Control of gene output by intron RNA structure

Intron removal through pre-mRNA splicing is a central step in gene expression across Eukarya. The process initiates with the recognition of intronic sequence elements (splice sites) by highly conserved RNA-protein components of the spliceosome.^1,2 Intron sequences themselves are not generally conserved beyond the splice sites, yet intronic mutations are often associated with genetic disease.³ Here we systematically test if and how intron RNA structure formation modulates gene output. We generated intron variant libraries that measure the impact of base pairing at every position across a natural intron. Using a massively parallel reporter assay (MPRA),⁴ we find that base pairing involving the splice sites modulates splicing across orders of magnitude. An additional intron region upstream of the branch site was also sensitive to structure, suggesting steric hindrance. Combining thermodynamic structure prediction with libraries designed to sequester splice sites in structures of varying stability, we show that machine learning models can nearly fully explain observed gene output. Informed by this, designed alterations in intron sequence that modulate base pairing are shown to improve inefficient splicing of human β-globin IVS1.⁵ Finally, intronic mutations that alter RNA structure emerge rapidly under selection pressure, providing eukaryotes with a simple evolutionary strategy to fine-tune gene output.