Skip to main content
. 2023 Mar 21;12:e83710. doi: 10.7554/eLife.83710

Figure 4. The hinge epitope is conserved across β-coronaviruses and variably accessible in authentic spike.

The hinge epitope recognized by 3A3 (SARS-2 amino acids 980–1006) is highly conserved across the spike (a) sequences and (b) structures of β-coronaviruses known to infect humans, including Alpha, Omicron BA.1, and other variants of concern (VOC; Beta, Gamma, Delta, Epsilon, Omicron BA.2 through BA.5). In (a), identical residues are indicated by a dot and similar residues are highlighted in gray. Residues conserved across all listed β-coronaviruses are in bold. Residues that lost binding to 3A3 when altered as shown in Figure 3 are in teal highlight and those whose disruption improved binding are orange. The location of the two proline mutations introduced to 2P variants are shown below the alignment. In (b), the structure of each epitope is displayed as follows: SARS-2 (6VSB) – red, SARS-1 (6CRV, RMSD = 0.8 Å) – magenta, MERS (5X5C, RMSD = 3.1 Å) – blue, HKU1 (5I08, RMSD = 0.5 Å) – teal, OC43 (6OHW, RMSD = 0.6 Å) – green. (c) Binding of full-length antibody 3A3 (black circles) and RAY53 (blue diamonds) to ancestral SARS-2, SARS-2 HexaPro (SARS-2 HP), SARS-2 4P, SARS-2 4P-DS, SARS-2 HexaPro Delta (SARS-2 HP Delta), SARS-2 HexaPro Omicron BA.1 (SARS-2 HP Omicron BA.1), MERS, SARS-1, HKU1, or milk (no coat) proteins by ELISA. Data are representative of duplicate biological replicates, each with duplicate technical replicates. The data midpoint is indicated with a bar. (d) Plasmids encoding full-length unstabilized spike proteins from SARS-2, SARS-2 Omicron BA.1, MERS, or SARS-1 were transiently transfected to Expi293 cells. The spike (blue or magenta histograms) or mock (grey histograms) transfected cells were stained with 100 nM RAY53 (top panels) or 10 nM control antibody S309 (bottom panels), followed by goat-anti-human Fc-PE secondary antibody, and flow cytometry scanning of 10,000 cells. The data shown are representative of triplicate experiments, with each condition repeated in technical duplicate. (e) Expi293 cells were transiently transfected with plasmids encoding SARS-2 spike and EGFP or EGFP only, then incubated with 3A3 (black circles) or RAY53 (blue diamonds) antibody (~1–300 nM) and anti-human Fc-PE before flow cytometric determination of the geometric mean fluorescence intensity (GMFI) in the PE channel for all green fluorescent cells. The GMFI of cells transfected with EGFP only was subtracted from the GMFI of cells expressing spike at each concentration, and the data fit to a three-parameter logistic curve to determine the effective Kd (Kd,eff) for antibody binding. The data shown are representative of triplicate experiments; ND, not detected.

Figure 4—source data 1. ELISA data and flow cytometry mean fluorescence intensity data.

Figure 4.

Figure 4—figure supplement 1. Antibody hu3A3 binds SARS-2 HexaPro similarly to 3A3 by ELISA.

Figure 4—figure supplement 1.

SARS-2 HexaPro spike was coated on high binding plates and allowed to bind dilutions of 3A3 or hu3A3 purified antibody, then anti-human Fc-HRP for detection.
Figure 4—figure supplement 2. Antibody hu3A3 yeast display libraries were enriched for binding to 4P-DS.

Figure 4—figure supplement 2.

Yeast surface display of irrelevant Fab or the hu3A3 Fab resulted in minimal binding to AF647 conjugated 4P-DS spike by flow cytometry. Site-directed (SD) or error-prone (EP) strategies to introduce diversity into the variable regions of hu3A3 Fab resulted in two libraries shown unsorted (R0), and after one (R1), two (R2), three (R3), or four (R4) rounds of selection. The 4P-DS-binding population was enriched by >70-fold during sorting of each library.
Figure 4—figure supplement 3. Antibody RAY53 retains epitope specificity while exhibiting higher affinity than 3A3 for SARS-2 HexaPro spike with reverted 2P changes.

Figure 4—figure supplement 3.

(a) ELISA data with spike 4P-DS or 4P coated on high binding plates, followed by the addition antibody 3A3 or RAY53 in a 1:5 dilution series and detection with anti-human Fc-HRP secondary antibody. (b) Raw SPR curves showing binding of immobilized 4P-DS to 3A3 or RAY53 as full-length human IgG1 at 500 nM or human Fab at 1000 nM. These curves are representative of duplicate data. (c) The same substitutions used for fine epitope mapping of 3A3 were analyzed for ELISA binding to RAY53 as described in Figure 3a.
Figure 4—figure supplement 4. Antibodies 3A3 and RAY53 have low-to-mid nanomolar affinities for stabilized SARS-2 spike variants.

Figure 4—figure supplement 4.

Binding of 3A3 Fab to HexaPro S2 was measured by (a) BLI with the HexaPro S2 trimer immobilized on the biosensor and (b) SPR with Fab immobilized on the chip. SPR was also used to evaluate binding of SARS-2 4P-DS to immobilized (c) IgG RAY53, (d) Fab RAY53 or (e) control IgG22. For BLI, Kd values were obtained using a 1:1 global fit model using the Octet instrument software. The SPR data were double-reference subtracted and fit to a 1:1 binding model using BIAevaluation software. (f) Binding of 3A3 and RAY53 IgG to additional spike variants was compared by incubating full-length IgGs immobilized on anti-Fc BLI sensors with serially diluted spike variants for an association step followed by a dissociation step in buffer only. Since the off-rates determined for this 3:1 spike protomer:antibody binding arm format using kinetic-based Kd were extremely slow due to rebinding (~10–12 s–1) and differential epitope accessibility is expected to influence the observed on-rate (Figure 2f), on-rates (kon) were compared. Vertical lines indicate the start of the dissociation phase; data shown in solid lines with fits (1:1 association then dissociation fit using GraphPad Prism software) shown in red lines.