Fig. 2

SRSF7 binds preferentially at pPASs and modulates their usage in a splicing-independent manner. a Metaprofiles of normalized iCLIP crosslink events of SRSF3 and SRSF7 around sPASs (left), pPASs and dPASs (right). Raw signal and loess smoothing are shown. Positions with a significant signal difference between SRSF3 and SRSF7 are marked below (false discovery rate [FDR] ≤ 0.01, two-proportions z-test). b Metaprofiles of normalized iCLIP crosslink events of SRSF3 and SRSF7 around pPASs and dPASs of transcripts with shorter 3′UTRs upon Srsf3 KD (SRSF3 targets) compared to randomly sampled PASs that are not affected by Srsf3 KD. Raw signal and loess smoothing are shown. Significant differences in SRSF3 and SRSF7 binding are computed by a z-score approach. Positions with FDR ≤ 0.01 are indicated below. c Genome browser view of the Ddx21 3′UTR: Top: RNA-seq coverage in control conditions and after Srsf3 and Srsf7 KD. Middle: MACE-seq coverage at the pPAS and dPAS. Bottom: iCLIP crosslink events of SRSF3 and SRSF7. Merged replicates are shown for all datasets. d Scheme of reporter genes. mCherry and Luciferase (Luc) were fused with endogenous 3′UTRs of Ddx21, Anp32e, and Rab11a. e Semiquantitative 3′RACE-PCR (top) of Luc-Ddx21 reporter upon SRSF3-GFP or SRSF7-GFP overexpression, validated by Western blot (bottom). Antibody against α-tubulin (α-TUB) was used as loading control. f Quantification of 3′RACE-PCRs (n = 2) of Luc-Ddx21 reporter transcripts upon SRSF3-GFP and SRSF7-GFP overexpression. Data are represented as mean ± standard deviation of mean. g Semiquantitative 3′RACE-PCR of transiently expressed mutated mCherry-Ddx21 reporters. h Quantification of 3′RACE-PCRs (n = 3) of mutated mCherry-Ddx21 reporter transcripts. Data are represented as mean ± standard deviation of mean. Student’s t test, *P value < 0.05, ***P value < 0.0005