Distinguishing Variants that Create an Ectopic Exonic Splice Site from Exon Skipping SAVs

Summary

Another of the possible splice-affecting outcomes of an internal exonic variant is the creation of an ectopic splice site which is then used in preference to that of the natural splice sites.

We used a set of 54 experimentally verified examples of this phenomenon from the DBASS5 and DBASS3 databases to discern features that distinguish ectopic splice-site activating SAVs from exon skipping SAVs as well as common HapMap polymorphisms.

We created a simple metric, ΔSS, which measures the maximum difference in splice site scores (using the MaxEnt splice site scoring program) between the wild-type sequence and the variant sequence for all possible 5' and 3' splite sites around the site of the variant. We also measured whether the strongest splice sites created by a variant was of an equal or greater score than the natural splice site.

We found that the vast majority (~85%) of ectopic SAVs have ΔSS scores of greater than 1 compared to just a fifth of exon skipping SAVs and 8% of HapMap SNPs. In addition to having a ΔSS score of ≥1, 54% of ectopic SAVs create splice sites that are equal or greater in strength than the natural splice site, compared to just 2% of exon skipping SAVs and 1% of HapMap SNPs. These features therefore allow a clear separation between exon skipping SAVs and ectopic splice site SAVs.

In Skippy, if a variant does not produce a ΔSS score of > 0 or the strongest splice site created does not have a score of ≥3 (a weak but viable splice site) then a ΔSS score of 0 is reported. Where the ΔSS score is 0 for both 5' and 3' splice sites, the box is shaded red to signify that, in combination with other features, this feature is associated with exon skipping (because these do not generally create ectopic splice sites).


ESR Changes Associated with Ectopic Splice Site SAVs

In addition, we found that ectopic splice site SAVs, like exon skipping SAVs, are significantly enriched for gains of ESSs (P=2.85x10-15) as well as ESS alterations (P=1.95x10-3) but not other ESR changes such as ESE losses.By looking at densities of ESEs and ESSs in and around natural splice sites averaged over ~25,000 internal constitutively spliced exons, we can see that the both the 3' and 5' splice sites cause sharp spikes in ESS density directly around the splice site. The creation of strong ectopic splice sites is therefore the most likely cause for this enrichment in ESS gains and alterations. Nevertheless, the real set of ectopic SAVs creates almost twice as many ESS motifs as a set of 55 HapMap SNPs with large ΔSS values and ectopic splice site scores that are greater than the natural splice sites. It is therefore possible that ectopic SAVs are also affecting exon definition stability that causes the ectopic splice site to replace the natural splice site.

Finally, when we compare the distribution of ectopic SAVs across the exon with those of the ectopic-like HapMap SNPs, made up in each case of around 80% of variants that activate (or look like they activate) a 5' splice site and 20% of 3' splice sites. We found that for real ectopic SAVs, the vast majority of 5' ectopic SAVs cluster towards the 5' splice site while those that activate 3' ectopic sites cluster towards the 3' splice site. We found the exact opposite ditribution for HapMap ectopic-like variants, suggesting that most of those sites are not real.