A coronavirus assembly inhibitor that targets the viral membrane protein

Ethics

Animal housing conditions and experimental procedures were approved by the ethics committee of animal experimentation of KU Leuven (licence P001/2021).

Compound synthesis and chemical characterization of CIM-834

The synthesis and characterization is described in the Supplementary Methods.

Antiviral molecules other than CIM-834

GS-441524 was obtained from MedChem Express (HY-103586). Hydroxychloroquine was purchased from Cell Signaling Technology (85523S). Nirmatrelvir (PF-07321332) was from Wuxi.

Cell lines

VeroE6–GFP cells (African monkey kidney cell line expressing green fluorescent protein; provided by M. van Loock, Janssen Pharmaceutica⁴⁰) were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% v/v heat-inactivated fetal bovine serum (FBS) + 0.5 mg ml⁻¹ geneticin. VeroE6–mCherry cells (generated as described in ref. ⁴¹) were maintained in DMEM, supplemented with 10% (v/v) heat-inactivated FBS and 10 μg ml⁻¹ blasticidin. The A549^ACE2+TMPRSS2 cells (a human lung carcinoma cell line overexpressing human ACE2 and human TMPRSS2 receptors), used for antiviral studies, were from InvivoGen (a549d-cov2r, A549-Dual hACE2-TMPRSS2 cells) and were cultured in DMEM supplemented with 10% v/v heat-inactivated FBS, 300 μg ml⁻¹ hygromycin, 0.5 μg ml⁻¹ puromycin and 10 μg ml⁻¹ blasticidin. A549^ACE2+TMPRSS2 cells, used for subcellular studies, were generated in-house from A549 obtained from ATCC (CCL-185), using lentiviral transduction with pWPI vectors encoding ACE2 (selected using 500 μg ml⁻¹ geneticin) and TMPRSS2 (selected using 1 μg ml⁻¹ puromycin) under EF1alpha promotor control. These cells were cultivated in DMEM supplemented with 10% (v/v) FBS, 100 U ml⁻¹ penicillin, 10 µg ml⁻¹ streptomycin and 1% non-essential amino acids. All assays using VeroE6–GFP, VeroE6–mCherry and A549^ACE2+TMPRSS2 involving virus replication were performed in the respective cell growth medium containing 2% (instead of 10%) FBS. All cell cultures were done at 37 °C and 5% CO₂. BHK-21 cells (Baby hamster kidney cell line obtained from ATCC, CCL10) were maintained in Glasgow MEM (Invitrogen) supplemented with 5% v/v FBS, 10% tryptose phosphate broth, 100 U ml⁻¹ penicillin, 100 µg ml⁻¹ streptomycin and 10 mM HEPES, pH 7.4. For transfection experiments, cells were maintained in Eagle’s minimal essential medium (EMEM), as described⁴². Huh7 cells were cultured in DMEM supplemented with 10% FBS and 1 mM Glutamax.

A549^TMPRSS2+DPP4 cells (A549 obtained from ATCC (CCL-185), developed in-house to overexpress human TMPRSS2 and DPP-4) were cultured in DMEM supplemented with 5% FBS, penicillin (100 U ml⁻¹), streptomycin (100 μg ml⁻¹), blasticidin (10 µg ml⁻¹) and geneticin (0.1 mg ml⁻¹). DMEM with 5% FBS, penicillin (100 U ml⁻¹) and streptomycin (100 μg ml⁻¹) was used for MERS-CoV infection experiments. HCT-8 cells (human colorectal carcinoma cell line obtained from ATCC, CCL-244) were cultured in DMEM supplemented with 5% FBS. DMEM + 3% FBS was used for HCoV-OC43 infection experiments. CRFK cells (Crandell-Rees feline kidney cell line obtained from ATCC, CCL-94) were cultured in DMEM supplemented with 5% FBS, penicillin (100 U ml⁻¹) and streptomycin (100 μg ml⁻¹). The same medium was used for experiments involving FIPV infection. LLC-MK2 cells (a rhesus monkey kidney epithelial cell line, obtained from ATCC, CCL-7) were cultured in MEM with Hanks’ and Earle’s salts supplemented with 5% FBS. MEM with Hanks’ and Earle’s salts and 3% FBS was used for HCoV-NL63 infection experiments. Vero cells (ATCC, CCL-81) were cultured in DMEM supplemented with 5% FBS, penicillin (100 U ml⁻¹) and streptomycin (100 μg ml⁻¹). DMEM with penicillin (100 U ml⁻¹), streptomycin (100 μg ml⁻¹) and 1 µg ml⁻¹ trypsin was used for PEDV infection experiments. All cell lines tested negative for mycoplasma contamination.

Viruses

The SARS-CoV-2 GHB strain (GHB-03021/2020, GISAID: EPI_ISL_407976 | 2020-02-03) was recovered from a nasopharyngeal swab taken from an asymptomatic patient returning from Wuhan, China. Virus stocks were inoculated on HuH-7 cells and then passaged seven times on VeroE6–eGFP cells. GHB has a ΔTQTNS deletion at residues 676–680 that is typical for SARS-CoV-2 strains passaged on VeroE6 cells⁴³. SARS-CoV-2 strains belonging to the alpha, beta, delta and omicron variants of concern were recovered from nasopharyngeal swabs of human cases confirmed by quantitative PCR with reverse transcription (RT–qPCR). Virus stocks were generated by passaging the virus on Calu-3 cells followed by production of a screening virus stock on A549^ACE2+TMPRSS2 cells. The sequences of the passage 0 of these strains are available through GISAID: Alpha/B.1.1.7 (hCoV19/Belgium/rega-12211513/2020; EPI_ISL_791333), beta/B.1.351 (hCoV19/Belgium/rega-1920/2021; EPI_ISL_896474); Delta/B.1.617.2 (EPI_ISL_2425097); Omicron BA.5 (EPI_ISL_14782497); and XBB1.5 (EPI_ISL_17273054). The sequences of BA.2.86 (SARS-CoV-2/hu/DK/SSI-H135) and EG.5.1 (SARS-CoV-2/hu/DK/SSI-H121) variants are available in the European Nucleotide Archive under the project number PRJEB67449 with accession numbers OY747653 and OY747654, respectively. The SARS-CoV-2 BavPat1/2020 strain was provided by C. Drosten (Charité Berlin) through the European Virology Archive (Ref-SKU: 026V-03883). This virus, which has a D614G mutation in the S protein and a mutation affecting the furin cleavage site, owing to passaging on VeroE6 cells, has been used previously for subcellular studies²². Recombinant SARS-CoV-2–mNeonGreen virus (Wuhan strain)⁴⁴ was a gift from V. Thiel (University of Bern). SARS-CoV-2 USA-WA1/2020 (EPI_ISL_404895) was used for the hamster efficacy study. This virus was obtained through BEI Resources (ATCC, NR-52281, batch 70036318). This isolate is closely related to the prototypic Wuhan-Hu-1 2019-nCoV (GenBank accession 112 number: MN908947.3) strain, as confirmed by phylogenetic analysis.

rSARS-CoV-2 WT-USA-WA1, rSARS-CoV-2 WT-USA-WA1-3CLpro: L50F, rSARS-CoV-2 WT-USA-WA1- 3CLpro: T21I-D263G, rSARS-CoV-2 WT-USA-WA1-3CLpro: L50F- E166A L167V have been described previously³⁰. SARS-CoV, strain 200300592, was obtained from the Centers for Disease Control and Prevention. Live SARS-CoV-2 and SARS-CoV work was done in the high-containment A3 and biosafety level 3 facilities of the KU Leuven Rega Institute (3CAPS) under licences AMV 30112018 SBB 219 2018 0892 and AMV 23102017 SBB 219 20170589, according to institutional guidelines. All work with live SARS-CoV (strain Frankfurt-1) and MERS-CoV (strain Jordan N3/2012) virus was done in a biosafety level 3 laboratory at the Leiden University Medical Center.

Other coronaviruses used in this study are: MERS-CoV isolate England 1, identifier: 1409231v, National Collection of Pathogenic Viruses, Public Health England, United Kingdom (studies performed at Jagiellonian University); HCoV-OC43 (ATCC VR-1558), FIPV strain 79-1146 (granted by G. Tekes, Justus Liebig University Giessen); HCoV-NL63 isolate Amsterdam 1 (GenBank: AY567487.2)⁴⁵; PEDV CV777 (granted by C. M. Lia van der Hoek, University of Amsterdam⁴⁶). A short description of the antiviral assays performed with these viruses is provided in Extended Data Table 1.

SARS-CoV-2 antiviral and toxicity assays

VeroE6–GFP cells were seeded at a density of 25,000 cells per well in 96-well plates (Greiner Bio One, 655090) and pretreated with three-fold serial dilutions of the compounds overnight in the presence of the MDR1 inhibitor CP-100356 (final concentration, 0.5 μM). The next day (day 0), cells were infected with SARS-CoV-2 inoculum at a multiplicity of infection (MOI) of 0.001 median TCID₅₀ per cell. The number of fluorescent pixels of GFP signal determined by high-content imaging on day 4 after infection was used as a read-out. The percentage inhibition was calculated by subtracting the background (the number of fluorescent pixels in the untreated − infected control wells) and normalizing to the untreated − uninfected control wells (also background subtracted). The EC₅₀ was determined by logarithmic interpolation. A similar protocol was used to determine the antiviral activity in A549^ACE2+TMPRSS2 (InvivoGen) cells, but no MDR1 inhibitor CP-100356 was used and the cell viability was determined four days after infection using viability staining with 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS). The percentage of antiviral activity was calculated by subtracting the background and normalizing to the untreated − uninfected control wells (also background subtracted), and the EC₅₀ was determined using logarithmic interpolation. In both cell lines, the potential toxicity of compounds was assessed in a similar set-up in treated − uninfected cultures in which metabolic activity was quantified at day 5 using the MTS assay. The 50% cytotoxic concentration (CC₅₀, the concentration at which cell viability reduced to 50%) was calculated by logarithmic interpolation.

Antiviral assays with other coronaviruses

For the antiviral assays with HCoV-NL63, FIPV, PEDV, HCoV-OC43 and MERS-CoV, cells were seeded the day before infection. The next day, serial compound dilutions were mixed with virus (1:1 v/v), incubated for 30 min at room temperature and overlaid on confluent cells. The cells were then incubated for 2 h with the relevant virus at 37 °C (for FIPV, PEDV and MERS-CoV) or 32 °C (HCoV-OC43 and HCoV-NL63). Cells were washed twice with PBS, after which compound containing medium was added and the cells were further incubated at 37 °C (for FIPV, PEDV and MERS-CoV) or 32 °C (HCoV-OC43 and HCoV-NL63) for 3–5 days and cell-culture supernatant was collected for RT–qPCR. For the HCoV-229E assay, three-fold serial dilutions of the compounds in assay medium were made in 96-well plates. After that, Huh7 cell suspension was added to assay wells at a density of 10,000 cells per well and cells were infected with HCoV-229E at an MOI of 0.01. The cell viability was determined 5 days after infection using viability staining with MTS. For the SARS-CoV CPE reduction assays, VeroE6 cells (11,000 cells per well), and for MERS-CoV CPE reduction assays, HuH-7 (10,000 cells per well), were seeded in 96-well plates the day before infection. The next day, two-fold serial dilutions of compound in infection medium with 2% FCS (and 0.25 µM p-GP inhibitor CP-100356 for VeroE6 cells) were added to cells (eight dilution series in parallel), and half of the plate was infected with SARS-CoV Frankfurt-1 (300 plaque-forming units per well) or MERS-CoV Jordan-N3/2012 (200 plaque-forming units per well). The other half of the plate was mock infected to determine the possible cytotoxic effect of the compounds in parallel. Three (for SARS-CoV) or two (MERS-CoV) days after infection, the CellTiter 96 aqueous non-radioactive cell-proliferation kit (Promega) was used to measure the cell viability of infected (n = 4) and non-infected cells (cytotoxicity, n = 4) by absorbance measurement at 495 nm with an Envision plate reader (Perkin Elmer). The values of the wells were normalized to the average signal of uninfected untreated cells (100%). The EC₅₀ and CC₅₀ values were determined by nonlinear regression.

Activity in SARS-CoV-2-infected SCID mice

Nirmatrelvir (from Excenen, batch EXA5024) was formulated as a 100 mg ml⁻¹ and 33.3 mg ml⁻¹ stock (for 300 mg per kg and 100 mg per kg dosing, respectively) in a vehicle containing 43% absolute ethanol and 27% propylene glycol (Sigma) in sterile distilled water. CIM-834 was formulated as 20 mg ml⁻¹ in 14% propylene glycol (Sigma), 1% Tween 80 (Sigma), 85% pH 5 citrate buffer. To evaluate in vivo efficacy, male SCID mice (CB-17/Icr-Prkdcscid/scid/Rj; Janvier Laboratories) 7–9 weeks old were treated by oral gavage with either the vehicle (n = 12, twice a day) or CIM-834 at 100 mg per kg (n = 12 twice a day and n = 12 once a day) or nirmatrelvir at 300 mg per kg (n = 12, twice a day) or 100 mg per kg (n = 6, twice a day), starting from day 0, just before infection with the beta variant B.1.351 (hCoV-19/Belgium/rega-1920/2021; EPI_ISL_896474, 2021-01-11). For virus infection, animals were anesthetized with isoflurane and inoculated intranasally with 40 µl containing 10⁵ TCID₅₀ SARS-CoV-2 beta variant (day 0). In the therapeutic set-up, animals were infected on day 0 and treatment with CIM-834 (100 mg per kg, twice a day) was initiated 24 h, 30 h or 48 h after infection. Mice were housed in individually ventilated cages with three mice per cage and monitored daily for weight changes and any clinical signs. At day 3 after infection, animals were euthanized by intraperitoneal injection of 100 µl Dolethal (200 mg ml⁻¹ sodium pentobarbital, Vétoquinol SA), and the lungs were collected. Infectious viral lung loads were quantified by end-point virus titration. To prevent carry-over of the compound during determination of infectious virus titres, the cells were washed and given fresh medium, immediately after incubation for 2 h with lung homogenates. Cells were subsequently incubated at 37 °C for three days before TCID₅₀ read-out.

Activity in SARS-CoV-2-infected Syrian hamsters

The hamster infection model of SARS-CoV-2 has been described previously⁴⁷. Female Syrian hamsters (Mesocricetus auratus) 8–10 weeks old from Janvier Laboratories were kept in twos in individually ventilated isolator cages (IsoCage N Bio-containment System, Tecniplast) at 21 °C, 55% humidity and 12 h:12 h day:night cycles. Housing conditions and experimental procedures were approved by the ethics committee of animal experimentation of KU Leuven (licence P065−2020). For infection, animals were anesthetized with ketamine/xylazine/atropine and inoculated intranasally with 50 µl containing 1 × 10⁴ TCID₅₀ of SARS-CoV-2 USA-WA1/2020. Hamsters were treated by oral gavage with either the vehicle, CIM-834+Ritonavir (100 + 50 mg per kg) or nirmatrelvir (300 mg per kg), twice daily starting from day 0, directly preceding the infection with SARS-CoV-2. All the treatments continued until day 3 after infection. On day 4, hamsters were euthanized by intraperitoneal injection of 500 μl Dolethal (200 mg ml⁻¹ sodium pentobarbital, Vétoquinol SA). Lungs were collected and viral RNA and infectious virus were quantified by RT–qPCR and end-point virus titration, and the left lung lobes were fixed in 4% formaldehyde for histopathological analysis. Two independent studies were performed, each with a total of n = 8 for the vehicle group, CIM compound group and nirmatrelvir group. In parallel, we assessed whether treatment of infected animals with CIM-834/Ritonavir could prevent transmission of infection to non-infected contact hamsters. For this, starting in the afternoon of day 1 after infection, each index infected hamster, treated with either vehicle (n = 4) or CIM-834 + Ritonavir (100 + 50 mg per kg, n = 8), was co-housed with untreated sentinel hamsters in one cage, and the co-housing continued until the day of euthanization, three days after the first exposure.

Histology

For histological examination, the lungs were fixed overnight in 4% formaldehyde and embedded in paraffin. Tissue sections (5 μm) were analysed after staining with haematoxylin and eosin and scored blindly for lung damage by a pathologist. The scored parameters, for which a score of 0–3 was attributed, were the following: congestion, intra-alveolar haemorrhagic, apoptotic bodies in the bronchus wall, necrotizing bronchiolitis, perivascular oedema, bronchopneumonia, perivascular inflammation, peribronchial inflammation and vasculitis. The sum of the different scores was then reported as a cumulative score.

In vitro resistance selection

SARS-CoV-2 B.1.1.7 was passaged in A549^ACE2+TMPRSS2 (InvivoGen) cells in the presence of increasing concentrations of CIM-834. Selection was initiated at 0.03 μM (around 1/3 of its EC₅₀) and, for each passage, the virus was cultured at the same concentration, and at 3×, 10× and 30× higher concentrations. The culture with the highest compound concentration that still showed virus breakthrough, as observed by a significant cytopathic effect, was then used for the next passage. At passage 5 (day 17), vRNA in the cell-culture medium was sequenced.

Genotypic analysis

Illumina sequencing

RNA was extracted from the cell-culture supernatant using a NucleoSpin RNA virus kit (Macherey-Nagel) according to the manufacturer’s instructions. Whole-genome sequencing was outsourced to Eurofins Genomics (ARTIC SARS-CoV-2 WGS, Konstanz), who performed reverse transcription, enrichment of the viral genome using a primer set similar to the ARTIC primers (more than 200 primer pairs, covering the full 29.9-kilobase viral genome), generation of libraries, Illumina sequencing (2 × 150 base-pair read mode) and sequence cleaning to remove adapters and poor-quality bases. Sequences were further analysed using Geneious Prime software (v.2022.2.1) by mapping to the SARS-CoV-2 reference sequence (NC_045512), and variant calling was performed as described by the software manufacturer (https://help.geneious.com/hc/ en-us/articles/360045070991-Assembly-of- SARS-CoV-2-genomes-from- tiled-amplicon-Illumina- sequencing-using-Geneious-Prime).

Generation and rescue of recombinant viruses

Reverse-engineered SARS‐CoV‐2 Wuhan with the M-protein P132S substitution was constructed by mutating nucleotide positions 26916–26918 from CCG to AGC in a bacterial artificial chromosome vector containing a full-length cDNA copy of the genome of SARS-CoV-2 strain SARS-CoV-2/human/NLD/Leiden-0008/2020 using the following primer pair for two-step en passant recombineering in Escherichia coli⁴⁸:

5′-TCTTCTCAACGTGCCACTCCATGGCACTATTCTGACCAGAAGCTTGCTAGAAAGTGAACTCGTAATGGCGTATCACGAGGCCCTTTCGTC-3′ and

5′-CACGAAGGATCACAGCTCCGATTACGAGTTCACTTTCTAGCAAGCTTCTGGTCAGAATAGTGCCATTAGGGATAACAGGGTAATGGCCTG-3′.

With the design of the primers, a silent marker mutation was introduced at nucleotide positions 26919 and 26921 changing the CUU codon to UUG. Two independent mutant BACs were sequenced across the M-protein gene and 5 µg DNA was linearized with NotI. Full-length RNA was obtained by in vitro transcription with T7 RNA polymerase followed by lithium chloride precipitation according to the manufacturer’s protocol (mMessage-mMachine T7 kit, Ambion). Synthetic mRNA encoding the SARS-CoV-2 N protein was generated as described previously⁴⁹. To launch SARS-CoV-2 wild type and mutant, 5 µg of full-length SARS-CoV-2 RNA and 10 µg of N-protein mRNA were electroporated into 2 × 10⁶ BHK-21 cells using the AMAXA Nucleofector 2b (program A-031) and Nucleofection T solution kit (Lonza). Transfected cells were mixed with 4 × 10⁶ VeroE6 cells, and cells were incubated at 37 °C until the full cytopathic effect was observed. From this passage 0 virus stock, a passage 1 stock was prepared on Calu-3 cells as described⁵⁰. The presence of the AGC codon at nucleotide positions 26916–26918, encoding M(P132S), was confirmed by Illumina sequencing (see above). This sequencing also showed that no other mutations had been accidently introduced in the viral backbone.

Reverse-engineered SARS‐CoV‐2 BF.7 (Omicron) was constructed, starting from the seven-plasmid reverse-genetics system⁵¹ for strain SARS-CoV-2/USA_WA1/2020 (a gift from P. Y. Shi through the World Reference Center for Emerging Viruses and Arboviruses). An RNA extract was made from a clinical isolate of SARS‐CoV‐2 Omicron subvariant BF.7 (donated by P. Maes, Rega Institute), using the QIAamp viral RNA kit (Qiagen). Next, cDNA was made with random hexamer primers and the Superscript IV RT First-Strand synthesis system (Invitrogen). The cDNA was used as a template to make seven PCR fragments covering the viral genome and containing an overlap with the target plasmids, using Platinum SuperFi PCR (Invitrogen). Using the NEBuilder HiFi DNA assembly cloning kit (New Biolabs), the WA1 sequences in the seven plasmids⁵¹ were replaced by the Omicron-encoding PCR fragments. The M-protein substitution P132S was introduced by Q5 site-directed mutagenesis (New England Biolabs). M-protein substitutions M91K, S99A and N117K were introduced by site-directed mutagenesis using Platinum SuperFi II DNA polymerase (Invitrogen).

All plasmids were validated by Sanger sequencing (Macrogen). Preparation of viral DNA, in vitro transcription and electroporation of BHK-21 cells was carried out as previously described⁵¹, except for the use of an ECM 830 Square Wave electroporation system (850 V, 3 pulses of 0.30 ms, a 3 s interval, BTX). The electroporated cells were added to A549^ACE2+TMPRSS2 cells (InvivoGen) in medium containing 10% FCS. After incubation for 6 h at 37 °C, the medium was replaced by medium with 0.2% FCS. Four days later, the virus stocks were collected, passaged once on A549^ACE2+TMPRSS2 (InvivoGen) and subjected to whole-genome sequencing (Oxford Nanopore Technologies, by B. Vanmechelen, Rega Institute) to verify the desired sequence.

SARS-CoV-2 studies in human nasal airway epithelial cultures

Previously described procedures⁵² were followed. In brief, human airway nasal epithelial cells (MucilAir pool of donors, product code EP02) were obtained from Epithelix in an air–liquid set-up. After arrival, the inserts were washed with pre-warmed 1× PBS and maintained in MucilAir medium (Epithelix, EP04MM) at 37 °C and 5% CO₂ for at least four days before use. On the day of the experiment, the cultures were pretreated with basal medium containing compounds at different concentrations for 1 h before infection with 100 μl SARS-CoV-2 inoculum (1,000 TCID₅₀ per insert) at the apical side for 1.5 h, after which the viral inoculum was removed. Viral release from the cultures was measured by washing the apical sides with 250 µl Mucilair medium and determination of the viral load by RT–qPCR or titration. Compound-containing medium in the basolateral side of the cultures was refreshed on day 2 after infection. All incubations from start of the infection were done at 35 °C and 5% CO₂.

Quantification of vRNA using RT–qPCR

Viral RNA from apical washes (air–liquid culture inserts) or from cells was isolated using the Cells-to-cDNA II buffer kit (Thermo Fisher Scientific, AM8723). In brief, 5 μl wash fluid was added to 50 μl lysis buffer or cells were lysed in Cell lysis buffer (2,500 cells per μl Cell Lysis II Buffer); samples were incubated at room temperature for 10 min and then at 75 °C for 15 min. Next, 150 μl nuclease-free water was added to the mixture before RT–qPCR. In parallel, a ten-fold serial dilution of corresponding virus stock was extracted using the same protocol, and it could then later function as a standard curve. The amount of viral RNA was quantified by RT–qPCR using an iTaq universal probes one-step kit (Bio-Rad, 1725141) and a commercial mix of primers for the N-protein gene (forward primer, 5′-GACCCCAAAATCAGCGAAAT-3′; reverse primer, 5′-TCTGGTTACTGCCAGTTGAATCTG-3′) and probes (5′-FAM-ACCCCGCATTACGTTTGGTGGACC-BHQ1-3′) from IDT Technologies (10006606). The reaction (a final volume of 20 μl) consisted of 10 μl one-step reaction mix 2×, 0.5 µl RT enzyme, 1.5 μl of a primers and probes mix, 4 μl nuclease-free water and 4 μl of sample. The RT–qPCR was executed on a Lightcycler 96 thermocycler (Roche), starting at 50 °C for 15 min and 95 °C for 2 min, followed by 45 cycles of 3 s at 95 °C and 30 s at 55 °C. By correlating the C_t values of the ten-fold serial dilution of the virus stock with known titre (standard curve), the amount of vRNA from the samples could be expressed as a TCID₅₀ equivalent per cell-culture insert (TCID₅₀ equivalent per insert).

Time-of-drug-addition assay

VeroE6 cells were seeded in 12-well plates the day before infection. Cells were infected with an MOI of 1 TCID₅₀ per cell (SARS-CoV-2 GHB) and compounds were added 0, 3, 5 and 7 h after infection. At 10 h after infection, virus-containing supernatant was removed and cells were either lysed for quantification of vRNA levels by RT–qPCR (see above) or cells were collected for determination of intracellular infectious virus titres by plaque assay (see below). For the time-of-drug-addition assay with reporter cells and reporter virus, VeroE6–mCherry cells were seeded in 96-well plates the day before infection. Cells were infected with SARS-CoV-2–mNeonGreen virus (at a final MOI of 0.1 TCID₅₀ per cell) and compounds were added −0.5, 0, 2, 4, 6 and 8 h after infection. At 10 h after infection, cells were imaged using high-content imaging and the number of infected cells was calculated, after which the percentage inhibition relative to the DMSO-treated control was calculated.

Determination of infectious virus titres

Cells were trypsinized and resuspended in PBS, after which they were freeze-thawed 3× on liquid nitrogen to release intracellular viral particles. Cellular debris was removed by centrifugation (5 min at 13,400 rpm and 4 °C) and 50 µl of the supernatant was used to determine the infectious titre by plaque assay.

For this, 5-fold dilutions of the cell lysate supernatant were added to VeroE6 cell monolayers in 12-well plates and incubated at 37 °C for 1 h. Subsequently, the inoculum mixture was replaced with 0.8% (w/v) methylcellulose in DMEM supplemented with 2% FBS. After three days of incubation at 37 °C, the overlays were removed, the cells were fixed with 3.7% PFA and stained with 0.5% crystal violet, and plaques were counted visually.

Plasmid construction for SARS-CoV-2 expression of M and E protein

The coding sequences of the SARS-CoV-2 M and E proteins were cloned between the BamHI and EcoRI restriction sites of a pCDNA3.1(+) vector, containing a coding sequence for an amino-terminal V5 tag or a carboxy-terminal HSV tag for the M and E proteins, respectively. To this end, cDNA obtained after reverse transcription of RNA extracted from SARS-CoV-2 (strain LillehCoV-19/France/HDF-IPL/2020, GenBank MW575140) infected cells was amplified by PCR using Q5 High-Fidelity 2X Master Mix (New England Biolabs) with forward primer 5′-CGGGATCCGCAGATTCCAACGGTACTATTA-3′ and reverse primer 5′-CAGAATTCTTACTGTACAAGCAAAGCAATA-3′ for M, and forward primer 5′-TCGGATCCGCCACCATGTACTCATTCGT-3′ and reverse primer 5′-TGGAATTCGACCAGAAGATCAGGAACTC-3′ for E. To allow better visualization of the M protein on western blot, an M-N5Q glycosylation site mutant was used in all VLP studies. This M-N5Q glycosylation site mutant was generated by site-directed mutagenesis PCR with forward primer 5′-CGGGATCCGCAGATTCCCAAGGTACTATTACCGTT-3′ and reverse primer 5′-CAGAATTCTTACTGTACAAGCAAAGCAATA-3′. The M(P132S) mutation was introduced by fusion PCR using the primers 5′-CTTTCTAGAAGCGATCTGGTCAGAATAGTGCCAT-3′ and 5′-TTCTGACCAGATCGCTTCTAGAAAGTGAACTCGT-3′. For E, a silent mutation removing the intrinsic EcoRI restriction site was inserted at nucleotide position 195 (G > A) by fusion PCR using the primers 5′-GAATTTAGATTTTTAACACGAGAGTA-3′ and 5′-AATCTAAATTCTTCTAGAGTTCCTGA-3′.

SARS-CoV-2 VLP assay and western blot analysis

Huh-7 cells⁵³ were seeded in 100-mm dishes at a concentration of 2 × 10⁶ cells per dish. The next day, cell-culture medium was changed to medium containing various concentrations of CIM-834 (0.2, 1.0 or 5.0 µM) or DMSO. Cells were co-transfected with 2 µg of plasmid encoding for wild-type or mutant (M(P132S)) protein and 1 µg of E-HSV-encoding plasmid using TransIT-LT1 Transfection Reagent (Mirus Bio). Then, 48 h after transfection, VLPs and cell lysates were collected and stored as described previously⁵⁴ for MERS-CoV VLPs. VLP and lysate samples were resuspended in reducing Laemmli loading buffer and separated on a 12% polyacrylamide gel by SDS–PAGE. M proteins were visualized by immunoblotting using a primary anti-V5 antibody (Thermo Fisher Scientific, R96025, clone SV5-Pk1, 1:1,000) and secondary HRP-labelled goat anti-mouse IgG antibodies (Jackson ImmunoResearch, 115-035-146, 1:10,000), as described previously⁵⁴. Tubulin was used as a loading control and stained using a mouse anti-β-tubulin IgG1 antibody (Sigma-Aldrich, T5201, clone TUB 2.1, 1:2,000). Western blot bands were quantified using Image J and its band quantification function. For unprocessed blots, see Supplementary Fig. 1.

Sample preparation for thin section and tomography TEM

A549^ACE2+TMPRSS2 cells (generated in-house) were cultured in 24-well plates on either glass coverslips for microwave-assisted chemical processing or on 3-mm flat sapphire discs 50 μm thick (Engineering Office M. Wohlwend, Switzerland) for high-pressure freezing. One hour after infection with SARS-CoV-2 (strain BavPat1/2020), the cells were treated with 1 µM CIM-834 or nirmatrelvir or a DMSO control. Mock-inoculated cells treated with DMSO or nirmatrelvir or CIM-834 were processed in parallel. Cells were collected at 10 h after infection by fixation in two steps. First, by adding a 2× concentrated EM fixative to the cell-culture medium in a 1:1 ratio for 15 min at room temperature, and subsequently, a 1× concentrated fixative for 120 min at room temperature. The composition of the 1× fixative was 2.5% glutaraldehyde and 1% formaldehyde in 50 mM Na-cacodylate buffer (pH 7.2) containing 50 mM KCl, 2.6 mM MgCl2, 2.6 mM CaCl2 and 2% sucrose. Plates were plunged in a bath of 6% EM-grade formaldehyde in PHEM buffer (pH 7.4) for virus inactivation for 30 min at room temperature and kept at 4 °C until further processing on the next day. Samples on coverslips were further processed at room temperature using a PELCO BioWave Pro+ microwave equipped with vacuum chamber (set to 20 hg) and water-cooling platform (set at 23 °C). Samples were rinsed six times with 100 mM Na-cacodylate for 10 min each, then post-fixed in a microwave in 65 mM Na-cacodylate buffer, 1% OsO₄ at 150 mW, cycling power on and off every 2 min for seven cycles, under vacuum. Samples were then rinsed four times in double-distilled water (twice for 10 min on the bench and twice for 40 s each time in the microwave at 250 mW, vacuum off) and subsequently treated in 1% uranyl acetate in double-distilled water in a BioWave for seven cycles at 2 min each, cycling power on (100 mW) and off at each cycle, under vacuum. Dehydration with an ethanol series (50%, 70%, 90% and 2 × 100%) was then performed for 40 s each step, at 100 mW, vacuum off. Coverslips were finally transferred to BEEM capsules on fresh 100% Epon 812 resin and polymerized at 60 °C for 48–76 h.

Cells prepared on sapphires for high-pressure freezing were prefixed and exported from BSL3 in 6% PFA, similarly to the samples on coverslips. The samples were then rinsed six times with PHEM buffer (100 mM), immersed in 100 mM PHEM with 15% BSA as a cryo-protectant and high-pressure frozen using a BalTec HPM-010 with carriers forming a 40 µm-deep cavity (3-mm aluminium carriers, type B 0/0.3 mm and type 748 0.04/0.020 mm, Engineering Office M. Wohlwend). High-pressure frozen samples were then subjected to freeze substitution in a Leica AFS II using cryo-tubes with screw caps and rubber sealing rings, containing 1 ml of fixative cocktail composed of 0.2% OsO₄, 1% uranyl acetate and 5% water in acetone. The AFS chamber temperature was increased over the course of 24 h as follows: 1 h at −90 °C; 8 h at −90 °C to −80 °C; 8 h at −80 °C to −50 °C; 2 h at −50 °C to −20 °C; 2 h at −20 °C to 0 °C. Samples were then rinsed for 5 min with dry acetone on ice and further processed using a microwave, rinsed twice with ethanol for 40 s at 250 mW and infiltrated with increasing Epon 812 concentrations in ethanol (10%, 30%, 50%, 70%, 90% and 2 × 100% for 3 min each in a BioWave vacuum cycling system). At the last 100% step in Epon 812, sapphires were transferred to AFS plastic moulds and polymerized at 60 °C for 48–72 h. Ultrathin and semithin sections of 70 nm or 300 nm, respectively, were generated for both chemically fixed and high-pressure frozen samples using a UC7 Leica ultramicrotome and a 30° diamond knife (Diatome) and collected on Pioloform-coated slot grids. Grids were post-stained for 5 min with 3% uranyl acetate in 70% methanol and 2 min with lead citrate. To locate the infected cells, the Serial-EM Navigator functionality and a procedure adapted from ref. ⁵⁵ were used to map the central section of a ribbon of 5 on a JEOL 1400 equipped with a TemCam-F416 camera. Cells displaying a DMV cluster were identified and stored in the map, and navigator maps and grids were then transferred to a JEOL 2100+ equipped with a Matataki sCMOS camera to acquire montages, covering the whole perinuclear region of the selected cells at a magnification of ×6,000 or ×15,500.

For tomography, semithin sections of 200 nm or 300 nm were screened and imaged using a Tecnai F30 microscope (Thermo Fisher Scientific) equipped with a Gatan OneView camera. Target positions were manually selected and acquired at ×15,500 magnification (−60° to +60° per axis; increment, 1°) by single-axis tomography (0.78 nm per pixel). The tilt series were reconstructed using IMOD^56,57,58. Segmentation of selected tomograms was done manually using the brush segmentation tool in Amira-Avizo software v.2020.1 (ThermoFisher), Volume renderings and animations were also computed with the same software.

Offline affinity selection mass spectrometry

We incubated 5 µM M protein for 30 min at 25 °C in buffer only (20 mM HEPES, pH 7.5, 150 mM NaCl, 0.001% LMNG, 0.0001% CHS, 0.00033% GDN, supplemented with 2% DMSO) or in buffer with CIM-834 (in DMSO, 2% final concentration). Unbound compound was removed from the bound compound to protein by Zeba Spin Desalting columns, 7 K MWCO, according to the manufacturer’s protocol. All experiments were done in triplicate. Mass-spectrometry measurements were done on an Agilent 6546 Quadrupole Time-of-Flight for liquid chromatography with tandem mass spectrometry. For data analysis, the Agilent Mass Hunter Qualitive Analysis software (target screening workflow) and the Find by Formula method for compound identification by mass/isotope pattern matching was used. An EIC peak area of [M + H]+, [M+Na]+ and [M + K]+ masses was extracted using a mass error tolerance window of 10 ppm.

Sample preparation for cryo-EM to study CIM-834–M interactions

For both the Fab–E and Fab–B complexes, M protein was combined with CIM-834 and either Fab–E or Fab–B, resulting in a final concentration of 44.44 µM M protein, 100 µM CIM-834 and Fab–E/B in a 2.2 molar excess of M protein. All components were diluted in a buffer solution comprising 20 mM HEPES-NaOH at pH 7.7, 150 mM NaCl, 0.0025% LMNG and 0.00025% CHS. Samples were mixed and incubated for 15 min on ice before vitrification. Approximately 3.5 µl of the sample was pipetted onto glow-discharged R1.2/1.3 200 mesh holey copper carbon grids (Quantifoil) and then plunge-frozen in liquid ethane using a Vitrobot mark IV (Thermo Fisher Scientific). Both datasets were collected at the Netherlands Center for Electron Nanoscopy. Grids were loaded into a Titan Krios electron microscope (Thermo Fisher Scientific) operating at 300 kV, equipped with a K3 direct electron detector and Bioquantum energy filter (Gatan). The slit width of the energy filter was set to 20 eV. Imaging was done at a nominal magnification of ×81,000 and ×105,000 for Fab–E and Fab–B, respectively, in super-resolution mode using EPU software (Thermo Fisher Scientific). A total of 5,058 and 5,037 movies were recorded for the Fab–E and Fab–B complexes, respectively. Detailed data-acquisition parameters are summarized in Extended Data Table 2.

Single-particle image processing to study CIM-834–M interactions

For the Fab–E and Fab–B complexes, patch motion correction, using an output F-crop factor of 0.5, and patch CTF estimation were performed in cryoSPARC⁵⁹. Micrographs with a CTF estimated resolution of worse than 10 Å were discarded, leaving 5,058 and 5,036 images for further processing, respectively. The blob picker tool was then used to select 4,327,548 and 4,497,785 particles, respectively, which were then extracted in an 80-pixel box (Fourier binned 4.5 × 4.5). For the Fab–E complex, two rounds of 2D classification were done, resulting in 270,970 particles being selected for further processing. Ab initio reconstruction generated one well-defined reconstruction of the Fab–E complex. Particles belonging to this class were then re-extracted in a 300-pixel box. During extraction, particles were Fourier binned by a non-integer value, resulting in a final pixel size of 1.00 Å. Subsequently, non-uniform refinement was done on the extracted particles with C2 symmetry imposed⁶⁰, yielding a final reconstruction with global resolution of 3.3 Å. Similarly, for the Fab–B complex, two rounds of 2D classification were done, resulting in 187,608 particles being selected for further processing. Ab initio reconstruction generated one well-defined reconstruction of the Fab–B complex. To identify all Fab–B complex particles from the dataset, all particles picked by the blob picking tool (4,497,785 particles) were used for five rounds of heterogeneous refinement, using the reconstruction of the Fab–B complex from the initial ab initio job as a reference volume for high-resolution particles to be further isolated. This was followed by two final rounds of ab initio to further isolate the best-quality particles for final reconstruction. The remaining 73,185 particles were then re-extracted in a 360-pixel box. During extraction, particles were Fourier binned by a non-integer value, resulting in a final pixel size of 0.836 Å. Subsequently, non-uniform refinement was done on the extracted particles with C2 symmetry imposed⁶⁰, yielding a reconstruction with a global resolution of 3.3 Å. Reference-based motion correction was successfully run on 72,998 of the re-extracted particles. These particles were then subjected to a final non-uniform refinement, with C2 symmetry imposed⁶⁰, yielding a final reconstruction with a global resolution of 3.2 Å. The gold standard Fourier shell correlation (FSC) criterion (FSC = 0.143) was used for calculating all resolution estimates, and 3D-FSC plots were generated in cryoSPARC⁶¹. To enable model building, globally refined maps were filtered by local resolution in cryoSPARC.

Modelling to study CIM-834–M interactions

Initially, a previously determined SARS-CoV-2 M structure (PDB: 8CTK; ref. ³⁴) and an Alphafold2 model⁶² of Fab–B were rigid-body fitted into the M–Fab–B-CIM map using the UCSF Chimera Fit in map tool⁶³. An initial round of molecular-dynamics flexible fitting was then done on the combined model using Namdinator⁶⁴ and then manually adjusted in Coot⁶⁵. During the latter step, the CIM-834 compound was added using the ligand builder tool. When selecting the pose of CIM-834, the following were taken into account. First, the morphology of the density. The flat plate-like morphology of the density orientated away from the M two-fold symmetry axis was consistent with the piperidine and pyrimidine rings of CIM-834. The density for the opposing end of the density was thinner and less well resolved (Extended Data Fig. 6), consistent with the increased flexibility of the molecule beyond the amide position. Second, the positioning of hydrogen-bonding partners. The selected orientation of CIM-834 positions the pyridazine ring in proximity to S99 and N117, allowing for potential direct or water-mediated hydrogen bonds. The model was then refined by performing iterative cycles of manual model building using Coot⁶⁵ and real-space refinement using Phenix⁶⁶; eLBOW was used to generate ligand restraints for CIM-834 (ref. ⁶⁷). Model validation was done using Molprobity^68,69,70.

Analysis and visualization to study CIM-834–M interactions

Interactions between M and CIM-834 were calculated using LigPlot⁷¹ and UCSF ChimeraX⁷². The UCSF Chimera MatchMaker tool was used to obtain root mean square deviation values using default settings. Figures were generated using UCSF ChimeraX⁷², and structural-biology applications used in this project were compiled and configured by SBGrid⁷³.

Spike pseudotyped vesicular stomatitis virus entry assay

A human codon-optimized gene encoding the spike protein of SARS-CoV-2 S ancestral Wuhan-Hu-1 (GenBank: NC_045512.2) was synthesized by GenScript. SARS-CoV-2 S pseudo-typed vesicular stomatitis virus (VSV) and neutralization assays were reframed as described previously⁷⁴. In brief, to produce SARS-CoV-2 S VSV, confluent HEK-293T cells were transfected with pCAGGS expression vectors encoding SARS-CoV-2 S with a C-terminal 18-residue cytoplasmic tail truncation, to increase cell surface expression. Then, 48 h after transfection, cells were infected with VSV G pseudotyped VSVΔG bearing the firefly luciferase reporter gene. Then, 24 h later, supernatant containing SARS-CoV-2 S pseudotyped VSV was collected and clarified, and the virus was titrated on Vero-E6 cells. For the virus-neutralization assay, CIM-834 and control monoclonal antibody 87G7 were serially diluted three-fold and five-fold, respectively, before pre-incubating with equal volumes of SARS-CoV-2 S pseudotyped VSV at room temperature for 1 h. After incubation, Vero-E6 cells were inoculated with the mixture and cells were incubated at 37 °C for 20 h. Subsequently, cells were washed once with PBS and lysed with passive lysis buffer (Promega). Firefly luciferase expression was measured on a Berthold Centro LB 960 plate luminometer with d-luciferin as a substrate (Promega). The percentage of virus neutralization was calculated as the ratio of the reduction in luciferase read-out in the presence of compound normalized to luciferase read-out in its absence. The half-maximal inhibitory concentrations (IC₅₀) were determined using four-parameter logistic regression (GraphPad Prism v.8.3.0).

Immunofluorescence

Huh-7 cells were seeded on coverslips in 24-well plates at a concentration of 8 × 10⁴ cells per well. The next day, 1 µM CIM-834 or DMSO was added to the medium and cells were transfected with a total of 250 ng plasmid encoding for V5-tagged WT-M using the TransIT-LT1 transfection reagent (Mirus Bio). Then, 16 h after transfection, cells were fixed with 4% paraformaldehyde in PBS for 20 min at room temperature, followed by permeabilization with 0.1% Triton-X100 in PBS for 3 min at room temperature. After a blocking step with 10% normal horse serum for 15 min at room temperature, V5-tagged M proteins were visualized by incubation with monoclonal anti-V5 antibodies (Thermo Fisher Scientific, 37-7500, clone 2F11F7, 1:500) in 10% normal horse serum, followed by incubation with Alexa Fluor 488-conjugated donkey anti-mouse IgG secondary antibodies (Jackson ImmunoResearch, 715-545-151, 1:500). The trans-Golgi network was visualized using polyclonal sheep anti-human TGN46 antibodies (AHP500, BioRad,1:500), followed by incubation with Cy3-conjugated donkey anti-sheep IgG antibodies (Jackson ImmunoResearch, 713-165-147, 1:500). Nuclei were visualized using 1 µg ml⁻¹ of 4′,6-diamidino-2-phenylindole (DAPI), and coverslips were mounted in Mowiol mounting medium. Images were acquired with an LSM 880 confocal laser scanning microscope (Zeiss) using a ×63 oil-immersion objective lens. The extent of colocalization between M and TGN46 was quantified by calculating the Pearson’s correlation coefficient using the JACoP plug-in of ImageJ. For both treated (1 µM CIM-834) and untreated (DMSO) conditions, 25 cells were analysed.

Expression and purification of recombinant proteins

The coding sequences for SARS-CoV-2 Wuhan-Hu-1 (GenBank: NC_045512.2) were used for recombinant protein expression. Nsp3 (PLpro domain), nsp15 (Mpro domain), nsp14 (N7-MTase), nsp10 and nsp16 (2′O-MTase) were cloned in fusion with an N-terminus hexa-histidine tag, as previously described⁷⁵. The proteins were expressed in E. coli and purified by affinity using IMAC cobalt bead. In brief, after cell sonication in a buffer containing 50 mM Tris, pH 6.8, 300 mM NaCl, 10 mM imidazole, 5 mM MgCl2 and 1 mM BME, supplemented with 0.25 mg ml⁻¹ lysozyme, 10 μg ml⁻¹ DNase and 1 mM PMSF, the proteins were purified through affinity chromatography with HisPur Cobalt resin 480 (Thermo Scientific), washed with an increased concentration of salt (1 M NaCl) and imidazole (20 mM), before elution in a buffer supplemented with 250 mM imidazole. The MTases were subsequently purified by size-exclusion chromatography (GE Superdex S200) in a final buffer of 50 mM Tris, pH 6.8, 300 mM NaCl, 5 mM MgCl2 and 1 mM BME. The proteins were concentrated using Vivaspin 20 centrifugal concentrators with a 10 kDa MWCO (GE Healthcare, VS2001), dialysed against the elution buffer in the absence of imidazole, and stored at −20 °C in a buffer containing 50% glycerol. The replication/transcription complex endowed with RdRp activity was purified and assembled, according to ref. ⁷⁶, with variations described in ref. ⁷⁷.

Determination of Mpro inhibition by CIM-834

A fluorescent substrate comprising the cleavage site of SARS-CoV-2 Mpro (Dabcyl-KTSAVLQ↓SGFRKM-E(Edans)-NH₂) and buffer composed of 20 mM HEPES, 120 mM NaCl, 0.4 mM EDTA, 4 mM DTT, 20% glycerol, pH 7.0, was used for the inhibition assay. In the fluorescence resonance energy transfer (FRET)-based cleavage assay⁷⁸, the fluorescence signal of the Edans generated as a result of the cleavage of the substrate by Mpro was observed at an emission wavelength of 460 nm with excitation at 360 nm using a Tecan Spark multimode microplate reader. The compound was diluted with 100% DMSO to prepare a stock solution. The IC₅₀ of CIM-834 was determined after incubation of 0.5 µM of SARS-CoV-2 Mpro and CIM-834 at various concentrations from 0 to 500 µM in reaction buffer at 37 °C for 10 min. At the end, the FRET substrate at a final concentration of 50 µM was added to each well at a final total volume of 100 µl to initiate the reaction. The IC₅₀ value was calculated with the help of the GraphPad Prism 9.2.0 software. Inhibitory activity of the compound was measured in triplicates and data are presented as mean ± s.d. Two positive control assays (nirmatrelvir and ensitrelvir) against Mpro were also performed simultaneously.

Determination of inhibition of PLpro by CIM-834

Activity and inhibition were determined by FRET, performed in black 384-well HiBase non-binding plates (Greiner Bio One, 784900), as previously described⁷⁵. In brief, increasing concentrations of inhibitor were incubated with purified PLpro protein (55 nM) in the presence of 5 μM of a fluorescent synthetic peptide (Dabcyl-FTLKGG↓APTK-Edans, Genscript) in HEPES buffer (20 mM, pH 6.5) containing 120 mM NaCl, 0.4 mM EDTA, 4 mM DTT and 10% glycerol. The final concentration of DMSO was adjusted to 0.5%. Cleavage of the fluorogenic peptide separates the Edans/Dabcyl fluorophore–quencher pair. The time courses of the enzymatic reaction were followed for 40 min by monitoring the increase in fluorescence emission at 493 nm (excitation at 335 nm) using a Tecan Safire2 fluorimeter. Enzymatic activities were estimated by calculating the slope of the linear part of the reaction curve and were normalized with respect to the activity measured in the absence of inhibitor.

RNA synthesis inhibition and IC₅₀ determination

The compound concentrations leading to 50% inhibition of polymerase-mediated RNA synthesis was determined in IC₅₀ buffer (50 mM HEPES, pH 8.0, 10 mM KCl, 2 mM MnCl₂, 2 mM MgCl₂ and 10 mM DTT) containing seven increasing concentrations of compound (from 1 µM to 100 µM) and 150 nM of nsp12 (ref. ⁷⁶) in complex with 450 nM nsp8 and 450 nM nsp8L7 (ref. ⁷⁷).

Reactions were done in a volume of 40 µl on a 96-well Nunc plate. All experiments were robotized by using a BioMek I5 automate (Beckman). Then, 2 µl of each diluted compound in 100% DMSO was added in wells to the chosen concentration (5% DMSO final concentration). For each assay, the (nsp8L7 + nsp8) mix was distributed in wells after an 8-min incubation at room temperature to pre-form the active complex. Nsp12 was then added to wells and incubated for 8 min. Reactions were started by adding the UTP + poly(A) template mix and were incubated at 30 °C for 20 min, using 350 nM of poly(A) template and 750 µM of UTP final concentration.

Reaction assays were stopped by adding 20 µl EDTA (100 mM). Positive and negative controls consisted of a reaction mix with 5% DMSO (final concentration) or EDTA (100 mM) instead of compounds, respectively. Reaction mixes were then transferred to a Greiner plate using a Biomek I5 automate (Beckman). PicoGreen fluorescent reagent was diluted to 1/800 in TE buffer, according to the manufacturer’s instructions, and 60 μl of reagent was distributed into each well of the Greiner plate. The plate was incubated for 5 min in the dark at room temperature and the fluorescence signal was then read at 480 nm (excitation) and 530 nm (emission) using a Tecan Safire2 and/or a ClarioStar.

The IC₅₀ was determined using this equation: percentage of active enzyme = 100/(1 + I²/IC₅₀), where I is the concentration of inhibitor and 100% of activity is the fluorescence intensity without the inhibitor. The IC₅₀ was determined from curve fitting using Prism software. For each measurement, results were obtained in triplicate.

N7 and 2′O-MTase assay

Purified SARS-CoV-2 nsp14 (N7-MTase) and nsp10/nsp16 (2′O-MTase) were incubated in a reaction mixture containing 40 mM Tris-HCl (pH 8.0), 1 mM DTT, 1 mM MgCl₂, 2 μM SAM and 0.1 μM 3H-SAM (Perkin Elmer), in the presence of either 0.7 μM GpppAC4 or 7m GpppAC4 synthetic RNA⁷⁹. The compounds were suspended in 50% DMSO (2.5% final DMSO concentration). Reactions were carried out at 30 °C and stopped after 30 min by ten-fold dilution in ice-cold water. Samples were transferred to a diethylaminoethyl filter mat (Perkin Elmer) using a Filtermat Harvester (Packard Instruments). The RNA-retaining mats were washed twice with 10 mM ammonium formate (pH 8.0), twice with water and once with ethanol. They were soaked with scintillation fluid (Perkin Elmer), and 3H-methyl transfer to the RNA substrates was determined using a Wallac MicroBeta TriLux liquid scintillation counter (Perkin Elmer).

Metabolic stability in mouse, hamster and human liver microsomes

Compound microsomal stability was determined using mouse microsomal fractions (Gibco; final protein concentration, 0.5 mg ml⁻¹), with hamster microsomal fractions (Xenotech; final protein concentration, 0.5 mg ml⁻¹) and with human microsomal fractions (Xenotech; final protein concentration, 0.5 mg ml⁻¹) at a substrate concentration of 1 μM with or without NADPH (final concentration, 1 mM). Incubations were done at 37 °C for 120 min in duplicates. At various time points (5, 15, 30, 60 and 120 min), 25 μl of the reaction mixture was sampled and quenched in 300 μl of acetonitrile containing internal standard. The resulting suspension was vortexed and centrifuged, and the supernatant was diluted with water (two-fold). The resulting solution was analysed by liquid chromatography with tandem mass spectrometry to determine the half-life of the compound.

Statistics and reproducibility

All statistical comparisons in the study were done in GraphPad Prism (v.6.07, v.8.3.0, v.9.2.0, v.10.1.2 and v.10.2.3) and the statistical tests used are indicated in the figure legends. The 2D EMs for quantification of DMVs were acquired from three independent preparations. Tomography was acquired on a selection of technical duplicates of the samples previously characterized in 2D, with the exception of the HPF samples that were prepared in duplicates in one preparation.

Preparation of figures

All graphs were prepared using GraphPad Prism (v.6.07, v.8.3.0, v.9.2.0, v.10.1.2 and v.10.2.3). Figures and schemes were created using BioRender.com and Adobe Illustrator 28.1 (Windows).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.