s: Session II 5 8 tion and physical isolation of such stages. All islets express the oncoprotein T antigen, but with decreasing frequency in hyperplastic islets (50–70%), angiogenic islets (10%) and end-stage tumors (2-4%). A long-term goal of our studies is to identify the genetic and epigenetic alterations that are required at each of these steps en route to tumor formation. Two genetic alterations have been identified in this model in a genome-wide screen for regions of loss of heterozygosity (LOH). One resides on chromosome 16 (LOH16) and is lost in the transition between hyperplastic islet and angiogenic islet. The other is on chromosome 9 (LOH9) and is lost as angiogenic islets develop into tumors. To identify the tumor suppressor genes contained within LOH9 and LOH16, we are mapping the fine structure of the two regions by analysis of simple sequence length polymorphisms and comparative genomic hybridization with BAC-based arrays. We have developed and validated the use of murine BACs spotted on glass slides as a sensitive platform for comparative genome hybridization, with which we can readily detect loss of LOH9 and LOH16 in islet tumors. To identify genes whose altered expression may be critical to tumor development we have also profiled expression of approximately 22,000 genes and expressed sequence tags on microarrays using messenger RNAs derived from distinct RIPTag tumor stages. We will present patterns of gene expression and potential candidates for functional studies.