Cyclic AMP Potentiates Vascular Endothelial Cadherin-Mediated Cell-Cell Contact To Enhance Endothelial Barrier Function through an Epac-Rap1 Signaling Pathway

Abstract

Cyclic AMP (cAMP) is a well-known intracellular signaling molecule improving barrier function in vascular endothelial cells. Here, we delineate a novel cAMP-triggered signal that regulates the barrier function. We found that cAMP-elevating reagents, prostacyclin and forskolin, decreased cell permeability and enhanced vascular endothelial (VE) cadherin-dependent cell adhesion. Although the decreased permeability and the increased VE-cadherin-mediated adhesion by prostacyclin and forskolin were insensitive to a specific inhibitor for cAMP-dependent protein kinase, these effects were mimicked by 8-(4-chlorophenylthio)-2′-O-methyladenosine-3′, 5′-cyclic monophosphate, a specific activator for Epac, which is a novel cAMP-dependent guanine nucleotide exchange factor for Rap1. Thus, we investigated the effect of Rap1 on permeability and the VE-cadherin-mediated cell adhesion by expressing either constitutive active Rap1 or Rap1GAPII. Activation of Rap1 resulted in a decrease in permeability and enhancement of VE-cadherin-dependent cell adhesion, whereas inactivation of Rap1 had the counter effect. Furthermore, prostacyclin and forskolin induced cortical actin rearrangement in a Rap1-dependent manner. In conclusion, cAMP-Epac-Rap1 signaling promotes decreased cell permeability by enhancing VE-cadherin-mediated adhesion lined by the rearranged cortical actin.

Endothelial cells lining blood vessels regulate endothelial barrier function, which restricts the passage of plasma proteins and circulating cells across the endothelial cells. Endothelial barrier dysfunction results in an increase in vascular permeability, thereby causing edema or inflammatory or metastatic cell infiltration. Inflammatory mediators such as thrombin and histamine induce intercellular gap formation, leading to an increase in endothelial permeability (1, 4). In contrast, angiopoietin 1 and sphingosine-1-phosphate (S1P) stabilize endothelial barrier integrity (17, 18). In addition, cyclic AMP (cAMP), a second messenger downstream of Gs-coupled receptor, improves endothelial cell barrier function (32, 39, 43). Consistently, cAMP-elevating G protein-coupled receptor (GPCR) agonists, adrenomedullin (AM), prostacyclin (PGI2), prostaglandin E2 (PGE2), and β-adrenergic agonists reduce endothelial hyperpermeability induced by inflammatory stimuli (15, 19, 25).

The endothelial cell barrier is structurally organized by adherens junctions (AJ) and tight junctions. Vascular endothelial (VE) cells express both VE-cadherin (also known as cadherin-5 and CD144) and neural (N)-cadherin (9, 33). VE-cadherin constitutes AJ, whereas N-cadherin formed the cell-cell contacts between endothelial cells and endothelial cell-supporting pericytes. VE-cadherin mediates calcium-dependent, homophilic intercellular adhesion. Its short cytoplasmic tail binds to three armadillo family proteins, β-, γ-, and p120-catenins. β- and γ-catenins associated with α-catenin link the VE-cadherin complex to the actin cytoskeleton and, therefore, strengthen the AJ adhesiveness (9).

Endothelial AJ are dynamic structures, and their adhesive property is finely regulated by several different mechanisms. Tyrosine phosphorylation of VE-cadherin, β-catenin, and p120-catenin correlates with weakened endothelial cell-cell adhesion. VE growth factors and inflammatory mediators such as histamine and thrombin induce tyrosine phosphorylation of AJ components, resulting in the weakened cell-cell contacts and increased endothelial cell permeability (1, 14, 40). In clear contrast, angiopoietin 1, which stabilizes cell-cell contacts, induces dephosphorylation of endothelial cell adhesion molecules, VE-cadherin, and platelet endothelial cell adhesion molecule 1 (17). It has been also reported that S1P induces AJ formation and enhances barrier function through a Rac-dependent cortical actin rearrangement (18). cAMP-dependent protein kinase A (PKA) is suggested to be crucial for cAMP-triggered stabilization of cell-cell contacts and for barrier integrity of endothelial cells (43). However, it has not been clear whether PKA-independent signaling is involved in the regulation of endothelial barrier function.

Rap1, belonging to Ras family GTPase, is involved in the formation and stabilization of AJ in Drosophila melanogaster (23). Rap1 becomes the GTP-bound active form by guanine nucleotide exchange factor (GEF) and the GDP-bound inactive form by GTPase-activating proteins (GAP), respectively. GEFs for Rap1 include C3G, CalDAG-GEFs, Epacs, and DOCK4 (reviewed in reference 6). DOCK4, which is disrupted in various types of human cancers, regulates the formation of AJ (41). Very recent reports also revealed that Rap1 activity is required for the formation of E-cadherin-based cell-cell contacts (20, 36). These findings prompted us to investigate how Rap1 is activated to stabilize cell-cell contacts and to examine the physiological consequence of stabilized cell-cell contacts by Rap1.

In the present study, we investigated the mechanism by which cAMP-elevating GPCR agonists potentiate endothelial barrier function and restrict cell permeability. We found that increased cAMP triggers Epac-Rap1 signaling to reduce permeability independently of PKA by augmentation of VE-cadherin-mediated cell-cell adhesion.

MATERIALS AND METHODS

Reagents and antibodies.

Human recombinant AM was kindly provided by Shionogi & Co. Ltd (31). Materials were purchased as follows: isoproterenol (Iso), PGE2, PGI2, thrombin, forskolin (FSK), and 3-isobutyl-1-methylxanthine (IBMX) from Wako Pure Chemical Industries; dibutyryl-cAMP (dbcAMP) from Sigma-Aldrich; H89 from Seikagaku Corporation; 8-(4-chlorophenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate (8-CPT-2′-O-Me-cAMP) from Tocris; fluorescein isothiocyanate (FITC)-labeled dextran (molecular weight, 42,000) and purified human immunoglobulin G (IgG) Fc protein from ICN Biologicals; vascular endothelial growth factor (VEGF) from R & D Systems. Anti-Rap1GAPII antibody was developed by immunization of glutathione S-transferase (GST)-tagged Rap1GAPII (amino acids 411 to 694 of Rap1GAPII). Other antibodies used here were purchased as follows: anti-VE-cadherin from Chemicon International and Transduction Laboratories; anti-β-catenin from Transduction Laboratories; anti-CREB and anti-phospho-CREB (Ser133) from Cell Signaling Technology; anti-Rap1 from Santa Cruz Biotechnology; anti-cortactin from Upstate Biotechnology, Inc.; rhodamine-phalloidin and Alexa 488-labeled goat anti-mouse IgG from Molecular Probes; horseradish peroxidase-coupled goat anti-mouse and goat anti-rabbit IgG from Amersham Biosciences.

Cell culture and transfection.

Human umbilical vein endothelial cells (HUVECs) and human arterial endothelial cells (HAECs) were purchased from Kurabo (Kurashiki, Japan). The cells were maintained in HuMedia-EG2 with a growth additive set as described previously (12) and used for experiments before passages 7 and 10, respectively. HEK293, 293T, and HeLa cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Nissui, Tokyo, Japan) supplemented with 10% fetal bovine serum and antibiotics (100 μg of streptomycin/ml and 100 U of penicillin/ml). HUVECs and 293T cells were transfected by using Lipofectamine Plus reagent (Invitrogen) and by the calcium-phosphate precipitation technique, respectively.

Plasmids and adenovirus.

pcDNA-VE-cad-Ect-Fc-His is a modified vector of pcDNA3.1-Fc-PECAM-1 (a kind gift from W. A. Muller, Cornell University) for producing the secreted form of the extracellular domain of VE-cadherin fused with Fc followed by a six-His tag. A DNA fragment encoding human Epac lacking the cAMP binding domain (amino acids 324 to 881) was amplified by PCR with pMT2SM-HA-Epac (a kind gift from J. L. Bos, Utrecht University, Utrecht, The Netherlands) as a template and ligated into the pCXN2 vector (12). pCXN2-FLAG-Rap1V12-IRES-EGFP expressed both FLAG-tagged Rap1V12 and internal ribosomal entry site (IRES)-driven enhanced green fluorescent protein (EGFP), and pCXN2-Rap1GAPII-IRES-EGFP expressed both FLAG-tagged Rap1GAPII and IRES-driven EGFP. pGL3 control vector was purchased from Promega Corp. Recombinant adenoviruses encoding Rap1GAPII (Ad-RapGAP) and LacZ (Ad-LacZ) were obtained from S. Hattori (The Institute of Medical Science, University of Tokyo) and M. Matsuda (Research Institute for Microbial Disease, Osaka University, Osaka, Japan), respectively. Adenoviruses expressing FLAG-tagged Rap1V12 and IRES-driven EGFP (Ad-Flag-Rap1V12-IRES-EGFP) were produced by using the Adeno-X system according to the manufacturer's protocol (Clontech). Endothelial cells were infected with adenoviruses at the appropriate multiplicities of infection (MOI) as described in the figure legends.

Permeability assay.

Permeability across the endothelial cell monolayer was measured by using type I collagen-coated transwell units (6.5-mm diameter, 3.0-μm-pore-size polycarbonate filter; Corning Costar Corporation). HUVECs plated at 10⁵ cells in each well were cultured for 3 to 4 days before experiments. After serum starvation in medium 199 containing 1% bovine serum albumin (BSA) for 1 h, the cells were treated with the agonists or drugs, as indicated in the figure legends, for 30 min. Permeability was measured by adding 1 mg of FITC-labeled dextran (molecular weight, 42,000)/ml together with or without 2 U of thrombin/ml to the upper chamber. After incubation for 30 min, 50 μl of sample from the lower compartment was diluted with 300 μl of phosphate-buffered saline (PBS) and measured for fluorescence at 520 nm when excited at 492 nm with a spectrophotometer F-4500 (Hitachi). HUVECs infected with adenovirus for 24 h after becoming confluent and kept for another 24 h in replaced medium were subjected to a cell permeability assay.

Immunocytochemistry.

Monolayer-cultured HUVECs grown on a 35-mm-diameter glass base dish (Asahi Techno Glass) were starved in medium 199 containing 0.5% BSA for 3 h and subsequently incubated with the stimulants indicated in the figure legends for 30 min. After stimulation, the cells were fixed in PBS containing 2% formaldehyde for 30 min at 4°C, washed with PBS, and permeabilized with 0.05% Triton X-100 for 30 min at 4°C. Cells were blocked with PBS containing 4% BSA for 1 h at room temperature (RT) and stained with rhodamine-phalloidin for 20 min, anti-VE-cadherin for 60 min, and anticortactin for 60 min at RT. Protein reacting with antibody was visualized with Alexa 488-labeled goat anti-mouse IgG. Images were recorded with a confocal microscope (BX50WI, Fluoview; Olympus) with a water immersion objective lens (LUMPlanF1 100X1.00W).

VE-cadherin translocation assay and Western blot analysis.

HUVECs plated in six-well plates were serum starved in medium 199 containing 1% BSA overnight. The cells were stimulated with PGI2 and FSK for the indicated time and fractionated with cytoskeleton-stabilizing buffer (10 mM HEPES [pH 7.4], 250 mM sucrose, 150 mM KCl, 1 mM EGTA, 3 mM MgCl₂, 1× protease inhibitor cocktail [Roche Diagnostics], 1 mM Na₃VO₄, 0.5% Triton X-100) by centrifugation at 15,000 × g for 15 min. The Triton X-100-insoluble fraction was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) followed by transfer to Immobilon-P (Amersham Biosciences) and immunoblotting with the indicated antibodies. Immunocomplexes were visualized by enhanced chemiluminescence detection (Amersham Biosciences) with species-matched peroxidase-conjugated secondary antibodies.

Purification of recombinant VE-cadherin ectodomain-Fc chimeric protein.

293T cells transfected with pcDNA-VE-cad-Ect-Fc-His were cultured in DMEM supplemented with 10% fetal calf serum for 24 h and subsequently kept in replaced medium (DMEM-F21 containing 1% fetal calf serum) for 7 days. VE-cadherin-Fc (VEC-Fc) protein secreted into the medium was collected every 2 days and centrifuged to remove floating cells and debris. VEC-Fc was collected on ProBond resin (Invitrogen) by gentle agitation overnight at 4°C. VEC-Fc protein bound to the beads was eluted with 500 mM imidazole, concentrated with Amicon Centriplus 30 (Millipore), and buffer exchanged into PBS containing 2 mM CaCl₂ and 2 mM MgCl₂ (PBS-Ca/Mg) by dialysis.

Cell adhesion assay.

Twenty-four-well tissue culture plates were coated with 10 μg of VEC-Fc or Fc protein/ml in PBS-Ca/Mg at 4°C overnight. After washing with PBS-Ca/Mg, the plates were blocked with 1% heat-inactivated BSA in PBS (heat inactivated at 85°C for 12 min) for 1 h at RT. To examine cell adhesion to the VEC-Fc- or Fc-coated dish, cells were suspended in 0.5% BSA-containing medium 199 and incubated for 30 min at 37°C. Cells (1.5 × 10⁵) were plated on each VEC-Fc- or Fc-coated well in the presence or absence of agonists, drugs, and 5 mM EGTA and adhered to the dish at 37°C for the indicated time. To analyze cell adhesion to a collagen-covered surface, cells were plated onto a collagen-coated six-well plate (Iwaki) and adhered to the dish in the presence or absence of 5 mM EGTA. After washing with PBS-Ca/Mg four times to remove nonadherent cells, adherent cells and input cells were quantified by measuring endogenous alkaline phosphatase activity as described elsewhere (35). Briefly, the cells were lysed in a buffer containing 100 mM Tris-citrate (pH 6.5) and 0.25% Triton X-100, and alkaline phosphate activity in the lysate was measured by using the AttoPhos AP fluorescent substrate system (Promega Corp.). To examine the effects of Rap1V12, EpacΔcAMP, and Rap1GAPII, HUVECs were transfected with plasmids encoding either Rap1V12, EpacΔcAMP, or Rap1GAPII together with the luciferase reporter construct (pGL3 control vector). The adhesion of cells expressing Rap1V12, EpacΔcAMP, or Rap1GAPII to the VEC-Fc-coated dish was normalized by measuring the luciferase activity of the cells and input cells (16).

Detection of GTP-bound form of Rap1.

Rap1 activity was assessed by a modified Bos's method as described previously (34). Briefly, HUVECs starved in medium 199 containing 1% BSA overnight were stimulated with the indicated agonists and drugs and lysed at 4°C in a pull-down lysis buffer (20 mM Tris-HCl [pH 7.5], 100 mM NaCl, 10 mM MgCl₂, 1% Triton X-100, 1 mM EGTA, 1 mM dithiothreitol, 1 mM Na₃VO₄, 1× protease inhibitor cocktail). GTP-bound Rap1 was collected on the GST-Rap1 binding domain of RalGDS precoupled to glutathione-Sepharose beads and subjected to SDS-PAGE followed by immunoblotting with anti-Rap1.

In vivo permeability assay.

In vivo permeability was quantified by a modified Miles assay as described previously (29). In brief, ICR mice (Japan SLC, Inc.) shaved 3 days before experiments were lightly anesthetized and intravenously injected with 150 μl of 1% Evans blue dye solution (in saline) passed through a 0.22-μm-pore-size filter. Fifteen minutes later, 20 μl of PBS, VEGF (50 μg/ml), and/or 8-CPT-2′-O-Me-cAMP (1 mM) were applied by intradermal injections with a 30-gauge needle. The sites of intradermal injection were photographed 60 min after the injection, carefully dissected, and weighed. To quantify the vascular permeability, extravasated blue dye was eluted from the dissected skin with formamide at 56°C, and optical density was measured by spectrophotometry at 620 nm.

RESULTS

cAMP enhances the barrier property of monolayer-cultured endothelial cell.

To evaluate the barrier function, we examined the permeability of FITC-labeled dextran across monolayer HUVECs. Expectedly, AM, Iso, PGE2, and PGI2 reduced basal endothelial permeability in HUVECs (Fig. 1A). PGI2 also reduced thrombin-induced vascular permeability (Fig. 1B). Other cAMP-elevating bio-ligands similarly reduced thrombin-induced permeability (data not shown). The bio-ligands for cAMP-elevating GPCR that we used in this study indeed increased cAMP in HUVECs (data not shown). Furthermore, IBMX (an inhibitor for phosphodiesterase), dbcAMP (a membrane-permeable cAMP analogue), and FSK (an adenylyl cyclase activator) resulted in a reduction of both basal and thrombin-induced endothelial permeability (Fig. 1; data not shown).

FIG. 1.

cAMP enhances barrier function of monolayer VE cells. (A) Vascular permeability, reflecting barrier function, was analyzed by measuring the fluorescence of FITC-labeled dextran across the monolayer-cultured HUVECs as described in Materials and Methods. HUVECs grown on transwell filters were incubated with control (Cont), 0.1 μM AM, 200 μM Iso, 200-ng/ml PGE2, 10-μg/ml PGI2, 1 mM IBMX, 1 mM dbcAMP, and 10 μM FSK for 30 min. Average permeability ± standard deviation is expressed as a percentage compared to the control. (B) The effects of PGI2 and FSK on vascular permeability were quantified in the presence (+) or absence (−) (Vehicle) of 2 U of thrombin (Thr)/ml. Average permeability ± standard deviation is expressed as the increase relative to that observed in unstimulated HUVECs in the vehicle. Data shown are the results from at least three independent experiments. Significant differences from the control (A) or between two groups (B) determined by Student's t test are indicated by a single asterisk (P < 0.05) or double asterisks (P < 0.01).

cAMP potentiates formation of AJ.

Endothelial barrier function is largely dependent upon endothelial cell junctions. To investigate how cAMP affects AJ formation, we examined AJ organization by immunostaining with anti-VE-cadherin before and after stimulation. When subconfluent HUVECs with intercellular gaps were stimulated with PGI2 or FSK, the cells extended the plasma membrane and established cell-cell contacts with neighboring cells (Fig. 2A). Similar results were obtained with AM and PGE2 (data not shown). Stimulation of HUVECs with PGI2 and FSK dramatically enhanced accumulation of VE-cadherin at cell-cell contacts (Fig. 2B).

FIG. 2.

cAMP induces AJ formation. (A) HUVECs cultured on a glass base dish were stimulated with 10 μg of PGI2/ml (upper panels) or with 10 μM FSK (lower panels) for 20 min and shown as phase-contrast images. Left and right panels show the cells before and after stimulation, respectively. The arrows indicate the sites of cell-cell contacts induced by PGI2 and FSK. The area boxed by the white broken line is enlarged in the right top of the panels. Bars, 50 μm. (B) Subconfluent HUVECs stimulated with vehicle (Cont), 10-μg/ml PGI2, and 10 μM FSK for 45 min were fixed, stained with anti-VE-cadherin antibody, and visualized with Alexa 488-conjugated secondary antibody through a confocal microscope (BX50WI; Olympus). Note that VE-cadherin (green) was accumulated at the cell-cell contact upon PGI2 and FSK stimulation. Bars, 50 μm. (C) Translocation of VE-cadherin was assessed by Triton X-100 solubility. HUVECs were stimulated with vehicle (top), 10-μg/ml PGI2 (middle), and 10 μM FSK (bottom) for the time indicated at the top and fractionated with cytoskeleton-stabilizing buffer as described in Materials and Methods. The Triton X-100-insoluble fraction was subjected to SDS-PAGE followed by Western blot analysis (WB) with anti-VE-cadherin.

The maturation of AJ requires homophilic binding of intercellular VE-cadherins and tight anchoring to the actin cytoskeleton via the cytoplasmic region through catenins. VE-cadherin anchored to the actin cytoskeleton is detected in detergent-insoluble fractions of cell lysates (26). We found an increase in VE-cadherin in the Triton X-100-insoluble fraction after stimulation with PGI2 or FSK (Fig. 2C). These results suggest that cAMP-elevating GPCR agonists potentiate AJ formation, which results in a cAMP-induced decrease in permeability.

cAMP promotes VE-cadherin-dependent endothelial cell adhesion.

VE-cadherin is required for AJ formation (9). To test the involvement of a homophilic interaction of VE-cadherin in cAMP-enhanced AJ formation, we directly examined VE-cadherin-mediated cell adhesion. To mimic the VE-cadherin-dependent cell adhesion, we used VEC-Fc chimeric protein, which consisted of the extracellular domain of VE-cadherin fused to the Fc portion of immunoglobulin. HUVECs were plated onto VEC-Fc-coated dishes and time-lapse imaged. Cells attached within 5 min to the VEC-Fc-coated dish, subsequently spread, and exhibited a typical fried-egg morphology characterized by a large circular lamellipodium (Fig. 3A). No cells attached to the Fc-coated dish (Fig. 3B and C). Since cadherin-dependent cell adhesion requires Ca²⁺, we examined the effect of Ca²⁺ chelation on cell adhesion to VEC-Fc-coated dishes. Cell adhesion to VEC-Fc-coated dishes was completely abolished by chelating extracellular Ca²⁺, although cell attachment to the collagen-coated dish was unaffected (Fig. 3C and D). Basal and FSK-augmented cell adhesion to VEC-Fc-coated dishes was inhibited by EGTA (Fig. 3C). Both HUVECs and HAECs expressing VE-cadherin adhered to the VEC-Fc-coated dish (Fig. 3E). In clear contrast, HeLa and HEK293 cells, which express N-cadherin, but not VE-cadherin (20, 42), did not adhere to the VEC-Fc-coated dish, although these cells could attach to the collagen-coated dish (Fig. 3E; data not shown). Collectively, these results indicate that endothelial cell adhesion to the VEC-Fc-coated dish depends upon the homophilic ligation of VE-cadherin.

FIG. 3.

Endothelial cells adhere to a VEC-Fc-coated dish through homophilic ligation of VE-cadherin. (A) HUVECs were plated onto the VEC-Fc-coated dish and time-lapse imaged at the time points (in minutes) indicated on the panels. Bar, 20 μM. (B) HUVECs were plated on the Fc-coated dish (top panel) or the VEC-Fc-coated dish (bottom panel) for 1 h and phase-contrast imaged after removal of nonadherent cells by washing with PBS-Ca/Mg. (C) HUVECs were plated onto either an Fc- or VEC-Fc-coated dish in the absence (−) or presence (+) of 5 mM EGTA and 10 μM FSK for 7 min. Cell adhesion was quantified as described in Materials and Methods. (D) Adhesion of HUVECs to a collagen-coated dish in the presence or absence of 5 mM EGTA was analyzed by a method similar to that described for panel C. (E) Adhesion of HUVECs, HAECs, and HeLa and HEK293 cells to the VEC-Fc-coated dish was examined as described in the legend for panel C. Cells adhering to the dishes of total input cells (percentage) is expressed as the mean ± standard deviation by measuring alkaline phosphatase activity of adherent cells divided by that of total input cells. Representative results from three independent experiments were shown in all panels.

We proceeded to investigate the effect of cAMP-elevating GPCR agonists on VE-cadherin-mediated cell adhesion. The adhesion of HUVECs plated in the presence of PGI2 or FSK was evaluated by the alkaline phosphatase activity of remaining cells after washing. PGI2 enhanced adhesion of HUVECs to the VEC-Fc-coated dish in a concentration-dependent manner (Fig. 4A) and in a time-dependent manner (Fig. 4B). In a time course analysis, we noticed that enhanced adhesion was observed 7 min after the plating (Fig. 4B). Other cAMP-elevating GPCR agonists, including AM, Iso, and PGE2, potentiated VE-cadherin-dependent cell adhesion (Fig. 4C). In addition, similarly enhanced cell adhesion to the VEC-Fc-coated dish was also observed in the cells treated with cAMP-elevating drugs such as IBMX, dbcAMP, and FSK (Fig. 4F). Like PGI2, the effect of FSK on cell adhesion to the VEC-Fc-coated dish was concentration dependent and time dependent (Fig. 4D and E). This cAMP-induced cell adhesion to the VEC-Fc-coated dish depends on the enhanced homophilic ligation of VE-cadherin because FSK did not augment endothelial adhesion to the Fc-coated dish or attachment to the VEC-Fc-coated dish in the absence of extracellular Ca²⁺ (Fig. 3C). These results indicate that cAMP potentiates VE-cadherin-dependent cell adhesion.

FIG. 4.

cAMP potentiates VE-cadherin-dependent cell adhesion. (A) HUVECs were plated onto a VEC-Fc-coated dish in the presence of PGI2 at the concentrations indicated at the bottom for 7 min. Cell adhesion was quantified as described in Materials and Methods. Mean adhesion activity ± standard deviation is expressed as the increase compared with that observed in unstimulated cells. (B) HUVECs were plated onto the VEC-Fc-coated dish in the absence (circle) or presence (square) of 10-μg/ml PGI2 for the time indicated at the bottom. The percent adhesion was calculated by measuring the alkaline phosphatase activity of adherent cells divided by that of total input cells. (C) HUVECs stimulated with cAMP-elevating ligands similar to that described in the legend to panel A were assessed for adhesion activity. The concentration of stimulants was the same as described in the legend to Fig. 1A. (D) The effect of FSK on cell adhesion was analyzed by a method similar to that described for panel A, except that cells were preincubated for 10 min before plating. (E) The effect of 10 μM FSK on time-dependent adhesion was analyzed as described in the legend to panel B, except that cells were preincubated for 10 min before plating. (F) HUVECs stimulated with the reagent indicated at the same concentration used as described in the legend to Fig. 1A were analyzed for cell adhesion by a method similar to that described for panel D. Data are expressed as means ± standard deviations of the results from three independent experiments in panels A, C, D, and F. Representative results from three independent experiments wereshown in panels B and E. A significant difference from the control determined by Student's t test is indicated with a single asterisk (P < 0.05) or double asterisks (P < 0.01).

cAMP augments endothelial barrier function in a PKA-independent manner.

PKA is suggested to be involved in cAMP-enhanced endothelial barrier function (43). Thus, we investigated the involvement of PKA in the regulation of endothelial barrier integrity by PGI2 and FSK. Unexpectedly, PGI2- and FSK-induced reduction of endothelial permeability was insensitive to a specific PKA inhibitor, H89 (7) (Fig. 5A and B). The reduction of thrombin- increased permeability by FSK was also unaffected by H89 (Fig. 5C). Consistently, H89 did not affect VE-cadherin-mediated cell adhesion enhancement by PGI2 and FSK (Fig. 5D and E). To confirm that H89 worked in HUVECs, we examined FSK-induced phosphorylation of CREB, a direct PKA substrate (38). Phosphorylation of CREB upon FSK stimulation was significantly inhibited by H89, indicating the effectiveness of this inhibitor in HUVECs (Fig. 5F). Therefore, these results apparently suggest a novel PKA-independent signaling pathway involved in cAMP-induced endothelial barrier function.

FIG. 5.

cAMP-enhanced VE-cadherin-dependent cell adhesion and endothelial barrier function does not depend upon PKA. (A) Permeability across monolayer HUVECs grown on transwell filters were assessed by measuring FITC-labeled dextran as described in the legend to Fig. 1A. The effect of 10-μg/ml PGI2 on cell permeability without pretreatment (Vehicle) or with pretreatment with 5 μM H89, a specific PKA inhibitor, for 10 min is indicated as the percent permeability compared to that observed in untreated cells. +, present; −, absent. (B) The effect of 10 μM FSK on cell permeability without pretreatment (Vehicle) and with pretreatment with H89 was assessed similar to that described for panel A. (C) The effect of pretreatment of HUVECs with 5 μM H89 on FSK-induced reduction of 2-U/ml thrombin-induced permeability was analyzed. Permeability indicates the increase relative to that observed in untreated cells. (D) HUVECs untreated or pretreated with H89 for 10 min prior to stimulation with 10-μg/ml PGI2 were analyzed for cell adhesion as described in the legend to Fig. 4A. (E) HUVECs untreated or pretreated with H89 for 10 min prior to stimulation with 10 μM FSK were analyzed for cell adhesion as described in the legend to Fig. 4D. For panels A to E, data are expressed as means ± standard deviations of the results from triplicate samples. Similar results were obtained in at least three independent experiments. Significant differences between two groups determinedby Student's t test are indicated by a single asterisk (P < 0.05) or double asterisks (P < 0.01). (F) HUVECs serum starved in 1% BSA-containing medium 199 for 6 h, followed by pretreatment with (+) or without (−) 5 μM H89 for 10 min, were stimulated with vehicle and 10 μM FSK for 10 min. Phosphorylation of CREB was assessed by Western blot analysis with anti-CREB (CREB) and anti-phospho-CREB-specific (pCREB) antibodies.

cAMP induces Rap1 activation.

Besides PKA, Epac (cAMP-GEF) was identified as a novel cAMP target and a Rap1-specific GEF (5, 21). We therefore hypothesized that cAMP-activated Epac-Rap1 signaling is involved in the enhancement of VE-cadherin-dependent cell adhesion and endothelial barrier function. To address this possibility, we tested whether cAMP-elevating GPCR agonists induce Rap1 activation in HUVECs. Rap1 activity was determined by a pull-down assay by using a GST fusion protein of Rap1-binding domain of RalGDS according to the Bos's method. Bio-ligands for cAMP-elevating GPCR activated Rap1 (Fig. 6A). PGI2 rapidly induced Rap1 activation, which peaked at 1 to 5 min after the stimulation and then declined to the basal level by 10 min (Fig. 6C). A second wave of Rap1 activation was also observed 15 to 45 min after the stimulation (Fig. 6C). PGI2-induced Rap1 activation occurred in a concentration-dependent manner (Fig. 6B), which was associated with enhancement of VE-cadherin-dependent cell adhesion (Fig. 4A). Similarly, dbcAMP, FSK, and IBMX activated Rap1 (Fig. 6D). FSK-induced Rap1 activation reached a maximal level 2 to 5 min after the stimulation, and the level was sustained for up to 15 to 30 min (Fig. 6E). Collectively, these findings indicate that cAMP induces Rap1 activation in endothelial cells.

FIG. 6.

cAMP induces Rap1 activation. (A) Serum-starved HUVECs kept in medium 199 containing 1% BSA overnight were stimulated with cAMP-elevating agonists for 2.5 min as indicated at the top and at the concentrations described in the legend to Fig. 1A. GTP-bound Rap1 was detected by pull-down assay as described in Materials and Methods. Activation indicates the ratio of the poststimulation GTP-Rap1 intensity of total Rap1 intensity to the prestimulation GTP-Rap1 intensity of total Rap1 intensity. (B) Rap1 activation was analyzed by detecting GTP-bound Rap1 with lysates from HUVECs stimulated with PGI2 for 2.5 min at the different concentrations indicated at the top. (C) Rap1 activation was analyzed by detecting GTP-bound Rap1 with lysates from cells stimulated with 10-μg/ml PGI2 for the time period indicated at the top. (D) Serum-starved HUVECs similar to those described in the legend to panel A were stimulated with the reagents indicated at the top for 10 min at the same concentrations described in the legend to Fig. 1A. Rap1 activation was assessed by a method similar to that described for panel A. (E) The effect of 10 μM FSK on time-dependent Rap1 activity was examined as described for panel C. Representative results from at least three independent experiments are shown for all panels.

Specific activation of Epac reduces endothelial permeability and enhances VE-cadherin-dependent cell adhesion.

To test whether the activation of endogenous Epac is sufficient to reduce endothelial permeability and to induce VE-cadherin-dependent cell adhesion, we used a recently developed cAMP analog, 8-CPT-2′-O-Me-cAMP, which specifically activates Epac without affecting PKA activity (13). As expected, 8-CPT-2′-O-Me-cAMP induced Rap1 activation in HUVECs (Fig. 7A), indicating that Epac is expressed in endothelial cells. 8-CPT-2′-O-Me-cAMP dramatically reduced basal endothelial permeability, as did FSK and dbcAMP (Fig. 7B). Thrombin-induced permeability was also inhibited by 8-CPT-2′-O-Me-cAMP (Fig. 7C). Furthermore, we examined the effect of 8-CPT-2′-Me-cAMP on in vivo vascular permeability. VEGF-induced vascular permeability was completely blocked by coinjection of 8-CPT-2′-O-Me-cAMP (Fig. 7D). In addition, adhesion of HUVECs to the VEC-Fc-coated dish was significantly enhanced by 8-CPT-2′-O-Me-cAMP (Fig. 7E). Hence, Epac activation is sufficient to enhance VE-cadherin-dependent cell adhesion and to augment endothelial barrier function in vitro and in vivo.

FIG. 7.

Activation of Epac is sufficient to enhance VE-cadherin-dependent cell adhesion and endothelial barrier function. (A) Serum-starved HUVECs in medium 199 containing 1% BSA were stimulated with 0.2 mM 8-CPT-2′-O-Me-cAMP (8CPT) for the indicated time. Rap1 activity was determined as described in the legend to Fig. 6A. The result is a representative from three independent experiments. (B) Permeability of cells treated with the reagents as indicated on the bottom for 30 min was analyzed as described in the legend to Fig. 1A. (C) The effect of 0.2 mM 8CPT-2′-O-Me-cAMP on 2-U/ml thrombin-induced permeability was analyzed as described in the legend to Fig. 1B. (D) Effect of 8CPT-2′-O-Me-cAMP on VEGF-induced permeability was assessed by intradermal Miles assay as described in Materials and Methods. Amounts of extravasation of Evan blue in mouse dermal skin were measured 60 min after intradermal injection of vehicle and VEGF together with (+) or without (−) 8CPT. Mean leakage ± standard deviation of the results from 6 mice per group is expressed as nanograms of weight of extravasated Evans blue per milligram of weight of dermal skin. A photograph on the bottom shows leakage of Evans blue in dermal skin. (E) HUVEC adhesion to the VEC-Fc-coated dish in the presence of 0.2 mM 8CPT and 1 mM dbcAMP for 7 min was analyzed as described in the legend to Fig. 4F. In panels B, C, and E, data are expressed as means ± standard deviations of the results from triplicate samples. A significant difference from the control in panels B and E or between two groups in panels C and D was determined by Student's t test and indicated by a single asterisk (P < 0.05) or double asterisks (P < 0.01).

Rap1 activation is essential for VE-cadherin-dependent cell adhesion and endothelial barrier function.

We next proceeded to investigate the role of Rap1 in VE-cadherin-dependent cell adhesion and endothelial barrier function. To examine the effect of Rap1 on cell permeability and VE-cadherin-mediated cell adhesion, we inactivated endogenous Rap1 by adenovirus-expressing Rap1GAPII (Ad-RapGAP), which specifically catalyzes the hydrolysis of GTP to GDP on Rap1 (30). As shown in Fig. 8A, endogenous Rap1 activity was almost completely suppressed by the expression of increasing amounts of Rap1GAPII in HUVECs. This Rap1 inactivation paralleled the increase in basal permeability (Fig. 8B) and the inhibition of cell adhesion to the VEC-Fc-coated dish (Fig. 8D). In contrast, a constitutively active Rap1, Rap1V12, reduced both basal and thrombin-increased cell permeability (Fig. 8C). VE-cadherin-mediated cell adhesion was also enhanced by Rap1V12 and EpacΔcAMP, a constitutively active mutant of Epac (Fig. 8D). Taken together, these results indicate that Rap1 activation is required for VE-cadherin-mediated cell adhesion and endothelial barrier function.

FIG. 8.

Rap1 plays a critical role in VE-cadherin-dependent cell adhesion and endothelial barrier function. (A) Rap1 inactivation was assessed by detecting GTP-Rap1 in HUVECs infected with different MOI of adenovirus-expressing Rap1GAPII (RapGAP) as indicated at the top. An adenovirus-expressing LacZ at an MOI of 50 was used as a control. GTP-bound Rap1 (GTP-Rap) was detected by pull-down assay as described in Materials and Methods. Rap1 (Rap) and Rap1GAPII (RapGAP) expression were examined by Western blot analysis. (B) The permeability of FITC-dextran across HUVECs infected with adenovirus as indicated at the bottom was analyzed as described in Materials and Methods. Data are the means ± standard deviations of the results from three independent experiments and are expressed as increases relative to those of LacZ-infected cells. (C) Monolayer HUVECs infected with either an adenovirus-expressing LacZ or that expressing Rap1V12 at an MOI of 50 for 24 h were medium changed and cultured for another 24 h. The permeability of cells upon 2-U/ml thrombin stimulation (Thr) after starvation for 1 h was analyzed as described in the legend to Fig. 1B. Data are the means ± standard deviations of the results from five independent experiments and are expressed as inductions relative to those of untreated HUVECs infected with the LacZ-expressing virus. (D) HUVECs were transfected with either empty vector (Mock), plasmids expressing Rap1GAPII (RapGAP), EpacΔcAMP, or Rap1V12 together with the luciferase reporter construct. Transfected cells were plated on the VEC-Fc-coated dish and allowed to adhere for 15 min. Cell adhesion was analyzed as described in Materials and Methods. Data are expressed as increases compared to those of mock-transfected cells. The results indicate the means ± standard deviations of the results from triplicate samples. Similar results were obtained in three independent experiments. Significant differences between two groups in panel C or from the control in panel D are determined by Student's t test and are indicated by a single asterisk (P < 0.05) or double asterisks (P < 0.01).

cAMP enhances VE-cadherin-dependent cell adhesion and endothelial barrier function by activating Rap1.

To test the requirement for Rap1 in endothelial barrier enhancement by cAMP-elevating GPCR agonists, we infected HUVECs with Ad-RapGAP and examined the effect of inactivation of Rap1 on PGI2- and FSK-induced reduction of cell permeability. Although basal endothelial permeability was reduced by PGI2 and FSK (Fig. 9A and B), overexpression of Rap1GAPII increased not only basal but also PGI2- and FSK-reduced endothelial permeability, indicating the requirement of Rap1 activity for PGI2- and FSK-induced barrier enhancement. We also investigated the involvement of Rap1 in PGI2- and FSK-induced VE-cadherin-dependent cell adhesion. PGI2 and FSK augmented VE-cadherin-dependent cell adhesion of HUVECs infected with control adenovirus (Ad-LacZ); however, their effects were dramatically suppressed by overexpression of Rap1GAPII (Fig. 9C and D). These data demonstrate that cAMP enhances VE-cadherin-dependent cell adhesion and endothelial barrier functions by activating Rap1.

FIG. 9.

Inactivation of Rap1 reduces PGI2- and FSK-induced barrier function and VE-cadherin-mediated cell adhesion. (A) Monolayer-cultured HUVECs grown on transwell filters were infected with either LacZ-expressing adenovirus (Ad-LacZ) or Rap1GAPII-expressing virus (Ad-RapGAP) at an MOI of 40 for 24 h. Medium was replaced with fresh medium after infection. Cells were cultured for an additional 24 h and treated with 10 μg of PGI2/ml for 30 min after serum starvation for 1 h. Permeability was analyzed as described in Materials and Methods. (B) The effect of 10 μM FSK on permeability in HUVECs infected with Ad-RapGAP was similarly analyzed. (C) HUVECs were infected with either Ad-LacZ or Ad-RapGAP at an MOI of 40 for 24 h. HUVECs resuspended in medium 199 with 0.5% BSA were plated onto VEC-Fc-coated dishes in the presence (+) or absence (−) of 10 μg of PGI2/ml for 7 min. Cell adhesion activity was quantified as described in the legend to Fig. 4A. (D) The effect of FSK on adhesion of HUVECs infected with Ad-RapGAP was analyzed similarly to that described for panel C. Resuspended HUVECs were preincubated with 10 μM FSK for 10 min before plating. Significant differences between two groups determined by Student's t test are indicated by a single asterisk (P < 0.05) or double asterisks (P < 0.01).

cAMP induces endothelial cortical actin rearrangement in a Rap1-dependent manner.

Endothelial barrier function is largely dependent upon the actin cytoskeleton supporting junctional adhesion molecules (10). Thus, we examined the effect of cAMP on cortical actin polymerization and assembly of polymerized actin in a monolayer of endothelial cells. Cortactin, an actin-binding protein, is known to be implicated in cortical actin rearrangement (8) and suggested to regulate S1P-induced endothelial barrier enhancement (11). PGI2, FSK, and 8-CPT-2′-O-Me-cAMP dramatically induced accumulation of polymerized actin and cortactin at cell-cell contacts (Fig. 10A). To explore the involvement of Rap1 in cAMP-mediated cortical actin rearrangement, an expression vector encoding Rap1GAPII was introduced into endothelial cells. FSK enhanced actin polymerization at cell-cell contacts in cells transfected with control vector encoding EGFP, whereas it did not in cells expressing Rap1GAPII (Fig. 10B). Cytochalasin D, an actin-depolymerizing agent, attenuated FSK-induced barrier enhancement (Fig. 10C) and inhibited FSK-induced VE-cadherin-dependent cell adhesion (Fig. 10D). These results suggest that the cortical actin rearrangement promoted by cAMP-Epac-Rap1 signaling may contribute to the potentiation of endothelial barrier function and VE-cadherin-dependent cell adhesion.

FIG. 10.

cAMP induces cortical actin rearrangement in a Rap1-dependent manner. (A) Monolayer-cultured HUVECs starved in 0.5% BSA-containing medium 199 for 3 h were stimulated with vehicle (top row), 10-μg/ml PGI2 (second row), 10 μM FSK (third row), and 0.2 mM 8-CPT-2′-O-Me-cAMP (8CPT) (bottom row) for 30 min. Fixed and permeabilized cells were stained with rhodamine-phalloidin (left column) and with anti-cortactin (center column). Rhodamine images to detect F-actin (red) and Alexa 488 images for cortactin visualized byAlexa 488-labeled secondary antibody (green) were obtained through a confocal microscope (BX50WI). Right panels show the merged images of rhodamine and Alexa 488 images. Bars, 20 μm. (B) HUVECs transfected with an EGFP-expressing vector (left) and pCXN2-Rap1GAPII-IRES-EGFP (right) were serum starved in 0.5% BSA-containing medium 199 for 3 h and stimulated with vehicle (top panels) and 10 μM FSK (bottom panels). Cells were fixed, permeabilized, and stained with Rhodamine-phalloidin. EGFP images (green) and rhodamine images showing F-actin (red) were obtained similar to those in panel A. Arrows and arrowhead indicate transfected and untransfected cells, respectively. Bars, 20 μm. (C) Cell permeability of HUVECs pretreated with 2 μM cytochalasin D (CytoD) for 30 min followed by 10 μM FSK stimulation for 30 min was analyzed as described in the legend to Fig. 1A. −, absent; +, present. (D) The effect of pretreatment of 2 μM cytochalasin D (CytoD) on adhesion of HUVECs stimulated with FSK was analyzed as described in the legend to Fig. 5E. A significant difference between two groups determined by Student's t test is indicated by double asterisks (P < 0.01).

DISCUSSION

cAMP is a well-known intracellular signaling molecule that is capable of restoring diminished endothelial barrier function. Previous reports suggested that cAMP-induced barrier enhancement occurs through PKA (27, 39). In this study, however, we demonstrated a novel PKA-independent signaling pathway, the cAMP-Epac-Rap1 signaling pathway, involved in cAMP-induced barrier function based on the following observations. PGI2- and FSK-reduced endothelial permeability was insensitive to H89. A specific activator for Epac, 8-CPT-2′-O-Me-cAMP, reduced both basal and thrombin-increased permeability. Plasma leakage in response to VEGF was also inhibited by 8-CPT-2′-O-Me-cAMP in vivo. We found that the activation of Rap1 leads to decreased permeability. Not only all cAMP-elevating bio-ligands we tested but also FSK, dbcAMP, and IBMX activated Rap1. Consistently, cAMP-dependent Rap1 activation upon stimulation by these ligands involved Epac in the regulation of barrier function. A previous report showed that Rap1 is phosphorylated by PKA in neutrophils and platelets, although the function of phosphorylated Rap1 has not been elucidated (37). So far, Epac is known to regulate several biological functions including integrin-dependent cell adhesion, insulin secretion, and calcium release through ryanodine-sensitive Ca²⁺ channels (reviewed in reference 5). In addition to these Epac-mediated functions, we show, for the first time, that Epac-Rap1 signaling is important for regulation of endothelial barrier function.

AJ assembly contributes to the regulation of barrier function. Rap1 is involved in the formation and maintenance of AJ constituted by cadherin (23, 41). Recently, it has been reported that homophilic ligation of E-cadherin induced Rap1 activation, which may be responsible for maturation of AJ (20). Consistently, suppression of endogenous Rap1 inhibits formation of E-cadherin-dependent cell adhesion (36), suggesting the critical role of Rap1 in the establishment of cadherin-based cell-cell contacts. Here, we demonstrate that Rap1 also acts downstream of cAMP-Epac to potentiate VE-cadherin-dependent cell adhesion, thereby improving barrier function. In addition to cAMP-elevating ligands, S1P, which enhances AJ formation and barrier function (18, 26), also activated Rap1 (our unpublished data). Thus, Rap1 may play a crucial role in barrier function induced by various types of barrier-improving factors.

Our data and previous studies show that cAMP protects thrombin-induced endothelial barrier dysfunction. cAMP does not limit the effect of thrombin on the initial loss of endothelial barrier (32). Instead, cAMP enhances the restoration of barrier function disrupted by thrombin. Recently, it was also reported that Cdc42 regulates the restoration of endothelial barrier function disrupted by thrombin (24). Thus, cAMP-Epac-Rap1 signaling may facilitate the formation of VE-cadherin-based cell-cell contacts, cooperatively or in parallel with Cdc42.

Rap1 enhances integrin-dependent cell adhesion in a variety of hematopoietic cells by modulating the affinity and avidity of integrin (6, 22). Cell adhesion to VEC-Fc-coated dishes was augmented by Rap1 activation, suggesting that the homophilic binding of VE-cadherin is also likely ascribed to the affinity and avidity of VE-cadherin modulated by Rap1-triggered inside out signaling. Hogan et al. reported that Rap1 activity is required for the targeting of E-cadherin molecules into nascent cell-cell contact sites, which in turn leads to the maturation of E-cadherin-based cell-cell contacts (20). Thus, cAMP-Epac-Rap1 signaling may also regulate the recruitment of VE-cadherin into maturing cell-cell contacts. Since downstream signaling of Rap1 that increases homophilic binding of VE-cadherin has not yet been characterized, the effector of cAMP-Epac-Rap1 signaling will need to be identified.

The actin cytoskeleton is a critical determinant of vascular integrity (10). PGI2, FSK, and 8-CPT-2′-O-Me-cAMP induced cortical actin rearrangement in a Rap1-dependent manner. FSK-induced VE-cadherin-dependent cell adhesion was inhibited by cytochalasin D. Thus, Rap1 may promote VE-cadherin-dependent cell adhesion by inducing cortical actin rearrangement. AF-6 may act downstream of Rap1 to regulate the actin cytoskeleton, since it binds to GTP-bound Rap1 and the actin cytoskeleton regulator, profilin, and is localized at AJ (2). Consistently, Canoe, the drosophila homolog of AF-6, and Rap1 function in the same molecular pathway during embryonic dorsal closure, which requires cell-cell contacts (3). S1P promotes endothelial barrier function by inducing Rac-dependent cortical actin rearrangement. S1P also induces Rap1 activation (our unpublished data). A previous report indicates that Rac can function downstream of Rap1 in the processing of the amyloid precursor protein (28). Taken together, Rac may act downstream of Rap1 to induce cortical actin rearrangement.

In conclusion, we have demonstrated that the cAMP-Epac-Rap1 signaling pathway promotes VE-cadherin-mediated cell adhesion and consequently improves endothelial barrier function.