INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE VASCULAR ENDOTHELIUM ACTS as a selective barrier between the vascular space and underlying tissues. Compromise of endothelial cell (EC) barrier integrity leads to an increase in vascular permeability, a cardinal feature of inflammation resulting in tissue edema, hypoxemia, and, often, increased morbidity and mortality. Substantial work, including work from our own laboratory (23), has verified that the EC barrier is regulated by contractile and tethering mechanisms, the effects of which are critically dependent on EC cytoskeletal components (microfilaments, intermediate filaments, and microtubules). For example, our laboratory has previously shown (21, 58) that specific edemagenic agents such as the serine protease thrombin induce EC barrier dysfunction, primarily via actomyosin-driven contraction initiated by myosin light chain (MLC) phosphorylation and tightly linked to microfilament reorganization. Thrombin increases Ca²⁺ and EC centripetal tension (41, 46), and 20-kDa MLC (MLC₂₀) phosphorylation peaks at 2 min and returns nearly to control levels by 60 min, indicating involvement of both MLC kinase (MLCK) and MLC phosphatase (PPase) activity in the thrombin response (21, 57). At least two responsible classes of MLCKs are potentially able to phosphorylate MLC in vivo (4, 21, 29), including the newly cloned endothelial Ca²⁺/calmodulin (CaM)-dependent MLCK (22) and Rho kinase, the activity of which depends on activation of the small G protein Rho.

It has become well established that Ras-related GTPases of the Rho family organize the actin cytoskeleton and regulate focal adhesion formation. Rho GTPases are inactive in the GDP-bound form and are activated by GDP/GTP exchange (17, 56). Rho can be specifically inactivated by bacterial toxins by either ADP-ribosylation (Clostridium botulinum C3 exotoxin) or glucosylation (Clostridium difficile, toxin B), which presumably block the interaction of Rho with downstream targets like Rho kinase (1, 2, 50). There is a significant controversy, however, as to the effect of Rho on EC contractile and barrier properties. For example, C3 exotoxin has been suggested to both attenuate EC thrombin-mediated contraction via reduction of the thrombin-induced decrease in myosin-specific PPase activity (20) and to disrupt EC barrier properties, with minimal effect on EC contractile properties (61). Our laboratory (26) has recently shown that C3 exotoxin completely abolished thrombin- and diperoxovanadate (DPV)-induced increases in MLC phosphorylation, suggesting the involvement of Rho activation in the EC contraction produced by these agonists. Downstream targets of the Rho/Rho kinase signaling cascade include the regulatory subunit of myosin-specific PPase (M 130), the phosphorylation of which by Rho kinase leads to dissociation from the catalytic subunit, thereby decreasing myosin-specific PPase 1 activity and subsequently enhancing the level of MLC phosphorylation (37, 39).

In contrast to the microfilamentous actin cytoskeleton, information about the microtubule network and its linkage to the contractile processes is limited. Disassembly of microtubules caused rapid and substantial strengthening of nonmuscle contractility and led to rapid assembly of microfilament bundles and focal adhesions (7, 15, 28), an effect abolished by microtubule stabilization (15). Isometric contraction in fibroblasts induced by microtubule disassembly correlated well with increases in MLC phosphorylation (40); however, the molecular basis for this finding has not been pursued. Furthermore, the role of the microtubule network in EC shape changes, tension development, and barrier properties is unknown. In this study, we examined the biochemical mechanisms by which microtubule disassembly increases endothelial contraction and barrier dysfunction and elucidated the involvement of increased MLC phosphorylation in this process. These studies confirm that MLC phosphorylation is a common mechanism by which microtubule-disrupting agents and receptor-mediated agonists induce nonmuscle contraction (40) and suggest an important physiological role of the microtubule network in EC barrier regulation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Reagents. EC cultures were maintained in medium 199 (GIBCO BRL, Chagrin Falls, OH) supplemented with 20% (vol/vol) colostrum-free bovine serum (Irvine Scientific, Santa Ana, CA), 15 µg/ml of EC growth supplement (Collaborative Research, Bedford, MA), 1% antibiotic-antimycotic (10,000 U/ml of penicillin, 10 µg/ml of streptomycin, and 25 µg/ml of amphotericin B; K. C. Biologicals, Lenexa, KA), and 0.1 mM nonessential amino acids (GIBCO BRL). Unless specified, reagents were obtained from Sigma (St. Louis, MO). PBS, Hanks' balanced salt solution without phenol red, and LipofectAMINE were purchased from GIBCO BRL (Grand Island, NY). DPV was kindly provided by Dr. V. Natarajan (Johns Hopkins University, Baltimore, MD). C3 exotoxin, vinblastine, and ML-7 were purchased from Calbiochem (La Jolla, CA). Toxin B was purchased from List Laboratories (Campbell, CA). MLC antibody was produced in rabbit against baculovirus-expressed and purified smooth muscle MLC by Biodesign International (Kennebunk, ME). Texas Red-phalloidin was purchased from Molecular Probes (Eugene, OR). Anti-MLCK D119 antibody was kindly provided by Dr. P. Gallagher (Indiana University, Indianapolis, IN).

Bovine pulmonary artery EC culture. ECs were obtained frozen at passage 16 from the American Type Culture Collection (CCL 209; Manassas, VA) and were used between passages 19 and 24 as our laboratory has previously described in detail (21, 59).

Endothelial monolayer resistance determinations. The electrical resistance of EC monolayers was measured with the electrical cell impedance sensor technique as previously described (24, 55). In this system (Applied Biophysics, Troy, NY), the cells are cultured on a small gold electrode (10⁴ cm²) in DMEM (GIBCO BRL) supplemented with 20% (vol/vol) colostrum-free bovine serum, antibiotics, and growth factors as described in Reagents. The cells act as insulating particles, and the total resistance across the monolayers is composed of the resistance between the ventral cell surface and the electrode and the resistance between cells. A 4,000-Hz AC signal with 1-V amplitude was applied to the ECs through a 1-M resistor, creating an approximate constant current source (1 µA). The lock-in amplifier attached to the electrodes detected changes in both magnitude and phase of the voltage appearing across the ECs and was controlled by an IBM-compatible personal computer that was used both to run the experiments and process the data. Electrical resistance increased immediately after the cells attached to and covered the electrodes, and the resistance achieved a steady state when the ECs became confluent. Thus experiments were conducted after the electrical resistance achieved a steady state. Resistance data were normalized to the initial voltage and plotted as a normalized resistance.

Isolated perfused ferret lung preparation. Commercially available ferrets were anesthetized with pentobarbital sodium (50 mg/kg ip), ventilated (20 breaths/min, tidal volume 12 ml/kg) via tracheostomy with warmed, humidified gas containing 40% O₂, and then rapidly exsanguinated as previously described (6). After exsanguination, the ventilatory gas mixture was switched to 16% O₂-5% CO₂, the ventilatory rate was adjusted to 10 breaths/min, and an end-expiratory pressure of 3 mmHg was added. Lungs were isolated by insertion of cannulas into the left atrium via the left ventricle and the pulmonary artery via the right ventricle, and then residual blood was flushed from the lungs with physiological salt solution (PSS) containing 5 mM dextrose, 3 g/dl of porcine albumin, and 2 g/dl of Ficoll (6). Isolated lungs were perfused at constant flow (50 ml · kg¹ · min¹) for 60 min with PSS-0.1% ethanol (control) or PSS containing 1 µM vinblastine. Pulmonary arterial, left atrial, and airway pressures were continuously monitored (Grass model 7) with Statham P50 transducers referenced to the left atrium. Glucose concentration and pH were monitored throughout the perfusion period and did not differ among preparations.

MLC phosphorylation. This assay was performed as our laboratory has previously described in detail (21, 55).

Cytotoxicity assay. A cytotoxicity assay was performed using a LIVE/DEAD viability/cytotoxicity kit (Molecular Probes) according to the manufacturer's protocol. This kit provides a two-color fluorescence cell viability assay that is based on the simultaneous dual determination of two recognized parameters of cell viability: intracellular esterase activity (calcein) and plasma membrane integrity (ethidium homodimer-1). Calcein produces an intense green fluorescence in viable cells, whereas ethidium homodimer-1 enters cells with damaged membranes and binds to nucleic acids, thereby producing a bright red fluorescence in nonviable cells.

Permeabilization of ECs. It is well known that C. botulinum C3 exoenzyme does not easily pass through the cell membrane under native conditions. To penetrate the cell membrane, EC monolayers (80-100% confluence) grown on 60-mm culture dishes were rinsed with OptiMEM-I medium and LipofectAMINE reagent (GIBCO BRL) added at a final concentration of 20 µg/ml for 1 h followed by the addition of C3 exoenzyme (2.5 µg/ml) for an additional 11 h as we have previously described in detail (9).

Preparation of silicone rubber substrates. To visualize cell contractility, flexible rubber substrates were generated as previously described (8, 31). Approximately 15 µl of silicone monomer (dimethyldiphenylpolysiloxane; Sigma) were applied to 18-mm glass coverslips and allowed to spread for 30 min. Cells were plated on the top of thin polymerized flexible rubber silicone films, and contractility was assessed by formation of wrinkles on the silicone substrate as observed with time-lapse videomicroscopy.

Western immunoblotting. Protein extracts were separated by SDS-PAGE, transferred to nitrocellulose or polyvinylidene difluoride membranes (30 V for 18 h or 90 V for 2 h), and reacted with specific antibodies of interest. Immunoreactive proteins were detected with the enhanced chemiluminescent detection system (ECL) according to the manufacturer's directions (Amersham, Little Chalfont, UK). The relative intensities of the protein in the bands were quantified by scanning densitometry.

Determination of myosin-specific and total serine/threonine PPase activities in endothelium. ECs from D100-mm dishes were treated with vehicle (0.1% methanol or DMSO), 0.2 µM vinblastine, or 10 nM calyculin for 10, 30, or 60 min. To prepare total cell lysates, the cells were rinsed twice with ice-cold Tris-buffered saline (20 mM Tris · HCl, pH 7.6, and 137 mM NaCl) and buffer A (50 mM Tris · HCl, pH 7.0, 0.1 mM EDTA, and 28 mM mercaptoethanol); then 400 µl of buffer A were added to the dishes, and the cells were quickly frozen at 70°C, scraped, and homogenized by passing the cell suspension several times through a 1-ml tuberculin syringe. Myosin-enriched fractions were prepared with 500 µl of total cell lysate as our laboratory has previously described (59). PPase activity was determined in a final volume of 20 µl with a Malachite Green (Upstate Biotechnology) 96-well plate assay as previously described (18, 30). Total cell lysate or myosin-enriched fraction (1 µl) was incubated with 0.25 mM phosphopeptide KRpTIRR (specific serine/threonine PPase substrate; Upstate Biotechnology, Lake Placid, NY) in PPase assay buffer (50 mM Tris · HCl, pH 7.0, 0.1 mM EDTA, 28 mM mercaptoethanol, and 0.1 mg/ml of BSA) at 30°C for 20 min. As a control, the cell suspension and phosphopeptide were incubated in a PPase assay buffer containing serine/threonine PPase inhibitors (1 mM EGTA and 5 µM okadaic acid). Reactions were terminated by the addition of 100 µl of Malachite Green solution, and the plates were incubated at room temperature for 15 min before measurement of absorbance at 620 nm. The phosphate released in the enzyme reaction was determined by comparing the absorbance over control value with the phosphate standard curve. Reactions were carried out in triplicate. To ensure linear rates of dephosphorylation, the extent of dephosphorylation was restricted to <25%. The number of proteins contained in cell lysate was assessed by TCA precipitation and then with a BCA protein assay kit (Pierce, Rockford, IL) used according to the manufacturer's protocol. Basal PPase activity in cell lysates was ~1,500 pmol phosphate · µg total protein¹ · min¹.

Immunofluorescent microscopy. ECs were grown to subconfluence on glass coverslips in DMEM. After treatment, cells were fixed in 4% paraformaldehyde for 10 min at room temperature. The cells were then washed three times with PBS, permeabilized with 0.25% Triton X-100 in PBS for 5 min, and blocked with 2% BSA in PBS for 30 min, and then actin was stained with Texas Red-phalloidin (Molecular Probes) for 1 h at room temperature. After three washes with PBS, the coverslips were mounted with a SlowFade antifade kit (Molecular Probes). Analysis of the stained cells was performed with a Nikon Eclipse TE 300 microscope equipped with ×20 to ×100 objective lenses.

Determination of MLCK activity. MLCK activity was determined in nondenaturing MLCK immunoprecipitates with exogenous MLC as a substrate, as our laboratory previously described in detail (58).

Statistics. ANOVAs with a Student-Newman-Keuls test were used to compare the means of kinase and PPase activities, the _alb, and the ratios of un-, mono-, and diphosphorylated MLCs of two or more different treatment groups. Results are expressed as means ± SE. Differences between two groups were considered statistically significant when P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effect of microtubule alteration on EC barrier function. To examine the effect of microtubule disruption on EC barrier function, we monitored electrical resistance across EC monolayers [(transendothelial electrical resistance (TER)] after treatment with the microtubule-disrupting agents nocodazole and vinblastine. Both nocodazole (Fig. 1A) and vinblastine (Fig. 1B) decreased TER in a dose-dependent manner, with half-maximal declines in TER after nocodazole (0.3-5.0 µM) occurring in ~15-30 min. Maximal vinblastine effect was observed at 0.5-2.5 µM; however, incubation of cells with even 200 nM of vinblastine significantly decreased electrical resistance (Fig. 1B). Vinblastine-induced EC permeability was also confirmed in vivo with the use of isolated perfused ferret lungs (Table 1) in which the _alb estimated in the control lungs was 0.7, a mean value similar to that previously obtained in uninjured isolated ferret lungs (6). In contrast, after 60 min of perfusion with vinblastine, the _alb decreased to 0.35 (Table 1) in association with significant lung weight gain and consistent with vinblastine-induced increased vascular permeability and EC barrier dysfunction. Pulmonary arterial and pulmonary airway pressures remained almost constant (7.0-8.5 and 5.5-6.0 mmHg, respectively) through 60 min of perfusion in the vinblastine preparation and were statistically indistinguishable from control values. Both nocodazole (Fig. 2) and vinblastine (data not shown) perturbed EC barrier properties without cytotoxicity. We next examined the effect of microtubule stabilization on barrier dysfunction induced by microtubule-disrupting agents. Figure 3A demonstrates that the microtubule-stabilizing agent paclitaxel did not affect basal TER but significantly attenuated the decrease in TER induced by nocodazole. Furthermore, paclitaxel reversed the effect of the microtubule-disrupting agents on TER, indicating that the effect of microtubule disruption on EC barrier dysfunction is reversible (Fig. 3B).

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Effect of microtubule disassembly on transendothelial electrical resistance (TER). Disruption of microtubule structure by microtubule inhibitors led to endothelial cell (EC) barrier compromise. A: TER was monitored for 2 h. Arrow, time at which cells were treated with either vehicle (0.1% DMSO) or nocodazole in complete medium. B: TER was monitored for 6 h. Cells were treated with either vehicle (0.1% methanol) or vinblastine in complete medium. Results are from representative experiments (n = 3 each).

View this table: [in this window] [in a new window]   Table 1.   Effect of microtubule disruption on alb in isolated perfused lungs

View larger version (69K): [in this window] [in a new window]   Fig. 2.   Effect of nocodazole on cell viability. Shown are immunofluorescent images (original magnification, ×40) of EC monolayers (80% confluence) stained with either calcein AM (1 µM for 45 min; A-C) for intracellular esterase activity or ethidium homodimer-1 (4 µM for 45 min; D-F) for plasma membrane integrity as described in MATERIALS AND METHODS. A and D: vehicle (0.1% DMSO)-treated ECs; B and E: ECs treated with 2 µM nocodazole for 45 min; C and F: ECs treated with 70% methanol for 30 min. Control and nocodazole-treated ECs show intense calcein staining, indicating the presence of ubiquitous intracellular esterase activity ("live" cells), whereas methanol-treated cells show intense ethidium staining, a characteristic of cells with damaged membranes ("dead" cells). Results are from a representative experiment (n = 3).

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Effect of microtubule stabilization on decrease in TER induced by microtubule inhibitors. EC monolayer resistance was monitored for 4 h. A: ECs were pretreated with either vehicle (0.1% DMSO) or paclitaxel (Pacli; 2.5 µM) for 75 min followed by challenge with either vehicle or nocodazole (Noco; 2 µM). B: ECs were first treated by either vehicle, vinblastine (Vin; 0.3 µM), or nocodazole (2 µM) for 30 min and then challenged with either vehicle or paclitaxel (10 µM). Results are from representative experiments (n = 3). Microtubule stabilization significantly attenuated and reversed the effect of microtubule inhibitors on TER.

Effect of microtubule disassembly on EC microfilament reorganization. Because our laboratory (23, 25) and others, (47) have demonstrated that agonist-induced EC barrier dysfunction is critically dependent on cytoskeletal changes, we next examined the effect of filamentous actin (F-actin) stabilization by phalloidin on vinblastine-induced EC barrier dysfunction (Fig. 4). Phalloidin significantly attenuated the decrease in electrical resistance induced by vinblastine, indicating an important role for the actin cytoskeleton in EC activation stimulated by microtubule disruption. To extend these findings, we next studied the direct effect of nocodazole on the EC actin cytoskeleton. Figure 5 shows immunofluorescent images of EC confluent monolayers stained for F-actin with Texas Red-phalloidin. As seen in Fig. 5A, control cells have F-actin organized primarily in a cortical ring, with few actin-associated stress fibers spanning the cell. Nocodazole produces dramatic cortical actin dissolution and a significant increase in stress fibers (Fig. 5B), reflecting contraction. Stress fiber formation induced by nocodazole leads to gap formation, reflecting barrier compromise (Fig. 5B).

View larger version (15K): [in this window] [in a new window]   Fig. 4.   Effect of microfilament stabilization on vinblastine-induced EC barrier dysfunction. ECs grown on gold microelectrodes were pretreated with the F-actin stabilizer phalloidin (Pha; 1 µM) for 60 min followed by vinblastine (0.3 µM). Arrows, time of addition. TER was assessed as described in MATERIALS AND METHODS. Results are from representative experiments (n = 3). Stabilization of actin filaments significantly attenuated the decrease in TER induced by microtubule inhibitors.

View larger version (66K): [in this window] [in a new window]   Fig. 5.   Effect of microtubule alteration on F-actin rearrangement. Immunofluorescent images (original magnification, ×60) show confluent ECs stained for F-actin with Texas Red-phalloidin. ECs were treated with either vehicle (0.1% DMSO; A) or nocodazole (B) for 30 min. Disruption of microtubules led to significant increases in stress fiber formation, indicating EC contraction with development of intercellular gaps (arrows) indicating barrier compromise.

Effect of microtubule disruption on MLC phosphorylation. Our laboratory (21) and others (66) have previously shown the critical involvement of MLC phosphorylation in agonist-induced barrier dysfunction in macro- and microvascular endothelium. Similar to DPV and thrombin treatments, both of which increase MLCK activity and MLC phosphorylation in ECs (22, 26, 57, 58), pretreatment of ECs with microtubule inhibitors produced significant enhancement of MLC phosphorylation in a time-dependent manner (Fig. 6A). The increases in MLC phosphorylation produced by microtubule disruption (up to 1 mol phosphate/mol MLC, peak at 30-60 min) were less than the levels of DPV-induced MLC phosphorylation (up to 1.8 mol/mol MLC) (26) but similar to the level of thrombin-induced MLC phosphorylation (up to 1.2 mol/mol MLC) (21). Neither vinblastine nor nocodazole at lower (up to 100 nM) concentrations had significant effects on the level of MLC phosphorylation (data not shown). The increased MLC phosphorylation produced by microtubule inhibitors was completely abolished by microtubule stabilization with paclitaxel (Fig. 6B). Increases in MLC phosphorylation correlated well with the time frame of decreased TER produced by microtubule assembly inhibition (shown in Fig. 1), suggesting the involvement of contractile mechanisms in EC barrier dysfunction induced by microtubule disruption. Cell contraction induced by microtubule disruption was confirmed by an increase in the number of wrinkles produced by EC monolayers grown on a thin silicone film (Fig. 7).

View larger version (38K): [in this window] [in a new window]   Fig. 6.   Effect of microtubule disruption on EC myosin light chain (MLC) phosphorylation. A: EC monolayers were pretreated with either 0.3 µM vinblastine (top), 2 µM nocodazole (bottom), or 5 µM diperoxovanadate (DPV; positive control; top and bottom) for indicated times. MLC phosphorylation was monitored by urea gel electrophoresis followed by Western immunoblotting with anti-MLC antibody as previously described by our laboratory (21, 55). un-P, unphosphorylated; mono-P, monophosphorylated; di-P, diphosphorylated. B: EC monolayers were pretreated by either vehicle (0.1% DMSO) or 10 µM paclitaxel for 60 min and then treated with vehicle or 2 µM nocodazole for 30 min, and the level of MLC phosphorylation was quantitated by densitometry of Western blots probed with anti-MLC antibody. Both microtubule inhibitors significantly increased MLC phosphorylation in a time-dependent manner, with maximal effect at 30-60 min. The increased MLC phosphorylation induced by microtubule disruption was completely abolished by microtubule stabilization.

View larger version (167K): [in this window] [in a new window]   Fig. 7.   Effect of nocodazole on EC contraction. Shown are phase- contrast images of living cells (original magnification, ×20) grown on silicone membranes as described in MATERIALS AND METHODS. A: ECs before stimulation. B: same field after treatment with nocodazole (2 µM) for 30 min. Note numerous wrinkles formed by focally contracting ECs on the silicone substrate (arrows).

Effect of microtubule disruption on MLCK and myosin PPase activities. Because MLC phosphorylation status is determined by the balance between MLCK and myosin PPase activities, we next determined the effect of microtubule disruption on EC MLCK activity measured in the MLCK immunoprecipitates in the presence of optimal Ca²⁺/CaM availability (maximal activity) in either the presence or absence of the specific MLCK inhibitor ML-7 or the Ca²⁺ chelator EGTA. The data in Fig. 8A demonstrate that inhibition of MLCK with ML-7 and EGTA completely abolished basal MLCK activity, confirming that MLCK activity assay is valid; however, neither form of microtubule inhibition altered MLCK activity above basal levels in the presence of Ca²⁺/CaM. Consistent with these results, ML-7 also failed to alter the MLC phosphorylation induced by nocodazole, whereas thrombin-induced MLC phosphorylation was abolished (Fig. 8B).

View larger version (29K): [in this window] [in a new window]   Fig. 8.   Effect of microtubule disruption on MLC kinase (MLCK) activity. A: confluent EC monolayers were pretreated with either vehicle (0.1% DMSO), 2 µM nocodazole, or 0.2 µM vinblastine for 30 min, and MLCK activity was assessed in nondenaturing MLCK immunoprecipitates. Purified MLC was used as the preferred substrate in the presence of either 0.3 mM CaCl2/1 µM calmodulin (Ca2+/CaM; maximal activity), the specific MLCK inhibitor ML-7 (10 µM), or the Ca2+ chelator EGTA (2 mM) as described in MATERIALS AND METHODS. Data are means ± SE; n = 3 experiments. B: ECs were pretreated with either vehicle (0.1% DMSO) or ML-7 (25 µM for 40 min) and then treated with thrombin (100 nM for 2 min) or nocodazole (2 µM for 30 min), and MLC phosphorylation was assessed by urea gel electrophoresis. Microtubule disruption had no effect on MLCK activity in endothelium. +, Presence; , absence.

View larger version (17K): [in this window] [in a new window]   Fig. 9.   Effect of vinblastine and calyculin on serine/threonine phosphatase (PPase) activity in endothelium. Confluent ECs from 100-mm dishes were treated with either vehicle (0.1% methanol or 0.1% DMSO), vinblastine (0.2 µM), or calyculin (10 nM) for indicated times. Cells were then lysed, and the myosin-enriched fraction was prepared as described in MATERIALS AND METHODS. Serine/threonine PPase activity of both total cell lysates (A) and myosin-enriched cell fractions (B) was measured with a serine/threonine PPase assay kit. Microtubule disruption led to significant decreases in myosin-specific PPase activity but not in total PPase activity in endothelium. Data are means ± SE normalized as percent control for each independent experiment; n = 4. *Significant difference from control.

Involvement of Rho in EC barrier dysfunction induced by microtubule disruption. In smooth muscle and nonmuscle cells including endothelium, myosin PPase activity is mediated by activation of the small GTP-bound protein Rho (20, 39). To examine the role of Rho in MLC phosphorylation induced by microtubule disruption, we first used the specific Rho inhibitor C3 exotoxin, which ADP-ribosylates and thereby inactivates Rho (2, 3). Because C3 exotoxin is not cell permeable, we initially permeabilized EC monolayers with LipofectAMINE to enhance C3 exotoxin access and found that the toxin completely abolished increases in MLC phosphorylation induced by either thrombin or vinblastine (Fig. 10A), indicating Rho involvement. Consistent with these data, the cell-permeable toxin B, which glucosylates and inactivates small Rho family G proteins (Rho, CDC 42, and Rac) (2, 13, 36), also completely abolished nocodazole-induced MLC phosphorylation (Fig. 10B). In addition, inhibition of the Rho downstream target Rho kinase by the specific cell-permeant inhibitor Y-27632 significantly attenuated nocodazole-induced decreases in TER (Fig. 11) and increases in stress fibers (data not shown) and completely abolished nocodazole-induced MLC phosphorylation (Fig. 11, inset), demonstrating the direct involvement of Rho kinase in nocodazole-induced MLC phosphorylation and permeability.

View larger version (39K): [in this window] [in a new window]   Fig. 10.   Effect of Rho inhibition on increase in MLC phosphorylation induced by microtubule disruption. Representative Western blots (n = 3) of MLC separated by urea gel electrophoresis are shown. A: ECs were first permeabilized by LipofectAMINE treatment in the presence and absence of the specific Rho inhibitor C3 exotoxin (2) and then treated with either 100 nM thrombin (Thr) for 2 min or 0.3 µM vinblastine for indicated time periods. B: ECs were pretreated with either DMEM (control) or the cell-permeable inhibitor of Rho family proteins toxin B (TB; 1 µM) for 60 min and then treated with either vehicle (0.1% DMSO) or nocodazole (2 µM) for 30 min. Rho inhibition completely abolished the increase in MLC phosphorylation induced by microtubule disruption.

View larger version (20K): [in this window] [in a new window]   Fig. 11.   Effect of Rho kinase inhibition on nocodazole-induced decrease in TER. TER was monitored for 5 h. At time indicated by left arrowhead, ECs were pretreated with either vehicle (0.05% DMSO) or the specific Rho kinase inhibitor Y-27632 (10 µM) followed by challenge (right arrowhead) with either vehicle (0.05% DMSO) or nocodazole (2 µM). Inhibition of Rho kinase by Y-27632 significantly attenuated but did not completely abolish the nocodazole-induced decrease in resistance of EC monolayers, consistent with the primary role for activation of Rho pathway in nocodazole-induced EC barrier dysfunction. Results from a representative experiment (n = 3) are shown. Inset: in the same experiment, Rho kinase inhibition completely abolished nocodazole-induced MLC phosphorylation.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, we have attempted to clarify the biochemical pathway by which changes in EC permeability are linked to microtubule assembly and disassembly. We used primarily a pharmacological approach to either disrupt or preserve microtubule organization (34) with agents such as nocodazole, a synthetic antitubulin agent that reversibly blocks the self-assembly of tubulin and depolymerizes preformed microtubules, and vinblastine, a vinca alkaloid that rapidly and reversibly binds to tubulin causing microtubule depolymerization and inhibition of microtubule assembly. Paclitaxel (taxol), a potent microtubule assembly-promoting and -stabilizing agent, prevented and reversed the effect of microtubule disruptors on tubulin assembly. We observed that both nocodazole and vinblastine caused significant dose-dependent decreases in TER in bovine pulmonary artery ECs, indicating EC barrier dysfunction. Stabilization of the microtubule network by paclitaxel significantly attenuated and reversed the decline in electrical resistance induced by microtubule disruption. Consistent with our results, microtubule disassembly has been noted to decrease electrical resistance across thyroid epithelial cells (65), increase microvascular permeability of the rat small intestine vasculature (51), and increase the permeability and rate of monocyte transendothelial migration across human macro- and microvascular ECs (38, 62). We further demonstrated that EC barrier dysfunction induced by microtubule disruption is linked to decreases in cortical actin and increases in stress fiber formation and endothelial contraction as confirmed by deformation of a silicone substrate (i.e., wrinkling). Microtubule-mediated contractility has recently been demonstrated in smooth muscle (5, 43, 52, 54). In nonmuscle cells such as fibroblasts, microtubule disruption appeared to activate actin polymerization and induce stress fibers and focal adhesions (7, 15, 16, 19, 35, 45). Cytoskeletal changes were accompanied by contraction as evidenced by either the measurement of isometric force generated from cells cultured within a collagen lattice (40) or deformation of a silicone substrate (15, 16).

The biochemical events linking the cytoskeletal changes initiated by microtubule disruption and nonmuscle contraction are uncertain but likely involve increased MLC phosphorylation and activation of the small GTP-binding protein Rho (40, 45). Our laboratory (21, 58) and others (29, 41, 48, 63, 64, 66) have previously shown a direct link between the level of MLC phosphorylation, EC retraction, and agonist-induced increases in EC permeability, indicating the importance of actomyosin-driven contraction in EC barrier dysfunction. Our present data demonstrate for the first time a correlation between microtubule disruption, activation of the contractile machinery (MLC phosphorylation, stress fiber formation), and a decrease in TER (barrier dysfunction). An increase in MLC phosphorylation after microtubule disruption in ECs (present study), previously noted in fibroblasts (40), suggests activation of biochemical cascades that lead to contraction and EC barrier failure. Importantly, fibroblasts precontracted with serum and having increased levels of MLC phosphorylation showed a decreased contractile response to microtubule disruption (40). This observation contradicts the "tensegrity" model, which suggests that microtubules function as compressive elements that oppose cellular contraction (10, 11, 32), and supports the active involvement of MLC phosphorylation and Rho activation in the initiation of contraction after microtubule disruption (16, 40).

Our laboratory (21, 57, 58) has previously shown that thrombin-mediated EC cytoskeletal rearrangement and barrier dysfunction are strongly correlated with increased Ca²⁺/CaM-dependent MLCK activity, myosin PPase inhibition, and a net increase in MLC phosphorylation. In contrast to thrombin, microtubule inhibitors did not increase MLCK activity, and specific inhibition of MLCK (with ML-7) did not affect nocodazole-induced MLC phosphorylation, indicating that MLCK is not involved in the activation of the EC contractile machinery that is induced by microtubule inhibition. Increases in MLC phosphorylation and the initiation of contraction in smooth muscle and nonmuscle cells also follow the activation of the small GTPase Rho (14, 53) and its target Rho kinase, which directly phosphorylates MLC in vitro and in vivo (4, 42, 44). In addition, Rho kinase phosphorylates the regulatory subunit of myosin-associated PPase, which leads to inhibition of myosin-associated PPase activity and an increase in MLC phosphorylation (37, 39, 49). Our laboratory (26) and others (12, 20, 27) have recently shown the involvement of Rho in agonist-induced EC barrier dysfunction, increases in MLC phosphorylation, and myosin PPase inhibition. Our present data indicate that specific Rho or Rho kinase inhibitors completely abolish increases in MLC phosphorylation induced by microtubule inhibitors, strongly implicating the involvement of the Rho pathway in the EC barrier dysfunction induced by microtubule disruption. Consistent with these data, Rho kinase inhibition significantly attenuated nocodazole-induced decreases in TER. Although microtubule depolymerization in fibroblasts also induces stress fiber formation via Rho activation (45), the precise mechanisms of Rho activation induced by microtubule disruption are not currently known. Our laboratory (Verin and Garcia, unpublished observations) demonstrated that Rho inhibition did not directly affect EC microtubule structure, indicating that Rho activation is a downstream event of microtubule disruption. To evaluate biochemical events leading to Rho-dependent increases in MLC phosphorylation and EC cytoskeletal rearrangement, we measured PPase activity in the myosin-enriched fraction, which includes myosin, actin, and regulatory and catalytic subunits of endothelial myosin PPase (57, 59, 60), and most likely represents myosin-specific PPase activity in endothelium. We found that the microtubule inhibitor vinblastine significantly decreased myosin-specific, but not total, PPase activity in bovine ECs, supporting the involvement of myosin PPase in the increased MLC phosphorylation, cytoskeletal rearrangement, and barrier dysfunction induced by microtubule disruption.

In summary, we have characterized the biochemical and physiological pathways leading to EC barrier dysfunction induced by microtubule disruption. Microtubule disassembly leads to increased MLC phosphorylation, stress fiber formation, and EC contraction via Rho GTPase-dependent, but not MLCK-dependent, mechanisms that are likely mediated via myosin PPase inhibition. Although the precise mechanism that links microtubules, Rho activation, and EC barrier dysfunction remains to be determined, our data demonstrate a significant role of microtubule dynamics in EC barrier regulation.