Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Multiple investigations have established an essential role for the vascular endothelium in regulating the diffusion integrity of the intravascular space. Derangement in endothelial cell (EC) barrier function is an important pathophysiologic mechanism underlying a wide variety of disease states, including atherosclerosis, diabetes mellitus, and acute lung injury syndromes. A variety of physical stimuli or bioactive agonists stimulate the formation of intercellular gaps in the confluent endothelium, providing paracellular pathways for the exudation of macromolecules and solutes from the intravascular space, thereby contributing to interstitial edema formation and subsequent organ dysfunction (1). The barrier function of the endothelium is under dual regulation provided by the contributions of axial contractile force generation, which promotes EC intercellular gap formation and barrier dysfunction (2, 3), and by cell-matrix and cell-cell adhesive interactions that appear to resist cellular retraction and promote integrity of the EC barrier, a process described as tethering (3). Alterations in vascular barrier integrity can arise as a result of changes in either EC axial contraction, or EC tethering properties, or both. Several laboratories, including ours, have demonstrated the central importance of myosin light chain kinase (MLCK) in regulating the contractile state of the endothelium (4) and in modulating EC barrier function (5). Kolodney and Wysolmerski have shown that thrombin-stimulated human umbilical vein endothelium generates a contractile force per unit of cross-sectional area that is only an order of magnitude less than that of smooth muscle and is dependent on the integrity of the microfilamentous cytoskeleton (8). We have shown that thrombin-induced EC barrier dysfunction is critically dependent on the activation of a novel high molecular weight isoform of MLCK that catalyzes the phosphorylation of the regulatory myosin light chains resulting in prominent barrier dysfunction (7, 9, 10).

Despite the clearly demonstrated importance of MLCK- dependent contractile forces in the model of thrombin-induced EC gap formation and barrier dysfunction, much less is known about the role of tethering forces in regulation of EC barrier function. It is clear, however, that models of EC barrier dysfunction exist for which MLCK activation is not a strict requirement. For example, treatment of EC monolayers with either phorbol myristate acetate (PMA) (11) or pertussis toxin (12) results in significant increases in EC permeability, albeit to a lesser degree than that produced by thrombin. However, in contrast to thrombin (7) or histamine (5), neither PMA nor pertussis toxin produce MLCK activation in endothelium as judged by the lack of agonist-induced phosphorylation of myosin light chains (7, 12). In addition, the proinflammatory cytokine interleukin 1 (IL-1) (13) and the calcium ionophore ionomycin (14), which also produce significant EC permeability, elicit EC barrier dysfunction in an MLCK-independent fashion. Together, these MLCK-independent pathways indicate a potential role for noncontractile modalities, such as tethering force disruption, as a relevant mechanism for EC barrier dysfunction.

Focal adhesions, structures that contribute importantly to EC tethering, mediate attachment of the actin cytoskeleton to the extracellular matrix through a complex of cytoplasmic proteins termed the focal adhesion plaque (for reviews see references 15). Focal adhesion plaques physically link F-actin filaments to the cytoplasmic domain of the ₁ subunit of transmembrane integrin receptors through a complex that is composed of structural proteins, including -actinin, vinculin, talin, and paxillin, and other molecules including protein kinases and proteases. In addition to their role in mediating attachment of the cytoskeleton to the extracellular matrix, increasing evidence indicates that focal adhesion plaques are participants in the transduction of transmembrane signals that may affect cytoskeletal organization (18, 19), and are key determinants of cytoskeletal integrity (20), a function that is critical to the maintenance of the EC barrier (21). Thus, via either the modulation of cell-substratum adhesion, or through the regulation of cytoskeletal organization, focal adhesion plaques appear to be excellent targets for alterations in EC tethering forces evoked by the activated thrombin receptor. As phosphorylation is a major mechanism of signal integration in diverse cellular systems, in this article we examine whether phosphorylation of proteins in EC focal adhesion plaques is associated with cytoskeletal changes that occur subsequent to thrombin challenge. Our data indicate that although thrombin challenge produces a dramatic cytoskeletal rearrangement and reorganization of focal adhesion plaques, neither the serine/threonine nor tyrosine phosphorylation of vinculin and talin are significantly altered in thrombin-challenged ECs. In contrast, thrombin produces rapid phosphorylation of p125^FAK, whereas ionomycin, a potent Ca²⁺ ionophore that produces time- and dose-dependent EC barrier dysfunction (14), induces rapid p125^FAK dephosphorylation. These results suggest that the phosphorylation status of p125^FAK may play a regulatory role in specific models of agonist-induced EC permeability.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Endothelial Cell Culture

Bovine pulmonary arterial ECs, culture line CCL-209, were obtained frozen at 16 passages (American Type Tissue Culture Collection, Rockville, MD) and were utilized at passage 19 to 24. Cells were cultured as described previously (22) in complete medium, consisting of Dulbecco's modified Eagle medium (DMEM; GIBCO, Grand Island, NY), 20% (vol/vol) colostrum-free bovine calf serum (Irvine, Santa Ana, CA), 15 µg/ml EC growth supplement (Collaborative Research, Bedford, MA), 1% antibiotic/antimycotic (10,000 U/ml penicillin, 10,000 µg/ml streptomycin, 25 µg/ml amphotericin B; GIBCO), and 0.1 mM nonessential amino acids (GIBCO). The ECs grew to contact-inhibited monolayers with typical cobblestone morphology. Cells from primary flasks were detached with trypsin- EDTA (ethylenediaminetetraacetic acid) and resuspended in fresh culture medium and passaged to additional 75-cm² flasks, 60- and 100-mm² dishes, glass coverslips, or wells for permeability determination.

Fluorescence Microscopy

Cells were grown onto gelatinized glass coverslips placed in 35-mm² dishes for 2 to 3 days prior to treatment with 100 nM thrombin. Monolayers were fixed with 5% (wt/vol) paraformaldehyde in TBS (137 mM NaCl, 0.1% NaN₃, 20 mM Tris-HCl [pH 7.5]) for 10 min on ice, rinsed with TBS, permeabilized in 0.2% (vol/vol) Triton X-100 for 3 to 5 min, and then rinsed with TBS. Each coverslip was incubated with primary antibody (1:100 VIN-11-5 [Sigma, St. Louis, MO] or 1:40 anti-p125^FAK [Transduction Laboratories, Lexington, KY]) in TBS-0.5% bovine serum albumin at room temperature, rinsed liberally with TBS, and then incubated with 1:100 fluorescein-conjugated donkey anti-mouse IgG (Jackson, West Grove, PA) plus 1:300 rhodamine-conjugated phalloidin (Molecular Probes, Inc., Eugene, OR) for 1 h at 37°C. Following rinsing with TBS, coverslips were mounted onto glass slides with Fluoromount-G (Fisher Scientific, Pittsburgh, PA). The cells were viewed with a MRC-1024 krypton-argon laser confocal microscope (Bio-Rad, Hercules, CA) equipped with a 3-µm aperture and a 60× oil emersion lens. Rhodamine staining was visualized using an excitation of 568 nm and emissions recorded in series of 0.5-µm planar sections with a photomultiplier tube and 598-nm filter. Fluorescein staining was subsequently imaged at the same planar depths, using an excitation of 488 nm and an emission filter of 522 nm. Images were digitally recorded by LaserSharp acquisition software (Bio-Rad), processed with Metamorph digital image analysis software (Universal Imaging Corp., West Chester, PA), and printed on a thermal dye diffusion printer (Kodak, Rochester, NY).

Transendothelial Electrical Resistance

Endothelial cells were seeded onto evaporated gold microelectrodes and grown to confluence as previously described (23). The EC monolayer-containing microelectrodes were then connected to an electrical cell-substrate impedance system (Applied Biophysics, Inc., Troy, NY) and rinsed three times with medium 199 (GIBCO) supplemented with 20 mM Hepes, pH 7.4 (M199H), and the transendothelial electrical impedance was monitored for 30 min to establish a baseline resistance. Bovine thrombin (100 nM) was then added and real-time transendothelial resistance measurements were collected. Resistance values from each microelectrode were normalized as the ratio of measured resistance to baseline resistance and plotted versus time.

Vinculin and Talin Purification

Bovine platelet vinculin was prepared from freshly isolated platelet-enriched bovine plasma, which was subjected to nitrogen cavitation, and chromatographed on DEAE-cellulose as described by Hathaway and Adelstein (24). NaCl was added to the vinculin-rich flow-through from this column to a final concentration of 150 mM NaCl, followed by stirring at 4°C for 30 min and centrifugation at 20,000 × g at 4°C for 20 min. The vinculin-containing supernatant was collected and 14.9 g of (NH₄)₂SO₄ per 100 ml was added with constant stirring over 30 min at 4°C, followed by centrifugation at 20,000 × g for 20 min at 4°C. The vinculin-containing supernatant was collected and 5.6 g of (NH₄)₂SO₄ per 100 ml was added with constant stirring at 4°C over 30 min, followed by centrifugation at 20,000 × g for 20 min at 4°C. The supernatant was discarded and the vinculin-containing pellet was resuspended in a minimal volume of buffer A (20 mM NaCl, 0.1 mM EDTA, 15 mM 2-mercaptoethanol, 20 mM Tris-HCl, pH 7.6), dialyzed overnight in excess buffer A at 4°C, and then loaded onto a column of DEAE-cellulose equilibrated with buffer A. After washing the column with 4 vol of buffer A, proteins were eluted with a NaCl gradient and 2-ml fractions collected. Fractions were evaluated by Western blotting, and the vinculin-enriched fractions (> 90% by sodium dodecyl sulfate-polyacrylamide gel electrophoresis [SDS-PAGE]) were pooled.

Talin was prepared from freshly obtained gizzards (150 g), which was then added to 600 ml of doubly distilled H₂O containing 650 µl of 1 M phenylmethylsulfonyl fluoride (PMSF). The gizzard suspension was homogenized in a Waring blender on high for three 10-s bursts, and then centrifuged at 7,000 rpm at 4°C for 10 min. The pellet was resuspended in 600 ml of doubly distilled H₂O containing 325 µl of 1 M PMSF and homogenized in a Waring blender (three 5-s bursts). The homogenate was centrifuged at 10,000 rpm at 4°C for 10 min. The pellets were resuspended in buffer B (2 mM Tris, 1 mM EGTA [ethylene glycol-bis (-aminoethyl ether)-N,N,N',N'-tetraacetic acid], 0.5 mM PMSF, pH 9.0), and extracted for 1 h at 37°C with occasional stirring. The resulting supernatant was collected and centrifuged at 10,000 rpm at 4°C for 10 min. The pellet was discarded and the volume of supernatant measured and transferred to a glass beaker; pH was adjusted to 7.0 by adding glacial acetic acid. Actin was precipitated by bringing the solution to 10 mM MgCl₂ and stirring for 15 min at 25°C. The suspension was centrifuged at 10,000 rpm at 4°C for 10 min and the pellet discarded. The volume of the supernatant was measured, and 18 g of (NH₄)₂SO₄ per 100 ml of supernatant was added. The solution was stirred at 4°C for 1 h followed by centrifugation at 10,000 rpm at 4°C for 10 min. The vinculin-containing pellet was set aside. An additional 5.6 g of (NH₄)₂SO₄ per 100 ml was added to the supernatant with stirring for 1 h at 4°C. The precipitated protein was collected by centrifugation, and then dissolved in and dialyzed against buffer A. The dialysate was chromatographed on a DEAE-cellulose column equilibrated in buffer A, and proteins eluted with a NaCl gradient. Fractions were analyzed by SDS-PAGE, and fractions enriched in 215-kD protein were dialyzed against buffer A. The dialyzed solution was chromatographed on a hydroxyapatite column equilibrated against buffer A, and eluted with a NaCl gradient. Fractions were again analyzed with SDS-PAGE, and 215-kD protein-containing fractions were pooled and stored at 4°C. Confirmation of talin isolation was obtained by Western blot analysis, using rabbit anti-talin (N681).

Purification of EC Protein Kinase C

Confluent bovine pulmonary artery endothelial cell (BPAEC) monolayers providing 25-50 mg of total cellular protein were used for purification of EC protein kinase C (PKC), as has been previously described (11). Culture medium was removed and cells were rinsed twice with lysis buffer (20 mM Tris-HCl [pH 7.4], 0.33 M sucrose, 2 mM EDTA, 0.5 mM EGTA, 1 mM PMSF), then scraped into 5 ml of lysis buffer. Cell suspensions were disrupted by probe sonication (Cole-Parmer, Niles, IL) on ice (five disruptions, 10 s each). After centrifugation at 1,000 rpm for 10 min at 4°C to remove whole cells, the homogenates were centrifuged at 100,000 × g for 90 min at 4°C. The resulting supernatants, representing the cytosolic fraction, were loaded onto DEAE-cellulose columns equilibrated with elution buffer (lysis buffer without sucrose). The PKC was eluted with elution buffer supplemented with 100 mM NaCl (two column volumes). The resulting PKC-enriched eluent was applied to a 1 × 5-cm threonine-sepharose column and rinsed with elution buffer. The PKC was then eluted with an NaCl gradient, using a high-performance liquid chromatography system (Perkin-Elmer, Norwalk, CT). Fractions were collected and assayed for Ca²⁺- and phospholipid-dependent [-³²P]adenosine triphosphate (ATP) phosphorylation of H1 histone. The purified PKC fractions obtained from threonine-sepharose chromatography were pooled, desalted, and concentrated by ultracentrifugation (Centricon-30; Amicon, Danvers, MA), and stored at 70°C after the addition of 0.01% Triton X-100.

Immunoprecipitation of Focal Adhesion Plaque Proteins and p125^FAK

Immunoprecipitation of vinculin and talin was carried out in ³²P-labeled cells, using a denaturing protocol to reduce background signals. Cells were grown to confluence in 60-mm² dishes and radiolabeled using ³²PO₄³ (New England Nuclear, Boston, MA) at 400 µCi/ml in phosphate-free DMEM (Sigma) supplemented with 1% bovine fetal calf serum and 20 mM Hepes (Sigma) for 2 h at 37°C. Labeling medium was removed by washing the dishes three times with phosphate-free DMEM, after which the dishes were treated with 100 nM thrombin as required by the experimental protocol. Reactions were stopped by scraping cells into 120 µl of denaturing stop buffer (1% SDS [wt/vol], 1% 2-mercaptoethanol [vol/vol], 137 mM NaCl, 20 mM Tris-HCl [pH 7.5 at 25°C], 1 mM EGTA, 1 mM EDTA), drawing the homogenates through a 26-gauge needle to shear DNA, and boiling for 5 min to denature proteins. The homogenates were then diluted to ~ 0.1% SDS by adding 1,080 µl of TBS.

Focal adhesion kinase (FAK) was immunoprecipitated from unlabeled EC monolayers using a nondenaturing protocol. The BPAECs were grown to confluency in 100-mm² dishes, treated with 100 nM thrombin, extracted in 1,200 µl of ice-cold extraction buffer (137 mM NaCl, 20 mM Tris-HCl [pH 7.4], 1 mM EGTA, 1 mM EDTA, 200 µM sodium orthovanadate, 1% Triton X-100 [Sigma], 0.5% NP-40 [Sigma], 0.5 mM PMSF, 2 mM benzamidine, 1 mM N-p-tosyl-L-lysine chloromethyl ketone [TLCK], 25 µg/ ml leupeptin) for 20 min at 4°C, and scraped and transferred to 1.5-ml centrifuge tubes. The samples were microcentrifuged for 10 min at 4°C to pellet detergent-insoluble proteins (cytoskeleton), and the supernatants were transferred to new tubes.

For both denaturing and nondenaturing protocols, nonspecific protein binding to protein A was cleared by incubation of samples with 50 µl of Pansorbin (Calbiochem, San Diego, CA) for 30 min at 25°C, followed by microcentrifugation for 15 min. To 500 µl of the resulting supernatant, either 3 µl of rabbit antiserum against talin (N681) or vinculin (G989), or 4 µg of FAK antibody (UBI, Lake Placid, NY), was added, followed by incubation for 1 h at 25°C. Antibody-antigen complexes were collected by adding 50 µl of 10% (wt/vol) solution of protein A-sepharose CL-4B (Pharmacia-LKB Biotechnology, Inc., Piscataway, NJ) to samples and incubating for 1 h at 25°C. The complexes containing bound antigen were pelleted and washed three times with TBS containing 1% Triton X-100 (Sigma) and 0.1% SDS. To the final pellet, 60 µl of sample buffer was added and the tubes were heated to boiling for 5 min. The resulting final supernatant, containing the immunoprecipitated protein, was split into equal 20-µl aliquots, and electrophoresed in parallel on 8% polyacrylamide slab gels. For autoradiography, one of the gels was fixed for 15 min in 7% acetic acid-10% methanol, dried, and then exposed to Kodak XAR-5 film with intensifying screens at 70°C. The remaining gels were prepared for Western immunoblotting.

In Vitro Phosphorylation

Phosphorylation experiments were carried out in a 250-µl reaction volume containing 20 mM Tris-HCl (pH 7.4), 10 mM magnesium acetate, 0.05 mM ATP, 0.008 µCi/µl [-³²P]ATP (New England Nuclear), and 25 µg of either histone III_s or bovine vinculin as the kinase substrate. Experimental conditions included 0.1 mg/ml L--phosphatidylserine (Sigma), 0.01 mg/ml sn-1,2-diacylglycerol (Sigma), and 1 mM CaCl₂. Control conditions included 5 mM EGTA. Reactions were started by the addition of 1 µg of protein yielded from the purification of EC PKC, and allowed to react at 25°C for 5 min. The reaction mixtures were stopped by the addition of 250 µl of ice-cold 10% trichloroacetic acid (TCA) and 25 µl of 0.01% bovine serum albumin. The TCA-precipitable materials were collected by centrifugation and the pellets were resuspended in 30 µl of SDS-PAGE sample buffer and 5 µl of saturated Tris-HCl. The samples were electrophoresed on 10% SDS-polyacrylamide slab gels and stained with Coomassie Blue to visualize the protein bands. The gels were dried and exposed to Kodak XAR-5 film using intensifying screens at 70°C.

Western Immunoblotting

Proteins separated on SDS-polyacrylamide slab gels were electrophoretically transferred to nitrocellulose membranes (Bio-Rad, Rockville Center, NY) at 100 V for 1 h in Tris-glycine buffer (25 mM Tris, 200 mM glycine). The membranes were blocked in 5% bovine serum albumin in TBS-T (137 mM NaCl, 20 mM Tris-HCl, pH 7.5 at 25°C, 0.1% Tween 20) for 1 h, washed extensively in TBS-T, and incubated for 1 h at room temperature in primary antibody solution, either mouse anti-FAK (Transduction Laboratories) or mouse anti-phosphotyrosine (UBI) in TBS-T containing 5% (wt/vol) bovine serum albumin. The membranes were then washed in TBS-T, and incubated for 1 h at room temperature in horseradish peroxidase-conjugated goat anti-mouse antibodies (Bio-Rad) at 1:10,000 dilution to detect antibody-antigen complexes. Following extensive washing in TBS-T, the membranes were developed utilizing an enhanced chemiluminescence protocol (Amersham, Arlington Heights, IL), and exposed to Hyperfilm-ECL (Amersham). Quantitation of the blots was performed with a Bio-Rad GL-670 scanning densitometer and Molecular Analyst version 1.5 software. Relative densitometric intensity (RDI) ratios were calculated from phosphotyrosine and FAK densitometric intensities that were normalized to corresponding signals of untreated control conditions for each experiment. The RDI ratios at each time point were expressed as the mean of the calculated ratios ± standard deviation.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Thrombin-induced Cytoskeletal and Focal Adhesion Plaque Rearrangement

Bovine pulmonary artery EC monolayers were grown to confluence and treated with thrombin (100 nM) prior to fixing and immunofluorescent staining. Figure 1A shows rhodamine-phalloidin fluorescent localization of F-actin bundles in vehicle-treated control cells. Actin fibers were arranged into a fine reticular cortical network, with dense peripheral actin bands evident at cell boundaries. Immunofluorescent localization of vinculin, which identifies focal adhesions, showed a fine diffuse pattern with central clearing (Figure 1B). Treatment of confluent monolayers with thrombin showed a dramatic time-dependent rearrangement and restructuring of the cellular F-actin network (Figure 1C). The EC monolayer no longer exhibited a cobblestone morphology, and cells appeared spindle shaped with intercellular gaps evident by rhodamine-phalloidin staining. Furthermore, thrombin induced a dissolution of the dense peripheral actin band, with the fine reticular cortical F-actin network replaced by thicker, more brightly staining stress fibers that were arranged in an orientation that paralleled the spindle axis of stimulated cells. Vinculin immunofluorescent localization in thrombin-treated BPAECs showed profound reorganization of vinculin staining with the fine diffuse vinculin distribution replaced by an overall rarefaction of vinculin, particularly near the cell-cell interface (Figure 1D). Dense, brightly staining clumps of vinculin were formed at areas that corresponded to the terminations of actin stress fibers. Studies were also performed to immunolocalize the p125^FAK under similar conditions (Figure 2). Similar to the vinculin distribution in control monolayers, FAK distribution in control monolayers demonstrated a diffuse, fine cellular distribution (Figure 2B), which was dramatically redistributed by thrombin challenge into aggregates predominantly located at the intercellular boundary (Figure 2D). We have previously shown that thrombin challenge induces an increase in transendothelial albumin clearance, which becomes statistically significant beyond 10 min of thrombin challenge (25). To provide additional physiologic characterization with which to correlate the observed immunocytochemical cytoskeletal alterations, real-time measurements of transendothelial electrical resistance were made during thrombin challenge (Figure 3). Thrombin challenge induced a very rapid early decline in electrical transendothelial resistance, which was > 50% maximal within the first 15 to 30 min of agonist stimulation. Thus, these data provide a positive correlation between the thrombin-induced rearrangements in the cytoskeletal proteins actin and vinculin and the signaling protein FAK with thrombin-mediated alterations in cell shape and adhesion as measured by changes in electrical impedance.

View larger version (186K): [in this window] [in a new window]   Figure 1.   The effect of thrombin on EC actin and vinculin organization. Bovine pulmonary artery EC monolayers were grown on glass coverslips, treated with 100 nM thrombin or vehicle for 10 min, followed by fixing, staining, and imaging as described in MATERIALS AND METHODS. (A) F-actin staining in nascent vehicle-treated control monolayers, which demonstrates thin cortical actin fibers and the dense peripheral actin band. (B) Vinculin immunolocalization in the same control cells, demonstrating fine, predominantly peripheral vinculin staining. (C) F-actin staining in EC monolayers following exposure to 100 nM thrombin for 15 min. There is an overall increase in cellular F-actin, which has been arranged in thick stress fibers that are arranged parallel to the spindle axis of contracted cells. (D) After 30 min there is a loss of the diffuse vinculin staining pattern, with particular loss of vinculin at cell-cell borders. Vinculin appears as dense, brightly staining clumps that correspond to the terminations of stress fibers.

View larger version (196K): [in this window] [in a new window]   Figure 2.   The effect of thrombin on EC actin and focal adhesion kinase organization. Bovine pulmonary artery EC monolayers grown to confluence on glass coverslips were treated with 100 nM thrombin, fixed, stained, and imaged as described in MATERIALS AND METHODS. Control monolayers again demonstrate a dense peripheral actin band (A) and fine, predominantly peripheral localization of FAK (C). Thrombin challenge induced the formation of intercellular gaps and reorganization of F-actin into stress fibers (B). Correspondingly, thrombin induced a redistribution of FAK into aggregates in a predominantly peripheral location corresponding to terminations of stress fibers (D).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Effect of thrombin on transendothelial electrical resistance. Bovine pulmonary artery ECs were grown to confluence on gold microelectrodes. The baseline transendothelial electrical resistance was determined across the monolayer for 100 min, following which 100 nM thrombin was added to the medium and real-time resistance determinations were obtained (shown is a representative tracing of normalized resistance values versus time from four experiments). The addition of thrombin induced a rapid and dramatic decline in transendothelial electrical resistance, which attained > 50% of the maximal change within the first 15 min of agonist challenge.

In Vitro and In Situ Phosphorylation of Vinculin and Talin

Protein kinase C is an important signaling enzyme that is known not only to be activated by thrombin receptor occupancy and to produce EC barrier dysfunction (11), but also to be colocalized with focal adhesion plaques in fibroblasts (26). Thus we hypothesized that thrombin-induced focal adhesion reorganization may be mediated by PKC-mediated phosphorylation of vinculin and talin. The potential for focal adhesion plaque proteins to serve as downstream signaling targets from the activated thrombin receptor via PKC was evaluated using purified bovine platelet vinculin (Figure 4A) and chicken gizzard talin. These proteins were added to an in vitro system containing purified EC PKC (Figure 4B). In the absence of the PKC activators, Ca²⁺, phospholipid, and diacylglycerol, there was no significant phosphorylation of vinculin (Figure 4B, lanes 1 through 4), whereas in the presence of activators there was significant phosphorylation of vinculin by EC PKC (Figure 4B, lane 5). Similarly, talin was also significantly phosphorylated by EC PKC in the presence of activators. These data confirm that vinculin and talin are in vitro substrates for phosphorylation by PKC.

View larger version (34K): [in this window] [in a new window]   Figure 4.   The effect of EC protein kinase C on the in vitro phosphorylation of vinculin. (A) Sequential steps in the purification of vinculin from bovine platelets: crude platelet lysate (lane 1); the vinculin-enriched pellet from the second ammonium sulfate precipitation step (lane 2); and pooled fractions from DEAE-cellulose chromatography (lane 3). Vinculin obtained from this procedure is > 90% pure. In (B) the purified vinculin was added to an in vitro system containing EC PKC with various combinations of PKC activators (L--phosphatidylserine, 1,2-sn-diacylglycerol, and Ca2+) or inhibitors (EGTA) as described below each lane. After reacting for 10 min at 30°C, the reactions were stopped with ice-cold TCA, run on SDS-polyacrylamide gels, and developed autoradiographically. Only in the presence of phosphatidylserine, diacylglycerol, and Ca2+, the only conditions under which PKC is active, is there significant phosphorylation of vinculin (lane 5). Bottom: The PKC phosphorylation of histone IIs, used as a positive control.

Vinculin and talin were next immunoprecipitated from confluent [³⁵S]methionine-labeled ECs (Figure 5A), and from ³²P-labeled ECs, to identify changes in total phosphorylation (Figure 5B). These studies revealed both vinculin and talin to be phosphorylated in situ under basal conditions; however, there was no appreciable increase in the total level of phosphorylation following treatment of EC monolayers with 100 nM thrombin. Thus, although vinculin and talin are in vitro substrates for phosphorylation by PKC and appear to be phosphorylated in situ, they do not appear to be substrates for thrombin-mediated serine/threonine phosphorylation.

View larger version (44K): [in this window] [in a new window]   Figure 5.   In situ vinculin and talin phosphorylation in thrombin-treated BPAECs. (A) Immunoprecipitation of vinculin and talin from [35S]methionine-labeled denatured EC homogenates, using specific antibodies as described in MATERIALS AND METHODS. Rabbit anti-vinculin specifically precipitates 130-kD protein corresponding to the molecular mass of vinculin, whereas rabbit anti-talin specifically precipitates a 215-kD protein corresponding to the molecular mass of talin. (B) The same experiment in 32P-labeled ECs stimulated with 100 nM thrombin for 5 and 15 min. Talin and vinculin are both phosphorylated under basal conditions in BPAECs, and thrombin treatment did not alter the level of phosphorylation of either vinculin or talin.

Tyrosine Phosphorylation of Focal Adhesion Proteins

Although the cloned thrombin receptor does not exhibit intrinsic tyrosine kinase activity, thrombin activates cellular tyrosine kinases and/or phosphatases in specific cell systems (27, 28). Furthermore, focal adhesion plaques colocalize with tyrosine kinases, and contain proteins that are substrates for endogenous tyrosine kinase activities. To examine tyrosine phosphorylation of specific focal adhesion proteins, vinculin and talin were immunoprecipitated from EC monolayers that had been exposed to 100 nM thrombin (1 to 30 min), and immunoblotted with a specific phosphotyrosine antibody. Neither vinculin nor talin contained significant levels of immunoreactive phosphotyrosine in the nascent, unstimulated state, with less than a 10% increase following treatment with 100 nM thrombin (1 to 30 min).

FAK is a protein tyrosine kinase that is a constituent of focal adhesion plaques, and is itself phosphorylated on tyrosine residues during activation (29). The p125^FAK was identified in unstimulated control BPAECs by Western blotting techniques and confirmed as a 125-kD protein (Figure 6, lane 1). Western immunoblotting with anti-phosphotyrosine antibodies demonstrated that in nascent, confluent EC monolayers, FAK is phosphorylated on tyrosine residues (Figure 6, lane 2). Following treatment of BPAEC monolayers with 100 µM orthovanadate, an inhibitor of cellular tyrosine phosphatases, there is a severalfold increase in the level of tyrosine phosphorylation (Figure 6, lane 3), indicating that FAK is an in situ substrate for endogenous tyrosine kinase and phosphatase activities. FAK immunoprecipitates were next obtained from thrombin-stimulated ECs and analyzed for phosphotyrosine content and immunoreactive FAK. These studies demonstrated a time-dependent decrease in both phosphotyrosine and FAK content (Figure 7A); however, the FAK phosphotyrosine content for each condition declined at a much slower rate than the decline in immunoreactive FAK. For example, the densitometrically determined FAK content in thrombin-treated ECs after 1 min of stimulation was approximately 20% relative to that of vehicle-treated controls, whereas phosphotyrosine content remained approximately 80% of controls. These data indicate that although immunoreactive FAK protein is lost after thrombin stimulation in a time-dependent manner from FAK immunoprecipitates, the ratio of phosphotyrosine content to FAK content is increased up to fivefold, consistent with FAK phosphorylation on tyrosine residues. Increased levels of immunoreactive FAK following thrombin stimulation were not found in the detergent-insoluble pellet, indicating that FAK redistribution to the cytoskeleton is not responsible for the decreased levels of FAK recovered by immunoprecipitation after thrombin treatment (data not shown). To confirm the observation that thrombin challenge induced tyrosine phosphorylation of FAK in ECs, two different strategies were utilized: (1) FAK was immunoprecipitated from ECs challenged with the proteolytically inactive thrombin receptor-activating peptide (TRAP) S-F-L-L-R-N, corresponding to the six residues of the new N terminus of the activated thrombin receptor (10); and (2) FAK was immunoprecipitated from thrombin-challenged ECs to which the potent thrombin-specific protease inhibitor, hirudin, was added only to the immunoprecipitation buffer following thrombin challenge. Although TRAP does not possess intrinsic proteolytic activity, it effectively reproduces thrombin effects on EC activation, including induction of phosphoinositide hydrolysis, activation of PKC, phosphorylation of myosin light chains, and EC barrier dysfunction (10). Phosphotyrosine immunoblots of FAK immunoprecipitates from TRAP-treated cells showed moderate increases in phosphotyrosine content at 5 and 15 min (Figure 7B). Corresponding FAK immunoblots demonstrate that, unlike thrombin, there is much less attenuation of immunoreactive FAK, with overall 1.5-fold increase in the phosphotyrosine-to-FAK ratio (Figure 7B). Although these data clearly confirm thrombin's capacity to induce FAK phosphorylation, a second strategy was used to confirm thrombin-induced tyrosine phosphorylation of FAK in ECs. Hirudin, a specific and potent inhibitor of thrombin's proteolytic activity, was added to the immunoprecipitation buffer during the immunoaffinity separation steps (Figure 7C). Phosphotyrosine immunoblots of FAK immunoprecipitated from thrombin-stimulated/hirudin-treated lysates demonstrated a large increase in phosphotyrosine content after 1 min of thrombin treatment (Figure 7C). In contrast, the corresponding FAK immunoblot demonstrated minimal loss of immunoreactive FAK. Thus, the rapid time-dependent loss of immunoreactive FAK seen in thrombin-challenged EC lysates is likely the consequence of nonspecific binding of thrombin to the endothelial monolayer surface during thrombin stimulation, resulting in proteolytically active thrombin liberated into the cellular detergent lysates. The validity of this speculation is supported by the lack of time-dependent attenuation of immunoreactive FAK levels in TRAP-treated EC, or in thrombin-treated ECs to which hirudin is added to the detergent lysate. Thus, these data further demonstrate that thrombin induces phosphorylation of FAK on tyrosine residues.

View larger version (32K): [in this window] [in a new window]   Figure 6.   FAK immunoblotting of immunoprecipitates from detergent-extracted BPAECs. BPAEC lysates were prepared by extracting cells in a nonionic detergent buffer for 20 min at 4°C. Aliquots of the resulting supernatants (500 µl, approximately 450 µg of total protein) were incubated with 4 µg of rabbit anti-p125FAK and immune complexes collected with protein A-conjugated sepharose beads. Immunoblotting the immunoprecipitates of control ECs with mouse anti-p125FAK antibody (F) specifically recognizes a protein with a molecular mass of 125 kD, corresponding to p125FAK (lane 1). The level of phosphorylation on tyrosine residues in nascent control cells was determined by immunoblotting with an anti-phosphotyrosine (PT) antibody (lane 2). Following treatment of BPAEC monoloayers with 100 µM orthovanadate for 30 min, phosphotyrosine immunoblotting of FAK immunoprecipitates shows a severalfold increase in tyrosine phosphorylation of FAK, confirming that FAK is an in situ substrate for BPAEC tyrosine kinase and tyrosine phosphatase activities (lane 3).

View larger version (20K): [in this window] [in a new window]   Figure 7.   The effect of thrombin on p125FAK tyrosine phosphorylation in FAK immunoprecipitates. Confluent BPAECs were treated with either vehicle or 100 nM thrombin for 1 to 15 min, followed by nonionic detergent lysis and immunoprecipitation with rabbit anti-p125FAK antiserum. The immunoprecipitate was solubilized in 70 µl of SDS-PAGE sample buffer, and 20-µl aliquots were electrophoretically separated in parallel and transferred to nitrocellulose for immunoblotting. A representative experiment immunoblotted for phosphotyrosine and a corresponding immunoblot of the same experiment for p125FAK is shown (A). As compared with vehicle-treated controls, at early time points thrombin induced a time-dependent decrease of FAK phosphotyrosine content (A, top), and a more time- dependent decrease of FAK immunoreactivity in FAK immunoprecipitates (A, bottom). BPAECs were then treated with the nonproteolytic thrombin receptor agonist peptide TRAP (100 µM) and p125FAK immunoprecipitates prepared (B). Immunoblotting demonstrated a modest increase in the FAK phosphotyrosine-to-FAK content ratio at 5 and 15 min. In contrast to (A) there was no apparent decline in FAK immunoreactivity in immunoprecipitates. Hirudin was subsequently added to the detergent lysates of thrombin-treated ECs (C). Hirudin blocked the thrombin-induced decrease in immunoprecipitated FAK content (C, bottom). Under these conditions, thrombin induced a threefold increase in FAK phosphotyrosine content, compared to vehicle-treated controls (C, top).

The potential Ca²⁺ dependency of tyrosine phosphorylation of FAK was subsequently examined by utilizing the Ca²⁺ ionophore, ionomycin, to effect changes in intracellular Ca²⁺ concentration independently of the thrombin receptor. Ionomycin challenge of EC produced two distinct effects on phosphotyrosine and FAK content in FAK immunoprecipitates: (1) 10 µM ionomycin demonstrated a rapid time-dependent decrease (within 1 min) in the phosphotyrosine content in FAK immunoprecipitates to less than 20% of vehicle-treated controls; and (2) 10 µM ionomycin challenge also produced a slower time-dependent loss of FAK protein in immunoprecipitates. The net result of these effects on the tyrosine phosphorylation of FAK is expressed in Figure 8. After 1 min of ionomycin challenge, FAK immunoreactive phosphotyrosine had decreased to less than 20% of vehicle-treated controls, whereas immunoreactive FAK remained at greater than 80% of vehicle-treated controls (Figure 8A). However, at later time points, the immunoreactive FAK signal steadily declined, so that after 15 min of 10 µM ionomycin challenge, immunoreactive FAK was approximately 20% of control and FAK phosphotyrosine was undetectable (Figure 8B). The following observations indicate that these effects are due to the activation of ionomycin-induced cellular mechanisms: (1) Time-dependent loss of FAK phosphotyrosine and FAK content in immunoprecipitates was not due to cytotoxicity, as 10 µM ionomycin challenge of BPAECs for up to 3 h did not result in significant cytotoxicity as judged by ⁵¹Cr and lactate dehydrogenase release (14); (2) the time-dependent loss of FAK was not due to loss of cellular protein from the detergent lysates, as protein concentrations of lysates harvested after 15 min of 10 µM ionomycin treatment were greater than 70% of vehicle-treated controls; and (3) the time-dependent decline in FAK content could not be explained by ionomycin-induced redistribution of FAK to the detergent-insoluble pellet, as FAK immunoblots of this fraction failed to show increases in FAK content (data not shown).

View larger version (38K): [in this window] [in a new window]   View larger version (11K): [in this window] [in a new window]   Figure 8.   Comparison of thrombin- and ionomycin-induced changes in p125FAK tyrosine phosphorylation. p125FAK immunoprecipitates of ECs treated with 10 µM ionomycin for 1 min were immunoblotted for both phosphotyrosine and FAK content (A). As compared with vehicle-treated control ECs, ionomycin induced a rapid decrease in the phosphotyrosine content of FAK immunoprecipitates (A, top). The level of FAK present is only slightly decreased after 1 min of treatment, compared to control (A, bottom). Phosphotyrosine/FAK RDI ratios from thrombin- and ionomycin-treated ECs were plotted versus time (B). Each time point represents the mean RDI ± standard deviation (n = 3 to 6). Both thrombin and ionomycin treatment produce divergent and distinct effects on the immunoreactive phosphotyrosine/FAK ratios compared to vehicle-treated control ECs, which did not change over time.

The net effects of thrombin and ionomycin on FAK tyrosine phosphorylation are compared in Figure 8B, which expresses the normalized RDI ratios for FAK immunoreactive phosphotyrosine per unit of immunoprecipitated FAK (PTyr/FAK). Vehicle-treated control ECs demonstrated no appreciable change in FAK phosphorylation, as the PTyr/FAK ratio remained level throughout the period (1 to 15 min). Thrombin treatment resulted in a rapid increase in the PTyr/FAK ratio that was maximal at 1 and 5 min of stimulation, indicating an increase in FAK tyrosine phosphorylation. Ionomycin, however, rapidly reduced the PTyr/FAK ratio at all time points examined, indicating a rapid and sustained dephosphorylation of FAK. Thus the effects of thrombin and ionomycin on net FAK phosphorylation are qualitatively very different.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Although the early signaling events that occur following thrombin receptor binding are well established (10, 30, 31), the molecular events that ultimately result in thrombin-induced EC activation and barrier dysfunction remain incompletely understood. Increasing information indicates that cytoskeletal proteins and other intracellular proteins that affect cytoskeletal organization are quite likely important targets for thrombin-activated signaling pathways (11, 32). Thus, potential mechanisms by which thrombin might produce EC paracellular gap formation and barrier dysfunction include (1) modulation of EC actomyosin contractility, and (2) alterations of EC tethering properties through the direct or indirect modulation of cytoskeletal organization, or through effects on cellular proteins that participate in cytoskeleton-dependent cell-cell or cell- substratum interactions. Recent reports have provided evidence that focal adhesions, defined as areas of very close association between plasma membrane and extracellular matrix, are key structures that participate in both cellular cytoskeletal organization and in cell-matrix and cell-cell attachment. For example, -actinin, which is present within focal adhesion plaques (33), participates in providing cytoskeletal integrity (20) because microinjection of non-actin-binding proteolytic -actinin fragments, which interact with the ₁ subunit of integrin but not with F-actin (34), produced stress fiber disassembly and loss of focal adhesions in fibroblasts. In addition, a functional role for cell-matrix adhesion in determining EC permeability was revealed by the observation that antibodies directed at extracellular matrix receptors increased EC permeability (35). Increasing information indicates that focal adhesions are also important signal transduction units capable of affecting cell differentiation and cytoskeletal organization (18, 36). A number of important cell-signaling molecules colocalize with focal adhesion plaques, including PKC (26), Ca²⁺-dependent protease II (37), and the tyrosine kinases pp60^c-src (38) and pp125^FAK (39). The abundance of regulatory and signaling molecules in this region indicates not only a critical role in determining cytoskeletal organization, but also underscores the ability of focal adhesions to serve as signal transduction units. Because the EC cytoskeletal integrity has a critical role in determining EC permeability, focal adhesion plaques may be central regulatory sites for modulating EC barrier function not only by virtue of their role in anchoring the cytoskeleton to the extracellular matrix, but also by affecting cytoskeletal organization and function via the key signaling molecules that localize at these structures. Thus, the focal adhesion plaque would appear to be an excellent potential target for the effector mechanisms of the activated thrombin receptor. In the present study, we examined the phosphorylation of specific focal adhesion plaque constituent proteins isolated from thrombin-stimulated BPAECs to address this hypothesis.

The findings of thrombin-induced rearrangement of F-actin, loss of dense peripheral actin bands, and reorganization of vinculin are similar to those of prior reports in another cell system (40). These immunohistologic changes are positively correlated with the temporal evolution of endothelial cell barrier dysfunction as measured by increases in transendothelial albumin clearance, as previously reported (25), and also by alterations in transendothelial electrical resistance, as reported in this study (Figure 3). Interestingly, significant thrombin-induced reductions in transendothelial resistance are observed earlier than increases in transendothelial albumin clearance (25). Transendothelial electrical resistance, a determination of the physical ability of the endothelium to form an electrically insulating layer, is dependent on the integrity of cell-cell and cell-matrix attachments and the closeness of adhesion between the monolayer and the microelectrode via focal contacts (41). Thus, the qualitative differences between the time course of thrombin- induced changes in electrical resistance and albumin clearance may reflect the early reorganization of the cytoskeleton, and concomitant alterations of focal contacts demonstrated in Figures 1 and 2.

Our data confirm that vinculin is an in vitro substrate for phosphorylation by EC PKC (42, 43), an important effector rapidly activated by thrombin (44). Because the  isotype of PKC colocalizes within focal adhesion plaques in other cellular types (26), and because changes in the levels of vinculin and talin phosphorylation occur in response to a variety of agonists (including phorbol esters) have been documented (45), we had speculated that focal adhesions contain substrates for PKC that are important in regulating the EC cytoskeleton. However, while the data indicate that both vinculin and talin are phosphorylated in situ under basal conditions, the level of phosphorylation was not further altered when ECs were stimulated with thrombin despite dramatic cytoskeletal reorganization. Although it is possible that phosphorylation of selective vinculin and talin isotypes may have been enhanced, several isotypes of vinculin are expressed in platelets without evidence of differential subcellular distribution into the cytoskeleton (50). The lack of a consistent association between vinculin or talin phosphorylation and focal adhesion or cytoskeletal rearrangement indicates that the phosphorylation of vinculin and talin in ECs may not have a direct role in regulating cytoskeletal organization.

Focal adhesion plaque proteins also contain important sites for tyrosine phosphorylation. For example, cells that are transformed with Rous sarcoma virus demonstrate p60^v-src-mediated tyrosine phosphorylation of both vinculin and talin (15). Targeting of p60^v-src to focal adhesion plaques was sufficient to induce a transformed phenotype, indicating that key substrates for tyrosine phosphorylation that regulate cell shape are present at these sites (51). When the EC focal adhesion plaque proteins vinculin and talin were immunoblotted from thrombin-stimulated ECs, neither was significantly phosphorylated on tyrosine residues in ECs in the basal state, and neither vinculin nor talin demonstrated further increases in tyrosine phosphorylation after thrombin. Thus, the lack of thrombin-induced tyrosine phosphorylation of vinculin and talin implies that the mechanisms of contractile agonist-mediated cytoskeletal rearrangement and force generation between ECs and tracheal smooth muscle may be distinct.

The tyrosine kinase p125^FAK colocalizes with focal adhesion plaques and is an early substrate for tyrosine phosphorylation in endothelial cells adhering to an extracellular matrix (52). Because the extracellular matrix composition is capable of affecting both EC permeability and cytoskeletal organization (53, 54), FAK may have a major role in transducing signals from the extracellular environment that consequently affect the F-actin cytoskeleton and EC permeability. FAK phosphorylation has been linked not only to integrin-mediated extracellular binding events, but also to stimulation with a variety of agonists including thrombin, bombesin, and cytokines (30, 55). Data collected by using bovine ECs indicate the ability of the thrombin-signaling cascade to result in an increase in the phosphotyrosine content of FAK. This finding was also seen with the proteolytically inactive thrombin receptor agonist TRAP, and with thrombin treatment followed by addition of hirudin to EC detergent lysates. The increase in FAK phosphorylation induced by TRAP was much weaker than that produced by thrombin challenge in these experiments. This may be explained by TRAP's weaker agonism of the thrombin receptor, as concentrations 100-fold or higher are required to elicit functional changes in thrombin receptor activation (10). Because the kinetics of this event matches the evolution of thrombin-induced EC barrier dysfunction, these data indicate a potential link between the phosphorylation of a major regulatory enzyme present within focal adhesion plaques and a well-defined EC physiologic event.

In an attempt to relate FAK phosphorylation to Ca²⁺-dependent cellular events, the Ca²⁺ ionophore ionomycin was used to increase the intracellular Ca²⁺ concentration. Ionomycin produces EC permeability in a dose- and time-dependent fashion that occurs independently of MLCK activation (14). Whereas thrombin produced a time-dependent increase in the immunoreactive PTyr/FAK ratio in these studies, ionomycin produced the opposite effect (Figure 8), inducing a rapid decrease in this ratio. As both agonists result in significant EC barrier dysfunction, this divergence complicates the analysis of a consistent relationship between EC barrier function and the tyrosine phosphorylation status of FAK. Thus, the observed cytoskeletal changes in thrombin-stimulated ECs may occur passively in response to the axial force generated by activation of the MLCK-actin-myosin system, or they may be orchestrated by other actin-binding proteins that are not localized in focal adhesion plaques. Nevertheless, considering the model of dual regulation of EC permeability, barrier dysfunction in thrombin-induced EC permeability appears to be potently determined by the contractile activity of the EC actomyosin system as the major effector mechanism, as suggested in studies of the role of MLCK activation and myosin 20-kD regulatory light chain phosphorylation in this system (7). Thus, the role of thrombin-induced FAK phosphorylation in the evolution of EC barrier dysfunction is unclear and may range from direct participation in the permeability response to providing a signal for limiting the duration of paracellular permeability evoked by thrombin. Investigation of additional mechanisms of cellular tethering in ECs, such as adherens junctions, may further define the role of tethering force disruption in the model of thrombin-induced EC barrier dysfunction. Because ionomycin produces EC barrier dysfunction in an MLCK-independent manner, ionomycin-induced dephosphorylation of FAK may be a relevant mechanism in modulating EC cytoskeletal organization and tethering, and requires further study.