Transposable element-like non-conserved miRNAs and their targets in rice: evolutionary and functional implications

MicroRNAs (miRNAs) are 20- to 24 nt riboregulators that play important roles in eukaryotic growth and development. Their evolution is poorly understood and there are no conserved miRNAs between plants and animals, yet transposon-like miRNAs have been described for both kingdoms. A recent report (Nature 464: 628 [2010]) established transposon-derived small RNAs (sRNAs) as causal for female gametophyte fate specification in the dicot Arabidopsis thaliana by an unknown mechanism. We hypothesize that transposon-like miRNAs may target protein-coding genes which function in a regulatory network analogous to small interfering RNAs (siRNAs) involved in leaf morphogenesis. Plant miRNAs act on messenger RNAs (mRNAs) by near-perfect base pairing, facilitating computational prediction of targets. We customized miRNA prediction program miRanda V2.0 and performed a comprehensive prediction of miRNA targets in genomes of Arabidopsis and rice Oryza sativa subsp. japonica. Among the predicted targets of rice miRNAs, more than 20% of hits were in the 5’ and 3’ untranslated (UTR) regions of protein coding genes, which was rare in Arabidopsis (< 7%) except the case of recently described miR5021 (25%) that has homology to a graminaceous monocot transposon TREP776 CACTA, "Caspar_453N11-1" (http://plantrepeats.plantbiology.msu.edu/search.html). The majority (~51%) of plant 3’-UTR targets are homologous to an abundant class of DNA miniature inverted repeat transposable element (MITEs), and this complementarity is restricted to the corresponding mature miRNAs, whereas this TE class makes up only ~ 7% and 9% of 5’ UTR- and coding region homologies, respectively. We found that rice UTRs predicted to be targets of miR1432 appear to be under purifying selection for the miRNA footprint based on maize, sorghum, and sugarcane target homologies, suggesting a functional significance for this element. We found extended homology between TE STOWAWAY2 and predicted 3’ UTR targets of miR1884b and miR808, with evidence of purifying selection for the mature miR1884b compared to flanking sequences. A miR1884 (24 nt) is expressed predominantly in rice roots, but not flowers. We propose a model of miRNA evolution in rice through the constraint of acquisition for siRNA binding sites mediated by the exaptation of DNA transposons. Subject to post-transcriptional gene silencing (PTGS) mechanisms, the translocation of MITEs into the 3’-UTRs of rice genes, which spawn or are targets of TEsiRNAs could put them under natural selection as targets of PTGS pathways. In this sense, the TE-associated siRNAs may represent intermediate molecular species between TEs and miRNAs.