The identification of tumor-associated T cell epitopes has contributed significantly to the understanding of the interrelationship of tumor and immune system and is instrumental in the development of therapeutic vaccines for the treatment of cancer. Most of the known epitopes have been identified with prediction algorithms that compute the potential capacity of a peptide to bind to HLA class I molecules. However, naturally expressed T cell epitopes need not necessarily be strong HLA binders. To overcome this limitation of the available prediction algorithms we established a strategy for the identification of T cell epitopes that include suboptimal HLA binders. To this end, an artificial neural network was developed that predicts HLA-binding peptides in protein sequences by taking the entire sequence context into consideration rather than computing the sum of the contribution of the individual amino acids. Using this algorithm, we predicted seven HLA A*0201-restricted potential T cell epitopes from known melanoma-associated Ags that do not conform to the canonical anchor motif for this HLA molecule. All seven epitopes were validated as T cell epitopes and three as naturally processed by melanoma tumor cells. T cells for four of the new epitopes were found at elevated frequencies in the peripheral blood of melanoma patients. Modification of the peptides to the canonical sequence motifs led to improved HLA binding and to improved capacity to stimulate T cells.

Small peptides bound by Major Histocompatibility Complex (MHC) class I molecules and recognized in this context by the T-cell receptor of CD8 þ T cells are known as T-cell epitopes and are of extraordinary importance for the development of new vaccines against cancer and viral infections. Several algorithms predicting a peptides binding capability to a given MHC class I molecule are currently available and have been successfully applied in the identification of new T-cell epitopes within proteins. Most of these newly identified epitopes obey to the empirically determined anchor residue patterns that are specific for the different MHC I alleles. However, in recent studies an increasing number of weakly binding T-cell epitopes could be identified that do not fit to these canonical amino acid patterns. Therefore there is a need for new prediction algorithms improving the prediction accuracy for weakly binding epitopes that are biologically relevant as they are presented by, e.g. antigen presenting cells. Here we describe the development and application of an Artificial Neural Network (ANN)-based T-cell epitope prediction strategy for the MHC class I allele HLA-A*0201, leading to the identification of seven new melanoma-associated T-cell epitopes, that do not fit to the canonical HLA-A*0201 recognition motif. In this strategy the ANN-based epitope prediction is combined with a filter algorithm that incorporates knowledge of peptide chemistry and a ranking procedure that enables the selection of candidate epitopes according to protein family and phylogenetic conservation.