Targeting HIF function: the debate continues

Paresh Vyas

In this issue of Blood, Velasco-Hernandez et al1 come to an important and at first sight, unexpected, conclusion that hypoxia inducible factor 1α (HIF-1α) may be a tumor suppressor gene.

Current data suggest leukemic stem cells (LSCs) are adapted to hypoxia, raising the possibility of therapeutic targeting of the transcriptional regulator HIF-1α. Direct oxygen measurement of human marrow shows it has a low overall oxygen partial pressure (∼55 mmHg, with an oxygen saturation of 87.5%). Furthermore, perfusion tracer experiments suggest that functional murine hematopoietic stem cells (HSC) are enriched in the lowest perfusion compartment.2 Recent elegant data confirmed what was long suspected: that there is regional variation in vascularization. The central murine marrow diaphysis is poorly vascularized and enriched for cells staining with pimonidazole, a chemical that makes thiol adducts in a low oxygen environment.3 This, coupled with observations that murine HSCs and human LSCs reside next to the endosteum,4,5 support the concept that HSCs and LSCs reside in a particularly hypoxic niche.

The heterodimeric transcriptional regulator HIF, which is composed of α and β subunits, mediates, in large part, adaption to hypoxia. There are 3 different α subunits, HIF-1α, HIF-2α, and HIF-3α, and a common HIF-1β subunit. HIF regulates the expression of many genes, which together allow cells to adapt to hypoxia and a low nutrient environment by switching energy production from oxidative to glycolytic pathways and reducing reactive oxygen species production that could have a deleterious mutagenic impact on the genome and promoting quiescence.2 Although all 3 HIF α subunits are expressed in murine HSCs, genetic manipulation of HIF-1α protein levels demonstrates that HIF-1α promotes murine HSC engraftment and quiescence when tested in transplantation assays and in aged mice.6 In contrast, HIF-2α does not appear to have a similar or additive role in HSCs.7

These observations, coupled with a large amount of literature on HIF function in cancers more generally, have promoted the concept of therapeutic targeting of HIF function. Credence for the notion that HIF-1α may be a therapeutic target in hematologic malignancies has come from in vitro and limited in vivo studies of primary human acute myeloid leukemia (AML) samples, in an experimental murine lymphoma model using the HIF-1α inhibitor echinomycin,8 and in an experimental murine chronic myeloid leukemia model in an HIF-1α−/− background.9 However, there are also data to support a more important role for HIF-2α, rather than HIF-1α, from knockdown studies that show reduced engraftment when 7 human AML samples were tested.10 Knockdown of HIF-2α also reduced short- and long-term engraftment of primary human CD34+ stem/progenitor cells, suggesting more work needs to be done to establish whether there is a therapeutic index for targeting HIF function.

Now add into the mix the paper from Cammenga’s laboratory, which asks the following question: is HIF-1α required for leukemic growth in 3 different murine AML models. They studied 3 AML models: a tetracycline inducible human mixed-lineage leukemia (MLL)–eleven nineteen leukemia (ENL) murine knock-in (KI) model and 2 transplantation models where mice were transplanted with bone marrow cells retrovirally transduced with either Hoxa9-Meis1 or Aml1-Eto9a. The choice of models was interesting as both MLL-ENL KI and Hoxa9-Meis1 are likely to directly activate HIF-1α through MEIS1 expression, whereas Aml1-Eto9a is not known to signal through HIF-1α. They studied the oncogenes in wild-type cells with Hif-1α alleles or where Hif-1α alleles could be conditionally deleted after engraftment. In all 3 models, the results were clear; there was no dependence on HIF-1α for leukemia initiation, propagation, and leukemia-initiating cell self-renewal in transplantation assays. If anything, the onset of leukemia was accelerated in cells deleted for Hif-1α in the Hoxa9-Meis1 model and when mice were secondarily transplanted with Aml1-Eto9a transduced leukemic cells. One obvious caveat is that compensation by HIF-2α may have obscured a physiologic role for HIF-1α. Although that may be the case, the data do suggest that simply targeting HIF-1α may not be sufficient. Studying AML initiation and propagation in cells with both Hif-1α and Hif-2α conditional alleles would address this question.

So where does this leave the field? Although there are still important mechanistic questions about the role of HIF and adaptation to hypoxia by normal stem/early progenitor cells, the bulk of evidence supports a critical role for HIF function in this area. Clearly, more work needs to be done to define any differential functional effects of HIF-1α and HIF-2α between humans and mice. In AML and other hematologic malignancies, the situation is likely to be more complex. The role of HIF (and specifically HIF-α subunits) may depend on a number of parameters. For example, the nature of oncogenic drivers (genetic and epigenetic) is likely to dictate genome integrity and genome robustness. One could hypothesize that loss of HIF function in some malignancies (and AML in particular) may make tumor initiating and propagating cells more vulnerable to genotoxic stress just like their normal hemopoietic stem/early progenitor counterparts, whereas this may not be true in cells with an altered TP53 function. Oncogenic drivers are also likely to influence self-renewal, the need (or lack of) for quiescence, and optimal metabolism for leukemia initiating and propagating cells. Taken together, this is likely to determine the nature of optimal niches and thus the requirement for HIF function. If these hypotheses are shown to be correct, it would also suggest that HIF requirement will not only vary between patients, but also within a patient at different stages of the disease. Thus, the data from Valasco-Hernandez et al should give pause for more thought and an opportunity to probe more deeply into the interaction between hypoxic adaption and function of cell populations that initiate and propagate AML and other cancers.


  • Conflict-of-interest disclosure: The author declares no competing financial interests.