Cryptic-site-specific antibodies to the SARS-CoV-2 receptor binding domain can retain functional binding affinity to spike variants

Abstract

Multiple SARS-CoV-2 variants of concern have emerged and caused a significant number of infections and deaths worldwide. These variants of concern contain mutations that might significantly affect antigen-targeting by antibodies. It is therefore important to further understand how antibody binding and neutralization are affected by the mutations in SARS-CoV-2 variants. We highlighted how antibody epitope specificity can influence antibody binding to SARS-CoV-2 spike protein variants and neutralization of SARS-CoV-2 variants. We showed that weakened spike binding and neutralization of Beta (B.1.351) and Omicron (BA.1) variants compared to wildtype are not universal among the panel of antibodies and identified antibodies of a specific binding footprint exhibiting consistent enhancement of spike binding and retained neutralization to Beta variant. These data and analysis can inform how antigen-targeting by antibodies might evolve during a pandemic and prepare for potential future sarbecovirus outbreaks.

Keywords: ACE-2 blocking; COVID-19; RBD; SARS-CoV-2; binding kinetics; biolayer interferometry; epitope binning; monoclonal antibodies; neutralizing antibodies; surface plasmon resonance.

Fig 1

Epitope binning on HexaPro separates the CoVIC panel into nine communities, with distinct completion profile, affinity range, and ACE-2 blocking ability. (a) Competition map of HexaPro binning communities from combining the results of two binning assays. Each colored square represents the categorization of normalized response for the specific pair of CoVIC constructs: red indicates competition (<0.5); green indicates sandwiching (>0.7); yellow indicates intermediate response (≥0.5 and ≤0.7); white indicates competition pairs where the data were not measured. The CoVIC IDs on the columns and rows correspond to the CoVIC IDs of the ligands (on chip surface) and the analytes (immune complexed with HexaPro), respectively. (b) Dendrogram derived from the competition map. The naming for each community and the corresponding color are shown in the legend (e.g., HexaPro-A, HexaPro-B). Sub-clusters of HexaPro-C, -E, and -H are indicated in panels a and b. (c) Binding K_D value for D614 HexaPro separated by HexaPro binning community. (d) Percent blocking of ACE-2 to RBD binding separated by HexaPro binning community, with error bars shown as gray vertical lines.

Fig 2

RBD-binder communities with more than two members can be separated into four categories. (a) A cartoon summary of the competition relationship among the four RBD-binding community categories is shown. Black texts indicate the competition relationship between each pair of community categories. (b-c) Distribution of K_D values for binding to D614 HexaPro (b) and percentage ACE-2 blocking (c) of CoVIC constructs in each of the four categories. The color code corresponds to the color scheme in (a).

Fig 3

Antibody constructs that target the top, cryptic side, and outer area of RBD can be grouped similarly among HexaPro binning communities, RBD binning communities, and the conventional classification system. (a-d) Illustration of the four main epitope footprints on RBD surface based on HexaPro epitope binning results. In the top row, RBM is colored in dark pink, RBD mutated residues in B.1.351 variant are colored in gold, and RBD mutated residues in BA.1 but not in B.1.351 are colored in cyan. In the bottom row, footprint illustrations of the main corresponding RBD communities and the corresponding conventional classifications are indicated for each footprint. RBD surface and footprint illustrations were generated from CoVIC-DB (https://CoVICdb.lji.org/) using PDB 7DDD (57).

Fig 4

RBD-inner face binders (HexaPro-A and -C) consistently show enhanced affinity for B.1.351 HexaPro relative to D614 HexaPro and have less significant loss of affinity for BA.1 HexaPro relative to D614 HexaPro when compared to RBD-top specific binders (HexaPro-H). (a, b) K_D value correlation between D614 HexaPro and B.1.351 HexaPro (a) or between D614 HexaPro and BA.1 HexaPro (b). The diagonal line is a theoretical line where the correlating affinities are identical. The areas above and below the diagonal line contain CoVIC IDs that have enhanced and weakened affinity, respectively, compared to D614 HexaPro. If binding by a specific CoVIC ID to B.1.351 or BA.1 is abolished, no data point for that construct is shown. The points are colored according to HexaPro community designations. CoVIC IDs without a community designation are in black. (c, d) Affinity ratios (D614 HexaPro K_D / B.1.351 HexaPro K_D (c) or D614 HexaPro K_D / BA.1 HexaPro K_D (d) organized by HexaPro epitope community. A K_D ratio >1 indicates an enhanced affinity for variant HexaPro relative to D614 HexaPro; a K_D ratio <1 indicates a weaker affinity for variant HexaPro relative to D614 HexaPro. In (c), each vertical line connects a ratio of 1 to the K_D ratio for a given CoVIC ID. In (d), each vertical line connects a ratio of 0.01 to the K_D ratio for a given CoVIC ID. A gray band in (d) with green borders indicates a range of ratio from 3 to 1/3. Lines without a capping point correspond to CoVIC constructs that had no B.1.351 or BA.1 binding.

Fig 5

CoVIC constructs that show enhanced affinities for B.1.351 HexaPro are more likely to retain neutralization activity against the B.1.351 variant. (a) Ratios of K_D values (D614G HexaPro K_D / B.1.351 HexaPro K_D) are shown. (b) Ratios of IC₅₀ values (D614G HexaPro IC₅₀ / B.1.351 HexaPro IC₅₀) for authentic neutralization assays are shown. IC₅₀ ratios were calculated based on values deposited in the CoVIC database contributed by the University of North Carolina (CoVIC-DB (60); see https://CoVICdb.lji.org/ for detailed values and descriptions of the method). (c) The correlation is shown between the ratio of K_D values and the ratio of IC₅₀ values for the constructs in panels a and b. In all panels, each CoVIC construct is colored based on its HexaPro community designation.

Fig 6

Fab fragments of CoVIC constructs targeting various binding sites can retain strong affinity to B.1.351. (a) The association rate constant (k_a), dissociation rate constant (k_d) and affinity values (K_D) for each CoVIC construct as a mAb or Fab binding to D614, B.1.351, and D614G HexaPro. Filled and open circles correspond to values for intact IgG and Fab fragments, respectively. No value is displayed for Fab fragments that showed no detectable binding to B.1.351. Constructs showing comparable binding to all three HexaPro proteins as Fabs are indicated within the rectangle. (b) The correlation between Fab binding affinity to D614G HexaPro and the IC₈₀ value for neutralization of authentic D614G virus is shown. (c) The correlation between Fab binding affinity to D614 HexaPro and the IC₈₀ value for pseudovirus neutralization bearing D614 is shown. (b, c) Neutralization data were obtained from CoVIC-DB (https://CoVICdb.lji.org/). R² values are from linear regression.