EPAS1 trans-Activation during Hypoxia Requires p42/p44 MAPK*

Hypoxia is a common environmental stress that regulates gene expression and cell function. A number of hypoxia-regulated transcription factors have been identified and have been shown to play critical roles in mediating cellular responses to hypoxia. One of these is the endothelial PAS-domain protein 1 (EPAS1/HIF2-α/HLF/HRF). This protein is 48% homologous to hypoxia-inducible factor 1-α (HIF1-α). To date, virtually nothing is known about the signaling pathways that lead to either EPAS1 or HIF1-α activation. Here we show that EPAS1 is phosphorylated when PC12 cells are exposed to hypoxia and that p42/p44 MAPK is a critical mediator of EPAS1 activation. Pretreatment of PC12 cells with the MEK inhibitor, PD98059, completely blocked hypoxia-inducedtrans-activation of a hypoxia response element (HRE) reporter gene by transfected EPAS1. Likewise, expression of a constitutively active MEK1 mimicked the effects of hypoxia on HRE reporter gene expression. However, pretreatment with PD98059 had no effect on EPAS1 phosphorylation during hypoxia, suggesting that MAPK targets other proteins that are critical for thetrans-activation of EPAS1. We further show that hypoxia-induced trans-activation of EPAS1 is independent of Ras. Finally, pretreatment with calmodulin antagonists nearly completely blocked both the hypoxia-induced phosphorylation of MAPK and the EPAS1 trans-activation of HRE-Luc. These results demonstrate that the MAPK pathway is a critical mediator of EPAS1 activation and that activation of MAPK and EPAS1 occurs through a calmodulin-sensitive pathway and not through the GTPase, Ras. These results are the first to identify a specific signaling pathway involved in EPAS1 activation.

Regulation of gene expression is a primary response by which cells adapt to changes in the environment. The mechanisms involved in regulation of gene expression in response to hypoxia are beginning to be understood. Transcription factors that are activated by hypoxia include the hypoxia-inducible factor (HIF1-␣), 1 c-fos, and CREB (1)(2)(3)(4). HIF1-␣ has been shown to be critical for hypoxia-induced regulation of a number of genes, including glycolytic enzymes, vascular endothelial growth factor, and erythropoieitin (5)(6)(7). Recently, endothelial PAS-domain protein 1 (EPAS1, also known as HIF2-␣, HLF, and HRF) was identified as a hypoxia-inducible transcription factor (8 -10). EPAS1 is a basic helix-loop-helix transcription factor, which shares 48% sequence identity with HIF1-␣ (8). EPAS1 protein levels, like HIF1-␣ levels, are relatively low under basal conditions and accumulate upon exposure of cells to hypoxia (11). These factors then translocate to the nucleus and trans-activate target genes containing the sequence 5Ј-GCCCTACGTGCTGTCTCA-3Ј, which is commonly referred to as the hypoxia response element (HRE) (8,12).
EPAS1 is expressed in many tissues and is particularly abundant in the type I oxygen-sensing cells of the carotid body (13). Type I cells act as the primary O 2 sensors in mammals and are responsible for matching changes in arterial pO 2 with appropriate changes in respiration (14). Our laboratory has used PC12 cells as a model system to study the biophysical and molecular properties of oxygen-sensing cells (15). There are a number of phenotypic similarities between type I and PC12 cells, including the presence of O 2 -sensitive K ϩ channels, which are inhibited by hypoxia (16,17). In addition, both PC12 cells and type I cells respond to hypoxia with an increase in tyrosine hydroxylase gene expression (18,19). Finally, both cell types depolarize and secrete the neurotransmitter dopamine in response to hypoxia (20 -22). We have therefore utilized PC12 cells to study the regulation of EPAS1.
The specific signaling pathways that are involved in HIF1-␣ and EPAS1 activation are almost completely unknown. In our previous studies, we measured the effects of hypoxia on the mitogen and stress-activated protein kinase pathways (MAPKs and SAPKs) (23). We found that moderate hypoxia (5% O 2 ) activates p42/p44 MAPK, two closely related protein kinases that can lead to the phosphorylation and activation of a number of transcription factors (24). We therefore hypothesized that the MAPK pathway may be important for EPAS1 activation during hypoxia. Results from the current study show that the MAPK pathway is critical for EPAS1 activation, as the specific MEK1 inhibitor, PD98059, prevents EPAS1 trans-activation of the HRE. Interestingly, PD98059 had no effect on EPAS1 protein levels, suggesting that the MAPK pathway is involved in the activation of EPAS1, rather than the accumulation of EPAS1. We also show, for the first time, that EPAS1 itself is phosphorylated during hypoxia. However, EPAS1 is not directly phosphorylated by MAPK, suggesting that MAPK mediates its effects indirectly, possibly by recruiting other proteins critical for EPAS1 trans-activation. Finally, we show that MAPK-activation of EPAS1 during hypoxia occurs via a calmodulin-sensitive pathway and not through a Ras-dependent mechanism.
PD98059 was obtained from New England Biolabs (Beverly, MA). EPAS1 polyclonal antibody, the HRE-Luc reporter gene, and the EPAS1 cDNA were generous gifts from Dr. Steven L. McKnight (University of Texas Southwestern, Dallas, TX). pFC-MEK1 was obtained from Stratagene (La Jolla, CA). Additional EPAS1 polyclonal antibody was obtained from Novus Biologicals (Littleton, CO), and similar results were obtained with both antibodies. RasN-17 was a gift from Dr. J. Silvio Gutkind (National Institutes of Health, NIDR, Bethesda, MD). A c-fos-luciferase fusion reporter gene (fos-Luc) was constructed from a c-fos-␤-galactosidase fusion gene construct, kindly provided by Dr. Tom Curran (St. Jude's Childrens Research Hospital, Memphis, TN). The ␤-galactosidase coding region was excised from the fos-lacZ plasmid (26) with NcoI and BamHI and replaced with the luciferase coding region from the pGL3-basic plasmid (Promega, Madison, WI). W13 was obtained from RBI (Natick, MA). Calmidazolium chloride was obtained from Calbiochem.
Reporter Gene Assays-PC12 cells were transfected with the hypoxia response element-luciferase (HRE-Luc) reporter gene using the Transfast transfection reagent according to the manufacturers recommended conditions (Promega). This reporter gene has been described previously (8,12). PC12 cells seeded in 24-well plates at 60% confluence were transfected in triplicate with 3 l of Transfast and 250 ng of HRE-Luc per well. In some experiments, 25-100 ng of EPAS1, pFC-MEK1, or RasN-17 was cotransfected with the HRE-Luc. In each transfection, pcDNA3 vector DNA was added to bring the total amount of DNA to 1 g of DNA/well. Cells were switched to serum-free medium for 18 h prior to the start of the experiment. The following day (48 h post-transfection), PC12 cells were exposed to normoxia or hypoxia (1% O 2 ) for 6 h. In other experiments, the effect of NGF on a c-fos reporter gene was evaluated. In these experiments, cells were cotransfected with 250 ng of the c-fosluciferase reporter gene and varying amounts of an N-17 Ras expression plasmid in 24-well plates. After 48 h, cells were incubated with nerve growth factor (50 ng/ml, Alomone Labs, Jerusalem, Israel) for 6 h. To perform luciferase assays, cells were washed with phosphate-buffered saline and lysed in 200 l of cell culture lysis reagent (Promega). Cell extracts were sonicated for 1 s with a microultrasonic cell disrupter (Kontes, Vineland, NJ). Twenty l of cell extracts were then aliquoted into luminometer tubes (Promega). Fifty l of luciferin substrate (Promega) was added to each tube and samples were analyzed in a luminometer (Turner Designs). We found previously that hypoxia inhibits expression of cytomegalovirus-␤-galactosidase, Rous sarcoma virus-␤-galactosidase, and SV40-␤-galactosidase reporter genes. 2 Therefore, as in previous studies, luciferase activity was normalized to micrograms of protein per well (3). Protein samples varied by less than 15% between samples.
Western Blotting-Western blotting was performed as described previously (3,23). For phospho-MAPK blots, membranes were immunolabeled with antibodies recognizing phospho-Tyr 204 MAP kinase (1:1000, New England Biolabs). EPAS1 protein expression was assayed using a rabbit polyclonal antibody directed against amino acids 1-10 of the EPAS1 protein at a dilution of 1:1000.
Phosphorylation and Immunoprecipitation-Experiments were performed essentially as described by Jewell-Motz et al. (27). Briefly, PC12 cells plated onto 100-mm dishes were washed twice with phosphate-free DMEM and then incubated at 37°C in phosphate-free DMEM (Life Technologies, Inc.) for 30 min. Phosphate-free medium (5 ml/dish) containing 1 mCi/ml of [ 32 P]orthophosphate and either Me 2 SO or PD98059 (50 M) was added to the cells. After preincubation for 1.5 h, cells were exposed to normoxia or hypoxia (1% O 2 , 6 h). Cells were harvested by washing with ice-cold phosphate-buffered saline and scraping in 1 ml of a lysis buffer containing 25 mM Tris, pH 7.4, 1% Triton X-100, 0.5 mM sodium vanadate, 25 mM ␤-glycerophosphate, 1 mM EDTA, 2 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, and 2 g/ml aprotinin. Whole cell lysates were precleared with 5 l of rabbit IgG (Sigma) and 50 l of a 10% (w/v) suspension of protein A-Sepharose beads. EPAS1 was immunoprecipitated using 10 g of an EPAS1 poly-clonal antibody (Novus Biologicals) followed by the addition of 50 l of a 10% (w/v) suspension of protein A-Sepharose beads. The reaction slurry was allowed to rock at 4°C for 2 h. Immunoprecipitates were washed three times with lysis buffer and then subjected to 7.5% PAGE analysis. The gel was dried and analyzed using a phosphorimager (Molecular Dynamics, Sunnyvale, CA).

RESULTS
As a first step toward characterizing the regulation of EPAS1 in PC12 cells, we evaluated EPAS1 protein levels following exposure to hypoxia. Fig. 1A shows that exposure to hypoxia (1% O 2 ) for 6 h resulted in a 12-fold increase in EPAS1 protein levels. It has been established previously that EPAS1 can trans-activate an HRE-Luc reporter gene (8). We found that titrating the level of hypoxia from 21% O 2 to 1% O 2 resulted in a dose-dependent increase in HRE-luciferase activity (Fig. 1B).
We have shown recently that hypoxia specifically regulates certain members of the SAPK and MAPK family (23). We reported that moderate hypoxia (5% O 2 ) induced a modest phosphorylation of MAPK. Fig. 2A shows results obtained when PC12 cells were exposed to more severe hypoxia (1% O 2 ), which caused a pronounced phosphorylation of p42/p44 MAPK. Because the MAPK pathway is known to regulate a number of transcription factors, including c-fos, jun-B, CREB, and Elk-1 (28 -30), we hypothesized that the MAPK pathway might be important for EPAS1 activation during hypoxia. To test this hypothesis, PC12 cells were cotransfected with the HRE-Luc reporter gene and a plasmid encoding the human EPAS1 cDNA or the empty expression vector, pcDNA3. Cells were then pretreated with either PD98059 (50 M) or vehicle and exposed to 2 D. Beitner-Johnson and D. E. Millhorn, unpublished observations. FIG. 1. EPAS1 protein accumulates and is activated by hypoxia. PC12 cells were exposed to normoxia (21% O 2 ) or hypoxia (1% O 2 , 6 h) followed by SDS-PAGE and immunoblotting with an ␣-EPAS1 antibody. A, immunoblot showing the effect of hypoxia on EPAS1 immunoreactivity. Results are representative of n ϭ 6 performed in two separate experiments. B, PC12 cells were seeded in 24-well dishes and transfected with the HRE-Luc reporter gene (250 ng/dish). 48-h posttransfection, cells were exposed to normoxia, or increasing levels of hypoxia, as indicated, and then assayed for luciferase activity as described under "Experimental Procedures." Data are representative of results performed in three experiments. normoxia or hypoxia (1% O 2 ) for 6 h. As reported by others (8) we found that expression of EPAS1 increased HRE-Luc activity under both normoxic and hypoxic conditions ( Fig. 2A). We also found that inhibition of MEK1, by PD98059, completely blocked the effect of hypoxia on both basal and EPAS1-stimulated HRE-Luc activity (Fig. 2A). These results strongly suggest that the MEK1-MAPK signaling pathway is critical for mediating EPAS1 activation of HRE-dependent gene expression.
To test this, we measured the effect of expressing a constitutively active MEK1 (pFC-MEK1) on basal and hypoxia-induced HRE-Luc activity. MEK1 is a dual specificity protein kinase that directly phosphorylates and activates MAPK (24). Fig. 2B shows that expression of pFC-MEK1-enhanced basal HRE-Luc activity during both normoxia and hypoxia. However, when coexpressed with EPAS1, pFC-MEK1 caused a much larger increase in the trans-activation of the HRE-Luc (Fig.  2B). The relative increase in HRE-Luc activity in the presence of pFC-MEK1 and EPAS1 was 13-fold higher than cells transfected with EPAS1 and exposed to normoxia (Fig. 2C). In contrast, transfection with EPAS1 alone, followed by hypoxia, resulted in only a 2-fold increase in HRE-Luc activity (Fig. 2C).
The increase in EPAS1 immunoreactivity induced by hypoxia was accompanied by a shift in the mobility of the EPAS1 protein (see Fig. 1, inset), suggesting that EPAS1 itself might be phosphorylated during hypoxia. To test this possibility, PC12 cells were pretreated with either PD98059 or vehicle, then metabolically labeled with [ 32 P]orthophosphate. Following normoxic or hypoxic exposure, EPAS1 was immunoprecipitated from whole cell lysates, and its phosphorylation state was evaluated by SDS-PAGE and PhosphorImager analysis. Fig.  3A shows that hypoxia does indeed induce phosphorylation of EPAS1. However, EPAS1 phosphorylation was not blocked by PD98059, in contrast to the effects of hypoxia on trans-activation of the HRE-Luc reporter gene by EPAS1. We also tested whether MAPK was involved in the induction of EPAS1 immunoreactivity by hypoxia. In these experiments, PC12 cells were pretreated with PD90859 or vehicle prior to exposure to hypoxia. Whole cell lysates were then immunoblotted for EPAS1. Fig. 3B shows that, while hypoxia induced a 13-fold increase in EPAS1 protein levels, inhibition of MEK1 with PD98059 had no effect on the hypoxia-induced accumulation of EPAS1.
Ras is an upstream activator of MAPK (24,31). In order to test whether Ras was involved in the EPAS1 trans-activation of the HRE-Luc, PC12 cells were cotransfected with the EPAS1 expression plasmid, the HRE-Luc plasmid, and increasing amounts of a dominant-negative Ras expression plasmid, RasN-17. Fig. 4A shows that increasing amounts of RasN-17 had no effect on the EPAS1 trans-activation of HRE-Luc. However, coexpression of the same amounts of RasN-17 did block activation of a c-fos-luc reporter gene by nerve growth factor (NGF) in PC12 cells (Fig. 4B). Thus, EPAS1 activation by hypoxia occurs via a Ras-independent mechanism.
Hypoxia results in depolarization and calcium influx into PC12 cells during hypoxia (17,32). Consistent with these findings, Egea et al. (33,34) have shown that depolarization of PC12 cells results in MAPK activation via a calmodulin-dependent mechanism. Thus, we hypothesized that calmodulin could be involved in the activation of MAPK and EPAS1 during hypoxia. Fig. 5A shows that pretreatment of PC12 cells with the calmodulin antagonist, W13 (20 g/l), caused a pronounced reduction in hypoxia-induced MAPK phosphorylation. These results are shown quantitatively in Fig. 5B. We also found that treatment with either W13, or calmidazolium chloride (CMZ, 1 M), another calmodulin antagonist, inhibited both endogenous HRE activity, as well as the EPAS1 transactivation of the HRE reporter gene (Fig. 5C). Thus, MAPK activation of EPAS1 occurs via a calmodulin-dependent pathway, rather than through the proto-typical mediator, Ras.

DISCUSSION
Regulation of gene expression by hypoxia is mediated by a number of signal transduction pathways (15,35). MAPK is known to be critical for the trans-activation of many genes and mediates its effects primarily through the phoshorylation of downstream transcription factors (24,25,31). The current study shows, for the first time, that EPAS1 is phosphorylated during hypoxia and that the MAPK pathway is critical for EPAS1 transactivation during hypoxia in PC12 cells. While our findings suggest that phosphorylation is an important regulatory step for EPAS1 activation, others have also shown that redox-sensitive changes are critical to the formation of the EPAS1 DNA-binding complex (36). It is likely that EPAS1 activation results from the integration of multiple signals and that the importance of specific signals varies in a cell type-specific manner.
We found that MAPK is required for EPAS1 trans-activation of the HRE-Luc reporter gene, as this was completely blocked by PD90859, a selective inhibitor of MEK1 (37), and enhanced by constitutively activated MEK1. However, neither hypoxiainduced phosphorylation nor accumulation of EPAS1 protein was inhibited by PD90859. Thus, although MAPK is critical for hypoxic regulation of EPAS1 function, it is not the kinase that phosphorylates EPAS1 during hypoxia. These results suggest that multiple MAPK-dependent and MAPK-independent signals are required for EPAS1 activation. One MAPK-independent signal leads to accumulation of the EPAS1 protein, presum-ably by inhibition of ubiquitin-proteasome degradation (38). A second MAPK-independent signal leads to the phosphorylation of EPAS1. Recent evidence by others suggests that multiple signals are involved in regulation of EPAS1 and identifies two domains of EPAS1 that are required for its activation during hypoxia. One of the critical EPAS1 domains is an internal domain that extends from amino acids 450 -571 and shares homology with the oxygen-dependent domain of HIF1-␣ (36,39,40). Both the EPAS1 (450 -571) and the HIF1-␣ (oxygen-dependent domain) domains were identified as being critical for the induction of their respective proteins during hypoxia. The second important EPAS1 regulatory domain is a C-terminal activation domain (amino acids 824 -876), which is the site of post-translational modification in EPAS1 during hypoxia (36). It is therefore tempting to speculate that phosphorylation of EPAS1 occurs within the C-terminal activation domain of the protein.
However, the functional consequences of EPAS1 phosphorylation are unknown and will require further investigation.
The mechanism of MAPK-dependent activation of EPAS1 is unknown. The fact that EPAS1 phosphorylation persists in the presence of PD98059 suggests that the MAPK pathway does not directly target EPAS1, but instead targets other protein(s) that are critical for the formation of the EPAS1 DNA-binding complex. Others have shown that CREB-binding protein (CBP) interacts with HIF1-␣ and EPAS1 and potentiates the activa- tion of these proteins (36,41). Janknecht et al. (42) have shown that C-terminal regions of CBP can be phosphorylated by MAPK in vitro. Furthermore, Liu et al. (43) showed that MAPK can directly regulate the transcriptional activity of CBP following NGF stimulation in PC12 cells. Thus, CBP might be a target of hypoxia-activated MAPK, which could then recruit EPAS1 to the DNA-binding complex. In addition to CBP, the von Hippel Lindau (VHL) tumor suppressor gene product has been shown recently to be involved in the regulation of HIF1-␣ and EPAS1 protein levels (44). Interestingly, pVHL was also shown to be present in the HIF1-␣ DNA-binding complex (44). Finally, it has been proposed that several "general transcription factors" are present in the EPAS1 DNA-binding complex (36). These proteins are also potential targets of MAPK regulation. Thus, it is likely that the MAPK-dependent activation of EPAS1 trans-activation involves the recruitment of proteins other than EPAS1 to the DNA-binding complex.
The prototypical mechanism of activation of MAPK is via activation of the Ras-Raf-MEK pathway (31). However, some stimuli, such as endothelin-1 and bacterial lipopolysaccharide, have been shown to activate MAPK in a Ras-independent manner (45,46). Our results indicate that hypoxia is similar to these stimuli, as expression of a dominant-negative Ras had no effect on the ability of EPAS1 to trans-activate the HRE-Luc reporter gene.
Since EPAS1 activation was Ras-independent, it was of interest to identify the upstream activators that lead to MAPK and EPAS1 activation. Egea et al. (33,34) have shown that, following depolarization of PC12 cells, MAPK is activated via a calmodulin-sensitive pathway. Exposure of PC12 cells to hypoxia also causes depolarization and calcium influx, via the inhibition of an oxygen-sensitive K ϩ channel (17,32). Our results demonstrate that calmodulin is critical to the activation of MAPK and EPAS1 during hypoxia. Calmodulin is known to activate a number of proteins, including the calcium/calmodulindependent family of protein kinases and the calcium/calmodulindependent protein phosphatase, calcineurin (47,48). Future experiments are aimed at defining the mechanism by which calmodulin activates MAPK under conditions of hypoxia. Finally, while the results of Egea et al. (33) bear some similarity to our own, their study showed that Ras was critical to the depolarization-induced activation of MAPK and that Ras activation likely resulted from phosphorylation of the epidermal growth factor receptor. In contrast to these results, we have found that MAPK activation by hypoxia was Ras-independent, and we were unable to demonstrate phosphorylation of the epidermal growth factor receptor by hypoxia. 3 These contrasting results illustrate that there may be important differences between hypoxia-induced depolarization and KCl-induced depolarization. In conclusion, our results show that MAPK and calmodulin are critical mediators of hypoxia-induced signal transduction and transcription factor activation. The importance of this calmodulindependent pathway is likely to be unique to PC12 cells and other excitable cells, as nonexcitable cells (HEP3B, HEPG2) do not depolarize when exposed to a hypoxic environment.
These results provide the first evidence to define a specific signaling pathway that leads to EPAS1 activation. We show that the MAPK pathway is a critical mediator of EPAS1 activation and that activation of MAPK and EPAS1 occurs through a calmodulin-sensitive pathway, but not through the GTPase, Ras. Further studies are aimed at determining the molecular mechanism by which MAPK regulates EPAS1 function and identifying the endogenous kinase that phosphorylates EPAS1. Such studies will facilitate our understanding of how excitable cells adapt and respond to low oxygen levels. We also show, for the first time, that EPAS1 is phosphorylated during hypoxia and that this phosphorylation is independent of MAPK.