OUP user menu

The regulation of leucocyte transendothelial migration by endothelial signalling events

Mar Fernandez-Borja, Jaap D. van Buul, Peter L. Hordijk
DOI: http://dx.doi.org/10.1093/cvr/cvq003 202-210 First published online: 12 January 2010


Leucocytes use sophisticated mechanisms to cross the endothelium lining the vasculature. This is initiated by chemokine- and adhesion molecule-induced intracellular signalling that controls adhesion, spreading, and motility. At the same time, adherent leucocytes trigger the endothelium, manipulating the barrier to promote their transmigration into the underlying tissues. Over the past years, our insights in the associated signalling events within the endothelium have increased considerably, albeit the order of events, their crosstalk, and the consequences for endothelial cells and leucocytes are only partially resolved. Here, we briefly review endothelial signalling that is initiated at the apical endothelial membrane, where the first contact with the leucocytes takes place and signal transduction is induced. In addition, we discuss subsequent events at endothelial cell–cell junctions insofar as they have been linked to transendothelial migration. Finally, we briefly touch upon the modulation of endothelial signalling by infectious pathogens, since these have developed additional, elegant ways to manipulate the endothelium and transendothelial migration that may provide new, relevant insights into this process.

  • Endothelial cells
  • Signal transduction
  • Transmigration
  • VE-cadherin
  • ICAM-1

1. Signalling at the apical endothelial membrane

The first step of leucocyte attachment involves the rolling of the leucocytes over the endothelium. This step is mediated by selectins. Yoshida et al.1 described that leucocyte clustering of E-selectin induces the linkage to the actin cytoskeleton. Interestingly, they show that the clustering induces the recruitment of several actin-binding proteins such as α-actinin, filamin, vinculin, paxillin, and focal adhesion kinase (FAK). Also the clustering of P-selectin results in changes in the actin cytoskeleton.2 For a thorough discussion on selectins, the reader is referred to an excellent review.3

At this point, the integrin ligands intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) become clustered in the plane of the membrane and concentrate in rings around the adherent leucocyte4,5 (Figure 1). The β2-integrins LFA-1 (lymphocyte function-associated antigen-, αL β2) and Mac-1 (macrophage-1 antigen, αM β2) bind to the first and third extracellular Ig-like domain of ICAM-1, respectively.6,7 The β1-integrin subunit when dimerized to the α4 subunit (VLA-4; very late antigen-1) binds the first and fourth Ig-like domain of VCAM-1. Also the α4β7 integrin binds to VCAM-1.8,9 ICAM-1 (CD54) can form dimers on the surface of the endothelium.10 However, ICAM-1 dimerization is not required for optimal binding to LFA-1.11

Figure 1

Schematic overview of proximal signalling events in endothelial docking structures, induced upon firm leucocyte adhesion. Activation of guanine nucleotide exchange factors by clustered integrin ligands such as ICAM-1 mediates the activation of a subset of Rho GTPases that are involved in feedback signalling towards the docking structure. In this signalling, RhoA is involved in additional clustering of ICAM-1/VCAM-1, and RhoG and Rac1 mediate membrane ruffling and the formation of the docking structure by the apical membrane. The DOCK/ELMO complex is presumed to mediate crosstalk between the RhoG and Rac1 GTPases.

The adhesion-induced clustering of ICAM-1 or VCAM-1 on the surface of the endothelium induces intracellular signalling in the endothelial cells.12,13 One of the molecules that is activated upon crosslinking of ICAM-1 is the small GTPase RhoA.14,15 Its activation results in the bundling of existing F-actin filaments into stress fibres.16 Nevertheless, blocking RhoA activity with C3-toxin from Clostridium botulinum does not affect adhesion of leucocytes to the endothelium but does prevent the clustering of ICAM-1 and thus the initiation of intracellular signalling.15,17 This indicates that ICAM-1 clustering is primarily regulated in an inside-out fashion by the endothelial cells, likely following an initial signal generated by engaged, but unclustered ICAM-1.

Crosslinking of ICAM-1 results in various effects, including the production of inositol-phosphate and the phosphorylation of phospholipase Cγ.18 Moreover, clustering of ICAM-1 rapidly triggers the mobilization of intracellular calcium. In addition, calcium-dependent activation of protein kinase C (PKC) regulates the phosphorylation of the Src tyrosine kinase and its substrate cortactin.19,20 In parallel, additional cytoskeletal and regulatory proteins become phosphorylated on tyrosine, including paxillin, FAK, and p130Cas downstream from ICAM-1 clustering. Interestingly, the phosphorylation of these proteins appears also to be dependent on RhoA activation, as was assessed using the C3-toxin.18 Finally, GST-pull-down assays showed that the adapter protein CrkII interacts with paxillin and p130Cas upon ICAM-1 clustering, which underscores the complexity of intracellular signalling initiated by clustered ICAM-1.

Using small microsphere beads (100 nm) coated with anti-ICAM-1 antibodies Muro et al.20 showed that ICAM-1 dimers are internalized. This internalization depends on signalling molecules PKC, Src, and RhoA, molecules that have been shown to play an important role in proper leucocyte transendothelial migration as well. Recently, Ridley and co-workers21 showed that ICAM-1 is involved in the transcellular migration of lymphocytes and that ICAM-1 remains bound to the leucocyte during the passage through the endothelial cell body. ICAM-1 thus ends up at the baso-lateral site of the endothelium, together with the leucocyte. This process depends on the membrane-associated signalling regulator caveolin-1, based on the finding that siRNA-mediated reduction of caveolin-1 levels prevented leucocyte diapedesis. Interestingly, very recently, Sverdlov et al.22 reported that filamin A regulates caveolin-1-mediated internalization of caveolae. Studies from our lab on ICAM-1 clustering have shown that filamin A and caveolin-1 bind to the intracellular tail of ICAM-123 and mediate docking structure formation and transendothelial migration. These data suggest that filamin A and caveolin-1 could be involved in the transcellular route of transmigration, i.e. through the cell body of the endothelium, rather than the paracellular route, i.e. through the endothelial cell–cell junctions. Earlier studies using electron microscopy already indicated that the transcellular pathway is used in vivo.24 Whereas the existence of both routes of transmigration is generally accepted, differences in their prevalence may depend on the type of leucocyte and its state of activation as well as the source of the endothelium. For more details the reader is referred to these excellent reviews.25,26

Upon binding of LFA-1 to ICAM-1 or antibody-mediated clustering, ICAM-1 is incorporated into detergent-insoluble membrane domains, referred to as lipid rafts.27 These domains may serve as signalling platforms and induce a variety of intracellular signals into the endothelium. Recent literature suggests that tetraspanins are involved in the clustering of ICAM-1 and VCAM-1 within the apical surface of the endothelium.28,29 The tetraspanins CD9 and CD151 are required for proper ICAM-1/VCAM-1 function and similarly, the tetraspanin CD81 was found to mediate ICAM-1 clustering.29 From a pathological point of view, it is interesting to note that CD81 RNA expression is increased in early atherosclerotic plaques.30 As a consequence, ICAM-1 function may be enhanced in areas of early lesions, resulting in increased leucocyte accumulation. CD9 and CD151 localize to the so-called ‘docking structures’ that are formed by the endothelial cells upon adhesion of leucocytes to ICAM-1 and/or VCAM-1.4,5 These structures are actin-rich membrane protrusions that form around an adhered leucocyte. So far, the mechanisms governing the regulation of these structures are not fully defined. The small GTPase RhoG, which is activated upon ICAM-1 clustering, mediates the formation of these structures.31 However, also the related GTPase Rac1 is activated upon ICAM-1 clustering. It is interesting to note that RhoG activation gives rise to dorsal membrane ruffles, whereas Rac1 is more involved in lateral ruffles.32,33 Both GTPases may thus play an important, complementary role in inducing the docking structures around the adhered leucocytes. Several studies have shown that RhoG can activate Rac1 through the protein complex ELMO-DOCK180 and the adapter protein CrkII.32,33 Interestingly, RhoG is involved in caveolae-mediated endocytosis and active RhoG co-localizes with caveolin-1.34 As discussed above, Millan et al.21 found ICAM-1 co-localizing with caveolin-1 and observed a translocation of these two proteins to the ends of F-actin stress fibres at the baso-lateral side of the endothelial cells. Thus, these data suggest a complex network of ICAM-1-induced signalling orchestrated by RhoA, RhoG, and Rac1, paralleled by ICAM-1 clustering and binding to specific adapters such as filamin A and caveolin-1. It remains to be demonstrated to what extent the signals that lead to the formation of docking structures in fact represent a starting point for transcellular leucocyte transmigration, or whether additional, unknown events in or on the endothelial cells determine the use of one route vs. the other.

Recently, invasive podosomes formed by the leucocytes have been implicated in transcellular migration.35 Shear force was found to promote the formation of invasive filopodia in lymphocytes, prior to transmigration. Formation of these structures was accompanied by the chemokine-induced, strong activation of LFA-1.36 The relationship between these filopodia and invasive podosomes is currently unclear and warrants further research. Following up on the, largely biochemical, data that underscore the role for specific GTPases downstream from ICAM-1, a next step will be to establish the localized signals that occur simultaneously in the adherent leucocyte and the endothelial cell at the subcellular level.

2. Signalling at endothelial junctions

Endothelial cell–cell adhesion is primarily mediated by two multiprotein complexes known as adherens and tight junctions.37 In contrast to epithelial cells where these complexes are spatially separated, in endothelial cells adherens and tight junctions are intermingled and regulate each other. Tight junctions have a more prominent role in brain microvasculature where they are essential to maintain the blood-brain barrier. Tight junctions contain several transmembrane adhesive proteins of the claudin, occludin, and junctional adhesion molecule (JAM) families and endothelial cell-selective adhesion molecule (ESAM). Intracellularly, these proteins are linked to the actin cytoskeleton through the scaffold proteins ZO-1 and ZO-2. JAMs and ESAM modulate leucocyte transendothelial migration and barrier function. JAM proteins are able to bind leucocyte integrins,38 thereby facilitating leucocyte transendothelial migration. Due to space limitations, the reader is referred to an excellent review on the role of the JAMs in inflammation.39

The major component of adherens junctions is vascular endothelial (VE)-cadherin. VE-cadherin is a transmembrane protein that establishes homotypic calcium-dependent interactions with its extracellular domain. Although many proteins have been implicated in endothelial cell–cell adhesion,37,40 VE-cadherin has a central role in the regulation of the integrity of the endothelial barrier and leucocyte transmigration as evidenced in vitro and in vivo (Figure 2). Antibodies directed against the extracellular domain of VE-cadherin effectively disrupt endothelial cell–cell adhesion, increase vascular permeability, and allow the passage of leucocytes, without affecting tight junctions.41,42 The cytoplasmic tail of VE-cadherin binds to catenin proteins, which are required for strong adhesion. The juxtamembrane domain binds p120 catenin, while the membrane distal domain binds β-catenin and plakoglobin in a mutually exclusive fashion that depends on cell–cell contact maturation. Finally, α-catenin alternately associates to β-catenin/plakoglobin or to the actin cytoskeleton.

Figure 2

(A) Stable junctions. In the steady state, endothelial intercellular junctions are sealed by the homotypic interaction of VE-cadherin molecules from adjacent cells. Association of the VE-cadherin intracellular tail with proteins of the catenin family (p120, β-catenin, plakoglobin and α-catenin) is required for strong adhesion. In addition, VE-cadherin interacts with the actin cytoskeleton via an as yet unknown protein, while α-catenin dynamically binds to actin or to β-catenin. The tyrosine phosphatase VE-PTP associates to VE-cadherin and maintains VE-cadherin/catenin complex in a non-phosphorylated state, which contributes to junction stability. (B) Junction disassembly. Leucocyte adhesion to endothelial cells triggers signalling from the adhesion receptors ICAM-1 and VCAM-1 leading to the dissociation of VE-cadherin from VE-PTP and the activation of tyrosine kinases Src and Pyk2, which results in the phosphorylation of components of the VE-cadherin/catenin complex and junction disassembly. ROS, produced by the activation of the Rac1/NADPH oxidase pathway, are also involved in VE-cadherin/catenin phosphorylation. In addition, activation of Rho and release of intracellular calcium regulate the formation of actin stress fibres through the phosphorylation of myosin II. Cell contraction mediated by actomyosin pulls disassembled VE-cadherin complexes apart. Additional mechanisms such as VE-cadherin endocytosis and degradation in lysosomes or shedding of the extracellular VE-cadherin domain by ADAM-mediated cleavage may also contribute to junction disassembly.

A crucial event promoting endothelial permeability and leucocyte transendothelial migration is the phosphorylation of tyrosine residues in the VE-cadherin intracellular tail and in the associated catenins.43 Phosphomimetic mutations of tyrosines 658 and 731 in VE-cadherin are sufficient to compromise barrier function in Chinese hamster ovary (CHO) cells.44 In the context of leucocyte transmigration, several reports have demonstrated that leucocyte binding to endothelial cells induces tyrosine phosphorylation of VE-cadherin and/or associated catenins, and that this is essential for leucocyte diapedesis in vitro.4549 The tyrosine kinases Src and Pyk2 have been implicated in the phosphorylation of tyrosines 658 and 731 of VE-cadherin during ICAM-1-dependent neutrophil transmigration.46,48 In contrast, VE-cadherin phosphorylation downstream of lymphocyte adhesion appears not to be mediated by Src but instead requires RhoA activity and intracellular calcium.47

How VE-cadherin phosphorylation affects intercellular adhesion is as yet unclear. Disassembly of the cadherin/catenin complex following VE-cadherin phosphorylation has been reported during vascular endothelial growth factor (VEGF)-induced endothelial permeability.50 Consistent with this, phosphomimetic mutations of tyrosines 658 and 731 in VE-cadherin prevent binding of p120 and β-catenin, respectively, and also impair barrier function.44 However, complex disassembly may not always follow tyrosine phosphorylation since, at least upon leucocyte adhesion or ICAM-1 ligation, the phosphorylated cadherin–catenin complex was found to remain intact.47 In conclusion, although tyrosine phosphorylation of the cadherin/catenin complex is certainly crucial for the opening of endothelial cell–cell junctions and leucocyte transendothelial migration, the connection between these two events needs further clarification.

The relevance of cadherin/catenin complex phosphorylation in the regulation of barrier function is further supported by the important role of phosphatases in the stability of junctional complexes. Several phospho-tyrosine phosphatases have been shown to localize to endothelial cell–cell contacts and to dephosphorylate VE-cadherin, thereby decreasing endothelial permeability. These include receptor protein tyrosine phosphatase μ,51 protein tyrosine phosphatase 1B (PTP1B),52,53 and vascular endothelial protein tyrosine phosphatase (VE-PTP).49 VCAM-1 ligation was shown to activate PTP1B via reactive oxygen species (ROS) and protein kinase Cα. Pharmacological inhibition of PTP1B resulted in diminished lymphocyte diapedesis under laminar flow.52 VE-PTP is an endothelium-specific phosphatase directly implicated in the regulation of leucocyte transendothelial migration. VE-PTP interacts with VE-cadherin in vitro and in vivo, being recruited to mature junctions from an endocytic compartment.49,54 Although VE-PTP was shown to dephosphorylate VE-cadherin in CHO cells,54 the main substrate for VE-PTP in endothelial cells appears to be plakoglobin.49 Silencing VE-PTP expression in endothelial cells enhanced plakoglobin phosphorylation and neutrophil transmigration. In turn, adhesion of lymphocytes or neutrophils to tumor necrosis factor (TNF)α-stimulated endothelial cells induced the dissociation of the VE-PTP/VE-cadherin complex. Unexpectedly, although TNFα stimulation of endothelial cells was required for this effect, none of the known endothelial leucocyte adhesion molecules appeared to be involved.49

VE-cadherin internalization from the plasma membrane has been proposed as an additional mechanism that operates downstream of permeability promoting stimuli.55 The p120 catenin has been shown to play a central role in the stabilization of VE-cadherin at the plasma membrane by preventing cadherin endocytosis. Initial evidence showed that VE-cadherin levels are regulated by p120 and that interaction of VE-cadherin and p120 is required for this effect.56 Later, p120 was shown to inhibit VE-cadherin clathrin-dependent endocytosis and subsequent VE-cadherin degradation in lysosomes.57 In the context of leucocyte transmigration, overexpression of p120 was found to inhibit neutrophil diapedesis.48 However, these authors could not relate p120 to VE-cadherin internalization dynamics but rather to the inhibition of VE-cadherin tyrosine phosphorylation. Another phosphorylation event, this time on serine 665, was found to regulate VEGF-induced endothelial permeability through VE-cadherin endocytosis via β2-arrestin.58 This pathway was dissected and found to involve the activation of Rac1 by Src and subsequent VE-cadherin phosphorylation on serine 665 by the Rac1 effector p21-activating kinase. β2-Arrestin recruitment by serine-phosphorylated VE-cadherin eventually drives cadherin endocytosis. However, a role for serine 665 in transendothelial migration has not been reported so far. It is possible that cadherin internalization has an exclusive role during cellular processes that require prolonged junction disruption, such as VEGF-induced angiogenesis. In contrast, lateral diffusion of VE-cadherin on the plasma membrane may be a more efficient mechanism for the transient opening of intercellular gaps during leucocyte transmigration.48

Recently, the disintegrin and metalloprotease ADAM10 has been proposed as a novel regulator of junction integrity and leucocyte transmigration. ADAM10 was shown to be responsible for the shedding of a VE-cadherin soluble ectodomain and to modulate T cell diapedesis across an endothelial cell monolayer.59 However, ADAM10 effects may not be specific for VE-cadherin since this metalloprotease cleaves several substrates, including the junctional protein JAM-A. In contrast to the positive effects on migration of ADAM10-mediated shedding of VE-cadherin, ADAM10 and ADAM17-generated soluble JAM-A was shown to antagonize neutrophil transmigration.60 Thus, although it is clear that ADAM metalloproteases modulate leucocyte transendothelial migration, the mechanisms underlying their action remain to be fully elucidated.

VE-cadherin complexes associate to the actin cytoskeleton. Although the molecular details of this interaction are yet unclear, cytoskeleton-driven cell contraction is necessary for the opening of junctions, likely through pulling cells apart following junction disassembly by VE-cadherin/catenin phosphorylation. Leucocyte adhesion and ICAM-1/VCAM-1 ligation with antibodies induce actomyosin-driven contractility via phosphorylation of myosin light chain, which leads to the formation of actin stress fibres. Activation of myosin light chain kinase and of the small GTPase Rho is implicated in this pathway.6165

In addition to the control of the cytoskeleton, Rho GTPases are implicated in the activation of additional pathways leading to endothelial permeability. Rho has recently been placed upstream of VE-cadherin phosphorylation induced by ICAM-1 ligation.47 The fact that actin depolymerization also blocked VE-cadherin phosphorylation suggests that Rho may mediate this effect through the regulation of cytoskeletal rearrangements. Furthermore, activation of Rac1 by VCAM-1 ligation was shown to impair monolayer integrity through the stimulation of NADPH oxidase activity and ROS production.64,66,67 ROS, in turn, induce tyrosine phosphorylation of one or more components of the VE-cadherin/catenin complex, which results in the weakening of junctional adhesion.66,6870

Platelet/endothelial cell adhesion molecule-1 (PECAM-1) is another adhesive molecule present at the endothelial borders that does not participate in the organization of adherens or tight junctions. PECAM-1 is also expressed on leucocytes, and the homophilic interaction of leucocyte PECAM-1 with endothelial PECAM-1 is absolutely essential for leucocyte extravasation. Phosphorylation of tyrosine residues in the two immunoreceptor inhibitory motifs of the cytoplasmic domain of PECAM-1 induces the recruitment of SHP-1/2 and SHIP phosphatases. Recently, phosphorylation of tyrosine 663 was shown to be required for SHP-2 recruitment and leucocyte transendothelial migration.71 The reader is referred to an excellent recent review for further details on PECAM-1.72

3. Pathogen-induced signalling that regulates leucocyte transendothelial migration

The exit of leucocytes from the blood stream is usually strongly associated with pathological conditions, in particular with acute or chronic infectious diseases. This results from the fact that leucocyte transendothelial migration is promoted by, for instance, pathogen-derived chemoattractants (e.g. formylMet-Leu-Phe or fMLP) or by pathogen-induced activation of extravasated leucocytes, such as monocytes or macrophages that in turn locally produce chemokines or stimulate the endothelium to do so, promoting recruitment of circulating leucocytes. In addition to this classical view on the role of pathogens in transendothelial migration, there has been a steadily growing insight into molecular aspects of more direct host-pathogen interactions at the level of the endothelial cells and the transmigrating leucocytes. Over the past 10 years or so, a series of studies showed that there exist different levels and mechanisms by which infectious agents modulate the transendothelial migration process to their own benefit, blocking host responses and promoting the spread of infection. This is true for pathogens such as bacteria and fungi, as well as for viruses. Here, we will briefly discuss recent findings in this area.

3.1 Bacteria and fungi

Bacterial infection of endothelial cells directly affects leucocyte transendothelial migration. This is particularly relevant for infections in the brain, such as meningitis, where blood-borne bacteria interact with and cross the endothelium of the brain which forms the blood-brain barrier. There is now significant insight in the effects of Neisseria meningitidis on the endothelial cells of its host. The initial interaction of these bacteria with brain endothelial cells requires the polysaccharide capsule as well as the type IV pili, dynamic polymeric structures protruding from the bacteria that are essential for adhesion to the endothelial cell surface.73 The adherent bacteria induce a variety of effects in the endothelial cells that partly resemble the effects induced by adherent leucocytes, described above. Although it is not entirely clear what the endothelial receptor for N. meningitidis is (possibly, but not exclusively CD46), its binding to the endothelium induces F-actin rearrangements and recruitment of specific proteins and enzymes. Intriguingly, the binding of N. meningitides, which usually adheres in clusters, induces the formation by the endothelium of membrane protrusions, similar to endothelial docking structures induced by leucocyte adhesion.4,74 Moreover, a selection of actin-binding proteins, such as ezrin and moesin but not paxillin or vinculin, is recruited to the adherent bacteria. Ezrin was found to mediate bacteria-induced actin polymerization and the formation of membrane protrusions by the endothelial cells. Additional components such as Cdc42 and RhoA, but not Rac1, were found to control the actin polymerization, but not the recruitment of ezrin.74

The Rac1 GTPase was later found to be required for the invasion of endothelial cells by N. meningitidis following the binding of the bacterial lipo-oligosaccharide to the endothelial cell. This pathway also requires phosphoinositide-3 kinase (PI3K) activity and mediates the recruitment of another actin-binding protein, cortactin.75 Cortactin, in parallel, is phosphorylated by Src, as a consequence of the bacteria-induced activation of the ErbB2 receptor tyrosine kinase.76 Cortactin phosphorylation is essential for bacterial invasion as it mediates the formation of the membrane protrusions on the endothelial cell surface.75 Thus, there are at least two parallel pathways, one through ezrin and the other through cortactin, that are activated upon adhesion of N. meningitidis to endothelial cells and that are required for effective invasion and ultimately bacterial transcytosis.

The interaction between N. meningitidis and the endothelium has, as a result of competition for intracellular components, important consequences for leucocyte transmigration and associated pathology. Using a bone-marrow-derived endothelial cell line, Doulet et al.77 showed that N. meningitidis adhesion to the endothelial cells resulted in the recruitment of cytoskeletal components as well as ICAM-1 and VCAM-1 to the bacterial clusters. As a consequence, formation of endothelial docking structures around nearby adherent leucocytes was inhibited and transendothelial migration of monocytes and neutrophils was severely reduced. Efficient leucocyte transendothelial migration could be restored by overexpression of ezrin or moesin, underscoring the critical role for these adapter proteins in these events.

Most recently, colonies of adherent N. meningitidis were found to also recruit the endothelial Par-polarity complex (comprising Par3/Par6/PKCζ) in a Cdc42-dependent fashion.78 This has interesting consequences as it results in the recruitment of junctional proteins, including VE-cadherin, p120, and β-catenin, as well as ZO-1,2 and claudin. The authors propose that, at the sites where the bacteria bind the endothelial cell surface, an ectopic junction-like cell–cell contact is formed. The activation of the PKCζ following bacterial adhesion is also required for a significant loss of endothelial integrity. Together, these studies suggest that N. meningitidis uses various sophisticated mechanisms to establish productive infection of the brain: (1) by reducing leucocyte adhesion and transendothelial migration by competing for the machinery for docking structure formation and (2) by reducing endothelial integrity which promotes bacterial crossing of the endothelium through the paracellular pathway (Figure 3).

Figure 3

Summary of different effects induced by N. meningitides on endothelial cells. Binding of bacteria to the apical endothelial cell membrane induces recruitment of ERM proteins (1) as well the adapter protein cortactin (2). This recruitment mediates the formation of structures around the adherent bacteria that are enriched in ICAM-1 and that compete for proteins that are required for neutrophil adhesion and transendothelial migration. As a result, neutrophil transendothelial migration is impaired. In addition, adherent bacteria recruit junctional proteins through activation of the Par polarity complex (3), which weakens cell–cell contacts and favours bacterial infection of the underlying tissues.

Although the studies on N. meningitidis are obviously among the most detailed and extensive in this field, the modulation of epithelial as well as endothelial barrier function and of leucocyte transmigration is a more common theme in infectious diseases. These effects can result from the induction of cytokine production, such as TNFα (e.g. by Rickettsia prowazekii79) or be induced by the bacterially derived lipopolysaccharide (LPS), a powerful activator of endothelial cells.80,81 In addition, bacterial infection may affect the surface expression of endothelial leucocyte adhesion molecules. The causative agent of tularaemia, Francisella tularensis, promotes the expression of endothelial ICAM-1 and VCAM-1 through an atypical LPS and also stimulates chemokine secretion by endothelial cells.82 As a consequence, neutrophil transendothelial migration is increased,82 although in a model based on lung endothelial cells, the neutrophils that had crossed infected endothelial cells, showed reduced responsiveness, which might contribute to effective spread of infection.

As for fungal infections, much less is known about concomitant changes in the endothelium or in transendothelial migration. Zink et al.83 showed that infection of endothelial cells with Candida albicans impairs barrier function of aortic endothelial cells, which could conceivably promote transmigration of leucocytes. The yeast-like fungus Cryptococcus neoformans has been shown to impair neutrophils transendothelial migration. The Cryptococcus capsular polysaccharide GXM (glucuronoxylomannan) appears to play a key role in this phenomenon as it has been suggested to inhibit neutrophil rolling as well as adhesion to endothelium,84,85 most likely by interfering with E-selectin–ligand interactions.

3.2 Viruses

Viral infection can affect leucocyte transendothelial migration both at the level of the endothelial cells, but also at the level of the infected leucocytes. Acquired immune deficiency syndrome-related dementia has been linked to effects on both brain endothelial cells as well as on astrocytes with respect to the induction of MCP-1 (monocyte chemotactic protein-1) production.86,87 This effect is induced by the human immunodeficiency virus (HIV)-derived and secreted tat protein which promotes the migration of infected monocytes into the brain. At the same time, HIV-infected monocytes may cross the endothelium of the brain more efficiently due to selective upregulation of chemokine receptors, as was found for the MCP-1 receptor CCR2.88 An additional relevant mechanism may be the inhibition of reverse transmigration, i.e. from the subendothelial space back into the bloodstream, of HIV-infected monocyte-derived macrophages.89 Although the mechanism of this remains to be identified, this effect was proposed to promote the establishment of a viral reservoir in tissues, ensuring persistence of infection. An earlier study by Ricardo-Dukelow et al.90 showed that HIV-infected monocyte-derived macrophages induce marked changes in protein expression in brain endothelial cells. The proteomic analysis of these endothelial cells showed many types of protein to be altered, including cytoskeletal and redox-regulating proteins, and these may well be relevant for HIV-infection-associated changes in endothelial permeability in the brain.

Such effects are not specific for HIV: cytomegalovirus (CMV) infection of human monocytes was found to induce an upregulation of monocyte β1- and β2-integrins as well as of the integrin ligand ICAM-1, via PI3K- and NFκB-dependent signalling.91 The consequence of these effects was enhanced monocyte transendothelial migration, which was paralleled by an increase in the intrinsic motility of the infected monocytes.

In T-lymphocytes infected with measles virus, similar events occur. Here, the expression as well as the activity of LFA-1 and VLA-4 was upregulated in infected cells, albeit that lymphocyte motility was not enhanced. In these cells, the adhesive capacity of the lymphocytes was increased to such level that transendothelial migration was effectively reduced. Moreover, the spread of virus to the endothelial cells could be observed upon prolonged co-culture, suggesting that the virus-induced increase in lymphocyte adhesion serves to promote spreading of the measles virus into the endothelium.92

Infection by the respiratory syncytial virus of endothelial cells from various tissues, including lung and umbilical vein, promotes neutrophil adhesion and transendothelial migration. This was due to virus-induced upregulation of ICAM-1 and VCAM-1, following activation of protein kinases A and C, PI3K, and p38 mitogen-activated protein kinase (MAPK). Similarly, CMV infection of endothelial cells promotes upregulation of ICAM-1 and VCAM-1, as well as of E-selectin and PECAM-1, which results in increased monocyte transendothelial migration.93 This was further promoted by a virus-induced reduction in transendothelial electrical resistance, loss of actin stress fibres, and a loss of expression of junctional proteins such as VE-cadherin, all indicative of reduced barrier function. Similarly, as described above for measles virus, viral transfer, in this case from endothelial cell to monocyte, was observed accompanying transendothelial migration. Thus, although the signalling pathways involved were not analysed, this data suggest that CMV infection in endothelial cells affects leucocyte adhesion molecules as well as junctional components to promote monocyte transendothelial migration and viral spread.

Endothelial signalling in the context of HIV infection of brain endothelium has been linked to the induction of chemokines and pro-inflammatory genes and to the activation of the Janus kinase (JAK)/signal transducer and activator of transcription1 (STAT1) pathway.94,95 STAT1 was found to be required for HIV1-induced interleukin-6 (IL-6) expression as well as for the impairment of the endothelial barrier function, also in brain vessels from HIV-1-infected individuals. A later study from the same group showed that secreted HIV-1 gp120 is capable of inducing STAT1 activation as well as IL-6 and IL-8 secretion from brain microvascular endothelial cells. Both PI3K and the MAPK activator MEK were implicated in gp120-mediated STAT1 activation, and in the consequent promotion of monocyte adhesion and transendothelial migration.96 These findings point to a role of STAT1-based signalling also in ‘normal’ chemokine-induced leucocyte transendothelial migration, albeit that there is no published data on this.

Infection with dengue virus (DV) is also associated with vascular leakage. DV infection of EA.hy 926 endothelial cells was recently found to decrease expression of adhesion molecules such as PECAM-1 as well as junctional proteins such as VE-cadherin, which may explain the loss of barrier function.97 DV binds to endothelial cells via the α5β3 integrin and productive infection further requires actin polymerization and the formation of filopodia, mediated by the Rac1 and Ccd42 GTPases.

Finally, West Nile virus, which also targets the brain vasculature, uses an alternative approach. In vitro infection of brain microvascular endothelial cells was found to upregulate claudin-1, VCAM-1, and E-selectin, but not ICAM-1 or PECAM-1.98 Cell-free virus was found to cross the endothelium without compromising the barrier function. The upregulation of adhesion molecules such as VCAM-1 may promote further virus infection by recruitment of virus-infected monocytes into the brain.98

In conclusion, pathogen-induced effects on endothelial cells include mechanisms of interfering or promoting leucocyte transendothelial migration that have not been previously recognized in studies on regular leucocyte transendothelial migration. These include the activation of JAK/STAT signalling as well as the competition effects, interfering with endothelial interactions with leucocytes. It is intriguing to consider that the mechanisms induced by bacterial or viral infection are endogenous to the endothelium and therefore are likely to be involved in, e.g. inflammatory responses as well. The information we can obtain from studying pathogens and their effects on transendothelial migration represents a rich source of additional mechanistic insights to be explored in the near future.

Conflict of interest: none declared.


JvB is funded by the Netherlands Organization for Scientific Research NWO (grant no. 916.76.053) and by the Dutch Heart Foundation (grant no. 2005T039).


  • Abbreviations:


View Abstract