Regulatory Targets of MicroRNAs

The discovery of hundreds of miRNA genes immediately raised the question of what all these tiny RNAs are doing. To address this question, we have developed methods of predicting miRNA targets without bringing in too many false-positive predictions. In plants, the miRNAs have extensive pairing to their targets, and the evolutionarily conserved targets are mostly genes that play important roles during development. In animals, the miRNAs usually recognize shorter sites (typically 7 or 8 nt in length), which match a short region of the miRNA containing the ‘seed’ sequence. Our mammalian predictions, obtained in collaboration with Christopher Burge, can be viewed at TargetScan.org.

Animal miRNAs have a great abundance and diversity of targets, with more than one-third of human genes under selective pressure to maintain pairing to miRNAs. When considering nonconserved targeting, the fraction of human genes regulated by miRNAs grows even higher. Experiments using reporter assays and mRNA expression arrays provide additional evidence that miRNAs have a widespread influence on both the expression and evolution of mammalian protein-coding genes. For example, mRNAs preferentially expressed in the same tissue as a highly expressed miRNA are strongly depleted in 7mer matches to that miRNA, presumably because these messages have important roles in that tissue, and during the course of evolution they have avoided acquiring sites to co-expressed miRNAs that would compromise their function. This selective avoidance of 7mer matches to miRNAs provides compelling evidence that 7mer sites are often sufficient for repression in animals.

Although a 7mer site matching a miRNA is often sufficient for mediating some repression, it is not always sufficient, indicating that other characteristics help specify targeting. Using both computational and experimental approaches, we uncovered five general features of site context that boost site efficacy. Combining these determinants, we constructed a model of target recognition that successfully predicts site performance, thereby providing an important resource for choosing which of the many miRNA-target relationships are most promising for experimental follow-up. Because our approach accurately distinguishes effective from ineffective sites without recourse to evolutionary conservation, it also identifies effective nonconserved sites and siRNA off-targets.