Deployment of non-canonical splicing in tunicate genomes is mediated by divergent U2AF function and re-patterning of snRNA m6A modification

In eukaryotes, splicing is a critical step of gene expression that produces transcripts competent for protein synthesis by removing introns from pre-mRNA. The spliceosome defines target splice sites based on sequences located around intron ends and in almost all cases, spliced segments begin with a GT and end with an AG. The key molecules for this function are the spliceosomal RNA (snRNA) and the U2AF heterodimer, which interact with splice signals found at the ends of introns. In chordates, tunicate genomes in the family Fritillariidae (Lohmann, 1915) form a remarkable exception by hosting numerous introns that escape the GT/AG rule, and they represent a material of choice to address the evolution of intron recognition. A comparison of chordate genomes revealed that the emergence of the unusual splicing code in Fritillaria borealis was accompanied by the duplication of its subunits U2AF1 and U2AF2. We found that the conserved duplicates bind to canonical splice signals on RNA, but different binding was seen with the divergent versions U2AF1β and U2AF2β, suggesting they could recognize other splice sites. When expressed in mammalian cells, U2AF2β promotes the removal of introns that have a weak polypyrimidine tract, and we also show that it interacts specifically with AHCYL1/2, a protein that can regulate the N6-methyladenosine (m6A) modification of snRNA U2 and U6. In striking contrast with other species, F. borealis has lost m6A on the U6 snRNA while U1 snRNA has gained a stable, 5′-terminal m6A. We used structure prediction to show how these modifications could play decisive roles during the recognition of non-GT/AG introns. Our study suggests that discrete changes implemented on the spliceosome during the evolution of Fritillariidae have contributed to a dramatic shift in splice site identity.