Splicing factors and chromatin organization enhance exon recognition by alleviating constraints generated by gene nucleotide composition bias

To characterize the rules governing exon recognition during splicing, we analyzed RNA-seq datasets and identified ~4,000 GC-rich and ~5,000 AT-rich exons, labelled GC-exons and AT-exons, respectively whose inclusion depends on different sets of splicing factors. We show that a high GC-load is associated with predicted RNA secondary structures at 5'ss and that GC-exons are dependent on U1 snRNP-associated proteins. Meanwhile, a high AT-load is associated with a large number of decoy splicing-related signals upstream exons such as the number of branchpoints and SF1- or U2AF65-binding sites and AT-exons are dependent on U2 snRNP-associated proteins. Nucleotide composition bias also influences local chromatin organization. Since the GC content of exons correlates with that of their hosting-genes, -isochores and – topologically-associated domains, we propose that regional nucleotide composition bias leaves a footprint locally, at the exon level, inducing, during splicing, constraints that are alleviated by the local chromatin organization and specific splicing factors. Samples siFUS, siGL2, siHNRNPC, siHNRNPH1, siHNRNPK and siTRA2A-B were made together and are part of the same experiment (simplicates). Samples siPP-1, siPP-2, siPP-3, siGL2-1, siGL2-2 and siGL2-3 were made together from different cell batchs and are part of the same experiment (triplicates).