Inhibiting Hypoxia-Inducible Factor 1 for Cancer Therapy

• HIF-1
• Hypoxia
• molecular target
• tumor microenvironment
• cancer therapy

Over the last few years, a better appreciation of the role that the tumor microenvironment plays in tumor progression has been associated with a paradigm shift in how to best develop therapeutic strategies for cancer treatment (1). As a consequence, renewed emphasis has also been placed on key features of the tumor microenvironment such as hypoxia. Indeed, it has long been recognized that hypoxia contributes to selecting cancer cells resistant to apoptosis and mediates resistance to chemotherapy and radiation therapy (2, 3). Nevertheless, it has not been until the discovery of the transcription factor hypoxia-inducible factor 1 (HIF-1) that a “molecular target” underlying the presence of hypoxia in solid tumors has been identified and proposed for development of cancer therapeutics (4-6).

HIF-1 is a master regulator of the transcriptional response of mammalian cells to oxygen deprivation. Since its discovery in the early 1990s, HIF-1 has rapidly attracted interest both for its involvement in fundamental biological processes, including but not limited to tumor metabolism, angiogenesis, metastasis, and inflammation, and for its potential role as therapeutic target (4-6). However, the complexity of the role that HIF-1 may play in cancer biology, its dual role in promoting cell survival and inducing apoptosis, and the presence of two HIF-α subunits (HIF-1α and HIF-2α), which may have unique and specific roles in different tumor types, have contributed to generating some early controversy about the role of HIF-1 as therapeutic target. Whether or not HIF-1 is a good target for development of cancer therapeutics has been addressed in a number of experimental models, which have led, not unexpectedly, to somewhat controversial results. Indeed, different results have been reported depending on the cell type used (7-9), the HIF-α subunit targeted (10-12), the site of tumor injection (e.g., s.c. versus orthotopic; refs. 13, 14), and the timing of HIF-1α inhibition, more active in early rather than late tumor growth (15). Because HIF-1 has not been directly implicated in oncogenic transformation, although evidence has been provided that it can cooperate with AKT in melanomagenesis (16), it has been difficult to translate results from in vitro studies to animal models. Such controversial results reflect, at least in part, the complexity of the involvement of HIF-1 in human cancers and the known limitations of experimental animal models. At least two conclusions can be drawn from these studies. The first is that the contribution of HIF-1 to tumor progression is highly cell type and context dependent, affected by the genetic complexity of the cell lines studied and influenced by the stromal component, which is neither properly represented in the xenograft models nor affected by genetic manipulation of HIF in cancer cells. The second conclusion is that manipulation of HIF-1 activity outside of the appropriate genetic context may actually generate misleading conclusions, as clonal expansion of HIF overexpressing cells, which occurs in human cancers, requires the temporal selection of specific genetic alterations and conducive environmental conditions.

How then can results of preclinical studies be translated to human cancer? In contrast to the results obtained in animal models, the studies conducted in cancer patients have been overall more consistent with a positive role of HIF in tumor progression. In fact, overwhelming evidence indicates that HIF-α is indeed overexpressed in the majority of human cancers (17-19), where it is associated with patient mortality and poor response to treatment (20-24), findings that can be hardly reconciled with a negative role of HIF-1 in tumor growth. Importantly, the association of several oncogenic and receptor tyrosine kinase–dependent signaling pathways, frequently deregulated in human cancers, with HIF-1α activation also suggests that many of these pathways converge on or implicate HIF-1 in mediating cell survival and growth (25-28). However, staining of HIF-1α in human cancers is variable, from few positive cells up to the majority of tumor tissue, and characterized by distinct patterns of expression, of which the implications and biological consequences are still poorly understood. HIF-1α expression in solid tumors has been detected in hypoxic perinecrotic areas, in stromal inflammatory cells, as well as in patterns unrelated to tissue oxygen concentrations, most likely dependent on genetic alterations and dysregulation of growth factors signaling (17-19). Then, rather than asking the question about whether or not HIF-1 should be inhibited, we ought to address the issue of how to do it. Most of the efforts to identify HIF inhibitors have converged on the discovery of small molecules, both in academic centers and pharmaceutical companies, despite the skepticism associated with targeting transcription factors. The identification of selective HIF-1 inhibitors would not only be useful for the potential therapeutic implications but also for their application as analytic tools to further define the role of HIF in human cancers. Earlier efforts in this area have been associated with hope that universal and specific inhibitors of HIF might be identified. Perhaps not unexpectedly, only a few examples of HIF inhibitors that potentially target selective pathways associated with HIF activation have been described, among which echinomycin (29) and synthetic polyamides (30) that inhibit HIF-1 DNA binding, and chetomin (31), which blocks recruitment of coactivator p300/CBP (Fig. 1 ). In addition, ongoing efforts in targeting dimerization of HIF-α with HIF-1β might lead to unanticipated results (32, 33), given the recent apparent success in the identification of small-molecule inhibitors of protein-protein interaction. However, none of these agents has been thus far proposed for clinical development. In contrast, a growing list of nonselective HIF-1 inhibitors has been generated over the last few years (34, 35). In fact, most of the described HIF-1 inhibitors do so by altering signal transduction pathways that are indirectly associated with HIF or that are part of more complex pathways relevant to human cancer, clearly limiting the specificity of their activity. Small-molecule inhibitors of HIF have been identified that target chaperone proteins (36, 37), microtubules (38), topoisomerase I (39), soluble guanylate-cyclase (40), and thioredoxin (41), as well as a number of signaling pathways that have been associated with HIF-1α activation including mammalian target of rapamycin (26, 28), AKT (42), Her2/Neu (43), epidermal growth factor receptor (27, 44), farnesyltransferase (45), Bcr-Abl (46), and histone deacetylase (ref. 47; Fig. 1). On the one hand, the involvement of HIF-1α in such a broad range of signaling pathways emphasizes its potential role as downstream “effector” molecule on which multiple pathways may converge; on the other hand, it raises the question about how to determine the contribution of HIF-1 inhibition to tumor growth in each individual case.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Possible targets of small-molecule inhibitors of HIF-1.

Then, how should nonselective HIF inhibitors be developed for treatment of human cancers? The fact that most HIF inhibitors described thus far are not selective does not necessarily rule out that these agents might be used in the clinic to target HIF. In fact, several of these agents, including 17-N-allylamino-17-demethoxygeldanamycin, 2-methoxyestradiol, and small molecules targeting receptor tyrosine kinase–dependent signaling pathways, are currently in clinical development for cancer therapy (48). Although for some of these agents (e.g., receptor tyrosine kinase inhibitors) HIF may not become the primary target of their clinical application, the recurrent implication of HIF-1 in receptor tyrosine kinase–dependent signaling pathways would suggest that HIF-1 inhibition might become a “biomarker” of biological activity. However, because most of the HIF inhibitors described affect multiple signaling pathways and/or targets indirectly associated with HIF, assessment of their activity as HIF inhibitors cannot be based on therapeutic efficacy, which might be unrelated to HIF inhibition. Instead, the preclinical development of HIF inhibitors should be guided by evidence in animal models that the target is modulated and that this inhibition is associated with meaningful biological consequences. Several approaches have been used in this regard, ranging from noninvasive imaging of hypoxia response element–dependent expression of bioluminescence (49) to dynamic contrast imaging techniques (50), to assessment of tissue molecular end points relevant to the HIF pathway (e.g., HIF-1α expression assessed by immunohistochenistry, HIF-target genes; refs. 49, 51). The identification and validation of pharmacodynamic end points associated with and relevant to HIF inhibition is also essential for early clinical development of HIF inhibitors, where tumor shrinkage alone cannot be expected to be a reliable determinant of activity on the HIF pathway. Early clinical trials of HIF inhibitors should then be designed to achieve (a) evidence of HIF inhibition in cancer tissue and (b) protracted schedules of administration in the absence of undesired or unmanageable adverse effects. Unlike cytotoxic agents that are given for short courses at the maximum tolerated dose to capitalize on killing of cancer cells, HIF-targeted therapeutics should be aimed at achieving sustained HIF inhibition. This conclusion, consistent with the current use of signaling targeting agents in cancer therapy, is also suggested by recent studies in which protracted, but not intermittent, administration of topotecan, a topoisomerase I poison that inhibits HIF-1, was associated with inhibition of tumor angiogenesis and cancer growth in xenograft models (49). A number of questions that have yet to be convincingly answered in preclinical models hamper the current development of novel HIF inhibitors. Is a HIF inhibitor expected to be active as single agent? Which are meaningful downstream end points that are associated with HIF inhibition? Are any of the available imaging techniques useful to assess the activity of HIF inhibitors? How should patients, in whom to test these agents, be selected? Should HIF expression be required for treatment or might HIF expression predict response to therapy? In an attempt to answer some of these questions, a clinical trial is being conducted at the National Cancer Institute (Bethesda, MD) in patients whose cancers express HIF-1α in more than 10% of tumor tissue.¹ Patients are treated with oral topotecan on a daily ×5 days ×2 weeks schedule every 28 days at a dose below the predicted maximum tolerated dose. Patients are selected based on tissue expression of HIF-1α, and the primary end point of the study is inhibition of HIF-1α expression, assessed by immunohistochemistry, in tumor tissue. A number of correlative studies, including imaging with 18F-fluorodeoxyglucose-positron emission tomography and dynamic contrast-enhanced magnetic resonance imaging to assess tumor metabolism and blood flow, respectively, mRNA expression of HIF-target genes, and soluble markers of angiogenesis, are also conducted to validate potential effects on the primary target. This clinical trial may provide useful information about the feasibility of such studies and the downstream biological effects associated with modulation of HIF-1 activity in tumor tissue.

Finally, two aspects relevant to the development of HIF inhibitors should be considered. The first is that HIF-1α expression is frequently detected in the inflammatory infiltrate of solid tumors, a feature poorly represented in animal models (52). Although the regulation of HIF-1 in stromal infiltrating cells is still poorly understood, recent evidence has shown that HIF-1α plays a critical role in myeloid cell (53) as well as endothelial cell (54, 55) functions. The effect of small-molecule inhibitors of HIF-1 on the cellular components of the tumor microenvironment has not been adequately addressed in preclinical models, and modulation of HIF activity in these cells may be instrumental for therapeutic success. The second aspect to consider is that HIF-1 inhibition alone may not be sufficient to halt angiogenesis and tumor growth, as HIF-independent pathways may bypass or overcome HIF inhibition (56). Thus, combination of HIF inhibitors with conventional therapies or emerging molecular targeted agents may be required and warranted. Combination therapies may be indicated not only for the potential synergistic inhibition of multiple and redundant signaling pathways, both HIF dependent and HIF independent, but also for the potential involvement of HIF in mediating resistance to radiation therapy (57) and chemotherapy (58, 59). The combination of HIF inhibitors with conventional therapeutic strategies and novel molecular targeted agents is also essential to finally integrate hypoxia-targeting approaches with standard therapeutics tools available to oncologists.