Regulation and Function of Non-Exonic Recursive Splicing
Removal of intron sequences from pre-mRNA by splicing is an essential step in the expression of most human genes. Recursive splicing is a mechanistic variant shown to mediate the stepwise removal of very long introns and a subset of alternative splicing events in Drosophila melanogaster. Recursive splice sites (RSSs) consist of juxtaposed 3’ and 5’ splice site motifs that define a zero-nucleotide exon whose 3’ and 5’ splice sites are coincident. Only 10-20% of RSSs in Drosophila are involved in alternative splicing, so most RSSs are non-exonic. Consequently, their biological role is unclear despite their high conservation and preferential association with long introns. Although RSSs can be predicted in human genes as well, their function has not been demonstrated. In this work, I study three problems: (1) How cis-elements contribute to the proper use of a non-exonic RSS. (2) The biological role of a non-exonic RSS in its natural context. (3) Verification of recursive splicing in human genes and some of its functional consequences.
It is unclear how RSSs can function without interference between their 3’ and 5’ss components. The consensus RSS motif suggests that these elements are biased to function initially as a 3’ss due to enhanced features of the 3’ss component. In addition, most non-exonic RSSs are (surprisingly) associated with silent downstream 5’ss motifs at a position where they would be expected to define an exon. This could also help activate the RSS as a 3’ss by interaction with the non-overlapped 5’ss (“pseudo-exon definition”), but some mechanism would have to redirect splicing subsequently to the regenerated 5’ss rather than the downstream site. Intriguingly, the silent downstream 5’ splice sites are at a distance from the RSS where regular 5’ splice sites exhibit a peak in the distribution of enhancers. Experimental dissection of an example associated with non-exonic RSS RP3 in the Ultrabithorax gene of Drosophila revealed that the downstream pseudo-‐5’ss is not required for activation of the RSS as a 3’ss. Instead, it functions as part of a conserved module that stimulates use of the 5’ss that is regenerated by the RSS. This prevents inappropriate use of competing alternative and cryptic sites. The result is consistent with the hypothesis that the enhanced 3’ss component and branch site are sufficient to ensure their activity, but that activation of the regenerated 5’ss requires assistance. The regulatory function of the pseudo-5’ss requires base-pairing with U1 snRNA, even though splicing does not occur at this position. Similar modules may assist the efficient and correct sequential activity of many RSSs. These arrangments may reflect a dynamic evolutionary history of interconversions between exonic and nonexonic RSSs.
To test the hypothesis that recursive splicing is important for the correct and/or efficient expression of genes with long introns, a two-step gene replacement strategy was used to delete the non-exonic RSS RP3 from within a 50 kb intron in the endogenous Ultrabithorax gene and to generate isogenic wild-type control chromosomes. A change in the alternative isoform ratios was detected by semi-quantitative RT-PCR, and phenotypic analyses indicate that deletion of RP3 leads to a mild loss of function. Additionally, a white marker gene inserted near RP3 is profoundly silenced but can be reactivated by deletions extending upstream, suggesting a repressive chromatin structure in this region.
A sample of predicted RSSs from human was tested using the same approaches as used previosuly in Drosophila. These tests made use of a recursive splicing reporter system or minigenes transfected into human cell lines. For three out of eight human RSS candidates tested, it was possible to detect the predicted recursive intermediates and a shift to use of an alternative 5’ss after mutation of the RSS 5’ss motif. A RSS associated with a novel ORF-truncating cassette exon (E3b) in the human dopamine reuptake transporter gene SLC6A3 was also validated. Alternative splicing of exon E3b was verified in endogenous transcripts in the substantia nigra of adult human brain and in reporter and minigene transcripts in a transfected neuronal cell line. E3b is flanked by single-nucleotide polymorphisms (SNPs) that appear to be associated with differential risk for schizophrenia. The risk-associated haplotype increases the inclusion of E3b in cell transfections assays and thus might be associated with reduced expression of dopamine reuptake transporter in vivo. An ORF-truncating exon E3b is present in all sequenced mammalian genomes except mouse, rat and rabbit, suggesting that recursive splicing of this cassette exon normally plays a role in regulating dopamine activity, and that genetically determined differences in regulation can underlie or exacerbate dopaminergic dysfunction.