Control of vascular permeability by adhesion molecules

Abstract

Vascular permeability is a vital function of the circulatory system that is regulated in large part by the limited flux of solutes, water, and cells through the endothelial cell layer. One major pathway through this barrier is via the inter-endothelial junction, which is driven by the regulation of cadherin-based adhesions. The endothelium also forms attachments with surrounding proteins and cells via 2 classes of adhesion molecules, the integrins and IgCAMs. Integrins and IgCAMs propagate activation of multiple downstream signals that potentially impact cadherin adhesion. Here we discuss the known contributions of integrin and IgCAM signaling to the regulation of cadherin adhesion stability, endothelial barrier function, and vascular permeability. Emphasis is placed on known and prospective crosstalk signaling mechanisms between integrins, the IgCAMs- ICAM-1 and PECAM-1, and inter-endothelial cadherin adhesions, as potential strategic signaling nodes for multipartite regulation of cadherin adhesion.

Introduction

Vascular permeability is an innate function of the circulatory system that regulates the flux of fluid, protein, and immune cells from blood to tissue. In most non-inflamed tissues, vascular permeability is controlled by the “barrier” comprised by the microvascular wall, which includes the endothelial glycocalyx, the endothelium, basement membrane, and any accessory cells (i.e. pericytes or smooth muscle cells) wrapped around the outer surface of the vessel. Though each of these components contributes to the permeability of the vascular wall, most studies have focused on the role of the endothelium, which forms an effective barrier to the movement of protein and cells from blood to tissue.^1,2

The endothelium occupies a unique physiological niche, receiving soluble signals from both the blood and tissue, and interacting, directly and indirectly, with cells from both compartments. Direct interactions between the endothelium and its immediate physical environment are mediated by adhesion receptors, which in addition to the cadherins, include the integrins, immunoglobulin-cell adhesion molecules (IgCAMs), junctional adhesion molecules (JAMs), claudins, and occludin. While direct regulation of cadherin adhesion is known to occur downstream of permeability-promoting soluble signals, recent evidence suggests that adhesion events mediated by other adhesion receptors, notably the integrins and IgCAMs, can modify cadherin signaling to effect vascular permeability. In this review, we discuss what is known about regulation of VE-cadherin based inter-endothelial junctions by endothelial adhesion signaling, what affect this has on the modification of vascular permeability in vivo (if known), and identify remaining questions that are critical to understanding the complex crosstalk between endothelial adhesion receptors.

Regulation of Inter-Endothelial Adherens Junctions Under Resting Conditions

The path which solutes, fluid, and cells take through the endothelial barrier is divided into 2 routes. Transcellular permeability occurs via clathrin- and caveaolae-mediated vesicular transport, whereas paracellular permeability occurs via dynamic regulation of inter-endothelial junctions. Regulation of inter-endothelial junctions is controlled at the level of homotypic VE-cadherin adhesion between neighboring cells.³ VE-cadherin expression on the plasma membrane is promoted and stabilized by the expression of cytoplasmic adaptor proteins p120-catenin and β−catenin which bind to the juxtamembrane and C-terminal portion of the VE-cadherin cytoplasmic domain, respectively. β−catenin also mediates the connection between VE-cadherin and the actin cytoskeleton via adaptor proteins such as α−catenin^4,5; this connection is absolutely required for junction maintenance. In addition, VE-cadherin adhesion is regulated by actin cytoskeletal dynamics. Confluent endothelial monolayers, as we would expect to find in vivo, exhibit predominantly circumferential actin fibers (also termed cortical actin) and few radial stress fibers. Many stimuli that affect vascular permeability also induce actin cytoskeletal turnover via the activation of RhoA, which is accompanied by loss of cortical actin and increased radial stress fibers (for review, see ⁶). In unstimulated endothelial cells, vascular permeability is maintained at basal levels by the low level action of Rac1 or Cdc42 GTPases. Under resting conditions, these enzymes stabilize the circumferential actin fibers that support robust VE-cadherin adhesions. Activation of Rac1 or Cdc42 have been noted downstream of S1P⁷ and cAMP/Rap1 signaling,^8,9 both of which are known to stabilize endothelial cell junctions.

Regulation of Inter-Endothelial Adherens Junctions During Inflammation

Much of our knowledge of endothelial junction regulation comes from the study of inflammation. During inflammation the endothelium responds to a complex array of signals and acquires new capacities, i.e. the endothelium becomes “activated.” Endothelial activation is marked by localized leakage of protein-rich fluid and recruitment and activation of circulating leukocytes, accompanied by a breakdown of intercellular junctions and a decrease in barrier function.^1,10 These events are also characteristic of the endothelial dysfunction seen in many disease states, including arthritis and atherosclerosis, and are even observed during the abnormal formation of new vessels.¹¹ Studying the regulation of vascular permeability during the inflammatory response has led to many breakthroughs in the field, however it is important to remember that permeability in and of itself is not pathological. Basal permeability is an essential function of the vascular system. Care should be taken to represent inflammation-induced permeability in terms of increased permeability over baseline, or, as some have termed it, as “hyper-permeability.”

Studies of inflammatory hyper-permeability have revealed that VE-cadherin adhesion is primarily down-regulated by mechanisms that induce the disassembly of the cadherin-catenin adhesion complex. Internalization of the VE-cadherin receptor can occur via clathrin- or cavaeolae- mediated pathways, where it can be targeted for either degradation or recycling. VEGF stimulation, for example, promotes the endocytosis of VE-cadherin downstream of active Src, which promotes the activation of p21-activated-kinase (PAK) and subsequent phosphorylation of VE-cadherin on Ser665. Phosphorylation of this residue promotes the recruitment of β−arrestin-2 and clathrin-mediated internalization.¹² Short-term stimulation with LPS also stimulates clathrin-mediated internalization, however longer treatment with LPS promotes the association of VE-cadherin with caveolin-1, and siRNA mediated knockdown of caveolin-1 rescues VE-cadherin plasma membrane localization.¹³

Junction disassembly can also induce phosphorylation of VE-cadherin or its associated adaptor proteins, p120- and β−catenin. Phosphorylation of VE-cadherin at Y658 or Y731, leads to dissociation of p120 and β−catenin respectively.¹⁴ Other phosphorylation sites on VE-cadherin that have been reported to affect junction stability include Y733, Y645, and Y685.¹⁵ Phosphorylation of p120 at S879 by PKC-α inhibits its association with the VE-cadherin cytoplasmic domain, and increases VE-cadherin internalization.¹⁶ β−catenin is phosphorylated on multiple residues; Y654 and Y489 are targeted by Src/RTK¹⁷ and Abl,¹⁸ respectively, and disrupt binding to VE-cadherin, whereas phosphorylation of Y142 disrupts binding to α−catenin.¹⁹ Phosphorylation of cadherin complex proteins is itself downregulated by the presence of junctional phosphatases, including VE-PTP,²⁰ SHP-2,²¹ and DEP1,²² thus increased expression or activation of these phosphatases can also regulate endothelial permeability.

Integrin signaling in endothelial cells

Integrins form a well-characterized class of adhesion molecules that bind to components of the extracellular matrix (ECM) and IgCAMs. Integrins are expressed as a heterodimer of α and β subunits, which form a calcium-dependent interface required for ligand binding. Integrin adhesion can be induced in a bidirectional manner, e.g., by activating the integrin molecule through intracellular signal transduction, or by ligand binding to the extracellular domains. Subsequent signaling can promote further integrin activation, in addition to many other signaling events. In epithelial monolayers, the occurrence of crosstalk signaling between cadherin and integrin-based adhesions is well established, where assembly of adherens junctions limits integrin expression to the basal surface in epithelial monolayers.^23,24 Integrins can be observed within epithelial cell-cell contacts, but integrin activation is restricted in close proximity to the adherens junction due to local down-regulation of integrin expression²⁵ and depletion of phosphatidylinositol-4,5-bisphosphate (PtdIns(4,5,)P₂,PIP₂).²⁶ While endothelial cell-cell contacts lack the strict segregation of epithelial layers, the same cadherin-adhesion dependent crosstalk mechanisms are assumed to occur in the endothelium.

Quiescent endothelial cells express a wide array of integrins, including modest levels of α₁β₁, α₂β₁, α₅β₁, and α₆β₁, as well as low levels of α₃β₁, α₆β₄, and α_vβ₅.²⁷ During angiogenesis, these cells highly up-regulate α₅β₁ and α_vβ₃, which have been considered targets for anti-angiogenic therapy. In endothelial cells, integrins are required for cell proliferation, migration, and sprouting angiogenesis, among other functions.^27-29

Regulation of the Endothelial Barrier and Vascular Permeability by Integrin Adhesion

β3 integrins have been known to indirectly regulate VE-cadherin adhesion and endothelial permeability via increasing VEGF signaling.^30,31 However, β₃ knockout mice exhibit increased endothelial leak (Miles assay) in response to LPS or VEGF, when compared to wildtype animals.^32,33 Curtis, et al³⁴ demonstrated that blocking β₁ and β₃ adhesion using a soluble peptide containing the β₁/β₃ integrin-binding motif RGD also weakened the transendothelial resistance of unstimulated cells. Application of the RGD peptide also blocked Rac activity and decreased the appearance of cadherin complexes.³⁴ These data suggest that integrin adhesion could promote endothelial barrier function, perhaps by promoting the stability of circumferential actin.

On the other hand, studies performed using either function blocking antibodies to α_vβ₅, or β₅ knockout animals, have shown a decrease in vascular permeability induced by VEGF, TGFβ, thrombin or LPS in vitro, or by mechanical ventilation in vivo.³⁵ β₅ knockout mice also showed increased survival in a mouse model of sepsis.³⁶ In addition, an interesting study by Alghisi, et al interrogated the effect of the α_vβ₃ antagonist cilengitide on endothelial leak. Cilengitide blocked α_vβ₃ adhesion to matrix, but unlike studies performed in β₃ knockouts, induced a redistribution of the integrin to cell edges. There, α_vβ₃ remained active, and induced phosphorylation of VE-cadherin on Y658 and Y731, causing a loss of VE-cadherin from cell contacts and an increase in monolayer leak.³⁷ Thus activation of β₅ or α_vβ₃ integrins appears to down-regulate endothelial barrier function, whereas activation of β₁ integrins promotes endothelial barrier function.

Additional clues can be gathered from the effects of extracellular matrix proteins on vascular permeability. In vivo, one could expect that changes in extracellular matrix expression and availability as ligand could change during vascular injury or in response to an inflammatory stimulus. Indeed, concentrations of fibronectin, a large ECM protein both expressed by the endothelium and a major plasma protein, are increased following vascular injury and during vascular development.^38-42 Though fibronectin is a ligand for α₅β₁ and α_vβ₃, addition of fibronectin to endothelial cell culture blocked TNF−α-induced endothelial leak.³⁴ As TNF−α inhibits β₁ integrin activation, this effect is likely due to an increase in β₁ integrin activity. Similarly, porcine coronary venules treated with an RGD containing peptide that inhibits fibronectin-integrin binding displayed a 2-3-fold increase in permeability to albumin.⁴³ In contrast, treatment of cultured endothelial cells with purified vitronectin or an integrin binding fragment of fibrinogen, both ligands for α_vβ₃, increased VE-cadherin internalization via activation of α_vβ₃.⁴³ In support, vitronectin knockout animals exhibit a blunted increase in permeability in response to ischemic injury,⁴⁴ however full-length fibrinogen, which binds to the endothelium via ICAM-1 and α₅β₁, also induces endothelial leak.⁴⁵ Thus the outcome of endothelial integrin-based adhesion is likely determined by the specific integrins and integrin-ligands expressed in the vascular environment.

Integrin-Adhesion Dependent Regulation of Cadherin Adhesion

Though endothelial integrin adhesion can clearly modulate VE-cadherin based adhesions, the mechanism or mechanisms by which this occurs remain obscure. Multiple signaling pathways are activated downstream of integrin ligation, including many also known to regulate vascular permeability downstream of inflammatory stimuli. These signaling mechanisms can be grouped into 2 large categories, those that regulate cadherin phosphorylation and junction disassembly, and those that mediate changes in the interaction of cadherin-based adhesions with the actin cytoskeleton. While these broad categories are imperfect, we will use these as a framework to briefly discuss what is known about the signals downstream of integrins that regulate vascular permeability.

Regulation of junction disassembly by integrin adhesion

Integrin adhesion activates a number of kinases, including Src,⁴⁶ focal adhesion kinase (FAK),⁴⁷ and integrin-linked kinase (ILK).⁴⁸ Src family kinases phosphorylate residues on VE-cadherin and β−catenin that are important for the cadherin/β−catenin interaction. Though the direct evidence is limited, activation of Src downstream of α_vβ₃ activation has been shown to induce phosphorylation of VE-cadherin at Y658 and Y731.⁴⁹ Integrin adhesion also induces activation of the Src family kinases Fer⁵⁰ and Fyn,⁵¹ which can mediate phosphorylation of Y654 and Y142 of β−catenin. However, in the absence of direct evidence, it is unclear whether Fer or Fyn activation downstream of integrins significantly regulates inter-endothelial junction stability.

Integrin-mediated Src activation also promotes activation of FAK, which traditionally promotes the turnover of integrin-based adhesions. FAK contains a kinase, and multiple protein-protein interaction domains, which are considered key to the temporal regulation of Rho and Rac GTPase. FAK is involved in TGFβ−induced vascular permeability along with its upstream regulator, Src.⁵² An extensive study by Jean, et al found that in tumor-associated endothelial cells, human lung endothelial cells, and human umbilical vein endothelial cells, FAK inhibition decreases VE-cadherin Y658 phosphorylation. FAK can also be activated downstream of growth factor receptors. FAK activation is required for VEGF-dependent phosphorylation of VE-cadherin Y658 and recent studies have demonstrated that FAK can directly phosphorylate this residue (Fig. 1). Pulmonary endothelial cells isolated from FAK kinase-dead knock-in mice showed decreased monolayer leak and decreased Y658 phosphorylation.⁵³

Figure 1.

Intracellular regulation of VE-cadherin by integrins. Activation of α_vβ₅ or α_vβ₃ causes activation of c-Src (Src) and subsequent activation of focal adhesion kinase (FAK). FAK phosphorylates tyrosine 658 on VE-cadherin, prompting loss of cadherin adhesion. FAK also phosphorylates paxillin (pax), which leads to the activation of Rho GTPase and the formation of radial stress fibers. Alternatively, the Src-family kinases Fer and Fyn phosphorylate tyrosines 654 and 142 on β−catenin. While Fer and Fyn can be activated downstream of integrin ligation, no evidence yet exists for integrin-mediated phosphorylation of β−catenin. Activation of β1 and β3 integrins conversely promotes cadherin adhesion. Activation of FAK downstream of sphingosine-1-phosphate recptor-1 (S1P1R) as well as integrin-dependent activation of vasodilator-stimulated phosphoprotein (VASP) triggers activation of Rac1 GTPase. Rac then promotes the stability of cortical actin and of cadherin adhesions.

On the other hand, numerous studies have linked FAK activation to barrier-promoting effects. Embryonic endothelial cells derived from FAK knockout mice exhibit increased permeability compared with wild type.⁵⁴ Loss of FAK expression in pulmonary artery endothelial cells, results in extended barrier disruption after thrombin stimulation.⁵⁵ In addition, tyrosine phosphorylation of FAK in pulmonary artery endothelial cells is associated with barrier enhancement,^56-58 and occurs downstream of the barrier-promoting activity of S1P⁵⁹ (Fig. 1). Parsing the barrier-promoting and -disrupting effects of FAK activation is a significant challenge, not the least because of the diversity of downstream signals modulated by FAK. It is extremely likely that integrin-mediated activation of FAK does play a role in regulation of the vascular barrier, however lack of direct evidence, and the absence of models in which the relative contribution of integrin-activated FAK can be measured, limits our ability to define its precise role.

Regulation of endothelial actin cytoskeletal turnover by integrin adhesion

Changes in actin turnover are a well-characterized outcome of integrin adhesion signaling. Integrin adhesion can activate Arp2/3 to promote actin polymerization and branching. Integrin signaling also activates multiple members of the Rho GTPase family, including RhoA, Rac1 and 2, and Cdc42. As previously discussed, Cdc42 and Rac activation are important for formation of circumferential actin bundles that are associated with stable cell-cell contacts. Heightened Rac activity can also block RhoA activation, which is associated with increased actin-myosin contractility and is generally believed to negatively regulate endothelial cell-cell contact and promote vascular permeability. However, RhoA-mediated activation of the formin mDia inhibits VEGF-stimulated permeability by sequestering Src and preventing VE-cadherin internalization,⁶⁰ suggesting that there may be conditions in which RhoA activation may not exert negative regulatory pressure on endothelial permeability.

In addition, many proteins that regulate actin filament dynamics and are important for endothelial barrier regulation are also components of the integrin cytoplasmic adhesion complex. For example, phosphorylation of paxillin via an Src-FAK-ERK mechanism stimulates increased permeability in HPAEC and is associated with activation of RhoA.⁶¹ Vasodilator stimulated phosphoprotein (VASP) is classically associated with actin filament polymerizaton downstream of integrin activaton. VASP also mediates cAMP-mediated Rac1 activation during cell-cell contact formation.⁶² VASP null endothelial cells exhibit impaired β₁ integrin dependent adhesion and VE-cadherin adhesion, though cadherin expression and localization appears normal. VASP deficiency also correlates with reduced actomyosin contractility and Rac dependent formation of cortical actin, which likely contributes to the loss of integrin and cadherin mediated adhesion.⁸

α−actinin is an actin crosslinking protein involved in the maturation of integrin-based adhesions. α−actinin can also link the VE-cadherin complex to the actin cytoskeleton by simultaneously binding α−catenin and actin and promoting cortical actin rearrangement.⁶³ Vinculin is a scaffolding protein that interacts with multiple proteins in the integrin cytoplasmic adhesion complex.⁶⁴ Vinculin is associated with mature focal adhesions where it stabilizes integrin-based adhesions and permits the transmission of actin-generated mechanical forces across the cell membrane.⁶⁵ Vinculin is also recruited to cadherin-based adhesions by α−catenin where it reinforces the connection of the actin cytoskeleton with VE-cadherin⁶⁶ (Fig. 1). The association of α−actinin and vinculin with both integrin- and cadherin-based adhesions suggests the possibility that there could be a competition-based crosstalk between the 2 types of adhesion receptors. This would depend on α−actinin and vinculin concentrations, as well as the number and stability of integrin and cadherin adhesions. Both α−actinin and vinculin are highly abundant, as are integrins and cadherins, in most cell types. However, local control of protein concentrations could be sufficient to exert negative regulatory pressure due to competition for these cytoplasmic actin-binding proteins. This hypothesis has yet to be explored to any significant extent, and parsing the individual and communal roles of α−actinin and vinculin in integrin and cadherin signaling will require future studies.

Endothelial IgCAM-Mediated Regulation of Endothelial Cell-Cell Contact and Vascular Permeability

Endothelial immunoglobulin superfamily cell adhesion molecules (IgCAMs) are key regulators of leukocyte-endothelial interaction. In the vascular endothelium, intercellular adhesion molecule -1 (ICAM-1) is expressed constitutively at low levels and acts as a ligand for β₂ integrins expressed on circulating leukocytes. ICAM-1 ligation by leukocyte integrins stimulates the firm adhesion required for extravasation. VCAM-1 is induced by cytokine stimulation of the endothelium, and is a ligand for α4β₁ and α4β₇ integrins expressed on leukocytes. VCAM-1 expression is important for leukocyte adhesion and rolling,^67-70 and to some extent leukocyte extravasation,^71-73 but its involvement in the regulation of vascular permeability is unclear. PECAM-1 is highly expressed by endothelial cells, as well as on the surface of platelets, monocytes, neutrophils, and some types of T-cells. PECAM-1 expression is concentrated at endothelial cell-cell contacts, where it mediates leukocyte diapedesis via homotypic PECAM-1 interactions with the transmigrating cell.⁷⁴ Similar to PECAM-1, junctional adhesion molecules (JAMs) are also found at endothelial cell-cell contacts, and mediate both endothelial-endothelial and endothelial-leukocyte interactions (reviewed in ⁷⁵). A thorough review of IgCAM adhesion signaling in inflammation was published earlier this year,⁷⁶ so our treatment of this topic will be brief.

ICAM-1

Activation of ICAM-1 signaling leads to increased leukocyte transmigration, increased vascular permeability and loss of the endothelial barrier, and rearrangement of the actin cytoskeleton. ICAM-1 activation also stimulates the downregulation of endothelial junction proteins VE-cadherin, occludin, and the tight junction adaptor proteins ZO-1 and ZO-2.⁷⁷ Most studies have examined the effect of ICAM-1 mediated signaling following the activation of inflammatory signaling pathways. However, recent evidence suggests that ICAM-1 signaling can be activated by ligation in the absence of inflammatory stimulus. Sumagin, et al⁷⁸ demonstrated that rolling leukocytes or antibody-mediated ICAM-1 crosslinking could stimulate an ICAM-1 dependent increase in vascular permeability in unstimulated mouse cremaster arterioles. This is supported by other work showing that over-expression ICAM-1 in unstimulated endothelial cells decreases endothelial barrier function.⁷⁹ Crosslinking ICAM-1 with ICAM-1 coated beads also promotes endothelial leak, as well as recruiting VCAM to sites of ICAM-1 clustering.⁸⁰ Treatment of cells with TNF−α or VEGF⁸¹ leads to an up-regulation of ICAM-1 expression after as little as 4 hours of treatment. Increased ICAM-1 expression is subsequently necessary for endothelial permeability and leukocyte transmigration. TNF−α stimulation has also been shown to induce a redistribution of ICAM-1 in vivo, where the heterogeneous distribution of ICAM-1 regulates the location of leukocyte adhesion.⁸²

Functionally, ICAM-1 activity appears to require both the extracellular and intracellular domains. Blocking antibodies directed against the extracellular domain of ICAM-1 reduce leukocyte adhesions and vascular permeability in both mouse pial and cremaster vessels.^83,84 Studies employing a truncation mutant of ICAM-1 lacking the cytoplasmic domain have demonstrated that this domain is required for ICAM-1-mediated leukocyte transmigration, but not leukocyte-endothelial adhesion.^85,86 The studies by Greenwood, et al and Sumagin, et al utilized a cell permeable cytoplasmic domain construct that inhibits endogenous ICAM-1 activity. As in the truncation experiments, treatment of brain microvascular cells or mouse cremaster venules with penetratin-ICAM-1 peptide strongly inhibited transendothelial migration but did not affect leukocyte adhesion.^85,87 Furthermore, treatment with this inhibitory peptide decreased the number of gaps in VE-cadherin staining observed in venules in response to fMLP.⁸⁷ However, in human dermal microvascular cells, the absence of the cytoplasmic domain was not able to inhibit ICAM-1-induced loss of endothelial barrier function.⁷⁹ Based on the relatively small number of studies, and the concept that transendotheilal migration is accompanied by an increase in solute translocation across the endothelium, it is not appropriate at present to conclude that cytoplasmic ICAM-1 signaling is uninvolved in the regulation of vascular permeability. However, just how the relatively short ICAM-1 cytoplasmic domain would mediate such regulation is not abundantly clear.

The ICAM-1 cytoplasmic domain has been reported to bind to several actin binding proteins, including α−actinin,⁸⁸ cortactin,⁸⁹ and ezrin.⁹⁰ Recent evidence suggests that binding to some of these partners is independent,⁸⁸ indicating that several distinct pools of ICAM-1 receptor may be present in the cell. Ligation of ICAM-1 transduces a signal that causes rearrangement of the actin cytoskeleton, likely via increasing cytoplasmic calcium levels,⁹¹ though the exact mechanism remains unclear. ICAM-1 ligation has been shown to activate Rac1, RhoA, PKC, Src family kinases, and the docking protein p130Cas.⁹² Inhibition of RhoA in ICAM-1 over-expressing cells mimics the effects of truncating the cytoplasmic domain on leukocyte transmigration and adhesion,⁸⁶ suggesting that RhoA signaling is a major regulator of ICAM-1-mediated leukocyte-endothelial interactions. Src activation downstream of ICAM-1 ligation induces phosphorylation of caveolin-1, which is required for transcellular vascular permeability following neutrophil adhesion.⁹³ Rac1 activation downstream of ICAM-1 mediates the activation of NADPH oxidase (Nox) and the production of reactive oxygen species (ROS). The rise in ROS levels promotes activation of the kinases Src and Pyk2, which can phosphorylate VE-cadherin on Y658 and Y731.⁹⁴ Interestingly, crosslinking VCAM-1 also stimulates calcium signaling and activation of Nox.⁹⁵ Analogous to ICAM-1 signaling, Vockel, et al demonstrated that VCAM-1-mediated ROS production also leads to activation of Pyk2⁹⁶ (Fig. 2). They show that this pathway stimulates the dissociation of VE-cadherin and VE-PTP to disrupt endothelial cell contacts, leaving open the question of whether VCAM-driven Pyk2 also stimulates the phosphorylation of VE-cadherin.

Figure 2.

Intercellular regulation of VE-cadherin by ICAM-1 and VCAM-1. Ligation of ICAM-1 recruits the actin binding proteins α-actinin (α−act), ezrin (ezr), and cortactin (cort). ICAM-1 activity leads to activation of Rac, which is a regulatory component of NADPH oxidases (NOX). Nox activation following binding of GTP-bound Rac stimulates the formation of reactive oxygen species (ROS). ROS can subsequently activate Src kinase and protein tyrosine kinase 2β (Pyk2). Src and Pyk2 then phosphorylate Y658 on VE-cadherin and stimulate junction disassembly. In addition, ICAM-1 activates Rho GTPase, leading to stress fiber formation, as well as PKCs (PKC), which mediate ICAM-1 dependent increased vascular permeability in the absence of an inflammatory stimulus. Ligation of VCAM-1 also activates Pyk2 to promote the loss of interaction between VE-PTP and VE-cadherin, increasing phosphorylation of junction proteins and stimulating junction disassembly.

PECAM-1

PECAM-1 expression, on the other hand, blocks neutrophil migration in response to TNF−α.⁹⁷ PECAM-1 expression is increased by barrier promoting substances like statins,⁹⁷ and reduced by barrier disrupting agents like TNF−α and ionizing radiation.⁹⁸ Knockdown of PECAM in endothelial cells in culture increases monolayer leak.⁹⁹ Blocking PECAM-1 using a function-blocking antibody decreases neutrophil-endothelial interactions and decreases permeability in multiple disease models.^100,101 PECAM-1 knockout mice demonstrate pulmonary fibrosis preceded by enhanced vascular leakiness and deposition of hemosiderin in the lung.¹⁰² PECAM knockouts also exhibit increased survival when challenged with endotoxin,¹⁰³ and an earlier incidence of experimental autoimmune encephalitis.¹⁰⁴ In addition, endothelial cells isolated from PECAM deficient animals demonstrate increased leak in response to LPS¹⁰³ and histamine.¹⁰⁴ While the intracellular pathways regulated by PECAM-1 are not completely defined, PECAM-1 mediated interactions can lead to activation of PLC–γ,¹⁰⁵ Akt and upregulation of eNOS via STAT3,¹⁰⁶ or increased Gαq/11 signaling.¹⁰⁷ One interesting signaling mechanism involves the Immunoreceptor Tyrosine-based Inhibitory Motif, or ITIM domain, in the PECAM-1 cytoplasmic tail. Phosphorylation of tyrosines 663 and 686 in the ITIM domain promotes binding of the tyrosine phosphatase SHP-2.¹⁰⁸ PECAM-1 can also bind to β−catenin. However, in contrast to SHP-2 binding, β−catenin tyrosine phosphorylation -rather than PECAM phosphorylation- was required for this association. This suggests that PECAM, β−catenin and SHP-2 form a ternary complex, which brings tyrosine-phosphorylated β−catenin in close proximity to a phosphatase. In fact, expression of an ITIM-defective PECAM-1 mutant severely increases β−catenin tyrosine phosphorylation, supporting the idea that PECAM-1 coordinates the dephosphorylation of β−catenin, and perhaps the subsequent stabilization of β−catenin/VE-cadherin association, via its interaction with SHP-2.

Regulation of Vascular Permeability by Cadherin Adhesion

Finally, we would like to quickly pay attention to a rather overlooked facet of cadherin signaling that may also play a significant role in the regulation of vascular permeability. Dissociation of p120 and β−catenin from VE-cadherin can stimulate feedback mechanisms to both promote and disrupt cadherin adhesion. As our understanding of endothelial junctions moves past the concept of junctions as static structural elements to one that encompasses the highly dynamic nature of the cadherin adhesion, it becomes clear that p120 and β−catenin likely function to modulate cadherin adhesion when both bound to- and dissociated from- the receptor.

p120 regulates local RhoA activation by recruiting the RhoA specific GTPase activating protein, p190RhoGAP, to the cadherin cytoplasmic complex.¹⁰⁹ However, in fibroblasts, cytoplasmic p120 can bind directly to RhoA and inhibit its intrinsic ability to dissociate guanosine diphosphate,¹¹⁰ similar to the action of RhoGDI. In endothelial cells, overexpression of p120 blocks neutrophil transmigration.¹¹¹ However, this study from the Luscinskas laboratory determined that this was not due to the binding of p120 and RhoA, but rather to the ability of cytoplasmic p120 to inhibit the association of active Src with VE-cadherin. Others have shown that in Chinese hamster ovary cells, p120 can bind to Vav2, an exchange factor for Rac1, Cdc42, and RhoA, and mediate Rac1 activation.¹¹² These data suggest that conditions that increase cytoplasmic p120 levels would limit RhoA activation and perhaps impair actin cytoskeletal reorganization. However, loss of p120 from the cadherin adhesion complex can also lead to the nuclear localization of p120.¹¹³ How p120 is translocated into the nucleus is unknown. What is known is that once there, association of p120 with the transcription factor Kaiso leads to increased expression of Rho in corneal endothelial cells.¹¹⁴ In bovine pulmonary artery endothelial cells and human brain endothelial cells, Kaiso and p120 co-immunoprecipitate. Knockdown of p120 increased the expression of a Kaiso reporter construct 2-fold but also significantly reduced Kaiso protein levels.¹¹⁵ The study by O’Donnell et al in brain microvascular and pulmonary artery endothelial cells supports the idea that p120 depletion increases the transcriptional activity of Kaiso but also suggests that loss of p120 activates NFκB and AP-1¹¹⁶. Further studies are needed to fully understand the role of nuclear p120 signaling in the endothelium. However, in epithelial cancers, loss of cadherin function is associated with an increase in nuclear Kaiso and a worse prognosis. This suggests the possibility that loss of VE-cadherin mediated adhesion could trigger changes in endothelial cell morphology or gene expression due to an increase in the cytoplasmic or nuclear localization of p120, respectively. More information is needed about the levels of p120 required to gain effective Rho inhibition, how those levels relate to VE-cadherin protein expression, and what genes are regulated by nuclear p120, before a realistic model of p120 regulation of vascular permeability, incorporating all cellular pools of p120 catenin, can be formed. Feel free to speculate wildly.

The binary signaling properties of β−catenin, on the other hand, have been well established, if not well mapped, in endothelial cells. In addition to its role as a cytoplasmic adapter for cadherin adhesion, β−catenin is the key second messenger in canonical Wnt (Wingless and INT-1) signaling. In this pathway, Wnt binding to the cell surface receptor Frizzled (Fzd) and co-receptor LRP5/6 causes intracellular association of disheveled (Dvl) with Fzd. Activation of Dvl causes inactivation of the constitutive β−catenin destruction complex (axin/adenomatous polyposis coli (APC)/glycogen synthase kinase 3β (GSK3β)) that normally targets cytoplasmic β−catenin for proteosomal degradation.^117,118 β−catenin then is free to translocate into the nucleus, where it interacts with the TCF/LEF family of transcription factors (reviewed in¹¹⁸). Wnt signaling in microvascular endothelial cells increases the proportion of cytoplasmic β−catenin, TCF/LEF transcriptional activity, proliferation and tube formation.^119,120 Expression of Wnt-1 in human umbilical vein endothelial cells stimulates proliferation and survival and promotes the expression of IL-8.¹²¹ In resting adult vasculature, β−catenin is rarely seen in the cytoplasm or nucleus.¹²² However, β−catenin is targeted to the nucleus by inflammatory stimuli,^113,123 suggesting nuclear β−catenin could play a role in the loss of the endothelial barrier and vascular permeability.

Disassembly of adherens junctions, such as after endothelial exposure to activated neutrophils, is accompanied by the release of β−catenin.^14,124 Some inflammatory mediators, such as PGE2¹²⁵ and thrombin,¹¹³ can inhibit GSK3β, thus blocking cytoplasmic β−catenin degradation. Others, such as VEGF, can increase nuclear β−catenin activity by stimulating nitrosylation of β−catenin.¹²⁶ Nitrosylation of β−catenin by VEGF-induced eNOS, or β−catenin nitration by macrophage derived NO,¹²⁷ leads to the shuttling of β−catenin into the nucleus. There, it can activate transcription by binding to the TCF/LEF family of transcriptional repressors. Interestingly β−catenin nitration appears to also promote the interaction of nuclear β−catenin with NFκB p65.¹²⁷ However, β−catenin nuclear localization following loss of VE-cadherin adhesion does not necessarily require inactivation of the degradation complex or additional post-translational modification of β−catenin. Taddei, et al showed that VE-cadherin null cells exhibit higher levels of nuclear β−catenin.¹²⁸ Our lab has shown that loss of the junctional scaffolding protein KRIT1 leads to loss of VE-cadherin adhesion and increased nuclear β−catenin in vitro, and increased vascular permeability in vivo.^129,130 Thus modification of inter-endothelial cell contacts is capable of switching the predominant mechanism of β−catenin signaling from junctional to nuclear.

β−catenin dependent transcriptional activity encompasses the up- and down-regulation of a wide variety of genes, in a cell context-dependent manner. Taddei et al showed that β−catenin, in concert with FOXO1, suppressed the expression of the tight junction protein claudin 1¹²⁸. In KRIT1 deficient endothelial cells, nuclear β−catenin promotes the expression of cyclin dD1, fibronectin, and VEGF-A¹³⁰. Other β−catenin target genes (e.g., c-myc, cyclinD1, cox-2, IL-8) have been linked to inflammation.^121,131,132 The β−catenin dependent transcriptome in endothelial cells is likely to vary significantly from that in epithelial cells, as well as between in vitro and in vivo conditions. However, there appears to be sufficient evidence to consider nuclear β−catenin signaling when examining the regulation of vascular permeability, particularly under pathological or stimulated conditions.

Concluding Remarks

Endothelial cells in vivo mediate their attachment to their environment through a panoply of adhesion receptors. Some of these are active only during inflammation, whereas others function in both stimulated and unstimulated conditions. Even in vitro, endothelial cells plated on tissue culture plastic are exposed to extracellular matrix molecules found in calf serum, and secrete an extensive matrix of their own within 24 hrs of plating (Dr. Glading, unpublished observation). Cell-cell contacts of cultured endothelial cells contain plentiful PECAM-1, and resting cells express low levels of ICAM-1. Thus the components necessary for “other” adhesion receptors to modify the responses measured in the laboratory are present and accounted for. The absence of an extensive body of literature describing in detail how other adhesion molecules regulate inter-endothelial cell contacts, endothelial barrier function, and vascular permeability rather underscores the inattention this area has received. When one considers how to approach this problem, the presence of common signaling nodes in inflammatory signaling pathways, integrin signaling, and IgCAM signaling spark a certain automatic focus. Activation of Src family kinases, alterations to the connection between cadherins and the actin cytoskeleton, and cytoskeletal regulation by RhoA and Rac1 appear to be key regulatory checkpoints in the regulation of endothelial cell-cell contact and vascular permeability. Fully understanding how each of these fundamental signaling nodes is regulated in vitro, whether in the context of adhesion receptor signaling, or during inflammation, may be the necessary basis needed to address the complexities of an in vivo system. Issues that should be considered when approaching these questions include variation in signaling due to endothelial cell origin, which should not be underestimated. Rather than being treated as unwanted sources of error, these variations likely provide vital clues to the physiological differences between vascular beds. In addition, cells may use different mechanisms to regulate basal permeability vs. inflammation-mediated hyper-permeability, thus the unstimulated condition should not be overlooked.

All cellular functions operate in the context of expression of a specific subset of genes. Stimulation of endothelial cells or intact vessels clearly triggers significant changes in the cellular transcriptome. While the nuclear functions of p120- and β−catenin have been known for quite some time, the impact of these binary signaling molecules on the endothelial barrier and vascular permeability has been artificially restricted to their function at cytoskeletal adaptors. Recent evidence suggests that nuclear functions of p120- and β−catenin could underlie a novel regulatory mechanism for vascular permeability.¹³⁰ Recent technological advances should make it possible to investigate nuclear p120- and β−catenin contributions to vascular permeability on a large scale. Nevertheless, the complexity of signaling mechanisms regulating the endothelial barrier and vascular permeability appear to be expanding exponentially, meaning that more and more sophisticated techniques, perhaps merging in vivo physiological measurements with computational biology, may be needed to assess this emerging regulatory system.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.