Cilostazol Induces PGI2 Production via Activation of the Downstream Epac-1/Rap1 Signaling Cascade to Increase Intracellular Calcium by PLCε and to Activate p44/42 MAPK in Human Aortic Endothelial Cells

Background Cilostazol, a selective phosphodiesterase 3 (PDE3) inhibitor, is known as an anti-platelet drug and acts directly on platelets. Cilostazol has been shown to exhibit vascular protection in ischemic diseases. Although vascular endothelium-derived prostaglandin I2 (PGI2) plays an important role in vascular protection, it is unknown whether cilostazol directly stimulates PGI2 synthesis in endothelial cells. Here, we elucidate the mechanism of cilostazol-induced PGI2 stimulation in endothelial cells. Methods and Results Human aortic endothelial cells (HAECs) were stimulated with cilostazol and PGI2 accumulation in the culture media was measured. Cilostazol increased PGI2 synthesis via the arachidonic acid pathway. Cilostazol-induced intracellular calcium also promoted PGI2 synthesis via the inositol 1,4,5-trisphosphate receptor. Using RNAi, silencing of PDE3B abolished the induction effect of cilostazol on PGI2 synthesis and intracellular cAMP accumulation. Inhibition of the exchange protein, which was directly activated by cyclic AMP 1 (Epac-1) and its downstream signal the Ras-like small GTPase (Rap-1), abolished cilostazol-induced PGI2 synthesis, but this did not take place via protein kinase A (PKA). Inhibition of downstream signaling, such as mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase (PI3K) γ, and phospholipase C (PLC) ε, suppressed cilostazol-induced PGI2 synthesis. Conclusions The PDE3/Epac-1/Rap-1 signaling pathway plays an important role in cilostazol-induced PGI2 synthesis. Namely, stimulation of HAECs with cilostazol induces intracellular calcium elevation via the Rap-1/PLCε/IP3 pathway, along with MAPK activation via direct activation by Epac-1/Rap-1 and indirect activation by Epac-1/Rap-1/PI3Kγ, resulting in synergistically induced PGI2 synthesis.


Introduction
Cilostazol [6-[4-(1-cyclohexyl-1H-tetrazol-5-yl) butyloxy]-3,4-dihydroquinolin-2-(1H)-one] is a selective phosphodiesterase 3 (PDE3) inhibitor, which has been shown to prevent platelet aggregation and peripheral vasodilation [1]. The PDE3 family, known for catalyzing cyclic adenosine monophosphate (cAMP), comprises two members, PDE3A and PDE3B, which exhibit different expression patterns. PDE3A is mainly present in the heart, platelets, vascular smooth muscles, and oocytes, whereas PDE3B is mainly found in adipocytes, hepatocytes, and spermatocytes [2]. Cilostazol similarly inhibits both PDE3A and PDE3B, with IC 50 values of 0.20 and 0.38 μM, respectively [3]. Cilostazol is the only medication with a class I indication approved by the Food and Drug Administration (FDA) for intermittent claudication [4]. Recent reports have demonstrated that cilostazol also exerts pleiotropic effects [5], due to unknown mechanisms, independent of its direct effects on platelets and smooth muscle cells. Vascular protection strategies, defined as augmentation of endothelial function, have focused on and proved effective in preventing ischemic vascular events [6]. In healthy vessels, endothelial cells produce the vasoactive hormones, nitric oxide (NO), and prostacyclin (PGI 2 ) [7]. NO and PGI 2 are regarded as key mediators of vascular protection and play important roles in the modulation of vascular tone, as well as anti-inflammatory and anti-thrombotic properties [8]. The loss or attenuation of NO and PGI 2 production is an early marker of endothelial dysfunction found in many ischemic diseases [9]. Both are coreleased by agonist-stimulated endothelial cells via intracellular calcium elevation, indicating that increased intracellular calcium activates endothelial nitric oxide synthase (eNOS) for NO synthesis and phospholipase A2 (PLA 2 ) to liberate arachidonic acid for PGI 2 production [7]. Various in vivo and in vitro studies have demonstrated that cilostazol exhibits vascular protection via eNOS activation, leading to beneficial impacts on ischemic diseases, including myocardial infarction [10], stroke [11], and limb ischemia [12]. Compared with the large volume of evidence for NO-involved vascular protection by cilostazol, the association between cilostazol and PGI 2 production remains unclear. However, it is reasonable to speculate that cilostazol activates PGI 2 production, as well as NO production. Igawa et al [13]. were the first to show the involvement of PGI 2 in cilostazol-exerted anti-platelet action, and that endothelial cells potentiated the inhibitory effect of cilostazol on platelet aggregation, which was antagonized by a cyclooxygenase (COX) inhibitor. However, PGI 2 synthesis in endothelial cells was not measured in their study, thus the goal of the present study was to address this question by examining whether and how cilostazol stimulates PGI 2 production in endothelial cells.

Cellular cAMP level
HAECs were plated in 96-well culture plates at a density of 5 × 10 3 cells/well and cultured overnight. After 15 min incubation with cilostazol, cells were lysed with lysis reagent (RPN225, Amersham Biosciences, Buckinghamshire, UK) cAMP concentration was determined using a cAMP EIA kit (Amersham Biosciences,) according to the manufacturer's instructions.

siRNA (small interfering RNA) transfection
HAECs were plated in 96-well culture plates at a density of 1 × 10 3 cells/well. After overnight incubation, HAECs were transfected with the indicated siRNAs (1.2 pmol/well) with Lipofectamine RNAiMAX Reagent (Life Technologies, Inc.) according to the manufacturer's instructions. After 4 h incubation, the transfection medium was replaced with EGM-2 complete medium and knockdown was assessed at 48 h. Knockdown of target proteins were verified by western blotting.

PGI 2 level
HAECs were plated in 24-well culture plates at a density of 5 × 10 4 cells/well and cultured overnight. Culture media were replaced with 250 μL of EGM-2 containing test drugs and incubated for 1 h. Supernatants were collected and stored at -80°C until further analysis. PGI 2 was assessed as 6-keto prostaglandin F 1α (6-keto PGF 1α ) using the 6-keto PGF 1α enzyme immunoassay (EIA) kit (Cayman Chemical, Michigan, USA) according to the manufacturer's instructions. Optical density was measured at 405 nm using a microplate reader (Soft max, Molecular Devices, Sunnyvale, CA, USA). Results are expressed as 6-keto PGF 1α concentration (pg/mL).

Intracellular calcium concentration
HAECs were plated on 8-well chamber glass slides at a density of 1 × 10 4 cells/well and cultured overnight. Then, cells were loaded with 2 μM fluo-4 AM (Molecular Probes, Life Technologies), a fluorescent calcium indicator, for 15 min. Cells were pretreated with or without BAP-TA-AM (100 μM) or 2-APB for 15 min, and then stimulated with test drugs. After treatment with cilostazol, cells were stimulated with 1 mM ionomycin to obtain a maximal response. The absorption shift of fluo-4 AM upon binding of Ca 2+ was determined by scanning the excitation light at 480 nm. Fluorescent images of individual cells were analyzed every 2 s with a confocal laser scanning microscope (TCS-SP5, Leica Microsystems GmbH, Wetzlar, Germany).

Inositol 1,4,5-trisphosphate (IP3) concentration
HAECs were plated in 96-well culture plates at a density of 1 × 10 4 cells/well and cultured overnight. Culture media were replaced with 250 μL of EGM-2 containing test drugs and incubated for 1 h. Supernatants were collected and stored at -80°C until further analysis. IP3 was assessed using human inositol 1,4,5-trisphosphate, IP3 ELISA Kit (Cusabio, Wuhan, China) according to the manufacturer's instructions. Optical density was measured at 450 nm using a microplate reader (Soft max, Molecular Devices). Results are expressed as IP3 concentration (pg/mL).

Immunofluorescence histochemistry
HAECs grown on 8-well chamber glass slides at a density of 1 × 10 4 cells/well were washed with PBS on ice and fixed with 4% paraformaldehyde for 30 min. Cells were washed with PBS, permeabilized, and blocked with 0.5% blocking reagent (PerkinElmer Inc., Waltham, MA, USA) in PBS for 30 min at room temperature. Cells were incubated overnight at 4°C with primary antibodies against PDE3B (1:200 dilution) or PDE3A (1:200 dilution) in the presence of anti-VE Cadherin (1:100 dilution). The primary antibodies were detected by incubation with Alexa fluor-conjugated secondary antibodies (1:500) for 60 min at room temperature. Cells were washed with PBS and mounted with Fluorescence Mounting Medium (Fluoromount/ Plus, Dako North America, Inc., Carpinteria, CA, USA). Fluorescent images were analyzed with a confocal laser-scanning microscope (TCS-SP5) equipped with a 64× water immersion objective.

Active Rap-1 pull-down assay
Active Rap-1 was assessed using the Active Rap-1 Pull-Down and Detection Kit (Thermo Fisher Scientific Inc.) according to the manufacturer's instructions. HAECs were starved overnight in EBM-2. After starvation, HAECs were treated with 30 μM cilostazol for 5 min and then lysed. The lysates (500 μg) were incubated with GST-RalGDS-RBD and Glutathione Resin. Samples were separated on a 4-20% SDS-PAGE and transferred to nitrocellulose membranes. After blocking with 0.5% blocking reagent (PerkinElmer Inc.) in PBS, the membranes were incubated overnight at 4°C with a rabbit monoclonal anti-Rap-1 antibody, followed by incubation with a peroxidase-conjugated goat anti-rabbit IgG (H+L) (dilution, 1:1000; Thermo Fisher Scientific Inc.). The membranes were developed using Super Signal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific Inc.) followed by exposure to a CCD camera, and analyzed using Image quant LAS4000 (GE Healthcare UK Ltd.).

Biacore experiments
All experiments were performed using Biacore S51 (GE Healthcare, Uppsala, Sweden) and carried out at 25°C with 5% DMSO-HBS-EP + (GE Healthcare) used as a continuous flow buffer. Synthetic peptides of PDE3B corresponding to the domain interacting with Epac-1 or PI3Kγ at 2 mg/mL were immobilized on sensor chip SA (GE Healthcare). For immobilization, biotinlabeled Epac-1-binding PDE3B peptide or PI3Kγ peptide was injected in running buffer for 120 s at a flow rate of 10 μL/min. Final immobilization levels were between 850 and 1250 resonance units (RU). For the direct binding and competition assays, test drugs were injected in running buffer for 120 s, and then an undisturbed dissociation phase was monitored for 180 s at a flow rate of 30 μL/min. In the direct binding assay, test drags (0.3125-5 μmol/L) were injected alone. In the surface competition assays, test drags (0.625-5 μmol/L) were injected in the presence or absence of PDE3B-binding Epac-1 peptide-1 or PDE3B-binding Epac-1 peptide-2 (50 nmol/L). The rate constants of association and dissociation were calculated by BIA evaluation software (GE Healthcare UK Ltd.).

Statistical Analysis
Values are expressed as the mean ± SEM of four to five experiments. Differences were considered statistically significant at p < 0.05. All analyses were performed with the Statistical Analysis System (SAS) software (Release 9.4, SAS).

Functional expression of PDE3 isozymes in HAECs
Western blot analysis and immunofluorescence microscopy showed that both PDE3 isoforms were expressed in HAECs, however, the PDE3A expression level was much lower than PDE3B ( Fig 1A). Correspondingly, silencing either PDE3A or 3B elicited a significant increase in the basal level of intracellular cAMP level compared with the control siRNA-transfected cells, but the increased intracellular cAMP level was less in PDE3A-depleted cells than in PDE3Bdepleted cells (1.7-fold and 2.4-fold, respectively; Fig 1B). Similar to the control siRNA-transfected cells, cilostazol (30 μM) significantly increased intracellular cAMP level in the PDE3Adepleted cells (1.73-fold and 1.57-fold respectively; Fig 1B), but not in the PDE3B-depleted cells. Silencing both PDE3A/B slightly increased intracellular cAMP level compared with PDE3A-or PDE3B-depleted cells.

Cilostazol increases PGI 2 production via the arachidonic acid cascade
Corresponding to intracellular cAMP levels, PGI 2 release increased in both PDE3A-and PDE3Bdepleted cells (1.75-fold and 5-fold respectively; Fig 1C). Treatment with cilostazol (30 μM) increased PGI 2 production in the control siRNA-transfected cells by 3.6-fold and in the PDE3Adepleted cells by 2.5-fold, but not in the PDE3B-depleted cells (Fig 1C). Cilostazol-induced PGI 2 production was significantly inhibited in a dose-dependent manner by the non-selective COX inhibitor, indomethacin, and was completely abolished at the concentration of 1 mM (Fig 2A). Mitogen-activated protein kinases (MAPKs) are key mediators of agonist-induced PGI 2 production via direct phosphorylation of cPLA 2 α, resulting in the release of arachidonic acid [14]. Western blot analysis showed that cilostazol induced p42/44 MAPK phosphorylation without changing expression levels ( Fig 2B). Furthermore, extracellular signal-regulated kinase (ERK) inhibitor completely abolished cilostazol-induced PGI 2 production (10 μM; Fig 2B). Treatment with the cPLA 2 inhibitor, AACOCF3 (50 μM), completely abolished cilostazol-induced PGI 2 production ( Fig 2C). The phosphorylation of cPLA 2α on Ser-505 was also enhanced by cilostazol in a dose-dependent manner without changing its total protein level.

Cilostazol-induced PGI 2 production is dependent on IP 3 receptormediated intracellular calcium elevation
The involvement of intracellular calcium elevation in cilostazol-induced PGI 2 production was examined by chelating intracellular calcium. Fluo-4 fluorescence images of the intracellular calcium response in HAECs showed that fluorescence was immediately elevated by cilostazol stimulation (30 μM; Fig 2D). This increase was completely suppressed by BAPTA-AM (100μM). Consistently, cilostazol-induced PGI 2 production was significantly decreased in a concentration-dependent manner using BAPTA-AM (Fig 3A), and 100 μM BAPTA-AM completely abolished cilostazol-induced PGI 2 production. In addition, cilostazol-induced intracellular calcium elevation was almost completely abolished by the IP3R antagonist, 2-APB (100μM, Fig 3B). Consistent with inhibition of calcium elevation, 2-APB significantly inhibited PGI 2 production in a dose-dependent manner, with complete inhibition at 100 μM (Fig 3C). In addition, cilostazol exerts a significant increase in IP3 levels ( Fig 3D).

Discussion
The current study demonstrated the effect of cilostazol on PGI 2 production and its mechanism in endothelial cells (Fig 8). We report several novel findings: first, cilostazol increases PGI 2 synthesis in endothelial cells by activating arachidonic acid metabolism via the COX/PGI 2 pathway. Second, the Epac-1/Rap-1, but not the PKA, signaling pathway is involved in cilostazolinduced PGI 2 production. Third, the mechanism of cilostazol-induced PGI 2 production involves increased intracellular calcium by releasing calcium from calcium stores via activation of the Epac-1/Rap-1/PLCε/IP3R pathway. Fourth, the Epac-1/Rap-1 mechanism of cilostazol directly activates MAPK and indirectly activates PI3Kγ. Because cilostazol is a potent PDE3 inhibitor (the IC 50 values of PDE3A and PDE3B are 0.20 and 0.38 μM, respectively) [3] and PDE3s are expressed in HAECs, we initially predicted that intracellular cAMP accumulation is involved in cilostazol-induced PGI 2 production. Indeed, under our experimental conditions, cilostazol increased both cAMP levels and PGI 2 synthesis. Corresponding to expression levels and cAMP-catalyzing activities, PDE3B acts predominantly on PGI 2 production. Thus, it seems reasonable to speculate that intracellular cAMP elevation is involved in the mechanism of cilostazol-induced PGI 2 production in the endothelium. Downstream functions of cAMP are mediated by PKA and Epac. PKA provided a link between stimulation of adenylyl cyclase, and Epac acts as a cAMP-activated guanine nucleotide exchange factor for Rap [15]. Interestingly, pharmacological activation or inhibition of PKA showed no impact on the basal level or cilostazol-induced PGI 2 production. In contrast, pharmacological activation and/or siRNAmediated silencing of Epac-1/Rap-1 revealed that inhibition of Epac-1/Rap-1 signaling only partially suppressed cilostazol-induced PGI 2 production, as the maximal inhibitions were only 30% and 36%, respectively. Furthermore, PI3K inhibition suppressed cilostazol-induced PGI 2 production to the same extent as inhibition of Epac-1/Rap-1 did, with a maximal inhibition of 42%. The finding that HDL-induced COX-2 expression and PGI 2 production were abolished by PI3K inhibitor in ECV304 endothelial cells with a maximal inhibition of 40% [16] supports our findings showing PI3K-mediated PGI 2 production in endothelial cells. Indeed, non-selective COX inhibitor, indomethacin, with an IC 50 for COX-1 and COX-2 of 0.063 μM and 0.48 μM, respectively [17], completely abolished cilostazol-induced PGI 2 production in HAECs. In endothelial cells, COX-1 and COX-2 are constitutively expressed [18]. Thus, it seems reasonable to suggest that cilostazol promotes PGI 2 production by activating COX-1 and COX-2 in HAEC. Further, Epac-1/Rap-1/PI3K signaling plays an important role in cilostazol-induced PGI 2 production. In endothelial cells, calcium is essential for PGI 2 synthesis [19]. That is, PGI 2 synthesis is initiated by catalyzing the cleavage of arachidonic acid from membrane-bound lipids via cPLA 2 activation depending on the intracellular calcium level [7]. A recent review on the physiological action of Epac [20] described new evidence showing that Epac directly interacts with intracellular calcium release channels, such as IP3 receptors via Epac-1 peptides-1 to PDE3B. The above-mentioned compounds' competition with PDE3B-binding Epac-1 peptides-1 (5 μM) binding to Epac-1-binding PDE3B peptide was evaluated (n = 4; ** p < 0.01 vs. 5 μM PDE3B-binding Epac-1 peptides-1, randomized Dunnett's test). (B) Cilostazol, cilostamide, and milrinone interfere with association of PDE3B-binding Epac-1 peptides-2 to PDE3B. These compounds' competition with Epac-1 peptides-2 (5 μM) binding to Epac-1-binding PDE3B peptide was evaluated (n = 4; * p < 0.05, ** p < 0.01 vs. 5 μM PDE3B-binding Epac-1 peptides-2, randomized Dunnett's test). (C) Direct bindings of cilostazol, cilostamide, and milrinone to PI3Kγ-binding PDE3B peptide. Relative responses of PI3Kγ-binding PDE3B peptide to drugs at concentrations of 0.3125, 0.625, 1.25, 2.5, and 5 μM (n = 4). Rap/PLCε. In human dermal microvascular endothelial cells (HMEC-1), β2-adrenoceptor activation induces machinery that mobilizes intracellular calcium elevation via the G-protein/adenylyl cyclase/cAMP/Epac-1/IP3 pathway [21]. Correspondingly, we observed that inhibition of PLCε affected cilostazol-induced PGI 2 production similarly to Epac-1/Rap1 inhibition with a maximal inhibition of 38%. Moreover, cilostazol increased intracellular calcium levels and IP3 release. Taken together, the mechanism of cilostazol-induced PGI 2 production is mediated by intracellular calcium via Epac-1/Rap-1/PLCε/IP3R activation. In contrast, MAPK also plays an important role in cPLA 2 activation by phosphorylating Ser-505, which acts synergistically with calcium to generate arachidonic acid [22], [23]. Contrary to Epac-1/Rap-1 and other downstream signaling inhibition, ERK1/2 inhibitor decreased cilostazol-induced PGI 2 production to a basal level, suggesting that MAPK signaling pathway plays a major role in cilostazol-induced PGI 2 production, and MAPK-mediated cilostazol-induced PGI 2 production could not be explained by Epac-1/Rap-1 signaling. Recently, Wilson et al [24]. discovered the novel signaling complex, PDE3B-tethered EPAC1/p84-p110γ, which regulates Epac-1 binding to cAMP and PI3Kγ downstream signals, such as ERK and PKB, in HAECs. We previously demonstrated that cilostazol induced PKB phosphorylation, which was abolished by the wide-range PI3K inhibitor, LY294002, in HAECs [25]. The present study also demonstrated that cilostazol induced PDK, PKB, and MAPK phosphorylation in HAECs. These observations suggest that Epac-1/Rap-1 and Epac-1/PI3K signaling synergistically activate MAPK to generate PGI 2 . However, the Biacore analysis provided evidence showing that cilostazol and other PDE3 inhibitors directly binds to the Epac-1-binging domain of PDE3B and interferes with formation of the PDE3B-Epac-1 complex in a similar fashion. Furthermore, none of the PDE3 inhibitors blocked formation of the PDE3B-PI3K complex. Moreover, Epac-1 activator, 007, showed no affinity for the PDE3B binding region of Epac-1. Nevertheless, 007 strongly induced PGI2 production. Taken together, it seems that the PDE3B/Epac-1/PI3K complex plays a minor role in PGI 2 production. Additionally, intracellular cAMP elevation has no impact on PGI 2 production. Addition of other PDE inhibitors or db-cAMP did not increase PGI 2 production. Our Schematic overview of cilostazo-induced PGI 2 production in HAECs. Cilostzol-induced MAPK activation, combined with intracellular calcium elevation, results in PGI 2 production in HAECs. Calcium elevation is triggered by inositol 3,4,5 triphosphate (IP3)-regulated calcium channels (IP3R) to activate cPLA2/COX signaling. In contrast, MAPK also plays a crucial role in cPLA2 activation. Activation of Epac-1signaling modulates both processes. PDE3 is partially involved in these processes. observations are consistent with earlier reports showing no correlation between global cAMP levels and PGI 2 synthesis in endothelial cells [26]. Milrinone, another PDE3 inhibitor that is structurally unrelated to cilostazol with a similar PDE3 inhibition potency [27], slightly decreased intracellular calcium levels and PGI 2 . In contrast, cilostamide slightly increased intracellular calcium levels and PGI 2 . We already mentioned that intracellular calcium elevation is essential for PGI 2 synthesis, and differences in PGI 2 synthesis by cAMP-elevating agents are likely due to differences in regulation of intracellular calcium elevation. These findings support our hypothesis that crosstalk between multiple signaling pathways initiates intracellular calcium elevation and MAPK activation via cilostazol. Hence, intracellular cAMP accumulation seems necessary, though it has a minor function in cilostazol-induced PGI 2 synthesis, and it appears that mechanisms other than cAMP accumulation also contribute to cilostazol-induced PGI 2 synthesis. Because cilostazol is also an adenosine uptake inhibitor, this indicates the possible existence of crosstalk between the Epac-1/Rap-1 pathway and the signaling cascade from adenosine receptors to extracellular adenosine elevation, to produce PGI 2 . Recent evidence supports our hypothesis that adenosine-mediated signaling is involved in prostaglandin synthesis in the endothelium and thatactivation of adenosine A1 receptor increases PGI 2 synthesis in the rat aorta [28] and rat aortic endothelial cells [28], [29]. Furthermore, milrinone is also known as an adenosine A1 receptor antagonist [30]. Therefore, it would be reasonable to speculate that milrinone inhibits the adenosine A1 receptor, thereby decreasing PGI 2 production. The involvement of adenosine receptor activation in cilostazol-induced PGI 2 production is now under consideration. However in this study, we concluded that HAEC stimulation with cilostazol induces increased intracellular calcium by activating calcium release from intracellular calcium stores via IP3 receptor activation, along with Epac-1/Rap-1/PLCε and Epac-1/Rap-1/MAPK activation, resulting in a synergistic increase in PGI 2 production. These results provide new evidence showing that the PDE3B/ Epac-1 signaling pathway mediates cilostazol-induced PGI 2 release from HAECs via an increase in intracellular calcium.