Changes in miRNA secondary structure can predict mutations associated with cancer and other diseases

MicroRNAs (miRNAs) are ubiquitous short RNAs regulating gene expression in many organisms, including humans. How the secondary structure (SS) of a mature miRNA affects its regulatory function remains an open question. Here we investigate this question through computational SS predictions of miRNA point mutants. We explore the mutational neighborhoods of miRNAs with association to human diseases, including cancer. We focus on possible SS changes independent of target-site complementarity, by leaving the seed region unchanged. We formulate metrics of the SS differences between such mutants and their wild types (WTs), and test whether these metrics predict disease association by comparing our results with the miRNASNP-v3 database. We find that disease-related mutants tend to have a higher probability of being fully unfolded than their WT; this and other SS-related measures are statistically significant at the database level. With the same approach, we identify a subset of individual miRNAs for which SS changes are most likely to predict disease-related mutations. These are hsa-miR-1269b, hsa-miR-4537, hsa-miR-4477b, hsa-miR-4641, and hsa-miR-6821-3p. In addition, we show that there are pairs of known miRNA WTs differing only by disease-related point mutations outside the seed region and exhibit very different SS. These pairs include hsa-miR-1269a--hsa-miR-1269b, and hsa-miR-3689a-3p--hsa-miR-3689b-3p.