Recursive sites aid in efficient splicing of long <i>Drosophila</i> introns.

(A) The percentage of splice junction reads (mean percentage across recursive introns; x-axis) that span the exon-exon boundary (5' splice site– 3' splice site; dark blue), a 5' splice and recursive site (light blue), two recursive sites (gold), and a recursive site and 3' splice site (yellow) across different labeling periods (y-axis). (B) Splicing half-lives for individual recursive segments (dark tan) and full non-recursive introns chosen to match recursive segment lengths (grey). (C) The delay in splicing half-life of the first recursive segment of an intron, relative to the time to transcribe the remainder of the intron (y-axis) for recursive introns with different numbers of recursive sites (x-axis). Increased splicing delays (> 1) are supporting of post-transcriptional splicing of the first segment, while decreased splicing delays ((D) Splicing accuracy measured by percentage of non-canonical unannotated reads for recursive introns (gold), non-recursive introns matched for intron length (grey), and all non-recursive introns (light grey, dotted). (E) Distribution of the highest maximum entropy score for an internal 5' splice site motif (x-axis) for initial recursive segments (not expected to contain an RS-exon; yellow); non-first recursive segments (potentially contain RS-exons; gold), and non-recursive introns matched for intron length (grey). Significance is indicated such that **: P < 0.01 and ***: P < 0.001, with a Kolmogorov-Smirnov test.