Structure of pre-miR-31 reveals an active role in Dicer processing

Abstract

As an essential post-transcriptional regulator of gene expression, microRNA (miR) levels must be strictly maintained. The biogenesis of many, but not all, miRs is mediated by trans-acting protein partners through a variety of mechanisms, including remodeling of the RNA structure. miR-31 functions as an oncogene in numerous cancers and interestingly, its biogenesis is not known to be regulated by protein binding partners. Therefore, the intrinsic structural properties of pre-miR-31 can provide a mechanism by which its biogenesis is regulated. We determined the solution structure of the precursor element of miR-31 (pre-miR-31) to investigate the role of distinct structural elements in regulating Dicer processing. We found that the presence or absence of mismatches within the helical stem do not strongly influence Dicer processing of the pre-miR. However, both the apical loop size and structure at the Dicing site are key elements for discrimination by Dicer. Interestingly, our NMR-derived structure reveals the presence of a triplet of base pairs that link the Dicer cleavage site and the apical loop. Mutational analysis in this region suggests that the stability of the junction region strongly influence both Dicer binding and processing. Our results enrich our understanding of the active role that RNA structure plays in regulating Dicer processing which has direct implications for control of gene expression.

Figure 1.. Conflicting secondary structure models for pre-miR-31 apical loop.

a) Secondary structure derived from in vitro DMS-MapSeq where coloring denotes reactivity of given bases. Red=high reactivity, orange=medium reactivity, black=low reactivity, gray=no data available. b) Secondary structure derived from NMR characterization. Coloring is based on identification of A-U base pairs (see panel e). c) Portion of a 2D ¹H-¹H NOESY spectrum of an A^2rG^rU^r-labeled FL pre-miR-31. Adenosine cross-strand NOEs consistent with helical stacking in the junction region are indicated. d) Secondary structure of the apical loop region highlighting NOEs noted in c with red arrows. e) Best-selective long-range HNN-COSY spectrum identifying A-U base pairs within FL pre-miR-31. Black peaks are adenosine H2-N1 correlations, red peaks are adenosine H2-uracil N3 correlations. Vertical lines indicate the detection of A-U base pairs. Unpaired adenosines are denoted in green, A-U base pairs in the stem region are denoted in black, junction A-U base pairs are denoted in cyan and purple.

Figure 2.. Tertiary structure of pre-miR-31.

a) NMR-derived secondary structure of FL-pre-miR-31. Dicer cleavage sites are indicated with scissors. Gray nucleotides were included in structural studies but are not present in a Dicing-competent WT pre-miR-31. b) Ensemble of 10 lowest energy structures after RDC refinement superimposed over residues 1–13 and 59–71. c) Lowest energy structure of pre-miR-31 with a transparent surface rendering. d) Enlarged view of the dicing site, colored orange. e) Enlarged view of the C•A mismatch, colored pink. f) enlarged view of the G•A mismatch, colored teal. g) Enlarged view of the A•A mismatch, colored green.

Figure 3.. Structure at the dicing site serves as a control element for Dicer processing.

a) Secondary structures of constructs designed to minimize the internal loop at the dicing site. Mutations are indicated with red lettering. b) Dicer processing efficiency for Δ43 and Δ43/U44A mutants normalized to WT pre-miR-31 at 10 min. c) Secondary structures of constructs designed to expand the internal loop at the dicing site. Mutations are indicated with red lettering. d) Dicer processing efficiency for G45C and G45C/C46G mutants normalized to WT pre-miR-31 at 10 min. e) Processing assay gels of hDicer (20 nM) with WT and dicing site mutant pre-miR-31 RNAs (2 nM) at pH = 7.5.

Figure 4.. Apical loop size is optimized for efficient Dicer binding and processing.

a) Secondary structures of constructs designed to minimize the pre-miR-31 apical loop. Mutations are indicated with red lettering. b) Quantification of the binding affinity of pre-miR-31 RNAs with Dicer. Solid lines represent best fits to a one site specific binding equation. c) Histogram quantifying the Dicer processing efficiencies of pre-miR-31 RNAs at 10 min. d) Secondary structures of constructs designed to extend the pre-miR-31 apical loop. Insertions are indicated with red lettering. e) Quantification of the binding affinity of pre-miR-31 RNAs with Dicer. Solid lines represent best fits to a one site specific binding equation. f) Histogram quantifying the Dicer processing efficiencies of pre-miR-31 RNAs at 10 min. For all binding and processing assays, average and standard deviation from n=3 independent assays are presented. Individual replicates shown with black circles.

Figure 5.. The junction region is a regulatory element within pre-miR-31.

a) Secondary structures of constructs designed to perturb the stability of the pre-miR-31 junction region. Mutations are indicated with red lettering. b) Histogram quantifying the Dicer processing efficiencies of pre-miR-31 RNAs at 10 min. c) Quantification of the binding affinity of pre-miR-31 RNAs with Dicer. Solid lines represent best fits to a one site specific binding equation. d) Inverse correlation between calculated binding affinity and measured thermal stability (melting temperature, T_m) for WT and junction region mutations. e) Correlation between Dicer binding affinity and Dicer processing efficiency for junction region mutations. For all binding and processing assays, average and standard deviation from n=3 independent assays are presented. Individual replicates shown with black circles.

Figure 6.. Secondary structure elements and their contribution to the regulation of pre-miR-31 processing.

The presence or absence of mismatches within the stem of pre-miR-31 had no impact on Dicer processing. More highly stabilized Dicing sites were processed as efficiently as the WT sequence, but pre-miRs with larger internal loops were not processed efficiently. Similarly, pre-miRs with either too small or too large apical loops were processed less efficiently than WT pre-miR-31. Interestingly, the WT pre-miR-31 has an inherently encoded structural switch at the junction region. Pre-miR-31 appears to sample both an open loop structure, which favors binding, and a closed loop structure, which promotes processing. This allows WT pre-miR-31 to maximize both binding with and processing by Dicer.