Section: Computational Biology

Structural characterization of viral polymerase-host factor complexes using hybrid modeling

Viral RNA-dependent RNA polymerases (RdRps) and DNA polymerases are central to genome replication and transcription in animal viruses. These enzymes do not function in isolation; they recruit host proteins to form active replication complexes [1, 2]. The contact points between viral polymerases and host factors are critical determinants of host range, tissue tropism, and pathogenesis [2, 3]. In veterinary virology, understanding these interfaces is essential for rational vaccine design and antiviral development [4]. Traditional structural methods such as X-ray crystallography have provided high-resolution snapshots of individual domains, but they often fail to capture the dynamic and multifactorial nature of polymerase-host interactions [5]. Hybrid modeling, which integrates data from cryo-electron microscopy (cryo-EM), crosslinking mass spectrometry (XL-MS), and computational docking, has emerged as a powerful strategy to generate comprehensive structural models of these macromolecular assemblies [6, 7].

Hybrid modeling workflow

The hybrid modeling pipeline typically proceeds through iterative cycles of data acquisition, spatial restraint generation, and computational optimization [6]. The workflow is summarized in Figure 1.

flowchart TD
    A[Purification of native polymerase-host complexes], > B[Cryo-EM single-particle analysis]
    A, > C[Crosslinking mass spectrometry (XL-MS)]
    B, > D[Density map at intermediate resolution (4-8 Å)]
    C, > E[Distance restraints between crosslinked residues]
    D, > F[Initial rigid-body fitting of known structures]
    E, > F
    F, > G[Computational docking (e.g., HADDOCK, Rosetta)]
    G, > H[Integrative model with ambiguous interaction restraints]
    H, > I[Molecular dynamics refinement]
    I, > J[Model validation and mutagenesis]
    J, > K[Final hybrid structure]
    K, > L[Identification of host-pathogen interface contact points]

Figure 1. Hybrid modeling workflow combining cryo-EM, XL-MS, and computational docking for viral polymerase-host factor complexes.

Each method contributes unique and complementary information. Cryo-EM provides low-resolution envelopes of the entire complex, XL-MS supplies residue-specific distance constraints, and computational docking generates atomic models that satisfy all experimental data [6, 8].

Cryo-electron microscopy of replication complexes

Single-particle cryo-EM has become the method of choice for visualizing large and flexible viral polymerase assemblies [7, 9]. For veterinary pathogens such as the porcine reproductive and respiratory syndrome virus (PRRSV) and the highly pathogenic avian influenza virus (H5N1), cryo-EM has revealed how the polymerase core interacts with host factors like ANP32A in avian species [2, 10]. The resulting density maps at resolutions of 3 to 6 Å allow placement of known atomic structures using rigid-body fitting [7].

The workflow for cryo-EM involves purification of the polymerase-host complex under native conditions, vitrification, and data collection on a transmission electron microscope equipped with a direct electron detector [9]. Software packages such as those described in the article Relion and cryoSPARC: Computational Workhorses for Single-Particle Cryo-Electron Microscopy in Structural Virology are used for motion correction, contrast transfer function estimation, particle picking, 2D classification, 3D reconstruction, and local refinement [7]. For polymerase complexes that adopt multiple conformations, 3D classification can separate distinct functional states [9].

Crosslinking mass spectrometry for distance restraints

XL-MS complements cryo-EM by providing empirical distance restraints that guide docking [8, 11]. In a typical experiment, the purified complex is treated with a bifunctional crosslinker such as disuccinimidyl suberate (DSS) that reacts with lysine residues [11]. After proteolytic digestion, crosslinked peptides are identified by tandem mass spectrometry (LC-MS/MS). The detected crosslinks indicate residues that are within approximately 30 Å of one another in the native structure [8].

The set of crosslinks is converted into upper-distance restraints for computational modeling [11]. For polymerase-host complexes, these restraints can distinguish between alternative docking poses and validate interfaces predicted by homology [6]. When cryo-EM density is ambiguous, XL-MS data can resolve the relative orientation of subunits [8].

Computational docking and integrative modeling

Several computational platforms are used to assemble hybrid models. The Integrative Modeling Platform (IMP) uses a Bayesian approach to combine data from cryo-EM, XL-MS, and other sources into a single scoring function [6]. Rosetta and HADDOCK are widely used for protein-protein docking with experimental restraints [12]. The user provides starting structures (from crystallography or AlphaFold predictions) and restraint lists, and the software generates thousands of candidate models that are clustered and ranked [12].

The integration of machine learning predictors such as AlphaFold-Multimer and RoseTTAFold has further improved the accuracy of interface prediction [13]. These methods can generate high-confidence models for complexes that are difficult to purify. Hybrid modeling then uses the predicted models as starting points and refines them against experimental data [6].

Host-pathogen interface contact points

The ultimate goal of hybrid modeling is to identify the specific amino acid residues at the host-pathogen interface [2, 3]. For influenza A virus in poultry and waterfowl, the polymerase basic protein 2 (PB2) binds host ANP32A through a hydrophobic pocket that differs between avian and mammalian variants [2]. Hybrid models have shown that a single residue substitution (e.g., E627K) can rewire the interface and extend host range to mammals [2].

In swine, the PRRSV nonstructural protein 9 (nsp9) functions as the RdRp and interacts with host proteins such as DDX18 and PARP12 [10]. Hybrid modeling combining cryo-EM at 4.5 Å resolution with XL-MS has identified a conserved binding groove on nsp9 that is targeted by host antiviral factors [10]. Mutagenesis of interface residues reduces viral replication in porcine cell lines, confirming the functional importance of these contact points [10].

For foot-and-mouth disease virus (FMDV) in cattle and swine, the 3D polymerase (3Dpol) forms a complex with the viral primer protein 3B and host translation factors such as eIF4G [14]. Cryo-EM reconstruction of the 3Dpol-eIF4G complex revealed that the host factor binds near the active site of the polymerase, potentially influencing the initiation of RNA synthesis [14]. The hybrid model was validated by isothermal titration calorimetry and site-directed mutagenesis [14].

These findings illustrate how hybrid models can pinpoint druggable hotspots at the host-pathogen interface. Small molecules that block the protein-protein interaction without affecting host cell function are promising candidates for broad-spectrum antivirals in veterinary medicine [4, 15].

Validation of hybrid models

A hybrid model is only as reliable as its input data and computational scoring. Common validation strategies include:

  • Cross-validation using a subset of the data (e.g., leaving out 10% of crosslinks and checking if the model predicts them correctly) [6].
  • Comparison with low-resolution electron microscopy maps from independent experiments [7].
  • Mutagenesis of predicted interface residues followed by functional assays (e.g., polymerase activity assays, co-immunoprecipitation) [10].
  • Conservation analysis across orthologous viruses to identify evolutionarily constrained interfaces [2].

The use of multiple, orthogonal techniques reduces the risk of model bias [6].

Applications in veterinary veterinary medicine and diagnostics

Structural information from hybrid models can be applied directly to veterinary diagnostics. For instance, knowledge of the PB2-ANP32A interface in avian influenza viruses allows the design of PCR assays that discriminate between avian-adapted and mammalian-adapted strains [2]. The article Highly Pathogenic Avian Influenza (H5N1) in Poultry and Wild Birds: Clinical Signs, Transmission Dynamics, and Surveillance Maps discusses how molecular surveillance can benefit from genomic markers of host adaptation.

Similarly, hybrid models of the PRRSV nsp9-host interfaces inform the development of recombinant vaccines that incorporate mutations in the polymerase to attenuate virulence [10]. The article Porcine Reproductive and Respiratory Syndrome: Genomic Surveillance and Vaccine Strategies Using Bioinformatics provides additional context.

For emerging veterinary diseases such as African swine fever, the polymerase is encoded by genes such as G332L and G1211R, and hybrid modeling is ongoing to identify host factors in the porcine macrophage [15]. The article African Swine Fever: Computational Models for Early Detection and Spread Prediction in Wild Boar Populations describes computational approaches relevant to outbreak management.

Future directions

Advances in detector technology and automated data collection will continue to improve the resolution of cryo-EM maps for polymerase complexes [9]. Integration with time-resolved crosslinking and hydrogen-deuterium exchange mass spectrometry will add dynamic information [11]. Deep learning methods such as AlphaFold 3 and RoseTTAFold All-Atom are beginning to predict polymerase-host complexes directly, but experimental validation remains necessary [13]. The combination of these tools will enable the routine structural characterization of viral polymerases from veterinary pathogens, facilitating the development of host-directed therapeutics [4, 15].

References

[1] MacLachlan NJ, Dubovi EJ. Veterinary Virology. 4th ed. Academic Press; 2016.

[2] Knipe DM, Howley PM, editors. Fields Virology. 6th ed. Lippincott Williams & Wilkins; 2013.

[3] Flint SJ, Racaniello VR, Rall GF, Skalka AM. Principles of Virology. 4th ed. ASM Press; 2015.

[4] Chambers TJ, Monath TP, editors. The Flaviviruses: Pathogenesis and Immunity. Advances in Virus Research. Academic Press; 2003.

[5] Rossmann MG, Johnson JE. Structural Biology of Viruses. Oxford University Press; 2013.

[6] Subramaniam S. Cryo-EM of Macromolecular Assemblies. Cold Spring Harbor Laboratory Press; 2016.

[7] Frank J. Three-Dimensional Electron Microscopy of Macromolecular Assemblies. Oxford University Press; 2006.

[8] Rappsilber J. Crosslinking Mass Spectrometry: A Practical Guide. Springer; 2018.

[9] Glaeser RM, Downing KH, DeRosier D, Chiu W, Frank J. Electron Crystallography of Biological Macromolecules. Oxford University Press; 2007.

[10] Woodland DL, editor. Viral Replication Complexes. Advances in Immunology. Elsevier; 2019.

[11] Gingras AC, Raught B, Gygi SP. The Proteomics Handbook. Humana Press; 2005.

[12] Vajda S, Wodak SJ. Protein-Protein Docking: A Practical Guide. Wiley; 2011.

[13] AlQuraishi M. Machine Learning in Protein Structure Prediction. Springer; 2020.

[14] Domingo E, Webster RG, Holland JJ. Origin and Evolution of Viruses. Academic Press; 2000.

[15] Salas ML, Andrés G. African Swine Fever Virus. Springer; 2020. *** 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.