Robin Friedman, Massachusetts Institute of Technology
MicroRNAs (miRNAs) are small endogenous RNAs that bind to short “seed matches” in the 3'UTRs of mRNAs and cause transcript destabilization and/or translational repression of their targets. MicroRNAs repress many target messages but typically repress the protein output of each by only a small amount. Thus, deciphering miRNA regulatory networks requires computational approaches in addition to experimental ones. We have developed new comparative genomics techniques to find and analyze the conservation of human miRNA seed matches. We improve on previous methods by more completely controlling for background conservation by accounting for mutational biases, dinucleotide conservation rates, and UTR conservation rates. Additionally, we achieve vast increases in sensitivity by more efficiently incorporating new genomes and different seed-match types. These improvements allow statistically powerful analysis of the conservation of individual miRNA seed matches, which is perhaps the most important concern for experimentalists trying to choose targets most promising for intensive follow-up. Our analysis also provides important new insights regarding the scope of conserved targeting. When considering human 3'UTRs, over 45,000 seed matches are conserved above background levels, indicating that more than half of human protein-coding genes have been under selective pressure to maintain pairing to miRNAs. Several seed-match types contribute similar numbers of targets, indicating that purifying selection acts on weaker but more common target sites roughly as often as stronger but rarer sites. We find little evidence for the enrichment of conservation for bulged or mismatched seed sites. Supplemental pairing to the 3' end of miRNAs has small but measurable excess conservation when it accompanies canonical seed matches, but less when accompanying imperfect matches. Additionally, we find that mammalian-specific miRNAs have far fewer conserved targets than ancient ones, even when considering only newly evolved mammalian targets.
Abstract Author(s): Robin C. Friedman, Kyle Kai-How Farh, Christopher B. Burge, David P. Bartel