A noncompeting pair of human neutralizing antibodies block COVID-19 virus binding to its receptor ACE2

1Coronavirus disease 2019 (COVID-19) caused by the novel COVID-19 virus has become a pandemic. The virus has spread worldwide, causing fever, severe respiratory illness, and pneumonia. Phylogenetic analysis indicates that the virus is closely related to severe acute respiratory syndrome coronavirus (SARS-CoV), but it appears to be more easily transmitted from person to person than SARS-CoV. To date, no specific drugs or vaccines are available for COVID-19.

2The COVID-19 virus belongs to the betacoronavirus genus, which includes five pathogens that infect humans. Among them, SARS-CoV and Middle East respiratory syndrome coronavirus (MERS-CoV) are two highly pathogenic viruses. As with other coronaviruses, the spike (S) glycoprotein homotrimer on the COVID-19 virus surface plays an essential role in receptor binding and virus entry. The S protein is a class I fusion protein—each S protomer consists of S1 and S2 domains, with the receptor binding domain (RBD) located within the S1 domain. Previous studies have revealed that the COVID-19 virus, similarly to SARS-CoV, uses the angiotensin-converting enzyme 2 (ACE2) receptor for cell entry. Numerous neutralizing antibodies have been found to target the RBDs of SARS-CoV or MERS-CoV. Therefore, screening for neutralizing antibodies that target the COVID-19 virus RBD is a priority.

3We expressed COVID-19 virus RBD protein as bait to isolate specific single memory B cells from COVID-19 patient peripheral blood mononuclear cells (PBMCs). The variable regions encoding the heavy and light chains were each amplified from separate single B cells and then cloned into a pCAGGS vector with the constant region to produce immunoglobulin G1 (IgG1) antibodies, as described previously. Seventeen paired B cell clones were amplified, three of which were identical (B5, B59, and H1). To identify the antibody binding abilities, the plasmids containing the paired heavy and light chains were cotransfected into human embryonic kidney–293T (HEK 293T) cells for monoclonal antibody (mAb) production. The supernatants were then screened for binding to the RBD by biolayer interferometry (BLI). An irrelevant anti–severe fever with thrombocytopenia syndrome virus Gn antibody and a SARS-specific antibody were used as controls. The supernatants from four different antibodies (B5, B38, H2, and H4) bound to COVID-19 virus RBD but not to SARS-CoV RBD (fig. S1), suggesting that the epitopes of the two RBDs are immunologically distinct. The usage of heavy chain (VH) and light chain (VL) variable genes in these four antibodies is listed in table S1.

4The dissociation constants (Kd) for the four antibodies binding to COVID-19 virus RBD, measured using surface plasmon resonance (SPR), ranged from 10−7 to 10−9 M (Fig. 1, A to D). We next studied the neutralizing activities of these four antibodies against COVID-19 virus (the BetaCoV/Shenzhen/SZTH-003/2020 strain). All four antibodies exhibited neutralizing activities, with median inhibitory concentration (IC50) values ranging from 0.177 to 1.375 μg/ml (Fig. 2, A to D). A cocktail of B38 and H4 exhibited synergetic neutralizing ability, even in the presence of a higher virus titer (Fig. 2E).

5To evaluate the ability of each antibody to inhibit binding between RBD and ACE2, we performed a competition assay using BLI and a blocking assay using fluorescence-activated cell sorting (FACS). For the BLI assay, streptavidin biosensors labeled with biotinylated RBD were saturated with antibodies, and then the test antibodies were flowed through in the presence of soluble ACE2. B38 and H4 showed complete competition with ACE2 for binding to RBD. In contrast, B5 displayed partial competition, whereas H2 did not compete with ACE2 for RBD binding (Fig. 1, E to H). The blocking assay by FACS presented the same result (Fig. 1I). To determine whether B38 and H4 target the same epitope, we performed an epitope competition assay by BLI. The nickel–nitrilotriacetic acid sensor labeled with the RBD was saturated with B38 IgG, and H4 IgG was flowed through, or the reverse (sensor saturated with H4 IgG, and B38 IgG flowed through). Although RBD was saturated with the first antibody, the second antibody could still bind to RBD, but with some inhibition. This suggests that B38 and H4 recognize different epitopes on RBD with partial overlap (Fig. 1, J and K).

6To explore the protection efficacy of B38 and H4 against challenge with COVID-19 virus in vivo, hACE2 transgenic mice were administered a single 25 mg/kg dose of B38 or H4 12 hours after viral challenge. The body weight of the B38 group decreased slowly and recovered at 3 days postinfection (dpi) compared with the phosphate-buffered saline (PBS) control group and the H4 group (Fig. 3A). The number of viral RNA copies in the lung were also measured at 3 dpi. The RNA copies of both the B38 group and the H4 group were significantly lower than those of the PBS group, with a reduction of 3.347 and 2.655 logs, respectively (Fig. 3B). These results show the same trends as the neutralization abilities. Histopathological examination indicated that severe bronchopneumonia and interstitial pneumonia could be observed in the mice of the PBS control group, with edema and bronchial epithelial cell desquamation and infiltration of lymphocytes within alveolar spaces (Fig. 3, C and F). Mild bronchopneumonia was observed in the H4 group (Fig. 3, E and H), whereas no lesions were observed in the B38 group (Fig. 3, D and G).

7As is consistent with the binding affinity between RBD and B38 or H4, stable complexes were obtained in both RBD-B38 and RBD-H4 mixtures (fig. S2). The complex crystal structure of RBD-B38 Fab was solved at 1.9-Å resolution (table S2). Three complementarity-determining regions (CDRs) on the heavy chain and two CDRs on the light chain are involved in interaction with RBD (Fig. 4, A, B, and G to K). The buried surface area of heavy and light chains on the epitope is 713.9 and 497.7 Å, respectively. There are 36 residues in the RBD involved in the interaction with B38, in which 21 residues and 15 residues interact with heavy and light chains, respectively (table S3 and Fig. 4B). Sequence alignment indicates that only 15 of the 36 residues in the epitope (defined as residues buried by B38) are conserved between COVID-19 virus and SARS-CoV (Fig. 4, D to F, and fig. S3). Notably, most contacts in the interface between B38 and RBD are hydrophilic interactions (table S4). Water molecules play an important role in the binding between COVID-19 RBD and B38 (Fig. 4, G and I to K). These differences explain the B38-specific binding to the COVID-19 virus rather than SARS-CoV.

8To explore the structural basis for B38 blocking the interaction between COVID-19 virus RBD and ACE2, the complex structures of RBD–B38-Fab and RBD-hACE2 were superimposed. Both the VH and VL of B38 would sterically hinder ACE2 binding (Fig. 4C). Notably, the RBDs in B38-bound form and hACE2-bound form have no notable conformational differences, with a Cα root mean square deviation of 0.489 Å (for 194 atoms). Further analysis indicated that 18 of the 21 amino acids on the RBD are involved in binding both B38 and ACE2 (Fig. 4D), which explains why B38 abolishes the binding between COVID-19 virus RBD and the receptor.

9As the COVID-19 outbreak continues to spread, characterization of the epitopes on the COVID-19 virus RBD will provide valuable information for vaccine development. Furthermore, the molecular features of the neutralizing antibody targeting epitopes are helpful for the development of small-molecule or peptide drugs and inhibitors. The neutralizing antibodies themselves are also promising candidates for prophylactic and therapeutic treatment against the COVID-19 virus.