MicroRNAs (miRNAs) are endogenous, small, non-coding RNAs that are involved in post-transcriptional gene silencing. In this study, an Arabidopsis miRNA gene, AtMIR171a, was characterized and used as a model to study the biogenesis and function of plant miRNAs.
AtMIR171a is a strongly conserved miRNA that is expressed at high levels in Arabidopsis. The transcription unit of the AtMIR171a gene is quite complex. It has at least three transcription start sites, Pri-I, -II and -III. It also contains alternative splice sites and alternative polyadenylation sites. Most transcripts of the AtMIR171a locus initiate at the transcription start site Pri-I located 74 bp upstream of the putative AtMIR171a-containing stem-loop. The complexity of the gene structure may reflect the low selection pressure exerted outside the stem-loop structure.
Mutations in three genes have been identified to affect miRNA accumulation in Arabidopsis. These genes are Dicer-like1 (DCL1), Hua Enhancer1 (HEN1) and Hyponastic Leaves1 (HYL1). DCL1 encodes an RNase III enzyme that cleaves miRNA precursors to generate mature miRNAs. HEN1 codes for a methyltransferase that prevents uridylation and degradation of small RNAs by methylating the ribose of the 3’-most nucleotides in the miRNA:miRNA* or siRNA:siRNA* duplex. HYL1 encodes a double-stranded RNA binding protein (dsRBP) which has two double-stranded RNA binding motifs (DSRM) in its N-terminal region. Three lines of evidence suggest that HYL1 and DCL1 act in the same step in miRNA biogenesis. First, the loss-of-function homozygotes of dcl1 and hyl1 exhibit similar development defects. Second, the accumulation of mature miRNA is reduced in both dcl1 and hyl1 mutant plants. Third, a previous study showed that a dsRBP R2D2 facilitates an RNase III protein DCR-2 in loading siRNA into the RNA-induced silencing complex (RISC) in Drosophila . This suggests that an RNase III enzyme interacts with a specific dsRBP in gene silencing. To test whether HYL1 acts with DCL1 in miRNA biogenesis, the amounts of AtMIR171a precursors were measured by northern blotting and RT-PCR.
The results showed that unlike hen1 homozygotes, which do not accumulate pri-AtMIR171a, both hyl1 and dcl1 mutants have much greater amounts of pri-AtMIR171a than wildtype plants. Therefore the paucity of mature AtMIR171a in hyl1 and dcl1 mutant plants is not attributable to either reduced miRNA stability or decreased transcription from the AtMIR171a gene, but is caused by a defect in the processing of pri-miRNA. In addition to the genetic evidence, the co-localization of HYL1 and DCL1 in the nucleoplasm and in perinucleolar bodies further supports the notion that these two proteins function together in miRNA biogenesis. The HYL1-DCL1-containing bodies are distinguishable from Cajal bodies, which contains the DCL3-dependent siRNA silencing machinery, because neither Arabidopsis Cajal body marker, AtCoilin or U2B”, co-localizes with HYL1. When HYL1 and SE, another protein involved in miRNA biogenesis, were co-expressed in Arabidopsis mesophyll protoplasts, the two proteins displayed partially overlapping distribution in the nucleus. Recent studies showed that SE, along with the two cap-binding proteins CBP20 and CBP80, are required for correct splicing of pre-mRNAs. CBP20 and CBP80 also control proper miRNA processing as mutations of the CBP20 and CBP80 genes lead to increased amounts of pri-miRNAs and decreased mature miRNA levels. Therefore, it appears likely that SE acts as a bridge to assemble pri-miRNAs with DCL1 and HYL1. It remains unknown whether the pri-miRNAs are transported to the DCL1-HYL1 bodies for cleavage or the DCL1 and HYL1 proteins are recruited to the sites of transcription to process pri-miRNA.
AtMIR171a has been shown to direct the cleavage of three SCL6-like genes, SCL6-II, SCL6-III and SCL6-IV. To determine the function of the SCL6-like genes, an AtMIR171a fragment was overexpressed in Col-0 background to simultaneously knockdown the SCL6-like transcripts. The silencing of the three SCL6-like genes causes many developmental defects such as abnormal inflorescence and flower development, abnormal phyllotaxy, reduced shoot branching from both cauline rosette leaf axils, epinastic dark green leaves, and reduced primary root length. There is a positive correlation between the AtMIR171a expression level and the severity of phenotypes. The pleiotropic phenotypes observed in the transgenic plants overexpressing the AtMIR171a fragment (171a-OX) may partially be caused by abnormal expression of WUS, as the WUS transcript level increases in 171a-OX seedlings.
To assess the biological importance of the AtMIR171-mediated post-transcriptional silencing of the SCL6-like transcripts, wildtype or AtMIR171-resistant SCL6-like::GUS genes were expressed from gene native promoters in Col-0. GUS staining assays detected all three wildtype or AtMIR171-resistant SCL6-like::GUS proteins in root tips. In the aerial organs, the expressions of the wildtype SCL6-like::GUS proteins were reduced compared to those of their AtMIR171a-resistant counterparts, indicating that AtMIR171 restricts the expression of the SCL6-like genes in shoots but not in roots.
A previous study has shown that the long, near-perfectly base-paired primary transcripts of several recently evolved miRNA genes are processed by DCL4, while the canonical miRNA precursors are cleaved by DCL1. These observations suggest that the canonical miRNA precursors have acquired the appropriate secondary structure that can be preferentially processed by the DCL1-dependent miRNA-biogenesis machinery. Canonical miRNA precursors have imperfectly base-paired stem-loop structures. To determine how the secondary structure of miRNA precursor determines the processing of miRNA, a series of mutations were introduced to alter the secondary structure of two pri-miRNAs, pri-AtMIR171a and pri-AtMIR167a. The mutated and wild-type pri-miRNAs were over-expressed in Arabidopsis and the phenotypes of the transgenic plants were analyzed. Over-expression of the wild-type pri-miRNA resulted in moderate to severe phenotypic effects, such as reduced shoot branching caused by overexpression of pri-AtMIR171a, or reduced silique length caused by over-expression of pri-AtMIR167a. When the base-pairing at the bottom of the miRNA-containing stem-loop was fully abolished, the transgenic plants were similar to those transformed with the vector alone. When the bulges and mismatches at the bottom of the miRNA-containing stem-loop were replaced by perfectly matched base pairs, the transgenic plants exhibited less severe phenotypes compared to those over-expressing wild-type pri-miRNAs. In contrast, removal of the bulges near the terminal loop in the miRNA-containing stem-loop did not significantly reduce the severity of the phenotypes observed upon over-expression of the wild-type pri-miRNA. Consistent with the phenotypic effects, lower levels of mature miRNA were observed in plants expressing primiRNA structural variants causing less severe phenotypes. Overall, these observations suggest that a stem with bulges and mismatches at the bottom of the miRNA-embedded hairpin is the most important structural feature for optimal miRNA processing.