Computational modeling of RNA-dependent RNA polymerase conformational dynamics

Introduction

RNA-dependent RNA polymerase (RdRp) is the central enzyme for the replication and transcription of RNA viruses, including numerous pathogens of veterinary importance such as bovine viral diarrhea virus (BVDV), Japanese encephalitis virus (JEV), influenza A virus in poultry, Nipah virus, Hendra virus, and noroviruses [1, 2, 3, 4, 5, 6]. The catalytic core of RdRp adopts a conserved right-hand fold composed of fingers, palm, and thumb subdomains, with seven conserved motifs (A through G) forming the active site [7]. Conformational changes within these subdomains govern nucleotide triphosphate (NTP) entry, catalysis, translocation, and product release. Understanding these dynamics at atomic resolution is critical for rational design of antiviral compounds, especially nucleoside analogs such as remdesivir and molnupiravir, which target the polymerase active site [8, 9]. Computational modeling has become indispensable for capturing these rapid motions that are often inaccessible to experimental techniques alone [7, 10]. This review systematically examines the biophysical principles, computational methods, and key findings of RdRp conformational dynamics, anchored strictly to the peer-reviewed literature listed in the references.

Structural architecture of RdRp

The RdRp catalytic core consists of three canonical subdomains: fingers (N-terminal), palm (central), and thumb (C-terminal) [11, 7]. Motifs A–F are located within the palm and fingers subdomains. Motif A contains the aspartate residues that coordinate catalytic metal ions (usually Mg2+ or Mn2+), while motif B contributes to template–primer positioning [7]. Motif C includes the GDD (Gly-Asp-Asp) sequence, a hallmark of all viral RdRps. Motifs D and E are involved in NTP binding and active site closure, and motif F, a flexible loop in the fingers domain, interacts with incoming NTPs [7]. Allosteric communication between the thumb and fingers domains regulates active site opening and closing [7, 12]. Conformational hotspots within the SARS-CoV-2 RdRp complex, particularly in the nsp8 cofactor, exhibit unique structural malleability that influences processivity [12]. Inter-subdomain cooperation between the fingers and thumb subdomains has been characterized through folding studies, revealing that the fingers domain folds first and templates the folding of the thumb [11]. For flaviviral RdRps (e.g., dengue virus, JEV), the NS5 protein harbors both methyltransferase and polymerase domains; structural dynamics of NS5 interactions with promoter stem-loop A (SLA) have been resolved using NMR and computational models [13, 2]. Comparative analyses between human pegivirus and hepatitis C virus (HCV) NS5B polymerase have highlighted conserved motifs amenable to drug repurposing [14]. The complete architecture of RdRp across different virus families is summarized in Table 1.

Conformational dynamics governing catalysis and translocation

Polymerase elongation requires a precise sequence of conformational transitions: open state (NTP entry), closed state (catalysis), and post-translocation state. Allosteric control of these transitions involves long-range coupling between the thumb and fingers subdomains [7]. Molecular dynamics (MD) simulations have revealed the free energy landscapes for these transitions. For example, substitution of Mg2+ with Ca2+ or Zn2+ in SARS-CoV-2 RdRp significantly alters active site geometry and nucleotide binding affinity, as demonstrated by classical MD and quantum mechanics/molecular mechanics (QM/MM) calculations [8]. The conformational stability of nucleotide analogs at the active site is influenced by ribose modifications and the presence of alpha-thiotriphosphate groups, which affect the hydrogen bonding network with motif B residues [19].

Translocation of the template–product duplex by one base pair is coupled to opening of the fingers domain. Coarse-grained simulations and targeted MD have been used to map this motion. Winston and Boehr reviewed the allosteric mechanisms that regulate RdRp fidelity, emphasizing that mutations in the thumb domain can alter the conformation of the fingers and affect nucleotide discrimination [7]. The induced intra- and intermolecular template switching, a mechanism exploited by some nucleoside analogs, has been studied computationally to understand how the polymerase can stall and realign the primer terminus [10].

Computational modeling pipeline

The typical computational workflow for studying RdRp conformational dynamics is depicted in Figure 1. Starting from a high-resolution crystal structure or a cryo-EM model (often obtained from the Protein Data Bank), researchers prepare the system by adding missing loops, solvating, and neutralizing with ions. After equilibration, multiple independent MD simulations are performed, often using accelerated sampling techniques (e.g., replica exchange, metadynamics) to enhance conformational exploration [15, 18, 21]. Binding free energy calculations (MM-GBSA or MM-PBSA) are used to rank inhibitors and map drug binding sites [5, 24, 32]. Per-residue energy decomposition identifies key residues for binding, which can be used to build pharmacophore models [23]. Normal mode analysis (NMA) provides insights into large-scale collective motions, such as the opening of the fingers domain upon NTP binding [12, 7]. The entire pipeline is applicable to both veterinary and zoonotic viruses, including those for which structural models are built via homology or AlphaFold [33]. For instance, AlphaFold-driven structure-guided identification of NS5 RdRp-targeting antiviral leads has been reported for Kyasanur Forest Disease Virus, a tick-borne flavivirus affecting non-human primates and potentially livestock [33].

flowchart TD
    A[Sequence/structure retrieval], > B[Model building and refinement]
    B, > C[System preparation: solvation, ions, force field assignment]
    C, > D[Equilibration (NVT, NPT)]
    D, > E[Production MD simulations]
    E, > F{Enhanced sampling needed?}
    F, >|Yes| G[Metadynamics, replica exchange, accelerated MD]
    F, >|No| H[Conventional MD trajectory analysis]
    G, > H
    H, > I[Binding free energy calculations (MM-GBSA, MMPBSA)]
    H, > J[Principal component analysis / normal mode analysis]
    I, > K[Pharmacophore modeling / per-residue decomposition]
    J, > L[Conformational landscapes and transition pathways]
    K, > M[Drug design and optimization]
    L, > N[Mechanistic insights into translocation and allostery]
    M, > N

Nucleotide analog binding and delayed chain termination

Nucleoside analogs such as remdesivir and molnupiravir are prodrugs that, upon intracellular phosphorylation, compete with natural NTPs for incorporation by RdRp. Once incorporated, they cause delayed chain termination (remdesivir) or lethal mutagenesis (molnupiravir) [9, 19]. Computational studies have elucidated the molecular basis of these drugs. For remdesivir, the incorporated analog after three additional nucleotides blocks further elongation; MD simulations show that the steric clash with motif B redistributes the active site into a catalytically incompetent state [8, 19]. Molnupiravir (a ribonucleoside analog of cytidine) can exist in two tautomeric forms, allowing it to base pair with either A or G, leading to error catastrophe. Ahmad et al. evaluated its efficacy against emerging Omicron variants using docking and MD, confirming stable binding to the RdRp active site [9]. The role of metal ions in binding has been explored by Chen et al., who found that replacing Mg2+ with Ca2+ reduces the affinity of nucleotide analogs, suggesting an alternative mechanism of resistance [8].

Several studies have identified non-nucleoside inhibitors (NNIs) that bind to allosteric pockets, often in the thumb domain, stabilizing a closed or open conformation and preventing translocation [23, 32]. Aziz et al. employed per-residue energy decomposition to generate pharmacophore models for NNIs targeting SARS-CoV-2 RdRp [23]. Jukič et al. identified thioether-amide and guanidine-linker classes through high-throughput virtual screening and free-energy calculations [32]. For flaviviruses, allosteric inhibition of dengue virus RdRp by Litsea cubeba phytochemicals was demonstrated using MD and MM-GBSA [18]. Similarly, bioflavonoids from Azadirachta indica were shown to bind JEV RdRp at a previously uncharacterized allosteric site [2]. The BVDV RdRp mutant F224S confers resistance to the NNI VP32947 by altering the conformation of a conserved loop; He et al. used MD and binding free energy decomposition to explain the resistance mechanism [1].

Mapping polymerase elongation trajectories for 3D visualization

To animate polymerase elongation, one can interpolate between known crystallographic snapshots (e.g., open, closed, post-translocated) using morphing tools such as the CHARMM or NAMD implementations of the targeted MD (TMD) method. Alternatively, a series of frames from an MD simulation can be concatenated to produce a continuous movie. For high-quality 3D visualization, the trajectories are loaded into molecular graphics software such as VMD, PyMOL, or ChimeraX. Key residues (e.g., the GDD motif, catalytic aspartates, and drug-binding residues) should be displayed as sticks and colored by subdomain. The motion of the fingers and thumb domains can be highlighted by drawing vectors from principal component analysis. Obi et al. demonstrated the utility of combining NMR and computational ensemble refinement to map the dengue NS5 polymerase interaction with SLA RNA, providing frames for an animated pathway of RNA binding [13]. For SARS-CoV-2, Gharui et al. identified conformational hotspots in the nsp8 cofactor; these regions exhibit the largest fluctuations and can be targeted for visualization [12]. Researchers can use the morph command in PyMOL between PDB structures of apo and NTP-bound states, or generate a low-frequency normal mode animation using elastic network models.

Challenges and future directions

Despite significant progress, several challenges remain. The large size and slow timescales of RdRp conformational changes (microsecond to millisecond) require advanced sampling methods such as Gaussian accelerated MD or Markov state models [12]. Force field accuracy for non-standard residues (e.g., phosphorylated nucleotide analogs, tautomeric forms) remains a limitation; QM/MM methods are often necessary to describe the catalytic step [8, 19]. Integration of cryo-EM data with computational modeling, as exemplified by studies on dengue and SARS-CoV-2 RdRp complexes, offers a path toward more realistic models [13, 11, 12]. For veterinary virology, the development of species-specific RdRp models (e.g., for avian influenza polymerase in chickens, or for Nipah virus polymerase) is essential for evaluating host-range barriers and designing broad-spectrum inhibitors [4, 5, 3]. The use of machine learning to predict mutational effects on conformational dynamics is an emerging frontier that may accelerate drug discovery [33, 21].

References

[1] He D, Li X, Wang S et al. Mechanism of drug resistance of BVDV induced by F224S mutation in RdRp: A case study of VP32947. Comput Biol Chem. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35751995/

[2] Dwivedi VD, Singh A, El-Kafraway SA et al. Mechanistic insights into the Japanese encephalitis virus RNA dependent RNA polymerase protein inhibition by bioflavonoids from Azadirachta indica. Sci Rep. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34518560/ *** 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.

[3] Abduljalil JM, Elfiky AA, Sayed ETA et al. Computational identification of drug-like marine natural products as potential RNA polymerase inhibitors against Nipah virus. Comput Biol Chem. 2023. URL: https://pubmed.ncbi.nlm.nih.gov/36907056/

[4] Lalu AC, Kundil VT, Joseph BB et al. Repurposing FDA-Approved Drugs as Hendra Virus RNA-Dependent RNA Polymerase Inhibitors: A Comprehensive Computational Drug Discovery Approach. Viruses. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/41472282/

[5] Samir A, Elshemey WM, Elfiky AA. Molecular dynamics simulations and MM-GBSA reveal a novel small molecule against Flu A RNA-dependent RNA polymerase. J Biomol Struct Dyn. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/38088748/

[6] Ebenezer O, Damoyi N, Jordaan MA et al. Unveiling of Pyrimidindinones as Potential Anti-Norovirus Agents-A Pharmacoinformatic-Based Approach. Molecules. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35056692/

[7] Winston DS, Boehr DD. Allosteric and dynamic control of RNA-dependent RNA polymerase function and fidelity. Enzymes. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34696831/

[8] Chen SH, Ho Y, Lin SJ et al. The investigation of the inhibition effects of substituting Mg(2+) with Ca(2+) and Zn(2+) on SARS-CoV-2 RNA-dependent RNA polymerase. J Biomol Struct Dyn. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41889305/

[9] Ahmad F, Basharat Z, Janjua A et al. An exploratory binding study of molnupiravir efficacy against emerging Omicron SARS-CoV-2 variants. Sci Rep. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/41120355/

[10] Janissen R, Woodman A, Shengjuler D et al. Induced intra- and intermolecular template switching as a therapeutic mechanism against RNA viruses. Mol Cell. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34687604/

[11] Sinha A, Tony AMC, Roy S. How fingers affect folding of a thumb: Inter-subdomain cooperation in the folding of SARS-CoV-2 RdRp protein. Biophys Chem. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/39490134/

[12] Gharui S, Sengupta D, Das A. Characterization of the Conformational Hotspots of the RNA-Dependent RNA Polymerase Complex Identifies a Unique Structural Malleability of nsp8. J Phys Chem B. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/39356135/

[13] Obi JO, Kihn KC, McQueen L et al. Structural dynamics of the dengue virus non-structural 5 (NS5) interactions with promoter stem-loop A (SLA). Npj Viruses. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40295851/

[14] Copenhaver KM, Hanson BA, Ziarek JJ et al. In Silico Characterization of Two Human Pegivirus Proteins Highlights Similarities with Hepatitis C Virus and Possible Therapeutic Repurposing. Viruses. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41754604/

[15] Alsaady IM, Gattan HS, Aljahdali SM et al. Conformational dynamics and binding free energy analyses unveil a stable flavonoid inhibitor of dengue virus NS5 polymerase. Sci Rep. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/41654575/

[16] Dhanasekaran S, Selvadoss PP, Manoharan SS et al. Regulation of NS5B Polymerase Activity of Hepatitis C Virus by Target Specific Phytotherapeutics: An In-Silico Molecular Dynamics Approach. Cell Biochem Biophys. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/39042185/

[17] Alobaida A, Abouzied AS, Younes KM et al. Analyzing energetics and dynamics of hepatitis C virus polymerase interactions with marine bacterial compounds: a computational study. Mol Divers. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/38856834/

[18] Panday H, Jha SK, Al-Shehri M et al. Allosteric inhibition of dengue virus RNA-dependent RNA polymerase by Litsea cubeba phytochemicals: a computational study. J Biomol Struct Dyn. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/38764132/

[19] Li Y, Yuan S. The Role of Ribose Modifications on the Structural Stability of Nucleotide Analogues with α-Thiotriphosphate at the Active Site of SARS-CoV-2 RdRp. J Chem Inf Model. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40476830/

[20] Zhao H, Chang X, Sun L et al. Identification of RdRp-NiRAN/JAK1 Dual-Target Drugs for COVID-19 Treatment. J Phys Chem B. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/39625254/

[21] Chinnamadhu A, Ramakrishnan J, Suresh S et al. Binding properties of selective inhibitors of P323L mutated RdRp of SARS-CoV-2: a combined molecular screening, docking and dynamics simulation study. J Biomol Struct Dyn. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/37301607/

[22] Bagabir SA. Investigating the potential of natural compounds as novel inhibitors of SARS-CoV-2 RdRP using computational approaches. Biotechnol Genet Eng Rev. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/36994810/

[23] Aziz S, Waqas M, Mohanta TK et al. Identifying non-nucleoside inhibitors of RNA-dependent RNA-polymerase of SARS-CoV-2 through per-residue energy decomposition-based pharmacophore modeling, molecular docking, and molecular dynamics simulation. J Infect Public Health. 2023. URL: https://pubmed.ncbi.nlm.nih.gov/36801630/

[24] M A Kawsar S, Hosen MA, Ahmad S et al. Potential SARS-CoV-2 RdRp inhibitors of cytidine derivatives: Molecular docking, molecular dynamic simulations, ADMET, and POM analyses for the identification of pharmacophore sites. PLoS One. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/36441684/

[25] Alamri MA, Tahir Ul Qamar M, Alabbas AB et al. Molecular and Structural Analysis of Specific Mutations from Saudi Isolates of SARS-CoV-2 RNA-Dependent RNA Polymerase and their Implications on Protein Structure and Drug-Protein Binding. Molecules. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/36235011/

[26] Prajapati KS, Singh AK, Kushwaha PP et al. Withaniasomnifera phytochemicals possess SARS-CoV-2 RdRp and human TMPRSS2 protein binding potential. Vegetos. 2023. URL: https://pubmed.ncbi.nlm.nih.gov/35729946/

[27] Mallick Gupta A, Mandal S, Mandal S et al. Immune escape facilitation by mutations of epitope residues in RdRp of SARS-CoV-2. J Biomol Struct Dyn. 2023. URL: https://pubmed.ncbi.nlm.nih.gov/35293850/

[28] Karthic A, Kesarwani V, Singh RK et al. Computational Analysis Reveals Monomethylated Triazolopyrimidine as a Novel Inhibitor of SARS-CoV-2 RNA-Dependent RNA Polymerase (RdRp). Molecules. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/35164069/

[29] Dhankhar P, Dalal V, Kumar V. Screening of Severe Acute Respiratory Syndrome Coronavirus 2 RNA-Dependent RNA Polymerase Inhibitors Using Computational Approach. J Comput Biol. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34847746/

[30] Skariyachan S, Gopal D, Deshpande D et al. Carbon fullerene and nanotube are probable binders to multiple targets of SARS-CoV-2: Insights from computational modeling and molecular dynamic simulation studies. Infect Genet Evol. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34823028/

[31] Sharma S, Sharma A, Bhattacharyya D et al. Computational identification of potential inhibitory compounds in Indian medicinal and aromatic plant species against major pathogenicity determinants of SARS-CoV-2. J Biomol Struct Dyn. 2022. URL: https://pubmed.ncbi.nlm.nih.gov/34766880/

[32] Jukič M, Janežič D, Bren U. Potential Novel Thioether-Amide or Guanidine-Linker Class of SARS-CoV-2 Virus RNA-Dependent RNA Polymerase Inhibitors Identified by High-Throughput Virtual Screening Coupled to Free-Energy Calculations. Int J Mol Sci. 2021. URL: https://pubmed.ncbi.nlm.nih.gov/34681802/

[33] Ajith Kumar K, Rajesh N, Dcunha L et al. AlphaFold-driven structure-guided identification of NS5 RdRp-targeting antiviral leads against Kyasanur Forest Disease Virus. J Mol Graph Model. 2026. URL: https://pubmed.ncbi.nlm.nih.gov/42142462/

[34] Pant S, Jena NR. Computational predictions of artificial nucleoside triphosphates as potent inhibitors of RNA-dependent RNA polymerase of the ZIKA virus. Hum Immunol. 2025. URL: https://pubmed.ncbi.nlm.nih.gov/40117673/

[35] Ezzemani W, Altawalah H, Windisch M et al. Identification of Zika virus NS2B-NS3 protease and NS5 polymerase inhibitors by structure-based virtual screening of FDA-approved drugs. J Biomol Struct Dyn. 2024. URL: https://pubmed.ncbi.nlm.nih.gov/37528667/