NS1 (ATCC TIB-18TM) was used for hybridoma fusion. ExpiSf9 [Thermo Fisher Scientific Inc.; Cat# A35243], Sf9 [Thermo Fisher Scientific Inc.; Cat# 11496015] and Expi293F™ [Thermo Fisher Scientific Inc.; Cat# A14528]) were used in the generation of recombinant proteins. BHK21 (ATCC CCL-10™) was used for the generation of pseudovirus. HEK293T-ACE2 [GeneCopeia Cat# SL221]) was used for pseudovirus neutralization test (pVNT). VeroE6/TMPRSS2 cells (JCRB cell bank of Okayama University; Cat# JCRB1819) was used for conventional live virus neutralization test (cVNT). All cell lines used in this study were routinely tested for mycoplasma and found to be mycoplasma-free.

Six-week-old female BALB/c mice (n=2) were immunized intramuscularly with S protein trimer of ancestral SARS-CoV-2 at 10 μg/mice (Acro Biosystems, Newark, USA; Cat# SPN-C52H9), which was mixed with an equal volume of QuickAntibody adjuvant (Biodragon Immunotechnologies Co., Ltd. Beijing, China; Cat# KX0210042), on day 0 and day 21, and then boosted intraperitoneally with 100 μg of recombinant RBD (Acro Biosystems, Newark, USA; Cat# SPD-C52H3) of ancestral SARS-CoV-2 at Day 35. On day 38, spleen was harvested. Hybridoma cells were produced by fusing splenic cells from immunized mice with myeloma cells and were selected using hypoxanthine-aminopterin-thymidine (HAT) medium. Hybridoma clones were screened for the presence of mAbs against SARS-CoV-2 by indirect enzyme immunoassay (EIA) with recombinant ancestral S protein as coated antigens. Cell subclones secreting the specific mouse mAbs were further selected by limiting dilution of positive hybridoma cells. Hybridoma cells specific for SARS-CoV-2 S protein were then injected intraperitoneally in mice, and the ascitic fluid were collected 10 days after inoculation. The mAb in the ascitic fluid was purified by precipitating with caprylic acid and ammonium sulfate. The mAb isotype was determined with the Mouse Monoclonal Antibody Isotyping Kit (Biodragon Immunotechnologies Co. Beijing, China; Cat# BF16002). The antibody titer was determined as the lowest antibody concentration that produced a positive test reaction by EIA against recombinant RBD protein.

The binding affinity of mAbs was determined using the biolayer interferometry (BLI) (Gator™ System, Gator Bio, Suzhou, China). Anti-mouse IgG Fc biosensors were pre-equilibrated in Q buffer and captured 10 μg/ml of mAbs for 240 s. After equilibrating in Q buffer for another 60 s, the sensors were incubated with serially-diluted recombinant ancestral RBD protein or Q buffer (as control) for 180 s, and finally dissociated in Q buffer for 300 s. Gator™ System Analysis software was used to calculate association (Ka) and dissociation (Kd) kinetic rate constants, and equilibrium dissociation constant (KD) of these mAbs.

We next determined the epitopes of these anti-RBD antibodies using a competition assay by BLI as described (Dong et al., 2020). Anti-His biosensors were pre-equilibrated in Q buffer and captured 10 μg/ml of recombinant His-tagged RBD protein for 180 s, followed by binding with 10 μg/mL of the primary antibody for 300 s and then the secondary antibody (or buffer control) for another 300 s. The wavelength shift of the primary and the secondary antibody on the probe surface were recorded and the competition rates between these two antibodies were calculated as following: % Competition = (1- (Shift_1st – Shift_2nd)/(Shift_1st - Shift_Control)) × 100. The graph was generated using the GraphPad Prism software.

Recombinant RBDs (residues 306-543) of ancestral strain and variant SARS-CoV-2 spike proteins were expressed and purified in insect cells as we described previously (Chen et al., 2022b). The ancestral RBD were designed based on the reference sequence Wuhan-Hu-1 (GenBank ID YP_009724390.1). The RBD mutants are listed in Supplementary Table S1 . Briefly, RBD gene sequences were baculovirus-codon-optimized and cloned into pFast dual baculovirus expression vector (Thermo Fisher Scientific, Waltham, MA, USA Cat # 10712024). The constructs were fused with an N-terminal gp67 signal peptide and C-terminal 6× His tag for secretion and purification. A recombinant bacmid DNA was generated using the Bac-to-Bac system (Thermo Fisher Scientific, Waltham, MA, USA). Baculovirus was produced by transfecting purified bacmid DNA into Sf9 cells using Cellfectin (Thermo Fisher Scientific, Waltham, MA, USA; Cat# 10362100), and subsequently used to infect ExpiSf9 cell suspension culture at a multiplicity of infection of 1 to 10. Infected ExpiSf9 cells were incubated at 27.5 °C with shaking at 125 r.p.m. for 96 hours for protein expression. The supernatant was collected and then concentrated using a 10 kDa MW cutoff Lab-scale TFF System (Millipore, Burlington, USA). The RBD protein was purified by Ni-NTA purification system, followed by size exclusion chromatography, and buffer exchanged into 1× phosphate buffered saline (PBS) pH 7.4. The concentration of purified RBD was determined by using the Bradford Assay Kit (Bio-Rad, California, USA; Cat# 5000002) according to the manufacturer’s instructions.

EIA for anti-S and anti-RBD mAbs was performed as we described previously with modifications (To et al., 2021). Briefly, 96-well Immuno plates (Nunc, Roskilde, Denmark; Cat# 243656) were coated with 100 μL/well (0.1 μg/well) of His-tagged SARS-CoV-2 S or RBD in 0.05 M carbonate bicarbonate buffer (pH 9.6) overnight at 4°C and then followed by incubation with a blocking buffer. After blocking, mAbs were diluted in a series of 2-fold dilution from 1 μg/ml. Then 100 μl diluted mAbs was added to the wells and incubated at 37°C for 1 hour. The attached mouse IgG was detected using horseradish-peroxidase-conjugated goat anti-mouse IgG antibody (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA; Cat# 31430). The reaction was developed by adding diluted 3,3’,5,5’-tetramethylbenzidine single solution and stopped with 0.3 N H₂SO₄. The optical density (OD) was read at 450 and 620 nm.

The neutralizing activity of the ten selected mouse mAbs was evaluated by detecting their inhibition for the binding of recombinant ancestral RBD protein to its human angiotensin-converting enzyme 2 (hACE2) receptor (Acro Biosystems, Newark, USA; Cat # AC2-H52H8) by BLI with Gator™ Label-Free Analysis System. All experiments were carried out at 25°C. Anti-His biosensors were first pre-equilibrated in Q buffer containing PBS (10 mM pH 7.4), 0.02% Tween-20, and 0.2% bovine serum albumin (BSA) for 300 s, and 1 μg/ml recombinant ancestral RBD of SARS-CoV-2 S protein was loaded onto these probes for 200 s. After equilibrating in Q buffer for another 30 s, the probes were incubated with serially diluted anti-RBD mAbs or dilution buffer (as control) for 180 s and followed by binding to 2 μg/ml of recombinant hACE2 protein for 180 s. The wavelength shift (indicative of binding signal) of hACE2 protein on the probe surface was recorded and the inhibition efficiency of the anti-RBD mAbs were calculated as follows: % Inhibition= (1- Shift_mAbs/Shift_Control) × 100. The IC₅₀ was calculated using log(inhibitor) vs. response (3 parameters).

Viral culture and conventional live virus neutralization test (cVNT) were performed in the Biosafety Level 3 facility at the University of Hong Kong as we described previously (Chu et al., 2020; Lu et al., 2022). In order to avoid mutations, viral culture was conducted using VeroE6/TMPRSS2 cells. Briefly, mouse mAbs were serially diluted in 2-folds starting from 200 µg/ml, while humanized WKS13 mAb was serially-diluted in 2-folds starting from 30 μg/ml. Then, 100 TCID₅₀ of virus isolates were mixed with mAb and incubated for 1 hour. Finally, 100 μl mAb-virus mixture was then added to VeroE6/TMPRSS2 cells. After incubation for 3 days, cytopathic effect was visually scored. All dilutions were performed in duplicates. The neutralization EC₅₀ titers were plotted using 5-parameter nonlinear fit in GraphPad Prism version 9.1.1. For mouse mAbs, a value of 0.12 μg/ml was assigned if no cytopathic effect was seen at a concentration of 0.24 μg/ml. For humanized WKS13, a value of 30 μg/ml was assigned if cytopathic effect was seen at a concentration of 15 μg/ml.

The spike gene of the Wuhan-Hu-1, Alpha, Beta, Delta, Gamma, and Omicron variant BA.1, BA.2, BA.2.75.2 and BA.5 sublineages were used for pseudovirus generation. The spike mutants are listed in Supplementary Table S2 . The spike gene was codon-optimized and cloned into the eukaryotic expression plasmid pCAGEN (Addgene, Watertown, Massachusetts, USA; Cat# 11160) to generate a vesicular stomatitis virus (VSV) pseudovirus plasmid pCAG-SARS-CoV-2-S as previously described (Chan et al., 2022a). BHK21 cell was transfected with spike plasmid using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA, USA; Cat# L3000015) following manufacturer’s instruction. After 24 hours, transfected cells were infected with VSV-ΔG/G*-luciferase. One hour after incubation, infected cells were washed three times with PBS, anti-VSV-G antibody (1 μg/ml) was added to neutralize excess VSV-ΔG/G*-luciferase. Twenty-four hours post-infection, culture supernatant was harvested and stored at −80°C in 0.5 ml aliquots until use. Serial 2-fold dilution of each pseudovirus was inoculated onto HEK293T-ACE2 cells on 96-well plates. After 24 h incubation, the culture supernatant was removed, and cell lysis buffer was added to lyse the cells for 15 min. Cell lysate was mixed with firefly luciferase substrate (Promega, Madison, USA; Cat# E4550) and transferred to White Opaque 96-well Microplate (PerkinElmer, Waltham, USA; Cat# 6005299) for the detection of luminescence by GloMax Explorer (Promega, Madison, USA).

HEK293T-ACE2 cells were seeded in a 96-well plate at a density of 5 × 10⁴ cells/well 24 h before assay. Each mAb was 4-fold diluted for 6 times, starting from 100 μg/ml. Each diluted mAb was mixed with 1 × 10⁶ RLU/ml rVSV-ΔG/S*-luciferase pseudoviruses. A mAb against Talaromyces marneffei Mp1p protein was used as a negative control (Wang et al., 2011). The mixture was incubated at 37°C for 1 h before being added to cells in 96-well plates. Twenty-four hours after infection, all samples were lysed and analyzed for luciferase readouts according to the manufacturer protocol. The inhibition of luciferase activity from each serum dilution could be calculated as follows: 100 – [(mean RLU from each sample (virus + serum) - mean RLU from CC)/(mean RLU from VC-mean RLU from CC) × 100]. The IC₅₀ (the half-maximal inhibitory concentration) values was calculated using nonlinear regression (log [inhibitor] vs response [four parameters]), with GraphPad Prism Version 9.1.1.

Female Syrian hamsters (4–6 weeks old) were obtained from the Chinese University of Hong Kong Laboratory Animal Service Centre through the University of Hong Kong (HKU) Centre for Comparative Medicine Research. The experimental procedures were approved by the Animal Ethics Committee on the Use of Live Animals in Teaching and Research of HKU (CULATR 5370-20) and were performed according to the standard operating procedures of Biosafety Level 3 animal facilities (Chan et al., 2020b; Chan et al., 2020c). Briefly, hamsters were intranasally inoculated with 10⁵ plaque-forming units (p.f.u.) SARS-CoV-2 Delta strain (B.1.617.2) (hCoV-19/Hong Kong/HKU-210804-001/2021; GISAID: EPI_ISL_3221329) as previously described (Chan et al., 2020b; Yuan et al., 2022). 24 hours after infection, the hamsters were either administered intraperitoneally with 5 mg/kg mouse WKS13, 5 mg/kg humanized WKS13, or equal volume of PBS. Three days later, one mouse WKS13-treated group, one humanized WKS13-treated group, and one control group were sacrificed for virological and histopathological analyses as previously described (Chan et al., 2022b), the other two groups were kept for body weight monitoring. The right lung was used for the determination of viral titer by using plaque assay, while the left lung was used for histopathological analysis and immunofluorescence staining (Lee et al., 2020).

Plaque reduction assay was performed in a 24-well tissue culture plates as we described previously with slight modifications (Yuan et al., 2021). Briefly, hamster lung was lysed by magnetic beads in 1 ml Dulbecco’s Modified Eagle Medium (DMEM), and then diluted 1:100, 1:1000 and 1:10000 with DMEM. The lung homogenate at different dilutions was added to VeroE6/TMPRSS2 cells that were seeded in 24-well plates 24h before infection. The plates were further incubated for 2 h at 37°C in 5% CO₂ before removal of unbound viral particles and washing once with DMEM. The cell monolayers were then overlaid with media containing 1% low melting agarose (Cambrex, East Rutherford, NJ, USA) in DMEM and incubated as above for 72 h. Next, the wells were fixed with 10% formaldehyde overnight. After removal of the agarose plugs, the cell monolayers were stained with 0.7% crystal violet, and the plaques were counted.

Fixed hamster lung tissues were processed, embedded, and cut to prepare 5 mm thick tissue sections on glass slides. Before staining, the slides went through dewaxed with xylene and serially decreased ethanol concentrations (100%, 95%, 70%). The tissue slides were stained with Gill’s hematoxyline and eosin (H&E) (Thermo Fisher Scientific, Waltham, MA, USA; Cat# 34002) as we previously described (Zhang et al., 2021).

For immunofluorescence staining, antigen retrieval of tissue slides were performed using Antigen Unmasking Solutions (Vector Labs, Newark, USA; Cat# H-3300-250). The slides were then incubated with an in-house rabbit antiserum against the SARS-CoV-2 nucleocapsid (N) protein. After incubation for 1 hour at room temperature, unbound antibodies were removed by washing with PBS-T six times. The bound anti-N antibodies were then detected by Alexa Fluor 488-conjugated goat anti rabbit IgG (H+L) cross-absorbed secondary antibody (Thermo Fisher Scientific, Waltham, MA, USA; Cat# A-11008) for 30 min at room temperature. After washing six times with PBST, the stained cells were mounted onto glass slides with VECTASHIELD mounting medium with 40,60-diamidino-2-phenylindole (DAPI) (Vector Labs, Newark, USA; Cat# H-1500-10). Images were acquired with the Olympus BX53 light microscope using a 20× objective.

Total RNA was extracted from Trizol-lysed WKS13 hybridoma cell lines and amplified by RT-PCR with primers in the constant antibody regions. The PCR products were sequenced by Genewiz (Azenta life sciences, Chelmsford, USA). The sequences of WKS13 antibody heavy and light chain V genes (VH/VL) were synthesized (Sangon Biotech, China, Shanghai) and cloned into the human IgG1 expression vector with restriction endonucleases AfeI and NheI (New England Biolabs, Ipswich, MA. USA; Cat# R0652L and R3131L) for the heavy chain, and AfeI and BsiWI (New England Biolabs, Ipswich, USA; Cat# R0553L) for the light chain. Equal amounts of heavy chain and light chain plasmids were transfected into Expi293F cells using 1 mg/ml polyethylenimine (PEI) (Polysciences, Inc. Warrington, USA; Cat# 9002-98-6, 26913-06-4), with cells grown to a density of 3 × 10⁶ cells/mL. Following transfection, cells were maintained in Expi293™ Expression Medium (Thermo Fisher Scientific, Waltham, MA, USA; Cat# A1435103) with GlutaMAX™ Supplement (Thermo Fisher Scientific, Waltham, MA, USA; Cat# 35050061) at 37°C in a humidified 8% CO₂ incubator rotated at 120 rpm. Six days after transfection, cell supernatant was harvested and clarified by centrifugation, and filtered through 0.22 µm filters and purified using chromatography cartridge protein A/G at room temperature. After washing with PBS, the antibodies were eluted from the chromatography cartridge protein A/G in chromatography columns using 0.1 M Na-Citrate (pH 3.0) and neutralized with 1M Tris-HCl (pH 8.8), followed by size exclusion chromatography and buffer exchanged into 1× PBS pH 7.4. The concentration of purified antibodies was determined by using the Bradford Assay Kit. The purity of recombinant mAb was verified using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE).

Statistical analysis was performed using PRISM version 9.1.1. Statistical comparisons between two experimental groups were performed using unpaired two-tailed Student’s t-tests. Comparisons among three or more experimental groups were performed using one-way or two-way ANOVA with Tukey’s multiple-comparison test. AUC values were calculated and analyzed using one-way ANOVA. Differences were considered to be statistically significant when P < 0.05.

To generate mAbs against the S protein, we first immunized two mice with ancestral S protein trimer on day 0 and day 21, and boosted with ancestral RBD protein on day 35. Splenic cells were isolated to generate hybridomas and produce mAbs ( Figure 1 ). Cell subclones that produced antibodies against the S protein were selected using an EIA assay. We have screened a total of 960 clones. Eighteen mAbs with high OD values in EIA were selected, including 10 IgG1, 4 IgG2a, 3 IgG2b, and 1 IgG3. ( Supplementary Table S3 ). Ten out of these 18 mAbs could bind to RBD.

Next, we determined the neutralizing activity of the 10 RBD-binding mAbs and the 8 remaining spike mAbs that bind outside RBD. In the cVNT, 5 of 18 mAbs (WKS10, WKS12, WKS13, WKS16 and WKS18) were able to neutralize ancestral virus at a concentration of 3.125 μg/ml ( Supplementary Table S3 ). The RBD-hACE blocking assay showed that all 10 RBD-binding mAbs could block the interaction between RBD and hACE2 with an IC₅₀ ranging from 0.4597 to 3.122 µg/ml ( Figure 2 ).

The binding kinetics of each RBD-binding mAb were measured using BLI. The binding kinetics of mAbs had low Kd values ranging from 0.026 nM (0.0039 μg/ml) to 1.150 nM (0.1725 μg/ml) ( Figure 3A ).

Next, we determined the binding epitopes of mAbs using a competitive EIA between 10 RBD-binding mAbs and 2 previously published mAbs (CR3022 and B38). mAbs that inhibit each other by >70% were considered to belong to the same group. Based on the pattern of competitive inhibition of mAb pairs, we classified our 10 RBD-binding mAbs into 5 groups ( Figure 3B ). WKS1, WKS14 and WKS17 could competitively inhibit CR3022, and therefore the epitopes of these mAbs were likely located at the inner face of the RBD (Hastie et al., 2021). WKS13 could compete with B38, and therefore likely target the receptor-binding motif (RBM) region of the RBD (Huang et al., 2022).

The entry of SARS-CoV-2 into host cells requires the binding between SARS-CoV-2 RBD and human ACE2 receptor. Mutations in the RBD can affect the binding activity of mAbs. We first used an EIA to determine how RBD mutations affect the binding of mAb to RBD. We generated RBD that has the amino acid sequence of ancestral strain (Wuhan-Hu-1) and 13 variants including all 5 Variants of Concern (VOCs). We found that WKS1 and WKS13 were least affected by the RBD mutations and were able to bind to all 18 RBDs ( Figure 4A ).

To assess whether the mAbs can inhibit different variants, we tested the ability of WKS10, WKS12, WKS13, WKS16 and WKS18 to neutralize ancestral virus and 5 variants, including Alpha, Beta, Gamma, Delta, and Omicron, using a pVNT. Only WKS13 was able to neutralize all variants tested, with IC₅₀ < 0.5 µg/ml for all variants ( Figure 4B ; Supplementary Figure S1 ).

Since mouse WKS13 exhibited broad-spectrum neutralizing activity in vitro, we tested the therapeutic effect of WKS13 in a hamster model ( Figure 5A ). Treatment with intraperitoneal administration of mouse WKS13 24 hours after Delta variant infection prevented body weight loss ( Figure 5B ) and reduced the lung viral load by about 3 log ( Figure 5C ). Immunofluorescent staining showed that the amount of virus was reduced in the lungs of WKS13-treated group when compared with the PBS group. SARS-CoV-2 N protein was abundantly expressed in the bronchioles in the control hamsters but was absent in the WKS13-treated animal lungs ( Figure 5D ). Histopathological analysis of the lung on day 4 post-infection showed multi-focal inflammation, diffuse alveolar destruction, protein-rich fluid exudate, hyaline membrane formation, marked mononuclear cell infiltration, cell debris-filled bronchiolar lumen, and alveolar collapse in the treated PBS group ( Figure 5E ). In contrast, there was a marked attenuation of the inflammatory response in the lung of WKS13-treated hamsters, with only mild pulmonary inflammatory infiltrates confined to the alveolar walls.

Since we plan to use WKS13 for use in humans, we replaced the mouse Fc fragment with human IgG1 Fc to avoid human anti-mouse antibody (HAMA) response ( Supplementary Figure S2 ). The humanized WKS13 had similar neutralizing activity against ancestral virus and different variants as the original mouse WKS13 ( Figures 6A, B ). We also verified the activity of humanized WKS13 against the Delta and Omicron variants with authentic virus with cVNT ( Figures 6C, D ). In the hamster model, we showed that treatment with humanized WKS13 reduced the lung viral load ( Figure 6E ).