The forkhead transcription factor FOXO4 induces the down-regulation of hypoxia-inducible factor 1 alpha by a von Hippel-Lindau protein-independent mechanism.

Tumors utilize hyperactivation of the phosphatidylinositol 3-kinase (PI3K)/AKT signaling pathway to cope with deleterious environmental conditions. Activation of the PI3K/AKT pathway has been shown to increase protein expression of the alpha subunit of the hypoxia-inducible factor (HIF) 1, a key regulator of oxygen homeostasis. Elevated levels of HIF-1 alpha induce expression of genes with critical roles in angiogenesis, erythropoiesis, and glucose metabolism, processes that are essential for tumor expansion. Here we examine the involvement of FOXO4 (also known as AFX), a member of the forkhead transcription factor superfamily that is negatively regulated by the PI3K/AKT pathway, in the regulation of HIF-1 alpha protein expression. Nuclear expression of FOXO4 results in the suppression of various responses to hypoxia, including decreased vascular endothelial growth factor, glucose transporter 1, and erythropoietin expression. Interestingly, FOXO4 down-regulates the HIF-1 alpha protein levels, consistent with the lack of hypoxia responsiveness. Previous results have revealed a role for prolyl hydroxylation and resultant von Hippel-Lindau protein (pVHL) interactions in the ubiquitin-proteasome-mediated degradation of HIF-1 alpha. However, neither inhibition of prolyl hydroxylases nor mutation of HIF-1 alpha-hydroxylated prolines involved with pVHL-mediated binding inhibits the observed FOXO4-mediated down-regulation of HIF-1 alpha. These results suggest a novel alternate mechanism for hypoxic regulation that is dependent upon the level of activation of FOXO4-mediated transcription.

Induction of the PI3K 1 signaling pathway results in a diversity of survival-enhancing functions in both normal as well as oncogenically transformed cells (1). A large body of work in a variety of systems has revealed that activation of this pathway results in the induction of a cascade of kinases that include PDK-1 and one or more of the AKT/protein kinase B kinases (2). A major end result of AKT/protein kinase B kinase activation is phosphorylation of several sites in a family of forkhead-type transcription factors that includes FOXO4 (3)(4)(5)(6)(7)(8)(9)(10). Phosphorylation of these transcription factors results in nuclear export and cytoplasmic sequestration by 14-3-3 proteins as well as other factors. Lack of nuclear localization of FOXO4-type forkhead family members is highly beneficial to the proliferative and survival states of the cell, because expression programs induced by this category of transcription factors include the expression of proteins that inhibit cell cycle progression and induce apoptosis (8,9,(11)(12)(13)(14)(15)(16)(17). Thus, inhibition of FOXO4type transcription factor nuclear localization and the subsequent lack of expression of a variety of proteins involved in cellular quiescence and death constitute a major end point for the activated PI3K pathway. This fact is highlighted by examination of a range of late-stage tumors, including glioblastoma multiforme, where this kinase cascade is hyperactivated by the mutational loss of PTEN, a lipid phosphatase that is the major negative regulator of this pathway (18 -22). The activation of this pathway in these tumors results in a high degree of resistance to many cytotoxic agents, including radiation and chemotherapy. It is, therefore, likely that examination of the diversity of functions regulated by this pathway may result in enhanced treatments for the many tumors that benefit from PI3K activation.
In addition to intracellular survival mechanisms such as the PI3K pathway, rapidly proliferating cells in embryonic and oncogenically transformed tissues require an increased blood supply to maintain a normal nutritive and oxygenated state. This is accomplished by the induction of angiogenic and hematopoietic factors, including VEGF and EPO. When the cell is exposed to reduced oxygen levels or hypoxia as a result of tissue expansion or tumor growth, a transcription factor called hypoxia-inducible factor 1 (HIF-1) becomes stabilized and activates the expression of VEGF, EPO, GLUT-1, and others that are involved in angiogenesis, erythropoiesis, and glucose metabolism (23). HIF-1 consists of two subunits, HIF-1␣ and HIF-1␤, and the steady-state quantity of the ␣ subunit is tightly controlled by degradative processes in response to changing oxygen concentrations. The molecular mechanisms that link oxygen and HIF-1␣ levels have recently been elucidated (24,25). The importance of the von Hippel-Lindau (VHL) tumor suppressor protein, a component of a ubiquitin ligase complex, in the ubiquitin-mediated degradation of HIF-1␣ has been demonstrated, and a diversity of tumors, particularly renal carcinomas, shows loss of function of VHL and greatly enhanced angiogenesis (26 -35). Furthermore, recent studies have demonstrated that a family of prolyl-4-hydroxylases are involved with sensing increased oxygen concentrations and converting this information into higher levels of HIF-1␣ prolyl hydroxylation (36 -40). The VHL protein (pVHL) interacts with these hydroxylated proline residues and induces the ubiquitination and subsequent proteasome-dependent degradation of HIF-1␣. These elegant regulatory mechanisms thus ensure that appropriate levels of cellular angiogenic responses are maintained under varying oxygen concentrations.
Because both the PI3K and the hypoxic response pathways are regulated by the sensing of extracellular nutrient or oxygen levels, respectively, it seems likely that these two pathways may be connected. Previous data revealed a clear linkage between the level of activation of the PI3K pathway and hypoxia response as well as the steady-state quantities of HIF-1␣ (41)(42)(43)(44)(45)(46)(47)(48). Inhibition of the PI3K pathway was accompanied by a clear decrease in HIF-1␣ protein levels and inhibition of VEGF induction. Here we demonstrate that the PI3K-regulated transcription factor FOXO4 can suppress cellular response to hypoxia at least in part by down-regulation of the steady-state levels of the HIF-1␣ protein. Analysis of the mechanism of this down-regulation revealed that, although mediated by the proteasome pathway, it appears to be independent of pVHL-induced ubiquitination. Our data provide not only a molecular explanation for the regulation of the angiogenic response by the PI3K pathway but also the first evidence for the involvement of forkhead transcription factor FOXO4 in the regulation of hypoxia response, suggesting alternate means to control the levels of HIF-1␣ and resultant angiogenesis.
TaqMan TM Quantitative PCR Analysis-Total RNA was isolated using Qiagen RNeasy mini kit (Qiagen), treated with DNase I (Amplification Grade, Invitrogen), and used at 100 ng per reaction. All the samples were performed in triplicates in each TaqMan TM experiment, and the mRNA levels were normalized to ␤-actin mRNA levels.
Luciferase Assays-293E cells were plated in 24-well plates at subconfluent density and transiently transfected with 160 ng/well pEG-FP.N3 or pEGFP.N3.TM-FOXO4, 36 ng/well pGL3p.VEGF or pGL3p.EPO promoter construct, and 4 ng/well pRLTK vector (Promega), using Effectene transfection reagent (Qiagen). When 16 ng/well pEGFP.N3.TM-FOXO4 was transfected, empty vector was added to equalize the final DNA contents. 9 h after transfection cells were incubated in the presence or absence of 100 M deferoxamine mesylate (DFO; Sigma) for 16 h and then assayed for luciferase activities, which were measured by the dual-luciferase reporter assay system (Promega) as described by the manufacture's protocol. The inducible firefly luciferase activity was normalized to the constitutive Renilla luciferase activity.

The PI3K Signaling Pathway Regulates HIF-1␣ Levels in
HeLa Cells-Studies in glioblastoma, prostate, and breast cancer cell lines have shown that activation of the PI3K/AKT signaling pathway results in increased expression of the HIF-1␣ protein and consequent up-regulation of the VEGF gene transcription (41)(42)(43)(44)(45)(46)(47). Inhibition of the PI3K/AKT activity by the PI3K inhibitors LY294002 and wortmannin, as well as by overexpression of the phospholipid phosphatase PTEN, demonstrated a requirement for activation of the PI3K/AKT pathway in the increase of the HIF-1␣ protein levels. To determine whether the PI3K/AKT activity is also necessary in HeLa cells for the induction of HIF-1␣-regulated genes under hypoxic condition, we treated the cells with LY294002 for 24 h in the presence or absence of a hypoxic mimic, deferoxamine mesylate (DFO), and analyzed the relative levels of VEGF mRNA by TaqMan TM -quantitative PCR (Fig. 1A). DFO highly activated the expression of VEGF transcripts (ϳ6-fold, purple bars), and this increase was significantly inhibited by the addition of the PI3K inhibitor LY294002 (blue bars). The reduction in VEGF transcript levels by LY294002 was clearly associated with a decrease in the level of HIF-1␣ protein expression (Fig. 1B). HeLa cells transfected with a HIF-1␣ expression plasmid showed a decreased HIF-1␣ protein level in the presence of LY294002. This decline was observed even in the presence of DFO, which normally stabilizes the HIF-1␣ protein. Thus, the inhibition of the PI3K/AKT activity by LY294002 blocked DFOinduced up-regulation of the HIF-1␣ protein level. In the absence of DFO, the expression of HIF-1␣ was almost undetectable, and to see the change in the normoxic HIF-1␣ protein levels, the Western blot was subjected to longer exposure (indicated by "long"). Thus, PI3K/AKT activity is required in HeLa cells to induce the expression of the HIF-1␣ protein and allow for increased VEGF transcript levels in response to hypoxia. Similar results were observed when BJAB cells were treated with LY294002 for 24 h (Fig. 1A, data not shown).
Nuclear FOXO4 Regulates HIF-1␣ Protein Levels-Because the forkhead transcription factor FOXO4 is one of the downstream targets of the PI3K/AKT signaling pathway, we examined if FOXO4 was involved in the regulation of the HIF-1␣ protein expression. To test this possibility, we used our previously reported HeLa Tet-on stable cell line, designated 15-14, which could be induced by doxycycline hydrochloride (DOX) to produce a constitutively active form of FOXO4 (9). All three of the AKT phosphorylation sites in this FOXO4 protein were mutated to alanines, allowing it to escape the phosphorylationinduced cytoplasmic sequestration by AKT and localize exclusively in the nucleus. This FOXO4 also contained a FLAG epitope tag at its amino terminus and the green fluorescent protein (GFP) at its carboxyl terminus. LY294002 treatment indicated that this inducible FOXO4 stable cell line 15-14 also required PI3K/AKT activity for hypoxia-induced increase in VEGF transcript levels (Fig. 1A). Nearly 8-fold up-regulation was seen in 15-14 cells when incubated with DFO, and addition of LY294002 suppressed not only up-regulation by DFO but also the basal level of VEGF transcript. Strikingly, induction of FOXO4 in 15-14 by DOX resulted in a dramatic down-regulation of the HIF-1␣ protein levels ( Fig. 2A). This down-regulation occurred even in the presence of DFO, indicating that the stabilizing effect of DFO on the HIF-1␣ protein could be overridden by FOXO4. In the absence of DOX, FOXO4 expression was not induced, and the levels of the transfected HIF-1␣-V5 fusion protein remained relatively constant as long as DFO was present. The down-regulation of the HIF-1␣ protein was not caused by DOX treatment, because the parental HeLa Tet-on cells, which did not express the mutant FOXO4 protein, did not exhibit down-regulation of the HIF-1␣ protein when treated with DOX (Fig. 2B). Transient transfection of the triple mutant FOXO4-GFP fusion protein (TM-FOXO4) into HeLa cells revealed a similar effect of FOXO4 on the expression of the HIF-1␣ protein (Fig. 2C). Cells transfected with a plasmid that expressed the mutant FOXO4-GFP protein showed much lower production of the co-transfected HIF-1␣ protein than cells transfected with a control GFP expression plasmid in both the presence and absence of the DFO. Expression of nuclear FOXO4 also induced the down-regulation of the HIF-1␣ protein under true hypoxic conditions (Fig. 2, D and E). Furthermore, in accordance with previous reports (reviewed in Ref. 24), we observed that hypoxia significantly induced the expression of HIF-1␣ in HeLa Tet-on cells; this up-regulation was unaffected by DOX treatment (Fig. 2E). However, when expression of nuclear FOXO4 was induced by DOX in 15-14 cells the hypoxiainduced up-regulation of HIF-1␣ protein level was completely inhibited (compare the change in HIF-1␣ levels between lanes 7 and 5 to that between lanes 8 and 5). Importantly, although PI3K/AKT activity increases the expression of the HIF-1␣ protein, the FOXO4 transcription factor, being negatively regulated by AKT, induces the down-regulation of the HIF-1␣ protein. Because no change in the HIF-1␣ transcript levels was observed in the mutant FOXO4-inducible 15-14 cells when treated with DOX, it is clear that nuclear FOXO4 does not modulate the transcription of the HIF-1␣ gene (data not shown). The most likely conclusion from these studies is that FOXO4 initiates a transcriptional program that regulates either the stability or the rate of synthesis of the HIF-1␣ protein.
Nuclear FOXO4 Inhibits the Induction of Hypoxia Response Genes-Because FOXO4 negatively regulates HIF-1␣ protein expression, we examined its effect on cellular response to hypoxia. Like the PI3K inhibitor LY294002, FOXO4 strongly suppressed the DFO-induced up-regulation of HIF-1␣-modulated genes (Fig. 3A). In the absence of DOX, exposure to DFO resulted in a significant increase of the endogenous VEGF and   results in a decrease of HIF-1␣ protein levels even in the presence of DFO. Tet-on-inducible FOXO4 cells  were transfected with a HIF-1␣ expression plasmid for 8 -9 h and then cultured in the presence of 100 M DFO overnight, allowing HIF-1␣ protein to accumulate. During the following days at the indicated time (hours) after treatment with or without 2 g/ml DOX in the presence of fresh 100 M DFO, both floating and adherent cells were harvested and analyzed by Western blotting. Anti-FLAG antibody reveals that a significant amount of FOXO4 protein is induced by 4 h of DOX treatment. Both anti-HIF-1␣ and anti-V5 antibodies, detecting the transfected HIF-1␣ protein, show that HIF-1␣ protein levels begin to decline 4 h after DOX treatment and completely disappear by 24 h (right panel). In the absence of FOXO4 induction, HIF-1␣ protein levels remain relatively constant over time in the presence of DFO (left panel). B, in HeLa Tet-on cells DOX treatment does not result in the down-regulation of HIF-1␣ protein levels. The experiment was performed as described in A. C, TM-FOXO4 but not GFP impedes the accumulation of the co-transfected HIF-1␣ protein in both the absence and presence of DFO. HeLa Tet-on cells were transfected with a HIF-1␣ expression plasmid along with the indicated amounts of TM-FOXO4 or GFP expression plasmid. 8 -9 h after transfection, cells were treated with or without DFO for 18 h and then harvested for Western blotting analysis. D, DOX induction of FOXO4 in 15-14 cells results in a decrease of HIF-1␣ protein levels under hypoxic condition (2% oxygen). Experiments were done essentially as described in A with the exception that cells were cultured with 2% oxygen instead of DFO. In the presence of FOXO4 expression, the transfected HIF-1␣.V5 fusion protein disappears by 16 h of DOX treatment as revealed by both anti-HIF-1␣ and anti-V5 antibodies. E, DOX induction leads to the expression of FOXO4 and the down-regulation of the HIF-1␣ protein levels under both normoxic and hypoxic (2% oxygen) conditions in 15-14 but not HeLa Tet-on cells. Cells were transfected with a HIF-1␣ expression plasmid and cultured in normoxic (N) or hypoxic (H) condition with or without DOX overnight; adherent cells were harvested for Western blotting analysis. Phospho-AKT antibodies (␣-P-Ser and ␣-P-Thr) indicated that AKT phosphorylation and thus activity are not significantly increased in hypoxia. The arrow designates the bands corresponding to the FOXO4 protein, and nonspecific bands detected by anti-FLAG antibodies are denoted by an asterisk. and HeLa Tet-on cells were treated with or without 2 g/ml DOX for 9 h and then cultured in the presence or absence of 100 M DFO with or without fresh DOX for 22 h. Total RNA was isolated and analyzed by TaqMan TM -quantitative PCR for the expression of two hypoxia-regulated genes, VEGF and GLUT-1. In 15-14 cells, DOX treatment, which induces the expression of FOXO4 (data not shown), suppresses the induction of both VEGF and GLUT-1 transcript levels by DFO. In contrast, in HeLa Tet-on cells, which do not carry the inducible FOXO4 construct, DOX treatment fails to block the DFO induction of VEGF and GLUT-1. After normalization to ␤-actin mRNA levels, the levels of VEGF and GLUT-1 transcripts in the cells treated with DFO but not DOX were arbitrarily set as 100%. All samples were analyzed in triplicate in each TaqMan TM experiment. Data shown are averages of three independent experiments. Error bars represent standard errors. B, FOXO4 blocks DFO-induced transcriptional activation of VEGF promoter. A 948-bp portion of the VEGF promoter (Ϫ1042 to Ϫ95) was isolated from human genomic DNA by PCR and subcloned upstream of a luciferase reporter gene in the pGL3 promoter vector. The VEGF promoter construct was co-transfected with the indicated amounts of GFP or TM-FOXO4 expression plasmid into 293E cells. 9 h after transfection cells were treated with or without 100 M DFO for 16 h and then assayed for luciferase activities. The dual-luciferase reporter assay system was used to measure the luciferase activities. Relative luciferase activities are shown after normalizing the inducible firefly luciferase activity to the constitutive Renilla luciferase activity. All samples were performed in quadruplicates in each luciferase experiment. Data shown are averages of three independent experiments. Error bars denote standard errors. C, FOXO4 also blocks DFO-induced transcriptional activation of the EPO promoter. A 327-bp portion of the EPO promoter was isolated from human genomic DNA by PCR and subcloned into the pGL3 promoter vector. Luciferase assays after co-transfection of the EPO promoter construct and GFP or TM-FOXO4 expression plasmid and DFO treatment were performed as described in B. Data shown are averages of two independent experiments in which samples were analyzed in quadruplicates. D, FOXO4 induction prevents the increase in VEGF transcript levels in response to hypoxic condition (2% oxygen). Tet-on-inducible stable cells  and HeLa Tet-on cells were treated with or without 2 g/ml DOX for 9 h and then cultured in either normoxic or hypoxic (2% O 2 ) condition with or without fresh DOX for 22 h. Total RNA was purified and analyzed by TaqMan TM -quantitative PCR. Data shown are averages of three independent experiments in which every sample was analyzed in triplicates. Relative VEGF mRNA levels are depicted after normalizing to ␤-actin mRNA levels.
We next determined whether the FOXO4-induced decrease in transcript levels of the HIF-1␣-regulated genes occurred at the level of transcription. Because FOXO4 inhibited the accumulation of the HIF-1␣ protein even in the presence of DFO, we would expect it to prevent the DFO-induced transcriptional activation of the VEGF and EPO promoters. As Fig. 3 (B and C) demonstrates, co-transfection of 160 ng of a control GFP expression plasmid allowed for ϳ2-fold activation of VEGF and EPO promoters in response to DFO treatment. Approximately 50% of this activation was inhibited with only 16 ng of the TM-FOXO4 expression plasmid. Co-transfection of 160 ng of TM-FOXO4 expression plasmid completely abolished the transcriptional activity of the promoters in both the presence and absence of DFO. Finally, to confirm that FOXO4 suppresses not only DFO-induced transcriptional activation but also up-regulation induced by true hypoxic conditions, we cultured 15-14 and HeLa Tet-on cells in 2% oxygen with or without DOX for 22 h and analyzed the relative levels of VEGF transcript by TaqMan TM -quantitative PCR. In the absence of DOX, VEGF transcript levels increased significantly in both 15-14 and HeLa Tet-on cells after treatment with 2% oxygen (Fig. 3D,  purple bars). However, in the presence of DOX, hypoxia induced up-regulation of the VEGF transcript in HeLa Tet-on but not in the FOXO4-expressing 15-14 cells (Fig. 3D, blue bars), not unlike the results obtained using DFO. Thus, FOXO4 induces down-regulation of the HIF-1␣ protein level and suppresses the hypoxia-induced transcriptional activation of HIF-1␣-regulated genes such as VEGF, GLUT-1, and EPO.
Nuclear FOXO4 Regulates HIF-1␣ Protein Levels through Protein Degradation Processes-Treatment with the pancaspase inhibitor zVAD-fmk revealed that the down-regulation of the HIF-1␣ protein expression was not merely a result of global protein degradation caused by FOXO4-induced apoptosis (Fig. 4A). The inclusion of zVAD-fmk, which blocks all caspase activity and resultant apoptosis, did not dramatically modify the rate of FOXO4-induced down-regulation of the HIF-1␣ protein. In both the presence and absence of zVAD-fmk, the HIF-1␣ protein levels began to decrease 4 h after DOX treatment and completely disappeared by 24 h (compare the right panel in Fig. 4A to the right panel in Fig. 2A). Thus, FOXO4-induced down-regulation of the HIF-1␣ protein does not require caspase activation and hence is not a consequence of apoptosis.
To elucidate the mechanism of FOXO4-induced down-regulation of the HIF-1␣ protein, we first determined whether the down-regulation was mediated by the proteasome pathway. Using a cell-permeable, irreversible proteasome inhibitor, clastolactacystin ␤-lactone, we found that FOXO4-induced downregulation of the HIF-1␣ protein required proteasome activity (Fig. 4B). In the absence of the proteasome inhibitor, FOXO4 induction resulted in a significant decrease in HIF-1␣ protein as seen previously (top left panel). Inclusion of clastolactacystin ␤-lactone allowed for the accumulation of the HIF-1␣ protein to levels similar to that observed in the absence of FOXO4 induction (compare the top right panel with the bottom panels in Fig.  4B). This result showed that the proteasome pathway was involved in FOXO4-induced down-regulation of the HIF-1␣ protein. Although it does not rule out the possibility that the FIG. 4. FOXO4-induced down-regulation of the HIF-1␣ protein requires proteasome activity. A, blocking caspase activities by the addition of 25 M zVAD-fmk (ZVAD) does not prevent the down-regulation of HIF-1␣ protein levels by FOXO4 induction. The experiment was performed essentially as previously described in Fig. 2A except zVAD-fmk was also included in the cultures. B, in the absence of clastolactacystin ␤-lactone (clasto), a cell-permeable, irreversible proteasome inhibitor, DOX induction of FOXO4 leads to the down-regulation of the HIF-1␣ protein (top left panel). This down-regulation is attenuated when the proteasome inhibitor is included in the culture (right top panel). In the absence of DOX-induced expression of FOXO4, the HIF-1␣ protein remains stable with or without clastolactacystin ␤-lactone (bottom panels). The Tet-on-inducible FOXO4-stable cells  were transfected with a HIF-1␣ expression plasmid for 8 -9 h and then cultured overnight with DFO. The time course was initiated when cells were exposed to media containing Me 2 SO or 20 M of clastolactacystin ␤-lactone with or without DOX. Fresh DFO was also included in the cultures. At the indicated times (hours), cells were harvested and analyzed by Western blotting. The arrow designates the bands corresponding to the FOXO4 protein, and nonspecific bands detected by anti-FLAG antibodies are denoted by an asterisk. FOXO4 Down-regulates HIF-1␣ Independent of pVHL decrease in the steady-state levels of the HIF-1␣ protein induced by FOXO4 was a consequence of a reduced rate of protein synthesis, the requirement for proteasome activity indicates that FOXO4-induced down-regulation of HIF-1␣ was mediated at least in part by a proteolytic process. Similar results were observed using another proteasome inhibitor, MG-132 (data not shown).
Nuclear FOXO4 Induces the Down-regulation of HIF-1␣ in a pVHL-independent Manner-An elegant series of experiments have shown that the von Hippel-Lindau (VHL) tumor suppressor protein is the recognition component of an E3 ubiquitin ligase complex that targets HIF-1␣ for proteasome-mediated degradation (28 -32). Furthermore, for the interaction between HIF-1␣ and pVHL to occur, the HIF-1␣ protein must be hydroxylated at proline residues 402 and 564 by one of three prolyl hydroxylase domain proteins (PHD1, PHD2, and PHD3 encoded by EGLN2, EGLN1, and EGLN3, respectively) (36 -40). Because FOXO4 induces the down-regulation of the HIF-1␣ protein in a proteasome-dependent manner, we examined if it is involved in the transcriptional activation of the VHL, EGLN2, EGLN1, and EGLN3 genes. If FOXO4 could activate transcription of any of these genes, excess expression of these gene products could potentially lead to the proteolysis of HIF-1␣. Using TaqMan TM -quantitative PCR, we found no change in the transcript levels of any one of these genes in DOX-induced 15-14 cells (data not shown). Western blotting also revealed that the VHL protein level remained relatively constant in the presence of increasing FOXO4 expression (Fig.  5A). In addition, although the EGLN1 promoter region appeared to contain several consensus FOXO4-binding sites, luciferase reporter experiments with the EGLN1 promoter provided no conclusive evidence that FOXO4 up-regulates EGLN1 promoter activity (data not shown). Thus, FOXO4 does not activate VHL, EGLN2, EGLN1, or EGLN3 transcription, suggesting that FOXO4 may induce the down-regulation of HIF-1␣ protein via an alternative mechanism independent of pVHL.
To determine if pVHL is required for FOXO4-induced downregulation of HIF-1␣, we examined if FOXO4 could still induce HIF-1␣ down-regulation in the presence of dimethyloxalylglycine, a potent inhibitor of prolyl hydroxylases (37). Because pVHL only recognizes and interacts with HIF-1␣ that has been hydroxylated on proline residues 402 and 564, blocking proline hydroxylation with dimethyloxalylglycine will prevent HIF-1␣ from interacting with pVHL and inducing its degradation. If FOXO4 failed to induce HIF-1␣ down-regulation in the presence of the hydroxylase inhibitor, it would indicate that interaction with pVHL was still necessary for FOXO4-induced down-regulation of HIF-1␣, even though VHL was not transcriptionally regulated by FOXO4. Strikingly, we found that FOXO4-induced down-regulation of HIF-1␣ still occurred in the presence of dimethyloxalylglycine (Fig. 5B), indicating that proline hydroxylation and hence interaction with pVHL were not required. In the absence of DFO, the transfected HIF-1␣ protein, detected by both anti-HIF-1␣ and anti-V5 antibodies, was almost completely degraded in HeLa Tet-on cells, presumably by an endogenous pVHL-mediated pathway (lanes 5 and 7). As expected for pVHL-mediated degradation, the inclusion of dimethyloxalylglycine prevented the degradation and allowed the HIF-1␣ protein to accumulate in HeLa Tet-on cells, regardless of DOX treatment (lanes 6 and 8). The expression of HIF-1␣ was also increased with the addition of dimethyloxalylglycine in 15 3 to lane 1). Although the addition of dimethyloxalylglycine resulted in some accumulation of the HIF-1␣ protein, the levels were significantly lower when FOXO4 was expressed (compare lane 4 to lane 2).
To further confirm that proline hydroxylation on HIF-1␣ was not essential for FOXO4-induced down-regulation of the HIF-1␣ protein, we mutated proline residue 402 to alanine and proline 564 to glycine and expressed this mutant HIF-1␣ protein in 15-14. If the observed down-regulation were due to pVHL, then the absence of these prolines would block the hydroxylation of HIF-1␣ protein and subsequent proteasomemediated degradation. Incidentally, this mutant form of HIF-1␣ has previously been shown to be stable under normoxic conditions and exhibit no pVHL-dependent ubiquitination (38). As shown in Fig. 5C, this mutant HIF-1␣ was still downregulated with similar kinetics as the wild-type protein in the presence of nuclear FOXO4 expression. Thus, FOXO4-induced down-regulation of HIF-1␣ did not require hydroxylation on the two consensus proline residues, an essential prerequisite for interaction with pVHL. Furthermore, consistent with our finding that FOXO4 induces proteasome-mediated down-regulation of the HIF-1␣ protein by a pVHL-independent mechanism, LY294002 treatment promoted the down-regulation of the HIF-1␣ protein in the absence of functional VHL protein.
Blocking the PI3K/AKT activity in the VHL-deficient renal carcinoma cell line RCC4 with LY294002, which would lead to the activation of FOXO4-type forkhead transcription factors, resulted in a significant reduction in the endogenous HIF-1␣ protein levels (Fig. 6). DISCUSSION A large body of literature indicates a link between the PI3K/ AKT signaling pathway and angiogenesis (41)(42)(43)(44)(45)(46)(47)(48). This link is logical for growth and maintenance of cells and tissues. It is likely that during periods of growth, for example in the presence of various factors that will activate the PI3K/AKT pathway, there should be enhanced angiogenesis to provide for the expanding tissue volume. In contrast, during conditions of nutrient deprivation, this pathway would be down-regulated, resulting in decreased or static vasculature, again a clearly advantageous situation for the organism. Thus, it is likely that the regulation of angiogenesis by PI3K/AKT modulation evolved to enable the organism to carefully regulate angiogenesis by sensing both external nutrient as well as oxygen levels.
The data reported herein provide a mechanistic explanation for the link between the PI3K/AKT signaling pathway and angiogenesis. Previous studies have demonstrated that activation of the PI3K pathway results in induction of a kinase cascade that eventually mediates the phosphorylation of a subfamily of forkhead transcription factors that includes FOXO4 (also known as AFX), FOXO1 (FKHR), and FOXO3a (FKHR-L1). Phosphorylated forms of these factors are sequestered in the cytoplasm, effectively inhibiting their transcriptional function (3)(4)(5)(6)(7)(8)10). Furthermore, this family of transcription factors was found to activate the expression of cell cycle regulatory and apoptotic genes (9,(11)(12)(13)(14)(15)(16)(17). In this report, we demonstrate that FOXO4 nuclear localization induces the down-regulation of the critical angiogenic mediator, HIF-1␣. Thus, in addition to suspending cell cycle progression and promoting apoptosis, the FOXO4-type forkhead transcription factors also likely play a role in negatively regulating angiogenesis. As with many other functions controlled by the PI3K/AKT pathway, angiogenic regulation appears to be an important output in response to environmental changes. To effectively stimulate angiogenesis in times of growth, the activated PI3K/AKT pathway prevents nuclear localization of the FOXO4-type forkhead transcription factors and hence FOXO4-induced down-regulation of the HIF-1␣ protein. However, under conditions of reduced nutrient  5. Down-regulation of the HIF-1␣ protein by FOXO4 appears to be independent of the pVHL pathway. A, the VHL protein levels are not affected by FOXO4 induction in the Tet-on-inducible FOXO4-stable cells . After growing in the presence or absence of DOX for the indicated times, the FOXO4-stable cells were lysed and analyzed by Western blotting. Anti-FLAG antibody reveals that FOXO4 is induced by DOX by 4 h and accumulates to high levels in 24 h. However, the VHL protein levels remain relatively constant in this time range even in the presence of DOX. B, inhibition of prolyl hydroxylase activity does not block FOXO4-induced down-regulation of the HIF-1␣ protein levels. After transfection with a HIF-1␣ expression plasmid for 8 h, HeLa Tet-on cells and Tet-on-inducible FOXO4 stable cells  were cultured in 100 M DFO overnight, allowing for the accumulation of the HIF-1␣ protein. During the following day, cells were then treated with or without 2 g/ml DOX in the presence of DFO for 4 h. At the end of 4 h, cells were washed once with medium to remove DFO and then exposed to Me 2 SO or 1 mM dimethyloxalylglycine, a potent prolyl hydroxylase inhibitor, with or without DOX for 6 h. Both floating and adherent cells were harvested and analyzed by Western blotting. In HeLa Tet-on cells, the addition of dimethyloxalylglycine prevents the degradation of HIF-1␣ in the absence of DFO, regardless of the presence or absence of DOX. However, in the FOXO4-stable cells, down-regulation of HIF-1␣ still occurs in the presence of dimethyloxalylglycine when DOX is included. C, mutations of the HIF-1␣ proline residues, Pro-402 and Pro-564, which mediate the interaction between HIF-1␣ and pVHL when hydroxylated, do not affect the susceptibility of the HIF-1␣ protein to FOXO4-induced down-regulation. In the absence of DOX and, hence, no FOXO4 expression as indicated by anti-FLAG, both wild-type and mutant HIF-1␣ proteins remain relatively stable (left panels) as shown by anti-HIF-1␣ and anti-V5 antibodies. In the presence of DOX, FOXO4 expression, induced by 4 h, is followed by the down-regulation of both wild-type and proline mutant HIF-1␣ proteins (right panels), indicating that prolyl hydroxylation and hence interaction with pVHL is not necessary for FOXO4-induced down-regulation of HIF-1␣. The Tet-on-inducible FOXO4-stable cells  were transfected with either a wild-type HIF-1␣ (HIF-1␣.V5) or a double proline mutant HIF-1␣ (HIF-1␣.P402A/P564G.V5) expression plasmid for 8 -9 h and then cultured in the presence of DFO overnight to allow for the accumulation of the HIF-1␣ proteins. At the start of the time course, cells were exposed to media containing DFO with or without DOX and at the indicated times (hours), both floating and adherent cells were harvested and analyzed by Western blotting. Arrowheads designate the bands corresponding to the FOXO4 protein, and nonspecific bands detected by anti-FLAG antibodies are denoted by an asterisk. levels, the inactivation of the PI3K/AKT pathway enables FOXO4 to enter the nucleus and activate one or more transcriptional programs that induce the down-regulation of the HIF-1␣ protein, slowing down the process of vasculature expansion. Thus, the PI3K/AKT/FOXO4 signaling pathway regulates the expression of the HIF-1␣ protein in response to nutrient (i.e. growth factor) levels, which results in increasing HIF-1␣ protein levels and permitting angiogenesis to occur only during nutrient abundance.
Interestingly, in accordance with previous reports (41,47), our results indicate that neither hypoxia nor the hypoxic mimic deferoxamine mesylate (DFO) increases AKT activity as measured by phosphorylation at Thr-308 and Ser-473 (Figs. 1B and 2E), even though the AKT activity is necessary for the HIF-1␣ protein expression and hypoxia response. We reason that the PI3K/AKT pathway regulates the HIF-1␣ protein expression in response to nutrient levels regardless of oxygen concentrations; AKT activity is, however, still necessary under hypoxia to inhibit the negative effect of FOXO4 on HIF-1␣ protein expression and hypoxic response, because our data here indicate that the stabilizing effect of DFO and hypoxia on the HIF-1␣ protein can be overridden by FOXO4. An alternative explanation for the observed lack of increase in AKT phosphorylation under hypoxia/DFO treatment is that the cells were cultured in complete media containing serum, which already maintained PI3K/AKT in the active state. Data in Figs. 1B and 2E showed that anti-phospho-AKT antibodies detected a significant amount of phosphorylation on AKT in the absence of DFO and under normoxic conditions. Thus, hypoxia/DFO treatment would not further increase the level of AKT phosphorylation.
Unfortunately, as with other important functions that maintain cellular viability, the PI3K/AKT pathway can be advantageously utilized by tumors to enhance their growth and survival. In contrast to other oncogenic pathways, the PI3K/AKT pathway can be activated by the mutational loss of PTEN, the major lipid phosphatase involved with the down-regulation of the enzymatic activity of PI3 kinases. The data reported here suggest that the loss of PTEN might provide the tumor with a reasonably simple method to stimulate angiogenesis, that is, by shutting down the transcriptional programs controlled by the FOXO4-type forkhead transcription factors. Rapidly growing tumors require a highly enhanced angiogenic response, and previous work has demonstrated that the mutational loss of the VHL proteins results in the up-regulation of the HIF-1␣ protein and subsequent increased production of angiogenic factors such as VEGF (28,34,35). The work reported here provides for an additional mechanism that allows for increased angiogenesis by stimulation of the PI3K/AKT pathway and resultant inhibition of the transcriptional activity of forkhead-type transcription factors such as FOXO4. Thus, in contrast to the mutation loss of VHL that results in the up-regulation of the HIF-1␣ protein levels, the mutation loss of PTEN leads to the increase in PI3K/AKT activity, which not only induces enhanced angiogenesis but also provides for a diversity of antiapoptotic responses, including the inhibition of the expression of pro-apoptotic genes, cell cycle inhibitory genes, and transcriptional repressors such as BCL-6 (8,9,(11)(12)(13)(14)(15)(16)(17)(42)(43)(44)(45). This is likely the reason why a diversity of late stage tumors upregulates the PI3K/AKT pathway via either the loss of PTEN and/or the up-regulation of upstream activators such as HER 2 and RAS (2).
Our data demonstrate that nuclear FOXO4 induces not only the down-regulation of the HIF-1␣ protein under hypoxia and in the presence of hypoxic mimic deferoxamine mesylate but also the down-regulation of a hyperstable form of the HIF-1␣ protein that cannot be proline-hydroxylated and suggest that FOXO4 can override the hypoxia-induced stabilization of HIF-1␣. This finding reveals that the FOXO4-mediated regulation of HIF-1␣ described here is a critical component of the angiogenic response, because FOXO4 can clearly cause a decrease of HIF-1␣ protein even under low oxygen concentrations such as might be found in rapidly growing tumors. Under hypoxic conditions, the lack of interaction between HIF-1␣ and the VHL tumor suppressor protein should be sufficient to inhibit the degradation of the HIF-1␣ protein by the pVHL-mediated proteasome pathway, resulting in the stable expression of the HIF-1␣ protein. However, the PI3K/AKT signaling pathway is still required under low oxygen conditions for the accumulation of the HIF-1␣ protein and hence the induction of HIF-1␣regulated genes. Our results provide a molecular explanation for this requirement by demonstrating that PI3K/AKT activity is necessary under hypoxic conditions to inhibit FOXO4 nuclear localization and the activation of one or more FOXO4mediated transcriptional programs that induce the degradation of the HIF-1␣ protein in a pVHL-independent manner. Thus, inhibitors of kinases in the PI3K/AKT pathway might be more beneficial as cancer therapeutics than previously hypothesized, because inhibition of the PI3K/AKT pathway will not only induce a diversity of apoptotic pathways but will also inhibit processes, such as angiogenesis and erythropoiesis, that are essential for tumor progression.
Because FOXO4-induced down-regulation of HIF-1␣ is not dependent on the pVHL pathway but still requires proteasome activity, we speculate that another E3 ubiquitin ligase could be up-regulated by FOXO4 nuclear localization and function to target HIF-1␣ for ubiquitination and subsequent proteasomemediated proteolysis. Interestingly, examination of the transcripts modulated by FOXO4 induction revealed that SMURF2, an E3 ubiquitin ligase known to regulate the stability of SMAD transcription factors as well as transforming growth factor ␤ receptors (49 -54), was significantly up-regulated in response to nuclear FOXO4 expression. 2 Furthermore, excess expression of wild-type but not catalytically inactive mutant SMURF2 prevented the accumulation of the HIF-1␣ protein in HeLa cells, suggesting a role of SMURF2 in the FOXO4 Down-regulates HIF-1␣ Independent of pVHL FOXO4-induced down-regulation of HIF-1␣. 3 Additional experiments are required to determine whether SMURF2 directly targets HIF-1␣ for ubiquitination and proteasome-mediated degradation.
Other mechanisms besides ubiquitination/proteasome-mediated proteolysis may also play a role in controlling the steadystate level of HIF-1␣. For example, it has been shown that the PI3K/AKT signaling pathway regulates HIF-1␣ protein level by controlling FK506 binding protein 12 rapamycin-associated protein 1 (FRAP) kinase whose known downstream targets include regulators of protein translation, p70s6 kinase, and eIF-4E binding protein (41,47). Thus, PI3K/AKT can also regulate the rate of HIF-1␣ translation. Although the result from our proteasome inhibitor experiment indicated a requirement for proteasome activity and hence a proteolytic process in the FOXO4-induced down-regulation of HIF-1␣, it did not rule out the possibility that FOXO4 activates a negative regulator of protein translation. Indeed, our survey of FOXO4-modulated genes showed that the level of ribosomal protein L7 was increased in response to nuclear FOXO4 expression. 2 Ribosomal protein L7 has been implicated in translational control in eukaryotic cells by its association with ribosomes and its ability to inhibit protein synthesis (55). It will be interesting to see whether excess expression of ribosomal protein L7 induced by FOXO4 is responsible for the observed down-regulation of HIF-1␣ by interfering with the synthesis of the HIF-1␣ protein.
Finally, we have also found that FOXO4 activates the expression of aryl hydrocarbon receptor (AHR) and p35srj. 2 Like HIF-1␣, AHR dimerizes with HIF-1␤, and p35srj interacts with p300/CBP (56 -59). Thus, excess expression of these two gene products may also lead to the down-regulation of HIF-1␣ by titrating away the binding partners of HIF-1␣ and exposing it to degradation. Further experimentation is necessary to elucidate the molecular mechanism(s) of FOXO4-induced downregulation of the HIF-1␣ protein.
Our data suggest that FOXO4-induced down-regulation of the HIF-1␣ protein level is responsible for the inhibition of hypoxia-activation of VEGF, EPO, and GLUT-1 transcription by FOXO4, because all three genes are commonly regulated by HIF-1␣. We cannot, however, rule out the possibility that other factors modulated by FOXO4 are involved in the FOXO4-mediated suppression of the hypoxia response. We and others have found that large number of genes are modulated by the FOXO4-type forkhead transcription factors (17). 2 At the present time we can only speculate on what genes might be involved in FOXO4-mediated regulation of hypoxia response and angiogenesis. For example, the up-regulation of AHR and p35srj by FOXO4 2 may interfere with the transcriptional activity of HIF-1 by competing away the necessary cofactors, HIF-1␤ and p300/CBP, respectively (59 -61). Another possible candidate is the Id1 protein, an inhibitor of a group of basic helix-loop-helix transcription factors. Id1 has been found to play an essential role in supporting tumor angiogenesis (62)(63)(64). Loss of Id proteins results in a reduction of VEGF expression and angiogenic defects in tumors (63). Furthermore, Volpert et al. (64) demonstrated that Id1 regulates angiogenesis through transcriptional repression of the angiogenesis inhibitor thrombospondin-1. Interestingly, we found that Id1 expression was reduced ϳ2-fold as a result of nuclear FOXO4 expression, 2 consistent with FOXO4 playing a potential role in negatively regulating angiogenesis. Ramaswamy et al. (17) also showed that Id1 transcript level was modulated by FOXO1. In summary, it is clear that FOXO4 affects the expression of the HIF-1␣ protein and inhibits VEGF, EPO, and GLUT-1 tran-scriptional activation in response to DFO and hypoxia. It will require additional experiments to determine which of the one or more FOXO4-modulated genes are essential factors in FOXO4-mediated regulation of HIF-1␣ levels during the hypoxia response.
In closing, the HIF-1␣ transcription factor seems particularly adept at utilizing a diversity of degradative mechanisms, including p53/MDM-and acetylation-regulated pathways (65,66) as well as transcriptional and translational mechanisms. Our results demonstrate that modulation of FOXO4 transcriptional activity is a novel and important regulator of HIF-1␣ stability and angiogenesis. Therefore, the PI3K/AKT signaling pathway induces angiogenic response not only by increasing the rate of HIF-1␣ protein synthesis, but also by preventing FOXO4 from entering the nucleus and initiating a set of transcriptional programs that result in the down-regulation of the HIF-1␣ protein levels and suppression of cellular response to hypoxia.