Evaluation of Melongosides as Potential Inhibitors of NS2B-NS3 Activator-Protease of Dengue Virus (Serotype 2) by Using Molecular Docking and Dynamics Simulation Approach

Dengue is a Flavivirus infection transmitted through mosquitoes of the Aedes genus, which is known to occur in over 100 countries of the world. Dengue has no available drugs for treatment; CYD-TDV is the only vaccine thus far approved for use by a few countries in the world. In the absence of drugs and a widely approved vaccine, attention has been focused on plant-derived compounds to the discovery of a potential therapeutic for DENV. The present study aimed to determine, in silico, the binding energies of the steroidal saponins, melongosides, to NS2B-NS3 activator protease of DENV-2, which plays an essential role in the viral replication. The blind molecular docking studies carried out gave binding energies (ΔG = −kcal/mol) of melongosides B, F, G, H, N, O, and P as 7.7, 8.2, 7.6, 7.8, 8.3, 8.0, and 8.0, respectively. All the melongosides interacted with the NS3 protease part of NS2B-NS3. Melongosides B, F, and N showed interactions with His51, while melongoside G interacted with Asp75 of NS3, to be noted, these are important amino acid residues in the catalytic site of the NS3 protease. However, the 200 ns molecular dynamic simulation experiment indicates significant stability of the protein-ligand interactions with the RMSD values of 2.5 Å, thus suggesting a better docking position and no disruption of the protein-ligand structure. Taken together, melongosides need further attention for more scientific studies as a DENV inhibitory agent, which if proven, in vivo and in clinical trials, can be a useful therapeutic agent against at least DENV-2.


Introduction
Dengue viruses belong to the family Flaviviridae and consist of four major serotypes, namely, dengue virus 1-4 (DENV 1-4), which are responsible for dengue fever [1]. A fifth serotype (DENV 5) has been reported in 2015, which follows the sylvatic cycle rather than the human cycle followed by the other DENV serotypes [2]. Dengue is a mosquito-borne viral infection which spreads when the virus is transmitted from humans to humans by mosquitoes belonging to the Aedes genera, like Aedes aegypti and Aedes albopictus [3].
Dengue fever in humans is characterized by fever, myalgia (muscle pain), arthralgia (joint pain), abdominal pain, rash, and thrombocytopenia (low counts of platelets). e more severe form of dengue causes dengue hemorrhagic fever, which can result in death [3]. As per the World Health Organization (WHO) fact sheet of 19 May 2021, the global incidence of dengue has grown in recent decades; about half the world's population is now at risk from dengue, and there are an estimated 100-400 million infected cases per year. Furthermore, there is no drug against the virus, and a single vaccine Dengvaxia ® (CYD-TDV) developed by Sanofi Pasteur was licensed in December 2015 and has now been approved by the regulatory authorities in only ∼20 countries [4]. e dengue virus carries a positive single strand RNA in its genome. e viral genome is encoded by three structural and seven nonstructural proteins (NSPs) [5]. Of the NSPs, NS2B/NS3 protease is considered to be an excellent therapeutic target for inhibiting the dengue virus because a serine protease domain present in the NS3 plays an integral role in the viral replication [6][7][8]. As a prime therapeutic target, the NS2B/NS3 has been widely studied for inhibition with natural compounds like panduratin, synthetic compounds like benzimidazole, and in silico methods, utilizing mostly the molecular docking programs [9,10]. In silico studies, in turn, have screened various phytochemicals for their binding affinity to various nonstructural proteins of DENV [11].
Plant-derived saponins, such as triterpene saponins [12,13] and glycosteroidal saponins [14,15], have been reported to exert the antiviral activity. Since plant-derived saponins have been reported for antiviral activities against both RNA [16] and DNA viruses [17], 3 steroidal saponins that were isolated from the roots of Solanum sisymbriifolium Lam. have shown significant antiviral effects against DENV-2 with EC 50 values ranging from 24.9 to 35.1 μg/mL [18]; the present study aims to evaluate the binding energies of the steroidal saponins, melongosides, to NS2B-NS3 activator protease of DENV-2 via in-silico approaches by using the molecular docking and molecular simulations techniques.

Receptors and Ligand Selection.
For this study, several bioactive phytochemicals with potential medicinal effects were considered, named as melongoside B, melongoside F, melongoside G, melongoside H, melongoside N, melongoside O, and melongoside P. e NS2B-NS3 protease of Dengue 2 (DENV-2) [19], PDB ID-2FOM, has been selected as the targeted receptor and the established inhibitor of DENV-2, and quercetin was considered as the standard reference for the study [20].

Protein Preparation and Generation of Receptor Grids.
e NS2B-NS3 protease of Dengue 2 (DENV-2), PDB ID-2FOM was used as the targeted receptor protein for this study.
e RCSB PDB Protein Data Bank (https://www. rcsb.org/) has been accessed to find the structure of the protein in the PDB and the Protein Preparation Wizard of Maestro Desmond version 12.5 (Schrödinger Release 2020-3 Schrödinger, LLC, New York, NY, 2020) was used to prepare the protein for molecular docking application. In order to achieve the desired results, the following criteria were used: assign bond orders, use the CCD database, add hydrogens, create zero-order bonds to metals, create disulfide bonds, fill missing side chains and loops by using prime, fixed cap termini, and delete waters beyond 5 from heat groups; and generate heat states of pH 7.0 ± 2.0 using Epik. e results were evaluated using the following criteria: assign bond orders, use the CCD database, add hydrogens, and generate zero. Using the refine tab and the OPLS3e force field, the H-bond was assigned to PROPKA pH level 7.0, as well as the devaluation, was constrained to RMAD 0.30 using the converging heavy atom. After that, the receptor grid was generated by targeting the natural ligand of the protein.

Ligand Compounds' Preparation.
All the three-dimensional (3D) structures of the phytochemicals and the standard were obtained in SDF format from the open-source PubChem database (https://pubchem.ncbi.nlm.nih.gov/) using the structure data format (SDF). All ligand structures were constructed for molecular docking using the LigPrep.
To make things easier, we used the OPLS3e force field to make things smaller and the Epik Ionizer to make them smaller. e maximum number of conformers per structure was 32 and the RMSD for each structure was 1.0Å.

Molecular Docking and Visualization.
e extra precision (XP) type of molecular docking was conducted by using the Maestro (Schrödinger Release 2021-2: Maestro, Schrödinger, LLC, New York, NY, 2020-3) and the PDB structure of all protein-ligands complexes was retrieved for post-docking analysis. Ligplot + version 2.2 was used to analyze the noncovalent (polar and hydrophobic) interactions between the protein-ligand complexes.
is visualization tool performed admirably because of the Java interface (Java SE Runtime Environment 8u271), which permitted only the combined PDB files created with the PYMOL tool [21]. Moreover, for the validation of complexes' structural bonding compactness, polar and nonpolar interaction bonds, Discovery Studio Visualizer (https:// media.accelrys.com/downloads/visualizer/45/DS45Client. exe) 64 bit was run.

Molecular Dynamic Simulation.
e 200 ns MD simulations analyzed the protein-ligand complex structures to determine the binding consistency of the selected three candidate ligand compounds to the targeted protein NS2B-NS3 protease of Dengue 2 (DENV-2), PDB ID-2FOM [22]. e molecular dynamic simulation of the protein-ligand complex structures was carried out by using the "Desmond v3.6 Program" in Schrödinger (https://www.schrodinger. com/) (paid version) in a Linux environment to analyze the thermodynamic stability of receptor-ligand complexes [23]. A predetermined TIP3P (transferable intermolecular potential with 3 points) water technique was developed for this framework to maintain a specified volume with an orthorhombic periodic bounding box shape separated by a distance of 10Å. For electricity neutralization in the framework, appropriate ions such as 0+ and 0 and 15 M salt have been selected and randomly distributed throughout the solvent system. Following the construction of the solvency protein systems with a ligand complex, the system framework was lowered and relaxed by employing the standard procedure performed by using force field constants OPLS3e within the Desmond package [22]. All NPT assemblies that made use of the temperature's combination of the Nose--Hoover thermostat and the isotropic approach were maintained at 300 K and at one-atmosphere pressure (1,01325 bar) and were accompanied by 50 PS capturing intervals with an efficiency of 1.2 kcal/mol. e simulation's entire event was evaluated using the Simulations Interaction Diagram (SID) of the Desmond modules included in the Schrödinger suite that was performed to evaluate the MD simulation's accuracy. e stability of the protein-ligand complex system was determined using the root-meansquare deviation (RMSD), protein-ligand contacts (P-L), intramolecular hydrogen bonds, solvent accessible surface area (SASA) value, radius of gyration (Rg) value, MolSA, and the polar surface area (PSA) values.

Interpretation of Molecular Docking.
A molecular docking technique using the Maestro package platform was used to identify phytochemicals that interact with the NS2B-NS3 protease of the Dengue 2 (DENV-2) protein.
e Maestro application produced the greatest possible docking score for macromolecules and ligands. Quercetin (control) was used as a control ligand in this investigation and it had a binding affinity of −8.2 Kcal/mol. Table 1 shows that the ligand melongoside P had the best fitting score of −9.5 Kcal/ mol, whereas other possible bioactive phytochemicals including melongoside O and melongoside G had the best docking values of −9.01 and −8.11 Kcal/mol, respectively.

Visualization of Post-Docking Protein-Ligand
Interactions.
e BIOVIA Discovery Studio Visualizer and Ligplot + Version 2.2 tools have been used to investigate the interactions between the four ligands that were specified and the target protein in this study. Ligplot + version 2.2 was performed to calculate the interactions (which were predominantly hydrophobic and noncovalent) for all the docked complexes as shown in (Table 1) and (Figures 1-4).

Molecular Dynamic Simulation.
Analysis of the biomolecular interactions and the analysis of the interface between the arrangement and activity of proteins can aid in the development of new drugs, and the performance data from the dynamic trajectory analysis named the molecular dynamic simulation (MDS) [24,25] are used to analyze the stability and the intermolecular interactions of a proteinligand complex in real-time. When a sophisticated system is subjected to an artificial environment, this technique can also be used to determine its conformational change. To better understand the conformational changes of the protein in complex, a 200 ns MD simulation of the protein in connection with the specific ligand was performed in this study. e terminal snapshots from the MDS trajectories were used to examine the intermolecular behavior initially.

RMSD Analysis.
With a range of 1-3Å, the average change in the root means square deviation (RMSD) of the protein-ligand interaction is acceptable. If the RMSD number is larger than 1-3, it means the protein structure has changed significantly. An MD simulation (200 ns) was run and the associated RMSD value was determined to assess the conformational change of the desired protein in the complex with the four ligand compounds, namely, CID-5280343, CID-131750951, CID-131750948, and CID-131752997. e average root means square deviation (RMSD) for the ligand compounds CID-131750951, CID-131750948, and CID-131752997 was 1.5-3. Another ligand molecule, CID-5280343, was shown to have a potent RMSD value in the 1.5-3.5 range. e compound's value change shows very little fluctuation, which is within the permitted range, indicating that the protein-ligand complex structure represented in ( Figure 5) is conformationally stable.

RMSF Analysis.
When the specific ligand chemicals interact with certain residues, the root mean square fluctuation (RMSF) can help characterize and determine the local alterations that occur within the protein chain. As a result, the Journal of Tropical Medicine RMSF values of the compounds CID-5280343, CID-131750951, CID-131750948, and CID-131752997 in complex with the Dengue 2 (DENV-2) NS2B-NS3 protease were calculated in order to investigate the change in protein structural flexibility caused by the attachment of specific ligand compounds to a specific residual position as shown in (Figure 6). e most rigid secondary structural components, such as alpha-helices and beta-strands, were shown to have a minimum observation rate of 5 to 290 amino acid residues. e majority of the variation is seen at the beginning and at the end of the protein due to the presence of the N-and C-terminal domains. As a result, the displacement of an individual atom in the simulated environment has a low fluctuation probability for the four ligand complexes investigated.

e Radius of Gyration (Rg) Analysis.
e arrangement of its atoms across its axis defines the radius of gyration (Rg) of a protein-ligand interaction system. e calculation of Rg is one of the most important indicators to look for when anticipating a macromolecule's structural functioning since it shows variations in complex compactness over time.
As shown in (Figure 7), the stability of CID-5280343, CID-131750951, CID-131750948, and CID-131752997 in interaction with the target protein was studied in terms of Rg throughout a 200 ns simulation duration. e average Rg values for the compounds' CID-5280343, CID-131750951, CID-131750948, and CID-131752997 were 3.9, 8.1, 8, and 6.75, respectively, indicating that the protein's binding site does not undergo major structural changes when the ligand compounds are bound.

Analysis of SASA, MolSA, and PSA.
e amount of solvent-accessible surface area (SASA) regulates the arrangement and activities of biological macromolecules. In most cases, amino acid residues on a protein's surface serve as active sites and/or interact with other molecules and ligands, allowing researchers to better understand a molecule's solvent-like behavior (hydrophilic or hydrophobic) and protein-ligand interaction components.
e SASA values for the protein complexes with CID-5280343, CID-131750951, CID-131750948, and CID-131752997 were calculated and are presented in (Figure 8).
e SASA values for the three compounds CID-131750951, CID-131750948, and CID-131752997 were 500 to 1200 A 2 on average, and 100 A 2 for CID-5280343, showing that in the complex systems, an amino acid residue was exposed to a high amount of the selected ligand molecules. e molecular surface area (MolSA) is the same as the van der Waals surface area calculated with a probe radius of 1.4. All of the ligand complexes CID-5280343, CID-131750951, CID-131750948, and CID-131752997 had the typical van der Waals surface area in our in-silico investigation ( Figure 9). Furthermore, only oxygen and nitrogen atoms contribute to a molecule's polar surface area (PSA). With the targeted protein, all the ligand molecules CID-6474309, CID-5280805, and CID-442658 had a high PSA value ( Figure 10).

Analysis of Intramolecular Bonds.
e complex structure of a protein with the selected ligands and their intermolecular interactions was investigated for a 200 ns simulation time using the simulation interactions diagram (SID). e hydrogen bond, the ionic bond, the water bridge bond, and the noncovalent interactions bond (hydrophobic bond) defines the interaction between the protein and the designated ligands, which were evaluated and displayed for CID-5280343, CID-131750951, CID-131750948, and CID-       VAL_72  LYS_73  LYS_74  ASP_74  LEU_76  TRP_83  LEU_85  GLU_86  GLY_87  GLU_88  TRP_89  THR_118  ASN_119  THR_120  THR_122  ILE_123  VAL_147  LEU_149  GLY_151  ASN_152  GLY_153  VAL_154  SER_163  ALA_164  ILE_165  ALA_166

Discussion
In recent years, the in silico drug design method has gained widespread acceptance due to its ability to accelerate quality drug development by evaluating the results of pharmacophore screening, molecular docking, analysis of postdocking interaction, molecular dynamic simulation (MDS), and prediction of noble drug compounds against a wide range of diseases in a computer-simulated environment [23]. ree out of the four phytochemicals utilized in this study were functionally selected as the active compounds, while another one was employed as a control.
Molecular docking is a technique for determining how two or more molecules will interact in the presence of the highest compositional confirmation and the lowest binding affinity conceivable [26,27]. Drug candidates that delivered the most significant and stable score were selected using the Maestro application (Schrödinger Release 2021-2: Maestro, Schrödinger, LLC, New York, NY, 2020-3.), which utilizes molecular docking to assign a score. e molecular docking study for the selected three natural bioactive compounds and the control drug with the NS2B-NS3 protease of Dengue 2 (DENV-2), PDB ID-2FOM reported the docking affinity of the control drug quercetin (Pubchem CID-5280343) was −8.2 Kcal/mol. Among the selected natural bioactive compounds, melongoside P (Pubchem CID-131750951) possessed the best binding affinity of −9.5, in which melongoside O (Pubchem CID-131750948) was −9.01, melongoside G (Pubchem CID-131752997) was −8.11, and all the docking affinity results with the chemical names of the ligands are depicted in Table 1. In the following step, Ligplot + (Version 2.2), an excellent investigative tool that normally operates through the Java interface, was employed to investigate the 2D protein-ligand interaction scheme. Furthermore, the Discovery Studio Visualizer tool v19.1.0.18287 (BIOVIA), an effective visualizer tool for drug discovery, was used to represent the postdocking receptorligands interactions with their animated structures (Figures 1-4).
Our study was conducted with the Schrödinger package software (Desmond Application) to run 200 ns molecular dynamic simulation (MDS) with the selected physiological and physicochemical parameters. is simulation trajectory of the simulation tool has also been used to perfectly analyze the root mean square deviation (RMSD), root means square fluctuation (RMSF), the radius of gyration (Rg), hydrogen bond number, and solvent-accessible surface area (SASA) [28]. e root means square deviation (RMSD) of the selected NS2B-NS3 protease of Dengue 2 (DENV-2) protein's alpha carbon and backbone was used to evaluate the protein structure's reliability and identify conformational changes; the lower value indicates the most stable compounds. e RMSD values of less than 1.5Å were typically indicative of greater consistency in docking since RMSD values over 1.5Å typically indicate the average binding positions. In our study, the RMSD values of protein-ligand interactions were within an appropriate range, namely, the average mean values of 2.5 (the lowest value for the selected ligand compounds was approximately 0.8 and maximum value was 3), suggesting a better docking position and no disruption of the proteinligand structure ( Figure 5). Individuals can quantify the average protein fluctuations from such a reference location using the RMSF and the RMSF plots demonstrate how the average protein fluctuations represent fluctuations at the residue level. An illustration of the RMSF of the c-alpha atoms can be found in (Figure 6).
is conformational stability was determined by counting the complete number of intermolecular bonds generated among the macromolecules and their ligands, and the highest number of intramolecular bonds were determined as melongoside P and melongoside O, which are more stable in conformation than that of the control compound ( Figure 11). e protein-ligands solvent-accessible surface area (SASA) was also determined using the simulation trajectories in order to determine the dimensional changes of the drug-like molecules along the simulation trajectory [29]. One of the greatest SASA values owes to the unstable structure, which contained hydrophobic amino acid residues close to the water molecule [30]. According to the SASA result from the MDS trajectory, melongoside P exhibited the highest SASA values, in which (−)-melongoside O, and melongoside G had the higher SASA values than the control drug, quercetin ( Figure 8). In the graph of MolSA and PSA validation, all the ligands' compounds possessed more potential value than the control drug (Figures 9 and  10).
Additionally, Rg is a measurement of the distance between the portion of the center mass and the end portion of the protein, and it possesses how far that distance is. As a result, this metric quantifies the protein molecule's compactness and provides further information about the protein's folding properties [31]. Furthermore, a larger Rg value denotes slack packing, whereas a lower Rg value denotes compact packing [32].
e Rg values are summarized in Figure 7 and show that all the ligands with the protein showed standard compactness compared to the control drug. However, there are insufficient clinical research findings based on these ligand compounds to suggest it as a therapeutic option for Dengue virus treatment. Consequently, these biologically active phytocompounds can be considered a potential option for SARS-CoV-2 medication when they will have been validated for anti-Dengue virus activity in in-vitro and in-vivo research models.

Concluding Remarks
A mosquito-borne viral disease, dengue, has grown dramatically in recent years and become a global burden that affects most Asian and Latin American countries. ere are no available specific drugs for dengue disease as well. Our present experiment that was evaluated by molecular docking and dynamics simulations techniques suggests further attention towards steroidal saponins (melongosides) as a DENV inhibitory agent, which if proven In Vivo and in clinical trials, can be a useful therapeutic target against at least DENV-2.