Involvement of Net and Hif1α in Distinct yet Intricately Linked Hypoxia-induced Signaling Pathways*

The present study compares negative Ets transcription factor (Net) and hypoxia-inducible factor 1α (HIF1α) regulation by hypoxia. Their protein stabilities are differently regulated by hypoxia, defining three periods in the kinetics: normoxia (high Net levels and low HIF1α levels), early hypoxia (high levels of Net and HIF1α), and late hypoxia (degradation of Net and HIF1α). Modulators of prolyl hydroxylase domain protein (PHD) activity induce a mobility shift of Net, similar to HIF1α, suggesting that post-translational modifications of both factors depend on PHD activity. The three PHDs have different roles in the regulation of Net protein levels; PHD1 and PHD3 are involved in the stabilization of Net, whereas PHD2 controls its degradation in late hypoxia. Net physically interacts with PHD2 in hypoxia, whereas PHD1 and PHD3 bind to Net in normoxia and hypoxia. Under the same conditions, PHD2 and PHD3 regulate both HIF1α stabilization in early hypoxia and its degradation at late hypoxia, whereas PHD1 is involved in HIF1α degradation in late hypoxia. We describe interconnections between the regulation of both Net and HIF1α at the protein level. Evidence is provided for a direct physical interaction between Net and HIF1α and indirect transcriptional regulation loops that involve the PHDs. Taken together our results indicate that Net and HIF1α are components of distinct signaling pathways that are intricately linked.

Hypoxia is a reduction in the normal level of tissue oxygen tension that occurs in many disease processes including cancer. It arises in solid tumors due to a mismatch between tumor growth and angiogenesis and is associated with an aggressive phenotype, resistance to radiation therapy and chemotherapy, as well as poor patient prognosis (1,2). The cellular response to hypoxia involves the induction of the hypoxia-inducible factor 1␣ (HIF1␣), 6 considered to be the major transcription factor involved in gene regulation by hypoxia (3). In normoxia, HIF1␣ is hydroxylated by the cellular oxygen "sensors" prolyl-hydroxylase domain proteins (PHD1, PHD2, and PHD3) and degraded by proteasomes (for reviews, see Refs. 4 and 5). The PHDs are not active in hypoxia, resulting in stabilization and activation of transcription by the non-hydroxylated and stabilized form of HIF1␣ (6,7).
We recently identified a new component of the hypoxic response, the ternary complex factor Net (Elk3) (8). Under basal conditions, Net is a strong repressor of transcription, but it can be converted to an activator by phosphorylation of its activation domain by the growth factor-Ras-mitogen-activated protein kinase pathway (9 -12). Net is involved in the regulation of various physiological processes, including cell migration, inflammation, wound healing, and angiogenesis (13)(14)(15). Loss of Net as a repressor and consequent activation of c-fos expression has been suggested to be a key event in human papillomavirusinduced carcinogenesis (16). Progression and treatment of cervical as well as other cancers implicate the hypoxic response (1,17). We previously reported that hypoxia enhances Net ubiquitylation, nuclear export, and subsequent proteasomal degradation (8). In a large scale analysis of RNA expression using microarrays in transformed mouse endothelial cells, we found that the majority of the genes induced in hypoxia require Net and HIF1␣, suggesting that the functions of these factors are closely linked (18). In our current study, we compared Net and HIF1␣ regulation in response to hypoxia in cells in which Net is a negative regulator (16). These cells (called "444") are one of the components of a cell-based model of cervical cancer progression (19,20). We demonstrate that the hypoxia-induced signaling pathways that involve Net and HIF1␣ have distinct features and that there are interconnections between Net and HIF1␣ at various levels. These results suggest that Net and HIF1␣ cross-talk in response to hypoxia and that the functional status of either factor will influence the way the complementary factor orchestrates the physiological outcome.
Co-immunoprecipitation-Co-transfected 444 and HeLa cells were washed thrice with ice-cold phosphate-buffered saline and lysed in immunoprecipitation buffer containing 20 mM Tris-HCl (pH 7.4), 0.3 M KCl, 0.25 mM EDTA, 0.125 mM EGTA, 0.025% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture (Roche Diagnostics). Immune complexes were formed by incubation of the lysates overnight at 4°C with anti FLAG (M2) agarose. The beads were washed four times with immunoprecipitation buffer, resuspended in non-denaturant Laemmli sample buffer, and processed for immunoblotting.
In Vitro Translation- 35 S-labeled in vitro translated GHO, GHO(P3 A), and Net were prepared using rabbit reticulocyte lysates (TNT Quick kit, Promega). After 45 min of reaction, the translation mixtures were supplemented or not with 100 M FeCl 2 , 1 mM 2-oxoglutarate, and 5 mM ascorbate for 45 min at 30°C. Reaction products were analyzed by SDS-PAGE and autoradiography.
In Vitro Shift Assays-HeLa and HEK293T cells were transfected with FLAG-Net or GHO. 48 h after transfection, the cells were treated with 200 M CoCl 2 for 5 h and then washed twice with ice-cold phosphate-buffered saline and lysed in Laemmli sample buffer for HEK293T or lysis buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.02% NaN 3 , 1% Nonidet P-40) for HeLa. Total cell extracts were analyzed by immunoblotting.

Hypoxia Has Different Effects on Net and HIF1␣ Protein
Levels-To study the integration of Net in the hypoxia/PHD/ HIF1␣ signaling pathway, we compared the hypoxic regulation  JULY 1A) that result from post-translational modification by phosphorylation and possibly other modifications (8,21). We confirmed that these bands are down-regulated by siRNAs against Net in 444 (data not shown), similar to other cell lines we have studied (8,21). As reported previously, there is an additional non-identified band (n.i.) that is variable and is not down-regulated by Net siRNAs. It does not compromise the measurements of Net levels by Western blotting. As shown in Fig. 1, A and B, Net protein levels decrease drastically after 12 h of hypoxia. In contrast, HIF1␣ protein levels follow a bell-shaped curve with a transient increase, peaking between 4 and 8 h of hypoxia, followed by a progressive decrease. We observed similar regulation of Net and HIF1␣ protein levels under hypoxia in total cell extracts (data not shown). To exclude that transcription accounts for the changes, we measured Net and HIF1␣ mRNA by real-time qRT-PCR. There were non-significant small variations in Net and HIF1␣ mRNA, which could not account for the observed changes at the protein level (Fig. 1C). Our results indicate that there are three time phases in the kinetics: 1) normoxia, high Net levels and low HIF1␣ levels; 2) early hypoxia (4 -8 h), high levels of Net and HIF1␣; and 3) late hypoxia (12-24 h), degradation of Net and HIF1␣ (Fig. 1B).

Net and HIF1␣ in Hypoxia Signaling Pathways
Modulators of PHD Activity Induce Mobility Shifts of Net Similar to HIF1␣-PHDs require oxygen, iron, ascorbate, and 2-oxoglutarate to hydroxylate HIF1␣ on Pro-564 in normoxia (22). Iron-displacing transition metals, such as cobalt, act as strong inhibitors of the activity of PHDs ( Fig. 2A). To investigate possible hydroxylation of Net, we compared the effect of modulating the activity of the PHDs in different sys-tems on the mobility of Net in comparison with GHO (Gal4-HIF1␣-(531-652)), which contains Pro-564 (23)(24)(25). As expected from previous reports, 35 S-labeled GHO produced by in vitro transcription and translation in rabbit reticulocyte lysate migrated as a doublet upon SDS-PAGE. Adding iron, ascorbate, and 2-oxoglutarate to the reaction induced a shift of migration, converting the slowly migrating non-hydroxylated species to the rapidly migrating hydroxylated form. The shift was not observed with GHO (PA) , in which Pro-564 is mutated to Ala (Fig. 2B). Interestingly, 35 S-labeled Net also migrated as a doublet, and the addition of the three cofactors shifted the mobility to the more rapidly migrating form (Fig. 2B). In transfected HEK293T cells, the hypoxia mimic, 200 M CoCl 2 , decreased the mobility of GHO, as expected from lack of hydroxylation. Similarly, the migration of full-length Net was also decreased in HeLa cells. The migration of ␤-actin remained unchanged, showing that the shift was not due to nonspecific effects on migration (Fig. 2C). Taken together our data suggest that Net undergoes post-translational modifications that are induced by regulators of PHDs activity.
PHDs Are Involved in Net Regulation in Normoxia-To investigate whether the PHDs regulate Net protein stability in 444 cells in normoxia, we either down-regulated the PHDs with specific siRNAs or overexpressed them with expression vectors. The PHDs were efficiently down-regulated, as shown by qRT-PCR (Fig. 3A). In normoxia, Net protein levels in the nucleus were decreased by down-regulation of PHD1 and PHD3 and increased by down-regulation of PHD2 (Fig. 3, B and C). When we overexpressed the three PHD isoforms, the opposite effect was observed. PHD1 and PHD3 increased and PHD2 decreased Net protein levels (Fig. 3, D and E). We did not find any significant differences in Net mRNA levels (data not shown), demonstrating that the three PHDs are involved in the regulation of Net protein stability in normoxia.
PHDs Are Involved in Net Regulation in Hypoxia-We then investigated how different levels of the PHDs would affect Net levels in hypoxia. The knockdown of PHD2 (Fig. 4, A and B) globally increased Net levels, and in particular, delayed Net down-regulation in late hypoxia (12 and 18 h) when compared with cells transfected with control siRNA targeting luciferase. Overexpression of PHD2 (Fig. 4, C and D) decreased Net levels, which was observed in particular early in hypoxia (4 -8 h). These results indicate that the loss of Net during hypoxia requires PHD2. Down-regulation of PHD1 and to some extent PHD3 (Fig. 4, A and B) globally decreased Net levels, which was particularly evident early in hypoxia (4 -8 h). Overexpression of PHD1 and PHD3 prevented loss of Net during hypoxia, which was most evident late in hypoxia (12 and 18 h). These results indicate that PHD1 and PHD3 help maintain Net levels during hypoxia. These opposing effects of PHD2 and PHD1 and PHD3 suggest that their relative levels of expression could finetune the dynamics of the response to hypoxia in terms of the levels of Net.
Comparison of HIF1␣ with Net in Terms of Their Regulation by the PHDs-To compare the role of the three PHDs in the regulation of HIF1␣ in the same experimental conditions as Net, we down-regulated and overexpressed the PHD isoforms in hypoxia. Silencing PHD2 and PHD3 significantly increased HIF1␣ protein level throughout the time course (Fig. 5, A and  B). PHD1 down-regulation had no significant effects on HIF1␣ protein levels in normoxia and early hypoxia (up to 8 h; 8 h was not statistically significant) but delayed degradation of HIF1␣ in late hypoxia (12-18 h, Fig. 5, A and B). Overexpression of PHD2 and PHD3 strongly inhibited HIF1␣ levels throughout the time course, which was the opposite of down-regulation (Fig. 5, C and D). Overexpression of PHD1 decreased HIF1␣ induction early in hypoxia (4 h), a time at which down-regulation had no effect. This would be expected from an increased activity of PHD1 due to overexpression. PHD1 overexpression also had an effect at 12 h (Fig. 5D), which is apparently paradoxical because down-regulation had the same effect. This suggests that there are complex regulatory mechanisms that involve factors other than PHD1 expression levels. In contrast to Net regulation, both PHD2 and PHD3 play a similar role in all three time phases of hypoxia, whereas PHD1 appears to be mainly involved in the late phase. Taken together our results show that both Net and HIF1␣ are regulated by PHDs but in different manners.
Cross-talk between Net and HIF1a at the Protein Level-To investigate cross-talk between Net and HIF1␣, we analyzed their reciprocal protein levels after inhibition by siRNA in normoxia (0 h) and early (6 h) and late (12 and 24 h) hypoxia. We found that down-regulation of Net impaired stabilization of HIF1␣ in early hypoxia and its degradation in late hypoxia when compared with the control (Fig. 6A). Down-regulation of HIF1␣ inhibited down-regulation of Net in late hypoxia (Fig.  6C). Overexpression of Net resulted in higher and prolonged induction of HIF1␣ (Fig. 6E). There were no significant changes at the mRNA level, as shown by qRT-PCR (Fig. 6, B, D, and F, right panels). These results indicate that Net is required for rapid induction of HIF1␣ by hypoxia, whereas HIF1␣ is necessary for Net degradation in hypoxia.
Net and HIF1␣ Regulate PHD Expression in Hypoxia-We investigated whether Net and HIF1␣ knockdown would affect expression of PHD at the mRNA level in normoxia (0 h) and early (6 h) and late (12 and 24 h) hypoxia. The specific siRNAs were efficient because they decreased Net and HIF1␣ mRNA levels by at least 80% (Fig. 7, A and E). Hypoxia stimulated the  JULY 9, 2010 • VOLUME 285 • NUMBER 28 expression of PHD2 2-3-fold and PHD3 about 6-fold, as expected (Fig. 7, C, D, G, and H), but did not affect PHD1 levels (Fig. 7, B and F). Net down-regulation had no significant effect on PHD1 and PHD2 levels in normoxia and hypoxia (Fig. 7, B and C) but decreased hypoxic induction of PHD3 (Fig. 7D). HIF1␣ down-regulation inhibited hypoxic induction of both PHD2 and PHD3 mRNA levels (Fig. 7, G and H) but had no significant effect on PHD1 levels (Fig. 7F). These results show that the Net and HIF1␣ regulate different PHDs; Net regulates PHD3, whereas HIF1␣ regulates both PHD2 and PHD3. In  addition, the subsequent effects of the PHDs on the levels of Net and HIF1␣ suggest that transcriptional regulation could contribute to the cross-talk between Net and HIF1␣ at the protein level.

Net and HIF1␣ in Hypoxia Signaling Pathways
Net Physically Interacts with the PHDs and HIF1␣-We also investigated physical interactions between Net and the PHDs. 444 cells were co-transfected with Net and one of the three isoforms of FLAG-tagged PHDs or with the empty vector pcDNA3 as a control. The PHDs were immunoprecipitated with FLAG antibody-coupled beads, and co-precipitated Net was detected by Western blotting. We found that, in normoxia, Net specifically interacted with PHD1 and PHD3 but not detectably with PHD2 (Fig. 8, A and C). Similar results were obtained with co-expressed Net and the three PHDs in HeLa cells (data not shown). Interestingly, under hypoxic mimic conditions (CoCl 2 ), Net interacted with PHD2 as well as with PHD1 and PHD3 (Fig. 8, B and C). We also investigated whether Net interacts with HIF1␣. We co-expressed HIF1␣ and FLAG-Net in HeLa cells and immunoprecipitated FLAG-Net using FLAG (M2) antibody-coupled beads. HIF1␣ specifically co-immunoprecipitated with FLAG-Net (Fig. 8D), indicating that Net physically interacts with HIF1␣ under these conditions. Taken together these data indicate that the PHDs, Net, and HIF1␣ can interact at the protein level and that the complexes that form depend upon hypoxia and the presence of HIF1␣.

DISCUSSION
In this study, we compared the hypoxia/PHDs/Net and hypoxia/PHDs/HIF1␣ pathways in human non-tumorigenic human papillomavirus-positive 444 cells. These cells were used because Net repression of the c-fos oncogene appears to account for their non-transforming properties (16). Hypoxia is a mechanism by which Net repression can be relieved (8), indicating that the 444 cell line is a good model to investigate the consequences of hypoxia-induced loss of Net on transformation. In addition, there are related transformed (HeLa, CGL3) and non-transformed (IMR90 fibroblasts) cells (19,20), which could be used to study the effects of transformation on the role of Net in hypoxia. We show that hypoxia regulates Net and HIF1␣ in 444 cells and that there are similarities and differences in this regulation. They exhibit similar mobility shifts in response to modulators of PHD activity. However the three PHDs regulate Net and HIF1␣ protein stability in different ways, suggesting that the hypoxia-Net and -HIF1␣ pathways are distinct. However, the two pathways are interlinked at the level of regulation of PHD expression and protein-protein interactions (see Fig. 9 for a schematic representation). These links could account for our previous observations that Net and HIF1␣ share a large number of target genes (18).
Our results show that Net and HIF1␣ are differently regulated during the time course of hypoxia, suggesting that they have different roles in the hypoxic response. Net levels are high in normoxia and early hypoxia, whereas HIF1␣ is induced in early hypoxia. Both factors are degraded in late hypoxia. HIF1␣ induction is required for the cellular response to hypoxia, and its degradation in late hypoxia protects cells against necrotic cell death and adapts them to chronic hypoxia (26). The intricate link between Net and HIF1␣ suggests that Net modulates these functions of HIF1␣. The PHDs are known to hydroxylate HIF1␣ on specific proline residues and thus regulate its stability (7,25), which causes a characteristic shift in mobility on SDS-PAGE (23,24). Net undergoes similar mobility shifts using in vitro and in vivo assays, indicating that it may also be hydroxylated. The multiple Net bands in SDS-PAGE might correspond to different posttranslational modifications (21,27), including hydroxylation. PHDs hydroxylate a number of proteins. PHD1 is involved in proline hydroxylation of RbpI, the large subunit of RNA polymerase II (28). The kinase activity of IB kinase ␤ may be inhibited by hydroxylation by PHD1 and PHD2 (29). The stability of activating transcriptional factor 4 (ATF4) (30) and myogenin (31) is regulated by PHD3. However, it has not been proven directly that either protein is hydroxylated. We did not detect hydroxylated prolines in Net by mass spectroscopy, but these studies are not conclusive and are hampered by the proline-rich nature of the protein. 7 We have shown that the three PHDs affect Net in different ways (Fig. 9). PHD1 and PHD3 contribute to stabilization, whereas PHD2 is involved in Net degradation. In addition, we demonstrate that Net physically interacts with PHD2 in hypoxia, whereas PHD1 and PHD3 bind to Net in normoxia and hypoxia. In contrast, PHD2 and PHD3 regulate HIF1␣ stabilization in early and late hypoxia, whereas PHD1 appears to be involved in HIF1␣ degradation in late hypoxia. HIF1␣ has been reported to be mainly regulated by PHD2, although some studies suggest that all three isoforms can hydroxylate HIF1␣ with different efficiency (32). Other proteins are also differentially regulated by the PHDs. PHD3 binds to myogenin and increases its stability (31) and also induces oxygen-dependent degradation of ATF4 (30). PHD1 and PHD2 co-immunoprecipitate with Rpb1 in response to oxidative stress, but PHD1 is necessary for RbpI hydroxylation and consequent degradation, whereas PHD2 has an inhibitory effect on this process (28).
We present evidence that the Net and HIF1␣ pathways are interconnected at several levels. Changes in the initial protein levels of Net influence hypoxic regulation of HIF1␣, suggesting that Net acts as a 'rheostat,' which modulates the HIF1␣-mediated hypoxic response in the cell. Induction of PHD2 and PHD3 in hypoxia partially counterbalances the diminution of oxygen, leading to increased hydroxylation and consequently proteasomal degradation of HIF1␣ (33). Net inhibition down-regulates PHD3 in hypoxia and may thereby delay degradation of HIF1␣. Similarly, HIF1␣ inhibition prevents the hypoxic induction of PHD2, which appears to be involved in Net down-regulation in late hypoxia. This could explain why HIF1␣ inhibition stabilizes Net in late hypoxia. However, the cross-talk between Net and HIF1␣ might also be a result of their physical interaction.
We found that HIF1␣ down-regulation inhibits the hypoxic induction of PHD2 and PHD3 mRNA levels ( Fig. 9), as expected from previous studies (33)(34)(35). Net down-regulation decreased PHD3 mRNA induction in hypoxia, suggesting that Net is a positive transcriptional regulator of its expression. In contrast, Net represses PHD2 and PHD3 expression in transformed mouse skin endothelial cells (SEND) (18), indicating that the role of Net in the hypoxic response is cell type-specific.
Our results show that Net physically interacts with PHD1 and PHD3 in normoxia and in hypoxia mimic (cobalt chloride; Fig. 9). These results suggest that Net could be a substrate for both enzymes. Interestingly, the interaction of PHD3 with myogenin increases its stability by preventing Von Hippel-Lindau-mediated degradation (31), which indicates that Net stabilization by PHDs could be also mediated by hydroxylationindependent mechanism. Interestingly, we found that PHD2 binds to Net only in hypoxic mimic conditions, which could be related to its involvement in the degradation of Net in late hypoxia. Moreover, we found that Net and HIF1␣ physically interact. This interaction could involve the PHDs, perhaps through the formation of a multiprotein complex. An interesting possibility is that HIF1␣ stabilization in hypoxia results in the recruitment of PHD2, which is involved in the degradation of Net in late hypoxia.
In summary, this study demonstrates that Net and HIF1␣ form different signaling pathways, which are intricately linked at several levels. Thus, we have added important insights into the understanding of the mechanism of hypoxic regulation of Net and its integration into HIF1␣ signaling. Pharmacological modulation of Net and HIF1␣ and the investigation of the effect on cell physiology will further increase our understanding of the complex interplay between both transcription factors in the regulation of cellular adaptation to hypoxia. FIGURE 9. Schematic representation of salient features of the hypoxia signaling pathways that involve Net and HIF1␣. The response to 1% O 2 can be divided into normoxia (0 h, low HIF1␣, high Net), early hypoxia (4 -12 h; high Net and HIF1␣), and late hypoxia (12-24 h; low Net and HIF1␣). In normoxia, PHD1/3 bind to and stabilize Net, whereas PHD2/3 degrade HIF1␣. In early hypoxia, Net binds to PHD2 and HIF1␣, in addition to PHD1/3, which is represented as one complex for simplicity. Net stabilizes HIF1␣ and regulates the expression of PHD3. HIF1␣ regulates the expression of PHD2/3. In late hypoxia, PHD2 and HIF1␣ degrade Net, whereas PHD1/2/3 degrade HIF1␣. The terms 'stabilize' and 'degrade' imply that the named proteins participate in these processes, probably with other components that have not been identified in this study.