Sphingosine-1-phosphate modulation of basal permeability and acute inflammatory responses in rat venular microvessels

Abstract

Aims

Although several cultured endothelial cell studies indicate that sphingosine-1-phosphate (S1P), via GTPase Rac1 activation, enhances endothelial barriers, very few in situ studies have been published. We aimed to further investigate the mechanisms whereby S1P modulates both baseline and increased permeability in intact microvessels.

Methods and results

We measured attenuation by S1P of platelet-activating factor (PAF)- or bradykinin (Bk)-induced hydraulic conductivity (L_p) increase in mesenteric microvessels of anaesthetized rats. S1P alone (1–5 µM) attenuated by 70% the acute L_p increase due to PAF or Bk. Immunofluorescence methods in the same vessels under identical experimental conditions showed that Bk or PAF stimulated the loss of peripheral endothelial cortactin and rearrangement of VE-cadherin and occludin. Our results are the first to show in intact vessels that S1P pre-treatment inhibited rearrangement of VE-cadherin and occludin induced by PAF or Bk and preserved peripheral cortactin. S1P (1–5 µM, 30 min) did not increase baseline L_p. However, 10 µM S1P (60 min) increased L_p two-fold.

Conclusion

Our results conform to the hypothesis that S1P inhibits acute permeability increase in association with enhanced stabilization of peripheral endothelial adhesion proteins. These results support the idea that S1P can be useful to attenuate inflammation by enhancing endothelial adhesion through activation of Rac-dependent pathways.

1. Introduction

Microvascular endothelium is the primary barrier to and regulator of solute and solvent exchange between blood and tissue. In individually perfused microvessels, various mediators induce an acute inflammatory reaction that is characterized by a transient breakdown of the barrier function of endothelium through localized loss of adhesion and transient opening of small gaps between the endothelial cells lasting on the order of 10min. Such permeability increases are a model for increased permeability in a variety of pathological states leading to oedema and compromised organ function. They also provide a useful model to investigate mechanisms that attenuate increased permeability. Reports from several laboratories including our own demonstrate that both elevated intracellular cAMP and activation of signalling pathways by sphingosine-1-phosphate (S1P) in endothelial cells attenuate increased permeability.^1–8 Although some of the actions of elevated cAMP can be attributed to a reduction of tension development in endothelial cells after exposure to agents such as thrombin, there is increasing evidence that the principal action of cAMP and S1P to attenuate increased vascular permeability involves regulation of adhesion between endothelial cells leading to junction complex stabilization.^9–12

An emerging concept is that mechanisms that strengthen endothelial cell–cell adhesion are a common and essential feature of the action of agents that protect the endothelial barrier after exposure to injury and inflammatory conditions.^4,13,14 Mechanisms to strengthen adhesion involve increased polymerization of cortical actin in endothelial cells, linkage of actin-binding proteins to the cortical network, and tighter adhesion between molecular complexes associated with the peripheral actin band and molecules forming the adherens junction and the tight junction.¹⁵ We previously demonstrated that acute inhibition of the small GTPase Rac1 using bacterial toxin (Clostridium sordellii lethal toxin) causes a large increase in permeability of venular microvessels, measured as hydraulic conductivity (L_p), as well as cultured endothelial monolayers.¹⁴ The increase in permeability is associated with a reduction of F-actin content by 50% and similar fractional reduction in VE-cadherin-mediated adhesion of coated beads to endothelium. In a separate study, preferential activation of Rac by the bacterial toxin cytotoxic necrotizing factor-1 (CNF-1) significantly attenuated the increased permeability induced by platelet-activating factor (PAF).¹⁶ Our observations that activation of Rac results in the protection of the endothelial barrier against acute inflammatory mediators is consistent with several recent reports that S1P, which also activates Rac, enhances the endothelial permeability barrier in vivo,⁴ in situ,⁶ and in vitro.⁵

The aim of the present experiments was to further evaluate the mechanisms whereby S1P modulates both baseline and elevated permeability in intact venular microvessels. Several important questions remain unanswered. For example, the most detailed investigations of S1P-dependent modulation of endothelial cell–cell adhesion have been carried out in cultured endothelial monolayers which have less well-formed junctions than intact mammalian microvessels. We therefore investigated changes in the distribution of VE-cadherin, occludin, and cortactin in intact microvessels in the control state and after exposure to inflammatory agents with and without the presence of S1P. Furthermore, preliminary investigations showed that S1P only partially attenuated increased permeability, in contrast with the action of elevated cAMP. cAMP acts via many pathways including activation of Rap1¹⁷ which may induce actin stabilization via Rac1.^18–20 To address the mechanism of S1P action, we demonstrated activation of Rac in a cultured endothelial cell system and correlated Rac activation with stabilization of peripheral cytoskeletal components, and then, we used nearly identical solutions and conditions to test for comparable cytoskeletal response in intact microvessels. We found using intact microvessels that S1P blocked the acute permeability response to both bradykinin (Bk) and PAF and that it induced a moderate reduction in baseline L_p. Only at the highest concentration tested (10 µM), did S1P moderately increase permeability. The permeability experiments as well as our immunofluorescence investigations were consistent with the hypothesis that the primary action of S1P is to stabilize the endothelial adhesion complex of intact microvessels through activation of Rac-dependent pathways and its effectiveness was not compromised except at very high concentration.

2. Methods

2.1. Animal preparation and measurements to characterize vessel wall permeability

Experiments were carried out on male rats anaesthetized with pentobarbital (100 mg/kg body weight, sc) and maintained by giving additional pentobarbital (30 mg/kg, sc) as needed. At the end of experiments, animals were euthanized with saturated KCl. The investigation conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). Animal protocols (07-13052) were approved by the Institutional Animal Care and Use Committee of the University of California, Davis, CA, USA. Experiments were performed on venular mesenteric microvessels (ca. 25–35 µm dia.). Hydraulic conductivity (L_p) was measured to characterize vessel wall permeability. Experiments were based on the modified Landis technique, which measures the volume flux of water crossing the wall of a microvessel perfused via a glass micropipette following occlusion of the vessel. Assumptions and limitations have been evaluated in detail.^1,21 See Supplementary material online for further details.

2.2. Confocal microscopy of immunolabelled mesenteries

Mesenteries were flooded with ice-cold fixative, incubated with primary antibodies against VE-cadherin, occludin, and cortactin, labelled with fluorescent secondary antibodies, and then mounted for confocal microscopy. Tissues were mounted whole to retain the three-dimensional structure of the vessels and, when possible, to enable separate collection of either front (near to lens) or rear half of each vessel. Stacks were projected onto a single plane for analysis. See Supplementary material online for further details.

2.3. Solutions and reagents

The following stock solutions were prepared: PAF (1-O-hexadecyl-2-acetyl-sn-glycero-3-phosphocholine; Calbiochem 511075) (1 mM) in EtOH, Bk (Sigma, B3259) (1 mM) in water, and S1P (1 mM) in EtOH.

2.4. Cell culture

Human dermal microvascular endothelial cells (HMVEC; Clonetics) were seeded at high density (near confluence) on collagen-coated tissue culture-treated polystyrene and cultured for up to 3 days. Immunofluorescence: HMVEC were serum starved for 16 h prior to treatment with S1P, then cells were fixed in freshly depolymerized paraformaldehyde and labelled with anti-VE-cadherin and anti-cortactin.

2.5. Rac activation assay

HMVEC were serum starved for 16 h prior to treatment with S1P or vehicle. Relative level of RacGTP was determined using a colorimetric ELISA-based assay (Rac1,2,3 activation assay, Cytoskeleton). Samples were processed according to the manufacturer's protocol.

2.6. Analysis and statistics

To examine the modulation of response to PAF or Bk, we tested the response in each vessel twice, first in the absence of other agents and second in the presence of a test reagent. Therefore, each vessel acted as its own control. Throughout, averaged L_p values were reported as mean ± SEM. The indicated statistical tests were performed assuming significance for probability levels <0.05.

3. Results

3.1. S1P inhibits inflammatory response to PAF

Representative experimental data from an individual experiment show L_p measured over 120 min (Figure 1A). After initial perfusion with vehicle solution, the vessel was pre-treated by perfusion with S1P (1 µM) for 30 min. Responsiveness to PAF (10 nM) was then tested in the continued presence of S1P and a small response was noted. After the L_p returned to near baseline during perfusion with vehicle solution, the response to PAF of the same vessel was tested in the absence of S1P and a much larger response was recorded. Some vessels were tested initially with PAF alone and then treated with S1P prior to a second PAF test. The results showed inhibition due to S1P comparable to those seen in Figure 1A and were pooled for analysis. A further group of experiments demonstrated that repeated PAF treatments in the absence of S1P elicited similar L_p peaks; the ratio of the first peak to the second peak was 1.00 ± 0.25 (first column of Figure 1B). The inhibition of the PAF response measured as the ratio of peak L_p in the presence of S1P to the peak L_p in the absence of S1P is summarized in Figure 1B. S1P at either 1 or 5 µM significantly attenuated the L_p response to near 35% of the peak L_p measured without inhibition.

Figure 1.

(A) S1P inhibits acute L_p response to PAF. Representative data from one experiment show that treatment with S1P (1 µM) inhibits the first test with PAF (10 nM), whereas a second test in the absence of S1P after recovery to baseline shows a typical PAF response. (B) Dose–response relation for inhibition of PAF by S1P. The ratio of the peak L_p response to PAF (10 nM) in the presence of S1P to the peak value in the absence of S1P is shown. Both 1 and 5 µM S1P inhibit about 65% of the acute L_p peak response. Control value shows that repeated treatment with PAF yields comparable peaks (*different from control, P < 0.05, Wilcoxon's signed-rank test, n for each group shown in parentheses).

3.2. S1P also inhibits inflammatory response to Bk

To test the generality of S1P inhibition, Bk was used as a second inflammatory mediator that also induces a transient L_p increase in perfused venules. Data from a representative experiment show that S1P produces a strong inhibition of the L_p response to Bk (Figure 2A). Data averaged from several experiments indicate that S1P (1 µM) inhibits about 65% of the peak L_p response to Bk (10 nM); in a separate set of experiments using higher concentrations of S1P (5 and 10 µM), the response to Bk was more strongly attenuated (Figure 2B). In the absence of S1P, repeated Bk treatments induced comparable L_p peaks in control experiments; the ratio of the first peak to the second peak was 0.94 ± 0.17 (first column of Figure 2B).

Figure 2.

(A) S1P inhibits acute L_p response to Bk. Using the same experimental design as in Figure 1 for PAF experiments, S1P (1 µM) is shown to partially inhibit the acute peak L_p response to Bk (10 nM) during the first Bk test L_p these data from a representative vessel. After recovery to baseline, the second Bk test responds with a typical rapid increase and recovery. (B) Dose–response relation for S1P inhibition of Bk. S1P at 1, 5, and 10 µM strongly reduces the peak L_p response to Bk (10 nM), but even at the highest concentration it does not fully block the response. Repeated tests in the absence of S1P show that multiple Bk tests yield similar peaks (*different from control, P < 0.05, Wilcoxon's signed-rank test, n for each group shown in parentheses).

3.3. Effect on baseline L_p

One published report using cultured endothelial cells indicated that S1P at high concentration (≥5 µM) induced a loss of barrier function in endothelial cells.²² This was in contrast to reports that physiological concentrations of S1P (10 nM to 2 µM) improved barrier function.⁴ Therefore, we tested whether S1P alone had an effect on baseline L_p. At the lowest concentration (1 µM), S1P after 30 min induced a small but statistically significant decrease in L_p (Figure 3A). We observed no change in L_p with either 5 or 10 µM S1P after 30 min. Only after 60 min of continuous perfusion with S1P at the highest concentration tested (10 µM), was there an approximate doubling of the L_p from 1.0 ± 0.2 × 10⁻⁷ cm/(scmH₂O) to 1.9 ± 0.4 × 10⁻⁷ cm/(scmH₂O) (P < 0.05, Wilcoxon's matched pairs test; n = 6). After 60 min of perfusion with either 1 or 5 µM S1P, there was no difference from initial control values.

Figure 3.

Effect of S1P on baseline L_p. (A) The ratio of L_p measured after 30 min of S1P relative to the value measured with vehicle solution alone is shown. Only at 1µM is the L_p significantly reduced to about 70% of control. At higher concentrations, there was no significant difference from control (*different from value of 1; P < 0.05, Wilcoxon's signed-rank test, n for each group shown in parentheses). (B) The ratio of L_p measured after 60 min of S1P relative to the L_p in the absence of S1P is shown. With 1 or 5 µM, there was no change from control. At 10 µM, S1P L_p increased by about two-fold (*different from value of 1; P < 0.05, Wilcoxon's signed-rank test, n for each group shown in parentheses).

3.4. Localization of junction components VE-cadherin and occludin in vivo

We previously documented that the elevated permeability state induced by PAF was associated with small discontinuities (ca. 1 µm) and rearrangement of both VE-cadherin and occludin revealed by immunofluorescence in rat mesentery microvessels.¹ Figure 4 shows that Bk induces similar rearrangement of these two junction molecules and that pre-treatment with S1P inhibits the rearrangement. Treatment with S1P alone did not produce any noticeable change from the continuous peripheral distribution of VE-cadherin or occludin that is seen in control images.

Figure 4.

S1P prevents rearrangement of VE-cadherin and occludin in vivo. Segment of a vessel wall dual labelled for VE-cadherin and occludin from a vessel perfused with vehicle solution (Ctrl) shows continuous label for both proteins. (The images are projections from stacks of images taken of three-dimensional vessel wall so that some junctions appear to be incomplete where they pass out of the collection volume.) A representative vessel that was perfused with Bk (10 nM, 5 min) shows discrete gaps (arrowheads) in both VE-cadherin and occludin, as well as spikes oriented perpendicularly to the junction (arrows). The L_p of the Bk-treated vessel was 14 × 10⁻⁷ cm/(s cmH₂O) immediately before fixation (L_p values of Ctrl, S1P-Bk, and S1P vessels were 1.2, 1.4, and 0.8, respectively, i.e. they were low and normal). In vessels pre-treated with S1P (5 µM, 30 min) prior to Bk, there were no prominent gaps or rearrangements of the junction protein patterns (S1P-Bk). Vessels treated with S1P alone (5 µM, 30 min) looked very similar to vehicle controls with continuous junctions and exhibited broad areas of VE-cadherin (asterisks).

3.5. Effect of S1P on Rac activation and cortactin localization in HMVEC culture

For cultured endothelium, S1P increases RacGTP (Rac1 activation) and induces peripheral actin.⁵ Although we had no assay for Rac activation that could be used for the perfused microvessels in situ, we wanted to test for cortactin localization in microvessels using S1P treatment conditions similar to those which activate Rac in cultured endothelial cells (see below). In order to establish the parallel treatments, we first tested the S1P effect on Rac activation and cortactin localization using cultured human dermal microvascular endothelium (HMVEC). The time and concentration dependence of Rac activation in these cells using S1P is shown in Figure 5. There was a strong Rac activation within 1 min relative to vehicle control that declined slowly, but remained above control for 30 min (Figure 5A). The S1P concentration dependence at 1 min showed a pronounced and nearly equal response for 1 and 5 µM (Figure 5B); both 1 and 5 µM S1P produce strong inhibition of the L_p response to Bk (Figure 2). We then used 1 µM S1P to test for effects on cortactin and VE-cadherin localization in HMVEC (Figure 5C). Quiescent cells showed very slight and incomplete peripheral cortactin and continuous peripheral VE-cadherin. Treatment with S1P for 1 min stimulated a strong peripheral localization of cortactin that remained well above control for 30 min. Distribution of VE-cadherin was not changed by S1P treatment in monolayers that were fully confluent and serum starved. These results confirm previous observations that S1P induces Rac activation associated with enhanced peripheral endothelial cortactin localization in cultured endothelial cells. The same S1P stock solutions, final concentration, and timing were used to treat mesenteric microvessels below.

Figure 5.

Effects of S1P on cultured HMVEC. (A) Rac activation, shown over time when exposed to S1P (1 µM), rapidly rises relative to vehicle treatment and remains elevated for at least 30 min (mean ± SEM). (B) Dose–response relationship for Rac activation in the presence of S1P (1 min) shows that level of activation at 1 µM is nearly equal to that at 5 µM. (C) Representative cultures of HMVEC double labelled for cortactin and VE-cadherin show that treatment with 1 µM S1P induces strong peripheral cortactin at both 1 and 30 min relative to vehicle control. VE-cadherin after S1P treatment was not different from fully confluent, quiescent monolayers. Scale bar 20 µm.

3.6. Effect of S1P on cortactin localization in microvessels

To examine the effects of S1P on the localization of cortactin in intact vessels, we compared Bk inflammatory stimulation (10 nM Bk, 5 min) with Bk stimulation in the presence of S1P (Figure 6). In Bk-treated vessels, we were unable to find peripheral cortactin; the VE-cadherin was peripheral, but had numerous small gaps and many lateral spikes as described above and similar to PAF treatment. In contrast, when vessels were pre-treated with S1P (5 µM, 30 min) and then exposed briefly to Bk (in continued presence of S1P), the cortactin was readily seen to be peripheral (Figure 6, arrows). We also examined control vessels treated with S1P alone and found that cortactin was peripherally located (data not shown). In summary, Bk-treated vessels lose peripheral cortactin; S1P prevents the rearrangement that we otherwise see with Bk.

Figure 6.

S1P localizes cortactin to periphery in vivo. Representative vessels, perfused with Bk (10 nM) and pre-treated with S1P (5 µM, 30 min) prior to Bk, respectively, were dual labelled for cortactin and VE-cadherin. The Bk vessel [final L_p was 34 × 10⁻⁷ cm/(scmH₂O)] shows distinctive gaps and lateral spikes (asterisks) in VE-cadherin, whereas the vessel pre-treated with S1P prior to Bk [final L_p was 2 × 10⁻⁷ cm/(scmH₂O)] had continuous VE-cadherin. Cortactin was diffuse in the vessel wall of the Bk vessel, but in the S1P pre-treated vessel, cortactin was clearly enhanced at the periphery of many endothelial cells (arrows) where it is seen near the VE-cadherin in the dual label image. (When these specimens were mounted, the vessels collapsed and therefore the labelled endothelial junctions from opposite sides of the vessel appear to cross one another.)

4. Discussion

Our principal results indicate that S1P attenuated the acute L_p response that was induced by either PAF or Bk in the venular microvessels of rat mesentery. The attenuation of the L_p response corresponded to an inhibition of the rearrangement of VE-cadherin and occludin that was otherwise induced by PAF or Bk. Treatment with S1P also preserved peripheral cortactin localization during Bk challenge in endothelial cells of the mesenteric microvessels. Treatment of microvessels with S1P at 1 µM for up to 30 min induced a partial decrease in baseline L_p, whereas higher concentrations showed no change in L_p. However, in vessels that were perfused with 10 µM S1P for 60 min, the L_p did increase about two-fold. These observations confirm and extend previous reports to show that S1P inhibits the acute permeability from increasing the effects of inflammatory modulators in intact vessels and that this is associated with an enhanced stabilization of peripheral actin-associated junction components. Further, under conditions of our experiments, S1P moderately increases permeability only at the highest concentration used (10 µM) and when present for 60 min.

This study provides further tests using intact microvessels of the hypothesis that stabilization of the endothelial–endothelial adhesion complex can protect against permeability increasing the effects of inflammatory modulators. Activation of Rac1 by action of S1P on endothelial receptors has been reported to both strengthen cultured endothelial monolayers and protect vascular permeability barriers in vivo.^4,6,8,23 The bacterial toxin CNF-1 is known to activate Rac1 in endothelial cells and we previously demonstrated in perfused microvessels that CNF-1 (300 ng/mL, 120 min) pre-treatment results in strong inhibition of the L_p response to PAF.¹⁶ Our present results demonstrate comparable attenuation of PAF-induced increases in permeability by S1P and CNF-1, even though these two agents act via different signalling pathways leading to Rac1 activation. In contrast, we note that one recent study suggests that Rac1 activation via a PAF receptor-activated pathway is associated with barrier breakdown in a cultured human umbilical vein model.²⁴ A weakness of our present results is that we did not demonstrate directly in vivo that Rac1 was activated as a result of S1P treatment. However, the strength of our approach was that we demonstrated, using endothelial cell culture, that the same stock solution of S1P (also dilutions and timing of treatment) that induced strong Rac activation and promoted cortactin localization in culture was also effective in vivo to block the acute L_p increase that otherwise was induced by either PAF or Bk.

We also demonstrated that S1P promoted peripheral cortactin localization and prevented re-organization of either VE-cadherin or occludin. This observation corresponds to previous studies showing that cortactin is translocated to the endothelial cell periphery in cultured cells by direct activation of Rac,¹⁶ by treatment with S1P,²⁵ and by increased cAMP.²⁶ Specifically, cortactin translocation and interaction with Arp2/3 in lamellipodia in endothelial migration has been shown to depend on interaction of S1P1 with Akt kinase.²⁷ To our knowledge, the present data are the first to demonstrate S1P-dependent cortactin translocation to the endothelial cell periphery in intact microvessels. Thus, the results are consistent with the hypothesis that S1P protects the vascular permeability barrier in vivo through stabilization of the adhesion complex.

On the basis of extensive cultured endothelial cell studies, S1P ligation of receptor S1P1 activates a Gi/PI3kinase/Akt-dependent pathway leading to Rac1-dependent stabilization of the peripheral actin cytoskeleton and enhanced endothelial cell–cell adhesion that provides attenuation of acute inflammatory stimulation.^4,27 In the present results, S1P alone provided an incomplete attenuation of the L_p response to PAF or Bk relative to the nearly complete inhibition that we, and others, have demonstrated by stimulation of cAMP (using a combination of rolipram and forskolin), suggesting that cAMP affects additional pathways not stimulated by S1P.^1,6,28 cAMP activates multiple signalling pathways including a Rap1 pathway and the well-known PKA-dependent pathways to both stabilize endothelial adhesion and strongly inhibit inflammatory stimuli.^9,11 The Rap1 pathway may also induce actin stabilization by interaction with the Rac1 pathway by inducing localization near the endothelial periphery of the Rac1 guanine nucleotide exchange factors Tiam1 and Vav2, thus further inducing inflammatory inhibition.^18,19,29 Although these data suggest that activation of the Epac/Rap1 pathway in addition to the S1P/Gi/PI3kinase/Akt/Rac1 pathway could provide inhibition equal to that of cAMP, they do not exclude the possibility of converging effects of cAMP on both PKA- and Epac-regulated mechanisms.^26,29 One interpretation of these results is that signalling pathways that stabilize tight junctions between endothelial cells act in parallel with Rac-dependent pathways that stabilize actin and adherens junctions. Alternative actions of cAMP have been suggested,³⁰ in which specific phosphorylation of vasodilator-stimulated phosphoprotein (VASP) acts to further stabilize endothelial barrier by facilitating Rac1 activation possibly within a scaffolding complex³¹ or phosphor-VASP may contribute to junction stability through integrin recruitment to the cell periphery.³² The action of cAMP to attenuate contractile mechanisms is not likely to be significant in these acute inflammatory responses as we have shown that inhibition of myosin light chain kinase, myosin ATPase, and Rho kinase do not block PAF-induced increases in permeability.¹ We speculate that regulation of Rac1 and Rap1 is modified in endothelial cells during chronic inflammation to induce an endothelial phenotype with reduced capacity to maintain normal permeability. In the absence of chronic inflammation, stabilization of junctions via cAMP pathways or S1P-dependent Rac1 activation provides strong inhibition against the acute permeability increases stimulated by PAF or Bk.

In contrast to the prominent inhibitory effect on acute inflammatory mediators, we found mixed results for the effect of S1P on basal L_p. At the lowest concentration (1 µM), there was a 30% decrease in L_p over 30 min relative to the initial vehicle control, whereas neither 5 nor 10 µM changed L_p over that time. Moreover, in experiments that ran to 60 min, the L_p of the 1 µM group returned to equal the vehicle control value, whereas the 10 µM group increased to twice the vehicle control (as discussed below). The 30% decrease measured with the 1 µM, 30 min treatment is similar to the decrease in L_p previously reported that is induced by increasing intracellular cAMP by treatment with a combination of rolipram (inhibitor of cAMP-specific phosphodiesterase 4) and forskolin (activator of adenylate cyclase).⁹ In the present study, we excluded vessels with a basal vehicle L_p greater than 2.5 × 10⁻⁷ cm/(scmH₂O) as being abnormally high. We found no correlation of the decrease in L_p using S1P (1 µM, 30 min) with the initial basal L_p. In contrast, the only other study of L_p in single vessels found no effect on basal L_p for vessels in the normal range, but did find that S1P was able to reduce the L_p of vessels having higher than normal basal L_p.⁶ Using a variety of cultured endothelial cell types, it was reported that S1P reduces basal permeability (measured as trans-monolayer electrical resistance).⁵ However, this effect in cultured cells appears to be in large part due to induction of cell spreading not associated with cell–cell adhesion.³³

We found that S1P treatment did not induce any change in the basal distribution of VE-cadherin or occludin in the intact microvasculature. Similarly, we recently found that neither cAMP (either stimulated by forskolin and rolipram) nor the cAMP analogue that preferentially activates the Epac/Rap1 pathway induces any change in the basal distribution of VE-cadherin or occludin.⁹ We interpret this to mean that for non-activated in situ endothelium, these adhesion proteins are maximally localized to the junctions. This is in contrast to reports that S1P increases the localization of VE-cadherin in cultured cells with treatments ranging from 10 to 60 min.^33–35 It is known that in many cell culture preparations, the basal state is not as tight as in situ endothelium and there is the ability to modulate the permeability to either lower or higher values. We were unable to find any change due to S1P treatment in the basal VE-cadherin distribution in our cultured HMVEC monolayers. The one element of our culture conditions that seemed ultimately to produce the best monolayers was to seed the cells at very high density (near confluence). Another possible explanation of the difference is that the dermal microvascular cells respond better to culture conditions than the cell types used in the studies mentioned above. If there were any increase in peripheral VE-cadherin (in response to S1P) in either our basal condition cultured monolayers or our in situ endothelium, it was small enough to be not detectable by immunofluorescence.

However, we also noted that when S1P was perfused at the highest concentration (10 µM) for 60 min, the L_p was moderately (about two-fold), but significantly, increased. One speculation is that down-regulation of S1P receptors could account partially for the increase in L_p at 60 min. We also note that Rac1 activation is an essential step in NADPH oxidase-dependent generation of reactive oxygen species (ROS),³⁶ which may be associated with endothelial barrier compromise.³⁷ If the latter is significant, then one predicts that scavenging ROS might prevent an increase in L_p due to the high S1P treatment. Whatever the mechanism, these data indicate that S1P increases permeability if given at a high enough concentration and, if generally true, would limit the potential utility of S1P in clinical therapy. Previous studies suggested that S1P at concentrations above 2 µM induced increased permeability and was associated with activation of stress fibres in cultured endothelial cells.²² Because the 10 µM concentration is well above that necessary to inhibit PAF or Bk, it appears not to compromise the potential utility of S1P to provide a protective effect on the endothelial barrier.

In summary, S1P appears to act in vivo by stimulation of Rac-dependent pathways to stabilize the peripheral endothelial adhesion complex. Although it blocks the acute permeability response to multiple inflammatory mediators, it does not do so as effectively as stimulation of intracellular cAMP. Only at high concentration and with long exposure, does S1P increase permeability. These results from intact microvessels provide further support for the general hypothesis that inflammation can be modulated by enhancing or maintaining endothelial adhesion and support the idea that S1P-dependent pathways are appropriate targets for inflammatory modulation.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Conflict of interest: none declared.

Funding

Supported by the Heart, Lung and Blood Institute at the National Institutes of Health (grant number HL28607).