Interactions of Mycobacterium tuberculosis with macrophages have long been recognized to be crucial to the pathogenesis of tuberculosis. The role of non-phagocytic cells is less well known. We have discovered a M. tuberculosis surface protein that interacts specifically with non-phagocytic cells, expresses hemagglutination activity and binds to sulfated glycoconjugates. It is therefore called heparin-binding hemagglutinin (HBHA). HBHA-deficient M. tuberculosis mutant strains are significantly impaired in their ability to disseminate from the lungs to other tissues, suggesting that the interaction with non-phagocytic cells, such as pulmonary epithelial cells, may play an important role in the extrapulmonary dissemination of the tubercle bacillus, one of the key steps that may lead to latency. Latently infected human individuals mount a strong T cell response to HBHA, whereas patients with active disease do not, suggesting that HBHA is a good marker for the immunodiagnosis of latent tuberculosis, and that HBHA-specific Th1 responses may contribute to protective immunity against active tuberculosis. Strong HBHA-mediated immuno-protection was shown in mouse challenge models. HBHA is a methylated protein and its antigenicity in latently infected subjects, as well as its protective immunogenicity strongly depends on the methylation pattern of HBHA. In both mice and man, the HBHA-specific IFN-gamma was produced by both the CD4(+) and the CD8(+) T cells. Furthermore, the HBHA-specific CD8(+) T cells expressed bactericidal and cytotoxic activities to mycobacteria-infected macrophages. This latter activity is most likely perforin mediated. Together, these observations strongly support the potential of methylated HBHA as an important component in future, acellular vaccines against tuberculosis.