Application Of PLS Method In The T Cell Epitopes Prediction

T cell epitopes could activate immune response of T cell, and play an important role in specific immune response. The production of T cell epitopes depends on the antigen processing and presentation pathway. There are some primary steps in process of producting the cytotoxic T lymphocytes (CTL) epitopes, i.e. endogenous antigen proteins need to be degradated by proteasome into endogenous peptides, peptides need to be transported by transporter associated with antigen processing (TAP) into endoplasmic reticulum (ER), the peptides need to bind with major histocompatibility complex (MHC) class I molecule, and peptide-MHC class I complexes need to be transported to the surface of antigen presenting cells (APC), recognized by T cell receptor (TCR) and activate CTL. There are some primary processes in production of the helper T cell (Th) epitopes, i.e. exogenous antigen proteins need to be degradated by lysosomal enzyme into exogenous peptides, peptides need to bind with MHC class II molecule, and peptide-MHC class II complexex need to be transported to the surface of APC and recognized by TCR and activate Th. Therefore, in the process of producting T cell epitope, the antigen processing and presentation pathway plays a key role. In order to further study the biological mechanism of the antigen processing and presentation, quantitative structure activity relationships (QSAR) method was used to theoretically study three important steps in the antigen processing and presentation pathway.1. In the endogenous antigen processing and presentation pathway, peptide-MHC class I complex plays a critical role in T cell activation. Only certain peptides could bind to MHCclass I molecule. Therefore, determining which peptides could bind to MHC class I molecule and predicting the affinity of peptide binding to MHC class I molecule accurately should be helpful to understand the mechanism of peptide binding MHC class I molecule and to show the mechanism of producting T cell epitope. In order to further study the specificity of MHC class I molecule binding antigen peptide, Three QSAR models of MHC class I ligand are built by PLS method. Comparison the prediction performance of the three model, it is known that the interactions between amino acid residues need to be considered when the QSAR model of MHC class I ligand is built. In addition, the specificities of MHC class I molecule binding antigen peptide were obtained by analysis the weight coefficients of QSAR model.2. The ubiquitin-proteasome system of the eukaryote plays an importance role in the process of endogenous antigen degradation. The QSAR model of proteasomal cleaving antigen protein is built in order to study the specificity of the proteasome cleaving antigen protein, and PLS method is used to solve the model. And the predictive accuracy of the model is 82.8%. Comparison with other models in the same test set, this model has the superior predictive performance. It is known that there are obvious specificities in the cleavage site and its adjacent positions. The result indicates that the proteasome cleaves the antigen protein selectively, but not randomly.3. In the exogenous antigen processing and presentation pathway, peptide-MHC classII complex plays an important role in activating immune response of helper T cell. Only certain peptides could bind to MHC class II molecule. In order to study the binding specificities between peptide and MHC class II molecule, four QSAR models of MHC class II ligand based on 9-mer core binding sequence and four QSAR models of MHC class II ligand based on 13-mer extended core binding sequence are built by CV-ISC-PLS. Through comparion among the predictive performance of models, it is known that the adjacent positons of core binding sequence need to be considered and suitable amino aicd descriptor is needed to be applied when the model is built. In addition, HLA DRBl*0101 is taken for example, the specificities of MHC class II molecule binding antigen peptide were obtained by analysis the weight coefficients of model. And the specificities agree with the experiment results.