Cancer cells have been shown to exhibit "markers," known as tumor antigens, which flag cancer cells for recognition by the immune system. Tumor antigens consist in small peptides of 8 to 10 amino acids, presented at the cell surface by class I molecules of the Major Histocompatibility Complex (MHC). These peptides result from the breakdown of cellular proteins by a large protease called the proteasome. At the cell surface, tumor antigens are recognized by a specific population of white blood cells called the cytolytic T lymphocytes, which can recognize and kill tumor cells. Novel anti-cancer therapeutic approaches, known as « cancer immunotherapy » aim at activating these T cells so that they can help patients destroy their tumor. However, many patients still fail to respond to such therapies, likely because they lack an appropriate anti-tumor response. There is therefore an urgent need to develop cancer vaccines or adoptive T-cell therapies efficiently targeting tumor antigens, in order to improve anti-tumor responses in patients. Our work aims at identifying such antigens, which would be optimal targets for cancer immunotherapy. To do so, we study how tumor antigens are produced within tumors and in particular, how the different proteasome subtypes influence the nature of the antigenic peptide repertoire. Although loss of the components of the MHC class I processing pathway often affects the antigenic peptide repertoire and generally favors cancer immune evasion, it sometimes also unveil new tumor antigens, which we are actively trying to characterize, as they might represent interesting targets for cancer immunotherapy.