Small molecule inhibition of hypoxia inducible factor-1α: A viable therapeutic approach? Free

In the constant search for new anticancer drugs, much recent attention has been devoted to the rational search for agents that target cell-signaling pathways. While this approach is attractive and has gained some recent success, downstream of the cytokine-induced kinase cascades lies a cadre of transcription factors which are, in and of themselves, not unattractive drug targets. In this issue of MCT, one such factor, hypoxia inducible factor 1α (HIF-1α), receives attention from Welsh and her colleagues as a target for small molecule intervention (1).

HIF-1 is functionally heterodimeric, composed of HIF-1β and one of three α subunits (HIF-1α, HIF-2α, or HIF-3α Ref. 2). All subunits are part of the basic helix-loop-helix superfamily of transcription factors, but its activity is primarily controlled by cellular levels of the HIF-1α subunit. Thus, to affect transcriptional regulation, interference with HIF-1α provides a logical approach to the design of a viable drug candidate. Furthermore, HIF-1α has an attractive profile with respect to cancer. It appears to be measurably expressed in many solid tumors, where surrounding normal tissue or benign lesions seem to lack expression (3, 4). There are also indications that high levels of HIF-1α can correlate with poor patient prognosis, failure to respond to chemotherapy, and highly aggressive tumor growth for a broad range of solid cancers (5–7). With such a story, it is perhaps not surprising that a number of existing drugs have been screened for their capacity to interfere with HIF-1-mediated transcription. Among those that show a capacity to reduce HIF-1 protein levels or inhibit HIF-1-mediated transcription are drugs that would normally be considered to target Hsp-90, thioredoxin, topoisomerases, microtubule complexes, histone deacetylases, kinases (such as PI-3 kinases and MEK1), and guanyl cyclase (8). It is a reasonable assumption that these agents do not have HIF-1 as a primary target. Indeed, HIF-1 inhibition can be achieved at concentrations that are generally 10-fold higher than those required to induce cell growth inhibition. Although antisense approaches have been successfully employed to attack HIF-1 (9), the need for a specific small molecule therapeutic inhibitor of HIF-1 seems readily apparent. Hence, the present commentary on the paper by Welsh et al. (1) in this issue.

What are the necessary approaches needed to advance the HIF-1 inhibitor field? Many investigations have shown that HIF-1 can be modulated in vitro, but a drug effective in animals with a good toxicity profile and impressive antitumor activity would certainly provide a significant step forward. Welsh et al. (1) clearly demonstrate that their new drug PX-478 (S-2-amino-3-[4′-N,N,-bis(2-chloroethyl) amino] phenyl propionic acid N-oxide dihydrochloride) has some of these characteristics. Perhaps the most impressive component of their study is the antitumor data. Using a reasonable i.p. schedule of Q1D5, statistically meaningful (but, perhaps more importantly, biologically significant) tumor growth delays were achieved in 10 solid tumor models grown in scid mice. Although the log₁₀ cell kill values ranged from 0.1 to 3.0, one small cell lung cancer was cured. As the authors mention, the tumor models chosen are among the most refractory to standard chemotherapeutic agents, so the responses reported are encouraging (1). Also encouraging is the enhanced response of larger established tumors to PX-478, a fact that has in the past augured well for successful development of a cancer drug. In terms of toxicity, if rodents can predict for humans, neutropenia and weight loss are not unmanageable in a clinical setting. As a further consideration, the rodent pharmacokinetics show that serum levels and half-life are not inconsistent with the concentrations required to achieve inhibition of HIF-1α in an in vitro setting. Indeed, these data are backed up by indications that HIF-1α is suppressed by drug treatments that cause tumor regressions and not by those that do not. They may also indicate that tumors with high levels of HIF-1α are the only ones that will respond to PX-478.

So what should be made of PX-478? Not surprisingly, a number of questions remain unanswered. In the absence of direct data, the implication that the drug acts directly on HIF-1α remains just that, an implication; although structural biology and/or modeling of a drug-protein complex might prove educational. What other intracellular targets may be causing the cytotoxic effects of the drug? Once again, the authors imply that anti-angiogenic effects are not relevant, but a transcriptional change in gene expression related to glucose transport and cellular intermediary metabolism may cause cytotoxicity (1). Cautionary to this conclusion is that a cause/effect relationship for direct drug action and these events has yet to be defined. What properties of a tumor make HIF-1 highly expressed? The factors that seem to influence expression include, hypoxia and redox balance. Both of these parameters are frequently subject to dysfunction in solid tumors, so this would seem to be an encouraging scenario. Is HIF-1 prevalent enough in human cancer to be a general target? Probably, and anyway, there are indications of a difference in expression between normal and malignant tissue, usually a sound rationale for improved therapeutic index. Specificity notwithstanding, the reported antitumor data are themselves encouraging enough to merit further development of the drug. Finally, is PX-478 here to stay? As is often quoted in the lay press, “scientists are optimistic, but caution that it is too early in development to tell.”