Computational Approaches for RNA Structure and Function Determination

Division Of Basic Sciences - Nci

Investigators

Linked publications, trials & patents

Abstract

In cell SHAPE prediction provides a new level of detail for determining RNA structure within cells. These results may vary from the more standard SHAPE techniques that do not take the cellular environment into account when producing potential structural predictions. We developed a method for the computational prediction of in cell SHAPE by training a neural network (which was optimized by hyper-paramaterization techniques) based on known in cell SHAPE measurements obtained from an E. coli database. Predictions, given a sequence, produce reasonably accurate results with a Pearson coefficient with experimental shape scores better than thermodynamic folding. As an example, we predicted the SHAPE scores around translation start sites in mRNAs. The method indicates that nucleotides immediately upstream of the translation start sites to be relatively unstructured. These results were found to be statistically significant, while in contrast, results based on thermodynamic folding were not. This is the first time that computational methods have been applied to the prediction of RNA structure within cells based on machine learning. ---In another project in collaboration with Mikhail Kashlev we determining motifs that during transcription are responsible for transcriptional termination. These motifs appear to go beyond the standard RNA hairpin that is normally involved in termination. The approach involves the use of MPGAFOLD, a massively parallel genetic algorithm that includes capabilities to predict RNA secondary structures that form during transcription, i.e. co-transcriptional folding as the RNA strand elongates. As it does local structures form. These structures in turn have the ability to form tertiary interactions which can influence the formation of termination motifs. These sequential secondary structure motifs are also modeled in 3D further verifying their potential formation and tertiary influence. A new paradigm for termination control may be indicated by these results.---Another project in collaboration with Stuart Le Grice, Kwaku Dayie, Jay Schneekloth, and Nathan Baird involves the development of a computational approach to determined binding sites and affinities of small molecules targeting various RNA structural motifs. An ultimate goal of this project is to aid in the screening of small molecules for their potential to be therapeutically beneficial in targeting viral RNAs or cancer causing genes. This problem is inherently more difficult than targeting protein pockets due to the flexibility of RNA and it relatively uniform binding surfaces. In the case of RNA, the small molecules will normally target comformationally complex motifs. One of the advantages of targeting RNA rather than DNA is that some proteins are untargetable. In addition, by targeting RNA you can inhibit or alter functionality at an earlier stage of the molecular life-cycle. The small molecules are initially derived from sets found by binding to experimental screening methods using small molecule microarrays. The computational pipeline as it currently stands is able to determine to a reasonable level of accuracy ligand poses as well as the conformation of the binding pockets. It also is able to discriminate between different levels of binding affinities for different ligands. The computational procedure relies on in-depth conformational sampling using molecular dynamics and docking programs to determine the best poses of the ligand and its target.The pipe-line has been applied to the epsilon region of the hepatitis delta virus and is being applied to the triple stranded PAN. We are able to get good agreement between computational, biophysical, NMR, and X-ray structure data respectively for these two significantly different sites. This methodology is opening the door to computational prediction of small molecule binding to RNA motifs for potential therapeutics purposes, a domain of research that has not been extensively explored.---In collaboration with Anne Simon, University of Maryland, we have been exploring the RNA motifs that are involved in alternative modes of translation in eukaryotic systems. Specifically we have been concentrating on those that do not contain 5' cap sites and lack a poly A tail, cap Independent translation elements (CITE, or PTE), which is not the normal mode of translation, but is a mechanism found in several RNA viruses. We have found elements, via computational 3D modeling and experimental verification such as site directed mutagenesis and SHAPE, that seem to be common for example, in Carmoviruses that stabilize structures beyond pseudoknot motifs that are conducive for translation factor binding and thus mimic 5' cap sites.