Natural Product-Derived Small Molecule Activators of Hypoxia-Inducible Factor-1 (HIF-1)

Abstract

Hypoxia-inducible factor-1 (HIF-1) is a key mediator of oxygen homeostasis that was first identified as a transcription factor that is induced and activated by decreased oxygen tension. Upon activation, HIF-1 upregulates the transcription of genes that promote adaptation and survival under hypoxic conditions. HIF-1 is a heterodimer composed of an oxygen-regulated subunit known as HIF-1α and a constitutively expressed HIF-1β subunit. In general, the availability and activity of the HIF-1α subunit determines the activity of HIF-1. Subsequent studies have revealed that HIF-1 is also activated by environmental and physiological stimuli that range from iron chelators to hormones. Preclinical studies suggest that HIF-1 activation may be a valuable therapeutic approach to treat tissue ischemia and other ischemia/hypoxia-related disorders.

The focus of this review is natural product-derived small molecule HIF-1 activators. Natural products, relatively low molecular weight organic compounds produced by plants, animals, and microbes, have been and continue to be a major source of new drugs and molecular probes. The majority of known natural product-derived HIF-1 activators were discovered through pharmacological evaluation of specifically selected individual compounds. The combination of natural products chemistry with appropriate high-throughput screening bioassays could provide an alternative approach to discover novel natural product-derived HIF-1 activators. Potent natural product-derived HIF-1 activators that exhibit a low level of toxicity and side effects hold promise as new treatment options for diseases such as myocardial and peripheral ischemia, and as chemopreventative agents that could be used to reduce the level of ischemia/reperfusion injury following heart attack and stroke.

Introduction

Over the course of time, multicellular organisms have evolved tightly regulated oxygen delivery systems to ensure oxygen dependent energy production. In the human body, high levels of oxygen (hyperoxia) can cause oxygen toxicity while low levels (hypoxia) are associated with hypoxia/ischemia-related diseases such as ischemic and neoplastic disorders. At the organism level, the body responds to changes in oxygen levels by altering respiration rate and blood flow. At the cellular level, changes in oxygen levels can trigger responses from a network of signaling pathways that lead to alterations in gene expression patterns.

Hypoxia-inducible factor-1 (HIF-1) is a transcription factor that is activated by hypoxic conditions [1]. It is composed of a HIF-1α subunit and a HIF-1β (also known as aryl hydrocarbon receptor nuclear translocator or ARNT) subunit, both are members of the basic helix-loop-helix (bHLH) PER-ARNT-SIM (PAS) family of transcription factors [2]. Over seventy genes have been identified as HIF-1 target genes and the list is still growing [3]. These genes encode proteins that are involved in many aspects of cellular physiology, ranging from cellular metabolism, cell proliferation/survival/death, cytoskeletal structure, cell adhesion/motility, angiogenesis, erythropoiesis, vascular tone, to drug resistance. It is no surprise that HIF-1 plays a crucial role in development, physiological processes such as wound healing, and pathological processes such as tumor progression and tissue ischemia [3–5].

As a key regulator of oxygen homeostasis, HIF-1 is tightly regulated by the level of oxygen. Initial studies revealed that the HIF-1α subunit is degraded rapidly under normoxic conditions and stabilized under hypoxic conditions, while the HIF-1β subunit is constitutively expressed [2]. One of the breakthroughs in elucidating the pathways and mechanisms involved in HIF-1 activation is the discovery that prolyl hydroxylation of HIF-1α protein is followed by the von Hippel-Lindau disease tumor suppressor (pVHL)-mediated, oxygen-dependent degradation of HIF-1α protein, Fig. (1) [6–10]. Under normoxic conditions, the HIF-1α subunit is hydroxylated by prolyl hydroxylases in the presence of oxygen and iron. The prolyl hydroxylated HIF-1α subunit is then recognized by an E3 ubiquitin ligase complex that contains pVHL, polyubiquitinated, and degraded by the 26S proteosome. When the cellular oxygen level is reduced, prolyl hydroxylation is inhibited. Inhibition of HIF-1α protein prolyl hydroxylation and degradation leads to an increase in HIF-1α protein level. Oxygen-dependent regulation of HIF-1 activity also occurs at the transactivation level [11–17]. The HIF-1α subunit is hydroxylated by asparaginyl hydroxylase in the presence of oxygen and iron [12–15]. Hydroxylation of the asparagine-803 in HIF-1α abrogates the interaction between HIF-1α and the coactivator CBP/p300 [CREB(cAMP-response element-binding protein)-binding protein/E1A-binding protein; 300kD] [16–17]. Upon reduction in oxygen concentration, the asparaginyl hydroxylation of HIF-1α protein is blocked. The stabilized and activated HIF-1α protein translocates into the nucleus where it heterodimerizes with the HIF-1β subunit and binds to hypoxia response elements (HREs) present on the promoters of HIF-1 target genes. This complex recruits coactivators such as CBP/p300 and enhances transcription. Kinetic studies suggest that the prolyl hydroxylases that modify HIF-1α protein are more sensitive to decreases in oxygen tension than the asparaginyl hydroxylase that modifies HIF-1α protein [18]. The pathways and mechanisms that connect oxygen sensing, hydroxylation, and HIF-1 are extensively discussed in a number of recent reviews [19–26]. In addition to hydroxylation, HIF-1 activity is also regulated by other post-translational modifications that include phosphorylation, acetylation, S-nitrosation, SUMOylation, and ubiquitination [26].

Fig. (1).

Oxygen-Dependent Regulation of HIF-1α Stabilization and Activation

From the aspect of drug discovery, target validation is a critical step towards finding effective therapeutic agents. Results from extensive preclinical and clinical studies support HIF-1 as a valid molecular target for anti-cancer drug discovery [3–4, 27–31]. Overexpression of the HIF-1α subunit is associated with advanced disease stages, poor prognosis, and treatment resistance among cancer patients, while inhibition of HIF-1 retards tumor growth in animal models. A number of recent reviews have summarized the progress towards the discovery of potential cancer therapeutic agents that inhibit HIF-1 [3, 27–30]. Intense research efforts directed at the discovery of novel HIF-1 inhibitors for the treatment of cancer are currently underway in academic, industrial, and government laboratories.

Agents that activate HIF-1 may prevent ischemia/reperfusion injuries and speed recovery from tissue ischemia. The focus of this review is natural product-derived activators of HIF-1 and their therapeutic potential. In contrast to the general impression among the lay public that “natural products” and “herbal medicines” are interchangeable terms, in the field of natural products chemistry and drug discovery the term “natural product” refers to small organic compounds that are generally thought to be “secondary metabolites” that play a role in chemical defense or otherwise enhance survival. These are typically low molecular weight compounds produced by animals, plants, and microbes. Several “primary metabolites” (compounds required for the normal biochemical and/or physiological function of an organism) such as certain steroids, carbohydrates, and a prostaglandin have been shown to regulate HIF-1. In light of the significant roles these compounds play on HIF-1 mediated-signaling pathways, primary metabolites that regulate HIF-1 activation are included in this review. Natural products have been a major source of new drugs for centuries and the chemical diversity offered by natural products has not been matched by other chemical approaches. Statistics show that over 60% of the approved anticancer, antihypertensive, and antimigraine medications are of natural origin (natural products or synthetic compounds based on natural product models) [32]. A large number of the molecular probes currently used in biomedical research are natural products (i.e. rapamycin, genistein, wortmannin, actinomycin D, cycloheximide, geldanamycin, etc.) or derived from natural products (i.e. PD98059, LY294002, etc.). In this review, natural product-derived HIF-1 activators, grouped by their chemical or structural classes, will be discussed in regard to their discovery, mechanism of action, and therapeutic potential.

Alkaloids and Other Amino Acid Derivatives

The first natural product shown to activate HIF-1 was deferoxamine (1) [33]. Deferoxamine (1) was first identified as a siderochrome from Streptomyces pilosus that chelates ferric ions [Fe³⁺; Fe (III)] [34]. The metal free mesylate salt form (desferoxamine mesylate, Desferal Mesylate^®) is used clinically as a heavy metal antagonist for the treatment of iron and aluminum poisoning [35]. Wang and coworkers demonstrated that 1 induces the DNA-binding activity of HIF-1 and increases erythropoietin (a HIF-1 target gene) mRNA levels in cultured cells [Hep3B human hepatoma and Chinese hamster ovary (CHO)] [33]. Deferoxamine (1) has since been widely used as a hypoxia mimetic to activate HIF-1. The prolyl and asparaginyl hydroxylases that destabilize and inactivate HIF-1α protein are Fe(II)- and 2-oxoglutarate dependent oxygenases. Since 1 is water soluble and has a chelation preference for Fe(III), it is likely that 1 inhibits HIF prolyl and asparaginyl hydroxylases by preventing the uptake of cellular iron ions and thus activates HIF-1.

The tuberculosis pathogen Mycobacterium tuberculosis secretes desferri-exochelins (high-affinity iron-binding siderophores) to extract iron ions from host proteins [36]. Treatment of cells with desferri-exochelin DFE 722 SM (2) induced HIF-1α protein, activated HIF-1, and increased the expression of known HIF-1 targets such as VEGF (vascular endothelial growth factor) and NIP3 (BCL2/adenovirus E1B 19-kD protein-interacting protein 3) in human breast tumor MDA468 cells [37]. The fact that iron-bound ferri-exochelin did not activate the HIF-1 pathway suggests that iron chelation is required for the activation of HIF-1. While both deferoxamine (1) and desferri-exochelins are high affinity Fe(III) chelators, desferri-exochelin DFE 722 SM (2) is at least ten times more potent than deferoxamine at HIF-1 activation [37]. Desferri-exochelin (2) is highly lipophilic, thus allowing it to readily enter cells and chelate intracellular iron ions. In contrast, 1 is hydrophilic, limiting its ability to penetrate cell membranes. This apparent difference in polarity may contribute to the difference in potency between these two iron chelators. Desferri-exochelin DFE 722 SM (2) most likely functions by inhibiting the Fe(II)-dependent prolyl and asparaginyl hydroxylases that modify, destabilize, and inactivate HIF-1α protein.

Ciclopirox olamine [3, Loprox^®, 6-cyclohexyl-1-hydroxy-4-methyl-2(1H)-pyridone 2-aminoethanol] is a commonly used antimycotic agent to treat fungal infections of the skin and nail [38]. In human heptoma HepG2 cells, 3 was shown to stabilize HIF-1α protein, induce HIF-1α protein nuclear translocation, facilitate the binding of HIF-1 to hypoxia-response elements, and promote HIF-1 activation [39]. Ciclopirox olamine (3) is at least ten times more potent than deferoxamine (1) at activating HIF-1 in vitro and ciclopirox olamine-induced HIF-1 activation can be blocked by Fe(II) and Al(III) ions [39]. It is likely that 3 also acts as an iron chelator and activates HIF-1 through the inhibition of HIF prolyl and asparaginyl hydroxylases that destabilize and inactivate HIF-1α protein under normoxic conditions. Compound 3 was shown to induce the expression of HIF-1 targets such as VEGF, GLUT-1 (glucose transporter-1), and aldolase in vitro (HepG2 cells), and stimulate angiogenesis (consequence of VEGF induction) in vivo [mouse skin wound model and chicken embryo chorioallantoic membrane (CAM) assay] [40]. However, 3 failed to induce HIF-1 target genes in an ex vivo organ model (isolated rat kidneys perfused with 3) [40].

The compound 8-methyl-pyridoxatin (4) is structurally related to ciclopirox olamine (3) and induces erythropoietin expression in human hepatoma Hep3B cells [41]. Compound (4) has been shown to activate HIF-1 in an engineered CHO cell-based reporter assay and induce HIF-1α protein in HepG2 cells [39]. Neither the effect of 4 on the expression of other HIF-1 target genes nor the mechanism of action has yet been reported.

The HIF prolyl and asparaginyl hydroxylases are members of the Fe(II)- and 2-oxoglutarate (2OG)-dependent oxygenase superfamily. Most members of this superfamily, including those that modify HIF, couple the oxidative decarboxylation of 2OG with the hydroxylation of substrate [20–24]. Analogues of 2OG such as N-oxaloylglycine (5, NOG) were originally developed as inhibitors of prolyl-4-hydroxylase, one of the Fe(II)- and 2OG-dependent oxygenases [42]. Both 5 and an ester of NOG known as DMOG (6, dimethyl-oxalylglycine) have been shown to induce HIF-1α protein through the inhibition of prolyl hydroxylation [7]. The ester form of NOG (6) is more potent in cell-based assays, presumably due to increased membrane permeability [17]. Subsequently, 6 was shown to inhibit the asparaginyl hydroxylation of HIF-1α protein and activate HIF-1 [12]. Analysis of X-ray crystallographic data of HIF asparaginyl hydroxylase (factor-inhibiting HIF, or FIH) complexed with either substrates or inhibitors indicates that 5 inhibits FIH by displacing the co-substrate 2OG from the active site and hindering the nucleophilic attack at the 2-carbonyl group [43]. In studies that employed recombinant FIH protein expressed by insect cells, 5 inhibited FIH and human type I collagen prolyl-4-hydroxylase (C-P4H-I) with the same potency, and inhibited HIF prolyl hydroxylases (HPH or PHD) with reduced efficiency (4- to 25-fold increase in Ki values, depending on each isoenzyme) [44, 45]. Among the known C-P4H-I inhibitors, compounds such as N-((3-hydroxy-6-chloroquinolin-2-yl)carbonyl)-glycine and 3-hydroxypyridine-2-carbonyl-glycine are selective inhibitors of HPHs relative to their affects on FIH [44, 45]. By contrast, 3,4-dihydroxybenzoate and pyridine-2,5-dicarboxylate are significantly more selective inhibitors of FIH, relative to their effect on HPHs, and pyridine-2,4-dicarboxylate has a comparable inhibitory effect on both FIH and HPHs [44, 45]. The alanine derivative N-oxalyl-2S-alanine (7), but not its epimer N-oxalyl-2R-alanine inhibited both HIF prolyl and asparaginyl hydroxylases [7, 43]. A structure-activity relationship study conducted using a cell-free assay system identified a number of additional 2OG analogues that inhibit HPHs [46]. The natural products alahopcin (8) and dealanylalahopcin (9) were first isolated from Streptomyces albulus subsp. ochragerus and both compounds exhibited weak inhibitory activity against the collagen prolyl hydroxylase [47–49]. Two dealanylalahopcin analogues 3-carboxymethylene N-hydroxy succinimide (10) and 3-carboxy-N-hydroxy pyrollidone (11) were subsequently shown to act as HPH inhibitors [50]. Since most of these HPH inhibitors were identified in cell-free assay systems, whether these substances can specifically activate the HIF-1 pathway in relevant biological systems remain to be demonstrated.

The level of HIF-1α protein has been shown to decrease over extended incubation (24 h) under either anoxic or hypoxic (0.1% O₂) conditions in human colon carcinoma RKO cells [51]. This same study revealed that two derivatives of the natural product indirubin (12) [5-iodoindirubin-3'-oxime (13) and 5-methylindirubin-3'-oxime (14)] prevented the decrease in HIF-1α protein through the inhibition of GSK3β (glucogen synthase kinase 3β) and increased HIF-1α protein translation. Neither indirubin derivative affected the level of HIF-1α protein under normoxic conditions. Whether this indirubin derivative-induced stabilization of HIF-1α protein is associated with increased HIF-1 activity was not addressed. Indirubin (12) is the active ingredient of Danggui Longhui Wan, a traditional Chinese remedy used to treat various diseases including chronic myelocytic leukemia [52]. It has been shown that indirubin derivatives such as compound (13) inhibit GSK3β, cyclin-dependant kinase-1/cyclin B (CDK1/cyclin B), and cyclin-dependant kinase-5/p25 (CDK5/p25) kinases with similar potency [53]. A structurally related compound indirubin-3'-monoxime inhibited a panel of kinases with IC₅₀ values in the nanomolar range [54]. The apparent lack of selectivity observed in these studies raises question as to the actual mechanism of action for these indirubin derivativres and whether or not the reported increase in HIF-1α protein will translate into an actual increase in HIF-1 activity.

One group of amino acid-derived natural products that activate HIF-1 is nitric oxide (NO) donors. Initial studies revealed that NO inhibits HIF-1 activation under hypoxic conditions [55–57]. Later, Kumara and coworkers discovered that the HIF-1 binding site in the human VEGF promoter actually mediates NO-induced activation of VEGF transcription under normoxic conditions [58]. The NO donor S-nitroso-N-acetyl-D,L-penicillamine (15, SNAP) induced HIF-1α protein accumulation, HIF-1 binding activities, and activated transcription from the VEGF promoter in A-172 human glioblastoma and Hep3B cells. Another NO donor, 3-(hydroxy-1-(1-methylethyl)-2-nitrosohydrazino)-1-propanamine (NOC5) also activated the VEGF promoter in a cell-based reporter assay. In bovine pulmonary artery endothelial and rat aortic smooth muscle cells, the NO donor diazen-1-ium-1,2-diolate (NOC-18), induced HIF-1α protein and HIF-1 DNA binding, augmented HIF-1β protein expression, and activated expression of the HIF-1 target gene heme oxygenase-1 (HO-1) [59]. The induction of HIF-1 by NOC-18 is both dose- and time-dependent (optimal induction: 500 μM, 3–4 h). It was observed that Angeli's salt (a NO donor that generates NO⁻, nitroxyl equivalents) failed to induce HIF-1 activity, while S-nitrosoglutathione (16, GSNO) (an endogenous NO donor that generates NO⁻, nitrosonium equivalents) exerted a similar ability to induce HIF-1 as NOC-18. This indicates that a NO⁻ equivalent-mediated electrophilic nitrosylation reaction may take place during NO-induced HIF-1 activation. Reversal of NOC-18-induced HIF-1 activation by dithiothreitol (DTT) suggests a mechanism that involves intracellular S-nitrosylation or oxidation of protein thiols. Subsequent studies have revealed that S-nitrosylation stabilizes HIF-1α protein and S-nitrosylation of Cys-800 promotes the interaction between HIF-1α protein and the co-activator p300, thus enhancing HIF-1 activation [60, 61]. Diazenium diolate NO donors that include spermine NONOate (17, a complex of NO with the natural product spermine), diethylamine NONOate, and diethyltryamine NONOate each induced HIF-1α protein in a dose- and time-dependent manner in multiple cell lines (proximal tubular LLC-PK1, human breast carcinoma MCF-7, MB231, and MB157) [62, 63]. Using a combination of NO donors that activate HIF-1 and pharmacological inhibitors of selected pathways, studies from numerous groups have revealed that NO donor-induced HIF-1 activation requires the presence of NO, is independent of the guanylate cyclase/guanosine 3',5'-monophosphate (cGMP) pathway, activates the PI3K/AKT/mTOR (phosphoinositol 3-kinase/protein kinase B/mammalian target of rapamycin) pathway to increase the synthesis of HIF-1α protein, and is sensitive to alterations in the cellular redox environment [59, 63–66]. Biochemical studies indicate that NO can bind to the iron in the active sites of HIF hydroxylases, block O₂ binding and inhibit the hydroxylation reaction [23]. Inhibition of the hydroxylases that destabilize and inactivate HIF-1α protein may also contribute to NO-mediated HIF-1 activation in cell-based studies. Both the concentration and duration of NO released by structurally different NO donors should be considered when interpreting the results of these HIF-1 activation studies. Various NO concentration thresholds have been demonstrated to activate different signaling pathways under normoxic conditions [63]. The activation of HIF-1 by NO is most likely the overall outcome of modulating distinct pathways with different mechanisms.

Microtubule-depolymerizing agents (MDA) disrupt the microtubule network and block tumor cell division. The plant alkaloids vinblastine (18) and colchicine (19), and the synthetic MDA nocodazole, each induced HIF-1α protein at the concentrations that also maximally disrupted microtubules in various cell lines (A549 human lung carcinoma, MCF-7, Jurkat human T-cell leukemia, and NIH-3T3 murine embryonic fibroblast) [67]. The induction of HIF-1α protein by MDAs requires microtubule disruption, nuclear factor-κB (NF-κB)-dependent transcription, and functional VHL protein. The MDA-induced HIF-1α protein is transcriptionally active, as evidenced by increased expression of the HIF-1 target iNOS ( inducible nitric oxide synthase). Taxol, an anti-tumor agent that promotes microtubule polymerization and stabilization, did not induce HIF-1 activation.

Phenolic Compounds

Dibenzoylmethane (20, DBM), a minor beta-diketone found in licorice (Glycyrrhiza glabra), was shown to stabilize HIF-1α protein in multiple cell lines that include, human prostate carcinoma LNCaP and PC-3, human embryonic kidney 293 (HEK 293), and rat neonatal primary cardiomyocytes [68]. The induction of HIF-1α protein correlated with increased HIF-1 activity (HEK 293 cells) and increased expression of the HIF-1 target gene VEGF at the secreted protein level (LNCaP cells and primary cardiomyocytes). Two structurally related compounds (dibenzoylpropane and curcumin) did not induce HIF-1α protein in HEK 293 cells. The fact that DBM-induced HIF-1α protein is non-ubiquitinated suggests that 20 may prevent HIF-1α protein degradation through the inhibition of prolyl hydroxylase. Dibenzoylmethane has been shown to inhibit 7,12-dimethylbenz[a]anthracene (DMBA)-induced DNA damage, mammary tumorigenesis, and lymphoma/leukemia in several animal models [69–71]. However, it is particularly important to note that the concentration of 20 that activates HIF-1 [68] is within the range of its IC₅₀ values for cytotoxicity in LNCaP and PC-3 cells [72]. Whether dibenzoylmethane (20) has therapeutic potential for the treatment of ischemic diseases remains to be addressed in a biologically relevant system.

The flavonoid quercetin (21) is commonly found in red wine, grapes, and many other plants. Under normoxic conditions, quercetin (3,3',4',5,7-pentahydroxyflavone) was shown to activate HIF-1 in HeLa cells and murine brain endothelial cells (MBEC) that overexpress HIF-1α (maximum activation: 50 μM in HeLa and 100 μM in MBEC cells) [73]. The same study demonstrated that 21 stabilizes HIF-1α protein in HeLa cells, induces HIF-1α protein nuclear translocation, and increases the expression of the HIF-1 targets VEGF and GLUT-1 in MBEC cells. A decrease in HIF-1 activity was observed when quercetin and hypoxia were combined. Since 21 is known to be a broad-spectrum protein kinase inhibitor [54, 74], it is unlikely inhibition of Ser/Thr kinases is the only mechanism for quercetin-induced HIF-1 activation, as proposed by the authors. A recent study demonstrated that quercetin (21) also inhibits factor-inhibiting HIF (FIH), the asparaginyl hydroxylase that inactivates HIF-1α protein under normoxic conditions [75]. Such FIH inhibitory activity may also contribute to the ability of quercetin to activate HIF-1.

Green tea [dried fresh leaves of the plant Camellia sinensis L. Ktze., (Theaceae)] is one of the most popular drinks worldwide. Green tea catechins that include (–)-epicatechin (EC), (–)-epigallocatechin (EGC), (–)-epicatechin-3-gallate (22, ECG), and (–)-epigallocatechin-3-gallate (23, EGCG) are believed to be responsible for the health promoting benefits of drinking green tea. In human breast carcinoma T47D cells, ECG (22) was found to activate HIF-1 at high concentrations (100 μM) [76]. The most widely studied catechin EGCG (23) also produced a modest increase in HIF-1 activity. Despite the fact that 22 and 23 are nearly identical structurally, except for one hydroxyl group, EGCG (23) is relatively unstable in aqueous solution [76–78]. Further investigation of the ECG-induced activation of HIF-1 revealed that 22 induces the accumulation of nuclear HIF-1α protein and activates the expression of HIF-1 targets that include VEGF, GLUT-1 and CDKN1A (cyclin-dependant kinase inhibitor 1A; p21^waf1/cip1) [76]. The observation that both the induction of HIF-1α protein and the activation of HIF-1 can be blocked by iron ions and ascorbate suggests that 22 may activate HIF-1 by chelating the iron ions that are required for the post-translational modifications that destabilize and inactivate HIF-1α protein [76]. Whether ECG (22), EGCG (23), and green tea extract products can activate HIF-1 in other models remains to be addressed. It is interesting to note that EGCG (23) and green tea extract have been shown to protect against ischemia/reperfusion injury, one of the intended therapeutic targets for HIF-1 activators [79–82].

Terpenes/Steroids

Three structurally related sesquiterpene-tropolones (pycnidione, epolone A and epolone B) were originally identified as natural products capable of inducing erythropoietin (EPO) in Hep3B cells [83]. Erythropoietin is one of the genes activated by HIF-1. Pycnidione (24) activated HIF-1 in a stable CHO cell line that expresses luciferase under the control of HREs from the transferrin gene (a target of HIF-1). The greatest levels of activation were observed with 24 at 8 μM under normoxic conditions and at 4 μM under hypoxic conditions (1% O₂) [39]. At higher concentrations, 24 is cytotoxic. Induction of HIF-1α protein was observed in the presence of 24 in HepG2 cells [39]. The effect of 24 on other HIF-1 target gene expression and the mechanism of action were not reported. Judging by their structural similarities and comparable abilities to induce EPO, it is likely that epolones A and B can function like pycnidione (24) to activate HIF-1.

The estrogen 17β-estradiol (E2) is normally metabolized by liver cytochrome P450 enzymes into 2-hydroxy estradiol (2-OHE2) and 4-hydroxy estradiol (4-OHE2, 25). The compound 4-OHE2 (25) was demonstrated to induce HIF-1α protein in a dose- and time-dependent manner (optimal induction: 100 μM, 3 hr) in human ovarian carcinoma OVCAR-3 and A2780-CP70 cells [84]. No induction of HIF-1α protein was observed in the presence of the structurally related metabolite 2-OHE2. An increase in secreted VEGF protein (the bioactive gene product) was observed in the presence of 25 in both OVCAR-3 and A2780-CP70 cells. Mechanism of action studies that employed molecular probes suggest that 25 induces HIF-1α protein and increases secreted VEGF protein via activation of the PI3K/AKT/mTOR pathway, not the mitogen-activated protein kinase (MAPK) pathway. Given the current knowledge on HIF-1 regulation, it is possible that 4-OHE2 (25) increases HIF-1α protein synthesis and activates HIF-1 by activating the PI3K/AKT/mTOR pathway. However, further studies on 4-OHE2-induced HIF-1α protein synthesis, nuclear translocation, HIF-1 activation, and the expression of other HIF-1 target genes are necessary to support this hypothesis. Since 17β-estradiol (100 nM) was shown to inhibit the hypoxic activation of HIF-1 in Hep3B cells [85], whether the activation of HIF-1 by the 17β-estradiol metabolite 4-OHE2 (25) at 100 μM (1000 times the concentration required for inhibition by 17β-estradiol ) is of any real significance remains to be addressed in a relevant animal-based in vivo model.

The testosterone metabolite dihydrotestosterone (26, 1 nM) induces HIF-1α protein and activates HIF-1 in androgen-dependent LNCaP cells, but not in androgen-independent PC-3 cells [86]. Similar effects were observed in the presence of the metabolically stable synthetic androgen methyltrienolone (27, R1881) at a concentration of 0.1 nM. Androgens activate HIF-1 by inducing growth factors such as epidermal growth factor (EGF), that in turn enhance the synthesis of HIF-1α protein via the activation of the receptor kinase/PI3K/AKT/mTOR pathway. Since androgens are not direct stimuli for HIF-1 activation, this response requires an extended incubation time in contrast to the action of direct stimuli such as hypoxia or iron chelators.

The diterpene ester phorbol 12-O-myristate 13-acetate (28, PMA, also known as 12-O-tetradecanoylphorbol 13-acetate or TPA) which was originally isolated from the seeds of Croton tiglium L. is a potent activator of protein kinase C. Phorbol 12-O-myristate 13-acetate (28) is used as a tumor promoting agent in model systems and as a molecular probe for pathway studies [87]. In human prostate carcinoma DU145, PC-3, PPC-1, and TSU cell lines, 28 (50 - 100 nM) was shown to induce nuclear HIF-1α protein accumulation, and inhibitors of the EGF/PI3K/PTEN (phosphatase and tensin homolog)/AKT/mTOR pathway blocked the effect of PMA on HIF-1α protein [88, 89]. Induction of nuclear HIF-1α protein by 28 is followed by an increase in HIF-1 activity in DU145 cells [88]. In cultured rat neonatal cardiomyocytes, 28 has been shown to induce HIF-1α expression at the mRNA level [90]. Studies conducted in genetically engineered mouse fibroblasts suggest that PMA-induced HIF-1 activation is independent of the receptor-interacting protein (RIP)/NF-κB pathway, while the induction of HIF-1 activation by tumor necrosis factor α (TNFα) depends on the RIP/NFκB pathway [91]. A recent study revealed that 28 induces a novel HIF-1α isoform (HIF-1α⁷⁸⁵) in HEK293 cells [92]. This HIF-1α variant is devoid of exon 11 that encodes part of the oxygen-dependent degradation domain (ODDD). Phorbol 12-O-myristate 13-acetate (28) stabilizes HIF-1α⁷⁸⁵ protein without affecting HIF-1α⁷⁸⁵ mRNA level and the activity of 28 requires at least redox-sensitive and mitogen-activated protein kinase kinase/extracellular signal-regulated kinase (MEK/ERK) pathways [92]. Since 28 modulates a number of cellular signaling pathways, its HIF-1 inducing activity is most likely the combined effects of modulating a network of pathways.

Prostaglandin E₂ (PGE₂)

Prostaglandin E₂ (29, PGE₂) was demonstrated to induce HIF-1α protein in human colon carcinoma HCT116 cells in a dose- and time-dependent manner [93]. Although modest in comparison to that triggered by hypoxia, maximal induction was observed after an 18–32 hr incubation in the presence of 29 (100 μM) . Prostaglandin E₂ (29) induces HIF-1α protein by increasing HIF-1α protein synthesis. Pathway studies that employed pharmacological agents suggest that 29 activates the G-protein coupled receptor EP₁, which then activates MEK/ERK and the v-src avian sarcoma (Schmidt-Ruppin A-2) viral oncogene homolog (c-src), followed by the activation of the PI3K/AKT/mTOR pathway, and leads to increased HIF-1α protein synthesis. The induction of HIF-1α protein by 29 is accompanied by an increase in the level of VEGF mRNA. In PC-3ML (a subline of PC-3) cells, 29 (1 μM) induced nuclear HIF-1α protein accumulation within 4 hr and 29 appears to be more potent than hypoxia (1% O₂) at inducing HIF-1α protein [94]. The effects of PGE₂ (29) on HIF-1α protein were mediated by EP₂ and EP₄ receptor subtypes, through activation of the MAPK and PI3K/AKT pathways.. The enzyme cyclooxygenase-2 (COX-2) has been identified as a mediator for the induction of HIF-1 and VEGF by the proinflammatory cytokine IL-1β in both A549 and colon carcinoma Caco-2 cells [95]. In A549 cells, PGE₂ (one of the major products of COX-2 catalyzed metabolism of arachidonic acid) induced nuclear HIF-1α protein accumulation at concentrations as low as 0.1 μM within 1 hr. These results suggest that the effect of 29 on HIF-1 and its target genes can be mediated by more than one pathway, depending on which specific tumor type is involved. Due to the wide range of concentrations and incubation times required in different cell lines to elicit a HIF-1 response from PGE₂ (29), particular caution should be taken when extrapolating these laboratory findings to the clinical setting.

Carbohydrates and Glycolysis Products

Hypoxic conditions and a state of aerobic glycolysis are commonly found in solid tumors. One group of HIF-1 target genes is glucose transporters and the glycolytic enzymes that promote glycolysis [3]. Identification of glucose and the glycolytic end products pyruvate and lactate as HIF-1 stimuli has helped construct a two-way connection between hypoxia and aerobic glycolysis [96]. In human glioma cells (U87, U373, and U251), glucose (30), pyruvate (31), and lactate (32) were each shown to induce nuclear HIF-1α protein accumulation in a dose- and time-dependent manner. Further study using pharmacological inhibitors identified pyruvate as the key glycolytic metabolite that induces HIF-1α protein by preventing its degradation. Activation of HIF-1 targets that include VEGF, EPO, GLUT3, and aldolase A was observed in the presence of either glucose or pyruvate [96]. At the final stage of glucose oxidation, reactions in the Krebs cycle (TCA cycle or citric acid cycle) oxidize the acetyl group of acetyl CoA into CO₂. One intermediate of the TCA cycle, oxaloacetate (33), induces HIF-1α protein, activates HIF-1, and induces the expression of HIF-1 target genes in U87 and U251 cells [97]. Induction of the HIF prolyl hydroxylases (HPH) HPH-1 and HPH-2 mRNAs was also observed in the presence of these endogenous 2-oxoacids (31 and 33), as well as by the other known HIF-1 stimuli that include hypoxia (1% O₂), the iron chelator DFO (1), the transition metal Co²⁺, and the 2-oxoglutarate (2OG) analogue dimethyl-N-oxalylglycine (6) [97]. This suggests the involvement of a potential negative feedback loop that regulates HIF-1 activity. In an effort to identify the link between succinate dehydrogenase (SDH) mutations and tumor formation, Selak and coworkers revealed that succinate (34) links mitochondrial dysfunction to oncogenesis via the activation of HIF-1 [98]. The enzyme SDH is localized to the inner mitochondrial membrane and catalyzes the conversion of 34 to fumarate. Both 34 and fumarate are intermediates of the TCA cycle. Inhibition of SDH leads to an increase in 34, which is transported from mitochondria to the cytosol. Succinate (34) is the product of 2OG-dependent oxygenases and a weak inhibitor of these enzymes. Succinate (34) inhibits HIF prolyl hydroxylases in the cytosol, stabilizes HIF-1α protein, and subsequently activates HIF-1. The concentrations of 34 that were shown to inhibit HPH activity in HEK293 cell extracts are within the range of 34 found in succinate dehydrogenase deficient cells. This SDH-succinate-HIF-1 link is supported by results from clinical studies that have demonstrated that SDH mutation-related tumors such as pheochromocytoma and renal cell carcinoma are highly vascular and have activated hypoxic signaling pathways [99].

Other Activators of HIF-1

Two groups of chemicals commonly used to induce chemical hypoxia are iron chelators and transition metals. Natural product-derived iron chelators that activate HIF-1 were discussed earlier in this review. Synthetic iron chelators such as 2,2'-pyridyl and 1,10-phenanthroline can also activate HIF-1 and the expression of HIF-1 target genes [40, 100]. A compound that is structurally related to 1,10-phenanthroline, known as FG-0041 inhibits the activity of HPH-2 (PHD-2; EGLN-1), induces HIF-1α protein, and activates the expression of the HIF-1 target VEGF [101]. Cobalt was the first transition metal shown to activate HIF-1 [102, 103]. Other metals such as nickel, chromium (VI), and copper have also been shown to activate HIF-1 [104–107]. The organomercurial compound mersalyl activates HIF-1 and its target genes via the insulin-like growth factor-1 receptor and MAPK pathway [108]. Recently, it was demonstrated that cobalt and nickel activate HIF-1 by depleting intracellular ascorbate, a co-factor for the HIF hydroxylases that destabilize and inactivate HIF-1α protein [109]. A separate study revealed that the Ca²⁺ chelator BAPTA (1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid) activates HIF-1 in SH-SY5Y human neuroblastoma cells [110].

There is increasing evidence that supports HIF-1 as a general regulator of cellular responses to environmental, extracellular, and intracellular signals. Non-hypoxic physiological stimuli such as growth factors, cytokines, and hormones can also induce HIF-1α protein and activate HIF-1. Activation of HIF-1 by insulin, insulin-like growth factor (IGF)-1 and IGF-2 leads to an increased expression of HIF-1 target genes such as IGF-2, IGF-binding protein (IGFBP)-2 and IGFBP-3 and forms an autocrine loop that promotes proliferation [111, 112]. Other growth factors that include EGF, basic fibroblast growth factor (bFGF), and heregulin have all been shown to activate HIF-1 in various cell lines [88, 112–114]. Both PI3K and MAPK signaling pathways are involved in mediating the induction and activation of HIF-1 by growth factors [112–117]. Activation of the PI3K/AKT/mTOR pathway relieves translational inhibition and enhances translation of HIF-1α mRNA [113]. These growth factors activate HIF-1 by increasing HIF-1α protein synthesis, in contrast to the mechanism of stimuli such as hypoxia and iron chelators that inhibit the degradation of HIF-1α protein. Cytokines such as tumor necrosis factor alpha (TNFα) and interleukin-1 beta (IL-1β) induce HIF-1α protein under normoxic conditions and the induction can be augmented by hypoxic treatments [118, 119]. TNFα-induced HIF-1α protein stabilization and HIF-1 activation involve reactive oxygen species (ROS)-sensitive pathways, activation of the PI3K/AKT and MAPK pathways, and receptor-interacting protein (RIP)-dependent activation of NFκB [65, 91, 120, 121]. A recent study demonstrated that TNFα increases HIF-1α protein synthesis [122]. Activation of HIF-1 by IL-1β requires pathways similar to those used by TNFα [95, 115, 123, 124]. Most of the studies indicate that IL-1β induces HIF-1α protein accumulation. One study demonstrated that IL-1 also rapidly increases HIF-1α mRNA levels [125]. The vascular hormones angiotensin II and thrombin activate HIF-1 via stimulation of both HIF-1α transcription and translation [126–128]. The mechanisms used by these hormones to activate HIF-1 were summarized in a recent review [129]. Recently, thyroid hormone and follicle-stimulating hormone were added to the list of hormones that activate HIF-1 [130, 131]. Other physiological factors such as the redox protein thioredoxin-1 (Trx-1) and oxidized low-density lipoprotein (oxLDL) can also induce HIF-1α protein and activate HIF-1, suggesting that HIF-1 may play a role in tumors that over-express Trx-1 and diseases related to oxLDL such as atherosclerosis [132, 133].

Studies conducted using genetically engineered mice deficient in HIF-1α established that HIF-1 plays a critical role in inflammatory responses [134–136]. These observations provide an explanation for the activation of HIF-1 by inflammatory stimuli that range from viral proteins to bacteria lipopolysaccharide (LPS) [137, 138].

Potential Therapeutic Applications of HIF-1 Activators

The intended therapeutic targets for HIF-1 activators are tissue ischemia and related diseases. These conditions may include peripheral, cerebral and myocardial ischemia, reperfusion injury following heart attack or stroke, wound healing, organ transplant, and retinopathy of prematurity in neonates. Current approaches to activate HIF-1 include HIF-1α gene therapy [139–141], small molecules that inhibit the HIF hydroxylases [40, 45, 101, 142], and peptides that block the degradation of HIF-1α protein [143, 144]. Results from studies in transgenic mice indicate that HIF-1 activation may have a therapeutic advantage over direct VEGF activation, in regard to the promotion of angiogenesis with lower incidence of side effects (i.e. vessel leakage associated with VEGF over-expression) [145].

As an example, the therapeutic potential of HIF-1 activation for the treatment of cardiovascular diseases is discussed. Atherosclerotic coronary artery disease progresses from vascular stenosis (narrowing of the vessel) and ischemia to vessel occlusion followed by myocardial infarction (death of the cells that depend on perfusion from the occluded vessel). One adaptive physiological response to chronic myocardial ischemia is to form new collateral blood vessels that function as replacement for the occluded vessels. In patients, the extent of coronary collateralization correlates with good prognosis after myocardial infarction [146]. Results from animal studies suggest that the ability to respond to ischemia and form new blood vessels is most likely influence by both genetic and environmental factors. For example, age plays a significant role in ischemia-induced neovascularization as older mice and rabbits exhibit impaired collateralization following femoral artery ligation [147–149]. Intervention with HIF-1 activators represents a potential therapeutic approach to compensate for the impaired ischemia-induced collateral vessel formation. In a rat acute myocardial infarction model, HIF-1α gene therapy with a HIF-1α/VP16 hybrid enhanced angiogenesis and reduced infarct size [140]. Reintroduction of blood flow to an ischemic organ causes reperfusion injury, which represents a second phase of injury in diseases such as heart attack and stroke. Activation of HIF-1 by intermittent hypoxia protected isolated rodent hearts against ischemia-reperfusion injury [150]. Constitutive expression of HIF-1α hybrids from recombinant adenoviral vectors protected cultured rat neonatal cardiomyocytes against ischemia-reperfusion injury [151]. These results suggest that HIF-1 activators may also prevent ischemia-reperfusion injuries through a preconditioning mechanism. Ironically, desferri-exochelin which was recently shown to activate HIF [37], has also been shown to prevent cardiac reperfusion injury in cultured rat adult cardiac myocytes and in isolated rabbit hearts [152]. Desferri-exochelin also inhibited human vascular smooth muscle cell proliferation in vitro and prevented restenosis in a porcine coronary artery restenosis model [153, 154]. The high incidence of restenosis has limited the success of coronary angioplasty for the treatment of obstructive coronary artery lesions. Whether HIF-1 activation can prevent the development of restenosis after vascular injury has not been conclusively demonstrated.

Perspectives

Natural products and natural product-derived compounds have contributed significantly as important molecular probes to elucidate the process of HIF-1 activation and may ultimately serve as therapeutically useful HIF-1 activators. It is important to emphasize that HIF-1 responds not only to hypoxia, but also to numerous physiological and non-physiological stimuli such as exogenous low molecular weight natural products. Hypoxia appears to be the universal stimulus for HIF-1 activation and other activators are relatively cell type-specific. Most of the reported HIF-1 activators have only been examined in cell culture-based systems. While some of these studies were performed at physiologically relevant concentrations, other studies were performed at concentrations far beyond those obtained physiologically. Several studies were even conducted at or above established toxic concentrations. One critical point that must be kept in mind is that the induction of HIF-1α protein does not necessarily translate into HIF-1 activation. Several investigators have defined HIF-1 activators as substances/conditions that induce HIF-1α protein accumulation. This may not be true, since proteasome inhibitors such as MG-132 and lactacystin induce HIF-1α protein by preventing proteasomal degradation but these compounds actually inhibit HIF-1 activation [155]. The known HIF-1 activators can be simply classified into two general classes based on their effect(s) on HIF-1α protein: 1) inhibitors of HIF-1α protein degradation and inactivation; and 2) activators of HIF-1α transcription and translation. The majority of the agents shown to activate HIF-1 have been identified from specific experiments to examine the response of individually selected test compounds, rather than from using bioassays to screen for novel activators. Therefore, enormous potential remains for the discovery of new classes of HIF-1 activators through the use of high-throughput screening-based drug discovery approaches that would examine either natural product-based or synthetic-based chemical libraries. The complexity of pathways that regulate the production and function of HIF-1 poses a challenge for the identification of new HIF-1 activators that are suitable for drug development. Both environmental and genetic factors related to each specific disease target need to be considered when establishing proper in vitro models and developing bioassays for the purpose of HIF-1 activator-based drug discovery. Since ischemic tissues have intrinsically impaired vascularization, drug delivery is critical to achieve the desired therapeutic efficacy of HIF-1 activators. The therapeutic activation of HIF-1 should be treated as a finely tuned process since excessive HIF-1 activation may produce undesired pathological consequences. The ideal therapeutic HIF-1 activator will be one that both prevents and treats tissue ischemia without producing significant side effects. Only time will tell if specific HIF activators will meet the expectations of ideal molecular-targeted drugs.

List of Abbreviations

2-OHE2

2-hydroxy estradiol

2OG

2-oxoglutarate

4-OHE2

4-hydroxy estradiol

AKT

Protein kinase B

ARNT

Aryl hydrocarbon receptor nuclear translocator

BAPTA

1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid

bFGF

basic fibroblast growth factor

bHLH

basic helix-loop-helix

CAM

Chicken embryo chorioallantoic membrane

CBP

CREB (cAMP-response element-binding protein)-binding protein

CDK1

Cyclin-dependant kinase-1

CDK5

Cyclin-dependant kinase-5

CDKN1A

Cyclin-dependant kinase inhibitor 1A

cGMP

Guanosine 3',5'-monophosphate

CHO

Chinese hamster ovary

COX-2

Cyclooxygenase-2

C-P4H-I

Human type I collagen prolyl-4-hydroxylase

c-src

v-src avian sarcoma (Schmidt-Ruppin A-2) viral oncogene homolog

DBM

Dibenzoylmethane

DMBA

7,12-dimethylbenz[a]anthracene

DMOG

Dimethyl-oxalylglycine

DTT

Dithiothreitol

E2

17β-estradiol

EC

Epicatechin

ECG

Epicatechin-3-gallate

EGC

Epigallocatechin

EGCG

Epigallocatechin-3-gallate

EGF

Epidermal growth factor

EPO

Erythropoietin

ERK

Extracellular signal-regulated kinase (same as MAPK)

FIH

Factor-inhibiting HIF; HIF asparaginyl hydroxylase

GLUT-1

Glucose transporter-1

GSK3β

Glucogen synthase kinase 3β

GSNO

S-nitrosoglutathione

HIF-1

Hypoxia-inducible factor-1

HO-1

Heme oxygenase-1

HPH

HIF prolyl hydroxylase (same as PHD)

HRE

Hypoxia response element

IGF

Insulin-like growth factor

IGFBP

Insulin-like growth factor binding protein

IL-1β

Interleukin-1β

iNOS

Inducible nitric oxide synthase

LPS

Lipopolysaccharide

MAPK

Mitogen-activated protein kinase (same as ERK)

MBEC

Murine brain endothelial cells

MDA

Microtubule-depolymerizing agent

MEK

Mitogen-activated protein kinase kinase; MAPKK; MAPK/ERK kinase

mTOR

Mammalian target of rapamycin

NFκB

Nuclear factor-κB

NIP3

BCL2/adenovirus E1B 19-kD protein-interacting protein 3

NO

Nitric oxide

NOC5

3-(hydroxy-1-(1-methylethyl)-2-nitrosohydrazino)-1-propanamine

NOC-18

Diazen-1-ium-1,2-diolate

NOG

N-oxaloylglycine

ODDD

Oxygen-dependent degradation domain

oxLDL

Oxidized low-density lipoprotein

PAS

PER-ARNT-SIM

p300

E1A-binding protein, 300 kD

PGE₂

Prostaglandin E₂

PHD

Prolyl hydroxylase domain-containing protein (same as HPH)

PI3K

Phosphoinositol 3-kinase

PMA

Phorbol 12-O-myristate 13-acetate (same as TPA)

PTEN

Phosphatase and tensin homolog

VHL

von Hippel-Lindau disease tumor suppressor

RIP

Receptor-interacting protein

SDH

Succinate dehydrogenase

SNAP

S-nitroso-N-acetyl-D,L-penicillamine

TNF-α

Tumor necrosis factor α

TPA

12-O-tetradecanoylphorbol 13-acetate (same as PMA)

Trx-1

Thioredoxin-1

VEGF

Vascular endothelial growth factor