Receptor-mimicking antibodies retain neutralization of recent SARS-CoV-2 variants Authors: Qi Tang, Liyuan Hu, Minxiang Xie, Ran Zhan, Qiao Wang, Rong Zhang Published: 2026 ABSTRACT Highlights•Scientific questions: As SARS-CoV-2 accumulates RBD mutations, we asked whether receptor-mimicking antibodies targeting ACE2-overlapping sites retain neutralization and spike binding against recent variants.•Evidence before this study: VIR-7229 and S2K146 can mimic ACE2 recognition and broadly neutralize SARS-CoV-2 variants, but the effects of recent RBM mutations on these antibodies remain unclear.•New findings: VIR-7229 retained the strongest activity among the tested antibodies, potently neutralizing Wuhan-Hu-1, BA.1, XBB.1.5, and BA.2.86 while maintaining considerable activity against KP.3.1.1, XEC, and XEC.25.1. ELISA, BLI, and structural analysis showed preserved spike binding and an intact VIR-7229-RBD interface.•Significance of the study: Receptor-mimicking antibodies can retain activity against recent SARS-CoV-2 variants by targeting constrained ACE2-overlapping epitopes, supporting their further evaluation and informing immunogen design. Introduction Since its emergence in late 2019, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has continued to spread and evolve, seriously affecting global public health [1], [2], [3]. The SARS-CoV-2 spike (S) protein mediates receptor binding and membrane fusion and is the main target of neutralizing antibodies. The receptor binding domain (RBD), especially the receptor binding motif (RBM), is a major target of neutralizing antibodies induced by infection or vaccination [4], [5]. Under immune pressure, the RBD, particularly the RBM, has become one of the main regions where mutations accumulate. These mutations have reduced the activity of many existing neutralizing antibodies and weakened immunity induced by vaccination [6], [7], [8]. Therefore, further evaluation of RBD antibodies with different functional mechanisms against recent variants is important for antibody evaluation and broad-spectrum immunogen design. In recent years, a group of SARS-CoV-2 RBD antibodies known as receptor-mimicking antibodies, which exhibit relatively broad neutralizing activity, has attracted increasing attention. Reported SARS-CoV-2 receptor-mimicking antibodies include VIR-7229, S2K146, BD55-1205, P2C-1F11, CB6, P4J15, and GAR05 [9], [10], [11], [12], [13], [14]. These antibodies often target angiotensin‐converting enzyme 2 (ACE2) binding sites that are functionally constrained. Mutations at these sites may impair receptor binding; therefore, these antibodies usually impose a high barrier to viral escape [11], [14]. In addition to structural mimicry, some receptor-mimicking antibodies have also been shown to mimic receptor function by inducing shedding of the spike S1 subunit, thereby further enhancing neutralization [10], [11], [14]. Notably, similar receptor mimicry strategies have also been reported for influenza virus, human immunodeficiency virus (HIV), and other pathogens, suggesting that receptor mimicry may represent a general mechanism for broad neutralization [15], [16], [17]. Among the reported receptor-mimicking antibodies, VIR-7229 and S2K146 are representative examples. VIR-7229 is notable for its activity against not only multiple SARS-CoV-2 variants but also many other sarbecoviruses [11]. S2K146 has also shown broad neutralizing activity against SARS-CoV and SARS-CoV-2-related sarbecoviruses that use ACE2 as the receptor [14]. In this study, we evaluated the neutralizing and binding activities of these two antibodies against recent variants. S2H13, a structurally characterized SARS-CoV-2 RBD-targeting antibody with limited neutralizing breadth against emerging variants, was used as a control [4]. We first compared the neutralizing activity of these antibodies against representative SARS-CoV-2 variants using pseudovirus neutralization assays. We then assessed their binding to S proteins from recent variants by enzyme-linked immunosorbent assay (ELISA) and biolayer interferometry (BLI). Finally, structural analysis was used to assess the possible effects of mutations in recent variants on VIR-7229 binding. These results may aid the evaluation of broadly neutralizing antibodies and the design of candidate immunogens against evolving SARS-CoV-2 variants. Materials and methods Antibody expression and purification The antibody heavy and light chain sequences were obtained from the corresponding Protein Data Bank (PDB) identifications (IDs): VIR-7229 (PDB: 9AU1), S2K146 (PDB: 7TAT), and S2H13 (PDB: 7JV6). Heavy and light chain sequences were optimized for Homo sapiens codon usage and synthesized [4], [11], [14]. The synthesized fragments were amplified and cloned into the pcDNA3.1 vector. Heavy and light chain expression plasmids were constructed for each antibody. Heavy and light chain plasmids were then co-transfected into HEK293F cells for transient antibody expression. Five days after transfection, culture supernatants were collected. Antibodies were purified on a Protein A affinity column (GenScript, L00464) and analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). Pseudovirus production HEK293T cells were seeded in T25 flasks at 3 × 10⁶ cells per flask. At 80% − 90% confluence, spike expression plasmids for Wuhan-Hu-1, BA.1, XBB.1.5, BA.2.86, KP.3.1.1, XEC, and XEC.25.1 were separately transfected into HEK293T cells using Ez Trans (Liji Biotech, AC04L092). The medium was replaced 12 h later. After another 12 h, cells were infected with scVSVΔG-Nluc-GFP [18]. After 1.5 h, cells were washed with phosphate buffered saline (PBS) and cultured in 10% Dulbecco's Modified Eagle Medium (DMEM). Supernatants were collected 24 h after infection, centrifuged to remove cell debris and stored at −80 °C. Pseudovirus neutralization assay Before neutralization assays, each SARS-CoV-2 pseudovirus was titrated to normalize the amount of virus used for infection. Vero E6 cells were seeded in 96-well plates at 2 × 104 cells per well and cultured overnight at 37 °C with 5% CO2. The next day, antibodies were prepared as threefold serial dilutions, with the highest concentration set at 1 μg/mL or 30 μg/mL, and mixed 1:1 with pseudovirus. The mixtures were incubated at 37 °C for 60 min and then added to the Vero E6 cells at 100 μL per well, with three replicate wells for each concentration. Virus-infected wells without antibody were used as controls. After 20 h, the medium was removed, and 100 μL of Luciferase Buffer with substrate (Meilunbio, MA052) was added to each well. Plates were incubated at room temperature for 15 min. The half maximal effective concentration (EC50) values were calculated in GraphPad Prism. S protein expression and purification To stabilize the trimeric conformation of the S protein, six proline substitutions were introduced at residues 817, 892, 899, 942, 986, and 987 [19]. The S1/S2 furin cleavage site at residues 682 to 685 (RRAR) was replaced with GSAS [20], and a T4 fibritin foldon trimerization domain was added at the C terminus [21]. Recombinant trimeric S proteins of KP.3.1.1, XEC, and XEC.25.1 carried a C-terminal His tag. The corresponding spike expression plasmids were transiently transfected into HEK293F cells. Culture supernatants were collected and purified by nickel affinity chromatography (GenScript, L00666). The purified proteins were analyzed by SDS-PAGE and used in ELISA and BLI assays. Enzyme-linked immunosorbent assay S proteins were diluted to 10 μg/mL in PBS and coated onto 96-well ELISA plates at 50 μL per well overnight at 4 °C. After coating, plates were washed three times with PBS with Tween-20 (PBST) and blocked with 2% bovine serum albumin (BSA) at room temperature for 2 h. Antibodies were prepared as threefold serial dilutions in PBS, with the highest concentration set at 10 μg/mL, and added at 50 μL per well for 1 h at room temperature. Plates were then washed and incubated with horseradish peroxidase (HRP) conjugated goat anti-human secondary antibody (Sigma, A6029) for 1 h at room temperature. After washing, 3,3′,5,5′-Tetramethylbenzidine (TMB) was added and incubated for 10 min (Thermo Fisher, 34028). Optical density at 450 nm (OD450) values were then measured. hE16, a human IgG1 antibody targeting West Nile virus envelope protein, was used as an antigen irrelevant negative control [22]. EC50 values were calculated in GraphPad Prism. Biolayer interferometry assay Binding kinetics were measured using a Pall ForteBio Octet RED96 system. Recombinant antibodies were loaded onto Protein A biosensors at 10 μg/mL for 300 s. The antibody-loaded biosensors were then transferred into S protein solutions for 300 s to record association, followed by dissociation in kinetics buffer for 180 s. Background signals were corrected using kinetics buffer without antigen. His-tagged recombinant human B7-1/CD80 protein (hB7.1-His) was used as an irrelevant protein control. For affinity fitting, S proteins were tested as threefold serial dilutions from 180 nmol/L (nM), with concentrations of 180, 60, 20, 6.667, and 2.222 nM. KD values were obtained by fitting the sensorgrams in Octet Data Analysis v11.1. Data were acquired using Octet Data Acquisition 9.0, and curves were plotted in GraphPad Prism. RBD sequence alignment and visualization RBD amino acid sequences from different variants were aligned using MAFFT v7.490. The alignment was visualized with the ESPript 3.2. Structural modeling and analysis of the VIR-7229 RBD complex The published complex structure of VIR-7229 antigen-binding fragment (Fab) bound to BA.2.86 RBD (PDB: 9ASD) was used as the template for structural analysis [11]. Missing residues in the BA.2.86 RBD were modeled with Modeller 10.8 [23] to generate a more complete complex template. Based on this template, models of VIR-7229 Fab bound to RBDs from different variants were generated using the Cartesian ddG protocol in Rosetta 3 [24]. The models were visualized in PyMOL v3.0.3, and local interactions at the antibody binding interface were analyzed. Surface electrostatic potentials were calculated using the APBS Electrostatics plugin in PyMOL. Prediction of changes in binding free energy The effects of RBD mutations on VIR-7229 Fab binding were predicted using the Rosetta Flex ΔΔG protocol [25]. BA.2.86, XEC, and XEC.25.1 RBDs were analyzed. Each variant was modeled based on the published VIR-7229-BA.2.86 Fab complex structure, with missing residues built using MODELLER 10.8, and the respective mutations were introduced to generate computational models for XEC and XEC.25.1. For each model, the Backrub algorithm was used to sample backbone conformations within 8 Å of the mutated residues. A total of 35,000 Backrub steps were performed, with a sampling interval of 17,500 steps, to explore the local conformational space. Energies were evaluated using the ref2015 all-atom energy function, followed by limited-memory Broyden-Fletcher-Goldfarb-Shanno (LBFGS) minimization with a maximum of 5,000 iterations. The change in binding free energy, ΔΔG, was calculated using the InterfaceDdGMover in Rosetta. For each RBD variant, 35 independent Backrub trajectories were generated. Data analysis Five parameter dose response curves were fitted in GraphPad Prism to determine EC50 values. Quantitative data are presented as mean ± standard deviation. Results VIR-7229 and S2K146 exhibit typical receptor-mimicking features We first summarized the binding epitopes of reported SARS-CoV-2 receptor-mimicking antibodies. These antibodies mainly target the RBM of the RBD, and their binding sites substantially overlap with the ACE2 binding site on the RBD (Fig. 1A). VIR-7229 has broad neutralizing activity. It neutralizes multiple SARS-CoV-2 variants and also shows activity against several other sarbecoviruses [11]. S2K146 also shows broad neutralizing activity against SARS-CoV and SARS-CoV-2 related sarbecoviruses that use ACE2 as the receptor [14]. S2H13 was identified in a structure-guided RBD antigenic mapping study [4] and was used to provide a reference for epitope mapping and neutralization analysis. Further mapping of the three epitopes onto the RBD surface showed that the epitopes of VIR-7229, S2K146, and S2H13 all overlapped with the ACE2 binding site to some extent (Fig. 1B). Steric hindrance analysis further showed that all three antibodies target the RBD and compete with human ACE2 (hACE2) for binding (Fig. 1D). VIR-7229 and S2K146 display receptor-mimicking binding features, preferentially recognizing spike trimers with two RBDs in the “up” conformation and inducing S1 subunit shedding, whereas S2H13 does not show these features (Fig. 1C).Fig. 1Epitope, structural, and functional features of representative SARS-CoV-2 receptor-mimicking antibodies. A) Epitope distribution of SARS-CoV-2 receptor-mimicking antibodies on the RBD. Based on published complex structures, RBD contact residues for ACE2 and different receptor-mimicking antibodies were mapped onto the linear RBD sequence (PDB IDs: 9ASD, 7TAS, 8XE9, 7CDI, 7C01, 8PQ2, and 7T72). Blue indicates ACE2-contact residues, light orange indicates antibody-contact residues, and green indicates the RBM. B) Epitope mapping of representative antibodies on the RBD surface. Antibody epitopes are shown in magenta, and the ACE2 binding footprint is outlined in black. C) Functional comparison of representative antibodies, including recognized RBD state, S1 shedding, and ACE2 competition. D) Steric hindrance analysis between ACE2 and different antibodies bound to the RBD. The structures of VIR-7229, S2K146, and S2H13 bound to the RBD are shown. ACE2 is shown in gray, the RBD is shown as a cyan cartoon, and the antibody heavy and light chains are shown in magenta and blue, respectively. Red asterisks indicate steric hindrance between the antibody and ACE2. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; ACE2, angiotensin‐converting enzyme 2; RBD, receptor-binding domain; RBM, receptor binding motif; PDB, Protein Data Bank; IDs, identifications; S1, spike S1. Epitope, structural, and functional features of representative SARS-CoV-2 receptor-mimicking antibodies. A) Epitope distribution of SARS-CoV-2 receptor-mimicking antibodies on the RBD. Based on published complex structures, RBD contact residues for ACE2 and different receptor-mimicking antibodies were mapped onto the linear RBD sequence (PDB IDs: 9ASD, 7TAS, 8XE9, 7CDI, 7C01, 8PQ2, and 7T72). Blue indicates ACE2-contact residues, light orange indicates antibody-contact residues, and green indicates the RBM. B) Epitope mapping of representative antibodies on the RBD surface. Antibody epitopes are shown in magenta, and the ACE2 binding footprint is outlined in black. C) Functional comparison of representative antibodies, including recognized RBD state, S1 shedding, and ACE2 competition. D) Steric hindrance analysis between ACE2 and different antibodies bound to the RBD. The structures of VIR-7229, S2K146, and S2H13 bound to the RBD are shown. ACE2 is shown in gray, the RBD is shown as a cyan cartoon, and the antibody heavy and light chains are shown in magenta and blue, respectively. Red asterisks indicate steric hindrance between the antibody and ACE2. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; ACE2, angiotensin‐converting enzyme 2; RBD, receptor-binding domain; RBM, receptor binding motif; PDB, Protein Data Bank; IDs, identifications; S1, spike S1. VIR-7229 and S2K146 retain neutralizing activity better than S2H13 To compare the neutralizing activity of different RBD antibodies, we selected seven representative variants for pseudovirus neutralization assays: Wuhan-Hu-1, BA.1, XBB.1.5, BA.2.86, KP.3.1.1, XEC, and XEC.25.1. These variants cover the major antigenic stages from the ancestral strain and early Omicron variants to recent circulating lineages (Fig. 2A). BA.2.86 is an important lineage in recent SARS-CoV-2 antigenic evolution. BA.2.86 carries more than 30 spike mutations relative to BA.2 and is associated with strong antibody escape, a feature also seen in its descendant lineages [26]. BA.2.86 further gave rise to JN.1 and several descendant lineages. KP.3.1.1 is one representative circulating lineage among JN.1 descendants [27], whereas XEC is a recombinant lineage derived from the JN.1-related lineages KS.1.1 and KP.3.3 [28]. XEC.25.1 is a later XEC derivative with an additional A435S substitution [29]. This substitution has also appeared in recent lineages such as NB.1.8.1 and BA.3.2, suggesting convergent evolution as SARS-CoV-2 continues to escape antibody pressure while maintaining receptor binding [30], [31], [32]. Because receptor-mimicking antibodies target key receptor binding regions, it is important to evaluate whether they can retain neutralizing activity during viral evolution. VIR-7229, S2K146, and S2H13 were successfully expressed and purified, as shown by SDS-PAGE (Fig. 2B).Fig. 2Comparison of neutralizing activity among different antibodies against selected SARS-CoV-2 variants. A) Schematic overview of the seven representative SARS-CoV-2 variants included in this study: Wuhan-Hu-1, BA.1, XBB.1.5, BA.2.86, KP.3.1.1, XEC, and XEC.25.1. These variants represent the ancestral strain and Omicron lineages at different stages of antigenic evolution. B) SDS-PAGE analysis of purified VIR-7229, S2K146, and S2H13 antibodies. C–E) Neutralization curves of VIR-7229, S2K146, and S2H13 against the seven representative variants. Data are presented as mean ± standard deviation from two independent experiments, each with three technical replicates. F) Summary of EC50 values for the three antibodies against different variants. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; EC50, the half maximal effective concentration; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Comparison of neutralizing activity among different antibodies against selected SARS-CoV-2 variants. A) Schematic overview of the seven representative SARS-CoV-2 variants included in this study: Wuhan-Hu-1, BA.1, XBB.1.5, BA.2.86, KP.3.1.1, XEC, and XEC.25.1. These variants represent the ancestral strain and Omicron lineages at different stages of antigenic evolution. B) SDS-PAGE analysis of purified VIR-7229, S2K146, and S2H13 antibodies. C–E) Neutralization curves of VIR-7229, S2K146, and S2H13 against the seven representative variants. Data are presented as mean ± standard deviation from two independent experiments, each with three technical replicates. F) Summary of EC50 values for the three antibodies against different variants. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; EC50, the half maximal effective concentration; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Pseudovirus neutralization assays revealed distinct neutralization patterns among the three antibodies (Fig. 2C–2E). VIR-7229 retained strong activity against Wuhan-Hu-1, BA.1, XBB.1.5, and BA.2.86. Its activity was weaker against KP.3.1.1, XEC, and XEC.25.1, but it still neutralized these variants. S2K146 neutralized Wuhan-Hu-1, BA.1, and XBB.1.5, but largely lost activity from BA.2.86 onward. S2H13 showed the narrowest profile, with only limited activity against Wuhan-Hu-1. The EC50 summary further quantified these differences (Fig. 2F). VIR-7229 showed EC50 values of 0.009, 0.003, 0.004, and 0.007 μg/mL against Wuhan-Hu-1, BA.1, XBB.1.5, and BA.2.86, respectively. For KP.3.1.1, XEC, and XEC.25.1, VIR-7229 still retained neutralizing activity, albeit with EC50 values increased to 1.667, 3.946, and 4.036 μg/mL, respectively. S2K146 showed EC50 values of 0.004, 0.018, and 0.113 μg/mL against Wuhan-Hu-1, BA.1, and XBB.1.5, respectively, whereas its EC50 values against BA.2.86 and later variants were all above 30 μg/mL. S2H13 showed only limited activity against Wuhan-Hu-1, with an EC50 of 0.96 μg/mL, and its EC50 values against all other variants were above 30 μg/mL. These results indicate that the two receptor-mimicking antibodies retained neutralizing activity better than the typical RBD antibody S2H13. Among them, VIR-7229 showed strong activity against early variants and BA.2.86, and still retained considerable neutralizing activity against later variants. VIR-7229 retains strong binding to S proteins from recent variants To further compare the binding profiles of the three antibodies against recent variants, recombinant soluble spike trimers corresponding to KP.3.1.1, XEC, and XEC.25.1 were expressed and purified (Fig. 3A).Fig. 3Binding kinetics of VIR-7229 antibody to spike proteins of SARS-CoV-2 variants. A) SDS-PAGE analysis of recombinant trimeric spike proteins. B–D) ELISA of antibody binding to KP.3.1.1, XEC, and XEC.25.1 spike proteins. Values in parentheses indicate the EC50 of antibodies. Data are presented as mean ± standard deviation from two independent experiments, each with three technical replicates. hE16, a previously humanized monoclonal antibody against West Nile virus E protein, was used as a negative control. E) Comparison of the binding kinetics of VIR-7229 to spike proteins from recent variants by biolayer interferometry assay. Protein A biosensors were first loaded with VIR-7229 and then incubated with 250 nM KP.3.1.1, XEC, or XEC.25.1 spike protein to record association and dissociation. His-tagged recombinant human B7-1/CD80 protein (hB7.1-His) was used as an irrelevant protein control; F–H) BLI analysis of VIR-7229 binding to KP.3.1.1, XEC, and XEC.25.1 spike proteins, with KD values calculated from the binding curves. Protein A biosensors were first loaded with VIR-7229 and then incubated with serially diluted spike proteins. Spike proteins were tested as threefold serial dilutions from 180 nM, with concentrations of 180, 60, 20, 6.67, and 2.22 nM. Vertical dashed lines indicate the transition from association to dissociation. BLI curves shown are representative results from one of two independent experiments. KD, equilibrium dissociation constant. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; BLI, biolayer interferometry; EC50, the half maximal effective concentration; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; nM, nmol/L. Binding kinetics of VIR-7229 antibody to spike proteins of SARS-CoV-2 variants. A) SDS-PAGE analysis of recombinant trimeric spike proteins. B–D) ELISA of antibody binding to KP.3.1.1, XEC, and XEC.25.1 spike proteins. Values in parentheses indicate the EC50 of antibodies. Data are presented as mean ± standard deviation from two independent experiments, each with three technical replicates. hE16, a previously humanized monoclonal antibody against West Nile virus E protein, was used as a negative control. E) Comparison of the binding kinetics of VIR-7229 to spike proteins from recent variants by biolayer interferometry assay. Protein A biosensors were first loaded with VIR-7229 and then incubated with 250 nM KP.3.1.1, XEC, or XEC.25.1 spike protein to record association and dissociation. His-tagged recombinant human B7-1/CD80 protein (hB7.1-His) was used as an irrelevant protein control; F–H) BLI analysis of VIR-7229 binding to KP.3.1.1, XEC, and XEC.25.1 spike proteins, with KD values calculated from the binding curves. Protein A biosensors were first loaded with VIR-7229 and then incubated with serially diluted spike proteins. Spike proteins were tested as threefold serial dilutions from 180 nM, with concentrations of 180, 60, 20, 6.67, and 2.22 nM. Vertical dashed lines indicate the transition from association to dissociation. BLI curves shown are representative results from one of two independent experiments. KD, equilibrium dissociation constant. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; BLI, biolayer interferometry; EC50, the half maximal effective concentration; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; nM, nmol/L. ELISA showed that both VIR-7229 and S2K146 bound to these three S proteins, but with different binding strengths (Fig. 3B–3D). VIR-7229 showed EC50 values of 0.008, 0.013, and 0.013 μg/mL for KP.3.1.1, XEC, and XEC.25.1, respectively, indicating strong binding to all three proteins. In comparison, S2K146 showed EC50 values of 0.044, 0.057, and 0.085 μg/mL, respectively, showing weaker binding than VIR-7229, with a further decrease for XEC.25.1. S2H13 showed binding signals close to background for all three S proteins. Overall, VIR-7229 showed stronger binding than S2K146 against these recent variants, whereas S2H13 largely lost recognition of these antigens. Although S2K146 lost effective neutralization from BA.2.86 onward, it still showed binding to the S proteins of the recent variants tested here. For VIR-7229, neutralization against KP.3.1.1, XEC, and XEC.25.1 was reduced, whereas binding to the corresponding S proteins remained strong. These results suggest that although the neutralizing activity of receptor-mimicking antibodies may decline as mutations accumulate, they can still retain partial epitope recognition, likely because their epitopes overlap with key regions required for receptor binding [33], [34]. BLI further showed that VIR-7229 bound to KP.3.1.1, XEC, and XEC.25.1 S proteins (Fig. 3E). Binding kinetics were then measured using serially diluted S proteins (Fig. 3F–3H). The KD values of VIR-7229 for KP.3.1.1, XEC, and XEC.25.1 were 5.46 nM, 0.864 nM, and 0.0164 nM, respectively (Fig. 3F–3H), indicating that VIR-7229 maintained high affinity for all three recent variant S proteins. Among them, VIR-7229 showed the highest affinity for XEC.25.1, followed by XEC and KP.3.1.1. These results show that VIR-7229 retained strong binding capacity against S proteins from recent variants. Structural analysis suggests a retained escape barrier for VIR-7229 Sequence alignment showed that KP.3.1.1 and XEC have identical RBD sequences, whereas XEC.25.1 carries an additional A435S substitution on the XEC background (Fig. 4A). Based on the VIR-7229-BA.2.86 RBD complex structure, we mapped the XEC substitutions onto the antibody binding interface and compared the local interactions. Although XEC carries L455S, F456L, and Q492E at this interface compared with BA.2.86, the key hydrogen bond interactions in this region were retained (Fig. 4B and 4C). This is consistent with the reported binding mode of VIR-7229, in which its tolerance to epitope variation is partly attributed to interactions with RBM backbone atoms [11]. Such a binding mode may make VIR-7229 less sensitive to side-chain mutations at the interface. However, in XEC RBD, residue 455 is changed from hydrophobic leucine to polar serine, which may weaken hydrophobic interactions with the VIR-7229 heavy chain variable region (VH). In addition, surface electrostatic analysis showed a weakly negative region on the VIR-7229 VH surface within 5 Å of Q492 in BA.2.86 RBD (Fig. 4D). In XEC RBD, Q492 is replaced by negatively charged glutamic acid, which may increase electrostatic repulsion with the VIR-7229 VH chain and reduce the binding affinity of VIR-7229 to XEC RBD. Based on the modeled complex of VIR-7229 with XEC.25.1 RBD, interface localization analysis showed that the XEC.25.1-specific A435S substitution was outside the direct VIR-7229 binding interface (Fig. 4E). This may explain why VIR-7229 showed similar neutralizing activity against XEC and XEC.25.1.Fig. 4Structural analysis of VIR-7229 binding to SARS-CoV-2 spike variants. A) RBD sequence alignment of different variants, visualized with ESPript 3.2. Green numbers below the alignment indicate cysteine residues. B–C) Local interactions of the VIR-7229 VH chain with residues 455, 456, and 492 in BA.2.86 RBD and XEC RBD, shown in PyMOL based on the structural model of VIR-7229 Fab bound to BA.2.86 RBD. Carbon atoms of RBD residues and VIR-7229 VH chain residues are shown in yellow and green, respectively. Nitrogen and oxygen atoms are shown in blue and red, respectively. Yellow dashed lines indicate hydrogen bonds, and numbers indicate interatomic distances in Å. D) Surface electrostatic potential distribution of the VH and VL chains of VIR-7229 Fab, shown using the APBS Electrostatics plugin in PyMOL. Q492, L455, and F456 in BA.2.86 RBD are displayed as sticks. Red indicates negative electrostatic potential, and blue indicates positive electrostatic potential. E) Structural comparison of XEC RBD and XEC.25.1 RBD around residue 435, shown in PyMOL. Residue 435 is shown as sticks to indicate the local change before and after the A435S substitution. The RBD is shown as a cyan cartoon, the VIR-7229 VH chain as a magenta cartoon, and the VIR-7229 VL chain as a blue cartoon. RBD residues within 5 Å of the VL chain are highlighted in orange. F) Binding free energy changes between VIR-7229 and BA.2.86, XEC, or XEC.25.1 RBD were calculated using the Flex ΔΔG protocol in Rosetta 3. The REU represents the relative energy score calculated by Rosetta to quantify the energetic states of protein conformations. Gray dots indicate independent calculations, and the red dashed line indicates the BA.2.86 RBD reference baseline. Data are summarized as the median with the interquartile range. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; RBD, receptor-binding domain; RBM, receptor binding motif; REU, Rosetta energy unit; VH, heavy chain variable region; VL, light chain variable region. Structural analysis of VIR-7229 binding to SARS-CoV-2 spike variants. A) RBD sequence alignment of different variants, visualized with ESPript 3.2. Green numbers below the alignment indicate cysteine residues. B–C) Local interactions of the VIR-7229 VH chain with residues 455, 456, and 492 in BA.2.86 RBD and XEC RBD, shown in PyMOL based on the structural model of VIR-7229 Fab bound to BA.2.86 RBD. Carbon atoms of RBD residues and VIR-7229 VH chain residues are shown in yellow and green, respectively. Nitrogen and oxygen atoms are shown in blue and red, respectively. Yellow dashed lines indicate hydrogen bonds, and numbers indicate interatomic distances in Å. D) Surface electrostatic potential distribution of the VH and VL chains of VIR-7229 Fab, shown using the APBS Electrostatics plugin in PyMOL. Q492, L455, and F456 in BA.2.86 RBD are displayed as sticks. Red indicates negative electrostatic potential, and blue indicates positive electrostatic potential. E) Structural comparison of XEC RBD and XEC.25.1 RBD around residue 435, shown in PyMOL. Residue 435 is shown as sticks to indicate the local change before and after the A435S substitution. The RBD is shown as a cyan cartoon, the VIR-7229 VH chain as a magenta cartoon, and the VIR-7229 VL chain as a blue cartoon. RBD residues within 5 Å of the VL chain are highlighted in orange. F) Binding free energy changes between VIR-7229 and BA.2.86, XEC, or XEC.25.1 RBD were calculated using the Flex ΔΔG protocol in Rosetta 3. The REU represents the relative energy score calculated by Rosetta to quantify the energetic states of protein conformations. Gray dots indicate independent calculations, and the red dashed line indicates the BA.2.86 RBD reference baseline. Data are summarized as the median with the interquartile range. Abbreviations: SARS-CoV-2, severe acute respiratory syndrome coronavirus 2; RBD, receptor-binding domain; RBM, receptor binding motif; REU, Rosetta energy unit; VH, heavy chain variable region; VL, light chain variable region. Rosetta binding free energy analysis further showed that, compared with BA.2.86, mutations in XEC and XEC.25.1 increased the predicted binding free energy of the VIR-7229 and RBD complex, with ΔΔG values of 2.486 ± 1.708 Rosetta energy unit (REU) and 2.782 ± 1.868 REU, respectively (Fig. 4F). These results suggest that the binding stability of VIR-7229 to the XEC and XEC.25.1 RBDs may be reduced, which may further explain why VIR-7229 showed comparable but weaker neutralizing activity against these two variants than against BA.2.86. In addition, recent studies have reported that circulating lineages carrying A435S, such as NB.1.8.1 and BA.3.2, still remain partially sensitive to VIR-7229 neutralization [35], [36]. Although the neutralizing activity was reduced compared with earlier strains, complete escape was not observed. Thus, these findings suggest that mutations in recent variants can affect local interactions between VIR-7229 and the RBD and reduce binding stability, but are still insufficient to completely disrupt the structural basis of VIR-7229 recognition. These results provide structural support for the partial resistance of VIR-7229 to immune escape by recent SARS-CoV-2 variants. Discussion During the continued spread of SARS-CoV-2, mutations have accumulated in the RBD. This region directly participates in ACE2 binding and is therefore constrained by the need to maintain viral entry [32], [37]. At the same time, it is under strong antibody selection pressure and serves as an important region for immune escape [38]. RBD evolution is therefore shaped by the balance between maintaining receptor binding and escaping antibody recognition [39]. As Omicron and its descendant lineages continue to evolve, many previously effective neutralizing antibodies have shown reduced activity against recent variants. In this context, evaluating whether and how receptor-mimicking antibodies retain activity during viral evolution is important for understanding SARS-CoV-2 antibody escape and assessing the potential of broad neutralizing antibodies. We therefore selected two representative receptor-mimicking antibodies, VIR-7229 and S2K146, both of which have broad activity and can neutralize across sarbecoviruses, and used the low-neutralizing RBD antibody S2H13 as a control. Neutralization assays showed that VIR-7229 and S2K146 retained activity across variants better than S2H13. Among them, VIR-7229 showed broader neutralization coverage in the tested variant panel. It retained neutralizing activity against BA.2.86 and its descendants, including KP.3.1.1, XEC, and XEC.25.1. In contrast, S2K146 showed a more pronounced loss of activity from BA.2.86 onward, whereas S2H13 had largely lost neutralizing activity against later variants at an earlier stage. These results suggest that VIR-7229 has an advantage in retaining neutralizing activity against recently emerged SARS-CoV-2 variants. To further examine antibody recognition of recently emerged variants, binding assays were performed using KP.3.1.1, XEC, and XEC.25.1 spike trimers. ELISA showed clear binding signals for both VIR-7229 and S2K146 against KP.3.1.1, XEC, and XEC.25.1 S proteins. Overall, VIR-7229 bound more strongly than S2K146, whereas S2H13 showed no detectable binding to these S proteins. Together with the pseudovirus neutralization results, these data show that although the neutralizing activity of VIR-7229 and S2K146 was reduced, their binding activity was still retained. These findings suggest that both VIR-7229 and S2K146 may retain antigen recognition by targeting functionally constrained ACE2 overlapping regions of the RBD. Although ELISA showed comparable VIR-7229 binding to the three S proteins, BLI revealed a clearer apparent affinity gradient, with XEC.25.1 showing the strongest binding, followed by XEC and KP.3.1.1. This difference may be due to the distinct assay formats. ELISA provides endpoint binding signals that are affected by antigen coating and multivalent avidity between full-length IgG and trimeric S proteins, whereas BLI measures real-time binding kinetics and may better capture differences in spike conformation. Computational structural analysis showed that, compared with BA.2.86, XEC and KP.3.1.1 carry the same substitutions near the VIR-7229 epitope, including L455S, F456L, and Q492E. However, key hydrogen bond interactions in this region were still maintained. Previous work showed that VIR-7229 can tolerate substitutions at RBD residue 455 by maintaining hydrogen bonds with the largely unchanged backbone atoms at this site [11]. In addition, the extra A435S substitution in XEC.25.1 is not located within the VIR-7229 and RBD binding interface. This is consistent with our neutralization and ELISA results, in which VIR-7229 showed similar activity against XEC and XEC.25.1. Rosetta binding free energy analysis also showed that mutations in XEC and XEC.25.1 reduced the predicted binding stability of the VIR-7229 and RBD complex to a similar extent. Notably, A435S is also present in other lineages, including NB.1.8.1 and BA.3.2, suggesting a possible convergent evolution pattern [30], [31]. Previous studies reported that VIR-7229 retained neutralizing activity against NB.1.8.1 and BA.3.2 [35], [36]. Together, these findings suggest that A435S does not directly disrupt the VIR-7229 binding interface, which may explain why some A435S-carrying variants remain sensitive to VIR-7229. This study provides a preliminary analysis of SARS-CoV-2 receptor-mimicking antibodies, such as VIR-7229 and S2K146, but it also has certain limitations. First, the number of SARS-CoV-2 variants included in this study was relatively limited. Although the selected variants covered the ancestral strain, early Omicron variants, and BA.2.86-related descendant lineages, they still represent only part of the antigenic evolution of SARS-CoV-2. Second, as a computational approach based on ΔΔG values, Rosetta analysis may not fully capture global conformational changes, solvent effects, or experimental conditions relevant to assessing binding stability. Third, only two representative receptor-mimicking antibodies were examined. Overall, our findings suggest that receptor-mimicking antibodies warrant continued attention as SARS-CoV-2 undergoes further antigenic evolution. Additional studies of these antibodies may help define conserved receptor binding epitopes and reveal how viral variants maintain receptor usage while escaping antibody pressure, thereby informing the evaluation of broadly neutralizing antibodies and the design of candidate immunogens.