Antibody-mediated SARS-CoV-2 entry in cultured cells

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) enters host cells by first engaging its cellular receptor angiotensin converting enzyme 2 (ACE2) to induce conformational changes in the virus-encoded spike protein and fusion between the viral and target cell membranes. Here, we report that certain monoclonal neutralizing antibodies against distinct epitopic regions of the receptor-binding domain of the spike can replace ACE2 to serve as a receptor and efficiently support membrane fusion and viral infectivity in vitro. These receptor-like antibodies can function in the form of a complex of their soluble immunoglobulin G with Fc-gamma receptor I, a chimera of their antigen-binding fragment with the transmembrane domain of ACE2 or a membrane-bound B cell receptor, indicating that ACE2 and its specific interaction with the spike protein are dispensable for SARS-CoV-2 entry. These results suggest that antibody responses against SARS-CoV-2 may help expand the viral tropism to otherwise nonpermissive cell types with potential implications for viral transmission and pathogenesis.

Keywords: SARS-CoV-2; antibody; receptor; viral entry.

Figure 1. Membrane fusion mediated by spike‐specific antibodies

1. Binding site locations of selected monoclonal antibodies and ACE2 binding site. Surface regions of the SARS‐CoV‐2 spike trimer in a top view targeted by eight selected antibodies on the RBD and NTD are highlighted by ellipses. Various domains of one protomer are colored (RBM in magenta, the rest of RBD in red and NTD in blue); the other two protomers in white and gray, respectively. The ACE2 binding site is marked with a yellow dashed line.

2. Schematic representation of antibody IgG (heavy and light chains) and FcγRI (α and γ chains) constructs, as well as a diagram for how FcγRI captures an IgG antibody on the surface of membrane based on the crystal structure PDB ID: 4W4O (Kiyoshi et al, 2015).

3. HEK293T cells with or without expressing FcγRI were decorated with eight selected monoclonal antibodies and tested for membrane fusion with the full‐length G614 S protein expressing cells in our standard cell–cell fusion assay. Cell–cell fusion led to reconstitution of α and ω fragments of β‐galactosidase yielding an active enzyme and thus the fusion activity was quantified by a chemiluminescent assay. Cells only and cells expressing FcγRI with no antibody added were negative controls. The experiment with three technical replicates (n = 3) was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

4. The cell–cell fusion assay was used to analyze seven polyclonal IgG antibodies purified from serum samples of vaccinated convalescent individuals reported previously (Chen et al, 2022). The experiment with three technical replicates (n = 3) was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

Data information: Statistical significance in (C and D) was determined by an unpaired two‐tailed t‐test with Welch's correction. P‐value format: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are available online for this figure.

Figure EV1. Membrane fusion mediated by antibody C63C8 and patient serum samples

1. Titration of C63C8 concentration in the assay for antibody/FcγRI mediated cell–cell fusion. HEK293T cells expressing FcγRI were decorated with C63C8 IgG at various concentrations and tested for membrane fusion with the full‐length G614 S protein expressing cells. Cell–cell fusion led to reconstitution of α and ω fragments of β‐galactosidase yielding an active enzyme and thus the fusion activity was quantified by a chemiluminescent assay. The experiment with three technical replicates (n = 3) each was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

2. The cell–cell fusion assay was used to analyze serum samples from 10 convalescent individuals, collected between May 2020 and April 2021. A serum sample for an uninfected healthy individual was used as a control. The ID₅₀ and ID₈₀ values were determined using a pseudovirus‐based neutralization assay (Xiao et al, 2021). The experiment with three technical replicates (n = 3) each was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

3. Cell–cell fusion of S‐expressing and FcγRI‐expressing cells decorated with monoclonal antibodies SP1‐77, C12C9 or C81D6 detected by our β‐galactosidase‐based cell–cell fusion assay. The wildtype ACE2 and C63C8 IgG were positive controls. The experiment with three technical replicates (n = 3) each was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

4. Images of cell association 15 min after S‐expressing and FcγRI‐expressing cells decorated with monoclonal antibodies SP1‐77, C12C9 or C81D6 were mixed, as indicated. Scale bar in black, 50 μm.

Data information: Statistical significance in (A–C) was determined by an unpaired two‐tailed t‐test with Welch's correction. P‐value format: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are available online for this figure.

Figure 2. Expression of endogenous ACE2 and cell association by mixing S‐expressing and receptor‐expressing cells

1. Schematic representation of the full‐length human ACE2 and design of antibody‐based expression constructs. Various segments for ACE2 include: catalytic peptidase domain, neck domain; TM, transmembrane anchor; and CT, cytoplasmic tail. Expression constructs of antibody‐ACE2 chimera, the Fab heavy chain of an antibody is fused with the neck domain, TM and CT of ACE2, coexpressed with the Fab light chain. A diagram showing how Fab is presented on the cell surfaces by the ACE2 TM anchor.

2. Expression of the endogenous ACE2 was monitored, by Western blot using an antibody recognizing the catalytic domain of ACE2, when cells were transfected with various antibody‐ACE2t constructs. Bands for the wildtype ACE2 and the sample loading control β‐actin are indicated.

3. Images of cell association 15 min after S‐expressing and antibody‐ACE2t‐expressing cells were mixed, as indicated. Scale bar in black, 50 μm.

Source data are available online for this figure.

Figure 3. Membrane fusion and viral entry mediated by antibody‐ACE2 chimeras

1. HEK293T cells transfected with eight different antibody‐ACE2 chimeric constructs were tested for membrane fusion with the full‐length S protein (G614 or Omicron subvariant BA.2) expressing cells in the β‐galactosidase‐based cell–cell fusion assay. The wildtype ACE2 was a positive control; no receptor/no S a negative control. The experiment with four technical repeats (n = 4) was repeated at least three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

2. Infection of HEK293T cells transfected with either ACE2 or various antibody‐ACE2 chimeric constructs by HIV‐based pseudotyped viruses using the full‐length G614 and BA.1 S constructs in a single cycle. Empty vector was used as a negative control. The experiment with two technical replicates (n = 2) was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. G614 and BA.1 data are individually normalized according to their measurement with the wildtype ACE2 receptor.

3. Virus foci in MDCK cells transfected with either ACE2 or various antibody‐ACE2 chimeric constructs followed by infection of the authentic SARS‐CoV‐2 G614 isolate. Empty vector was used as a negative control. The experiment with three technical replicates (n = 3) was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

Data information: Statistical significance was determined in (A and C) by an unpaired two‐tailed t‐test with Welch's correction. P‐value format: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are available online for this figure.

Figure EV2. Foci images of receptor‐like antibody expressing MDCK cells infected with authentic SARS‐CoV‐2

Nonpermissive MDCK cells were first transfected with ACE2 or various antibody‐ACE2t constructs and subsequently infected with the authentic SARS‐CoV‐2 G614 (B.1) virus at MOI of 0.005. Virus foci as a cluster of cells expressing viral antigen were imaged and counted using AID vSpot ELISPOT reader. The number of virus foci (dark spots) automatically identified by AID vSpot is shown in the lower corner of each well. The control images with ACE2 and empty vector are the same as those in Fig EV4.Source data are available online for this figure.

Figure 4. Membrane fusion and expression of antibody‐ACE2 chimeras

1. Schematic representation of the C‐terminally GFP‐tagged full‐length human ACE2 and antibody‐ACE2t expression constructs.

2. HEK293T cells transfected with GFP‐tagged ACE2 and eight different GFP‐tagged antibody‐ACE2t constructs were tested for membrane fusion with the full‐length S protein (G614 or Omicron subvariant BA.2) expressing cells in the β‐galactosidase‐based cell–cell fusion assay. The GFP‐tagged ACE2 was a positive control and no receptor/no S a negative control. The experiment with four technical replicates (n = 4) was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

3. Expression levels of GFP‐tagged constructs were quantified by flow cytometry. The experiment with two technical replicates (n = 2) was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate.

4. Expression levels of GFP‐tagged constructs were quantified using an anti‐Fab secondary antibody conjugated with a dye by flow cytometry. The experiment with two technical replicates (n = 2) was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate.

Data information: Statistical significance in (B) was determined by an unpaired two‐tailed t‐test with Welch's correction. P‐value format: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are available online for this figure.

Figure 5. Membrane fusion and viral entry mediated by BCRs

1. Schematic representation of the membrane‐bound B cell receptor (BCR) expression constructs and a diagram showing how Fab is presented on the cell surfaces by the BCR TM anchor.

2. HEK293T cells transfected with eight different membrane‐bound BCR constructs were tested for membrane fusion with the full‐length S protein (G614 or Omicron subvariant BA.2) expressing cells in the β‐galactosidase‐based cell–cell fusion assay. The wildtype ACE2 was a positive control and no receptor/no S a negative control. The experiment with four technical replicates (n = 4) was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

3. Infection of HEK293T cells transfected with either ACE2 or various BCR constructs by HIV‐based pseudotyped viruses using the full‐length G614 and Omicron BA.1 S constructs in a single cycle. Empty vector was used as a negative control. The experiment with two technical replicates (n = 2) was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate. G614 and BA.1 data are individually normalized according to their measurement with the wildtype ACE2 receptor.

4. Virus foci in MDCK cells transfected with either ACE2 or various BCR constructs followed by infection of the authentic SARS‐CoV‐2 G614 isolate. Empty vector was used as a negative control. The experiment with three technical replicates (n = 3) each was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d.

Data information: Statistical significance in (B and D) was determined by an unpaired two‐tailed t‐test with Welch's correction. P‐value format: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Source data are available online for this figure.

Figure EV3. Expression of endogenous ACE2 and various BCR constructs

1. Expression of the endogenous ACE2 was monitored, by Western blot using an antibody recognizing the catalytic domain of ACE2, when cells were transfected with various BCR constructs. Bands for the wildtype ACE2 and the sample loading control β‐actin are indicated.

2. Expression levels of BCR constructs were quantified using an anti‐Fab secondary antibody conjugated with a dye by flow cytometry. The experiment with two technical replicates (n = 2) each was repeated twice with similar results, and one representative experiment is shown. Each dot represents a technical replicate.

Source data are available online for this figure.

Figure EV4. Foci images of receptor‐like antibody expressing MDCK cells infected with authentic SARS‐CoV‐2

Nonpermissive MDCK cells were first transfected with ACE2 or various BCR constructs and subsequently infected with the authentic New York‐G614 (B.1) virus at MOI of 0.005. Postinfection supernatants were titrated by a focus‐forming assay. Foci as a cluster of cells expressing viral antigen were imaged and counted using AID vSpot Spectrum. The number of foci (dark spots) automatically identified by AID vSpot is shown in the lower corner of each well. The control images with ACE2 and empty vector are the same as those in Fig EV2.Source data are available online for this figure.

Figure 6. Inhibition of SARS‐CoV‐2 entry mediated by receptor‐like antibodies

1. A, B

   Inhibition of viral infectivity by protease inhibitors. Pseudovirus (G614 S) infection of HEK293T cells transfected with ACE2 or antibody constructs with/without TMPRSS2 were treated with either E‐64d (a cathepsin L inhibitor) or Camostat (TMPRSS2 inhibitor). DMSO, organic solvent used to dissolve E‐64d. The experiment with two technical replicates (n = 2) each was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate.

2. C–E

   Antibody neutralization of pseudoviruses containing the G614 S protein was determined using IgG antibodies (C63C8 and SP1‐77 in red, S2H97 and C63C7 in light red; G32B6 and C12A2 in magenta; and C12C9 in blue) with ACE2 (in C), C63C8‐ACE2t (in D) and C63C8/BCR (in E) stable cell lines. Percent neutralization was calculated from average relative light units (RLUs) from duplicate wells (technical replicates). Assays were performed twice with similar results.

Source data are available online for this figure.

Figure EV5. Inhibition of SARS‐CoV‐2 entry mediated by receptor‐like antibodies

Inhibition of viral infectivity by protease inhibitors. Pseudoviruses infection of HEK293T cells transfected with ACE2 or antibody constructs with/without TMPRSS2 was treated with E‐64d (a cathepsin L inhibitor) or Camostat (TMPRSS2 inhibitor). DMSO, organic solvent used to dissolve E‐64d. Average relative light units (RLUs) from duplicate wells (technical replicates) were plotted against virus dilutions. Assays were performed twice with similar results.Source data are available online for this figure.

Figure 7. Entry of SARS‐CoV‐2 variants mediated by receptor‐like antibodies

C63C8‐ACE2t and C63C8/BCR mediate infection by MLV‐based pseudoviruses of various SARS‐CoV‐2 variants of concern. The experiment with four technical replicates (n = 4) each was repeated three times with similar results, and one representative experiment is shown. Each dot represents a technical replicate. Error bars, mean ± s.d. Statistical significance was determined by an unpaired two‐tailed t‐test with Welch's correction. P‐value format: ns, not significant; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.Source data are available online for this figure.

Figure 8. S1 dissociation induced by ACE2 or monoclonal antibodies

The purified full‐length G614 S trimer solubilized in detergent was incubated with soluble ACE2 or various monoclonal IgG antibodies as indicated and resolved by gel‐filtration chromatography on a Superose 6 column. Peaks fractions were analyzed by SDS–PAGE with bands for the uncleaved S, ACE2, S1 fragment and S2 fragment, as well as heavy (H) and light (L) chains for antibody indicated. Representative 2D averages by negative stain EM of the peak fractions are also shown. The box size of 2D averages is ~448 Å. Bottom row, the G614 S trimer was reconstituted in lipid nanodiscs, incubated with either C63C8 or G32B6, and resolved by gel‐filtration chromatography. Peaks fractions were analyzed by SDS–PAGE with bands for the uncleaved S, S1 fragment and S2 fragment, MSP (membrane scaffold protein) as well as heavy (H) and light (L) chains for antibody indicated. Representative 2D averages by negative stain EM of the peak fractions are also shown. The box size of 2D averages is ~448 Å. Scale bar in white, 100 Å. Data information: The experiment was repeated twice with a different set of biological samples giving similar results. Source data are available online for this figure.