Impact of RNA Sequence and Structural Context on Protein-RNA Interactions

Regulation of RNA processing is principally driven by RNA:protein interactions, yet how RNA binding proteins (RBPs) select their targets among the complex cellular pool of RNAs remains incompletely understood. We developed a high-throughput version of RNA Bind-n-Seq, an in vitro method capable of determining the RNA sequence, RNA structural and sequence context preferences of RBPs. In this assay, randomized RNA regions (typically of length 20 or 40) are synthesized and incubated with recombinant RBPs. RNA:protein complexes are isolated and subjected to high-throughput sequencing yielding ~10-20 million unique reads preferentially bound to protein. We carried out RBNS on a diverse set of 78 human RBPs containing several RBD types with varying amino acid similarity between domains, most with previously uncharacterized specificities. Binding assays for each of these RBDs were carried out at five protein concentrations, totaling more than 400 binding assays and yielding over 6 billion protein-bound RNA reads. A comprehensive analysis of these datasets revealed a high degree of overlapping sequence specificities across RBPs and a strong preference for low complexity sequence motifs irrespective of RBD type. Consistent with previous studies, we observed that most RBPs prefer unstructured RNA substrates. However, we also identified factors including RBM22 and ZNF326 that preferentially bound structured motifs. Another lab subset of RBPs preferentially interact with pairs of submotifs spaced apart by one to ten bases; spaced motif interactions were more prevalent for multi-domain proteins. These sequence, structure and bi-partite motif specificities can often be observed in enhanced crosslinking and immunoprecipitation (eCLIP) data and are predictive of regulation in cells. This work establishes binding preferences for a set of human RNA binding proteins that extends beyond sequence motifs, highlighting complex mechanisms of RNA:RBP interactions