Stabilization of Hypoxia-inducible Factor-1α by Prostacyclin under Prolonged Hypoxia via Reducing Reactive Oxygen Species Level in Endothelial Cells*

Hypoxia-inducible factor-1 (HIF-1) takes part in the transcriptional activation of hypoxia-responsive genes. HIF-1α, a subunit of HIF-1, is rapidly degraded under normoxic conditions by the ubiquitin-proteosome system. Hypoxia up-regulates HIF-1α by inhibiting its degradation, thereby allowing it to accumulate to high levels with 3–6 h of hypoxia treatment and decreasing thereafter. In vascular tissues, prostacyclin (prostaglandin I2 (PGI2)) is a potent vasodilator and inhibitor of platelet aggregation and is known as a vasoprotective molecule. However, the role of PGI2 in HIF-1 activation has not been studied. In the present study, we investigated the effect of PGI2 on HIF-1 regulation in human umbilical vein endothelial cells under prolonged hypoxia (12 h). Augmentation of PGI2 via adenovirus-mediated gene transfer of both cyclooxygenase-1 and PGI2 synthase activated HIF-1 by stabilizing HIF-1α in cells under prolonged hypoxia or the hypoxia-normoxia transition but not under normoxia. Exogenous H2O2 abolished PGI2- and catalase-induced HIF-1α up-regulation, which suggests that degradation of HIF-1α under prolonged hypoxia is through a reactive oxygen species-dependent pathway. Moreover, PGI2 attenuated NADPH oxidase activity by suppressing Rac1 and p47phox expression under hypoxia. These data demonstrate a novel function of PGI2 in down-regulating reactive oxygen species production by attenuating NADPH oxidase activity, which stabilizes HIF-1α in human umbilical vein endothelial cells exposed to prolonged hypoxia.

Hypoxia induces a number of cellular responses, such as angiogenesis, erythropoiesis, and glycolysis, through both gene regulation and post-translational modification of proteins. Hypoxia-inducible factor-1 (HIF-1) 2 takes part in the transcriptional activation of hypoxia-responsive genes through binding to the hypoxia-responsive element (HRE) in the promoter or enhancer regions and activating a number of genes (1)(2)(3)(4). HIF-1 is a heterodimer composed of HIF-1␣ and HIF-1␤. HIF-1␣ is rapidly degraded under normoxic conditions by the ubiquitin-proteosome system, whereas HIF-1␤ is constitutively expressed (5). Under hypoxia, HIF-1␣ has been shown to be up-regulated to high levels at 3-6 h and decrease thereafter (6,7). A number of studies have focused on the mechanism of HIF-1␣ stabilization under hypoxia for 4 -6 h. Recently, a natural antisense HIF-1␣ was suggested to down-regulate HIF-1␣ in A549 cells under prolonged hypoxia (6). However, regulation of HIF-1␣ under prolonged hypoxia (12 h) in endothelial cells remains largely unknown.
Under normoxic conditions, HIF-1␣ is regulated through hydroxylation of proline residues by prolyl hydroxylase enzymes (8,9). The von Hippel-Lindau tumor suppressor protein (pVHL) associates with hydroxylated HIF-1␣ and targets it for ubiquitination and rapid degradation (10). Under hypoxia, prolyl hydroxylase is inactivated through oxygen-sensing mechanisms, and the unmodified HIF-1␣ accumulates, permitting dimerization with HIF-1␤ (3,11). Various redox-dependent signaling pathways for oxygen sensing and HIF-1␣ stabilization have been proposed. Under hypoxia, the level of reactive oxygen species (ROS) generated by NADPH oxidase is attenuated because of low pO 2 , which leads to stabilization of HIF-1␣ (11,12). However, contradictory results suggest that the NADPH oxidase inhibitor has differential inhibitory effects on HIF-1␣ regulation under hypoxia (13,14). Moreover, overexpression of NADPH oxidase 1 up-regulates HIF-1␣ (15), whereas stabilization of HIF-1␣ is not altered in gp91 phox knock-out mice (16). These results demonstrate controversial HIF-1␣ regulatory mechanisms via ROS generated by NADPH oxidase under hypoxia. Another model proposes that increased ROS production in mitochondria results in HIF-1␣ stabilization under hypoxia (17,18). However, paradoxical results have been reported in cells lacking mitochondria (19) or a functional mitochondrial respiratory chain (20). Although results are conflicting, they suggest that redox-dependent signaling is involved in the HIF-1 activation pathway. Nevertheless, whether ROS signaling is involved in HIF-1␣ regulation under prolonged hypoxia remains elusive.
Prostacyclin (prostaglandin I 2 (PGI 2 )) is the major eicosanoid produced in vascular endothelial and smooth muscle cells. PGI 2 is a potent vasodilator and inhibitor of platelet aggregation and monocyte attachment and known as a vasoprotective molecule. It is synthesized by a series of enzymatic reactions whereby cytosolic phospholipase A 2 cleaves arachidonic acid from phospholipids, cyclooxygenase (COX) converts arachidonic acid to PGH 2 , and PGI 2 synthase (PGIS) catalyzes PGH 2 to PGI 2 (21,22). Two COX isoforms (COX-1 and COX-2) have been identified. COX-1 is constitutively expressed, whereas COX-2 is inducible (22,23) and is the primary source of PGI 2 biosynthesis, espe-cially in the cardiovascular system (24). Previously, we have shown that overexpression of both COX-1 and PGIS by adenovirus-mediated gene transfer can selectively increase PGI 2 production in endothelial cells without a concurrent overproduction of other prostanoids (25,26).
In the present study, we investigated the effect of PGI 2 on HIF-1␣ regulation in human umbilical vein endothelial cells (HUVECs) under prolonged hypoxia. PGI 2 up-regulated HIF-1␣ under prolonged hypoxia via down-regulating ROS production. This finding demonstrates a novel function of PGI 2 and sheds light on the regulatory mechanism of HIF-1␣ under prolonged hypoxia.

EXPERIMENTAL PROCEDURES
Cell Culture-HUVECs in passages 3-5 were isolated from freshly obtained umbilical veins and cultured in 95% air and 5% CO 2 at 37°C as described previously (25). Hypoxia culture was in a gas-controlled chamber (Forma Scientific) maintained at 1% oxygen, 94% N 2 , and 5% CO 2 at 37°C. PGE 2 and iloprost were purchased from Cayman Chemicals (Ann Arbor, MI). Cyclohexamine (10 g/ml) (Invitrogen) was added before the cells were exposed to normoxia, and MG132 (10 M) was added 2 h before the cells were exposed to normoxia.
Recombinant Adenovirus Preparation-Recombinant adenoviruses containing a human phosphoglycerate kinase (PGK) promoter and a polyadenylation signal of bovine growth hormone were constructed by homologous recombination in human embryonic kidney 293 cells (American Type Culture Collection CRL number 1573) as described previously (25). Adenoviruses carrying both COX-1 and PGIS (Ad-COP1) and catalase (Ad-catalase) were prepared as described previously (26,27). The adenovirus with no insert, Ad-PGK, was used as a control. HUVECs were infected with recombinant adenovirus at a multiplicity of infection (plaque-forming unit/cell) of 25 for 48 h for all experiments.
Luciferase Reporter Assay-An HRE reporter construct, pHRE-Luc, was prepared as follows. The luciferase gene was excised from pGL3-Basic vector (Promega, Madison, WI) and cloned into an HRE-cytomegalovirus vector (a gift of Dr. L. Y. Chau, Academia Sinica, Taipei, Taiwan), which contains 4 HREs and a minimal cytomegalovirus promoter. pRL-tk, which carries a Renilla gene, was used as an internal control of transfection. Transfection was conducted by 15 ml of Lipofectamine 2000 (Invitrogen) with 10 g of DNA in a 35-mm dish. The transfection efficiency was ϳ18% by transfection of pEGFP-N1, and the fraction of enhanced green fluorescent protein-expressing cells was measured by flow cytometry. HUVECs were co-transfected with pHRE-Luc and pRL-tk, infected with Ad-PGK or Ad-COP1 at a multiplicity of infection of 25, and cultured in complete medium for 48 h. The cells were then exposed to normoxic or hypoxic conditions. At the end of the incubation, the cells were washed with phosphate-buffered saline and lysed, and luciferase activity was measured with the use of luciferase assay reagent (Promega). Relative promoter activity was measured by the ratio of luciferase to Renilla activity.

Reverse Transcription-Polymerase Chain Reaction (RT-PCR) and
Real-time PCR-RNA of HUVECs was isolated by the TRIzol RNA isolation system (Invitrogen). A total of 1.5 g of RNA was used for cDNA synthesis with oligo(dT) and SuperScript II reverse transcriptase (Invitrogen). cDNAs were used for PCR amplification with the primers 5Ј-TGCCAAGTGGTCCCAG-3Ј and 5Ј-GTGAGGTTTGATC-CGC-3Ј for vascular endothelial growth factor (VEGF, 277 bp) and 5Ј-GTTTACTAAAGGACAAGTCACC-3Ј and 5Ј-TTCTGTTTGTT-GAAGGGAG-3Ј for HIF-1␣ (193 bp). PCR of ␤-actin was used as a reference with the primers 5Ј-GGGRCAGAAGGATTCCTATG-3Ј and 5Ј-GGTCTCAAACATGATCTGGG-3Ј (238 bp). The regimen of RT-PCR amplification was as follows: 5 cycles of 95°C for 40 s, 58°C for 40 s, and 72°C for 40 s followed by 20 cycles of 95°C for 40 s, 55°C for 40 s, and 72°C for 40 s for VEGF. Quantitative real-time PCR was performed in a 10-ml reaction volume by the standard protocols of the Roche Applied Science LightCycler system under the regimen of 40 cycles of 95°C for 20 s, 52°C for 10 s, and 72°C for 10 s for HIF-1␣. The relative gene expression was obtained by ⌬C T assay (⌬C T ϭ C T (Target gene) Ϫ C T␤-actin ). All reactions were performed in triplicates and normalized by reference gene expression.
ROS Measurement-Intracellular ROS generation was assessed with use of 2Ј,7ЈϪdichlorofluorescein diacetate (Molecular probes, Eugene, OR). ROS in cells causes oxidation of 2Ј,7ЈϪdichlorofluorescein diacetate, yielding the fluorescent product 2Ј,7ЈϪdichlorofluorescein. The cells were cultured in medium containing 2Ј,7ЈϪdichlorofluorescein diacetate (10 mol/liter) for 30 min. The medium was then removed, cells were lysed and centrifuged, and the level of 2Ј,7ЈϪdichlorofluorescein in the supernatant was measured with the use of a spectrophotometer at an excitation:emission ratio of 488:525 nm.
Statistical Analysis-Unpaired Student's t test was used for comparisons between treatments, and data are expressed as mean Ϯ S.E. A p value Ͻ 0.05 was considered significant.

RESULTS
Effect of PGI 2 on HIF-1␣ Regulation in HUVECs Exposed to Hypoxia-HUVECs were infected with an adenovirus carrying both COX-1 and PGIS, Ad-COP1, at a multiplicity of infection of 25 for 48 h. Western blot analysis showed both COX-1 and PGIS protein levels highly increased in Ad-COP1infected cells compared with cells mock-infected with Ad-PGK under both hypoxia and normoxia (Fig. 1, A and B). The level of 6-keto-PGF 1␣ , a stable hydrolyzed product of PGI 2 , in Ad-COP1-infected cells was augmented to ϳ10ϫ that in mock-infected and control cells (Fig. 1C). To examine the effect of PGI 2 on HIF-1␣ regulation, HUVECs were infected with Ad-PGK or Ad-COP1 and exposed to hypoxia for 3-15 h. HIF-1␣ was up-regulated in cells exposed to hypoxia for 3-9 h but not in cells exposed to normoxia ( Fig. 2A). In control cells, HIF-1␣ levels were down-regulated to a very low level at 12 and 15 h. Surprisingly, HIF-1␣ was significantly up-regulated in cells infected with Ad-COP1 under hypoxia for 12 and 15 h (Fig. 2, A and B). Up-regulation of HIF-1␣ was also detected in cells treated with iloprost, a stable analogue of PGI 2 (Fig. 2B), whereas exogenous PGE 2 did not alter HIF-1␣ expression (Fig. 2C).
HIF-1 activity was measured by HRE reporter assay in HUVECs transfected with pHRE-Luc and infected with Ad-PGK or Ad-COP1. Ad-COP1 infection did not up-regulate HIF-1 activity in HUVECs under normoxia (Fig. 3A). However, the activity was increased in control cells exposed to hypoxia for 12 h and further up-regulated in cells infected with Ad-COP1. Because HIF-1 is the major mediator of VEGF up-regulation, augmented transactivation activity of HIF-1 was also supported by RT-PCR results showing the VEGF mRNA level up-regu-lated in Ad-COP1-infected cells as compared with Ad-PGK-infected cells (Fig. 3B). These data imply that augmentation of PGI 2 up-regulates HIF-1 activity in HUVECs under prolonged hypoxia. To investigate whether HIF-1␣ up-regulation was through gene transactivation or protein stabilization, real-time PCR revealed that HIF-1␣ mRNA levels did not significantly differ between cells infected with Ad-PGK and those with Ad-COP1 (Fig. 3C). Thus, up-regulation of HIF-1␣ by PGI 2 is through protein stabilization but not DNA transactivation.

PGI 2 Stabilizes HIF-1␣ in HUVECs Treated with Hypoxia and Then
Normoxia-Because HIF-1␣ is rapidly degraded through a proteosome pathway in cells exposed to hypoxia and then normoxia (5), HUVECs were incubated under hypoxia for 6 h and then exposed to normoxia for 15-120 min to investigate whether PGI 2 stabilizes HIF-1␣ in such cells. Western blot analysis showed that HIF-1␣ in Ad-PGK-infected cells was degraded to a very low level in cells exposed to hypoxia and then normoxia for 15 and 30 min (Fig. 4A). However, a significant amount of HIF-1␣ was detected in cells infected with Ad-COP1 exposed to hypoxia and then normoxia for up to 60 min, markedly reduced at 90 min, and down-regulated to a very low level at 120 min (Fig. 4B). A similar effect was identified in Ad-COP1-infected cells treated with cyclohexamine, a translational inhibitor, that a high level of HIF-1␣ was detected at 45 min, gradually decreased at 60 and 90 min, and downregulated to a very low level at 120 min (Fig. 4C). These results suggest that overexpression of both COX-1 and PGIS prolongs HIF-1␣ stability in the hypoxia-normoxia transition.
Interaction of pVHL with HIF-1␣ was investigated by co-immunoprecipitation of HIF-1␣ and pVHL. Ad-COP1 infection abolished the interaction of pVHL with HIF-1␣ in cells exposed to prolonged hypoxia and in control cells exposed to hypoxia for 6 h, whereas high levels of pVHL protein co-immunoprecipitated with HIF-1␣ in control and Ad-PGK-infected cells (Fig. 4D). Because pVHL mediated HIF-1␣ degradation, this result suggests that Ad-COP1 infection abolishes pVHL interaction with HIF-1␣ and, hence, inhibits its degradation.
Stabilization of HIF-1␣ by PGI 2 Is ROS-dependent-To determine whether stabilization of HIF-1␣ is related to the ROS signal, we found  . PGI 2 stabilizes HIF-1␣ in HUVECs exposed to hypoxia for 6 h and then normoxia. Shown are Western blot analyses of HIF-1␣ levels in cells infected with Ad-PGK or Ad-COP1. Cells were exposed to hypoxia for 6 h and then normoxia for 15 (H63 N15m), 30 (H63 N30m) (A and C), or for 45-120 min (B and C) with or without MG132 treatment (10 mol/liter) (B and C). C, cells were treated with cyclohexamine (CHX) before being exposed to normoxia (N). D, nuclear extracts of HUVECs were immunoprecipitated with HIF-1␣ antiserum, and the immunoprecipitate was analyzed by Western blotting with pVHL and HIF-1␣ antibodies.
that exogenous H 2 O 2 abolished the PGI 2 effect on HIF-1␣ up-regulation (Fig. 5A), whereas HIF-1␣ was up-regulated in cells infected with Ad-catalase (Fig. 5B). Moreover, HIF-1␣ levels were not altered by H 2 O 2 or catalase overexpression in cells under normoxia. Thus, under prolonged hypoxia, HIF-1␣ degradation is signaled by ROS and can be stabilized by reduction of endogenous H 2 O 2 by catalase. To examine whether PGI 2 attenuates ROS production, as implied above, HUVECs infected with Ad-PGK or Ad-COP1 were exposed to hypoxia and their ROS levels measured. As shown in Fig. 5C, ROS levels were increased in Ad-PGK-infected cells exposed to hypoxia for 6 h, further elevated in control cells under hypoxia for 12 h, and markedly attenuated in Ad-COP1-infected and iloprost-treated cells, whereas Ad-COX-1 infection did not alter the ROS level. These results suggest an antioxidant effect of PGI 2 under hypoxia but not normoxia. To examine whether the antioxidant effect of PGI 2 was through up-regulation of SOD or catalase, Western blot analysis showed that Ad-COP1 infection did not significantly alter SOD and catalase levels under prolonged hypoxia (Fig. 5, D and E).
Measurement of ROS generation in cells exposed to the hypoxianormoxia transition showed significant up-regulation in control cells exposed to normoxia for 15 min and a further increase at a longer incubation (Fig. 6A). However, Ad-COP1 infection markedly attenuated the up-regulation of ROS in cells exposed to normoxia for 15-30 min, which was slightly increased at 60 min and up-regulated to a higher level at 120 min. To investigate whether reduction of ROS stabilized HIF-1␣ in cells exposed to hypoxia and then normoxia for 90 and 120 min, infection of Ad-catalase further attenuated ROS production in cells exposed to the hypoxia-normoxia transition (Fig. 6A), which resulted in up-regulation of HIF-1␣ in cells exposed to hypoxia and then normoxia for 90 and 120 min (Fig. 6B). Moreover, MG132, a proteosome inhibitor, up-regulated HIF-1␣ in the hypoxia-normoxia transition (Figs. 4, B and C; and 6B), which confirms that HIF-1␣ downregulation is through a proteosome degradation pathway. PGI 2 Attenuated NADPH Oxidase Activity-Because NADPH oxidase is one of the major sources of ROS in HUVECs, we examined whether augmentation of PGI 2 suppressed NADPH oxidase activity in HUVECs under prolonged hypoxia. As shown in Fig. 7A, Ad-COP1 infection reduced NADPH oxidase activity in HUVECs as compared with cells infected with the control virus under normoxic conditions. Surprisingly, under prolonged hypoxia, NADPH oxidase activity was markedly attenuated to a very low level in cells infected with Ad-COP1, which supports the antioxidant effect of PGI 2 shown in Figs. 5C and 6A. For an unknown reason, NADPH oxidase activity was reduced in cells infected with the control virus under prolonged hypoxia. Moreover, HIF-1␣ expression was not affected in cells treated with exogenous diphenylene iodonium, an NADPH oxidase inhibitor, and exposed to normoxia or hypoxia for 6 h (Fig. 7B). However, under prolonged hypoxia, HIF-1␣ was significantly up-regulated in such cells and in those infected with Ad-COP1 (Fig. 7B). Thus, inhibition of NADPH oxidase activity up-regulates HIF-1␣ in HUVECs under prolonged hypoxia.
To investigate the molecular mechanism of NADPH oxidase attenuation by PGI 2 , protein levels of NADPH oxidase subunits were analyzed by Western blotting in cells exposed to prolonged hypoxia. p67 phox was  up-regulated under hypoxia, and infection with Ad-COP1 significantly suppressed the up-regulation (Fig. 8A). The p67 phox level in Ad-COP1infected cells was similar to that in cells under normoxia, which implies that regulation of p67 phox does not contribute to marked attenuation of NADPH oxidase activity (Fig. 7A). In addition, p47 phox and Rac1 were markedly down-regulated in cells infected with Ad-COP1 as compared with those infected with Ad-PGK (Fig. 8, B and C). These data suggest that PGI 2 attenuates NADPH oxidase activity by down-regulating Rac1 and p47 phox in HUVECs infected with Ad-COP1 under prolonged hypoxia.

DISCUSSION
In the cardiovascular system, the up-regulation of HIF-1 activates VEGF and many other genes, which may protect endothelial cells or surrounding organs by promoting angiogenesis, anti-apoptosis (29), and cell proliferation (30). In this study, we identified a novel function of PGI 2 in down-regulating ROS production and activating HIF-1 by stabilizing HIF-1␣ under prolonged hypoxia in HUVECs. Fig. 9 delineates the model of PGI 2 function and is supported by the major new findings of this study. 1) Overexpression of COX-1 and PGIS via adenovirusmediated gene transfer augments PGI 2 production, which stabilizes HIF-1␣ in cells under prolonged hypoxia and hypoxia followed by normoxia. 2) Stabilization of HIF-1␣ by PGI 2 or Ad-catalase infection is mediated through reduction of ROS production under prolonged hypoxia. 3) PGI 2 attenuates NADPH oxidase activity by suppressing Rac1 and p47 phox expression under prolonged hypoxia. Although antisense HIF-1␣ has been suggested to down-regulate HIF-1␣ in A549 cells exposed to prolonged hypoxia (6), our data suggest a different regulatory mechanism, because the mRNA level of HIF-1␣ was not altered (Fig. 3C).
Up-regulation of ROS by hypoxia or the hypoxia-normoxia transition was associated with HIF-1␣ degradation, which was attenuated by PGI 2 production, catalase overexpression, or exogenous diphenylene iodonium. These results indicate that up-regulation of ROS mediates HIF-1␣ degradation in such conditions, which explains why exogenous H 2 O 2 pretreatment suppressed HIF-1␣ stabilization in cells exposed to hypoxia for 8 h (11). Although NADPH oxidase activity was not up-reg-ulated under hypoxia, increased ROS levels might be generated by the complex III of mitochondria (17). In addition, diphenylene iodonium has no effect on HIF-1␣ expression in cells under normoxia or 6-h hypoxia (Fig. 7B), which implies that HIF-1␣ up-regulation in HUVECs exposed to hypoxia for 6 h is not mediated through suppression of NADPH oxidase activity. Conversely, up-regulation of HIF-1␣ by accumulated H 2 O 2 has been reported in immortalized JunD Ϫ/Ϫ cells via suppression of PGIS and other antioxidant gene expression, as well as transactivation of NADPH oxidase 4 (18). Moreover, HIF-1␣ stabilization by Ad-COP1 infection was not detected in human lung cancer (H1299) and hepatoma (HepG2) cells under prolonged hypoxia (data  not shown). These data indicate a differential regulatory effect of ROS in endothelial and tumor cells and suggest a cell type-dependent regulation of HIF-1␣.
Because NADPH oxidase is one of the major sources of ROS in endothelial cells, inhibition of ROS production reduces oxidative products, such as oxidized low density lipoprotein, and their damage is related to atherosclerosis and cell apoptosis (27). Moreover, up-regulation of VEGF might have a protective effect in inhibiting atherogenesis and cardiovascular damage (31). Transactivation of VEGF by HIF-1 because of PGI 2 augmentation may increase angiogenesis around the hypoxic region to increase blood and nutritional supply and protect cells from hypoxic damage. These data suggest an important role for PGI 2 in the cardiovascular system, not only as a potent vasoprotective molecule to cause vasodilatation and reduce platelet and monocyte activation but also to attenuate ROS production and enhance angiogenesis in endothelial cells under hypoxia. This function may be related to the deleterious effect of COX-2-selective inhibitors, which suppress PGI 2 production and increase the risk of myocardial infarction and stroke (32,33). In addition, hypoxia and a constitutively active form of HIF-1␣ markedly induces COX-1 and PGIS expression in endothelial cells (4) and up-regulates PGI 2 production in the ischemic brain (26), which suggests a positive feedback regulation of HIF-1␣ and an important role of PGI 2 under hypoxia. Moreover, although PGI 2 may stabilize the HIF-1␣ in the endothelial cells of tumors under ischemia and enhance angiogenesis, the PGIS promoter was shown to be hypermethylated and its expression suppressed in colorectal cancers and cell lines (34), therefore the function of PGI 2 in tumorigenesis remains unclear.
Under physiological conditions, PGI 2 is rapidly hydrolyzed to 6-keto-PGF 1␣ . Therefore, analogues of PGI 2 , such as OP-2507, beraprost, and iloprost have been used in most studies to investigate the function of PGI 2 . Our findings of the inhibition of NADPH oxidase activity by PGI 2 is supported by results of studies involving PGI 2 analogues. OP-2507 down-regulates NADPH oxidase activity in neutrophils exposed to phorbol myristate acetate (35), whereas sodium beraprost inhibits p47 phox translocation and phosphorylation in neutrophils stimulated with formyl-methionyl-leucyl-phenylalanine but not phorbol myristate acetate (36). Recently, Muzaffar et al. (37) suggested that iloprost antagonizes ROS production induced by the thromboxane A 2 analogue and cytokines by attenuating NADPH oxidase activity through down-regulation of gp91 phox in the smooth muscle and endothelial cells of pig pulmonary arteries. These results all suggest that PGI 2 inhibits NADPH oxidase activity but with different mechanisms. These discrepancies may be due to the nonspecific activity of PGI 2 analogues; they may cross-react with other prostaglandin receptors such as the PGE 2 receptor (38). To circumvent this problem, we achieved augmentation of PGI 2 production by adenovirus-mediated gene transfer of COX-1 and PGIS. Our results demonstrate that Ad-COP1 infection augmented PGI 2 level to ϳ300 pg/ml, which is within the range of results of clinical studies showing that under various stresses, the PGI 2 level can vary from Ͻ50 to Ͼ500 pg/ml (39,40).
Because Rac1 expression was markedly suppressed to Ͻ20% the level of controls, suppression of NADPH oxidase activity by PGI 2 is mainly through down-regulation of Rac1. Intriguingly, for unclear reasons, NADPH oxidase activity was greatly diminished by PGI 2 in cells exposed to prolonged hypoxia but only partially reduced under normoxia. Moreover, both Ad-COP1 infection and iloprost treatment upregulated HIF-1␣ but not PGE 2 , which suggests that stabilization of HIF-1␣ under hypoxia is through PGI 2 but not PGE 2 . Although PGE 2 has been suggested to induce VEGF expression via stabilization of HIF-1␣ under normoxic conditions in human prostate tumor and colon carcinoma cells (41,42), PGI 2 and PGE 2 did not up-regulate HIF-1␣ under normoxia in HUVECs in our study. Because iloprost alone attenuated ROS production but Ad-COX-1 did not alter the ROS level (Fig.  5C), the antioxidant effect in cells infected with Ad-COP1 is mediated through PGI 2 but not the increased expression of COX-1.
In conclusion, as illustrated in Fig. 9, our results demonstrate a novel function of PGI 2 in attenuating NADPH oxidase activity, which results in ROS reduction and activation of HIF-1 by stabilizing HIF-1␣ in HUVECs exposed to prolonged hypoxia or the hypoxia-normoxia transition but not normoxia. PGI 2 inhibits NADPH oxidase activity by suppressing the protein expression of Rac1 and p47 phox . Under prolonged hypoxia, HIF-1␣ degradation is signaled by H 2 O 2 , in that reduction of H 2 O 2 stabilizes HIF-1␣. These results reveal that PGI 2 not only protects the cardiovascular system against thrombosis and atherosclerosis but also attenuates ROS production and extends HIF-1 activation under prolonged hypoxia.