Pancreas cancer is the fifth leading cause of cancer death in the United States, and arguably the most lethal of human malignancies. Of the 29,000 patients diagnosed with the disease in 1998, all but 300 are expected to die within five years of diagnosis. Efforts to develop more effective therapies have been hampered by the lack of a pancreas cancer animal model where the contribution of individual genetic lesions to the neoplastic phenotype can be explored and the sensitivity to various treatment strategies can be assessed. We are developing such a model both to facilitate preclinical trials and as an engine of gene discovery. Our initial model is theTGFa/ink4a compound mutant mouse, containing mutations in two major regulatory pathways implicated in the development of nearly all human pancreatic cancers. Preliminary observations indicate that the two mutations cooperate to generate a novel, neoplastic pancreatic ductal phenotype. This phenotype will be fully characterized, and pancreatic cell lines established from this and related compound mutants screened for novel mutations that may play a pivotal role in the development of this disease. In addition, we will use transgenic technology to facilitate the selection, propagation, and transformation in culture of highly purified populations of pancreatic duct cells, the cell type of origin of most human pancreatic cancers.