Global signatures of protein binding on structured RNAs in Saccharomyces cerevisiae

Abstract

Protein binding is essential to the transport, decay and regulation of almost all RNA molecules. However, the structural preference of protein binding on RNAs and their cellular functions and dynamics upon changing environmental conditions are poorly understood. Here, we integrated various high-throughput data and introduced a computational framework to describe the global interactions between RNA binding proteins (RBPs) and structured RNAs in yeast at single-nucleotide resolution. We found that on average, in terms of percent total lengths, ~ 15% of mRNA untranslated regions (UTRs), ~ 37% of canonical non-coding RNAs (ncRNAs) and ~11% of long ncRNAs (lncRNAs) are bound by proteins. The RBP binding sites, in general, tend to occur at single-stranded loops, with evolutionarily conserved signatures, and often facilitate a specific RNA structure conformation in vivo. We found that four nucleotide modifications of tRNA are significantly associated with RBP binding. We also identified various structural motifs bound by RBPs in the UTRs of mRNAs, associated with localization, degradation and stress responses. Moreover, we identified >200 novel lncRNAs bound by RBPs, and about half of them contain conserved secondary structures. We present the first ensemble pattern of RBP binding sites in the structured non-coding regions of a eukaryotic genome, emphasizing their structural context and cellular functions.