Computational Analysis of RNA-binding Protein Target-site Selection and Function

Gene expression is extensively regulated by the binding of RNA-binding proteins (RBPs) to cis-regulatory elements encoded in mRNA. A robust literature has emerged regarding the stability and intracellular distribution of transcripts genome-wide. However, clear binding preferences have only been reported for a few RBPs, and these binding data collectively explain only a small portion of these post-transcriptional events. I developed the RNA Regulatory Element Analysis and Discovery (RNA-READ) pipeline, which takes as input positive (e.g., regulated transcripts) and negative gene lists (e.g., the co-expressed transcripts not affected by the same post-transcriptional event), and outputs the RNA cis-regulatory element that distinguishes the positive from the negative transcripts. First, RNA-READ tests for enrichment of previously reported RNA motifs, then it performs novel RNA motif discovery to identify the consensus RNA motif that best discriminates between the positive and negative transcripts. An important innovation of the RNA-READ pipeline is that it considers both sequence and structural constraints on binding of RBPs. I consider two binding classes: (1) the RBP binds to mRNAs with primary sequence-specificities; (2) the RBP binds to elements in mRNAs defined completely by structure (i.e., shape recognition). I have shown that computationally estimated target-site accessibility improves prediction of sequence-specific binding for various RBPs, with >22% average relative decrease in error versus using only sequence information. The predictive power is further increased with the introduction of structural-context constraints to the single-stranded target sites. Furthermore, I showed that computationally estimated intrinsic mRNA secondary structure is also helpful for determining RBP binding via shape recognition. I identified specific structural elements enriched in dsRBP Staufen targets versus non-targets. I applied the RNA-READ pipeline to the datasets that measured translation, stability and localization of transcripts in the Drosophila early embryo and identified numerous significant associations between the presence of motif matches and specific regulatory outcomes.Ph