Deep mutational scanning and structural modeling of SARS-CoV-2 receptor binding domain: Predicting escape from monoclonal antibodies

Introduction

The receptor binding domain (RBD) of the SARS-CoV-2 spike glycoprotein is the primary target of neutralizing antibodies elicited by infection or vaccination [1, 2]. The RBD mediates attachment to the angiotensin-converting enzyme 2 (ACE2) receptor on host cells, a step essential for viral entry [3, 4]. Continuous antigenic drift in the RBD, driven by selective pressure from polyclonal and monoclonal antibody responses, leads to the emergence of escape variants that can evade neutralization [5, 6]. Understanding the mutational landscape of the RBD and its impact on antibody recognition is critical for designing effective therapeutic antibodies and for predicting viral evolution [7, 8]. This article reviews the integration of deep mutational scanning (DMS) experiments with computational structural modeling to systematically map antibody escape mutations in the SARS-CoV-2 RBD. The methods described are broadly applicable to other coronaviruses of veterinary importance, including those affecting companion animals and livestock, where similar RBD-antibody interfaces exist [9, 10].

Deep mutational scanning of the RBD

Deep mutational scanning is a high-throughput experimental approach that quantifies the functional impact of every possible single amino acid substitution in a protein of interest [1, 2]. For the SARS-CoV-2 RBD, DMS libraries are typically generated by site-directed mutagenesis and expressed on the surface of yeast or mammalian cells, followed by selection with a monoclonal antibody (mAb) [3, 4]. The relative enrichment or depletion of each variant after selection is measured by high-throughput sequencing, yielding a comprehensive escape map [5, 6]. DMS has revealed that the RBD exhibits extensive mutational tolerance at many positions, but critical residues for ACE2 binding and antibody recognition are highly constrained [7, 8]. Epistatic interactions between mutations can shift the fitness landscape over time, as demonstrated by [1] in their analysis of recent SARS-CoV-2 variants. The authors showed that amino acid preferences at certain hotspot residues change depending on the genetic background, highlighting the importance of context-dependent mutational effects [1].

DMS has been applied to multiple SARS-CoV-2 lineages, including Omicron subvariants JN.1 and XEC, to map the antibody escape and infectivity landscape [2]. Those experiments demonstrated that the RBD of these variants harbors mutations that simultaneously reduce antibody binding and maintain or enhance ACE2 affinity [2]. The breadth of DMS data allows for the construction of detailed fitness landscapes that can be used to predict future evolutionary trajectories [3, 4]. For example, [3] developed a protein language model informed by DMS data that predicts the spatiotemporal evolution of the RBD with high accuracy. Similarly, [4] used machine learning on DMS-derived fitness landscapes to simulate the emergence of escape variants under different antibody pressures.

Structural modeling of RBD-antibody interfaces

Computational structural modeling complements DMS by providing a three-dimensional context for mutational effects [5, 6]. The RBD adopts a compact, antiparallel beta-sheet structure with a receptor binding motif (RBM) that directly contacts ACE2 [11, 12]. Antibodies typically bind to the RBD at the ACE2 interface (class 1 and 2) or at more distal epitopes (class 3 and 4) [5, 6]. Structural models of RBD-antibody complexes, derived from crystallography, cryo-electron microscopy, or computational prediction tools such as AlphaFold2 and Rosetta, allow detailed mapping of the paratope-epitope interface [7, 8]. [11] performed a comprehensive in silico structural analysis of the SARS-CoV-2 RBD, identifying key molecular determinants for virus-host interaction. Their work highlighted the role of specific hydrogen bonds and hydrophobic contacts in ACE2 binding and antibody recognition [11]. [12] modeled the dynamics of coronavirus spike proteins, showing that conformational fluctuations in the RBD can influence epitope accessibility and immune escape.

The integration of structural models with DMS data enables the identification of residues that are both critical for antibody binding and structurally permissive for mutation [5, 6]. For instance, mutations at the antibody interface that cause steric clashes or loss of hydrogen bonds are likely to confer escape, while mutations that destabilize the RBD fold are less tolerated [9, 10]. [9] used a combination of DMS and structural analysis to characterize how clonal interference and changing selective pressures shape the escape of SARS-CoV-2 from hundreds of antibodies. They found that escape mutations often cluster in a few epitope regions that are structurally vulnerable [9]. [10] further demonstrated that protein language models can capture the evolutionary potential of the RBD by learning from both sequence and structural data.

Integrating DMS and structural data for escape prediction

The combination of DMS and structural modeling yields a powerful framework for predicting antibody escape [1, 2]. The workflow typically involves the following steps: (1) generation of a DMS library for the RBD, (2) selection with a panel of mAbs, (3) high-throughput sequencing to quantify variant frequencies, (4) mapping of enriched variants onto the three-dimensional structure of the RBD-antibody complex, and (5) computational validation using energy calculations or molecular dynamics simulations [3, 4]. [5] developed a predictive model of immune escape that combines DMS data with structural features, achieving high accuracy in grouping antigenic variants. Their model uses a random forest classifier trained on residue-level escape scores, structural distance to the antibody, and local flexibility [5]. [6] extended this approach by incorporating serum polyclonal antibody escape data, demonstrating that deep mutational learning can dissect the contributions of individual epitopes to overall neutralization escape.

The structural context is especially important for interpreting mutations that are distal to the antibody binding site but still affect escape through allosteric mechanisms [7, 8]. [7] reviewed how RBD evolution from ACE2 binding optimization to immune epitope remodeling involves both direct and indirect effects. Mutations outside the epitope can alter the conformational dynamics of the RBD, making it less accessible to antibodies [8]. [8] developed a modeling framework that integrates receptor binding and immunity to predict the fitness landscape of SARS-CoV-2, showing that trade-offs between ACE2 affinity and antibody evasion are key determinants of variant success.

Validation through sequence surveillance

Predictions of antibody escape from DMS and structural modeling must be validated against real-world sequence data derived from viral surveillance [1, 2]. Large-scale sequencing efforts, such as those deposited in the GISAID database, provide a rich resource for tracking the emergence and spread of RBD mutations over time [3, 4]. [1] used DMS data from recent variants to show that changes in amino acid preferences at epistatic hotspots are reflected in circulating sequences. [2] validated their DMS-based escape maps by comparing them with the mutational profiles of JN.1 and XEC lineages observed in global surveillance. Similarly, [3] demonstrated that their DMS-informed language model could predict the dominance of specific RBD mutations months before they became prevalent in the population.

The concordance between experimental DMS results and sequence data strengthens confidence in the predictive power of these approaches [5, 6]. Discrepancies can arise due to differences in the selection pressure applied in vitro versus in vivo, or due to the influence of polyclonal sera that target multiple epitopes simultaneously [7, 8]. [9] addressed this by analyzing the escape landscape from hundreds of individual antibodies, showing that the combination of multiple escape mutations is required to evade polyclonal responses. [10] used sequence data from the entire pandemic to train a protein language model that captures the evolutionary trajectory of the RBD, further validating the relevance of DMS-derived fitness landscapes.

Implications for therapeutic antibody design

The insights gained from DMS and structural modeling have direct implications for the design of monoclonal antibodies with improved breadth and durability [11, 12]. By identifying conserved epitopes that are mutationally constrained, researchers can prioritize antibodies that target regions of the RBD that are essential for ACE2 binding and thus less likely to escape [13]. [13] determined the molecular determinants of an antibody cocktail that prevents SARS-CoV-2 escape, showing that targeting two non-overlapping epitopes reduces the probability of resistance. Structural modeling of the RBD-antibody interface can guide the engineering of antibodies with higher affinity and broader neutralization profiles [11, 12]. For veterinary applications, similar approaches can be applied to animal coronaviruses such as feline infectious peritonitis virus (FIPV) or porcine epidemic diarrhea virus (PEDV), where RBD-targeting antibodies are used therapeutically [9, 10]. The methods described here are also relevant for designing antibodies against other viral glycoproteins, such as influenza hemagglutinin [14].

Workflow diagram

The following Mermaid diagram illustrates the integrated workflow for predicting antibody escape from the SARS-CoV-2 RBD using DMS and structural modeling.

flowchart TD
    A[Generate RBD DMS Library], > B[Select with mAb Panel]
    B, > C[High-Throughput Sequencing]
    C, > D[Compute Enrichment Scores]
    D, > E[Map Escape Mutations to Structure]
    E, > F[Structural Modeling of RBD-Ab Complex]
    F, > G[Predict Escape Candidates]
    G, > H[Validate with Sequence Surveillance]
    H, > I[Iterate with New mAbs]

Conclusion

Deep mutational scanning and computational structural modeling together provide a robust framework for systematically mapping antibody escape mutations in the SARS-CoV-2 RBD. The integration of experimental fitness data with three-dimensional structural context enables the prediction of mutations that are likely to emerge under selective pressure from monoclonal antibodies. Validation against global sequence surveillance data confirms the relevance of these predictions for real-world viral evolution. These approaches inform the design of therapeutic antibodies with reduced susceptibility to escape and are extendable to other coronaviruses of veterinary significance. Continued advances in DMS technology, structural prediction algorithms, and machine learning will further refine our ability to anticipate viral antigenic drift.

References

[1] Taylor AL, Starr TN. Deep mutational scanning of recent SARS-CoV-2 variants highlights changing amino acid preferences within epistatic hotspot residues. PLoS Pathog. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42330076/

[2] Shao C, Yang L, Xiao C, et al. Deep mutational scanning reveals the antibody escape and infectivity landscape of SARS-CoV-2 Omicron JN.1 and XEC receptor-binding domains. Emerg Microbes Infect. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42324717/

[3] Yang S, Luo X, Luo J, et al. A deep mutational scanning-informed protein language model predicts SARS-CoV-2 evolution dynamics with spatiotemporal resolution. Nat Microbiol. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42204343/

[4] Durumeric AEP, McCarty S, Smith J, et al. Machine Learning-Driven Simulations of the SARS-CoV-2 Fitness Landscape from Deep Mutational Scanning Experiments. J Chem Inf Model. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42089465/

[5] Nasir A, Lee D, Avena LE, et al. Predictive modeling of immune escape and antigenic grouping of SARS-CoV-2 variants. J Virol. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42037411/

[6] Shlesinger D, Sadilek V, Minot M, et al. Dissecting serum polyclonal antibody escape to SARS-CoV-2 variants by deep mutational learning. Cell Rep Methods. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42030951/

[7] Soliman OA, Shahine Y, Baecker D, et al. Beyond the Mutation Abyss: Revisiting SARS-CoV-2 Receptor-Binding Domain Evolution from ACE2 Binding Optimization to Immune Epitope Remodeling. Pathogens. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41901725/

[8] Ding Z, Yuan HY. The role of receptor binding and immunity in SARS-CoV-2 fitness landscape: A modeling study. iScience. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41809055/

[9] Haddox HK, Abdel Aziz O, Galloway JG, et al. Clonal interference and changing selective pressures shape the escape of SARS-CoV-2 from hundreds of antibodies. Virus Evol. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41767406/

[10] Lamb KD, Hughes J, Lytras S, et al. From single-sequences to evolutionary trajectories: protein language models capture the evolutionary potential of SARS-CoV-2. Nat Commun. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41714330/

[11] Nelson-Sathi S, Umasankar PK, Sreekumar E, et al. Mutational landscape and in silico structure models of SARS-CoV-2 spike receptor binding domain reveal key molecular determinants for virus-host interaction. BMC Mol Cell Biol. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/34991443/

[12] Kunkel G, Madani M, White SJ, et al. Modeling coronavirus spike protein dynamics: implications for immunogenicity and immune escape. Biophys J. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34767789/

[13] Ku Z, Xie X, Davidson E, et al. Molecular determinants and mechanism for antibody cocktail preventing SARS-CoV-2 escape. Nat Commun. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/33473140/ *** Disclaimer: This article is for educational and informational purposes only. It is not intended to substitute for professional veterinary advice, diagnosis, treatment, or regulatory guidance. Always consult a licensed veterinarian or qualified specialist regarding animal health, disease diagnosis, and therapeutic decisions.