Dr. Chris Burge: The impact of RNA-binding proteins on human genomic variation

Live Blogger: Henry Ertl

Editor: Liz Tidwell

This piece was written live during the 6th annual RNA Symposium: Towards our Future of RNA Therapeutics, hosted by the University of Michigan’s Center for RNA Biomedicine. Follow MiSciWriter’s coverage of this event on Twitter with the hashtag #umichrna. 

The genetic code for amino acids was cracked in the mid-20^th century. Since then, biologists have had much more difficulty deciphering the code of gene regulation–when, where, and how much a gene is expressed. This problem is made difficult in part due to the relative complexity of gene regulation, which is primarily carried out by both protein-DNA and protein-RNA interactions. Chris Barge’s lab at MIT works on the protein-RNA part of this problem by applying experimental and computational approaches to ask: what are the genetic determinants and consequences of RNA binding protein (RBP) binding to RNAs? 

The Barge lab is part of a large collaboration – ENCODE RBP – which aims to systematically catalog RBPs and their function. In recent years, these efforts have revealed over a thousand RBP-encoding genes that bind to many different mRNAs to aid in processes such as splicing, polyadenylation, and stability. These conclusions underscore the importance of RBP-RNA interactions in the cell, leading the Barge lab to dissect the mechanisms of RBP-RNA interactions more finely in their own research program. 

From the ENCODE RBP project, the Barge lab had already collected empirical data to determine the sequences preferably bound by 79 diverse human RBPs. They used the RNA Bind-N-Seq approach in which (1) a given RBP is mixed with a random set of RNA sequences, (2) the RBP-RNA complexes are bound to beads and purified, and (3) the bound-RNAs are deeply sequenced. These data showed that, surprisingly, RBP sequence specificity is not that diverse, rather it is further encoded in structural features (e.g., loops, stems, bulged stems) and complex binding modes (e.g., bipartite motifs). 

The Barge lab then used the empirical RBP binding sequence data to build a biophysically based model to predict RBP affinity spectra while incorporating information on multiple motifs, RBP concentration, binding site saturation, and secondary structure. The predicted RBP binding events were highly correlated with empirical measurements from eCLIP data, confirming the accuracy of their model predictions. This model then allowed them to experiment in silico by testing model performance with and without certain biophysical parameters. These experiments showed that binding saturation and multiple local motifs are the most important factors for model performance, further supporting the finding that non-sequence features are major determinants of RBP–RNA binding specificity. 

Once molecular interactions have been empirically characterized and/or confidently predicted – as with RBP–RNA interactions here – the next question is whether these molecular events are functionally important. To answer this question, the Barge lab used human population genomics resources to measure signatures of negative selection in RBP-RNA binding regions relative to other functional regions in the genome. If a certain sequence determines RBP binding, which is in turn important for organismal function and survival, we expect negative selection to produce a population-based pattern in which very few individuals are genetically variable at this sequence. Shockingly, the Barge lab found that variation in high-affinity, RBP-bound sequences exhibits similar – in some cases even stronger – signatures of negative selection than missense variation in coding regions. These results not only underscore the organismal importance of RBP-RNA interactions but also put them on par with other protein functions throughout the organism. 

Chris Barge received both his BS and PhD in Biology and Computational Biology, respectively, from Stanford University. In 2002, he joined the Department of Biology at MIT, and has since been awarded the Overton Prize for Computational Biology in 2001 and the Schering-Plough Research Institute Award in 2007. He is currently the Whitehead Career Development Associate Professor of Biology and Biological Engineering at MIT, where his lab uses a combination of computational and experimental approaches to ultimately understand the code for RNA splicing.