We will develop and apply widely applicable, methods, based on advanced computational techniques, to the problems involved in elucidating the structural basis of receptor-ligand interactions. Although we expect the methods to be generally useful, the proposed project will focus specifically on early events in the activation of cytotoxic T cells, and to some extent on helper T cells. Both cell types play major roles in normal and pathological responses. The cytotoxic response is central to destroying host cells that have been infected by virus. Common to the early phases of both responses is recognition of foreign peptides (e.g., the remnants of proteins from an invading organism) by host major histocompatibility complex (MHC) products on the infected cell- called class I for the cytotoxic response and class II for the helper response. A given individual has very few MHC structures, but must be able to recognize a wide range of peptides. Developing a predictive understanding of the structural basis of recognition i.e., being able to predict which peptides will and which will not form stable complexes with a given MHC allele, is central to a wide range of fundamental and applied problems including manipulation of the immune response by vaccination or other means. Class I structures will be predicted using the known crystal structure of a human class I molecule. The method for computationally determining unknown structures from a closely homologous sequence of known structure will be validated using a structural database of highly homologous molecules. Similarly structural data from other systems is available to test the docking algorithm that will predict where and how stably a given peptide sequence binds to class I. The methods will be further validated against a large and rapidly growing database of peptides and their analogues known to bind particular effects of site specific changes in either peptide or MHC can quickly be made and evaluated. This will be done and the predictions tested experimentally.
Showing the most recent 10 out of 27 publications
