The dynamic states of hepatitis B virus capsid monomers under the impact of different class of capsid-assembly modulators

Abstract

Hepatitis B virus (HBV) remains a global health challenge. Capsid assembly modulators (CAMs) represent a promising class of antiviral agents that disrupt HBV core antigen (HBcAg) function. Understanding the structural and dynamic impact of CAMs on HBcAg is crucial for the development of next-generation antiviral therapies. This study employed molecular dynamics (MD) simulations to evaluate the conformational behavior of capsid monomers in unbound and ligand-bound states. Different classes of CAMs, Heteroaryldihydropyrimidine (HAP), Sulfamoylbenzamide (SBA), and Ciclopirox, were analyzed to assess their impact on HBcAg stability, flexibility, and interaction energy. RMSD analysis revealed that HAP binding stabilized HBcAg, reducing backbone fluctuations, whereas SBA and PPA increased HBcAg flexibility. RMSF calculations demonstrated that CAM interactions influenced loop and terminal region dynamics. PCA suggested ligand-specific alterations in HBcAg’s essential motions, with Sulfamoylbenzamide inducing the highest variance. Salt bridge analysis indicated that Ciclopirox formed the strongest electrostatic interactions, stabilizing its binding. DSSP secondary structure analysis showed that CAMs disrupted α-helical content, with Sulfamoylbenzamide and Ciclopirox exhibiting the most pronounced structural rearrangements. This study provides novel insights into CAM-induced conformational changes in HBcAg. While HAP stabilizes the core protein, SBA and Ciclopirox increase flexibility, potentially leading to misassembled or destabilized capsids. These findings contribute to the rational design of CAM-based antiviral therapies and highlight key structural determinants for future drug optimization.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-18339-6.

Introduction

Hepatitis B virus (HBV) remains a global health concern, affecting over 296 million individuals chronically and leading to approximately 820,000 deaths annually due to liver cirrhosis and hepatocellular carcinoma^1,2. Despite the availability of vaccines and nucleos(t)ide analog therapies, these treatments do not eliminate covalently closed circular DNA (cccDNA) and are challenged by resistance issues³. Therefore, innovative therapeutic strategies are imperative for managing chronic HBV infection (CHB). HBV capsid assembly modulators (CAMs) have emerged as a promising class of antiviral agents, targeting the HBV core antigen (HBcAg) essential for viral replication and cccDNA formation.

The capsid protein (Cp) or HBcAg, an icosahedral structure formed by 120 homodimers of Cp, relies on precise interactions at the dimer-dimer interface for its assembly. CAMs exploit these interactions by either misdirecting assembly into non-functional capsids or stabilizing incomplete intermediates⁴. CAMs like Heteroaryldihydropyrimidines (HAPs), Isoquinolinone derivatives, and AB-836, function by inducing defective or empty capsids, thereby hindering the encapsidation of pregenomic RNA (pgRNA) and viral polymerase, critical for viral replication and persistence⁴. CAMs can suppress extracellular HBV RNA, including pgRNA and spliced variants, an effect not observed with nucleos(t)ide analogs⁵. Furthermore, preclinical data from CAMs like RG7907 demonstrated a promoted nuclear aggregation of core proteins, triggering apoptosis of infected hepatocytes and reducing cccDNA reservoirs in preclinical models⁶. While no CAM has yet received FDA approval, a Phase 1b Clinical Trial of ABI-4334, a next-generation CAM demonstrated potent antiviral activity in a 28-day trial, achieving mean HBV DNA reductions of 3.2 log₁₀ IU/mL at 400 mg daily. Pharmacokinetic data support once-daily dosing, with trough concentrations exceeding protein-adjusted EC₅₀ for both viral replication inhibition and cccDNA formation suppression⁷. CAMs uniquely reduce extracellular HBV RNA, a potential biomarker for functional cure. However, their inability to directly eliminate cccDNA necessitates combination strategies. Early CAMs faced limitations in potency and durability, driving development of next-generation agents like ABI-4334, which targets both viral replication and cccDNA establishment^7,8.

In the mechanistical point of view, direct-acting CAMs have distinct mechanism of actions. HAPs may bind to a hydrophobic pocket at the dimer-dimer interface, inducing allosteric changes that accelerate improper assembly or destabilize preformed capsids. HAP-1, in particular, destabilizes mature capsids while promoting the formation of aberrant structures⁹. Sulfamoylbenzamides (SBAs) and phenylpropenamides (PPAs) offer distinct mechanisms of action. SBAs, such as AB-836, inhibit pgRNA-containing nucleocapsid formation, thereby reducing cccDNA production and viral replication¹⁰. Molecular dynamics (MD) simulations and structure-activity relationship (SAR) studies have further elucidated the binding dynamics of these CAMs. Hydrophobic interactions and van der Waals forces are essential for stabilizing inhibitor-protein complexes, while specific substitutions on CAM scaffolds enhance efficacy¹¹.

The CAMs’ effects are multifaceted and it is not completely understood. It has been reported that CAMs impact capsid stability by inducing structural rearrangements in the spike and base regions of Cp¹². CAM-A compounds increase dynamics within the spike region, while CAM-E compounds stabilize inter-subunit interactions, promoting tighter capsid assembly¹³. These effects are critical for disrupting the HBV lifecycle, as they prevent encapsidation of pgRNA and inhibit reverse transcription. HAP18 disrupts the hydrogen bonding networks within HBV dimers and stabilizes capsid structures, demonstrating allosteric communication that modulates capsid dynamics⁴. Isoquinolinone-based CAMs have been shown to suppress HBV replication by binding at the dimer: dimer interface of the HBV core protein, locking the preferred conformation and enhancing drug efficacy¹³. Moreover, AB-836, a potent Class II CAM, inhibits viral replication by preventing the formation of cccDNA, thereby suppressing HBV antigen production and transcription during de novo infection⁹. Therefore, the therapeutic potential of CAMs extends beyond direct antiviral activity. By targeting HBV capsid assembly, CAMs effectively suppress antigen production, reduce immune evasion, and limit cccDNA replenishment. Their specificity for the HBV core protein minimizes off-target effects, making them suitable for combination therapies with existing antivirals. In our last study¹⁴it observed that HAP dynamically stabilize Y132A mutant of HBcAg and compound with structurally similar features, like Lovastatin and Simvastatin, also induce similar impacts. Furthermore, Class I CAMs like SBAs, which characterized by a core scaffold combining a sulfonamide group with a benzamide moiety. The first-in-class SBA, NVR 3-778, binds at the dimer-dimer interface of the HBV Cp, inducing aberrant capsid assembly and suppressing viral replication by preventing pgRNA encapsidation and cccDNA formation^15,16. Unlike classical Class II CAMs, some SBAs induce time-dependent formation of tubular capsid structures, revealing a hybrid mechanism between Class I and II modulators. Also, Ciclopirox an FDA approved antifungal, was recently repurposed as an HBV CAM. it binds the hydrophobic pocket at the Cp149 dimer interface, inhibits capsid assembly¹⁷. By leveraging computational approaches, this study aim to uncover dynamic of HBcAg in complex with these unique CAMs and contribute to the rational design of next-generation antivirals.

Results

Molecular dynamic simulation

This atom-level MD simulation study was performed, considering it an efficacious approach for studying the behaviour of HBcAg monomer in apo and holo systems with Ciclopirox, Sulfamoylbenzamide, and Heteroaryldihydroprymidine. MD analysis would provide valuable information regarding the dynamics of the complexes. Also, we could see how HBcAg monomers are conformationally affected by the binding of different types of HBV CAMs. Therefore, the trajectories were analyzed within 100 ns all-atom MD simulations.

Trajectory analysis of HBcAg Apo and holo systems

The stability of HBcAg-Heteroaryldihydroprymidine, HBcAg-Ciclopirox, and HBcAg-Sulfamoylbenzamide holo systems was monitored using the GROMACS rms module to estimate their respective RMS values throughout the 100 ns simulation runs. Additionally, the RMSD fluctuation of the HBcAg apo system (0.35 ± 0.04 nm) was assessed to determine the convergence time at which initial conditions and notable fluctuations are stabilized, indicating system equilibration. Also, RMSD fluctuations were measured at the backbone of HBcAg systems (Fig. 1).

Fig. 1.

RMSD analysis of HBcAg in unbound and CAM-bound states. (a) RMSD of HBcAg alone and in complex with CAMs over a 100 ns MD simulation. The black dashed line at 15 ns indicates the assumed convergence time, beyond which all systems exhibited stable fluctuations. (b) Ligand RMSD fluctuations within their respective binding pockets, indicating binding stability. Sulfamoylbenzamide exhibited transient dissociation from HBcAg, suggesting weaker interaction stability.

The backbone RMSD fluctuations of HBcAg and its complexes with Ciclopirox, Sulfamoylbenzamide, and Heteroaryldihydroprymidine were analyzed over a 100 ns molecular dynamics (MD) simulation (Fig. 1a). The assumed convergence time was determined at approximately 15 ns (see the Black dashed line in Fig. 1), beyond which all systems exhibited stable fluctuations. The HBcAg apo system maintained an average RMSD of 0.35 ± 0.04 nm. The HBcAg_Ciclopirox complex exhibited slightly higher structural deviations (0.41 ± 0.04 nm), implying a stable yet more flexible binding state. The HBcAg_Sulfamoylbenzamide complex showed the highest RMSD (0.47 ± 0.05 nm), indicating notable structural perturbations upon ligand binding. In contrast, the HBcAg_Heteroaryldihydroprymidine complex showed the lowest HBcAg flexibility (RMSD of 0.29 ± 0.07 nm). The ANOVA analysis (Table 1) showed a significant differences of RMSD means between and among systems.

Table 1.

One‑way repeated‑measures ANOVA of systems’ mean RMSD.

Source	SS	DF	MS	F	p-value
Measures	187.619	3	62.540	29882.13	< 0.0001
Subjects	51.080	10,000	0.005
Error	62.787	30,000	0.002
Total	301.485	40,003	0.008

H0: µ₁ = µ₂ = µ…

The mean of the populations are all equal.

H1: µ_i ≠ µ_j for at least one i, j

The mean of the populations are not all equal.

Ligand stability within the binding pocket was assessed by evaluating the RMSD of individual ligands (Fig. 1b). The Heteroaryldihydroprymidine ligand exhibited the lowest RMSD (0.07 ± 0.02 nm), indicating a highly stable binding mode with minimal fluctuation. Conversely, Sulfamoylbenzamide displayed a higher RMSD (0.195 ± 0.039 nm), suggesting moderate movement within the pocket. Ciclopirox exhibited RMSD fluctuations around 0.084 ± 0.036 nm, indicating a relatively stable yet dynamic interaction.

The radial distribution function (RDF) analysis revealed distinct patterns of ligand distribution around HBcAg. Heteroaryldihydropyrimidine exhibits the smallest average radial distance from the protein (1.84 nm) and an average radial (r) distribution of 2.70, achieving the highest cumulative number of ligand molecules (24.66) within that radius. By contrast, Ciclopirox showed the largest average g(r) of 4.80 at an distance of 2.06 nm, indicating strong local enrichment at specific separations; however, its cumulative occupancy (14.92) is the lowest among the three ligands, implying that it does not remain as persistently in close proximity to the protein (see Fig. 2). Quantitative analysis of ligand-protein center-of-mass (COM) distances reveals distinct binding stability profiles. Sulfamoylbenzamide demonstrates a dissociation over the simulation timeframe, with mean distance from the reference HBcAg protein significantly increasing from 3.38 nm (15–30 ns) to 3.86 nm (90–100 ns) (F(5,11010) = 405.03, p < 0.0001). This drift is accompanied by high spatial variability (SD = 3.01–3.76 nm), indicating transient sampling of unbound states. The dissociation trajectory shows three distinct phases, including an initial weakening (15–45 ns: +0.38), a peak in radial distribution numbers (45–60 ns: 3.87), and a sustained displacement (> 3.80 beyond 60 ns), which are consistent with the ligand transient dissociation (Supplementary Video 1). In contrast, Heteroaryldihydropyrimidine maintains constricted binding (2.69–2.73, Δ = 0.04) with minimal fluctuation (SD ≈ 2.15 nm). Though ANOVA showed statistical significance (F(5,9165) = 208.81, p < 0.0001), the effect size was negligible (η²=0.00005), confirming binding stability. Also, Ciclopirox exhibits non-dissociative fluctuation as observed in the simulation. In this regard, Ciclopirox radial distributions increased to 5.16 (60–75 ns) before partially rebinding (4.14 at 90–100 ns). The significant statistical result (F(5,10270) = 626.14, p < 0.0001) might reflect this oscillation rather than dissociation.

Fig. 2.

RDF and probability density distributions of ligand-center-of-mass (COM) distances from the protein binding site for (left-to-right) Ciclopirox, Sulfamoylbenzamide, and Heteroaryldihydropyrimidine. Sulfamoylbenzamide exhibits progressive dissociation (distance increase: 3.38 → 3.86 nm, F = 405.03, p < 0.0001), while Ciclopirox shows non-monotonic fluctuations and Heteroaryldihydropyrimidine maintains stable binding with consistent radial distribution patterns at different time spans with sharp unimodal peaks characteristic of stable binding at a COM distance more close to HBcAg (1.32 nm). Data derived from 15-100-ns MD simulations; bin width = 0.1 nm; reference: protein COM; selection: ligand COM.

Backbone RMSF analysis

To see if the lower HBcAg’s backbone RMSD was due to the size of the ligands, per-residue and per-atom (Fig. 3) fluctuations were evaluated using the gromacs rmsf module after system convergence time. Accordingly, The HBcAg apo system (blue line) exhibited the lowest fluctuations, indicating the unbound protein rigidity. The HBcAg complex with the Ciclopirox complex displayed slightly increased fluctuations, particularly in loop regions. The HBcAg-Sulfamoylbenzamide complex exhibited higher fluctuations, particularly in regions near the ligand binding site. Moreover, The HBcAg-Heteroaryldihydroprymidine system showed the highest backbone fluctuations, particularly in terminal and loop regions, indicating localized structural flexibility upon ligand binding. It was also observed that all systems exhibited the highest fluctuations at the N-terminal region (M1-I13), central loop region (N74-L84), and C-terminal residues (A131-S141), which are typically flexible regions of the HBcAg protein.

Fig. 3.

RMSF analysis of HBcAg in unbound and CAM-bound states. RMSF analysis highlights per-residue atomic fluctuations, revealing increased flexibility in specific regions upon ligand binding. The highest fluctuations were observed at the N-terminal (M1-I13) and C-terminal (A131-S141) residues, as well as central domains. Further structure alignments of systems are depicted in 3D using PyMOL v3.1.1. Consequently, a random frame from the last 10 ns (90 ns to 100 ns) of the trajectory was transformed into a pdb file for HBcAg. For other holo systems, a corresponding frame was achieved for structural alignment. In this context, all holo structures were aligned to the HBcAg apo system as a reference. It is evident that α-helical structures convert into coils in all holo systems. To clarify, crystallographic structures of ligands-complexed HBcAg are illustrated (small images in the top-right panel) to show that α-helical structures were evident in the original crystallographic data (ligands are omitted for improved visualization). Amino acid residues for the loop domain (73 to 83) are depicted in blue in 3D structures for clarity. The alignment results show that Heteroaryldihydroprymidine led to greater distance residues in the loop domain compared to the original state in HBcAg.

To assess ligand stability, RMSF per atom of each ligand was calculated (Fig. 4). Heteroaryldihydroprymidine exhibited the lowest average RMSF (0.07 nm), indicating highly stable binding. The highest fluctuations were observed at C30 (0.0788 nm), O31 (0.1442 nm), and O32 (0.1504 nm), suggesting minor flexibility in functional groups. Sulfamoylbenzamide displayed higher fluctuations (0.195 nm on average), suggesting movements within the binding pocket. Significant fluctuations were observed in C02 (0.2395 nm), F18 (0.2166 nm), and H2 (0.3352 nm), and some degree of instability was observed in these regions. Also, Ciclopirox had relatively stable interactions, with an average RMSF of 0.084 nm. Higher fluctuations were detected at H3 (0.1141 nm), H5 (0.1133 nm), and H8 (0.0946 nm), suggesting that hydrogen atoms in its structure may contribute to flexible interactions.

Fig. 4.

Ligand atomic fluctuation analysis. RMSF per atom for each ligand in complex with HBcAg, showing relative stability within the binding pocket. Heteroaryldihydropyrimidine exhibited the lowest atomic fluctuations, suggesting a highly stable binding mode, whereas Sulfamoylbenzamide displayed significant movement, correlating with its transient dissociation.

Interaction energy analysis

Molecular mechanics calculations were performed to investigate the binding interaction energies between HBcAg and CAMs. The short-range Coulombic (Coul-SR) and Lennard-Jones (LJ-SR) interaction energies, along with the 1–4 electrostatic (Coul-14) and 1–4 van der Waals (LJ-14) interactions, were analyzed over the convergence time (15–100 ns).

The HBcAg-Ciclopirox complex exhibited a moderate Coulombic short-range interaction energy (−24.76 kJ/mol) and a strong van der Waals stabilization (−79.46 kJ/mol), indicating that hydrophobic interactions contribute significantly to the ligand binding. Interestingly, the Coul-14 interaction energy was highly negative (−102.19 kJ/mol). This could be due to a strong electrostatic attraction, likely facilitated by salt bridge formation between Ciclopirox’s charged functional groups and basic HBcAg residues. The LJ-14 component was slightly positive (49.67 kJ/mol), indicating minor steric repulsions within the binding site.

The Heteroaryldihydroprymidine complex exhibited the most favorable interaction energy profile among the three ligands, with a highly stabilizing Coulombic short-range energy (−52.08 kJ/mol) and strong van der Waals interactions (−128.32 kJ/mol). These findings suggest that the ligand fits well within the binding pocket, forming extensive non-covalent interactions. However, the Coul-14 interaction was highly positive (665.89 kJ/mol), implying the presence of electrostatic repulsion between the ligand and nearby protein residues. Despite this, the negative LJ-SR (−128.32 kJ/mol) and favorable LJ-14 (86.97 kJ/mol) interactions suggest that van der Waals forces help stabilize the complex.

Moreover, the Sulfamoylbenzamide complex showed moderate Coulombic stabilization (−39.19 kJ/mol) but a robust Heteroaryldihydroprymidine-comparable van der Waals interactions (−127.38 kJ/mol), indicating significant hydrophobic contributions to ligand binding. The Coul-14 energy (−1221.26 kJ/mol) was the most negative among all complexes, suggesting extensive charge-charge interactions, possibly involving multiple salt bridges (SBs) with charged residues in the binding pocket (see Fig. 5, Table panel). There were significant mean differences among different calculated interaction energies between and among systems (for more detail see Table 2).

Fig. 5.

Interaction energy analysis of HBcAg-ligand complexes. Two distinct interaction types were observed: (i) van der Waals and electrostatic short-range interactions stabilizing all three complexes and (ii) a strong 1–4 electrostatic interaction in Ciclopirox and Sulfamoylbenzamide complexes, contributing to their binding stability.

Table 2.

One‑way repeated‑measures ANOVA of LJ‑SR and Coul‑SR.

Energy	Source	SS	DF	MS	F	p-value
LJ‑SR	Measures	1.33E + 07	2	6.64E + 06	21595.27	< 0.0001
	Subjects	3.28E + 06	8500	385.09
	Error	5.23E + 06	17,000	307.42
	Total	2.18E + 07	2550	853.94
Coul‑SR	Measures	3.18E + 06	2	1.59E + 06	2822.52	< 0.0001
	Subjects	5.36E + 06	8500	630.48
	Error	9.57E + 06	17,000	563.06
	Total	1.81E + 07	25,502	710.13

H0: µ₁ = µ₂ = µ…

The mean of the populations are all equal.

H1: µ_i ≠ µ_j for at least one i, j

The mean of the populations are not all equal.

Salt bridge analysis

Owing to the meaningful Coulombic 1–4 interaction energies between ligands and HBcAg, index files containing positively (Lys, Arg, and His) and negatively (Glu and Asp) charged HBcAg residues were generated. The files were used to evaluate the number of SB formations and possible hydrogen bondings between ligands and the receptor through Gromacs mindist and hbond modules. The SB formation was analyzed at a maximum distance of 3.5 Å and 30° angle.

As expected, Ciclopirox and Sulfamoylbenzamide form high amounts of SBs with positively and negatively charged amino acids of HBcAg (see Fig. 6). The HBcAg-Ciclopirox complex exhibited the highest SB formation among the three ligands. The average number of SBs with negatively charged residues was 13.99 ± 7.40, while positive residues formed an even higher numbers (25.91 ± 11.53). This suggests that Ciclopirox forms strong electrostatic interactions with both acidic and basic residues in the binding pocket, likely contributing to its observed highly negative Coul-14 energy (−102.19 kJ/mol). The significant fluctuations in the number of contacts over time indicate dynamic interactions, possibly due to ligand flexibility or conformational shifts observed in the ligand RMSD.

Fig. 6.

Salt bridge formation analysis between CAMs and HBcAg. Ciclopirox exhibited the highest number of salt bridges with positively and negatively charged HBcAg residues, suggesting strong electrostatic stabilization. Sulfamoylbenzamide formed moderate salt bridges, while Heteroaryldihydropyrimidine relied more on hydrophobic interactions.

The Sulfamoylbenzamide complex displayed moderate SBs, but its interactions were more selective. The average number of SBs with negatively charged residues was only 0.38 ± 1.48, while interactions with positively charged residues were higher (7.80 ± 8.84). However, the significantly lower interaction with negatively charged residues indicates a lack of strong anionic interactions, which could explain why its binding was not as stable as Ciclopirox. Interestingly, Heteroaryldihydroprymidine formed no SB with positively charged residues (0 contacts at > 3.5 Å), while it displayed weak interactions with negatively charged residues (0.41 ± 1.01). This suggests that electrostatic forces are not the primary contributors to the stability of this complex. Instead, van der Waals interactions (as shown by LJ-SR: −128.32 kJ/mol) likely play a dominant role in ligand stabilization. The lack of SB formation may explain why Heteroaryldihydroprymidine exhibited positive Coul-14 interactions (665.89 kJ/mol), indicating repulsion rather than attraction. Detailed descriptive and ANOVA analyses of number of hydrogen bindings are provided in Fig. 7; Table 3.

Fig. 7.

Descriptive statistics of number of hydrogen bindings of salt bridges in the HBcAg holo systems. Data indicates frequent hydrogen bonding of Ciclopirox with positively and negatively charged residues.

Table 3.

One‑way repeated‑measures ANOVA of number of hydrogen contacts < 0.35 nm in different HBcAg holo systems.

Bonds	Source	SS	DF	MS	F	p-value
Neg. Charged Res.	Measures	1047814.1	2	523907.0	26670.40	< 0.0001
	Subjects	159869.9	8500	18.8
	Error	333943.9	17,000	19.6
	Total	1541627.9	25,502	60.5
Pos. Charged Res.	Measures	3004050.8	2	1502025.4	21106.25	< 0.0001
	Subjects	586694.7	8500	69.0
	Error	1209804.5	17,000	71.2
	Total	4800550.0	25,502	188.2

H0: µ₁ = µ₂ = µ…

The mean of the populations are all equal.

H1: µ_i ≠ µ_j for at least one i, j

The mean of the populations are not all equal.

We have further investigated the hydrogen bond formation between the ligand and the proposed residues in the range of salt bridge formation. Accordingly, an additional index file was generated for each system. For the HBcAg-Ciclopirox complex, the index file was composed of Glu42, Arg38, Arg55, and Arg27. For Sulfamoylbenzamide and Heteroaryldihydroprymidine, the index file comprised Arg149 and Asp28, respectively. As depicted in Fig. 6, Ciclopirox makes high numbers of binding events with positively and negatively charged HBcAg residues. However, it only establishes salt bridges with one negatively charged amino acid residue, Glu42 (0.25 ± 0.09 nm), and three positively charged residues, Arg38 (0.28 ± 0.13 nm), Arg55 (0.30 ± 0.11 nm), and weakly with Arg27 (0.51 ± 21) (Fig. 8). For Sulfamoylbenzamide (Fig. 9), only one positively charged residue, Arg149 (0.46 ± 0.31 nm), was observed in the range of 3.5 Å. In addition, a weakly possible salt bridge was observed between Heteroaryldihydroprymidine and a negatively charged residue, Asp28 (0.44 ± 0.09 nm) (see Fig. 10).

Fig. 8.

Salt bridge and hydrogen bonding analysis between Ciclopirox and HBcAg. (Top) Time evolution of the distances between negatively charged (left) and positively charged (right) residues of HBcAg and Ciclopirox ligands over 15 to 100 ns of MD simulation. The fluctuations indicate dynamic ligand-residue interactions, with some residues maintaining stable interactions while others exhibit transient binding. (Bottom) Average distances of negatively (left) and positively (right) charged residues from the ligand binding site, highlighting preferential distance of hydrogen bondings.

Fig. 9.

Salt bridge and hydrogen bonding analysis between Sulfamoylbenzamideand and HBcAg. (Top) Time evolution of the distances between negatively charged (left) and positively charged (right) residues of HBcAg and Sulfamoylbenzamide ligands over 15 to 100 ns of MD simulation.

Fig. 10.

Salt bridge and hydrogen bonding analysis between Heteroaryldihydroprymidine and HBcAg. (Top) Time evolution of the distances between negatively charged (left) and positively charged (right) residues of HBcAg and Sulfamoylbenzamide ligands over 15 to 100 ns of MD simulation.

Cluster analysis

Cluster analysis was performed to evaluate the RMS distributions across unbound state of HBcAg and its complexes with Ciclopirox, Heteroaryldihydropyrimidine, and Sulfamoylbenzamide. The mean RMS values for the primary clusters (likely representing intra-cluster stability) were relatively low across all groups, ranging from 0.173 ± 0.101 nm (Ciclopirox) to 0.227 ± 0.132 nm (Sulfamoylbenzamide), with 0.178 ± 104 nm (HBcAg), and 0.212 ± 0.124 nm (Heteroaryldihydropyrimidine) falling within this spectrum (Fig. 11).

Fig. 11.

Cluster analysis of HBcAg in unbound and CAM-bound states. The histograms illustrate the distribution of RMS values for HBcAg (blue), Ciclopirox (yellow), Heteroaryldihydropyrimidine (red), and Sulfamoylbenzamide (green) throughout the molecular dynamics simulations. Each peak represents a distinct conformational cluster, indicating the predominant structural states adopted during the simulation. The differences in clustering suggest that ligand binding modulates the structural dynamics of HBcAg, with Ciclopirox and Sulfamoylbenzamide inducing more flexible conformational transitions.

Also, the mean number of clusters was similar (0.356 nm) for all groups. However, variability in these clusters was markedly greater, with standard deviations ranging from 0.623 nm (Ciclopirox and Sulfamoylbenzamide) to 0.683 nm (HBcAg).

SASA analysis

As shown in Fig. 12, analysis of Solvent Accessible Surface Area (SASA) showed differences in solvent exposure profiles. The average SASA values ranged from 90.64 ± 1.67 nm² (HBcAg) to 101.03 ± 2.39 nm² (Sulfamoylbenzamide-complex), with intermediate values observed for HBcAg-complexed Heteroaryldihydroprymidine (93.87 ± 1.75 nm²) and HBcAg-complexed Ciclopirox (99.55 ± (2.20 nm²). HBcAg protein exhibited the highest mean solvent accessibility in complex with Sulfamoylbenzamide. The result showed a relatively compact or shielded surface of HBcAg when it is in complex with Heteroaryldihydroprymidine.

Fig. 12.

SASA analysis of HBcAg and its complexes with CAMs. The graph illustrates the changes in solvent exposure during the MD simulations, with HBcAg (blue) exhibiting the lowest SASA, indicating a compact structure. Sulfamoylbenzamide (green) resulted in the highest SASA, suggesting a more open or destabilized conformation, while Ciclopirox (yellow) and Heteroaryldihydropyrimidine (red) exhibited intermediate effects.

Secondary structure calculation

DSSP analysis quantified profound ligand-induced destabilization of HBcAg’s native α-helical fold (Table 4). The apo HBcAg maintained a dominant α-helical conformation (60.5% of states; mean = 5143.53 ± 3778.69), with minor contributions from bends (7.6%), turns (9.6%), and coils (11.9%). Significantly, all holo systems exhibited complete α-helix ablation (0% occupancy), accompanied by extensive reorganization into disordered states. The demonstrated key structural shifts. Accordingly, bend dominance in all complexes showed > 82% bend occupancy (vs. 7.6% in apo), indicating adoption of kinked, non-helical conformations. Also, coil content increased by 35–40% in complexes (15.8–16.3% vs. 11.9% apo), confirming global flexibility. Moreover, complete order loss was observed by fully abolished β-strands, turns, 3₍₁₀₎-helices, and π-helices in ligand-bound states (Fig. 13).

Table 4.

Descriptive statistics of secondary structure changes from Apo to holo systems of HBcAg.

Secondary structure	HBcAg (Apo)	Sulfamoylbenzamide	Ciclopirox	Heteroaryldihydropyrimidine
α-Helix	5143.53 ± 3778.69	0 ± 0	0 ± 0	0 ± 0
Bends	642.92 ± 1928.10	7016.55 ± 2927.19	7078.04 ± 2947.72	7112.24 ± 2892.35
β-Strands	0 ± 0	0 ± 0	0 ± 0	0 ± 0
Turns	818.74 ± 1647.56	0 ± 0	0 ± 0	0 ± 0
3(10)-Helix	436.49 ± 1125.12	0 ± 0	0 ± 0	0 ± 0
π-Helix	358.26 ± 1273.04	0 ± 0	0 ± 0	0 ± 0
Polyproline Helix	91.01 ± 508.70	143.39 ± 686.62	108.39 ± 596.24	144.88 ± 605.88
Coils	1010.84 ± 2489.00	1341.06 ± 2684.22	1314.57 ± 2742.64	1243.88 ± 2596.55
Total States	8502	8501	8501	8501

Fig. 13.

DSSP-derived secondary structure map of unbonded state of HBcAg and its complexes with Ciclopirox, Sulfamoylbenzamide, and Heteroaryldihydroprymidine from 15 to 100 ns of simulation. Each column represents a simulation snapshot, and each row corresponds to a protein residue. Colors indicate the number of secondary structure for each residue at different time frame (Blue the highest numbers).

Residue-specific destabilization occurred at critical assembly motifs. In this regard, a complete α→bend transition at Y118–P135 in all complexes (orange boxes) was observed, disrupting C-terminal dimerization interfaces. Also, central helix (Q57–L84): 95% helix loss at residues 70–80 (dimer-dimer contact zone), replaced by polyproline helices (1.7% occupancy). These shifts confirm that CAM binding induces a near-complete α→disorder transition, directly compromising structural elements essential for capsid assembly.

PCA analysis

PCA of the HBcAg systems revealed differences in overall atomic fluctuations and the distribution of essential motions upon ligand binding. The sum of eigenvalues, representing the total variance in atomic displacements, was 13.82 nm² for unbound HBcAg, 10.70 nm² for the Ciclopirox complex, 23.90 nm² for the Sulfamoylbenzamide complex, and 14.89 nm² for the Heteroaryldihydroprymidine complex. Focusing on the first ten principal components—which captured over 80% of the variance in the unbound state—the cumulative variance accounted for was 81.77% for HBcAg alone, 79.05% for the Ciclopirox complex, 87.54% for the Sulfamoylbenzamide complex, and 85.99% for the Heteroaryldihydroprymidine complex. These results indicate that while the essential dynamics of HBcAg are largely confined to the top ten PCs, ligand binding alters the overall fluctuation landscape in a ligand-dependent manner, potentially influencing the protein’s functional motions and stability.

The evolution time of HBcAg’s projection onto the first ten eigenvectors showed all PCs are actively contributing in atomic projections. Except one major projection at 50 nm time-point in PC1, the system was stable (Fig. 14A). Similar behaviours were also observed among the HBcAg complexes in the PC5 to PC10. But in the first five PCs the projections were diverse, indicating the involvement of the earlier PCs in reflecting system projections. However, for the sake of reaching 80% or higher trajectory data capture by PC, ten components were selected. Furthermore, the mode of motion for different component of HBcAg and its complex with the ligands is depicted in Fig. 14B. As a finding, except of PC1, PC2, and PC4, mostly C-terminal of the protein has some peak of total (Black lines) atomic motions mostly in horizontal (x and y) axis. Also, most of the central domain atomic movements were also observed almost in all PCs. For Ciclopirox-complexed HBcAg, the atomic motions were more pronounced in both C-terminal (atoms 370 to 420) and central domains (atoms 120 to 150 and 200 to 250). For Sulfamoylbenzamide complex, first 20 atoms at the N-terminal domain were significantly mobile mostly in 2D horizontal axis. In this regard, similar motion like that observed in Ciclopirox complex at central domain was observed but not in all of PCs. However, the C-terminal domain was significantly motile at the ending atoms compare to HBcAg unbound state and Ciclopirox complex (see Fig. 14B). In the contrary to these, the complex of HBcAg to Heteroaryldihydroprymidine was very stable at the C-terminal domain. This could also explain the induced flexibility of HBcAg in complex by Heteroaryldihydroprymidine as observed in the RMSD analysis. Nevertheless, like other complex the HBcAg was mobile at the N-terminal atoms and central domain similarly.

Fig. 14.

(A) Time evolution of the PC projections for unbound HBcAg (black) and HBcAg complexed with Ciclopirox (orange), Sulfamoylbenzamide (green), and Heteroaryldihydropyrimidine (red). The x-axis shows simulation time (ns), and the y-axis is the projection amplitude (nm) onto the indicated PC, illustrating how each system samples that collective mode over time. (B) Per-atom contributions to the PCs for each system. In each subplot, the black line represents the total displacement amplitude, while the red, green, and blue lines show the x, y, and z components, respectively, plotted against the atom index from N- to C-terminus. Peaks indicate regions of the protein exhibiting the largest motions along the corresponding eigenvector.

The RMSF of individual atoms was also evaluated for the given PCs. In accordance to the eigenvalues analysis for atomic motion, the RMSF findings revealed significant atomic peaks at the already mentioned domain. This finding also supported RMSF of amino acid residues mentioned earlier (Fig. 15).

Fig. 15.

RMSF profiles of HBcAg (top left) and its complexes with Ciclopirox (top right), Sulfamoylbenzamide (bottom left), and Heteroaryldihydropyrimidine (bottom right) along selected PCs. Each row depicts the RMSF of individual atoms (x-axis) in nanometers (y-axis) for a specific eigenvector. Peaks highlight regions of the protein exhibiting the greatest atomic displacements in each collective motion, revealing how ligand binding alters the spatial distribution of fluctuations relative to unbound HBcAg.

Further conformational substates of HBcAg were assessed (Fig. 16). According to the 2D projection analysis over the first PC with highest motions and the last PC with more stable state, HBcAg spend most of the time of simulation in a stable state at two places with almost similar motions as observed in both PCs. The less condensed bridge between these two clusters (black snapshots between two condensed clouds; Fig. 16) is also indicates a possible conformational transformation between these two states. This finding was used as a base conformational structure for making the possibility of comparing the apo state with conditions that HBcAg is complex with CAMs. Accordingly (Fig. 15), upon Ciclopirox binding, the bridge mentioned between two states is almost disappeared. This possibly confers HBcAg to has a rapid conformational transform state, and structurally rearrange in the second place. Even more interestingly, HBcAg had a more structurally flexible transitions when it was in complex with Sulfamoylbenzamide, since it was observed that the bridge was extended for almost 2 nm, and this made a new cluster in space. Moreover, the first structure was more condensed in a new space almost 1 nm away from the one observed in apo state of HBcAg. These new conformational states or substates are possibly involve in the structural flexibility induced by Sulfamoylbenzamide. HBcAg was more stable and the trajectory snapshots were more condensed in the first structure in its complex with Heteroaryldihydroprymidine (Fig. 16). However, the second place was also nested within the second conformation but less dense. Moreover, a new place in space was also observe that HBcAg met in this complex. This finding suggested that HBcAg preserved and stable when bound to Heteroaryldihydroprymidine, and also has a new substate (3rd), which was similar to the one induced by Sulfamoylbenzamide. This supports how Sulfamoylbenzamide reduces the flexibility of HBcAg as observed in RMSD analysis.

Fig. 16.

Two-dimensional projection of HBcAg conformational substates during MD simulations. Ligand binding influenced conformational space, with Sulfamoylbenzamide introducing additional conformational substates, whereas Heteroaryldihydropyrimidine stabilized HBcAg with minimal fluctuation.

Discussion

HBV remains a significant global health concern, with chronic infections leading to severe liver diseases, including cirrhosis and hepatocellular carcinoma^2,18–20. A fundamental component in the HBV lifecycle is the capsid, a protein shell that encases the viral genome and plays essential roles in replication and assembly^14,21. Targeting the capsid assembly process has emerged as a promising therapeutic strategy, leading to the development of CAMs²². In this study it was aimed to dynamically evaluate the HBV capsid monomer in unbound apo or complexed holo states.​ The MD simulations and trajectory analysis provided detailed insights into the structural perturbations induced by distinct classes of CAMs, Heteroaryldihydropyrimidine, Ciclopirox, and Sulfamoylbenzamide. These findings contribute to the growing body of evidence that highlights the efficacy and mechanistic of CAMs in disrupting HBV replication and capsid stability.

HBV capsid assembly is a critical step in the viral lifecycle, with the capsid protein forming an icosahedral structure^4,22. The stability and structural integrity of these capsids are essential for the encapsidation of pgRNA and subsequent reverse transcription. Capsid assembly inhibitors function through two primary mechanisms: (1) misdirecting assembly into non-functional structures or aberrant capsid formation (Class-A CAMs) and (2) stabilizing immature capsids to prevent RNA encapsidation or empty capsid formation (Class-E CAMs)^9,13,23. This study revealed distinct effects of CAMs on HBcAg conformation and stability, aligning with previous studies that highlight the potential of CAMs as therapeutic agents²⁴. Accordingly, RMSD analysis indicated differential impacts of CAMs on the HBcAg backbone. HAP exhibited the lowest fluctuations, suggesting a strong stabilizing effect. Conversely, Sulfamoylbenzamide and Ciclopirox induced a relatively higher RMSD, indicating significant structural perturbations. These findings align with previous reports that classify CAMs into different mechanistic categories. HAPs, such as BAY 41-4109, are known to stabilize capsids while promoting aberrant structures at high concentrations²⁵. Similarly, the misdirecting effects observed in Sulfamoylbenzamide interactions are consistent with reports on bis-ANS, a small molecule known to misdirect capsid assembly²⁶.

The binding affinity and stability of CAMs within the HBcAg binding pocket were assessed through RMSD and RDF analyses. HAP remained the most tightly bound ligand in close contact, exhibiting the smallest average radial distributions and the highest cumulative ligand occupancy. These findings suggest strong and persistent interactions, reinforcing its role as a capsid stabilizer²⁷. Sulfamoylbenzamide, despite moderate binding affinity, exhibited transient dissociation events, consistent with its proposed sliding and surface-exploring behavior. This aligns with studies showing that sulfamoylbenzamides, including NVR 3-778, effectively inhibit pgRNA-containing nucleocapsid formation²⁷. The backbone RMSF analysis demonstrated that ligand binding induced structural flexibility in the HBcAg protein, particularly in the loop and terminal regions. Among the ligands, HAP induced the highest backbone fluctuations, particularly in loop regions. These findings align with previous reports indicating that HBV capsid inhibitors can alter protein flexibility, affecting capsid stability and viral replication^25,28,29. Further analysis of per-atom ligand RMSF revealed that HAP exhibited the lowest average RMSF, indicating a stable binding within the HBcAg binding pocket. Conversely, Sulfamoylbenzamide displayed higher fluctuations, particularly at functional groups involved in hydrophobic interactions. Ciclopirox maintained relatively stable interactions but showed some flexibility at specific atomic positions. These variations in RMSF suggest that the ligands differ in their ability to anchor within the binding site, a factor that is critical for HBV inhibitor efficacy.

Binding interaction energies provided insight into the stabilizing forces governing CAM-protein interactions. HAP showed the most favorable interaction energy profile, with a strong short-range Coulombic interaction and significant van der Waals forces. However, its highly positive Coul-14 energy suggests electrostatic repulsion, which may affect its overall stability within the binding pocket. Also, CHARMM36 force field inaccuracies in modeling electrostatic repulsions may affect energy quantification^30–32. In contrast, Sulfamoylbenzamide exhibited moderate Coulombic stabilization but robust van der Waals interactions, comparable to HAP. The moderate interaction energies observed for Ciclopirox suggest that its binding is driven by hydrophobic interactions rather than electrostatic forces. Previous studies indicated that electrostatic interactions play a pivotal role in CAM binding efficacy. BAY 41-4109 forms electrostatic contacts with positively charged HBcAg residues³³mirroring our findings for Sulfamoylbenzamide. The presence of highly negative Coul-14 interaction energy suggests the formation of electrostatic interactions, potentially via salt bridges. Similar findings have been observed in previous studies where hydrophobic interactions were found to be essential in stabilizing HBV core protein inhibitors^34,35.

The salt bridge analysis further elucidated the nature of ligand-protein interactions. Ciclopirox exhibited the highest number of salt bridges, particularly with both positively and negatively charged residues. This finding implies strong electrostatic interactions that enhance its binding stability. The moderate salt bridge formation observed in the Sulfamoylbenzamide complex, particularly with positively charged residues, indicates selective electrostatic interactions. Interestingly, HAP formed no salt bridges with positively charged residues, relying predominantly on van der Waals interactions for stability. These findings are in agreement with previous studies highlighting the role of electrostatic interactions in HBV capsid inhibition^23,36. Also, cluster analysis confirmed that all ligand-bound HBcAg systems exhibited similar RMSD, suggesting that the overall structural integrity of the protein remained stable. However, the HBcAg-HAP complex exhibited slightly higher deviations, consistent with the high fluctuations observed in RMSF analysis. SASA analysis further confirmed differences in ligand-induced conformational changes. The HBcAg-Sulfamoylbenzamide complex exhibited the highest solvent accessibility, suggesting a less compact structure, while HAP binding resulted in a more shielded protein surface.

Secondary structure calculations via DSSP analysis revealed that HBcAg occupied with a predominantly α-helical conformation as of crystallographic structures. However, upon ligand binding, Ciclopirox exhibited a marked reduction in regular secondary structure, favoring more flexible conformations. This disruption of α-helical architecture especially at C-terminal dimerization domain and central α-helical structure suggests that CAMs binding induce significant changes in HBcAg structures, potentially altering HBcAg functionality and modulating HBV capsid formation. Such ligand-induced structural alterations have been previously reported as a mechanism by which HBV inhibitors misdirect capsid assembly, ultimately disrupting viral replication^25,26,37,38. Loss of the α-helical structure in the HBcAg disrupts its ability to form a stable icosahedral capsid, which is crucial for encapsulating the viral pgRNA and polymerase during viral replication³⁹. This structural instability also impacts HBcAg’s role in nuclear trafficking and antigen presentation, as shown in studies involving HBc mutants⁴⁰. Accordingly, a loss of this structure can lead to misassembly or destabilization of the capsid. Interestingly, helix-coil transitions were demonstrated in all holo systems. Ciclopirox and Sulfamoylbenzamide caused near-complete helix loss. This structural destabilization implies that ligand binding induces conformational shifts. Also, the increased solvent exposure in the Sulfamoylbenzamide complex (SASA: 101.03 ± 2.39 nm²) suggests partial unfolding or loosened packing, which may enhance proteolytic susceptibility or alter antigenicity, a phenomenon documented in HBV core protein mutants³⁹.

Further conformational state of HBcAg was evaluated by PCA. PCA along with cluster analyses revealed ligand-dependent alterations in HBcAg’s essential motions. Sulfamoylbenzamide induced the highest variance, reflecting pronounced flexibility, while Ciclopirox restricted motions, likely due to its stabilizing salt bridges. The emergence of distinct conformational substates in ligand-bound systems, particularly the extended “bridge” in the Sulfamoylbenzamide complex, suggests that ligand chemistry dictates accessible conformational ensembles. These findings align with theories positing that ligands act as “allosteric modulators” by reshaping energy landscapes⁴¹. Moreover, it was observed that HAP stabilize HBcAg in space where the protein was already met, but in a more density manner. It was also observed that the protein meet a place where the protein has not met before that was also bridge by transient transformation between other two structures. This finding suggests that HAP stabilize HBcAg with more rapid rearrangements between three stable location in space. These fluctuations probably indicate changed inter-dimer interactions, which might lead to improper capsid assembly.

The helix-coil transition of HBcAg’s induced by all ligands raises questions about functional consequences. While disordered states may impair viral assembly or secretion⁴²excessive flexibility (as seen with Sulfamoylbenzamide) could reduce target engagement longevity. This could also partly induced by the Y132A mutation within the crystallographic structures of HAP and Sulfamoylbenzamide. This mutation, however, was not presented in the wild-type HBcAg and the Ciclopirox-complexed system. Ciclopirox’s balanced interaction profile, combining electrostatic and hydrophobic forces, positions it as a promising scaffold for optimization. Future work should explore covalent stabilization of salt bridges or incorporation of hydrophobic moieties to enhance HAP’s efficacy. The findings also provide mechanistic insights into the interactions of HBV core protein inhibitors and their potential antiviral effects. The variations in RMSF, interaction energy, salt bridge formation, and secondary structure disruption suggest that each ligand influences HBcAg differently. The strong electrostatic interactions and stability observed with Ciclopirox suggest its potential as a potent inhibitor. HAP, despite favorable interaction energy, exhibits electrostatic repulsion, which may affect its binding efficiency. Sulfamoylbenzamide, while stable, interacts selectively with HBcAg, which could influence its effectiveness as a capsid assembly inhibitor.

CAMs represent a promising strategy for disrupting HBV replication by interfering with capsid assembly and stability. The results of our study reinforce the therapeutic potential of targeting capsid assembly through distinct molecular mechanisms. Our monomeric simulations imply critical implications for capsid assembly. In this regard, Increased flexibility in the central domain (residues 70–90) and C-terminus (residues 130–140) upon Sulfamoylbenzamide binding would impair (i) helix 3/4 interactions at the dimer interface and (ii) interdimer ‘spike’ formation (mediated by helix 4 tip residues W102/T109). Also, α→coil transitions in helices would destabilize the four-helix bundle critical for dimerization, leading to reduced capsid stability. HAP appear particularly effective in stabilizing incomplete capsids, making them attractive candidates for antiviral development⁴³. Meanwhile, Sulfamoylbenzamides exhibit properties, which could be leveraged to prevent the formation of functional viral particles. But the interaction of Sulfamoylbenzamides was not that stable and visually it was observed that it transiently disassociates HBcAg (see Supplementary Video 1 file).

The findings of the present study can be translated in drug development applications. Accordingly, optimizations of the compounds can led to enhanced interaction energy and in the case of Sulfamoylbenzamide, reduce dissociation. Also, development of HAP-derivatives could be promising due to the close and stable interaction with HBcAg. Furthermore, synergistic effects of different classes of CAMs together or in combination with nucleos(t)ide analogs could be promising and yield dual-mechanism inhibitors resistant to escape mutations. While our MD simulations provide atomistic insights into ligand-induced dynamics, key limitations warrant acknowledgment. The 100-ns timescale may not capture slow dissociation events beyond milliseconds, and the monomeric HBcAg system cannot replicate capsid assembly dynamics. Force field inaccuracies (CHARMM36) in modeling electrostatic repulsions (e.g., Coul-14) may affect energy quantification. Also, the Y132A mutation in crystallographic structures for Sulfamoylbenzamide and Heteroaryldihydropyrimidine limits extrapolation to wild-type capsids.

Conclusion

This study demonstrated the atomistic mechanisms by which distinct classes of CAMs, Ciclopirox, Sulfamoylbenzamide, and Heteroaryldihydropyrimidine, induce structural changes in HBcAg, thereby disrupting capsid assembly. The findings revealed that all three ligands provoke near-complete α-helix ablation, replacing ordered helices with bend- and coil-dominated conformations, thereby compromising residues critical for dimer-dimer interfaces like residues Y118–P135 and central loop at residues 70–80. Sulfamoylbenzamide exhibited transient dissociation, evidenced by progressive center-of-mass distance drift and elevated spatial variability, which facilitates sequential misincorporation during capsid assembly. Ciclopirox, despite moderate Coulombic stabilization, drives structural disorder via salt bridge networks, while Heteroaryldihydropyrimidine paradoxically combines strong van der Waals interactions with electrostatic repulsion, hindering inter-dimer salt bridges. These perturbations translate to capsid dysfunction, including helix 3/4 destabilization impairs icosahedral curvature, and C-terminal flexibility disrupts maturation-critical “clasps”. Key findings were including, the first quantification of ligand-specific secondary structure collapse (α→bend transitions), identification of Coul-14 repulsion as a design liability in HAPs, and mechanistic validation of Sulfamoylbenzamide’s “sliding” behavior. These findings rationalize optimization strategies to extend interaction time, to enhance salt bridge stability, and to mitigate electrostatic clashes. Future work should prioritize cryo-EM validation of aberrant capsid morphologies and SPR assays to correlate dissociation rate with experimental k_off. This study also provides a mechanistic blueprint for engineering dual-mechanism CAMs that resist escape mutations by synergistically targeting assembly and stability.

Methods

Data gathering of HBcAg and Ciclopirox

The crystallographic structure of HBcAg of HBV genotype D subtype adyw complexed with Ciclopirox was obtained from a protein data bank (PDB) with a PDB ID of 6J10 and 2.30 Å resolution¹⁷. Further crystallographic structures of HBcAg Y132A mutant complexed with HAP with the PDB ID of 5WRE and 1.95 Å resolution⁴⁴. Also, crystallographic structures of Sulfamoylbenzamide in complex with HBcAg Y132A (PDB ID of 5T2P) was obtained with 1.69 Å resolution⁴⁴. The structure of HBcAg was cleaned from water molecules and other non-standard fragments, as previously shown^45,46. Molecular docking was not conducted, and the crystallographic structures of HBcAg bound to the ligands were utilized for MDS.

Molecular dynamic simulation

MDSs were conducted on Hepatitis B core antigen (HBcAg) alone and on three more complex systems, including HBcAg with Heteroaryldihydroprymidine, Cyclopirox, and Sulfamoylbenzamide. The simulations were performed using the GROMACS simulation software, version 2021.3 ⁴⁷. Simulations were run on a system equipped with an Intel(R) Core(TM) i7-4510U CPU @ 2.00 GHz, with a total of 4 cores and four logical cores. SIMD instructions used were AVX2_256 with a memory model of 64-bit. GPU support was disabled for these simulations. The software was compiled using the GNU 10.2.0 C and C + + compilers with flags -mavx2 -mfma for AVX2 instruction sets and -O3 for optimization.

Simulations used the CHARMM36 force field for the protein and lipids, with the CHARMM-compatible TIP3P water model. The system was solvated in a cubic box (7.68 nm side length) with periodic boundary conditions. PME electrostatics (1.2 nm cutoff) and switch-modified vdW interactions (1.0–1.2 nm) were applied. Furthermore, ligands’ parameters were generated using the CHARMM General Force Field (CGenFF) via the ParamChem server⁴⁸. The ligand’s 3D structure were submitted in.mol2 format, and CGenFF returned a.str file containing atom types, charges, and bonded/non-bonded parameters. The.str output was then converted to GROMACS-compatible.itp and.top files using the cgenff_charmm2gmx.py script (https://github.com/Lemkul-Lab/cgenff_charmm2gmx), ensuring compatibility with CHARMM36 force fields during GROMACS simulation setup. The penalty score checked for ligand validation, and scores < 10 indicated reliable analogy.

Initial models of HBcAg complexed with each ligand were subjected to energy minimization to remove any close contacts or high-energy conformations. The system was then solvated with explicit water molecules. Any charges present in the system were neutralized using appropriate counter-ions. Accordingly, The HBcAg-Ciclopirox complex was neutralized by 4 Na⁺ ions. Additionally, The HBcAg-Sulfamoylbenzamide compound complex was neutralized with 4 NA⁺ ions. The simulations were carried out under periodic boundary conditions. Each system was equilibrated at 300 K and 1 bar using a canonical (NVT) ensemble followed by an isothermal-isobaric (NPT) ensemble. The simulations were run for 100 nanoseconds with a time step of 2 femtoseconds. The analysis of the MD simulation trajectories was performed with GROMACS built-in tools.

Trajectory analysis

Molecular dynamics (MD) simulation trajectories were subjected to an extensive range of analytical procedures conducted using GROMACS utilities to decipher the system’s key structural and dynamic aspects. Structural stability and conformational alterations were evaluated through root mean square deviation (RMSD) calculations (gmx rms). Protein flexibility at the residue level was investigated via root mean square fluctuation (RMSF) computations (gmx rmsf). System compactness was gauged by determining the radius of gyration (Rg) using gmx gyrate.

The strength of receptor-ligand interactions over the simulation period was elucidated by quantifying interaction energies via gmx energy. Hydrogen bonding dynamics within the molecular system were comprehended using the gmx hbond utility. Evolutionary changes in secondary structures were scrutinized employing gmx do_dssp.

A principal component analysis (PCA) was performed using the gmx covar and gmx anaeig utilities to identify predominant motion patterns in the system. The solvent-accessible surface area (SASA) was computed to continuously monitor the protein’s exposure to the solvent during the simulation. A cluster analysis was conducted to identify the conformational states most frequently adopted by the system. Additionally, radial distribution functions (RDFs) were calculated to characterize the spatial relationships between various entities within the system. All analyses were performed as reported before^49–53.

Statistical analysis

All quantitative analyses were performed using Microsoft Excel (2019) with the XLSTAT add‑in (Addinsoft, 2025, a comprehensive statistical and data analysis solution) as described before^54,55. Descriptive statistics were computed for each system, reporting the number of frames (N), the mean, mean standard error (Mean SE), and standard deviation (SD). Data were assessed for normality of residuals and homogeneity of variances using XLSTAT’s built‑in diagnostic tools prior to inferential analysis. For each ANOVA, the null hypothesis (H₀: all group means equal) was tested against the alternative (H₁: at least one group mean differs), with α = 0.05 and 95% confidence interval (CI). The p‑values < 0.05 across all analyses led to rejection of H₀. Post‑hoc comparisons with Tukey’s HSD were conducted in XLSTAT when needed to identify specific pairwise differences.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Alireza Mohebbi conceptualized the study, designed the research methodology, and supervised the project. Molecular dynamics simulations were performed by Fatemeh Sana Askari and Alireza Mohebbi. Alireza Mohebbi and Fatemeh Sana Askari contributed to data analysis and interpretation. All authors contributed in manuscript writing and figure preparation. All authors reviewed and approved the final manuscript. Both authors reviewed and approved the final manuscript.

Data availability

All data generated or analysed during this study are included in this published article and its Supplementary Information file, Supplementary Video 1.

Declarations

Competing interests

The authors declare no competing interests.