Molecular Docking and Molecular Dynamics Simulation Studies to Predict Flavonoid Binding on the Surface of DENV2 E Protein
Dengue infections are currently estimated to be 390 million cases annually. Yet, there is no vaccine or specific therapy available. Envelope glycoprotein E (E protein) of DENV mediates viral attachment and entry into the host cells. Several flavonoids have been shown to inhibit HIV-1 and hepatitis C virus entry during the virus–host membrane fusion. In this work, molecular docking method was employed to predict the binding of nine flavonoids (baicalin, baicalein, EGCG, fisetin, glabranine, hyperoside, ladanein, quercetin and flavone) to the soluble ectodomain of DENV type 2 (DENV2) E protein. Interestingly, eight flavonoids were found to dock into the same binding pocket located between the domain I and domain II of different subunits of E protein. Consistent docking results were observed not only for the E protein structures of the DENV2-Thai and DENV2-Malaysia (a homology model) but also for the E protein structures of tick-borne encephalitis virus and Japanese encephalitis virus. In addition, molecular dynamics simulations were performed to further evaluate the interaction profile of the docked E protein–flavonoid complexes. Ile4, Gly5, Asp98, Gly100 and Val151 residues of the DENV2-My E protein that aligned to the same residues in the DENV2-Thai E protein form consistent hydrogen bond interactions with baicalein, quercetin and EGCG during the simulations. This study demonstrates flavonoids potentially form interactions with the E protein of DENV2.
KeywordsDengue virus MalaysiaDENV2FlavonoidsBaicalinBaicaleinEGCGFisetinGlabranineHyperosideLadaneinQuercetinDockingMolecular dynamics simulationsEnvelope glycoprotein E
Dengue virus (DENV) infection transmitted by infected Aedes mosquitoes, particularly Aedes aegypti [1, 2], is now becoming a global health threat estimated to infect 390 million people annually in tropical and subtropical countries [3, 4]. The virus causes dengue fever and in severe cases leads to lethal hemorrhagic fever/shock syndrome. In Malaysia, dengue infection is a serious public health problem, and a major outbreak occurs every 4 years . In 2007, there were more than forty-eight thousand dengue infection cases in Malaysia alone. Yet, there is no vaccine or specific antiviral available.
DENV is a member of Flaviviridae family and categorized in the subfamily of flavivirus. There are four DENV serotypes (DENV1, DENV2, DENV3 and DENV4), which are distinct but closely related to clinical manifestations and epidemiology . Other well-known viruses in the same subfamily are West Nile virus (WNV), Japanese encephalitis virus (JEV) and tick-borne encephalitis virus (TBEV). The mature DENV virion consists of 180 copies of envelope glycoprotein E and M on the external shell that anchored onto the viral lipid bilayer. The RNA genome encodes a single polyprotein that is co-translationally and posttranslationally processed into three structural (capsid, prM and E proteins) and seven non-structural proteins (NS1, NS2a, NS2b, NS3, NS4a, NS4b and NS5) [7, 8].
The E protein plays a vital role in the viral entry by mediating the viral attachment to the host cell receptors and involves in the fusion process between the viral and the host membrane . The virus entry process initially starts in the endosome, and it is triggered by low pH environment (~6.5 and below), which induces histidine protonation of E protein. Due to that, the E proteins of DENV undergo major rearrangement involving the domain I and domain II [10, 11]. A hydrophobic stretch, also known as the fusion peptide, that is buried between the domain I and II during the viral prefusion state is then exposed and reoriented at one end after the transition to the postfusion trimeric structure. In this conformation, the fusion peptide attaches to the target cell membrane and initiates the viral membrane fusion process [12–14]. Thus, targeting the E protein to design the DENV entry inhibitors is one of promising strategies to inhibit the virus. Current in silico screenings of small molecule inhibitors mostly target the β-N-octyl-glucoside (β-OG) binding site, which is a pocket found to occupy a detergent molecule in a DENV E protein crystal structure . There were potential lead compounds discovered [16–18]; however, most of the compounds failed to proceed to the subsequent experimental stages due to their poor druggability properties. Antiviral drugs targeting the entry stage of the virus were already discovered for HIV-1 [19, 20] and hepatitis B virus (HBV) , but for the dengue virus it is still very challenging.
Flavonoids are one of the most studied natural compounds that possess various pharmacological activities, such as antioxidative , antibacterial  and antimutagenic properties . A study by Calland et al.  showed epigallocatechin gallate (EGCG) inhibited the hepatitis C virus (HCV) by preventing the virus attachment to the cell surface. Similarly, ladanein is a phenolic compound from Marrubiumperegrinum L (Lamiaceae) also showed to inhibit HCV during the viral entry stage . Baicalin, a flavonoid purified from a medicinal plant, was reported to inhibit the HIV-1 infection by interfering with the interaction of HIV-1 E protein and chemokine receptors during HIV-1 entry . However, there is no clear study of flavonoids and DENV interacting during the viral entry stage. Several studies showed flavonoid inhibitions during the DENV replication stage. Fisetin, quercetin and baicalein are among the flavonoids that were reported to inhibit the DENV2 replication by targeting the viral RNA [28–30].
In this work, several computational methods were employed to study the binding potential of flavonoids to the soluble region of DENV2 E protein. For the purpose of our interest, we constructed a homology model of the DENV2 E protein isolated from Malaysia (DENV2-My). Nine flavonoids were selected for this study based on their known biological activities against viruses. We performed molecular docking method to evaluate the interaction of flavonoids and predict their binding location on the surface of the E protein. The flavonoids were intensively docked to the E proteins of the DENV2-My and the DENV2-Thai structures. Additionally, the flavonoids were also docked to the E proteins of the other members of the flavivirus (DENV3, TBEV and JEV). We also performed molecular dynamics simulations for selected E protein–flavonoid complexes of both DENV2-My and DENV2-Thai. Our main goal is to identify the consensus binding site and amino acid interactions of the flavonoids that may serve as initial data for other DENV structural studies.
2 Materials and Methods
2.1 Sequence Analyses and Structure Preparation
The primary amino acid sequence of DENV2 isolated from Malaysia; DENV2-My (GenBank ID: ACN94866.1) was retrieved from GenBank . The crystal structures and sequences of E protein for DENV2, DENV3, TBEV and JEV were obtained from Protein Data Bank (PDB) : DENV2 (PDB ID: 3J27), DENV3 (PDB ID: 1UZG), TBEV (PDB ID: 1SVB) and JEV (PDB ID: 3P54). The DENV2-My sequence was subjected to the BLAST search analysis using the NCBI BLAST database  (www.blast.nlm.nih.gov) to search for the local similarity region between the target protein and the crystal structures. The alignment between the target and template was computed using the CLUSTAL OMEGA (http://www.ebi.ac.uk/Tools/msa/clustalo/) , which aligned multiple sequences using a progressive pairwise alignment algorithm. Minor adjustment of the alignments was performed using the JALVIEW program .
2.2 Homology Modeling of DENV2-My
Homology modeling method was used to build the soluble ectodomain model of DENV2-My. The aligned sequences between the target protein (DENV2-My) and the template (DENV2-Thai: PDB ID: 3J27) were used for the model construction using the MODELLER 9v9 program . The stereochemical and structural properties of the model were evaluated using the PROCHECK program . The sequence-structure compatibility of the models was evaluated based on the Z-score using the ProSA web server . Root mean square deviation (RMSD) was computed using the DaliLite server .
2.3 Molecular Docking
High-resolution structures of the E proteins from DENV2-Thai, JEV, and TBEV were used for the molecular docking to study their potential interactions with the flavonoids. The docking results were compared among the studied structures as well as with the model structure of DENV2-My E protein.
The 2D structures of the ligands were downloaded from the PubChem server (https://pubchem.ncbi.nlm.nih.gov/) and converted to a 3D format using the SMILES Translator (http://cactus.nci.nih.gov/services/translate/) . Flavonoids chosen for the docking study are baicalin (PubChem ID: 64982), baicalein (PubChem ID: 5281605), EGCG (PubChem ID: 65064), fisetin (PubChem ID: 5281614), flavone (PubChem ID: 10680), glabranine (PubChem ID: 124049), hyperoside (PubChem ID: 5281643), ladanein (PubChem ID: 3084066) and quercetin (PubChem ID: 5280343).
Molecular docking was performed using Autodock Vina 1.1 . Autodock Tools (ADT) were utilized to prepare the input file for the receptors and ligands. Gasteiger charges were assigned, and nonpolar hydrogen atoms were merged. All torsions were allowed to rotate during the docking process. The auxiliary program AutoGrid is used to generate the grid maps for the protein. Each grid was centered at the receptors of the corresponding inhibitors. In the first stage, the blind docking method was performed on both DENV2-Thai and DENV2-My E protein structures in order to search for the common binding pocket of flavonoids. The grid dimensions used for the blind docking were covering the whole structure of the E protein. In the second docking stage, the targeted docking focused on the consensus region identified by blind dockings. The size of the grid box was set to 30 Å in each dimension. Analysis of the protein–ligand complexes was based on the binding energy score, hydrogen bond interactions and orientation of the docked compound within the binding pocket .
2.4 Molecular Dynamics Simulation
Molecular dynamics simulations of E protein–flavonoids complexes were carried out using GROMACS 4.5.5 package [43, 44] with GROMOS96 43A1 force field to describe the atoms’ interactions [45, 46]. The starting protein–ligand structures for the simulations were obtained from the docking results. The topology and parameter files for the ligands were generated using the PRODRG server . Each protein–ligand complex was placed in a triclinic box solvated with a simple point charge (SPC) type of water molecules. The distance between protein and box was set to 1.2 nm, and van der Waals cutoff was 1.0 nm. The systems were neutralized with counterions, and the salt concentration of 0.1 M NaCl was added to the system. Bond lengths involving the hydrogen atoms were constrained using LINCS algorithm, and van der Waals interactions were evaluated with a cutoff distance of 1.0 nm. Particle mesh Ewald (PME) was employed to treat long-range electrostatics interactions with a Coulomb cutoff of 1.0 nm .
A total of 50 000 steps of steepest descent minimization were applied to the solvated system to allow the solvent to adjust around the protein–ligand complex. The systems were equilibrated with NVT and NPT ensemble protocols for about 100 ps each. The temperature of the simulation system was set to 300 K. The production run of the molecular dynamics simulations was performed for a total of 20,000 ps (20 ns). The protein–ligand, solvent and ions were separately coupled to a Berendsen temperature (τT = 0.1 ps) and pressure (τP = 2.0 ps) baths  at 300 K and 1 bar, respectively. Simulations were run with 2-fs time steps. Trajectory analysis was performed using GROMACS 4.5.5 tool [43, 44].
3 Results and Discussions
3.1 Sequence and Structure of E Protein DENV2
3.2 Docking of Flavonoids
Docking binding affinity of the nine flavonoids to the E proteins of flaviviruses
Binding affinity (kcal/mol)
Although the DENV3 E protein has high sequence similarity to the DENV2 E protein, flavonoids docked to E protein of DENV3 showed different results compared to the other E proteins. Only baicalin, glabranine and hyperoside docked into the flavonoid consensus binding pocket. The other three compounds baicalein, EGCG and flavone docked to another pocket located on another E protein subunit (chain A). Fisetin, ladanein and quercetin docked to a region between the E protein dimer interface that was approximately 11 Å distant away from the flavonoid consensus pocket (Supplementary Figure S2). The explanation to this behavior is highly likely due to the different orientations of the domain I and II of the DENV3 E protein crystal structure (1UZG.pdb) .
3.3 Molecular Dynamics Simulations
Molecular dynamics simulations were performed in order to evaluate the dynamic behavior of E protein–flavonoid complexes. We selected four E protein–flavonoid complexes from both DENV2-My and DENV2-Thai based on the consistent binding mode observed in the docking results and known experimental data of the flavonoids interacting with the host cells during the virus entry. Initial starting structures of the complexes (E protein, baicalein, quercetin, EGCG and baicalin) were taken from the docking results. In addition, molecular dynamics simulations were also performed for the unbound E proteins of both DENV2-My and DENV2-Thai. All simulations were subjected to 20,000 ps molecular dynamics simulations.
Radius of gyration (Rg) of the backbone atoms describes the compactness of the E protein complexes and the unbound E protein structures. As shown in Supplementary Figure S3 (a), in the unbound E proteins, the Rg value was stabilized as early as 1500 ps while the E protein–flavonoid complexes reach stability at around 7000 ps. Rg for all complexes decreased along the simulation time, indicating an increase in the compactness that might be due to the increase in interactions between the flavonoid and the E protein. The flavonoid consensus binding pocket is located in between the two subunits of the E protein dimer. Therefore, the interactions of the flavonoids to the residues in the binding pocket brought both E protein subunits closer to each other. This observation can also be explained by the reduction in distance between the both subunits of E protein dimers during the simulations (Supplementary Figure S3 (b)).
Hydrogen bond analysis between the flavonoids and E protein of DENV2-Thai and DENV2-My
Baicalein: 6-OH—Asp98: O
Baicalein: 7-OH—Gly100: NH2
Baicalein: 7-OH—Ile4: O
Baicalein: 7-OH—Gly5: O
Baicalein: 6-OH—Val151: O
Baicalein: 1-O—Asp98: O
Baicalein: 7-OH—Gly100: O
Baicalein: 6-OH—Ile4: O
Baicalein: 7-OH—Gly5: O
Baicalein: 6-OH—Val151: O
Quercetin: 3′-OH—Asp98: O
Quercetin: 3′-OH—Gly100: NH2
Quercetin: 7-OH—Ile4: NH2
Quercetin: 4′-OH—Gly5: O
Quercetin: 7-OH—Val151: O
Quercetin: 3′-OH—Asp98: O
Quercetin: 1-O—Gly100: NH2
Quercetin: 7-OH—Ile4: O
Quercetin: 7-OH—Gly5: O
Quercetin: 5-OH—Val151: O
EGCG: 5′-OH—Asp98: NH2
EGCG: 3″-OH—Gly100: NH2
EGCG: 5″-OH—Ile4: O
EGCG: 5″-OH—Gly5: O
EGCG: 4″-OH—Val151: O
EGCG: 4″-OH—Asp98: O
EGCG: 1 = O—Gly100: NH2
EGCG: 3′-OH—Ile4: O
EGCG: 3″-OH—Gly5: O
EGCG: 7-OH—Val151: O
Baicalin: carbonyl oxygen of beta-D-glucopyranose acid—Asn103: NH2
Baicalin: 4-OH of beta-D-glucopyranose acid—Gly5: O
Baicalin: carbonyl oxygen of beta-D-glucopyranose acid—Asn153: NH2
Baicalin: 4-OH of beta-D-glucopyranose acid—Gly152: O
Baicalin: carbonyl oxygen of beta-D-glucopyranose acid—Asp154: NH2
Baicalin: 6-OH—Asp98: O
Baicalin: 2-OH of beta-D-glucopyranose acid—Arg99: NH2
Baicalin: 4 = O—Lys247: NH2
Baicalin: 6-OH—Ser7: O
Baicalin: 4-OH of beta-D-glucopyranose acid—Asn153: NH2
3.4 DENV2 E Protein–Baicalein Complex
In traditional Chinese medicine, Scutellaria baicalensis is one of the most popular herbs that are being used to treat common cold and other virus infections. Baicalein (5, 6, 7-trihydroxyflavone) is one of the flavones isolated from Scutellaria reported to show potent inhibitory activity against the influenza A virus during the viral entry [52, 53]. We showed in the previous section that baicalein docked to the same binding pocket of the DENV2, TBEV and JEV E proteins. The docking results provided high binding affinity to the E protein of the DENV2-Thai and DENV2-My E proteins (Table 1).
3.5 DENV2 E Protein–Quercetin Complex
Quercetin is a phenolic type of flavonoids commonly found in many fruits and vegetables. Although no specific study relates this compound with DENV E protein, however, there were reports showed that quercetin caused concentration-dependent reduction in viral infectivity in other type of viruses . This suggests potential activity of quercetin during the viral entry. In the molecular docking section, the result showed quercetin consistently docked to the consensus binding pocket of E proteins of DENV2, JE and TBEV except for the DENV3. Therefore, the E protein–quercetin complexes of DENV2-My and DENV2-Thai were further investigated using the molecular dynamics simulations.
3.6 DENV2 E Protein–EGCG Complex
EGCG (also known as epigallocatechin-3-gallate), a compound present in green tea extract, was shown to be a potent HCV entry inhibitor that blocked the virus attachment to the host cells . Similar to other selected complexes, EGCG docked to the same flavonoid consensus pocket; therefore, we selected the E protein–EGCG complexes for further analysis using the molecular dynamics simulations.
3.7 DENV2 E Protein–Baicalin Complex
Baicalin (7-glucuronic acid,5, 6-dihydroxyflavone) is also one of the flavones that can be extracted from the Scutellaria species. Structurally, baicalin has an extra bulky glucuronic acid moiety on its 7-OH baicalein moiety. This compound has similar inhibitory activities as the baicalein against influenza A virus . In addition, baicalin was also shown to inhibit HIV-1 entry . Docking of baicalin to the E proteins produced the highest binding affinity among the nine studied flavonoids and higher in the DENV2 E proteins compared to the other E proteins (Table 1).
In this study, docking and molecular dynamic simulations showed consistent results. Moreover, the DENV2-My E protein structure, which is a homology model, produced highly similar results as the DENV2-Thai E protein structure. These data describe that the constructed E protein model is in good agreement with the template structure. Consistently, a recent study constructed a DENV1 E protein model based on the DENV2 E protein structure successfully developed a lead inhibitor. This potential DENV1 entry inhibitor was shown to potently inhibit the viral membrane fusion .
The flavonoid consensus pocket observed in this study is proposed as an alternative binding pocket. This pocket is present during the E protein dimeric state and consists of residues known to form the fusion loop, which are conserved among the flaviviruses. This conserved region could be useful for the antiviral development targeting the four serotype of DENV. However, several studies reported the presence of other cavities on the surface of the E protein that exists during different states of the virus. For example, Yennamali et al. [18, 56] identified potential inhibitor binding pockets using another crystal structure of DENV2 E protein. Studies using molecular dynamics simulations also reported different pockets on the surface of the DENV2 E protein dimeric prefusion structure that were observed during different histidine protonation states [57, 58].
The flavonoids that were observed to dock into the consensus pocket could be used as the potential leads to design the DENV entry inhibitor. Residues Ile4, Gly5, Asp98, Gly100 and Val151 located in the potential binding pocket were consistently observed to form interactions with the E proteins of DENV2. These residues can be used as the references for the bioassay and mutagenesis studies in order to validate this computational prediction. Despite the observed binding consensus between DENV2-My and DENV2-Thai E proteins, there are possibilities that these flavonoids could also bind to different E protein cavities that exist during the rearrangement process or they could interact with other non-structural proteins at the later stage of the viral life cycle. Such binding events could be addressed more specifically in the future studies. We also note that some of these flavonoids are categorized as amphiphilic phytochemical, which may affect the host membrane lipid perturbations rather than specifically bind to the target protein .
We have identified an alternative binding pocket on the surface of the DENV2 E protein based on the consensus binding site of the flavonoids. We describe here the important amino acid residues of the E protein that may interact with the flavonoids. For the DENV2-Thai and DENV2-My E proteins, Ile4, Gly5, Asp98, Gly100 and Val151 residues were identified to form interactions with baicalein, quercetin and EGCG. These data are useful for the future mutagenesis and structural studies of DENV E protein and the development of the DENV entry inhibitor.
The authors thank Dr. Ozlem Demir for the helpful discussion and careful reading of the manuscript. This research was supported by the Universiti Teknologi MARA (UiTM) Dana Cluster 600-RMI/DANA 5/3/CG (2/2012). NAI was funded by the scholarship from the Ministry of Higher Education (Malaysia) through the MyBrain15 program. We thank Faculty of Pharmacy, UiTM Puncak Alam Campus for providing the computing facilities in the Bioinformatics Lab, as well as Research Management Institute (RMI), UiTM and Ministry of Science and Technology Malaysia (MOSTI) for the financial and administrative support.
Compliance with Ethical Standards
Conflict of interest
The authors declare that they have no conflict of interest.
This study does not contain any studies with human participants or animals performed by any of the authors.