Discordant Antigenic Properties of Soluble and Virion SARS-CoV-2 Spike Proteins

Abstract

Efforts to develop vaccine and immunotherapeutic countermeasures against the COVID-19 pandemic focus on targeting the trimeric spike (S) proteins of SARS-CoV-2. Vaccines and therapeutic design strategies must impart the characteristics of virion S from historical and emerging variants onto practical constructs such as soluble, stabilized trimers. The virus spike is a heterotrimer of two subunits: S1, which includes the receptor binding domain (RBD) that binds the cell surface receptor ACE2, and S2, which mediates membrane fusion. Previous studies suggest that the antigenic, structural, and functional characteristics of virion S may differ from current soluble surrogates. For example, it was reported that certain anti-glycan, HIV-1 neutralizing monoclonal antibodies bind soluble SARS-CoV-2 S but do not neutralize SARS-CoV-2 virions. In this study, we used single-molecule fluorescence correlation spectroscopy (FCS) under physiologically relevant conditions to examine the reactivity of broadly neutralizing and non-neutralizing anti-S human monoclonal antibodies (mAbs) isolated in 2020. Binding efficiency was assessed by FCS with soluble S trimers, pseudoviruses and inactivated wild-type virions representing variants emerging from 2020 to date. Anti-glycan mAbs were tested and compared. We find that both anti-S specific and anti-glycan mAbs exhibit variable but efficient binding to a range of stabilized, soluble trimers. Across mAbs, the efficiencies of soluble S binding were positively correlated with reactivity against inactivated virions but not pseudoviruses. Binding efficiencies with pseudoviruses were generally lower than with soluble S or inactivated virions. Among neutralizing mAbs, potency did not correlate with binding efficiencies on any target. No neutralizing activity was detected with anti-glycan antibodies. Notably, the virion S released from membranes by detergent treatment gained more efficient reactivity with anti-glycan, HIV-neutralizing antibodies but lost reactivity with all anti-S mAbs. Collectively, the FCS binding data suggest that virion surfaces present appreciable amounts of both functional and nonfunctional trimers, with neutralizing anti-S favoring the former structures and non-neutralizing anti-glycan mAbs binding the latter. S released from solubilized virions represents a nonfunctional structure bound by anti-glycan mAbs, while engineered soluble trimers present a composite structure that is broadly reactive with both mAb types. The detection of disparate antigenicity and immunoreactivity profiles in engineered and virion-associated S highlight the value of single-virus analyses in designing future antiviral strategies against SARS-CoV-2.

Keywords: Covi-mAbs; SARS-CoV-2 spike; anti-HIV glycan mAbs; antigenicity; fluorescence correlation spectroscopy; in vitro binding assay.

Figure 1

A647-dye-labeled anti-spike mAbs exhibit variable binding efficiencies in reactions with the SARS-CoV-2 spike protein of the Wuhan variant. Representative FCS autocorrelation plots used to calculate binding efficiencies (see text) are shown in panels (a,b). Antibodies were analyzed with or without exposure to recombinant spike protein. (a) mAb CR3022; broadly reactive positive control. (b) Non-specific human IgG1; negative control. (c) FCS autocorrelation plots such as in panels (a,b) were used to determine the binding efficiencies (% antibody bound) of various labeled mAbs with the SARS-CoV-2 spike protein. All experiments were performed three times. Data are presented as the mean of three experiments ± sem.

Figure 2

Anti-spike mAbs exhibit variable reactivity with diverse SARS-CoV-2 spike proteins. FCS was used to analyze mAb-binding efficiencies in reactions with the spike proteins of (a) Alpha, (b) Beta, (c) Gamma, (d) Delta, (e) Omicron (Lineage B.1.1.529), (f) Omicron (Lineage BA.2), (g) Omicron (Lineage BA.2.75), (h) Omicron (Lineage BQ.1.1), and (i) Omicron (Lineage BA.4). (j) mAb-binding efficiencies in reactions with the spike proteins of Wuhan, D614G mutant, and Hexapro mutant were compared. Data for mAb-binding efficiencies in reactions with the spike proteins of Wuhan variant from Figure 1c was repeated for comparison. Non-specific IgG1 and CR3022 mAb were used as negative and positive controls, respectively. Data are presented as the mean of triplicate measurements ± SEM. Each experiment was performed three times with similar results. * and ** represent p values of <0.05 and <0.01 respectively.

Figure 3

Sequence alignment of variants of concern (VOC) RBD’s mutations and percent binding of mAbs (Covi-10, 11, and 17) to soluble spike protein of different variants. (a) Sequence alignment of RBD mutations in VOCs (left) compared with the percent mAb binding as measured in solution by FCS (right). * Hexapro is a vaccine reagent and is not a VOC. Shown are RBD mutations found on the outer face (purple), inner face (green), and top surface (yellow). VOC RBD mutations deviating from consensus are shown as light purple, light green, and orange. The mAb-binding efficiencies in reactions with the spike proteins of different variants as determined by FCS are color-coded as follows: 75–100% (red), 42–74% (yellow), 0–41% (grey–white). (b) Surface representation of RBD mutations. VOC RBD mutations are mapped onto model (PDB:7A94) and colored corresponding to outer (purple), inner (green), and top (yellow) RBD surface views (as previously defined by Callaway et al. [79]). Covi-11-binding epitope [57] on RBD is outlined and shown in cyan (bottom right).

Figure 4

Binding of anti-spike mAbs to SARS-CoV-2 varies between pseudovirions versus gamma-irradiated inactivated virions of Wuhan variant. Autocorrelation plots were used to determine the binding efficiencies (% antibody bound) of a panel of A647-labeled anti-spike mAbs reacted with (a) SARS-CoV-2 pseudovirions or (b) gamma-irradiated inactivated SARS-CoV-2 virions. Relationships between mAb-binding efficiencies with recombinant Wuhan SARS-CoV-2 spike trimers vs. Wuhan (c) pseudovirions (p = 0.08 (ns), R² = 0.48) or (d) gamma-irradiated virion (p = 0.03 (*), R² = 0.62). (e) Correlation between mAb-binding efficiencies with pseudovirion vs. gamma-irradiated virion (p = 0.01 (*), R² = 0.73). All experiments were performed three times. Data are presented as the mean of three experiments ± SEM.

Figure 5

Relationships between mAb-binding efficiency and in vitro neutralization activity against SARS-CoV-2 pseudovirus of Wuhan variant. (a) Dose–effect neutralization curves of Covi-series mAbs in assays with the SARS-CoV-2 pseudovirus of Wuhan variant and HEK293T-ACE2 cells (see Methods). Data are presented as the mean of triplicate measurements ± SEM. Relationships between mAb-binding efficiency measured with pseudoviruses (b) or recombinant spike protein (c) were compared with their neutralization effect at 0.15 µg/mL. Each dot in (b) or (c) represents a single mAb.

Figure 6

SARS-CoV-2 reactivity with anti-HIV-1 glycan mAbs varies according to the strain and context of spike proteins. In all experiments, CR3022 is included as a positive control and nonspecific human IgG1 as negative control. (a) Comparative binding efficiencies of A647-labeled mAbs 2G12, PGT121, PGT126, and Fab 2G12 with soluble spike proteins. mAb 2G12 and PGT121 reactivity was further tested against spike proteins extant on whole pseudovirions or released from particles by detergent lysis. Diffusion coefficients were used to determine the spike protein disposition (virion-bound or released into solution; see text) reactive with the mAb. Fluorescence autocorrelation plots of (b) mAb 2G12; or (c) mAb PGT121 under the indicated reaction conditions. Autocorrelation plots and diffusion coefficients of signals were used to determine (see text) the binding efficiencies of mAb 2G12 (d) or PGT121 (e). All experiments were performed three times. Data are presented as the mean of three experiments ± SEM. **** represents a p value of <0.0001.