Structural Prediction and Binding Dynamics of Zoonotic Spillover: Computational Modeling of Bat Coronavirus Spike-Receptor Interactions

Introduction

The emergence of zoonotic coronaviruses from bat reservoirs represents a persistent threat to mammalian health, including domestic animals and wildlife [1, 2]. Bat-borne coronaviruses (CoVs) possess spike glycoproteins that mediate host cell entry by binding to specific receptors, most notably angiotensin-converting enzyme 2 (ACE2) [3, 4]. The molecular determinants of cross-species transmission reside largely in the receptor-binding domain (RBD) of the spike protein, where specific amino acid residues govern host tropism and binding affinity [5, 6, 7]. Computational modeling of spike-receptor interactions has become an indispensable tool for predicting spillover risk, as it enables high-throughput screening of viral variants and animal receptors without the biosafety constraints of live-virus experiments [8, 9, 10].

This article reviews the principal computational methodologies used to predict bat coronavirus spike protein structures and to model their binding dynamics with host receptors. Emphasis is placed on the biophysical principles underlying these techniques, the key residues that drive host adaptation, and the integration of structural predictions with experimental cryo-electron microscopy (cryo-EM) validation [11, 12]. Data sources such as the GISAID database, the [Protein Data Bank](/knowledge/bioinformatics/protein-data-bank-formats-archival-validation 2) (PDB), and UniProt are essential for sequence-structure mapping and will be discussed in the context of their utility for veterinary virology [13].

Computational Methods for Spike Structure Prediction

Homology Modeling and Template-Based Approaches

When high-resolution experimental structures are unavailable, homology modeling remains a reliable first-line approach for constructing spike protein three-dimensional models. This method relies on aligning the target bat CoV spike sequence with one or more template structures of related coronaviruses (e.g., SARS-CoV or SARS-CoV-2 spike proteins) obtained from the PDB [14]. The quality of the resulting model depends heavily on sequence identity between target and template; for bat sarbecoviruses, identities typically exceed 70% in the RBD, permitting the generation of plausible structural ensembles [10]. Tools such as MODELLER or SWISS-MODEL are commonly used, with iterative refinement loops that optimize loop geometries and side-chain rotamers [15, 13].

Deep Learning-Based Prediction: AlphaFold2 and ESMFold

The advent of deep learning-based structure prediction has revolutionized the field. AlphaFold2 and the later ESMFold model have demonstrated remarkable accuracy in predicting the folded states of viral glycoproteins from primary sequence alone, without requiring homologous templates [16, 8]. For bat coronavirus spike proteins, these methods generate models that often align closely with experimentally derived cryo-EM structures, particularly in the core RBD fold [11, 9]. The key advantage of deep learning methods is their ability to capture conformational ensembles, which is critical for understanding the dynamic behavior of the spike during receptor engagement [8].

A comparison of structure prediction accuracy for bat sarbecovirus RBDs is summarized below.

Table 1. Comparison of computational methods for bat coronavirus spike RBD structure prediction. RMSD values are approximate and based on published benchmarks [8, 11, 10].

Molecular Dynamics Simulations of Spike-Receptor Complexes

Once a structural model of the spike RBD is obtained, molecular dynamics (MD) simulations are employed to study the binding dynamics and stability of the complex with host receptors such as ACE2 [7, 4]. All-atom MD simulations using force fields such as CHARMM or AMBER allow the tracking of atomic-level interactions over nanosecond to microsecond timescales [9]. Key observables include root-mean-square fluctuation (RMSF) of binding interface residues, hydrogen bond occupancy, and solvent accessible surface area (SASA) changes upon complex formation [15].

For zoonotic risk assessment, MD simulations can reveal how mutations at critical residues (e.g., position 403 in the RBD) alter binding affinity and conformational dynamics [3]. Simulations of bat SARS-like coronavirus WIV1 spike demonstrated that an adaptive mutation at residue 498 enhances ACE2 binding by stabilizing a specific loop conformation [9]. Similarly, MD of a MERS-like mink coronavirus (MCoV) confirmed that its spike uses ACE2 rather than DPP4, a finding that was first predicted by docking and later validated by experimental binding assays [6].

Protein-Protein Docking and Binding Affinity Prediction

Docking algorithms such as ZDOCK, HADDOCK, or ClusPro are routinely used to predict the geometry of the spike RBP-receptor complex when a crystal structure is not available [5, 14]. The docking solutions are then scored using physics-based energy functions or knowledge-based potentials. To improve accuracy, ensemble docking can be performed using multiple conformations of the spike RBD extracted from MD trajectories [10]. Binding free energies are subsequently estimated using methods like Molecular Mechanics Generalized Born Surface Area (MM-GBSA) or Poisson-Boltzmann Surface Area (MM-PBSA) [7].

For bat coronaviruses, docking studies have identified species-specific ACE2 residues that modulate binding: for example, residue 31 of human ACE2 (Glu31) forms a crucial salt bridge with spike RBD residue Lys403, and its conservation across mammals influences host range [3, 4]. Screening of candidate intermediate hosts, such as those performed for porcine respiratory coronavirus, relies on docking these spike proteins to ACE2 orthologs from various animal species [5].

The following Mermaid diagram outlines a typical computational workflow for evaluating bat coronavirus spillover risk.

flowchart TD
    A[Bat Coronavirus Spike Sequence], > B[Structural Prediction: AlphaFold2 / Homology Modeling]
    B, > C[Host Receptor Sequence (e.g. ACE2 from target species)]
    C, > D[Protein-Protein Docking]
    D, > E[MD Simulation of Complex]
    E, > F[Binding Free Energy Calculation (MM-GBSA)]
    F, > G{Key Interface Mutations Identified?}
    G, >|Yes| H[Assess Impact on Affinity via in silico Mutagenesis]
    G, >|No| I[Compare with Experimental Cryo-EM / Binding Data]
    H, > I
    I, > J[Zoonotic Spillover Risk Assessment]

Figure 1. Computational workflow for predicting bat coronavirus spike-receptor binding dynamics and zoonotic risk.

Key Residues Driving Host Tropism

The evolutionary trajectory of bat coronavirus spikes is dictated by a relatively small set of amino acid positions within the receptor-binding motif (RBM) [3, 2]. Comprehensive mutational scanning and structural modeling have pinpointed residues that act as molecular switches for host range [7, 13]. The table below summarizes critical RBD positions and their reported effects on ACE2 binding across different coronaviruses.

Table 2. Key RBD residues influencing bat coronavirus spike binding to human ACE2.

Residue 403 has been shown to be a particularly sensitive determinant: substitution of lysine with a neutral residue at this position completely abolishes binding of certain bat sarbecovirus spikes to human ACE2 [3]. Conversely, mutation of residue 498 from glutamine to histidine, observed in WIV1-CoV, enhances human ACE2 binding by 3-fold as measured by surface plasmon resonance, a result that correlates with MD-derived binding free energy differences [9].

Beyond ACE2, some bat coronaviruses utilize alternative receptors. For instance, a MERS-like mink coronavirus (MCoV) was shown to use ACE2 rather than DPP4, highlighting the importance of broad-receptor docking screens [6]. Sialic acid binding via the hemagglutinin-esterase (HE) protein also contributes to host tropism by modulating virion avidity, and co-evolution of HE and spike has been documented in betacoronaviruses [17, 12].

Case Studies in Computational Spillover Prediction

Bat Sarbecovirus WIV1

The bat SARS-like coronavirus WIV1 (closely related to SARS-CoV) has been extensively studied using computational tools. Cryo-EM structures of its locked spike trimer reveal a pre-fusion conformation similar to SARS-CoV, and MD simulations across multiple host ACE2 orthologs highlight dynamic differences in the RBM that control species specificity [11]. A specific adaptive mutation (Q498H) was identified through phylogenetic analysis and then shown by MD and free energy calculations to stabilize the RBM loop and enhance human ACE2 binding [9]. Protein interaction mapping between bat and human cells further revealed rewired networks that govern immune evasion, which in turn can affect the probability of spillover [16].

Deltacoronavirus and Alternative Receptors

Deltacoronaviruses, such as porcine deltacoronavirus (PDCoV), exhibit a broader receptor usage that includes aminopeptidase N (APN). MD simulations of PDCoV spike-receptor complexes have demonstrated that receptor affinity-selective differential dynamics of membrane fusion initiation govern cross-species transmission [15]. The simulations suggest that subtle changes in the spike's fusion peptide region, rather than the RBD alone, can influence host range by altering the kinetics of membrane fusion [15, 17].

MERS-like Mink Coronavirus

A novel MERS-like coronavirus isolated from mink was initially assumed to use DPP4 as its receptor. However, computational docking with human ACE2 followed by MD simulations predicted a stable complex, which was subsequently confirmed by experimental entry assays [6]. This case underscores the value of unbiased docking screens against multiple potential receptors, especially when intermediate host species (e.g., mink) are implicated [5].

Implications for Zoonotic Spillover Risk Assessment

Integrating structural prediction and binding dynamics into risk assessment frameworks allows veterinary virologists to prioritize surveillance efforts. Computational pipelines can rapidly evaluate the spike sequences of newly discovered bat coronaviruses against a panel of domestic animal and human ACE2 orthologs [5, 8, 1]. The emergence time of SARS-CoV-2, for example, has been estimated using molecular clock analyses combined with binding affinity predictions, providing a timeline for potential spillover events [1].

Key data resources for such analyses include:

• GISAID: repository of genomic sequences and metadata for coronaviruses, including bat isolates.
• Protein Data Bank (PDB): source of experimental structures for spike proteins and receptor complexes.
• UniProt: annotated protein sequences and functional information for host receptors across species.

These databases supply the raw material for computational modeling, and their regular curation is essential for accurate predictions [14, 2].

Conclusion

Computational modeling of bat coronavirus spike-receptor interactions has matured into a powerful discipline that informs zoonotic spillover risk assessments in veterinary medicine. Homology modeling and deep learning methods (AlphaFold2, ESMFold) provide accurate structural predictions, while MD simulations and docking reveal the dynamic and energetic determinants of host tropism. Key residues such as spike positions 403, 498, and 501 act as molecular switches that control ACE2 binding affinity, and their mutation profiles can be monitored through continuous surveillance [3, 10].

The combination of computational predictions with experimental validation (e.g., cryo-EM, surface plasmon resonance) remains the gold standard, but even in silico-only screens can prioritize which bat coronaviruses warrant further investigation [11, 9]. As high-throughput sequencing of bat populations accelerates, these computational tools will become increasingly vital for preemptively identifying viruses with pandemic potential and for guiding veterinary public health interventions [16, 4].

References

[1] Samson S, Lord É, Makarenkov V. Assessing the emergence time of SARS-CoV-2 zoonotic spillover. PLoS One. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/38574109/

[2] Malik YS, Ansari MI, Kattoor JJ, et al. Evolutionary and codon usage preference insights into spike glycoprotein of SARS-CoV-2. Brief Bioinform. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/33377145/

[3] Zech F, Schniertshauer D, Jung C, et al. Spike residue 403 affects binding of coronavirus spikes to human ACE2. Nat Commun. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34824253/

[4] Rynkiewicz P, Lynch ML, Cui F, et al. Functional binding dynamics relevant to the evolution of zoonotic spillovers in endemic and emergent Betacoronavirus strains. J Biomol Struct Dyn. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/34286673/

[5] Sootichote R, Chamkasem A, Toniti W, et al. Screening candidate intermediate hosts for porcine respiratory coronavirus using molecular docking. Comp Immunol Microbiol Infect Dis. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42361779/

[6] Wang N, Ji W, Jiao H, et al. A MERS-CoV-like mink coronavirus uses ACE2 as an entry receptor. Nature. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40306315/

[7] Rajendran M, Babbitt GA. Persistent cross-species SARS-CoV-2 variant infectivity predicted via comparative molecular dynamics simulation. R Soc Open Sci. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/36340517/

[8] Balogun TA, Kearns FL, Calvó-Tusell C, et al. Structural dynamics and allosteric communication of a SARS-like bat coronavirus spike glycoprotein. Biophys J. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42026866/

[9] Tse AL, Lasso G, Berrigan J, et al. Bat sarbecovirus WIV1-CoV bears an adaptive mutation that alters spike dynamics and enhances ACE2 binding. PLoS Pathog. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/41100556/

[10] Rodrigues JPGLM, Barrera-Vilarmau S, M C Teixeira J, et al. Insights on cross-species transmission of SARS-CoV-2 from structural modeling. PLoS Comput Biol. 2020. URL: https://pubmed.ncbi.nlm.nih.gov/33270653/

[11] Liu C, Zheng J, Wang Y, et al. Cryo-EM structure of locked spike glycoprotein from bat SARS-like coronavirus WIV1, molecular dynamics and biophysics across host range. Proc Natl Acad Sci U S A. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41706884/

[12] Lang Y, Li W, Li Z, et al. Coronavirus hemagglutinin-esterase and spike proteins coevolve for functional balance and optimal virion avidity. Proc Natl Acad Sci U S A. 2020. URL: https://pubmed.ncbi.nlm.nih.gov/32994342/

[13] Prates ET, Garvin MR, Pavicic M, et al. Potential Pathogenicity Determinants Identified from Structural Proteomics of SARS-CoV and SARS-CoV-2. Mol Biol Evol. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/32941612/ *** 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.

[14] Elkazzaz M, Ahmed A, Abo-Amer YE, et al. In Silico Discovery of GPCRs and GnRHRs as Novel Binding Receptors of SARS-CoV-2 Spike Protein Could Explain Neuroendocrine Disorders in COVID-19. Vaccines (Basel). 2022. URL: https://pubmed.ncbi.nlm.nih.gov/36146578/

[15] An D, Peng Q, Ma YH, et al. Receptor affinity-selective differential dynamics of membrane fusion initiation govern deltacoronavirus cross-species transmission. Sci China Life Sci. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/41144156/

[16] Batra J, Rutkowska M, Zhou Y, et al. Coronavirus protein interaction mapping in bat and human cells reveals network rewiring governing immune evasion and zoonotic potential. Cell Host Microbe. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42134328/

[17] Kuchipudi SV, Nelli RK, Gontu A, et al. Sialic Acid Receptors: The Key to Solving the Enigma of Zoonotic Virus Spillover. Viruses. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/33567791/