An efficient strategy for the determination of the three-dimensional architecture of ribonucleoprotein complexes by the combination of a few easily accessible NMR and biochemical data: intermolecular recognition in a U4 spliceosomal complex.

Ribonucleoprotein (RNP) complexes are involved in several cellular processes, including RNA processing, transcription and translation. RNP structures are often dynamic in nature, undergoing significant remodeling during the course of their function. Visualization of the three-dimensional arrangement of single components in the complex and characterization of the intermolecular interactions are essential for understanding the mechanisms of operation. Crystallization either is not always achievable for these highly dynamic RNP particles or requires trimming the complex to a stable, well-structured core that lacks the flexible, regulatory domains. Alternative techniques that can provide structural information for complexes in solution under native conditions, where they retain their natural dynamic properties, are needed. In this study, we explored the possibility of using a combination of NMR, biochemical data and molecular modeling to generate an accurate high-resolution model of RNP complexes. We applied this strategy to the ternary hPrp31 (human Prp31)-15.5K-U4 5'-SL (stem-loop) spliceosomal complex, which, due to its large size and instability and because of the difficulty in obtaining isotopically labeled hPrp31, is not amenable to complete structure determination by NMR. We designed a protocol where the protein-protein interaction surface is defined for 15.5K by NMR data, while the relative orientations of the U4 RNA and the hPrp31 protein are described by mutational and cross-linking data. Using these data in a restrained ensemble docking protocol, we obtained a model for the ternary complex that reveals a novel rationale for the hierarchical assembly of the complex. Comparison of the docking model with the crystal structure recently obtained for a trimmed version of the complex reveals the high accuracy of the docking model, even down to an atomic level. This work shows that the architecture of large RNP complexes is within reach by NMR investigation in solution even for those cases where a traditional structural determination cannot be performed.