Heme enables proper positioning of Drosha and DGCR8 on primary microRNAs

Abstract

MicroRNAs regulate the expression of many proteins and require specific maturation steps. Primary microRNA transcripts (pri-miRs) are cleaved by Microprocessor, a complex containing the RNase Drosha and its partner protein, DGCR8. Although DGCR8 is known to bind heme, the molecular role of heme in pri-miR processing is unknown. Here we show that heme is critical for Microprocessor to process pri-miRs with high fidelity. Furthermore, the degree of inherent heme dependence varies for different pri-miRs. Heme-dependent pri-miRs fail to properly recruit Drosha, but heme-bound DGCR8 can correct erroneous binding events. Rather than changing the oligomerization state, heme induces a conformational change in DGCR8. Finally, we demonstrate that heme activates DGCR8 to recognize pri-miRs by specifically binding the terminal loop near the 3' single-stranded segment.

Fig. 1

Heme binding reverses the orientation of Microprocessor on pri-miRs. a Domain organization and construct diagrams for Drosha and DGCR8. Recombinant protein complexes were purified after coexpressing the highlighted regions (magenta). Coexpression of Drosha with DGCR8^Heme, DGCR8^Apo, and DGCR8^C352S yields MP^Heme, MP^Apo, and MP^C352S, respectively. CED central domain, RIIID ribonuclease III domain, dsRBD double-stranded RNA-binding domain, HBR heme-binding region, CTT C-terminal tail. b In vitro pri-miR processing assay using 5′ end-radiolabeled pri-miR-143. Reactions involving MP^Heme (130 nM), MP^Apo (130 nM) and MP^C352S (260 nM) are shown. Correct product is shown with a black asterisk, incorrect product is shown with a red asterisk. c In vitro pri-miR processing assays used to orient the RIIID domains of Drosha on a pri-miR, with diagram shown to the right. Substrate shown is pri-miR-21. d Model depicting the heme-induced reversal of Drosha/DGCR8 complex

Fig. 2

Heme dependence varies among miRs. a 5′ end-labeled in vitro processing assays comparing activities of MP^Heme (130 nM) and MP^C352S (130 nM) for a series of pri-miRs. Substrates are arranged from least heme dependent (left) to most heme dependent (right). b Cotransfection scheme for in vivo processing assays shown in c–h. Constructs containing pri-miR-9-1 and pri-miR-125a in tandem, or pri-miR-21 and pri-miR-125a in tandem, were cotransfected with either full-length FLAG-DGCR8^WT or FLAG-DGCR8^C352S into 293T cells. c Taqman quantitative PCR assay results from transfections in b. Mature miR levels were normalized to miR-125a expression levels. Data are represented as mean ± standard deviation from three biological replicates. Also see Supplementary Fig. 2a–d. d, e Representative Northern blots for pre-miRNAs indicated. U6 levels confirm normalized loading in each lane. f, g Quantitation of Northern blots as shown in d and e. Data are shown as mean ± standard deviation, for a total of three biological replicates. h Quantitative PCR of primary transcript levels for transfections in b. Levels were normalized to pri-miR-125a levels. i, j Representative splinted ligation assay results detecting miR-125a and miR-21 i or miR-9-1 j. U6 levels from Northern blots were used for normalization. k, l Statistical analyses of splinted ligation results, where miR levels with mutant DGCR8 were normalized to those with wild-type DGCR8. Data are represented as mean ± standard deviation, for a total of three biological replicates. *p < 0.05, **p < 0.005, n.s. (not significant), p > 0.05, Student’s t-test (two-sided, paired)

Fig. 3

Heme enables DGCR8 to guide Drosha to the correct junction. a 5′ end-labeled in vitro processing assays of pri-miR-21, using MP^Heme (130 nM), MP^C352S (130 nM), MP^ΔHBR (260 nM), and MP^CTT (1.04 μM). Correct (black asterisk) and incorrect (red asterisk) products are indicated. DGCR8^ΔHBR lacks the HBR but contains the tandem dsRBDs and C-terminal tail. Deletion of HBR and dsRBDs yields the C-terminal tail (DGCR8^CTT). b, c EMSAs comparing binding affinities of MP^ΔHBR to basal b and apical c junctions of pri-miR-21. Schematic diagrams of the RNA are shown next to free RNA bands (arrow), with mature sequence shown in light blue. The basal junction RNAs are annealed RNA duplexes lacking the terminal loop. The apical junction RNAs are the pre-miR fragments. Protein concentrations are (left to right): 3.33, 6.67, 10, 13.32, and 16.67 nM. d 5′ end-labeled in vitro processing assays of pri-miR-125a, similar to a. e, f EMSAs similar to b and c, for pri-miR-125a

Fig. 4

Heme induces a conformational switch in DGCR8 that facilitates RNA binding. a, b Representative negative stain electron micrographs of a MP^Heme and b MP^C352S after crosslinking with glutaraldehyde, with zoomed-in view of the selected particles shown underneath. c Thermal denaturation of DGCR8 in heme-saturated vs. heme-deficient states shows that the melting temperature (T_m) increases by 1.5 °C in the presence of heme. d EMSA with 5′ end-labeled pre-miR-143 shows that DGCR8^Heme binds with an increased affinity compared to that of the DGCR8^C352S mutant. “−” lane contains no protein. Protein concentrations are (left to right): 0.13, 0.26, 0.52, 1.04, and 2.08 μM

Fig. 5

Heme binding allows DGCR8 to recognize the terminal loop structure. a SHAPE analysis of pri-let-7d, showing hyperreactivity associated with MP^Heme binding. Secondary structure diagram (right) depicts the position of the SHAPE “hotspot” (red). Mature sequence is shown in light blue. Control reactions containing no SHAPE reagent (benzoyl cyanide) are shown in lanes 1–3. b SHAPE analysis of pri-let-7d with MP^Heme by capillary electrophoresis. c SHAPE analysis of pri-let-7d with DGCR8. Only the reactions containing the SHAPE reagent are shown, and (unshifted) nucleotide marker lanes are shown to the right. d Diagram summarizing SHAPE results for a series of pri-miRs, aligned according to the site of highest reactivity marked with red font on yellow background. The sequence of the apical fragment released by Dicer is shown for each miR. SHAPE gels for additional miRs are shown in Supplementary Fig. 5c–i. e, f EMSAs comparing the binding affinities of DGCR8^Heme and DGCR8^C352S for e pre-miR-9-1 and f pre-miR-9-1 containing a central break in the terminal loop. Diagrams of RNA sequences are shown to the left. Protein concentrations are (left to right): 33, 50, 67, 83, 100, 117, 133 , 150, 167, 183, 200, 217, 233, and 250 nM. All secondary structure diagrams were designed using mfold

Fig. 6

Model for heme-dependent pri-miR recognition by Drosha and DGCR8. Each pri-miR contains two ssRNA/dsRNA junctions where one is a stronger Drosha-binding site (indicated by “strong” or “weak”). Heme induces a conformational switch in DGCR8 that enables it to bind the terminal loop with higher specificity and affinity. Pri-miRs containing weak basal junctions are more heme dependent because placement of Drosha at the proper junction is driven by heme-bound DGCR8. For pri-miRs containing strong basal junctions, the interaction between the HBR and terminal loop can reinforce Drosha/pri-miR interactions to enhance processing efficiency