Antiangiogenic Effect of Rosiglitazone Is Mediated via Peroxisome Proliferator-activated Receptor γ-activated Maxi-K Channel Opening in Human Umbilical Vein Endothelial Cells*

Recent evidence shows that peroxisome proliferator-activated receptor γ (PPARγ) ligands induce the antiangiogenic effect in endothelial cells and tumors. In the present study, we elucidated the involvement of maxi-K channel activation in the antiangiogenic effect of rosiglitazone, a well known PPARγ ligand in human umbilical vein endothelial cells. We found that the antiangiogenic effects of rosiglitazone were reversed by either bisphenol A diaglycidyl ether, a PPARγ antagonist, or iberiotoxin, a maxi-K channel blocker. Knockdown of maxi-K channel expression also reversed the antiangiogenic effects. Iberiotoxin reversed the rosiglitazone-induced hyperpolarization while having no effect on the endogenous PPARγ activation, suggesting that rosiglitazone activates maxi-K channel via PPARγ. In the rosiglitazone-induced antiangiogenic process, endothelial nitric-oxide synthase-Ser1179 phosphorylation and NO production were significantly elevated, and treatment with the NOS inhibitor NG-monomethyl-l-arginine acetate abolished the antiangiogenic and apoptotic effects of rosiglitazone, indicating NO as a key mediator of the rosiglitazone actions. In conclusion, rosiglitazone significantly inhibited VEGF165-induced angiogenesis by a proapoptotic mechanism via PPARγ-mediated NO production, followed by maxi-K channel opening.

Angiogenesis, the growth of new capillaries from preexisting microvessels (1), plays an important role in a variety of physiological processes, including embryonic development, wound healing, and tissue or organ regeneration (2) and in the pathological development and progression of various cancers, rheumatoid arthritis, diabetic retinopathy, and psoriasis (3). Among many factors stimulating angiogenesis, vascular endothelial growth factor (VEGF) 2 is a key regulatory factor of the angiogenic process in either physiological or pathological conditions (4,5). Since the progression of tumor growth and metastasis appears to be dependent on angiogenesis, antiangiogenic drugs are clinically relevant for cancer therapy.
Recent investigations suggested that PPAR␥ ligands have inhibitory effects on tumor cell lines, but the effects appear not to be entirely elicited by the direct action on tumor cells. Similarly, Panigrahy et al. (12) reported that PPAR␥ is expressed in tumor endothelium and in cultured endothelial cells and that rosiglitazone inhibits angiogenesis in vitro and in vivo via its effects on endothelium. In addition, 15-deoxy-⌬-12,14-prostaglandin J2, an endogenous PPAR␥ ligand, inhibited VEGF-induced angiogenesis in the rat cornea (13) and induced caspasemediated endothelial cell apoptosis via a PPAR-dependent pathway (9). Possible mechanisms of antiangiogenic actions by PPAR␥ activation include the inhibition of mitogen-activated protein kinase-dependent activation (14); up-regulation of angiogenesis inhibitor maspin and CD36 (15), the receptor for antiangiogenic thrombospondin; inhibition of matrix metalloproteinase, VEGF, and VEGF receptor expression; and increase of plasminogen activator inhibitor and matrix metalloproteinase inhibitor expression (12,13,16). However, the direct mechanism for the antiangiogenic effects of PPAR␥ activation in the endothelial cells remains currently unclear.
The endothelium plays a pivotal role in a variety of vascular functions such as blood pressure control, vascular remodeling, and angiogenesis, where ion channels, particularly large conductance Ca 2ϩ -activated K ϩ channels (also known as the BK channel or maxi-K channel), are the key mediators (17). Maxi-K channels are distributed in various tissues, such as human brain tumors, vascular smooth muscle cells, and endothelial cells, and their activations were shown to inhibit migration of human glioma cells (18). The maxi-K channels are synergistically regulated by various intracellular second messengers, such as cytoplasmic Ca 2ϩ concentration, cGMP, hydrogen peroxide, and nitric oxide (NO) (19 -21). NO is produced by a group of enzymes called nitric-oxide synthases (NOS), and endothelial NOS (eNOS or NOS3) is responsible for the production of NO in vascular endothelial cells. Although the expression and activity of eNOS seems to be constitutive, the activity was shown to be regulated by serine phosphorylation (22).
In this study, we investigated the possible involvement of maxi-K channels in the antiangiogenic effects of rosiglitazone in human umbilical vein endothelial cells (HUVECs). The inhibitory action of rosiglitazone on VEGF-stimulated angiogenesis was blocked either by iberio-toxin or by knockdown of maxi-K channel expression, and the effect was associated with apoptotic action as shown by DNA fragmentation and Bcl-2 and Bax regulation. The mediator involved in the maxi-K channel opening by rosiglitazone appears to be NO, produced by PPAR␥-activated eNOS. These findings provide important understanding that maxi-K channel plays a critical role in the rosiglitazone-induced antiangiogenesis in HUVECs.

EXPERIMENTAL PROCEDURES
Cell Cultures-HUVEC (ATCC CRL-1730; endothelial cell line derived from vein of human umbilical cord) were cultured in Kaighn's F12K medium supplemented with 10% heat-inactivated fetal bovine serum, 0.1 mg/ml heparin sodium, 0.03-0.05 mg/ml endothelial cell growth supplement, and 1% antibiotics (100 units/ml penicillin and 100 g/ml streptomycin). Cells were grown to confluence at 37°C in 5% CO 2 on 0.1% gelatin-coated culture dishes and used for experiments at no greater than passage 8.
Angiogenesis Assays-The tube formation assay was performed on 24-well plates coated with 250 l of Matrigel Basement Membrane Matrix (10 mg/ml)/well and polymerized for 30 min at 37°C. Cells treated with either bisphenol A diaglycidyl ether (BADGE) or iberiotoxin for 30 min at 37°C were plated onto a layer of Matrigel at a density of 1 ϫ 10 5 cells/well and followed by the addition of either rosiglitazone or NS-1619. Matrigel cultures were incubated at 37°C for 18 h and photographed (ϫ100).
An in vivo Matrigel plug assay was carried out by injecting 0.5 ml of Matrigel containing VEGF 165 (5 ng/ml) and heparin (40 units/ml) into C57BL/6J mice subcutaneously. After 5 days, mice were sacrificed, and Matrigel plug was recovered, fixed with 10% formaldehyde/phosphatebuffered saline (pH 7.4) and examined with hematoxylin/eosin stain after paraffin embedment. To quantify the formation of new blood vessels, the amount of hemoglobin was measured using the total hemoglobin kit. The concentration of hemoglobin was calculated from a known amount of hemoglobin assayed in parallel.
Cell Proliferation Assay-HUVECs seeded at a density of 1 ϫ 10 4 cells/well in 96-well plates were incubated for 3 h in Kaighn's F12K medium containing 1% fetal bovine serum and exposed to rosiglitazone or NS-1619 (0.1-10 M) for 1 h followed by VEGF 165 (10 ng/ml) stimulation for 4 days. After incubation, an MTT assay was carried out (23), and the optical density was measured at 570 -630 nm using a microplate spectrometer (Bio-Rad).
Apoptosis Assays-Cells were treated with either rosiglitazone (10 M) or NS-1619 (10 M) for 3 h in the presence or absence of either BADGE (20 M), iberiotoxin (0.3 M), clotrimazole (10 M), or glibenclamide (10 M). Then cells were stimulated with 10 ng/ml VEGF 165 for 2 days. DNA fragmentation was detected by electrophoresis of DNA (15-20 g/well) on 1.6% agarose gel, followed by visualization using ethidium bromide staining, and gel pictures were taken by UV transillumination.
Apoptotic index was further quantified by acridine orange/ethidium bromide uptake, as previously reported (24). Cells treated as described above were stained by using acridine orange (final 25 g/ml) and ethidium bromide (final 25 g/ml) mixture for 10 min, and then cells were photographed (ϫ100).
Membrane Potential Measurement Using Voltage-sensitive Fluorescent Dye-Cells (1.5 ϫ 10 4 cells/well) were incubated with 200 nM DiBAC in Dulbecco's phosphate-buffered saline for 30 min at 37°C and exposed to BADGE (20 M), iberiotoxin (0.3 M), or N G -monomethyl-L-arginine acetate (L-NMMA, 10 M) for 1 h and then stimulated by rosiglitazone or NS-1619 (10 M) for 3 h, followed by VEGF 165 (10 ng/ml) for 1 h. The fluorescence intensities of DiBAC molecules were recorded at 545 nm using a fluorescence plate reader (Bio-Tek Instruments, Inc. Winooski, VT). Hyperpolarization resulted in the extrusion of the dye from cells, causing a subsequent decrease in fluorescence intensity.
Maxi-K Channel Knockdown Using Stealth TM RNAi Oligonucleotide-The Stealth TM RNAi oligonucleotide was synthesized by Invitrogen with the following sequence complementary to human maxi-K mRNA (GenBank TM accession number nm 002547, beginning at the position 1726 target sequence site): sense, 5Ј-CCG AAG AUA AGA AUC AUC ACU CAA A; antisense, 5Ј-UUU GAG UGA UGA UUC UUA UCU UCG G. The Stealth TM RNAi negative control Duplex (Invitrogen) was used as a control oligonucleotide. Transfection efficiency was monitored using a fluorescent oligonucleotide (BLOCK-iT fluorescent oligonucleotide; Invitrogen) and estimated to be 80 -90%.
The Stealth TM RNAi molecules were transfected into HUVECs using Lipofectamine and Plus reagent following Invitrogen's protocols. The final concentration of 100 nM Stealth TM RNAi oligonucleotide was empirically determined to maximally suppress target RNA expression, and the Stealth TM RNAi oligonucleotide was transfected to the medium 48 h prior to the treatment of rosiglitazone. The ability of the Stealth TM RNAi oligonucleotide to knock down maxi-K channel expression was analyzed by Western blot and real time PCR on whole cell extracts.
Measurement of Cytosolic Ca 2ϩ and NO Concentrations-Cells transfected with either control RNAi or Maxi-K channel Stealth TM RNAi were seeded at 5 ϫ 10 4 cells/well in 12-well tissue culture plates and incubated with Kaighn's F12K medium containing 1% fetal bovine serum plus BADGE or iberiotoxin for 1 h and then exposed to rosiglitazone or NS-1619 for 3 h. Thereafter, VEGF 165 was added and incubated for 3 h. Cytosolic Ca 2ϩ concentration was determined by measuring the fluorescence of fluo-3 AM (Molecular Probes, Inc., Eugene, OR) at 545 nm using a fluorescence plate reader (Bio-Tek Instruments). NO production by HUVECs was measured as nitrite (a stable metabolite of NO) concentrations in cell culture supernatants using Griess reagent system following Promega's protocols.
Drugs-Recombinant human VEGF 165 was purchased from R&D Systems. Rosiglitazone was synthesized in the Korea Research Institute of Chemical Technology (Daejeon, Korea). NS-1619, iberiotoxin, BADGE, clotrimazole, endothelial cell growth supplement, heparin, acridine orange, ethidium bromide, and MTT were from Sigma. The Hemo-s reagent kit was from YD Diagnostics (Seoul, Korea). L-NMMA and glibenclamide were from Tocris. Matrigel was from BD Biosciences. Rosiglitazone, NS-1619, BADGE, iberiotoxin, clotrimazole, glibenclamide, and L-NMMA were dissolved in dimethyl sulfoxide as a 20 mM stock solution and then diluted with phosphate-buffered saline.
Statistics-The results are expressed as means Ϯ S.E. Student's t test was used for analyzing values between vehicle groups and compoundtreated groups. p Ͻ 0.05 was considered to be statistically significant.

Effects of Maxi-K Channel Modulators and Rosiglitazone on VEGF-induced
Angiogenesis-Matrigel-plated HUVECs elongated and migrated in the presence of VEGF, forming a tubular network, as evidenced by morphological changes. Both rosiglitazone and NS-1619, a maxi-K channel opener, markedly suppressed the formation of tube-like structures at 10 M (Fig. 1). The suppression of tube formation by rosiglitazone (10 M) was reversed by either BADGE (20 M), a PPAR␥ antagonist, or iberiotoxin (0.3 M), a maxi-K channel blocker. The concentrations of BADGE and iberiotoxin were selected as 20 and 0.3 M, respectively, which showed maximum effect without cytotoxicity.
Five days after implantation of Matrigel containing VEGF 165 (5 ng/ml) in mice, histological examination and measurement of hemoglobin content of Matrigel were carried out. The microvessel formation and hemoglobin content were significantly increased in the Matrigel containing VEGF 165 (5 ng/ml) (Fig. 2, B and F). Rosiglitazone (10 M) significantly reduced the formation of VEGF 165 -stimulated neomicrovessels and hemoglobin contents (57% inhibition; VEGF 165 only, 6.12 Ϯ 0.60 g/dl; rosiglitazone ϩ VEGF 165 , 2.44 Ϯ 0.47 g/dl), which was completely antagonized by either BADGE or iberiotoxin (Fig. 2F). These results suggest that rosiglitazone elicited a strong in vivo antiangiogenic activity via PPAR␥-dependent maxi-K channel opening.
Effects of Maxi-K Channel Modulators and Rosiglitazone on the VEGF-induced Cell Proliferation-When HUVECs were incubated in the medium containing 10 ng/ml VEGF 165 for 4 days, cell proliferation was increased to about 2-fold. Rosiglitazone (0.1-10 M) suppressed the cell proliferation in a concentration-dependent manner with about 70% inhibition at 10 M (Fig. 3). The effect of rosiglitazone was partially reversed by BADGE (20 M) and iberiotoxin (0.3 M), suggesting that  the antiproliferative effects of rosiglitazone may be, at least in part, mediated by PPAR␥ activation and by maxi-K channel opening (Fig. 3). The extent of the observed inhibitory effects of rosiglitazone was similar to that described in previous reports (13,25). In parallel, NS-1619 (0.1-10 M) significantly suppressed the cell proliferation induced by VEGF 165 (10 ng/ml) in a concentration-dependent manner, which was reversed by iberiotoxin but was not reversed by BADGE.
Apoptotic Effects of Rosiglitazone on HUVECs-To examine whether antiangiogenic and antiproliferative activities of rosiglitazone are related with apoptosis, we measured the effects of rosiglitazone on DNA fragmentation. Rosiglitazone (10 M) induced prominent oligonucleosomal DNA fragmentation, indicating the apoptotic action of rosiglitazone (Fig. 4A). Apoptotic cell death induced by rosiglitazone was further confirmed by acridine orange and ethidium bromide staining. As shown in Fig. 4C, control and VEGF 165 -treated cells had a uniform green color in nuclei, but rosiglitazone-treated cells had bright orange areas of condensed chromatin in the nucleus. The rosiglitazone-induced apoptosis was inhibited by either BADGE (20 M) or iberiotoxin (0.3 M), suggesting that the apoptosis seems to be mediated by PPAR␥-dependent maxi-K channel opening.
NS-1619 (10 M) significantly induced the DNA fragmentation and stained orange in the nucleus, which was reversed by iberiotoxin (0.3 M) but not by BADGE (20 M) (Fig. 4). Either BADGE or iberiotoxin alone had no effect on oligonucleosomal DNA fragmentation.
On the other hand, clotrimazole (10 M), an intermediate conductor of Ca 2ϩ -activated K ϩ channel blocker, and glibenclamide (10 M), an ATP-sensitive K ϩ channel blocker, were not capable of inhibiting rosiglitazone-induced DNA fragmentation, suggesting the specific involvement of maxi-K channel in the rosiglitazone-induced apoptotic action (Fig. 4B).
The DNA fragmentation by rosiglitazone was accompanied by suppression of Bcl-2 (antiapoptotic protein) expression and increased Bax (proapoptotic protein) expression (Fig. 5), further suggesting that the suppression of endothelial cell viability is associated with apoptotic effects. These effects were also reversed by BADGE (20 M) and iberiotoxin (0.3 M).
Effects of Knockdown of Maxi-K Channel Expression on the Rosiglitazoneinduced Antiangiogenic and Antiproliferative Actions-The transfection of Stealth TM RNAi oligonucleotide in HUVECs resulted in the reduction of maxi-K channel expression to 10% of control (Fig. 6A). In contrast to the antiangiogenic and antiproliferative effects of rosiglitazone in negative control Stealth TM RNAi-transfected HUVECs, rosiglitazone (10 M) was ineffective on VEGF 165 -induced tube formation and cell proliferation in maxi-K channel-knocked down cells (Fig. 6, B and C). These results provide conclusive evidence that maxi-K channel activation is required for the antiangiogenic and antiproliferative effects of rosiglitazone on endothelial cells.
Rosiglitazone-induced Maxi-K Channel Opening via PPAR␥ Activation-To directly demonstrate that rosiglitazone induces maxi-K channel opening, we examined the effect of rosiglitazone on the membrane potential of HUVECs using DiBAC fluorescence. Rosiglitazone caused endothelial hyperpolarization, which was reversed by either BADGE (20 M) or iberiotoxin (0.3 M) (Fig. 7A). These results provide the direct evidence that rosiglitazone activates maxi-K channel opening in HUVECs. NS-1619 also resulted in the endothelial hyperpolarization being reversed by iberiotoxin but not reversed by either BADGE or L-NMMA.
To confirm that maxi-K channel activation is the downstream effect of rosiglitazone-induced PPAR␥ activation, the effect of iberiotoxin on the rosiglitazone-induced endogenous PPAR␥ activation was determined. Rosiglitazone increased endogenous PPAR␥ expression as expected, but this effect was not affected by iberiotoxin (Fig. 7B), indicating that activation of maxi-K channel is indeed the consequence of PPAR␥ activation by rosiglitazone. In addition, iberiotoxin had no effect on the endogenous PPAR␥ transcriptional activity stimulated by rosigli-tazone when determined with a transcription reporter assay using PPAR-responsive element reporter genes (data not shown).
Effect of Rosiglitazone on eNOS Phosphorylation and NO Release-To determine whether rosiglitazone opened the maxi-K channel via NO production, we examined the eNOS phosphorylation by rosiglitazone. The eNOS-Ser 1179 phosphorylation was significantly and concentration dependently elevated by rosiglitazone (1 and 10 M). Increased eNOS-Ser 1179 expression was antagonized by 20 M BADGE but not antagonized by 0.3 M iberiotoxin (Fig. 8A). On the other hand, eNOS-Thr 497 phosphorylation was little affected by rosiglitazone treatment.
Concurrent with increased eNOS-Ser 1179 phosphorylation, rosiglitazone increased NO production, as shown in Fig. 8B. Rosiglitazone (10 M) increased NO production to about 900 nM, which approximates the required concentration to activate maxi-K channels (19). Increased NO production by rosiglitazone was antagonized by either 20 M BADGE or 10 M L-NMMA (data not shown) but not antagonized by 0.3 M iberiotoxin or by knockdown of maxi-K channel expression (Fig. 8B). In addition, L-NMMA pretreatment pre-vented rosiglitazone-induced hyperpolarization (Fig. 7A), suggesting that NO is an important player for PPAR␥-mediated maxi-K channel activation. The DNA fragmentation and suppression of in vitro tube formation (Fig. 8, C and D) by rosiglitazone were antagonized by 10 M L-NMMA, further indicating that NO is involved as a key mediator of rosiglitazone-induced apoptotic and antiangiogenic effects. L-NMMA itself had no effect on VEGF 165 -induced in vitro tube formation and DNA fragmentation. The concentration of 10 M of L-NMMA was chosen, because 10 M L-NMMA exhibited maximum effect without cytotoxicity.
Effect of Rosiglitazone on Cytosolic Ca 2ϩ Concentrations-We studied the effects of rosiglitazone and NS-1619 on the cytosolic Ca 2ϩ concentration based on the previous observation that maxi-K channel activation resulted in the increase of intracellular Ca 2ϩ concentration (26). Rosiglitazone elevated cytosolic Ca 2ϩ concentration (45 Ϯ 3% increase),   (Fig. 9). In parallel, NS-1619 treatment also increased intracellular Ca 2ϩ concentration (results not shown).
To determine whether increased Ca 2ϩ is the upstream or downstream effect of maxi-K channel activation by rosiglitazone, we checked the effects of rosiglitazone on the cytosolic Ca 2ϩ in the maxi-K channel RNAi-transfected cells. The increased Ca 2ϩ by rosiglitazone was abolished in maxi-K channel RNAi-transfected cells, suggesting that increased Ca 2ϩ may be the consequence of the maxi-K channel activation by rosiglitazone.

DISCUSSION
PPAR␥ has been recognized as a potential therapeutic target for the treatment of pathologic neovascularization (13,25), since various PPAR␥ ligands inhibited growth and/or migration of vascular endothelial cells, smooth muscle cells, monocytes, and certain tumor cells (27,28). The present study demonstrates the novel finding that the antiangiogenic effect of rosiglitazone is mediated by PPAR␥-induced maxi-K channel opening and subsequent apoptosis in HUVECs.
Among many cell types involved in the PPAR␥-mediated antiangiogenic action, recent investigations supported the importance of endothelial cells in the inhibitory activity of PPAR␥ ligands with the identification of functionally active PPAR␥ expression in the tumor endothelium and in the immortalized endothelial cells (9,12,13). An endogenous ligand, 15-deoxy-⌬-12,14-prostaglandin J-2, and thiazolidinediones inhibited endothelial differentiation into tube-like structures and proliferation, and suppressed VEGF-induced angiogenesis (13,29). Similarly, troglitazone and rosiglitazone inhibited VEGF-induced proliferation, migration, and tube formation of bovine choroidal endothelial cells (13), and rosiglitazone directly suppressed growth of a variety of primary tumors and metastatic invasion by antiangiogenic effect (12). Our results confirmed the previous reports (13) that rosiglitazone, a PPAR␥ agonist, induced a strong concentration-dependent inhibition of neovascularization in response to VEGF 165 in vitro and in vivo. Consistent with the reports showing PPAR␥-mediated apoptosis in endothelial cells (9), we also observed that the antiproliferative effects of rosiglitazone were accompanied by DNA fragmentation with decreased Bcl-2 and increased Bax expression. In addition, the apoptotic cell death by rosiglitazone was further confirmed by acridine orange and ethidium bromide staining. Control and VEGF 165 -treated cells had a uniform green color in nuclei, but rosiglitazone-treated cells had bright orange areas of condensed chromatin in the nucleus. The effects of rosiglitazone were abolished by pretreatment with BADGE, a PPAR␥ antagonist, suggesting that rosiglitazone-induced anti-angiogenesis via apoptosis is predominantly mediated through PPAR␥ activation in HUVECs.
Maxi-K channels, widely expressed channels in various cell types, regulate membrane action potential, neurotransmitter release, and cell death in neurons (30,31), and may induce cell shrinkage and apoptosis due to increased loss of cytosolic K ϩ in unexcitable endothelial cells. Thus, we postulated that maxi-K channels might be involved in the inhibitory action of rosiglitazone in VEGF 165 -induced angiogenesis of HUVECs. As confirmed from previous reports (12,32), maxi-K channels as well as PPAR␥ were expressed in HUVECs by Western blot in our experiments (data not shown).
Several lines of our experiments supported the involvement of maxi-K channels in rosiglitazone-induced anti-angiogenesis in endothelial cells in a PPAR␥-dependent manner. First, the antiangiogenic effect by rosiglitazone was reversed by either iberiotoxin, a maxi-K channel blocker, or BADGE, a PPAR␥ antagonist. Second, NS-1619, a specific maxi-K channel opener exhibited similar antiangiogenic effects, which was reversed by iberiotoxin but not by BADGE. Third and most importantly, knockdown of maxi-K channel expression by RNAi abolished the antiangiogenic effects of rosiglitazone. On the other hand, either clotrimazole, an intermediate conductor of Ca 2ϩ -activated K ϩ channel blocker, or glibenclamide, an ATP-sensitive K ϩ channel blocker, was ineffective on the action of rosiglitazone, strongly indicating that PPAR␥ activation induced specifically the maxi-K channel opening, at least in part, resulting in the proapoptotic and antiangiogenic effects in endothelial cells. after rosiglitazone treatment in HUVECs. A, cells were exposed to BADGE or IBTX for 1 h and then to rosiglitazone for 3 h, followed by VEGF 165 (10 ng/ml) for 1 h. A, Ⅺ, eNOS-Ser 1179 phosphorylation; f, eNOS-Thr 497 phosphorylation. Results are expressed as means Ϯ S.E. of three different experiments. **, p Ͻ 0.01; ***, p Ͻ 0.001 versus VEGF 165 alone; † † †, p Ͻ 0.001 versus rosiglitazone alone. B, effects of maxi-K channel knockdown on rosiglitazone-induced NO formation. HUVECs transfected with either maxi-K channel Stealth TM RNAi oligonucleotide or negative control Stealth TM RNAi oligonucleotide were exposed to BADGE or IBTX for 1 h and then to rosiglitazone for 3 h. After incubation, VEGF 165 (10 ng/ml) was added and incubated for 3 h. Results are expressed as means Ϯ S.E. of two experiments done with quadruplicate. ***, p Ͻ 0.001 versus VEGF 165 alone; † † †, p Ͻ 0.001 versus 10 M rosiglitazone alone. C, effect of L-NMMA on rosiglitazone-induced DNA fragmentation in HUVECs. Cells were exposed to L-NMMA (10 M) for 1 h and then were stimulated by rosiglitazone (10 M) for 3 h. After incubation, VEGF 165 (10 ng/ml) was added and incubated for 2 days. Lane 1, control; lane 2, 10 ng/ml VEGF 165 ; lane 3, 10 ng/ml VEGF 165 ϩ 10 M rosiglitazone; lane 4, 10 ng/ml VEGF 165 ϩ 10 M rosiglitazone ϩ 10 M L-NMMA; lane 5, 10 ng/ml VEGF 165 ϩ 10 M L-NMMA. Experiments were carried out three times. D, effect of L-NMMA on the rosiglitazone-suppressed capillary tube formation of HUVECs. HUVECs were plated on Matrigel-coated wells at a density of 1 ϫ 10 5 cells/ well without (control) or with rosiglitazone. BADGE, IBTX, or L-NMMA was given to the cells for 30 min at 37°C before seeding in a 1.5-ml tube. Photographs were taken after 18 h in culture (ϫ100). Experiments were carried out three times.
The reversal of rosiglitazone action by maxi-K channel inhibition (i.e. iberiotoxin and RNAi) could be due to either the suppression of the rosiglitazone-induced activation of PPAR␥ or the downstream effects of PPAR␥ activation by rosiglitazone. In the present study, we observed that the effects of rosiglitazone on PPAR␥ activation were not interfered with by iberiotoxin. Furthermore, we demonstrated that rosiglitazone opened maxi-K channel, as determined by membrane hyperpolarization being reversed in the presence of either iberiotoxin or BADGE. All of these results suggest that rosiglitazone induced maxi-K channel opening via PPAR␥ activation, subsequently exhibiting antiangiogenic effects in endothelial cells.
Maxi-K channels are regulated by various intracellular second messengers, including NO (19 -21). For example, NO appears to stimulate maxi-K channel activity either via indirect cGMP generation or via direct fashion (19,20), although the regulatory role of NO in the endothelial maxi-K channel is still controversial (21,33). Recently, several studies reported that PPAR␥ activation increased endothelial NO production (22,34), possibly via regulation of eNOS phosphorylation at serine 1179 and at threonine 497. In comparison, other studies suggested that eNOS phosphorylation is not involved in the maxi-K channel-mediated NO production (26). Thus, we further elucidated the signaling pathway of rosiglitazone-induced maxi-K channel opening by examination of eNOS phosphorylation and NO production. The eNOS-Ser 1179 phosphorylation was significantly and concentration-dependently elevated by rosiglitazone, resulting in the increased NO production, whereas eNOS-Thr 497 phosphorylation was not altered. NO production by rosiglitazone was antagonized by BADGE but not antagonized by either iberiotoxin or knockdown of maxi-K channel expression. In addition, L-NMMA blocked rosiglitazone-induced hyperpolarization, implying that NO acts as a mediator of rosiglitazone for maxi-K channel opening. On the other hand, the ineffectiveness of L-NMMA on the NS-1619-induced hyperpolarization is consistent with the previous results (26), and the previously found NO production downstream of the maxi-K channel opening indicates that NO may also play a role as a downstream mediator.
A localized increase in cytoplasmic Ca 2ϩ concentration was observed after maxi-K channel opening and then induced apoptosis in vascular smooth muscle cells (35,36). Consistent with previous results (26,36), the present study showed that NS-1619 and rosiglitazone increased cytosolic Ca 2ϩ , but this increase was not observed in maxi-K channel RNAi-treated cells, indicating that the increase of intracellular Ca 2ϩ may be the downstream effect of rosiglitazone-induced maxi-K channel activation.
Taken together, the present study shows that rosiglitazone significantly inhibited VEGF 165 -induced angiogenesis in endothelial cells by a proapoptotic mechanism via PPAR␥-mediated NO production, followed by maxi-K channel opening. Contrary to the established idea that PPAR␥ ligand-mediated NO production is linked to the vascular protective effects (22), the present study suggests that NO produced by rosiglitazone acts as an important mediator of apoptotic cell death through the pathway involving maxi-K channel activation. We believe that these data represent the first evidence of maxi-K channel activity in the rosiglitazone-induced anti-angiogenesis in endothelial cells. Although in vivo antiangiogenic activity of rosiglitazone seems to be a coordinated phenomenon from the effects on many different cell types, the results of the present study would provide an important insight on the antiangiogenic mechanism of rosiglitazone in the endothelial cells. Furthermore, the novel compounds acting as maxi-K channel openers in endothelial cells would potentially provide a useful approach for antiangiogenic therapy.