Fig. 4

Evolution of-tissue specific APA across Drosophila species. a Reannotation of D. yakuba and D. virilis 3′ UTRs from deep 3′-seq data yields thousands of 3′ UTR extensions. b Percentage of genes with either one annotated end or more than one for the current D. yakuba annotation (r1.05) or D. virilis annotation (r1.06) and our revised 3′-seq-based atlas. c, d Examples of expression of genes with orthologs in D. melanogaster, D. yakuba, and D. virilis. RNA-seq and 3′-seq of head, ovary, and testis are shown as illustrated in the legend. 3′-Seq is overlaid in light green onto the RNA-seq track. c Example of a gene with conserved tissue-specific patterns of 3′ UTR expression. Extension of the annotation of D. yakuba and D. virilis is shown in light blue. d Example of a gene that shows de novo expression of a dominant short 3′ UTR isoform in the testis of D. virilis. e–h Weighted 3′ UTR length comparison between tissues. Weighted 3′ UTR length is obtained taking the average of all 3′ UTR isoform lengths per gene weighted by the contribution of each isoform expression. Genes are expressed at a minimum of 5 RPM in all samples. The genes for which weighted 3′ UTR length differs by 100 bp or more between samples are shown colored: red, longer weighted length in the sample on the x-axis; blue, longer weighted length in the sample on the y-axis. Comparisons are shown for D. yakuba and D. virilis. e, f Head vs testis. g, h Testis vs ovary. i Conservation of tissue-specific APA. The overlap of the genes that express a weighted 3′ UTR length of more than 100 bp in head compared to both testis and ovary, have orthologs in all three species, and are expressed at least at 5 RPM was taken and is represented as a bar graph. Schematic Venn diagram and dots below the bar graph show the nature of each intersection. Horizontal bar graph shows the number of genes considered for each species. j Preferred conservation of head extensions across Drosophilid phylogeny. This analysis was done as in i, except that the gene set was first conditioned on genes with specific evidence for head-extension in D. melanogaster from comparison to carcass (e.g., Fig. 3l), then were overlaid onto the gene sets with differential 3′ UTRs between head and testis 3′-seq datasets across species. Nearly all of these loci exhibit a conserved pattern