An exploratory binding study of molnupiravir efficacy against emerging Omicron SARS-CoV-2 variants

Abstract

SARS-CoV-2 (severe acute respiratory syndrome causing coronavirus 2) caused an epidemic that swept the globe and resulted in large number of casualties. It is still sporadically causing cases and has a long-term impact on the health of once infected individuals. Molnupiravir binds RNA dependent RNA polymerase (RdRp) of SARS-CoV-2 as well as spike protein. In this study, we assessed the mutated spike protein of BA.5 variant and BQ.1.1 subvariant of COVID-19 and tested their binding with it. Multiple sequence and structural alignment of homologous structures revealed highly conserved amino acid residues at the active site of the domain. The molecular docking of Molnupiravir with the active site of the domain, comprised conserved motifs (motif A-G), and exhibited considerable binding affinity against variant and subvariant protein targets. Molnupiravir exhibited stability in its interactions with the Omicron and BQ.1.1 spike proteins, preserving constant engagement within the active site. The protein and Ligand reached An equilibrium with An RMSD of 10.46 Å after 100 nanoseconds, whereas the Ligand measured 8.0 Å. Fluctuations were noted between 40 And 75 nanoseconds, stabilizing from 80 to 100 ns. In simulations including the BQ.1.1 subvariant, the RMSD values demonstrated considerable stability, exhibiting Little variations. The ligand demonstrated flexibility, altering its binding orientation over time, resulting in An average RMSD of 18.72 Å. Herein, investigation of molecular dynamics trajectories elucidated the conformational stability of Molnupiravir, emphasizing its interactions with active residues and the hydrogen bond acceptor and donor environments. The results highlighted the crucial function of protein loops in offering flexibility and enabling ligand binding within the active site. It is concluded that Molnupiravir has the potential to function as an inhibitor of both omicron and its subvariant BQ.1.1.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-19353-4.

Introduction

The COVID-19 pandemic produced by the SARS-CoV-2 virus has waned, but it is becoming endemic, And reports of outbreaks still exist. According to the World Health Organization, there have been over 769 million confirmed cases of COVID-19 globally as of August 6, 2023, with over 6.9 million fatalities. In parallel to this, the vaccination campaigns continue, with over 12 billion vaccine doses delivered¹. Some countries have begun to relax their public health guidelines, particularly when it comes to wearing masks outside². This is because the number of COVID-19 patients has decreased. Here, we focus on Omicron variant, BA.5 and its sub-variant. These variants were behind the recent increase in COVID-19 cases in South Africa, the United States, and Europe³. The enormous number of mutations (> 50) in these variants, to escape antibodies may jeopardize the current immunization effort⁴.

Around 30 of these are spike protein mutations. The spike protein of SARS-CoV-2 is a trimeric, glycosylated class I fusion protein that protrudes from the viral envelope, forming the distinctive “corona” morphology and consists of two main subunits: S1 and S2. S1 is responsible for receptor recognition via its receptor-binding domain (RBD) and S2 mediates membrane fusion with host cells⁵. Proteolytic cleavage and conformational shifts in the spike protein enable membrane fusion, receptor binding, and immune evasion. Due to its surface exposure and immunogenicity, the spike protein is the primary target of neutralizing antibodies, antiviral drugs, and current vaccines⁶. Variants such as Omicron sub-lineages (e.g., BA.5, BQ.1.1) carry approximately 30 mutations in the spike protein, which enhance immune escape, modify receptor affinity, and alter cell entry pathways⁷. These changes have led to shifts in viral tropism and pose challenges for therapeutic and vaccine development.

The strong transmission potential of Omicron is owing in part to the variant’s extremely high affinity for Angiotensin converting enzyme 2 (ACE2) in comparison to the wildtype (WuhanHu1) strain⁸. Other variants like BA.2.75, BA.4.6, XBB, and BQ.1 have also been reported⁹. The European Centre for Disease Prevention and Control (ECDC) has identified the novel Omicron sub-variant BQ.1.1 as a variant of interest¹⁰. BQ.1.1 sub-variant remained the most common one for some time in the US And most European nations, accounting for 27.9% of all SARS-CoV-2 cases in the US. It was discovered for the first time in early September 2022 And has since been recognized in 73 nations¹¹. The Omicron variant (BA.5 lineage) and its subvariant BQ.1.1 have a high transmissibility and potential immune evasion. BA.5 and BQ.1.1 differ in their spike protein mutations, immune escape capabilities, and epidemiological impact. BA.5, first identified in early 2022, harbors spike mutations that enhance transmissibility and immune evasion¹². BQ.1.1, a sub-lineage of BA.5 that emerged later in 2022, retains core mutations of BA.5 but includes additional changes such as K444T, R346T, and N460K, that further enhance immune evasion¹³. Both subvariants are highly transmissible, with BQ.1.1 showing a growth advantage and rapidly becoming dominant in multiple regions. Laboratory studies indicate that BQ.1.1 escapes neutralizing antibodies to a greater extent than BA.5, significantly reducing the efficacy of monoclonal antibody therapies and increasing breakthrough infections, even in vaccinated individuals¹⁴. Clinically, both cause similar symptoms, typically milder than earlier variants, though BQ.1.1 may exhibit slightly more pathogenic traits in animal models. Updated bivalent vaccines targeting BA.4/BA.5 offer partial protection, but BQ.1.1 continues to pose challenges to current therapeutic and immunization strategies¹⁵.

Docking gives a detailed analysis of how the ligand interacts with the protein, which can optimize the lead compounds for drug development¹⁶. For this purpose, molecular docking has become a key tool for drug discovery and molecular modelling applications. The result gives a score of the interaction, making it more reliable for predicting the ligand pose and, through that pose binding site of the ligand can easily be determined¹⁷. This is followed by molecular dynamic (MD) simulations studies which infer the best mode of action by identifying best binding poses with reference to time intervals¹⁸. Molnupiravir (prodrug of the ribonucleoside analog β-D-N4-hydroxycytidine (NHC)) is a new oral antiviral medication and has been licensed for the treatment of COVID-19 in several countries¹⁹. Owing to its tautomeric flexibility, NHC induces frequent mispairing during viral replication, leading to G-to-A and C-to-U transitions and ultimately causing lethal mutagenesis. Unlike chain-terminating antivirals, Molnupiravir permits RNA synthesis to continue but yields nonviable, hypermutated viral genomes. Clinically, this mechanism has been shown to reduce viral load and limit disease progression, with efficacy demonstrated in both in vitro studies and human trials, supporting its therapeutic use in mild to moderate COVID-19²⁰. This study was designed to validate the drug binding of Molnupiravir, to inhibit the mechanistic behavior of the target spike protein involved in binding, internalization and hence, pathogenicity of COVID-19 omicron and sub-variant BQ.1.1.

Materials and methods

The current study has been described in the flow chart as shown in Fig. 1.

Fig. 1.

Methodology for the study undertaken is represented as a flow chart.

Sequence retrieval and mutation

Spike protein sequences of SARS-CoV-2 Omicron subvariants BA.5 and BQ.1.1 were retrieved from the ViralZone database (https://viralzone.expasy.org/9556). The BA.5 spike protein harboured the following mutations vs. Wuhan HU-1 strain: T19I, LPPA24–27 S, Δ69–70, G142D, V213G, G339D, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, L452R, S477N, T478K, E484A, F486V, Q498R, N501Y, Y505H, D614G, H655Y, N679K, P681H, N764K, D796Y, Q954H, and N969K (https://www.bv-brc.org/view/VariantLineage/#view_tab=lineage&loc=BA.5). The BQ.1.1 spike protein contains the following mutations vs. Wuhan HU-1 strain: T19I, LPPA24–27 S, H69del, V70del, V213G, G142D, G339D, R346T, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, K444T, L452R, N460K, S477N, T478K, E484A, F486V, Q498R, N501Y, Y505H, D614G, H655Y, N679K, P681H, N764K, D796Y, Q954H, and N969K (https://www.bv-brc.org/view/VariantLineage/#view_tab=lineage&loc=BQ.1.1). It shared many of these mutations with BA.5 but also included additional substitutions: R346T, K444T, and N460K (Supplementary Fig. 1). These full-length spike sequences were subjected to BLAST analysis against the Protein Data Bank (PDB), and structures with ≥ 99% sequence similarity were identified. The crystal structure 7XNQ_A was selected for both BA.5 and BQ.1.1 due to its high similarity. Identified mutations were manually introduced into the structure using PyMOL software, and the modified spike proteins were saved in PDB format for downstream analysis²¹.

Protein Preparation and physicochemical properties

The Linear chain that comprised of 1−1268 amino acids (A chain) was taken and prepared. Extra constituents present as side chain C and the Nitrogen and water components were removed through Pymol²². The physicochemical properties of the protein which are molecular weight, isoelectric point, amount of negative and positive residues, extinction coefficient, instability index, aliphatic index and GRAVY were determined using ProtParam, which is a tool of ExPAsy²³.

In Silico binding

Docking is a common practice to streamline the virtual screening of Ligands that can act as a drug. The 3D structure of omicron variant BA.5 and its subvariant BQ.1.1 spike proteins were energy-minimized to improve structural refinement prior to molecular docking with the selected inhibitor (Molnupiravir) to find the best orientation in the active pocket of protein. The scoring functions of docked complex depicted the best pose of protein and ligand complex. For docking purpose, PyRx was used that implements genetic algorithm. Ligands were imported into PyRx and subjected to energy minimization using the Open Babel MMFF94 force field to optimize their geometry. Following minimization, the Ligand structures were converted into PDBQT format for compatibility with the docking engine. The docking grid parameters were configured using the Vina Wizard within PyRx, with a grid box size typically set to 25× 25 × 25 Å in the X, Y, and Z dimensions to ensure adequate space for full Ligand flexibility within the binding site. Protein was also imported And converted to pdbqt format. AutoDock Vina was employed as the docking engine, using a default exhaustiveness value of 8 to balance accuracy and computational efficiency. Binding affinities were calculated and expressed in kcal/mol, and the pose with the lowest binding energy was selected for subsequent visualization and interaction analysis. Complexes were visualized through UCSF (University of California at San Francisco) Chimera²⁴ and Discovery Studio visualizer²⁵.

MD simulations

To support the intrinsic atomic interactions and binding conformations of selected antiviral drug in dynamic nature, the Schrödinger LLC platform Desmond was applied to run MD simulations for 100 nanoseconds (ns). The system was preprocessed through wizard module by optimizing the protein structure. The optimization completed through energy minimization for which optimized potentials for the liquid simulations (OPLS)−2005 force field was used and further refinement was justified by the addition of hydrogen atoms²⁶. Energy minimization was performed to achieve Fmax < 1000 kJ/mol/nm using the steepest descent approach for 5000 iterations to create stable conformations. Both systems were equilibrated using canonical ensembles (NVT) and isobaric-isothermal ensembles (NPT) at a constant temperature of 310 K And a constant pressure of 1 atm for 300 ps. The molecular dynamics simulation was configured to operate at a temperature of 310 K for a duration of 100 ns^27,28. The system was solvated through TIP3P water model²⁹. 10 Å buffer region was created by orthorhombic water box while temperature was kept constant at 300 K using NPT ensemble class³⁰. The generated trajectories were analyzed for their stability and the dynamic behavior of complex within the system.

Ligand properties analysis

Hydrogen bonding

Hydrogen Bonding was mapped as it displays a directionality and interaction specificity between a protein and its ligand (protein, effector, inhibitor, nucleic acid or substrate) that is a key aspect of molecular recognition³¹. Therefore, the energy and kinetics of hydrogen bonds must be optimal to allow rapid sampling and folding kinetics. This confers stability protein structure and specificity required for selective macro-molecular interactions³¹.

Solvent accessible surface area (SASA)

The extent of solvent-accessible surface area (SASA) plays a pivotal role in governing the configuration and functions of biological macromolecules. Typically, the amino acid residues situated on a protein’s surface act as active sites and/or engage in interactions with various molecules and ligands. This enables to gain deeper insights into the molecule’s behavior in a solvent environment, distinguishing between its hydrophilic or hydrophobic characteristics and elucidating the elements involved in protein-ligand interactions³². SASA was also determined.

Molecular surface area (MolSA)

Ligand properties were assessed using the Molecular Surface Area (MolSA), which employs a calculation method identical to that of the van der Waals surface area, utilizing a probe radius of 1.4³².

MMGBSA analysis

The MM-GBSA (Molecular Mechanics, Generalized Born model, and Solvent Accessibility) study was conducted to examine the free binding energies of the protein-ligand complexes. The primary module of Schrödinger program was used to compute the optimum binding energy of the chosen complexes with the lowest docking score. The study used the VSGB 2.0 model, including the OPLS-AA force field along with an implicit solvent model and physics-based modifications for π-π interactions, hydrophobic interactions, and hydrogen bonding self-contact interactions³³.

Results

Identification of functional domains of protein

Functional domains are the active part of the protein which are used for interaction with other proteins And substances. The InterPro job ID for finding the functional domains of 7XNQ_A was retrieved via https://www.ebi.ac.uk/interpro/result/InterProScan/iprscan5-R20240105-021238-0086-98890539-p1m/. The target protein was checked for family, domains And homologous superfamily. The target protein showed a high conservation with reference protein. Total protein residues were composed of 1268 bp which was homologous to Beta corona spike protein. 16 domain sites were observed in the target protein And 8 in homologous superfamily, investigated with other unregulated and site regions as shown in Fig. 2.

Fig. 2.

Spike protein depicting the functional domain, family, and homologous superfamily of the target variants with other variants of COVID-19.

Molecular docking

PyRx was used to check the binding affinities of the inhibitor And protein. The docking sites automatically gave calculation of the size, center, And sites of bonding in 5 different poses of interaction. The binding affinity calculated for the BA5 docked complex is −5.70 and BQ.1.1 was − 5.30 kcal/mol. The best pose with the minimum score in Kcal/mol was taken And 2D interaction was observed in the Discovery Studio (hydrophobic and hydrogen bond interactions among the protein-ligand complex system) (Fig. 3A and B) and Chimera. These analyses revealed that Molnupiravir formed stable and specific hydrogen bonds with key spike protein residues in both variants. For the BA.5 variant (Fig. 3A), strong hydrogen bonds were observed between the ligand and Gly1042 and Gly1044, along with additional interactions involving van der Waals contacts, carbon hydrogen bonds, and π-alkyl interactions with residues such as Tyr1045 and Val1038. In the BQ.1.1 variant (Fig. 3B), the ligand engaged in multiple conventional hydrogen bonds with Tyr394, Arg353, Ser512, Thr428, and Asp426, interacting through both its oxygen and nitrogen atoms. These interactions suggest a robust and specific binding mode, indicating potential inhibitory effects against both spike protein variants. However, the BQ.1.1 complex exhibited a higher interaction density, indicating a stronger overall binding affinity. This observation aligns well with the docking results, where BQ.1.1 showed a more favorable docking score.

Fig. 3.

Interacting residues of the docked complexes. (A) BA5-Molnupiravir against the omicron spike protein (B) BQ.1.1-Molnupiravir against the BQ.1.1 spike protein.

MD simulations

Stability analysis

The MD simulations provide data on the RMSD and RMSF of Molnupiravir bound with studied Spike protein variants, as well as the Rg and ligand-protein interactions. In the BA.5 complex, Molnupiravir remained stably bound within the active site throughout the 100 ns simulation. The system reached equilibrium with a protein RMSD of 10.46 Å and a Ligand RMSD of approximately 8.0 Å (Fig. 4A). Transient fluctuations were observed between 40 And 75 ns, with the system stabilizing from 80 ns onward. The highest average RMSD (~ 15.84 Å) was recorded in residues 465–504 during the 45–60 ns window, likely reflecting localized flexibility near the C-terminal region where the ligand repositioned within the pocket. Despite these internal adjustments, Molnupiravir maintained continuous interactions with adjacent residues, remaining tightly bound until the end of the simulation.

Fig. 4.

(A) RMSD plot of the Cα atoms of the Omicron BA.5 spike protein in complex with Molnupiravir over a 100 ns simulation. The trajectory shows noticeable fluctuations around 37 ns, followed by stabilization after approximately 62 ns, indicating the system reached equilibrium and maintained structural stability thereafter. (B) RMSD plot of both the BQ.1.1 spike protein And bound Molnupiravir across the 100 ns simulation. The protein backbone remains stable after ~ 25 ns, but the ligand exhibits a notably higher RMSD (~ 20 Å), suggesting possible rearrangement or a shift in binding mode during the simulation.

In a parallel simulation with the BQ.1.1 variant, the receptor exhibited An average RMSD of 13.0 Å (maximum 15.69 Å), indicating global stability. However, the Ligand demonstrated greater mobility, with An average RMSD of 18.72 Å (Fig. 4B). This higher deviation suggests that Molnupiravir underwent a conformational shift, transitioning from its initial binding orientation toward the β-core sheet region near the N-terminal site between 25 And 57 ns. This movement Likely represents An adaptive search for a more energetically favorable conformation, and despite its flexibility, the Ligand remained associated with the receptor throughout the trajectory, reflecting a dynamically stable interaction. This interpretation is supported by the interaction density observed in the 2D diagram (Fig. 3B), where multiple residues contribute to anchoring the ligand, despite its higher mobility.

MD simulation trajectories were further analyzed to gain insights into the conformation stability of Molnupiravir within the spike protein binding pockets of the Omicron variants (Fig. 5A-B). The surface representation of the BA.5 spike protein (Fig. 5A) highlights the ligand-binding region located near the C-terminal domain. Notably, residues such as ASP1039, TYR1045, and SER1035 contribute to the firm encapsulation of Molnupiravir through a combination of hydrogen bonds and hydrophobic contacts (Fig. 5B).

Fig. 5.

The simulated complexes and their binding orientation inside the active pocket. (A) BA5 variant of COVID-19 with inhibitor binds to its C-terminal site. (B) Close up pose of inhibitor at the end of the simulation time interval with its binding residues and hydrogen bond acceptor and donor atoms. (C) COVID-19 BQ.1.1 with inhibitor binding to its N-terminal (D) Hotspot residues with ligand (final orientation) at the end of simulation time interval.

A comparative structural snapshot of the BQ.1.1 spike protein complex is shown in Fig. 5C, where the binding pocket is located nearer the N-terminal region. Residues ASP426 and ARG353 play key roles in stabilizing the ligand through hydrogen bonding and van der Waals interactions. Figure 5D further emphasizes the dynamic aspects of ligand binding, showing how side chain flexibility in residues such as HIS513 and GLU427 may influence the local binding environment. These conformational changes reflect the adaptive nature of the binding pocket in BQ.1.1 and may account for the higher ligand RMSD observed in the MD analysis.

Subsequently, the RMSF was calculated to evaluate the flexibility of the protein-ligand complexes over the course of the simulations (Fig. 6). RMSF specifically measured the dynamic flexibility of Cα atoms across the protein structure. For the BA.5 complex, the average RMSF was 1.7 Å, with fluctuations ranging from a minimum of 0.7 Å to a maximum of 5.2 Å.

Fig. 6.

(A) Graphical representation of RMSF of BA.5 complex, (B) RMSF of Molnupiravir inside the pocket of BA.5. (C) RMSF graphical fluctuations of BQ.1.1 complex (D) RMSF of Molnupiravir inside the pocket of BQ.1.1.

In the BQ.1.1 complex, simulated over 50 ns, the average RMSF was 1.61 Å, with the highest peak reaching 4.2 Å. Pronounced fluctuations were observed at residues 76, 103, And 203–279, as well as at the terminal regions 518 And 599, which primarily correspond to loop regions. The RMSF profile (Fig. 6) also highlighted the localized movements, reflecting the natural flexibility within non-secondary structure elements during the simulation.

The flexibility of ligand atoms of all the complexes was also observed. Protein targets during simulation time intervals show high flexibility and maximum level of fluctuation due to their loop regions. The COVID-19 variant with BA5 and BQ.1.1 protein target had mean RMSF values of 5.38 Å, And 4.34 Å, respectively (Fig. 6A-C). A high degree of concurrence regarding intermolecular stability was suggested by these values. The ligands exhibited a significant propensity for fluctuations, which was observed at a rapid rate and continued thereafter (Fig. 6B-D). The core, N-like domain, catalytic binding domain, and C-terminal domain were found to contain a significant proportion of flexible loops, which compelled these regions to exhibit more dynamic behavior. This behavior of the protein might be an inherent mechanism designed to provide flexibility to facilitate the substrate/ligand molecule’s correct accommodation within the pocket and enable the catalytic mechanism to be executed.

Interactions between proteins and ligands

The helical motion of certain flexible bonds in Molnupiravir contributed to an increase in its RMSD during the simulation. Despite this conformational flexibility, Molnupiravir consistently maintained key molecular interactions within the active site, as confirmed by interaction analysis. The protein-ligand contact histogram is presented in Fig. 7, illustrating the persistence and types of interactions formed during the trajectory. In the BA.5 complex (Fig. 7A), Molnupiravir established both conventional hydrogen bonds and water-bridged interactions within the simulation environment. While hydrophobic contacts were also present, hydrogen bonding was the predominant interaction type observed throughout the simulation. Key interacting residues included Arg1037, Asp1039, Cys1041, Gly1042, Lys1043, Gly1044, and Gln1069, which formed stable hydrogen bonds, often reinforced by bridging water molecules. In contrast, the BQ.1.1 complex (Fig. 7B) featured strong hydrogen bond interactions involving residues such as Gly379, Asp426, Thr428, Ser512, Leu515, Thr545, Arg565, and Ala568. These were further supported by hydrophobic interactions and water-mediated contacts, indicating a robust and adaptive binding profile for Molnupiravir in both variants.

Fig. 7.

(A) Histogram plot shows binding linkages among Molnupiravir and BA5. (B) Linkages of Molnupiravir with subvariant BQ.1.1.

Molnupiravir maintained continuous interactions with active site residues throughout the entire simulation period, as illustrated in Fig. 8. These interactions primarily involved hydrogen bonds, hydrophobic contacts, and water bridges, contributing to the overall stability of the complexes. In BA.5 interaction, key residues such as Arg1037, Asp1039, Cys1041, Gly1042, and Lys1043 played prominent roles in forming stable hydrogen bonds with the ligand, indicating a high interaction intensity throughout the trajectory (Fig. 8A).

Fig. 8.

The graphs (orange lines) show the contacts of ligand with different residues in each trajectory frame while blue lines show the total number of contacts that ligand form throughout the simulation period. (A) Graphical presentation of ligand interaction via hydrogen bonds against BA5 (B) Presenting the hydrogen bonds interactions with BQ.1.1 with 100 ns simulations time intervals.

In contrast, BQ.1.1 complex exhibited a temporary disruption in ligand interactions as Molnupiravir relocated deeper into an alternative binding cavity. Initially, it formed hydrogen and hydrophobic interactions with Gly379, Asp426, Thr428, Ser512, and Leu515. Upon repositioning within the cavity, the ligand established new interactions with Thr545, Arg565, and Ala568, primarily via hydrophobic contacts and water bridges. Additional stabilization was observed through salt bridge and hydrogen bond formation with Asp36 and Tyr34, indicating a dynamic yet sustained engagement with the protein. The flexible movement of Molnupiravir within the binding pocket allowed it to interact with multiple residues over time, reflecting an adaptive binding mode (Fig. 8B).

Ligand stability

Ligands’ Root Mean Square Deviation (RMSD), radius of gyration (rGyr), molecular surface area (MolSA), solvent-accessible surface area (SASA), and polar surface area (PSA) were also assessed. Figure 9 depicted the computed ligand RMSD over simulation intervals. Our investigation yielded the following fascinating findings: Molnupiravir with BA5, with fewer rotatable bonds, demonstrated significant stiffness. It did, however, exhibit periodic instability at certain intervals, with the Ligand attached to the active domain displaying mobility. It had maximum RMSD values of 1.5 Å And a mean value of 1.2 Å. Molnupiravir with BQ.1.1, on the other hand, had RMSD values that were typically more than 1.8 Å (Fig. 9A), but remained steady throughout the simulation (Fig. 9B). Despite its multiple rotatable bonds, it exhibited oscillating movement over the active domain surface (Fig. 9B). Rg values, which were inversely linked to protein compactness, were critical in determining ligand stability. Stable protein folding is characterized by little variation in Rg values, indicating static ligand behavior within the cavity. The Rg values of Molnupiravir in BA5 pocket varied from 4.25 Å to 3.5 Å from 0 to 100 ns Fig. 9A, whereas, for BQ.1.1, Rg of Molnupiravir was between 4 Å And 3.5 Å from 0 to 100 ns (Fig. 9B). This Rg depicted more stability as compared to BA-5, which resulted in a better inhibitory response towards the new variant of COVID-19 (Fig. 9B). The intramolecular hydrogen bonds in the BA5-Molnupiravir complex were absent but observed in BQ.1.1 subvariant and Molnupiravir complex.

Fig. 9.

The properties of the ligand like root mean square deviation (RMSD), radius of Gyration (rGyr), Molecular surface area (MolSA), solvent accessible surface area (SASA) and Polar Surface Area (PSA) calculated over the trajectory for (A) BA.5 and (B) BQ.1.1.

The polar surface area (PSA) quantified the surface area of the solvent generated by polar groups during the simulation, such as nitrogen and oxygen atoms of the Ligand. It was observed that the PSA fluctuated between 240 And 270 Ų for BA5 and BQ.1.1(Fig. 9A). Values considerably greater than 270 Ų were observed for complex 2 (Fig. 9B). This was not only due to the ligand’s increased dimensions but also to the substantial phosphate group and numerous basic nitrogen atoms that were present within it. The SASA, which provided a quantitative calculation of ligands containing protein-implicit water molecules, stood for solvent accessibility. Significant variations existed among the ligands under investigation with respect to this parameter. SASA for BA5 complex fluctuated between 100 And 300 Ųwhereas SASA for BQ.1.1 complex, despite having same molecular masses, was found to be nearly two times larger. This was the result of considerably greater flexibility and the molecules’ less compact configuration. In conclusion, it could be observed that BQ.1.1 complex exhibited a considerably greater degree of flexibility, and a less compact shape compared to BA5. Nevertheless, a decline in flexibility and an increase in compactness were also detected during the simulation of BA5 complex (Fig. 8A). The 100% increase in SASA observed in BQ.1.1 during the simulation could be attributed to the ligand’s displacement from the binding site to the protein surface after the dissociation of the complex. This procedure was initiated following a 20 ns simulation and then moved deep into the pocket at the end of the time interval which becomes stable (Fig. 9B).

The MolSA quantified the area of the solvent in water that Ligand molecules absorbed. The surface area was determined in the MolSA study through the utilization of a probe radius of 1.4 Ų. This radius was roughly equivalent to both the surface area And radius of a single water molecule. The MolSA oscillated at approximately 280, 288, 296, And 304 Ųcorrespondingly, for BA5 complex (Fig. 8A) and for BQ.1.1 complex, it oscillated among 270, 285 And 300 Ų (Fig. 9B). The reduction in MolSA for BQ.1.1 complex could be attributed directly to the ligand’s conformational change, as evidenced by the decline in radius of gyration.

Ligand torsion analysis inside the active pocket

As illustrated in Fig. 10, the ligand torsion plot presented a concise overview of the conformational alterations witnessed in every rotatable bond (RB) of the Ligand throughout the complete simulation trajectory spanning from 0 to 100ns. The upper section illustrated a two-dimensional representation of the ligand, with distinct colours used to emphasize the RB (torsion) regions. Figure 10A and B present a comparative depiction of the spatial and angular distributions of the BA5 variation and the BQ.1.1 subvariant, respectively. Herein, Fig. 10A presents a scatter plot and polar plot that depicts the geographical distribution of data points for the BA5 variation, while the histograms quantify the angular distribution, emphasizing characteristics that may affect its dissemination or behavior. Likewise, Fig. 10B emphasizes the BQ.1.1 subvariant, illustrating its geographical distribution via scatter and polar plots, and its angular distribution through histograms. These visualizations are essential for comprehending the behavioral variations and dissemination patterns between the two genotypes, providing significant insights into their epidemiological effects. In the same colour, a dial plot and corresponding bar plots accompany each RB torsion. The dial diagrams illustrate the conformational changes that occurred in the torsions throughout the simulation. The initiation of the simulation was denoted at the center of the radial depiction, whereas the progression of time is represented in the radial orientation. In contrast to the dial plots, the bar plots presented the probability density of torsion at various time intervals in a concise manner. Moreover, if data regarding the torsional potential was accessible, the potential energy of the RB was incorporated into the plot through the summation of the potential energies of the pertinent torsions. The values of the potential energy, denoted in kcal/mol, were illustrated along the Y-axis on the left side of the chart. An examination of the histogram and the correlation between torsional potential and torsion values might provide valuable information regarding the strain that the ligand underwent while attempting to preserve its conformational bond with the protein.

Fig. 10.

The ligand torsion profile (A) Presenting the torsion of the inhibitor inside the variant BA5. (B) Shows torsion and flexibility of the compound inside the active cavity of BQ.1.1 subvariant as density of torsion increases.

Binding free energies calculations

The MM/GBSA binding energy data and the residue-level interaction profiles shown in the figure, Molnupiravir demonstrates significantly stronger and more stable binding to the BQ.1.1 subvariant of the SARS-CoV-2 spike protein (Fig. 7B; Table 1) compared to the BA.5 variant (Fig. 7A; Table 2). Against BQ.1.1, Molnupiravir obtains a very favorable total binding free energy of − 616.49 kcal/mol, driven by strong van der Waals (–576.41 kcal/mol), lipophilic (–171.27 kcal/mol), and electrostatic (–194.1 kcal/mol) interactions, combined with substantial hydrogen bonding (–20.31 kcal/mol). These interactions are reinforced by substantial residue-level contacts, notably with ASP-99, GLU-100, ASP-101, GLU-108, and GLU-109, which provide water bridges and polar stability. In contrast, Molnupiravir’s binding to the BA.5 version is substantially weaker, with a total ΔG_Bind of − 57.72 kcal/mol and fewer stabilizing interactions, particularly involving PHE-103 and GLU-100. The reduced residue engagement and larger energy swings in the BA.5 complex imply a less stable and less favorable binding conformation. Overall, both MM/GBSA energetics and residue-level mapping reveal that Molnupiravir binds more efficiently to the BQ.1.1 subvariant, underlining its potential therapeutic significance against developing SARS-CoV-2 mutations.

Table 1.

Binding energies of BQ.1.1 in complex with molnupiravir after simulations.

Variable	Binding Energies (Kcal/mol)	Mean ± SD
ΔG_Bind	−616.4	−616.49 ± 50.13
ΔG_Bind_Coulomb	−194.1	−194.1 ± 36.29
ΔG_Bind_Covalent	19.4	19.44 ± 24.75
ΔG_Bind_Hbond	−20.3	−20.31 ± 9.07
ΔG_Bind_Lipophilic Contribution	−171.2	−171.27 ± 26.12
ΔG_Bind_Packing	−6.8	−6.8 ± 3.04
ΔG_Bind_Solv_GB	217.8	217.81 ± 25.45
ΔG_Bind_vdW	−576.4	−576.41 ± 58.9

Table 2.

Binding energies of BA.5 in complex with molnupiravir after simulations.

Variables	Binding Energies (Kcal/mol)	Mean ± SD
ΔG_Bind	−57.7	−57.72 ± 60.63
ΔG_Bind_Coulomb	−86.9	−86.93 ± 78.85
ΔG_Bind_Covalent	23.6	23.67 ± 29.39
ΔG_Bind_Hbond	−13.4	−13.48 ± 10.05
ΔG_Bind_Lipophilic Contribution	−30.5	−30.52 ± 16.28
ΔG_Bind_Packing	−5.3	−5.35 ± 4.54
ΔG_Bind_Solv_GB	−28.1	−28.1 ± 100.41
ΔG_Bind_vdW	−227.8	−227.84 ± 80.83

Discussion

With the emergence of the human SARS-CoV-2, the biological community was indulged in finding answers to the reservoirs of the virus, its spread and its effect on the human race³⁴. It has been found that upon transcription, the beta coronavirus produces 800 KD polypeptide, which is cleaved by papain-like protease And 3-chymotrypsin like protease to generate various proteins which are involved in viral replication³⁵. The main protease protein having an essential role in the replication of virus serves as an amazing drug target site³⁶. Many drugs have been used against BA5 and BQ1.1 and few vaccines are developed for the inhibition of virus replication. Molnupiravir has been commonly used in Pakistan and in other countries. It has shown vital results during respiratory, chronic inflammatory disorders and bronchiolitis³⁷. Molnupiravir appears to be a better drug candidate against the reported protein structure, particularly when compared to other antiviral agents such as nirmatrelvir/ritonavir. It tends to have a superior safety profile, with substantially fewer drug-drug interactions and adverse events, making it more suitable for patients with multiple comorbidities or those on complex medication regimens.

In this study, docking was performed where Molnupiravir exhibited a binding score of −5.70 kcal/mol in the BA.5 complex, forming one hydrogen bond and ten hydrophobic interactions. In comparison, in the BQ.1.1 complex, Molnupiravir showed a slightly less favorable binding score of −5.30 kcal/mol, but formed a greater number of interactions, including seven hydrogen bonds and nine hydrophobic contacts. This observation underscores that stronger binding affinity is not solely dependent on the number of hydrogen or hydrophobic interactions. Instead, binding energy reflects the overall thermodynamic favorability, which includes factors such as the spatial complementarity of the ligand within the pocket, the desolvation penalty, entropic contributions, and the dynamic stability of the complex during simulation. In the case of BA.5, the lower binding score despite fewer hydrogen bonds suggests a more optimal fit and interaction geometry, resulting in a more energetically favorable complex.

According to MD simulations, the drug receptor complex became stable in the physiochemical environment as its shape altered over time. Despite minor alterations inside the chain and loop mobility, the inhibitor remained stable. The structural stability of the docked complex following simulation studies implies that the chosen ligand might be a viable lead chemical. Protein in complex with Molnupiravir attained high stability throughout the simulations time interval against both the variants. The findings suggested that it exhibited favorable protein-ligand contacts and optimal RMSD and RMSF values. Despite its movement in BQ.1.1 pocket, Molnupiravir maintained persistent interactions with key residues, indicating dynamic but stable engagement. Extending the simulation time or replica runs could further clarify whether the observed mobility reflects true induced fit dynamics, partial unbinding, or equilibrium among multiple binding poses. Overall, Molnupiravir exhibits potential as a viable therapeutic agent against both BA.5 and BQ.1.1 variants.

While this study provides valuable insights into the binding interactions of Molnupiravir with the targeted viral protein structures, several limitations must be acknowledged to contextualize the findings. The conclusions are based entirely on computational methodologies, including molecular docking and MD simulations, and lack experimental biochemical or cellular validation. The simulations relied on a single static crystal structure for each spike protein variant, which may not fully represent the protein conformational landscape under physiological conditions. Protein flexibility, especially in loop or receptor-binding regions, and potential allosteric effects remain underexplored. Additionally, docking protocols assume a mostly rigid receptor model and utilize simplified scoring functions. While these methods provide useful approximations, they may introduce inaccuracies when assessing flexible ligands or highly dynamic binding sites. Although MD simulations improve upon static models by incorporating flexibility and solvent effects, they are constrained by force field accuracy, ion models, system setup, and simulation timescales. Events such as long-timescale rearrangements or rare binding/unbinding transitions may not be captured within the 100 ns timeframe. Furthermore, protonation states and tautomeric forms of ligands, which can significantly influence docking results, were selected based on standard assumptions and not exhaustively sampled. The protein structures used may also contain unresolved regions or missing residues that could impact binding site accuracy. Moreover, the computational framework does not account for the full complexity of a cellular or physiological context, including drug metabolism, bioavailability, transport mechanisms, or off-target effects. As such, even favorable docking and MD results may not predict real-world antiviral potency. The significantly lower ΔG_Bind for BQ.1.1 (–616.49 kcal/mol) compared to BA.5 (–57.72 kcal/mol) indicates Molnupiravir markedly stronger and more stable interaction with the BQ.1.1 spike protein. Hence, the findings of this study must be interpreted within the scope of the above limitations and should be further substantiated through experimental validation.

Conclusion

Molnupiravir demonstrates potential as a therapeutic agent against the BA.5 and BQ.1.1 variants of SARS-CoV-2. The observed binding profiles suggest that its inhibitory effect may be attributed to a combination of hydrogen bonding and hydrophobic interactions with key residues in the spike protein, contributing to the stability of the drug-protein complex. Hydrogen bonds play a central role in molecular recognition, potentially stabilizing the complex and limiting the conformational changes required for viral entry and replication. Even in the absence of extensive hydrogen bonding, hydrophobic contacts with nonpolar regions further support the interaction stability. Binding affinity values and interaction patterns observed through docking and molecular dynamics simulations highlight the favorable interaction profile of Molnupiravir, although these results remain predictive in nature. Enhanced van der Waals, electrostatic, and hydrogen bonding contributions cause Molnupiravir’s better binding to BQ.1.1, as evidenced in the much lower ΔG_Bind. Since no experimental validation was conducted in this study, definitive claims about antiviral activity cannot be made. Instead, the findings provide a basis for future in vitro or in vivo studies that can verify and expand upon these computational predictions. However, the elucidation of Molnupiravir binding mechanism to spike proteins offers insights that could inform the structure-based design or optimization of antiviral agents targeting emerging SARS-CoV-2 variants. Continued research integrating experimental and computational approaches will be essential to fully assess the therapeutic potential and clinical relevance of these findings.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Conceptualization, F.A.; Z.B.; Y.W.; methodology, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O., Y.W.; software, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O., Y.W.; validation, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O; Y.W.; formal analysis, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O., Y.W.; investigation, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O.; Y.W.; resources, M.H.N.; F.N.A.; T.H.;; data curation, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O.,Y.W.; writing—original draft preparation, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O., Y.W.; writing—review and editing, F.A.; Z.B.; A.J.; M.H.N.; F.N.A.; T.H.; D.U.O,.; Y.W. supervision, Y.W.; project administration, M.H.N.; Y.W.; funding acquisition, M.H.N.; F.N.A.; T.H.; Y.W.; All authors have read and agreed to the published version of the manuscript.

Data availability

All data generated or analyzed during this study are included in this published article (and its Supplementary Information file). Additional data will be provided on suitable request to corresponding author.

Declarations

Conflict of interest

The authors declare no conflicts of interest.