Modelado computacional del procesamiento y presentación de antígenos

Resumen

The identification of T cell epitopes is crucial for understanding diseases pathogenesis and aetiology. Moreover, it is also crucial for the development of epitope-based vaccines against infectious agents and treatments for allergic, autoimmune diseases and cancer. CD8 T cell epitopes are peptides presented on the surface of infected or damaged cells bound to MHC I molecules that are recognized by the T cell receptor (TCR). These peptides derive from foreign protein antigens that are degraded in the cytosol by the proteolitic activity of the proteasome. Some of the peptides are translocated by TAP into the endoplasmic reticulum where they can bind to nascent MHC I molecules. Subsequently, peptide loaded MHC I molecules are then displayed on cell surface for recognition by T cells. Traditionally, the identification of T cell epitopes requires the synthesis of overlapping peptides spanning the entire length of a protein, followed by experimental assays over each peptide. This method is expensive and time consuming. Therefore, it is key to develop alternative computational approaches for the prediction of T cell epitopes to decrease the experimental burden associated with epitope identification. In this Thesis, we have modeled the classical processing pathway of MHC I antigens. We have analyzed the location of 190, 249 and 78 CD8 T cell epitopes of Hepatitis C Virus, Human Immunodeficiency Virus and Influenza A Virus, respectively, in the viral proteins. We found that capsid and matrix proteins encompass, significantly, more epitopes than the expected by their size. We have also modeled the specificity of the cleavage site of the proteasome, using N-grams. We have developed two different models for the proteasome and the immunoproteasome from two distinct sets of MHC I-restricted peptides. The proteasome model was developed using a sets of peptides eluted from human MHC I molecules, whereas the immunoproteasome model was trained using CD8 T cell epitopes naturally processed. In addition, we have also studied the peptide affinity to TAP using support vector machines trained on single residue positions and residue combinations drawn from a large dataset consisting of 613 nonamer peptides of known affinity to TAP. Finally, we have developed two different web tools that are instrumental for epitope selection: PVS and TEPIDAS. PVS (Protein Variability Server) is a useful tool for the identification of conserved T and B cell epitopes through the sequence variability analysis. This sever estimates the variability using different methods, like the Shannon entropy, the Simpson diversity index and the Wu-Kabat variability coefficient. PVS can also plot the variability in the MSA and display it in a relevant 3D-structure. TEPIDAS is a DAS Annotation Server that includes CD8 T cell epitopes specific of human pathogenic organisms, the MHC I restriction elements (experimentally determined or predicted) and the associated cumulative phenotypic frequency.