Natural selection on human microRNA binding sites inferred from SNP data.

A fundamental problem in biology is understanding how natural selection has shaped the evolution of gene regulation. Here we use SNP genotype data and techniques from population genetics to study an entire layer of short, cis-regulatory sites in the human genome. MicroRNAs (miRNAs) are a class of small noncoding RNAs that post-transcriptionally repress mRNA through cis-regulatory sites in 3' UTRs. We show that negative selection in humans is stronger on computationally predicted conserved miRNA binding sites than on other conserved sequence motifs in 3' UTRs, thus providing independent support for the target prediction model and explicitly demonstrating the contribution of miRNAs to darwinian fitness. Our techniques extend to nonconserved miRNA binding sites, and we estimate that 30%-50% of these are functional when the mRNA and miRNA are endogenously coexpressed. As we show that polymorphisms in predicted miRNA binding sites are likely to be deleterious, they are candidates for causal variants of human disease. We believe that our approach can be extended to studying other classes of cis-regulatory sites.