Regulation of the Hypoxia-inducible Factor 1α by the Inflammatory Mediators Nitric Oxide and Tumor Necrosis Factor-α in Contrast to Desferroxamine and Phenylarsine Oxide

Hypoxic/ischemic conditions provoke activation of the hypoxia-inducible factor-1 (HIF-1), which functions as a transcription factor. HIF-1 is composed of the HIF-1α and -β subunits, and stability regulation occurs via accumulation/degradation of HIF-1α with the notion that a prolyl hydroxylase accounts for changes in protein level. In addition, there is evidence that HIF-1 is up-regulated by diverse agonists during normoxia. We investigated the impact of inflammatory mediators nitric oxide (NO) and tumor necrosis factor-α (TNF-α) on HIF-1α regulation. For comparison, LLC-PK1 cells were exposed to hypoxia, stimulated with desferroxamine (DFX, known to mimic hypoxia), and the thiol-cross-linking agent phenylarsine oxide (PAO). Although all stimuli elicited HIF-1α stabilization with differences in the time-dependent accumulation pattern, significant variations appeared with regard to signaling. With the use of a superoxide anion (O 2 − ) generator, we established an O 2 − -sensitive pathway that blocked HIF-1α stabilization in response to NO and TNF-α while DFX- and PAO-evoked HIF-1α stabilization appeared O 2 − -insensitive. NO and TNF-α signaling required phosphorylation events, especially activation of the phosphatidylinositol 3-kinase/Akt, which is in contrast to DFX and PAO. Based on HIF-1-dependent luciferase reporter gene analysis, it was found that, in contrast to NO and TNF-α, PAO resembled a stimulus that induced a dysfunctional HIF-1 complex. These data indicate that diverse agonists activate HIF-1α under normoxic conditions by employing different signaling pathways.

The transcription factor hypoxia-inducible factor-1 (HIF-1) 1 is a heterodimer composed of the helix-loop-helix/Per-Arnt-Sim protein HIF-1␣ and the aryl hydrocarbon nuclear translocator, also known as HIF-1␤ (1)(2)(3). An active HIF-1 complex accumulates in the nucleus; binds to a specific DNA sequence, the HIF-1 binding site within the hypoxia response element (HRE); and enhances transcription of hypoxia-inducible genes, such as erythropoietin or vascular endothelial growth factor. The availability of HIF-1 is mainly determined by the presence/absence of HIF-1␣ (4,5). In many cell types, both HIF-1␣ and HIF-1␤ appear to be constitutively expressed at the mRNA level, whereas, on protein level, HIF-1␣ is degraded under normoxic conditions, which contrasts with permanently expressed HIF-1␤. Studies in von Hippel-Lindau-defective renal cell carcinomas indicated that the von Hippel-Lindau protein fulfills a critical function in HIF-1␣ degradation, thus accounting for the extremely short protein half-life (6). However, accumulation of HIF-1␣ that promotes active HIF-1 complex formation by hypoxia is not fully understood. Oxygen species such as superoxide (O 2 Ϫ ) or hydrogen peroxide (H 2 O 2 ) have been proposed to limit HIF-1␣ stability (7). A postulated source for these species is a NAD(P)H-metabolizing membrane-bound type b cytochrome quite similar to the respiratory burst oxidase in phagocytes. In addition to these intracellular redox changes, phosphorylation cascades such as mitogen-activated protein kinases have been ascribed to stabilize HIF-1␣, but precise signaling mechanisms and their cross-talk have not been not fully defined (8 -10).
Activation of HIF-1 as an adaptive response was first described for conditions of decreased oxygen pressure. Therefore, most mechanistic and functional studies on HIF-1 regulation refer to hypoxic conditions. More recent evidence suggests that HIF-1 can be activated by growth factors, cytokines, hormones, or nitric oxide (NO) as well with very little information on signal transduction pathways being involved (11)(12)(13)(14)(15). Zhong et al. (11) verified a role of phosphatidylinositol 3-kinase (PI3K), Akt phosphorylation, and FRAP activation for HIF-1␣ induction in response to hypoxia and epithelial growth factor treatment. This pathway is negatively regulated by PTEN (phosphatase and tensin homologue deleted in chromosome ten) and loss of function correlated with tumor angiogenesis. For the signaling molecule NO, we noticed that the general kinase inhibitor genistein and, more specifically, the PI3K inhibitors wortmannin or Ly 294002 blocked HIF-1␣ accumulation, whereas mitogen-activated protein kinases were not involved (16).
Herein, we have compared different stimuli for HIF-1␣ stability regulation and examined the involvement of phosphorylation events with a focus on the PI3K/Akt pathway. We concentrated on two inflammatory mediators, NO and tumor necrosis factor ␣ (TNF-␣). Both agents have been shown to activate HIF-1, but signaling mechanisms largely remained unclear. We compared our results with hypoxia and the iron chelator desferroxamine (DFX) often chosen to mimic hypoxia. For further consideration we included a fourth agent, phenylarsine oxide (PAO), known to cross-link vicinal -SH groups since modifications of the -SH moiety of cysteine by S-nitrosation or oxidation are discussed for NO signaling during activation/inhibition of transcription factors such as nuclear factor B or AP-1 (17,18).
Although all stimuli elicited HIF-1␣ accumulation, their signaling pathways differed significantly. The HIF-1␣ response was blocked by O 2 Ϫ generation or by interrupting the PI3K/Akt pathway in case of hypoxia and the inflammatory mediators NO and TNF-␣, whereas the DFX or PAO responses were not affected at all. Interestingly, PAO induced a strong HIF-1␣ protein accumulation but failed to form an active HIF-1 DNAbinding complex or to provoke reporter gene activation.
Cell Culture-Proximal tubular LLC-PK 1 cells were cultured in Dulbecco's modified Eagle's medium with 1 g/liter glucose, supplemented with 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal calf serum. Cells were transferred two times a week, and medium was changed prior to experiments. Cells were kept in a humidified atmosphere of 5% CO 2 in air at 37°C. For hypoxic stimulation, dishes were placed in an incubator with 1% O 2 , 5% CO 2 , and 94% N 2 in a humidified atmosphere, for times indicated.
Transfections-4 ϫ 10 5 LLC-PK 1 cells were seeded in 6-cm dishes 1 day before transfection. With about 70% confluence, cells were transfected with 3 g of the different plasmids. Therefore, 35 l of a 5 mM polyethylenimine solution was mixed with 85 l of medium without serum and 3 g of the selected DNA. Mixtures were vortexed for 5 s, incubated for 30 min at room temperature, and added dropwise to the cells containing 2.5 ml of medium with supplements. 4 h later, medium was changed and 24 h after transfection, cells were stimulated as indicated. Transfections of HepG2 cells and luciferase reporter gene analysis, in response to hypoxia, NO, TNF-␣, DFX, or PAO, was performed as described previously (20).
Gel Shift Assay-Oligonucleotides for gel shift assays were synthesized by MWG (Ebersberg, Germany) and contained the sequence of the HIF-1 binding site derived from the human transferrin gene. The sequences were as follows: sense, 5Ј-TTCCTGCACGTACACACAAAGCG-CACGTATTTC-3Ј; and antisense, 5Ј-GAAATACGTGCGCTTTGTGTG-TACGTGCAGGAA-3Ј. For radioactive labeling, 1.25 pmol of annealed oligonucleotides were incubated with 2.5 l of 10ϫ polynucleotide kinase phosphorylation buffer, 1 l of phosphatase-free T4 polynucleotide kinase, and 50 Ci of [␥-32 P]ATP in a final volume of 20 l at 37°C for 25 min. Labeling reactions were terminated by the addition of 1 l of 0.5 M EDTA, and unlabeled oligonucleotides were separated by using Chromaspin-10 columns. The efficiency of radioactive labeling was measured with a counter. Afterward, HIF-1 binding reactions were set up in a volume of 20 l and nuclear extracts (5 g protein) were incubated in a buffer with a final concentration of 50 mM KCl, 10 mM Tris, pH 7.7, 5 mM DTT, 1 mM EDTA, 1 mM MgCl 2 , 5% glycerol, 0.03% Nonidet P-40, 400 ng of salmon testes DNA, and 40,000 cpm 32 P-labeled oligonucleotide. Incubations were overnight at 4°C, and samples were resolved by electrophoresis on 5% polyacrylamide gels (polyacrylamide:bisacrylamide, 29:1) at room temperature with 110 V. For supershift experiments, 0.25 g of HIF-1␣ antibody was added to the reactions 1 h before running the gel.
Statistical Analysis-Each experiment was performed at least three times, and representative pictures are shown. Values presented are mean Ϯ S.E.

RESULTS
HIF-1␣ Accumulation and the Interference by the Redox Cycler DMNQ-A major target cell type affected during renal ischemia/reperfusion injury and several inflammatory processes in the kidney are tubular epithelial cells, whereas in the liver hepatocytes are affected. Therefore, we used tubular LLC-PK 1 or HepG2 cells in this study. To examine the impact of the inflammatory mediators NO and TNF-␣ on HIF-1␣ accumulation, we applied 100 M amounts of the chemically distinct NO donors S-nitrosoglutathione or SpNO, or recombinant human TNF-␣ at a concentration of 500 ng/ml. Furthermore, we exposed cells to 1% hypoxia, and to imitate hypoxia we employed DFX at a concentration of 100 M. To study the impact of vicinal -SH group modification, PAO at a concentration of 1 M was used. Agonist treatments were neither apoptotic nor necrotic to the cells (data not shown).
GSNO at a concentration of 100 M elicited a fast but transient HIF-1␣ response, as determined by Western blot analysis. HIF-1␣ accumulation revealed a maximum between 1 and 2 h with a declining signal afterward (Fig. 1A). TNF-␣ represented a slow inducer for HIF-1␣. Protein accumulation in response to TNF-␣ started around 6 h but remained stabilized up to 18 h (Fig. 1B). Moreover, we noticed HIF-1␣ stabilization between 2 and 4 h after DFX or PAO stimulation (Fig. 1, C and D) with the further notion that HIF-1␣ stayed elevated up to 8 h (data not shown). The relatively fast accumulation pattern observed after GSNO or DFX treatment resembled the one described for hypoxia (data not shown). Although the appearance of HIF-1␣ varied in time, we showed that diverse stimuli promoted a strong HIF-1␣ accumulation during normoxia.
It has been proposed that the intracellular level of oxygen species (ROS) determine degradation or stabilization of HIF-1␣. To investigate whether ROS interfere with signal transduction of our stimuli, we used the redox cycler DMNQ. DMNQ penetrates the plasma membrane and releases O 2 Ϫ inside the cell (22). For the following experiments, we used defined concentrations and time points, as revealed by the experiments mentioned above, to achieve maximal HIF-1␣ induction in combination with 0, 1, 5, or 10 M DMNQ. HIF-1␣ accumulation was monitored by Western blot analysis. In the case of GSNO and TNF-␣ treatment, increasing concentrations of DMNQ attenuated and further on abolished the HIF-1␣ signal at higher concentrations (Fig. 2, A and B). Under conditions of DFX or PAO supplementation, ROS generation failed to modulate HIF-1␣ accumulation (Fig. 2, C and D). These results point to signal transduction pathways as either being ROS-sensitive or -insensitive, with the observation that DMNQ by itself did not stabilize HIF-1␣.
HIF-1␣ Accumulation and the Interference by Phosphorylation Cascades-Besides the regulatory role of intracellular ROS, phosphorylation events have been linked to HIF-1␣ stabilization. With the use of the unselective kinase inhibitor genistein, we probed for the involvement of phosphorylation cascades. In these experiments, genistein was preincubated for 30 min to assure cell membrane penetration. Genistein at a concentration of 100 M showed no effect on HIF-1␣ accumulation by its own (data not shown), but completely attenuated GSNO-induced HIF-1␣ accumulation detected after 1, 2, and 4 h (Fig. 3A). When using TNF-␣, DFX, and PAO, the effect of genistein was tested at 4 or 6 h, only. Similar to GSNO, genistein significantly attenuated the HIF-1␣ signal triggered by TNF-␣ (Fig. 3B), whereas the DFX-or PAO-elicited HIF-1␣ accumulation appeared unaffected by genistein (Fig. 3, C and  D).
Zhong et al. (11) depicted the PI3K/Akt pathway to be essential for hypoxia-induced HIF-1 induction, and we described similar results for NO signaling and HIF-1␣ stabilization. Stimulation of LLC-PK 1 cells with GSNO led to Akt activation/ phosphorylation in close correlation to HIF-1␣ accumulation as inhibition of the PI3K with Ly 294002 suppressed Akt phosphorylation and HIF-1␣ stabilization (16). However, there is little or no information on the role on the PI3K/Akt pathway in the response of HIF-1␣ to TNF-␣, DFX, or PAO. Therefore, we analyzed the HIF-1␣ response elicited by two chemically diverse NO donors GSNO and SpNO, as well as hypoxia itself. Wortmannin at a concentration of 100 nM resembles a specific PI3K inhibitor with no effect on HIF-1␣ by its own (Fig. 4A). Similar to genistein, wortmannin was preincubated for 30 min prior to agonist addition. Wortmannin blocked HIF-1␣ accumulation in response to GSNO or SpNO when assayed after 1, 2, or 4 h (Fig. 4A). The PI3K inhibitor appeared effective after TNF-␣ stimulation as well (Fig. 4B) and, as expected, also blocked HIF-1␣ stabilization in response to hypoxia (Fig. 4C)  (11). The experiments with DFX and PAO in the absence or presence of even an increased concentration of 10 M wortmannin failed to affect HIF-1␣ accumulation (Fig. 4, D and E), which neglects the requirement of a wortmannin-sensitive pathway for these agonists. These observations confirmed our results obtained with genistein.
To further establish the role of PI3K and Akt for NO-and TNF-␣-induced HIF-1␣ regulation, LLC-PK 1 cells were transfected with the pSR␣ plasmid encoding the wild type form of the regulatory PI3K subunit p85 (WTp85) or a dominant-negative p85 form (⌬p85), which lacks amino acids 479 -513 (19). Comparable amounts of p85 due to overexpression of WTp85 or ⌬p85 was documented in all experiments by Western blot analysis with antibodies against p85 (herein, only shown in combination with GSNO; Fig. 5A). Transfections of WTp85 did not alter HIF-1␣ accumulation in response to all stimuli tested. This is exemplified for GSNO stimulation when HIF-1␣ was detected after 1, 2, and 4 h (Fig. 5A). Opposite effects on the HIF-1␣ signal were achieved by overexpressing the dominantnegative form ⌬p85. GSNO-and TNF-␣-evoked HIF-1␣ stabilization was significantly attenuated by ⌬p85 expression (Fig.  5, B and C). In case of TNF-␣, the 16-h time point was chosen in these experiments to achieve a stronger HIF-1␣ signal, whereas GSNO responses are examined after 1, 2, and 4 h. As seen in unstimulated controls, plasmid transfections did not influence basal expression of HIF-1␣. ⌬p85 transfection experiments in combination with the agonist DFX or PAO underlined the results ascertained with wortmannin (Fig. 5, D and   E). Apparently, an inactive PI3K failed to interrupt the signaling pathway of DFX or PAO that leads to HIF-1␣ accumulation.
Whether PI3K signaling is mediated via Akt, also known as protein kinase B (PKB), was tested by overexpression of the dominant-negative Akt protein. Dysfunctional Akt was achieved by mutating the ATP-binding site Lys 179 to Ala (23). In addition, a consensus sequence for both myristoylation and palmitylation (m/p) was hooked to the construct as activated Akt is recruited to the membrane via its PH domain. m/p has been shown to be sufficient to localize a number of cytosolic proteins to the plasma membrane (24). Therefore, only Akt activation is blocked but not its translocation (23). Control experiments with the empty pCMV5 vector showed no interference with HIF-1␣ accumulation. Overexpression of PKB␣K179A suppressed HIF-1␣ stabilization initiated by 100 M GSNO (Fig. 6A). Suppression of HIF-1␣ accumulation in cells overexpressing PKB␣K179A was confirmed toward hypoxia or spermine NO (Fig. 6B). Along that line, TNF-␣-evoked HIF-1␣ stabilization was attenuated in PKB␣K179A-transfected cells (Fig. 6C). In contrast, DFX-or PAO-mediated HIF-1␣ accumulation remained unaffected by PKB␣K179A overexpression (Fig. 6, D and E). Equal amounts of PKB expression due to transfection of the PKB expression vectors were assured by Western blot analysis with an antibody against Akt under all conditions. This is exemplified for GSNO stimulation (Fig. 6A). Our results suggest that accumulation of HIF-1␣ by the inflammatory mediators NO and TNF-␣, in analogy to hypoxia, is achieved through phosphorylation cascades, specifically via PI3K and Akt, whereas DFX-and PAO-evoked pathways do not include these phosphorylation events.
Impact of Inflammatory Mediators on HIF-1 DNA Binding and HIF-1-dependent Transcriptional Activity-HIF-1 DNA binding and formation of an active HIF-1 complex have been shown previously for NO, TNF-␣, and DFX (12,14). In case of PAO, this is the first report describing a strong increase in the HIF-1␣ protein levels. In order to assess HIF-1 DNA binding, we performed electrophoretic mobility shift assays with radioactive-labeled oligonucleotides containing a HIF-1 binding site. As expected, GSNO, TNF-␣, and DFX addition not only displayed HIF-1␣ protein accumulation but also HIF-1 binding to the oligonucleotide. Specificity of the HIF-1 complex was assured by supershifts with 0.25 g of the HIF-1␣ antibody (Fig.  7).
Stimulation with 1 M PAO for 1, 2, or 4 h revealed no HIF-1 binding to the radioactive oligonucleotide, although Western blot analysis of the same samples used for shift assays revealed prominent HIF-1␣ stabilization. Therefore, we studied the activation of a HIF-1-dependent luciferase reporter gene construct containing three hypoxia-response elements from the erythropoietin gene in response to hypoxia: GSNO, PAO, and TNF-␣ (Table I). For a more pronounced and longer lasting HIF-1 signal, we used 1 mM of the NO donor in these studies. As expected, hypoxia and GSNO promoted strong and significant reporter gene activation in HepG2 cells. With a somewhat lower activity, a positive response was shared by TNF-␣. In contrast, PAO was unable to evoke reporter gene activation.
These results are in analogy to information obtained from gel shift assays. We conclude that HIF-1␣ stabilization and HIF-1 activity in response to hypoxia, GSNO, TNF-␣, DFX, and PAO are distinguished based on the sensitivity to O 2 Ϫ , phosphorylation events, DNA binding, and reporter gene activation. DISCUSSION Hypoxia is of major (patho)physiological importance in evoking stabilization and activation of HIF-1. It is also appreciated that HIF-1 is subjected to complex modulation under normoxia as well. Herein, we concentrated on stability regulation of HIF-1␣ by the inflammatory mediators NO (derived from GSNO or spermine NO) as well as TNF-␣ and compared the results with classical agonists such as hypoxia or DFX. For mechanistic considerations we included the thiol-modifying agent PAO to obtain information on signaling pathways in regulating HIF-1␣ accumulation and HIF-1 transcriptional activity.
Initial reports pointed to an inhibitory action of NO on hypoxia-induced HIF-1␣ stabilization and/or HIF-1 target gene activation (25)(26)(27). Subsequently, it turned out that the choice of the NO donor, the concentration, and probably the time of application appeared important in affecting HIF-1␣ stabilization. Several groups, including our own, have shown in different cell types that, with the exception of sodium nitroprusside, chemically diverse NO donors, transfection of NO synthase, or macrophage-derived NO evoked HIF-1␣ accumulation and HIF-1 DNA binding followed by target gene expression (14,15,28,29). Signaling pathways that promoted HIF-1␣ stabilization in response to NO pointed to a genistein-sensitive phosphorylation cascade, which appears in analogy to the impact of genistein on hypoxia-evoked HIF-1␣ stabilization (10). Along that line, our results show the involvement of the PI3K/Akt pathway as the inhibitor wortmannin or more specifically transfections of inactive kinases attenuated GSNO-induced HIF-1␣ accumulation (Figs. 4 -6). Information about cytokines and HIF-1 appeared similarly diverse. Although interleukin-1␤ has been shown to initiate HIF-1␣ accumulation and to potentiate hypoxia-evoked HIF-1 DNA binding (12), we failed to detect HIF-1␣ stabilization when stimulating tubular LLC-PK 1 cells with 25 units/ml interleukin-1␤ up to 24 h. In addition, interferon-␥ and lipopolysaccharide were ineffective as well (data not shown), whereas we found TNF-␣ to be a slow but strong HIF-1 inducer. This is in contrast to observations of Hellwig-Bü rgel et al. (12). They noticed an additive effect of TNF-␣ during hypoxic stimulation but no impact of TNF-␣ on HIF-1␣ mRNA, protein level, or HIF-1 activity by itself when performing experiments in tubular and hepatoma cells for 4 h. Thornton et al. (30) used fibroblasts to show an increase of HIF-1␣ mRNA at 3 h after TNF-␣ with no data on protein accumulation or DNA binding of HIF-1. In part these results point to some cell specificity but, in the case of tubular cells, underscore a time-dependent effect. We did not recognize TNF-␣-evoked HIF-1␣ accumulation before 6 h, but a maximal effect was not achieved until 10 -18 h (Fig. 1). In general, the stimulatory action of cytokines and GSNO on HIF-1␣ accumulation may imply a functional role of HIF-1 during inflammatory settings. Herein, we then elucidated signal transducing pathways. Under hypoxic conditions the formation of oxygen species, specifically H 2 O 2 , attenuated HIF-1 activation (2)  which should be clarified with the use of authentic ONOO Ϫ in further experiments. However, the impact of oxygen or nitrogen species on HIF-1␣ accumulation presently is not fully understood. Obviously, signaling cascades that depend on phosphorylation cascades to achieve a HIF-1␣ signal such as NO or TNF-␣ appeared ROS-sensitive, whereas those of DFX and PAO did not.
The action of DMNQ-derived species seems to contradict studies that show a requirement of mitochondria-derived ROS in stabilizing HIF-1␣ during hypoxia (3,32,33). Chandel et al. (33) reported a wortmannin-sensitive PI3K pathway to account for HIF-1␣ stabilization during ROS signaling. One may speculate that a defined cellular redox environment senses redox changes elicited by either NO or O 2 Ϫ and transmits these changes via phosphorylation cascades into a functional HIF response. Cells equipped with variable amounts of defense systems to fight radical formation may then reveal a variable HIF response to NO or O 2 Ϫ formation.
In further experiments we provided evidence that PI3K and Akt are essential signaling components for GSNO and TNF-␣, whereas DFX and PAO signaled PI3K independently. Previously, the involvement of the PI3K/Akt pathway in affecting HIF-1␣ stability was identified for hypoxia in addition to activation of the MAPK cascade. To our knowledge, the role of MAPK for NO-, TNF-␣-, DFX-, or PAO-triggered HIF-1␣ accumulation has not been investigated. In case of GSNO, neither p42/p44, p38, nor c-Jun N-terminal kinases were activated and MAPK inhibitors such as PD 98058 or SB 203580 did not attenuate HIF-1␣ stabilization. This is in contrast to angiotensin II signaling (13), where HIF-1 activation was attenuated by blocking PI3K or MAPK pathways.
Stimulation with PAO revealed a unique pattern of HIF-1␣ accumulation and HIF-1 activation. First, HIF-1␣ protein stabilization appeared with a time pattern remarkably similar to that described for hypoxia, DFX, or cobalt chloride. Interestingly, neither O 2 Ϫ production nor kinase inhibition disturbed PAO-elicited HIF-1␣ accumulation. PAO binds with high affinity to vicinal -SH groups. Predicted on the human primary sequence, Cys 334 and Cys 337 may be targeted in HIF-1␣, but this protein region is not considered to be essential for HIF-1␣ stabilization, interaction with other proteins, or DNA binding. However, it cannot be excluded that other potential -SH groups provide targets in the three-dimensional structure. Direct bind-FIG. 6. Dysfunctional Akt kinase blocked the NO, TNF-␣, and hypoxia pathway but appeared ineffective in case of DFX and PAO. 4 ϫ 10 5 LLC-PK 1 cells were seeded 1 day before the transfection with 3 g of either pCMV5. or pCMV5.-m/p-PKB␣K179. For details, see "Experimental Procedures." 24 h after transfection, LLC-PK 1 cells were stimulated with vehicle, 100 M GSNO, hypoxia (1%), 250 M SpNO, 500 ng/ml TNF-␣, 100 M DFX, or 1 M PAO for times indicated. HIF-1␣ protein levels were determined by Western blot analysis, followed by reprobing the blot against Akt. Each experiment was performed at least three times, and representative data are shown.
FIG. 7. PAO induced a dysfunctional HIF-1 complex in contrast to GSNO, TNF-␣, and DFX. LLC-PK 1 cells were treated with vehicle, 100 M GSNO, 500 ng/ml TNF-␣, 100 M DFX, or 1 M PAO for times indicated. Nuclei were prepared and incubated overnight with a radioactive-labeled oligonucleotide containing a HIF-1 binding site. Specific (HIF-1) and nonspecific (n.s.) bands are indicated. Supershifting (SS) of the HIF-1-hypoxia-responsive-element complex was achieved as described. For details, see "Experimental Procedures." Data are representative for at least three independent experiments. GNSO, PAO, and TNF-␣ HepG2 cells were transiently transfected with luciferase reporter genes containing three copies of the HRE of the erythropoietin gene (pGLEPOHRE). Transfected cells were stimulated for 16 h with hypoxia (1% oxygen), 1 mM GSNO, 1 M PAO, and 500 ng/ml TNF-␣. The increase in luciferase activity was related to induction of the control vector pGL3 (control) set as 100% (n ϭ 3-5). ing of PAO to HIF-1␣ would explain a transcriptional dysfunctional HIF-1 complex (Fig. 7, Table I) and the lack of interference by other signaling pathways. Further studies are necessary to explore how binding of PAO stabilizes HIF-1␣. Obviously, PAO will be a useful agent to study stabilization/ regulation of HIF-1␣ in more detail.
Our results establish that GSNO and TNF-␣, besides hypoxia, stabilize HIF-1␣ either in tubular LLC-PK 1 or HepG2 cells. Evidently, the role of HIF-1 is not restricted to hypoxic conditions and thus may contribute to inflammatory episodes that are characterized by massive NO and/or TNF-␣ formation. What kind of signal transduction pathways, except of PI3K/ Akt, MAPK, or ROS signaling are involved needs further investigation. Signaling cross-talk will influence HIF-1-dependent target gene expression to orchestrate hypoxic and inflammatory settings.