Engrafted human cells generate adaptive immune responses to Mycobacterium bovis BCG infection in humanized mice

BACKGROUND: Currently used mouse models fail to fully reflect human immunity to tuberculosis (TB), which hampers progress in research and vaccine development. Bone marrow-liver-thymus (BLT) mice, generated by engrafting human fetal liver, thymus, and hematopoietic stem cells in severely immunodeficient NOD/SCID/IL-2Rγ(-/-) (NSG) mice, have shown potential to model human immunity to infection. We engrafted HLA-A2-positive fetal tissues into NSG mice transgenically expressing human leukocyte antigen (HLA)-A2.1 (NSG-A2) to generate NSG-A2-BLT mice and characterized their human immune response to Mycobacterium bovis bacillus Calmette-Guerin (BCG) infection to assess the utility of this model for investigating human TB. RESULTS: NSG-A2-BLT mice were infected intravenously with BCG and the immune response of engrafted human immune cells was characterized. After ex vivo antigenic stimulation of splenocytes, interferon (IFN)-γ-producing cells were detected by ELISPOT from infected, but not uninfected NSG-A2-BLT mice. However, the levels of secreted IFN-γ, determined by ELISA, were not significantly elevated by antigenic stimulation. NSG-A2-BLT mice were susceptible to BCG infection as determined by higher lung bacillary load than the non-engrafted control NSG-A2 mice. BCG-infected NSG-A2-BLT mice developed lung lesions composed mostly of human macrophages and few human CD4(+) or CD8(+) T cells. The lesions did not resemble granulomas typical of human TB. CONCLUSIONS: Engrafted human immune cells in NSG-A2-BLT mice showed partial function of innate and adaptive immune systems culminating in antigen-specific T cell responses to mycobacterial infection. The lack of protection was associated with low IFN-γ levels and limited numbers of T cells recruited to the lesions. The NSG-A2-BLT mouse is capable of mounting a human immune response to M. tuberculosis in vivo but a quantitatively and possibly qualitatively enhanced effector response will be needed to improve the utility of this model for TB research.