Computational Design Yields Multi-Epitope Tuberculosis Vaccine Candidate Fused to Bacterial Enterotoxin Fragment

Researchers have used reverse vaccinology and immunoinformatic screening to design a new multi-epitope vaccine candidate against Mycobacterium tuberculosis, the bacterium that infects roughly a quarter to a third of the world’s population. The study, published in Genomics & Informatics, combined a pan-genome analysis of 256 complete M. tuberculosis genome sequences with epitope predictions and molecular docking, and introduced seven shortlisted immunogenic proteins and seven immunoreactive epitopes as candidate vaccine components. Like all purely computational vaccine designs, the construct has not yet been tested in the laboratory, and the authors state that experimental validation in vitro and in vivo is the necessary next step.

The motivation rests on well-documented shortcomings of current tuberculosis control. According to the 2024 World Health Organization Global Tuberculosis Report cited by the authors, 8.2 million new tuberculosis cases were diagnosed in 2023, and global tuberculosis mortality fell by only 23 percent between 2015 and 2023, well short of the WHO End TB Strategy target of a 75 percent reduction by 2025. The Bacille Calmette-Guérin vaccine, the only approved tuberculosis vaccine, is moderately effective in infants but has shown controversial efficacy in adolescents and adults across several trials. Multidrug-resistant strains, resistant to both rifampicin and isoniazid, account for approximately a quarter of deaths caused by antibiotic-resistant infections, adding urgency to the search for next-generation vaccines.

The team, led by Narjes Noori Goodarzi and Farzad Badmasti, began with the protein content of 256 M. tuberculosis strains with complete whole-genome sequences, analyzed using the BPGA pan-genome pipeline. They identified 2,134 core proteins, 4,354 pan-genome proteins and 1,720 accessory proteins, with core proteins making up 49 percent of the pan-proteome. A B-parameter of 0.0173889 indicated an almost closed pan-genome, meaning that screening core proteins offers targets likely to be shared across circulating strains. In parallel, they examined all 3,901 coding sequences of strain H112, a hypervirulent Beijing-lineage isolate.

From these starting sets, the researchers identified 80 surface-exposed proteins, 67 extracellular and 13 cell-wall associated, of which 69 had a prevalence above 90 percent across the 256 strains. Successive filters narrowed this pool: 65 proteins had one or fewer transmembrane helices, 48 were antigenic by the VaxiJen predictor with a score above 0.5, seven allergens were removed, and ten proteins with sequence similarity to the human proteome were excluded. Thirty-one proteins survived these screens, and seven with linear B-cell epitope ratios and MHC II binding site ratios both above 0.1 were shortlisted.

The seven final immunogenic targets span several functional classes: two PE/PPE family proteins, MT2404 and PPE40, whose precise functions remain unknown; the stress-responsive chaperone Acr2, a small heat shock protein; BlaC, a class A beta-lactamase; the peptidoglycan-associated proteins RipC and RipA; and a C40 family peptidase. RipC and MT2404 carried the highest numbers of conformational B-cell epitopes in structural predictions, and all seven proteins showed high sequence conservation among M. tuberculosis strains, an attribute the authors consider essential for a protective antigen intended to cover diverse circulating lineages.

The second arm of the study drew on experimentally validated data from the Immune Epitope Database. Comparing 4,718 immunoreactive linear B-cell epitopes of M. tuberculosis against the 80 surface-exposed core proteins yielded 2,284 matching epitopes, of which 719 were non-redundant with 100 percent identity. One epitope resembled the human proteome and was discarded, leaving 718. Cross-referencing with IEDB assay data identified 460 epitopes associated with interferon-gamma release, a cytokine the authors highlight as central to protective cell-mediated immunity against tuberculosis because it promotes phagosome maturation in infected macrophages.

Further filtering progressively narrowed the epitope set: 152 epitopes had conservancy scores above 95 percent, 29 of those were antigenic with VaxiJen scores above 1, and 14 were non-allergenic. Finally, comparison against T-cell receptor binding epitopes using the TCRMatch tool, with coverage and identity both above 50 percent, left seven epitopes with significant homology to TCR-recognized peptides. These were derived from the antigens Ag85A, Ag85C, EsxB and a PPE family protein, and the authors propose them as novel immunogenic candidates for immunization against M. tuberculosis.

To build the vaccine construct, the seven epitopes were joined with GPGPG flexible linkers, and the 194–319 amino acid C-terminal fragment of Clostridium perfringens enterotoxin was fused to the C-terminus via a rigid EAAAK linker. This fragment lacks the cytotoxicity of the full toxin but retains receptor-binding capacity toward claudins, tight-junction proteins at epithelial cell contacts, and has previously shown efficacy in inducing nasal immune responses. The rationale is mucosal delivery: because M. tuberculosis enters through the respiratory tract, the authors argue that a vaccine mimicking that route of entry may induce stronger immunity at the site of infection than intradermal BCG, which generates systemic but not mucosal responses.

The resulting 27.85 kDa construct scored 0.96 on the VaxiJen antigenicity predictor, was predicted non-allergenic, soluble, non-toxic and hydrophilic, with no similarity to the human proteome. Structural prediction identified 10 MHC class I and 30 MHC class II binding epitopes within the construct. Molecular docking of the top-scoring epitope AKDIEKEIL against the human allele HLA-B39 produced 29 hydrophobic interactions, 13 hydrogen bonds and one salt bridge, with a docking score of −106.37, while the epitope TGQKYRILASKIVDF docked against HLA-DRB1*01:01 with 30 hydrophobic interactions, five hydrogen bonds and a score of −172.29. An immune simulation on the C-ImmSim platform, modeling three doses at 14-day intervals, predicted class switching from IgM to IgG with booster doses and cytokine peaks of IL-2, interferon-gamma and TNF-alpha suggestive of a Th1-skewed response.

The authors emphasize that the study’s distinctive contribution is the use of the CPE fragment as a mucosal-targeting adjuvant in a tuberculosis vaccine design, described as a first, alongside the proposal of PE/PPE proteins and TCR-recognized immunoreactive peptides as promising vaccine components. They also acknowledge the inherent limits of the approach: variations in computational tools, algorithms and selection thresholds often produce discrepant candidate lists across studies, and the immunoreactivity of the designed construct in a living system remains unconfirmed. The predicted protective efficacy, mucosal performance and safety of the multi-epitope vaccine will therefore depend on the experimental evaluations the team plans for future work.

Subject of Research: Computational design of a multi-epitope vaccine candidate against Mycobacterium tuberculosis using reverse vaccinology and immunoreactive peptides

Article Title: Design of a multi-epitope vaccine against Mycobacterium tuberculosis using reverse vaccinology and immunoreactive peptides

Article References: Noori Goodarzi, N., Fereshteh, S., Shahbazi, B., Rezaie, N., Khazani Asforooshani, M., & Badmasti, F. (2026). Design of a multi-epitope vaccine against Mycobacterium tuberculosis using reverse vaccinology and immunoreactive peptides. Genomics & Informatics, 24(1), Article 13. https://doi.org/10.1186/s44342-026-00075-6

Image Credits: AI Generated

DOI: 10.1186/s44342-026-00075-6

Keywords: Mycobacterium tuberculosis, multi-epitope vaccine, reverse vaccinology, immunoinformatics, BCG, epitope prediction, molecular docking, PE/PPE proteins, mucosal immunity, IEDB, interferon-gamma, vaccine design