Understanding the Structural Dynamics of Zoonotic Spillover: Computational Modeling of Receptor-Binding Domain Evolution in Bat Coronaviruses

Introduction

Bat coronaviruses constitute a vast reservoir of genetic diversity with demonstrated capacity for cross-species transmission into mammalian hosts. The receptor-binding domain (RBD) of the viral spike glycoprotein governs host cell entry via interaction with species-specific receptors such as angiotensin-converting enzyme 2 (ACE2) and dipeptidyl peptidase 4 (DPP4) [1]. Predicting which bat coronaviruses possess the structural potential to productively bind receptors of novel hosts remains a central challenge in veterinary virology and pandemic preparedness. Computational modeling approaches, including molecular dynamics (MD) simulations, free energy perturbation, and deep learning, now enable quantitative assessment of RBD evolution and its implications for zoonotic spillover [1, 2].

This review focuses on computational methods applied to bat coronavirus RBD structural dynamics, emphasizing the biophysical principles underlying RBD-receptor interactions. The discussion is framed within a veterinary context, with particular attention to bat-derived viruses and their potential to infect livestock and companion animals. For a broader overview of receptor-binding dynamics, readers are directed to the article Predicting Zoonotic Spillover: Computational Modeling of Receptor-Binding Dynamics in Emerging Bat Coronaviruses.

Structural Biology of the Bat Coronavirus Spike RBD

The spike glycoprotein of betacoronaviruses adopts a trimeric prefusion conformation. Each protomer contains an S1 subunit housing the RBD and an S2 subunit responsible for membrane fusion [1]. The RBD exists in two conformational states: a "standing" or "up" conformation that exposes the receptor-binding motif (RBM) and a "lying" or "down" conformation that occludes the RBM. Allosteric communication between RBDs within the trimer modulates the equilibrium of these states [1].

Functional binding dynamics relevant to zoonotic spillover have been characterized through large-scale MD simulations [2]. Comparative analyses of multiple betacoronavirus strains (hCoV-OC43, hCoV-HKU1, MERS-CoV, SARS-CoV-1, and SARS-CoV-2) revealed a conserved interaction site between the N-terminal helices of ACE2 and the viral RBD [2]. A second, more dynamically complex interaction site involving ACE2 residues K353 and Q325, as well as a novel motif (AAQPFLL, residues 386-392), was identified in strains associated with recent cross-species spillover [2]. This motif was absent in hCoV-OC43, suggesting a potential evolutionary marker for enhanced zoonotic potential.

Bat coronavirus HKU4, a suspected progenitor of MERS-like viruses, was modeled in complex with both ACE2 and CD26 (the MERS receptor). Simulations showed that the "two-touch" interaction sites governing RBD binding activity in human strains also modulated binding in HKU4 hybrids [2]. These findings underscore the structural plasticity of bat coronavirus RBDs and their capacity to engage multiple host receptors.

Computational Methods for Modeling RBD Evolution

Molecular Dynamics Simulations

MD simulations provide atomistic description of RBD-receptor complexes. Force fields such as CHARMM and AMBER are applied to solvated systems, and trajectories are analyzed for root-mean-square fluctuations (RMSF), hydrogen bond occupancy, and binding free energies [1, 2]. Replicate simulations with GPU acceleration allow statistical comparison of atom fluctuations in bound versus unbound states [2]. The article GROMACS Molecular Dynamics: Setting Up, Simulating, and Analyzing Protein-Water Systems provides practical protocols for such simulations.

Free Energy Calculations

Binding free energy can be estimated using Molecular Mechanics Poisson-Boltzmann Surface Area (MM-PBSA) or Generalized Born Surface Area (MM-GBSA) methods. These approaches decompose interaction energetics by residue, identifying "hot spots" critical for binding affinity [1]. Alanine scanning mutagenesis in silico further validates these hot spots by quantifying the energetic impact of side chain truncation [2].

Structural Prediction with Deep Learning

AlphaFold2 enables high-accuracy prediction of RBD structures when experimental templates are absent. This is particularly valuable for newly discovered bat coronaviruses with low sequence identity to known strains [1]. Predicted structures serve as starting points for MD simulations and docking. For further reading, see Structural Prediction of Viral Envelope Glycoproteins Using AlphaFold2: Implications for Host Receptor Binding and Vaccine Design.

Computational Mutagenesis

Computational mutagenesis systematically mutates RBD residues and recalculates binding affinity to assess the impact of naturally occurring or hypothetical substitutions. Rynkiewicz et al. employed this technique to confirm the functional relevance of the ACE2 K353 and AAQPFLL sites [2]. Such scans can be combined with sequence surveillance to prioritize variants with elevated spillover risk.

The table below summarizes key computational methods and their application to bat coronavirus RBD studies.

Key Studies on Bat Coronavirus RBD Dynamics

Allosteric Communication in SARS-like Bat Coronavirus Spike

Balogun et al. investigated the structural dynamics and allosteric communication of a SARS-like bat coronavirus spike glycoprotein [1]. Using MD simulations and network analysis, they identified residue subnetworks that transmit conformational signals between the RBD and the S2 fusion machinery. Perturbations in the RBD, such as mutations or receptor binding, altered the free energy landscape of the trimer, influencing the equilibrium between prefusion and postfusion states [1]. This allosteric network provides a mechanism by which RBD evolution can modulate the overall spike stability and fusion competence, factors critical for host cell entry.

Functional Binding Dynamics and the "One Touch/Two Touch" Model

Rynkiewicz et al. performed large replicate sets of GPU-accelerated MD simulations to compare atom fluctuations of ACE2, ACE1, and CD26 in the presence and absence of RBDs from six betacoronavirus strains [2]. Their analysis revealed a common interaction site at the N-terminal helices of ACE2 across all strains (hCoV-OC43, hCoV-HKU1, MERS-CoV, SARS-CoV-1, SARS-CoV-2). A second, more complex interaction site involving ACE2 residues K353, Q325, and the AAQPFLL motif (386-392) was associated with the more recent and pathogenic strains [2]. Bat coronavirus HKU4, when modeled with both ACE2 and CD26, also engaged these two sites.

From these observations, the authors proposed a "one touch/two touch" model of viral evolution [2]. According to this model, ancestral strains rely primarily on the conserved N-terminal helix interaction ("one touch"), whereas more recent strains have evolved supplementary contacts ("two touch") that enhance binding affinity and facilitate cross-species spillover. Computational mutagenesis confirmed that disruption of the second site significantly reduced RBD-ACE2 binding in SARS-like strains but had minimal effect on hCoV-OC43 [2].

The workflow below illustrates a typical computational pipeline for assessing zoonotic spillover potential from bat coronavirus RBD sequences.

flowchart TD
    A[Bat Coronavirus RBD Sequence], > B[Sequence Alignment & Phylogenetic Analysis]
    B, > C[Template Selection & Homology Modeling\nor AlphaFold2 Prediction]
    C, > D[Model Refinement & Validation\n(Ramachandran, pLDDT)]
    D, > E[MD Simulations of RBD-Receptor Complex\n(GROMACS, CHARMM force field)]
    E, > F[Trajectory Analysis\n(RMSF, hydrogen bonds, PCA)]
    F, > G[Binding Free Energy Calculation\n(MM-PBSA/MM-GBSA)]
    G, > H[Computational Mutagenesis\n(Alanine scanning, saturation mutagenesis)]
    H, > I[Identification of Key Residues\n& Allosteric Networks]
    I, > J[Machine Learning Classification\n(Zoonotic risk score)]
    J, > K[Experimental Validation\nRecommended]
    K, > L[Risk Assessment & Surveillance\nPrioritization]

Integration with Machine Learning for Spillover Prediction

Machine learning models trained on structural and energetic features derived from MD simulations can classify bat coronavirus strains by spillover risk. Features such as binding free energy, number of interfacial hydrogen bonds, and conformational entropy of the RBM are used as inputs [1, 2]. Deep learning approaches further incorporate sequence embeddings and predicted structural properties (see Deep Learning-Driven Prediction of Viral Receptor-Binding Domain Mutations: A Computational Virology Approach to Zoonotic Risk Assessment).

AlphaFold2 has been used to generate structures for poorly characterized bat coronaviruses, enabling docking against host ACE2 orthologs from livestock species (e.g., swine, bovine, camel) [1]. The combination of structural prediction with MD-based refinement and machine learning scoring constitutes a powerful pipeline for proactive surveillance. For a broader discussion of machine learning applications, refer to Machine Learning-Driven Prediction of Receptor-Binding Dynamics in Emerging Zoonotic Coronaviruses.

Implications for Veterinary Surveillance

Bats harbor coronaviruses genetically related to pathogens that cause significant disease in livestock, such as porcine epidemic diarrhea virus (PEDV) and transmissible gastroenteritis virus (TGEV). Although these viruses utilize aminopeptidase N rather than ACE2, the principles of RBD structural evolution apply [1]. Computational modeling can identify bat coronaviruses carrying RBDs capable of binding to livestock receptors.

Furthermore, the "one touch/two touch" model [2] offers a framework for monitoring RBD sequence changes in bat populations. Surveillance efforts can prioritize strains that acquire second-site interacting residues analogous to those enhancing ACE2 binding in human pathogens. Such strains may pose elevated risk for spillover into domestic animals. The article Zoonotic Spillover Pathways and Receptor Binding Evolution in Bat Reservoirs provides additional context on reservoir ecology.

For interactive three-dimensional visualization of RBD-ACE2 complexes and the effect of key mutations, the site's 3D Protein Viewer can be employed. Users are encouraged to load models of bat coronavirus RBDs docked to different host ACE2 orthologs and to apply in silico mutagenesis tools to explore structural consequences in real time.

Future Directions

Several methodological advances are poised to enhance the predictive power of computational RBD modeling. Enhanced sampling techniques such as Markov state models and replica exchange MD can capture rare conformational events relevant to receptor binding [1]. Integration with deep mutational scanning datasets allows training of sequence-to-function models. For a review of Markov state approaches, see Markov State Models in Molecular Dynamics Simulations.

The development of end-to-end differentiable models that directly predict binding affinity from sequence, bypassing explicit MD, is an active area of research. Such models, when trained on large corpora of bat coronavirus RBD variants, could accelerate real-time risk assessment during field surveillance [2].

Conclusion

Computational modeling of bat coronavirus RBD structural dynamics provides mechanistic insight into the molecular determinants of zoonotic spillover. All-atom MD simulations have revealed both conserved and strain-specific interaction sites on ACE2 [1, 2]. The "one touch/two touch" model offers a testable hypothesis for how RBD evolution enhances cross-species binding [2]. By integrating molecular dynamics, free energy calculations, deep learning structural prediction, and computational mutagenesis, veterinary virologists can systematically evaluate the spillover potential of bat coronaviruses. These tools, combined with field surveillance and experimental validation, constitute a robust framework for predicting and mitigating emerging zoonotic threats to animal and public health.

References

[1] Balogun TA, Kearns FL, Calvó-Tusell C, et al. Structural dynamics and allosteric communication of a SARS-like bat coronavirus spike glycoprotein. Biophysical Journal. URL: https://pubmed.ncbi.nlm.nih.gov/42026866/

[2] Rynkiewicz P, Lynch ML, Cui F, et al. Functional binding dynamics relevant to the evolution of zoonotic spillovers in endemic and emergent Betacoronavirus strains. bioRxiv. URL: https://www.semanticscholar.org/paper/0401d2fc2c4d55c7627251d9dd2e7450127a3c9b *** 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.