Fig. 7: Affinity–specificity landscapes for RNA binding proteins (RBPs). | Nature Communications

Fig. 7: Affinity–specificity landscapes for RNA binding proteins (RBPs).

From: A comprehensive thermodynamic model for RNA binding by the Saccharomyces cerevisiae Pumilio protein PUF4

Fig. 7

a The landscape of RNA affinity for two RBPs, where the color coding indicates the specificity (bar under plot). bd Mock comparison of affinities for two RBPs binding, RBP #1 (orange) and RBP #2 (green), to all RNAs in a transcriptome. Three plots are shown that represent a continuum of possibilities, ranging from fully distinct targets (b) to identical specificities (d) with c showing an intermediate case with specificity overlap. Plots in ad are in terms of ΔΔG, which provides affinities relative to a reference RNA. e, f PUF4/PUF3 affinity–specificity landscapes for the highest affinity binding site for the respective RBP of all S. cerevisiae 3′UTRs in the Saccharomyces genome database (e)56 and for PUF3 and PUF4 3′UTR targets identified from previous in vivo experiments (f)9. The mathematical binding models from this and prior work34 were used to calculate the highest affinity binding site within the 3′UTR of each transcript. To obtain estimates for PUF3 using the highly homologous PUM2 data, an additional term was added to account for PUF3’s favorable interaction with C at position −2 (see Supplementary Table 2 and “Methods”)51. Values are shown as ∆∆G relative to the respective consensus sequences. Points are given for each binding site (N = 5089) and shaded based on the density of points with the corresponding value. In f, PUF3 targets are depicted by an orange triangle; targets that bind PUF3 and PUF4 with a yellow square, and PUF4 targets with a green circle. N is the number of known targets for each. g Thermodynamic affinity predictions compared to PUF4 RNA-IP enrichment from Saccharomyces cerevisiae9. The enrichment (log2 ratio) for PUF4 targets was reported in ref. 9 for significantly enriched targets (from four independent experiments). The enrichment was estimated as 0 for sites in the yeast genome with predicted affinities <4 kcal/mol that were not identified in the target list. The mean of the enrichment was taken for bins of 0.5 kcal/mol with error bars indicate 95% CIs of the mean.

Back to article page