Multi-Scale Modeling of Antigen Presentation With Applications to Tuberculosis.

Abstract

Antigen presentation is the process by which cells of the immune system display peptides from pathogens on their surface after binding the peptides to major histocompatibility complex (MHC) molecules. T helper cells recognize peptides from pathogens in this context then secrete cytokines that activate other cells, initiating an immune response. Antigen presentation is therefore a requisite for immunity to several pathogens including Mycobacterium tuberculosis (Mtb). To approach questions related to antigen presentation and disease, I represented antigen presentation at different scales using a series of mathematical and statistical models. At the molecular scale, I asked whether heterogeneity in peptide length affects binding to MHC class II, the class of MHC responsible for binding peptides from bacteria such as Mtb. By developing statistical models of peptide-MHC binding, I found that length has a nonlinear effect on binding affinity and that this information, or a more accurate representation of register shifting, could improve the accuracy of binding prediction. At the cellular scale, I asked why Mtb possesses multiple mechanisms to inhibit antigen presentation on the cell surface. My mathematical model shows that these mechanisms may be acting on different timescales and therefore complementary rather than merely redundant. Finally, at the multi-cellular level, I asked how polymorphisms in multiple genes related to antigen presentation might affect T cell response and susceptibility to infectious diseases such as tuberculosis. Using a multi-scale model representing both an antigen-presenting cell and T cell, I found that polymorphisms in two different genes may exert the same influence on the output, potentially canceling out their effects. Future work with these models may include evaluation of candidate peptide-based vaccines to ensure high-affinity binding, T cell response, and broad efficacy in diverse populations.