Fig 1.
Effects of mutations in Omicron BQ.1.1 and XBB.1.5 receptor-binding domains on ACE2 binding.
(A) Diagram of the RBD substitutions that distinguish BA.2 from Wuhan-Hu-1 (top), and BQ.1.1 and XBB.1.5 from BA.2 (bottom). Dashed black lines illustrate the propagation of substitutions from Wuhan-Hu-1 to BA.2 in BQ.1.1 and XBB.1.5. Italicized mutations in BQ.1.1 and XBB.1.5 show secondarily mutated (D339H) or reverted (R493Q) substitutions that originally changed from Wuhan-Hu-1 to BA.2. Wuhan-Hu-1 reference spike numbering is used throughout the manuscript. (B) Heatmaps illustrate the impacts of all mutations and single-codon deletions in the BQ.1.1 and XBB.1.5 RBDs on ACE2-binding affinity as determined from FACS-seq experiments with yeast-displayed RBD mutant libraries. Prior measurements in the Wuhan-Hu-1 and Omicron BA.2 background from [6,18] included for reference. ACE2 contact residues defined as RBD residues with non-hydrogen atoms <5Å from ACE2 in the Wuhan-Hu-1 (PDB 6M0J), BQ.1.1 (PDB 8IF2), and/or XBB.1 (PDB 8IOV) structures. Antibody escape residues defined as those with >0.1 average antibody escape from aggregated deep mutational scanning data [30]. See S1–S3 Figs for additional experimental details. Individual measurements are reported in S1 Data, and an interactive version of these heatmaps is available at https://tstarrlab.github.io/SARS-CoV-2-RBD_DMS_Omicron-XBB-BQ/RBD-heatmaps/.
Fig 2.
Effects on ACE2-binding and evolution of RBD amino acid deletions.
(A, B) For each RBD site, comparison of the effect on ACE2 binding of single-residue deletion and the median (A) or minimum (B) impact of amino acid mutations at the site as measured in the XBB.1.5 dataset. (C, D) Mapping of the effect on ACE2 binding of the median amino acid mutation (C) or single-residue deletion (D) onto the ACE2-bound RBD structure. (E) Select alignment of ACE2-utilizing sarbecovirus RBDs illustrating patterns of deletion in loops at the ACE2-binding interface. Alignment numbering according to SARS-CoV-2 Wuhan-Hu-1 reference. Sarbecovirus variants are clustered and colored according to phylogenetic clades of RBD sequence [41]. (F) Comparison of the frequency of deletions at each RBD residue in the GISAID database of SARS-CoV-2 genomic sequences with the measured impact on ACE2 binding in the XBB.1.5 background. Sites of strong antibody escape, as defined in Fig 1B, are highlighted in red.
Fig 3.
Epistatic shifts in mutational effects on ACE2 binding.
(A) Epistatic shift in the effects of mutations on ACE2 binding at each RBD position as measured in Wuhan-Hu-1, BQ.1.1, or XBB.1.5 compared to those measured in Omicron BA.2. Wuhan-Hu-1 and Omicron BA.2 datasets are included from prior publications [6,18]. Interactive plot is available at https://tstarrlab.github.io/SARS-CoV-2-RBD_DMS_Omicron-XBB-BQ/epistatic-shifts/. Gray lines indicate sites of strong antibody escape, as defined in Fig 1B. Structures at right map the epistatic shift metric to the ACE2-bound RBD. (B) Mutation-level plots of epistatic shifts at sites of interest. Each scatterplot shows the measured ACE2-binding affinity of each amino acid in the XBB.1.5 versus BA.2 background (see S4A and S4B Fig for additional sites and backgrounds). Red dashed lines mark the parental RBD affinities, and the gray dashed line indicates the additive (non-epistatic) expectation. (C) Zoomed structural view of the ACE2-bound XBB.1 RBD structure (PDB 8IOV), illustrating the structural proximity of residue 493 with residues 453–456. (D) Double mutant cycle diagrams illustrating epistatic interactions in the XBB.1.5 (top) or BA.2 (bottom) background (representative binding curves shown in S4C Fig). Transparent points indicate per-replicate measurements, and solid points and lines connect the averaged binding values for each genotype. Colored lines show the impact of introducing Y453W, L455W, or F456L in the wildtype versus Q493R/R493Q backgrounds, with gray lines indicating the complementary change in effects of 493 mutations. Asterisk indicates expected double-mutant binding affinity assuming additivity of mutational effects.
Fig 4.
The epistatic landscape of SARS-CoV-2 evolution.
(A) Multidimensional scaling projection of variant backgrounds based on pairwise dissimilarities in per-site epistatic shifts. Dissimilarities between each RBD background were computed as the root-mean-squared epistatic shift across RBD sites (e.g., Fig 3A). (B) Layout from (A), with variants colored (and labeled in parentheses) by the number of amino acid differences from the SARS-CoV-2 Wuhan-Hu-1 RBD. (C) Layout from (A), with variants colored by antigenic distance from SARS-CoV-2 Wuhan-Hu-1 RBD. Antigenic distances were collated from the literature [27,28], and although displayed in arbitrary units, a one-unit change corresponds approximately to a two-fold loss of neutralization titer.