Rapid characterization of spike variants via mammalian cell surface display

Abstract

The SARS-CoV-2 spike protein is a critical component of vaccines and a target for neutralizing monoclonal antibodies (nAbs). Spike is also undergoing immunogenic selection with variants that increase infectivity and partially escape convalescent plasma. Here, we describe Spike Display, a high-throughput platform to rapidly characterize glycosylated spike ectodomains across multiple coronavirus-family proteins. We assayed ∼200 variant SARS-CoV-2 spikes for their expression, ACE2 binding, and recognition by 13 nAbs. An alanine scan of all five N-terminal domain (NTD) loops highlights a public epitope in the N1, N3, and N5 loops recognized by most NTD-binding nAbs. NTD mutations in variants of concern B.1.1.7 (alpha), B.1.351 (beta), B.1.1.28 (gamma), B.1.427/B.1.429 (epsilon), and B.1.617.2 (delta) impact spike expression and escape most NTD-targeting nAbs. Finally, B.1.351 and B.1.1.28 completely escape a potent ACE2 mimic. We anticipate that Spike Display will accelerate antigen design, deep scanning mutagenesis, and antibody epitope mapping for SARS-CoV-2 and other emerging viral threats.

Keywords: COVID-19; N-terminal domain; cell display; receptor-binding domain; variants.

Graphical abstract

Figure 1

Biophysical characterization of spikes displayed on human cells (A) Spike ectodomains are displayed on the surface of HEK293T cells. An automated cloning pipeline is coupled with flow cytometry to enable high-throughput screening. Biophysical characterization is performed with spikes cleaved from cell surfaces. (B) Immunostaining confirms that SARS-CoV-2 (6P-D614G) spikes are localized to cell membranes and bind ACE2 (RBD-directed), REGN10933 (RBD-directed), or 4A8 (NTD-directed). Scale bar, 10 μm. (C) Negative-stain electron microscopy micrograph (left), 2D class averages (middle), and a 3D model of surface-displayed spikes in the pre-fusion conformation. The majority of particles show a “one RBD up” configuration. Scale bar, 100 nm. (D) Relative Spike Display signal correlates with recombinant spike expression levels for engineered and clinical spike variants (Hsieh et al., 2020; Long et al., 2020). For both axes, the signal is normalized to spike-2P (two prolines) expression. Pearson correlation is used for statistical analysis. (E) ACE2 soluble domain binding by diverse coronavirus-family spikes displayed on mammalian cell surfaces. (F) Normalized binding for RBD- (REGN10987), NTD- (4A8), and S1-directed (2-43) neutralizing antibodies was measured using either the full SARS-CoV-2 spike or the isolated RBD. (G) Titration of NTD-binding nAbs using Spike Display and flow cytometry. (H) Titration of ACE2, RBD-directed (REGN10933 and REGN10987), and S1-directed (2-43) nAbs using Spike Display and flow cytometry. All measurements in (E)–(H) are an average of three biological replicates. Error bars, SD.

Figure 2

A high-resolution map of NTD-targeting nAb epitopes (A) Spike domain map (top) (Wrapp et al., 2020) and distribution of all non-synonymous mutations (total = 866,373) found in GISAID (accessed on February 24, 2021) (bottom). The NTD harbors 46% of all mutations while making up 23% of the protein. (B) Spike trimer structure (PDB: 7DDN [Zhang et al., 2021]) with domains colored as in (A). An enlarged structure of the NTD (blue) with alanine scan positions (white) is shown on the right. (C) The effect of single alanine substitutions on antibody and ACE2 binding measured by flow cytometry (see STAR Methods and supplemental information for the mean and SD). Red, decreased binding; blue, increased binding, relative to the reference spike (6P-D614G). The RBD is included as a negative control for all NTD-binding antibodies (last row). NTD loops 1–5 are annotated on the right. (D) Co-structure (PDB: 7C2L [Chi et al., 2020]) of the 4A8 Fab (light chain, violet; heavy chain, pink) in complex with the NTD (blue). Alanine scan binding data for 4A8 is superimposed on the NTD and represented on a 0 to −7 binding scale. (E) Combining multiple alanine mutations abrogates 4A8 binding. RBD is included as a negative control (gray). Mean ± SD of at least two biological replicates.

Figure 3

Clinical NTD mutants evade nAb binding (A) The effect of single amino acid substitutions on antibody or ACE2 binding measured by flow cytometry. Red, decreased binding; blue, increased binding, relative to spike (6P-D614G). RBD is a negative control for all NTD-binding antibodies (last row). (B) Nearly all nAb-evading NTD mutants occur at a low frequency in the GISAID. Notable mutations and the antibodies they evade are highlighted in red. (C) Pearson correlation matrix comparing the loss of binding across all antibodies and point mutants (left). The ten NTD-binding antibodies tested cluster in four groups based on r values > 0.5 (right). (D) Histogram of all deletions located in the spike protein with common NTD deletions (blue). Inset: distribution of all spike substitutions, deletions, and inserts in the GISAID. (E) Consequences of NTD indels on ACE2 (top) and antibody (bottom) binding (n = 3 biological replicates, mean ± SD reported in supplemental information).

Figure 4

Most variants of concern evade NTD-directed nAbs (A) Normalized spike expression for four variants of concern and their mutations, measured by flow cytometry. Red, decreased expression; blue, increased expression, relative to spike (6P-D614G). Gray indicates the absence of a mutation in a lineage (mean ± SD of at least twelve biological replicates reported in supplemental information). (B–F) Relative antibody and ACE2 binding to B.1.1.7 (B), B.1.351 (C), B.1.1.28 (D), B.1.427/B.1.429 (E), and B.1.617.2 (F) as compared to spike (6P-D614G). NTD binders, blue; RBD binders, yellow; S1 binders, red; ACE2, green. Mean ± SD of three biological replicates. (G–K). Relative antibody binding measured for each variant’s mutations (n = 3 biological replicates, mean ± SD reported in supplemental information). (L) Pseudovirus neutralization curves comparing D614G (light blue) and B.1.1.7 (dark blue) variants using three neutralizing antibodies (4A8, CM30, and 4-18). VSV-G (black) was included as an infection control (Figures S7F–S7H).

Figure 5

ACE2 mimics lose effectiveness against some variants of concern (A) Schematic of the LCB1 peptide and ACE2 competing for RBD binding. (B) Co-structure (PDB: 7JZL [Cao et al., 2020]) of LCB1 (green) binding the RBD (yellow) with contacting residues (white) and four notable VOC mutations (orange) highlighted. (C) Competition assay with a constant concentration of ACE2 and increasing concentrations of LCB1 using spike displayed VOCs (error bars, SD of three biological replicates as measured by flow cytometry).