermixed the OPN1LW and OPN1MW genes, generating a large variety of genetic haplotypes. Some combinations of the 8 variant nucleotides in exon 3 have been associated with vision disorders. A major effect of the exon 3 haplotype is on exon 3 inclusion in the final messenger RNA. The mechanism by which nucleotide variations promote exon skipping are mysterious. A strong understanding of splicing signals is required to develop an accurate bioinformatics tool that accurately predicts exon skipping, and such a tool would be useful clinically. Here we tested HAL, a new exon skipping prediction web tool that was created by molecular phenotyping millions of alternatively spliced mini-genes containing random sequences (Rosenberg et al. Cell 163, 693-7111, 2015). We assessed the ability of HAL to predict accurately exon 3 skipping in OPN1LW.
Minigenes contained an OPN1LW cDNA in which introns 2 and 3 were inserted. 112 minigenes that differed in the haplotypes of exon 3 with regard to 8 variant nucleotide positons were generated and transfected into HEK293 cells in duplicate. The percentage of recovered messenger RNA (mRNA) that excluded exon 3 was measured. The observed data was compared to predictions made by HAL for a control OPN1LW exon 3.
There was a highly significant correlation between the predicted and observed degree of exon 3 skipping for our dataset, with a Spearman coefficient of 0.6466 (corrected for ties, 95% CI 0.5199 to 0.7455), and p < 0.0001. There were some variants of exon 3 for which the extent of exon 3 skipping was well predicted by HAL, but there were also notable failures.
The rich variety of exon 3 haplotypes in the L and M cone photopigment genes (OPN1LW and OPN1MW) provide a naturally occurring system in which to identify molecular signals for splicing and the interplay between nearby signals. The dataset used to develop HAL suggested that interactions between adjacent hexamer elements that influence splicing were additive, however, the pattern of HAL’s failures in predicting exon 3 skipping in OPN1LW variants suggest there are significant non-linear interactions. Further investigations of the OPN1LW and MW genes promise to improve our understanding of splicing signals.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.
This PDF is available to Subscribers Only