In silico study on baicalein and baicalin as inhibitors of dengue virus replication

Abstract

The dengue virus (DENV) is an important human arbovirus that belongs to the Flaviviridae. Currently, there is no vaccine or effective antiviral agent against DENV. Therefore, finding an efficient antiviral agent against this virus is crucial. We have previously reported the anti-DENV activity of two flavonoids, namely, baicalein and baicalin, against different stages of the virus replication cycle in Vero cells. In this study, we aimed to predict the possible interactions between viral proteins that are important for DENV replication and these two flavonoids as potential candidates for anti-DENV drug discovery with known in vitro anti-DENV activity. In this study, the interactions between compounds of interest and three important proteins of DENV were predicted using appropriate software. Moreover, the binding energy between these compounds and selected proteins was calculated as one of the main criteria for molecular docking studies. The results showed that both compounds of interest, as ligands, can bind with chosen viral proteins, as receptors, through hydrogen bonding and other interactions such as pi–pi interactions, pi–sigma interactions and pi–cation interactions. The obtained data showed significant affinity between the tested compounds and the NS2B–NS3 protease of DENV. Therefore, an in vitro anti-protease assay was conducted and the results showed significant anti-DENV protease activity for both compounds, especially baicalein. In conclusion, our obtained data showed that both tested compounds can affect DENV intracellular replication and internalization. However, these results support our previous findings from in vitro studies and encourage us to further studies towards finding the mechanism of action of these compounds.

---

Introduction

The dengue virus (DENV) is a member of the Flaviviridae family with four distinct serotypes^1,2 and recently a fifth serotype of DENV has also been reported in Sarawak, Malaysia.³ DENV causes a range of clinical manifestations including dengue fever, dengue haemorrhagic fever (DHF) and dengue shock syndrome (DSS). DENV, as an important mosquito-borne virus, is transmitted by Aedes aegypti and Aedes albopictus. Dengue infection has become a serious problem and health issue in more than 100 countries around the world, mainly in tropical and subtropical regions. However, numerous cases of dengue infection have been reported recently from Pakistan, Saudi Arabia, Sudan, Yemen and Madagascar as newly infected regions.⁴ Until now, there has been no licensed vaccine or effective antiviral drug available against dengue infection. Therefore, many investigators are trying to find effective compounds with significant anti-DENV activity.

DENV has a linear single-stranded RNA as its genome, which encodes structural and non-structural proteins including envelope (E), capsid (C) and PreM as structural proteins and NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5 as non-structural proteins.⁴

NS3 protein is a serine protease, as well as an RNA helicase and RTPase/NTPase. The N-terminal domain of NS3 (aa 1–169) is a chymotrypsin-like serine protease that cleaves the viral polyprotein in both cis and trans forms.^5,6 NS3 protease needs NS2B as a cofactor for its activity.^7–9 NS2B is a 14 kDa fundamental membrane protein with three domains: a central area with 47 amino acids (spanning amino acids 49–96), which acts as an essential cofactor of the NS3 protease, and two trans-membrane segments located at the N and C termini¹⁰ (Fig. 1A). The flavivirus NS3 protein is not soluble and cannot be an active catalyst as a protease in vitro, which means that this protein needs NS2B as a cofactor for proper folding, which must be provided in either cis¹¹ or trans form.^12–14 The NS3 protease residues consist of six β-strands ordered into two β-barrels. Between these two β-barrels the catalytic triad (HIS51, ASP75 and SER135) can be found, of which the activity depends on the presence of the NS2B cofactor. NS2B acts as a cofactor by wrapping around the NS3 protease domain to create the active site. The rest of the NS3 residues (180–618) form the three subdomains of the DENV helicase. A six-stranded parallel β-sheet surrounded by four α-helices constitutes subdomains I and II, and subdomain III is composed of four α-helices surrounded by three shorter α-helices and two antiparallel β-strands. Several drugs targeting the NS3 protease of hepatitis C virus (HCV) have received U.S. Food and Drug Administration (FDA) approval, and a few promising candidates are in clinical trials.¹⁵ This could further stimulate the development of antiviral drugs targeting the flavivirus NS2B–NS3 protein.

Fig. 1 DENV NS3/NS2B protein vs. baicalein and baicalin. (A) DENV NS3/NS2B (2FOM) protein structure. (B) DENV NS3/NS2B (2FOM) protein with baicalein with nine different possibilities of binding sites. (C) DENV NS3/NS2B (2FOM) protein with baicalin with nine different possibilities of binding sites.

DENV E protein, as a vital protein, is important for the initial attachment of viral particles to host cell receptors. Therefore, any compound with activity against E protein can be considered for further investigation towards anti-DENV drug discovery as well. There are some cell surface molecules which could interact with DENV E protein such as ICAM3-grabbing non-integrin,¹⁶ 2CD209,¹⁷ Rab 5,¹⁸ GRP 78,¹⁹ and mannose receptor,²⁰ which have been shown to be important for mediating virus binding and entry. E protein consists of three structural domains (Fig. 3A).²¹ Domain III (DIII) of E protein is responsible for receptor binding, whereas domain II (II) of E protein facilitates insertion of the virus into a host cell membrane following conformational changes in the structure of E protein because of changes in pH.²² Dengue virus E protein adopts two main conformations: dimeric and trimeric conformations. The dimeric form can be found in the mature virion and the trimeric form can be seen while the virion is still in the immature stage and during the adoption of the fusogenic conformation inside the endosome.^21,23–25 Antiparallel dimers lay flat on the surface of the virion in the dimeric conformation and, in order for the virion to adopt a spiky appearance, trimers point away from the viral membrane in the form of conical rods. TRP101, LEU107 and PHE108 are three highly conserved amino acid residues, which are located in the fusion loop that is exposed at the tip.^23,24 Although the dimeric conformation is reversible, the transition to the trimeric conformation is irreversible.

The conformational changes leading to the formation of the E protein trimer are thought to be initiated when low pH in the endosome is sensed by five histidine residues, which are conserved among flaviviruses. These histidines are distributed in the three domains of the protein and in the stem region.^24,26–28 Histidine 323, which is located at the DI–DIII interface, functions as a pH sensor and also stabilizes the postfusion conformation of the trimer. This stabilization might be due to the presence of a salt bridge between HIS323 and GLU373.²⁷ When HIS323 transfers a proton to a molecule, the salt bridge separates and weakens the DI–DIII interface.^24,27 There are two widely conserved histidines located in E glycoprotein (HIS209 and HIS7) that may also act as pH sensors. Hydrophobic interactions between E glycoprotein and M protein stabilize E glycoprotein and are important for conformational changes. When these histidines are in their protonated forms, E glycoprotein dissociates from M protein, which allows the formation of the E trimer.²⁶ ASN153 in domain I and ASN67 in domain II are other important structural characteristics of E protein.^25,26 The ASN153 residue interacts with the dimeric form of the protein, which facilitates DENV infectivity, and is conserved among flaviviruses. The ASN67 residue is only found in DENV strains and it is thought that its glycosylation is important for virus assembly or virus release.^29,30

DENV NS5 protein, which has a molecular weight of 104 kDa with 900 residues, is the largest DENV protein. It is an RNA-dependent RNA polymerase at the C-terminal end (residues 320–900). NS5 protein is the most highly conserved DENV protein and shares almost 67% amino acid sequence identity across the four DENV serotypes. According to structural and biochemical studies, NS5 protein has three functional domains. The domain spanning amino acid residues 1 to 296, which is the N-terminal S-adenosylmethionine methyltransferase (MTase) domain, has been expressed as a soluble active enzyme. A central S-adenosylmethionine-dependent MTase core structure sits within a “cradle” shaped by N-terminal and C-terminal subdomains. Interestingly, a segment of 20 amino acids, which is strictly conserved in all flaviviruses, also lies between amino acid residues 320 and 368 of NS5 (Fig. 2A). This region of NS5, which is involved in binding β-importin,^31,32 is also thought to interact with NS3.³² Furthermore, the association between NS5 and NS3 is expected to modulate their respective enzymatic activities. Indeed, NS5 was reported to stimulate the nucleotide triphosphatase and RNA triphosphatase activities³³ of NS3. The C-terminal region of NS5 contains five amino acid sequence motifs that form the signature of RNA-dependent RNA polymerases (RdRps).³⁴ In conjunction with other viral proteins and unidentified host cell proteins, the NS5 polymerase domain is responsible for synthesizing a transient double-stranded replicative RNA intermediate. Therefore, finding and/or designing compounds that can bind and interact with the above-mentioned viral proteins efficiently is crucial in antiviral drug discovery. On the other hand, there are numerous reports on the antiviral activity of different natural products.^35–40 Flavonoids, which are polyphenolic natural products that are found mainly in plants, are well known due to their different biological properties including antiviral activity. We have previously reported some flavonoids with significant anti-DENV activity such as quercetin, baicalein, baicalin and fisetin.^38,41 However, baicalin and baicalein exhibited more potent in vitro anti-DENV activities compared with fisetin and quercetin, with higher selectivity indices based on our previous findings.^38,41 Also, a series of dialkylated flavanones and chartaceones isolated from Cryptocarya chartacea Kosterm., which is a plant from the Lauraceae family, demonstrated inhibitory activity against DENV NS5 RNA-dependent RNA polymerase, with IC₅₀ values ranging from 1.8 to 4.2 μM.⁴² In addition, some secondary metabolites isolated from plants such as flavonoids, chalcone derivatives, and biflavonoids showed inhibitory activity against DENV-2 NS2B–NS3 serine protease.^43,44 4-Hydroxypanduratin A and panduratin A isolated from Boesenbergia rotunda (L.) Mansf., which is a species that belongs to the Zingiberaceae, were characterized as competitive inhibitors of DENV-2 NS2B–NS3 (Ki = 21 and 25 μM, respectively), and pinostrobin and cardamomin were shown to be non-competitive inhibitors.⁴³ The allosteric pocket of DENV-2 NS2B–NS3 protease, which is close to its catalytic triad, has been shown to be a promising drug target, although few non-competitive inhibitors of dengue virus serine protease have been discovered so far.^30,45,46 In this article, we present data from molecular docking studies of baicalein and baicalin with NS2B–NS3 protease, NS5 and E protein of DENV-2. However, the antiviral activity of baicalein and baicalin against DENV has been reported by the same laboratory^38,40 and this study aims to add more information and details about the probable mechanism of action of these two compounds via in silico study.

Fig. 2 DENV NS5 protein vs. baicalein and baicalin. (A) DENV NS5 (2J7U) protein structure. (B) DENV NS5 (2J7U) protein with baicalein with nine different possibilities of binding sites. (C) DENV NS5 (2J7U) protein with baicalin with nine different possibilities of binding sites.

Fig. 3 DENV E protein vs. baicalein and baicalin. (A) DENV E (1OKE) protein structure. (B) DENV E (1OKE) protein with baicalein with nine different possibilities of binding sites. (C) DENV E (1OKE) protein with baicalin with nine different possibilities of binding sites.

Results and discussion

In this work, we used baicalein and baicalin, which are two flavonoids active against three vital DENV proteins, namely, NS3–NS2B, NS5 and E proteins, for molecular docking experiments using AutoDock Vina. The results of docking studies show that both baicalein and baicalin interact with NS3–NS2B, E and NS5 proteins. AutoDock Vina ranked the results for interaction according to the affinity energy (Tables 2–4). Baicalein exhibits H-bonding and pi–pi interactions with NS3–NS2B (2FOM) dengue protease (Table 2) (Fig. 4). Baicalein bonds to the active site of NS3–NS2B protein as well as baicalin through close contact via the LYS74, LEU76, ASN152, TRP83, LEU149, GLY148 and GLU88 NS3/NS2B residues. H-bonding between baicalein and NS3–NS2B can be seen between ASN152-HD21 1 and LEU149-HN1 (Fig. 4). Besides hydrogen bonding, interestingly, baicalein exhibits a pi–pi interaction with TRP83 (Table 2). Fig. 1B shows the nine different conformations of baicalein with different sites of NS2B/NS3 protein, which were ranked by AutoDock Vina. The interaction affinity of baicalein with NS3–NS2B is equal to −7.5 kcal mol⁻¹. Baicalin, as a metabolite of baicalein, has a stronger interaction affinity equal to −8.0 kcal mol⁻¹. Baicalin shows close contact with the GLU88, TRP86, GLY148, LEU149, TRP83, LEU76 and ASN152 NS3/NS2B residues. The hydrogen bonds between baicalin and NS3–NS2B are between the TRP86-HE1 1, LEU128-HN1, TYR161-HN1, ARG54-HH22 1 and GLY153-HN1 residues (Table 3) (Fig. 7). There is a pi–pi interaction between the compound and TYR161, HIS51 and TYR150 (Table 5). The nine different conformations of baicalin with NS3/NS2B are shown in Fig. 1C.

Fig. 4 Interaction of baicalein with DENV NS3/NS2B protein. The interaction between NS3/NS2B (2FOM) protein and baicalein. (A) 3D scheme of the interaction of baicalein with 2FOM, where a hydrogen bond can be found between baicalein and the 2FOM residues ASN152 and LEU149, which are active sites of 2FOM protein. A pi–cation interaction can be observed with residue LYS74. (B) 2D scheme of the interaction of baicalein with NS3/NS2B (2FOM), including H-bonding, pi–cation interaction and close contact.

NS5 protein, as one of the important DENV proteins with RNA polymerase activity, was also chosen for our molecular docking study. The results of the study of molecular docking between NS5 and baicalein show close contact via the GLN350, VAL579, VAL358, ASP538, THR539, TRP302, ARG598, VAL577, PHE354, PHE398, LYS401, VAL603, GLY604, ASN492, GLN602, ARG481, ASN609, ASP663, TYR606 and HIS796 residues. Strong hydrogen bonding was observed between baicalein and NS5 residues such as ASN492:HD21 1, GLY604:HN1, TYR606:HN1, ASN609:HD21 1, ASP663:HN1 and ARG581:HE1 (Table 3) (Fig. 5). Also, there are pi–pi interactions between baicalein and NS5 via TRP477 and HIS229 (Table 5). The different conformations of baicalein with different sites of NS5 are shown in Fig. 2B. Baicalin and NS5 show close contact via the THR539, VAL358, TRP302, ARG598, SER600, LYS575, GLN602, TRP477, ARG481, VAL450, LYS357, VAL577, GLN360, SER710, HIS798, CYS709, SER661, ASP663, TYR606, THR605, ASN609, GLY601, GLY536, ASP538, ASN405, ARG792, THR404, ASN492, LYS401, VAL402 and PHE398 residues. The binding affinity is −8.6 kcal mol⁻¹ and hydrogen bonding was observed between ASN609:ND2 1, SER710:OG 1, GLY604:N1, THR605:N1, TYR606:N1, THR604:OG1 1 and LYS401:NZ1 (Table 3) (Fig. 8). Fig. 2C shows the different conformations of baicalin with different sites of NS5 protein.

Fig. 5 Interaction of baicalein with DENV NS5 protein. The interaction between NS5 (2J7U) protein and baicalein. (A) 3D scheme of the interaction of baicalein with 2J7U, where a hydrogen bond can be found between baicalein and the 2J7U residues ARG581 and PRO684. A pi–pi interaction is observed with the residue HIS292. (B) 2D scheme of the interaction of baicalein with NS5 (2J7U), including H-bonding, pi–pi interaction and close contact.

Fig. 6 Interaction of baicalein with DENV E protein. (A) 3D scheme of the interaction of baicalein with E protein, with two pi–pi interactions between baicalein and the E protein residue HIS27. There is an H-bond at the E protein residue THR200. (B) 2D scheme of the interaction of baicalein with E (1OKE) protein, including H-bonding, pi–pi interaction and close contact.

Fig. 7 Interaction of baicalin with DENV NS3/NS2B protein. (A) 3D scheme of the interaction of baicalin with 2FOM, where we can see hydrogen bonding between baicalin and the 2FOM residues ASP75 and GLY153 as active sites of 2FOM protein. There is a pi–sigma interaction with the residues HIS51 and LEU128; also, a pi–pi interaction can be observed at the TYR161 residue. (B) 2D scheme of the interaction of baicalin with NS3/NS2B (2FOM), including H-bonding, pi–sigma interaction, pi–pi interaction and close contact.

Fig. 8 Interaction of baicalin with DENV NS5 protein. (A) 3D scheme of the interaction of baicalin with 2J7U, where there are hydrogen bonds between baicalin and the 2J7U residues ASN609, ASP663 and SER661. (B) 2D scheme of the interaction of baicalin with NS5 (2J7U), including H-bonding and close contact.

Another important protein that we studied was DENV E protein. Baicalein interacted with E protein with an affinity of −7.1 kcal mol⁻¹. Close contact between baicalein and DENV E protein was observed at domain I and domain III (Fig. 3B). Results from AutoDock Vina and Discovery Studio show that the close contact occurred between baicalein and these residues of DENV E protein: ARG2, SER29, THR359, PRO364, VAL356, ASN366, ASP10, ARG9, ILE4, VAL24, VAL31, ASP22, LYS284, VAL354, ARG188, THR189, GLY190, PRO187, LEU135, PHE193, PRO132, THR280, VAL321, HIS27, ILE48, GLY28, THR359, SER363, VAL358, ALA267, VAL208, HIS209, MET196, GLY318 and GLN316. Hydrogen bonding between E protein and baicalein could be seen via ARG9:HH11 1, LYS284:HZ1 1, GLY190:HN1, THR280:HG1 1, SER29:HG1, THR359:HN1 and THR359:HG1 (Table 4) (Fig. 6). Baicalin displays better binding affinity with E protein compared with baicalein, with an affinity equal to −8.0 kcal mol⁻¹. The close contact that baicalin makes with E protein residues occurs via THR182, ARG288, TRP20, MET289, ASP10, VAL24, VAL31, ILE4, VAL354, ASN355, LYS334, GLU269, THR268, THR280, GLU314, THR315, ALA313, LYS310, VAL321, ARG323, LYS88, LYS234, VAL91, CYS92, HIS94, GLU85, ILE232, GLN233, LEU351, ASN355, ILE6, GLU360, PRO364, ARG2, GLY28, SER29, ILE357, VAL356, THR359, VAL358, LYS393, GLY318, LEU207, ALA267, VAL208, HIS209, LYS394, MET1, SER363, HIS27, LYS47 and ASN366. H-bonding between baicalin and E protein was observed between the molecule and the following residues: ARG288:HE1, LYS334:HZ1 1, ASN355:HN1, CYS92:HN1, LYS234:HN1, LYS334:HZ1 1, THR359:HN1, GLU360:HN1, LYS394:HN1, THR359:HG1 1, THR280:HG1 1, ARG323:HH12 1 and ASN366:HN1 (Table 4) (Fig. 9). There is a pi–pi interaction between baicalin and TRP20, which shows strong binding between the compound and E protein (Table 5). Besides the pi–pi interaction, there is a pi–cation interaction between baicalin and the ARG2 residue (Table 5). Nine different conformations, which were ranked by AutoDock Vina, between baicalin and E protein are shown in Fig. 3C.

Fig. 9 Interaction of baicalin with DENV E protein. (A) 3D scheme of the interaction of baicalin with E protein, with a pi–pi interaction between baicalin and the E protein residue TRP20. There are H-bonds at the E protein residues ASP290, ARG288 and GLY18. (B) 2D scheme of the interaction of baicalin with E (1OKE) protein, including H-bonding, pi–pi interaction and close contact.

All the binding energies between the two compounds and the three dengue proteins were calculated using Discovery Studio 2.5 and are shown in Table 1.

Table 1 Binding energies for the interaction of baicalein and baicalin with each protein obtained using Discovery Studio 2.5. The binding energies for both compounds were calculated using Discovery Studio in order to determine the energy required for the interaction between the protein and ligands

Name	Potential energy (kcal mol^−1)	van der Waals energy (kcal mol^−1)	Electrostatic energy (kcal mol^−1)	Interaction energy (kcal mol^−1)	VDW interaction energy (kcal mol^−1)	Electrostatic interaction energy (kcal mol^−1)
2FOM-baicalein	−850.3	−147.5	−702.8	−850.3	−147.5	−702.8
2FOM-baicalin	−900.6	−172.4	−728.1	−900.6	−172.4	−728.1
2J7U-baicalein	−807.9	−46.1	−761.8	−807.9	−46.1	−761.8
2J7U-baicalin	−954.5	−170.2	−784.3	−954.5	−170.2	−784.3
1OKE-baicalein	−549.9	−27.6	−522.3	−549.9	−27.6	−522.3
1OKE-baicalin	−639.1	−14.7	−621.3	−636.1	−14.7	−621.5

---

As discussed, baicalin and baicalein have both been reported to be in vitro anti-DENV agents and it has been proven that both of these flavone compounds have a direct effect on the virus replication cycle using a post-infection assay. Despite the post-infection effect of baicalin and baicalein, these two compounds displayed neutralizing activity against extracellular DENV particles, which could be due to their interactions with DENV E protein.^38,40 Our previous in vitro antiviral experimental data showed that baicalein and baicalin inhibited virus replication with IC₅₀ values of 6.46 and 13.5 μg mL⁻¹, respectively, when they were added to infected Vero cells following virus internalization. It was also shown that baicalein and baicalin showed activity against binding of the virus to cell receptors during the virus adsorption time, with IC₅₀ values of 7.14 and 18.7 μg mL⁻¹, respectively, in addition to their potent neutralizing activity against extracellular DENV particles. All the above-mentioned data could support a hypothesis about the interactions of baicalin and baicalein with different DENV proteins including E protein, as a structural protein responsible for the binding of the virus to the host cell, as well as NS2B–NS3 protease and NS5 protein, as two important non-structural proteins with critical roles in intracellular virus replication. Therefore, these interesting results encouraged us to find out more about the mechanisms of action of baicalin and baicalein against important DENV proteins using a molecular docking study before we further proceeded to an in vitro mechanistic study.

All DENV proteases have the ability to perform cleavage at a number of sites, such as NS2A/NS2B, NS2B/NS3, NS3/NS4A and NS4B/NS5. They also involve in cleaving upstream of the signal sequence, at the junction of C-prM and NS4A/NS4B, and between NS2A and NS3 itself.⁴⁷ The replication and polyprotein processing of the flavivirus are handled by NS3 protease.⁴⁸ NS3 protease is a serine protease, which has a catalytic triad that comprises the residues histidine 51 (HIS51), aspartic acid 75 (ASP75), and serine 135 (SER135).⁹ The protease activity of NS3 is handled by its N-terminal region, and the enzymatic functions of nucleoside triphosphatase and RNA helicase activity are handled by the C-terminal two-thirds of NS3.⁴⁹ It has been mentioned that the catalytic activity of NS3 depends upon activation by the NS2B cofactor.⁵⁰ The NS2B/NS3pro protease complex is a better option, due to it being a more structurally relevant target, than NS3 protein alone for anti-DENV drug discovery. However, scientists have yet to find the mechanism by which NS2B contributes to the activation of NS3 protein. An explanation of the necessary role of NS2B will lead to the discovery and design of new drugs against dengue infection.⁷ Both baicalin and baicalein have displayed good binding affinity to the NS3–NS2B complex, with docking energies of −8.0 kcal mol⁻¹ for baicalin and −7.5 kcal mol⁻¹ for baicalein, where baicalin displayed better binding affinity compared with baicalein. Baicalin exhibited a strong binding interaction with the active site of NS3–NS2B protein, where H-bonding interaction with ASP75 could be seen. There is also a pi–sigma interaction with HIS51, which is another important residue of the active site (Table 5). Baicalein, as well as baicalin, interacted with the NS3–NS2B active site, which is the SER135 residue. All the hydrogen bonds with baicalin and baicalein were formed with lengths of 1.8 Å to 2.4 Å, which means that the hydrogen bonding is moderate or quite strong (Table 2).

Table 2 Hydrogen bonding of baicalein and baicalin with DENV NS3/NS2B protein (2FOM). This table shows the H-bonding interaction of baicalein and baicalin with different residues of NS3/NS2B protein at different affinities ranked by AutoDock Vina 1.5.6. The third column shows the length of each H-bond between the molecule and residue. The last column shows the affinity energy, which was ranked by AutoDock Vina 1.5.6

Compound	Interaction	Distance (Å)	Affinity (kcal mol^−1)
Baicalein	Baicalein:UNK1:H9–NS3–NS2B:GLU88:OE2	2.4	−7.5
	NS3NS2B:LEU149:HN–baicalein:UNK1:O17	1.8	−7.2
	NS3NS2B:ASN152:HD21–baicalein:UNK1:O18	2.1	−7.2
	NS3NS2B:ARG55:HH21–baicalein:UNK1:O20	2.3	−7.1
	NS3NS2B:VAL59:HN–baicalein:UNK1:O20	2.3	−7.1
	Baicalein:UNK1:H8–NS3NS2B:ASP58:OD1	2.4	−7.0
	Baicalein:UNK1:H8–NS3NS2B:GLY151:O	2.5	−7.0
	Baicalein:UNK1:H9–NS3NS2B:SER135:OG	2.2	−7.0
	Baicalein:UNK1:H10–NS3NS2B:TYR150:OH	1.9	−7.0
	Baicalein:UNK1:H8–NS3NS2B:PHE130:O	2.0	−7.0
	NS3NS2B:GLU90:HN–baicalein:UNK1:O19	2.4	−6.9
	NS3NS2B:ASN105:HD21–baicalein:UNK1:O10	2.1	−6.9
	NS3NS2B:ARG107:HH21–baicalein:UNK1:O20	2.2	−6.9
	Baicalein:UNK1:H9–NS3NS2B:GLY103:O	1.8	−6.9
	NS3NS2B:TRP83:HE1–baicalein:UNK1:O17	2.2	−6.6
Baicalin	NS3NS2B:TRP83:HE1–baicalin:UNK1:O30	2.3	−8.0
	Baicalin:UNK1:H18–NS3NS2B:GLU88:OE2	1.9	−8.0
	Baicalin:UNK1:H6–NS3NS2B:VAL72:O	2.1	−7.9
	Baicalin:UNK1:H17–NS3NS2B:GLY151:O	1.8	−7.9
	NS3NS2B:TRP83:HE1–baicalein:UNK1:O30	2.0	−7.8
	Baicalin:UNK1:H6–NS3NS2B:ASN167:O	2.2	−7.8
	Baicalin:UNK1:H17–NS3NS2B:LEU85:O	2.3	−7.8
	NS3NS2B:SER127:HG–baicalin:UNK1:O9	2.4	−7.8
	NS3NS2B:LEU128:HN–baicalin:UNK1:O8	2.0	−7.8
	NS3NS2B:TYR161:HN–baicalin:UNK1:O10	1.8	−7.8
	Baicalin:UNK1:H17–NS3NS2B:PHE130:O	2.5	−7.8
	Baicalin:UNK1:H7–NS3NS2B:LYS73:O	1.8	−7.8
	NS3NS2B:ARG54:HH22–baicalin:UNK1:O9	2.1	−7.8
	NS3NS2B:GLY153:HN–baicalin:UNK1:O10	2.2	−7.7
	Baicalin:UNK1:H17–NS3NS2B:LYS73:O	2.4	−7.7
	Baicalin:UNK1:H18–NS3NS2B:ASN152:OD1	2.4	−7.7
	Baicalin:UNK1:H17–NS3NS2B:TYR150:OH	2.0	−7.6
	Baicalin:UNK1:H6–NS3NS2B:ASP75:OD1	2.4	−7.4

---

NS5 protein, which is known to be an RNA-dependent RNA polymerase (RdRp), is responsible for the replication of the DENV positive-strand RNA genome in an asymmetric and semi-conservative process in which the anti-genome is only present in a double-stranded RNA replication intermediate.⁴⁷ It has been proven by experiment that NS5 facilitates NS3's RNA 5′-triphosphatase (5′-RTPase) and NTPase activities,^51,52,54 and NS3 also stimulates NS5's guanylyltransferase (Gtase) activity.^53,55 NS5 plays a very important role in DENV replication by activating some important enzymes, which makes NS5 a good candidate for potential antiviral drug discovery. Our obtained data show that there are several hydrogen bonds, interactions and close contacts between NS5 and both compounds of interest. However, baicalin displays better interaction, with higher affinity equal to −8.6 kcal mol⁻¹, compared with baicalein, which has an affinity equal to −7.3 kcal mol⁻¹. The binding distance for both compounds in H-bonds ranges from 1.8 to 2.4 Å (Table 3). Furthermore, there are pi–cation interactions and pi–sigma interactions in some of the binding, which makes the binding of these compounds to the protein stronger (Table 5). These data support our previous findings regarding the intracellular antiviral activity of these two compounds against DENV-2 after internalization of the virus into Vero cells.^51,52 However, further investigation is needed to reveal the real biological mechanism of action against the in vitro replication of DENV due to treatment with baicalein and baicalin.

Table 3 Hydrogen bonding of baicalein and baicalin with DENV NS5 protein (2J7U). This table shows the H-bonding interaction of baicalein and baicalin with different residues of NS5 protein at different affinities ranked by AutoDock Vina 1.5.6. The third column shows the length of each H-bond between the molecule and residue. The last column shows the affinity energy, which was ranked by AutoDock Vina 1.5.6

Compound	Interaction	Distance (Å)	Affinity (kcal mol^−1)
Baicalein	NS5:ASN492:HD21–baicalein:UNK1:O20	2.0	−7.3
	NS5:GLY604:HN–baicalein:UNK1:O17	2.1	−7.3
	NS5:THR605:HN–baicalein:UNK1:O18	2.4	−7.3
	NS5:TYR606:HN–baicalein:UNK1:O18	2.4	−7.3
	NS5:TRP477:HE1–baicalein:UNK1:O19	2.4	−7.3
	NS5:GLN602:HE21–baicalein:UNK1:O18	2.3	−7.3
	Baicalein:UNK1:H9–NS5:VAL450:O	2.4	−7.3
	NS5:ASN609:HD21–baicalein:UNK1:O20	2.1	−7.0
	NS5:ASP663:HN–baicalein:UNK1:O17	1.9	−7.0
	Baicalein:UNK1:H8–NS5:ARG598:O	2.4	−6.6
	NS5:ARG581:HE–baicalein:UNK1:O17	1.9	−6.5
Baicalin	NS5:ARG481:HH22–baicalin:UNK1:O9	2.4	−8.6
	NS5:GLN602:HE21–baicalin:UNK1:O9	2.2	−8.6
	Baicalin:UNK1:H18–NS5:SER661:O	2.5	−8.4
	NS5:ASN609:HD21–baicalin:UNK1:O30	1.9	−8.4
	NS5:ASP663:HN–baicalin:UNK1:O31	2.0	−8.4
	Baicalin:UNK1:H6–NS5:CYS709:O	2.1	−8.2
	Baicalin:UNK1:H17–NS5:SER661:O	2.5	−8.2
	NS5:TRP477:HE1–baicalin:UNK1:O9	2.2	−7.9
	NS5:GLY601:HN–baicalin:UNK1:O2	2.5	−7.9
	NS5:ASP538:HN–baicalin:UNK1:O31	2.1	−7.8
	Baicalin:UNK1:H7–NS5:VAL450:O	2.5	−7.7
	NS5:SER600:HN–baicalin:UNK1:O32	2.4	−7.7
	Baicalin:UNK1:H7–NS5:SER600:O	2.5	−7.7
	NS5:GLY604:HN–baicalin:UNK1:O8	2.3	−7.7
	NS5:THR605:HN–baicalin:UNK1:O8	2.5	−7.7
	NS5:TYR606:HN–baicalin:UNK1:O8	1.8	−7.7
	Baicalin:UNK1:H7–NS5:ASN405:O	2.3	−7.5
	Baicalin:UNK1:H8–NS5:THR404:OG1	2.4	−7.5
	NS5:LYS401:HZ3–baicalin:UNK1:O9	2.3	−7.5
	NS5:ASN492:HD21–baicalin:UNK1:O31	2.1	−7.5
	NS5:ARG792:HH12–baicalin:UNK1:O10	2.4	−7.5

---

Envelope protein is one of the critical DENV proteins, which facilitates viral infection within host cells, as well as virus internalization. This study showed that most of the binding interactions between the compounds of interest and E protein could have occurred at domain III and domain I of this protein. Domain III is responsible for binding to receptors and domain I acts as a small hairpin to rotate between domain II and domain III and plays a role in virus entry and fusion.²³ However, with baicalin there is an interaction with domain II, which is a hydrophobic domain that facilitates virus internalization (Fig. 3C). Previous studies have proven the effects of baicalin and baicalein on extracellular DENV particles, as well as virus entry.^38,40 The molecular docking study provides supportive data about a possible mechanism for interfering with virus internalization based on interaction between the compounds of interest and DENV E protein. Our obtained data showed that there are hydrogen bonds between baicalin and all the domains of E protein with a high affinity of −8.0 kcal mol⁻¹. Baicalein interacts with domain I and domain III of E protein through hydrogen bonding with an affinity of −7.3 kcal mol⁻¹ (Fig. 3B). The lengths of hydrogen bonds for both baicalin and baicalein range from 1.8 Å to 2.4 Å (Table 4). Pi–pi interactions and pi–cation interactions are observed between both compounds and E protein, which makes the binding stronger and more stable (Table 5). These findings are consistent with our previous data regarding the potent activity of baicalin and baicalein against the binding of DENV-2 to Vero cells and their neutralizing properties against extracellular DENV-2 particles.

Table 4 Hydrogen bonding of baicalein and baicalin with DENV E protein (1OKE). This table shows the H-bonding interaction of baicalein and baicalin with different residues of E protein at different affinities ranked by AutoDock Vina 1.5.6. The third column shows the length of each H-bond between the molecule and residue. The last column shows the affinity energy, which was ranked by AutoDock Vina 1.5.6

Compounds	Interaction	Distance (Å)	Affinity (kcal mol^−1)
Baicalein	E:SER29:HG–baicalein:UNK1:O17	2.0	−7.1
	E:ASN366:HN–baicalein:UNK1:O18	2.4	−7.1
	E:ARG9:HH11–baicalein:UNK1:O17	2.1	−7.0
	Baicalein:UNK1:H8–E:ASP10:OD1	2.3	−7.0
	E:LYS284:HZ1–baicalein:UNK1:O19	2.0	−6.6
	E:GLY190:HN–baicalein:UNK1:O17	2.5	−6.4
	E:THR280:HG1–baicalein:UNK1:O10	2.1	−6.3
	E:ARG2:HH11–baicalein:UNK1:O10	2.4	−6.1
	E:THR359:HN–baicalein:UNK1:O18	2.2	−6.1
	E:GLY318:HN–baicalein:UNK1:O18	2.3	−5.9
	Baicalein:UNK1:H10–E:GLN316:O	1.9	−5.9
Baicalin	E:GLY18:HN–baicalin:UNK1:O23	2.4	−8.0
	E:ARG288:HE–baicalin:UNK1:O8	2.0	−8.0
	Baicalin:UNK1:H9–E:ASP290:OD1	1.8	−8.0
	E:LYS334:HZ1–baicalin:UNK1:O9	2.0	−7.8
	E:ASN355:HN–baicalin:UNK1:O8	1.9	−7.8
	E:THR268:HG1–baicalin:UNK1:O11	2.3	−7.6
	E:GLU269:HN–baicalin:UNK1:O9	2.3	−7.6
	E:THR280:HG1–baicalin:UNK1:O30	2.4	−7.6
	Baicalin:UNK1:H9–E:GLU314:O	1.9	−7.6
	E:CYS92:HN–baicalin:UNK1:O32	1.8	−7.4
	E:GLN233:HN–baicalin:UNK1:O30	2.5	−7.4
	E:LYS234:HN–baicalin:UNK1:O30	1.9	−7.4
	Baicalin:UNK1:H6–E:GLU85:OE1	2.2	−7.4
	Baicalin:UNK1:H9–E:CYS92:O	2.1	−7.4
	E:SER29:HG–baicalin:UNK1:O10	2.4	−7.2
	E:THR359:HN–baicalin:UNK1:O9	2.2	−7.2
	E:GLU360:HN–baicalin:UNK1:O9	2.2	−7.2
	Baicalin:UNK1:H18–E:GLY28:O	2.3	−7.2
	E:LYS394:HN–baicalin:UNK1:O10	1.9	−7.2
	Baicalin:UNK1:H6–E:GLY318:O	2.4	−7.2
	Baicalin:UNK1:H6–E:MET1:O	2.1	−7.0
	E:THR280:HG1–baicalin:UNK1:O30	1.8	−7.0
	E:ARG323:HH12–baicalin:UNK1:O2	2.1	−7.0
	E:ASN366:HN–baicalin:UNK1:O8	2.1	−7.0
	Baicalin:UNK1:H6–E:PRO364:O	2.1	−7.0

---

Table 5 Pi interactions with baicalein and baicalin. Pi–pi interactions, pi–sigma interactions and pi–cation interactions of baicalin and baicalein with the three dengue proteins E protein (1OKE), NS3/NS2B (2FOM) and NS5 (2J7U). The distance of interaction is shown in the fourth column

Compound	Protein	Binding	Distance (Å)	Interaction
Baicalein	NS3NS2B	Baicalein:UNK1–NS3NS2B:LYS74:NZ	6.1	Pi–cation interactions
		Baicalein:UNK1–NS3NS2B:LYS42:NZ	5.8	Pi–cation interactions
		Baicalein:UNK1–NS3NS2B:LEU76:HD11	2.8	Pi–sigma interactions
		Baicalein:UNK1–NS3NS2B:TYR161	4.1	Pi–pi interactions
		Baicalein:UNK1–NS3NS2B:ASN105:HB1	2.8	Pi–sigma interactions
		Baicalein:UNK1–NS3NS2B:LYS74:NZ	6.4	Pi–cation interactions
		Baicalein:UNK1–NS3NS2B:LYS74:NZ	6.1	Pi–cation interactions
	NS5	Baicalein:UNK1–NS5:LYS401:NZ	6.3	Pi–cation interactions
		Baicalein:UNK1–NS5:VAL603:HA	2.6	Pi–sigma interactions
		Baicalein:UNK1–NS5:LYS401:NZ	6.1	Pi–cation interactions
	E	Baicalein:UNK1–E:ARG2:NE	6.4	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	5.1	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	5.3	Pi–cation interactions
		Baicalein:UNK1–E:PHE193:HB1	2.7	Pi–pi interactions
		Baicalein:UNK1–E:HIS27	3.8	Pi–pi interactions
		Baicalein:UNK1–E:HIS27	4.5	Pi–pi interactions
		Baicalein:UNK1–E:ARG2:NE	5.6	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	5.1	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	4.5	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	4.2	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	6.0	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	4.8	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	6.0	Pi–cation interactions
		Baicalein:UNK1–E:ARG2:NE	4.8	Pi–cation interactions
Baicalin	NS3NS2B	Baicalin:UNK1–NS3NS2B:LYS74:NZ	6.0	Pi–cation interactions
		Baicalin:UNK1–NS3NS2B:TYR161	4.4	Pi–pi interactions
		Baicalin:UNK1–NS3NS2B:LYS74:NZ	6.0	Pi–cation interactions
		Baicalin:UNK1–NS3NS2B:LYS74:NZ	6.2	Pi–cation interactions
		Baicalin:UNK1–NS3NS2B:LEU76:HD21	2.7	Pi–sigma interactions
		Baicalin:UNK1–NS3NS2B:TYR161	3.9	Pi–pi interactions
		Baicalin:UNK1–NS3NS2B:GLY153:HA2	2.7	Pi–sigma interactions
		Baicalin:UNK1–NS3NS2B:TYR161	4.3	Pi–pi interactions
		Baicalin:UNK1–NS3NS2B:LEU128:HD12	2.4	Pi–sigma interactions
		NS3NS2B:HIS51–baicalin:UNK1:C4	3.6	Pi–sigma interactions
		Baicalin:UNK1–NS3NS2B:LEU128:HD23	2.6	Pi–sigma interactions
	NS5	Baicalin:UNK1–NS5:LYS401:NZ	4.6	Pi–cation interactions
		Baicalin:UNK1–NS5:LYS401:NZ	5.5	Pi–cation interactions

---

According to our molecular docking results, we designed an in vitro study as an anti-DENV protease (NS2B/NS3) assay. The reason why we chose NS2B/NS3 protein is because both our compounds, baicalin and baicalein, interacted with the active site of DENV protease, as mentioned above. This aroused our interest in performing an in vitro study with DENV protease. The compounds were weighed and dissolved in DMSO for the preparation of a stock solution. Both baicalein and baicalin displayed considerable inhibitory effect towards dengue NS2B/NS3pro. The test compounds were added to the dengue protease reaction at different concentrations and the inhibition profile was plotted, as shown in Fig. 10. The reaction velocity of NS2B/NS3pro decreased with an increase in the concentrations of the test compounds. This finding clearly indicated that the compounds inhibited the catalytic activity of NS2B/NS3pro (Fig. 10). A kinetic assay study showed that the compounds inhibited the activity of dengue NS2B/NS3pro with IC₅₀ values of 68.65 μg mL⁻¹ for baicalein and 144.7 μg mL⁻¹ for baicalin (Fig. 10).

Fig. 10 Inhibitory effect of baicalein and baicalin against dengue NS2B/NS3pro. The inhibitory profile of the test compounds was plotted against the enzyme activity of dengue NS2B/NS3pro. This assay was performed using increasing concentrations of the inhibitor while all other conditions were kept constant. The data were analysed using the Michaelis–Menten model with a non-linear regression curve fit in Graph Pad Prism (version 5.01) software to calculate the expected value of IC₅₀ at normal human physiological temperature (37 °C).

The results of the in vitro anti-DENV protease study prove our findings from the molecular docking study, namely, that both baicalin and baicalein can inhibit the dengue replication cycle via one of the critical DENV genes. In summary, this study proved that baicalin and baicalein, as metabolites of each other, are potential inhibitors of dengue by inhibiting one of the vital DENV proteins and affecting the replication cycle.

Experimental

Flavonoids

Baicalein and baicalin were purchased from Sigma Chemical Company (Sigma, St Louis, USA) and subsequently dissolved in dimethyl sulfoxide (DMSO) (Sigma-Aldrich, St Louis, MO, USA) prior to use.

Receptor and ligand preparation

The three-dimensional structures of the DENV-2 NS2B/NS3 (2FOM),^56,57 NS5 (2J7U)^58,59 and E (1OKE)^26,60 proteins were downloaded from the Protein Data Bank (http://www.rscb.org/pdb). These structures were then minimized using the CHARMM27 force field using the Discovery Studio 2.5 software package. The ligand structures were constructed using ChemDraw (CambridgeSoft) and then imported into the Discovery Studio 2.5 software package to be minimized before beginning the docking process. All the water molecules and ligands were deleted from all the receptors before being used for docking in Discovery Studio 2.5. AutoDock Vina was used to prepare both ligands and receptors for running and to save both compounds as PDBQT files.

Molecular docking

In this study, we used AutoDock Vina 1.5.6 software to examine the binding of the ligands to the target proteins, including the NS3/NS2B, NS5 and E proteins of DENV. All compounds and proteins of interest had hydrogen molecules added before being saved as PDBQT files. In this study, we used blind molecular docking. Before running the prepared docking files, all the receptors were placed inside the grid box and all the grid information was entered in the text file. The grid that was used to run NS3/NS2B was center-X = −1.976, center-Y = −15.106, center-Z = 16.323, size-X = 52, size-Y = 54, size-Z = 46. The grid box information for NS5 was center-X = 24.078, center-Y = 57.301, center-Z = 16.616, size-X = 74, size-Y = 74, size-Z = 72. The grid box information for E protein was as follows: center-X = −10.916, center-Y = 36.317, center-Z = −16.089, size-X = 60, size-Y = 44, size-Z = 118. The grid box was chosen carefully to make sure that the whole protein of interest fitted inside the box.

Analysing the molecular docking results

AutoDock Vina 1.5.6 and Discovery Studio 2.5 were used to analyse the obtained docking results. The output PDBQT file was opened in AutoDock software in order to see the different conformations of our compounds with each protein. AutoDock Vina ranks all the conformations based on their affinity energy. AutoDock Vina is able to provide data on hydrogen bonding and all close contact interactions between ligands and receptors, as well as on pi–pi interactions and pi–cation interactions. In order to obtain more information about the interaction energies (i.e., van der Waals and electrostatic energies) and also to calculate the binding energy, we have to use Discovery Studio 2.5. PyMOL software was used in order to make all conformations into a single file before use for calculation in Discovery Studio 2.5. The sum of the interaction energy and binding energy was defined as the complexation energy. This energy serves as a criterion for selecting the most active inhibitors.

In vitro study

Dengue NS2B–NS3 protease assay. This assay was carried out to identify the inhibitory activities of the test compounds against dengue NS2B–NS3pro, as previously described.⁶¹ In brief, dengue NS2B–NS3pro was produced by E. coli as a recombinant protein. The end point reaction mixture was prepared in black 96 well plates and consisted of 2 μM recombinant NS2B–NS3pro, 100 μM fluorogenic peptide substrate (Boc-Gly-Arg-Arg-AMC; Minoh-shi, Osaka, Japan, lot no.: 580907), and the test compound at varying concentrations prepared in DMSO (0, 25, 50, 100, 200 and 400 μg mL⁻¹), buffered at pH 8.5 with 200 mM Tris–HCl with a total volume of 200 μL. The reaction mixture without the test compound, substrate with the compound, enzyme and different concentrations of compound were used as controls. Thereafter, all reaction mixtures were incubated at 37 °C for 30 minutes, and the substrate was added to the specific reaction mixtures and incubated at the same temperature for another 30 minutes. Measurements were performed in triplicate using a Tecan Infinite M200 Pro fluorescence spectrophotometer (Tecan Group Ltd., Switzerland). Substrate cleavage was normalized against buffer only (control) at an emission wavelength of 440 nm upon excitation at 350 nm. Reaction velocities at steady state were calculated from the slope of reaction progression curves by non-linear regression of the initial velocities using Graph Pad Prism (version 5.01) software.

Conclusion

In conclusion, the results of this study support our previous findings on the antiviral activity of baicalein and baicalin against DENV, as in the current study we show the possible interactions between these compounds and three DENV proteins that are important for virus replication. The data obtained from this study are consistent with previous reports, which showed that baicalein and baicalin can interfere with different stages of the replication cycle of DENV in vitro.^39,41 However, baicalein displayed better interaction affinity with NS3/NS2B (−7.5 kcal mol⁻¹) compared with the other dengue proteins, and baicalin displayed better interaction affinity with NS5 (−8.6 kcal mol⁻¹). Nevertheless, in general baicalin exhibits stronger interactions with all three tested proteins compared with baicalein. As baicalin is the main metabolite of baicalein, this study therefore supports an anti-DENV role for both compounds, which can encourage scientists to further investigation towards the development of effective anti-DENV agents from these compounds in the future.

Author contribution

KZ and SAB conceived and designed the experiments. MS collected the data, MS and PH performed the experiments, and MS analyzed the data and wrote the paper.

Conflict of interest

The authors declare that they have no competing interests.

Acknowledgements

The authors would like to thank the Ministry of Higher Education, Malaysia for a High Impact Research (HIR) MOHE Grant (E000013-20001) and the Fundamental Research Grant Scheme (FP054-2014B). We would also like to thank the University of Malaya for the Postgraduate Research Fund (PG024-2014B). We would also like to acknowledge the team of the Tropical Infectious Disease Research Centre, Department of Medical Microbiology, Faculty of Medicine, University of Malaya for providing help and support throughout the study.

References

1. P. M. Armstrong and R. Rico-Hesse, Am. J. Trop. Med. Hyg., 2003, 68, 539–544.
2. A. Nisalak, T. P. Endy, S. Nimmannitya, S. Kalayanarooj, R. M. Scott, D. S. Burke, C. H. Hoke, B. L. Innis and D. W. Vaughn, Am. J. Trop. Med. Hyg., 2003, 68, 191–202.
3. M. Mustafa, V. Rasotgi, S. Jain and V. Gupta, Med. J. Armed Forces India, 2015, 71, 67–70.
4. M. G. Guzman and E. Harris, Lancet, 2015, 385, 453–465.
5. S. L. Alcaraz-Estrada, M. Yocupicio-Monroy and R. M. del Angel, Future Virol., 2010, 5, 575–592.
6. T. J. Chambers, C. S. Hahn, R. Galler and C. M. Rice, Annu. Rev. Microbiol., 1990, 44, 649–688.
7. J. Li, S. P. Lim, D. Beer, V. Patel, D. Wen, C. Tumanut, D. C. Tully, J. A. Williams, J. Jiricek and J. P. Priestle, J. Biol. Chem., 2005, 280, 28766–28774.
8. B. Falgout, M. Pethel, Y. Zhang and C. Lai, J. Virol., 1991, 65, 2467–2475.
9. L.-R. Jan, C.-S. Yang, D. W. Trent, B. Falgout and C.-J. Lai, J. Gen. Virol., 1995, 76, 573–580.
10. R. Yusof, S. Clum, M. Wetzel, H. K. Murthy and R. Padmanabhan, J. Biol. Chem., 2000, 275, 9963–9969.
11. S. Clum, K. E. Ebner and R. Padmanabhan, J. Biol. Chem., 1997, 272, 30715–30723.
12. J. Lescar, D. Luo, T. Xu, A. Sampath, S. P. Lim, B. Canard and S. G. Vasudevan, Antiviral Res., 2008, 80, 94–101.
13. C. W. Kim and K.-M. Chang, Clin. Mol. Hepatol., 2013, 19, 17–25.
14. W. Y. Phong, N. J. Moreland, S. P. Lim, D. Wen, P. N. Paradkar and S. G. Vasudevan, Biosci. Rep., 2011, 31, 399–409.
15. C.-F. Wu, S.-H. Wang, C.-M. Sun, S.-T. Hu and W.-J. Syu, J. Virol. Methods, 2003, 114, 45–54.
16. K. A. Salam and N. Akimitsu, BioMed Res. Int., 2013, 2013(9), 467869.
17. E. Navarro-Sanchez, R. Altmeyer, A. Amara, O. Schwartz, F. Fieschi, J. L. Virelizier, F. Arenzana-Seisdedos and P. Desprès, EMBO Rep., 2003, 4, 723–728.
18. B. Tassaneetrithep, T. H. Burgess, A. Granelli-Piperno, C. Trumpfheller, J. Finke, W. Sun, M. A. Eller, K. Pattanapanyasat, S. Sarasombath and D. L. Birx, J. Exp. Med., 2003, 197, 823–829.
19. M. N. Krishnan, B. Sukumaran, U. Pal, H. Agaisse, J. L. Murray, T. W. Hodge and E. Fikrig, J. Virol., 2007, 81, 4881–4885.
20. S. Jindadamrongwech, C. Thepparit and D. Smith, Arch. Virol., 2004, 149, 915–927.
21. J. L. Miller, B. De Wet, L. Martinez-Pomares, C. M. Radcliffe, R. A. Dwek, P. M. Rudd and S. Gordon, PLoS Pathog., 2008, 4, e17.
22. F. A. Rey, F. X. Heinz, C. Mandl, C. Kunz and S. C. Harrison, Nature, 1995, 375, 291–298.
23. S. L. Allison, J. Schalich, K. Stiasny, C. W. Mandl and F. X. Heinz, J. Virol., 2001, 75, 4268–4275.
24. Y. Modis, S. Ogata, D. Clements and S. C. Harrison, Nature, 2004, 427, 313–319.
25. S. Bressanelli, K. Stiasny, S. L. Allison, E. A. Stura, S. Duquerroy, J. Lescar, F. X. Heinz and F. A. Rey, EMBO J., 2004, 23, 728–738.
26. Y. Modis, S. Ogata, D. Clements and S. C. Harrison, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 6986–6991.
27. X. Zhang, P. Ge, X. Yu, J. M. Brannan, G. Bi, Q. Zhang, S. Schein and Z. H. Zhou, Nat. Struct. Mol. Biol., 2013, 20, 105–110.
28. R. Fritz, K. Stiasny and F. X. Heinz, J. Cell Biol., 2008, 183, 353–361.
29. V. Nayak, M. Dessau, K. Kucera, K. Anthony, M. Ledizet and Y. Modis, J. Virol., 2009, 83, 4338–4344.
30. Y. Zhang, W. Zhang, S. Ogata, D. Clements, J. H. Strauss, T. S. Baker, R. J. Kuhn and M. G. Rossmann, Structure, 2004, 12, 1607–1618.
31. J. A. Mondotte, P.-Y. Lozach, A. Amara and A. V. Gamarnik, J. Virol., 2007, 81, 7136–7148.
32. A. J. Brooks, M. Johansson, A. V. John, Y. Xu, D. A. Jans and S. G. Vasudevan, J. Biol. Chem., 2002, 277, 36399–36407.
33. M. Johansson, A. J. Brooks, D. A. Jans and S. G. Vasudevan, J. Gen. Virol., 2001, 82, 735–745.
34. C. Yon, T. Teramoto, N. Mueller, J. Phelan, V. K. Ganesh, K. H. Murthy and R. Padmanabhan, J. Biol. Chem., 2005, 280, 27412–27419.
35. A. Bartholomeusz and P. Wright, Arch. Virol., 1993, 128, 111–121.
36. I. Sanchez, F. Gómez-Garibay, J. Taboada and B. Ruiz, Phytother. Res., 2000, 14, 89–92.
37. M. Muhamad, L. Y. Kee, N. A. Rahman and R. Yusof, Int. J. Biol. Sci., 2010, 6, 294.
38. M. Alen, T. De Burghgraeve, S. Kaptein, J. Balzarini, J. Neyts and D. Schols, PLoS One, 2011, 6, e21658.
39. K. Zandi, B.-T. Teoh, S.-S. Sam, P.-F. Wong, M. R. Mustafa and S. AbuBakar, BMC Complementary Altern. Med., 2012, 12, 214.
40. K. Zandi, R. Lani, P.-F. Wong, B.-T. Teoh, S.-S. Sam, J. Johari, M. R. Mustafa and S. AbuBakar, Molecules, 2012, 17, 2437–2445.
41. E. Moghaddam, B.-T. Teoh, S.-S. Sam, R. Lani, P. Hassandarvish, Z. Chik, A. Yueh, S. Abubakar and K. Zandi, Sci. Rep., 2014, 4, 5452.
42. A. J. Stevens, M. E. Gahan, S. Mahalingam and P. A. Keller, J. Med. Chem., 2009, 52, 7911–7926.
43. K. Zandi, B.-T. Teoh, S.-S. Sam, P.-F. Wong, M. R. Mustafa and S. AbuBakar, J. Virol., 2011, 8, 560.
44. P.-M. Allard, E. T. H. Dau, C. C. Eydoux, J.-C. Guillemot, V. Dumontet, C. Poullain, B. Canard, F. O. Guéritte and M. Litaudon, J. Nat. Prod., 2011, 74, 2446–2453.
45. R. Othman, T. S. Kiat, N. Khalid, R. Yusof, E. Irene Newhouse, J. S. Newhouse, M. Alam and N. A. Rahman, J. Chem. Inf. Model., 2008, 48, 1582–1591.
46. S. Idrees and U. A. Ashfaq, Genet. Vaccines Ther., 2012, 10, 6.
47. A. Mukhametov, E. I. Newhouse, N. Ab Aziz, J. A. Saito and M. Alam, J. Mol. Graphics Modell., 2014, 52, 103–113.
48. M. Yildiz, S. Ghosh, J. A. Bell, W. Sherman and J. A. Hardy, ACS Chem. Biol., 2013, 8, 2744–2752.
49. A. S. d. Oliveira, M. L. d. Silva, A. F. Oliveira, C. C. d. Silva, R. R. Teixeira and S. O. De Paula, J. Braz. Chem. Soc., 2014, 25, 1759–1769.
50. M. Bollati, K. Alvarez, R. Assenberg, C. Baronti, B. Canard, S. Cook, B. Coutard, E. Decroly, X. de Lamballerie and E. A. Gould, Antiviral Res., 2010, 87, 125–148.
51. H. Li, S. Clum, S. You, K. Ebner and R. Padmanabhan, J. Virol., 1999, 73, 3108–3116.
52. Z. Zuo, O. W. Liew, G. Chen, P. C. J. Chong, S. H. Lee, K. Chen, H. Jiang, C. M. Puah and W. Zhu, J. Virol., 2009, 83, 1060–1070.
53. A. A. Khromykh, H. Meka, K. J. Guyatt and E. G. Westaway, J. Virol., 2001, 75, 6719–6728.
54. A. D. Davidson, Adv. Virus Res., 2009, 74, 41–101.
55. M. Issur, B. J. Geiss, I. Bougie, F. Picard-Jean, S. Despins, J. Mayette, S. E. Hobdey and M. Bisaillon, RNA, 2009, 15, 2340–2350.
56. P. Erbel, et al., Structural basis for the activation of flaviviral NS3 proteases from dengue and West Nile virus, Nat. Struct. Mol. Biol., 2006, 13, 372–373.
57. R. Raut, et al., A small molecule inhibitor of dengue virus type 2 protease inhibits the replication of all four dengue virus serotypes in cell culture, J. Virol., 2015, 12, 16.
58. L. L. G. Ariza, G. A. T. Ramirez, H. F. C. Hernández, L. P. Sanabria and J. C. C. Osorio, in Advances in Computational Biology, Springer, 2014, pp. 273–278.
59. T. L. Yap, et al., Crystal structure of the dengue virus RNA-dependent RNA polymerase catalytic domain at 1.85-angstrom resolution, J. Virol., 2007, 81, 4753–4765.
60. T. Kampmann, et al., In silico screening of small molecule libraries using the dengue virus envelope E protein has identified compounds with antiviral activity against multiple flaviviruses, Antiviral Res., 2009, 84, 234–241.
61. H. A. Rothan, et al., Protegrin-1 inhibits dengue NS2B–NS3 serine protease and viral replication in MK2 cells, J. Biomed. Biotechnol., 2012, 251482.

---