Peroxisomal Proliferator-activated Receptor-α-dependent Inhibition of Endothelial Cell Proliferation and Tumorigenesis*

The peroxisomal proliferator-activated nuclear receptor-α (PPARα), the target for most hypolipidemic drugs in current clinical use, regulates the transcription of genes involved in lipid metabolism and transport, and energy homeostasis. More recently, PPARα and its ligands have been implicated in inflammatory responses and the regulation of cell proliferation. PPARα also regulates the expression of Cyp4a fatty acid ω-hydroxylases and Cyp2c arachidonic acid epoxygenase genes. To study the role of the PPARα receptor and of its Cyp2c epoxygenase gene target in tumorigenesis, we treated mice injected with tumor cells with Wy-14,643, a PPARα-selective ligand. Compared with untreated controls, Wy-14643-treated animals showed marked reductions in tumor growth and vascularization, which were accompanied by decreases in the plasma levels of pro-angiogenic epoxygenase metabolites (EETs), hepatic EET biosynthesis, and Cyp2c epoxygenase expression. All these Wy-14643-induced responses were absent in PPARα-/- mice and are thus PPARα-mediated. Primary cultures of mouse lung endothelial cells treated with Wy-14643 showed reductions in cell proliferation and in the formation of capillary-like structures. These effects were absent in cells obtained from PPRAα-/- mice and reversed by the addition of EETs. These results identify important anti-angiogenic and anti-tumorigenic roles for PPARα, characterize the contribution of its Cyp2c epoxygenases gene target to these responses, and suggest potential anti-cancer roles for this nuclear receptor and its ligands.

The peroxisomal proliferator activated nuclear receptors (PPARs), 2 namely PPAR␣, PPAR␤/␦, and PPAR␥, regulate the transcription of several genes involved in lipid metabolism, as well as energy utilization and storage (1)(2)(3)(4). PPAR␥ is expressed mostly in the adipose tissue, where it controls genes involved in adipogenesis (2), PPAR␦ is ubiquitously expressed and regulates lipolytic functions in several extrahepatic tissues. PPAR␣ is predominantly expressed in liver, kidney, heart, and vascular tissues (2)(3)(4) and controls expression of lipolytic genes and members of the cytochrome P450 (P450) CYP4A and -2C gene subfamilies of fatty acid monooxygenases (5)(6)(7). PPAR␣ is a target for fibrates (8), a class of synthetic PPAR␣ ligands that are clinically effective hypolipidemic agents with limited side effects (9 -11). The recognition that these nuclear receptors regulate the expression of genes associated with diseases such as obesity, diabetes, inflammation, and cancer has raised intense interest in the study of their gene targets, signaling mechanisms, as well as their physiological and pathophysiological roles. Recent studies indicate that PPAR␣ and PPAR␥ ligands regulate endothelial cell growth, migration, and angiogenesis (12)(13)(14)(15) and influence the progression of vascular inflammation and tumorigenesis (16 -18). In this regard, proand anti-tumorigenic activities have been described for PPAR␣ and its ligands. In rodents, extended exposure to fibrates causes PPAR␣-mediated liver hypertrophy and hepatocarcinomas (1,19), however these effects have not been observed in humans, even after extended use (19). On other hand, PPAR␣ ligands decrease intestinal polyp formation in Apc-deficient mice (18), and inhibit vascular smooth muscle cell proliferation by promoting the expression of the tumor suppressor p16 gene (16).
The reported pro-angiogenic properties of the EETs (22)(23)(24) and the known regulation of the CYP2C epoxygenase expression by PPAR␣ and its ligands (5-7) suggested a role for this nuclear receptor and these enzymes in tumor angiogenesis. We report here that Wy-14643, a selective PPAR␣ ligand (27,28), reduces tumor angiogenesis and growth in vivo, and show that its anti-tumorigenic effects are PPAR␣-dependent. In addition, studies with primary endothelial cells show that Wy-14643 decreases the proliferative and tubulogenic capacity of these cells in a PPAR␣and EET-dependent fashion and thus alters their angiogenic potential. Furthermore, in vivo studies with three different tumor cell lines identify a role for PPAR␣ in tumor growth and angiogenesis and suggest that its ligands could be utilized as safe and well tolerated anti-tumorigenic drugs.

EXPERIMENTAL PROCEDURES
Cell Culture-Primary murine pulmonary microvascular endothelial cells were isolated from sex-and age-matched adult wild-type and PPAR␣ Ϫ/Ϫ (PPAR␣KO) mice (a gift from Dr. Frank Gonzalez, NCI, National Institutes of Health, Bethesda, MD) (29) and cultured in EGM-2-MV (Clonetics) containing 5% fetal calf serum as described (30). Temperature-sensitive, conditionally immortalized pulmonary microvascular endothelial cells were isolated in the same manner from H-2Kb-tsA58 SV40 large T Ag wild-type transgenic mice (31) and propagated at 33°C in the presence of 100 IU/ml ␥-interferon. For experiments, cells were cultured at 37°C without ␥-interferon for at least 4 days before use, because this is the optimal time for the cells to acquire a phenotype similar to freshly isolated primary endothelial cells (32). Due to limitations in the number of primary endothelial cells that can be isolated and cultured from mouse lungs, conditionally immortalized wildtype cells were used as surrogates for the analysis of epoxygenase protein expression and EET/DHET biosynthesis. The large T antigen/Ras/Myc-transformed fibroblast line p60.5 was cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum as described (30). Mouse colon carcinoma cells CT26 were cultured in Dulbecco's modified Eagle's medium containing 10% fetal calf serum as described (33). The human non small cell lung cancer cells A459 were purchased from ATCC and cultured according to the supplier's instructions.
To analyze the effects of Wy-14643 on cellular epoxygenase activity, tumor cells or temperature-sensitive endothelial cells were cultured in media with or without different concentrations of Wy-14643 (ChemSyn Laboratories). After 4 days, the cells were incubated in serum-free medium containing AA (10 M, final concentration) and, 4 h after, the cells and the media were removed from the plates, and the EETs and DHETs present in the cells and the media were extracted, purified, and quantified as described below.
Proliferation Assay-Cells were plated (5 ϫ 10 3 cells/well; 96-well plate) in complete medium alone or containing either: and each individual EET regioisomer (1 M final concentration). EETs were used at 1 M because this was the concentration producing maximal responses (24). Two days after plating, the medium was replaced with media containing [ 3 H]thymidine (1 Ci/well), the cells were incubated for another 48 h, and [ 3 H]thymidine incorporation was determined as described (30).
Matrigel-based Capillary Formation Assay-Capillary-like tube formation was analyzed as described (34). Briefly, 96-well plates were coated with 50 l of Matrigel and incubated for 30 min at 37°C. Primary endothelial cells were cultured in complete media with or without Wy-14643 (25 or 100 M, final concentration). After 4 days, the cells were serum-starved for 24 h, suspended in serum-free media (200 l final volume) with or without Wy-14643 (25 or 100 M, final concentration), and then plated (1.5ϫ10 4 cells/well) over solidified Matrigel in the presence or absence of EETs (1 M, final concentration). The formation of capillary-like structures was recorded (three images per treatment per well) hourly for the next 10 h. Fig. 3 shows representative images taken 6 h after plating. To quantify capillary-like network formation, cellular nodes were defined as junctions linking at least three cells, and they were counted from digital images. Four independent experiments were performed with a total of 12 images analyzed per treatment.
Analyses of Primary Tumor Growth-All animal protocols were reviewed and approved by Vanderbilt University Medical Center Institutional Animal Care and Use Committee. Threemonth-old 129SvJ male wild-type and PPAR␣⌲⌷ mice were given two dorsal subcutaneous injections of p60.5 cells (3 ϫ 10 5 cells in 200 l of phosphate-buffered saline per site). Threemonth-old male wild-type BALB/c and immunodeficient nude mice were given two dorsal subcutaneous injections of CT26 cells (3 ϫ 10 5 cells in 200 l of phosphate-buffered saline per site) and A549 cells (1 ϫ 10 6 cells in 200 l of phosphate-buff-ered saline per site), respectively. Immediately afterward, the mice were separated into groups receiving water or a solution of Wy-14643 (0.02% w/v) as their only fluid supply (36). Mice were sacrificed after two (129SvJ and BALB/c mice) or four (immunodeficient nude mice) weeks and the tumor volume, weight, and number evaluated. Averaged tumor volumes were calculated using the following formula: tumor volume (mm 3 ) ϭ (length ϫ width 2 )/2 (37).
Analysis and Quantification of Epoxygenase Metabolites-Adult (10 -20 weeks of age) male wild-type and PPAR␣KO mice or wild-type BALB/c mice were fed commercial solid diets and allowed free access to either water or a 0.02% (w/v) solution of Wy-14643 for 7-10 days. Liver microsomes were isolated and suspended in 50 mM Tris-Cl (pH 7.4) containing 0.25 M sucrose (0.5-1 mg of protein/ml), and incubated at 37°C with [1-14 C]AA (50 Ci/mol, 70 -100 M final concentration) in the presence of NADPH (1 mM, final concentration), and an NADPH-generating system (38). Organic soluble products were extracted and resolved, and the epoxygenase metabolites (EETs plus DHETs) were quantified by using on-line ␤-detection (38).
The EETs and DHETs present in cultured cells, mouse liver, or plasma were extracted, purified, analyzed, and quantified by the isotope ratio method using gas chromatography/mass spectrometry as described (41). All biological samples were extracted in the presence of equimolar mixtures of synthetic 2 H 3 -labeled 8,9-, 11,12-, and 14,15-EET and 2 H 8 -labeled 8,9-, 11,12-, and 14,15-DHET (5 ng each). Blood samples were collected in the presence of EDTA, and the plasma fractions were extracted as described (42). Mouse livers and plasma were extracted immediately after isolation.
Statistical Analysis-One-tailed Student's t test was used for group comparisons and for the analysis of variance using Sigma-Stat software for statistical differences between multiple groups. p Ն 0.05 was considered statistically significant.

RESULTS
The identification of 5,6-and 8,9-EETs as angiogenic lipids (22)(23)(24) suggested a role for these metabolites in de novo vascularization and tumor angiogenesis. The regulation of CYP2C epoxygenase expression and EET biosynthesis by the PPAR␣ nuclear receptor and its ligands (5)(6)(7), offered an opportunity to study the role of PPAR␣ and the epoxygenase pathway in tumorigenesis. For these purposes we utilized Wy-14643, a selective PPAR␣ ligand, that binds and trans-activates the receptor with high affinity (29,36) and regulates the expression of hepatic CYP2C genes in a PPAR␣-dependent manner (7). We first documented the endothelial effects of PPAR␣ activation using lung endothelial cells isolated from wild-type and PPAR␣KO mice and then the role of PPAR␣ in tumor vascularization and growth. Three different combinations of tumor cell lines and hosts were used for these studies: (a) p60.5-transformed tumor cells and isogenic 129SvJ wild-type and PPAR␣KO mice, (b) CT26 cells and isogenic BALB/c mice, and (c) human A549 non-small cell lung cancer cells and immunocompromised nude mice recipients.
Wy-14643 Inhibits the Proliferation of Endothelial Cells Isolated from Wild-type but Not from PPAR␣KO Mice-To determine whether Wy-14643 and the PPAR␣ receptor had an effect on endothelial cell proliferation, primary cultures of lung endothelial cells isolated from wild-type and PPAR␣KO mice were incubated with different concentrations of Wy-14643, and their proliferation was evaluated by measuring [ 3 H]thymidine incorporation. Compared with vehicle-treated (ethanol) controls, wild-type cells incubated with Wy-14643 showed marked reductions in proliferation that were concentration-dependent, and reached a maximum at 100 M Wy-14643 (Fig. 1A). Importantly, Wy-14643 had no significant effects on the proliferation of endothelial cells isolated from PPAR␣KO mice ( Fig. 1A) demonstrating, unequivocally, that its anti-proliferative effects were PPAR␣-dependent. The observation, that wild-type and PPAR␣KO mice show comparable levels of [ 3 H]thymidine incorporation in the absence of an exogenous ligand (Fig. 1A), indicates that the receptor affects DNA replication upon ligand binding activation.
To explore the role of the epoxygenases in the Wy-14643mediated reductions in endothelial cell proliferation, we performed rescue experiments in which [ 3 H]thymidine incorporation was measured in cells incubated with Wy-14643 in the presence or absence of added EETs. All four EETs (1 M, final concentration each) restored the proliferation activity of Wy-14643-treated wild-type cells to levels similar to those of cells treated with EET only (Fig. 1B). Furthermore, at 100 M, Wy-14643 had no effect on the PPAR␣KO endothelial cells (Fig. 1B), and the EETs stimulated the proliferation of PPAR␣KO cells in the presence or absence of Wy-14643 (Fig.  1B). These results suggested that the Wy-14643-mediated reductions in cell proliferation could be due to its effects on the expression and/or activity of cellular epoxygenases and EET levels. To investigate the effects of Wy-14643 on endothelial cell epoxygenase expression, we analyzed by semi-quantitative RT-PCR the levels of Cyp2c29, -2c38, -2c40, and -2c44 transcripts present in cells cultured in the presence or absence of Wy-14643. As shown in Figs. 2 (A and B), Wy-14643 downregulated, in a PPAR␣-dependent manner, the expression of Cyp2c38 and, to a lesser extent, that of Cyp2c29, -2c40, and -2c44.
Western blot analysis of microsomes from primary endothelial cells using antibodies raised against rat CYP2C11 (immunoreactive toward mouse Cyp2c29, -2c38, and -2c40) or mouse Cyp2c44, while indicative of Wy-14643-mediated reductions in Cyp2c38 and -2c44 levels in wild-type cells, were inconclusive due to the low yield of microsomal fractions that can be obtained from these cells and the high variability of the results. To overcome this limitation, conditionally immortalized wild-type endothelial cells were used as surrogates for analysis of protein expression and EET biosynthesis and hydration to dihydroxyeicosatrienoic acids (DHETs) (20,21). As shown for lung primary endothelial cells (24), these immortalized cells also respond to EETs by increasing their proliferation, migration, and tubulogenesis (not shown). Moreover, Wy-14643 decreased the cellular levels of microsomal proteins immunoreactive toward anti-CYP2C11 and anti-Cyp2c44 antibodies with the expected mobilities (ϳ55 kDa) (Fig. 2C) and reduced 50 -80% the biosynthesis of endogenous EETs.
Wy-14643 Inhibits the Formation of Capillary Tubule-like Structures by Endothelial Cells from Wild-type but Not PPAR␣KO Mice-For studies of PPAR␣ roles in the formation of capillary-like structures, primary endothelial cells were cultured in the presence (25 and 100 M, final concentration) or absence of Wy-14643. After 4 days, the cells were plated on Matrigel and cultured in serum-free medium without or with Wy-14643 (25 and 100 M, final concentration). Wy-14643 markedly reduced the capacity of wild-type cells to generate tubule-like structures. These effects were dose-dependent and resulted in a ϳ75% reduction in cell branching and sprouting at 100 M Wy-14643 (Fig. 3, A and B). In contrast, none of these effects were observed when Wy-14643 was added to PPAR␣KO cells (Fig. 3, A and B), demonstrating that these functional responses to the ligand required a functional PPAR␣. In an earlier report it was shown that endothelial cells cultured in the presence of an epoxygenase inhibitor lose the capacity to generate capillary-like structures and that this could be reversed by the addition of synthetic EETs (24). To examine whether decreased EET biosynthesis was the cause of Wy-14643-mediated inhibition of tubulogenesis, cells were cultured with or without Wy-14643 and in the presence or absence of EETs (1 M, final concentration). Each of the exogenously added EETs rescued the cells from the inhibitory effects of Wy-14643, and  restored the formation of tubule like structures as shown in Fig.  3, C (for 11,12-EET) and D. Importantly, all tubulogenesis experiments were done under conditions in which cell proliferation is minimal, i.e. serum-starved cells plated in serum-free media, and short culture times (measurements done within the first 6 h of cell plating). In summary, these results demonstrate that the Wy-14643-mediated reductions in endothelial cell proliferation and tubulogenesis are PPAR␣-dependent and likely to involve Wy-14643 receptor binding and activation. Furthermore, they indicate that reduced EET biosynthesis is a key step in the Wy-14643/PPAR␣mediated inhibition of endothelial cell proliferation and tubulogenesis. Finally, endothelial cells isolated from wild-type and PPAR␣KO mice show similar basal proliferative and tubulogenic activities, suggesting that, under non-stimulated conditions, the participation of PPAR␣ and its endogenous ligands in the control of these cellular functions is limited. The identity of the endogenous PPAR␣ ligands and the degree of receptor/ligand occupancy remains unknown.
Wy-14643 Down-regulates Hepatic Epoxygenase Expression, EET Biosynthesis, and Plasma EET Concentrations-The role of PPAR␣ and its ligands in the control of hepatic Cyp2c and -4a protein levels, and plasma lipid homeostasis is well documented (7,29,36). Furthermore, circulating EETs are associated with human and rat plasma lipoproteins, and a role for the liver epoxygenases in their biosynthesis has been reported (42). Because the endothelium is in contact with the plasma and its lipid components and lipoprotein uptake could be a source of endothelial EETs, wild-type and PPAR␣KO mice were treated with Wy-14643 for 8 -14 days, and their levels of hepatic Cyp2c epoxygenase transcripts and protein, microsomal EET synthase activity, and circulating plasma EETs were analyzed and quantified. Treatment of wild-type mice with Wy-14643 down-regulated the expression of Cyp2c29, -2c40, and -2c44 transcripts, but not of Cyp2c38 (Fig. 4A). In addition, the levels of microsomal proteins immunoreactive toward anti-CYP2C11 and anti-Cyp2c44 were markedly decreased (Fig. 4B), and a ϳ45% reduction in microsomal EET biosynthesis was also observed ( Table 1). In parallel with its effects on the hepatic epoxygenases, Wy-14643 reduced the plasma concentrations of epoxygenase metabolites by 74% (Table 2). In contrast, none of these effects were observed when Wy-14643 was administered to PPAR␣KO mice (Figs. 4A and 4B, and Tables 1 and 2), showing that they are PPAR␣-dependent. Compared with wildtype, the levels of plasma epoxygenase metabolites are higher in untreated PPAR␣KO mice (Table 2), perhaps reflecting known effects of PPAR␣ on hepatic EET biosynthesis (36), hydration (43), and oxidation (36), as well as on plasma lipoprotein expression and levels (1-3).
The aforementioned in vitro effects of PPAR␣ activation in endothelial cell function and EET synthase activity, as well as the in vivo down-regulation by Wy-14643 of hepatic EET biosynthesis and secretion, suggested that PPAR␣ and its ligands could inhibit tumor angiogenesis and growth and thus, have beneficial anti-cancer effects. To test this hypothesis, we uti-  lized three different tumor cell lines injected subcutaneously into isogenic or immunocompromised nude mice, and tumor growth and vascularization were then evaluated in groups of untreated and Wy-14643-treated mice. The large T antigen/ Ras/Myc-transformed p60.5 cells were chosen to initiate these studies, because they are one of the few available cell lines that generate primary tumors in 129SvJ mice (30), the genetic background of our PPAR␣KO mice. Thus, these cells provide the opportunity to distinguish between host and tumor PPAR␣associated effects. Prior to the initiation of the in vivo studies, we characterized the responses of the p60.5 cells to Wy-14643.
Wy-14643 Does Not Alter Cultured p60.5 Cell Proliferation, Cyp2c Epoxygenase Expression, or EET Biosynthesis-To evaluate the effects of Wy-14643 and EETs on p60.5 cell proliferation, cells were cultured in the presence of different concentrations of Wy-14643 or synthetic EETs, and proliferation estimated by measurements of [ 3 H]thymidine incorporation. Compared with vehicle-treated cells, neither Wy-14643 nor the synthetic EETs cause significant changes in p60.6 cell proliferation (Figs. 5, A and B). Enzymatic and RT-PCR analyses showed that p60.5 cells catalyze the epoxidation of AA to EETs and contain transcripts coding for Cyp2c29, -38, -40, and -44 (Fig. 5, C and D). However, neither the levels of Cyp2c epoxygenase transcripts nor cellular EET biosynthesis were significantly altered by Wy-14643 (Fig. 5, C and D), although a reduction in Cyp2c44 transcripts was observed at 100 M Wy-14643 (not shown). We concluded that, in contrast to endothelial cells, p60.5 cells are not a target for growth regulation by Wy-14643 or EETs. Organ and/or tissue-specific effects for PPAR␣ ligands on Cyp2c epoxygenase expression have been reported (5)(6)(7)44).

Wy-14643 Reduces Tumor Angiogenesis and Growth in Vivo-
To study the role of PPAR␣ in tumor angiogenesis and growth, p60.5 cells were injected subcutaneously into isogenic 129SvJ wild-type and PPAR␣KO mice, and thereafter each genotype was divided into a group left untreated and another administered Wy-14643, immediately after the injection. All animals were sacrificed after 2 weeks (the time at which the tumors in untreated mice reach ϳ15% of body weight), and their tumors were removed. As a measure of the effectiveness of Wy-14643, we compared the size of livers from treated and untreated mice. As shown in Fig. 6A, liver hypertrophy (1,19) was observed in wild-type, but not PPAR␣KO Wy-14643-treated mice.
Notably, Wy-14643 caused marked reductions in tumor size, volume, and mass when administered to wild-type mice (Fig. 6, B-D), whereas, in contrast, it had no effects on tumor-bearing PPAR␣KO mice (Fig. 6, B-D). These results show that the anti-tumorigenic effects of Wy-14643 are mediated by the host PPAR␣ nuclear receptor, identify PPAR␣ as a host antitumorigenic gene, and provide the   first unequivocal demonstration of a role for this receptor in extrahepatic tumorigenesis (19,29). Although it has been reported that bezafibrate (a mixed PPRA␣/ PPAR␦ ligand) and pioglitazone (a PPAR␥ ligand) suppress intestinal polyp formation in Apc 1309 and Apc min mice, none of those studies addressed the role of the PPAR␣ receptor in the effects of these drugs on colon cancer initiation (18,45).
To determine whether decreased angiogenesis could account for the anti-tumorigenic effects of Wy-14643, frozen tumor sections were stained with anti-CD31 and anti-von Willebrand factor antibodies, as endothelial cell markers, and the extent of vascularization analyzed by fluorescence microscopy. Wy-14643 caused significant reductions in the vascularization of tumors grown in wild-type mice as shown by significant reductions in the levels of CD31 and von Willebrand immunoreactive proteins (shown for CD31 associated fluorescence in Fig. 7, A and B), which paralleled reductions in tumor growth (Fig.  6, B-D). Importantly, Wy-14643 had no effect on tumor vascularization when the hosts were PPAR␣KO mice (Fig. 7, A and B), showing that its anti-tumorigenic and anti-angiogenic effects required a functional, host expressed, PPAR␣ receptor. Finally, frozen tumor sections were co-stained with anti-PPAR␣ and anti-CD31 antibodies and, as seen in Fig. 7C the tumors grown in wild-type mice express PPAR␣ in both endothelial and tumor cells, whereas in those grown in the PPAR␣KO host the expression of PPAR␣ is restricted to the tumor cells (Fig. 7C). These results indicate that the tumor cell PPAR␣ receptor does not play a significant role in the Wy-14643 anti-tumor activity.
The Anti-tumorigenic Effects of Wy-14643 Are Independent of Mouse Strain or Tumor Cell Type-To studythegeneralityoftheWy-14643mediated anti-tumorigenic effects,  wild-type BALB/c mice were injected subcutaneously with mouse colon carcinoma CT26 cells and either left untreated or treated with Wy-14643 as indicated above. CT26 cells were used for the study since: (a) they form highly vascularized tumors when injected subcutaneously into BALB/c mice (33,37,46); (b) their growth is enhanced by prostaglandin E2 (33) but not added EETs (Fig. 8A); and (c) Wy-14643 does not cause significant changes in their proliferation (Fig. 8B). Similar to the findings with the 129SvJ strain, Wy-14643-treated BALB/c mice showed marked reductions in hepatic epoxygenase expression (Fig. 9A) and Ͼ60% reduction in hepatic EET biosynthesis and plasma EET concentrations (not shown). Most importantly, compared with the untreated animals, the Wy-14643-treated BALB/c mice showed significant reductions in tumor volume and vascularization (Fig. 9, B-D), demonstrating that the anti-cancer effects of Wy-14643 appear to be independent of tumor cell type and mouse strain.
Finally, to further document the generality of the Wy-14643 anti-tumorigenic effects and to explore their potential relevance to tumors of human origin, nude mice were injected with human A549 nonsmall cell lung cancer cells and treated as above. Similar to p60.5 or CT26 cells, these cells generate highly vascularized tumor when injected into nude mice (47), and their growth is not affected by EETs or Wy-14643 (not shown). As with the previous two models, the administration of Wy-14643 to nude mice injected with A549 cells led to marked reductions in tumor volume (Fig. 9E), confirming the generality of the Wy-14643 effects and extending their potential relevance to human-derived tumors.

DISCUSSION
PPAR␣ is recognized as a key transcriptional regulator of lipolytic pathways such as fatty acid mito-  chondrial, peroxisomal, and microsomal oxidation (1,2,19,29). However, recent studies have documented tissue-specific proand anti-growth-promoting activities for this receptor and its ligands (16,18,48). In rodent liver, sustained exposure to PPAR␣ligandsincreases,inaPPAR␣-dependentfashion,cyclindependent kinase expression, DNA synthesis, and the incidence of liver tumors (19). In contrast, PPAR␣ ligands, but not necessarily PPAR␣ (49), have been reported to inhibit endothelial and vascular smooth muscle cell proliferation and migration (12)(13)(14)(15)(16) and to inhibit vascular epidermal growth factor receptor expression and signaling (13), suggesting they possess anti-angiogenic properties (12,15). In this study, we confirm that Wy-14643, a selective PPRA␣ ligand, inhibits endothelial cell proliferation and tubulogenesis and, importantly, demonstrate that the functional effects of Wy-14643 are absent in cells from PPAR␣KO mice and are thus mediated by this nuclear receptor.
Because the CYP2C epoxygenases are known PPAR␣ target genes (5-7) and because several reports have characterized the EETs as pro-angiogenic lipids (22)(23)(24) and documented the expression of CYP2C epoxygenases in the endothelium (24), we investigated whether the anti-angiogenic effects of Wy-14643 were associated with changes in Cyp2c expression. Three lines of evidence indicate that the anti-angiogenic effects of Wy-14643 result from changes in the expression of Cyp2c epoxygenase isoforms and/or EET levels: (a) phenotype-rescue experiments show that the inhibitory effects of Wy-14643 on endothelial cell proliferation and tubulogenesis can be reversed by the addition of synthetic EETs (Figs. 1 and 3), (b) semi-quantitative RT-PCR analysis of RNAs from endothelial cells cultured in the presence or absence of Wy-14643 showed that the ligand caused marked PPAR␣-mediated reductions in Cyp2c29, -2c38, and -2c40 epoxygenase transcripts, and (c) Western blots of microsomes from temperature-sensitive immortalized endothelial cells, using anti-Cyp2c44 and anti-CYP2C11 (a rat homologue of murine Cyp2c29, -2c38, and -2c40) antibodies, show that Wy-14643 reduced the cellular levels of anti-CYP2C11 and 2c44 immunoreactive epoxygenase proteins. Based on the RT-PCR data in Fig. 2, we tentatively identify the CYP2C11 immunoreactive protein band as a mixture of Cyp2c29, -2c38, and -2c40. We and others have identified rodent and human CPY2Cs as predominant endothelial cell AA epoxygenases (22,24) and, using a Cyp2c-selective inhibitor, as the enzymes responsible for EET biosynthesis in these cells (24). However, catalytic assignments within the members of the Cyp2c gene subfamily are complicated by their extensive structural homology and overlapping catalytic selectivities (39,40). For example, murine Cyp2c29, -2c38, and -2c40 catalyze AA epoxidation with variable degrees of efficiency and regioselectivity (40), and Cyp2c44 is an active AA epoxygenase (39). Finally, the observation that added EETs are sufficient to rescue the endothelial cells from the Wy-14643-induced phenotypes, indicates that changes in free AA and/or increases in prostanoid or HETE formation play only a limited, if any, role in the responses to the PPAR␣ ligand.
The endothelium is in constant and direct contact with the circulating plasma and receives from it, among other factors, fatty acids, lipids, and pro-angiogenic molecules. The earlier demonstration of lipoprotein-associated EETs in rat and human plasma (42) suggested these could serve as a source of EETs for the endothelium. Moreover, the liver has been proposed as a source of plasma EETs (42), and the PPAR␣-mediated negative regulation of hepatic CYP2C genes is well established (5)(6)(7). The analysis of the effects of Wy-14643 on the expression of liver epoxygenases demonstrated that, together with a ligand-mediated down-regulation of Cyp2c29, Cyp2c40 and Cyp2c44 transcripts, there were marked reductions in the corresponding proteins, microsomal EET biosynthesis, and plasma EETs. Notably, all of the above Wy-14643-induced changeswereabsentinPPAR␣KOmiceandarethereforePPAR␣dependent. Compared with wild-type, untreated PPAR␣KO mice show increased basal concentrations of plasma EETs, suggesting that the receptor may participate in the regulation of steady-state EET liver biosynthesis, hydration, or secretion. In this regard, the roles for PPAR␣ in the regulation of hepatic lipoprotein secretion, cytosolic epoxide hydrolase expression, and fatty acid ␤and -oxidation are published (3).
Several facts are suggestive of a functional association between the epoxygenase and PPAR␣ activation. For example, the anti-growth and/or anti-angiogenic properties of PPAR␣ ligands have been linked to changes in vascular epidermal growth factor receptor levels and signaling (13), Akt and p38 activity (14,50,51), or p16 INK4a expression (16). On the other hand, published data indicate that EETs mediate growth factor signaling (22,52), activate ERK (extracellular signal-regulated kinase), p38, and phosphatidylinositol 3-kinase cascades (24), and promote neovascularization in vivo (24). Taken together, the published evidence and the data shown here provide strong support for a role of the epoxygenase and its metabolites in the changes elicited in endothelial cell proliferation and tubulogenesis by PPAR␣ ligand activation.
Tumor growth and invasiveness are highly dependent on the blood supply, and thus, the identification of novel angiogenic mediators is an area of research that has lead to the characterization of antagonists and/or inhibitors of angiogenesis as promising and attractive therapeutic targets (53,54). The studies reported here incorporate to the list of anti-angiogenic molecules the PPAR␣ ligands, as a class of conceptually novel antiangiogenic and anti-tumorigenic agents. Although PPAR␣ and PPAR␥ ligands have been shown to suppress polyp formation in Apc-deficient mice (18,45), and Wy-14643 to reduce mammary tumor progression in rats treated with dimethylben(a)anthracene (55), the present results demonstrate unequivocally, and for the first time, that the anti-tumorigenic effects of Wy-14643 require a functional, host-expressed, PPAR␣ receptor and thus, identify PPRA␣ as an anti-tumorigenic, anti-angiogenic gene. Furthermore, our initial mechanistic studies indicate that the anti-angiogenic and anti-tumorigenic properties of Wy-14643 are associated with PPAR␣-dependent decreases in epoxygenase expression and in the levels of pro-angiogenic EETs. Finally, the demonstration that the anti-tumorigenic effects of Wy-14643 are evident, in three different mouse strains and three different tumor cell types, strongly indicates that its anticancer properties might be of a general nature.
PPAR␣ plays an established role in the regulation of lipid homeostasis (1-4), and there are documented associations between obesity, high fat diets, and plasma dyslipidemia with the incidence of cancer in humans, particularly colon cancer (56 -58). Indeed, the colon polyp prone Apc mice develop severe hypertriglyceremia and hypercholesterolemia (18,45), and the reported anti-polyp effects of bezafibrate have been linked to its effects on plasma lipids (18,45). On the other hand, PPAR␣ ligands are used extensively for the management of lipid disorders and plasma dyslipidemia, and, after several years of continuous clinical use, these drugs have been proven to be safe and well tolerated, to have low toxicity, and to have limited side effects. Although the goal of most current anti-cancer therapies is the inhibition of tumor cell proliferation, low tolerance and high toxicity continues to be a major limitation for most available anti-cancer therapies. The data presented here supports the concept that tumor angiogenesis is an effective therapeutic target. Furthermore, the demonstration, that the antiangiogenic effects of Wy-14643 require a host-expressed PPAR␣ gene and are associated with its effects on the expression of Cyp2c epoxygenases and EET biosynthesis, identifies two host-associated targets that could provide novel avenues for the development of better tolerated and safer anti-cancer therapies. Finally, it would be of interest to perform epidemiological studies of potential associations between the use of these drugs and the incidence of cancer and/or its progression. However, it is important to recognize that the ligand binding affinity of most fibrates in clinical use is comparatively low and that there are important differences between rodent and humans in PPAR␣ expression levels, signaling efficiency, and gene targets (59,60).