Inactivated vaccine-elicited potent antibodies can broadly neutralize SARS-CoV-2 circulating variants

Study design

We recruited individuals who were planning to immunized with COVID-19 vaccines in May 2021. Twenty-eight healthy volunteers were enrolled for blood donation and followed regularly to evaluate the immune response to approved COVID-19 vaccines. Vaccination was conducted as part of routine care outside of the cohort study. Participants received the primary two-dose regimen (4 weeks apart) of the BBIBP-CorV (or Covilo) vaccine and were boosted at a median of 9 months after the second dose with the same vaccine based on information provided by the participants during regular follow-up visits. BBIBP-CorV, an inactivated COVID-19 vaccine, was developed by Sinopharm’s Beijing Institute of Biological Products (China) and became the first whole inactivated virus COVID-19 vaccine to receive emergency use authorization by the WHO. The HB02 strain with optimal replication and virus yields was selected and passaged in Vero cells to generate vaccine production. The BBIBP-CorV stock was inactivated by thoroughly mixing with β-propionolactone at a ratio of 1:4000 at 2–8 °C. The vaccine was manufactured as a liquid formulation containing 4 μg total protein with aluminum hydroxide adjuvant (0.45 mg/ml) per 0.5 ml⁸⁴. Eligible participants were healthy people 18 to 65 years of age, and key exclusion criteria included contraindications for vaccines, previous diagnosis of COVID-19, previous vaccination with any coronavirus vaccine, known acute and chronic infectious diseases, severe chronic and bleeding disorders, pregnancy, and lactation. Blood of the participants was collected before and after immunization (for details see Supplementary Table 1). The work was approved by the Institutional Review Board of Tsinghua University (20210061). Written informed consent was obtained from all participants.

Blood sample processing and storage

Blood samples were collected from participants at the study visit and processed within 12 h. Briefly, the plasma and blood cells were separated by centrifugation, and the blood cells were subjected to Ficoll Paque Plus (GE Healthcare, 17144002) after a 1:1 dilution in PBS (Gibco, C10010500BT) to isolate peripheral blood mononuclear cells (PBMCs) according to the manufacturer’s instructions. The plasma was divided into aliquots and stored at -80 °C. PBMCs were stored in liquid nitrogen in the presence of 10% dimethyl sulfoxide (DMSO) (Sigma, D2650-100 ml) in fetal bovine serum (FBS) (HyClone, SH30084.03). Before experiments were performed, aliquots of the plasma samples were heat-inactivated (56 °C for 1 h) and then stored at 4 °C.

Enzyme-linked immunosorbent assay (ELISA)

The SARS-CoV-2 S trimer, RBD, S2 and nucleocapsid protein (N) (Acro Biosystems SPN-C52H9, SPD-C52H3, S2N-C52H5, and NUN-C5227, respectively) were coated on 96-well plates at 0.5 μg/ml in PBS overnight at 4 °C. After the plates were blocked with buffer (1× PBS with 3% BSA (Solarbio, A8020) and 0.05% Tween-20 (Sigma, P9416)) for 1 h at 37 °C, plasma samples were added and incubated for 1 h at 37 °C. The plasma samples were assayed at a 1:100 starting dilution and 7 additional threefold serial dilutions in blocking buffer (1× PBS with 1% BSA (Solarbio, A8020) and 0.05% Tween-20). The plates were washed 5 times with washing buffer and then incubated with anti-human IgG (Abcam, ab97225) or IgM (Abcam, ab97205) secondary antibody conjugated to horseradish peroxidase (HRP) in blocking buffer at a 1:50,000 dilution (and IgG) or 1:20,000 dilution (IgM) for 0.5 h at 37 °C. After the plates were washed 5 times, the HRP substrate TMB (Solarbio, PR1200) was added for 10 min, followed by the addition of 50 μl of 1 M H₂SO₄ (Solarbio, C1058) to stop the reaction. The absorbance was measured at 450 nm with an ELISA microplate reader (TECAN Infinite 200 PRO).

SARS-CoV-2 pseudotype neutralization assays

SARS-CoV-2 and variant pseudotypes were generated by cotransfection of the S-glycoprotein-encoding plasmid and human immunodeficiency virus (HIV) backbone expressing firefly luciferase (pNL4-3.luc.RE) into HEK293T cells (ATCC, CRL3216) using Lipofectamine 2000 transfection reagent (Life Technologies, 11668-019) according to the manufacturer’s instructions⁸⁵. Pseudotyped virus stocks were collected 48 h after transfection, filtered and stored at −80 °C. Viral titers were measured based on luciferase activity determined by relative light units (Luciferase Assay Systems, Promega Biosciences). Plasma samples or antibodies were serially diluted threefold and then incubated with SARS-CoV-2 pseudovirus for 1 h at 37 °C. HeLa-hACE2 cells (Prof. Qiang Ding, Tsinghua University; HeLa, ATCC CCL-2) (1.3 × 10⁴ per well) were directly added to the antibody-virus mixture. After 48 h, the cells were washed with PBS and lysed with Luciferase Cell Culture Lysis 5× reagent (Promega, E1531), and the luciferase activity was measured using the Luciferase Assay System (Promega, E1501). Plasma or mAbs were tested in duplicate wells, and the assay was independently repeated at least twice for sets of plasma samples and individual mAbs. The neutralization ID₅₀ or IC₅₀ was calculated using nonlinear regression (log [inhibitor] vs. normalized response, variable slope) (GraphPad Prism v.8.0).

Authentic SARS-CoV-2 neutralization assays

An authentic SARS-CoV-2 neutralization assay was performed using a cytopathic effect (CPE) assay in a biosafety level 3 laboratory. Briefly, each antibody was serially diluted twofold starting at 2 μg/ml (Beta), 4 μg/ml (WT, Delta, Omicron BA.2), and 16 μg/ml (Omicron BA.1). Triplicate or duplicate preparations of each antibody dilution were incubated with the same volume of 100 TCID₅₀ of authentic SARS-CoV-2 WT (IME-BJ01 strain, GenBank No. MT291831), Beta (CSTR: 16698.06.NPRC2.062100001), Delta (CSTR.16698.06.NPRC6.CCPM-B-V-049-2105-6), Omicron BA.1 (SARS-CoV-2 strain Omicron CoV/human/CHN_CVRI-01/2022), and BA.2 (SARS-CoV-2 strain Omicron CoV/human/CHN_CVRI-04/2022) strains at 37 °C for 1 h. The mixtures were then transferred to 96-well plates seeded with Vero cells (ATCC, CCL-81, >80% density)⁸⁶. After culturing at 37 °C for 4 days, CPEs caused by virus infection were scored for each well in a blinded fashion. The results were then converted into the percentage of neutralization at a given antibody concentration, and the averages ± s.d. (triplicates) or averages (duplicates) were plotted using nonlinear regression (log [inhibitor] vs. normalized response, variable slope) (GraphPad Prism v.8.0).

SARS-CoV-2-specific memory B-cell analyses

PBMCs from vaccinated individuals were thawed and blocked with Human TruStain FcX Fc (Biolegend, 422302) for 10 min at 4 °C, followed by incubation in cell staining buffer (1× PBS, 2% FBS) with Biotinylated SARS-CoV-2 S protein (Acro Biosystems, SPN-C82E9), Biotinylated SARS-CoV-2 Spike Trimer (T19R, G142D, EF156-157del, R158G, L452R, T478K, D614G, P681R, D950N) (Acro Biosystems, SPN-C82Ec), or Biotinylated SARS-CoV-2 Spike Trimer (B.1.1.529/Omicron) (Acro Biosystems, SPN-C82Ee) for 60 min at 4 °C and then incubation for 30 min at 4 °C with the following anti-human antibodies (all at a 1:100 dilution): anti-CD19-FITC (Biolegend, 363008, Clone HIB19), anti-CD3-Pacific Blue (Biolegend, 300431, Clone UCHT1), anti-CD8-Pacific Blue (Biolegend, 301023, Clone RPA-T8), anti-CD14-Pacific Blue (Biolegend, 325616, Clone HCD14), anti-CD27-PerCP/Cyanine5.5 (Biolegend, 356408, Clone M-T271), streptavidin-APC (Biolegend, 405207) and streptavidin-PE (Biolegend, 405203). The CD3⁻CD8⁻CD14⁻CD19⁺CD27⁺Spike-PE⁺Spike-APC⁺ B cells were quantified using a CytoFLEX LX (Beckman Coulter) and CytExpert (v.2.4) for analysis.

Single-cell sequencing

The frequency distributions of the human V genes, CDR3 length and nucleotide SMH in individuals before immunization with the inactivated SARS-CoV-2 vaccine were characterized by single-cell sequencing. PBMCs from three participants in the cohort were selected to enrich B cells with a pan-B-cell isolation kit (Miltenyi Biotec, 130-101-638). After being blocked with Human TruStain FcX Fc (Biolegend, 422302) for 10 min at 4 °C, the enriched B cells from different participants were stained with distinct TotalSeq C antibodies (all at 1:250 dilution), TotalSeq-C0251 Anti-Human Hashtag 1 Antibody (Biolegend, 394661), TotalSeq-C0252 Anti-Human Hashtag 2 Antibody (Biolegend, 394663), and TotalSeq-C0253 Anti-Human Hashtag 3 Antibody (Biolegend, 394665), and incubated in cell staining buffer (1× PBS, 2% FBS) for 30 min at 4 °C with the following anti-human antibodies (all at 1:100 dilution): anti-CD19-FITC (Biolegend, 363008, Clone HIB19), anti-CD3-Pacific Blue (Biolegend, 300431, Clone UCHT1), anti-CD8-Pacific Blue (Biolegend, 301023, Clone RPA-T8), anti-CD14-Pacific Blue (Biolegend, 325616, Clone HCD14), anti-CD27-PerCP/Cyanine5.5 (Biolegend, 356408, Clone M-T271). Single CD3⁻CD8⁻CD14⁻CD19⁺CD27⁺ B cells were gated and sorted into Eppendorf tubes containing PBS with 10% FBS using a MA900 Cell Sorter (Sony).

Cells were counted and prepared for constructing 5′-mRNA, VDJ, and feature barcode libraries using the 10× Chromium System (10X Genomics) according to the manufacturer’s instructions. The Chromium Next GEM Single Cell 5′ Kit v2 (10X Genomics, PN-1000266), Library Construction Kit (10X Genomics, PN-1000196), Chromium Next GEM Single Cell 5ʹ Gel Bead Kit v2 (10X Genomics, PN-1000267), 5′ Feature Barcode Kit (10X Genomics, PN-1000256), BCR Amplification Kit (10X Genomics, PN-1000253), Chromium Next GEM Chip K Single Cell Kit (10X Genomics, PN-1000287), Dual Index Kit TT Set A (10X Genomics, PN-1000215), and Dual Index Kit TN Set A (10X Genomics, PN-1000250) were used. All the libraries were quantified by using Fragment Analyzer (Agilent) and sequenced on a NovaSeq 6000 (Illumina) with 10 cycles for the i7 index and i5 index. The average sequencing depth aimed at the mRNA library was 20,000 read pairs per cell and 5000 read pairs per cell for the VDJ libraries and for feature barcode libraries.

RBD-specific single B-cell sorting

B cells were enriched among PBMCs from vaccinated individuals using a pan-B-cell isolation kit (Miltenyi Biotec, 130-101-638) according to the manufacturer’s instructions. The enriched B cells were blocked with Human TruStain FcX Fc (Biolegend, 422302) for 10 min at 4 °C, and incubated for 60 min at 4 °C in cell staining buffer (1× PBS, 2% FBS) with biotinylated SARS-CoV-2 spike RBD (Acro Biosystems, SPD-C82E9), and then incubated for 30 min at 4 °C with the following anti-human antibodies (all at a 1:100 dilution): anti-CD19-FITC (Biolegend, 363008, Clone HIB19), anti-CD3-Pacific Blue (Biolegend, 300431, Clone UCHT1), anti-CD8-Pacific Blue (Biolegend, 301023, Clone RPA-T8), anti-CD14-Pacific Blue (Biolegend, 325616, Clone HCD14), anti-CD27-PerCP/Cyanine5.5 (Biolegend, 356408, Clone M-T271), streptavidin-APC (Biolegend, 405207) and streptavidin-PE (Biolegend, 405203). Single CD3⁻CD8⁻CD14⁻CD19⁺CD27⁺RBD⁺ B cells were gated and sorted into 96-well PCR plates containing 4 μl of lysis buffer (0.5× PBS, 10 mM DTT, 10 units of RNase Inhibitor (New England Biolabs, M0314L)) per well using an MA900 Cell Sorter (Sony) for acquisition and Cell Sorter Software (v.3.1.1) for analysis. The plates were snap-frozen on dry ice and then immediately used for subsequent RNA reverse transcription or stored at −80 °C.

Antibody amplification, cloning and expression

Human antibody heavy and light chain variable genes were generated as previously described⁸⁷. RNA from single B cells was reverse transcribed using High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, 4368813), followed by nested PCR for amplification of the variable IGH, IGL and IGK genes. The products of the second round of PCR were purified and cloned into antibody expression vectors encoding the constant regions of human IgG1 by enzymatic assembly⁸⁸. The IGBLAST program (https://www.ncbi.nlm.nih.gov/igblast/igblast.cgi) was used to analyze germline genes, germline divergence or the degree of somatic hypermutation (SHM), the framework region (FR) and the loop length of CDR3 for each antibody clone.

The paired heavy and light chain constructs were cotransfected into HEK293T cells grown in 12-well plates. The transfected culture supernatants were directly tested for binding and neutralization. Expression plasmids of mAbs showing neutralizing activity were transiently transfected into HEK293F cells (Thermo Fisher Scientific, R79007) with polyetherimide (PEI) (Polysciences, 24765) and the supernatant was purified using Protein G bead columns (Solarbio, R8300) according to the manufacturer’s protocol. Variable gene sequences of anti-SARS-CoV-2 antibodies (CB6, C121, COV2-2130, COVA1-16, and ADG-2) were synthesized (Tsingke) and produced in a HEK293F cell system.

Design, expression, and purification of bispecific antibodies

BsAbs were designed in the IgG-ScFv format. BI-2C5B and BI-5B2C were constructed from the sequences of the mAbs 6-2C and 10-5B. For BI-2C5B, the 10-5B ScFv was connected to the C-terminus of the 6-2C heavy chain with (G4S)₅ linkers to form the heavy chain, which was paired with the light chain of 6-2C. For BI-5B2C, the 6-2C ScFv was linked to the C-terminus of the 10-5B heavy chain with (G4S)₅ in the heavy chain, which was paired with the light chain of 10-5B. Similarly, BI-2C1C and BI-1C2C were constructed from the sequences of the mAbs 6-2C and 13-1C. The antibodies were produced by transient PEI transfection into HEK293F cells, and the supernatant was purified using Protein G bead columns according to the manufacturer’s protocol.

Biolayer interferometry (BLI)

Antibody affinity and the competitive binding of antibodies and hACE2 (or between two antibodies) were assessed using an Octet RED384 system (FortéBio). For apparent affinity (K_D, app) determination, 10 µg/ml recombinant biotinylated RBD of wild type (Acro Biosystems, SPD-C82E9), Omicron BA.1 (Acro Biosystems, SPN-C82E4), Omicron BA.2 (Acro Biosystems, SPN-C82Eq), and Omicron BA.4/BA.5 (Acro Biosystems, SPN-C82Ew) were loaded (1.0–1.2 nm) onto streptavidin biosensors (Molecular Devices, 18-5019), respectively. After a baseline step in PBS (Gibco, C10010500BT) for 60 s, the antigen-loaded biosensors were exposed to the mAbs for 200 s and then dipped (200 s) into PBS to measure any dissociation of antibodies from the biosensor surface. Data for which the binding responses were >0.1 nm were aligned, interstep corrected (to the association step) and fitted to a 1:1 binding model using FortéBio data analysis software, version 12.1.

For the hACE2 competition assay, 10 µg/ml recombinant biotinylated RBD (Acro Biosystems, SPD-C82E9) was immobilized (1.0–1.2 nm) onto streptavidin biosensors (Molecular Devices, 18-5019). After a baseline step in PBS buffer for 60 s, the mAbs (300 nM) were incubated with the RBD-coated biosensor for 300 s. After another baseline step in 1× PBS for 60 s, the biosensors were incubated with the hACE2 receptor (150 nM) (Sino Biological, 10108) for 300 s. The maximum binding of hACE2 was normalized to a PBS-only control. The percent binding of hACE2 in the presence of the antibody was compared to the maximum binding of hACE2. A reduction in the maximal signal to less than 20% was considered hACE2-blocking.

Epitope binning was performed with an in-tandem assay with streptavidin biosensors (Molecular Devices, 18-5019). The loaded biosensors were immersed in PBS for 60 s and then associated with the first antibody (Ab1, 300 nM) for 300 s. After a 60 s baseline step in 1× PBS, the sensors were associated with the second antibody (Ab2, 150 nM) for 300 s. Curve fitting was performed using FortéBio data analysis HT v12.1.

In vivo efficacy in a humanized ACE2 mouse model

Studies with authentic SARS-CoV-2 Delta in a humanized ACE2 mouse model were performed in biosafety level 3 laboratories and approval was obtained from the Institutional Animal Care and Use Committees of the Guangzhou Medical University. Female nine- to twelve-week-old humanized ACE2 mice were purchased from GemPharmatech (T037630). Animals were housed in a negative pressured isolator under 12 h light–dark cycles with a temperature at 22 °C and humidity set points of 50–60%. For the prophylactic experiment, the mice were intraperitoneally administered one dose of 20 mg/kg mAb 10-5B, 6-2C, or 10-5B + 6-2C (1:1 ratio) or PBS alone as a control. After 24 h, the mice were anesthetized and intranasally inoculated with 5 × 10⁴ FFU of authentic SARS-CoV-2 Delta variant. Their body weight was measured daily, and the mice were euthanized to collect lung tissue at 3 days post infection (dpi).

The virus titers of the right lung homogenate were measured by FFU assay. Serially diluted homogenates were added to Vero E6 cells in 96-well plates and incubated for 1 h at 37 °C. Afterward, the inoculum was removed, and overlay medium (MEM containing 1.6% carboxymethylcellulose) was added. Twenty-four hours later, the cells were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton X-100. After incubation with a rabbit anti-SARS-CoV-2 nucleocapsid protein polyclonal antibody (Sino Biological, 40143-T62) at a 1:3000 dilution, the cells were labeled with an HRP-labeled goat anti-rabbit secondary antibody (Jackson ImmunoResearch Laboratories, 111-035-144) at a 1:10000 dilution. The foci were visualized with TrueBlue Peroxidase Substrate (KPL, 50-78-02) and counted with an ELISPOT reader (Cellular Technology).

In vivo efficacy in a K18-hACE2 mouse model

Studies with authentic SARS-CoV-2 Omicron BA.2 (SARS-CoV-2 strain Omicron CoV/human/CHN_CVRI-04/2022) were performed in biosafety level 3 laboratories. Female seven to eight-week-old K18-hACE2 transgenic mice were purchased from GemPharmatech (T037657) and randomly allocated to groups. For the prophylactic experiment, the mice were intraperitoneally administered one dose of 10 mg/kg mAb BI-2C5B or PBS alone as a control. After 24 h, the mice were anesthetized and intranasally inoculated with 3.6 × 10³ TCID₅₀ of authentic SARS-CoV-2 Omicron BA.2 variant. Their body weight was measured daily, and the mice were euthanized 3 days post infection (dpi) to harvest lung tissues for virological assessment or histological examination.

SARS-CoV-2E gene sgRNA was quantified by quantitative reverse transcription-PCR (qRT‒PCR) assays using primers and probes as previously described^89,90. Briefly, RNA samples collected from challenged mice were reverse transcribed using HiScript III RT SuperMix (Vazyme, R323-01) followed by PCR using AceQ qPCR Probe Master Mix (Vazyme, Q112-02) with 400 nM concentrations of each of the primers, as well as 200 nM of probe (GenScript). Reactions were performed on a CFX96 Touch real-time PCR detection system (Bio-Rad). Standard curves were used to calculate the level of sgRNA in copies per gram.

Histology

The lung tissues were fixed with neutral buffered formalin for 7 days before further processing. The tissues were embedded in paraffin, and sections were stained with hematoxylin and eosin⁹¹. Images were scanned using a Pannoramic MIDI (3D HISTECH). To detect the viral antigen, a rabbit anti-SARS-CoV/SARS-CoV-2 nucleocapsid antibody (Sino Biological, 40143-R004) at a 1:200 dilution was used, followed by incubation with an Alexa Fluor 488 goat anti-rabbit antibody (Wuhan Bioqiandu Technology Co., Ltd, B100805) at a dilution of 1:200. Images were acquired on a NIKON DS-U3 Imaging system and analyzed with Image-Pro Plus 6.0 (Media Cybernetics).

Crystallization and data collection

The 6-2C Fab fragments were mixed with the SARS-CoV-2 spike RBD at a molar ratio of 1:1.2. Then, we purified the complex by gel-filtration chromatography. The purified complex was concentrated to 14 mg/ml in HBS buffer (10 mM HEPES, pH 7.2, 150 mM NaCl) for crystallization. The crystallization temperature was set at 18 °C. The sitting drop vapor diffusion method was used by mixing 0.2 μl of protein with 0.2 μl of reservoir solution. Crystals of RBD-Fab complexes were successfully obtained in 0.1 M BIS-TRIS pH 6.5, 16% w/v polyethylene glycol 10,000. Diffraction data were collected at the BL02U1 beamline of the Shanghai Synchrotron Research Facility (SSRF) and processed with HKL2000 v721.4⁹².

Structural determination and refinement

Structures were determined by the molecular replacement method using PHASER in CCP4 Program Suite v7.0⁹³. The search models were the SARS-CoV-2 RBD structure (PDB: 6M0J) and the heavy and light chain variable domain structures available in the PDB with the highest sequence identities. Subsequent model building and refinement were performed using COOT⁹⁴ and PHENIX⁹⁵, respectively. All structural figures were generated using PyMOL 2.0⁹⁶.

Cryo-EM sample preparation and data collection

Aliquots of complexes (4 μl, in buffer containing 20 mM Tris, pH 8.0, and 150 mM NaCl) of the SARS-CoV-2 spike ectodomains (WT 1.65 mg/ml, BA.1 2 mg/ml, BA.4 1.3 mg/ml, respectively) and Fab (6-2C Fab, 10-5B Fab) were applied to glow-discharged holey carbon grids (Quantifoil grid, Cu 300 mesh, R1.2/1.3). The Fab fragments were mixed with the SARS-CoV-2 spike trimer at a molar ratio of 1.2:1. The grids were then blotted for 2 s and immediately plunged into liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). The cryo-EM data of the complexes were collected with an FEI Titan Krios microscope (Thermo Fisher Scientific) at 300 kV with a Gatan K3 Summit direct electron detector (Gatan Inc., Pleasanton, CA, USA) at Tsinghua University. A total of 5494 movies were collected with SerialEM v 4.0.4, with a magnification of 29000 and defocus range between −1.3 and −1.5 μm. Each movie had a total accumulated exposure of 50 e-/Ų fractionated in 32 frames of 2.13 s exposures. The stacks were binned twofold, resulting in a pixel size of 0.97 Å/pixel.

Cryo-EM data processing

Motion correction (MotionCor2 v.1.2.6), CTF estimation (GCTF v.1.18), and nontemplated particle picking (Gautomatch v.0.56; http://www.mrc-lmb.cam.ac.uk/kzhang/) were automatically executed using the TsingTitan.py program^97,98. Sequential data processing was performed in cryoSPARC v3.3.1^99,100. Details of the data collection and processing are shown in Supplementary Figs. 13–16 and Supplementary Tables 6–7.

Cryo-EM model building and refinement

The initial model of the SARS-CoV-2 spike ectodomains (PDB 7DWY) and 10-5B Fab was fit into the map using UCSF Chimera v.1.15¹⁰¹. Manual model rebuilding was performed with COOT v.0.9.2 and refined with PHENIX v.1.18.2 real-space refinement^94,95. The quality of the final model was analyzed with PHENIX v.1.18.2⁹⁵. The validation statistics of the structural models are summarized in Supplementary Table 7. All structural figures were generated using PyMOL 2.0 and Chimera v.1.15^96,101.

Quantification and statistical analysis

No statistical methods were used to predetermine the sample size. The animals were randomly allocated into different groups. A description of the statistical analysis is provided in the figure legends. The analyses were performed using Prism 8.0 software (GraphPad, La Jolla, CA, USA). p < 0.05 was considered statistically significant.

Reporting summary

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