Cell lines

Human embryonic kidney HEK-293T cells (American Type Culture Collection (ATCC) CRL-3216) and the derivative expressing hACE2, that is, 293T/hACE2.cl22 (ref. 39), Caco-2 cells (ATCC HTB-37), human hepatoma-derived Huh-7.5 cells48, Vero E6 cells and a derivative expressing TMPRSS2 (ref. 49) and A549 cells (adenocarcinomic human alveolar basal epithelial cells) were maintained in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% foetal bovine serum (Sigma F8067) and gentamycin (Gibco). All cell lines used in this study were monitored periodically to ensure the absence of retroviral contamination and mycoplasma.

Generation of hACE2-specific human mAbs

ACE2-binding mAbs with human variable regions were generated using AlivaMab Mouse (Ablexis) transgenic mouse strains. Specifically, AlivaMab mice were immunized subcutaneously with recombinant human ACE2 extracellular domain (1–740aa) fused to human IgG1 Fc and/or a polyhistidine tag. Mice were immunized at 3 week intervals at least four times, using 10 µg subcutaneous injections at different sites. Mice with sera exhibiting SARS-CoV-2 infection inhibiting activity in the pseudotype virus assay were acute boosted before fusion of splenocytes with SP2/0 cells for hybridoma generation. Hybridomas expressing antibodies that bound to hACE2 were identified by ELISA using plates coated with purified hACE2. Anti-hACE2 hybridoma supernatants that contained antibodies with human ACE2 binding activity were tested for inhibition of SARS-CoV-2 pseudotyped virus infection of Huh-7.5 cells. This analysis indicated ten hybridoma antibodies that were positive for binding and potently inhibited SARS-CoV-2 pseudotype virus infection that were chosen for hybridoma cell subcloning and expansion. Antibodies were purified from these hybridoma culture supernatants and were further tested for potency ranking in the SARS-CoV-2 pseudotype virus inhibition assay.

Human mAb expression plasmids

DNA encoding the variable regions of the heavy (VH) and light (VL) from hybridomas was PCR amplified from DNA extracted from the hybridoma cell lines. For the 05B04 and 05D06 antibodies, the DNA sequences encoding VH and VL were human codon-optimized using GenSmart Codon Optimization, synthesized by Integrated DNA Technologies (IDT). For each mAb, DNA encoding VH were fused to complementary DNA encoding the Fc domain of human IgG1, in which Fc domain was modified to include the substitutions at L234 L235 (LALA) that abolish FcR–gamma interaction, and substitutions at M428 N434 (LS) that enhance interaction with the neonatal Fc receptor to prolong mAb half-life in humans23,24,25. To construct the expression plasmids for heavy and light chain antibody expression50, PCR amplicons or synthetic DNA encoding variable regions were subcloned using AgeI and XhoI (for LC), or AgeI and SalI (for HC), respectively, using NEBuilder HiFi DNA Assembly.

Human ACE2 and sarbecovirus spike expression plasmids

Plasmids expressing the spike proteins from SARS-CoV, SARS-CoV-2 (Wuhan-hu-1, Beta (B.1.351), Delta (B.1.617.2) and Omicron (B.1.1.529) variants), the pangolin (Manis javanica) coronaviruses from Guangdong, China (pCoV-GD) and Guanxi, China (pCoV-GX) were previously described14,39,51. Human codon-optimized cDNAs encoding spike proteins from the rufous horseshoe bat (Rhinolophus sinicus) coronaviruses Rs4231 and Rs7327 were generated using GenSmart Codon Optimization, synthesized by IDT as gBlocks, and inserted into the pCR3.1 expression vector using NheI and XbaI and NEBuilder HiFi DNA Assembly.

To construct the plasmids expressing NanoLuc-fused to conformationally stabilized versions of the SARS-CoV-2 Wuhan-hu-1 or Omicron variant spike proteins, the HexaPro (6P) modified cDNA was fused at its C-terminus with DNA encoding a trimerization domain, a GGSGG spacer sequence, NanoLuc luciferase (NLuc), a human rhinovirus 3C protease cleavage site and a polyhistidine tag (8XHis). This cDNA, termed (S-6P-NanoLuc), was inserted into the pCR3.1 expression vector.

To construct a plasmid expressing catalytically inactive His-tagged or IgG1 Fc-fused soluble ectodomain of hACE2 (1–740aa), H374N and H378N substitutions were introduced by overlap extension PCR into a hACE2 cDNA and a His-tag was fused to its C-terminus, and the purified PCR product was inserted into the pCAGGS expression vector (hACE2-1–740aa-His). To construct the expression plasmid encoding hACE2 (1–740aa)-NanoLuc-8XHis, the hACE2 1-740aa and NanoLuc-8XHis fragments were PCR amplified using hACE2-1–740aa-His and S-6P-NanoLuc as templates, respectively, followed by Gibson assembly and insertion into the pCR3.1 expression vector. Oligonucleotide sequences used during molecular construction are provided in Supplementary Data 1.

Protein and antibody expression and purification

To express the monomeric, His-tagged human ACE2 extracellular domain (residues 1–740) used as immunogen, His-tagged hACE2 (1–740aa)-NanoLuc or soluble hACE2 used for cryo-EM studies (residues 1–614), Expi293 cells were transfected with the expression plasmid hACE2-1–740aa-8xHis, hACE2(1–740aa)-NanoLuc-8XHis, or hACE2-1–614aa-8xHis using ExpiFectamine 293 (Thermo Fisher Scientific), respectively. Four days later, the supernatant was filtered with 0.22 μm membrane filter and loaded on Ni-NTA agarose (Qiagen) and, after washing, hACE2 proteins were eluted with 200 mM imidazole in phosphate-buffered saline (PBS). For cryo-EM structural studies, a subsequent size-exclusion chromatography step on a Superdex 200 10/300 column (Cytiva) was performed against PBS, and fractions corresponding to monomeric soluble hACE2 were pooled and stored at 4 °C. Dimeric, Fc-fused hACE2 extracellular domain was also expressed in Expi293 cells in the same way. The secreted proteins in supernatant were first incubated with Protein G Sepharose 4 Fast Flow overnight at 4 °C, loaded into column and, after washing, eluted with 0.1 M glycine, pH 2.9 into tubes containing 1/10th volume of 1 M Tris, pH 8.0.

To express mAbs, Expi293 cells were transfected with the corresponding light chain and heavy chain expression plasmids at the ratio of 1:1 using ExpiFectamine 293. Four days later, the mAbs in the supernatant were purified through Protein G Sepharose 4 Fast Flow and eluted with 0.1 M glycine, pH 2.9 as described above.

To express S-6P-NanoLuc proteins, Expi293 cells were transfected with S-6P-NanoLuc expression plasmids that included for the original Wuhan-hu-1 or Omicron spike variants using ExpiFectamine 293. Three days later, the supernatant was collected and loaded on Ni-NTA agarose and, after thorough wash, S-6P-NanoLuc proteins were released after HRV 3C protease (TaKaRa) treatment overnight at 4 °C.

All recombinant proteins, including purified mAbs, were dialysed against PBS before used in further experiments.

Sarbecovirus spike-bearing pseudotypes and infectivity inhibition assay

To generate HIV-1 virions pseudotyped with sarbecovirus spikes, including SARS-CoV, SARS-CoV-2 (Wuhan-hu-1, Beta, Delta and Omicron variants), pangolin coronavirus pCoV-GD, pangolin coronavirus pCoV-GX and bat coronaviruses SARSr-CoV 4231 and SARSr-CoV 7327, ten million 293T cells in a 15 cm dish were transfected with 25 μg of an HIV-1 envelope-deficient proviral plasmid expressing NanoLuc along with 7.5 μg of spike expression plasmids, in which the C terminal 19aa was truncated (Δ19) (ref. 39). Cells were washed twice with PBS the next morning, and virions were collected at 48 h post transfection, filtered (0.22 μm) and purified by Lenti-X Concentrator (TaKaRa). To measure the infectivity, viral stocks were twofold serially diluted and added to Huh-7.5 cells39, which express hACE2 (ref. 52) and are permissive to SARS-CoV-2 (ref. 53), in 96-well plates seeded 1 day before infection. Cells were then collected at 48 h post infection for measuring NanoLuc activity using the Nano-Glo Luciferase Assay System and GloMax Navigator Microplate Luminometer (Promega).

To measure antiviral activity, the hACE2 mAbs were fourfold serially diluted (beginning with 2 μg ml−1) in 96-well plates over seven dilutions and incubated with Huh-7.5 target cells for 1 h at 37 °C. Thereafter, the mAb-treated Huh-7.5 cells were infected with sarbecovirus spike pseudotyped viruses. Cells were collected 48 h post infection and NanoLuc luciferase activity measured in infected cells as described above.

SARS-CoV-2 virus stocks and titration

SARS-CoV-2 strains USA-WA1/2020 and the Omicron variant B.1.1.529 were obtained from BEI Resources (catalogue nos. NR-52281 and NR-56461, respectively). The original virus (WA1/2020) was amplified in Caco-2 cells, which were infected at a multiplicity of infection of 0.05 PFU per cell and incubated for 5 days at 37 °C. The Omicron variant B.1.1.529 was amplified in Vero E6 cells (ATCC) that were engineered to stably express TMPRSS2. Vero-TMPRSS2 cells were infected at a multiplicity of infection of 0.05 PFU per cell and incubated for 4 days at 33 °C. Virus-containing supernatants were subsequently collected, clarified by centrifugation (3,000g × 10 min), filtered with a 0.22 μm membrane and stored at −80 °C. To measure the virus stock titres by standard plaque assay, 500 µl of serial tenfold virus dilutions in Opti-MEM were used to infect 4 × 105 Vero E6 cells (from Ralph Baric) in six-well plates. After 1.5 h adsorption, the virus inoculum was removed, and cells were overlaid with DMEM containing 10% FBS with 1.2% microcrystalline cellulose (Avicel). Cells were incubated for 4 days at 33 °C, followed by fixation with 7% formaldehyde and crystal violet staining for plaque enumeration. All SARS-CoV-2 experiments were performed in a biosafety level 3 laboratory.

SARS-CoV-2 inhibition assays

The day before infection, Vero E6/Huh-7.5 cells were seeded at 1 × 104 cells per well into 96-well plates. Antibodies were serially diluted in DMEM, mixed with target cells and incubated for 60 min at 37 °C. Subsequently a constant amount of SARS-CoV-2 was added to achieve 40–50% virus-positive cells. Cells were fixed 18–24 h after infection by adding an equal volume of 7% formaldehyde to the wells, followed by permeabilization with 0.1% Triton X-100 for 10 min. After extensive washing, cells were incubated for 1–2 h at room temperature with blocking solution of 5% goat serum in PBS (005-000-121; Jackson ImmunoResearch). A rabbit polyclonal anti-SARS-CoV-2 nucleocapsid antibody (GTX135357; GeneTex) was added to the cells at 1:1,000 dilution in blocking solution and incubated at 4 °C overnight. A goat anti-rabbit Alexa Fluor 594 (A-11012; Life Technologies) at a dilution of 1:2,500 was used as a secondary antibody. Nuclei were stained with Hoechst 33342 (62249; Thermo Scientific) at a concentration of 1 μg ml−1. Images were acquired with a fluorescence microscope and analysed using ImageXpress Micro XLS and MetaXpress software (Molecular Devices).

MAb binding measurements using SPR

SPR experiments were performed using a Biacore 8K instrument (GE Healthcare). Human mAbs 2G7A1, 05B04, 05H02 and hybrid mAb 05B04LC/05D06HC were captured with a Series S Sensor ship Protein G (Cytiva) at a concentration of 20 nM at a flow rate of 10 μl min−1 for 60 s. Flow cell 1 was kept empty and used as a negative control. A concentration series of His-tagged hACE2 1–740aa proteins (fourfold dilutions from a maximum concentration of 500 nM) was injected at 30 µl min−1 for 240 s followed by a dissociation phase of 2,400 s at a flow rate of 30 µl min−1. Binding reactions were allowed to reach equilibrium, and KD values were calculated from the ratio of association and dissociation constants (KD = kd/ka), which were derived using a 1:1 binding model that was globally fit to all curves in a dataset. Flow cells were regenerated with 10 mM glycine pH 1.5 at a flow rate of 30 μl min−1 for 30 s.

Spike hACE2 binding and binding inhibition assay

To evaluate whether anti-hACE2 mAbs inhibit spike–hACE2 interaction, 400 ng of each mAb (2G7A1, 05B04, 05H02 or 05B04LC/05D06HC) and a twofold serial dilution thereof over seven dilutions were mixed with 100 ng of His-tagged hACE2 1-740aa proteins in PBS containing 2% bovine serum albumin. After 1 h incubation at 4 °C, the mixture was incubated with 10 ng of Wuhan-1 or Omicron S-6P-NanoLuc proteins for 1 h at 4 °C. Then 1 μl of Dynabeads His-Tag Isolation and Pulldown magnetic beads (Thermo Fisher Scientific) was added into each well. After 1 h incubation at 4 °C, the beads were washed three times and bound NanoLuc activity measured using Nano-Glo Luciferase Assay System and a GloMax Navigator Microplate Luminometer (Promega).

Cryo-EM sample preparation, data collection and structure refinement

Purified 05B04 Fab was mixed with monomeric soluble hACE2 (residues 1–614) at an equimolar concentration for 1 h at room temperature. Fab-hACE2 complex was concentrated to 4 mg ml−1 and deposited on a freshly glow discharged 300 mesh, R1.2/1.3 Quantifoil grid (Electron Microscopy Sciences). Samples were vitrified in 100% liquid ethane using a Mark IV Vitrobot (Thermo Fisher) after blotting at room temperature and 100% humidity for 3 s with Grade 595 filter paper (Ted Pella).

Single-particle cryo-EM data were collected on a Titan Krios transmission electron microscope equipped with a Gatan K3 direct detector, operating at 300 kV and controlled using SerialEm automated data collection software54. A total dose of 60 e− Å−2 was accumulated on each movie comprising 40 frames with a pixel size of 0.867 Å and a defocus range of −1.0 to −2.6 µm. Data processing was carried out as previously described49 using cryoSPARC v3.2 (ref. 55) and summarized in Supplementary Table 3. Briefly, 6,599 movies were patch motion corrected for beam-induced motion and contrast transfer function (CTF) estimates were performed on non-dose weighted micrographs in cryoSPARC v3.2. After curation to remove images with poor CTF fits and ice contamination, particles were automatically picked using blob picker, extracted 4x-binned and 2D classified. Class averages that showed secondary structure features were pooled and used to generate ab initio volumes (n = 4). Particles corresponding to volumes that resembled the Fab–sACE2 complex were re-extracted at 2x-bin, pooled and subsequently 3D classified in cryoSPARC. Particles corresponding to the 3D classes with well-defined features were pooled, unbinned and subjected to CTF and non-uniform refinement (C1 symmetry) in cryoSPARC. This resulted in a global map with an estimated resolution of 3.5 Å based on gold standard FSC calculations. A mask was generated to exclude Fab constant domains from focused refinement, which yielded an overall map resolution for the locally refined volume of 3.3 Å as calculated using the gold-standard Fourier shell correlation of 0.143 criterion.

Structure modelling, refinement and analysis

Coordinates for initial complexes were generated by docking individual chains from reference structures into cryo-EM density using UCSF Chimera56. An initial model for the 05B04 Fab–hACE2 structure was generated from coordinates of Protein Data Bank (PDB) 7S0B (chain A: 05B04 heavy chain), PDB 7DPM (chain B: 05B04 light chain) and PDB 6VW1 (chain A: hACE2). Models were refined using one round of rigid body refinement with morphing followed by real space refinement in Phenix57. Sequence-updated models were built manually in Coot58 and then refined using iterative rounds of refinement in Coot and Phenix. Glycans were modelled at potential N-linked glycosylation sites in Coot. Validation of model coordinates was performed using MolProbity59.

Structure figures were made with UCSF ChimeraX60. Local resolution maps were calculated using cryoSPARC v3.2. BSAs were calculated using PDBePISA61 and a 1.4 Å probe. Potential hydrogen bonds were assigned as interactions that were <4.0 Å and with A–D–H angle >90°. Potential van der Waals interactions between atoms were assigned as interactions that were <4.0 Å. Hydrogen bond and van der Waals interaction assignments are tentative due to resolution limitations. hACE2 epitope residues were defined as residues containing atom(s) within 4.5 Å of a 05B04 Fab atom.

hACE2 enzymatic activity assay

To measure the effect of anti-hACE2 mAbs on the catalytic activity of hACE2, various concentrations of 2G7A1, 05B04, 05H02 and 05B04LC/05D06HC (2 μg ml−1, 10 μg ml−1 and 50 μg ml−1) were mixed with hACE2 at 0.2 μg ml−1. Human ACE2 enzymatic activity was then measured using the ACE2 Inhibitor Screening Assay Kit (BPS Bioscience) following the manufacturer’s instructions. The intensity of the fluorescent product of the hACE2 reaction product was detected at 555 nm/585 nm (excitation/emission) with Clariostar Plus Microplate Reader (BMG Labtech). MLN-4760 (Sigma, #5306160001) served as a positive control ACE2 inhibitor.

Flow cytometric analysis of cell surface hACE2 binding by human anti-hACE2 mAbs

To evaluate the ability of anti-hACE2 mAbs to bind to cell surface ACE2 by flow cytometry, A549 cells (human alveolar basal epithelial cells) were engineered to express ACE2. The hACE2 or macaque (mac)ACE2-expressing cells were detached from plates with 10 mM ethylenediaminetetraacetic acid in PBS and 105 cells incubated in the absence or the presence of human anti-hACE2 mAbs (2 μg ml−1) for 2 h at 4 °C. After washing, the cells were incubated with Alexa Fluor 488 conjugated goat anti-human IgG (Thermo Fisher Scientific). Flow cytometry was performed using Attune NxT Acoustic Focusing Cytometer (Thermo Fisher Scientific, V3.1.2 and V5.1.0). The same procedure was applied to parental, unmodified A549 cells as a negative control for non-specific cell surface binding.

To assess the impact of hACE2 variation on the interaction between hACE2 and human anti-hACE2 antibodies, human 293T cells were transfected with plasmids expressing 18 hACE2 variants, including S19P, K26R, K26E, T27A, E35K, E37K, K68E, M82I, P84T, E329G, D355N, P389H, P426A, D427Y, R559S, S692P, N720D, L731F and wild-type hACE2, respectively, along a GFP-expressing plasmid. Cells transfected with plasmids carrying GFP only served as the negative control. Two days later, cells were collected and stained with anti-hACE2 antibodies followed by incubation with Alexa Fluor 647 conjugated goat anti-human IgG (Thermo Fisher Scientific). GFP-positive populations were gated to measure anti-hACE2 mAbs binding.

hACE2 internalization assay

To determine whether the anti-hACE2 mAbs induced hACE2 internalization, live A549 cells expressing hACE2 receptor with an HA-epitope tag appended to its intracellular C-terminus were incubated with anti-hACE2 mAbs (1 μg ml−1) for 1 h at 37 °C. Then, cells were fixed with 4% paraformaldehyde/PBS, treated with 10 mM glycine and permeabilized with 0.1% Triton X-100. Total hACE2-HA was then detected with mouse anti-HA.11 antibody (BioLegend, cat 901503, clone 16B12, 1 μg ml−1). The internalization of the hACE2-HA protein and anti-hACE2 mAbs was then evaluated by staining with goat anti-mouse Alexa Fluor 594 (to detect the HA-tagged hACE2) (1/500 dilution) and/or goat anti-human Alexa Fluor 488 antibodies (to detect the anti-hACE2 mAbs) (1/500 dilution) (Thermo Fisher Scientific). As a control, anti-CD44 antibody conjugated with FITC (clone F10-44-2 from abcam, cat. no. ab30405, 1/20 dilution) was used. Wheat germ agglutinin conjugated with Alexa Fluor 594 (Thermo Fisher Scientific, cat. no. W11262, 5 μg ml−1) were used to visualize the cell surface. Images were captured using a DeltaVision OMX SR imaging system (GE Healthcare).

Analysis of anti-hACE2 mAbs pharmacokinetics in mice

Six-week-old hACE2-knock-in female mice, in which human ACE2 cDNA replaces the endogenous mouse ACE2 sequences, were obtained from Jackson Labs (B6.129S2(Cg)-Ace2tm1(ACE2)Dwnt/J, strain 035000). The mice were housed at a temperature of 22 °C and a humidity of 30–70% under a 12 h–12 h light–dark cycle with ad libitum access to food and water. After acclimatization for 2 weeks, these mice received subcutaneous injections of 250 μg of human anti-hACE2 mAbs per mouse (n = 5). Mice were bled on day 0, day 1, day 3, day 7 and day 14 with blood collected into Microvette CB 300 Serum (Sarstedt).

Serially diluted mouse plasma (fivefold serial dilution over four dilutions from a maximum volume of 0.5 μl) was diluted in PBS buffer containing 2% bovine serum albumin and mixed with 30 ng of His-tagged hACE2(1-740aa)-NanoLuc protein. After 1 h incubation at 4 °C, the mixture was incubated with 3 μl of Dynabeads Protein G magnetic beads (Thermo Fisher Scientific). After 1 h rotation at 4 °C, the beads were washed three times and bound NanoLuc activity was measured using Nano-Glo Luciferase Assay System and a GloMax Navigator Microplate Luminometer (Promega). To construct standard calibration curves for measurement of mAb levels in plasma, 100 ng of mAbs (2G7A1, 05B04, 05H02 or 05B04LC/05D06HC) were fivefold serially diluted over seven dilutions and mixed with hACE2(1-740aa)-NanoLuc proteins. MAb:hACE2(1-740aa)-NanoLuc complexes were captured and quantified in parallel with those formed using the plasma samples from mAb-infused mice.

All of the animal procedures and experiments were performed by following protocols approved by the Rockefeller University Institutional Animal Care and Use Committee.

SARS-CoV-2 challenge experiments in hACE2-expressing mice

Six-week-old hACE2-knock-in female mice, in which human ACE2 cDNA replaces the endogenous mouse ACE2 sequences, were obtained from Jackson Labs (B6.129S2(Cg)-Ace2tm1(ACE2)Dwnt/J, strain 035000). The mice were housed at a temperature of 22 °C and a humidity of 30–70% under a 12 h–12 h light–dark cycle with ad libitum access to food and water. After acclimatization for 2 weeks, the mice (five mice per treatment group) were injected retro-orbitally with 250 μg (equivalent to ~12.5 mg kg−1) of anti-hACE2 mAbs. At 2 days after mAb injection, mice were challenged intranasally with SARS-CoV-2, USA_WA/2020 P3, 2 × 105 PFU per mouse (virus titres determined on Vero E6 cells). At 3 days after infection, mouse lungs were dissected and homogenized in TRIzol. Chloroform was added to induce phase separation. Then after centrifugation, RNA in the aqueous phase was precipitated with isopropanol and, after wash with ice-cold 75% ethanol, dissolved in nuclease-free water. The number of viral genomes per microgram of total lung RNA was measured by qRT–PCR, using Power SYBR Green RNA-to-CT 1-Step Kit (Thermo Fisher Scientific) on StepOne Plus Real-Time PCR system (Applied Biosystems). The primers used target RNA sequences encoding the nucleocapsid protein: 2019-nCoV_N1-F: 5′-GACCCCAAAATCAGCGAAAT-3′ and 2019-nCoV_N1-R: 5′-TCTGGTTACTGCCAGTTGAATCTG-3′. The standard was obtained from IDT (2019-nCoV_N_Positive Control 10006625).

Infection experiments in mice, and all procedures involved therein, were approved by the Rockefeller University Institutional Animal Care and Use Committee.

Selection of rVSV/SARS-CoV-2 variants in the presence of antibodies

HEK-293T/ACE2cl.22 cells were incubated with anti-hACE2 mAbs at concentrations of 50, 10 or 1 μg ml−1 for 1 h at 37 °C. Viral populations (rVSV/SARS-CoV-2/GFP2E1) containing 1 × 106 infectious units were then added to the cells. As a control, the rVSV/SARS-CoV-2/GFP2E1 populations were incubated with the spike-specific mAb C144 (10 or 1 μg ml−1) for 1 h before infection. After 24 h, the medium was replaced with fresh medium containing the respective concentration of each mAb. After another 24 h, the virus-containing supernatant was filtered (0.22 μm), and 100 μl of the supernatant was added to HEK-293T/ACE2cl.22 cells, which had been pre-incubated with anti-ACE2 mAb (50, 10 or 1 μg ml−1) for 1 h at 37 °C as indicated above. Two passages at 50 μg ml−1 were or four passages at 10 and 1 μg ml−1 were carried out. RNA was extracted from 100 μl of filtered p4 supernatant and reverse transcribed using the SuperScript VILO cDNA Synthesis Kit (Thermo Fisher Scientific). Sequences encoding the extracellular domain of the spike protein were amplified using KOD Xtreme Hot Start Polymerase (Sigma-Aldrich, 719753). These PCR products were sequenced to identify escape substitutions.

Reporting summary

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