Novel Bioinformatics Approaches for MicroRNA Detection and Target Prediction.

Abstract

MicroRNAs (miRNAs) are regulators of gene expression at the post-transcriptional level. Scientists have not been able to fully unlock their therapeutic potential because their functions and mechanisms of action have not been fully characterized. In this thesis we address shortcomings and provide solutions for detecting miRNAs in a high-throughput manner and for predicting miRNA targets - areas key to understanding miRNA function. Profiling expression of miRNAs using microarrays has its limitations owing to diverse melting temperatures and high sequence similarities, which affects sensitivity and specificity. A simple yet effective strategy that we employ involves base changes to probes complementary to mature miRNAs. Using nearest-neighbour thermodynamic principles we determine the best probes for all mature miRNAs that serve to eliminate cross-hybridization and create a uniform melting temperature profile. We present a set of probes that are designed for the human let-7 family and demonstrate their power to resolve these similar sequences in a microarray experiment using both spiked-in and true biological samples. The second problem that is tackled in this thesis involves improving miRNA target prediction, a key to understanding miRNA function in various biological processes. We use a combination of thermodynamic and sequence-based searches to identify endogenous sites on 5′-UTRs. There are two aspects that make our approach unique compared to other target prediction methodologies. First, we not only consider seed-matches on the 3′-UTR but also 5′-UTR matches with 3′-ends of miRNAs. Second, we show that non-conserved sites on the 5′-UTR can possibly contribute to species-specific targeting. We verify our claims through in vitro experiments using two predicted miRNA-target pairs: hsa-miR-34a and its target AXIN2, and cel-lin-4 and its target lin28. Extending results from the target prediction study, we show that upstream AUGs (uAUGs), which are known to post-transcriptionally regulate gene expression, are probable binding sites for miRNAs. We show that the cell- or tissue-specific repression of genes that harbour uAUGs can be explained by the expression of targeting miRNAs in those cells. The approaches suggested here will help further our understanding of how these tiny RNAs regulate gene expression.