Peroxisome Proliferator-activated Receptor γ Ligands Are Potent Inhibitors of Angiogenesis in Vitro and in Vivo *

Peroxisome proliferator-activated receptor γ (PPARγ) is a nuclear receptor that functions as a transcription factor to mediate ligand-dependent transcriptional regulation. Activation of PPARγ by the naturally occurring ligand, 15-deoxy-Δ12,14-prostaglandin J2(15d-PGJ2), or members of a new class of oral antidiabetic agents, e.g. BRL49653 and ciglitizone, has been linked to adipocyte differentiation, regulation of glucose homeostasis, inhibition of macrophage and monocyte activation, and inhibition of tumor cell proliferation. Here we report that human umbilical vein endothelial cells (HUVEC) express PPARγ mRNA and protein. Activation of PPARγ by the specific ligands 15d-PGJ2, BRL49653, or ciglitizone, dose dependently suppresses HUVEC differentiation into tube-like structures in three-dimensional collagen gels. In contrast, specific PPARα and -β ligands do not affect tube formation although mRNA for these receptors are expressed in HUVEC. PPARγ ligands also inhibit the proliferative response of HUVEC to exogenous growth factors. Treatment of HUVEC with 15d-PGJ2also reduced mRNA levels of vascular endothelial cell growth factor receptors 1 (Flt-1) and 2 (Flk/KDR) and urokinase plasminogen activator and increased plasminogen activator inhibitor-1 (PAI-1) mRNA. Finally, administration of 15d-PGJ2 inhibited vascular endothelial cell growth factor-induced angiogenesis in the rat cornea. These observations demonstrate that PPARγ ligands are potent inhibitors of angiogenesis in vitro and in vivo, and suggest that PPARγ may be an important molecular target for the development of small-molecule inhibitors of angiogenesis.

Angiogenesis, the formation of new blood vessels, is not only critically involved in a number of normal physiological processes such as embryonic development, ovulation, and wound healing, but is also a critical step in many pathological conditions including solid tumor growth, diabetic retinopathy, and age-related macular degeneration (1). The complex steps involved in new vessel formation have been well characterized in recent years and involve the degradation of the basement membrane by cellular proteases, the penetration and migration of endothelial cells into the extracellular matrix, endothelial proliferation, and the resultant formation of patent, interconnected vascular networks (2,3). It is now well established that the progression of solid tumor growth and metastasis is dependent on angiogenesis, with the resultant global effort by many laboratories to identify new angiostatic therapies for the treatment of cancer.
Peroxisome proliferator-activated receptors (PPARs) 1 are members of the steroid receptor superfamily and, as such, are ligand-activated transcription factors (4,5). Three subtypes of PPAR, ␣, ␤ (also known as ␦ or NUCI), and ␥ have been identified and cloned. Like other members of this superfamily, PPARs mediate transcriptional regulation by their central DNA binding domain that recognizes response elements in the promoters of specific target genes (6,7). Activation of PPAR␥ has been linked to adipocyte differentiation and regulation of glucose homeostasis in rodents and humans (8). Recent evidence also shows that the natural receptor ligand for PPAR␥, 15d-PGJ 2 (9,10), and synthetic antidiabetic thiazolidinedione drugs (e.g. BRL49653 and ciglitizone), inhibit macrophage and monocyte activation (11,12) and suppress tumor cell growth (14 -16). PPAR␥ can heterodimerize with at least one other member of the steroid receptor superfamily, retinoid acid receptor (RXR) (6,17). Specific ligands for the PPAR␥ and RXR have been shown to act synergistically to induce terminal differentiation of human liposarcoma cells (18) in vitro and to enhance insulin sensitivity in diabetic animals (19). Activation of PPAR␣ has been demonstrated to mediate lipid catabolism. Treatment of animals with PPAR␣ activators results in the proliferation of peroxisomes and the induction of hepatic genes involved in the ␤-oxidation of fatty acids (20). Mice that lack functional PPAR␣ accumulate lipid droplets in their livers (21). Several selective PPAR␣ activators have been described including WY 14643 and clofibrate (22). Inhibition of human aortic smooth muscle cell activation by PPAR␣ ligands has also been reported recently (13). Although PPAR␤ has been found to be ubiquitously expressed, specific functions for this receptor are not known at the present time.
In this study we report on the potent and novel inhibitory activity of PPAR␥ ligands on HUVEC differentiation into tubelike structures and proliferation in vitro, and the inhibition of VEGF elicited angiogenesis in vivo. These studies demonstrate that PPAR␥ is an important molecular target for the development of small molecule inhibitors of angiogenesis, which may be useful therapeutic agents in the treatment of cancer and other vasculoproliferative disorders.

EXPERIMENTAL PROCEDURES
Materials-PGA 1 , PGA 2 , PGB 1 , PGB 2 , PGD 1 , PGD 2 , PGE 1 , PGE 2 , ciglitizone, WY 14643, clofibrate, and 15d-PGJ 2 were from Cayman * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Chemical (Ann Arbor, MI). BRL49653 was kindly synthesized at Genentech by Dr. Jim Marsters (South San Francisco, CA). HUVEC were purchased from Clonetics (San Diego, CA) and maintained in Clonetics EGM medium supplemented with 10% fetal bovine serum (FBS) and endothelial cell growth supplements provided by the company. Type I rat tail collagen and recombinant basic fibroblast growth factor (bFGF) were purchased from Collaborative Research (Becton Dickinson Labware, Bedford, MA). Recombinant VEGF was from Genentech. 10ϫ medium 199 (M199) and PMA were purchased from Sigma. FBS was from Hyclone (Logan, Utah). ITS (insulin, transferrin, and selenium A), trypsin, and versene were from Life Technologies, Inc. The mouse monoclonal antibody to PPAR␥ (E-8, SC-7273) and goat polyclonal antibodies to PPAR␣ (N- 19, SC-1985) and PPAR␤ (C-20, SC-1983) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Magnetic protein A beads (Dynabeads) were from Dynal (Lake Success, NY).
Endothelial Tube Assay-Collagen gels were formed by mixing together ice-cold gelation solution (10ϫ M199, H 2 O, 0.53 M NaHCO 3 , 200 mM L-glutamine, type I collagen, 0.1 M NaOH, 100:27.2:50:10:750:62.5 by volume) and cells in 1ϫ basal medium (see below) at a concentration of 3 ϫ 10 6 cells/ml at a ratio of 4 volumes gelation solution:1 volume of cells. After gelation at 37°C for 30 min, the gels were overlaid with 1ϫ basal medium consisting of M199 supplemented with 1% FBS, 1ϫ ITS, 2 mM L-glutamine, 50 g/ml ascorbic acid, 26.5 mM NaHCO 3 , 100 units/ml penicillin, and 100 units/ml streptomycin supplemented with 40 ng/ml bFGF, 40 ng/ml VEGF, and 80 nM PMA. All drugs were added to the 1ϫ basal medium immediately after gelation. To quantitate tube formation, the number of tubes per high power (20X) field was determined at 48 h after addition of the basal medium. A tube was defined as an elongated structure comprised of one or more endothelial cells that exceeded 100 m in length (long axis). Five independent fields separated by 100 m optical sections were assessed for each well, and the average number of tubes/20X field determined. Cytoxicity was assessed using cell proliferation kit II from Boehringer Mannheim.
Reverse Transcription-Polymerase Chain Reaction (RT-PCR) Assays-Synthesis of complementary DNA (cDNA) was performed using 6 g of total RNA extracted from human liver (positive control) or HU-VEC cultured on three-dimensional collagen gels for 24 h by random primer according to the manufacturer's instructions (Stratagene Co., San Diego, CA). Subsequent amplifications of the partial cDNA encoding PPAR␣, ␤, and ␥ were performed using different amounts of reverse transcribed mixture (4 l for liver ␣, ␤, and ␥ and 5 l for HUVEC ␣ and ␤ and 6 l for HUVEC ␥) as templates with specific oligonucleotide primers as follows: PPAR␣ sense 5Ј-CCAGTATTTAGGACGCGGTC-C-3Ј and antisense 5Ј-AAGTTCTTCAAGTAGGCCTGCG-3Ј; PPAR␤ sense 5Ј-AACTGCAGATGGGCTGTGAC-3Ј and antisense 5Ј-GTCTCG-ATGTCGTGGATCAC-3Ј and PPAR␥ sense 5Ј-TCTCTCCGTAATGGA-AGACC-3Ј and antisense 5Ј-GCATTATGAGACATCCCCAC-3Ј. The expected sizes of PCR products for PPAR ␣, ␤, and ␥ were 492, 484, and 474 base pairs, respectively. Negative controls for reverse transcription and PCR amplifications were included. The PCR mixtures were subjected to 35 cycles of amplification by denaturation (30 s at 94°C), hybridization (30 s at 55°C), and elongation (45 s at 55°C). The PCR products were analyzed by 1.5% agarose gel electrophoresis with ethidium bromide.
Western Blot Assays-Confluent HUVEC grown on type I collagencoated flasks were washed once with ice-cold phosphate-buffered saline, then scraped and pelleted by centrifugation. Pellets were lysed in icecold lysis buffer (20 mM Tris/HCl, 150 mM NaCl, 2 mM EDTA, 1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride, 40 units/ml aprotinin, 15 g/ml leupeptin, 0.36 mM 1,10-phenanthroline, 1% Nonidet P-40, and 1% Triton X-100, pH 7.5). An aliquot equivalent to 100 g of protein was incubated for 16 h at 4°C with 20 ng of anti-PPAR antibody in a total volume of 500 l. Protein-A agarose beads (50 l) were then added, and the mixture incubated for 2 h at 4°C. The beads were washed three times with 500 l of lysis buffer, and the proteins remaining on the beads were solubilized by heating at 100°C for 5 min in 2ϫ SDS sample buffer. The precipitated complexes were then analyzed by Western blotting. Proteins were visualized using the ECL chemiluminescence kit (Amersham Pharmacia Biotech).
Proliferation Assay-HUVEC were seeded on collagen-coated 96-well plates at 6,000 cells/cm 2 in Clonetics EGM containing endothelial cell growth supplements and 10% FBS, 2 mM L-glutamine, 100 units/ml penicillin, and 100 units/ml streptomycin and allowed to attach for 4 h. Medium was then replaced with 1ϫ basal medium consisting of M199 supplemented with 1% FBS, 1ϫ ITS, 2 mM L-glutamine, 50 g/ml ascorbic acid, 26.5 mM NaHCO 3 , 100 units/ml penicillin, and 100 units/ml streptomycin supplemented with 40 ng/ml bFGF, 40 ng/ml VEGF, and 80 nM PMA. Cells were cultured in the above medium in the presence of drugs or vehicle for 4 h. 5 l (100 M) of 5Ј-bromo-2Јdeoxyuridine (BrdUrd) was added in a final volume of 100 l/well, and cells were incubated for another 20 h. BrdUrd incorporation was evaluated by an enzyme-linked immunosorbent assay kit from Boehringer Mannheim.
Corneal Angiogenesis Assay-A 1.5-mm incision was made approximately 1 mm from the center of the cornea of isoflurane-ketamine (60 -80 mg/kg) xylazine (10 -15 mg/kg) anesthetized Sprague-Dawley rats. Using a curved spatula, the incision was bluntly dissected through the stroma toward the outer canthus of the eye. A hydron pellet (2 ϫ 20 mm) containing VEGF (200 ng), sucralfrate (100 ng) with or without (control) 15d-PGJ 2 (10 g/pellet) was inserted into the base of the pocket. After surgery the eyes were coated with gentamicin ointment. Animals were observed at 24 -48 h for the occurrence of nonspecific inflammation and then daily thereafter. At day 6, the animals were euthanized and injected with fluorescein isothiocyanate-dextran to allow for visualization of the vasculature. Corneal whole mounts were made of the enucleated eyes and analyzed for neovascular area using computer-assisted image analysis (Image Pro-Plus 2.0, Silver Spring, MD).
Statistics-Data are expressed as the mean Ϯ S.E. Statistic analysis was performed using one-way analysis of variance (INSTAT, Graph Pad Software, Sorrento Valley, CA). Multiple comparisons against the control were analyzed using Bonferroni modification of Student's t test to determine differences between groups. A p value Ͻ 0.05 was accepted as significant.

RESULTS AND DISCUSSION
Endothelial Tube Formation in Three-dimensional Collagen Gels-In this study we used a model of in vitro angiogenesis initially described by Davis et al. (23). In this model, endothelial cells suspended in three-dimensional collagen lattices comprised of type I collagen in the presence of VEGF, bFGF, and PMA undergo rapid morphogenesis. Numerous vacuoles are observed in the majority of endothelial cells within 4 h; the formation of tube-like structures can be observed at 24 h, and an interconnected network of tube-like structures is observed (Fig. 1A) at 48 h. Inhibitors of mRNA synthesis (actinomycin D) and protein synthesis (cycloheximide) completely blocked the formation of tube-like structures (data not shown), demonstrating the requirement for new mRNA and protein synthesis.
Effects of Eicosanoids on Tube Formation by HUVEC-It has recently been shown that a number of eicosanoid metabolites can serve as PPAR ligands (8 -10, 22), and activation of PPARs has been linked to numerous functions in a number of cell types (8,(11)(12)(13)(14)(15)(16). To determine whether PPARs could affect HUVEC differentiation, we investigated the effect of 15d-PGJ 2 , a naturally occurring endogenous ligand of PPAR␥ on HUVEC tube formation. HUVEC were grown in three-dimensional collagen gels in medium containing VEGF, bFGF and PMA with addi-tion of 15d-PGJ 2 (10 M) for 48 h and were analyzed for tube formation. As shown in Fig. 1B, 15d-PGJ 2 markedly reduced HUVEC tube formation. The majority of cells in the gels remained rounded and devoid of vacuoles or lumen-like structures. Vehicle-treated control groups did not show any significant inhibitory effect (data not shown). We extended our observations to a variety of metabolites of 20:4 and 20:3 fatty acids for their ability to inhibit tube formation in three-dimensional collagen gels (Fig. 1C). PGJ 2 and PGD 2 and their derivatives/analogs demonstrated dose-dependent inhibition of tube formation over the concentration range of 1 to 100 M. PGB 1 and PGB 2 possessed only weak inhibitory effects. In contrast, two prostanoids known to activate cell surface G-protein-coupled receptors, namely PGE 1 and PGE 2 , had no significant effect over this concentration range. Among the tested drugs, 15d-PGJ 2 was the most potent inhibitor of HUVEC tube formation. Data from parallel cytotoxicity assays indicated that none of the tested prostanoids exhibited significant toxicity over the time course (48 h) of the experiment (data not shown). Previous studies (24,25) have demonstrated that PGD 2 can spontaneously convert to PGJ 2 , and in the presence of serum or albumin, to 15d-PGJ 2 , a natural selective ligand for PPAR␥. Therefore, activation of PPAR␥ may modulate endothelial morphogenesis.
PPAR␥ Is Expressed in HUVEC-RT-PCR was used to demonstrate the expression of all three PPAR (␣, ␤, and ␥) mRNAs in HUVEC cultured in the three-dimensional collagen gels. In two independent experiments conducted with different HU-VEC populations derived from different donors, the mRNA expression of three isoforms of PPAR were demonstrated ( Fig.  2A). By using real time quantitative RT-PCR (Taqman) (26,27), the relative mRNA expression levels of each isoform were determined to be PPAR␤ Ͼ PPAR␣ Ͼ PPAR␥ (Fig. 2B). Preliminary observations also suggested that expression of PPAR␥ mRNA was not altered in medium containing VEGF, bFGF and PMA at 4, 24 or 48 h of incubation with 15d-PGJ 2 or vehicle in HUVEC cultured in three-dimensional collagen gels or on collagen-coated surfaces (data not shown). The expression of PPAR␥ protein in HUVEC was confirmed by Western blot analysis (Fig. 2C), and PPAR␥ protein levels were not significantly altered by the growth factor mixture when HUVEC were cultured on collagen-coated surfaces (Fig. 2C). Using commercially available antibodies, we were unable to demonstrate protein expression of PPAR␣ and -␤ in HUVEC (data not shown). The expression of PPAR␣ mRNA in HUVEC has been previously reported (28); however, this study demonstrates for the first time the expression of PPAR␥ mRNA and protein and PPAR␤ mRNA by HUVEC.
PPAR␥ Mediates the Inhibitory Effect on HUVEC Tube Formation and Proliferation-To determine which PPAR isoform(s) mediated the inhibition of HUVEC tube formation, the effect of specific PPAR ligands was assessed. As shown in Fig.  3A, treatment of HUVEC with the specific PPAR␥ ligands, 15d-PGJ 2, BRL49653, and ciglitizone, dose dependently inhibited tube formation with half-maximal inhibition concentrations (IC 50 ) of 2.8, 6.2, and 2.7 M, respectively. In contrast, the selective agonists for PPAR␣ (WY 14643 and clofibrate) and PPAR␤ (erucic acid) (29) did not significantly inhibit tube formation (Fig. 3B). These observations demonstrated that activation of PPAR␥ results in inhibition of endothelial cell differentiation into tube-like structures in vitro. We further investigated whether PPAR␥ ligands also regulate the proliferative response of HUVEC to growth factors. BrdUrd incorporation assays were performed to detect proliferation of HUVEC cultured on type I collagen-coated surface in medium containing VEGF, bFGF, and PMA in the presence of vehicle or PPAR␥ ligands. As shown in Fig. 3C, 15d-PGJ 2 and BRL49653 dose dependently inhibited HUVEC proliferation with IC 50 of 2.4 and 15.7 M, respectively. In contrast, PGE 1 and PGE 2 did not affect HUVEC proliferation (data not shown). These results demonstrate that PPAR␥ ligands also repress growth factorinduced HUVEC proliferation in vitro.
15d-PGJ 2 Regulates Gene Expression Events Associated with Angiogenesis-Flk/KDR and Flt-1 are two structurally related endothelial cell tyrosine kinase receptors for VEGF. The importance of these two receptors during angiogenesis has been clearly demonstrated by the findings that KDR functions as a

FIG. 2. Expression of PPAR in HUVEC.
A, RT-PCR analysis using specific primers for PPAR␣, -␤, or -␥ as described under "Experimental Procedures." Specific cDNA were synthesized from human liver RNA (positive control) and from HUVEC RNA using random primer in the presence of 50 units of Moloney murine leukemia virus reverse transcriptase. PCR products were resolved on a 1.5% agarose gel. The DNA ladder was included as a marker to indicate the sizes of the PCR products of PPAR␣ (lanes 1 and 4), -␤ (lanes 2 and 5), and -␥ (lanes 3 and  6). B, quantitative analysis of mRNA expression of PPARs in HUVEC by real time RT-PCR (Taqman). Total RNA (100 ng) were extracted from HUVEC grown in three-dimensional collagen gels in medium containing VEGF, bFGF and PMA. The relative expression of PPAR isoforms were normalized to GAPDH levels measured in the same RNA preparation. Data shown are from three independent experiments analyzed in duplicate and are expressed as mean Ϯ S.E. C, Western blot analysis of PPAR␥ expression in HUVEC. Tissue extract from mouse adipose tissue (20 g of protein, positive control) or HUVEC cell lysates (100 g of protein) prepared from cells cultured on collagen-coated surfaces in 1ϫ basal medium without (lane 1) or containing (lane 2) VEGF, bFGF, and PMA for 24 h were immunoprecipitated by PPAR␥ monoclonal antibody. Following SDS-polyacrylamide gel electrophoresis and transfer to nitrocellulose, the proteins were immunoblotted with the same PPAR␥ monoclonal antibody. Similar results were obtained from two independent experiments with different HUVEC cultures. HUVEC grown in three-dimensional collagen gels (Fig. 4, A  and B, open bars). 15d-PGJ 2 inhibited the induction of both VEGF receptor mRNA in HUVEC cultured in three-dimensional collagen gels (Fig. 4, A and B, hatched bars). Similar data were obtained when HUVEC were grown on type I collagen-coated surface with growth factor stimulation (not shown). It is also well known that the production of proteases (e.g. plasminogen activators) and their inhibitors (e.g. plasminogen activator inhibitor I, PAI-1) is correlated with endothelial cell degradation of extracellular matrix and migration, the two critical steps of the angiogenic processes (33). We analyzed the effects of 15d-PGJ 2 on gene expression of uPA and PAI-1 in three-dimensional collagen gels. Treatment of HUVEC with 15d-PGJ 2 reduced uPA mRNA at 4 h and increased PAI-1 gene expression at 24 h, respectively (Fig. 4, C and D, hatched bars). Taken together, these observations indicate some possible molecular mechanisms by which PPAR␥ ligands mediate inhibition of angiogenesis.
15d-PGJ 2 Demonstrates Anti-angiogenic Activity in Vivo-To determine the potential anti-angiogenic effect of PPAR␥ ligand 15d-PGJ 2 in vivo, hydron pellets containing 200 ng of recombinant VEGF with or without 10 g of 15d-PGJ 2 were implanted into the corneas of Sprague-Dawley rats. Summary data from the experiment show that pellets containing the combination of 15d-PGJ 2 and VEGF demonstrated a significant reduction in vessel area compared with the VEGF only positive controls (Fig. 5, A, B, and C). Results from this experiment were repeated in a second independent experiment. These in vivo data strongly support our in vitro observations that activation of PPAR␥ results in marked inhibition of angiogenesis.
Recent evidence has demonstrated that PPAR␥ is highly expressed in most colon cancer cells and breast cancer cells (14,16). Activation of the PPAR␥ resulted in growth arrest and induction of differentiation of cancer cells and reduction of tumor cell growth rate, suggesting potential implications of PPAR␥ as a target for treatment of human cancers. In this study we have documented, for the first time, the expression of PPAR␥ in human endothelial cells. Activation of PPAR␥ with either naturally occurring ligand or synthetic selective ligands results in potent inhibition of endothelial differentiation into tube-like structures and proliferation in vitro and suppression of VEGF-induced angiogenesis in vivo. PPAR␥ activation also inhibits the expression of at least three important genes in the angiogenic process, the VEGF receptors Flk/KDR, Flt-1, and the protease uPA. This study thus provides a new insight into the mechanism of PPAR␥ ligands as inhibitors of solid tumor growth and suggests that such drugs may also provide novel means to control other angiogenic disorders.