We have established the structures of 10 human microRNA (miRNA) precursors using biochemical methods. Eight of these structures turned out to be different from those that were computer-predicted. The differences localized in the terminal loop region and at the opposite side of the precursor hairpin stem. We have analyzed the features of these structures from the perspectives of miRNA biogenesis and active strand selection. We demonstrated the different thermodynamic stability profiles for pre-miRNA hairpins harboring miRNAs at their 5-and 3-sides and discussed their functional implications. Our results showed that miRNA prediction based on predicted precursor structures may give ambiguous results, and the success rate is significantly higher for the experimentally determined structures. On the other hand, the differences between the predicted and experimentally determined structures did not affect the stability of termini produced through "conceptual dicing." This result confirms the value of thermodynamic analysis based on mfold as a predictor of strand section by RNAi-induced silencing complex (RISC).
MicroRNAs (miRNAs)1 are a large family of short 20 -25-nt single-stranded noncoding RNAs recently identified in many eukaryotes from nematode to human (1-4). The best known founding members of this family are lin-4 (5) and let-7 (6) of Caenorhabditis elegans that trigger the translational inhibition of their target mRNAs by partial base-pairing within the 3Ј-untranslated region. According to the results of recent surveys performed using the experimental (3, 7) and bioinformatic approaches (8, 9), the miRNA genes may contribute ϳ1% to the total gene content of the investigated organisms making this regulatory mechanism more common than previously thought.Primary transcripts of the miRNA genes, pri-microRNAs, are processed in the nucleus to pre-microRNAs by the ribonuclease Drosha (10) and exported from the nucleus by Exportin-5 (11). The 60 -90-nt miRNA precursors form the stem and loop structures, and the cytoplasmic ribonuclease class III enzyme Dicer (12-14) excises miRNAs from the pre-miRNA hairpin stem. Dicer, either alone or with the help of Drosha, cleaves both strands of the precursor to form a double-stranded microRNA/microRNA* duplex (10, 15, 16) but only this strand accumulates which enters the RNAi-induced silencing complex (RISC) (17, 18). Based on the mechanism of the precursor hairpin processing and the similarities between the miRNA and RNAi pathways, active siRNA duplexes could be predicted with a high success rate (17, 18). It appears that designing siRNAs so that their properties resemble those of putative double-stranded miRNA intermediates, produces highly effective siRNAs. The strand whose 5Ј-end is less tightly paired to its complement is selected to enter into the RNAi-induced silencing complex, whereas the opposite strand is degraded. Thus, the structure of miRNA precursors and their short lived processing intermediates turned out to be the key for successful siRNA design (17,18). In this ...