Precursor RNA structural patterns at SF3B1 mutation sensitive cryptic 3' splice sites

SF3B1 is a core component of the spliceosome involved in branch point recognition and 3' splice site selection. The SF3B1 K700E mutation (lysine to glutamic acid) is common in myelodysplastic syndrome and other blood disorders. SF3B1 K700E mutants utilize novel cryptic 3' splice sites; however, the properties distinguishing SF3B1-sensitive splice junctions from other alternatively spliced junctions are unknown. We identify a subset of 192 cryptic 3' splice junctions with significantly altered use in SF3B1 K700E cells, termed SF3B1-sensitive cryptic 3' splice sites, and 2800 cryptic 3' splice sites used in SF3B1 wild-type, termed SF3B1-resistant. We find that SF3B1-sensitive cryptic 3' splice sites are embedded in extended polypyrimidine tracts. Furthermore, canonical splice sites paired to SF3B1-sensitive cryptic 3' splice sites are significantly weaker than canonical 3' splice sites paired to SF3B1-resistant cryptic 3' splice sites. We test whether SF3B1-sensitive splice sites are structurally different from SF3B1-resistant 3' splice sites using chemical probing. We develop experimental RNA structure data for 83 SF3B1-sensitive junctions and 39 SF3B1-resistant junctions. We find that the pattern of structural accessibility at the NAG splicing motif in cryptic and canonical 3' splice sites is similar. However, the magnitude of accessibility differences is less in paired SF3B1-sensitive splice sites than in paired SF3B1-mutant splice sites. Additionally, SF3B1-sensitive splice junctions are more flexible than SF3B1-resistant junctions. Our results suggest that SF3B1-sensitive splice junctions have unique structure and sequence properties, containing poorly differentiated, weak splice sites that lead to altered 3' splice site recognition in the presence of SF3B1 mutation.