Folding landscape analysis of RNA molecules

RNA is the key enzymatic component in several essential cellular processes including splicing and translation. Beyond these ancient, fundamental roles, RNA has found important applications in modern biotechnology and medicine, ranging from small interfering RNAs and protein-binding RNA aptamers to RNAs with specific catalytic functions. Functional RNAs thus can serve as effective tools in health and technology. As a prerequisite for the efficient utilization of RNA as the biopolymer of choice, it is crucial to have detailed answers for key questions arising in any RNA design task: How can RNAs avoid kinetic traps along their folding pathways and finally fold into desired structures? How can RNAs switch between alternative structures in response to changes in the cellular environment? The common theme of all these issues refers to the structure of the molecule`s folding landscape. Indeed, RNA folding is a complicate kinetic process and therefore the purely static view provided by equilibrium ensembles commonly used in structure prediction is insufficient to provide detailed insights into the dynamic behaviors of RNAs. A sufficiently sophisticated analysis of the folding energy landscape, on the other hand, can provide the relevant information and hence becomes a subject of utmost importance in computational biology. Recently, single molecule fluorescence resonance energy transfer (smFRET) has been utilized to monitor the distance between two dye-labeled nucleotides in order to reveal details of RNA folding kinetics in real time. The smFRET technique is becoming a powerful tool for distinguishing discrete states (folded versus unfolded versus intermediates) of an RNA folding reaction. A major obstacle is, at present, there is no general and efficient way to interpret smFRET measurements in terms of explicit molecular structures. We propose that directly analyzing folding landscapes with suitable coarse-graining methods, will not only present a new perspectives to explore RNA folding kinetics, but, in combination with smFRET data, will enable us to model RNA folding kinetics with sufficient accuracy for most applications. In this proposal, we intend to first develop computational algorithms to approximate the folding landscapes of RNA molecules with a novel coarse-grained model we call B(asin)H(opping)-graph``. A BH-graph is a graph whose vertices are the local minima, representing the corresponding basins in the landscape, and edges are introduced between neighboring basins when direct hops between them are ``energetically favorable``. Based on this model, we will investigate both the geometric and the kinetic features of folding landscapes. In particular, we will focus on studying the kinetic influence of individual RNA structures. Finally, we will employ this new model as a platform for interpreting the process of RNA molecular transitions by converting smFRET signal trajectories into structural pathways in the RNA folding landscapes via combinatorial optimization methods.

Instead of only predicting the most stable structure for an RNA molecule, our project is focused on interpreting RNA folding as a complicate kinetic process. This is because the most stable structure provides only a static view of the molecule folding mechanism and rearrangements happened during the folding processes may lead to severe diseases include SARS, the common cold, influenza, hepatitis C, and measles. To gain a thorough understanding of the folding processes of RNA molecules, we need a sophisticated analysis of their folding landscapes since folding processes can be naturally modeled as Markov processes in their landscapes. To this end, we introduce a novel coarse-grained model of folding landscapes called the Basin Hopping Graph (BHG). BHGs can be approximated accurately and efficiently for RNA molecules. In practice, BHGs can be computed for large RNAs that are not accessible by existing algorithms. This BHG modeling method is particularly applied to study the folding paradigm of RNAs containing pseudoknots. Further, single-molecule Förster Resonance Energy Transfer (smFRET) is a powerful tool for understanding RNA molecules' essential structure-dynamics-function relationships. Using BHG model for eucliding the folding landscape and Ernwin for 3D structural predictions, we introduce a hidden Markov model approach termed FRETtranslator to predict RNA secondary structure transitions based on input smFRET experimental data.