Mechanistic insights into the Japanese encephalitis virus RNA dependent RNA polymerase protein inhibition by bioflavonoids from Azadirachta indica

Japanese encephalitis (JE) virus is a flavivirus causing encephalitis causing neurological damage. RNA-dependent-RNA-polymerase (RdRp) is responsible for genome replication making it excellent anti-viral target. In this study, the crystal structure of JE RdRp (jRdRp) and bioflavonoids reported in Azadirachta indica were retrieved from specific databases. Structure-based virtual screening was employed using MTiOpenScreen server and top four compounds selected with the most negative docking scores. Conformations were redocked using AutoDock Vina; these complexes showed mechanistic interactions with Arg474, Gly605, Asp668, and Trp800 residues in the active site of jRdRp, i.e., guanosine-5′-triphosphate. Furthermore, 100 ns classical molecular dynamics simulation and binding free energy calculation showed stability of docked bioflavonoids in the active jRdRp pocket and significant contribution of van-der-Waals interactions for docked complex stability during simulation. Therefore, this study predicted the anti-viral activity of Gedunin, Nimbolide, Ohchinin acetate, and Kulactone against jRdRp and can be considered for further antiviral drug development.


Materials and methods
Virtual screening and ADMET predictions. Crystal Structure of viral Japanese encephalitis virus RNA dependent RNA polymerase (jRdRp) protein in complex with guanosine-5′-triphosphate (GTP) (PDB ID: 4HDG) 44 solved at 2.38 Å resolution was used as the receptor molecule for structure based virtual screening with the natural compounds from neem plant (Azadirachta indica) at MTi-OpenScreen webserver 45 . Herein, a total of 43 bioactive bioflavonoids reported in Azadirachta indica were searched in the literature and their 3D conformations were downloaded from the PubChem database (https:// pubch em. ncbi. nlm. nih. gov/) 46 . For structure based virtual screening, initially the receptor was processed by following steps: (1) removal of crystalized water molecules and heteroatoms, (2) assignment of Gasteiger charges, (3) addition of polar hydrogen atoms and (4) removal of non-polar hydrogen atoms using default parameters of Dock Prep tool in UCSF Chimera-1.14 47 . Following, collected bioactive compounds were screened at the catalytic site in jRdRp using MTi-OpenScreen webserver. After that, top four bioactive compounds with highest negative docking score were considered for further ADMET (Absorption, distribution, metabolism, excretion, and toxicity) analysis using SwissADME (http:// www. swiss adme. ch) 48 and AdmetSAR (http:// lmmd. ecust. edu. cn/ admet sar1/) servers 49 . Re-docking simulation and binding pose profiling. Molecular docking simulation between the jRdRp and potential natural compounds was performed by Chimera-AutoDock Vina plugin setup to decipher the most interacting residues in the active site of jRdRp against reference molecules, as reported earlier 50 . Briefly, 3D structure of protein and potential bioflavonoids as ligand were subjected to minimization under the default parameters in structure minimization tool in UCSF Chimera-1.14 47 . Later, both receptor and ligands were primed for docking in Dock prep tool in Chimera with default parameters, where native ligand from the crystal structure was removed, polar hydrogen atoms hydrogen atoms and charges were added. Finally, molecular docking simulations were performed using AutoDock Vina 51 as plugin under default setting at the nucleotide binding site in jRdRp protein by adjusting the grid size of 30 × 30 × 30 Å along both three (X, Y, and Z) axes, covering all the essential residues centre at − 38.85, − 0.98, − 43.61 Å region, to distribute profuse space to the conformations of ligands during docking processes. Herein, at least 10 docked poses for the ligands were collected and conformation with highest negative docking score and least root mean square deviation (by default 0 in AutoDock Vina) were extracted for further binding pose analysis under default parameters using 2D interaction diagram Classical molecular dynamics simulation. Molecular dynamics simulation method was employed to analyse the stability and intermolecular interactions of the selected protein-ligand complexes obtained from the docking experiments. MD simulations for the docked complexes were performed under Linux environment on HP Z2 Microtower workstation using Desmond-maestro 2018.4 52 . The simulation system was prepared using the system builder module in Desmond and TIP4P model was employed for the solvation of the docked complex by placing in a 10 Å × 10 Å × 10 Å orthorhombic box. Later, the complete system was neutralized by adding Na + or Cl − counter ions to maintain the neutrality and further minimized under default parameters using minimization tool. Finally, the whole system was subjected to simulation at a temperature of 300 K and pressure of 1.01325 bar using Nose−Hoover thermostat and Martyna−Tobias−Klein method, respectively under default parameters. Each system, including reference complex, were simulated for 100 ns under default parameters with OPLS (optimized potentials for liquid simulations)-2005 force field, where a total of 5000 frames were collected at 20 ps interval during the simulation interval using molecular dynamics tool of Desmond-maestro 2018.4 52 . where ΔG bind denotes the binding free energy, ∆G complex, indicates the free energy of the complex, ∆G receptor and ∆G ligand exhibits the energy for receptor and ligan, respectively.

Results and discussion
Structure based virtual screening. A total of 43 bioflavonoids (Table S1) documented in Azadirachta indica were collected from the literature and used in virtual screening against the catalytic site in jRdRp protein.
All the compounds showed considerable binding affinity between − 11.6 and − 5.2 kcal/mol with the active residues of viral jRdRp protein (Supplementary Table S1). Following top four compounds, i.e., Gedunin, Nimbolide, Ohchinin acetate, and Kulactone, with higher docking scores were considered for further re-docking and intermolecular interaction analysis against native crystalized ligand, i.e., guanosine-5′-triphosphate (GTP) and Adenosine triphosphate (ATP) as reference molecules (Fig. 1). ADME profiling. The intrinsic properties, such as drug-likeness and pharmacological characteristics, have been marked as an important factor for compounds under medical applications. Hence, ADMET properties for top four selected compounds, i.e., Gedunin, Nimbolide, Ohchinin acetate, and Kulactone, were predicted using SwissADME and admetSAR, which help to analyse the pharmacokinetic and toxic properties of compounds ( Fig. 2, Supplementary Table S2). All four selected bioflavonoids show negative AMES toxicity test and non-carcinogenic profile which reveals the non-mutagenic nature of compounds predicted using admetSAR. Cytochrome P450 2D6 (CYP2D6) is an enzyme plays vital role in the metabolism of drugs and xenobiotics and inhibition of this enzyme can lead to drug-drug interactions. Interestingly, all the selected bioflavonoids were found to be non-inhibitors of CYP2D6 using SwissADME along with other cytochromes; list of the complete ADME (absorption, distribution, metabolism, and excretion) profiling for the selected bioactive compounds is given in Supplementary  Table S2). However, it is important to mentioned that rules for drug-likeness are not applicable to the natural bioactive compounds as these agents can be identified by the active transport system of the cells when contemplating "druggable chemical entities" 54,55 . In addition, other properties such as pharmacokinetics and medicinal chemistry friendliness were computed for the potent compounds (Supplementary Table S2). Conclusively, the screened compounds were suggested with ideal medical properties.

Re-docking and intermolecular interaction analysis. Structure based virtual screening algorithms
are fast and comparatively less accurate; hence, typically best conformations of the ligands has been suggested to  Table S3). These results suggested the considerable binding affinity of bioactive compounds in the active pocket of jRdRp protein.
Interestingly, the re-docking of GTP as reference molecule in the same active pocket reveals − 9.0 kcal/mol and interactions with key residues, including seven hydrogen bonds formation (Asp 541 (2), Asp 668 , Asp 669 , and Ser 715 ), pi-cation (Arg 474 ), salt bridge formation (Arg 474 , and Lys 463 (2)), hydrophobic (Trp 54 , Tyr 610 , Cys 714 , and   Figure S2). In re-docked complexes of reference molecules, the orientation of GTP and ATP was silghty different in their original binding pocket of jRdRp when compared to their crystal structures (Supplementary Figure S3-4). Of note, interacting residues were also noted in the crystal structure of jRdRp-GTP and jRdRp-ATP complex and essentially required to conduct the replication of viral RNA genome 44 . Overall, molecular contact analysis showed interaction of docked bioactive compounds with essential residues of the A, B, C, E and F motifs and the priming loop in viral RdRp protein (Fig. 3, Supplementary www.nature.com/scientificreports/ Table S3). Hence, these findings suggested the screened bioactive compounds in the order of Ohchinin acetate, Nimbolide, Gedunin, and Kulactone as potential inhibitors of jRdRp by comparison to reference ligands.

Classical molecular dynamics simulation analysis.
In computational drug discovery, molecular dynamics (MD) simulation is used to predict the docked complexes stability and formation of intermolecular interaction in reference to time 56,57 . In this study, the stability of docked complexes was monitored by means of root mean square deviation (RMSD), root square means fluctuation (RMSF), and protein-ligand contact mapping, extracted from respective 100 ns simulation trajectories. Typically, RMSD and RMSF are used to observe the structural fluctuations which are essentially required to establish the dynamic stability of the system. Furthermore, the protein and ligands contact maps are analysed to calculate the intermolecular interaction in docked poses as function of time during simulation interval scrutinize the stability of docked ligands at the active pocket of viral protease.
RMSD and RMSF analysis. Initially, docked complexes of potential bioactive compounds with JRdRp were analyzed for protein and ligand RMSD with respect to initial pose as reference frame (Fig. 4). In all the docked complexes, RMSD for jRdRp showed deviations < 2.8 Å till 60 ns, except JRdRp-Kulactone showed deviations between 4.5 and 5.8 Å till end of the 100 ns simulation interval. However, jRdRp docked with reference ligands, i.e., GTP, exhibits acceptable deviations (~ 1.50 to 2.0 Å) and stability during the simulation interval (Supplementary Figure S5) while docked complex with ATP showed deviations < 3.2 Å till 40 ns followed by variations and a maximum of < 5.6 Å was noted at the end of 100 ns MD simulation interval (Supplementary Figure S6).
These observations were further supported by the calculated RMSF values (< 3.5 Å), except higher deviations (< 5-7.2 Å) were noted in the C-terminal of jRdRp protein docked with Nimbolide and Ohchinin acetate. Also, acceptable deviations (< 3.5 Å) were observed in the residues of jRdRp protein interacting with the respective docked ligands (Supplementary Figure S7). Collectively, these observations suggested the considerable rigidity and stability in jRdRp protein docked with bioactive compounds during the simulation without major structural deformations. Besides, docked bioactive compounds, i.e., Gedunin, Nimbolide, Ohchinin acetate, and Kulactone, at the active site of jRdRp protein showed considerable RMSD values till 100 ns with acceptable deviations (< 3-3.8 Å), indicates the stability of docked ligands in the active site of jRdRp protein (Fig. 4). Likewise, reference molecule, viz. GTP, also exhibits acceptable RMSD (< 3 Å) fit at the active site of jRdRp protein revealed the substantial stability of respective docked complex (Supplementary Figure S5). In case of reference molecule ATP, no change in the protein structure was observed while ligand showed stability after 60 ns (Supplementary Figure S6). Furthermore, each ligand showed acceptable RMSF values (< 3 Å) fit on the protein during 100 ns simulation, inferred the significant stability of docked bioactive compounds at the active site of jRdRp protein (Supplementary Figure S8).
Protein-ligand interaction profiling. The docked complexes of viral protein with potential compounds were also considered for protein-ligand interaction profiling against reference complex during the course of 100 ns simulation interval in terms of hydrogen bonding, hydrophobic interactions, ionic interactions, and water bridge formation (Fig. 5, Supplementary Figure S9-10).
In the case of jRdRp-Gedunin, docked complex exhibits hydrogen bond formation for 90% of the simulation time at Gly 605 residue, Tyr 610 and Trp 800 residues noted for hydrophobic interactions for more than 50% of the simulation time, and Asp 669 and Ser 715 residues participated in water bridges for more than 30% of the MD time (Fig. 5a). Also, jRdRp-Nimbolide complex indicates hydrophobic interactions via Trp 800 for 80% of the simulation time, Arg 460 residue was noted for hydrogen bond formation in 30% of the simulation time, Arg 474 and Ser 604 residues contributed to water mediated interactions for 30% of the total simulation time (Fig. 5b). Moreover, jRdRp-Ohchinin complex exhibits the most stable hydrogen bond interaction at Arg 474 with more than 100% of the simulation time and water bridges formation at Asp 668 residue for 75% of the simulation interval (Fig. 5c). Furthermore, significant water bridge interactions via Arg 484 (40%) and Asp 668 (90%) were noted for the jRdRp-Kulactone docked complex during 100 ns simulation interval (Fig. 5d). However, reference complexes, viz. jRdRp-GTP and jRdRp-ATP, showed the most significant interaction at Arg 474 via both hydrogen bonds and water bridge formation for 50% of the simulation interval (Supplementary Figure S11-12). Interestingly, the interactive residues were also observed in the respective docked complexes and essentially required for the replication of virus (Supplementary Table S2). Hence, these results support the considerable stability of selected compounds at the active pocket of jRdRp by formation of strong hydrogen bonding and hydrophobic interactions; hence, can be used as potent inhibitors of jRdRp protein.
Additionally, intermolecular interactions between the residues of jRdRp protein and potential bioactive compounds, i.e., Gedunin, Nimbolide, Ohchinin acetate, Kulactone, and reference ligands, viz. GTP and ATP, were calculated at total 30% interval of 100 ns simulation, which revealed considerable binding of respective ligands with active residues (Fig. 6, Supplementary Figure S8). Interestingly, all the selected ligands were observed for hydrogen bonding and water bridge interactions, suggested the stability of selected compounds at the active site of jRdRp protein. Hence, based on 100 ns molecular dynamics simulation analysis, docked complexes can be arranged in order of stability, i.e. jRdRp-Ohchinin acetate, jRdRp-Gedunin, jRdRp-Kulactone, and jRdRp-Nimbolide.
Binding free energy analysis. The binding affinities of the four selected protein-ligand complexes were estimated using MM/GBSA approach. Herein, poses of jRdRp-bioactive compounds, i.e., Gedunin, Nimbolide, Ohchinin acetate, and Kulactone, were extracted from 100 ns MD simulation trajectory at every 10 ns interval for average free binding energy calculations against reference complex, viz. jRdRP-GTP complex (Fig. 7 Figure S9). All the docked complexes of bioactive compounds with jRdRp exhibits significant binding free energy (> − 50 kcal/mol). Of note, highest ΔG Bind (− 61.13 ± 3.26 kcal/mol) was recorded for jRdRp-Nimbolide complex against jRdRp-GTP (ΔG Bind = − 78.9 ± 9.05 kcal/mol) and jRdRp-ATP (− 33.34 ± 3.61 kcal/ mol). Moreover, calculation of dissociation energy components for each complex exhibits substantial support of ΔG Bind Lipo , and ΔG Bind vdW in the stability of docked complexes while ΔG Bind Solv GB was observed in contribution for instability of the respective complexes (Fig. 7, Supplementary Figure S13). Conclusively, results suggested the

Conclusion
The role of RNA dependent RNA polymerase (jRdRp) in the replication and survival of RNA genome viruses inside a host and absence of human homolog makes it an attractive molecular target for anti-viral drugs. Besides, board range of secondary metabolites as natural bioactive compounds in Azadirachta indica have been reported with medicinal benefits. Hence, this study evaluated the potential of known bioactive compounds in Azadirachta www.nature.com/scientificreports/ indica as potential inhibitors of jRdRp protein using complex molecular simulation, drug likeness profiling, and end-point binding free energy calculations. Herein, among the screened 43 bioactive compounds at the active site of jRdRp protein, four bioflavonoid compounds, viz. Gedunin, Nimbolide, Ohchinin acetate, and Kulactone, with substantial docking score and druglikeness were considered for further mechanistic inhibition analysis of jRdRp protein. The analysis of docked complexes showed that selected bioactive compounds occupied the active site by both hydrogen bonds, hydrophobic interactions, and other considerable intermolecular interactions with essential residues of jRdRp protein. Moreover, molecular dynamics simulation and free energy calculation further revealed the substantial stability and contribution of van der Waals interaction in the stability of respective docked complexes, respectively. The low penetration of the screened compounds to the BBB might affect their efficacy, however BBB penetration can be improved using SAR of the screened compounds during lead optimization. The investigated flavonoids might have potential effect against other flaviviruses which needs further investigation against specific viruses. In conclusion, these results supported the Gedunin, Nimbolide, Ohchinin acetate, and Kulactone as acceptable inhibitors of jRdRp protein and advised to studied for further in vitro and in vivo experiments to develop potential treatment against Japanese encephalitis virus.