Background: Tuberculosis (TB) remains a major global health challenge, highlighting the urgent need for more effective vaccines. This study aimed to develop an artificial intelligence-guided epitope prediction and prioritization pipeline to identify immunodominant peptides from Mycobacterium tuberculosis (Mtb) and to evaluate the immunogenicity and protective efficacy of the resulting vaccine candidates. Methods: An AI-guided framework was used to predict and prioritize immunodominant Mtb epitopes, leading to the generation of 72 recombinant immunogens. Among these, RI-13, RI-20, and RI-31 were selected as the leading candidates. Their protective efficacy was assessed in a zebrafish TB infection model and in BALB/c mice following DNA vaccination. Humoral and cellular immune responses were further evaluated in C57BL/6 mice. Results: RI-13, RI-20, and RI-31 markedly reduced infection-associated pathology and lowered bacterial burden by up to 1.5 log10 in the zebrafish TB infection model, outperforming benchmark antigen combinations, including the Ag85A plus ESAT6/CFP10 cocktail used in the phase III vaccine candidate GamTBvac. In BALB/c mice, DNA vaccination with each construct reduced pulmonary mycobacterial burden by approximately 0.3 log10 and alleviated lung tissue damage. In addition, all three candidates elicited robust humoral and cellular immune responses, with RI-13 showing the strongest overall immunogenicity and inducing a balanced Th1, Th2, and Th17 response profile in C57BL/6 mice. Conclusions: These findings identify RI-13, derived from Rv1174c, as a promising next-generation TB vaccine candidate. More broadly, this study supports the utility of an AI-guided framework for the rational design and preclinical prioritization of novel TB immunogens.
Fang et al. (Fri,) studied this question.