Fig. 8: Experimental validation for the regulatory impact of RBPs on microRNA targeting.

a Fold repression of 3′UTRs with a miRNA target site (MTS) near an RBP-binding site (RBS) measured by luciferase reporter assay with the design of constructs shown left. Four different designed constructs were used: (1) wild-type MTS and RBS (MTS^WTRBS^WT), (2) mutated MTS (MTS^MUTRBS^WT), (3) mutated RBS (MTS^WTRBS^MUT), and (4) mutated MTS and RBS (MTS^MUTRBS^MUT). The normalized fold repressions were compared between MTS^WTRBS^WT and MTS^WTRBS^MUT (two-sided Wilcoxon’s rank-sum test, *P < 0.05, **P < 0.01, ***P < 0.001, n = 12, See Methods for normalization procedures). The changes in the minimum free energy of the 3′UTRs (ΔG_miR − ΔG₀ and ΔG_RBP+miR − ΔG_RBP) are listed at the bottom. KATNA1 was used as a technical control. The mean values ±95% confidence intervals are displayed. P values are provided in Source Data. b The design of constructs and the expected changes in the secondary structures of mRNAs are shown (top left): parental, RBP KO, and rescue conditions of the deleted RBP are depicted by blue, orange, and green colors, respectively, with striped colors representing the mutated MTS. The protein abundance was quantified by western blot (top right). For the rescue experiment of each RBP KO, FLAG-tagged RBP was quantified (top right). GAPDH levels serve as a loading control. The MTS^WT values were normalized to MTS^MUT and were compared between parental and KO and between KO and rescue (two-sided Wilcoxon’s rank-sum test, *P < 0.05, **P < 0.01, ***P < 0.001, n = 12). Otherwise as in (a). c Transcriptome-wide response of miRNA targets after RBP removal. mRNA expression fold changes after miRNA overexpression were measured for IGF2BP1 or PCBP2 KO cells and compared to HEK293T parental cells, with the x-axis indicating the difference of the log₂(mRNA fold change) values. Target 3′UTRs with a single 7, 8mer MTS were selected if the distance between an MTS and the nearest RBS, denoted as d_MTS-RBS, is short (<100, orange) or long (≥100, blue). After controlling for confounding features of MT (right), distributions of log₂(mRNA fold change) were compared between the subgroups (left, two-sided Kolmogorov–Smirnov test). Distribution of miRNA non-targets (‘No-site’, gray) was plotted for comparison. The mean values of confounding features are displayed and the error bars represent 95% confidence intervals.