Prediction of B and T cell epitope-based peptide vaccines from highly conserved regions in Enterovirus D68 capsid protein VP1: a computational approach

Tahirah Yasmin Tahirah Yasmin*1¹

1. Introduction

Human enterovirus D68 (EV-D68) is a member of species enterovirus D (EV-D), which belongs to the genus Enterovirus and the family of Picornaviridae. Picornaviruses constitute a large family of small, icosahedral viruses with a single, positive-stranded RNA genome and an external diameter of about 300Å [1] and can cause a wide range of clinical symptoms, ranging from mild febrile illness to fatal meningitis and encephalitis [2]. Although EV-D68 was first isolated from samples obtained in California in 1962 from four children with pneumonia and bronchiolitis [3], it has been rarely reported since then.

However, an upsurge of EV-D68 cases in the past few years worldwide including Japan, the Philippines and the Netherlands [4] as well as several clusters in the USA has implicated EV-D68 as an emerging respiratory pathogen. In August 2014 an outbreak of mild to severe respiratory illnesses occurred among thousands of young children in the United States of which 1116 cases have been confirmed to be caused by EV-D68. This virus has also been associated with occasional neurological infections [5]. Although EV- D68 has emerged as a considerable global public health threat, there is no available vaccine or effective antiviral treatment [1]. Due to the ever rising spread of this viral infection, the development of vaccines or antiviral drugs against this pathogen is crucial and an immediate need.

The capsids of EVs consist of 60 copies of each of four different viral proteins, VP1, VP2, VP3 and VP4. The antigenic diversity among the enteroviruses is caused by variations within capsid proteins VP1 to VP3, but neutralization epitopes are most densely clustered on VP1 [6]. In addition, the hydrophobic pocket of VP1 contains a lipid moiety termed the “pocket factor,” which likely stabilizes the mature virion. As observed in poliovirus and certain picornaviruses, receptor binding at this junction site triggers the uncoating process, which is characterized by the delivery of the viral genome into the host cell compartment for replication and transcription [7].

Based on the consensus that VP1 is an important site for interaction with host cells [8], in this study VP1 capsid protein has been targeted in order to predict informative epitopes which can be helpful for future vaccine designing. The concept of peptide vaccines is based on identification and chemical synthesis of B- cell and T-cell epitopes which are immunodominant and can induce specific immune responses. Well characterized protective epitopes designed from this protein can be a great help for offering consistent, cost effective and quality therapeutics against this pathogen.

Conventional techniques for vaccine development are laborious and time consuming. As a result, computational methods [9] for predicting epitopes have attracted attention of the researchers to reduce the cost and time of vaccine development to fight with the rapidly growing devastating organisms. Current immunoinformatics tools are able to predict human B-cell and T-cell epitopes with high accuracy. These tools are playing a vital role in understanding the molecular basis of immunity and, notably in the development of epitope basedpeptide vaccines, immunotherapy against cancer and autoimmune diseases [10]. In comparison to the conventional vaccines, peptide or epitope based vaccines are also easy to develop, chemically stable, more 427 specific, and free of any infectious or oncogenic potential hazard [11].

Therefore this present study was undertaken with a view to applying immunoinformatics by using currently available online algorithms in order to identify potentially immunogenic T and B cell epitopes in VP1 capsid protein and design possible candidate peptide vaccines for EV-D68.

2. Materials and Methods

2.1 Retrieval of VP1 protein sequence and identification of conserved regions

The amino acid sequence of VP1 capsid protein of Enterovirus D68 from different strains was retrieved from NCBI database (www.ncbi.nlm.nih.gov/) in FASTA format.

To find the conserved region, retrieved sequences were aligned using ClustalW tool that uses Gonnet matrix in MEGA software (v 6.0) along with 1000 bootstrap value and default parameters. For the analysis of the evolutionary divergence in the capsid protein of EVD68, a phylogenetic tree was also constructed using MEGA software.

The minimum length of the conserved regions was set at 15 as binding predictions for MHC class II alleles use to create 15mers. The conserved regions were then used for B and T cell epitope prediction.

2.2 Prediction of putative B cell epitopes, their antigenicity and transmembrane properties

Prediction of potentially immunogenic epitopes in a given protein sequence may significantly reduce wet lab effort needed to discover the epitopes required for the design of vaccines. The aim of the prediction of the B cell epitope was to find the potential antigen that would interact with B lymphocytes and initiate an immune response [12].

For this, antigenicity characteristics of each conserved region were screened using an online antigen prediction server, VaxiJen v2.0 (http://www.ddgpharmfac.net/vaxijen/). The cut off value for being a probable antigen was set at 0.5. On the other hand, each sequence was also subjected to TMHMM v0.2 server [13] in order to find out whether they contained any transmembrane topology. In order to predict linear B cell epitopes two different online tools were employedBepipred and ABCpred server (http://www.imtech.res.in/raghava/abcp red/). BepiPred predicts the location of linear B-cell epitopes using a combination of a hidden Markov model and a propensity scale method [14]. Default threshold value 0.35 was used for analysis. On the other hand, ABCpred server uses artificial neural network [15] and here cutoff value was set at 0.51.

2.2.1 Prediction of surface accessibility,

hydrophilicity, flexibility and beta-turn of the predicted epitopes

A B-cell epitope is characterized by being antigenic, hydrophilic, and accessible in a flexible region of an immunogen [16]. Therefore to find out whether the predicted epitopes possessed these desired characteristics, several tools all hosted by IEDB-AR were utilized. The tools includeEmini surface accessibility prediction [17], Parker hydrophilicity scale [18], Karplus and Schulz flexibility scale [19] and Chou and Fasman beta-turn prediction tool [20] all with default parameters. The results from all these sites were cross-referenced and apparently common findings were taken as the most probable B-cell 428

2.2.2 Structure analysis

The 3D structure of VP1 capsid protein was constructed by homology modeling using the SWISS-MODEL (http:// swissmodel.expasy.org) work space. Being a server for automated comparative modeling of three-dimensional (3D) protein structures SWISS-MODEL provides several levels of user interaction where template selection, alignment and model building are done completely automated by the server [21]. Finally, the predicted model was evaluated by using two software tools, PROCHECK [22] and QMEAN [23].

2.3 Prediction of putative T cell epitopes

Firstly, the NetCTL v1.2 server was used for predicting potential cytotoxic T lymphocyte (CTL) epitopes from the conserved peptides which integrates prediction of peptide MHC class I binding, proteasomal C terminal cleavage and TAP transport efficiency [24]. MHC-I binding and proteasomal cleavage were carried out through artificial neural networks and the weight matrix was used to estimate the TAP transport efficiency. The threshold value for epitope identification was set at 0.75 for maintaining sensitivity and specificity of 0.80 and 0.97, respectively during the analysis. The epitope prediction was confined to 12 MHC-I super types.

For cross checking, a similar tool from IEDB analysis resources which combines MHC class I binding, transporter of antigenic peptides (TAP) transport efficiency, and proteosomal cleavage prediction, was employed to predict a total score (http://tools.immuneepitope. org/processing/) for each conserved peptide's intrinsic potential of being a T cell epitope [25]. Here, to calculate the half-maximal inhibitory concentration (IC values Stabilized Matrix Method 50) (SMM) was applied. All the available MHC alleles were selected and the peptide lengths were set at 9.0. The top scoring peptides were selected by setting cut-off values of IC⁵⁰ within 100 nM and the overlapping peptides found in both NetCTL and IEDB analysis were subjected to further analysis.

2.3.1 MHC-I and MHC-II restriction analysis

The candidate peptides were analyzed by IEDB MHC class I binding prediction tool (http://tools.immuneepitope.org/mhci/) for all the available class I alleles. The conserved region of each protein was also tested for predicting epitopes that interact with MHC class II molecules by selecting all the alleles in IEDB MHC class II binding prediction tool (http://tools.immuneepito pe.org/mhcii/). In both cases, SMM-align method was employed to find out good binders and the cut-off value of IC was 50 set at 100 nm.

2.3.2 Analysis of population coverage

Population coverage for the selected epitopes was assessed by the IEDB population coverage calculation tool [26]. Here the allelic frequency of the interacting HLA alleles was used for the prediction of the population coverage of the corresponding epitopes.

2.4 Allergenicity and disorder analysis of the conserved regions

The web-based Allerdictor server was used to predict the allergenicity of the predicted epitopes for vaccine development. It is a fast and accurate sequence-based allergen prediction tool that models protein sequences as text 429

documents and employs support vector machine in text classification for allergen prediction [27]. Furthermore, for predicting the disorder among the amino acid sequences, DISOPRED [28] server was used.

3. Results

retrieved from NCBI GenBank sequence database (www.ncbi.nlm.nih.gov/genbank /) in FASTA format. To find the conserved regions, multiple sequence alignment (MSA) by CLUSTALW analysis in MEGA software was performed. CLUSTALW program generated several conserved sequences with varying lengths. The conserved sequences (length more than 430

3.2 Prediction of putative B cell epitopes from the conserved regions

On the basis of VaxiJen score (VaxiJen score ≥ 0.5) and transmembrane topology, among the conserved sequences, two regions-“GLYGINPADTIGNLCVRIVNEHQP” and “GLPDLTLQAMFVPTGALTP” were considered to be the most potential antigenic peptides which fulfilled the criteria of exomembrane characteristics as well (Table 1).

These two regions were then subjected to Bepipred [14] and ABCPred [15] analysis and subsequently “INPADT” epitope from “GLYGINPADTIGNLCVRIVNEHQP” and “GALTP” from “GLPDLTLQAMFVPTGAL TP” were found to be potential B cell To further strengthen the prediction, several additional tools from IEDB resource portal were employed. The surface accessibility of the peptides was determined by Emini surface accessibility prediction tool [17] using 1.000 threshold levels and one epitope- “INPADT” was found to be above the threshold level

the yellow region. The red horizontal line indicates surface accessibility cutoff

(1.000).

(Table 2; Figure 2). The same epitope was also found to cross the threshold to be hydrophilic, flexible and have beta turn region determined by Parker hydrophilicity scale [18], Karplus and Schulz flexibility scale [19] and Chou and Fasman beta-turn prediction tool [20], respectively (Table 2; Figure 3, 4 and 5). However, “GALTP” epitope did not fulfill all the criteria.

The three dimensional structure of the capsid protein was obtained by SWISS- MODEL workspace [21] through automated homology modeling which is shown in Figure 6A and PROCHECK server validated the stereo chemical quality of the model through Ramachandran Plot (Figure 6B) [22], and QMEAN server [23] also assessed the tertiary structure, with a Qmean4 score of -2.54 ( Figure 6C). The B cell epitope is marked by red color and it can be seen that the epitope is on the surface of the protein’s 3D structure hence proving its

the green region.

431

below the cut off (red line) are in the green region.

are below the cut off (red line) are in the green region.

3.3. Prediction of T Cell Epitopes

T-cell epitopes were primarily selected by using the NetCTLv1.2 [24] server where the epitope prediction was confined to 12 MHC-I super types. The conserved sequences were also analyzed by a combined algorithm in IEDB analysis integrating MHC class I binding, TAP transport efficiency and proteasomal cleavage prediction (binding affinity kept within 100 nM) [25]. In total, 8 epitopes were found common in both analyses which are listed in Table 3.

These processed peptides were then analyzed by SMM based IEDB MHC I prediction tool and here again the cut off IC value was set at 100nM. Thus, among 50 these peptides, only those peptides which interacted with minimum four MHC class I alleles as well as were found to be overlapping with 15mers that interacted with a good number of MHC class II alleles, were selected finally (Table 4).

It was found “YMSIANANY”, “LVSKRSFEY”, “ENFLSRAAL” and “AMFVPTGAL”- these four epitopes were most likely to induce T cell immune response as they interacted with highest number of MHC class I and class II alleles.

3.4 Analysis of population coverage

INQHGVSETLVENFLSRAALVSKRSFEYK GLPDLTLQAMFVPTGALTP IEDB population coverage tool [26] analyzed the population coverage of the proposed epitopes. The combined MHC-I and MHC-II class were assessed against the whole world population with the selected MHC-I and MHC-II alleles (Table 4). The epitope with the highest population coverage was found to be “LVSKRSFEY”.

3.5 Conservancy, allergenicity and disorder analysis

Since all the epitopes were predicted from conserved regions of VP1 protein they were 100% conserved among all the isolates found in various parts of the world.

In addition, it was found from Allerdictor server [27] all of the most suitable T cell and B cell epitopes were potentially nonallergen. Also, DISOPRED server [28] predicted the disorder of the conserved peptide “INQHGVSETLVENFLSRAALVSKRSFEYK” in order to get insight about the disorder among the sequence, which is depicted in Figure 7. It was found that the two most probable T cell epitopes- “LVSKRSFEY” and “ENFLSRAAL” lie outside of the disordered region to secure their potentiality as effective epitopes.

LVSKRSFEY 80 A1,A3, B58,B62 NQHGVSETL 62 B39,B44, B62 SETLVENFL 67 B44 ENFLSRAAL 72 B39 AMFVPTGAL 157 A2, B62 433

scheme would be most desirable for the 434

The outside location of the epitope indicates its surface accessibility. B) Ramachandran plot

of the predicted model shows that most of the residues (83%) are in the allowed region of the plot, proving the validity of the model. C) QMEAN4 plot of the predicted model

435 With time, due to antigenic drift, any foreign particle can escape the antibody memory response; however, the T cell immune response often provides longlasting immunity [31]. Therefore, in the present study both B cell and T cell epitopes were designed using in silico computational approaches.

Inhibition of viruses at the stage of viral attachment provides a target for therapeutic intervention [32]. In order to predict a peptide that can be used in vaccine development to prevent viral entry or its interaction with host cell, this study focused on VP1 capsid protein. This protein is required by enterovirus for mediating viral adsorption and uncoating process [33] and subsequently becomes a good target for vaccine designing.

Here, multiple prediction methods were applied to determine a potential B-cell epitope considering several criteria like hydrophilicity, surface accessibility, flexibility and beta-turn. The proposed epitope “INPADT” has met all the criteria of the above B-cell prediction methods. Furthermore, the three-dimensional model of the capsid protein ensured the exact location of the epitope outside of the protein surface and the model validity was evaluated by Ramachandran Plot, where it was seen 83 % amino acid residues were within the favored region.

Along with the B-cell epitope, in this study, T-cell epitopes were also designed. Due to lack of a consensus mapping protocol with immunoinformatics tools, a combined prediction method was applied according to the hypothesis of Trost et al. [34] who proposed that greater prediction accuracy can be achieved by combining the predictions from several algorithms rather than relying on just one. Both NetCTL and IEDB analysis tools were used to find epitopes for the activation of T-cell immunity with potential antigenicity. By examining the output it was predicted that “YMSIANANY”, “LVSKRSFEY”, “ENFLSRAAL” and “AMFVPTGAL”- these four epitopes would be the best epitope candidates. They interacted with the highest numbers of MHC class I alleles, were the core peptides of a good number of MHC class II binding predictions and also demonstrated large world population coverage. Again, conservancy is another very important criterion of an epitope to consider it for vaccine development and all the proposed epitopes were 100% conserved since they were predicted from conserved region analysis.

However, allergenicity is one of the prominent obstacles in vaccine development since today most vaccines stimulate the immune system into an “allergic” reaction, [35] through induction of type-2 T helper (Th2) cells and immunoglobulin E (IgE). In silico analysis revealed that all the vaccine candidates predicted in this work were nonallergenic in nature.

Thus, it is believed that these suggested B and T cell epitopes will definitely reduce time, cost and labor during in vivo and in vitro studies to be carried out for developing a vaccine against Enterovirus D-68.

Conclusion

In this study, an attempt was made to design epitope based vaccines against Enterovirus D68 which could be tested for their efficacy in eliciting immunity through humoral and cell mediated immune responses. The results are based on sequence analysis and computational predictions. However, to prove the effectiveness of mounting an immune response, both in vitro and in vivo studies are required along with this in silico study.

Acknowledgement

Author deeply expresses her sincere thanks and gratitude to Professor AHM 436 Nurun Nabi of the Department of Biochemistry and Molecular Biology at the University of Dhaka for his continuous encouragement.

References

References

1. Yue Liu, Ju Sheng, Andrei Fokine, Geng Meng, Woong-Hee Shin, Feng Long, Richard J. Kuhn, Daisuke Kihara, and Michael G. Rossmann. Structure and inhibition of EV-D68, a virus that causes respiratory illness in children. Science. 2015 Jan 2; 347(6217): 71–74.
2. Rafal Tokarz, Cadhla Firth, Shabir A. Madhi, Stephen R. C. Howie, Winfred Wu, Amadou Alpha Sall, Saddef Haq, Thomas Briese, and W. Ian Lipkin. Worldwide emergence of multiple clades of enterovirus 68. J Gen Virol. 2012 Sep; 93(Pt 9): 1952–1958.
3. Schieble J. H., Fox V. L., Lennette E. H. A probable new human picornavirus associated with respiratory diseases. Am J Epidemiol (1967) 85, 297–310
4. Tokarz R, Firth C, Madhi SA, Howie SR, Wu W, Sall AA, Haq S, Briese T, Lipkin WI Worldwide emergence of multiple clades of enterovirus 68. J Gen Virol. 2012;93: 1952–1958.
5. Kreuter JD, et al. A fatal central nervous system enterovirus 68 infection. Arch Pathol Lab Med.2011;135: 793–796.
6. Rueckert RR. In: Virology. Fields BN, Knipe DM, editor. Lippincott-Raven, New York; Picornaviruses and their replication; 1990. pp. 507–548.
7. Helenius A. 2007. Virus entry and uncoating, p P100–P118 In Knipe DM, Howley PM, Griffin DE, Lamb RA, Martin MA, Roizman B, Straus SE, editors. (ed), Fields virology, 5th ed, vol 1 Lippincott Williams & Wilkins, Philadelphia, PA
8. Chen H. F., Chang M. H., Chiang B. L., Jeng S. T. 2006. Oral immunization of mice using transgenic tomato fruit expressing VP1 protein from enterovirus 71. Vaccine 24: 2944–2951
9. E. M. Lafuente and P. A. Reche, “Prediction of MHC-peptide binding: a systematic and comprehensive overview,” Current Pharmaceutical Design, vol. 15, no. 28, pp. 3209–3220, 2009.
10. Hasan MA, Hossain M, and Alam MJ: A Computational Assay to Design an epitope- Based Peptide Vaccine Against Saint Louis Encephalitis Virus. Bioinform Biol Insights 2013, 7: 347–355.
11. Holland J, Domingo E .Origin and evolution of viruses. Virus Genes (1998) 16(1):13– 21
12. Nair DT, Singh K, Siddiqui Z, Nayak BP, Rao KV, Salunke DM. Epitope recognition by diverse antibodies suggests conformational convergence in an antibody response. J Immunol. 2002; 168(5):2371–2382
13. Krogh A, Larsson B, Heijne GV and Sonnhammer ELL: Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. Journal of Molecular Biology2001, vol. 305, no. 3, pp. 567–580.
14. Jens Erik Pontoppidan Larsen, Ole Lund and Morten Nielsen. Improved method for predicting linear B-cell epitopes. Immunome Res. 2006; 2: 2.
15. Saha, S and Raghava G.P.S. Prediction of Continuous B-cell Epitopes in an Antigen Using Recurrent Neural Network. Proteins, (2006) 65(1),40-48
16. Fieser TM, John A, Tainer H, et al. Influence of protein flexibility and peptide conformation on reactivity of monoclonal anti-peptide antibodies with a protein ahelix. Proc Natl Acad Sci. 1987; 84(23):8568–72.
17. Emini EA, Hughes JV, Perlow DS, Boger J: Induction of hepatitis A virus-neutralizing antibody by a virus-specific synthetic peptide. J Virol 1985, 55(3):836-839.
18. Parker JM, Guo D, Hodges RS: New hydrophilicity scale derived from high performance liquid chromatography peptide retention data: correlation of predicted surface residues with antigenicity and X-ray-derived accessible sites. Biochemistry 1986, 23:25.
19. Karplus PA, Schulz GE: Prediction of Chain Flexibility in Proteins - A tool for the 437 Selection of Peptide Antigens. Naturwissenschafren 1985, 72:212-3.
20. Chou PY, Fasman GD. Prediction of the secondary structure of proteins from their amino acid sequence. Adv Enzymol Relat Areas Mol Biol. 1978; 47:45–148.
21. Torsten Schwede, Jürgen Kopp, Nicolas Guex, and Manuel C. Peitsch. SWISS- MODEL: an automated protein homologymodeling server. Nucleic Acids Res. 2003 Jul 1; 31(13): 3381–3385.
22. Laskowski R A, MacArthur M W, Moss D S, Thornton J M (1993). PROCHECK - a program to check the stereochemical quality of protein structures. J. App. Cryst., 26, 283-291.
23. Benkert P, Biasini M, Schwede T .Toward the estimation of the absolute quality of individual protein structure models. Bioinformatics(2011) 27(3):343–350
24. Larsen MV, Lundegaard C, Lamberth K, Buus S, Lund O, Nielsen M. Large-scale validation of methods for cytotoxic T- lymphocyte epitope prediction. BMC Bioinformatics. Oct 31; 8: 424. 2007
25. Tenzer S, Peters B, Bulik S, Schoor O, Lemmel C, Schatz MM, Kloetzel PM, Rammensee HG, Schild H, and Holzhutter HG. Modeling the MHC class I pathway by combining predictions of proteasomal cleavage, TAP transport and MHC class I binding. Cell Mol Life Sci. 2005 62:1025- 1037.
26. Bui HH, Sidney J, Dinh K et al .Predicting population coverage of T-cell epitopebased diagnostics and vaccines. BMC Bioinformatics (2006) 17(7):153
27. Dang, H. X., & Lawrence, C. B.. Allerdictor: Fast allergen prediction using text classification techniques. Bioinformatics, (2014) btu004.
28. Ward JJ, McGuffin LJ, Bryson K et al. The DISOPRED server for the prediction of protein disorder. Bioinformatics 20 (2004) (13):2138–2139
29. Tadatsugu Imamura and Hitoshi Oshitani. Global reemergence of enterovirus D68 as an important pathogen for acute respiratory infections. Rev Med Virol. 2015 Mar; 25(2): 102–114.
30. Shrestha B, Diamond MS. Role of CD8+ T cells in control of West Nile virus infection. J Virol 2004, 78(15):8312– 8321.
31. Oany AR, Emran AA and Jyoti TP: Design of an epitope-based peptide vaccine against spike protein of human coronavirus: an in silico approach. Drug Des Devel Ther 2014, 8: 1139–1149.
32. Chee Wah Tan, Yoke Fun Chan, Kooi Mow Sim, Eng Lee Tan, and Chit Laa Poh. Inhibition of Enterovirus 71 (EV-71) Infections by a Novel Antiviral Peptide Derived from EV-71 Capsid Protein VP1. PLoS One. 2012; 7(5): e34589.
33. Shih SR, Tsai MC, Tseng SN, Won KF, Shia KS, et al. Mutation in Enterovirus 71 capsid protein VP1 confers resistance to the inhibitory effects of pyridyl imidazolidinone. Antimicrob Agents Chemother. 2004; 48: 3523–3529.
34. Trost B, Bickis M, Kusalik A: Strength in numbers: achieving greater accuracy in MHC- I binding prediction by combining the results from multiple prediction tools. Immunome Res 2007, 3: 5–15.
35. McKeever TM, Lewis SA, Smith C, Hubbard R. Vaccination and allergic disease: a birth cohort study. Am J Public Health. 2004; 94(6):985–989. 438