Dronabinol as an answer to flavivirus infections: an in-silico investigation

Abstract Flavivirus infections are common in several parts of the world. Two major types of flaviviruses are dengue and zika viruses. Both these two viral infections have caused many fatalities around the world. There is an absence of a vaccine and an effective medication against these viruses. In this study, we analyzed the ability of dronabinol to act as a potential cure against these viral infections. We performed the docking of dronabinol with several viral proteins followed by molecular dynamics simulation, MM/PBSA and PCA analysis. We checked the ability of the polyphenol dronabinol to interfere with the binding of viral helicases to their cellular targets. We performed 2 D-QSAR studies, drug likeliness, ADMET and target prediction studies. From our study, we observed that dronabinol had the best docking ability against the helicase proteins of dengue and zika. Molecular dynamics simulation and MM/PBSA investigation confirmed the stability of the binding while PCA investigation showed a lowering of molecular motions in response to dronabinol docking to the helicases. Dronabinol interfered in the binding of the helicases to RNA. 2 D QSAR studies revealed a low IC50 value for dronabinol. Dronabinol showed favorable drug-likeness, ADMET properties and target prediction results. Thus we propose dronabinol be further investigated in-vitro as a cure against dengue and zika virus infections. Communicated by Ramaswamy H. Sarma


Introduction
The family flaviviruses are a group of viruses that are the causative agents of several epidemics, along with a high number of fatalities.Flavivirus have RNA as their genetic material.These viruses can also infect animals important in the farming and dairy industry.In all of these infections, the viruses require an arthropod intermediary.Two of the epidemics causing flaviviruses are dengue and zika virus (Pierson & Diamond, 2020).South-East Asia, in general, and India, in particular, had a high caseload of dengue.India alone contributes to over 30% of the global cases of dengue (Murhekar et al., 2019).India has also reported zika virus outbreaks on multiple occasions (Gupta et al., 2020).
The life cycle of flaviviruses involves the coordinated action of several viral proteins.The primary contact necessary for the entry of flaviviruses into the human cell is the viral envelope (E) protein.The flaviviral genome encodes for various nonstructural proteins which are needed at various stages of the multiplication of the viral genome (Pierson & Diamond, 2020).The flaviviral nonstructural protein NS3 contains protease activity at its N-terminal end and helicase activity at its C-terminal end.The protease protein cleaves the viral polyprotein into individual proteins leading to their activation.The activated helicase domain of NS3 protein plays its part in viral replication by unwinding the doublestranded RNA (dsRNA) to produce the single RNA strands that serve as the genetic material of flavivirus particles (Davidson et al., 2018).The nonstructural protein 5 (NS5) consists of the methyltransferase activity at the N-terminal end and RNA dependent RNA polymerase (RdRP) activity at the C-terminal end.The methyltransferase protein carries out the 5 0 capping of the unwinded RNA molecule.The actions of methyltransferase allows for the stable interaction between the RNA and the RdRP protein leading to continued replication (Rusanov et al., 2018).
All the proteins discussed above have been prime targets for building drugs against dengue and zika viruses.The current drugs are aimed at ameliorating the symptoms arising out the infections.Vaccines have not had success against dengue and zika viruses (Boldescu et al., 2017;Carro & Cherry, 2020).Thus considering the huge population that these viruses can severely impact and the lack of effective cures, it becomes imperative that we look at different sources for finding a cure against these viral infections.In this study, we investigate the prospect of dronabinol to be considered; as a drug candidate against dengue and zika virus infections.Dronabinol belongs to the cannabinoid class of compounds, and it is a clinically prescribed compound used for issues ranging from neuropathic pain and cancer (Schimrigk et al., 2017, Bajtel et al., 2022).From our results, we can infer that dronabinol could; be potentially repurposed; as a drug against dengue and zika virus infections, provided it shows efficacy in in-vivo testings.

Preparation of ligand for docking procedure
The three dimensional information of the structure of the selected ligand, dronabinol was obtained from the PubChem database in the SDF format.The structure was converted to PDB file format using the Open Babel software (O'Boyle et al., 2011).PDB was used as input for docking in the software Autodock 4.0 used in the current study.Ligand was prepared using the default parameters and polar hydrogens were added followed by computation of Gasteiger charges.The number of torsions was set at 5. The overall parameters for the ligand are saved in an extended PDB format termed PDBQT, which saves atomic partial charges and atom types for the ligand (Forli et al., 2016).

Preparation of protein for docking
The 3 D structures of the proteins of interest were obtained from the Protein Data Bank (www.https//rscb.org/).The PDB IDs used for the study were 5XC7 (dengue helicase) and 5JMT (zika helicase), 1L9K (dengue methyl transferase) and 5KQR (zika methyl transferase), 5F3Z (dengue RdRP) and 5UO4 (zika RdRP), 2M9Q (dengue protease) and 5Z0B (zika protease).All these PDB IDs focused on the active site of the proteins.To prepare the PDBQT format of these proteins, additional water molecules and heteroatoms in the protein structures were removed leaving behind amino acid coordinates only.Polar hydrogens were added and the Kollman charges were computed.

Grid selection
AutoDock requires generation of a grid map for pre-calculation of atomic affinities to be done by the AutoGrid procedure.The grid map consisted of a three dimensional lattice so as to accommodate all the amino acids of interest which were chosen to carry out the molecular docking.The grid maps were created for each of the proteins by selecting amino acid residues that were present in the active site of the proteins and the grid dimensions were adjusted to incorporate the selected amino acids.The grid spacing was kept at 0.375 Å.The grid dimensions were saved as a grid parameter file (.gpf) (Forli et al., 2016).

Molecular docking using Autodock
The AutoDock function of MGL tools was used to carry out the molecular docking of dronabinol with the proteins of interest.We performed the AutoGrid calculation to generate the grid maps evaluating the atomic affinity potentials for each atom type in the ligand molecule involved in docking followed by Autodock calculations which provide the energetics of a particular ligand configuration using values from the grid.The Lamarckian Genetic Algorithm (LGA) method was used as the search algorithm for the docking procedure.AutoDock was run for ten times to get several docked conformations.The current version of Autodock uses an empirical free energy scoring function to evaluate the conformations during docking.The scoring function along with the search algorithm is used by the Autodock to compute binding energy of the ligand and the protein complex.The Discovery studio software was used for the analysis of the protein-ligand interactions (Forli et al., 2016).

Molecular dynamics (MD) simulation
The dronabinol-helicase complex of the dengue and zika complexes were subjected to MD simulation studies using the Groningen Machine for Chemicals Simulations (GROMACS) 2021.3 package.The GROMACS program pdb2gmx was used to add hydrogens to the system.All the MD simulations were performed using CHARMM36 force field and solvated explicitly by the three-point transferable intermolecular potential water (TIP3P) model in a cubic box with 1 nm distance maintained between solute and the edge of the box.The ligand topology parameters were generated using CHARMM General Force Field (CGenFF) program.The whole simulation system was solvated in 23238 water molecules and neutralized by the addition of one Cl -counter ion to ideally maintain the physiological conditions.Energy minimization was completed by the steepest descent approach and subsequent equilibration within the NVT ensemble for 1 ns, followed by NPT ensembles at constant pressure (1 Bar) and temperature (298 K) using Berendsen coupling for 5 ns.The equilibrated system was then prepared for a 100 ns MD simulation at a time-step of 2 fs.Pressure of the protein-ligand system was maintained at 1 bar by isotropic pressure coupling to a Berendsen with a time constant s ¼ 2.0ps.All the electrostatic calculations during the course of simulation were performed using Particle-mesh Ewald (PME) summation.Constraints were imposed via the LINCS algorithm.
The generated trajectories were analyzed for structural deviations in terms of root mean square deviation (RMSD) and radius of gyration (Rg) using inbuilt GROMACS utilities gmxrmsd and gmxgyrate respectively.Solvent accessible surface area (SASA) of protein both in the presence and absence of ligand was evaluated using the gmxsasa program.gmxhbond program of Gromacs utilities was used to calculate the number of bonds between ligand and protein in the docked complexes during the course of simulation.Important residues involved in interaction were evaluated with the help of mindist program of the GROMACS utility.All conformations were visualized using PYMOL (Abraham et al., 2015).

MM/PBSA analysis
The g_mmpbsa tool of GROMACS was used to calculate molecular mechanical (MM) energy, polar salvation energy and apolar solvation energy, all of which make up the binding energy.g_mmpbsa was used to extract conformations of the protein-ligand complexes to calculate the binding profile of helicases from the last 50 ns with 1 ns interval.Polar solvation energy is calculated based on Assisted Poison Boltzmann Solver (APBS) program while Solvent Accessible Surface Area (SASA) model was used for apolar solvation energy calculation.MM energy comprises mainly of energy contributions by electrostatic (E elec ) and van der Waals (E vdW ) interactions.Additionally, the binding free energy calculation is calculated by equations (i) and (ii): where G can be further expanded into following components: In addition to total binding energy, residue-wise free energy was evaluated using the g_mmpbsa coupled python program, MmPbSaDecomp.pyscript which relies on energy contributions evaluated from g_mmpbsa (Kumari et al., 2014).For this purpose we choose only the residues around the active site.Only binding free energy contributions greater than 0.1 kJ/mol and less than À 0.1 kJ/mol were recorded as significant.

Principal component analysis
Principal component analysis (PCA) or essential dynamics provides conformational insights during a particular timeframe of the simulation trajectory (Kitao et al, 1999).The dynamics are traced by a covariance matrix from which eigen-values are resolved, corresponding to principal components (PC) of the system.These components define the motion of the trajectory.In the current work, we have performed PCA to understand relative motions in the helicases affected in the presence of ligand, dronabinol.The PCA calculations were performed by in-built gromacs programs, gmx covar and gmx anaeig in respective orders.

Helicase-RNA docking study
We performed helicase-RNA interaction studies for dengue and zika helicases in the absence and presence of the polyphenol dronabinol based on a previous study by Basu et al. (2020) with minor modifications.We used the HDOCK webserver for carrying out the study (Huang & Zou, 2008;Huang & Zou, 2014;Yan et al., 2017a;Yan et al., 2017b;Yan et al., 2020a;Yan et al., 2020b).We analyzed the net binding energy obtained after the docking study.

2 D-Quantitative-structure activity relationship (QSAR) study for dronabinol
For this QSAR study, different datasets of inhibitors of the flaviviral NS3 helicase were retrieved from the ChemBL database (Rassias et al., 2019;Millies et al., 2019).The pIC 50 (-LogIC 50 ) value of each inhibitor was calculated, followed by the conversion of their 2 D structure to 3 D structure and computation of 2 D descriptors using the MarvinSketch software version 21.11 and PaDEL-Descriptor software version 2.20, respectively (Yap, 2011).The data pretreatment, dataset division, model building, model validation and prediction steps of QSAR studies were performed using the Drug Theoretics and Cheminformatics (DTC)-QSAR software version 1.0.5.Datasets were divided into training and test datasets by applying Kennard-Stone's algorithm (Kennerd and Stone, 1969).To select the variables, the Genetic Algorithm method was applied in this study.The QSAR models, generated by the multiple linear regression (MLR) analysis, were validated and used to predict the pIC 50 value of inhibitors as well as Dronabinol.In the case of the internal validation, statistical metrics such as determination coefficient (R 2 ), leave-one-out cross-validated correlation coefficient (Q 2 (LOO) ), average Rm 2 (LOO) , D Rm 2 (LOO) and Mean Absolute Error (MAE) were determined.In the case of the external validation, statistical metrics such as Q 2 F1 , Q 2 F2 , average Rm 2 (Test) , DRm 2 (Test) and MAE were determined to assure the significance of the developed mode (Tropsha, 2010).Moreover, a Y-Randomization test was performed to ensure the robustness of Model 1.

Drug likeliness and ADMET studies
We checked the drug-likeness properties of dronabinol and checked its pharmacological properties like absorption, metabolism and toxicity using the SwissADME and admetSAR web tools (Diana et al., 2017;Yang et al., 2019;Cheng et al., 2012).

Target prediction study
To check for the probability of dronabinol to bind to macromolecular targets in humans we carried out a target prediction study using the SwissTarget webserver (Diana et al., 2019).

Docking of dronabinol with target proteins
Docking study of dronabinol with the target proteins showed that dronabinol bound to the proteins favorably but with different binding energies.The binding energies and the inhibition constants obtained from the docking studies of dronabinol are mentioned in Table 1.We observed that dronabinol showed the best binding with the helicase proteins of both dengue (-8.43 kcal/mol) and zika (-8.91 kcal/mol) virus (Figures 1A and 2A).The binding energy of dronabinol against the dengue and zika methyl transferase enzymes stood at À 6.63 kcal/mol and À 6.54 kcal/mol respectively (Table 1, Figures 1B and 2B).The RNA-dependent RNA polymerase for dengue and zika virus showed binding affinities of À 5.93 and À 5.85 kcal/mol, respectively (Table 1, Figures 1C and 2C).The protease protein of zika showed comparatively better binding of À 7.68 kcal/mol as compared to the binding energy of À 6.77 kcal/mol obtained with the dengue protease (Table 1, Figures 2D and 1D).A comparative docking study with of the binding of the undertrial anti-flaviviral drug Celgosivir (Lim et al, 2013;Garc� ıa et al., 2017) with the helicases yielded in lesser favorable binding energies of À 5.41 kcal/mol (against dengue helicase) and À 5.60 kcal/mol (against zika helicase).

Molecular simulation studies of protein and dronabinol
As we obtained the best docking values with the complexes of dronabinol-dengue helicase protein and dronabinol-zika helicase protein we checked for the stability of the complexes in an in-vivo and in-vitro setting by using molecular dynamics simulation (Nasution et al., 2018).Figure 3 depicts decreasing RMSD for both the helicases when in complex with dronabinol.Both proteins have attained an average RMSD of 0.3 to 0.4 nm.A gradual drop in the RMSD to approximately 0.2 to 0.3 nm is observed for the dronabinol-helicase complexes (Figures 3A and 3B).The radius of gyration (Rg) value for the dengue virus NS3 Helicase shows minor fluctuations between 2.25 and 2.35, which does not account for significant changes in the Rg in the protein in different conditions, i.e. absence or presence of ligand (Figure 4A).In the case of zika helicase, there is a wide range of differences in compactness.There is an increase of gyration from 1.5 nm to more than 2 nm for the protein when in complex with the ligand, clearly visible in Figure 4B.This is indicative of the advent of higher ligand-protein interactions in the course of 100 ns.The analysis of the RMSF values of the dengue helicase protein revealed that most of the fluctuations occur toward the N-terminal.These mostly involved residue numbers 200 to 300 and C-terminal residues numbered between 500 and 550 (Figure 5A).On further carrying out a residue specific analysis of residues which are within 5 Å of dronabinol, we     observed a gradual increase in proximity of the chosen residues toward dronabinol (Figure 6A).For zika helicase, most of the fluctuations happened in the amino acid ranges of 380 and 420 (Figure 5B).A residue specific analysis showed the increased proximity of PRO320, GLN396, ASN463 and PRO464 toward dronabinol.At the same time, ASP193, PRO224, THR318 and PRO319 fluctuate largely and gradually settle closer to the ligand backbone (Figure 6B).The pictorial representations of the residues of interest and dronabinol for dengue and zika helicases have been shown Figures 7A and  7B, respectively.The analysis of the solvent accessible surface area (SASA) showed a decreasing trend in dengue helicase protein in complex with dronabinol as compared to the dengue helicase protein alone (Figure 8A).For Zika NS3 Helicase, it is observed that the SASA has increased as a function of time with respect to conformations bound with the ligand.Plotted in Figure 8B, the SASA increases exponentially from an area of approximately 205 nm 2 to 230 nm 2 .We have calculated the number of hydrogen bonds between the ligand and the protein.For dengue helicase (Figure 9A), the hydrogen bonds were not uniformly sustained.A maximum of 2 hydrogen bonds were observed between dronabinol and dengue helicase during the period of simulation.For the dronabinol-zika helicase complex (Figure 9B), the presence of a number of hydrogen bonds fluctuated from one to three to no hydrogen bonds in the initial timeframe of 20 ns.However, a minimum of one and a maximum of two numbers of hydrogen bond was perpetually maintained between dronabinol and zika helicase throughout the simulation after 40 ns.

Binding energy components of the interaction between dronabinol and helicase proteins-
Estimated binding free energy calculation in Table 2 shows that zika helicase and dronabinol complex showed lower binding free energy (-87.551± 26.230 kJ/mol) than dengue helicase complex (-36.954± 59.414 kJ/mol).The non-polar interactions were the    major contributors to the binding energy in both the complexes.The residue wise free energy decomposition clearly depicted that residues like K388 and E541 contributed significantly toward the binding affinity between dengue helicase and dronabinol (Figure 10A).Similarly, Figure 10B shows that residues like K200, R202 and E286 had a major involvement in the favorable binding of zika helicase and dronabinol.

Principal component analysis of dronabinol and helicase proteins
Principal component analysis (PCA) describes the most accessible motions over a range of time scales.These protein conformations are best characterized by a vector space spanning many dimensions equivalent to the number of degrees of freedom selected to represent the motions.PCA was applied to the backbone Ca atoms in the protein simulation systems both in presence and absence of ligand.Two-dimensional projections of PC1 and PC2 between the two largest eigenvalues in the different protein conformations are represented in Figure 11.It is evident from the figure that the conformational space differs for the protein in absence and presence of the ligand.Both the proteins depict a significantly reduced space in the bound states with dronabinol which indicates the presence of ligand has drastically affected the protein motion, which in lieu may also affect the functional attributes of the helicases.Figure 11 displays porcupine plots of the top two eigenvectors.Combinations of both clockwise and anticlockwise motions are observed.It is observed that there was significant loss of motion observed in both the dengue (Figures 11A and 11C) and zika helicases (Figures 11B and 11D) in the presence of dronabinol.Overall, the most dominating motions are lowered in the docked complexes of both the proteins which may be attributed to the presence of dronabinol (Figures 11C and 11D).

Effect of dronabinol on binding of proteins to their receptor-
To check the effect of dronabinol on the ability of the helicase proteins from dengue and zika viruses to bind to their respective targets we utilized the HDOCK webserver.Dronabinol caused a    decrease in the binding affinity of the two helicases from both dengue and zika, with double-stranded RNA (Table 3).The lowering of the binding energies by dronabinol indicates the unsteadiness that dronabinol causes to the binding of the viral proteins with their respective targets (Basu et al., 2020).

2 D QSAR study of dronabinol
Initially, a total of 563 descriptors were calculated from inhibitors of flavivirus NS3 helicase.Based on the training datasets of the target protein inhibitors, QSAR models were generated, and Model 1, as the best model, was selected.Model 1 indicated a significant correlation between a set of three descriptors; namely, AATSC3i   Q 2 F2 ¼ 0.5371, average Rm 2 (Test) ¼ 0.6334, D Rm 2 (Test) ¼ 0.057 and MAE ¼ 0.2168) predictability of the QSAR model.In the Y-Randomization test against Model 1, an average R 2 value of 0.2325 and average Q 2 (LOO) value of À 0.3191 were obtained (Table 5).The prediction of inhibitory activity of Dronabinol by Model 1 yielded the predicted pIC50 value of À 2.57 against the flaviviral NS3 helicase (Table 6).

Druglikeliness and ADMET properties of dronabinol-
Dronabinol was able to clear more than one drug likeliness filter (Table 7).Dronabinol showed a high GI absorption, Caco2 cell permeability and was capable of crossing the Blood-Brain Barrier (Table 8).Dronabinol also proved to be a substrate of the metabolic enzyme CYPD26 while not inhibiting OCT1 and OCT2 proteins.Dronabinol showed no hepatotoxicity while testing negative for mutagenesis, carcinogenesis and eye irritation (Table 8).

Target prediction studies of dronabinol
The target prediction studies using the SwissTarget webserver showed that dronabinol had a higher probability of binding to the Cannabinoid receptor1 and 2 (Table S1, Supplementary material) while showing low binding probability toward other proteins.

Discussion
We embarked upon this study, to check for the potential of using dronabinol as a cure against these viral infections.Dronabinol docked with critical amino acids of all the proteins in our study.In case of the dengue helicase protein, the conserved residues ASP541 and LYS388 were involved in hydrogen bonding with dronabinol.These amino acids are involved in forming the RNA binding site of the helicase (Swarbrick et al., 2017).Dronabinol docked with GLY197, LYS200 and ARG202 that make up the NTPase active site of the zika helicase protein (Tian et al., 2016).Flavonoids have previously been reported to hamper zika helicase activity by docking with critical residues like GLU286 and GLY197 (Kumar et al., 2020).
Dronabinol docked with the amino acids THR104, LYS105, VAL132 and ILE147, all of which are part of the S-adenosyl Lmethionine (SAM) containing domain (Egloff et al., 2002).In the case of the zika virus, dronabinol bound largely with the stretch of amino acid residues from position 208 to 218 of the zika methyl transferase.These amino acids form the cleft for the binding of 7-methyl guanosine (Coloma et al., 2016).Both of these two sites in the flaviviral methyl transferases have shown potential as a site for drug targeting (Benmansour et al., 2017;Coloma et al., 2016).Dronabinol bounded with amino acids which make up the catalytic finger domain of the dengue and zika RdRP protein (Lim et al., 2013;Godoy et al., 2017).Several potential drugs have been targeted against the finger domain of the dengue virus (Pathania et al., 2022;Wu et al., 2015).Dronabinol docked with the dengue protease protein and docked with amino acids present in the catalytic region of the protease.Dronabinol docked with HIS51 and SER135, both of which are members of the catalytic triad for dengue protease.Aprotinin is a standard serine protease inhibitor and it binds to the catalytic triad region of the dengue protease (Noble et al., 2012).Dronabinol docked with the substrate binding site of the zika protease.Several peptide inhibitors also have been shown to bind to the substrate binding sites of zika protease (Phoo et al., 2018).
From the molecular simulation analysis, the lower RMSD value of helicase-dronabinol complexes indicates the stabilizing conjugation of the ligand and the protein, which is a mandate for successful drug action.The lack of any sudden fluctuation in the RMSD values indicates the stability of the interaction between dronabinol and the dengue and zika helicases (Tran et al., 2022).With a common backbone difference of around 0.1 nm (10 Å) in both proteins, it is evident that dronabinol has induced structural changes from the initial state of both proteins in the study (Li et al., 2021).The overall constant value of the radius of gyration indicated the preservation of the structural integrity of both the dengue and the zika helicase protein in the presence of dronabinol (Haider et al., 2020).
The RMSF values for the protein-ligand complexes for both the helicases as compared to the respective unbound proteins showed significant fluctuations in the RNA binding regions of the helicases.The distance analysis of the residues showing high fluctuations has further indicated that the residues have moved away from the RNA binding site and shifted closer to dronabinol in both the helicase-ligand complexes (Swarbrick et al., 2017;Tian et al., 2016).These may lead to the lowered catalytic efficiency of the helicases as observed previously with other ligand-induced increases in RMSF values (AlZahrani et al., 2022;Xu & Meng, 2020).
The decreasing trend of the SASA values for the dengue helicase-dronabinol complex showed the structural stability of the complex during the course of the simulation.The complex of dronabinol-zika helicase showed an increase in SASA values toward the later part of the simulation.This is indicative of the fact that the ligand has altered interactions of the protein with the solvent (Dash et al., 2019).The presence of hydrogen bonding between dronabinol and the helicases indicated the thermodynamic stability of the binding of dronabinol and the helicases (Majewski et al., 2019).
The outcomes of the binding free energy calculations and residue wise free energy decomposition analysis assure that both the ligands had significant binding affinities with the ligand molecule, dronabinol.These values indicated the thermodynamic viability of the complex between dronabinol and the helicases and further testified to the potential of dronabinol as a flaviviral helicase inhibitor (Rastelli et al., 2010).The lowering of the dominating motions in the helicases when bound to dronabinol may also be indicative of the fact that the ligand may be restricting motions contributing to protein function altogether.The lowering of motions might allow dronabinol to stabilize its binding with the helicases (Haider et al., 2008).The lowering of the motion of the loops in the presence of dronabinol could be the basis of the lowered RMSD values obtained for the complex of dronabinol with the dengue and zika helicases respectively (Mart� ınez-Archundia et al., 2019).
From the docking studies involving the viral helicases and their RNA targets in the host cell, we could observe that the binding energies of the viral helicases with RNA decreased in the presence of dronabinol.The lowered binding energy values in the presence of dronabinol indicate a decrease in the stability of the binding complexes of protein-RNA.Similar studies have been used to check the anti-viral efficacy of other polyphenols like hesperidin, emodin and chrysin (Basu et al., 2020).In our QSAR study, Model 1 underlines the contribution of three 2 D descriptors showing the positive contribution of AATSC3i and the negative contribution of SpMax2_Bhm and SpMin3_Bhe.The high values of R 2 and Q 2 (LOO) recommends that the developed model is acceptable in terms of stability, predictive ability and fitness (Golbraikh & Tropsha, 2002).Moreover, the Y-Randomization test also proved that Model 1 is robust and not obtained by any chance as average R 2 and average Q 2 (LOO) values were lower than their corresponding original values (Mahmud et al., 2020).Interestingly, the predicted pIC50 value of Dronabinol calculated by the model also suggests that Dronabinol is a more potent inhibitor of flaviviral NS3 helicase than compounds of the training dataset.Dronabinol cleared several filters of drug likeliness and showed favorable ADMET properties, thereby justifying its current status of being a clinically administered drug (Schimrigk et al., 2017).The possibility of dronabinol to act as a potential antiviral drug has been further upheld as it showed a very low probability to bind to other human proteins except for the cannabinoid receptor (Gfeller et al., 2014).

Conclusion
From our study we can conclude that dronabinol showed favorable docking with several proteins of dengue and zika viruses like helicase, methyltransferase, RNA dependent RNA polymerase and protease.Out of the above, dronabinol showed the best docking with the helicase proteins of both dengue and zika viruses.The stability of the docking of dronabinol with both the helicases was further confirmed by molecular dynamics simulation as evident from the steady RMSD, RMSF, Rg and SASA values.These results were further complimented by the MM/PBSA analysis which showed favorable binding energies for the complex of dronabinol with both the helicases.The binding of dronabinol to the helicases lowered the molecular motions of the helicases as evident from the PCA analysis.Dronabinol could potentially interfere in the binding of the helicases to RNA. 2 D-QSAR study of dronabinol showed a low pIC50 value as compared to the standard compounds involved in flaviviral helicase inhibition.Dronabinol passed all the drug likeliness filters and showed low probability of binding to human proteins.Thus dronabinol can be further investigated for its efficacy against dengue and zika viruses.

Eye corrosion
Eye irritation Hepatotoxicity X X X x X

Figure 3 .
Figure3.RMSD values of (A) Dengue helicase-dronabinol complex in comparison with dengue helicase protein, (B) Zika helicase-dronabinol complex in comparison with zika helicase protein.

Figure 4 .
Figure 4. Radius of gyration (Rg) values of (A) Dengue helicase-dronabinol complex in comparison with dengue helicase protein and (B) Zika helicase-dronabinol complex in comparison with zika helicase protein.

Figure 5 .
Figure5.RMSF values of (A) dengue helicase-dronabinol complex in comparison with dengue helicase protein and (B) zika helicase-dronabinol complex in comparison with zika helicase protein.

Figure 6 .
Figure 6.Distance analysis plot of residues in closest proximity (5 Å). (A) and (B) are distance plots of residues in dengue and zika helicases respectively that have gradually come closer to the ligand over 100 ns simulation time.

Figure 7 .
Figure 7. (A) and (B) are pictorial representations of the residues and ligand at 100 ns for dengue and zika helicases, respectively.

Figure 8 .
Figure 8. SASA values of (A) dengue helicase-dronabinol complex in comparison with dengue helicase protein, (B) zika helicase-dronabinol complex in comparison with zika helicase protein.

Figure 9 .
Figure 9. Variations in number of Hydrogen bonds between ligand and protein across 100 ns of simulation time.(A) dengue helicase and (B) zika helicase.

Figure 11 .
Figure 11.Porcupine plots of showing overall loss of significant motions across the first principal component in (A) dengue helicase and (C) zika helicase in absence of ligand.Considerable changes and loss of motions are observed in the presence of ligand in (B) dengue helicase and (D) zika Helicase.The higher length of spikes displays a higher amount of motion.

Figure 12 .
Figure 12.Plot of the predicted pIC50 values against the observed pIC50 values of training and test dataset inhibitors of zika virus NS3 helicase.

Table 1 .
Docking results of dengue and zika proteins against dronabinol.

Table 2 .
Computed binding free energy and the contribution of energy elements (kJ/mol) for the two helicases and Dronabinol complex.

Table 3 .
H-Dock derived docking properties of viral helicases with their cellular target.

Table 4 .
Observed pIC50 values, 2 D descriptor values and predicted pIC50 values of training and test dataset inhibitors of flaviviral NS3 helicase.

Table 6 .
2 D descriptor values and predicted pIC50 values of dronabinol.

Table 8 .
Table for ADMET properties of dronabinol.