Interplay between Vascular Endothelial Growth Factor (VEGF) and Nuclear Factor Erythroid 2-related Factor-2 (Nrf2)

Background: Several studies have suggested that decreasing VEGF levels might result in placental oxidative stress in preeclampsia. Results: VEGF activates Nrf2 in an ERK1/2-dependent manner, protecting against oxidative stress, and, in turn, up-regulates VEGF expression. Conclusion: Reduced VEGF bioavailability may lead to aggregation of oxidative stress and result in preeclampsia. Significance: Nrf2 activation might be considered as an adjunct therapeutic strategy to combat preeclampsia. Several recently published studies have suggested that decreasing VEGF levels result in placental oxidative stress in preeclampsia, although the question as to how decreased VEGF concentrations increase oxidative stress still remains unanswered. Here, we show that VEGF activated Nrf2, the main regulating factor of the intracellular redox balance, in the cytotrophic cell line BeWo. In turn, this activated the production of antioxidative enzymes thioredoxin, thioredoxin reductase, and heme oxygenase-1, which showed a decrease in their expression in the placentas of preeclamptic women. Nevertheless, this activation occurred without oxidative stress stimulus. As a consequence, the activation of Nrf2 protected BeWo cells against H2O2/Fe2+-induced oxidative damage. We further show that VEGF up-regulated the expression of itself. A positive feedback loop was described in which VEGF activated Nrf2 in an ERK1/2-dependent manner; the up-regulation of HO-1 expression by Nrf2 augmented the production of carbon monoxide, which in turn up-regulated VEGF expression. In conclusion, VEGF induces the Nrf2 pathway to protect against oxidative stress and, via a positive feedback loop, to elevate VEGF expression. Therefore, decreased VEGF bioavailability during preeclampsia may result in higher vulnerability to placental oxidative cell damage and a further reduction of VEGF bioavailability, a vicious circle that may end up in preeclampsia.

Several recently published studies have suggested that decreasing VEGF levels result in placental oxidative stress in preeclampsia, although the question as to how decreased VEGF concentrations increase oxidative stress still remains unanswered. Here, we show that VEGF activated Nrf2, the main regulating factor of the intracellular redox balance, in the cytotrophic cell line BeWo. In turn, this activated the production of antioxidative enzymes thioredoxin, thioredoxin reductase, and heme oxygenase-1, which showed a decrease in their expression in the placentas of preeclamptic women. Nevertheless, this activation occurred without oxidative stress stimulus. As a consequence, the activation of Nrf2 protected BeWo cells against H 2 O 2 /Fe 2؉ -induced oxidative damage. We further show that VEGF up-regulated the expression of itself. A positive feedback loop was described in which VEGF activated Nrf2 in an ERK1/2-dependent manner; the up-regulation of HO-1 expression by Nrf2 augmented the production of carbon monoxide, which in turn up-regulated VEGF expression. In conclusion, VEGF induces the Nrf2 pathway to protect against oxidative stress and, via a positive feedback loop, to elevate VEGF expression. Therefore, decreased VEGF bioavailability during preeclampsia may result in higher vulnerability to placental oxidative cell damage and a further reduction of VEGF bioavailability, a vicious circle that may end up in preeclampsia.
Preeclampsia is unique to human pregnancy and is characterized by proteinuric hypertension in up to 3-5% of all pregnancies (1). In developing countries where access to health care is limited, preeclampsia is a leading cause of maternal mortality, with estimates of Ͼ60,000 maternal deaths per year (2). Despite intensive research, the etiology of preeclampsia still remains unclear. Pathogenic factors affecting both fetus and mother are currently under intensive investigation: in the fetus, the principal pathology appears to be placental oxidative stress resulting from insufficient uteroplacental blood supply (3), whereas the underlying maternal pathology is a severe systemic inflammatory response involving leukocyte and endothelial cell activation (4 -6).
VEGF expression takes place in the villous trophoblast (7)(8)(9). Upon gene activation, the resulting VEGF receptor initiates a cascade of cellular protein phosphorylation reactions by several protein kinases and leads to a variety of cellular responses (10 -12). One family of these downstream kinases includes MAPKs p42 and p44, also referred to as ERK1 and ERK2. ERK1 and ERK2 are phosphorylated and activated by MAPK kinase (MAPKK or MEK) in the cytosol, translocate to the nucleus, and subsequently stimulate transcription of early response genes regulating cell proliferation and survival (10 -12).
Maynard et al. (13) reported that the soluble VEGF receptor sFlt-1 (soluble fms-like tyrosine kinase-1; soluble VEGF receptor-1) is involved in the pathophysiology of preeclampsia. More recently, it was reported that elevated maternal sFlt-1 and decreased VEGF concentrations result in increased oxidative stress, which contributes to vascular dysfunction during pregnancy, although the question as to how decreased VEGF concentrations increase oxidative stress still remains unanswered (14).
Oxidative stress in syncytiotrophoblasts of women with preeclampsia is well documented (3,6). These cells are especially sensitive to oxidative stress partly because of their location in the outermost layer of placental villi, where they are exposed to high oxygen concentrations, and partly because they contain surprisingly low concentrations of antioxidant enzymes (15,16).
A battery of genes encoding antioxidant enzymes is orchestrated upon exposure to reactive oxygen species (ROS). 2 This coordinated response is regulated via the antioxidant response element (ARE) contained within the regulatory regions of socalled "safeguard" genes such as NQO1 (NAD(P)H:quinone oxidoreductase-1), TXNRD1 (thioredoxin reductase-1), glutathione peroxidase, and heme oxygenase-1 (HO-1) (17,18). Activation of Nrf2 (nuclear factor erythroid 2-related factor-2) as a consequence of oxidative stress initiates and enhances transcription of these safeguard genes, thus protecting cells against oxidative stress as well as a wide range of other toxins (19 -23).
Mann et al. (24) were the first to discuss a link between Nrf2, vascular homeostasis, and preeclampsia. Recently, Wruck et al. (25) provided the first experimental data that Nrf2 is active exclusively within cytotrophoblasts of preeclamptic placenta, strongly suggesting that these cells suffer from oxidative stress caused by ROS. In this work, we hypothesized that oxidative stress during preeclampsia results in increased expression and transfer of antioxidant enzymes from cytotrophoblasts into syncytiotrophoblasts via enhanced syncytial fusion, thereby increasing cytotrophoblast proliferation and syncytial knot formation (necrotic and aponecrotic subcellular syncytial fragments).
Thus, the principal aim of this study was to investigate whether VEGF activates Nrf2, which counters oxidation stress. The second objective was to study whether the activation of Nrf2, which leads to an increase in cellular carbon monoxide, raises VEGF levels.
These cells were subsequently stimulated with 10 ng/ml human VEGF 165 (R&D Systems) or were left unstimulated as a control. VEGF 165 , an isoform of VEGF-A, was used in this study because it is supposed be one of the most potent angiogenic isoforms (26). For positive control of Nrf2 activation, some cells were treated with 1 M sulforaphane (Sigma-Aldrich) (27,28). Oxidative stress was induced with hydrogen peroxide (Carl Roth GmbH) in the presence of 10 M iron(II) sulfate (Sigma-Aldrich) (29).
For experiments with MAPK, p38, c-JNK, and PI3K inhibitors, the cells were seeded into 6-or 96-well plates. After reaching confluence, the cells were incubated with 50 M vitamin C before they were exposed to kinase inhibitors. The inhibitor concentrations used were as follows: 1-50 M MEK1 inhibitor PD98059 and 1-10 M MEK1/2 inhibitor U0126 (Cell Signaling Technology) and 1-40 M p38 inhibitor SB203580, 1-50 M c-Jun N-terminal kinase inhibitor SP600125, and 0.1-5 M PI3K inhibitor wortmannin (Calbiochem). Diphenyliodonium chloride (DPI) at concentrations of 1-20 M was used as a specific inhibitor of NAD(P)H oxidase (Sigma-Aldrich). The cells were pretreated for 30 min with inhibitors before VEGF was added for an additional 1 h prior to the assay.
To test the effect of carbon monoxide, the tricarbonyldichlororuthenium(II) dimer (CORM-2; Sigma-Aldrich) was applied. The cells were incubated in Ham's F-12 medium containing 10% FCS and the indicated concentration of CORM-2 for 24 h. To scavenge CO, human Hb (Sigma-Aldrich) was first dissolved in PBS under an argon atmosphere and agitated at room temperature for 2 h prior to filter sterilization. Various doses of Hb were added to Ham's F-12 medium at the same time as the administration of VEGF 165 . After the stimulation, the cells were lysed, and VEGF protein concentrations were measured by ELISA.
Cell Lysate Preparation and Immunoblot Analysis-Cells were lysed in buffer containing 10 mM HEPES (PromoCell GmbH), 1.5 mM MgCl 2 (Sigma-Aldrich), 10 mM KCl, 0.5 mM DTT, 0.05% Nonidet P-40 (pH 7.9), and protease inhibitors, and the lysate were left on ice for 10 min. Cytosolic fractions were obtained as the supernatant after centrifugation at 950 ϫ g for 10 min at 4°C. The pellets were resuspended in nuclear extraction buffer containing 5 mM HEPES, 1.5 mM MgCl 2 , 0.2 mM EDTA (Invitrogen), 0.5 mM DTT, 26% glycerol, and protease inhibitors (pH 7.9). NaCl (4.6 M) was added, followed by a 30-min incubation on ice. Nuclear proteins were found in the supernatant after a 20-min centrifugation at 24,000 ϫ g at 4°C (following the Abcam protocol). The protein (6 g) from the nuclear fractions and the protein (15 g) from the cytosolic fractions were subjected to 12.5% discontinuous SDS-PAGE, separated by electrophoresis, and then electroblotted onto PVDF membranes (Millipore GmbH).
The densities of the bands were measured using PCBAS 2.0 software, and the ratio between the protein and the corresponding loading control (histone H3 and ␤-actin for Nrf2; ␤-actin for thioredoxin, TXNRD1, and HO-1; total ERK1/2 for phospho-ERK1/2; p38 for phospho-p38; and Akt for phospho-Akt) was calculated.
tions. Twenty-four hours after transfection, the cells were seeded in a 96-well plate. Firefly and Renilla luciferase activities were determined 48 h after transfection with the Dual-Luciferase reporter gene assay system (Promega) in a 96-well plate reader (GloMax-96 microplate luminometer, Promega). The firefly luminescence signals were normalized to the corresponding Renilla luciferase signals acting as an internal control. Nrf2 Silencing-BeWo cells were transfected with 10 g of control shRNA and Nrf2 shRNA. The plasmids were part of the MISSION shRNA assortment purchased from Sigma-Aldrich. These aforementioned cells were transfected using jetPRIME TM transfection reagent according to the manufacturer's instructions. After 24 h, the medium was replaced with fresh Ham's F-12 medium containing 10% FBS; puromycin (Carl Roth GmbH) was added 48 h after transfection. The cells were grown in the medium containing 1 g/ml puromycin to obtain a sufficient selection. Stable cell lines were established once all of the cells in the negative control plate were killed. These stable cell lines were continuously grown in the medium containing 1 g/ml puromycin. After ϳ10 passages, the gene expression levels of Nrf2 were measured by quantitative RT-PCR to check the knockdown from Nrf2 transcription in these cells.
RT-PCR-Cells were harvested in peqGOLD TriFast TM (PEQLAB Biotechnologie GmbH) for extracting RNA according to the manufacturer's protocol. The RNA concentration was determined by photometric analysis with the NanoDrop 1000 system (PEQLAB Biotechnologie GmbH), and 1 g of total RNA was then digested with DNase I (Roche GmbH) and transcribed into cDNA by reverse transcription with RevertAid TM reverse transcriptase (Fermentas Life Sciences). Real-time PCRs were processed in triplicate using the ABI StepOnePlus TM apparatus (Applied Biosystems) in a total volume of 20 l containing 100 ng of cDNA, gene-specific primers, and SYBR Green I reagent (Applied Biosystems). Gene expression was determined by normalizing the target gene C t values to the expression of the housekeeping gene, which was 18 S in this study (Eurofins MWG Operon). The sequences for the 18 S primer were 5Ј-TCAACTTTCGATGGTAGTCGCC-3Ј (forward) and 5Ј-ATGTGGTAGCCGTTTCTCAGGC-3Ј (reverse), with an annealing temperature of 61.5°C, and those for the Nrf2 primer (Eurofins MWG Operon) were 5Ј-TCCAGT-CAGAAACCAGTGGAT-3Ј (forward) and 5Ј-AATGTCTGC-GCCAAAAGCTG-3Ј (reverse), with an annealing temperature of 60.5°C.
Assay for Intracellular Redox State-Intracellular redox state levels were measured using the fluorescent dye 2Ј,7Ј-dichlorofluorescein diacetate (Invitrogen) as described previously (30). Briefly, cells were washed once with PBS and incubated in the same buffer containing 10 g of 2Ј,7Ј-dichlorofluorescein diacetate for 30 min at 37°C. Intracellular fluorescence was detected at Ex 485 /Em 530 using a SpectraMax Gemini EM system (Molecular Devices).
Cytotoxicity Assay-The CytoTox-Glo TM cytotoxicity assay (Promega) was used to measure cytotoxicity. The cells were seeded in 96-well plates and starved overnight by adding FCSfree medium supplemented with vitamin C. The cells were treated with or without 10 ng/ml VEGF 165 for 3 h, and all cells were then incubated with 500 M H 2 O 2 and 10 M Fe 2ϩ for 6 h. The CytoTox-Glo TM cytotoxicity kit was used according to the manufacturer's recommendations. The luminescence was measured in a 96-well plate reader (GloMax-96 microplate luminometer).
ELISA-The cells were seeded in 6-well plates; after reaching confluence, the cells were treated with VEGF 165 for 6 h. Subsequently, the cells were washed twice with PBS and lysed or supplied with new medium (without FCS) for 24 h. The levels of VEGF protein in the supernatant or in the cell lysate were assessed by a sandwich ELISA (R&D Systems) that detects all VEGF splice forms. Human recombinant VEGF 165 was used for standard curve determination.
Statistical Analysis-All measurements were performed at least in triplicates, and the results are expressed as the mean Ϯ S.E. Statistical analyses were performed using Student's unpaired t test for dual comparisons. Mean differences were considered to be significant when p was Ͻ0.05. All statistical graphs and analyses were created with GraphPad Prism 5.0 (GraphPad Software, La Jolla, CA).

VEGF 165 Increases Nrf2 Activation and Nrf2 Target Gene
Expression-To investigate the efficacy of VEGF 165 to activate the cis-acting element ARE, a Dual-Luciferase reporter gene assay was performed with the ARE from the NQO1 gene. ARE activation was determined in a dose-response assay at concentrations up to 10 ng/ml VEGF 165 in BeWo cells after 6 h of incubation. According to Fig. 1A, the relative ARE activity of VEGF 165 (10 ng/ml) was approximately twice that of the control. This was also true for the positive control, 1 M sulforaphane.
We then tested whether this ARE activation is Nrf2-dependent. Therefore, an shRNA against mRNA coding for human Nrf2 was designed, whereby a BeWo cell line containing Nrf2 shRNA (BeWo-shNrf2) and a control BeWo cell line containing a scrambled control shRNA (BeWo-shControl) were produced and stably transfected in BeWo cells. Real-time RT-PCR was used to test the efficacy of the shRNA against Nrf2. Real-time RT-PCR analysis showed an Nrf2 knockdown of Ͼ90% in both untreated and 1 M sulforaphane-treated BeWo-shNrf2 cells. In contrast, sulforaphane treatment showed a significant activation of Nrf2 mRNA in BeWo-shControl cells (Fig. 1B, inset).
The Nrf2-specific shRNA totally abolished the VEGF 165 -and sulforaphane (positive control)-dependent ARE activation. These results demonstrate that Nrf2 is the transcription factor that activates ARE expression in response to VEGF 165 (Fig. 1B).

JOURNAL OF BIOLOGICAL CHEMISTRY 42865
Nrf2 is a critical regulator of the intracellular antioxidants and phase II detoxification enzymes as an adaptive response to oxidative stress or pharmacological stimuli. To investigate whether VEGF 165 induces Nrf2 activation, 10 ng/ml VEGF 165 was administered to the cells at various time points. As early as 1 h following the administration of VEGF 165 , Nrf2 accumulated in the nucleus of the BeWo cells and remained at an elevated level for at least for 6 h (Fig. 1C). Fig. 1E presents the mean of three independent experiments. Subsequently, the protein expression of three Nrf2 target genes, i.e. thioredoxin, TXNRD1, and HO-1, was elevated 3 h after the administration of VEGF 165 (Fig. 1D). The mean of three independent experiments is shown in Fig. 1F.
Nrf2 Activation by VEGF 165 Is ERK1/2-dependent-To address the role of MAPK pathways in ARE gene regulation by VEGF 165 , the effects of various kinase inhibitors were examined. Hence, the activation of ERK1/2 was found to be a prerequisite for the activation of Nrf2 by VEGF 165 because VEGF 165mediated Nrf2 activation was exclusively inhibited by the MEK1/2 inhibitor PD98059 (50 M) as well as the MEK1/2 inhibitor U0126 (10 M) (Fig. 2A). The c-Jun N-terminal kinase inhibitor SP600125 (1-50 M), the p38 MAPK inhibitor SB203580 (1-40 M), and the PI3K inhibitor wortmannin (0.1-5 M) did not affect ARE activation in VEGF 165 -stimulated BeWo cells ( Fig. 2A).
To confirm these results, the effect of VEGF on both p38 and Akt activation were tested. VEGF stimulation did not up-regulate p38 phosphorylation over time. In contrast, phorbol 12-myristate 13-acetate (PMA), which was used as a positive control, induced phosphorylation of p38, and this phosphory- to activate the Nrf2/ARE system in an ARE Dual-Luciferase assay, as did sulforaphane as the positive control. Inset, Nrf2 was down-regulated by 90% compared with the control via stably transfected Nrf2-specific shRNA. C, Western blot analysis to measure Nrf2 activation in response to VEGF stimulation in BeWo cells. Nrf2 translocated into the nucleus in response to VEGF stimulation in BeWo cells. D, Western blot analysis to measure the effects of VEGF on the expression of antioxidative stress genes in BeWo cells. The Western blot shows up-regulation of HO-1, thioredoxin (Trx), and TXNRD1 protein levels. E, the densities of corresponding bands (Nrf2 and the loading control histone H3) were measured, and the ratio of Nrf2 to histone H3 was calculated. The mean of three independent experiments is shown. F, the densities of corresponding bands and the loading control ␤-actin were measured, and the ratio was calculated. The mean Ϯ S.E. of three independent experiments is shown. Data are presented as -fold induction after treatment compared with untreated cells (control ϭ 1). *, p Ͻ 0.05; **, p Ͻ 0.005; ***, p Ͻ 0.001; #, p Ͻ 0.0001 versus the control (one representative Western blot is shown; n ϭ 3). lation was blocked by 20 M SB203580 (supplemental Fig. 1, A  and B). On the other hand, hydrogen peroxide was used as a positive control for Akt activation (31). VEGF stimulation did not up-regulate Akt phosphorylation over time. In contrast, hydrogen peroxide induced phosphorylation of Akt, and this phosphorylation was blocked by 1 M wortmannin (supplemental Fig. 1, C and D).
Next, DPI was used as an NAD(P)H oxidase inhibitor to elucidate the role of NAD(P)H oxidase and ROS generation in the signal transduction of VEGF ( Fig. 2A). To test if stimulation with VEGF induces ROS production, we monitored the levels of ROS with the fluorescent dye 2Ј,7Ј-dichlorofluorescein diacetate. VEGF treatment showed no induction of ROS production up to 6 h after stimulation (Fig. 2F). We used the phorbol ester PMA as a positive control for ROS production (32) and to test the effectiveness of the NAD(P)H oxidase inhibitor DPI (Fig.  2F). The results clearly showed that NAD(P)H oxidase and ROS generation are not involved in VEGF-induced Nrf2 activation.
To determine the role of ERK1/2 in VEGF 165 -mediated Nrf2 activation, the effect of VEGF on ERK1/2 activity was exam-  (1-20 M)). Data are presented as percentage -fold induction of luciferase activity after treatment compared with the control (treated cells without inhibitor; control ϭ 100%). B, immunoblot analysis with anti-phospho-ERK1/2 antibodies (pERK1/2) was used to measure ERK1/2 activation. BeWo cells were pretreated with either PD98059 or U0126 for 30 min, followed by treatment with VEGF 165 (10 ng/ml) for 1 h. The membrane was stripped and probed with anti-ERK1/2 antibodies as a loading control. C, the densities of corresponding phospho-ERK1/2 and ERK1/2 bands were measured, and the ratio was calculated. The mean of three independent experiments is shown. Data are presented as -fold induction after treatment compared with the vehicle control (control ϭ 1). Pos., positive. D, role of ERK1/2 in Nrf2 activation in response to VEGF treatment measured by Western blotting. BeWo cells were pretreated with PD98059 or U0126 for 30 min, followed by treatment with VEGF 165 (10 ng/ml) for 1 h. The immunoblots of BeWo nuclear extracts with anti-Nrf2 antibodies were used to analyze Nrf2 nuclear translocation. After Nrf2 immunoblotting, the membrane was stripped and probed with anti-␤-actin antibodies as a loading control. E, band intensity was calculated as the ratio of untreated control cells after normalization to the amount of the loading control.  DECEMBER 16, 2011 • VOLUME 286 • NUMBER 50 ined. ERK1/2 was activated by dual phosphorylation of threonine and tyrosine residues located in the "activation lip" of the conserved core kinase sequence, and the activated species could be detected by antibodies against phosphorylated peptides contained in these residues. BeWo cells were treated with 10 ng/ml VEGF 165 , and cell extracts were analyzed for phosphorylated and total ERK1/2 by Western blotting. The loading control via total ERK1/2 Western blotting showed no significant differences (Fig. 2B). The mean of three independent experiments is shown in Fig. 2C. The results clearly show that VEGF 165 activates ERK1/2 and that 50 M PD98059, as well as 10 M U0126, inhibits this ERK1/2 activation in comparison with cells treated with VEGF 165 , which was supposed to be the positive control.

VEGF Activates Nrf2
Consequently, inhibition of ERK1/2 either by 50 M PD98059 or by 10 M U0126 significantly blocked the VEGF 165 -induced nuclear accumulation of Nrf2 (Fig. 2D), and the cells treated with VEGF 165 without inhibitor were considered to be the positive control. The mean of three independent experiments is shown in Fig. 2E.
Nrf2 Activation by VEGF 165 Confers Cytoprotection against Oxidative Stress-The hypothesis that activation of Nrf2 could protect BeWo cells from lesions caused by oxidative stress via H 2 O 2 treatment was tested. BeWo cells were preincubated with 10 ng/ml VEGF 165 for 3 h and then treated with 500 M H 2 O 2 and 10 M Fe 2ϩ . Cell viability was measured by the WST-1 assay, and cell death was measured by the CytoTox-Glo TM cytotoxicity assay 6 h after H 2 O 2 administration. Preincubation of BeWo cells with 10 ng/ml VEGF 165 effectively protected the cells from H 2 O 2 -induced toxicity in both the WST-1 assay (Fig.  3A) and the CytoTox-Glo TM cytotoxicity assay (Fig. 3C).
A causal relationship between Nrf2 activation and cytoprotection mediated by VEGF 165 was further investigated. As shown in Fig. 3 (A-D), VEGF 165 could protect BeWo-shControl (but not BeWo-shNrf2) cells from oxidative toxicity. After incubation with 500 M H 2 O 2 and 10 M Fe 2ϩ , the cell viability of BeWo-shNrf2 cells declined by 50% (Fig. 3B) versus the 20% decline that was found with BeWo-shControl cells (Fig.  3A). Consistently, BeWo-shControl cells showed a significant 30% decrease in cell death after treatment with 500 M H 2 O 2 and 10 M Fe 2ϩ in the presence of VEGF (Fig. 3C), whereas VEGF treatment had no effect in BeWo-shNrf2 cells (Fig. 3D).
VEGF Up-regulates VEGF via Nrf2/HO-1/CO-To evaluate whether treatment with VEGF 165 has an effect on the production of VEGF protein itself in BeWo cells, both BeWo-shControl and BeWo-shNrf2 cells were treated with 10 ng/ml VEGF 165 for 6 h. 24 h after stimulation, both cell lines showed elevated levels of VEGF protein itself in the supernatants (Fig.  4A, black bars). Nrf2 silencing reduced the VEGF-induced increase in VEGF itself by 30% in comparison with BeWo-shControl cells, which infers that partial up-regulation of VEGF protein expression is attributed to at least one Nrf2-mediated product (Fig. 4A, stippled bars). We next tested if Nrf2 activators induce VEGF expression in our system. BeWo-shControl cells stimulated with 1 M sulforaphane and 1 mM bipyridyl (33) showed a 2-fold increase in VEGF expression compared with control cells. In contrast, BeWo-shNrf2 cells were not able to up-regulate VEGF expression in response to sulforaphane (Fig. 4B).
Gene expression for the VEGF protein is induced by carbon monoxide produced by exogenous sources or the degradation of heme by HO-1 (34). Therefore, the effect of CO on VEGF expression in BeWo cells was examined. CORM-2 treatment significantly increased VEGF protein levels in a low dose-dependent manner (Fig. 4C, left y-axis) in comparison with untreated cells. Because a higher dose of CORM-2 (50 M) was toxic, it did not further induce VEGF expression (Fig. 4C, right y-axis).
We then examined whether CO is essential for VEGF-induced VEGF up-regulation. Cell treatment with the CO scavenger Hb (10 -100 M) showed a dose-dependent inhibition of VEGF-induced VEGF expression (Fig. 4D). These results indicate that VEGF-up-regulated VEGF expression depends on Nrf2 and HO-1.

DISCUSSION
Several recently published studies have suggested that decreasing VEGF levels might result in placental oxidative stress in preeclampsia (13,35,36). Experimental animal studies have demonstrated that elevated maternal sFlt-1 and decreased VEGF concentrations result in increased oxidative stress (14). The question as to how decreased VEGF concentrations increase oxidative stress still remains unanswered. A recent study suggested a possible role of Nrf2 in mechanisms underlying preeclampsia. The study used genome-wide transcriptional profiling of preeclamptic and normal pregnancies and showed that the Nrf2-mediated oxidative stress response is overrepresented in preeclampsia (37). Thus, these data strengthen our previous published results (25) and our hypothesis that Nrf2 has a critical role in the etiology and progression of preeclampsia.
Nrf2 plays a key role in the adaptive response to oxidative and electrophilic stress by maintaining the cellular self-defense. In this study, we have provided substantial experimental evidence that VEGF activates Nrf2 and up-regulates Nrf2-dependent gene products within the well established cytotrophic cell line BeWo (Fig. 1) (38).
Several studies have shown that protein kinases are involved in the activation of Nrf2 (39). Because it is also known that binding of VEGF to VEGF receptor-2 activates downstream effectors, including the PKC, Raf, and ERK1/2-PI3K-focal adhesion kinase pathways. Therefore, we examined whether kinases are involved in the signal transduction of VEGF leading to Nrf2 activation. Thus, the effects of various kinase inhibitors on Nrf2 activation were tested. Of all tested inhibitors, only the MEK1/2 inhibitors PD98059 and U0126 inhibited Nrf2 activation. Consequently, ERK1/2 activation is a prerequisite for Nrf2 activation by VEGF. However, Nrf2 might not be a direct sub- strate of ERK1/2. Instead, it has been suggested that ERK1/2 phosphorylates the nuclear transcription co-activator CBP (cAMP-responsive element-binding protein-binding protein) and that CBP enhances the Nrf2 transcriptional response (35,36). In addition, BeWo cells were stimulated with VEGF, and the phosphorylation status of ERK1/2 was analyzed to substantiate the activation status of the kinases ( Fig. 2A).
It is generally accepted that oxidative or electrophilic stress is a requirement for Nrf2 activation (19,20). Thus, it was assumed that this activation may occur via VEGF-induced NAD(P)H oxidase activation because VEGF supposedly utilizes ROS as a second messenger after VEGF receptor-2 stimulation (40). However, NAD(P)H oxidase inhibition had no effect on the VEGF-triggered Nrf2 activation (Fig. 2, A, OE, and F), so oxidative stress can be excluded here. In this work, we have shown that VEGF induced Nrf2 activation and that this activation occurred without oxidative stress stimulus. This means, in effect, that VEGF stimulation prevents oxidative stress. Hence, we hypothesized that decreased VEGF bioavailability during preeclampsia results in reduced basal defense against oxidative stress and in a higher vulnerability of the cells to oxidative damage.
To test this hypothesis, BeWo cells were stimulated with VEGF before challenge with H 2 O 2 , implying oxidative stress. As expected, VEGF-stimulated cells showed less cell death and lower vulnerability after oxidative stress treatment with H 2 O 2 compared with unstimulated control cells (Fig. 3, A and C).
These results clearly show a protective effect of VEGF against oxidative stress, comparable with the described effects of EGF (41). To elucidate the role of Nrf2 in the protective effect of VEGF, a BeWo cell line carrying a stably transfected shRNA against Nrf2 mRNA (BeWo-shNrf2) was utilized. This cell line was no longer able to activate the Nrf2/ARE system (Fig. 1B). Thus, shRNA technology instead of dominant-negative Nrf2 overexpression was applied because shRNA knocks down only Nrf2. By contrast, dominant-negative Nrf2 blocked binding to the ARE by competitive inhibition and thus blocked all factors with ARE affinity. In these Nrf2-deficient BeWo cells, VEGF no longer protected against oxidative stress (Fig. 3, B and D), which infers that Nrf2 has a pivotal function in cytoprotection mediated by VEGF. These results indicate that cell death may be induced in villous trophoblasts following ROS exposure and demonstrate the placental protective effect of VEGF.
As a transcription factor, Nrf2 unfolds its protective effect via up-regulation of genes coding for antioxidant enzymes. Therefore, we studied whether VEGF up-regulates the expression of Nrf2 target genes in the applied system, particularly thioredoxin, TXNRD1, and HO-1 because these enzymes have been shown to be expressed at lower levels in the placentas of women suffering from preeclampsia (42)(43)(44). Supporting the proposed hypothesis, all three corresponding Nrf2 target genes were upregulated in response to VEGF treatment (Fig. 1, D and F).
In particular, decreased gene expression for HO-1 and the resultant decline in CO production have been shown to initiate FIGURE 5. Postulated model of the signal transduction of VEGF 165 -induced Nrf2 activation. VEGF receptor (VEGF-R) activation-induced ERK1/2 phosphorylation leads to Nrf2 liberation from the Nrf2 inhibitor Keap1. In the nucleus, Nrf2 binds to the cis-acting ARE and up-regulates the transcription of cytoprotective genes: HO-1 and other antioxidant enzymes. HO-1 then contributes via biliverdin to protect the cell against oxidative damage (49) and via CO to promote VEGF expression by increasing hypoxia-inducible factor-1␣ (HIF1␣) protein levels as shown elsewhere (#) (34,50). HRE, hypoxia response element. pathological processes throughout gestation (40,42). Cudmore et al. (45) demonstrated that the HO-1/CO pathway inhibits sFlt-1 and soluble endoglin release, providing compelling evidence for a protective role of HO-1 in pregnancy. Also Zhao et al. (46) showed that HO-1 deficiency is associated with elevation of maternal diastolic blood pressure and plasma sFlt-1 levels. Recently, Lin et al. (47) showed that HO-1/CO is able to induce VEGF expression in myocytes. We have proven, for the first time, a positive feedback loop of VEGF inducing the expression of itself via Nrf2/HO-1 activation. A sketch of the assumed signaling pathway of VEGF-induced HO-1 expression via Nrf2 is shown in Fig. 5.
These findings also prove a protective role of the Nrf2/HO-1/CO system during pregnancy. In this context, HO-1 up-regulation via Nrf2 activation shows great potential to serve as an effective drug target in future pharmacological therapy for preeclampsia (48).
In conclusion, VEGF-induced Nrf2 activation prevents placental oxidative stress. Therefore, it can be hypothesized that decreased VEGF bioavailability during preeclampsia results in insufficient Nrf2 activation, reduced basal defense against oxidative stress, and higher vulnerability to placental oxidative cell damage. The resulting damage to the placenta causes disproportionate release of toxic placental factors that are manifested as preeclampsia and endanger maternal health. Specific attempts to strengthen the fetal endogenous defense system against oxidative stress with an Nrf2 inducer, i.e. sulforaphane, methsyticine, or andrographolide, at an early gestation stage could prove to be a possible therapeutic option and may, in turn, reduce the risk of preeclampsia and associated perinatal complications.