Fig 1.
RBPs analyzed are involved in post-transcriptional regulation and bind mainly at 3’UTR edges.
a) Scatter plot summarizing the most significant GO-terms associated with the RBPs analyzed. Bubble sizes indicate the frequency of the GO-term in the GO database. The color of the bubbles represents how significant the term is in the set of RBPs analyzed. b) Distribution of significant clusters (FDR < 0.01) across gene regions. The height of the bars represents the amount of significant clusters in 3’UTRs (blue), coding region (green) and 5’UTR (orange). c) Heatmap showing the distribution of significant clusters across standardized 3’UTRs. The colors range from red (higher frequency) to blue (lower frequency).
Fig 2.
RBP enrichment on 3’UTRs is significantly correlated for most RBP pairs.
a) Schematic representation of the z-score calculation for each RBP pair (top). Pearson correlation coefficients of cluster enrichments are calculated for each pair of proteins (purple line) in a distance from -200 to 200 nt and compared against the positional correlations obtained from shuffled clusters in the same region (dashed green lines show the top and bottom 5% random distributions). The random clusters are then used to calculate the z-score of the Pearson correlation coefficient at each position. The colors in the main panel represent the z-scores of the correlations and range from blue (lower z-score) to red (higher z-score). All against all positional correlations calculated in this way are summarized in the heatmap (bottom). For each RBP in a column we show the significance of the positional binding of each of the RBPs in the rows around its binding sites. b) Zoom on the positional correlations calculated around AGO2 binding sites. The Pearson correlation coefficients of cluster enrichments (left) and their z-scores (right) are shown. The strongest positional correlations are observed around AGO2 binding sites. On the right panel, the z-scores of the correlations are shown, row-normalized.
Fig 3.
RBP binding sites are enriched on target sites of highly expressed miRNAs.
a) Characteristic CLIP enrichment profiles for AGO2, HuR, EWSR1 and WDR33 RBPs around miRNA target sites that are highly (red), moderately (green) and lowly (blue) expressed. The grey and the black lines show the maximum and the minimum enrichment values for the 90% confidence intervals around random miRNA target sites. b) Heatmap showing the distribution of CLIP enrichment values around hisites. The colors range from red (high enrichment) to blue (low enrichment). c) Scatter plot summarizing the correlation values between RBP enrichment on miRNA target sites and miRNA expression (y-axis) and their p-values (x-axis). d) Barplot summarizing the fraction of RBP clusters on hisites (bars) and the fraction of hisites overlapped by RBPs (points). Colored bars (blue) and points (yellow) highlight the cases in which the fraction of RBPs or miRNA target sites is higher than expected by chance (empirical p-value < 0.01 using 100 random miRNA target site sets).
Fig 4.
PolyA complex RBPs are enriched on distal hisites.
CLIP enrichment of RBPs from the polyA complex around proximal (blue), medial (green) and distal (yellow) hisites. All the RBPs show a higher enrichment on distal hisites compared to proximal and medial hisites. These binding profiles can be grouped in two categories. In the first group, which includes CF-Im25, CF-Im59, CF-Im68, CPSF-160 and Fip1, the profiles are similar to that of AGO2. In the second set, which includes CstF-64 and WDR33, the enrichment changes from low to high across hisites.
Fig 5.
Hotspot sequences are highly conserved, accessible and contain AU-rich elements.
a) Relation between accessibility (top), sum of minor allele frequencies (middle), and conservation (bottom) with the number of RBPs in a window. The colored lines show the behavior of windows that overlap expressed miRNAs in HEK293 (light green), non-expressed miRNAs in HEK293 (blue) or that do not overlap miRNA target sites (dark green). The grey dashed lines mark the 2.5% and 97.5% percentiles of the distribution obtained by generating 100 random miRNA datasets. The presence of miRNA target sites in the windows does not affect the correlation observed between accessibility and sum of minor allele frequencies. In contrast, windows that overlap miRNA target sites (both expressed and non-expressed) show higher conservation regardless of the amount of RBPs that bind in the window. b) Cumulative distribution function showing the fraction of windows (y-axis) bound by a number of RBPs (x-axis) for all windows (blue), windows containing miRNA target sites (dashed green line) and random miRNA target sites (grey lines). c) Scatter plot showing the correlation between the ranks of the 7-mers identified using cWords [40] ordering the windows according to mean phyloP scores (y-axis) or the number of RBPs in a window (x-axis). Highlighted are the words that are in the top-20 in both analyses. The identified ARE UAUUUAU is highlighted in red. d) Scatter plot showing the correlation between the z-scores of the 7-mers identified ordering the windows according to the number of RBPs in all windows (y-axis) or only in windows containing miRNA target sites (x-axis). The words with the highest z-scores in both datasets (cut offs 17.5 and 8 for all and miRNA-containing windows respectively) are highlighted. In red is highlighted the ARE UAUUUAU.
Fig 6.
RBPs compete with each other and with miRNAs for binding on hotspots.
Cumulative fraction plot showing the effect of having 0, 1 or 2 or more a) hisites b) HuR binding sites overlapping RBP hotspots. As an additional control genes lacking a) hisites and b) HuR binding sites are shown. The x-axis shows the distribution of log2FC upon a) AGO2 or b) HuR KD. The y-axis shows the cumulative fraction of transcripts. *, ** and *** denote a p-value < 0.05, 0.01 and 0.001 respectively. The sets containing 1 or more sites overlapping hotspots are compared to set without sites overlapping hotspots. The set without sites overlapping hotspots is compared to the control set.