Thrombin Inactivates Myosin Light Chain Phosphatase via Rho and Its Target Rho Kinase in Human Endothelial Cells*

The role of Rho GTPase and its downstream targets Rho kinase and myosin light chain phosphatase in thrombin-induced endothelial cell contraction was investigated. The specific Rho inactivator C3-transferase from Clostridium botulinum as well as microinjection of the isolated Rho-binding domain of Rho kinase or active myosin light chain phosphatase abolished thrombin-stimulated endothelial cell contraction. Conversely, microinjection of constitutively active V14Rho, constitutively active catalytic domain of Rho kinase, or treatment with the phosphatase inhibitor tautomycin caused contraction. These data are consistent with the notion that thrombin activates Rho/Rho kinase to inactivate myosin light chain phosphatase in endothelial cells. In fact, we demonstrate that thrombin transiently inactivated myosin light chain phosphatase, and this correlated with a peak in myosin light chain phosphorylation. C3-transferase abolished the decrease in myosin light chain phosphatase activity as well as the subsequent increase in myosin light chain phosphorylation and cell contraction. These data suggest that thrombin activates the Rho/Rho kinase pathway to inactivate myosin light chain phosphatase as part of a signaling network that controls myosin light chain phosphorylation/contraction in human endothelial cells.

A variety of pathological conditions including the early stages of atherosclerosis, acute inflammation, and anaphylactic shock are associated with increased vascular permeability (1,2). Thrombin generated under these pathological conditions induces endothelial cell contraction and increases vascular permeability through activation of a specific receptor that is coupled via heterotrimeric G-proteins of the Gq family to phospholipase C␤ that cleaves phosphatidylinositol-4,5-bisphosphate to yield inositol 1,4,5-trisphosphate. Inositol 1,4,5-trisphosphate 3 mobilizes Ca 2ϩ from intracellular stores and thus increases intracellular Ca 2ϩ concentration (3)(4)(5). It is well established that the thrombin-induced increase in intracellular Ca 2ϩ con-centration leads to activation of Ca 2ϩ /calmodulin-dependent myosin light chain kinase (MLCK), 1 which phosphorylates Thr-18 and Ser-19 of the light chain of myosin II (MLC) (3). Phosphorylation induces a conformational change in MLC that enables actin-myosin interaction and activates the Mg 2ϩ -ATPase activity of myosin (6). Besides MLC kinases, myosinassociated MLC phosphatase (PP1M) also seems to play a major role in the control of MLC phosphorylation/dephosphorylation in endothelial cells. This is demonstrated by the finding that pharmacological inhibitors of protein phosphatase 1 (PP1) increased MLC phosphorylation and cell contraction (7), whereas microinjection of active PP1 decreased MLC phosphorylation and disturbed actin/myosin interaction (8). PP1 is composed of three components, a 37-38-kDa catalytic subunit, a 130-kDa regulatory subunit, and a 20-kDa subunit (9 -11). Recently, it was shown that the regulatory subunit can be phosphorylated and inactivated by Rho kinase, a specific target protein of the GTPase Rho (12). Previous work had indicated that Rho mediates cell contraction induced by thrombin, but the Rho targets involved have not been identified thus far (13)(14)(15).
The Rho family of Ras-like GTPases that consists of more than 10 members has been implicated in actin cytoskeleton organization and cellular shape changes in a variety of cell types (16,17). Rho-GTPases cycle between a GTP-bound active state and a GDP-bound inactive state, and this cycle is controlled by guanine nucleotide exchange factors and GTPaseactivating proteins (18 -20). ADP-ribosylation and inactivation of Rho by C3-transferase from Clostridium botulinum specifically inhibits the cellular effects of Rho (21)(22)(23)(24)(25)(26)(27)(28). Microinjection of constitutively active V14Rho induced formation of stress fibers and focal adhesion sites as well as a contractile phenotype in fibroblasts (21,28). It has also been reported that GTP␥S-and aluminum flouride-mediated Ca 2ϩ sensitation of smooth muscle contraction is mediated by Rho (29). A number of target proteins that interact with GTP-bound but not with GDP-bound Rho have been identified (16). These include the closely related Ser/Thr kinases ROK␣/Rho kinase and ROK␤/ p160 (30,31,(33)(34)(35)(36)(37). ROK␣/Rho kinase consists of multiple domains including a catalytic domain at the amino terminus, a coiled coil domain including the Rho binding domain (RBD), and a C-terminal pleckstrin homology domain (34). Microinjection of isolated Rho kinase domains into fibroblasts or HeLa cells indicated that Rho kinase is the target protein by which Rho forms stress fiber and focal adhesions (36,37). Consistent with the involvement of Rho kinase in contractile events, overexpression of constitutively active V14Rho in fibroblasts caused phosphorylation of the MBS, inactivation of myosin phosphatase, and an increase in MLC phosphorylation (12).
Here, we provide evidence that thrombin uses the Rho/Rho kinase pathway to inactivate PP1M in human endothelial cells. Inactivation of PP1M seems to be coordinated with activation of Ca 2ϩ -calmodulin-dependent MLCK to maximally increase MLC phosphorylation in the early phase of thrombin-induced endothelial cell contraction.

EXPERIMENTAL PROCEDURES
Materials-The inhibitors okadaic acid, KT5926, and tautomycin were from Calbiochem (Bad Soden, Germany); all other materials not specifically indicated were from Sigma.
Cell Culture-Human umbilical vein endothelial cells (HUVEC) were obtained and cultured as described previously (24). Briefly, cells harvested from umbilical cords were plated onto collagen-coated (24 h, 100 g/ml collagen G; Biochrom, Berlin, Germany) plastic culture flasks and cultured in endothelial growth medium (Promo Cell, Heidelberg, Germany), containing endothelial cell growth supplement/heparin and 10% fetal calf serum. For all experiments, cells were plated at a density of 2 ϫ 10 4 cells/cm 2 and grown to confluency for 10 days. Confluent monolayers were stimulated for the indicated time periods with ␣-thrombin from human plasma (Boehringer, Mannheim, Germany).
Measurement of Endothelial Permeability-Horseradish peroxidase diffusion through HUVEC monolayers was determined as described previously with some modifications (38). Briefly, cells were plated (2 ϫ 10 4 cells/cm 2 ) on collagen-coated polyethylene terephtalate cell culture inserts (3-m pore size, Becton Dickinson), which were set into 24-well Falcon companion TC plates (Becton Dickinson). HUVEC were cultured for 10 days with medium changes every 2 days. For thrombin stimulation, medium was replaced with 500 l of culture medium containing thrombin. For controls, thrombin was omitted, but otherwise cells were treated identically. After 15 min of stimulation, 500 l of medium was filled into the lower compartment, and the medium in the upper compartment was replaced with fresh medium containing horseradish peroxidase (0.34 mg/ml, IV-A type, 44,000 M r ; Sigma, Deisenhofen, Germany). After 1 min, 60 l of medium was collected from the lower compartment and mixed with 860 l of reaction buffer (50 mM NaH 2 PO 4 , 5 mM Guaiacol) and 100 l of freshly made H 2 O 2 solution (0.6 mM in H 2 O). The reaction was allowed to proceed for 15 min at room temperature, and absorbance was measured at 470 nm.
Immunofluorescence-For fluorescence staining, HUVEC were plated (2 ϫ 10 4 cells/cm 2 ) on Eppendorf Cellocate glass coverslips (Eppendorf, Hamburg, Germany) coated with 100 g/ml collagen G (24 h) and grown to confluency for 10 days. To label F-actin, cells were fixed for 10 min with 3.7% formaldehyde solution in PBS containing 1 mM Ca 2ϩ and 1 mM Mg 2ϩ , permeabilized for 5 min in cold acetone (Ϫ20°C), and air dried. Coverslips were then incubated for 20 min with rhodamine phalloidin (Molecular Probes; 1/20 in PBS) and mounted in Mowiol (Calbiochem, Bad Soden, Germany) containing 0.2% p-phenylenediamine (Sigma, Deisenhofen, Germany) as anti-fading agent. All steps were performed at room temperature with three washes in PBS/2% BSA between antibody incubations. Fluorescence microscopy was performed with a Leica RDM 3 microscope, and microphotographs were recorded on Kodak T-Max 400 film.
Recombinant Proteins-Recombinant C3-transferase and V14 RhoA and RBD were expressed as glutathione S-transferase fusion proteins in Escherichia coli and purified on glutathione-Sepharose beads as described (21,39). The fusion proteins were cleaved by thrombin. Thrombin was removed by incubation for 2 h with benzamidine beads, and the released proteins were concentrated and dialyzed against microinjection buffer (see below). Purity and complete removal of thrombin was checked by SDS-PAGE and Coomassie staining. Extracellular application of V14Rho even at amounts exceeding the microinjected volume by 9 orders of magnitude (10 Ϫ15 versus 10 Ϫ6 liters) did not produce any thrombin-like effect, excluding thrombin contamination of recombinant protein. Protein concentrations were determined with the BCA protein assay kit (Pierce) using BSA as a standard. As tested by SDS-PAGE and Coomassie staining, protein preparations showed essentially only one band. Rho kinase was expressed in SF9 cells as described previously (37).
The microinjected volume was about 1-3 ϫ 10 Ϫ15 liters per cell. PP1 catalytic subunit (␣-isoform; Calbiochem, Bad Soden, Germany) was diluted with phosphatase buffer (100 mM K ϩ glutamate, 39 mM K ϩ citrate, pH 7.3) and was injected at a concentration of 200 units/ml. Control injections carried out with microinjection or phosphatase buffer, respectively, did not produce any significant effect on cell morphology or actin organization. After injection, cells were returned to the incubator for 30 min. Injected cells were identified by labeling a coinjected marker protein (5 mg/ml rat IgG) with fluorescein isothiocyanate-conjugated goat anti-rat IgG (Dianova, Hamburg, Germany) and by relocation on the Cellocate coverslips using the microgrid. For each experiment, about 100 cells were injected and examined by fluorescence microscopy.
Myosin Light Chain Phosphorylation-MLC phosphorylation was analyzed by urea PAGE separation of the mono-and diphosphorylated forms as described in detail elsewhere (40). HUVEC were stimulated with thrombin as indicated, and reaction was terminated by immediate addition of 1.5 ml of ice-cold 10% trichloroacetic acid. Cells were scraped and then centrifuged for 20 min at 14,000 ϫ g. Supernatants were discarded, and pellets were washed with ddH 2 0 to remove trichloroacetic acid and resolved in 1.5 ml of sample buffer (6.7 M urea, 20 mM Tris, 22 mM glycine, 10 mM dithiothreitol, pH 9.0). Samples (75 g per lane) were applied to urea gel electrophoresis (top gel 3.5% acrylamide, bottom gel 10% acrylamide) and run at 9 mA for approximately 45 min until the marker dye had come out of the bottom gel. The gels were stained for 1 h with 0.05% Coomassie Brilliant Blue 250, 10% acetic acid, 30% methanol, 60% H 2 O. The stoichiometry of MLC phosphorylation (mol phosphate/mol MLC) was determined by densitometric analysis of the wet gels, using a Sharp XL-325 densitometer and Pharmacia Image Master software, and calculated using the formula P 1 ϩ 2xP 2 /P 0 ϩ P 1 ϩ P 2 .
Measurement of Cytosolic Ca 2ϩ Concentration-Cytosolic [Ca 2ϩ ] i was measured as described previously (41). Cells grown on collagen-coated glass coverslips were loaded for 30 min at 37°C with fura-2 AM (2.5 M; Calbiochem, Bad Soden, Germany) and resuspended in HEPES buffer (20 mM HEPES, 120 mM NaCl, 2.7 mM KCl, 1.4 mM MgSO 4 , 0.5 mM CaCl 2 , 1.4 mM KH 2 PO 4 , 25 mM NaHCO 3 , 10 mM glucose, pH 7.4). Ca 2ϩ -dependent fluorescence was measured using a double excitation spectrofluorimeter (AMKO Light Technology, Tornesch, Germany) with the emission wavelength set at 510 nm and the emission wavelength rapidly alternating between 340 and 380 nm. [Ca 2ϩ ] i was quantified applying the equation Preparation of Myosin-enriched Cell Fractions-Myosin-enriched fractions of HUVEC were prepared as described previously (7). Briefly, HUVEC were plated on collagen-coated 100-mm plates (Falcon) and cultivated for 10 days. Monolayers were washed two times with ice-cold PBS (Sigma), and 200 l of homogenization buffer containing 50 mM Tris-aminomethane, pH 7.5, 0.1 mM EDTA, 28 mM ␤-mercaptoethanol, and 1 g/ml each of leupeptin, pepstatin, pefabloc, and aprotinin as protease inhibitors (1 g/ml each) were added to the cells. Plates were immediately cooled down to Ϫ80°C, scraped with a rubber policeman, and homogenized by passing the suspension several times through a syringe. Homogenates were then treated with high salt buffer (0.6 M NaCl, 0.1% Tween 20), containing protease inhibitors (1 g/ml of each) leupeptin, pepstatin, and pefabloc for 1 h at 4°C and subsequently centrifuged at 4500 ϫ g for 30 min at 4°C. The supernatant was diluted 10-fold with assay buffer (50 mM Tris, 0.1 mM EDTA, 28 mM ␤-mercaptoethanol, pH 7.0) and centrifuged at 10,000 ϫ g at 4°C. The resulting pellet was resolved in 10 l of high salt buffer. This myosin-enriched cell fraction contains only PP1 and essentially no PP2 activity (7). To inhibit residual PP2 activity, the myosin-enriched cell fractions were supplemented with okadaic acid (1 nM) to completely inhibit PP2 (IC 50 0.1 nM), whereas PP1 is only affected at concentrations 100-fold higher (IC 50 10 nM).
Measurement of Myosin-associated Phosphatase Activity-For measuring myosin-associated phosphatase activity in myosin-enriched cell fractions, we used the protein phosphatase assay system (Life Technologies, Inc.) according to the instructions of the manufacturer. This assay system is based on the method described by Cohen et al. (42). Briefly, phosphorylase b (0.1 mM) was in vitro phosphorylated by phosphorylase kinase (0.1 mg/ml) in the presence of [␥-32 P]ATP (5 mCi/ml) in phosphorylation buffer (250 mM Tris-HCl, pH 8.2, 16.7 mM MgCl 2 , 1.67 mM ATP, 0.83 mM CaCl 2 , 133 mM 2-mercaptoethanol) for 1 h at 30°C. Reaction was stopped with 90% ammonium persulfate solution (4°C). Reaction tube was then kept on ice for 1 h and subsequently centrifuged at 12,000 ϫ g for 10 min. The resulting protein pellet was resuspended with ammonium persulfate solution (45% saturated). The protein pellet was washed four times in this way. Protein solution was then concentrated to a final concentration of 3 mg/ml using Amicon Centricon-30 concentrators.
Phosphatase activities of myosin-enriched cell fractions were then quantified by measuring release of radioactivity from [ 32 P]phosphorylase a in the presence of myosin-enriched cell fractions. For this purpose, myosin-enriched fractions were diluted with 30 l of assay buffer (50 mM Tris, 0.1 mM EDTA, 28 mM ␤-mercaptoethanol, 6.25 mM caffeine, pH 7.0) and mixed with 20 l of radioactive phosphatase substrate. Reaction was allowed to proceed for 10 min at 30°C and stopped with ice-cold 20% trichloroacetic acid. Samples were then incubated on ice for 10 min and centrifuged at 12,000 ϫ g for 3 min. The radioactivity released in the supernatant was measured using a Wallack 1410 liquid scintillation counter. To prove that phosphorylase b dephosphorylation in fact measures MLC-phosphatase, we also used smooth muscle 32 P-MLC as a substrate (7,9). MLC (0.8 mg/ml) was phosphorylated for 1 h at room temperature with MLCK (50 g/ml) in buffer containing 50 mM Tris-HCl, pH 7.4, 5 mM magnesium acetate, 0.1 mM CaCl 2 , 0.1 mg/ml calmodulin, 0.35 M NaCl, 50 M [ 32 P]ATP, passed through 2 ϫ 1-ml spin columns with Sephadex G-25 (Pharmacia, Uppsala, Sweden), equilibrated with assay buffer, and dialyzed for 18 h at 4°C with 2 ϫ 5 liters of assay buffer.
Western Blot Analysis of PP1/MBS Load of Myosin-enriched Cell Fractions-The PP1 and MBS load in myosin-enriched cell fractions was tested by Western blot analysis. Briefly, myosin-enriched cell fractions were dissolved in Laemmli buffer containing protease inhibitors (1 g/ml leupeptin, pepstatin, aprotinin) and applied to SDS, 10% PAGE. Proteins were then blotted onto polyvinylene difluoride membranes for 1 h at 80 V. Membranes were blocked with 10% low-fat milk dissolved in Tris-buffered saline. PP1 was detected using rabbit polyclonal anti-PP1 antibody (Upstate Biotechnology, Lake Placid, NY) or rabbit polyclonal anti-MBS antibody (12), both diluted 1/1000 in Tris-buffered saline, and horseradish peroxidase-labeled secondary antibody (Amersham, Braunschweig, Germany), diluted 1/4000 in Tris-buffered saline, developed with enhanced chemiluminescence reagent (Amersham), and then exposed to hyperfilm ECL (Amersham).

Inactivation of Rho by C3-transferase Blocks Thrombin-induced Endothelial
Cell Contraction-To test whether Rho is involved in endothelial cell contraction, we measured the thrombin-stimulated increase of transendothelial horseradish peroxidase diffusion in the absence and presence of the specific Rho inactivator C3-transferase from C. botulinum (45,46). The results presented in Fig. 1 demonstrate that confluent monolayers of HUVEC show a low transendothelial diffusion of horseradish peroxidase. Stimulation with thrombin (0.1-1 unit) dose dependently increased horseradish peroxidase permeability up to 10-fold. The thrombin-induced increase in permeability could almost completely be abolished by pretreatment of the cells with C3-transferase (24 h, 5 g/ml). To confirm that the underlying mechanism for this C3-transferase effect was inhibition of cell contraction, we performed actin staining. As can be seen in Fig. 2a, actin is mainly concentrated in a dense peripheral band along cell-cell contacts in intact endothelial monolayers. When stimulated with thrombin, cells contracted, expressed actin fibers, and formed numerous intercellular gaps (Fig. 2b). In cells pretreated with C3-transferase, these thrombin effects were abolished, and cells remained flat and spread out (Fig. 2c). The actin band at cell-cell contacts was thinned out by the C3 treatment but remained essentially intact. These results suggest that activation of Rho is a key mechanism in thrombin-induced endothelial cell contraction/ increase of endothelial permeability. The Rho subtype involved likely is RhoA because we recently found that RhoA is by far the predominant substrate of C3-transferase in HUVEC and is ADP-ribosylated to about 70 -80% under the conditions employed (27). Consistent with this, microinjection of constitutively active V14RhoA produced cell contraction and stress fiber formation similar to thrombin (Fig. 2f).
Evidence for the Involvement of Rho Kinase and Myosin PP1M in Thrombin-induced Cell Contraction-To obtain evidence that the Rho target protein Rho kinase mediates the thrombin/Rho effect on contraction, we microinjected isolated RBD of Rho kinase, which has been shown to inhibit interaction of Rho with Rho kinase (37). Microinjection of RBD blocked thrombin-induced cell contraction similar to C3-transferase (Fig. 2d). Conversely, when we microinjected the recombinant catalytic domain of Rho kinase, endothelial cells contracted and showed a shape change (Fig. 2g) similar to stimulation with thrombin ( Fig. 2b), microinjection of V14RhoA (Fig. 2f), or treatment with the PP1M inhibitor tautomycin (not shown). These data suggest that Rho kinase is the Rho target protein by which thrombin exerts its effects on cell contraction. To investigate whether inhibition of the Rho kinase target protein PP1M contributes to thrombin-induced endothelial cell contraction, we microinjected the constitutively active catalytic domain of PP1. As can be seen in Fig. 2e, microinjection of PP1 inhibited the thrombin-induced cell contraction similar to C3transferase (Fig. 2c) or RBD (Fig. 2d). These data are consistent with the idea that thrombin uses a pathway that involves activation of Rho and Rho kinase as well as inactivation of PP1M to regulate cell contraction.
Thrombin Inactivates PP1M Activity via Rho-We reasoned that if Rho kinase is in fact activated by thrombin, an inhibition of PP1M activity should be detected. We therefore determined PP1M activity in cells stimulated with thrombin for different time periods by assaying dephosphorylation of phosphorylase b (9). The results presented in Fig. 3 clearly show that thrombin produced a transient decrease of PP1M activity between 30 s and 3 min, which was followed by a return to base-line values after 5 min (Fig. 3a). In the C3-transferasetreated cells, the thrombin-induced decrease in PP1M activity was abolished, further supporting the notion that Rho regulates PP1M. To unambiguously demonstrate that [ 32 P]phos-  2. Thrombin-induced endothelial cell contraction is mediated by Rho, Rho kinase, and PP1. HUVEC were not stimulated (a), stimulated with thrombin (1 unit/ml; 15 min) (b), pretreated with C3-transferase (24 h, 5 g/ml), and then stimulated with thrombin (c), phorylase b dephosphorylation reflects MLC-phosphatase (PP1C) activity, we also used 32 P-MLC as a substrate. The inset in Fig. 3a shows that PP1C activity determined with 32 P-MLC as substrate was essentially identical to the PP1C activity determined with [ 32 P]phosphorylase b as substrate. Furthermore, we tested by Western blot whether thrombin caused dissociation of the catalytic subunit (PP1C) or the regulatory subunit (MBS) of PP1M from the myosin-enriched fractions. The data in Fig. 3c show that PP1C and MBS did not dissociate from myosin, suggesting that the MLC-phosphatase activity measured was because of the holoenzyme.
In addition, we reasoned that if C3-transferase inhibits thrombin effects by preventing the decrease of PP1M activity, then external inhibition of the PP1M should restore thrombininduced endothelial contraction. In fact, the PP1M inhibitor tautomycin could completely restore the block in thrombininduced endothelial cell contraction caused by the C3-transferase treatment (Fig. 4). For this experiment, we chose a tautomycin concentration (6 nM) that by itself did not induce cell contraction (Fig. 4). Interestingly, C3-transferase not only abolished the thrombin-induced decrease of PP1M activity but also produced a dose-dependent (0 -10 g/ml) increase in PP1M activity in unstimulated endothelial cells, suggesting that there is a basal pool of active Rho/Rho kinase (Fig. 5). We previously reported that in HUVEC the degree of in situ ADPribosylation of Rho is directly proportional to C3-transferase concentrations ranging from 0.3 to 10 g/ml (27). Taken together, these data indicate that when endothelial cells are stimulated with thrombin, Rho is activated and interacts with Rho kinase, which inhibits PP1M.
Inhibition of Rho Blocks Thrombin-stimulated MLC Phosphorylation-Phosphorylation of the light chain of myosin II (MLC) is a crucial mechanism by which thrombin signals are converted into the mechano-chemical force for cell contraction (47,48). To investigate whether the observed thrombin-induced inhibition of PP1M is relevant for MLC phosphorylation, we stimulated control and C3-pretreated (24 h, 5 g/ml) endothelial cells for different times with thrombin (1 unit/ml) and separated un-, mono-, and diphosphorylated MLC on 10% urea polyacrylamide gels. Densitometric quantitation of these MLC forms revealed that thrombin caused phosphate incorporation into MLC with a peak after 1 min (Fig. 3c). The level of phosphorylated MLC then dropped to a plateau above base line between 5 and 15 min, indicating that MLC is partly dephosphorylated after the initial peak. The peak in phosphorylation exactly correlated with maximal PP1M inhibition, whereas the drop to plateau phosphorylation correlated with the increase of PP1M activity back to base line. In cells pretreated with C3, the thrombin-stimulated peak in MLC phosphorylation was essentially abolished (Fig. 3c). These data suggest that inhibition of PP1M activity via Rho/Rho kinase is an essential mechanism by which thrombin yields a peak level in MLC phosphorylation.
Rho Is Not Involved in Thrombin-stimulated Ca 2ϩ Mobilization in Endothelial Cells-It is well established that thrombin elevates intracellular Ca 2ϩ concentration and thereby activates Ca 2ϩ /calmodulin-dependent MLCK (48). Interestingly, C3-transferase was shown to inhibit thrombin-stimulated Ca 2ϩ mobilization in fibroblasts (32). To test whether this is the mechanism by which C3-transferase prevents thrombin-induced MLC phosphorylation in HUVEC, we loaded control or C3-treated endothelial cells with the Ca 2ϩ indicator fura-2 AM and determined cytosolic-free Ca 2ϩ concentration using fluorescence spectrometry. We found that the C3 treatment affected neither basal Ca 2ϩ concentration nor the thrombin-stimulated increase in peak (after 30 s) or plateau (after 3 min) Ca 2ϩ concentration (Fig. 6). We conclude that Rho is not involved in the thrombin-induced cytosolic Ca 2ϩ increase in endothelial cells.
C3-transferase Does Not Prevent Thrombin-induced Release of Catenins from the Cytoskeleton-In tightly confluent HU-VEC, the VE-cadherin/catenin-based adherens junctions are associated with the actin cytoskeleton to stabilize the endothelial barrier. It has been suggested that thrombin-induced increase in endothelial permeability might be partly because of a release of catenins from the Triton X-100 insoluble cytoskeletal cell fraction (43,44). To exclude that the effect of C3-transferase on thrombin-induced increase in endothelial permeability was the result of prevention of this shift, we performed detergent solubility assays of catenins. As shown in Fig. 7, we found that plakoglobin and ␤-catenin are associated with the Triton X-100 insoluble fraction in confluent HUVEC. After thrombin treatment, both ␤-catenin and plakoglobin (␥-catenin) lost their association with the cytoskeleton and shifted to the cytoplasm. This shift was not prevented by C3-transferase treatment. This result indicates that prevention of the release of catenins from the cytoskeleton is not the mechanism by which C3-transferase inhibits thrombin-induced increase in endothelial permeability.

DISCUSSION
Our data suggest an additional signal pathway by which thrombin regulates myosin light chain phosphorylation and the subsequent increase in endothelial cell contraction/vascular permeability. We propose that thrombin activates Rho, which then interacts with its target Rho kinase that in turn inactivates PP1M, most likely by phosphorylation of the 130-kDa regulatory subunit (12). The transient inactivation (within 0.5-5 min) of PP1M demonstrated here most likely produces the peak in MLC phosphorylation seen within the first 5 min of thrombin stimulation. Others have found a similar peak of MLC phosphorylation in thrombin-stimulated cells (49). Interestingly, the peak in MLC phosphorylation also correlates with the transient peak in intracellular Ca 2ϩ elevation obtained after thrombin stimulation (41). After 5 min of thrombin stimulation, MLC phosphatase activity reversed to near base-line values and, in parallel MLC phosphorylation, dropped to a plateau. This plateau MLC phosphorylation correlates with a plateau in intracellular Ca 2ϩ concentration (41). Taken together, these data suggest that the Rho-induced inhibition of MLC phosphatase activity is coordinated with a peak in Ca 2ϩ mobilization to produce maximal MLC phosphorylation. Along this line, it has been reported that GTP␥S-and aluminum flouride-mediated Ca 2ϩ sensitation of smooth muscle cell contraction is dependent on Rho (29). In these smooth muscle cells, myosin light chain phosphatase was inhibited by a Tritonsoluble membrane-bound effector, which was not Rho. This effector could be Rho kinase.
We demonstrated that C3-transferase could completely block thrombin-stimulated MLC phosphorylation and cell contraction as well as the increase in endothelial permeability. We want to emphasize that this does not contradict the idea that Rho-induced inhibition of MLC phosphatase is mainly responsible for the transient peak in MLC phosphorylation. As shown microinjected (asterisks) with Rho binding domain of Rho kinase and then stimulated with thrombin (d), microinjected (asterisks) with PP1 catalytic subunit and then stimulated with thrombin (e), microinjected (arrow) with V14RhoA (f) or recombinant Rho kinase (arrows, g). Cells were stained for actin using rhodamine-phalloidin. Representative experiments are shown. Bar represents 30 m.
in Fig. 5, C3-transferase by itself increased MLC phosphatase activity in unstimulated cells, most likely by inhibiting a basal Rho activity. This somewhat artificially elevated MLC phosphatase activity could blunt MLC kinase activity brought about by the thrombin-induced Ca 2ϩ signal. The fact that the MLC phosphatase inhibitor tautomycin completely reversed the C3transferase effect on cell contraction also supports this idea.
Besides phosphorylating and inactivating PP1M, it was demonstrated that Rho kinase can directly phosphorylate MLC in vitro, i.e. can act as a MLC kinase (50). At present, we have no indication that this mechanism is relevant in endothelial cells.
Recently, it was reported that C3-transferase inhibited lysophosphatidic acid-induced MLC phosphorylation and contraction in fibroblasts. It was speculated that this inhibition is because of enhanced myosin phosphatase activity (13). Our results obtained in thrombin-stimulated endothelial cells seem to support this notion. Furthermore, we noticed that contraction induced by thrombin, V14Rho, or active Rho kinase precedes formation of stress fibers, 2 which is consistent with the idea that contraction drives stress fiber formation (13). We noticed, however, that stress fibers were not as efficiently pro-  3). b, HUVEC were pretreated without (Ϫ) or with (ϩ) 5 g/ml C3-transferase (C3) from C. botulinum for 24 h and stimulated with thrombin (1 unit/ml) for different time periods. Un-(P 0 ), mono-(P 1 ), and di-(P 2 )-phosphorylated MLCs were separated by urea gel electrophoresis and Coomassie stained. Specific bands were quantified by densitometric analysis. Stoichiometry phosphate incorporation was calculated as described under "Experimental Procedures." Results are representative of three experiments Ϯ S.E. Thrombin maximally increased MLC phosphorylation within the first 2 min. This peak in MLC phosphorylation was prevented by C3-transferase. (c) The protein level of the catalytic subunit (PP1C) and the regulatory subunit (MBS) of PP1C in the myosin fractions was determined by Western blot. A representative experiment is shown. duced by Rho kinase as by V14Rho, indicating that additional Rho targets contribute to efficient stress fiber formation.
Endothelial cells flatten and spread out when Rho is inactivated and contract when Rho is activated. A similar behavior has been found in neuronal cells (15), human and mouse macrophages (51,52), and HeLa cells (36). In contrast, other cells including NIH 3T3 fibroblasts and Vero cells round up when Rho is inactivated and spread out when it is activated (53,54). Presumably, this behavior depends on the relative importance of Rho-dependent focal adhesion/integrin cluster formation versus Rho-dependent contractility in the respective cell type.
In fibroblasts, Rho seems to directly trigger Ca 2ϩ mobilization, most likely by providing phosphatidylinositol 4,5-bisphosphate through stimulation of a PI(5) kinase activity (32). The reason why we did not find an effect of Rho inhibition on Ca 2ϩ mobilization in endothelial cells might lie in the recruitment of different target proteins by Rho, depending on the cell type (16). Fig. 8 depicts the presumptive signal pathway by which thrombin induces cell contraction. The Ca 2ϩ -triggered activation of MLCK induces in concert with inhibition of MLC phosphatase by Rho/Rho kinase an increase in MLC phosphorylation and finally cell contraction.
The mechanism by which Rho is activated by thrombin in HUVEC remains to be determined. One possibility is that thrombin activates Rho via G 13 and the epidermal growth factor receptor tyrosine kinase, as was recently shown for lysophosphatidic acid in fibroblasts (55).
Here, we describe a pathway involving Rho/Rho kinase by which thrombin inactivates PP1M and thus controls MLC phosphorylation and contraction in human endothelial cells. This pathway is coordinated with the well established Ca 2ϩ / calmodulin-dependent pathway of MLC kinase activation and presumably with yet another pathway regulating adherens junction disassembly. A complex signaling network of thrombin-controlled increase in endothelial permeability is evolving.