(A) Domain architectures of representative RNase III proteins. The N-terminal domain (NTD) unique to budding-yeast RNase III enzymes is indicated.
(B) Activity of recombinant Dcr1 proteins under standard single-turnover conditions. Body-labeled 500 bp dsRNA was incubated with no enzyme (−), full-length K. polysporus Dcr1, or K. polysporus Dcr1ΔC for the indicated time. For comparison, the substrate was incubated with K. polysporus whole-cell extract (Extract) or with the buffer used for extracts (Buffer). Also shown are radiolabeled synthetic 22 and 24 nt RNAs (left) and the migration of other RNA standards (right).
(C) Length distributions of products with the indicated 5' nucleotides. Top and middle: Substrate-matching sequencing reads from analysis of in vitro reactions using the indicated purified proteins. Bottom: For reference, genome-matching small RNAs sequenced from K. polysporus, redrawn from Drinnenberg et al. (2009).
(D) Correlation between cleavage products generated by purified K. polysporus Dcr1 and Dcr1ΔC. Plotted is the read count (including a pseudocount of 1) for each sequenced product from the Dcr1ΔC reaction (y-axis) compared to the count for the corresponding product from the Dcr1 reaction (x-axis). Colors indicate the length of the sequenced products: 14–22 nt (blue), 23 nt (red), and 24–30 nt (green).
(E) Comparison of in vitro activity with product accumulation in strains expressing the corresponding Dcr1 variants. Top: Body-labeled 500 bp dsRNA was incubated with extracts from S. castellii strains with the indicated deletions, additions, and replacements. Bottom: An RNA blot with samples from the same strains was probed for an endogenous siRNA, then reprobed for U6 small nuclear RNA.
See also Figure S1 and Table S1.

(A) Sequence alignment of RNase III domains from each class of RNase III enzymes. Highlighted are the previously identified catalytic residues (green), two newly identified catalytic residues (red), and additional well-conserved amino acids (blue; intensity indicates degree of conservation). Residue numbers and the secondary structure of K. polysporus Dcr1 are indicated below the alignment. Tandem RNase III domains present in Drosha and canonical Dicer are designated a and b. Tm, Thermotoga maritoma; Mt, Mycoplasma tuberculosis; Ec, Escherichia coli; Aa, Aquifex aeolicus; Hs, Homo sapiens; Ce, Caenorhabditis elegans; Sp, Schizosaccharomyces pombe; At, Arabidopsis thaliana; Gi, Giardia intestinalis; Scer, Saccharomyces cerevisiae; Scas, S. castellii; and Kp, K. polysporus.
(B) Left: Close-up view of the Dcr1 catalytic sites showing RNase III domains (ribbons), conserved catalytic residues (sticks), and Mg²⁺ ions (spheres). Right: Metal-ion coordination in the active sites shown in stereo view. Octahedral coordination bonds and hydrogen bonds are drawn as solid and dotted lines, respectively. Fo–Fc simulated annealing omit maps of Mg²⁺ metal ion (magenta) and four water molecules (lime) are contoured at 4.0σ (purple wire mesh).
(C) Stereo view of the superimposed active sites of apo-Dcr1ΔC (green) and apo-AaRNase III (tan; PDB code 1RC5). Bonds are drawn as in (B).
(D) Dicing activity of active-site mutants under standard single-turnover conditions. Body-labeled 500 bp dsRNA was incubated for 5 s with buffer only (−), wild-type Dcr1ΔC (WT), or Dcr1ΔC variants with the indicated mutations.
See also Figure S3.

(A) Effects of changing active-site spacing. Top: The schematic depicts wild-type (green) and active-site mutant (red) Dcr1ΔC dimers assembled on dsRNA to generate different products with lengths varying by 23 nt increments. Bottom: Body-labeled 500 bp dsRNA present at 140 pM was incubated without (−) or with (+) 30 nM wild-type Dcr1ΔC in the absence (−) or presence of increasing concentrations of E224Q Dcr1ΔC (3-fold dilution series from 0.33 nM to 243 nM). Products were resolved by denaturing 10% PAGE.
(B) Length of dicing intermediates. Top: The schematic shows the predicted length variation of intermediates (cyan) for substrates of increasing length. Trapezoids represent dimers of RNase III domains. Bottom: Body-labeled dsRNA substrates of the indicated length (present at ~45 pg/µl) were incubated with buffer only (−), a mixture of 30 nM wild-type Dcr1ΔC and 3 nM E224Q Dcr1ΔC (KpDcr1), or H. sapiens Dicer (HsDicer).
(C) Substrate length requirements for preferentially generating 23 nt products. Top: Docking models depict a pair of dimers bound to the indicated substrates, either anchoring the first dimer at the dsRNA terminus (left) or centering the dimer pair (right). Bottom: Body-labeled dsRNA substrates of the indicated length were incubated for 2 min either without (−) or with (+) Dcr1ΔC under standard single-turnover conditions.
(D) Dicer activities on open and closed substrates. Top: Schematic of substrates, which contained 70 bp of dsRNA flanked by either short ssRNA overhangs or loops. Bottom: Body-labeled substrates were incubated without (−) or with (+) Dcr1ΔC (KpDcr1) or D. melanogaster Dcr-2 (DmDcr-2) under multiple-turnover conditions (30 nM substrate and 10 nM protein). Size markers were estimated based on RNA standards in Figure S5C.
(E) Dicer activities on substrates resembling endogenous yeast substrates. Top: Schematic of substrates, which contained 161 bp of dsRNA within either a perfect duplex (substrate D), a palindromic RNA (substrate E), or an internal duplex (substrate F; red, 7-methylguanosine cap; green, poly(A) tail). Bottom: Body-labeled substrates were reacted with Dicer enzymes as in (D). Percent product was normalized to account for radiolabeled phosphodiester linkages occurring outside of the dsRNA region of substrate E.
See also Figure S5.

See text for description. Off-pathway cleavage products (gray) are disfavored by both cooperative binding and slow product release.
See also Figure S7.

(A) Crystal structure of a Dcr1ΔC dimer at 2.3 Å resolution, showing a pair of NTDs (red and pink), a pair of RNase III domains (blue and silver) and a single dsRBD1 (green). The other dsRBD1 had poor density and is not shown. Disordered loops are shown as dotted lines, and VL-1 and VL-2 are labeled.
(B) Structure of Dcr1ΔC dsRBD (middle), flanked by close-up views of representative segments with good (left) or poor (right) electron density maps (2Fo–Fc). This domain is stabilized through interactions with a symmetry-related NTD of another Dcr1ΔC molecule (red sticks).
(C) Crystal structure of an E224Q Dcr1Δ2d dimer at 1.97 Å resolution. Domain designations and colors are as in (A).
(D) Topology of the A. aeolicus RNase III domain dimer (PDB code 2NUG). Individual monomers are colored in blue and silver, highlighting glycine residues at the kink point between helices α7 and α8 (red stick models).
(E) Topology of K. polysporus E224Q Dcr1Δ2d. Helices α13 and α14 (numbered according to the entire model) participate in a domain swap, with kink-point glycine residues indicated as in (D). The expanded region shows an Fo–Fc simulated annealing omit map of the loops between helices α12 and α13 contoured at 3.2σ (orange wire mesh). The α12–α13 loops are shown as stick models, and the other regions are shown as ribbon models.
See also Figure S2 and Table S2.

(A) Structural alignment of apo-Dcr1ΔC (green) and AaRNase III–dsRNA complex (tan) based on their RNase III domains.
(B) Close-up view of the VL-1 and VL-2 loops. The dsRNA is represented with a sphere model. The numbering of secondary structure elements is as in Figure 3A.
(C) Dicing activity of variable-loop mutants under standard single-turnover conditions. Body-labeled 500 bp dsRNA was incubated for 5 s or 30 min with buffer only (−), wild-type Dcr1ΔC (WT), or Dcr1ΔC variants with the indicated GiDicer loop substitutions.
(D) Cleavage reactions following body- or end-labeled 70 bp dsRNA substrates. Substrates were incubated with buffer only (−), K. polysporus Dcr1ΔC (Kp; standard single-turnover conditions), or H. sapiens Dicer (Hs). HsDicer generates ~21–22 nt siRNA products and ~40–50 nt intermediates as indicated; a ~25 nt product of unknown origin is also observed (*).
(E) A pair of Dcr1ΔC dimers (green and yellow) modeled with dsRNA (sphere model). The anticipated 23 nt siRNA product is highlighted (cyan). The dsRBD is not shown for clarity.
(F) Protein cross-linking analyses of Dcr1ΔC oligomerization. E224Q Dcr1ΔC was incubated with buffer only (−) or the indicated nucleic acid before crosslinking with the indicated glutaraldehyde (GA) concentration. Reactions were analyzed by SDS-PAGE and visualized by silver staining. Also shown is the migration of protein standards with the indicated molecular weights.
See also Figure S4.

(A) Product accumulation under multiple-turnover conditions. Body-labeled 70 bp dsRNA present at 100 nM (implying a ~300 nM concentration of non-overlapping 23 bp binding sites) was incubated with 30 nM Dcr1ΔC for the indicated time.
(B) Quantitative analysis of multiple-turnover kinetics. Body-labeled 70 bp dsRNA present at 100 nM was incubated with the indicated concentration of Dcr1ΔC for the indicated time. Plotted are average values (n = 3; error bars, standard deviation). Solid lines, least-squares fit to the burst equation. Dotted lines, extrapolation of the steady-state rate to the y-axis.
(C) siRNA products bound to Dcr1ΔC. Body-labeled 500 bp dsRNA was incubated with buffer only (−), wild-type Dcr1ΔC (WT), or an active-site mutant of Dcr1ΔC (E224Q) under standard single-turnover conditions. Reactions were placed on ice (−) or heat-denatured at 90°C for 2 min (+) before fractionation. Products were resolved by denaturing 15% PAGE (left) or native 6% PAGE (right). siRNA, duplex of 23 nt siRNAs; ssRNA, 23 nt ssRNA.
(D) Schematic illustrating that for an enzyme that does not recognize the end of a duplex, siRNA generation under multiple-turnover conditions implies binding cooperativity.
(E) Representative gel-shift analysis of E224Q Dcr1ΔC binding to 70 bp dsRNA. Trace amounts of body-labeled 70 bp dsRNA were incubated with increasing amounts of protein.
(F) Binding isotherms for E224Q Dcr1ΔC. Data points represent average values (n = 3 for all substrates except 30 bp, for which n = 2; error bars, standard deviation). Solid lines show the best fit to the Hill equation, which produced the parameters shown to the right. For the 20 bp substrate, the maximum fraction bound was assumed to be 1.0 in order to obtain an estimate of the Hill coefficient (*).
See also Figure S6.