Corrigendum: Computational Approach Towards Exploring Potential Anti-Chikungunya Activity of Selected Flavonoids

Scientific Reports 6: Article number: 24027; published online: 13 April 2016; updated: 31 May 2016 The original version of this Article contained an error in the spelling of the author Esaki Muthu Shankar, which was incorrectly given as Shankar Esaki Muthu. This has now been corrected in the PDF andHTML versions of the Article.

also reported against CHIKV, including silymarin 16 and luteolin 17 . However, none of these compounds are presently approved for the treatment of CHIKV infection 4 . Nevertheless, compounds such as ribavirin and chloroquine have been tested in clinical trials 1,18 .
The conventional method of performing in vitro cell culture-based assays previously used by researchers to discover new lead compounds with antiviral activity is costly for many investigators, especially if several compounds will be screened. To fill this gap in the field, instead of using the costly conventional method, we can use bioinformatics tools as a cost-effective solution for primary virtual screening of potential compounds.
Molecular docking, a branch of bioinformatics that accelerates the drug design process, is used in the biopharmaceutical industry to discover and develop new lead compounds 19 . The purpose of this assay is to predict a valid pose from a receptor conformation and a set of ligand conformations using scoring based on their binding affinity 20 . We recently began a series of in vitro studies on different potential flavonoids against CHIKV and have preliminarily observed in vitro anti-CHIKV activity of three selected flavonoids, namely baicalin, naringenin and quercetagetin, that is still being investigated (unpublished data). Therefore, we have designed the present study using computational approaches to discover the potential of these three flavonoids targeting nsP3 as one of the most essential viral elements for CHIKV intracellular replication.

Results
The ligand conformations were ranked according to their predicted binding affinities using the default scoring function in AutoDock Vina. The nsP3 residues that formed close contacts with ADP-ribose are listed in Table 1 and shown in Fig. 1. The best docking conformation of ADP-ribose showed a binding affinity of − 8.7 kcal mol −1 , whereas among the three other ligands tested, baicalin showed the most potent antiviral activity with a binding affinity of −9.8 kcal mol −1 . The binding affinities and their interaction energy are tabulated in Table 2. The intermolecular hydrogen bonds formed between each compound and nsP3, together with their distances, are presented in Table 3, which shows that the majority of the hydrogen-bond donors came from the protein residues and that the corresponding acceptors were derived from the ligands. It is evident that there were interactions of the ligands with 10 residues in the active site of nsP3 (baicalin: LEU108, TYR142, SER110, THR111; naringenin: SER110, THR111; quercetagetin: CYS34, LEU108, ARG144, ASP145). There was also one pi-pi interaction between baicalin with nsP3 residue TYR114 ( Table 4). The best docking pose for each ligand was also recorded (Figs 2-4).

Discussion
To date, only a little information is available on CHIKV nsP3 and its potential inhibitors. According to Rashad et al. 21 , the most promising targets from a chemical and biological standpoint are CHIKV nsP2 and E protein.
Only one molecular docking study targeting nsP3 can be found in the literature; it reports on the combination of molecular docking, virtual screening and molecular dynamics simulations to identify the potential inhibitors of CHIKV nsP3 22 . Other molecular docking studies targeted CHIKV nsP2 23 , nsP4 24,25 and viral E2 protein 26 .
Considering the role of nsP3 in viral replication, the discovery of an nsP3 inhibitor is a major advance towards finding an active compound that can interfere with intracellular CHIKV replication. In the present study, we performed molecular docking using AutoDock Vina to study the antiviral activity of baicalin, naringenin and quercetagetin. The inhibitory effects of these three flavonoids against CHIKV in vitro replication were revealed during our other ongoing experimental studies (unpublished data), but their mechanisms of action remain unknown. Thus, the present study was conducted to evaluate the potential mechanism of action of these flavonoids based on the hypothesis that flavonoids can interfere with the viral replication cycle.   Table 4. Pi-pi interaction between baicalin and CHIKV nsP3 residue, TYR114. During the initial stage of the present study, ADP-ribose was re-docked into the nsP3 protein and managed to reproduce the important closely contacting residues in the ADP-ribose binding site in reference to the complex crystal structure from the Protein Data Bank (PDB ID: 3GPO). The results of the re-docking procedure comprising the closely interacting residues were used to evaluate the potential pharmacokinetic properties resulting from the binding sites of the subsequently tested ligands. Of the three tested ligands, baicalin showed the strongest interaction with nsP3 and showed the most potent potential anti-CHIKV activity, with a binding affinity of −9.8 kcal/mol and low Ki value of 0.064 μM, followed by quercetagetin and naringenin, with respective binding affinities of −8.6 and −8.4 kcal/mol. The Ki value or inhibitor constant is an indicator of the potency of an inhibitor, with a highly potent inhibitor being indicated by a low Ki value. Drugs with a Ki value <1 mM are normally considered to be effective 27 . Baicalin showed interaction with four residues of nsP3, LEU108, TYR142, SER110 and THR111, with distance ranges from 1.81482 to 2.42882. The most stable H-bond is that with an approximately 180° angle, i.e. close to linear 28 . A review by Szatylowicz 29 classified the energy borders setting for strong, moderate and weak H-bonds, with 1.2-1.5 considered strong, >1.5-2.2 moderate and >2.2 weak. All intermolecular hydrogen bonds between baicalin and nsP3 in this study fell under the moderate bond group with one exception, interaction with the THR111 residue, which was categorised as a weak bond. Most of the hydrogen-bond donors came from the protein residues, in agreement with the study reported by Nguyen et al. 22 .

Interacting residue Distance (Å) H bond (D-H-A)
We also discovered one pi-pi interaction between baicalin and TYR114 of nsP3. In addition to hydrogen bonding and the pi-pi interaction, the van der Waal forces and electrostatic interaction energy were also documented. The van der Waal interaction is a weak intermolecular force between molecules that occurs when there is a fluctuation in the electron cloud of a nucleus that affects the transient dipole moment and electron cloud of nearby atoms. Despite its weak energy, large numbers of interaction can occur with the existence of good steric and electrostatic complementarity between an enzyme's binding pocket and a ligand's structure 30 . Electrostatic  interaction occurring between a charged ligand and a receptor's binding pocket helps to provide the thermodynamic driving forces involved in forming protein-ligand complexes.
The macro domain of CHIKV nsP3 contains the ADP-ribose binding site, which has been suggested to play an influential role in the metabolism of ADP-ribose 1″ -phosphate and/or other ADP-ribose derivatives possessing regulatory activities in the cell 31 . In reference to the observed interacting residues in the ADP-ribose binding site, we found that all but one hydrogen bond (LEU108, THR111, TYR142) and the residue involved in the pi-pi interaction (TYR114) of baicalin with the nsP3 protein occurred within the active site of nsP3. Similarly, the hydrogen bonds formed when naringenin (THR111) and quercetagetin (CYS34, LEU108, ARG144) were docked to the active site of nsP3 were also observed when ADP-ribose was re-docked to the protein. Hence, it can be hypothesised that the binding of these three ligands will potentially result in a pharmacokinetic effect via their interactions with the ADP-ribose binding site.
Baicalin is a metabolite of baicalein, which can be extracted from the root of Scutellaria baicalensis, a Chinese medicinal herbal plant. Many pharmacological and in vitro antiviral activities of this flavonoid have been described. Other than having the ability to inhibit dengue virus replication and internalisation 11 , baicalin also serves as a potential antiviral agent against influenza viruses, where it acts as a neuraminidase inhibitor 32 . Another study in China presented a potent antiviral effect of this flavonoid against enterovirus 71 by inhibiting EV71/3D polymerase expression and the Fas/Fasl signalling pathways 33 .
In the medical industry, protein-ligand docking, rather than other docking types, has drawn special attention due to its fundamental role in structure-based drug design. Apart from being used to screen large databases to find a new antiviral lead compound, it is also useful for other purposes. A group of researchers studied the potential of cyclopentadeca-4, 1,2-dienone as an antidiabetic drug using Schrodinger software, with the result showing 52.3% α -amylase inhibitory activity 34 . Another docking study using AutoDock Vina was conducted to recognise the probable inhibitor against transketolase enzyme in Plasmodium falciparum, the protozoan responsible for malaria. The results of this study showed that 6′-methyl-thiamin diphosphate, an organic compound, gives the best pose with a binding affinity of −6.6 kcal/mol. Qiu et al. 35 also documented a computational study to examine the molecular interaction between the active site of ginger and the human terminal oxidase enzyme cytochrome P450. The findings of this study led to the resolution of herbal medicine issues and safety concerns.
Molecular docking aims to predict the optimal ligand-receptor complex orientation and conformation in two steps, first by gathering conformation of ligands in the active site of the protein and then by ranking the conformations via a scoring function to predict binding tightness for each individual orientation 36 . Many commonly used molecular docking software programs are available, including AutoDock Vina, GOLD, FlexX, FRED and DOCK. However, AutoDock Vina was found to be the most useful software in blind docking pose prediction by consistently performing better than the other docking programs 37 . In addition, its accuracy and speed are approximately double that of its predecessor, AutoDock4 38 .
In conclusion, we found that among the three different ligands tested, baicalin exhibited the most potent potential antiviral activity against CHIKV nsP3 with a binding affinity of −9.8 kcal/mol, followed by quercetagetin and naringenin with respective affinities of −8.6 and −8.4 kcal/mol. Considering the involvement of CHIKV nsP3 in the intracellular CHIKV replication cycle, this result suggests that baicalin can potentially interfere with the post-entry stage(s) of CHIKV infection, which should prompt further investigations to reveal the mechanism of action of baicalin against CHIKV in vitro replication.

Methods
Receptor and Ligand Preparation. Molecular docking and virtual screening were conducted using AutoDock Vina. In this study, the protein was kept rigid while the ligands were fully flexible. The protein was prepared by retrieving the three-dimensional crystal structure of the nsP3 from the Protein Data Bank (PDB ID: 3GPG) and this was used as the receptor for molecular docking. The ligands, ADP-ribose, baicalin, quercetagetin and naringenin, were drawn using ChemDraw. The protein was subsequently cleaned by removing the water molecules, followed by management of its conformer and the minimisation process. The CHARMM27 force field was then applied in the Accelrys Discovery Studio 2.0 software package. Molecular Docking Using AutoDock Vina. The input files for AutoDock Vina were prepared using AutoDock Tools v. 1.5.6. After the minimising process, the protein was placed in a grid box measuring 26.85 Å × 28.17 Å × 24.53 Å along the x, y and z axes, respectively, using AutoDock Vina at 1.00 A to define the binding site. The configuration file used for the docking process was also prepared along with the addition of hydrogen bonds and the Gasteiger charge. ADP-ribose was first re-docked into the ADP-ribose binding site of nsP3 22 , and the resulting interactions were compared with those found in the subsequent focused dockings of baicalin, quercetagetin and naringenin into the similar active site using the same grid box. The docking procedure was performed using the instructed command prompts. The docking results included the binding energy value given in kcal/mol, the locations of hydrogen bonds, pi-pi interactions and closely interacting residues.
Analysing and Output Visualisation using Discovery Studio. The docking poses were ranked according to their docking scores. The scoring function in AutoDock was used to predict the binding affinity of one ligand to the receptor molecule. The conformation with the lowest binding affinity was selected for further analysis after the docking process. Ki was calculated by the equation: Ki = exp [(ΔG*1000)/(R*T)], where ΔG is docking energy, R (gas constant) is 1.98719 cal K −1 mol −1 and T (temperature) is 298.15 K. Only the best pose (the one with the lowest binding energy) was considered for each ligand. The molecular visualisation of the docked complexes was performed using the Accelrys Discovery Studio software package 31,38 .