Accepted Article
Received Date : 12-Jan-2014 Accepted Date : 13-Feb-2014 Article type
: Research Article
Peptide Inhibitors against Dengue Virus Infection Aussara Panya 1,2, Kunan Bangphoomi 3, Kiattawee Choowongkomon 3, Pa-thai Yenchitsomanus 1,*
1
Division of Molecular Medicine, Department of Research and Development, Faculty
of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand 2
Department of Biochemistry, Faculty of Medicine Siriraj Hospital, Mahidol
University, Bangkok, Thailand 3
Department of Biochemistry, Faculty of Science, Kasetsart University, Bangkok,
Thailand
*Corresponding author. Tel: +66 2 4192777; Fax: +66 2 4110169 E-mail Address: [email protected], [email protected]
Abstract Dengue virus (DENV) infection has become a public health problem worldwide. The development of anti-DENV drug is urgently needed because neither licensed vaccine nor specific drug is currently available. Inhibition of DENV attachment and entry to host-cells by blocking DENV envelope (E) protein is an attractive strategy for antiDENV drug development. A hydrophobic pocket on the DENV E protein is essential This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an 'Accepted Article', doi: 10.1111/cbdd.12309 This article is protected by copyright. All rights reserved.
Accepted Article
for structural transition in the membrane fusion and inhibition of this process is able to inhibit DENV infection. To search for a safe anti-DENV drug, we identified short peptides targeting to the hydrophobic pocket by molecular docking. In addition, the information of predicted ligand-binding site of reported active compounds to DENV2 hydrophobic pocket was also used for peptide inhibitors selection. The di-peptide, EF, was the most effective on DENV2 infection inhibition in vitro with a half maximal inhibition concentration (IC50) of 96 μM. Treatment of DENV2 with EF at the concentration of 200 μM resulted in 83.47% and 84.15% reduction of viral genome and intracellular E protein, respectively. Among four DENV serotypes, DENV2 was the most effective for the inhibition. Our results provide the proof-of-concept for development of therapeutic peptide inhibitors against DENV infection by the computer-aided molecular design.
1. Introduction Dengue virus (DENV) infection has become a public health worldwide.
It is
estimated that approximately 50 million people are infected and over 20,000 people died each year [1]. DENV comprises 4 related serotypes, DENV1-4, transmitted via mosquito vectors.
DENV infection causes a spectrum of clinical manifestations
ranging from a mild dengue fever (DF) to a more severe dengue hemorrhagic fever (DHF) and life-threatening dengue shock syndrome (DSS) [2]. Currently, no licensed vaccine or specific drug against DENV infection is available.
The correlation
between magnitude of DENV titer and disease severity [3,4,5,6,7] suggests that the reduction of DENV level in the patients will be able to ameliorate the disease outcome and anti-DENV agents should be beneficial to the DENV-infected patients in the viremic phase. DENV and host proteins controlling the DENV life cycle are
This article is protected by copyright. All rights reserved.
Accepted Article
attractive targets for the development of anti-DENV agents. A number of inhibitors targeting to these proteins showed inhibitory activities to DENV infection. These inhibitors include small-molecule compounds [8,9,10,11,12,13], siRNAs [14,15], nucleoside analogs [16,17], antisense oligomers [18], antibodies [19,20,21,22], and peptide inhibitors [23,24,25,26,27]. DENV envelope (E) protein that has a major function in the viral entry is a key target for the development of anti-DENV agents that inhibit DENV entry by interfering with either the viral attachment or membrane fusion process. Currently, a number of anti-DENV molecules targeting to DENV E protein such as monoclonal antibodies [19,21,22], peptide inhibitors [24,25,26,27], synthetic compounds [8,9,11,13], and drug derivatives [28] have been reported. Previously, the study of structural difference between E homodimer in the absence or presence of a detergent, namely n-octyl-β-D-glycoside (BOG), revealed an area of hydrophobic pocket which serves as a hinge region responsible for structural transition [29]. Since mutations that affect pH threshold for membrane fusion are also mapped to this region, the hydrophobic pocket has been proposed as the target for fusion inhibitors [29]. This was proven by the fact that synthetic compounds targeting to this pocket showed inhibitory effects on membrane-fusion in vitro [8,13]. Peptide inhibitors have been considered as a class of therapeutic agent with diverse chemical properties and biological activities. An advantage of peptide inhibitors is that they have less toxicity, compared to synthetic compounds. They are naturally one of first-line defense in wide classes of life ranging from insects, ichthyoids, amphibians, mammals and humans to protect against bacteria, fungal, and also viral pathogens [30,31]. Accordingly, they are studied in many diseases of humans both infectious diseases such as viral infections and non-infectious diseases such as cancers
This article is protected by copyright. All rights reserved.
Accepted Article
[32]. The peptidic antiretroviral Enfuvirtide (Fuzeon) is the first clinically FDAapproved antiviral peptide inhibitor of virus-cell fusion that shows potent antiretroviral activity and significantly improves medical outcomes in highly treatment-experienced patients with HIV-1 infection [33]. This represents an example of peptide inhibitor that is developed for clinical use in the viral infection. In the case of DENV infection, a number of peptide inhibitors targeting E protein have been reported. These include peptide inhibitors derived from the stem region of E protein (residues 412-447) [24,25], peptide inhibitors optimized from original E sequences using computational optimization method [26], and peptide inhibitors de novo designed to target putative receptor-binding site [27]. These peptide inhibitors had ability to interfere with DENV2 entry and thus potentially attenuate rate of infection. However, the use of in silico design for small peptide inhibitors specific to the hydrophobic pocket of DENV E protein has not yet been reported. Targeting to this pocket requires a minimal length of peptides to optimally fit into its cavity. Thus, this study aims to identify a set of small peptide inhibitors targeting to the hydrophobic pocket of DENV2 by using the ligand-binding site information of reported active compounds and drugs combined with the computational screening. Out of seven peptides that were selected from hundreds of peptides for testing the anti-DENV effect, one peptide has striking inhibitory effects on DENV2 replication, protein synthesis, and virus production.
2. Materials and methods 2.1 Preparation of receptor and ligands for computer analysis To identify small peptides that specifically bind to the n-octyl-β-D-glycoside (BOG) or hydrophobic pocket, the molecular docking methodology was performed by using
This article is protected by copyright. All rights reserved.
Accepted Article
two docking programs; AutoDOCK v4.2 [34] and CDOCKER Discovery Studio v2.1 [35]. DENV2 E protein in complexes with BOG (PDB: 1OKE) was used as a receptor. The preparation of receptor by using AutoDOCK v4.2 and CDOCKER v2.1 programs was slightly different. Briefly, the preparation by using AutoDOCK was began by adding the charges and salvation term to the receptor. The binding site or grid box was located using “center to ligand” mode in AutoDOCKTools in dimension of x-size = 40, Y-size = 40, Z-size = 40. The AutoGrid program was performed to pre-calculate the grid maps for individual atom type of ligand. For CDOCKER, the receptor was subjected to a restrained energy minimization with CHARMm forcefield and MMFF 94 partial charge in Discovery Studio v2.1. The binding site or sphere was located by using the “Define sphere form selection” mode and expanded the sphere size up to 20 Å in radius. Ligands in this study including amino acids, dipeptides, and tri-peptides were derived from “Build and edit protein” function in DS. The structures were optimized by using Dreiding-like forcefield in “Clean geometry” mode and exported in the PDB format.
2.2 Molecular docking on DENV E protein Docking simulation against BOG pocket on the DENV E protein was carried out by using the AutoDOCK and CDOCKER programs. All 20 amino acids were firstly docked into the BOG pocket where the parameters were set as default values following the program instructions. The top 7 hits identified from both programs according to the scoring function of the particular docking software were collected and used as di-peptide building block. The dockings of di-peptides were performed to select for the top 5, which were used as scaffolds. The scaffolds were grown at the Nor C-terminus randomly dependent on the position of di-peptide at the key interaction
This article is protected by copyright. All rights reserved.
Accepted Article
site on the pocket to generate tri-peptides. The dockings of tri-peptides were again performed to collect the ligands with high binding-scores. The peptides with topranked scores from both programs with the predicted ligand-binding residue contraction were considered.
Finally, only seven peptides were selected for
experimental validation. The figures were prepared by using PyMOL programs.
2.3 Determination of ligand-binding residues The site-moiety map (SiMMap) algorithm [36] was used to identify ligand-binding residues on the BOG pocket of DENV E protein. The conformational poses derived from docking of the active drugs and compounds including doxycyclin [28], relitetracyclin [28], compound A5 [8], compound 36 [9], and compound 11 [9] against the BOG pocket by using the AutoDOCK v4.2 and CDOCKER (DS v2.1) programs were input to the SiMMap software for the analysis.
2.4 Cell culture and virus preparation DENV1 (strain Hawaii), DENV2 (strain 16681), DENV3 (strain H87), DENV4 (strain Loc3 00574/99) were propagated in Aedes albopictus (C6/36) cells cultured in L-15 medium containing 1% fetal bovine serum (FBS) and 10% tryptose phosphate broth by using multiplicity of infection (MOI) of 0.1. After incubation for 5 days at 28°C, the culture supernatant was collected and cellular debris was removed by centrifugation. The virus was stored at -70°C until it was used for experiments. Vero cells were cultured in minimum essential media (MEM) supplemented with 10% FBS and antibiotics (penicillin G and streptomycin) and were maintained at 37°C in an incubator at 5% CO2.
This article is protected by copyright. All rights reserved.
Accepted Article
2.5 Peptide synthesis Peptides were custom-made by Selleck Chemicals (Huston, TX, USA). Seven peptides that were selected and synthesized were solubilized in sterile distilled water to a final concentration of 25 mM and stored in a freezer at -25°C.
2.6 Cytoxicity test The cytotoxic effects of peptides were tested on cultured Vero cells by using PlestoBLUETM Cell Viability Reagent (Invitrogen, CA, USA). The peptides at the maximal concentration of 0.5 mM prepared in the cell culture medium (a final volume of 90 µL) were added to the Vero cells (2x104 cells per well of a 96-well plate) and incubated at 37°C in 5% CO2. After incubation for 72 hours, an assay was performed by following the manufacturer’s instructions. Briefly, the PrestoBLUETM reagent (10X) was added (10% v/v) to culture medium. The reaction was incubated at 37°C for 30 minutes and then measured absorbance at 570 nm with a reference wavelength set at 600 nm.
2.7 Virus infection and treatment with peptides The peptides at indicated concentrations were incubated with DENV (200 FFU in a 96-well plate and 1000 FFU in a 24-well plate) for 30 minutes at 37°C before adding to Vero cells (2 x 104 cells per well for a 96-well plate and 1 x 105 cells per well for a 24-well plate) for viral adsorption for 2 hours at 37°C. The unbound viruses were removed
and
the
cells
were
over-layered
by
the
medium
containing
carboxylmetylcellulose for foci-forming assay or were added with fresh medium for other assays. The infected cells were incubated at 37°C until the experiment was performed.
This article is protected by copyright. All rights reserved.
Accepted Article
2.8 Foci-forming reduction assay The foci-forming reduction assay was carried out to determine inhibitory effects of peptides on DENV infection to Vero cells. At 72 hours post infection, the numbers of foci were determined by the protocol as previously described [8]. The Vero cells were fixed with 4% paraformaldehyde for 15 minutes and washed twice with phosphate buffered saline (PBS). Cells were permeabilized with 0.2% Triton X-100 for 15 minutes at room temperature, blocking with 1% Bovine Serun Albumin (BSA) in PBS for 30 min, and incubated for 3 hours with monoclonal anti-DENV E antibody, 4G2, at 37°C. Cells were washed 5 times with PBS containing 0.5% Tween-20 (PBST) and further incubated for 1 hours with rabbit anti-mouse conjugated horseradish peroxidase (HRP).
After PBST washing, 3’ di-amino-
benzidine (DAB) peroxidase substrate (Sigma, USA) was added to stain the DENV foci, which were manually counted under a light microscope. The IC50 of each peptide was analyzed from dose-response curve by using non-linear regression function of GraphPad Prism software.
2.9 Quantitative real-time polymerase chain reaction Quantitative real-time PCR (qRT-PCR) was carried out to determine the effects of the peptides on DENV replication. After DENV infection, the cells were harvested at 48 and 72 hours and RNA was extracted by using TRIzol reagent (Invitrogen, CA, USA) following the manufacturer’s protocol. RNA (1.5 μg) was changed into cDNA by using eAMV reverse transcriptase (Promega).
qRT-PCR was conducted in a
LightCycler 480 (Roche, Mannheim, Germany) machine by using 1 μL of cDNA, DENV E specific primers; D2R (5′-CCGGCTCTACTCCTATGATG-3′) and D2L (5′ATCCAGATGTCATCAGGAAAC-3′) [37] and 2x SYBR Green I Master Mix
This article is protected by copyright. All rights reserved.
Accepted Article
(Roche, Mannheim, Germany). The data was normalized with that of the β-actin and the relative value of 1.0 derived from the infected condition without the treatment with peptide was assigned as the calibrator.
2.10 Cell-based flavivirus immunodetection To examine the inhibitory effect of peptide on viral protein synthesis, DENV protein level was determined by using cell-based flavivirus immunodetection (CFI) assay. The DENV E protein was detected by the CFI method using 4G2 antibody as previously described [10] at 72 hours post infection. Briefly, the Vero cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, and blocking with 1% Bovine Serun Albumin (BSA). Cells were incubated for 3 hours with monoclonal anti-DENV E antibody, 4G2, at 37°C, washed 5 times with PBS containing 0.5% Tween-20 (PBST) and further incubated for 1 hours with rabbit antimouse conjugated horseradish peroxidase (HRP). The 3,3’,5,5’-tetramethylbenzidine or TMB was added and incubated at room temperature for 15 minutes before adding sulfuric acid to stop the reaction. The absorbance was measured at OD492 by a plate reader set by using the mock control as blank.
2.11 Immunofluorescence assay Immunofluorescence assay (IFA) was performed to examine the percentage of DENV-infected cells after the peptide treatment.
At 72 hours of infection, the
infected cells were harvested and stained for DENV E protein with monoclonal 4G2 antibody and Alexa Fluor® 488 Goat Anti-Mouse IgG while cell nuclei were stained with Hoechst, by the method as previously described [12,38].
This article is protected by copyright. All rights reserved.
Accepted Article
2.12 Statistic analysis In every experiment, at least three independent assays were performed. The average quantitative values (means) and their standard deviation were determined, and student’s t tests were performed by using GraphPad Prism software. P value of less than 0.05 was considered to be statistically significant.
Results and discussion Identification of ligand-binding hot spots The identification of specific sites (or hot spots) for ligand binding on the receptor protein is helpful for designing inhibitors that interfere with its biological function. Hot spots at the hydrophobic (BOG) pocket of DENV E protein were firstly characterized by testing the interactions between each of the five reported compounds (doxycyclin, relitetracyclin, compound A5, compound 36, and compound 11) and the hydrophobic pocket of DENV E protein by using SiMMap program. Three distinct hot spots (H1, H2, and H3) at the hydrophobic pocket of DENV E protein, which were expected to be critical for binding and involving in the inhibitory effect of these compounds, were identified (Figure 1). The H1 and H3 hot spots located at the entrance of the pocket but they situated at the opposite direction. The H1 hot spot, containing Lys128 and Gln200, has partially positive charge whereas the H3 hot spot, containing The48, Glu49, and Ala50, has partially negative charge. The H2 hot spot lined at the bottom of the pocket with hydrophobic amino acids, Ile270, Leu277, and The280. From the inspection of the interaction between the ligands and the pocket, it revealed that one end of the ligands inserted into the pocket to interact with H2 suggesting the preference of H2 in the interaction. This might be attributable to the narrow space at the middle of pocket and the ligands were possibly locked to the H2
This article is protected by copyright. All rights reserved.
Accepted Article
position. On the other hand, the wider space at the entrance of the pocket might allow the flexibility of interaction between ligands and H1 or H3 through the favorable electrostatic property that might contribute to their binding affinities. The analysis of the ligand-pocket interaction suggested that the binding of two third of the hot spots, H1-H2 or H3-H2, could contribute to the inhibitory properties of these known compounds. To explore this concept, the interactions of ligands with either H1-H2 or H3-H2 were also considered as an extra criterion for selection of peptide inhibitors in this study.
Design of small peptides To validate the docking methodologies by AutoDOCK and CDOCKER programs, the two programs were used to simulate the interaction of BOG ligands to the hydrophobic pocket of DENV E protein.
The superimposes of docking results
obtained from the simulation using these two programs to BOG in the reported crystal structure revealed the overlaps of molecules with root mean square deviation (RMSD) of 0.667 and 1.61 Å, respectively (data not shown). As the resulted RMSDs were less than 2 Å, the docking methodologies by both programs were considered to be acceptable [35]. To identify small peptides binding to the hydrophobic pocket of DENV E protein, the top 7 hits of amino acids obtained from the docking simulation were taken to construct the peptides containing two amino acids. The peptides were then used as scaffolds to create the peptides containing three amino acids. When all peptides were used for docking to the hydrophobic pocket by both programs, altogether 304 hits were obtained. In the final selection, only the peptides that formed bonds with the predicted hot spots in the hydrophobic pocket were chosen. Seven peptides with high
This article is protected by copyright. All rights reserved.
Accepted Article
docking scores including Lys-Glu-Asn (KEN), Trp-Glu-Asn (WEN), Glu-Glu-Gly (EEG), Trp-Glu-Thr (WET), Glu-Gln (EQ), Glu-His (EH), and Glu-Phe (EF), which bound to at least one hot spot, were finally picked for further studies.
Inhibition of DENV2 infection by small peptides Seven small peptides (KEN, WEN, EEG, WET, EQ, EH, and EF) were custom-made by Selleck Chemicals (Huston, TX, USA). To evaluate the cytotoxic effect of these peptides, individual peptide at the concentration of 500 μM was tested by adding to Vero cells – a permissive cell line for DENV2 infection [39], and cell viability was measured at 72 hours. The results showed that cell viabilities were not decreased (Figure 2A), suggesting that these peptides had no cytotoxic effect to the cells at the high concentration as 500 μM.
The peptides were then tested for their abilities to inhibit foci formation of DENV2. Almost all peptides could reduce the foci formation of DENV2 (Figure 2B). Among these peptides, EF and KEN could reduce the foci formation to approximately 90% while other peptides reduced it to 40-60%. However, EQ could not reduce the foci formation of DENV2.
The correlation between peptide concentrations and
percentages of virus production inhibition was examined (Figure 3A and 3B) and the half maximal inhibitory concentration (IC50) of peptides was also determined. EF and KEN had IC50 of 96.50 μM and 331.9 μM, respectively, which were concordant to their predicted inhibition constant (Ki) of 98 μM and 455 μM, respectively, as predicted by the AutoDOCK program (supporting information, Table 1). The reason why EF had a better efficiency than KEN might be explained by that EF could interact with the H1 and H3 hot spots at the entrance of hydrophobic pocket and
This article is protected by copyright. All rights reserved.
Accepted Article
also with the H2 hot spot at the bottom of the pocket while KEN could interact with only the H1 and H3 hot spots of the pocket (Figure 4A). Hydrogen bonds between side chains of EF and those of the H1 and H3 hot spots of the pocket are crucial for the interaction. EF interacted with H1 and H3 by using hydrogen bond between the carboxylate-group side chain of Glu (E) and the NH-group side chain of Lys128 of H1 in addition to the NH-group main chain of Phe (F) to carboxylate-group main chain of The48 of H3 (Figure 4B, 4D). In addition, phenyl ring of Phe (F) made contact with H2 containing Ile270, Leu277, Phe279, and The280 at the bottom of the hydrophobic pocket. KEN formed hydrogen bonds with H1 and H3 but did not have hydrophobic interaction with H2 (Figure 4C, 4E). The carboxylate-group side chain of Glu (E) in KEN formed hydrogen bonds to the NH-group side chains of Lys128 and Gln200 of H1 whereas the NH-group side chain of Asn (N) formed hydrogen bonds to carboxylate-group main chain of The48 of H3 and nearby Ser274. In contrast to EF, KEN could only interact to the upper hydrophilic part of the pocket. The hydrophilic feature of KEN is likely to prevent it to enter into the hydrophobic environment within the pocket, making it loosely binding to the pocket. Additionally, a large side chain of KEN might prevent it to deeply insert into the pocket. The result of molecular docking showed that the occupation of the EF peptide into the hydrophobic pocket did not entirely fit, especially at the upper part of pocket; hence, it is possible for any modification to improve peptide affinity, specificity, and half-life. Additionally, the peptide for oral administration is a challenge because peptide inhibitor often requires intravascular route of administration as it has low gastrointestinal absorption.
This article is protected by copyright. All rights reserved.
Accepted Article
Reduction of DENV RNA, protein, and virus production by EF peptide The EF peptide was designed specific to the hydrophobic pocket on the DENV E protein – the site that affects DENV entry to the host cells [8,10,13]. Inhibition of DENV entry by binding of EF peptide to this critical site would consequently result in the reduction of DENV intracellular activities including RNA replication, protein synthesis, and virus production. To study these effects, we treated DENV2 with different concentrations of the EF peptide before adding to the Vero cells. After incubation for 72 hours, the cells were harvested and the viral genomic RNA was determined by real-time qRT-PCR method while the viral E protein was measured by cell-based flavivirus immunodetection (CFI) assay. As shown in Figure 5A and 5B, the EF peptide could reduce DENV2 RNA and protein in dose dependent manner. At a high concentration of EF (200 μM), the amount of viral RNA relative to beta-actin – a house-keeping gene, was decreased by 83.47% and the amount of viral protein by 84.15%. Treatment with individual amino acids, Glu (E) or Phe (F), or their combination at the equalmolar concentration to DENV2 did not significantly affect DENV E protein levels (supporting information, Figure S1). These results indicate that the binding of EF peptide to the hydrophobic pocket on the DENV E protein affects virus entry to the cells and reduces viral RNA and protein production.
To examine the effect of EF peptide treatment on DENV production, the virus in culture supernatant of DENV2-infected Vero cells was measured by foci-forming reduction assay. Pretreatment of DENV2 with the EF peptide before infection to Vero cells caused reduction of virus production in the culture supernatant in dose dependent manner. At the peptide concentration of 200 μM, the reduction of virus production was less than 20 folds. Concordantly, after the EF peptide treatment,
This article is protected by copyright. All rights reserved.
Accepted Article
intracellular DENV2 E protein in the infected Vero cells was markedly decreased when it was detected by immunofluorescence assay (IFA) (Figure 5C).
These
findings emphasize that the EF peptide is potent to inhibit DENV2 entry and intracellular protein production.
Inhibition of other DENV serotypes by EF peptide The envelope (E) protein of Flaviviruses shared approximately 40% homology [40] and the hydrophobic pocket is thought to be conserved [8]. We thus investigated whether the EF peptide could inhibit other DENV serotypes. After treatment of each DENV serotype with the EF peptide, intracellular DENV E protein was measured by CFI assay. At the concentration of 100 μM of the EF peptide, the E proteins of DENV 1, 3, and 4 were significantly decreased in the range of 20-40% but it could decrease the E protein of DENV 2 by approximately 70% (Figure 6A). The variable inhibitory effects of the EF peptide for different DENV serotypes were not unexpected because similar finding has previously been reported for the effects of small compound inhibitors to different DENV serotypes [10]. Notably, the ligand accessibilities or binding preferences might be determined by the slight difference of amino acid compositions at the hydrophobic pocket. Interestingly, less conserved areas (
Received Date : 12-Jan-2014 Accepted Date : 13-Feb-2014 Article type
: Research Article
Peptide Inhibitors against Dengue Virus Infection Aussara Panya 1,2, Kunan Bangphoomi 3, Kiattawee Choowongkomon 3, Pa-thai Yenchitsomanus 1,*
1
Division of Molecular Medicine, Department of Research and Development, Faculty
of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand 2
Department of Biochemistry, Faculty of Medicine Siriraj Hospital, Mahidol
University, Bangkok, Thailand 3
Department of Biochemistry, Faculty of Science, Kasetsart University, Bangkok,
Thailand
*Corresponding author. Tel: +66 2 4192777; Fax: +66 2 4110169 E-mail Address: [email protected], [email protected]
Abstract Dengue virus (DENV) infection has become a public health problem worldwide. The development of anti-DENV drug is urgently needed because neither licensed vaccine nor specific drug is currently available. Inhibition of DENV attachment and entry to host-cells by blocking DENV envelope (E) protein is an attractive strategy for antiDENV drug development. A hydrophobic pocket on the DENV E protein is essential This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an 'Accepted Article', doi: 10.1111/cbdd.12309 This article is protected by copyright. All rights reserved.
Accepted Article
for structural transition in the membrane fusion and inhibition of this process is able to inhibit DENV infection. To search for a safe anti-DENV drug, we identified short peptides targeting to the hydrophobic pocket by molecular docking. In addition, the information of predicted ligand-binding site of reported active compounds to DENV2 hydrophobic pocket was also used for peptide inhibitors selection. The di-peptide, EF, was the most effective on DENV2 infection inhibition in vitro with a half maximal inhibition concentration (IC50) of 96 μM. Treatment of DENV2 with EF at the concentration of 200 μM resulted in 83.47% and 84.15% reduction of viral genome and intracellular E protein, respectively. Among four DENV serotypes, DENV2 was the most effective for the inhibition. Our results provide the proof-of-concept for development of therapeutic peptide inhibitors against DENV infection by the computer-aided molecular design.
1. Introduction Dengue virus (DENV) infection has become a public health worldwide.
It is
estimated that approximately 50 million people are infected and over 20,000 people died each year [1]. DENV comprises 4 related serotypes, DENV1-4, transmitted via mosquito vectors.
DENV infection causes a spectrum of clinical manifestations
ranging from a mild dengue fever (DF) to a more severe dengue hemorrhagic fever (DHF) and life-threatening dengue shock syndrome (DSS) [2]. Currently, no licensed vaccine or specific drug against DENV infection is available.
The correlation
between magnitude of DENV titer and disease severity [3,4,5,6,7] suggests that the reduction of DENV level in the patients will be able to ameliorate the disease outcome and anti-DENV agents should be beneficial to the DENV-infected patients in the viremic phase. DENV and host proteins controlling the DENV life cycle are
This article is protected by copyright. All rights reserved.
Accepted Article
attractive targets for the development of anti-DENV agents. A number of inhibitors targeting to these proteins showed inhibitory activities to DENV infection. These inhibitors include small-molecule compounds [8,9,10,11,12,13], siRNAs [14,15], nucleoside analogs [16,17], antisense oligomers [18], antibodies [19,20,21,22], and peptide inhibitors [23,24,25,26,27]. DENV envelope (E) protein that has a major function in the viral entry is a key target for the development of anti-DENV agents that inhibit DENV entry by interfering with either the viral attachment or membrane fusion process. Currently, a number of anti-DENV molecules targeting to DENV E protein such as monoclonal antibodies [19,21,22], peptide inhibitors [24,25,26,27], synthetic compounds [8,9,11,13], and drug derivatives [28] have been reported. Previously, the study of structural difference between E homodimer in the absence or presence of a detergent, namely n-octyl-β-D-glycoside (BOG), revealed an area of hydrophobic pocket which serves as a hinge region responsible for structural transition [29]. Since mutations that affect pH threshold for membrane fusion are also mapped to this region, the hydrophobic pocket has been proposed as the target for fusion inhibitors [29]. This was proven by the fact that synthetic compounds targeting to this pocket showed inhibitory effects on membrane-fusion in vitro [8,13]. Peptide inhibitors have been considered as a class of therapeutic agent with diverse chemical properties and biological activities. An advantage of peptide inhibitors is that they have less toxicity, compared to synthetic compounds. They are naturally one of first-line defense in wide classes of life ranging from insects, ichthyoids, amphibians, mammals and humans to protect against bacteria, fungal, and also viral pathogens [30,31]. Accordingly, they are studied in many diseases of humans both infectious diseases such as viral infections and non-infectious diseases such as cancers
This article is protected by copyright. All rights reserved.
Accepted Article
[32]. The peptidic antiretroviral Enfuvirtide (Fuzeon) is the first clinically FDAapproved antiviral peptide inhibitor of virus-cell fusion that shows potent antiretroviral activity and significantly improves medical outcomes in highly treatment-experienced patients with HIV-1 infection [33]. This represents an example of peptide inhibitor that is developed for clinical use in the viral infection. In the case of DENV infection, a number of peptide inhibitors targeting E protein have been reported. These include peptide inhibitors derived from the stem region of E protein (residues 412-447) [24,25], peptide inhibitors optimized from original E sequences using computational optimization method [26], and peptide inhibitors de novo designed to target putative receptor-binding site [27]. These peptide inhibitors had ability to interfere with DENV2 entry and thus potentially attenuate rate of infection. However, the use of in silico design for small peptide inhibitors specific to the hydrophobic pocket of DENV E protein has not yet been reported. Targeting to this pocket requires a minimal length of peptides to optimally fit into its cavity. Thus, this study aims to identify a set of small peptide inhibitors targeting to the hydrophobic pocket of DENV2 by using the ligand-binding site information of reported active compounds and drugs combined with the computational screening. Out of seven peptides that were selected from hundreds of peptides for testing the anti-DENV effect, one peptide has striking inhibitory effects on DENV2 replication, protein synthesis, and virus production.
2. Materials and methods 2.1 Preparation of receptor and ligands for computer analysis To identify small peptides that specifically bind to the n-octyl-β-D-glycoside (BOG) or hydrophobic pocket, the molecular docking methodology was performed by using
This article is protected by copyright. All rights reserved.
Accepted Article
two docking programs; AutoDOCK v4.2 [34] and CDOCKER Discovery Studio v2.1 [35]. DENV2 E protein in complexes with BOG (PDB: 1OKE) was used as a receptor. The preparation of receptor by using AutoDOCK v4.2 and CDOCKER v2.1 programs was slightly different. Briefly, the preparation by using AutoDOCK was began by adding the charges and salvation term to the receptor. The binding site or grid box was located using “center to ligand” mode in AutoDOCKTools in dimension of x-size = 40, Y-size = 40, Z-size = 40. The AutoGrid program was performed to pre-calculate the grid maps for individual atom type of ligand. For CDOCKER, the receptor was subjected to a restrained energy minimization with CHARMm forcefield and MMFF 94 partial charge in Discovery Studio v2.1. The binding site or sphere was located by using the “Define sphere form selection” mode and expanded the sphere size up to 20 Å in radius. Ligands in this study including amino acids, dipeptides, and tri-peptides were derived from “Build and edit protein” function in DS. The structures were optimized by using Dreiding-like forcefield in “Clean geometry” mode and exported in the PDB format.
2.2 Molecular docking on DENV E protein Docking simulation against BOG pocket on the DENV E protein was carried out by using the AutoDOCK and CDOCKER programs. All 20 amino acids were firstly docked into the BOG pocket where the parameters were set as default values following the program instructions. The top 7 hits identified from both programs according to the scoring function of the particular docking software were collected and used as di-peptide building block. The dockings of di-peptides were performed to select for the top 5, which were used as scaffolds. The scaffolds were grown at the Nor C-terminus randomly dependent on the position of di-peptide at the key interaction
This article is protected by copyright. All rights reserved.
Accepted Article
site on the pocket to generate tri-peptides. The dockings of tri-peptides were again performed to collect the ligands with high binding-scores. The peptides with topranked scores from both programs with the predicted ligand-binding residue contraction were considered.
Finally, only seven peptides were selected for
experimental validation. The figures were prepared by using PyMOL programs.
2.3 Determination of ligand-binding residues The site-moiety map (SiMMap) algorithm [36] was used to identify ligand-binding residues on the BOG pocket of DENV E protein. The conformational poses derived from docking of the active drugs and compounds including doxycyclin [28], relitetracyclin [28], compound A5 [8], compound 36 [9], and compound 11 [9] against the BOG pocket by using the AutoDOCK v4.2 and CDOCKER (DS v2.1) programs were input to the SiMMap software for the analysis.
2.4 Cell culture and virus preparation DENV1 (strain Hawaii), DENV2 (strain 16681), DENV3 (strain H87), DENV4 (strain Loc3 00574/99) were propagated in Aedes albopictus (C6/36) cells cultured in L-15 medium containing 1% fetal bovine serum (FBS) and 10% tryptose phosphate broth by using multiplicity of infection (MOI) of 0.1. After incubation for 5 days at 28°C, the culture supernatant was collected and cellular debris was removed by centrifugation. The virus was stored at -70°C until it was used for experiments. Vero cells were cultured in minimum essential media (MEM) supplemented with 10% FBS and antibiotics (penicillin G and streptomycin) and were maintained at 37°C in an incubator at 5% CO2.
This article is protected by copyright. All rights reserved.
Accepted Article
2.5 Peptide synthesis Peptides were custom-made by Selleck Chemicals (Huston, TX, USA). Seven peptides that were selected and synthesized were solubilized in sterile distilled water to a final concentration of 25 mM and stored in a freezer at -25°C.
2.6 Cytoxicity test The cytotoxic effects of peptides were tested on cultured Vero cells by using PlestoBLUETM Cell Viability Reagent (Invitrogen, CA, USA). The peptides at the maximal concentration of 0.5 mM prepared in the cell culture medium (a final volume of 90 µL) were added to the Vero cells (2x104 cells per well of a 96-well plate) and incubated at 37°C in 5% CO2. After incubation for 72 hours, an assay was performed by following the manufacturer’s instructions. Briefly, the PrestoBLUETM reagent (10X) was added (10% v/v) to culture medium. The reaction was incubated at 37°C for 30 minutes and then measured absorbance at 570 nm with a reference wavelength set at 600 nm.
2.7 Virus infection and treatment with peptides The peptides at indicated concentrations were incubated with DENV (200 FFU in a 96-well plate and 1000 FFU in a 24-well plate) for 30 minutes at 37°C before adding to Vero cells (2 x 104 cells per well for a 96-well plate and 1 x 105 cells per well for a 24-well plate) for viral adsorption for 2 hours at 37°C. The unbound viruses were removed
and
the
cells
were
over-layered
by
the
medium
containing
carboxylmetylcellulose for foci-forming assay or were added with fresh medium for other assays. The infected cells were incubated at 37°C until the experiment was performed.
This article is protected by copyright. All rights reserved.
Accepted Article
2.8 Foci-forming reduction assay The foci-forming reduction assay was carried out to determine inhibitory effects of peptides on DENV infection to Vero cells. At 72 hours post infection, the numbers of foci were determined by the protocol as previously described [8]. The Vero cells were fixed with 4% paraformaldehyde for 15 minutes and washed twice with phosphate buffered saline (PBS). Cells were permeabilized with 0.2% Triton X-100 for 15 minutes at room temperature, blocking with 1% Bovine Serun Albumin (BSA) in PBS for 30 min, and incubated for 3 hours with monoclonal anti-DENV E antibody, 4G2, at 37°C. Cells were washed 5 times with PBS containing 0.5% Tween-20 (PBST) and further incubated for 1 hours with rabbit anti-mouse conjugated horseradish peroxidase (HRP).
After PBST washing, 3’ di-amino-
benzidine (DAB) peroxidase substrate (Sigma, USA) was added to stain the DENV foci, which were manually counted under a light microscope. The IC50 of each peptide was analyzed from dose-response curve by using non-linear regression function of GraphPad Prism software.
2.9 Quantitative real-time polymerase chain reaction Quantitative real-time PCR (qRT-PCR) was carried out to determine the effects of the peptides on DENV replication. After DENV infection, the cells were harvested at 48 and 72 hours and RNA was extracted by using TRIzol reagent (Invitrogen, CA, USA) following the manufacturer’s protocol. RNA (1.5 μg) was changed into cDNA by using eAMV reverse transcriptase (Promega).
qRT-PCR was conducted in a
LightCycler 480 (Roche, Mannheim, Germany) machine by using 1 μL of cDNA, DENV E specific primers; D2R (5′-CCGGCTCTACTCCTATGATG-3′) and D2L (5′ATCCAGATGTCATCAGGAAAC-3′) [37] and 2x SYBR Green I Master Mix
This article is protected by copyright. All rights reserved.
Accepted Article
(Roche, Mannheim, Germany). The data was normalized with that of the β-actin and the relative value of 1.0 derived from the infected condition without the treatment with peptide was assigned as the calibrator.
2.10 Cell-based flavivirus immunodetection To examine the inhibitory effect of peptide on viral protein synthesis, DENV protein level was determined by using cell-based flavivirus immunodetection (CFI) assay. The DENV E protein was detected by the CFI method using 4G2 antibody as previously described [10] at 72 hours post infection. Briefly, the Vero cells were fixed with 4% paraformaldehyde, permeabilized with 0.2% Triton X-100, and blocking with 1% Bovine Serun Albumin (BSA). Cells were incubated for 3 hours with monoclonal anti-DENV E antibody, 4G2, at 37°C, washed 5 times with PBS containing 0.5% Tween-20 (PBST) and further incubated for 1 hours with rabbit antimouse conjugated horseradish peroxidase (HRP). The 3,3’,5,5’-tetramethylbenzidine or TMB was added and incubated at room temperature for 15 minutes before adding sulfuric acid to stop the reaction. The absorbance was measured at OD492 by a plate reader set by using the mock control as blank.
2.11 Immunofluorescence assay Immunofluorescence assay (IFA) was performed to examine the percentage of DENV-infected cells after the peptide treatment.
At 72 hours of infection, the
infected cells were harvested and stained for DENV E protein with monoclonal 4G2 antibody and Alexa Fluor® 488 Goat Anti-Mouse IgG while cell nuclei were stained with Hoechst, by the method as previously described [12,38].
This article is protected by copyright. All rights reserved.
Accepted Article
2.12 Statistic analysis In every experiment, at least three independent assays were performed. The average quantitative values (means) and their standard deviation were determined, and student’s t tests were performed by using GraphPad Prism software. P value of less than 0.05 was considered to be statistically significant.
Results and discussion Identification of ligand-binding hot spots The identification of specific sites (or hot spots) for ligand binding on the receptor protein is helpful for designing inhibitors that interfere with its biological function. Hot spots at the hydrophobic (BOG) pocket of DENV E protein were firstly characterized by testing the interactions between each of the five reported compounds (doxycyclin, relitetracyclin, compound A5, compound 36, and compound 11) and the hydrophobic pocket of DENV E protein by using SiMMap program. Three distinct hot spots (H1, H2, and H3) at the hydrophobic pocket of DENV E protein, which were expected to be critical for binding and involving in the inhibitory effect of these compounds, were identified (Figure 1). The H1 and H3 hot spots located at the entrance of the pocket but they situated at the opposite direction. The H1 hot spot, containing Lys128 and Gln200, has partially positive charge whereas the H3 hot spot, containing The48, Glu49, and Ala50, has partially negative charge. The H2 hot spot lined at the bottom of the pocket with hydrophobic amino acids, Ile270, Leu277, and The280. From the inspection of the interaction between the ligands and the pocket, it revealed that one end of the ligands inserted into the pocket to interact with H2 suggesting the preference of H2 in the interaction. This might be attributable to the narrow space at the middle of pocket and the ligands were possibly locked to the H2
This article is protected by copyright. All rights reserved.
Accepted Article
position. On the other hand, the wider space at the entrance of the pocket might allow the flexibility of interaction between ligands and H1 or H3 through the favorable electrostatic property that might contribute to their binding affinities. The analysis of the ligand-pocket interaction suggested that the binding of two third of the hot spots, H1-H2 or H3-H2, could contribute to the inhibitory properties of these known compounds. To explore this concept, the interactions of ligands with either H1-H2 or H3-H2 were also considered as an extra criterion for selection of peptide inhibitors in this study.
Design of small peptides To validate the docking methodologies by AutoDOCK and CDOCKER programs, the two programs were used to simulate the interaction of BOG ligands to the hydrophobic pocket of DENV E protein.
The superimposes of docking results
obtained from the simulation using these two programs to BOG in the reported crystal structure revealed the overlaps of molecules with root mean square deviation (RMSD) of 0.667 and 1.61 Å, respectively (data not shown). As the resulted RMSDs were less than 2 Å, the docking methodologies by both programs were considered to be acceptable [35]. To identify small peptides binding to the hydrophobic pocket of DENV E protein, the top 7 hits of amino acids obtained from the docking simulation were taken to construct the peptides containing two amino acids. The peptides were then used as scaffolds to create the peptides containing three amino acids. When all peptides were used for docking to the hydrophobic pocket by both programs, altogether 304 hits were obtained. In the final selection, only the peptides that formed bonds with the predicted hot spots in the hydrophobic pocket were chosen. Seven peptides with high
This article is protected by copyright. All rights reserved.
Accepted Article
docking scores including Lys-Glu-Asn (KEN), Trp-Glu-Asn (WEN), Glu-Glu-Gly (EEG), Trp-Glu-Thr (WET), Glu-Gln (EQ), Glu-His (EH), and Glu-Phe (EF), which bound to at least one hot spot, were finally picked for further studies.
Inhibition of DENV2 infection by small peptides Seven small peptides (KEN, WEN, EEG, WET, EQ, EH, and EF) were custom-made by Selleck Chemicals (Huston, TX, USA). To evaluate the cytotoxic effect of these peptides, individual peptide at the concentration of 500 μM was tested by adding to Vero cells – a permissive cell line for DENV2 infection [39], and cell viability was measured at 72 hours. The results showed that cell viabilities were not decreased (Figure 2A), suggesting that these peptides had no cytotoxic effect to the cells at the high concentration as 500 μM.
The peptides were then tested for their abilities to inhibit foci formation of DENV2. Almost all peptides could reduce the foci formation of DENV2 (Figure 2B). Among these peptides, EF and KEN could reduce the foci formation to approximately 90% while other peptides reduced it to 40-60%. However, EQ could not reduce the foci formation of DENV2.
The correlation between peptide concentrations and
percentages of virus production inhibition was examined (Figure 3A and 3B) and the half maximal inhibitory concentration (IC50) of peptides was also determined. EF and KEN had IC50 of 96.50 μM and 331.9 μM, respectively, which were concordant to their predicted inhibition constant (Ki) of 98 μM and 455 μM, respectively, as predicted by the AutoDOCK program (supporting information, Table 1). The reason why EF had a better efficiency than KEN might be explained by that EF could interact with the H1 and H3 hot spots at the entrance of hydrophobic pocket and
This article is protected by copyright. All rights reserved.
Accepted Article
also with the H2 hot spot at the bottom of the pocket while KEN could interact with only the H1 and H3 hot spots of the pocket (Figure 4A). Hydrogen bonds between side chains of EF and those of the H1 and H3 hot spots of the pocket are crucial for the interaction. EF interacted with H1 and H3 by using hydrogen bond between the carboxylate-group side chain of Glu (E) and the NH-group side chain of Lys128 of H1 in addition to the NH-group main chain of Phe (F) to carboxylate-group main chain of The48 of H3 (Figure 4B, 4D). In addition, phenyl ring of Phe (F) made contact with H2 containing Ile270, Leu277, Phe279, and The280 at the bottom of the hydrophobic pocket. KEN formed hydrogen bonds with H1 and H3 but did not have hydrophobic interaction with H2 (Figure 4C, 4E). The carboxylate-group side chain of Glu (E) in KEN formed hydrogen bonds to the NH-group side chains of Lys128 and Gln200 of H1 whereas the NH-group side chain of Asn (N) formed hydrogen bonds to carboxylate-group main chain of The48 of H3 and nearby Ser274. In contrast to EF, KEN could only interact to the upper hydrophilic part of the pocket. The hydrophilic feature of KEN is likely to prevent it to enter into the hydrophobic environment within the pocket, making it loosely binding to the pocket. Additionally, a large side chain of KEN might prevent it to deeply insert into the pocket. The result of molecular docking showed that the occupation of the EF peptide into the hydrophobic pocket did not entirely fit, especially at the upper part of pocket; hence, it is possible for any modification to improve peptide affinity, specificity, and half-life. Additionally, the peptide for oral administration is a challenge because peptide inhibitor often requires intravascular route of administration as it has low gastrointestinal absorption.
This article is protected by copyright. All rights reserved.
Accepted Article
Reduction of DENV RNA, protein, and virus production by EF peptide The EF peptide was designed specific to the hydrophobic pocket on the DENV E protein – the site that affects DENV entry to the host cells [8,10,13]. Inhibition of DENV entry by binding of EF peptide to this critical site would consequently result in the reduction of DENV intracellular activities including RNA replication, protein synthesis, and virus production. To study these effects, we treated DENV2 with different concentrations of the EF peptide before adding to the Vero cells. After incubation for 72 hours, the cells were harvested and the viral genomic RNA was determined by real-time qRT-PCR method while the viral E protein was measured by cell-based flavivirus immunodetection (CFI) assay. As shown in Figure 5A and 5B, the EF peptide could reduce DENV2 RNA and protein in dose dependent manner. At a high concentration of EF (200 μM), the amount of viral RNA relative to beta-actin – a house-keeping gene, was decreased by 83.47% and the amount of viral protein by 84.15%. Treatment with individual amino acids, Glu (E) or Phe (F), or their combination at the equalmolar concentration to DENV2 did not significantly affect DENV E protein levels (supporting information, Figure S1). These results indicate that the binding of EF peptide to the hydrophobic pocket on the DENV E protein affects virus entry to the cells and reduces viral RNA and protein production.
To examine the effect of EF peptide treatment on DENV production, the virus in culture supernatant of DENV2-infected Vero cells was measured by foci-forming reduction assay. Pretreatment of DENV2 with the EF peptide before infection to Vero cells caused reduction of virus production in the culture supernatant in dose dependent manner. At the peptide concentration of 200 μM, the reduction of virus production was less than 20 folds. Concordantly, after the EF peptide treatment,
This article is protected by copyright. All rights reserved.
Accepted Article
intracellular DENV2 E protein in the infected Vero cells was markedly decreased when it was detected by immunofluorescence assay (IFA) (Figure 5C).
These
findings emphasize that the EF peptide is potent to inhibit DENV2 entry and intracellular protein production.
Inhibition of other DENV serotypes by EF peptide The envelope (E) protein of Flaviviruses shared approximately 40% homology [40] and the hydrophobic pocket is thought to be conserved [8]. We thus investigated whether the EF peptide could inhibit other DENV serotypes. After treatment of each DENV serotype with the EF peptide, intracellular DENV E protein was measured by CFI assay. At the concentration of 100 μM of the EF peptide, the E proteins of DENV 1, 3, and 4 were significantly decreased in the range of 20-40% but it could decrease the E protein of DENV 2 by approximately 70% (Figure 6A). The variable inhibitory effects of the EF peptide for different DENV serotypes were not unexpected because similar finding has previously been reported for the effects of small compound inhibitors to different DENV serotypes [10]. Notably, the ligand accessibilities or binding preferences might be determined by the slight difference of amino acid compositions at the hydrophobic pocket. Interestingly, less conserved areas (
