Pre-mRNA splicing is an integral part of metazoan gene expression, involving a vast array of chromatin-associated actors; including transcriptional machinery, chromatin remodelling complexes, and splicing factors. Through these agents, splicing is differentially regulated throughout development and differentiation, but is subject to widespread dysregulation in malignant neoplasia. Given the number and diversity of splicing regulators and the auto-regulatory role of many splicing networks, deconvoluting the regulatory effect of individual factors on co-transcriptional splicing is critical to addressing oncogenic splicing in the clinic.
Here, we leverage large-fragment RNA sequencing of RNA polymerase II-bound primary transcripts (POINT-Seq) [1] to assess the dynamics of co-transcriptional splicing in HeLa cells. We developed a bias-controlled computational pipeline to measure the kinetics and order of co-transcriptional splicing, and to measure the mechanistic influence of different genotypic, physiological and pharmacological factors from nascent-RNA sequencing. Using our pipeline, we identify more than 8 million splicing intermediates and quantify the kinetics and order of splicing across ~65,000 adjacent exon pairs in the human transcriptome. We demonstrate that the order of exon ligation is largely variable at most intron pairs, indicating that while the pre-spliceosome may assemble in the order of transcription, splicing catalysis is subject to additional regulation and often occurs in different orders on the same primary transcripts. While most intron pairs can splice in any order, we identify a kinetically-distinct class of ‘polar’ intron pairs where the splicing of an follower-intron can only complete following the splicing of an adjacent leader-intron. Finally, we demonstrate an association between altered co-transcriptional splicing order and alternative isoform usage. Our results provide critical evidence that co-transcriptional splicing coordination underlies alternative splicing throughout the human transcriptome.