Acute myeloid leukemia (AML) is the most common form of acute leukemia in adults, with AML deaths in the US reaching nearly 11,000 in 2014. Unfortunately, the clinical outcome of AML is poor, with the majority of patients experiencing relapse within three years. Frequent occurrence of comorbidities precludes treatment with a standard chemotherapy regimen, and the field continues to seek novel therapies that may target leukemic stem cells (LSCs), felt to be the root cause of relapse and refractory disease. However, ideal targets for drug therapies remain elusive.

Hypoxia-inducible factor-1α (HIF-1α) is a subunit of a heterodimeric transcription factor containing a basic helix-loop-helix domain and represents the key transcriptional regulator of cellular and developmental response to low oxygen level (hypoxia). AML is initiated within the bone marrow (BM), likely under local hypoxic conditions. Recently, Dominique Bonnet and colleagues reported that shRNA-based down-regulation of mediators of cellular responses to hypoxia, such as HIF-1α or HIF-2α, induces apoptosis and prevents leukemic engraftment upon transplantation into mice (the functional property that defines LSCs). This observation suggested that HIF-1α or HIF-2α is required for the maintenance of LSCs and may represent potential therapeutic targets for AML. In contrast, the group of Jörg Cammenga has recently shown that genetic deletion of Hif-1α has no effect on mouse AML maintenance, introducing a contradiction in the role of HIFs in AML disease. A complete understanding of the role of HIFs in AML, however, is lacking and calls for sophisticated genetic approaches.

In this issue, Vukovic et al. develop a conditional genetic model to examine the impact of deletion of Hif-2α or both Hif-1α and Hif-2α at different stages of Meis1/Hoxa9- or MLL-AF9–driven leukemogenesis. The authors reveal that Hif-2α suppresses the development of LSCs but has no impact on AML propagation in a Meis1/Hoxa9-induced murine AML model. Hif-2α deletion accelerates LSC development but does not affect LSC maintenance of Mll-AF9–driven AML. The lack of requirement for Hif-2α in AML propagation was surprising; thus, the Kranc group used the recently available CRISPR-Cas9–mediated genome editing approach to determine the potential impact of HIF-2α ablation in human leukemic cells. HIF-2α ablation had no effect on human AML lines, including on cell cycle regulation. In the context of therapeutic targeting, they used BAY 87–2243, a compound able to impair HIF-1α and HIF-2α protein accumulation under hypoxic conditions, to show that inhibition of the HIF pathway has no effect on human AML cell survival or proliferation.

This study represents the first genetic evidence that Hif-2α may act as a tumor suppressor in AML development and/or initiation but is dispensable for LSC-based disease maintenance. Consistent with this idea, the authors show that Hif-1α and Hif-2α deletion promoted a gene expression signature that facilitated survival and proliferation of preleukemic cells and suggest that this is responsible for the selective effects of HIF pathway on leukemic initiation and not on AML established by LSC trasnplantation. Accordingly, this study questions the use of targeting HIFs in AML after critical events of leukemic initiation have occurred.

Before neglecting molecular or pharmacological targeting of HIF in AML, other caveats to these observations need to be explored. Like most drugs, the compound inhibitor used in this study, BAY 87–2243, has many modes of action, including the ability to affect mitochondrial functions. The importance of mitochondrial capacity and AML has been previously demonstrated by the Schimmer group, but it remains to be examined in a conditional deletion model similar to the one used by Vukovic et al. This may provide insights on the role of the HIF pathway and the metabolome of AML LSCs that have yet to be uncovered. Furthermore, as AML is a genetically heterogeneous cancer and is characterized by several molecular alterations, the current studies, like many recent reports for AML disease in vivo, are restricted to the Mll-AF9–Meis1/HoxA9–driven leukemia that represents only a fraction of human AMLs. Before any conclusions can be drawn, the role of HIFs and their pharmacological inhibition will need to be tested in other clinical models that capture the heterogeneity of patients and their response to chemotherapy. Mouse xenografts of human AML cells may serve as an experimental alternative to test the role of the HIF pathway in leukemia-initiating cells (LICs), using in vitro manipulation of the pathway to quantitatively determine the effects on frequency of transplanted human LICs from a diverse set of AML patients. Because HIFs are known to regulate SDF-1 in the BM niche, it would also be interesting to target this pathway in mice with established human AML and investigate the effect of the tumor microenvironment on LSC function in vivo.

Until then, distinguishing the effects on leukemia initiation versus maintenance by targeting the HIF pathway and demonstrating the selectivity for “preleukemic cells,” as opposed to normal HSCs, is where the low-hanging oxygen may lie in the treatment or prevention of AML.


, et al
J. Exp. Med.