p42/p44 Mitogen-activated Protein Kinases Phosphorylate Hypoxia-inducible Factor 1α (HIF-1α) and Enhance the Transcriptional Activity of HIF-1*

Hypoxia-inducible factor-1 (HIF-1) controls the expression of a number of genes such as vascular endothelial growth factor and erythropoietin in low oxygen conditions. However, the molecular mechanisms that underlie the activation of the limiting subunit, HIF-1α, are still poorly resolved. Results showing that endogenous HIF-1α migrated 12 kDa higher than in vitrotranslated protein led us to evaluate the possible role of phosphorylation on this phenomenon. We report here that HIF-1α is strongly phosphorylated in vivo and that phosphorylation is responsible for the marked differences in the migration pattern of HIF-1α. In vitro, HIF-1α is phosphorylated by p42 and p44 mitogen-activated protein kinases (MAPKs) and not by p38 MAPK or c-Jun N-terminal kinase. Interestingly, p42/p44 MAPK stoichiometrically phosphorylate HIF-1α in vitro, as judged by a complete upper shift of HIF-1α. More importantly, we demonstrate that activation of the p42/p44 MAPK pathway in quiescent cells induced the phosphorylation and shift of HIF-1α, which was abrogated in presence of the MEK inhibitor, PD 98059. Finally, we found that in a vascular endothelial growth factor promoter mutated at sites previously shown to be MAPK-sensitive (SP1/AP2–88-66 site), p42/p44 MAPK activation is sufficient to promote the transcriptional activity of HIF-1. This interaction between HIF-1α and p42/p44 MAPK suggests a cooperation between hypoxic and growth factor signals that ultimately leads to the increase in HIF-1-mediated gene expression.

The growth of new blood vessels in the adult is termed angiogenesis. Angiogenesis occurs in natural situations such as the female reproductive cycle, or in pathological situations such as wound healing, retinopathy, tumor proliferation, and metastasis. In the latter cases, these events share a common characteristic of occurring in a hypoxic environment.
Recently, three independent laboratories succeeded in the knockout of the mouse HIF-1␣ gene (HIF-1␣ Ϫ/Ϫ ) (19 -21). They observed that HIF-1␣ Ϫ/Ϫ mice were not viable (death occurring at embryonic day 10.5). They also showed that HIF-1␣ Ϫ/Ϫ mice were deficient in vascularization and had cardiac and neuronal abnormalities. Tumors formed from HIF-1␣ Ϫ/Ϫ ES cells were retarded in growth, and vascularization was absent. These results clearly show the irrevocable role that HIF-1␣ plays in neovascularization and embryogenesis. A large quantity of information concerning HIF-1␣ has been published in a very short period. However, molecular signals underlying activation of HIF-1␣ are still unknown. Very little is known about the post-translational modifications of HIF-1␣ and, more precisely, the possible role of phosphorylation. In this work, we show that HIF-1␣ is highly phosphorylated in vivo and that HIF-1␣ phosphorylation induces strong changes in its electrophoretic migration pattern. We also show that, in vitro, p42/p44 MAPK duplicate this phosphorylation. More interestingly, direct activation of the p42/p44 MAPK pathway in quiescent cells induces the same striking shift in molecular mass. Finally, this activa-tion of p42/p44 MAPK can promote the transcriptional activation of HIF-1.

MATERIALS AND METHODS
Cell Culture-The established Chinese hamster fibroblast cell line CCL39 and their corresponding transfected cells, HA-p44 MAPK, HA-p38 MAPK, and HA-JNK cells, were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 7.5% fetal calf serum (FCS), penicillin (50 units/ml), and streptomycin (50 g/ml) (Life Technologies, Inc.). Mouse embryo fibroblasts were cultured on gelatin-coated culture plates in the same medium as above. Raf-1:ER clonal cell line (clone 19) is a derivative of the CCL39 cell line, which stably expresses a fusion protein between the catalytic domain of Raf-1 and the ligand binding domain of the estrogen receptor (22,23). This clone allows the exclusive activation of the p42/p44 MAPK pathway upon estradiol treatment, and its characteristics have been previously described (24). These cells were cultured in the same medium as described above in the absence of phenol red in order to reduce the basal activity of the Raf-1:ER chimera. Human 293, HeLa, and HepG2 cell lines were cultured in DMEM containing 7.5% heat-inactivated FCS. Mouse endothelial 1G11 cells were cultured on gelatin-coated plates in the DMEM medium supplemented with 20% FCS, 100 g/ml heparin, and 100 g/ml endothelial cell growth supplement (25). Cell growth was arrested by total deprivation of serum for 16 -20 h if not otherwise indicated. Hypoxic conditions were obtained placing the cells in a sealed "Bug-Box" anaerobic workstation (Ruskinn Technologies, Leeds, UK/Jouan, Saint Herblain, France). The oxygen in this workstation was maintained at 1-2% with the residual gas mixture being 93-94% nitrogen and 5% carbon dioxide.
Plasmids-The hemagglutinin (HA)-tagged HIF-1␣ and HIF-1␤ were cloned by reverse transcription-PCR. Briefly, poly(A) RNA was extracted from hypoxic 293 cells with the use of the mRNA isolation kit (Roche Molecular Biochemicals) and reverse transcribed using the Expand reverse transcriptase system (Roche Molecular Biochemicals). HIF-1␣ and HIF-1␤ were amplified with the Expand high fidelity PCR system (Roche Molecular Biochemicals) using the primers: sense 5Ј-A-TGGAGGGCGCCGGCGGCGAG-3Ј and antisense 5Ј-GTTAACTTGAT-CCAAAGCTCTGAG-3Ј for HIF-1␣ and sense 5Ј-TGGCGGCGACTACT-GCCAACCCC-3Ј and antisense 5Ј-TTCTGAAAAGGGGGGAAAC-3Ј for HIF-1␤. Conditions for PCR amplification were: 35 cycles with 30 s at 95°C, 1 min at 55°C, 2.5 min at 72°C, and a last cycle of elongation at 72°C for 10 min. Blunt ended fragments were 3Ј A-tailed with Taq DNA polymerase, purified, and ligated into the PCR fragment cloning vector pTag (Ingenius; R&D Systems). To construct the HA-tagged forms of HIF-1␣ and HIF-1␤, a new PCR reaction was performed using the p-Tag constructions as templates and the same sense (initiation codons ATG were mutated to GAG) and antisense primers (containing a SmaI and a XbaI restriction site, respectively) previously utilized. The PCR products were digested with SmaI and XbaI and subcloned into the pECE/HA expression vector (26). Finally, the two tagged forms were subcloned into the pcDNA3 expression vector (Invitrogen). The complete sequence was verified by Eurogentec (Liège, Belgium). The luciferase reporter used in these experiments is a modified version of the original VEGF promoter gene construct that was generously provided by Dr. Werner Risau. Two SP1 and one AP2 binding sites, which have been recently implicated in p42/p44 MAPK-mediated VEGF expression, were point mutated as previously reported (6). GST-ATF2 construct was a generous gift from Dr. Roger Davis.
Transient Transfection and Luciferase Assay-Raf-1:ER cells or parental CCL39 cells in 12-well plates (1.5 ϫ 10 5 cells/well) were transiently transfected by CaPO 4 precipitation. 100 ng/well reporter plasmid were used along with the indicated concentrations of expression vector, 100 ng/well virus ␤-galactosidase as a control for transfection efficiency and the empty vector, pcDNA3, to normalize the quantity of DNA transfected to a total of 1 g/well. Four hours after transfection, cells were FCS-starved for 5 h, followed by stimulation for 16 h. Cells were then washed twice with cold PBS, and luciferase assays were performed as follows. Cells were lysed in a lysis buffer (25 mM Trisphosphate (pH 7.8), 2 mM DTT, 2 mM 1,2-diaminocyclohexane-N,N,NЈ,NЈ-tetraacetic acid, 10% glycerol, and 1% Triton X-100) for 15 min at room temperature, and the lysate was cleared by centrifugation. Antibodies-Anti-HIF-1␣ antiserum 2087 was raised in rabbits immunized against the last 20 amino acids of the C-terminal end of human HIF-1␣. Mouse monoclonal anti-hemagglutinin (HA) antibody 12CA5 was from BabCO (Richmond, CA). Rabbit anti-phospho-p44/p42 MAPK polyclonal antibody was from New England Biolabs and horseradish peroxidase-coupled anti-mouse and anti-rabbit antibody were from Promega. Rabbit p42 MAPK antiserum E1B4 was produced and characterized in our laboratory (48).
In Vitro Phosphorylation-Bacterially expressed and purified active p42 MAPK was generously provided by Dr. Melanie Cobb (27). HAtagged p44 MAPK, p38 MAPK, and JNK were obtained by immunoprecipitation from stably expressing cell lines developed in our laboratory (28). HA-HIF-1␣ or HA-HIF-1␤ were translated in vitro using the TNT-coupled reticulocyte lysate system (Promega). For radiolabeled HIF-1␣, translation was performed in the presence of 40 Ci of [ 35 S]methionine (Amersham Pharmacia Biotech). For unlabeled HIF-1␣, translation was performed in the presence of 20 M L-methionine (Sigma). 5 l of the translation mixture (for each condition) were immunoprecipitated using anti-HA antibody and used as substrate. The kinase assay was performed in a buffer containing 20 mM Hepes, 10 mM MgCl 2 , 0.5 mM DTT, 5 mM p-nitrophenyl phosphate, 50 M ATP with or without 2 Ci of [␥-32 P]ATP for the desired time at 30°C. Proteins were then resolved in SDS-polyacrylamide gels (7.5%) and revealed by autoradiography and Western blotting.
Dephosphorylation Assay-Endogenous HIF-1␣ from HeLa cells was used as substrate. HIF-1␣ was immunoprecipitated using an anti-HIF-1␣ antiserum and washed three times with the previously mentioned lysis buffer. Immunoprecipitates were then washed twice with the phosphatase buffer supplied by the manufacturer and incubated with phosphatase (New England Biolabs) for 1 h at 37°C. Proteins were resolved in SDS-polyacrylamide gels (7.5%) and revealed by Western blotting.

HIF-1␣ Is Strongly Modified by Phosphorylation-
We developed a high affinity anti-HIF-1␣ antibody that specifically allows us to evaluate HIF-1␣ expression in vitro and in whole cell extracts. Human HIF-1␣, translated in vitro in a rabbit reticulocyte system, migrated in a 7.5% SDS-polyacrylamide gel as a sharp band, which corresponded to a molecular mass of 104 kDa (Fig. 1). We then compared the in vitro expressed protein to endogenous HIF-1␣ from different exponentially growing cell lines. In all the cases tested, HIF-1␣ was strongly induced by hypoxia. The migration patterns of the endogenous HIF-1␣ are different between cell lines. However, one common occurrence is noted: induced HIF-1␣ migrated with a very diffused pattern (104 -116 kDa) as compared with the in vitro translated protein (Fig. 1). Since very few post-translational modifications can occur in vitro, these results suggested that HIF-1␣, expressed in a whole cell system, undergoes strong posttranslational modifications.
It has often been demonstrated that phosphorylation can markedly modify the migration pattern of proteins in SDSpolyacrylamide gels. Therefore, we wanted to evaluate whether these changes in electrophoretic migration (about 12 kDa) were due to phosphorylation. Endogenous HIF-1␣ from HeLa cells was dephosphorylated with the use of a nonspecific protein phosphatase, phosphatase. As shown in Fig. 2, when HeLa cells were incubated in hypoxic conditions and HIF-1␣ was immunoprecipitated with anti-HIF-1␣ antibody and revealed with the same antibody, hypoxia strongly induced HIF-1␣, which migrated as two distinct bands: the first at 104 kDa and the second at 116 kDa. Incubation of immunoprecipitated HIF-1␣ with phosphatase drastically decreased the molecular mass of HIF-1␣ to a value that corresponded exactly to in vitro translated HIF-1␣, i.e. 104 kDa. These results show that HIF-1␣ is phosphorylated in vivo and that phosphorylation is responsible for increase in HIF-1␣'s apparent molecular mass.
p42/44 MAPK Can Phosphorylate HIF-1␣ in Vitro-We next wanted to identify a kinase capable of phosphorylating HIF-1␣ and inducing this 12-kDa shift in molecular mass. Results from our laboratory showed that treatment of CCL39 cells with the potent serine/threonine phosphatase inhibitor okadaic acid led to a clear increase in HIF-1␣'s molecular mass (data not shown). This suggested that serine/threonine protein phosphatase(s) dephosphorylate(s) HIF-1␣ and therefore, serine/ threonine protein kinase(s) should be responsible for the mobility shift of HIF-1␣.
p42/44 MAPK are two serine/threonine protein kinases activated by mitogenic stimulation and modulate the activity of a number of transcription factors (for review, see Ref. 29). Recently, Elk-1 was shown to be phosphorylated during hypoxic conditions by MAPK, inducing the transcription of the c-fos gene (30). We therefore wanted to evaluate the capacity of p42/p44 MAPK to phosphorylate HIF-1␣. HIF-1␣ was translated in vitro with [ 35 S]methionine and incubated for different periods of time with active p42 MAPK. Surprisingly, incubation of HIF-1␣ with p42 MAPK induced a rapid shift in the electrophoretic mobility of HIF-1␣ with a t 1/2 of 5 min and a complete shift of the molecule at 20 min (Fig. 3). The molecular mass of the shifted HIF-1␣ was strikingly similar to the uppermost band seen for endogenous HIF-1␣ expressed in human cells, i.e. 116 kDa. As a control, HIF-1␤ was also translated in vitro and incubated with p42 MAPK. In contrast to HIF-1␣, the migration of HIF-1␤ was not modified by p42 MAPK (Fig. 3). These results suggest that HIF-1␣ is phosphorylated in vitro by p42 MAPK and, more importantly, that this phosphorylation is able to reproduce the SDS gel mobility shift observed in vivo.
To this point, we have evaluated the phosphorylation of in vitro translated HIF-1␣ by p42 MAPK. However, we next wanted to examine whether this phosphorylation was reproducible on endogenous HIF-1␣ expressed in a intact cell system. Therefore, we performed phosphorylation assays with active p42 MAPK on phosphatase dephosphorylated HIF-1␣ from HeLa cells. As previously shown in Fig. 2, dephosphorylated HIF-1␣ was shifted to a lower molecular mass of approximately 104 kDa, which corresponds to the unmodified in vitro translated HIF-1␣. After dephosphorylation, HIF-1␣ was incubated in the presence of p42 MAPK for 20 min at 30°C. As expected, HIF-1␣ now migrated at a molecular mass that corresponded to the uppermost band of untreated HIF-1␣ from HeLa cells (Fig. 4). In these conditions, the shift is complete, all of HIF-1␣ migrating as the higher molecular mass band. This suggests a good stoichiometric phosphorylation of HIF-1␣ by p42 MAPK, since every molecule is phosphorylated on the site(s) responsible for this shift. The same results can be seen with HIF-1␣ from other cell types and transiently transfected HA-HIF-1␣ (data not shown). These results demonstrate that endogenous HIF-1␣ can be phosphorylated by p42 MAPK. We also show that this phosphorylation can reproduce the maximal SDS gel shift detected for HIF-1␣ induced in intact cells.
Phosphorylation of HIF-1␣ by MAPKs Is Specific to p42/44 MAPK (ERKs)-The MAPK family of protein kinases includes the mitogen-stimulated p42/p44 MAPK or ERKs and also the stress-activated kinases p38 MAPK and JNK (31). We there- FIG. 1. Comparison of in vitro translated and endogenous HIF-1␣. Different cell lines were maintained under normoxic (21% O 2 ) or hypoxic (1% O 2 ) conditions for 3 h. Whole cell extracts (50 g) were analyzed by SDS-PAGE (7.5% gel) and immunoblotting using an anti-HIF-1␣ antiserum. In vitro translated HA epitope-tagged human HIF-1␣ was obtained with the use of the TNT-coupled reticulocyte lysate system (Promega). 1 l of the reaction mixture was used in this experiment. fore evaluated the specificity of the p42/p44 MAPK phosphorylation by assessing whether p38 MAPK or JNK could phosphorylate HIF-1␣. In this experiment, HA-tagged kinases, stimulated or not, were immunoprecipitated from CCL39 cells stably transfected with the corresponding expression vector and incubated with in vitro translated HIF-1␣ as substrate. In the case of p44 MAPK, cells were incubated for 5 min with 10% FCS, a condition that gives a maximal activation of p42/p44 MAPK. As for p38 MAPK and JNK, cells were maximally activated by addition of 1 g/ml anisomycin for 20 min. Incubation of unlabeled HIF-1␣ with [␥-32 P]ATP and activated p44 MAPK strongly phosphorylated HIF-1␣ (Fig. 5, autoradiography) and induced a marked shift in its molecular mass (Fig. 5, Western blot). Neither p38 MAPK or JNK were able to phosphorylate HIF-1␣ or induce changes in HIF-1␣'s migration pattern. These results indicate that the phosphorylation of HIF-1␣ by MAPK is specific to p42/44 MAPK.
Hypoxia Does Not Activate p42/p44 MAPK in CCL39 Fibroblasts-We now wanted to evaluate if hypoxia could activate p42/p44 MAPK, which would then phosphorylate HIF-1␣ and induce the molecular mass shift. Hypoxia has been shown to activate p42/p44 MAPK in HeLa cells (30). The HeLa cell line is a difficult model to evaluate p42/44 MAPK activity, since these cells are highly transformed and have a high basal level of p42/p44 MAPK activity even in FCS-starved conditions (32). In contrast, CCL39 cells are an excellent model for studying p42/p44 MAPK activation. When deprived of FCS, CCL39 cells arrest well and show very low levels of p42/p44 MAPK activity. Therefore, we evaluated the effect of hypoxia on p42/p44 MAPK activation in CCL39 cells. Since dual phosphorylation of p42/ p44 MAPK is a clear indication of activation (33,34), we used an anti-phospho-p42/p44 MAPK antibody to evaluate variations in p44/p42 MAPK activity. As seen in Fig. 6, stimulation of quiescent CCL39 cells with 10% FCS (v/v) induced a strong activation of p42 MAPK. However, no active p42 MAPK could be detected after exposure to hypoxia (1% oxygen) for 5-60 min while the levels HIF-1␣ rapidly and steadily increased under the same conditions. Longer times were also performed (up to 24 h of hypoxia) without any detectable phosphorylation of p42 MAPK (data not shown). p42 MAPK protein levels did not vary during hypoxia. It is important to note that in this model system and with the antibody used, detection of active p44 MAPK is very faint even if p44 MAPK is present in a relatively similar quantity as p42 MAPK (data not shown). These results suggested that p42/p44 MAPKs are not activated by hypoxia. To confirm these results, p42/p44 MAPK activity was also analyzed by a kinase assay. Serum stimulation (10% v/v) induced a strong phosphorylation of MBP (over 10-fold) while hypoxia (from 5 min to 24 h in 1% oxygen) did not significantly increase the phosphorylation of MBP over basal levels (data not shown). Taken together, these results demonstrate that p42/ p44 MAPK activity is not increased by hypoxia in growtharrested CCL39 cells. Therefore, the phosphorylation of HIF-1␣ by p42/p44 MAPK and the changes in molecular mass induced by this phosphorylation are not those induced by hypoxia.
p42/p44 MAPK Activation Induces HIF-1␣ Phosphorylation in Vivo-Since hypoxia did not activate p42/p44 MAPKs, the induction of HIF-1␣ mobility shift should not be apparent in quiescent hypoxic cells and would be induced by an activation of the p42/p44 MAPK pathway. To investigate this possibility, we used a derivative of the CCL39 cell line that stably expresses the chimera Raf-1:ER. This protein is a fusion between the catalytic domain of Raf-1, an upstream activator of the p42/p44 MAPK cascade, and the ligand binding domain of the estradiol receptor. Estradiol stimulation of Raf-1:ER cells leads to a rapid and exclusive p42/p44 MAPK activation. As seen in HeLa cells were maintained under normoxic (21% O 2 ) or hypoxic (1% O 2 ) conditions for 3 h. Whole cell extracts (1 mg) were immunoprecipitated using an anti-HIF-1␣ antiserum. The immunoprecipitates were treated with phosphatase for 1 h at 30°C followed by extensive washing. Active recombinant p42 MAPK was then added for 20 min at 30°C. The samples were analyzed by SDS-PAGE (7.5% gel) and immunoblotting using an anti-HIF-1␣ antiserum.

FIG. 5. Phosphorylation of HIF-1␣ by MAPKs is specific to p42/p44 MAPK.
In vitro translated HA-HIF-1␣ (unlabeled) was obtained with the use of the TNT-coupled reticulocyte lysate system. 5 l of the mixture was immunoprecipitated using an anti-HA antibody and incubated with [␥-32 P]ATP in the presence of active or inactive p44 MAPK, p38 MAPK and JNK for 20 min at 30°C, followed by SDS-PAGE (7.5% gel). Proteins were transferred to an Immobilon-P membrane and analyzed by autoradiography and immunoblotting using an anti-HA antibody. As a control of kinase activity, p44 MAPK was incubated with the substrate myelin basic protein (MBP), while p38 MAPK and JNK were incubated with their known substrate ATF2. Samples were analyzed by SDS-PAGE (12.5% gel) followed by autoradiography (data not shown). mass does not occur in quiescent cells. When these cells were treated with estradiol for different intervals before the end of the 3-h hypoxic period, a form of HIF-1␣ appeared at 116 kDa, which corresponded exactly to the molecular mass of HIF-1␣ phosphorylated in vitro by p42/p44 MAPK. The induction of the HIF-1 shift closely followed the activation of p42/p44 MAPK. Essentially the same results were obtained using another stimulus shown to activate p42/p44 MAPK, 10% FCS (data not shown). Furthermore, this shift was inhibited by treatment of cells with the specific p42/p44 MAPK pathway inhibitor, PD 98059 (Fig. 7B). However, the mobility shift of HIF-1␣ after estradiol stimulation was not complete. We suspected that strong phosphatase activity in the nucleus was the cause of the partial HIF-1␣ phosphorylation and that if we could increase the level of p42/p44 MAPK activity, we could induce a complete shift of HIF-1␣. We therefore incubated the cells with a potent tyrosine phosphatase inhibitor, the vanadate-derived compound, bpV(phen) (Calbiochem). Treatment of cells with bpV-(phen) alone strongly activated p42/p44 MAPK activity in CCL39 cells (Fig. 7B, phospho-p42/p44 MAPK) and also induced a partial HIF-1␣ mobility shift. As with the estradiol stimulation, the shift induced by bpV(phen) was blocked by the PD 98059 compound. More interestingly, when the cells were treated with both estradiol and bpV(phen), this had an additive effect on p42/p44 MAPK activity and induced a complete shift of the HIF-1␣ molecule. In these conditions, the strong increase in p42/p44MAPK activity along with the HIF-1␣ shift could only be partially blocked with 50 M PD 98059. However, at 100 M a complete inhibition could be achieved by PD 98059 (data not shown). Taken together, these results strongly suggest that HIF-1␣ is phosphorylated in vivo by p42/p44 MAPK.
Activation of p42/44 MAPK Promotes HIF-1-dependent Transcriptional Activation-The ultimate goal of HIF-1 induction is the transcription of target genes such as VEGF and erythropoietin (35). Phosphorylation by p42/p44 MAPK has been shown to modulate the activity of a number of transcription factors. We therefore wanted to evaluate the possible effect of a phosphorylation of HIF-1␣ by p42/p44 MAPK on HIF-1 transcriptional activity. To perform these experiments, we used a luciferase reporter plasmid driven by the VEGF promoter in which a mutation (Mut Sp1/AP2) has eliminated the previously shown site for p42/p44 MAPK activation (6). This reporter is completely insensitive to p42/p44 MAPK stimulation (Ref. 6; Fig. 8, reporter only). This construct was co-transfected with HIF-1␣ and HIF-1␤ constructions in the Raf-1:ER expressing cell line. Since overexpression of HA-HIF-1␣ allows us to detect HIF-1␣ protein even in normoxic conditions (data not shown), this system permitted us to directly evaluate the role of MAPK activation on the HIF-1 complex independently of HIF-1␣'s hypoxic induction. In these conditions, if the stimulation of p42/p44 MAPK increases the expression of the VEGF/ luciferase reporter, this effect can only be through the HIF-1 complex. As shown in Fig. 8, FCS-starved Raf-1:ER cells transiently expressing HIF-1␣ and HIF-1␤ only slightly increased the activation of the mutated VEGF promoter under normoxic conditions. However, addition of 100 nM estradiol strikingly activated the mutated VEGF promoter by almost 5-fold over untreated cells without affecting basal levels in cells transfected with the reporter only. To a lesser degree, an increase in VEGF promoter expression could also be seen when cells were transfected with only HIF-1␣. However, when cells were transfected with HIF-1␤ only, no increase in reporter expression could be seen. This demonstrates that HIF-1␣ expression is essential for this induction. Finally, estradiol had no effect on FIG. 7. p42/p44 MAPK activation induces HIF-1␣ phosphorylation in vivo. CCL39 cells stably expressing the Raf-1:ER chimera (Raf-1:ER cells) were FCS-starved for 24 h before 3 h of hypoxia. A, before the end of the hypoxic period, Raf-1:ER cells were stimulated with 100 nM estradiol for the times indicated. B, before the end of the hypoxic period, Raf-1:ER cells were pretreated or not with 50 M PD 98059 for 30 min followed or not by a stimulation with either 100 nM estradiol for 30 min, 1 mM bpV(phen) for 15 min or a combination of both. Whole cell extracts (50 g) were analyzed by SDS-PAGE (7.5% gel) and immunoblotting using an anti-HIF-1␣ antiserum or an anti-phospho-p44/p42 MAPK polyclonal antibody.
VEGF promoter activity in parental CCL39 cells that did not express the Raf-1:ER chimera (data not shown). These results demonstrate that strong activation of p42/p44 MAPK is sufficient to effectively promote the transcriptional activity of HIF-1.

DISCUSSION
Phosphorylation and dephosphorylation activities have been suggested to be critical in the signaling pathway leading to HIF-1 activation (18,35). Wang et al. (36) demonstrated that treatment of cells with the tyrosine protein kinase inhibitors genistein and herbimycin A, the serine/threonine protein kinase inhibitor, 2-aminopurine and the serine/threonine protein phosphatase inhibitor sodium fluoride blocked the induction of HIF-1␣. Inversely, sodium orthovanadate, a tyrosine phosphatase inhibitor increased basal HIF-1␣ levels and activity. Interestingly, okadaic acid, another serine/threonine phosphatase inhibitor, did not inhibit HIF-1␣ expression but rather increased the proportion of the high molecular mass form of HIF-1␣ (data not shown). All these results suggested that different protein kinases and phosphatases are capable of regulating HIF-1␣. However, no studies have directly investigated the possible phosphorylation of HIF-1␣ and the effect on its activity. A number of studies have shown that the electrophoretic migration of HIF-1␣ is very diffuse (Refs. 12, 37, and 38; our present work). This was suggested to be caused by post-translational modifications of the HIF-1␣ protein (12). In accordance with this proposal, in vitro translated HA-tagged human HIF-1␣, which is not modified post-translationally, migrates as a sharp band at 104 kDa. Our hypothesis was that these post-translational modifications were caused by phosphorylation.
In this work, we have shown evidence that HIF-1␣ is phosphorylated in vivo and that this phosphorylation can dramatically affect the migration pattern of HIF-1␣. We also show that the differently migrating bands of HIF-1␣ protein were due to varying levels of phosphorylation. These changes in molecular mass induced by phosphorylation are clearly independent of hypoxia since they can be seen when HIF-1␣ is overexpressed even in normoxic conditions (Ref. 37; data not shown). These results suggest that phosphorylation of the HIF-1␣ molecule occurs once HIF-1␣ is produced in the cell. However, we can not exclude that additional phosphorylations are implicated in the hypoxic response. Subsequent phosphorylations could possibly occur during hypoxia to induce the full activation of the HIF-1 complex.
Active p42/p44 MAPK induced a rapid phosphorylation of HIF-1␣, either in vitro translated or immunoprecipitated and dephosphorylated from HeLa cells. This phosphorylation increases HIF-1␣'s molecular mass to a value similar to the uppermost band detected after HIF-1␣ induction in many cell systems. This same phosphorylation can be seen in an intact cell system when the p42/p44 MAPK pathway is activated. Taken together, these results strongly suggest that p42/p44 MAPK do phosphorylate HIF-1␣ in vivo. Interestingly, p38 MAPK and JNK did not phosphorylate nor affect the migration pattern of HIF-1␣, suggesting that this effect is specific to p42/p44 MAPK. We also demonstrated that HIF-1␤ is not shifted when incubated with p42 MAPK. These results suggest that HIF-1␤ does not undergo the same type of modifications as HIF-1␣ and strengthen the specificity of p42/p44 MAPK on HIF-1␣ phosphorylation.
At this point, the HIF-1␣ sites which are phosphorylated by p42/p44 MAPK have not been elucidated. Two p42/p44 MAPK consensus sites (PXSP) exist on human HIF-1␣ (positions 515 and 687). In order to evaluate the role of these residues, we point mutated the two serines into alanine and glycine, respectively. However, the ability of p42 MAPK to induce modifications in HIF-1␣'s migration pattern remained unchanged (data not shown). We are currently investigating other possible phosphorylation sites.
The finding that p42/p44 MAPKs are not activated by hypoxia correlates well with the previously mentioned results showing that overexpressed HIF-1␣ is phosphorylated and migrates in a diffuse pattern even in normoxia (Ref. 37; data not shown). It is simply possible that the basal level of p42/p44 MAPK activity is responsible for the phosphorylation of induced HIF-1␣. This would explain the different migration patterns observed between the different exponentially growing cell types. Interestingly, cells with a low basal level of p42/p44 MAPK activity, like CCL39 cells, showed less of the uppermost migrating HIF-1␣ band. However, 293 and HeLa cell lines, which are highly transformed cells with a elevated basal p42/ p44 MAPK activity, demonstrate an increased level of the high migrating HIF-1␣. In the same context, it is important to note that CCL39 derivatives expressing constitutively active Ras and MEK (v-Ras and MEK SS/DD) showed an augmented level of the higher migrating HIF-1␣ as compared with parent CCL39 cells (data not shown).
MAPK activity has been shown to regulate the induction and/or degradation of certain transcription factors (39 -42). However, the p42/p44 MAPK pathway does not seem to be implicated in the stabilization of the HIF-1␣ molecule or the regulation of its ubiquitin-proteasome-mediated degradation. Two unpublished experiments from our laboratory support this statement. First, the induction or degradation kinetics of HIF-1␣ are essentially the same in the case of exponentially growing cells in the presence of 10% FCS and serum-starved quiescent cells. Second, strong activation of the p42/p44 MAPK pathway with the Raf-1:ER chimera does not modify the rate of HIF-1␣ induction or degradation. Therefore, if phosphorylation events are implicated in these phenomena, they are independent on the phosphorylation by p42/p44 MAPK.
The last and major finding in this work is that strong and sustained p42/p44 MAPK activation induces the expression of HIF-1-dependent reporter genes in normoxic cells when the HIF-1 is forcibly induced by overexpression of the HIF-1␣ and HIF-1␤ proteins. Increases in reporter gene expression can also be detected in FCS-stimulated cells, but to a lesser degree (data not shown). This is possibly due to the temporal activation of p42/p44 MAPK with estradiol, which is much more prolonged and sustained than with FCS. In addition, these experiments show that MAPK activation alone is sufficient to promote HIF-1-mediated transcriptional activation. These results are in agreement with previous work, which demonstrated that treatment with PD98059, an upstream inhibitor of the p42/p44 MAPK pathway, inhibited HIF-1-mediated target gene activation (43). The mechanism underlying this activation is still unknown. Phosphorylation of HIF-1␣ by p42/p44 MAPK may favor dimerization with partners like HIF-1␤ to form the HIF-1 complex. We have performed co-immunoprecipitation experiments to evaluate this possibility. We have not seen any differences in the level of HIF-1␤ that co-immuprecipitates with HIF-1␣ when cells are stimulated with estradiol. Another possibility is that phosphorylation may increase the interaction of the complex with its DNA binding site. In this context, it has been shown that treatment with PD98059 does not alter the DNA binding activity of the HIF-1 complex (43). However, these experiments were not done in conditions of strong MAPK activation. Another possibility is that phosphorylation of HIF-1␣ augments its interaction with the basal transcriptional machinery. Finally, phosphorylation may modify the interaction between HIF-1␣ and another partner(s) to increase HIF-1-mediated gene activation. HIF-1␣ has been shown to interact with CBP/p300 and p53 (44 -47). These interactions modulate the transcriptional activity of the HIF-1 complex. It is still not known whether phosphorylation is implicated in this interaction. These research avenues are currently being investigated.
In summary, we have demonstrated that HIF-1␣ is a highly phosphorylated protein in vivo and that this phosphorylation of HIF-1␣ induces strong changes in the HIF-1␣'s migration pattern. We also show that in vitro, p42/p44 MAPK can reproduce this phosphorylation. In quiescent cells, strong activation of p42/p44 MAPK induces the phosphorylation of HIF-1␣ and increases HIF-1-dependent transcriptional activity. The further comprehension of these phosphorylation processes will undoubtedly be a major contribution in the understanding of the signaling pathways that modulate HIF-1 activity and HIF-1-mediated gene expression.