Results and Discussion
The NS3 protein (accession No. YP_009227202) of ZIKV has 617 amino acids residues and NS3 protein of DENV (accession No. NP_740321.1) has 618 amino acid residues in its protein sequence. The NS3 protein of both ZIKV and DENV have structurally similar with three conserve domain, i.e., FLAVIVIRUS_NS3PRO, Flavivirus NS3 protease (NS3pro) domain; HELICASE_ATP_BIND_1, superfamilies 1 and 2 helicase ATP-binding type-1 domain; and HELICASE_CTER, superfamilies 1 and 2 helicase C-terminal domain (predicted from ScanProsite tool of ExPASy) [33]; however, there are 67% identities and 100% similarities in their amino acid sequences obtained from BLASTP result [34].

Protein structure modeling and validation
Three-dimensional structure of both NS3 proteins (Fig. 1A) were modeled using Phyre2 server which has taken chain A of 2VBC (crystal structure of the NS3 protease-helicase from DENV) as template for modeling having 100% confidence and 67% identity. From the ProCheck analysis obtained from Ramachandran plot (Fig. 1B), it was observed that 88.8% residues were located in most favorable region, 9.8% were in additional allowed region, 0.4% residues in generously allowed regions whereas 0.9% of the residues located in disallowed region. The compatible Z score value of –10.71 (Fig. 1C) obtained from ProSA-web evaluation revealed that the modelled structure is fit within the range of native conformation of crystal structures [29]. Further, the overall residue energies of NS3 was largely negative (Fig. 1D). The modelled NS3 protein of ZIKV showed Levitt-Gerstein (LG) score of 5.060 by Protein Quality Predictor (ProQ) tool, it implies the high accuracy level of the predicted structure and taken to account as extremely good model. The ProQ LG score of more than 2.5 is needed to proposing a predicted structure is of very good quality [30]. By using ERRAT plot same assumptions were achieved, where the overall quality factor is 67.002%. All of the above out comes recommended the reliability of the proposed model.

Protein-ligand docking
Different studies have already reported that the nonstructural NS3 protein is very essential for the development of the viral polyprotein and is a suitable target for designing antiviral inhibitors [32]. Analysis of the amino acid sequence of the NS3 protein of hepatitis C virus suggested that this viral protein contains a trypsin-like serine protease domain that functions in the processing of the viral polyprotein [35]. NS3 protein of ZIKV contains catalytic triad of His51, Asp75, and Ser135 [9]. The active site of serine proteases in variably contains three residues, histidine, aspartate, and serine that maintain the same relative spatial position in all the known structures of these enzymes. Thus, we performed molecular docking of four FDA approved drugs using AutoDock Vina [26] around the active site residues of NS3 protein of both ZIKV and DENV. From the docking analysis, it was observed that among all the drugs berberine showed the lowest binding energy in both ZIKV and DENV of –5.8 kcal/mol and –6.6 kcal/mol respectively. In ZIKV, berberine has formed hydrogen bond with Ser135 (Fig. 2). Likewise other three drugs, i.e., amodiaquine, prochlorperazine, and Quinacrine also docked with NS3 protein of both ZIKV and DENV with considerable binding energy (Table 1). From ZIKV protein-ligand docking we select the ligand having lowest binding energy then proceed for screening to get more similar drug like compounds.

Virtual screening and validation of docking results
The virtual screening technique is an economical, reliable and time saving method for screening large set of lead molecules. In our study, we have used PyRx 0.8 [24] for screening lead molecules against NS3 protein was used to solve the purpose. The NS3 protein of ZIKV was used as receptor to screen 4,121 drug-like compounds (small molecules) based on properties of berberine such as molecular weight of 336.367 g/mol, Xlog p-value of 0.20, net charge of 1, rotatable bonds 2, polar surface area 41Å2, hydrogen donor 0, hydrogen acceptor 5, apolar desolvation 8.91 kcal/mol, and polar desolvation of 8.91 kcal/mol. Further, an approximate range molecular weight ranges from (300 to 400 g/mol); Xlog p (0.10 to 0.50); net charge (–1 to 3); rotatable bonds (1 to 5); polar surface area (20 to 60 Å2); hydrogen donor (0 to 5); hydrogen acceptor (2 to 8); apolar desolvation (5 to 15 kcal/mol); and polar desolvation ranges from (–40 to –10 kcal/mol) was taken to retrieve similar drug compounds from ZINC database. AutoDock Vina [26] in PyRx 0.8 was employed for screening the ligands which generates nine distinct poses of each ligand. Further, based on the hydrogen bond interaction with active site residues of NS3, the docking pose for each ligand was selected.
Top ten drug like compounds (Table 2) obtained by virtual screening bound to catalytic site of NS3 protein with lowest binding energy were taken as promising candidates for further analysis. All the ligand molecules observed to follow the Lipinski's rule of five (molecular weight not more than 500 Da; hydrogen bond donor not more than 5; hydrogen bond acceptors not more than 10; log p-value not greater than 5). The chemical structure of these ten compounds along with known inhibitor berberine are given in Fig. 3.
All the proposed lead molecules were observed to form hydrogen bonds with active site residues of NS3 protein (Fig. 4). The Lead 1 molecule is ZINC53047591(2(benzylsulfanyl)-3-cyclohexyl-3H-spiro[benzo[h]quinazoline-5,1'-cyclopent an]-4(6H)-one); having molecular weight 456.64218 (g/mol) showed the lowest binding of –7.1 kcal/mol and observed to interact with active site residues of NS3 by forming hydrogen bonds with SER135 and ASN152. The other nine lead molecules are also observed to inhibit NS3 protein with significant lower binding energy as compared to berberine. Thus, the higher binding affinity of selected lead molecules compared to berberine in respective docking complexes proposed that the selected ligands would probably bind more competitively in the active site of NS3 protein (Table 2).
The participation of selected lead molecules have very important role to inhibit the NS3 protein through hydrogen bonding interaction with its active site residues justified them as possible inhibitors against the important enzyme of ZIKV. Therefore, the given inhibitors may be demonstrated through biochemical assays.
In conclusion, the ZIKV belongs to Flaviviridae family, which is similar to DENV, West Nile virus and yellow fever virus. Currently, there are no specific drugs for prevention and treatment of ZIKV. As, the NS3 protein of this virus perform a central part in the viral life cycle, it is of appreciable interest in designing new potential drug candidate to deal with conquer the challenges of ZIKV infection. A high quality 3D structure of NS3 protein was predicted and validated through computational approach and molecular docking technique was employed to observe the interaction of the existing drugs used against other flavivirus, against ZIKV NS3 protein. Further, for identifying novel inhibitors, high-throughput virtual screening approach was employed in this study, would be of significant in designing new drugs for ZIKV infection.