Predicting promiscuous antigenic T cell epitopes of Mycobacterium tuberculosis mymA operon proteins binding to MHC Class I and Class II molecules

Infection, Genetics and Evolution 44 (2016) 182–189

Contents lists available at ScienceDirect

Infection, Genetics and Evolution journal homepage: www.elsevier.com/locate/meegid

Research paper

Predicting promiscuous antigenic T cell epitopes of Mycobacterium tuberculosis mymA operon proteins binding to MHC Class I and Class II molecules Iti Saraav a, Kirti Pandey a, Monika Sharma a, Swati Singh a, Prasun Dutta b, Anshu Bhardwaj b, Sadhna Sharma a,⁎ a b

D S Kothari Centre for Research and Innovation in Science Education, Miranda House, and Department of Zoology, Miranda House, University of Delhi, Delhi 110007, India Open Source Drug Discovery Unit, Council of Scientific and Industrial Research (CSIR), Delhi 110001, India

a r t i c l e

i n f o

Article history: Received 13 February 2016 Received in revised form 24 June 2016 Accepted 3 July 2016 Available online 5 July 2016 Keywords: Antigenic T-cell epitopes Mycobacterium tuberculosis mymA operon Tuberculosis Vaccine candidates

a b s t r a c t Limited efficacy of Bacillus Calmette–Guérin vaccine has raised the need to explore other immunogenic candidates to develop an effective vaccine against Mycobacterium tuberculosis (Mtb). Both CD4+ and CD8+ T cells play a critical role in host immunity to Mtb. Infection of macrophages with Mtb results in upregulation of mymA operon genes thereby suggesting their importance as immune targets. In the present study, after exclusion of self-peptides mymA operon proteins of Mtb were analyzed in silico for the presence of Human Leukocyte Antigen (HLA) Class I and Class II binding peptides using Bioinformatics and molecular analysis section, NetMHC 3.4, ProPred and Immune epitope database software. Out of 56 promiscuous epitopes obtained, 41 epitopes were predicted to be antigenic for MHC Class I. In MHC Class II, out of 336 promiscuous epitopes obtained, 142 epitopes were predicted to be antigenic. The comparative bioinformatics analysis of mymA operon proteins found Rv3083 to be the best vaccine candidate. Molecular docking was performed with the most antigenic peptides of Rv3083 (LASGAASVV with alleles HLA-B51:01, HAATSGTLI with HLA-A02, IVTATGLNI and EKIHYGLKVNTA with HLA-DRB1_01:01) to study the structural basis for recognition of peptides by various HLA molecules. The software binding prediction was validated by the obtained molecular docking score of peptide-HLA complex. These peptides can be further investigated for their immunological relevance in patients of tuberculosis using major histocompatibility complex tetramer approach. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Tuberculosis (TB) is a global emergency, with ∼8 million new cases and ∼3 million deaths each year (Raja, 2004). Due to the failure of the MVA85A vaccine, expansion of antigenic repertoire is required to identify new potential TB vaccine candidates (Tameris et al., 2013; Singh et al., 2014). Genes of Mycobacterium tuberculosis (Mtb) involved in maintaining appropriate cell wall architecture have been found to play an important role in its virulence (Brennan, 2003). The mymA operon comprises of seven genes (Rv3083-Rv3089) and is named after the first gene mymA (Rv3083). The proteins of this operon are known to play a role in fatty acid metabolism and modification of mycobacterial cell wall (Singh et al., 2003, 2005; Kumar et al., 2009; Fisher et al., 2002). Rv3083 and Rv3086 genes of mymA operon are known to be cell wall

⁎ Corresponding author at: Department of Zoology, Miranda House, University of Delhi, Delhi 110007, India. E-mail addresses: [email protected] (I. Saraav), [email protected] (K. Pandey), [email protected] (M. Sharma), [email protected] (S. Singh), [email protected] (P. Dutta), [email protected] (A. Bhardwaj), [email protected] (S. Sharma).

http://dx.doi.org/10.1016/j.meegid.2016.07.004 1567-1348/© 2016 Elsevier B.V. All rights reserved.

associated which suggests that they might be immunogenic (Wolfe et al., 2010). This was validated in our previous studies, wherein Rv3083 was found to be capable of activating macrophages via Toll Like Receptor 2 (TLR2) in THP1 cell line (Saraav et al., 2014a, 2014b). Similar results were also obtained in the Peripheral Blood Mononuclear Cells (PBMCs) of Bacillus Calmette–Guérin (BCG) vaccinated healthy individuals. Until now, only 7% of the Mtb genome has been mined for T cell specific epitopes in humans (Geluk et al., 2014; Tang et al., 2010). Thirty Open Reading Frames (ORFs) have been searched so far, which accounts for only 65% of the epitopes reported (Chaitra et al., 2005; Geluk et al., 2014; Tang et al., 2010). Role of CD4+ and CD8+ T cells in limiting mycobacterial infection is well documented (Woodworth and Behar, 2006). Various studies on mouse and human models have identified specific roles of CD4+ and CD8+ T cell mediated immune responses (Prezzemolo et al., 2014). The protective immune response in TB involves T cell activation where the T cell receptor recognizes a complex formed by specific peptide fragment and a Major Histocompatibility Complex (MHC) molecule (Kruh et al., 2010). Specificity of the peptides during complex formation decides the outcome of the immune response. An experimental examination of each antigen for its vaccine

I. Saraav et al. / Infection, Genetics and Evolution 44 (2016) 182–189

candidature is expensive, tedious and long drawn process, an alternative in silico approach is cost and time effective (De Groot et al., 2002). Various tools are available for the identification of T-cell epitopes from proteins (Ivanyi and Ottenhoff, 2014; Rappuoli, 2007; Pandey et al., 2015). In the present work, we have used Bioinformatics and Molecular Analysis Section (BIMAS) and NetMHC 3.4 software for the CD8+ T cell epitope prediction and ProPred and Immune Epitope Database (IEDB) software for CD4+ T cell epitope prediction. Due to significant polymorphism in the peptide binding groove of MHC molecule it is important to determine the promiscuous epitopes (Frahm et al., 2007). Potential promiscuous epitopes of each protein were also confirmed for antigenicity by the VaxiJen server. Molecular docking was performed for the top VaxiJen scoring promiscuous epitopes of Rv3083 using CABSdock server and Hex program. These antigens can be tested using MHC tetramer approach to study their interactions with CD4 + and CD8 + cells in an in vitro/ex-vivo assay for their inclusion in future recombinant BCG or subunit vaccines. 2. Methods An outline of the methodology used in this study has been summarized in Fig. 1. 2.1. Retrieving protein sequences The complete amino acid sequences of mymA operon proteins were retrieved from tuberculist (http://tuberculist.epfl.ch/) in FASTA format. 2.2. Identification of self-peptides The mymA operon peptides predicted to bind to Human Leukocyte Antigen (HLA) alleles were analyzed for the exact match in the human proteome peptide library. 15-mer, 14-mer, 13-mer, 9-mer, 8-mer and 7-mer peptides were generated keeping a step-size of 1 amino acid (aa) across the entire protein length for mymA operon proteins using

183

an indigenously developed PERL script software. Likewise, the human proteome was downloaded from National Center for Biotechnology Information FTP and similar library of peptides was generated. Each of these peptides of mymA operon proteins was checked for identity with the peptides derived from 45,513 human ORFs. 2.3. Comparative analysis of the peptides of mymA operon proteins for their binding profile to MHC Class I and II molecules As MHC Class I epitopes are usually 8–10 aa long, nonamer was selected as the length of the peptide binding to MHC Class I. In order to cover maximum number of alleles with accuracy, two different software BIMAS (Parker et al., 1994) and NetMHC 3.4 server (Lundegaard et al., 2008; Nielsen et al., 2003, 2004) were used. In BIMAS, prediction was done using algorithm HLA BIND. For BIMAS a T1/2 of 100 min was chosen as a cut-off point in order to select the relatively high-affinity binding peptides. NetMHC 3.4 software predicts binding of peptides to different HLA molecules using Artificial Neural Networks. Only strong binders with half maximal inhibitory concentration b 500 (≤IC500) were selected from the total epitopes generated. BIMAS software covers 33 different alleles of HLA class I which includes 9 HLA-A alleles, 20 HLA-B alleles and 4 HLA-C alleles while NetMHC 3.4 covers 77 different HLA alleles which includes 34 HLA-A alleles, 33 HLA-B alleles and 10 HLA-C alleles. Two HLA-A, 6 HLA-B and 3 HLA-C alleles were found to be common in both the software. For MHC Class II, 9–18 aa long peptides were selected and two different software ProPred (Singh and Raghava, 2001) and IEDB (Vita et al., 2015) were used. ProPred is a graphical web tool for predicting MHC Class II binding regions in the antigenic protein sequences. The top scoring peptides were chosen by selecting the peptides with threshold N 1%. In the case of IEDB, SMM-align method was employed to obtain potential MHC Class II binders. Cut-off value of IC50 within 250 nM was chosen for the predicted binders. ProPred software covers 52 different alleles of MHC Class II which includes 52 HLA-DR alleles while IEDB covers 29 different HLA alleles which includes 8 HLA-DP alleles, 6 HLA-DQ alleles and

Fig. 1. Flowchart summarizing the protocol used for prediction of most probable T cell epitopes of mymA operon proteins of Mtb.

184

I. Saraav et al. / Infection, Genetics and Evolution 44 (2016) 182–189

Table 1 Possible overlapping nonamers generated from mymA operon proteins with step-size of one amino acid and their percent similarity with human proteins. Gene name

Rv number

mymA

Rv3083

lipR

Rv3084

sadhD

Rv3085

adhD

Rv3086

chp

Rv3087

chp

Rv3088

fadD13

Rv3089

Product Probable monooxygenase Probable acetyl-hydrolase Probable short-chain type dehydrogenase Probable zinc-type alcohol dehydrogenase Possible triacylglycerol synthase Putative triacylglycerol synthase Probable chain -fatty-acid CoA ligase

No. of overlapping nonamers with one aa frameshift

100% similarity

80% similarity

70% similarity

486

2

4

21

299

0

0

14

260

2

6

18

267

1

2

15

463

0

1

20

465

0

0

15

494

0

0

17

15 HLA-DR alleles. Eleven HLA-DR alleles were found to be common in both the software. 2.4. Prediction of antigenicity of promiscuous epitopes For MHC Class I alleles, epitopes which bound to ≥3 alleles in BIMAS and ≥5 alleles in NetMHC 3.4 were considered promiscuous. In the case of MHC Class II alleles, epitopes which bound to ≥ 5 alleles in both ProPred and IEDB were selected. The promiscuous epitopes obtained were tested for their antigenicity using VaxiJen server (Doytchinova and Flower, 2007) where epitopes showing score ≥0.4 were predicted as good antigens. 2.5. Molecular docking of the promiscuous epitopes binding to MHC Class I and Class II molecules The highest VaxiJen scoring promiscuous epitope of Rv3083 protein predicted through BIMAS was docked with the HLA-B51:01 (Protein Data Bank (PDB) ID: 1E27) and that selected through NetMHC 3.4 was docked with the HLA-A02 (PDB ID: 1S9Y). The 3D structure of HLAB51:01 complexed with a HIV immunodominant peptide “LPPVVAKEI” and HLA-A02 complexed with a peptide “SLLMWITQS” derived from a commonly expressed tumor antigen were retrieved from PDB database. These peptides “LPPVVAKEI” and “SLLMWITQS” were used as control peptides for HLA-B51:01 and HLA-A02 respectively. For MHC Class II alleles molecular docking was performed with the HLA-DRB1_01:01 (PDB ID: 1EKG0). The 3D structure of HLA-DRB1_01:01 complexed with an Epstein-Barr virus protein was retrieved from PDB database. Docking was performed first with the CABSdock server (Kurcinski and Kolinski, 2007) where both the HLA molecules and peptides were subjected to energy minimization followed by the molecular docking. Then, the peptides and the alleles were separated from the obtained docked structures and were again docked with the Hex program (Macindoe et al., 2010) for obtaining energy scores. A comparative analysis of the binding energy (Kcal/mol) of the test and control peptide with the HLA molecule was also done. The docked structures obtained from both the software were visualized using Discovery Studio Visualizer 4.1 (Accelyrs Inc.) and the binding sites were analyzed for the hydrogen bonds and hydrophobic interactions between the amino acid residues of the peptide and the HLA molecule. 3. Results 3.1. Sequence retrieval Sequences of 7 proteins of mymA operon of Mtb were retrieved and a total of 2833 possible overlapping 9-mer and 2798 15-mer peptides were generated from the entire protein length of all the proteins. This

library comprised of peptides that are generated using a step size of 1 (Table 1). P-BLAST was done to find sequence similarity among several sequenced mycobacterial species. It was found that the proteins of mymA operon shared 99% sequence similarity with other family members of Mtb complex like M. bovis, M. africanum, M. avium, M. canetti therefore, no differences were obtained in the aa's which could affect the predicted epitopes and their selection. 3.2. Self- or partially self-peptides among binding nonamers Binding peptides from mymA operon proteins were analyzed for similarities with the library of peptides generated from the human proteome. It was observed that for the nonameric peptides a total of 5

Table 2 Comparative analysis of the peptides of mymA operon proteins binding to MHC Class I molecules. Proteins of mymA operon arranged in descending order using BIMAS and NetMHC3.4 on the basis of (a) Number of epitopes generated (b) Class I HLA allele covered by the peptides (c) Number of promiscuous epitopes generated. (a) Software used

Number of epitopes generated

BIMAS

Rv3089 N Rv3087 N Rv3083 N Rv3088 N Rv3086 N Rv3084 N

NetMHC 3.4

Rv3085 Rv3088 N Rv3087 N Rv3083 N Rv3089 N Rv3084 N Rv3086 N Rv3085

(b) Software used

Class I HLA allele covered by the peptide

BIMAS

Rv3083 N Rv3089 N Rv3088 = Rv3087 N Rv3086 N Rv3085 N

NetMHC 3.4

Rv3084 Rv3083 N Rv3087 N Rv3088 N Rv3089 N Rv3084 N Rv3086 N Rv3085

(c) BIMAS NetMHC 3.4 Gene ID generating promiscuous epitopes

Number of peptides binding to one allele

Number of peptides binding to ≥three alleles

Number of peptides binding to one allele

Number of peptides binding to ≥five alleles

Rv3083 Rv3084 Rv3085 Rv3086 Rv3087 Rv3088 Rv3089

44 23 24 19 46 33 50

7 7 5 6 13 6 12

50 47 33 33 53 60 51

8 5 5 4 12 16 8

I. Saraav et al. / Infection, Genetics and Evolution 44 (2016) 182–189

185

Prediction of the binding profile of the generated peptides with the MHC Class I molecules was done at various levels as shown below:

3.3.2. HLA alleles binding of the peptides A peptide binding to more HLA alleles has a greater potential as a vaccine candidate therefore, HLA alleles covered by the peptides of each protein were calculated. In both BIMAS and NetMHC 3.4, epitopes of Rv3083 bound to maximum 19 and 53 HLA alleles respectively (Table 2b). In BIMAS, HLA-B locus alleles showed higher affinity of binding to the generated peptides as compared to alleles of A or C locus. In NetMHC 3.4, alleles of HLA-A locus showed higher affinity of binding to the generated peptides as compared to alleles of B or C locus (Fig. 2a & b).

3.3.1. Number of epitopes generated by mymA operon proteins In BIMAS, 551 epitopes were generated which bound to 33 Class I alleles. In NetMHC 3.4, 1452 strong binding epitopes were generated which bound to 77 Class I alleles. Rv3089 protein generated maximum number of 106 epitopes in BIMAS while Rv3084 generated maximum number of 324 epitopes in NetMHC 3.4 (Table 2a).

3.3.3. Promiscuity of peptide - HLA binding Most of the mymA operon peptides bound to single allele. However, it was found that in BIMAS Rv3087 generated maximum promiscuous epitopes while in NetMHC 3.4, Rv3088 gave highest number of promiscuous epitopes (Table 2c). The peptide “APLVAFSGL” of Rv3088 bound

peptides had 100% similarity, 15 peptides had 90% similarity and 120 peptides had 80% similarity (Table 1). These peptides were eliminated before further analysis. For the 15-mer peptides no self or partially self-peptides were found. 3.3. Prediction of the T cell epitopes binding to the MHC Class I molecules

Fig. 2. Binding profile of mymA operon proteins to MHC Class I and Class II molecules. (a) In BIMAS, HLA-B locus alleles showed higher affinity of binding to the generated peptides. Maximum number of nonamers was found to be recognized by the allele HLA-B27:05 (145 peptides) followed by HLA-B51:01 (b) In NetMHC 3.4, alleles of HLA-A locus showed higher affinity of binding to the generated peptides. The largest number of nonamers was found to be recognized by the allele HLA-A02:50 (151 peptides) followed by HLA-A02:11 (c) In IEDB, HLA-DR locus alleles showed higher affinity of binding to the generated peptides. Maximum number of nonamers was found to be recognized by the allele HLA-DRB1_01:01 (1119 peptides).

186

I. Saraav et al. / Infection, Genetics and Evolution 44 (2016) 182–189

to maximum 4 alleles in BIMAS and the peptide “HMADNGYTV” of Rv3083 bound to 13 alleles in NetMHC 3.4.

3.3.4. Prediction of the antigenicity of promiscuous epitopes Fifty six promiscuous epitopes were obtained from mymA operon proteins using both the software. These epitopes were then subjected to the VaxiJen server. Out of these, 22 epitopes from BIMAS and 19 epitopes from NetMHC 3.4 were found to be antigenic. The top two VaxiJen scoring promiscuous epitopes of proteins of mymA operon generated using both the software have been listed in Supplementary Table S1.

3.4. Prediction of the T cell epitopes binding to the MHC Class II molecules Similarly, prediction of the binding profile of the generated peptides with the MHC Class II molecules was also done at various levels as shown below:

3.4.1. Number of epitopes generated by mymA operon proteins Proteins of mymA operon generated a total of 1089 epitopes that bound to 52 Class II HLA alleles using ProPred and 1010 epitopes that bound to 29 Class II HLA alleles using IEDB. In ProPred, Rv3083 protein generated maximum of 233 epitopes while in IEDB, Rv3088 generated maximum of 994 epitopes (Table 3a).

3.4.2. HLA alleles binding of the peptides In both ProPred and IEDB, epitopes of Rv3083 bound to maximum MHC Class II alleles (Table 3b). In ProPred, almost all peptides of mymA operon could bind to MHC Class II alleles present. In IEDB, HLADR locus alleles showed higher affinity of binding to the generated peptides (Fig. 2c). Table 3 Comparative analysis of the peptides of mymA operon proteins binding to MHC Class II molecules. Proteins of mymA operon arranged in descending order using ProPred and IEDB on the basis of (a) Number of epitopes generated (b) MHC Class II allele covered by the peptides (c) Number of promiscuous epitopes generated. (a) Software used

Number of eptiopes generated

ProPred

Rv3083 N Rv3088 N Rv3089 N Rv3086 N Rv3085 N Rv3087 N

IEDB

Rv3084 Rv3088 N Rv3083 N Rv3087 N Rv3089 N Rv3084 N Rv3086 N Rv3085

(b) Software used

Class II HLA allele covered by the peptide

ProPred

Rv3083 = Rv3088 = Rv3089 N Rv3087 N Rv3086 N Rv3085 N

IEDB

Rv3084 Rv3083 N Rv3086 N Rv3087 N Rv3084 = Rv3088 N Rv3089 N RV3085

(c) ProPred IEDB Gene ID generating promiscuous epitopes

Number of peptides binding to one allele

Number of peptides binding to ≥five alleles

Number of peptides binding to one allele

Number of peptides binding to ≥five alleles

Rv3083 Rv3084 Rv3085 Rv3086 Rv3087 Rv3088 Rv3089

9 5 3 1 4 4 9

19 6 6 12 5 17 10

71 42 43 70 74 171 126

57 39 31 29 46 52 7

3.4.3. Promiscuity of peptide - HLA binding Seventy five promiscuous epitopes were obtained in ProPred and 261 promiscuous epitopes were obtained in IEDB (Table 3c). For MHC Class II epitope “VRLAQALGA” of Rv3085 bound to 32 alleles in ProPred and to 47 alleles in IEDB. 3.4.4. Prediction of the antigenicity of promiscuous epitopes Out of 261 promiscuous epitopes, 36 epitopes from ProPred and 106 epitopes from IEDB were found to be antigenic. The top two VaxiJen scoring promiscuous epitopes of each protein have been listed in Supplementary Table S2. 3.5. Confirmation of MHC- peptide interactions by molecular docking studies Comparative analysis of mymA operon proteins on different parameters discussed above showed Rv3083 as the best vaccine candidate. Hence molecular docking of the most antigenic and promiscuous Rv3083 peptides was done with MHC Class I (HLA-B51:01, HLA-A02) and Class II molecules (HLA-DRB1_01:01) which are prevalent globally as well as found in North Indian population. These alleles also bound to maximum number of epitopes of mymA operon proteins in our study (Fig. 2a, b, c). All the antigenic promiscuous peptides of mymA operon proteins gave similar energy scores as the control peptides for both MHC Class I and Class II alleles. Interestingly, both the software gave similar binding site interactions of the peptides with the MHC molecules. Results of the energy scores of the molecular docked structures obtained from Hex program along with their interactions have been listed in Table 4 and depicted in Figs. 3 and 4. 4. Discussion The proteins of mymA operon play an important role in the intracellular survival of Mtb. They are involved in the modification of the cell wall (Singh et al., 2003, 2005). Microarray analysis revealed that genes of mymA operon are among the highest induced genes after acid shock (Fisher et al., 2002). The immunogenic potential of Rv3083 has been studied in our lab, in both THP1 cell line and PBMCs where Rv3083 was found to activate immune cells via TLR2. Therefore, we proposed that the proteins of mymA operon might prove to be potential vaccine candidates using an immunoinformatics and molecular modeling approach of wider population coverage. An effective vaccine for TB should be a good inducer of protective cellular immune response (Kruh et al., 2010). Various studies have reported that majority of TB infected individuals control Mtb in a T celldependent manner. Both CD4 + and CD8 + T cells are considered to play a major role in protection against Mtb (Prezzemolo et al., 2014). Only a fraction of the Mtb proteome has been mined for the identification of antigens capable of stimulating T cells (Mehra and Kaur, 2003). For this reason, new bioinformatics approach has been used to identify Mtb epitopes that can be presented by MHC Class I and Class II molecules to T cells (Cho et al., 2000; Ehlers and Schaible, 2013). In the present work, we have done T cell epitopes prediction of mymA operon proteins for both MHC Class I and Class II. For MHC Class I BIMAS and NetMHC 3.4 software were used. For MHC Class II ProPred and IEDB software were used. Among the HLA- Class I binding peptides of MymA operon proteins, 135 peptides were found to have partial similarity to human proteins. Five peptides (2 peptides “VLIIGAGLS” and “VIGSGATAV” of Rv3083, 2 peptides “TGAGSGIGR” and “GAGSGIGRA” of Rv3085 and 1 peptide “AAAKVDELT” of Rv3086) were found to have complete homology with the human proteins and are designated self peptides. For MHC Class II no self or partially self-peptides were found. The self peptides were removed from the analysis as no self peptide specific T cells are expected to be present in healthy population. A total of 2833 overlapping nonamers were generated from mymA operon proteins. In BIMAS, 551 HLA binding peptides had a

I. Saraav et al. / Infection, Genetics and Evolution 44 (2016) 182–189

187

Table 4 Molecular modeling results showing binding energy scores and the interactions of antigenic promiscuous peptides of Rv3083 with MHC Class I and Class II molecules using Hex program. Residues involved in hydrogen bond

Residues involved in Hydrophobic bond

Peptide residue –allele residue

Peptide residue –allele residue Leu 1 - Trp 95, Ile 124, Tyr 116, Trp 133 Ala 2 - Trp 147, Tyr 159 Ala 5 - Ile 366 Val 8 - Met 5 , Phe 33, His 171 Val 9 - Ile 52, Trp 167 Ala 2 - Trp 95

Gene Rv3083

Amino acid sequence of the top VaxiJen scoring promiscuous epitope of the protein

Binding HLA allele

Energy score (Kcal/mol)

BIMAS

LASGAASVV

HLA-B51:01

−424

Leu 1 - Trp 95 Se 3 - Asn 77 Gly 4 - Asn 70 Ser 7 - Asn 73

Net MHC 3.4

HAATSGTLI

HLA-A02

−534

Ser 5- Ala 377, Asn 70 Leu 8 - Gly 381

Ala 3 – Tyr 116, Trp 147 Leu 156 Leu 8 - Tyr 99, Cys 164

ProPred

IVTATGLNI

HLA-DRB1_01:01 −451

Ile 1- Phe 30 Thr 3- Glu53, Phe 52

IEDB

EKIHYGLKVNTA

HLA-DRB1_01:01 −719

Lys 2His 4Gly 6Leu 7-

Asn 60 Glu 7 Asp 235 Glu 264

Ile 9 - Met 5 , Tyr 7, Val 34, His 171 Ile 1- Ile 29, Phe 30 Ala 4- Tyr 256, Phe 22 Leu7- Phe 191 Ile 9- Cys 208, Trp 187, Val 216 Ile 3- Phe 20, Phe 49 Val 9- Phe 191 Trp 11- Phe 235, Tyr 238

Fig. 3. Docking of the antigenic promiscuous peptide of Rv3083 with MHC Class I molecule. (a) Rv3083 peptide “LASGAASVV” in the binding groove of the HLA-B51:01 (b) The interactions between the peptide and the HLA molecule are shown where hydrogen bonds are in green broken lines and hydrophobic interactions are in orange broken lines. Peptide residues are in orange color and residues of HLA are in purple color.

188

I. Saraav et al. / Infection, Genetics and Evolution 44 (2016) 182–189

Fig. 4. Docking of the antigenic promiscuous peptide of Rv3083 with MHC Class II molecule. (a) Rv3083 peptide “EKIHYGLKVNTA” in the binding groove of HLA-DRB1_01:01 (b) The interactions between the peptide and the HLA molecule are shown where hydrogen bonds are in green broken lines and hydrophobic interactions are in orange broken lines. Peptide residues of are in orange color and residues of HLA in purple color.

T1/2 ≥ 100 min. Interestingly, 105 peptides from these had a higher T1/2 of ≥ 500 min indicating their greater binding affinity. In NetMHC 3.4, it was found that majority of the peptides of mymA operon proteins bound to 77 HLA alleles present in the software. Similarly for MHC Class II most of the peptides bound to HLA alleles present in ProPred and IEDB software. The stability of T Cell Receptor (TCR)HLA-peptide complex plays a crucial role in the determination of specific immune response (Vani et al., 2006). Hence, binding profile of the peptides with the HLA molecules was studied at three levels; first at the protein level by calculating the total number of epitopes generated; then, at the peptide level by calculating number of HLA alleles bound; and finally, at allele level by analyzing promiscuity for various epitopes. Of the total peptides generated from mymA operon proteins, 83% bound to HLA-B alleles in BIMAS and 63% bound to HLA-A alleles in NetMHC 3.4 which clearly indicated their respective affinity towards HLA-B or HLA-A locus. Interestingly, it was observed that one of the promiscuous epitopes “LASGAASV” of Rv3083 was common in both the software of MHC Class I. In case of ProPred and IEDB, most peptides bound to HLA-DR alleles. The epitope “WAPRYVGSSDPRL” of protein Rv3084 was found to be overlapping between the two software of MHC Class II. The peptide “LLPLPMFHV” of Rv3089 protein had the highest VaxiJen score 2.6012 for Class I and peptide “SATIVGGHEGSG” of Rv3086 had the highest score 3.1097 for MHC Class II making them highly antigenic. However, it was observed that the peptides with highest affinity to the MHC Class I or Class II molecules were mostly promiscuous but not necessarily antigenic. Combining the data at each level Rv3083 was found to be the top vaccine candidate showing good population coverage. Molecular

docking of the top VaxiJen scoring promiscuous epitopes of Rv3083 protein was performed with the HLA-B51:01, HLA-A02 and HLADRB1_01:01 due to their high global prevalence. The population coverage of these alleles globally is 15–25% for HLA-B51:01, 45–55% for HLA-A2 and of 8–21% for HLA-DRB1_01:01 (http://www. allelefrequencies.net). In North Indian population the coverage of HLA-B51:01 is 10%, of HLA-A2 is 40% and of HLA-DRB1_01:01 is 10% (Mehra and Kaur, 2003; Balamurugan et al., 2004). The molecular modeling of selected antigenic peptide with HLA further confirmed the usefulness of these peptides as important vaccine candidates. The binding energy scores of the peptides with the alleles were found to be close to that of the control immunogenic peptides “LPPVVAKEI” and “SLLMWITQS”. These identified antigenic CD4 + and CD8 + T cell epitopes of Rv3083 protein with global population coverage can be followed for their specificity in PBMC of active and latent TB patients by using MHC tetramer approach. Conflict of interest The authors have no competing interests to declare. Acknowledgement We are thankful to Dr. Pratibha Jolly, Principal and PI, D S Kothari Centre for Research and Innovation in Science Education, Miranda House, and University of Delhi for her support and encouragement. The Senior and Junior Research Fellowships to Iti Saraav, Swati Singh and Kirti Pandey respectively by Indian Council of Medical Research (ICMR) and the Department of Biotechnology (DBT) (DBT- BT/

I. Saraav et al. / Infection, Genetics and Evolution 44 (2016) 182–189

PR5638/MED/29/580/2012) and the financial support of Council of Scientific & Industrial Research (OSDD, CSIR) (OSDD/HCP001/12FYP/20122013/FIN/2408) Government of India, to carry out the project is gratefully acknowledged. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.meegid.2016.07.004. References Balamurugan, A., et al., 2004. Human leukocyte antigen class I supertypes influence susceptibility and severity of tuberculosis. J. Infect. Dis. 189, 805–811. Brennan, P.J., 2003. Structure, function, and biogenesis of the cell wall of Mycobacterium tuberculosis. Tuberculosis (Edinb) 83, 91–97. Chaitra, M.G., et al., 2005. Defining putative T cell epitopes from PE and PPE families of proteins of Mycobacterium tuberculosis with vaccine potential. Vaccine 23, 1265–1272. Cho, S., et al., 2000. Antimicrobial activity of MHC class I-restricted CD8+ T cells in human tuberculosis. Proc. Natl. Acad. Sci. U. S. A. 97, 12210–12215. De Groot, A.S., et al., 2002. Immuno-informatics: mining genomes for vaccine components. Immunol. Cell Biol. 80, 255–269. Doytchinova, I.A., Flower, D.R., 2007. VaxiJen: a server for prediction of protective antigens, tumour antigens and subunit vaccines. BMC Bioinforma. 8, 4. Ehlers, S., Schaible, U.E., 2013. The granuloma in tuberculosis: dynamics of a host-pathogen collusion. Front. Immunol. 3, 411. Fisher, M.A., et al., 2002. Microarray analysis of the Mycobacterium tuberculosis transcriptional response to the acidic conditions found in phagosomes. J. Bacteriol. 184, 4025–4032. Frahm, N., et al., 2007. Extensive HLA class I allele promiscuity among viral CTL epitopes. Eur. J. Immunol. 37, 2419–2433. Geluk, A., et al., 2014. Innovative strategies to identify M. tuberculosis antigens and epitopes using genome-wide analyses. Front. Immunol. 5, 256. Ivanyi, J., Ottenhoff, T.H., 2014. Significance of antigen and epitope specificity in tuberculosis. Front. Immunol. 5, 524. Kruh, N.A., et al., 2010. Portrait of a pathogen: the Mycobacterium tuberculosis proteome in vivo. PLoS One 5, 13938. Kumar, P., et al., 2009. The Mycobacterium tuberculosis protein kinase K modulates activation of transcription from the promoter of mycobacterial monooxygenase operon through phosphorylation of the transcriptional regulator VirS. J. Biol. Chem. 284, 11090–11099. Kurcinski, M., Kolinski, A., 2007. Steps towards flexible docking: modeling of three-dimensional structures of the nuclear receptors bound with peptide ligands mimicking co-activators' sequences. J. Steroid Biochem. Mol. Biol. 103, 357–360. Lundegaard, C., et al., 2008. Accurate approximation method for prediction of class I MHC affinities for peptides of length 8, 10 and 11 using prediction tools trained on 9mers. Bioinformatics 24, 1397–1398.

189

Macindoe, G., et al., 2010. HexServer: an FFT-based protein docking server powered by graphics processors. Nucleic Acids Res. 38, W445–W449. Mehra, N.K., Kaur, G., 2003. MHC-based vaccination approaches: progress and perspectives. Expert Rev. Mol. Med. 5, 1–17. Nielsen, M., et al., 2003. Reliable prediction of T-cell epitopes using neural networks with novel sequence representations. Protein Sci. 12, 1007–1017. Nielsen, M., et al., 2004. Improved prediction of MHC class I and class II epitopes using a novel Gibbs sampling approach. Bioinformatics 20, 1388–1397. Pandey, K., et al., 2015. Analysis of the DosR regulon genes to select cytotoxic T lymphocyte epitope specific vaccine candidates using a reverse vaccinology approach. Int. J. Mycobacteriol. 5, 34–43. Parker, K.C., et al., 1994. Scheme for ranking potential HLA-A2 binding peptides based on independent binding of individual peptide side-chains. J. Immunol. 152, 163–175. Prezzemolo, T., et al., 2014. Functional signatures of human CD4 and CD8 T cell responses to Mycobacterium tuberculosis. Front. Immunol. 5, 180. Raja, A., 2004. Immunology of tuberculosis. Indian J. Med. Res. 120, 213–232. Rappuoli, R., 2007. Bridging the knowledge gaps in vaccine design. Nat. Biotechnol. 25, 1361–1366. Saraav, I., et al., 2014a. Cell wall-associated Mycobacterium tuberculosis rRv3083 protein stimulates macrophages through toll-like receptor-2 (TLR2). Int. J. Mycobacteriol. 4, 176. Saraav, I., et al., 2014b. Outcome of Mycobacterium tuberculosis and toll-like receptor interaction: immune response or immune evasion? Immunol. Cell Biol. 92, 741–746. Singh, H., Raghava, G.P., 2001. ProPred: prediction of HLA-DR binding sites. Bioinformatics 17, 1236–1237. Singh, A., et al., 2003. mymA operon of Mycobacterium tuberculosis: its regulation and importance in the cell envelope. FEMS Microbiol. Lett. 227, 53–63. Singh, A., et al., 2005. Requirement of the mymA operon for appropriate cell wall ultrastructure and persistence of Mycobacterium tuberculosis in the spleens of Guinea pigs. J. Bacteriol. 187, 4173–4186. Singh, S., et al., 2014. Immunogenic potential of latency associated antigens against Mycobacterium tuberculosis. Vaccine 32, 712–716. Tameris, M.D., et al., 2013. Safety and efficacy of MVA85A, a new tuberculosis vaccine, in infants previously vaccinated with BCG: a randomised, placebo-controlled phase 2b trial. Lancet 381, 1021–1028. Tang, S.T., et al., 2010. Genome-based in silico identification of new Mycobacterium tuberculosis antigens activating polyfunctional CD8+ T cells in human tuberculosis. J. Immunol. 186, 1068–1080. Vani, J., et al., 2006. A combined immuno-informatics and structure-based modeling approach for prediction of T cell epitopes of secretory proteins of Mycobacterium tuberculosis. Microbes Infect. 8, 738–746. Vita, R., et al., 2015. The immune epitope database (IEDB) 3.0. Nucleic Acids Res. 43, 405–412. Wolfe, L.M., et al., 2010. Proteomic definition of the cell wall of Mycobacterium tuberculosis. J. Proteome Res. 9, 5816–5826. Woodworth, J.S., Behar, S.M., 2006. Mycobacterium tuberculosis-specific CD8+ T cells and their role in immunity. Crit. Rev. Immunol. 26, 317–352.