PAF- and bradykinin-induced hyperpermeability of rat venules is independent of actin-myosin contraction

R. H. Adamson, M. Zeng, G. N. Adamson, J. F. Lenz, F. E. Curry


We tested the hypothesis that acutely induced hyperpermeability is dependent on actin-myosin contractility by using individually perfused mesentery venules of pentobarbital-anesthetized rats. Venule hydraulic conductivity (Lp) was measured to monitor hyperpermeability response to the platelet-activating factor (PAF) 1-O-hexadecyl-2-acetyl-sn-glycero-3-phosphocholine or bradykinin. Perfusion with PAF (10 nM) induced a robust transient high Lp [24.3 ± 1.7 × 10-7 cm/(s·cmH2O)] that peaked in 8.9 ± 0.5 min and then returned toward control Lp [1.6 ± 0.1 × 10-7 cm/(s·cmH2O)]. Reconstruction of venular segments with the use of transmission electron microscopy of serial sections confirmed that PAF induces paracellular inflammatory gaps. Specific inhibition of myosin light chain kinase (MLCK) with 1–10 μM 1-(5-iodonaphthalene-1-sulfonyl)-1H-hexahydro-1,4-diazepine hydrochloride (ML-7) failed to block the PAF Lp response or change the time-to-peak Lp. ML-7 reduced baseline Lp 50% at 40 min of pretreatment. ML-7 also increased the rate of recovery from PAF hyperpermeability measured as the decrease of half-time of recovery from 4.8 ± 0.7 to 3.2 ± 0.3 min. Inhibition of myosin ATPase with 5–20 mM 2,3-butanedione 2-monoxime also failed to alter the hyperpermeability response to PAF. Similar results were found using ML-7 to modulate responses. These experiments indicate that an actin-myosin contractile mechanism modulated by MLCK does not contribute significantly to the robust initial increase in permeability of rat venular microvessels exposed to two common inflammatory mediators. The results are consistent with paracellular gap formation by local release of endothelial-endothelial cell adhesion structures in the absence of contraction by the actin-myosin network.

  • bradykinin
  • platelet-activating factor
  • inflammatory gaps
  • vascular endothelium
  • ML-7
  • myosin light chain kinase

the acute inflammatory response in venular microvessels is characterized by an initial high permeability to water and macromolecules, an opening in the endothelial layer of both paracellular gaps at the junctions, and, under some conditions, transcellular holes near endothelial cell-cell junctions. Extensive investigations in cultured endothelial cells conform to the hypothesis that gaps form between adjacent endothelial cells when Ca2+-dependent contractile forces within endothelial cells overcome the cell-cell adhesion. Both contractile and adhesion mechanisms are themselves modulated by a variety of signaling cascades, and it is therefore not surprising that there is increasing evidence from cultured endothelial cells that the balance of contractile and adhesive forces can vary in endothelial cells from different vessels and with different inflammatory mediators on the same monolayer. There is also evidence that cell-cell adhesion declines with increasing passage number, suggesting that culture conditions themselves may modify the balance of force (6, 8, 41). Furthermore, we (7) have recently suggested that a key regulator of the contractile mechanisms in cultured endothelial cells, the small GTPase RhoA, may be upregulated in some cultured endothelial cells, thereby accentuating the contribution of contractile forces to gap formation in cultured endothelial cell monolayers. Thus it is difficult to extrapolate from cultured endothelial cell studies to mechanisms modulating permeability in intact venular microvessels. The overall goal of the present experiments was to further evaluate the contribution of contractile mechanisms to increase permeability in intact venular microvessels in rat mesentery by modulating the activity of actin-myosin contraction.

On the basis of studies (9, 24, 41) that investigate the signal transduction pathways regulating acute inflammatory responses in cultured endothelial cells, there is strong evidence to support a model of active actin-myosin contraction modulated by an endothelial cell-specific form of myosin light chain (MLC) kinase (MLCK) (3). In the contraction hypothesis, the intracellular response to mediators such as bradykinin (BK), platelet-activating factor (PAF), or thrombin depends on a rapid increase of intracellular Ca2+ from intracellular stores and influx from the extracellular space (4, 23). The release from internal stores in intracellular Ca2+ concentration ([Ca2+]i) is in response to inositol 1,4,5-trisphosphate generation stimulated by phospholipase C activation (22). The increase in intracellular Ca2+ stimulates MLC phosphorylation through activation of Ca2+/calmodulin-dependent MLCK, resulting in actin-myosin contraction and opening of intercellular gaps (51). The phosphorylation state of MLC is also regulated by MLC phosphatase (45). In thrombin-stimulated cells, the myosin-binding subunit of MLC phosphatase is phosphorylated and thereby inactivated by Rho-dependent kinase (ROCK), implicating Rho A as a positive effector of inflammatory modulators (19, 43, 46). Moreover, MLC is a substrate for ROCK, further supporting the hypothesis that Rho A and ROCK stimulate endothelial contraction (2). Indeed, the results of one recent study (48) indicate that ROCK phosphorylation of MLC predominates over MLCK activity in endothelium stimulated with NaF.

Studies using intact microvessels have confirmed some, but not all, of the Ca2+ contraction hypothesis identified using cultured endothelial cells. Endothelial cells of intact venules respond to a variety of inflammatory mediators (ATP, BK, and VEGF) with a rapid rise in [Ca2+]i due to the involvement of both Ca2+ influx (16) and mechanisms involving phospholipase C stimulation as a key step leading from histamine (53) or PAF (18), suggesting involvement of inositol 1,4,5-trisphosphate signaling to intracellular stores. In one study that specifically tested the contraction hypothesis, inhibition of MLCK was found to partially inhibit the permeability responses induced in venular microvessels by phorbol 12-myristate 13-acetate, sodium nitroprusside, or cGMP (54). However, at the highest concentrations used, the MLCK inhibitor 1-(5-iodonaphthalene-1-sulfonyl)-1H-hexahydro-1,4-diazepine hydrochloride (ML-7) also inhibits other kinases (including PKC and PKA) known to modulate increased permeability. Furthermore, as a test of the contraction hypothesis, we recently demonstrated, in venular microvessels of the rat mesentery, that inactivation of the small GTPase Rho A, or inhibition of its main target, ROCK, did not attenuate the increase in permeability caused by the inflammatory agents PAF and BK (1). Those results suggested that the contraction hypothesis of endothelial barrier modulation, derived primarily from thrombin-stimulated cultured endothelial cell monolayers, may not be representative of the response of intact venular microvessels to mediators, such as BK and PAF. Furthermore, the development of leaky sites on venules has been shown to be dependent on nitric oxide signaling and PKC in response to BK (27, 32) and PAF (36, 37) and the role of these pathways in modulating contractile mechanism in endothelium is not known. The specific aim of the present study was to further test the hypothesis that contraction of the actin-myosin cytoskeleton, dependent on the activity of MLCK, is a key step in the development of transient high permeability after inflammatory stimulation. To do this, we evaluated the contribution of contractile mechanisms in intact microvessels exposed to PAF or BK in the presence and absence of inhibitors of actin-myosin contraction.

The role of the modulation of adhesion structures located at the endothelial cleft has also received additional attention because reduced adhesion in the absence of active contraction could be sufficient to open local gaps. There is evidence from cultured cell studies for the adhesion modulation hypothesis through regulation of VE-cadherin (20, 40, 44), catenins (5, 20, 34, 40), occludin (47), and the claudin tight junction protein family (21). Disruption of VE-cadherin organization has been associated with inactivation of Rho family proteins by toxin B from Clostridium difficile (1) and protein kinase C-dependent pathways in thrombin-treated cells (35, 39). The adhesion modulation hypothesis has not been extensively investigated using intact vessels. In response to TNF-α and IFN-γ there was found a redistribution of VE-cadherin near sites of increased solute flux on rat mesenteric venules suggestive of adhesion modulation, but not necessarily exclusive of contraction (50). Similarly suggestive, our recent results with the use of toxin B from C. difficile to inactivate Rho family proteins resulted in redistribution of VE-cadherin corresponding with an increase in venular hydraulic conductivity (Lp) (1).

Whereas the primary goal of these investigations is to understand paracellular gap formation between adjacent endothelial cells, the contribution of transcellular holes to increased permeability under the conditions of our experiments must be evaluated (29). With the use of three-dimensional reconstructions, two groups have demonstrated transcellular holes or vesiculovacuolar organelles (VVO) after exposing venular microvessels of mesentery and skin to common inflammatory mediators (11, 31). However, in the rat trachea, inflammatory gaps are almost exclusively paracellular (28). Both transcellular and paracellular structures could account for inflammatory permeability increases, but these different structural end points may be associated with different signaling pathways. We therefore used the same three-dimensional reconstruction methods used to examine transcellular holes to examine the gaps formed during our experiments.

We found that inhibition of actin-myosin tension development did not prevent either the increase in permeability of rat venular microvessels measured as Lp or the formation of gaps that formed exclusively between the cells. However, the conditions that elevate cAMP that can both inhibit actin-myosin tension development and potentiate intercellular adhesion, completely blocked the PAF-induced hyperpermeability. These experiments are consistent with a mechanism of paracellular gap formation by local release of endothelial-endothelial cell adhesion structures in the absence of contraction by the actin-myosin network.


Animal preparation. Experiments were carried out on male Sprague-Dawley rats (350–450 g, Hilltop Laboratory Animals) anesthetized with pentobarbital (65 mg/kg body wt sc). Anesthesia was maintained with additional pentobarbital (subcutaneous 3 mg/dose) as needed. At the end of experiments, the animals were euthanized with a pentobarbital overdose. All of the animal protocols in this study were approved by the Institutional Animal Care and Use Committee of the University of California at Davis. Each rat was anesthetized as above and placed on a heating pad to maintain normal body temperature. A midline surgical incision (2–3 cm) was made in the abdominal wall, and the mesentery was gently taken out from the abdominal cavity and spread over a quartz pillar for Lp measurements. The upper surface of the mesentery was continuously superfused with Ringer solution maintained at 35–37°C during preparation and experimentation. Experiments were performed on straight nonbranched segments of venular microvessels typically 25–35 μm in diameter. Before cannulation, all vessels selected for experiments had brisk blood flow and were free of leukocytes sticking or rolling on the vessel wall.

Measurement of Lp of microvessel wall. All measurements were based on the modified Landis technique, which measures the volume flux of water across the wall of a microvessel perfused via a glass micropipette after occlusion of the vessel. The assumptions and limitations of the measurement have been evaluated in detail (29). The initial transcapillary water flow per unit area of the capillary wall [(Jv/S)0] was measured at predetermined capillary pressures of 30–60 cmH2O. Microvessel Lp was calculated as the slope of the relation between (Jv/S)0 and applied hydraulic pressure. For most experiments, (Jv/S)0 was estimated from a single occlusion at one hydraulic pressure with the assumption that the net effective pressure determining fluid flow was equal to the applied hydraulic pressure -3 cmH2O, the approximate oncotic pressure contributed by the BSA in all perfusates. Experimental reagents (e.g., PAF) were added to the perfusate in the respective experiments and delivered via the micropipette continuously during Lp measurement. Changes in perfusate were accomplished by withdrawal of the initial micropipette and replacement of it with a second micropipette filled with new perfusate solution of the appropriate composition.

Experimental protocols. Each vessel was initially perfused with a control solution containing 10 mg/ml BSA (catalog no. A4378; Sigma-Aldrich) in a Ringer solution. To establish a control Lp, between 5 and 10 occlusions at 50 cmH2O were used over a 10- to 20-min period. The first pipette was then removed and a second pipette containing the test solution was introduced at the same cannulation site. Multiple occlusions over the subsequent 15–60 min enabled estimation of Lp during exposure to the test solution. Occlusions were made every 20–30 s during the first 5 min of test perfusion to check for rapid initial change in (Jv/S)0 and then made less frequently (2–3 occlusions every 5–10 min) for the remainder of the experiment. The tissues to be used for electron microscopy were flooded with ice-cold glutaraldehyde immediately after the final occlusion and processed as described below.

For some protocols, we used a pipette refilling apparatus that enabled exchange in ∼60 s of one perfusion solution for the next without removing the pipette from the cannulation site (33). The apparatus consisted of a push-pull syringe pump used to deliver new solution within the micropipette tip while the prior solution was removed at an equal rate from the distal end of the micropipette. Small bore polyethylene-50 tubing segments were drawn out by hand, introduced through side ports in the micropipette holder, and were connected to both the pushing (fill side) and pulling (removal side) syringes. The rear port of the micropipette holder was in free communication with a water manometer, which was used to control the hydraulic pressure within the micropipette at all times. A segment of larger bore tubing (1/8 in. inner diameter) positioned between two stopcocks on the fill side acted as a perfusate reservoir. With the pump off and stopcocks opened, the content of the perfusate reservoir was changed by injection through a syringe at one stopcock and removal as waste at the second. With the stopcocks returned to closed positions, the pump was turned on and new perfusate was introduced to the micropipette. The solution within the micropipette was exchanged at a rate of 0.7 ml/min enabling a complete exchange to the new solution within ∼60 s, as determined by a dye-filling measurement.

Solutions and reagents. Mammalian Ringer solution was composed of (in mM) 132 NaCl, 4.6 KCl, 2 CaCl2, 1.2 MgSO4, 5.5 glucose, 5.0 NaHCO3, and 20 HEPES. The ratio of acid-HEPES to Na-HEPES was adjusted to achieve a pH of 7.40–7.45 for Ringer solutions. All perfusates were mammalian Ringer solution that also contained 10 mg/ml BSA. The following stock solutions were prepared in advance and diluted into the final Ringer-albumin solution immediately before use. The PAF (1-O-hexadecyl-2-acetyl-sn-glycero-3-phosphocholine) was bought from CalBiochem (catalog no. 511075) and prepared at 1 mM in ethanol and stored for no longer than 2 wk. BK (catalog no. 05-23-0500, Calbiochem) was prepared at 0.5 mM in Ringer solution fresh each day and diluted to working concentration immediately before use. Rat thrombin (catalog no. T-5772, Sigma) was diluted to 200 U/ml in Ringer, frozen in small aliquots, and diluted to the working concentration of 10 U/ml on the day of use. The MLCK inhibitor ML-7 (catalog no. I-2764, Sigma-Aldrich), was prepared at 10 mM in 50:50 ethanol-water. The inhibitor of myosin ATPase activity 2,3-butanedione 2-monoxime (BDM) was prepared as a 1 M aqueous stock. Rolipram (catalog no. PD-175, Bio-Mol), a cAMP-specific phosphodiesterase type 4 inhibitor, was prepared at 10 mM in ethanol. Forskolin, an adenylate cyclase activator (catalog no. CN-100, Bio-Mol) was prepared at 5 mM in ethanol. Stock solutions of ML-7, rolipram, and forskolin were stored for up to 2 mo and diluted to working concentrations on the day of use.

Electron microscopy. To investigate the structural nature of the inflammatory response, three vessels exposed to PAF (1 nM) were fixed after a substantial rise in Lp (between 5 and 8 min of PAF exposure). After the final determination of (Jv/S)0, fixation for electron microscopy was started by dripping ice-cold glutaraldehyde (3% vol/vol in 0.1 mM cacodylate buffer) on the mesentery. This was done while the vessels were still perfused and occluded to ensure that the final perfusion conditions and geometry were maintained during fixation. After 1–2 min, the perfusion pipette was removed while immersion fixation continued. A diagram of the area enabled later identification of the perfused vessel segment. After 10-min in situ fixation, tissue was cut away from the animal and glutaraldehyde fixation continued on ice for a total of 1 h before changing to OsO4 (1% in veronal acetate, pH 7.2) for 1 h. Tissue was treated with tannic acid (0.5% in maleate buffer, pH 6) for 30 min to enhance membrane staining and placed in uranyl acetate (2% in maleate buffer, pH 6) overnight. After dehydration in a graded series of acetone, the tissues were flat embedded in epoxy resin.

Ribbons of serial sections (typically 20–40 sections per ribbon with 50-nm section thickness) from transversely sectioned vessels were collected on formvar-coated slots. Sections were stained with uranyl acetate and lead citrate. For each inflammatory gap found, micrographs (minimum electronic magnification ×46,000) were taken of all sections including the gap and several sections preceding and following the gap. Each gap was examined to determine whether it was transcellular, i.e., an opening through a single endothelial cell, or paracellular, i.e., an opening between two cells resulting from a loss of endothelial-endothelial adhesion.

Analysis and statistics. Lp measurements during the control period were averaged to establish a single value for Lp control for each vessel. Peak Lp values attained (Math) were the single highest measurement recorded after treatment with an inflammatory mediator. The time to reach the Math (tpeak) was the time from initial perfusion with mediator to the start of the Math measurement, determined from videotape to the nearest 0.1 min. During experiments to examine the time course of recovery, Lp measurements were made ∼1 min. Measurements were binned to the nearest 1 min using the time of Math as a reference. For most experiments, groups of three measurements were made at ∼10-min intervals and averaged in 10-min time bins. Throughout, averaged values were reported as means ± SE. The indicated statistical tests were performed assuming significance for P levels <0.05.


Increased Lp in microvessels exposed to PAF. The results of Fig. 1A are typical of experiments showing that PAF causes a robust transient increase in Lp of venular microvessels. Each vessel was first perfused with BSA-Ringer to establish control Lp. Perfusate was switched to PAF (10 nM), and there was a rapid increase in Lp that peaked within 7–12 min and then decreased toward the control level. When vessels were exposed to PAF for only 2 min using the pipette refilling technique, the Lp went through a similar transient increase, but returned more rapidly toward control (Fig. 1B). There was no difference (P > 0.25) in the time to reach Math (9.7 ± 0.8 min, n = 7, for continuous exposure and 8.4 ± 0.8 min, n = 9, for 2-min exposure) or the value of Math [21.3 ± 1.8 × 10-7 cm/(s·cmH2O), n = 7, for continuous exposure, and 24.8 ± 3.0 × 10-7 cm/(s·cmH2O), n = 9, for 2-min exposure] for the two techniques. Therefore, results concerning tpeak and Math from the two techniques are combined throughout except where indicated. For the 10 nM PAF treatment group (n = 16), the mean control Lp was 1.6 ± 0.1 × 10-7 cm/(s·cmH2O) and the mean Math was 24.3 ± 1.7 × 10-7 cm/(s·cmH2O) (Fig. 2). PAF responses were also tested at 0.1 and 1 nM to establish a dose response relation. The Math response to PAF at 0.1 nM [2.8 ± 1.2 × 10-7 cm/(s·cmH2O)] was not significantly different from the mean of paired control Lp values [1.2 ± 0.2 × 10-7 cm/(s·cmH2O)] (n = 6, P > 0.05, Wilcoxon's matched pairs test). Both the 1 and 10 nM groups showed reproducibly large increases in Math after PAF stimulation. Math measured with 1 nM PAF [17.8 ± 2.8 × 10-7 cm/(s·cmH2O)] was not quite significantly different from that measured at 10 nM (P > 0.05, Kruskal-Wallis test with Dunn's post tests). The control Lp for the 1 nM PAF group (n = 11) was 1.2 ± 0.1 × 10-7 cm/(s·cmH2O). The 10 nM PAF group responded with a smaller variability in Math than the 1 nM PAF group. One of the vessels in the 1 nM group had a less than threefold increase in Lp in response to PAF. Therefore, we chose to use 10 nM in subsequent protocols, reasoning that 10 nM would provide a more reliable test for inhibitors than lower concentrations where some vessels respond weakly even in the absence of inhibition.

Fig. 1.

Effect of platelet-activating factor (PAF) 1-O-hexadecyl-2-acetyl-sn-glycero-3-phosphocholine seen in representative vessels with and without 1-(5-iodonaphthalene-1-sulfonyl)-1H-hexahydro-1,4-diazepine hydrochloride (ML-7) pretreatment. A: data showing that on recannulation with solution containing PAF (1 nM), the hydraulic conductivity (Lp) rose to a peak in ∼10 min and then slowly returned toward control levels. B: effect of a 2-min exposure to PAF (10 nM) using the refilling pipette is shown in a second vessel. Initial response to PAF is similar in time course and peak value. Lp returns more rapidly toward control than during continuous exposure. C: vessel pretreated with ML-7 (10 μM) shows no inhibition in initial responsiveness to 10 nM PAF for 2 min.

Fig. 2.

Dose-response relationship between PAF and peak Lp response (Formula). Paired Lp controls are showed immediately to the left of each PAF group.

Ultrastructure of endothelial cell gaps in vessels treated with PAF. To characterize the morphology of the high-permeability state induced by PAF, we investigated some vessels with the use of transmission electron microscopy. We examined the ultrastructure of two microvessels fixed with glutaraldehyde near the Math after exposure to PAF (1 nM) and found numerous inflammatory gaps. The Math responses in these two vessels were 11 × 10-7 and 29 × 10-7 cm/(s·cmH2O). By reconstructing the gaps using electron micrographs of serial sections, we determined that of the seven gaps that could be clearly reconstructed, all were continuous with the paracellular space (Fig. 3). The smallest of these was 350 nm long by 350 nm wide; the largest was 1,000 nm by 600 nm. We did not find transcellular holes such as those described by Neal and Michel (31). Nor did we identify chains of vesicles or vacuoles like those found by Dvorak's group (11). However, because our primary focus was on the paracellular gaps, the absence of VVOs in these samples is not strong evidence against the possible contribution of VVOs to permeability under other conditions. No gaps were found in sham-perfused microvessels or in vessels pretreated with rolipram and forskolin before PAF exposure. Therefore, the acute permeability increase in these microvessels exposed to PAF is associated with the formation of paracellular gaps up to 1 μm across.

Fig. 3.

Reconstruction of an inflammatory gap induced by PAF. A: six sections from a serial of 40 taken of the region of an intercellular gap in the wall of a venule perfused with PAF (1 nM) and fixed during high Lp response [11 × 10-7 cm/(s·cmH2O)]. In sections 1-11 and 35-40, cell a overlaps cell b and forms intact tight junctions. In intervening sections, cell a is seen as two pieces and there is an open path from the lumen (top) to the interstitium. B: reconstruction of the overlap of cells a and b as it would be seen looking toward the vessel wall from the lumen. The middle bar identifies section number and approximate thickness and location of the sections along the dashed lines in the diagram (right). The micrometer scale applies to the diagram as well as the micrographs. The opening between the endothelial cells was ∼700 × 300 nm.

MLCK inhibition fails to block PAF-induced hyperpermeability. We then tested whether the response of vessels to PAF was dependent on the activity of MLCK. Venular microvessels first pretreated with ML-7 (25 min, 10 μM) and then exposed to 10 nM PAF responded with Math of 22.0 ± 1.3 × 10-7 cm/(s·cmH2O) (n = 9) (see Fig. 1C). This was not different from the response with PAF alone [24.3 ± 1.7 × 10-7 cm/(s·cmH2O); n = 16]. A second group (not shown) pretreated with ML-7 for a longer time (45 min, 10 μM, n = 9) also had a robust Math [21.0 ± 3.9 × 10-7 cm/(s·cmH2O)] that was not different from the group treated with PAF alone. At a higher concentration in a final group, ML-7 (25 min, 100 μM) partially inhibited the PAF response [Math 9.6 ± 2.6 × 10-7 cm/(s·cmH2O)]. Concentrations of ML-7 <10 μM are close to the IC50 for inhibition of MLCK; at concentrations >10 μM ML-7 also inhibits other kinases, including PKC and PKA. Figure 4A summarizes results from experiments testing the responsiveness to PAF (10 nM) in the presence of ML-7 at 0, 10, and 100 μM. These results demonstrate that specific inhibition of MLCK does not alter the magnitude of the acute Lp response to PAF.

Fig. 4.

Effect of ML-7 on venule response to PAF. A: Formula response to PAF (10 nM) during specific inhibition of MLCK by ML-7 (10 μM) is not different from Formula in the absence of ML-7. B: time to reach Formula after the beginning of PAF exposure is not altered by ML-7 at either 10 or 100 μM. *P < 0.05.

As a further test of the dependence of the acute Lp response on MLCK, we asked whether inhibition with ML-7 could alter the time course of the response. With the use of these same sets of experiments, we measured the time from initial PAF exposure to tpeak. The mean value of tpeak was between 8 and 10 min in all three groups, with no significant difference between the groups (P > 0.05). MLCK inhibition did not change tpeak, even at 100 μM (Fig. 4B). The results shown in Fig. 4, A and B, are not consistent with a significant role for MLCK to increase microvessel permeability in response to PAF.

Recovery phase analysis of PAF treatment. An extension of the hypothesis that inflammatory gaps form through the action of actin/myosin contraction is that an active mechanism could hold gaps open during the recovery phase. Alternatively, an active actin/myosin mechanism could be necessary to close the gaps. To test these ideas, we examined the time course of recovery from the high Lp state after 2-min exposure to PAF (10 nM) in the presence or the absence of ML-7 pretreatment (10 μM for 25 min). The results of Fig. 5A show the mean values of Lp at 1-min intervals up to 25 min after the peak Lp response. Within 1–4 min after the peak Lp response to PAF, we found that the Lp recovery phase of each vessel, over a part of its time course, could be fit to a single exponential of the form Math where Lp,s represents the span between Math and Lp,f, Lp,f is the plateau value where the Lp is falling, and k is the time constant (Fig. 5, B and C). The half-time for recovery (t1/2) equals 0.69/k. The mean values of these constants enable statistical comparison of the recovery behavior in the two groups. Most importantly, we find that inhibition of MLCK reduces the t1/2 from 4.8 ± 0.7 min (n = 7) in the PAF group to 3.2 ± 0.3 min (n = 6) in the ML-7 pretreated group (P < 0.05, Mann-Whitney test). MLCK inhibition enabled more rapid Lp recovery after PAF stimulation. The Lp,s values were not different in the two groups (P > 0.4, Mann-Whitney test). The mean plateau value, Lp,f, for the ML-7-treated group [1.1 ± 0.1 × 10-7 cm/(s·cmH2O)] was lower than for the PAF-only group [1.8 ± 0.2 × 10-7 cm/(s·cmH2O)] (P < 0.05, Mann-Whitney test). This difference suggested an effect of ML-7 on basal permeability, which we investigated more fully in the following series of experiments.

Fig. 5.

ML-7 speeds recovery from PAF stimulation. A: pattern of Lp recovery after exposure to PAF (2 min, 10 nM) is shown for vessels pretreated with or without ML-7 (10 μM, 25 min). B: exponential phase of the recovery was fit for each vessel in the PAF-only group and mean one-half time to recovery (t1/2) was 4.8 min. C: exponential phase of the recovery is shown for vessels pretreated with ML-7; mean t1/2 was 3.2 min. Note log scale in B and C.

ML-7 reduces Lp in nonstimulated microvessels. The failure of ML-7 to block the PAF-induced increase in permeability is not the result of a failure of ML-7 to modify endothelial cell function. In fact, in some vessels, ML-7 at 10 μM reduced the Lp to close to 50% of the initial resting value over a 25-min period of exposure. We tested the ability of ML-7 (10 μM) to reduce the permeability of nonstimulated microvessels by continuous perfusion with solution containing BSA and ML-7 (10 μM) for 60 min and monitored Lp every 10 min (Fig. 6A). In 60 min, ML-7 reduced the mean Lp to 42% of the initial value (n = 8, P < 0.05, two-way ANOVA). In a separate set of 15 sham-perfused vessels, the Lp at 60-min perfusion was 92% of control. We also examined the effect of 100 μM ML-7 on Lp over 20 min in a further set of six nonstimulated vessels. The mean Lp in that group [1.2 ± 0.2 × 10-7 cm/(s·cmH2O)] decreased to 56% of control and was not different from the mean Lp [1.2 ± 0.1 × 10-7 cm/(s·cmH2O)] in the 10 μM ML-7 group for 20 min (Mann-Whitney, P > 0.60) (Fig. 6B). Thus the action of ML-7 to reduce resting permeability and modulate recovery is in striking contrast to the lack of effect on the initial increase in Lp.

Fig. 6.

Effect of myosin light chain kinase (MLCK) inhibition on basal Lp. A: perfusion with ML-7 (10 μM) decreases basal Lp by >50% within 60 min. Perfusion with vehicle control solution in a separate set of vessels showed a nonsignificant change in Lp. In the ML-7 group, Lp is significantly lower than in the time-matched controls for all times >10 min (two-way ANOVA, Bonferroni post tests, *P < 0.05). B: ML-7 at 100 μM has similar effect to 10 μM on basal Lp. Treatment with 10 μM ML-7 for 20 min reduced basal Lp by 36% compared with paired controls (open bars, n = 9). Treatment with 100 μM ML-7 for 20 min in a separate group of vessels reduced Lp by 44% from paired control values (solid bars, n = 6). †P < 0.05, different from matched controls (Wilcoxon signed-rank test).

MLCK inhibition fails to block BK-induced hyperpermeability. We also tested the effects of ML-7 on response to BK, a second inflammatory mediator that we have previously shown to consistently cause an increase in Lp of rat mesentery venules (15). Neither the Math response to BK nor the time to reach the peak response was significantly different in vessels treated with ML-7 compared with controls (Fig. 7A). The Math in vessels pretreated for 50 min with ML-7 (10 μM) and then recannulated and perfused with BK (10 nM) was 37.7 ± 6.4 × 10-7 (n = 5). Vessels perfused with BK (10 nM) in the absence of ML-7 responded with Math of 34.8 ± 9.5 × 10-7 (n = 7). These responses were not different (P < 0.05; Mann-Whitney test). Similarly, tpeak was not different between the two groups (Fig. 7B).

Fig. 7.

Lack of effect of ML-7 on response to bradykinin (BK). A: Formula response to 10 nM BK during inhibition of MLCK by ML-7 (10 μM) is not different from Formula in the absence of ML-7. B: time to reach Formula after the beginning of BK exposure is not altered by ML-7 (10 μM).

Analysis of recovery from BK-induced hyperpermeability. Lp data from vessels treated with BK were examined using the same analysis as vessels treated with PAF (above). Between 1 and 10 min after passing through Math, the Lp of each vessel recovered for several minutes toward initial values. Data from the recovery phase of each experiment were fit to a one-phase exponential decay. Vessels treated with ML-7 recovered more quickly than untreated vessels (Fig. 8). For the ML-7 group, the t1/2 of the recovery phase was 2.6 ± 0.3 min (n = 5), significantly faster than that of the control group, for which the t1/2 was 4.5 ± 0.8 (n = 7; P < 0.05, Mann-Whitney test). The BK data were very similar to the PAF data in that the most apparent effect of ML-7 was to speed recovery from the hyperpermeability state, suggesting that MLCK actively opposes the closure of inflammatory gaps.

Fig. 8.

Analysis of Lp recovery after BK hyperpermeability. A: exponential phase of the recovery was fit to data from each vessel in the BK group; the mean t1/2 was 4.5 min. B: exponential phase of the recovery is shown for vessels treated with ML-7; mean t1/2 was 2.6 min.

Myosin inhibition with BDM does not block acute hyperpermeability. As a further test of myosin contraction in the response to inflammatory mediators, we tested whether direct inhibition of myosin function could block the acute Lp response to PAF. The inhibitor BDM blocks myofibril contraction through inhibition of myosin ATPase (17) and has an IC50 for neutrophil motility between 10 and 20 mM (10). It prevents cytochalasin B-induced increase in tight junction permeability of epithelial cells (25) and inhibits stress fiber formation and cell contraction induced by TNF-α in endothelial cells (49) at 20 and 5 mM, respectively. After measurement of Lp with control solution, vessels were pretreated with BDM for 20 min before the effect of PAF (10 nM) was tested (Fig. 9A). Neither the peak Lp response to PAF in the presence of 5 mM BDM [14.2 ± 2.0 × 10-7 cm/(s·cmH2O), n = 8] nor 20 mM BDM [16.7 ± 6.4 × 10-7 cm/(s cmH2O)] was significantly different from a control group [19.6 ± 5.6 × 10-7 cm/(s·cmH2O), n = 9; Kruskal-Wallis test, Fig. 9B]. Similarly, the time to peak was unaffected by BDM (Kruskal-Wallis test; Fig. 9C). Note that the PAF stock for this series of experiments was made from a different lot than that for previous experiments and the mean Lp response to PAF was somewhat lower than that shown in Fig. 2.

Fig. 9.

Effect of 2,3-butanedione 2-monoxime (BDM) on PAF-induced hyperpermeability. A: data from a representative vessel pretreated for 20 min with BDM (5 mM) and stimulated with PAF (10 nM). B: mean value of peak Lp in response to PAF (10 nM) is not changed by BDM (5 or 20 mM). C: time to reach peak Lp is not changed by BDM (5 and 20 mM).

Thrombin does not increase rat microvessel Lp. Most direct evidence for a role of contractile mechanisms to modulate permeability comes from studies using thrombin to increase the permeability of cultured endothelial cell monolayers (12, 13, 24). Figure 10 shows an experiment on a single vessel exposed to BK (1 nM) and then to thrombin (10 U/ml). Thrombin failed to increase Lp, although BK did increase Lp. The results from 10 microvessels exposed to thrombin are summarized in Table 1. The was there a clear transient increase in permeability in only 2 of the 10 microvessels due to thrombin, and in one of those the increase was less than twofold. The failure of thrombin to increase permeability is not due to lack of thrombin protease activity because thrombin always caused clotting distal to the perfused vessels where it came in contact with the animals' whole blood. These results indicated, contrary to cell culture results, that thrombin does not cause an increase in permeability typical of an acute inflammatory agent such as BK, histamine, PAF, or ATP in individually perfused rat mesentery vessels (14, 15, 30).

Fig. 10.

Data from a representative vessel showing lack of response to thrombin. Lp is shown as a function of time during perfusion first with vehicle control solution containing only BSA (10 mg/ml) in Ringer solution. The Lp responded when the vessel was perfused with BK (1 nM). After 45-min washout with control solution, the vessel did not respond when perfused with solution containing thrombin (10 U/ml).

View this table:
Table 1.

Thrombin has little effect on acutely measured Lp

Elevated intracellular cAMP inhibits PAF-induced hyperpermeability. We also investigated other mechanisms that can modify permeability by acting to modulate MLCK. We have shown previously 8-BrcAMP (1 mM) pretreatment inhibits the BK-induced acute increase in permeability (15). Therefore, we compared the increase in Lp of venular microvessels exposed to PAF with the response of vessels exposed to conditions that increase intracellular cAMP (5 μM forskolin to stimulate adenylate cyclase and 10 μM rolipram to inhibit phosphodiesterase IV, both for 30 min). Elevated cAMP effectively abolished the response to PAF. Figure 11 summarizes the action of elevated cAMP to inhibit increased microvessel permeability due to PAF (1 and 10 nM).

Fig. 11.

Stimulation of cAMP blocks PAF. PAF (1 and 10 nM) stimulates a transient increase in Lp (same data as in Fig. 2). Pretreatment (30 min) with rolipram (10 μM) and forskolin (5 μM) to elevate intracellular cAMP (RF) strongly blocks the response at both concentrations. *P < 0.05, different from matched PAF.


The main conclusion from these experiments is that an actin-myosin contractile mechanism modulated by MLCK does not contribute significantly to the robust initial increase in permeability in rat venular microvessels exposed to two common inflammatory mediators, BK and PAF. First, this conclusion is based on the observation that the MLCK inhibitor ML-7, at a concentration (10 μM) where it has specificity for MLCK, has no significant effect on the initial permeability increase. Second, in the same microvessels, ML-7 reduced resting permeability and increased the rate of recovery after exposure to these inflammatory mediators. Third, inhibition of myosin ATPase activity by BDM failed to block the hyperpermeability response to PAF. The results are consistent with the hypothesis that an intact actin-myosin system regulated by MLCK modulates resting permeability and recovery, but plays no significant role in the initial rise in permeability under the conditions of our experiments. Furthermore, vessels exposed to PAF clearly form gaps between adjacent endothelial cells. Although our data cannot exclude some circumstances where there is formation of transcellular holes as reported by Michel and Neal (31) or VVOs, as reported by Dvorak's group (11), the present observations clearly indicate that PAF can stimulate the formation of paracellular gaps as initially proposed by Majno and Palade (26). Thus under the conditions of our experiments, the mechanisms that modulate cell-to-cell adhesion appear to be more important than active actin-myosin contraction to increase permeability.

There are several unique aspects of our experimental design that provide insight into the possible mechanisms underlying our results. First, the resting permeability of all microvessels (measured as the hydraulic conductivity) was measured before exposure to the inflammatory agent so that the initial permeability state of each vessel was precisely determined. All resting values of Lp fell within the range of normal values for rat microvessels measured in our laboratory, and these were at least one order of magnitude smaller than the hydraulic conductivity of both cultured endothelial monolayers and some microvessels where contractile mechanisms have been described. Thus the microvessels investigated in the present series are as close as we can determine to be in a state where there has been no recent exposure to inflammatory conditions which may induce a state of a sustained increase in permeability. Second, thrombin did not increase the permeability in the microvessels that we investigated. Almost all investigations of contractile mechanism in cultured endothelial monolayers are carried out after thrombin stimulation. Thus the endothelial cells forming the permeability barrier in these intact microvessels lacked either thrombin receptors or key elements of the thrombin-activated signaling pathway leading to contraction. These two observations are consistent with the hypothesis that some endothelial cells in intact microvessels and many endothelial cells in culture, express a more contractile phenotype than that found in the venular microvessels investigated in the present study. This hypothesis will be evaluated in more detail below. First, we evaluate our experimental strategy and experimental methods and design.

Experimental design. Our overall strategy to evaluate the contribution of contractile and adhesion mechanisms to increased permeability in recent years has been to carry out experiments in intact microvessels to parallel studies on contractile mechanisms in cultured cells. Because many investigations of contractile mechanism were done using thrombin as an inflammatory stimulus, this agent was our original agent of choice to increase venular microvessel permeability and then investigate contractile and adhesive mechanisms. The result that thrombin did not increase permeability in rat microvessels which had a normal baseline permeability forced us to modify our approach. In one series of experiments, now completed, we tested the effect of inhibiting Rho A and ROCK, key components of the thrombin signaling pathways leading to contraction in cultured endothelial cells on the permeability of intact rat and mouse vessels (1). We found that inhibition of Rho A and ROCK did not prevent the increase in permeability after stimulation with BK or PAF, indicating that the Rho A pathways to stimulate contractile mechanisms was not involved to acutely increase permeability after BK or PAF. In these vessels, the redistribution of VE cadherin indicated that changes in permeability were more directly associated with cell-cell adhesion modulation than contraction. The present investigations further extend this approach with the use of BK and PAF, both of which cause calcium-dependent increases in permeability in intact rat microvessels and in cultured monolayers, and using inhibitors of actin-myosin contraction. The observations that neither ML-7 nor BDM significantly attenuate the initial increase in permeability are consistent with the results of the investigations to inhibit Rho A and ROCK in that neither MLCK nor key agents thought to increase myosin phosphorylation increase permeability. An important assumption is that ML-7 has reasonable specificity for MLCK at concentrations close to 10 μM. The effective IC50 is in the range of 1–10 μM (10, 38) and investigations in vivo are consistent with this result. At concentrations of 100 μM, ML-7 also inhibits PKA and PKC. Because the acute response to both PAF and BK are thought to be PKC dependent (18, 32), a reduction in permeability in the presence of 100 μM ML-7 cannot be used as evidence in support of MLCK activity.

Contractile phenotypes. Much of the evidence to support a contractile mechanism in intact vessels is indirect. The most detailed studies are by Yuan et al. (52, 54) in coronary venular microvessels. The investigators demonstrated a dose-dependent reduction in albumin permeability increase by exposing the vessels to nitric oxide donors and analogs of cGMP in the range of 0.1–100 μM ML-7. ML-7 at concentrations in the range of 1–10 μM dose dependently reduced resting permeability, and inhibited ∼50% of the increase in albumin permeability. Thus one interpretation of these data is that in isolated coronary venules, contractile mechanism may contribute up to 50% of the increase in albumin permeability with the remainder accounted for by changes in cell-cell adhesion. Another possibility is that only some of the endothelial cells in the microvessels studied by Yuan et al. (52, 54) have a contractile phenotype and contribute up to 50% of the increase in permeability, whereas other endothelial cells do not express a contractile phenotype.

The idea that there can be a distribution of contractile and noncontractile phenotypes within a microvascular bed, and even within a single microvessel is further suggested by investigations based on measurements of localized leak sites. An important observation is that leaky sites reported by many investigators are quite rare (often only 1–5 per 1,000 μm of vessel length, i.e., 1–5 leaky sites per 100 endothelial cells). This means that not all the endothelial cells within a single microvessel respond in the same way to an inflammatory stimulus and many microvessels do not respond by forming large leaky sites at all. We suggest that the large fraction of endothelial cells that are not involved with the formation of localized leaky sites may have properties similar to those investigated in the present experiments and respond by the formation of much smaller gaps distributed along the junctions. Conversely, the endothelial cells associated with isolated leaky sites may be more representative of the contractile phenotype. It is interesting to note that a preliminary report by Valeski and Baldwin (42) described the inhibition of relatively rare thrombin-stimulated leaky sites in rat mesentery at 10 μM ML-7. Most vessels did not respond to thrombin, just as we report. This observation would be explained if a thrombin-stimulated contractile endothelial cell phenotype were present predominantly in the locations where localized leaky sites formed.

Mechanisms that cause endothelial cells to change from a noncontractile to a contractile phenotype are not well understood, but the possible role of prior exposure to inflammatory agents, or injury that might result when cells are isolated and grown under some culture conditions, needs to be evaluated. In summary, investigations using inhibitors of myosin contraction do not conform to the hypothesis that actin-myosin interacting via MLCK is the principal mechanism to cause acute increases in microvessel permeability. It appears that most of the endothelial cells forming the microvessel wall are not the contractile phenotype characterized by thrombin-stimulated increases in permeability in cultured cells. The idea that endothelial cells in culture, and possibly some endothelial cells after microvascular injury, have increased contractile phenotypes requires further investigation.


We thank Detlev Drenckhahn and Jens Waschke for the suggestion to use BDM and for many useful discussions. Yan-yan Jiang provided expert technical assistance.

This study was supported by National Heart, Lung, and Blood Institute Grants HL-44485 and HL-28607.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract