key: cord-002463-qhtj1pef
authors: Dash, Raju; Das, Rasel; Junaid, Md; Akash, Md Forhad Chowdhury; Islam, Ashekul; Hosen, SM Zahid
title: In silico-based vaccine design against Ebola virus glycoprotein
date: 2017-03-21
journal: Adv Appl Bioinform Chem
DOI: 10.2147/aabc.s115859
sha: 
doc_id: 2463
cord_uid: qhtj1pef

Ebola virus (EBOV) is one of the lethal viruses, causing more than 24 epidemic outbreaks to date. Despite having available molecular knowledge of this virus, no definite vaccine or other remedial agents have been developed yet for the management and avoidance of EBOV infections in humans. Disclosing this, the present study described an epitope-based peptide vaccine against EBOV, using a combination of B-cell and T-cell epitope predictions, followed by molecular docking and molecular dynamics simulation approach. Here, protein sequences of all glycoproteins of EBOV were collected and examined via in silico methods to determine the most immunogenic protein. From the identified antigenic protein, the peptide region ranging from 186 to 220 and the sequence HKEGAFFLY from the positions of 154–162 were considered the most potential B-cell and T-cell epitopes, correspondingly. Moreover, this peptide (HKEGAFFLY) interacted with HLA-A*32:15 with the highest binding energy and stability, and also a good conservancy of 83.85% with maximum population coverage. The results imply that the designed epitopes could manifest vigorous enduring defensive immunity against EBOV.

Ebola virus (EBOV) is an antisense-strand RNA virus from the Filoviridae family, and it is structurally filamentous. 1 Although the initial discovery of EBOV was in 1976, till now more than 24 epidemics have been reported from Africa, mostly with the Zaire species (http://who.int/mediacentre/factsheets/fs103/en/). [2] [3] [4] The genome of EBOV enciphers the seven structural proteins, ie, nucleoprotein (NP), viral structural proteins (VP35, VP40, VP30, and VP24), glycoprotein (GP), and RNA-dependent RNA polymerase (L). 5 Among these, three different versions of glycoprotein are transcribed by the GP gene. [6] [7] [8] [9] Both attachment protein (GP1) and entry/fusion protein (GP2) are expressed from the full length of the GP chain, which are synthesized from messenger RNAs (mRNAs), containing an additional nontemplated adenosine. The soluble GP (sGP) is synthesized from the unedited RNA transcript. On the contrary, small soluble GP (ssGP) is translated during this process by adding two additional adenosine residues. 10 The GPs are expressed virally on the virion surface, which plays a crucial role in the catalysis of membrane fusion and amalgamation to host cells. As a result, it is considered not only a crucial component for vaccines but also an essential target for developing inhibitors and antibodies of attachment and fusion. [11] [12] [13] Protein sequence retrieval, evaluation analysis, and antigenic protein identification All available sequences of the GP of EBOV were extracted from the UniProt database. 20 After that, multiple sequence alignment was performed by using the ClustalW2 tool, and a phylogenetic tree was assembled by MEGA 6.0 21 software. And then, VaxiJen v2.0 22 was used to predict most efficient antigenic protein from the available protein sequences. Top scored eiptope subjected to 100 ns MD simulation **RMSF **RMSD **Hydrogen bond occupency analysis Secquence, having highest vaxijen score Prediction of B cell epitope, using-**T cell epitope prediction by proteasomal C terminal cleavage, TAP transport efficiency and MHC class 1 binding **Epitopes with IC50 value less than 50 for their binding to MHC class 1 molecule from IEDB analysis along with binding to highest number of alleles in both analyses were chosen **Epitope conservancy analysis **Population coverage analysis **Kolaskar and Tongaonkar antigenicity scale 48 **Emini surface accessibility prediction 47 **Karplus and Schulz flexibility prediction 49 **Bepipred linear epitope prediction 50 **Chou and Fasman beta turn prediction 52 Vaxijen analysis with a threshold score of >0. 5 Secquence, having highest vaxijen score

Vaxijen analysis with a threshold score of >0.5 

In silico-based vaccine design against EBOV GP T-cell epitope identification and conservancy analysis T-cell identification was done using the NetCTL 1.2 server, 22 setting thresholds at 0.5, 0.89, and 0.94 for sensitivity and accuracy. MHC-I binding of the identified epitopes and epitope conservancy were then calculated using tools from the immune epitope database (IEDB). [24] [25] [26] These tools calculate the half maximal inhibitory concentration (IC 50 ) value of epitope binding to human leukocyte antigen (HLA) molecules using the stabilized matrix base method. 26, 27 The restriction for epitope identification was set to 12 MHC-I supertypes. Prior to the run, all the alleles were considered, and the length of the peptides was set at 9.0.

The population coverage tool from IEDB was applied to determine the population coverage for every single epitope by selecting HLA alleles of the corresponding epitope. Allergenicity of the predicted epitope was calculated using AllerHunter, 27 which can predict both nonallergens and allergens with a high level of accuracy , by comparing the input sequence with the sequence of known allergen. 29 Molecular simulation analysis of HLA allele interaction Design of the three-dimensional structure of epitope and HLA protein

The three-dimensional structures of all the five epitopes were predicted by a PEP-FOLD web-based server. 30 For each sequence, this server predicted the five most provable structures, the best of which, having the lowest energy model, was chosen for further analysis.

To validate the binding of identified epitope and HLA molecule, we considered the homology modeling as there is no relevant structure available in the protein data bank. We selected homology modeling using the most popular online protein fold recognition server, Phyre2, 31 to generate the three-dimensional structure of HLA-A*32:15 32 (accession id: AM422702). Then, ModRefiner 33 was used to minimize and correct the hypothetical structure. The validation of the predicted structure was done using PROCHECK, 34 verify 3D, 35 ERRAT, 36 PROVE, 37 and QMEAN. 38 

Molecular docking analysis was performed using AutoDock Vina, 39 by considering HLA molecule as a protein and identified epitopes as ligands. First, we used the protein preparation wizard of UCSF Chimera 40 to prepare the hypothetical protein for docking analysis by adding hydrogens and Gasteiger-Marsili charges. 41 The prepared file was then converted into pdbqt format. The parameters used for the docking simulation were set to default. The size of the grid box in AutoDock Vina was kept at 36.3095, 54.3374, and 48.025, respectively, for X, Y, and Z. The energy range was kept at 4, according to the default setting. AutoDock Vina was implemented via the shell script offered by AutoDock Vina developers. Docking results were observed by negative score in kcal/mol, as binding affinity of ligands. 39 Binding energy estimation and molecular dynamics (MD) simulation

The binding free energy of HLA-epitope complexes were calculated by using MM (CHARMm) 42 -Generalized Born Surface Area (GBSA) and Poisson-Boltzmann Surface Area (PBSA) protocols, implemented in Accelrys Discovery Studio 2.5. Using implicit solvent models of GBSA and PBSA, the binding free energy (ΔG bind ) for each epitope was calculated by maintaining salt concentration of 0.15 M. Default value was set for conformational entropy and ligand minimization. The distance cutoff value was set to 14.0 Å. The binding energy was calculated by using following equation:

The entire dynamics simulation study for the HLAepitope complex was accomplished in YASARA Dynamics software. Prior to simulation, the complex was cleaned and optimized the hydrogen bond network. 43 After that, a cubic simulation cell was created with a periodic boundary condition, and the atoms of the complex were typed using the AMBER14 44 force field. The pKa (acid dissociation constant) values of protein titratable amino acids were calculated and solvated the simulation box using the transferable intermolecular potential3 points (TIP3P) water model (density: 0.997 g/L -1 ). The system consistent with 46406 atoms was energy minimized using the steepest gradient approach (5000 cycles) followed by simulated annealing method. Restrained and unrestrained all-atom molecular dynamics simulation were performed in solvent using the PME method to describe long-range electrostatic interactions at a cut off distance of 8 Å at physiological conditions (298 K, pH 7.4, 0.9% NaCl). 45 A multiple time step algorithm together with a simulation time step interval of 2.50 fs was chosen. 46 Molecular dynamics simulations of 100 ns long were performed at constant temperature using a Berendsen thermostat and constant submit your manuscript | www.dovepress.com

Dash et al pressure. The MD trajectories were saved every 250 ps for analysis. The trajectories generated from the simulation were analyzed for the stability by various evaluative measures viz. RMSD, RMSF (RMS fluctuations), and initial and final protein backbone comparisons using YASARA structure built in macros and VMD software.

To detect B-cell epitope, various tools from IEDB were used to identify the B-cell antigenicity, together with the Emini surface accessibility prediction, 47 Kolaskar and Tongaonkar antigenicity scale, 48 Karplus and Schulz flexibility prediction, 49 and Bepipred linear epitope prediction analysis. 50 Since antigenic parts of a protein belong to the beta turn regions, 51 the Chou and Fasman beta turn prediction tool 52 was also used.

A total of 46 GP sequences from the different variants of EBOV were collected from the UniProtKB database. Multiple sequence alignment analysis was then performed, and a phylogenetic tree (Figure 2 ) was constructed thereby. Using the unweighted pair-group method with arithmetic mean, a phylogram was constructed using the bootstrap with 1,000 replications in MEGA6. 53 From the multiple sequence alignment analysis, it is clearly seen that protein sequences that isolated from various strains were having a close relationship. Also, from the multiple comparison result, the selected sequences of EBOV of the same subtype have 78%-99% similarity. This result also confers the possibilities of mutation in glycoprotein of all strains, which demonstrates a good agreement with the results from Veljkovic et al. 54 Antigenic protein prediction Protein sequences in this study were considered to screen out using VaxiJen web server for the identification of potent antigenic protein. As a corollary, UniProtKB id: Q9YMG2 was identified as the most potent antigenic protein having a maximum total prediction score of 0.5390. Here the threshold of 0.5 is considered as the potent antigenicity. 55 This sequence was used for further analysis.

On the basis of the high combinatorial score, the five best epitopes were predicted by the NetCTL server from the selected protein sequence in a preselected environment. The identified epitopes are represented in Table 1 . In combination with several methods such as proteasomal cleavage/transporter associated with antigen processing (TAP)/MHC-I combined predictor, MHC-I processing of the NetCTL server calculates an overall score for each peptide's intrinsic potential from a protein for the designing of T-cell epitope. Peptides with a higher score represent higher processing capabilities.

The five T-cell epitopes were subjected to MHC-I binding prediction, using the stabilized matrix base method. The epitopes that elicited higher affinity (IC 50 <200 nM) were subjected to afterward analysis (Table 2) . Notably, proteins are transformed into peptides by proteasome complex, which cleaved the peptide bonds. By combining with Class I MHC molecules, these peptides were deported to the cell membrane, where they were introduced to T helper cells.

As shown in Table 2 Furthermore, this epitope retained the highest conservancy of 83.85%, according to the IEDB conservancy analysis, as tabulated in Table 2 . As population coverage in vaccine design generally plays a crucial role, it was calculated in this study.

The cumulative percentage of population coverage was obtained for the predicted epitope HKEGAFFLY. As shown in Table 3 , the population coverage for East Africa was found to be 66.98%; in West and North Africa, it was 69.50% and 63.89%, respectively; and for Central Africa it was observed to be 75.93%. The population coverage was recorded at 55.88% for the East Asian region, which was a major hotspot for viral infection. For North America, the population coverage was found to be 58.69%. In current vaccine design pipeline, allergenicity is considered the most prominent barrier in vaccine designing, since most vaccines convert the immune system into an "allergic" reaction 56 by inducting Type 2 T helper cells and immunoglobulin E. That is why we predicted allergenicity of the selected epitope by the AllerHunter web server, where the probability is >0.06. The epitope HKEGAFFLY was scored 0.00 (sensitivity =91.6%, specificity =89.3%), and was thus considered a nonallergen, according to the Food and Agriculture Organization/World Health Organization evaluation system of allergenicity prediction. 

Dash et al protein model having >90% of the residues in the core and allowed regions can be considered a high-quality model. 57 The hypothetical model was further analyzed using ERRAT and Verify 3D. 58 For a good model, structure should retain an ERRAT score >80.00, against which the model in this study obtained an ERRAT score of 89.859. 55 Verify 3D graph indicates that 100.00% of residues of this model had an averaged 3D-1D score of 0.2, which is good. 59 Along with the QMEAN analysis, the protein model in our interest resulted in a Z-score of −1.33, and the total score was 0.636. This value denotes a higher quality of the model, where the acceptable score ranges between 0 and 1 ( Figure 3B ). 38 On the basis of the results obtained from the aforementioned structural validation programs, the model ( Figure 3C ) showed much reliability and was considered for further study. Molecular docking simulation revealed that the proposed epitopes bound in the cleft of the HLA-A*32:15 ( Figure S2 ), where the highest binding affinity was −7.6 kcal/mol (Table S2 , observed for the HKEGAFFLY epitope). The Chimera 40 program was used to visualize the interactions of docked HLA-A-epitope complexes, as shown in Figures 4 and S2. Then, binding energy calculation was carried out to understand the binding of HLA with epitopes. Here the binding free energies of MM-GBSA and MM-PBSA are approximate free energies of binding, so a more negative value denotes stronger binding. From MM-GBSA analysis, the highest binding free energy was observed for HLA-A*32:15 with epitope (HKEGAFFLY) of -63.89 kj/mol (Table S1 ). On the contrary, the lowest binding free energy was obtained for ATEDPSSGY epitope, i.e. -44.86 kj/mol. In contrast of MM-GBSA, the HKEGAFFLY epitope was also resulted the highest binding energy of -38.48 kj/mol, while the lowest binding free energy was seen for TEDPSS-GYY epitope, -20.98 kj/mol. Since HKEGAFFLY epitope obtained the highest docking affinity and binding free energy, its complex subjected for molecular dynamics simulation. Table 2 Interaction, binding, and conservancy of identified T-cell epitopes Validation of predicted T-cell epitope

As described in the "Materials and methods" section, the hypothetical structure of HLA-A*32:15 protein was generated using the homology technique. The structure was then analyzed through various web-based protein validation software. As shown in Figure 3 , the Ramachandran plot generated by the PROCHECK 34 server showed that about 98.9% of the residues of protein are located in the most favored region, as against 0% in the outlier region and 1.1% in the generously allowed region. It should be noted that the and remained stable in the range from 2.0 Å to 3 Å. In case of epitope, similar RMSD pattern was observed, where the order of magnitude was seen to fluctuate in some range. The average energy of the simulation was -578125.270 kj/mol; the average Coulombic charge and van der Waals interactions was -694749.662 kj/mol, 77122.511 kj/mol, respectively. We also calculated the contribution of each residue for both HLA and epitope in the simulation, in terms of RMSF and RMSD. As seen in Figure 6A , highest RMSD was observed

The 100 ns MD simulation of HLA-epitope (HLA-A*32:15-HKEGAFFLY) complex was carried out using AMBER14 force field, following the energy minimization protocol. The stability of the HLA-epitope complex by means of RMSD was calculated and rendered in Figure 5A . From the results, it is revealed that the HLA molecule was stabilized after 5 ns simulation and tended to remain in plateau phase thereafter for rest of the period. The RMSD value of HLA was observed to grow up quickly from 0. for ARG residue at the position of 180 in HLA, while lowest RMSD observed for CYS100. However, this residue was also resulted highest RMSF value of 7.181 Å, while the rests of the residues were in lowest fluctuation. In case of epitope, the histidine residue at the first position and the tyrosine residue in 9th position were seen to be very much flexible, as these residues were resulted with highest RMSD and RMSF ( Figure 6B ). In the meanwhile, we calculated the number of hydrogen bond formed between the epitope and HLA molecule during the simulation. The results represented in Figure 5B , showed that hydrogen bond at initial stage was 236, and the range decreased to 160. During the simulation, the number of hydrogen bond was at a range of 160-210, 

In silico-based vaccine design against EBOV GP potentiality to express the B-cell response. Furthermore, the surface accessibility of the protein was also analyzed using the Emini surface accessibility prediction methods, since a potent B-cell epitope should be accessible through the surface. 47 As shown in Figure S4 and Table 5 , higher accessibility was found in regions 9-17 and 186-223 amino acid residues. Figure S5 represents the β-turns region identified by Chou and Fasman β-turn methods. 52 According to the result, the region from 200 to 220 (in the region of 200-220 and 105-150) is regarded as β-turns as well as hydrophilic in nature. These are two properties required to be a potent B-cell epitope. 60 Experimentally, antigenicity is related to the protein flexibility. 61 That is why we implemented the Karplus and Schulz flexibility prediction method, where it was evident that the regions of 255-280 and 200-220 were regarded as the most flexible ( Figure S6 ). Finally, based on the Hidden Markov model, the Bepipred linear epitope prediction tool was utilized to predict linear B-cell epitopes. The predicted result is rendered and tabulated in Figure S7 and Table 6 . Hence, by comparing the foregoing results, the peptide sequences ranging from 186 to 220 are which indicates the strong binding of epitope-HLA complex. Hence, all analyses lead to the conclusion that HKEGAFFLY is one of the most prominent T-cell epitopes for GP based designing of vaccine.

For the identification of potential B-cell epitopes, amino acid scale base methods have been used in this study. Consistent with this protocol, we used diverse investigation processes for the calculation of an incessant B-cell epitope.

According to the analysis of Kolaskar and Tongaonkar's 48 antigenicity prediction method, the average antigenicity was 1.028, while 1.225 and 0.894 were the maximum and minimum, respectively. The Kolaskar and Tongaonkar 48 antigenicity prediction uses a semiempirical method to predict antigenicity on the basis of physicochemical properties of the residues in a protein and their diversity in experimentally known epitopes, where values >1.00 were considered to denote a potential antigen. As summarized in Table 4 and Figure S3 [62] [63] [64] [65] [66] However, the information representing the population coverage in the worldwide are still limited. In such case, computational based epitiope screening is very much efficient in context of HLA class I molecules, 67 and also much safe, high specificity and cost effective. Therefore, this study incorporated various immunoinformatics and molecular modelling tools to identify potential epitopes present in EBOV GPs.

Initially, a set of 46 glycoprotein sequences from the different strains of EBOV has been subjected to perform multiple sequence alignment. Previous GP sequences analysis of different strains of each EBOV species revealed a high degree of sequence similarity, 68, 69 and thereby, it is believed that targeting GP from old strain could provide strong and cross reactive immunity against the new strain and previous outbreaks in 2014. 70 Interestingly, in our molecular analysis, we have found ~98-99% conservation for the amino acid sequences of different strains within the species, which confers the degree of 1  14  14  F  1  2  57  59  LSS  3  3  73  106  NGVATDVPSATKRWGFRSGVPPKVVNYEAGEWAE  34  4  114  131  KKPDGSECLPAAPDGIRG  18  5  141  148  VSGTGPCA  8  6  175  176  TF  2  7  191  193  KDF  3  8  198  215  PLREPVNATEDPSSGYYS  18  9  223  229  TGFGTNE  7  10  261  270  YTSGKRSNTT  10  11 279 285 PEIDTTI 7 1  4  11  TGILQLPR  8  2  17  56  TSFFLWVIILFQRTFSIPLGVIHNSTLQVSDVDKLVCRDK  40  3  63  69  LRSVGLN  7  4  76  82  ATDVPSA  7  5  89  99  RSGVPPKVVNY  11  6  118  126  GSECLPAAP  9  7  132  154  FPRCRYVHKVSGTGPCAGDFAFH  23  8  156  172  EGAFFLYDRLASTVIYR  17  9  177  189  AEGVVAFLILPQA  13  10  194  202  FSSHPLREP  9  11  211  221  SGYYSTTIRYQ  11  12  233  247  LFEVDNLTYVQLESR  15  13  249  259  TPQFLLQLNET  11  14  274  280  IWKVNPE  7 able to provoke the immune response as B-cell epitope for GP-based designing of vaccine.

In recent trends, the primary focus of vaccine development is very much rely on GPs, as they are involved in cell attachment, fusion and entry as well as assist in invasion; and thus plays the role of pathogenesis of disease. The central role 

In silico-based vaccine design against EBOV GP similarity and support the previous analysis. From this set of GPs, the most antigenic protein sequence was determined by Vaxijen server. Based on auto cross covariance (ACC), the Vaxijen server transform the protein sequence into uniform vectors of physicochemical properties of proteins. With 91% sensitive, 82% accuracy and 72 specificity, the l00-CV (leave one -out cross validation) was used to identify antigenicity of protein for viral species. 71 The resultant antigenic protein (VaxiJen score ≥0.5) was then subjected for various immunoinformatics analysis, followed by IEDB web server. At the beginning five potent 9-mer epitopes have been predicted from NetCTL 1.2 server and selected for further study. Using the threshold of 0.5, the NetCTL 1.2 server predicts maximum number of epitopes without compromising the specificity or sensitivity levels, covering all 12 MHC class I supertypes. 23 The five most potent epitopes are represented in Table 1 , and the scores are the predicted MHC class I affinities in the form of -logIC50 and IC50 value.

For MHC-I binding prediction, peptides with IC50 values <50 nM are considered high affinity, <500 nM for intermediate affinity, and <5000 nM for low affinity. Therefore, we selected maximum alleles having binding affinity <200 nM. 72 It is advocated that T-cell epitope binding to specific multiple HLA supertypes are termed as promiscuous in vaccine design, since they effectively increase the coverage of higher proportions of human populations. 73, 74 According to the results, both HKEGAFFLY and LFEVDNLTY bind to the highest number of alleles. However, HKEGAFFLY represents highest conservancy and was hence considered as epitope of choice.

We also validated each epitope by molecular docking simulation and MM-GBSA/MM-PBSA studies with HLA-A*32:15 protein, as it was found common in the results from MHC-I binding interaction analysis. Prior of docking simulation, the three dimensional structure of HLA molecule was prepared by using the Phyre2, followed by intensive mood. As a result, 179 residues (99%) of HLA-A*32:15 modelled at >90% accuracy. In these study, 2BCK_Chain A, crystal structure of HLA-A*2402 showed the highest similarity of 90% ( Figure S1 ). The selected model has been chosen from the twenty models generated by Phyre2, on the basis of similarity and confidence level. Phyre2 is one of the best protein prediction servers that allows remote fold reorganization and homology detection. Using hidden Markov model (HMM), this server predicts the structure of given protein sequence by constructing backbone, loop modelling and adding side chains. 75 However in intensive mode, additional ab initio approach is used for reconstruction of missing region, backbone and side chain. 75 In docking simulation, among the other epitopes, HKEGAFFLY obtained highest binding affinity (Table S1 , supplementary material). In addition, MM-PBSA and MM-GBSA techniques are frequently to re-rank docking poses from molecular docking study, as they achieve a much better performance than docking scoring functions. 76 Nevertheless, the success rate of the absolute binding free energy prediction strongly depends on the systems. 77 Thence, we used both of these solvation models to predict binding energy more accurately, where the results from MM-PBSA examine the accuracy and reliability of the results from MM-GBSA. Results of binding energy calculation are shown in Table S1 . The relative magnitude of binding free energy obtained from GB methods is found to be consistent with those calculated using PB method, despite of the differences in the absolute value of salvation energy. As a corollary, these results also demonstrated the consistence with relative stabilities of HLA-epitope complexes. In previous published reports regarding in silico epitope identification of EBOV, [78] [79] [80] [81] the studies are limited to sequence-based scoring function techniques and some extend to docking simulation. These techniques have certain limitations, though these are very useful. 82 In docking simulation of peptide and protein, it faces problem like peptide's flexibility. 83, 84 Whereas, energy based approach like molecular mechanics and interaction energy scoring can add valuable information to sequence based results. 85, 86 Therefore, we have performed MD simulation study of 100 ns long to enhance the predictive power of the peptide affinity calculations to MHC molecule. In molecular dynamics simulation, both epitope and HLA protein were seen to achieve equilibration, while different fluctuations of RMSD were seen by the time evolution. Higher RMSD values of epitope indicate the flexibility in binding with HLA molecule, during the simulation. These results were further confirmed by the analysis of per residue contributions in dynamics simulation by means of RMSF and RMSD. Low values of RMSF indicate the core region of the HLA protein was stable, while high values of RMSD demonstrated the motion of the protein during the simulation. In like manner, the RMSD and RMSF profiles of eptiope confirm the synergic conformation changes to accommodate the binding pocket of HLA. The hydrogen bond occupancy analysis between the HLAepitope further confirmed the stability of the complex during the simulation. Overall, these results evidently demonstrate that both HLA and epitope have remarkable conformation changes to facilitate the binding and formed stable complex in thermodynamic environment (Figure 7 ). submit your manuscript | www.dovepress.com Dash et al It is one of important factors in vaccine design that the distribution of HLA varies according to the diverse ethnic groups and geographic regions around the world. Therefore, wide range of population coverage must be considered during the designing of an effective design. According to the results from population coverage analysis, the epitope HKEGAFFLY showed wide range of population coverage in different regions of the world (Table 3) , where the highest coverage was observed Central Africa; one of the most EBOV infected areas. This result indicates that it will specifically bind with the prevalent HLA molecules in the target population, where the vaccine will be employed.

In other aspects, the B-cell epitope stimulates minimal immune unity, which is very much strong enough to elicit a potent humoral immune response, causing no harmful side effects to human body. Thereby, we are also calculated and found that the sequences ranging from 186-220 as a B-cell epitope, by taking consideration of amino acid property, hydrophilicity, accessibility, flexibility, turns, exposed surface, polarity and antigenic propensity. This study could provide a solid base for vaccine design.

In recent years, most vaccines have been developed based on B-cell immunity; however, the current strategy relies mostly on T-cell epitope owing to long-lasting immunity. Both B-cell and T-cell epitopes are offered in this study for stimulating immunity in several ways. The resulting peptides showed B-cell and T-cell selectivity, better conservancy, population coverage, and significant interaction with MHC-1 allele with good affinity. Above all, the predicted epitopes are anticipated to offer long-term and high protective immunity against EBOV.

The authors report no conflicts of interest in this work. Computational Biophysics; Chemoinformatics and Drug Design; In silico ADME/Tox prediction. The manuscript management system is completely online and includes a very quick and fair peer-review system, which is all easy to use. Visit http://www.dovepress.com/testimonials. php to read real quotes from published authors.

Dash et al 

Ebola, Marburg and Disease

Ebola viral disease: what should be done to combat the epidemic in 2014?

Treatment of ebola virus disease

Ebola viral disease outbreak -West Africa

Genome structure of Ebola virus subtype Reston: differences among Ebola subtypes

Genomic RNA editing and its impact on Ebola virus adaptation during serial passages in cell culture and infection of guinea pigs

Ebola virus RNA editing depends on the primary editing site sequence and an upstream secondary structure

Deep sequencing identifies noncanonical editing of Ebola and Marburg virus RNAs in infected cells

Ebolavirus delta-peptide immunoadhesins inhibit marburgvirus and ebolavirus cell entry

The multiple roles of sGP in Ebola pathogenesis

Ebolavirus glycoprotein structure and mechanism of entry

Structure of the Ebola virus glycoprotein bound to an antibody from a human survivor

Prediction and identification of mouse cytotoxic T lymphocyte epitopes in Ebola virus glycoproteins

A highly conserved WDYPKCDRA epitope in the RNA directed RNA polymerase of human coronaviruses can be used as epitope-based universal vaccine design

More than one reason to rethink the use of peptides in vaccine design

Immunogenicity and safety of a novel therapeutic hepatitis C virus (HCV) peptide vaccine: a randomized, placebo controlled trial for dose optimization in 128 healthy subjects

Conserved epitopes of influenza A virus inducing protective immunity and their prospects for universal vaccine development

Approaching rational epitope vaccine design for hepatitis C virus with meta-server and multivalent scaffolding

Construction and immunological evaluation of multivalent hepatitis B virus (HBV) core virus-like particles carrying HBV and HCV epitopes

In silico-based vaccine design against EBOV GP

UniProt: the Universal Protein knowledgebase

Clustal W and Clustal X version 2.0

VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines

Large-scale validation of methods for cytotoxic T-lymphocyte epitope prediction

Development of an epitope conservancy analysis tool to facilitate the design of epitope-based diagnostics and vaccines

Sensitive quantitative predictions of peptide-MHC binding by a 'Query by Committee' artificial neural network approach

Generating quantitative models describing the sequence specificity of biological processes with the stabilized matrix method

Modeling the MHC class I pathway by combining predictions of proteasomal cleavage, TAP transport and MHC class I binding

AllerHunter: a SVM-pairwise system for assessment of allergenicity and allergic cross-reactivity in proteins

Combining pairwise sequence similarity and support vector machines for detecting remote protein evolutionary and structural relationships

PEP-FOLD: an updated de novo structure prediction server for both linear and disulfide bonded cyclic peptides

Protein structure prediction on the Web: a case study using the Phyre server

Identification and characterization of three novel HLA alleles, HLA-A*240214, HLA-A*3215 and HLA-DQB1*060302

Improving the physical realism and structural accuracy of protein models by a two-step atomic-level energy minimization

AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR

VERIFY3D: assessment of protein models with three-dimensional profiles

Verification of protein structures: patterns of nonbonded atomic interactions

Deviations from standard atomic volumes as a quality measure for protein crystal structures

QMEAN: A comprehensive scoring function for model quality assessment

AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading

UCSF Chimera--a visualization system for exploratory research and analysis

Rotamer libraries in the 21st century

CHARMM General Force Field (CGenFF): A force field for drug-like molecules compatible with the CHARMM all-atom additive biological force fields

Assignment of protonation states in proteins and ligands: combining pKa prediction with hydrogen bonding network optimization

Lipid14: the amber lipid force field

Fast empirical pKa prediction by Ewald summation

New ways to boost molecular dynamics simulations

Induction of hepatitis A virus-neutralizing antibody by a virus-specific synthetic peptide

A semi-empirical method for prediction of antigenic determinants on protein antigens

Prediction of chain flexibility in proteins

Improved method for predicting linear B-cell epitopes

Structural evidence for induced fit as a mechanism for antibody-antigen recognition

Empirical predictions of protein conformation

MEGA6: molecular evolutionary genetics analysis version 6.0

In silico analysis suggests interaction between Ebola virus and the extracellular matrix

Identification of novel potential vaccine candidates against tuberculosis based on reverse vaccinology

Vaccination and allergic disease: a birth cohort study

Computational Analysis and Binding Site Identification of Type III Secretion System ATPase from Pseudomonas aeruginosa

A method to identify protein sequences that fold into a known three-dimensional structure

Rational design, synthesis, and biological evaluation of 7-Azaindole derivatives as potent focused multi-targeted kinase inhibitors

Turns in peptides and proteins

Antigenic determinants in proteins coincide with surface regions accessible to large probes (antibody domains)

Efficacy and effectiveness of an rVSV-vectored vaccine expressing Ebola surface glycoprotein: interim results from the Guinea ring vaccination clusterrandomised trial

An adenovirus vaccine expressing Ebola virus variant makona glycoprotein is efficacious in Guinea Pigs and nonhuman primates

Potent neutralizing monoclonal antibodies against Ebola virus infection

Mechanism of binding to Ebola virus glycoprotein by the ZMapp, ZMAb, and MB-003 cocktail antibodies

Safety and immunogenicity of novel adenovirus type 26-and modified vaccinia ankaravectored Ebola vaccines: a randomized clinical trial

Computational prediction and identification of HLA-A2. 1-specific Ebola virus CTL epitopes

Conservancy of mAb epitopes in Ebolavirus glycoproteins of previous and 2014 outbreaks

Detection and molecular characterization of Ebola viruses causing disease in human and nonhuman primates

Clinical development of Ebola vaccines

In silico prediction of B-and T-cell epitope on Lassa virus proteins for peptide based subunit vaccine design

An in silico approach predicted potential therapeutics that can confer protection from maximum pathogenic Hantaviruses

Development of a DNA vaccine designed to induce cytotoxic T lymphocyte responses to multiple conserved epitopes in HIV-1

A combined immuno-informatics and structure-based modeling approach for prediction of T cell epitopes of secretory proteins of Mycobacterium tuberculosis

The Phyre2 web portal for protein modeling, prediction and analysis

The MM/PBSA and MM/GBSA methods to estimate ligand-binding affinities

Computations of Standard Binding Free Energies with Molecular Dynamics Simulations

Highly conserved regions in Ebola virus RNA dependent RNA polymerase may be act as a universal novel peptide vaccine target: a computational approach

Computational elucidation of potential antigenic CTL epitopes in Ebola virus

Epitope-based peptide vaccine design and target site depiction against Ebola viruses: an immunoinformatics study

A highly conserved GEQYQQLR epitope has been identified in the nucleoprotein of Ebola virus by using an in silico approach

Conformational flexibility in designing peptides for immunology: the molecular dynamics approach

Managing protein flexibility in docking and its applications

Protein flexibility: Multiple molecular dynamics simulations of insulin chain B

Toward an atomistic understanding of the immune synapse: Large-scale molecular dynamics simulation of a membrane-embedded TCR-pMHC-CD4 complex

MHC-peptide binding is assisted by bound water molecules