This competitive renewal proposes to continue our efforts characterizing acute myelogenous leukemia (AML) with studies that build on the significant findings that we achieved in the current funding period in both mouse models and human leukemia. In 1998 we showed that mouse AML was a multistage process involving clonal expansion of cells that resist programmed cell death. In 2000 we found that only multipotent progenitor (MPP) stage cells in human AML with AML1-ETO translocations are leukemia cells; more importantly, in the same patients, the translocation is present in normal self-renewing blood stem cells (HSC). We proposed that pre-leukemic stages of development can only occur in self-renewing clones of HSC, and that some clones progress to leukemia stem cells (LSC) at the stage of MPP. These LSC are the only cells truly responsible for driving the disease. The MPP, but not precursor HSC or progeny blast cells, transfer the leukemia to immunodeficient mice. Recent evidence has revealed cancer stem cells (CSC) to be more resistant to chemo- and radiotherapy that target their progeny proliferating cells. Thus, tumors often shrink in response to chemotherapy, but almost universally recur due to resistance of the cancer stem cells. A key implication and requirement of the cancer stem cell model is that only therapies that kill LSC/CSC can cure the patient. Thus, in order to develop CSC-targeted therapies, it is necessary to identify molecules specific to the LSC and dysregulated molecules and pathways that are critical for their pathogenesis. We propose here 3 aims designed to increase our understanding of human AML LSC in order to therapeutically target them. First, in light of new reports and our own findings, we will conduct a rigorous identification and quantification of LSC from patient samples of de novo AML, as well as investigate the normal counterparts to these LSC. Second, we will investigate the functional role of CD96 in the progression and pathogenesis of human AML, and investigate potential methods of targeting CD96 as a therapeutic strategy. During the current funding period, CD96 was identified as a candidate LSC-specific marker in two independent studies. Finally, we will examine whether CD47 expression (a 'don't eat me' signal) on leukemic cells renders them invisible to the innate (macrophage) and adaptive (dendritic cell presentation of tumor antigens) immune systems by examining the ability of various subsets of macrophages and dendritic cells to phagocytose leukemic cells and present antigen to the adaptive immune system. We will test whether a blocking anti-CD47 antibody can overcome this effect and render them susceptible to immune clearance.