Spike-Dependent Opsonization Indicates Both Dose-Dependent Inhibition of Phagocytosis and That Non-Neutralizing Antibodies Can Confer Protection to SARS-CoV-2

Abstract

Spike-specific antibodies are central to effective COVID19 immunity. Research efforts have focused on antibodies that neutralize the ACE2-Spike interaction but not on non-neutralizing antibodies. Antibody-dependent phagocytosis is an immune mechanism enhanced by opsonization, where typically, more bound antibodies trigger a stronger phagocyte response. Here, we show that Spike-specific antibodies, dependent on concentration, can either enhance or reduce Spike-bead phagocytosis by monocytes independently of the antibody neutralization potential. Surprisingly, we find that both convalescent patient plasma and patient-derived monoclonal antibodies lead to maximum opsonization already at low levels of bound antibodies and is reduced as antibody binding to Spike protein increases. Moreover, we show that this Spike-dependent modulation of opsonization correlate with the outcome in an experimental SARS-CoV-2 infection model. These results suggest that the levels of anti-Spike antibodies could influence monocyte-mediated immune functions and propose that non-neutralizing antibodies could confer protection to SARS-CoV-2 infection by mediating phagocytosis.

Keywords: SARS – CoV – 2; antibodies; antibody binding; antibody function; in vivo model; phagocytosis; spike (S) protein.

Figure 1

Convalescent patient plasma reduces Spike-monocyte interaction. (A) Biotinylated Spike protein was conjugated to fluorescent (APC) streptavidin microspheres and was opsonized with three convalescent patient plasma concentrations (1%, 0.1%, and 0.01%). The beads were then mixed with THP-1 cells at a ratio of 2:1, and the association was measured using flow cytometry. Cells that had signal in the APC channel were considered positive. The gating strategy is shown in the top right. (B) The same samples of THP-1 cells and beads from (A) were fixed with methanol and stained with a fluorescent (FITC) Fab anti-human Fab secondary antibody. The samples were analyzed for human antibody (opsonin) binding to the Spike-beads using flow cytometry. The gating strategy is shown in the top right. The data presented are from three independent experiments. Error bars represent the SD. Statistical significance was assessed using two-way ANOVA with Dunnett’s multiple comparison correction. ** denotes for p ≤ 0.01, *** for p ≤ 0.001 and **** for p ≤ 0.0001. ns, not significant. Created with BioRender.com.

Figure 2

Generation of Spike-reactive human monoclonal antibodies. (A) Human monoclonal antibodies were generated from convalescent donor B cells through single-cell sequencing technology. 96 antibodies derived from Spike-reactive human B cells were produced in HEK293F cells. (B) Cell culture supernatants containing the antibodies were assayed by ELISA for reactivity against immobilized Spike protein. Serum from a COVID19 patient was used as a positive control (external patient, not from the 20-patient cohort used in Figure 1 ). The data represent three replicate ELISAs where reproducibly reactive antibodies are indicated with their names above the heatmap. (C) Antibodies which were Spike-reactive in (B) were assayed for reactivity to Spike immobilized on beads. Fluorescent (APC) Streptavidin beads coated with biotinylated Spike protein were incubated with HEK293F-produced antibodies at a concentration of 1 µg/ml. The beads were then stained with a fluorescent (FITC) secondary anti-Fab antibody. The beads were analyzed by flow cytometry. Antibodies that shifted the beads into the FITC-positive gate were deemed reactive. Created with BioRender.com.

Figure 3

Epitope mapping and structural mass spectrometry identify antibody binding sites. (A) Antibody binding to Spike domains was analyzed using ELISA as shown with HRP signal as readout. Antibodies were titrated at 0.2-30 nM, and the integrated signal was calculated. The relative binding to each Spike domain is shown. nd, not determined. (B) Epitope mapping was performed using scFvs targeting RBD epitopes as shown. Antibody blocking of scFv binding was measured using anti-FLAG HRP signal. Representative of two independent experiments. (C) Spike protein was mixed with anti-spike antibodies and the complex was cross-linked with DSS, allowing for inter and intra cross-links. After trypsinization, mass spectrometric analysis was performed. The table to the right displays the number of inter-protein cross-links detected between Spike and its corresponding antibody. (D, E) The binding sites for Ab66 and Ab94 were determined by TX-MS using the cross-links from c, and the data was modelled using Rosetta. Models for Ab66 (D) binding the Spike protein in both its open and closed conformations as well Ab94 (E) are shown. The cross links between Spike and antibodies are shown in yellow. Created with BioRender.com.

Figure 4

Neutralization assays identify one monoclonal as blocking the ACE2-Spike protein interaction. (A) SPR analysis of the binding of monoclonal antibodies to the RBD domain of the Spike protein. PBS served as a negative control, and the intact RBD was our positive control for ACE2 binding. (B) The 96 antibodies which we produced were assayed for neutralization potential in a Spike-bead-based neutralization assay. Spike-beads [such as the ones used in (B)] were opsonized with the antibodies in 96 well plates. The beads were then centrifuged, reconstituted in fresh media, and added to HEK293-ACE2 cells at a ratio of 20 beads per cell and imaged with automated microscopy. The data is from 4 pooled experiments and is presented as bead association normalized percentage. Error bars indicate the SEM for the replicate experiments. (C) The 10 Spike-ELISA reactive antibodies were assayed for pseudovirus neutralization. A firefly luciferase encoding pseudotype lentivirus was used to infect HEK239-ACE2 cells. Antibody serial dilutions were used to block the viral entry into the HEK293-ACE2 cells. Nonlinear regression lines were fitted for the three antibodies that showed a higher than 50% reduction of infectivity. Those antibodies were highlighted in green (Ab57), blue (Ab36), and red (Ab59). Created with BioRender.com.

Figure 5

High levels of human monoclonal antibodies reduce Spike-monocyte interaction. (A) Spike-reactive monoclonal antibodies at concentrations of 100, 10, 1, 0.1, and 0.01 µg/ml were used to opsonize Spike-beads. Plasma was used at serial dilutions of 10%. The beads were then incubated with THP-1 cells at a ratio of 2 beads/cell. The cells were then analyzed by flow cytometry for association with the fluorescent Spike-beads. The data show the % of bead-associated cells and is pooled from three independent experiments. Error bars represent the SD. (B) The cells used in (A) were fixed with methanol and restained with a fluorescent (FITC) Fab anti-human secondary antibody. The samples were assessed for human antibody (opsonin) binding to the Spike-beads using flow cytometry. The data are from three independent experiments. Error bars represent the SD. * denotes p ≤ 0.05, ** for p ≤ 0.01, and **** for p ≤ 0.0001. ns, not significant. Created with BioRender.com.

Figure 6

Non-neutralizing antibodies can protect against SARS-CoV-2 infection. (A) Humanized ACE2 mice were infected intranasally with SARS-CoV-2 (Wuhan strain). One day after infection, the animals (N=7 per group) were treated intraperitoneally with antibodies. Relative body weights were recorded and tabulated. (B) Body weights relative to each individual mouse over the time course of viral infection and treatment. (C) Quantitative PCR analysis was performed on the BAL fluids obtained from the infected mice at the experiment termination. The cycle threshold (Ct) values were detected and graphed. Error bars represent the SD. Statistical significance was assessed using two-way ANOVA with Dunnett’s multiple comparison correction both within each treatment group and across the groups for each day. * denotes p ≤ 0.05, ** for p ≤ 0.01, *** for p ≤ 0.001 and **** for p ≤ 0.0001. ns, not significant. Created with BioRender.com.