LAUSR

Simple Summary MicroRNAs are ~23 nt, highly conserved non-coding RNA molecules involved in the regulation of target gene expression. Most of the microRNA-target prediction algorithms rely heavily on seed rules and evolutionary conservation. However, such strategies suffer from missing the non-canonical target sites. The aim of this study is to identify the general features of non-canonical targets and their interactions with microRNAs. We found that the bulge-targets were preferentially associated with the microRNAs containing CG dinucleotides in their seed region. This finding indicates that non-canonical targets could be rich due to high mutation frequency of CG within the target mRNAs. Multi-step validation, which included evolutionary, overexpression, correlation, and CLASH data analysis, supports the interactome between the microRNAs with CG dinucleotides in the seed region and their bugle targets. Thus, a major novelty of this work is the identification of a sequence motif, CG dinucleotides, in the seed region of microRNAs, is strongly correlated to bulge targeting patterns. Abstract MicroRNAs (miRNAs) are endogenous ~23 nt RNAs which regulate message RNA (mRNA) targets mainly through perfect pairing with their seed region (positions 2–7). Several instances of UTR sequence with an additional nucleotide that might form a bulge within the pairing region, can also be recognized by miRNA as their target (bugle-target). But the prevalence of such imperfect base pairings in human and their roles in the evolution are incompletely understood. We found that human miRNAs with the CG dinucleotides (CG dimer) in their seed region have a significant low mutation rate than their putative binding sites in mRNA targets. Interspecific comparation shows that these miRNAs had very few conservative targets with the perfect seed-pairing, while potentially having a subclass of bulge-targets. Compared with the canonical target (perfect seed-pairing), these bulge-targets had a lower negative correlation with the miRNA expression, and either were down-regulated in the miRNA overexpression experiment or up-regulated in the miRNA knock-down experiment. Our results show that the bulge-targets are widespread in the miRNAs with CG dinucleotide within their seed regions, which could in part explain the rare conserved targets of these miRNAs based on seed rule. Incorporating these bulge-targets, together with conservation information, could more accurately predict the entire targets of these miRNAs.