Prospective mapping of viral mutations that escape antibodies used to treat COVID-19

Abstract

Antibodies are becoming a frontline therapy for SARS-CoV-2, but the risk of viral evolutionary escape remains unclear. Here we map how all mutations to SARS-CoV-2's receptor-binding domain (RBD) affect binding by the antibodies in Regeneron's REGN-COV2 cocktail and Eli Lilly's LY-CoV016. These complete maps uncover a single amino-acid mutation that fully escapes the REGN-COV2 cocktail, which consists of two antibodies targeting distinct structural epitopes. The maps also identify viral mutations that are selected in a persistently infected patient treated with REGN-COV2, as well as in lab viral escape selections. Finally, the maps reveal that mutations escaping each individual antibody are already present in circulating SARS-CoV-2 strains. Overall, these complete escape maps enable immediate interpretation of the consequences of mutations observed during viral surveillance.

Figure 1.. Complete maps of escape mutations from the REGN-COV2 antibodies and Ly-CoV016.

(A) Maps for antibodies in REGN-COV2. Line plots at left show total escape at each site in the RBD. Sites of strong escape (purple underlines) are shown in logo plots at right. The height of each letter is proportional to how strongly that amino-acid mutation mediates escape, with a per-mutation “escape fraction” of 1 corresponding to complete escape. The y-axis scale is different for each row, so for instance E406W escapes all REGN antibodies but it is most visible for the cocktail as it is swamped out by other sites of escape for the individual antibodies. See https://jbloomlab.github.io/SARS-CoV-2-RBD_MAP_clinical_Abs/ for zoomable versions. Letters are colored by how mutations affect the RBD’s affinity for ACE2 (7), with yellow indicating poor affinity and brown indicating good affinity; see Fig. S2 for maps colored by how mutations affect expression of folded RBD. (B) Map for LY-CoV016. (C) Validation of key mutations in neutralization assays using pseudotyped lentiviral particles. Each point indicates the fold-increase in inhibitory concentration 50% (IC50) for a mutation relative to the unmutated “wildtype” (WT) Wuhan-Hu-1 RBD. The dotted blue line indicates wildtype-like neutralization sensitivity, and the dashed gray lines indicate upper and lower bounds on detectable fold changes. Point shapes / colors indicate if escape was expected at that site from the maps. Full neutralization curves are in Fig. S3.

Figure 2.. Escape maps are consistent with viral mutations selected in cell culture and a persistently infected patient.

(A) Viral escape mutations selected by Regeneron with spike-pseudotyped VSV in cell culture in the presence of antibody (10). (B) Escape maps like those in Fig. 1A but showing only mutations accessible by single-nucleotide changes to the Wuhan-Hu-1 sequence, with non-gray colors indicating mutations in cell culture (red), in the infected patient (blue), or both (purple). Fig. S5 shows these maps colored by how mutations affect ACE2 affinity or RBD expression. (C) Dynamics of RBD mutations in a patient treated with REGN-COV2 at day 145 of his infection (black dashed vertical line). E484A rose in frequency in linkage with F486I, but since E484A is not an escape mutation in our maps it is not shown in other panels. See also Fig. S4. (D) The escape mutations that arise in cell culture and the infected patient are single-nucleotide accessible and escape antibody binding without imposing a large cost on ACE2 affinity. Each point is a mutation with shape / color indicating whether it is accessible and selected during viral growth. Points further to the right on the x-axis indicate stronger escape from antibody binding; points further up on the y-axis indicate higher ACE2 affinity.

Figure 3.. Antibody escape mutations in circulating SARS-CoV-2.

For each antibody or antibody combination, the escape score for each mutation is plotted versus its frequency among the 180,555 high-quality human-derived SARS-CoV-2 sequences on GISAID (25) as of November 12, 2020. Escape mutations with notable GISAID frequencies are labeled.

Figure 4.. Structural context of escape mutations.

(A) Escape maps projected on antibody-bound RBD structures. (REGN10933 and REGN10987: PDB 6XDG (9); LY-CoV016: PDB 7C01 (11)). Antibody heavy-and light-chain variable domains are shown as blue cartoons, and the RBD surface is colored to indicate how strongly mutations at that site mediate escape (white indicates no escape, red indicates strongest escape site for that antibody / cocktail). Sites where no mutations are functionally tolerated are colored gray. (B) For each antibody, sites were classified as direct antibody contacts (non-hydrogen atoms within 4Å of antibody), antibody-proximal (4–8Å), or antibody-distal (>8Å). Each point indicates a site, classified as escape (red) or non-escape (black) (dashed gray line, see Methods). Red and black numbers indicate how many sites in each category are escape or non-escape, respectively. Interactive visualizations are at https://jbloomlab.github.io/SARS-CoV-2-RBD_MAP_clinical_Abs/ and additional static views are in Fig. S6.