Antigen Structure-Based Supervised Learning for CD4+ T-cell Epitope Prediction CD4+ T cells provide numerous protective functions as part of the adaptive immune response, including cytokine-mediated and contact-mediated signals to B cells, CD8+ T cells, and innate- immune cells, as well as direct modes of attack on pathogenic agents. Nevertheless, their most critical roles in defense against intracellular bacterial pathogens are especially poorly understood. A major barrier to both study and vaccine design has been the lack of epitope- specific reagents for counting and tracking CD4+ T cells. We propose to develop a novel and well-validated algorithm for CD4+ T-cell epitope prediction that will enable progress in one of the most promising, yet presently hindered fields of immunology and vaccinology. CD4+ T-cell epitope dominance has been much less predictable than CD8+ T-cell epitope dominance because the class II MHC antigen-presenting protein is less selective and because proteolytic antigen processing has a major influence on the availability of peptide ligands. In the endocytic compartments, proteases act on mostly natively folded antigens, whose 3D structure directs proteolysis to the flexibly disordered segments. Thus, potential MHC-binding sequences in the flexible segments are destroyed, and sequences in the stable antigen segments are preferentially loaded and presented to CD4+ T cells. We will incorporate this bias toward epitope dominance in the stable antigen segments into a computational tool that significantly improves upon existing sequence-based methods for epitope prediction. We will develop our stability-based method using several possible supervised learning techniques, including hidden Markov models and position-specific scoring matrix methods. For soluble antigens, our feature set will be conformational stability data including crystallographic b-factor, surface-accessibility, COREX residue stabilities, and sequence entropy. In order to validate our method, we will use it to predict novel epitopes for 5 soluble secreted antigens from Salmonella typhimurium and Burkholderia pseudomallei, organisms for which CD4+ T-cell immunity is essential. Peptides corresponding to the 80th- percentile of predicted-epitopes will be tested for responses in individual C57BL/6 mice, following two types of exposure to the intact antigens, bacterial infection and subunit vaccination.
CD4+ T cells constitute a major arm of the immune system, recognizing foreign antigens only when small pieces (epitopes) of the antigens are presented to them by other cells. Knowing what epitopes are going to be presented is key to understanding how the immune system works and may tell us how to design vaccines. Existing methods for predicting computational epitope prediction have not considered by the machinery that breaks the antigen into epitopes. We propose to use machine learning techniques to model how antigens are processed using experimental immune response data. The primary outcome of this project is an improved method for CD4+ T cell epitope prediction that can be combined with existing tools.
| Landry, Samuel J; Moss, Daniel L; Cui, Da et al. (2017) Structural Basis for CD4+ T Cell Epitope Dominance in Arbo-Flavivirus Envelope Proteins: A Meta-Analysis. Viral Immunol 30:479-489 |
| Mettu, Ramgopal R; Charles, Tysheena; Landry, Samuel J (2016) CD4+ T-cell epitope prediction using antigen processing constraints. J Immunol Methods 432:72-81 |
