No statistical methods were used to predetermine sample size. The experiments were not randomized and the investigators were not blinded to allocation during experiments and outcome assessment.
Protein expression and purification
The gene encoding full-length Drosophila melanogaster Dcr-2 (UniProt: A1ZAW0) was cloned from the recombinant pFastBac-Dcr-2 plasmid (gifted by the Q. Liu laboratory). Full-length DmLoqs-PD (UniProt: M9MRT5) was PCR amplified from Drosophila cDNA and cloned into a modified pET28a (with a 6×His-SUMO tag). The constructs of WT Dcr-2, Loqs-PD and other mutations were generated using a standard PCR-based cloning strategy and cloned into the corresponding vectors, and their identities were confirmed by sequencing analysis.
Dcr-2 or its mutants was expressed using the Bac-to-Bac baculovirus expression system (Invitrogen) in sf9 cells at 27 °C. One litre of cells (2 × 106 cells per ml, medium from Expression Systems) was infected with 20 ml baculovirus at 27 °C. After growth at 27 °C for 48 h, the cells were collected, resuspended in buffer A (150 mM NaCl, 20 mM Tris-HCl pH 8.0, 10% glycerol, 20 mM imidazole) with 0.5 mM PMSF and protease inhibitors, and lysed by adding 0.5% Triton X-100 and shaken gently for 30 min at 4 °C. Dcr-2 was purified to homogeneity using Ni-NTA affinity, Hitrap Q column (Cytiva), 2nd Ni-NTA affinity and size-exclusion chromatography using the Superdex 200 10/300 Increase column (Cytiva) (in that order).
Loqs-PD and its mutants were expressed in Escherichia coli BL21 (DE3). Loqs-PD was first purified by Ni-NTA affinity chromatography. Using protease ULP1 to remove the 6×His–SUMO tag, and dialysis was applied to remove imidazole. The sample was then applied to a second Ni-NTA chromatography and the flow-through was collected for size-exclusion chromatography using the Superdex 200 16/600 column (Cytiva). Fractions corresponding to the apo Loqs-PD were collected and concentrated to about 10 mg ml−1.
Preparation of dsRNAs
The dsRNAs were in vitro transcribed using T7 RNA polymerase. The pUC19 plasmids containing target sequences with 3′-HDV ribozyme sequences were linearized by EcoRI, extracted with phenol–chloroform and precipitated with isopropanol. The in vitro transcription reaction was performed at 37 °C for 5 h in the buffer containing 100 mM HEPES-K (pH 7.9), 10 mM MgCl2, 10 mM dithiothreitol (DTT), 6 mM NTP each, 2 mM spermidine, 200 μg ml−1 linearized plasmid and 100 μg ml−1 T7 RNA polymerase. For the 5′-monophosphate RNA, 40 mM GMP was added in the transcription reactions. EDTA at a final concentration of 20 mM was added to the samples containing palindromic transcripts. The samples were heated to 95 °C for 5 min and then slowly cooled to room temperature. The annealed transcripts were purified by 8% denaturing urea PAGE, eluted from gel slices and precipitated with isopropanol. After centrifugation, the RNA precipitant was collected, washed twice with 70% ethanol and air-dried, and the RNA was dissolved in ultrapure water. We next used T4 PNK (NEB, M0201) to remove the 2′,3′ cyclic phosphate at the 3′ end of the RNA. The FAM-labelled dsRNA was produced by Silencer siRNA Labeling kit-FAM according to the manufacturer’s instructions.
Pull-down assays were performed to detect Dcr-2–Loqs-PD interactions using His-tagged proteins purified from bacterial or insect cells. First, 1.25 μM His-tagged Loqs-PD and 0.6 μM untagged Dcr-2 were mixed and incubated on ice for 30 min. The protein mixture was then incubated with 15 μl Ni-NTA Agarose (Qiagen, 30210) in a total volume of 500 μl in the binding buffer (200 mM NaCl, 20 mM Tris pH 8.0, 5% glycerol, 20 mM imidazole) at 4 °C for 1 h with gentle rotation. After centrifugation at 500i for 1 min, the supernatant was removed and the beads were washed five times using wash buffer (200 mM NaCl, 20 mM Tris pH 8.0, 5% glycerol, 20 mM imidazole, 0.1% NP-40) by centrifugation, followed by SDS–PAGE analysis.
In vitro dsRNA cleavage assays
Dicer-2–Loqs-PD cleavage assays were performed in cleavage buffer (50 mM HEPES pH 7.2, 100 mM NaCl, 1 mM DTT, 5 mM ATP) with dsRNA. Dcr-2–Loqs-PD and dsRNA were preincubated at 25 °C for 15 min, then added with ATP and MgCl2 to a final concentration of 5 mM to start the reactions. The reactions were stopped with equal volume of 2× formamide loading buffer (95% formamide, 20 mM EDTA, 0.1% SDS, 0.005% xylene cyanol, 0.005% bromophenol blue). Samples were separated by 12% denaturing PAGE, visualized on Typhoon FLA-9000 (GE Healthcare) system.
BS3/EDC-mediated cross-linking mass spectrometry
The purified complexes were incubated with 0.25 mM bis(sulfosuccinimidyl)suberate (BS3; Thermo Fisher Scientific, 21580) in the reaction buffer containing 50 mM HEPES pH 7.5, 80 mM NaCl and 5% glycerol at 25 °C for 2 h or 5 mM 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC; Thermo Fisher Scientific, PG82073) in the reaction buffer containing 50 mM HEPES pH 7.2, 80 mM NaCl and 5% glycerol at 25 °C for 2 h. Cross-linked complexes were further purified to remove oligomer and glycerol by size-exclusion chromatography. The proteins (10 μg) were precipitated and digested for 16 h at 37 °C by trypsin at an enzyme-to-substrate ratio of 1:50 (w/w). The tryptic digested peptides were desalted and loaded on an in-house packed capillary reverse-phase C18 column (40 cm length, 100 µM ID × 360 µM OD, 1.9 µM particle size, 120 Å pore diameter) connected to an Easy LC 1200 system. The samples were analysed with a 120 min high-performance liquid chromatography gradient from 6% to 35% buffer B (buffer A: 0.1% formic acid in water; buffer B: 0.1% formic acid in 80% acetonitrile) at 300 nl min−1. The eluted peptides were ionized and directly introduced into a Q-Exactive mass spectrometer using a nano-spray source. Survey full-scan MS spectra (m/z = 300–1,800) were acquired in the Orbitrap analyzer with resolution r = 70,000 at m/z = 400. Cross-linked peptides were identified and evaluated using pLink2 software30.
Cryo-EM sample preparation and data collection
We used the same specimen preparation and data collection method for all of the cryo-EM datasets. An aliquot of 4 μl of purified or reaction sample was applied to a custom-made graphene grid31 (Quantifoil Au 1.2/1.3, 300 mesh), which were glow-discharged (in a Harrick Plasma system) for 10 s at middle level after 2 min evacuation. The grids were then blotted by a couple of 55 mm filter papers (Ted Pella) for 0.5 s at 22 °C and 100% humidity, then flash-frozen in liquid ethane using the FEI Vitrobot Mark IV. Cryo-EM data were collected on different Titan Krios electron microscopes, all of which were operated at 300 kV, equipped with a Gatan K3 direct electron detector and a Gatan Quantum energy filter. All data were automatically recorded using AutoEMation32 or EPU (post-dicing state dataset) in counting mode and defocus values ranged from −1.5 μm to −2.0 μm. The other parameters of each dataset are provided in Extended Data Table 1.
Image processing and 3D reconstruction
For all of the datasets, the image processing was adopted in similar steps. All of the raw dose-fractionated image stacks were 2× Fourier binned, aligned, dose-weighted and summed using MotionCorr2 (ref. 33). The following steps were then processed in RELION (v.3.1)34. The contrast transfer function parameters were estimated using CTFFIND4 (ref. 35). Approximately 2,000 particles were manually picked and 2D-classified to generate initial templates for automatic picking. A large number of particles were then automatically picked from raw micrographs on the basis of our templates. After one round of reference-free 2D classification and several rounds of 3D classification, using the initial 3D reference models obtained by ab initio calculation in RELION v.3.1, particles from good 3D classes, with better overall structure features, were selected for 3D refinement. The final high-resolution homogeneous refinement was performed in CryoSPARC36. The resolutions were determined by gold-standard Fourier shell correlation. Local resolution distribution was evaluated using blocres command in the Bsoft software package37. The detailed image processing of each dataset is provided in Extended Data Figs. 2 and 3.
Model building and refinement
The highest resolution EM density map of dimer status was used for initial model building, in which the quality of density was sufficient for de novo model building in COOT38. The initial model was separated into three parts (helicase-LoqsPD, DUF283 and other domains) and docked into EM 3D density maps of other states in Chimera39 and then adjusted manually in ISOLDE40 in Chimerax41 and COOT. Finally, all of the models were refined against the EM map by PHENIX42 in real space with secondary structure and geometry restraints. The final models were validated in PHENIX software package. The model statistics are summarized in Extended Data Table 1.
Statistics and reproducibility
For Extended Data Fig. 1a–c, experiments were repeated at least three times. For Extended Data Figs. 1d, 2a,f, 3a,j, 9a and 10a–c, experiments were repeated at least twice.
Further information on research design is available in the Nature Research Reporting Summary linked to this paper.