Integrated computational prediction and experimental validation identifies promiscuous T cell epitopes in the proteome of Mycobacterium bovis

Abstract

The discovery of novel antigens is an essential requirement in devising new diagnostics or vaccines for use in control programmes against human tuberculosis (TB) and bovine tuberculosis (bTB). Identification of potential epitopes recognised by CD4+ T cells requires prediction of peptide binding to MHC class-II, an obligatory prerequisite for T cell recognition. To comprehensively prioritise potential MHC-II-binding epitopes from Mycobacterium bovis, the agent of bTB and zoonotic TB in humans, we integrated three binding prediction methods with the M. bovisproteome using a subset of human HLA alleles to approximate the binding of epitope-containing peptides to the bovine MHC class II molecule BoLA-DRB3. Two parallel strategies were then applied to filter the resulting set of binders: identification of the top-scoring binders or clusters of binders. Our approach was tested experimentally by assessing the capacity of predicted promiscuous peptides to drive interferon-γ secretion from T cells of M. bovis infected cattle. Thus, 376 20-mer peptides, were synthesised (270 predicted epitopes, 94 random peptides with low predictive scores and 12 positive controls of known epitopes). The results of this validation demonstrated significant enrichment (>24 %) of promiscuously recognised peptides predicted in our selection strategies, compared with randomly selected peptides with low prediction scores. Our strategy offers a general approach to the identification of promiscuous epitopes tailored to target populations where there is limited knowledge of MHC allelic diversity.

Design of an Epitope-Based Vaccine Ensemble for Animal Trypanosomiasis by Computational Methods

“…Therefore, the next step to take will be to experimentally validate that the predicted epitopes are presented and recognized in the cow immunological context, and that they are capable of eliciting immunogenic responses. Despite being based on predictions, the goodness of currently available tools for this purpose has been acknowledged by several authors and the use of prediction tools is seen as a great aid for the identification of new epitopes [19,[36][37][38]. Since the protein space of trypanosomes, particularly T. congolense, T. evansi, and T. vivax remains largely unexplored, the majority of the antigens of origin of the predicted epitopes were either uncharacterized proteins or had low annotation scores inferred by homology.…”

“…For the prediction of CD4 T-cell epitopes we used the IEDB recommended method, the MHC II binding predictions tool (http://tools.iedb.org/mhcii/) [18], over the invariant proteome file. Given that no bovine lymphocyte antigen (BoLA) alleles are available for this tool we ran the program against the set of human leukocyte antigens (HLAs), as described elsewhere [19]. We considered for further analysis only the 0.01% top scoring CD4 T-cell epitopes that were predicted on the T. cruzi-masked proteome file.…”

“…The MHC-II binding tool at IEDB does not allow to specifically select class II bovine leukocyte antigens for the binding predictions so we used the available HLAs and the IEDB recommended set of tools [18]. Although the validated data are limited, MHC-II predictions using human alleles have been shown to be helpful towards the identification and validation of Mycobacterium bovis epitopes in cows [19]. We applied these MHC binding tools onto the invariant proteome of AAT-causing trypanosomes.…”

Computational Analysis of African Swine Fever Virus Protein Space for the Design of an Epitope-Based Vaccine Ensemble

“…Since there are no available SLA class II alleles for this tool, we used the HLA class II reference set consisting of 27 alleles available at IEDB [ 81 ]. It has been described that HLA-II can be used to approximate predictions for other mammals [ 82 ].…”

Antigenic outer membrane proteins prediction of Pasteurella multocida serotype B:2

“…Majority of MHC class II epitope binding algorithm used prediction towards human leukocyte antigens (HLA) as compared to other leukocyte antigens particularly bovine leukocyte antigens (BoLA). However, it was found that there was a certain pseudo-sequence (representation of the MHC-II complex polymorphic binding pockets) similarity between both species' alleles to approximate the protein binding in bovine (Farrell et al, 2016). Hence, the application of the selected HLA alleles that were found shown to be similar to BoLA by Farrel et al (2016).…”