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

BACKGROUND: Lassa fever is a severe, often-fatal and one of the most virulent disease in primates. However, the mechanism of escape of virus from the T-cell mediated immune response of the host cell is not explained in any studies yet. In our studies we had aimed to predict B- and T- cell epitope of Lassa virus protein, for impaling the futuristic approach of developing preventive measures against this disease, further we can also study its presumed viral- host mechanism.
MATERIALS AND METHODS: Peptide based subunit vaccine was developed from all four protein against Lassa virus. We adopted sequence, 3D structure and fold level in silico analysis to predict B-cell and T-cell epitopes. The 3-D structure was determined for all protein by homology modeling and the modeled structure validated.
RESULTS: One T-cell epitope from Glycoprotein (WDCIMTSYQ) and one from Nucleoprotein (WPYIASRTS) binds to maximum no of MHC class I and MHC class II alleles. They also specially bind to HLA alleles namely, A*0201, A*2705, DRB*0101 and DRB*0401.
CONCLUSIONS: Taken together, the results indicate the Glycoprotein and nucleoprotein are most suitable vaccine candidates against Lassa virus.

INTRODUCTION

Lassa fever is an acute viral zoonotic illness caused by Lassa virus, a member of the Arenaviridae family and responsible for a severe hemorrhagic fever characterized by fever, sour throat, muscle pain, nausea. Lassa fever was first described in Sierra Leone in 1950s but the virus responsible for the disease was not identified until 1969 when two missionary nurses died in Nigeria, West Africa, and the cause of their illness was found to be Lassa virus, named after the town in Nigeria (Lassa in the Yedseram River valley) where the first cases were isolated.[12] There are an estimated 300,000 to 500,000 cases of Lassa fever each year[34] with a mortality rate if 15–20% for hospitalized patient. Mortality rate has become high as 50% during epidemic and 90% in third month pregnancy for the expectant mother and the fetus both.[32] Since then, a number of outbreaks of Lassa virus infection were reported in various parts of Nigeria including Jos, Zonkwua, Onitsha, Owerri, Abo Mbaise, Lafiya and Epkoma.[45678] Epidemics of Lassa fever were also documented in other West African countries including Liberia, Sierra Leone, Guinea, Mali and Senegal.[179] A few cases of the importation of Lassa virus into other parts of the world for example by travelers were documented.

Morphologically, Lassa virus consists of enveloped particles that vary in diameter from approximately 60 to more than 300 nm, with mean particle size of 92 nm. The virus is approximately spherical, enveloped particles that rang in diameter from 50 to 300 nm. Lassa virus having ambisense genomic organization (two viral genes separated by an intergenic region), negative sense, bisegmented, ssRNA genome S (small, ~3.4 kb) and L (large, ~7.2 kb) segments.[1011] The small segment encodes the 75 KDa glycoprotein precursors (GPC) and the 63 KDa nucleoprotein (NP). After part translation modification GPC is cleaved into GP1 and GP2, the large segment of RNA encodes the 11KDa Z protein, which binds zinc and as Matrix protein and the 200KDa L protein, which acts as matrix protein. Replication and transcription of the genome occur in the cytoplasm of an infected cell and both take place within Rib nucleoprotein complex.[1112]

Lassa virus is transmitted to human being from the rodent reservoir Mastomys natalensis, by direct content with infected tissue, food contaminated with excreta. Mostomys natalensis, is a common rodent in village houses, is therefore primary human infection and common (r) Lassa fever may also spread through person to person contact. Lassa virus trasmition occurs when a healthy person comes in contact with virus in the blood, secretion, tissue or excretion of any infected individual. The virus cannot be transmitting through skin to skin contact without exchange of body fluid. Lassa virus enters the cell via the alpha-dystroglucan receptor.[13] Furthermore, ribavirin should be made available in hospitals and health centers in the endemic areas particularly in rural communities. This would help to control the disease. The aim of this study is to design the peptide based vaccines for Lassa virus.

MATERIALS AND METHODS

Proteome data retrieval and analysis

Z protein (NP_694871), Nucleoprotein (NP_694869), L protein (NP_694872), and Glycoprotein (NP_694870) are the protein of Lassa Virus available in the NCBI Protein data base and hence it is used for this analysis. Bioinformatics tools were used for the analysis of proteome of Lassa virus. The expected molecular weight, isoelectric point (pI) highly repeated amino acids (%) of repetition values were calculated using ExPaSy (http://www.expasy.org/).[1415] All four proteins were analyzed for antigenicity using VaxiJen web server [Figure 1].

Figure 1

Insilco methodology for B and T-cell epitope prediction

B-cell and T-cell epitope prediction

All of these targeted proteins of the Lassa virus strain found namely Z protein Nucleoprotein L protein and Glycoprotein were analyzed for the B-cell epitopes using BCPred.[16] Identify common B-cell epitope by using both BCPred and AAP prediction method of BCPred.[17] B-cell epitope with >0.8 BCPred and >0.4 Vaxijen score were re-elected for the identification of T cell epitope. These selected epitope further subjected to ProPred1 and ProPred analysis.[18192021] Both (Propred1 and Propred) are matrix-based method that allows prediction of MHC binders for various alleles based on the multiplication and additional matrices, proteosome cleavage site, simultaneously. ProPred1 allow predicting MHC class I binding peptide (CTL epitope) for 47 Allele and ProPred to predict MHC class II binding peptide (HTL peptide) for 51 alleles. Common epitope that bind both the MHC classes were selected for further analysis.

These epitopes were analyzed with VaxiJen v2.0 web server. The IC50 (inhibitory concentration 50) value of corresponding epitope was deduced from MHCpred server.[22] Epitope having less than 1000 nM IC50 values for DRB1*0101 of MHC class II were selected. T epitope designer and MHCpred are the second screening method. T-epitope Designer is a structure based QSAR simulation method to predict HLA-peptide binding based on virtual binding pocked using X-ray crystal structure of HLA-peptide complex.[23] In the second screening, the following selection criteria were used: i) Binding with large number of HLA alleles (>1000), (ii) must bind to A*0201, A*0204 and A*2705. Predicted epitope well further analyzed for fold level topology.

Molecular modeling and fold level analysis

The 3D structure of Glycoprotein was not available in database PDB. Query glycoprotein and template 3MKO after alignment were used as input in Modeller program, giving five models for target. Modeller objective function and dope score helped in the validation of the model of glycoprotein. The validated glycoprotein models were chosen for further studies and refinement. The most acceptable model of NDM-1 was finalized by Ramachandram Plot, providing the position of residue in particular segment based on phi (Ø) and psi (ψ) angles between N-Cα and Cα-C atoms of residue. The theoretical model generated was validated by using the programs PROCHECK[2425] and ProSA.[26] PROCHECK is a suite of programs to check the stereo chemical quality of protein structure. It includes parameters such as bond length, bond angle, main chain and side chain properties and residue-by residue properties to assess whether the geometry of the residues in a given protein structure is normal or unusual.[25] ProSA (Protein Structure Analysis) program exploits interactive web- based applications to check the three-dimensional models of protein structures for potential errors by displaying scores and energy plots.

Pepitope servers were used to analyze the folding and cluster of selected epitopes in folded protein PepSurf and Mapitope algorithms used by this server to predict fold level topology. PepSurf algorithm helps to map the epitope onto the surface of the antigen.[27] By Mapitope algorithms epitopes shared by the entire set of peptides are detected by the following steps user carried out to predict fold level topology. i) Prediction algorithm was executed and ii) 3D structure of predicted epitope are visualized using 3D structure viewer.

Characterization of epitopes

DISTELL server was used to design the 3D structures of the predicted binding peptides based on the similarity with PDB temples.[28] After designing the structures epitopes were then validated with PROCHECK and ProSA. For prediction of domain, motif and functionality of epitope ProFunc, Motif Scan and InterProScan are used. ProteinDigest web server was used to predict pI, molecular weight and enzyme degradation site.[29]

Molecular docking studies

After designing the epitopes structures molecular docking of selected alleles and epitopes was performed with the help of Autodock.[30] This program used a simulated annealing approach to explore the conformation space between the ligand and target protein. The energy evaluation process is done by using grid-based molecular affinity potential, the docking was performed on the basis of Lamarckian Genetic Algorithm. The PMV (Python Molecular Viewer) was used for the visualization of Binding, position, H-bonding between the selected peptides and alleles.

RESULTS

Physicochemical analysis and antigenicity prediction

The L protein has the highest molecular weight of about 253431.9 KDa which consists of Leucine (L) a neutral nonpolar amino acid residue has the highest percentage of repetition (11.9%). The physicochemical properties of putative proteins are given in Table 1. The pI value of any protein indicates the stability of protein in that particular isoelectric point. Isoelectric points of these proteins were ranged 6.17 to 8.95.

Table 1

It comprises the data of Lassa virus proteins, molecular weight, pI and percentage of highly repeated amino acid residues in individual protein with antigenisity score

Prediction of B and T-cell epitope

Epitopes which are capable to induce both type immunity (B-cell and T-cell) are known to be good vaccine candidate. For identification of these epitopes amino acid sequence of all four proteins were subjected to BCPred for B- cell epitope prediction. B-cell epitope prediction is the initial step for vaccine construct. BCPred is web based method that uses a novel method of subsequence kernel which was used to predict linear B-cell epitope from each protein. BCPed and AAP (amino acid pair) methods are used to predict fixed length epitope [Table 2]. Finally, two out of three B-cell epitope from Z protein, seven out of ten epitope from Glycoprotein, seven out of fourteen epitope from Nucleoprotein and sixteen out of twenty five epitope from L protein were selected for further analysis. These selected epitopes were analyzed for T-cell epitope identification. On the first level, sequence based 2D screening propred 1, propred and MHCPred were used to identify the potential T-cell epitope. Default ProPred1 and Propred parameters were used to determine the T-cell epitope [Table 3].

Table 2

B-cell epitope prediction using BCPred (BCPred+AAP) and anigenicity of peptide using Vaxijen

Table 3

Common epitope (B and T-cell) along with their various parameters (selected epitope are highlighted in bold)

Common epitopes that bind with both MHC class I and II, and interact with DRB1*0101 are shown in Table 4. Epitopes with best Vaxijen and MHCPred score were selected for further analysis. T-epitope Designer was used to predict identified epitope binding abilities to >1000 MHC allele. T-epitope Designer is implemented based on a MHC-peptide prediction model described recently.

Table 4

3D QSAR based T-cell epitope prediction using T epitope designer

Modeling and fold level analysis

Modeller 9.10 generated the 3-D model of query glycoprotein protein at resolution of 1.80 Å based on template protein 3MKO with sequence query coverage 26%. PROCHECK analysis of the model structure from Modeller 9.10 through Ramachandran plot showed that 96.1% of the residues were in favored and additional allowed regions [Figure 2]. Qualitative assessment of the model through ProSA analysis revealed that the model matched with the NMR region of the plot with Z score of -8.02. GROMOS96 which was used for energy minimization optimized the model structure from initial energy -3568.30 KJ/mol to final energy of -9434.68 KJ/mol.

Figure 2

(a) 3D model of Glycoprotein, (b) Ramachandram plot for Glycoprotein model

Positional fold level topologies of predicted epitope on theoretical models of proteins were predicted using Pepitope web server. Figure shows that the epitopes were present within the cluster on the surface of the protein. The glycoprotein is antigenic and one epitope WDCIMTSYQ from cluster I (score: 8.1608, Residue No. 8) was found to be antigenic (Vaxijen score: 1.1117) and can bind 64 MHC molecule of both MHC I and II molecule. The IC50 values of this epitope for DRB1*0101 and DRB1*0401 were 70.96 and 186.64, respectively, which indicates a good inhibition this epitope has also been found to bind all selected MHC molecules (A*0201, A*0204, B*2705) and 96.57% HLA molecule of T epitope designer shown in Table 5. The Nucleoprotein is also antigenic and one epitope WPYIASRTS from cluster I (Score: 10.301 Residue No. 8) was found to be antigenic (VaxiJen score: 1.3451) and can bind 64 MHC molecule of both MHC class I and II. The IC50 values of this epitope for DRB1*0101 and DRB1*0401 were 293.76 and 146.89, respectively, which indicates a good inhibition this epitope has also been found to bind all selected MHC molecules (A*0201, A*0204, B*2705) and 96.57% HLA molecule of T epitope designer [Figure 3].

Table 5

The selected epitopes showing MHC binding and inhibition values predicted from 3D QSAR based T-epitope designer and MHCPred server

Figure 3

Fold level topology of epitope analyzed by Pepitope server (a) Glycopeotein epitope (WDCIMTSYQ) shown in red colour, (b) Nucleoprotein epitope (WPYIASRTS) shown in red color

Epitope characterization

After designing the 3D structures both the tools (PROCHECK and ProSA-web) are used for model validation. Furthermore no motif or domain could be assigned using ProFunc, Motif Scan and InterProScan for both proteins. Calculated M. wt. and pI of the WDCIMTSYQ from glycoprotein were 1146.30 Da and 3.08 respectively. Protein was found to be un digested with Tripsin, Clostripain, Proline Endopept, Staph Protease, Trypsin K, Trypsin R, as analyzed by ProteinDigest server for epitope WPYIASRTS from nucleoprotein, M. wt. and pI were 1080.21 and 8.75 respectively, and found to be undigested by Trypsin, Cyanogen_Bromide, Staph_Protease, Trypsin_K, AspN.

Molecular docking analysis

Some of the conserved peptides (WDCIMTSYQ, WPYIASRTS) with their interaction energies in kcal mol-1 are given in Table 6 and shown in Figures 4 and 5. These peptides are promiscuous HLA binders. It will be useful to include these peptides in a chimeric constructs containing both cytotoxic and helper epitopes. It is expected that though this T-cell vaccine would not prevent Lassa virus infection, it would aid in quick clearance of the virus and prevent the severe infection.

Table 6

The Docking results of interactions obtained with docking energy, build hydrogen bond and active site residues

Figure 4

Two best docking conformation ((a). WDCIMTSYQ with, (b). WDCIMTSYQ with 1KLG) analyzed by Python Molecular Viewer (docked ligand shown by balls and sticks while hydrogen bonds shown by black sticks)

Figure 5

Two best docking conformation ((a). WPYIASRTS with 1J8H, (b). WPYIASRTS with 1AOS) analyzed by Python Molecular Viewer (docked ligand shown by balls and sticks while hydrogen bonds shown by black sticks)

DISCUSSION

In the present study, three putative proteins of Lassa virus were used for the physicochemical analysis such as molecular weight, isoelectric point (pI value) and antigenic nature. Vaxijen was used to predict antigenisity, which is based on auto cross covariance (ACC) transformation of protein sequence into uniform vectors of principal aminoacid properties. The l00-CV (leave one – out cross validation) was used to identify antigenicity of protein with 91% sensitive, 82% accuracy and 72 specificity for viral sheces. The resultant protein was antigenic.

Epitopes which are capable to induce both type immunity (B-cell and T-cell) are known to be good vaccine candidate. Propred1 and ProPred are matrix-based methods that allow prediction of MHC binders for various alleles based on the multiplication and additional matrices, proteosome cleavage site, simultaneously. This is based on the observations made in previous studies which demonstrate that MHC binders having proteosome cleavage site at their C terminus have high potency to become T-cell epitopes.

MHCPred runs as a CGI server, and uses partial least square base approach for the identification of binding affinity to MHC molecule. MHCPred server generated IC50 nM values as output. A lover value of IC50 shows higher affinity with MHC molecule. The epitope bind with maximum number of alleles was selected for molecular docking analysis. The Pepitope server is a web based tool to predict discontinuous epitope based on set of peptide that have affinity against a monoclonal antibody or peptide. The server aligns a linear peptide sequence on to a 3D protein structure. The DISTILL model server was used to design the 3D structures of the predicted binding peptides. It is important to identify those peptides which are conserved across the various strains of Lassa virus and in this study that has been shown for the conserved peptides present in the constituent proteins of Lassa virus. The analysis reveals that there are number of suitable peptides from all four which may be included in the construction of poly epitopes T-cell vaccine.

CONCLUSION

The screening of putative epitopes using bioinformatics tools thus suggests that Glycoprotein and Nucleoprotein protein of Lassa virus could be used for preparation of immunological constructs. Molecular simulation and binding tests also suggest that the two nonameric epitopes WDCIMTSYQ and WPYIASRTS predicted and reported for the first time have considerable binding with MHC molecules and low energy minimization values providing stability to the peptide-MHC complex. These peptide construct will further undergo for wet lab studies, for the development of targeted vaccine against Lassa virus strains. Using a similar approach the short listing of candidate epitopes for vaccine design using other proteins can also be targeted that would reduce time and experimental expense.