The long-term survival of patients with acute myeloid leukemia (AML) is dismally poor. A permanent cure of AML requires elimination of leukemic stem cells (LSCs), the only cell type capable of initiating and maintaining the leukemic clonal hierarchy. We report a therapeutic approach using an activating monoclonal antibody directed to the adhesion molecule CD44. In vivo administration of this antibody to nonobese diabetic-severe combined immune-deficient mice transplanted with human AML markedly reduced leukemic repopulation. Absence of leukemia in serially transplanted mice demonstrated that AML LSCs are directly targeted. Mechanisms underlying this eradication included interference with transport to stem cell-supportive microenvironmental niches and alteration of AML-LSC fate, identifying CD44 as a key regulator of AML LSCs. The finding that AML LSCs require interaction with a niche to maintain their stem cell properties provides a therapeutic strategy to eliminate quiescent AML LSCs and may be applicable to other types of cancer stem cells.

Cancer stem cells (CSCs) represent a population of cancer cells that possess unique self-renewal and differentiation characteristics required for tumorigenesis and are resistant to chemotherapy-induced apoptosis. Lung CSCs can be enriched by several markers including drug-resistant side population (SP), CD133pos and ALDHhigh. Using human non-small cell lung adenocarcinoma cell lines and patient-derived primary tumor cells, we demonstrate that SP cells represent a subpopulation distinct from other cancer stem/progenitor cell (CS/PC) populations marked by CD133pos or ALDHhigh. The non-CS/PCs and CS/PCs of each subpopulation are interconvertible. Epithelial-mesenchymal transition (EMT) promotes the formation of CD133pos and ALDHhigh CS/PC subpopulations while suppressing the SP CS/PC subpopulation. Rac1 GTPase activity is significantly increased in cells that have undergone EMT, and targeting Rac1 is effective in inhibiting the dynamic conversion of non-CS/PCs to CS/PCs, as well as the CS/PC activity. These results imply that various subpopulations of CS/PCs and non-CS/PCs may achieve a stochastic equilibrium in a defined microenvironment, and eliminating multiple subpopulations of CS/PCs and effectively blocking non-CS/PC to CS/PC transition, by an approach such as targeting Rac1, can be a more effective therapy.