The RNA-Protein Interactions in an Unstructured Context

Nucleic acids and proteins are two of the most fundamental types of biologically important molecules and without them, life as we know it would simply not be possible. What is special about nucleic acids and proteins is that they constantly interact with each other through weak, mostly temporary binding, and affect each others function. However, the fundamental principles guiding such binding are still not understood. This, in particular, concerns a very common class of such interactions, that between ribonucleic acids (RNAs) and unstructured proteins. A part of the reason for this is that unstructured proteins do not have a permanent, fixed structures and are very difficult to study using the classical techniques of structural biology. Our proposal aims to address this challenge and in parallel explore the basic principles of binding between RNA and protein molecules in the unstructured context as well as develop a computational method that could be used to predict which RNAs and proteins are expected to bind to each other, where exactly and how strongly. The central premise behind our approach is that the basic interaction propensities between nucleic acids and proteins should be related to the intrinsic interaction propensities between their fundamental building blocks, nucleobases and amino acids, respectively. Importantly, in our recent work, we have derived the latter propensities using several orthogonal methods, creating a rigorous, quantitative foundation for the present work. Moreover, in a preliminary analysis, we have shown that the proposed research paradigm indeed can provide novel insight in several well-studied experimental systems. In addition to examining the binding potential between all RNAs and all proteins in different organisms, here we will specifically focus on analyzing the binding in the context of the ribosome as well as between a physiologically and biomedically important RNA molecule named Xist and all of the proteins known to bind to it. While the ribosome is an essential cellular machine in charge of protein synthesis, Xist participates in controlling the level of expression from female X chromosomes. Understanding the basis of their function could lead to major advances in treating diverse diseases. Overall, our project aims to uncover novel principles behind the cellular organization and interactions in general, and provide a deeper understanding of several specific systems in particular.

The relationship between ribonucleic acids (RNAs) and proteins is one of the key hallmarks of life. However, the physicochemical principles guiding the interaction between these important molecules are still not fully understood, especially in the unstructured state. Recently, it has been proposed that proteins prefer to interact with the very own messenger RNAs (mRNA) that code for them, especially if unstructured, reflecting the influences behind the origin of the genetic code. Moreover, it was proposed that proteins interact not only with their own mRNAs, but also other RNAs that are compositionally related to them. The present project was aimed at establishing a computational foundation for assessing the potential of such interactions, and testing whether and in what contexts they indeed occur by comparing against experiment. First, we have developed several approaches for predicting the potential of RNAs and proteins to interact in the unstructured state using intrinsic nucleobase/amino-acid interaction preferences and have applied them to different biological systems. Moreover, we have released a public webserver for the analysis, comparison and visualization of physicochemical profiles of biopolymers, allowing one to assess their ability to interact. As a central achievement, we have used these tools and advanced statistical analysis to show that both of the above theoretical predictions are strongly supported by experiment. In particular, we demonstrated that ~67% of studied proteins directly interact with own mRNAs with a strong preference for coding sequence. Furthermore, we showed that the interaction specificity between an unstructured domain of a key enzyme RNA polymerase II with mRNAs is partly related to the genetic code. Similarly, we demonstrated that some ribosomal RNAs are compositionally related to the mRNAs of the ribosomal proteins they interact with. Also, we showed that the locations in viral RNA genomes where they interact with viral capsid proteins can be well predicted by considering the intrinsic potential of the partners to interact in the unstructured state. Finally, by comparing with experimental eCLIP data, we showed that the RNA partners of different RNA-binding proteins are compositionally related to the mRNAs of those proteins. Overall, our project has uncovered novel principles behind the large-scale trends concerning RNA-proteins interactions in general, and has enabled a more thorough understanding of several specific systems in particular. Considering the fundamental importance of RNA-protein interactions in biology, we expect that our results will have impact in both fundamental research and practical applications, especially in the biomedical context.