Insights from the predicted epitope similarity between Mycobacterium tuberculosis virulent factors and its human homologs

Mycobacterium tuberculosis is known to be associated with several autoimmune diseases such as systemic lupus erythematous, rheumatoid arthritis and multiple sclerosis. This is attributed to sequence similarity between virulent factors and human proteins. Therefore, it is of interest to identify such regions in the virulent factors to assess potential autoimmune related information. M. tb specific virulent factors were downloaded from the VFDB database and its human homologs were identified using the sequence comparison search tool BLASTP. Both virulent proteins and their corresponding human homologs were further scanned for epitopes (B cell and HLA class I and II allele specific) using prediction programs (BCPRED and NETMHC). Data shows the presence of matching 22 B-cell, 79 HLA class II and 16 HLA class I specific predicted epitopes in these virulent factors having human homologs. A known peptide (HAFYLQYKNVKVDFA) associated with autoimmune atopic dermatitis is shown in the superoxide dismutase homolog structures of the bacterium (PDB ID: 1IDS) and human (PDB ID: 2QKC). This data provides insight into the understanding of infection-associated auto-immunity

: A workflow showing steps involved in the identification of epitopes in the virulent factors of M. tb having human homologs. Epitopes were predicted in both virulent factors and their corresponding homologs. Both virulent proteins and their corresponding human homologs were further scanned for epitopes (B cell (BCPRED with score > 0.9) and HLA class I and II alleles (NETMHC with binding score < 50nM) prediction.

Phobius 1.0.1
Phobius was used to identify and exclude the signal region of the homologous proteins. [23].

Conserved Domain Database (CDD)
This database was used to identify conserved domains in homologous proteins of M.tb virulent factors and human.

B-cell epitope prediction server (BCPREDS)
Prediction of B cell epitopes (Table 1) for M.tb specific virulent factors and its corresponding human homologs using B cell epitope prediction server (BCPREDS) [24].

T-cell epitope prediction
Prediction of HLA class-I ( proteins were run through Phobius to remove predicted Nterminal signal peptides from the protein sequence. Then sequences are run through CDD for getting domain coordinates. Further the collected sequences are run through BCPRED server for B cell epitopes of 20 amino acids length and the classifier specificity was 75% and overlap filter was used for analysis. Based on prior BLAST results, regions of amino acids (small peptides) that were similar between the human and M. tuberculosis proteins were selected for further analysis. BCPRED score of greater than 0.9 is considered for blast matched peptides in both pathogen and host homologs.
NetMHC (version 2.2) was used for HLA class II and NetMHC (version 3.0) for HLA class I binding peptide prediction. Peptides were selected based on IC50 values <50 nM as high affinity, <500 nM as intermediate affinity and <5000 nM as low affinity [27]. The matched peptides in both pathogen and host with a binding score less than IC50 ≤ 50 are considered as strong binders. 3D structures of protein sequences matched to host are viewed and aligned structurally to find out whether these peptides are on the surface of the protein. The similarity between the predicted epitopes of virulent factors was found by multiple structural alignments using the STAMP algorithm in VMD. The detection of epitopes is shown in Figure 1. All calculations were performed using the local Linux server.

Results & Discussion:
The analysis of data obtained from the search between M. tuberculosis virulence factors and the human proteome revealed considerable similarities in sequences. A total of 25 best-hit homologous proteins with E-value cut off 1-E02 and similar regions of 9 or more amino acids were identified. The classification of the homologous virulent factor proteins are 21 metabolic proteins 3 membrane associated protein and a protein kinase (Table 3). Binding affinities of M.tb virulent factors vs. B cell epitopes and HLA class II and I alleles were measured by BCPRED and NetMHC. A peptide was considered having significant affinity to virulent factors if it had a BCPRED score ≥ 0.9 for B cell epitope (Table 4) and IC50 value ≤ 50 for HLA class II and I epitopes. The analysis of binding affinities of HLA class II peptides is 83% as compared to HLA class I (17%). Of 79 HLA class II host-pathogen epitopes highest affinity was to HLA-DRB10101 (57% followed byHLA-DRB10701 (14%), HLA-DRB10401 (11%), HLA-DRB10301 (6%), HLA-DRB11101 (5%), HLA-DRB10302 (2.5%) and HLA-DRB11501 (2.5%). The analysis of HLA class I peptides indicated a maximum affinity of peptides binding to allele A*0201 (44%) followed by B*0702 (31%), 2 for A*1101 (13%) and 6% each for A*0101 & A*2402 ( Figure 2, Table 1 & 4). HLA class II has significant number of high affinity binding peptides which could be involved in dys-regulation of T cell function and or autoimmunity [28]. The virulent factors binding to host tissue antigens could influence signaling and immune evasion [29]. The myco-bacterial virulent proteins of this study were classified into categories such as structural, metabolic, catalytic, kinases and transport proteins ( Table 4, 1 & 2). Majority of the virulent factor epitopes having binding affinity for B and HLA class I and II alleles were involved in (i) lipid, protein and nucleotide metabolism/degradation pathways (ii) free radical mediated damage pathway (iii) ion transport (iv) degradation of proteins glycosylation/phosphorylation pathways. These similarities could impact metabolic rate of reactions, interfere in homeostasis of cell and could trigger cell damage by free radical mediated reactions [29]. Peptides, which have binding capacity to more than one allele of HLA class-I and-II, are called promiscuous peptides. Promiscuous peptides for HLA class II were 24% (19/79) and none for HLA class I molecules. Interestingly, the presence of promiscuous peptides for HLA class II suggest that these peptides could have role in presentation of antigens for immune recognition and amplification of response against M.tb (Table 3 and  HAFYLQYKNVKVDFA, bound to allele HLA-DRB1*15:01 allele with high affinity is identified (Table 3). HLA-DRB1*15:01 is known to be responsible for susceptibility to tuberculosis. Further there was a high structural similarity of M.tb SOD and human MnSOD at both primary and tertiary structure level Figure 3 [35]. Clinical studies identified MnSOD cross-reactive autoimmune antibodies in patients with atopic dermatitis (AD) and has been implicated in disease pathogenesis [31]. This epitope is conserved and well investigated in Aspergillus fumigatus Mn SOD (1KKC) in relation to various autoimmune conditions [31][32][33][34]. Identifying the key homologous peptides of host pathogen similarity could help us design highly selective peptide blockers, which would be a valuable addition to complement the understanding of autoimmune diseases.
PDB crystal structures of superoxide dismutase M.tb, human and Aspergillus fumigatus were available. Superposition revealed a high measure of structural conservation and similarity with low RMSD having Qres value of 0.9 and showing high measure of the similarity of the 'C-C alpha' distances between residues of aligned proteins (Figure 3) [36]. These structurally similar regions of these three epitopes (which is known to cause atopic dermatitis) could be significant in tuberculosis in causing immune inflammatory processes characteristic of TB (Figure 3). It can be noted that many other mycobacterial antigens have been associated with autoimmune diseases [37][38][39]. There is no clear evidence that M.tb virulent factors are involved and further clinical investigations on epitope specificities involved in autoimmunity are warranted.
Although, computational tools have been used in the past to examine molecular mimics in other diseases [40]; the understanding of these epitopes need to be further probed. Utilizing these methods, we have identified potential autoreactive B cell, HLA class II and class I epitopes that may elicit autoimmune response during M. tuberculosis infection. The findings of this study are as follows: (i) there were 95 auto reactive B cell, HLA class II and class I epitopes that are similar to peptides of myco-bacterial virulent factors; (ii) 22% of similarities were promiscuous that are binding to HLA class II cell epitopes (iii) high Qres score of 0.9 suggesting structural similarity between M.tb SOD and human Mn SOD and the epitope has an established evidence of autoimmunity. The similarities were observed across the spectrum of metabolic activities of host cell suggesting M.tb could use multiple split approach in causing tuberculosis.

Conclusions:
We report regions in the M.tb virulent factors having human homologs sharing predicted B-cell and T-cell epitopes. Data shows the presence of 22 B-cell, 79 HLA class II and 16 HLA class I specific predicted peptides in these virulent factors having human homologs. A known peptide (HAFYLQYKNVK VDFA) associated with autoimmune atopic dermatitis is shown in the superoxide dismutase homolog structures of the bacterium (PDB ID: 1IDS) and human (PDB ID: 2QKC). This data provides insights in understanding infection-associated auto-immunity.