Branch point mapping

HEK293 cells were grown in DMEM with or without DBR1 siRNAs (Supplementary Table S1) and harvested 72 h after (mock) transfection for RNA extraction. The final concentration of each duplex was 40 nM. DBR1 encodes a debranching enzyme that hydrolyzes 2′-5′ branched phosphodiester bonds, converting lariats into linear molecules for degradation (37). A lack of debranching activity in vivo leads to accumulation of excised lariat introns. Total RNA was extracted using TRI-reagent and treated with DNase. One microgram of purified RNA was reverse transcribed with the SuperScript™ III cDNA synthesis kit (LifeTechnologies) and primer R1 (Supplementary Table S1). For exon Ab, the first-strand cDNA was amplified with outer primers F1 and R1 in the first round of PCR, which was divided into multiple second rounds of PCR with inner primers F2 and R2 (Supplementary Table S1). For exon 3, we employed primers R1/F1-3 and R2/F2-3. Each step was carried out at several annealing temperatures. Amplicons were gel-purified, ligated into pGEM-T Easy (Promega) and sequenced.

RNA-Seq data generated from cultures treated with or without siRNAs targeting U2AF1 isoforms (ArrayExpress accession number E-MTAB-2682) were searched for 15- and 20-nt sequence strings at the 5′ss of U2AF1 intron 2 (three mismatches allowed). In addition, we analysed ENCODE RNA-Seq data from 14 cell lines sequenced using Illumina GA and GAII (38) and Illumina Body Map data of 16 different human tissues sequenced using Illumina HiSeq 2000.