Structure-based prediction and characterization of photo-crosslinking in native protein-RNA complexes

Abstract

UV crosslinking of protein-RNA complexes in direct contacts has been widely used to map protein-RNA interactions and yet the underlying mechanisms of photo-crosslinking are poorly understood. This knowledge gap is due to the challenge to precisely map the crosslink sites in protein and RNA simultaneously in their native sequence and structural contexts. Here we developed a computational method named PxR3D-map to analyze protein-RNA interactions and photo-crosslinking by bridging crosslinked nucleotides and amino acids mapped by different assays with experimentally resolved 3D structures of protein-RNA complexes. PxR3D-map can reliably predict crosslink sites using structural information characterizing protein-RNA interaction interfaces. Amino acid-nucleotide crosslinking is strongly associated with and likely facilitated by several structural features. Importantly, we found that base stacking facilitating protein-RNA crosslinking can occur not only with aromatic residues, but also dipeptide bonds that involve glycine, and distinct mechanisms underlie crosslinking of RNA with different types of RNA-binding domains. Our work suggests protein-RNA photo-crosslinking in the cellular environment is highly selective, which can be utilized to better interpret data generated based on UV-crosslinking and guide further technology development to map and characterize protein-RNA complexes.