Phorbol Ester Stimulates the Nonhypoxic Induction of a Novel Hypoxia-Inducible Factor 1α Isoform: Implications for Tumor Promotion

As the first step in the cellular adaptation to hypoxia, the transcriptional activation of genes essential for cell survival is mediated by the binding of hypoxia-inducible factor 1 (HIF-1) to the hypoxia response elements of these hypoxia-inducible genes (1). HIF-1 is composed of HIF-1α and HIF-1β, both of which belong to the per-arnt-sim family of basic helix-loop-helix transcription factors (2). HIF-1 activity depends primarily on the amount of HIF-1α protein (3). At the protein level, HIF-1α is markedly increased by hypoxia, whereas HIF-1β is constantly present regardless of oxygen tension.

HIF-1α is composed of 826 amino acids (2). Its NH₂-terminal contains the basic helix-loop-helix and the per-arnt-sim domains, which are essential for dimerization and DNA binding (4), and its COOH-terminal contains two transactivation domains and a nuclear localization signal (5, 6). In the middle section of HIF-1α lies a Pro-Ser-Thr rich oxygen-dependent degradation domain (ODDD, amino acid 401–603), which is responsible for stability of HIF-1α protein (7). Under normoxic conditions, HIF-1-prolyl hydroxylases (PHD1–3) hydroxylate two proline residues in the amino acid motif LXXLAP in the NH₂-terminal (amino acid 390–417) and the COOH-terminal (amino acid 557–576) of the ODDD (8, 9, 10, 11, 12). The von Hippel-Lindau tumor suppressor protein, a part of the E3 ubiquitin ligase protein complex, binds directly to the modified amino acid motifs of HIF-1α, resulting in the ubiquitination and proteasomal degradation of HIF-1α (13, 14). Because the enzymatic reaction of prolyl hydroxylation requires molecular oxygen as a substrate, hypoxia limits the hydroxylation, thereby precluding the binding of von Hippel-Lindau tumor suppressor protein and stabilizing HIF-1α. In addition, the acetylation of a lysine residue (Lys⁵³²) within the ODDD is another mechanism regulating HIF-1α stability. The lysine acetylation enhances the interaction between HIF-1α and von Hippel-Lindau tumor suppressor protein. Because the expression of ARD1, a HIF-1α acetyltransferase, decreased under hypoxic conditions, HIF-1α escapes from the acetylation and is stabilized (15).

A growing body of evidence indicates that HIF-1 contributes to tumor progression and metastasis. In human tumors, HIF-1α is overexpressed as a result of intratumoral hypoxia and genetic alterations affecting key oncogenes and tumor suppressor genes (16, 17). The expression level of HIF-1α, in biopsies of various solid tumors, is correlated with the tumor aggressiveness, vascularity, treatment failure, and mortality (18). In addition, tumor growth and angiogenesis in grafted tumors also depend on HIF-1 activity or the expression level of HIF-1α (19). Our recent study (20) demonstrated that the growth of grafted tumors was halted by the inhibition of HIF-1α and strongly supports the causative role of HIF-1α in tumor promotion.

In this study, we identified a novel splice variant of HIF-1α in mammalian cells. Deprived of exon 11, the variant, designated HIF-1α⁷⁸⁵, produced a HIF-1α of 785 amino acids. The variant was strikingly up-regulated by phorbol 12-myristate 13-acetate (PMA) through a redox-dependent mitogen-activated protein/extracellular signal-regulated kinase-1 (MEK-1) pathway and was also induced by heat and oxidative stresses under normoxic conditions. In a xenograft experiment, HIF-1α⁷⁸⁵-expressing tumors grew faster than HIF-1α-expressing tumors. HIF-1α⁷⁸⁵ seems to play an important role in tumor promotion and in other cellular processes related to HIF-1 activation.

PMA, wortmannin, rapamycin, KT5720, KT5823, calphostin C, and herbimycin A were purchased from Alexis Biochemicals (Lausen, Switzerland). SB203580 and PD98059 were purchased from Calbiochem (San Diego, CA) and [α-³²P]CTP (500 Ci/mmol) was from NEN Life Science (Boston, MA). Other chemicals were obtained from Sigma-Aldrich Corp (St. Louis, MO) unless otherwise indicated. Culture media and the FCS were purchased from Life Technologies, Inc. (Grand Island, NY).

Most cancer cell lines were obtained from American Type Culture Collection and SNU601 cell line, a stomach cancer cell line was obtained from Korean Cell Line Bank (Seoul, Korea). Hep3B and other cell lines were cultured in α-modified Eagle’s medium and in DMEM, supplemented with 10% heat-inactivated FCS, 100 units/ml penicillin, and 100 μg/ml streptomycin in a humidified atmosphere containing 5% CO₂ at 37°C. O₂/CO₂ levels in the incubator (Vision Sci Co., Seoul, Korea) were either 20%/5% (normoxic) or 1%/5% (hypoxic).

RNAs were isolated using Trizol (Life Technologies, Inc.) and reverse transcribed using the avian myeloblastosis virus reverse transcriptase system (Promega, Madison, WI). Reverse-transcription (RT)-PCR was done in a reaction involving reverse transcription at 48°C for 1 h and 25 PCR cycles at 94°C for 30 s, 53°C for 30 s, and 68°C for 30 s ∼2 min. PCR products (5 μl) were electrophoresed on 1 or 2% agarose gel containing ethidium bromide and directly sequenced. On the basis of the nucleotide sequence of the HIF-1α gene (GenBank no. AH006957 for the human gene; GenBank no. AH006789 for the mouse gene; GenBank no. AF057308 for the rat cDNA), eight pairs of PCR primers were designed to amplify HIF-1α and its variant cDNAs, as shown in Fig. 1.

Full-length HIF-1α cDNAs of ∼2.5 kb were amplified by RT-PCR using S3 and A2 primers and cloned using a pCR2.1-TOPO cloning kit (Invitrogen, Carlsbad, CA). To select a colony of Escherichia coli transformed with the HIF-1α cDNA variant, colony PCRs using S1 and A4 primers were performed. The HIF-1α cDNA variants were amplified and sequenced.

To quantitate mRNAs, a highly sensitive, semiquantitative RT-PCR was performed as described previously (21). One μg of total RNAs was reverse transcribed at 48°C for 1 h, and the cDNAs were amplified over 20 PCR cycles (94°C for 30 s, 53°C for 30 s, and 68°C for 30 s) in a 50 μl of reaction mixture containing 5 μCi [α-³²P]dCTP and 250 nm of each primer set. The PCR products (5 μl) were electrophoresed on a 4% polyacrylamide gel, and dried gels were autoradiographed. The nucleotide sequences of the primer pair (5′ to 3′) for ARD1 were ATGAACATCCGCAATGCG and AGGCTCTAGGAGGCTGAG. Primers for vascular endothelial growth factor, aldolase A, enolase 1, and β-actin were constructed as described previously (20).

Hemagglutinin (HA)-tagged HIF-1α expression plasmid (pcDNA3) was constructed as described previously (3). HA-tagged HIF-1α variant (HIF-1α⁷⁸⁵) expression plasmid was made using a PCR-based mutagenesis kit (Stratagene, Cedar Creek, TX). Mutagenesis using specific oligonucleotides (forward, 5-GACACAGATTTAGACTTGGAGATG-3; reverse, 5-CTCAGGTGAACTTTGTCTAGTGCTTC-3) was used to delete nucleotides derived from exon 11. To replace a lysine residue (amino acid 532) in HIF-1α with arginine, the bases AAG in pcDNA3-HA-tagged HIF-1α were replaced with the bases CGA, by using a QuickChange site-directed mutagenesis kit (Stratagene). All constructs were verified by DNA sequencing.

To establish stable transfectant cell lines, ∼40% of the confluent HEK 293 cells in 60-mm cell culture dishes were transfected with 4 μg of plasmid using the calcium phosphate method. Cells were allowed to stabilizing 36 h and then treated with 0.45 mg/ml G418. Thirty stable transfectants from three different transfections were pooled to avoid bias in gene expression because of variable chromosomal integration.

A luciferase reporter gene, containing the HIF-1-binding erythropoietin enhancer region, was constructed as described previously (3). To assay HIF-1 activity, HEK293 cells were cotransfected with 0.6 μg each of reporter gene and plasmid cytomegalovirus-β-gal and/or plasmids of HIF-1α or its variant, using the calcium phosphate method. pcDNA was added to ensure that the final DNA concentrations in both the control and experimental groups were at the same level. After being allowed to stabilize for 48 h, the cells were incubated under either normoxic or hypoxic conditions, in the absence or in the presence of PMA for 16 h, and lysed to determine luciferase and β-gal activities.

Total cell lysate was separated on a 6.5% SDS/polyacrylamide gel and transferred to an Immobilon-P membrane (Millipore, Bedford, MA). Membranes were blocked with 5% nonfat milk in Tris-buffered saline containing 0.1% Tween 20 (TTBS) at room temperature for 1 h and then incubated overnight at 4°C with rabbit anti-HIF-1α (22) diluted 1:1000, rabbit anti-HA (Biodesign International, Saco, ME) diluted 1:4000, or rat anti-ARD1, generated against bacterially expressed full-length peptide of human ARD1, diluted 1:5000 in 5% nonfat milk in TTBS. Horseradish peroxidase-conjugated antirabbit or antirat antiserum was used as a secondary antibody (1:5000 dilution in 5% nonfat milk in TTBS, 2 h of incubation) and the antigen antibody complexes were visualized using an Enhanced Chemiluminescence Plus kit (Amersham Biosciences Corp., Piscataway, NJ). β-Actin was used as an internal standard and quantified using mouse monoclonal anti-β-actin antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA).

Male nude (BALB/cAnNCrj-nu/nu) mice were purchased from Charles River Japan, Inc. (Shin-Yokohama, Japan). Animals were housed in a specific pathogen-free room under controlled temperature and humidity. All animal procedures were performed according to the procedures in the Seoul National University Laboratory Animal Maintenance Manual. Eighteen mice ages 7–8 weeks were injected s.c. in the flank with 5 × 10⁶ viable cells expressing HIF-1α or variant. Tumor volumes were measured with calipers and calculated using the formula: volume = a × b²/2, where a is the width at the widest point of the tumor and b is the width perpendicular to a.

HIF-1α cDNAs were amplified in HEK293 cells that had been treated under normoxic, hypoxic, or with CoCl₂. In addition to the wild-type cDNA fragment, a smaller product was also reproducibly amplified (Fig. 2,A). DNA sequencing revealed that the small fragment was a result of the omission of exon 11 and the splicing of exons 10 and 12. Both PCR 2 and the PCR 3, which leave out exon 11, revealed a single band of 847 and 1578 bp, respectively (second lanes in Fig. 2, B and C), indicating the absence of a splicing site other than that at exon 11 in the variant cDNA. Primer specificity on the variant cDNA was verified because the wild-type HIF-1α DNA was not amplified (third lanes). Because the intensity of the 28-cycle fragment in PCR 5 (variant) was approximately the same as that of the 25 cycle PCR 4 (wild type), the amount of variant mRNA was estimated to be 1/8 [=1/2^(28–25)] of that of the wild type (Fig. 2,D). Among 20 colonies harboring full-length HIF-1α cDNAs, 17 contained wild-type cDNA and 3 contained variant cDNA (Fig. 2,E). Of note, the 3–20 colony ratio is comparable with the mRNA ratio estimated in Fig. 2,D. The variant mRNA was detected by PCR 7 in both human and murine cell lines, where the wild-type mRNA was also detected by PCR 8 (Fig. 2 F). The murine cell lines produced a larger PCR 8 product because mouse exon 11 is 42-bases longer than human HIF-1α.

The structure of the spliced variant is summarized in Fig. 3. The variant cDNA was identical to HIF-1α cDNA, except for the deletion of the 123-bp exon 11. This variant generates a shorter mRNA without altering the reading frame and produces a 785-amino acid polypeptide composed of amino acid residues 1–512 and 554–826, which is designated here as HIF-1α⁷⁸⁵. HIF-1α⁷⁸⁵ preserves most of the functional domains, including both von Hippel-Lindau tumor suppressor protein-interacting subdomains (amino acids 390–417 and 557–576). However, HIF-1α⁷⁸⁵ is deprived of a small part (amino acid 513–553) of the ODDD, which contains the lysine residue (Lys⁵³²) responsible for the regulation of HIF-1α stability by acetylation. Because the ODDD is partially missing, it is possible that the protein stability of HIF-1α⁷⁸⁵ is regulated in a different manner.

HIF-1α is not regulated by oxygen tension alone; it is also regulated by various other stimuli such as PMA, hyperthermia, and reactive oxygen species (ROS; Refs. 23, 24). When cells transfected with pHA-HIF-1α or pHA-HIF-1α⁷⁸⁵ were subjected to hypoxia or treated with PMA, HIF-1α⁷⁸⁵ was markedly induced by PMA in a dose-dependent manner, whereas HIF-1α was induced by hypoxia (Fig. 4,A). The level of HIF-1α⁷⁸⁵ was much greater in PMA-treated cells than in hypoxia-treated cells. In addition, HIF-1α⁷⁸⁵ was induced by heat in a time-dependent manner, whereas HIF-1α and β-actin were unaffected (Fig. 4 B). These results indicate that HIF-1α⁷⁸⁵ is an inducible isoform under normoxic conditions.

In terms of the mobility and bandwidth of the proteins, HIF-1α⁷⁸⁵ was distinguished from HIF-1α. HIF-1α⁷⁸⁵ showed a faster mobility and a sharper band pattern on SDS-PAGE than HIF-1α (Fig. 4,C). To determine whether PMA induces endogenous HIF-1α⁷⁸⁵, we examined the mobility and band pattern of HIF-1α induced by hypoxia and PMA. Fig. 4,D shows that PMA-induced HIF-1α migrated faster and was sharper than hypoxia-induced HIF-1α. This is comparable with the characteristics of expressed HIF-1α⁷⁸⁵, as shown by Fig. 4 C. These results suggest that the different HIF-1α isoforms are induced by hypoxia, whereas PMA specifically induces HIF-1α⁷⁸⁵.

HIF-1α⁷⁸⁵ expression by PMA appears to be regulated at the protein level because its mRNA level was unaltered by PMA (Fig. 4,E). To further examine whether PMA treatment increases the stability of HIF-1α⁷⁸⁵ protein, we measured the half-life of the protein after blocking de novo protein synthesis with cycloheximide. Fig. 4 F shows that the half-life of HIF-1α⁷⁸⁵ was prolonged to 15.5 min by PMA treatment, whereas the half-life in untreated cells was only 5.4 min.

Both hypoxia and PMA increased the luciferase activity of erythropoietin enhancer reporter in pcDNA-transfected cells (Fig. 5,A). In HIF-1α-transfected cells, however, only hypoxia and not PMA further augmented reporter activity. In contrast, HIF-1α⁷⁸⁵-transfected cells showed enhanced reporter activity by PMA but not by hypoxia. To confirm the functional role of HIF-1α⁷⁸⁵ in response to PMA, we examined the mRNA levels of hypoxia-inducible genes, namely, vascular endothelial growth factor, aldolase A, and enolase 1 (Fig. 5 B). In pcDNA-transfected cells, all of the mRNAs of the hypoxia-inducible genes were increased by 10 nm PMA. In HIF-1α-transfected cells, the basal levels of these mRNAs were elevated in the control, but their expressions were not additionally increased by PMA. In HIF-1α⁷⁸⁵-transfected cells, however, the levels of these mRNAs were markedly increased by an even lower concentration (2 nm) of PMA. Theses results suggest that HIF-1α⁷⁸⁵ plays an important role in the PMA-induced expression of hypoxia-inducible genes under normoxic conditions.

The distinctive nonhypoxic induction of HIF-1α⁷⁸⁵ may be accounted by the lack of exon 11. Recently, the lysine residue (Lys⁵³²) within exon 11 was identified as the target of ARD1 acetyltransferase, which destabilizes HIF-1α (15). To determine whether Lys⁵³² inhibits the nonhypoxic induction of HIF-1α by PMA and hyperthermia, we changed lysine at 532 to arginine (K532R), as shown in Fig. 6. Surprisingly, the point mutation of Lys532 was sufficient to render HIF-1α responsive to both PMA and hyperthermia (Fig. 6,A), whereas HIF-1α was not induced (Fig. 6 B). Thus, Lys⁵³² is likely to play an inhibitory role in stabilization of HIF-1α by PMA and hyperthermia.

Concerning the nonhypoxic induction of HIF-1α⁷⁸⁵, we hypothesize that PMA and hyperthermia provide some signal for the stabilization of HIF-1α isoforms but that wild-type HIF-1α is not stabilized because of acetylation of the exon 11 domain by ARD1. In contrast, HIF-1α⁷⁸⁵ lacking the exon 11 domain is stabilized. This hypothesis is based on the premise that the HIF-1α acetylation process is active in cells treated with PMA and hyperthermia. As reported previously (15), the expression of ARD1 mRNA and protein markedly decreased under hypoxic conditions. However, the ARD1 expression was not suppressed by PMA or hyperthermia (Fig. 6,C), which suggests that HIF-1α is acetylated and degraded under these normoxic conditions. To confirm this hypothesis, we examined whether the stability of HIF-1α⁷⁸⁵ was affected by ARD1. The PMA induction of HIF-1α⁷⁸⁵ was not suppressed in pARD1-transfected cells (Fig. 6,E), whereas the hypoxic induction of HIF-1α was almost completely blocked (Fig. 6 D). These results suggest that blockade of acetylation in HIF-1α⁷⁸⁵ provides protein stability under normoxic conditions.

To determine the nature of the signal transduction pathway mediating the PMA induction of HIF-1α⁷⁸⁵, HEK293 cells and pHIF-1α⁷⁸⁵-transfected cells were incubated with various inhibitors such as N-acetylcysteine, a reducing thiolic compound; wortmannin, a phosphatidylinositol 3′-kinase inhibitor; SB203580, a p38 mitogen-activated protein kinase inhibitor; PD98059, a MEK-1 inhibitor; rapamycin, an FK506-binding protein 12-rapamycin associated protein inhibitor; KT5720, a protein kinase A inhibitor; KT5823, a protein kinase G inhibitor; calphostin C, a protein kinase C inhibitor; herbimycin A, a tyrosine kinase inhibitor. N-Acetylcysteine and PD98059 reduced the induction of endogenous HIF-1α⁷⁸⁵ (Fig. 7,A, left panel), although these inhibitors did not affect the hypoxic induction of HIF-1α. In contrast, HIF-1α expression was reduced by SB203580 (Fig. 7,A, right panel). N-Acetylcysteine and PD98059 also inhibited the induction of expressed HIF-1α⁷⁸⁵ (Fig. 7,B). In addition, another reducing thiolic compound, DTT, inhibited the PMA induction of HIF-1α⁷⁸⁵ (Fig. 7,C), suggesting that the redox state regulates the stability of HIF-1α⁷⁸⁵. Furthermore, HIF-1α⁷⁸⁵ expression was up-regulated by hydrogen peroxide, and this effect was completely abolished by N-acetylcysteine (Fig. 7 D). The hyperthermic induction of HIF-1α⁷⁸⁵ was also suppressed by N-acetylcysteine (data not shown). These results suggest that the HIF-1α⁷⁸⁵ stabilization by PMA is regulated by redox state and by a MEK-1 pathway, whereas HIF-1α stabilization by hypoxia is regulated by a p38 mitogen-activated protein kinase pathway.

HIF-1 plays a causative role in tumor growth (20). In addition, PMA and oxidative stress, both of which induce HIF-1α⁷⁸⁵, are also well-known tumor promoters (25). To investigate the possibility that HIF-1α⁷⁸⁵ promotes tumor growth, we grafted HIF-1α- or HIF-1α⁷⁸⁵-expressing cells in nude mice. As expected, HIF-1α-expressing cells successfully formed tumors, whereas the cells having empty vector did not. However, tumors expressing HIF-1α⁷⁸⁵ grew faster than tumors expressing HIF-1α (Fig. 8). Excised HIF-1α⁷⁸⁵ tumors were larger and more hypervascular than HIF-1α tumors. Thus, HIF-1α⁷⁸⁵ expression seems to be associated with tumor growth.

In this study, we identified a novel splice variant of HIF-1α, HIF-1α⁷⁸⁵, in human and murine cancer cells. This variant lacks exon 11 and produces a 785 amino acid HIF-1α without altering the reading frame and retains normal transactivation activity. Moreover, its expression was found to be strikingly up-regulated by PMA. The expression of HIF-1α⁷⁸⁵ was blocked by reducing agents and enhanced by hydrogen peroxide. Mutational and pharmacological studies revealed that HIF-1α⁷⁸⁵ can escape from lysine acetylation and that this results in the PMA-induced stabilization of HIF-1α⁷⁸⁵ protein. Moreover, HIF-1α⁷⁸⁵-enhanced tumor growth in vivo. Taken together, our results show that HIF-1α⁷⁸⁵ is induced by PMA under normoxic conditions via a redox-dependent MEK-1 pathway and suggest that HIF-1α⁷⁸⁵ plays an important role in the nonhypoxic activation of HIF-1 and in tumor promotion.

To date, four splice variants of human HIF-1α mRNA have been reported. One has an additional 3 bp at the junction of the exons 1 and 2 without a frameshift (26). The second loses exon 14 and translates into a 736-amino acid polypeptide (HIF-1a⁷³⁶). This variant is regulated by oxygen tension and transactivates the vascular endothelial growth factor promoter (26). The third loses exon 12 and produces a short form of HIF-1α⁵⁵⁷, which was induced specifically by the zinc ion (21), and the fourth lacks exons 11 and 12 and produces a shorter form of HIF-1α⁵¹⁶ (27). These truncated variants, which contain the per-arnt-sim domain, inhibit HIF-1 activity by sequestering HIF-1β and reduce the mRNA expression of hypoxia-inducible genes. Similarly, a splice variant of HIF-3α, which also contained the per-arnt-sim domain, inhibited HIF-1 activity in the mouse cornea (28, 29). In this study, we found the fifth splice variant of human HIF-1α, HIF-1α⁷⁸⁵. Compared with the protein structures of truncated variants previously found, HIF-1α⁷⁸⁵ retains all of the essential domains of HIF-1α and functions as a transcription activator.

Hypoxia is not the only player in HIF-1α stabilization. Many other stimuli have been found to play important roles in controlling HIF-1α regulation in nonhypoxic environments. Hyperthermia induces the nonhypoxic expression of HIF-1α. Katschinski et al. (24) demonstrated that the expression of HIF-1α was up-regulated in several tissues and in cultured hepatoma cell lines after hyperthermic incubation. They also found a low molecular weight form of HIF-1α, which had a sharper band pattern in cells exposed to hyperthermia. Although they suggested that the molecular weight difference was due to reduced phosphorylation of heat-induced HIF-1α protein, the identity of the HIF-1α isoform induced by hyperthermia has not been determined. On the basis of the mobility and bandwidth of heat-induced HIF-1α, it is possible that heat-induced HIF-1α is HIF-1α⁷⁸⁵.

In terms of the mechanism underlying the HIF-1α induction by PMA, the activation of the phosphatidylinositol 3′-kinase pathway is possibly involved (30). However, PMA stimulates the expression of HIF-1α⁷⁸⁵ in a different way because HIF-1α⁷⁸⁵ expression was not suppressed by a phosphatidylinositol 3′-kinase inhibitor. Instead, an oxidative process seems to be responsible for HIF-1α⁷⁸⁵ stabilization (Fig. 7). Similarly, reducing agents also suppressed HIF-1α⁷⁸⁵ induced by hyperthermia or hydrogen peroxide. These results suggest that the stabilization of HIF-1α⁷⁸⁵ is achieved via an oxidative pathway. This is supported by reports that PMA and hyperthermia stimulate the oxidation process. PMA is known to produce oxidative stress via stimulating NADPH oxidase in nonphagocyte cells as well as phagocytes (31). Hyperthermia also produces oxidative stress in cancer and muscle cells due to increased energy metabolism (32). Therefore, PMA, hyperthermia, and hydrogen peroxide could stimulate an oxidation process to stabilize HIF-1α⁷⁸⁵.

The role of oxidative stress in the regulation of HIF-1α is controversial. Several experiments have demonstrated that strong oxidizing reagents impair the expression of HIF-1α in hypoxic cells (3). In contrast, Chaddel et al. (33, 34) demonstrated that HIF-1 activation is associated with increased production of ROS from mitochondrial complex III. In addition, inflammatory cytokines producing ROS have been reported to stabilize HIF-1α (35). Why is the effect of ROS on HIF-1α expression different? This question may be derived from the misconception that the same isoform of HIF-1α is induced by hypoxia and ROS. On the basis of our results, we speculate that under hypoxic conditions, wild-type HIF-1α is suppressed by oxidative stress but that under normoxic conditions HIF-1α⁷⁸⁵ is induced by oxidative stress.

We also demonstrated the possibility that two mitogen-activated protein kinase pathways, MEK-1/extracellular signal-regulated kinase and p38 mitogen-activated protein kinase, selectively regulate HIF-1α⁷⁸⁵ and HIF-1α. Several studies (36, 37) have demonstrated that these pathways mediate the stabilization or the transcriptional activation of HIF-1α. However, the effects of specific pathway inhibitors were variable and depended on the conditions of HIF-1α induction. In the present study, inhibitors of the MEK-1 and p38 pathways, PD98059 and SB203580, also produced different effects on the expression of HIF-1α isoforms in PMA- and hypoxia-treated cells. These results additionally support our hypothesis that PMA and hypoxia regulate different isoforms of HIF-1α via different mechanisms.

A growing body of evidence indicates that HIF-1 contributes to tumor progression and metastasis (16, 20). Because PMA and ROS, which are well-known tumor promoters, were found to enhance HIF-1 activity by inducing HIF-1α⁷⁸⁵, it is possible that HIF-1α⁷⁸⁵ contributes to tumor promotion by these stimuli. Our results, as shown in Fig. 8, support this possibility. In addition, it has been reported that HIF-1 also contributes to other cellular processes that occur under normoxic conditions such as the development of normal tissues, the determination of cell death or survival, immune responses, and adaptation to mechanical stresses (23). Therefore, HIF-1α⁷⁸⁵ could participate in these normoxic processes.