Chapter 115: Vascular Function in Hemostasis

The endothelium represents a dynamic interface between flowing blood and the vessel wall, and produces a variety of factors that regulate blood fluidity (Fig. 115–1). Endothelial cells are subject to unique shear stress forces, to soluble factors in the blood, and to signals emanating from cells in the circulation, vascular wall, and tissues, all of which create region-specific phenotypes.^1,2,3 In addition to modulating vascular permeability and fragility, the endothelium regulates the fluid state of blood through its thromboresistant nature, profibrinolytic properties, and antiinflammatory potential. These activities maintain vascular patency.⁴

ENDOTHELIAL CELL HETEROGENEITY

The heterogeneity of endothelial cells is mediated by two mechanisms.^5,6 First, extracellular biochemical and biomechanical signals trigger posttranscriptional and/or posttranslational changes that vary across the vascular tree. Second, certain site-specific properties of the endothelium are genetically programmed, and therefore, independent of the extracellular milieu. This phenotypic variability serves at least two important purposes: (1) It allows endothelial cells to meet the specific metabolic needs of the surrounding tissue. For example, the tight junctions of the blood–brain barrier protect neurons from fluctuations in composition of the aqueous blood supply, whereas the fenestrated discontinuous endothelium of hepatic sinusoids allows ready access of nutrient-rich portal venous blood for the metabolic systems in hepatocytes; and (2) phenotypic variability provides endothelial cells with site-specific mechanisms for thriving within many different microenvironments. For example, endothelial cells in the inner medulla of the kidney must survive the relatively hypoxic and hyperosmolar local environment, whereas endothelial cells in the pulmonary capillary bed have adapted to an oxygen-rich environment.

A rapid endothelial cell response is required for sudden environmental perturbations. Translational control mechanisms, which are more immediate than transcriptional changes, provide regulatory responses for up to 10 percent of genes expressed in endothelial cells.⁷ Because of their close association with both flowing blood and solid tissues, endothelial cells are subject to a broad spectrum of agonistic and inhibitory external signals that frequently require rapid functional and phenotypic responses. Clinically, such stimuli are associated with sepsis, inflammation, ischemia–reperfusion injury, and direct mechanical vascular trauma induced clinically by stents, balloon catheters, and graft procedures.

ENDOTHELIAL PRODUCTION OF THROMBOREGULATORY MOLECULES

Thromboregulatory compounds, such as eicosanoids, nitric oxide, and the ecto-ATP/Dase-1/CD39, control platelet and vascular reactivity during the early stages of thrombus formation (Table 115–2).⁸ Eicosanoids are hydrocarbon compounds derived from essential fatty acids in the diet. The most important endothelial eicosanoid is prostacyclin (PGI₂), which blocks platelet reactivity, induces vascular relaxation, and stimulates cytokine production.⁹ Nitric oxide (NO) is a naturally occurring gas released from vascular endothelial cells in response to binding of vasodilators to endothelial cell membrane receptors. Thus, it is a short-lived vasodilator and inhibitor of platelet reactivity. By activating guanylate cyclase, the resulting increase in cyclic guanylate monophosphate (GMP) inhibits platelet function and induces vascular relaxation.^10,11 Endothelial cell ecto-ATP/Dase-1/CD39 is a membrane-associated apyrase that metabolizes adenosine diphosphate (ADP) in the primary platelet releasate, preventing further platelet activation and recruitment.^12,13

Late thromboregulators produced by endothelial cells act either to prevent excessive thrombin generation or to promote lysis of intravascular thrombi (see Table 115–1). Antithrombin, a natural anticoagulant, acts as an inhibitor of thrombin and factor Xa in the circulation. Endothelial cell heparan proteoglycans act as cofactors for antithrombin. The tissue factor pathway inhibitor (TFPI) inhibits the complex between factor VIIa and tissue factor (TF). The thrombomodulin/endothelial cell protein C receptor (EPCR)/protein C system in the vascular wall regulates hemostasis through inactivation of procoagulant cofactors, and antiinflammatory activity.¹⁴ The fibrinolytic system is intimately involved with the vascular endothelium because endothelial cells not only synthesize and secrete tissue plasminogen activator (t-PA), but also regulate formation of plasmin from its precursor, plasminogen, through the expression of receptors.¹⁵ Impairment of fibrinolytic potential can play a central role in the etiology of occlusive vascular disease.¹⁶ Finally, endothelial cell adhesion molecules, including the cell adhesion molecules (CAMs: mucosal addressin cell adhesion molecule [MAdCAM]-1, intercellular adhesion molecule [ICAM]-1, vascular cell adhesion molecule [VCAM]-1, and platelet endothelial cell adhesion molecule [PECAM]-1) and the selectins (P- and E-selectin), are glycoproteins that modulate multiple interactions between the endothelium and various classes of circulating leukocytes, thereby modulating vascular patency.¹⁷ Together, these mechanisms define thromboregulation, the processes by which blood cells and cells of the vessel wall, through their close proximity, interact to facilitate or inhibit thrombus formation.¹⁸

The physiologic defense systems that render endothelial surfaces and blood cells antithrombotic can be overwhelmed by excessive shear stress, increased turbulence, injury, inflammation, and severe atherosclerosis.¹⁹ These events transform the endothelial cells into a prothrombotic and antifibrinolytic phenotype,²⁰ which is accompanied by upregulation of leukocyte and endothelial CAMs, increased expression of TF, and accumulation of monocytes/macrophages in the vessel wall.²¹ These events commonly occur at the site of fissured atherosclerotic plaques in the coronary and cerebrovascular circulation.²² Because the eicosanoids such as PGI₂ (prostaglandin [PG]I₂), as well as NO, and the ecto-ATP/Dase-1/CD39 group reach peak activity very early in the hemostatic/thrombotic cascade (Figs. 115–2, 115–3, 115–4), they represent potential targets for therapeutic intervention in the sequence of events beginning with platelet activation, and leading to coagulation, thrombosis, and atherogenesis.^21,22 Finally, functional and physical contacts between platelets and endothelial cells are of critical importance for the maintenance of vascular integrity and cell permeability.^1,23

Dietary fatty acids give rise to arachidonic acid, the starting point for synthesis of other eicosanoids. Originating from different cells, intermediates in the arachidonic acid pathway can interact with each other to produce new products with new biologic activities. Oxygenation and further enzymatic transformation of arachidonic acid gives rise to eicosanoids (formerly classified as PGs) and hydroxy acids, such as the leukotrienes. Eicosanoids are autocoids, an important group of transient, physiologically active endogenous substances, that act within the immediate environment of the cell, where they promote or inhibit a biologic function.²⁴ These autacoids have a very short life span and may act within a few seconds, a phenomenon that is clinically important but difficult to study experimentally.

In 1975, Hamberg and colleagues discovered that a new eicosanoid, PGI₂, was derived from arachidonic acid in endothelial cells.²⁵ Soon thereafter, Moncada realized that the effects of PGI₂ opposed those of thromboxane, namely vasodilation and inhibition of platelet aggregation.^26,27 The biologic half-life of PGI₂ was found to be 10 to 20 seconds. In addition, it was determined that the first step in arachidonic acid oxidation and conversion, which is carried out by cyclooxygenase (COX)-1, is inhibited by aspirin (acetylsalicylic acid), which donates an acetyl group that inactivates COX-1 and inhibits platelet function.²⁸

BIOSYNTHESIS OF PROSTACYCLIN IN ENDOTHELIAL CELLS

PGI₂ is the major and most important eicosanoid produced by endothelial cells. A broad range of stimuli, including hormones, biochemicals, or physical forces such as shear stress can elicit release of PGI₂. Kinetic studies revealed two distinct patterns of PGI₂ production: (1) rapid release, independent of new COX-1 mRNA or protein synthesis, and (2) slower production reflecting increased COX-2 expression.

In the case of rapid stimulation of PGI₂ production, as induced by thrombin, histamine, bradykinin, and ionophore, the response plateaus at 10 minutes.²⁹ These agonists activate phospholipase C which generates inositol trisphosphate (IP₃) and diacylglycerol (DAG). The released IP₃ induces an elevation of intracellular calcium, which translocates phospholipase A to the outer portion of the nuclear envelope and endoplasmic reticulum. Phospholipase A then couples functionally to COX-1, which is located on the luminal membrane. Prostacyclin synthase (PGIS) colocalizes with COX-1 in endothelial cells. Activated phospholipase A2 (cPLA2) catalyzes the release of arachidonic acid from membrane phospholipids, and the free arachidonate interacts with COX-1 and is converted to the endoperoxide PGH₂. PGIS converts prostaglandin H₂ (PGH₂) to PGI₂. The half-life of COX-1 is approximately 10 minutes, whereupon it autoinactivates.

Stimulation of PGI₂ production by proinflammatory cytokines and growth factors, such as lipopolysaccharide (LPS), interleukin (IL)-1β, tumor necrosis factor (TNF)-α, and platelet-derived growth factor (PDGF), is a slower, more sustained process.²⁹ In response to these agonists, PGI₂ production occurs within 30 to 60 minutes, and parallels the time course of production induced by COX-2, but not COX-1.

THE TWO ISOFORMS OF PROSTACYCLIN G/H SYNTHASE

The recognition that there was a constitutive and an inducible cyclooxygenase (COX-1 and COX-2, respectively), was a major advance.³⁰ Cloning studies of an immediate to early response gene from 3T3 fibroblasts revealed that the COX-2 complementary DNA was highly homologous to that of COX-1.^{30,31,32,33,34} COX-2 is inducible in endothelial cells by prothrombotic, inflammatory, or mitogenic stimuli, and in neutrophils by inflammatory stimuli.^35,36

Within a specific species, there is approximately 60 percent homology between deduced amino acid sequences of COX-1 (576 residues) and COX-2 (587 residues). The C-terminal sequence of 18 amino acids in COX-2 is absent in COX-1. Therefore, antibodies directed at this C-terminal sequence can identify COX-2 in tissues by immunoblot. The catalytic activity of both COX enzymes are similar and all amino acids critical for COX-1 activity are conserved in COX-2. The active site in COX-1 is slightly larger than that of COX-2, a fact that has impacted design of COX inhibitors. COX-2 contains mannose, and an N-glycosylation site within the 18-amino acid C-terminal sequence. An N-glycosylation site at Asn410 is required for COX-1 to fold into its active conformation.

The gene for COX-1 is located on chromosome 9 and spans 22 kb of genomic DNA, while the gene for COX-2 is located on chromosome 1 and spans 8 kb of DNA. Transcription of COX-2 proceeds via several signaling mechanisms initiated by cyclic adenosine monophosphate (cAMP)/protein kinase A, protein kinase C, tyrosine kinases, and pathways activated by growth factors, endotoxin, and cytokines.^33,37,38,39 The discoveries of COX-1 and COX-2 were of great importance and have led to new concepts concerning the structure and function of COX-induced autacoids.⁴⁰

PROSTACYCLIN AS AN AUTACOID

PGI₂ is released from stimulated endothelial cells by a broad range of agonists, and plays a critical role in the maintenance of vascular integrity by promoting thromboresistance and inhibiting inflammatory responses in the vasculature. Production of PGI₂ is dynamically regulated to meet the challenges arising from frequent prothrombotic and proinflammatory events.²⁹ As an autacoid, PGI₂ has a half-life of 3 minutes, whereupon it undergoes chemical hydrolysis to 6-keto-PGF₁α. It acts on the type I platelet PG receptor (IP) by increasing cAMP levels in a paracrine manner.⁴¹ IP is a 7-transmembrane, G-protein– and adenylyl cyclase–coupled receptor. The latter binds to and activates protein kinase A (PKA), resulting in inhibition of platelet activation and recruitment.⁴² Physical or chemical perturbation of endothelial cells results in enhanced PGI₂ production, which increases platelet cAMP resulting in abolition of platelet shape change, inhibition of platelet secretion and recruitment, and impaired binding of von Willebrand factor (VWF) and fibrinogen to the platelet surface. PGI₂ also inhibits platelet adhesion to subendothelium, especially at high shear rates.⁴³

The discovery of PGI₂ revealed that the vascular endothelium had a protective effect on blood fluidity.^2,8 It also meant that PGI₂ released from endothelial cells could counteract the effect of excessive thromboxane formation. In addition, it was appreciated that intermediates in the synthesis of PGI₂ from arachidonic acid could interact with other cells and tissues. Thus, PGI₂ could be synthesized from platelet-derived endoperoxides by cultured human endothelial cells.⁴⁴ Because of a low threshold for toxicity (hypotension and diarrhea), PGI₂ does not display a satisfactory therapeutic window. An interesting compendium of eicosanoid-related disorders is described in a review on eicosanoids in health and disease.⁴⁵

In vascular endothelial cells, NO synthase (NOS) catalyzes formation of NO from L-arginine, in the presence of nicotinamide adenine dinucleotide phosphate (NADPH) and oxygen.⁴⁶ The L-arginine is subsequently converted to citrulline and NO. The endothelial cell isoform of NO synthase (eNOS or the NOS3 gene product) functions constitutively, and is further activated by receptor-agonists that elevate intracellular calcium. Major stimuli include ADP, thrombin, bradykinin, and shear stress.⁴³ Shear forces induce transcriptional activation of the eNOS gene because its promoter contains a shear response consensus sequence (GAGACC). The NO that forms activates guanylate cyclase, thereby generating cyclic GMP. NO becomes oxidized to nitrite and then to nitrate, which is measurable in blood samples. NO in the circulation is rapidly inactivated by erythrocytes.^11,47,48 NO has a vasodilatory effect on the pulmonary vasculature, and, in patients with congestive heart failure, its inhalation decreases pulmonary hypertension and increases pulmonary ventilation.^{10,11,47,48,49,50,51,52,53,54} Acetylcholine released by activated nerve terminals in the vessel wall activate the endothelial cell to produce and release NO. This NO effect also explains the action of nitroglycerin, which has long been used to treat patients with angina resulting from coronary artery disease.⁵⁴

Importantly, production of NO by endothelial cells is impaired in the presence of the thiol-containing amino acid, homocysteine. Cynomolgus monkeys with diet-induced hyperhomocysteinemia demonstrated reduced blood flow in the lower extremity and an impaired response to endothelial cell-dependent vasodilators.⁵¹ Similarly, production of NO by endothelial cells in vitro is significantly inhibited in the presence of homocysteine, possibly by a mechanism involving impairment of the enzyme glutathione peroxidase.^52,53

STRUCTURE AND BIOCHEMICAL PROPERTIES OF NITRIC OXIDE SYNTHASE

There are two isoforms of NOS, the constitutive form (eNOS), synthesized by the endothelial cell and regulated by Ca²⁺ and calmodulin, and the cytokine-inducible, posttranscriptionally regulated form (iNOS).⁴⁷ Both constitutive and inducible forms are mainly cytosolic, although a membrane-bound constitutive NOS isoform containing a myristoylation consensus sequence has been isolated from bovine aortic endothelial cells.⁴³ eNOS has a molecular mass of 144 kDa and shares 57percent amino acid sequence identity with neuronal NOS. The cofactor (6R-tetrahydro-L-biopterin [H₄B]) participates in inducible and constitutive NOS isoform reactions. It is thought that H₄B stabilizes the enzyme in a manner allowing for maximum activity of the NOS subunit to which the pterin binds.^10,11,47,54

BLOCKADE OF PLATELET AGGREGATION AND SECRETION BY NITRIC OXIDE

Platelet activation and recruitment in response to all agonists, such as ADP, collagen, epinephrine, and thrombin, is blocked by NO. Blockade also occurs in vivo via formation of NO from endothelium.¹⁰ Importantly, the inhibitory action of NO is not affected by aspirin either in vivo or ex vivo. Therefore, NO production is not caused by participation of endothelial cell eicosanoids.

In addition to eNOS, the NOS3 gene product, endothelial cells stimulated by agonists such as cytokines express the inducible form of NO synthase, iNOS, the NOS2 gene product. Through this mechanism, NO can further inhibit platelet reactivity and reduce basal vessel tone by inducing relaxation of vascular smooth muscle. The biochemical basis for the reaction is that NO binds to the heme prosthetic group of guanylyl cyclase. The inhibitory effect of NO on platelet activation can be monitored by measuring surface expression of P-selectin. The ability of NO to inhibit mobilization of intracellular platelet calcium results in reduction of the conformational changes in platelet membrane glycoprotein (GP)IIb/IIIa, an absolute requirement for fibrinogen binding and subsequent platelet aggregation. There is a broad spectrum of other effects of NO, including inhibition of leukocyte adhesion to endothelial cell surfaces, inhibition of smooth muscle migration, and reduction of smooth muscle cell proliferation. These phenomena suggest that secretion of NO into the microenvironment is a major component of the response to vascular injury⁴³

In addition to the platelet inhibition by PGI₂ and NO, endothelial cells inhibit platelet function via the action of endothelial cell ecto-ATP/Dase-1/CD39, an ecto-apyrase with ADPase and adenosine triphosphatase (ATPase) activities. The cluster designation symbol for this compound is CD39, the product of the ENTPD1, ectonucleotide triphosphate diphosphohydrolase gene.⁵⁵ CD39 is localized mainly in endothelial cells and leukocytes. In endothelial cells, CD39 is located on the cell surface with the major portion of the molecule facing the vessel lumen.^12,13,56 The enzyme has both N- and C-terminal transmembrane regions with small cytosolic portions anchoring the molecule.⁵⁷ In addition to CD39, CD73 (5’-nucleotidase) is present on vascular cells and converts the adenosine monophosphate (AMP) generated from CD39 metabolism to adenosine (Fig. 115–5). In contrast to all other known platelet inhibitors, acting in concert with CD73, CD39 can convert the local environment from a prothrombotic ADP/ATP-rich entity to an antithrombotic adenosine-rich environment.⁵⁸ This phenomenon was evident from observations that platelets became unresponsive to all agonists when in motion or in proximity to endothelial cells, even when eicosanoid and NO production were blocked.² Importantly, CD39 and CD73 do not exert their action on the platelet per se but act in series to metabolize ATP and ADP secreted from activated platelets to AMP and hence to adenosine.^13,59 ADP released from activated platelets is metabolized by CD39, thereby inhibiting ADP-induced platelet activation, release and aggregation (Fig. 115–5).

Most platelet agonists initiate secretion of dense granule contents within 15 to 20 seconds. The enhanced metabolism of ATP and ADP by therapeutically administered soluble CD39 would also reduce secondary autoamplification and recruitment, and, consequently, thrombus formation.^9,27,60 Because CD39 and CD73 are probably acting together, they will theoretically increase levels of endogenous adenosine and elevate the threshold for platelet activation in the local microenvironment. In a murine model, soluble CD39 administration ameliorates the extent of stroke and reverses excessive platelet reactivity without bleeding complications, even if administered 3 hours following stroke induction.⁶¹ Therapeutic benefit of soluble CD39 has also been demonstrated in animal models of cardiac ischemia,⁶² in the development of atherosclerosis,⁶³ regulation of leukocyte proinflammatory activity,⁶⁴ inhibition of metastasis,⁶⁵ and in transplantation medicine.⁶⁶ That the preclinical therapeutic use of soluble CD39 could abrogate thrombosis without inducing the hemorrhage seen with the use of existing antiplatelet therapies⁶⁷ could provide a therapeutic advantage over existing therapies for thrombotic disorders, including those who are resistant to existing therapeutic paradigms.⁹ CD39 represents a major control system for blood fluidity.⁶⁸

The protein C pathway⁶⁹ plays a critical role in the prevention of thrombosis and is an integral part of the host inflammatory response. This pathway is initiated on the endothelial cell surface when thrombin combines with the endothelial receptor protein thrombomodulin (TM). Although thrombin is capable of slowly activating protein C, this reaction is markedly inhibited in the presence of physiologic concentrations of calcium ions. Upon binding of thrombin to TM, the rate of protein C activation is dramatically enhanced and becomes dependent on the presence of calcium. The detailed biochemistry of this activation reaction has been reviewed elsewhere.⁷⁰ Another protein found predominantly in large vessels, the EPCR, can bind protein C and further augment its activation by the thrombin–TM complex.⁷⁰ Activated protein C (APC) can dissociate from EPCR and interact with protein S on either the endothelial cell or other membrane surface to exert its anticoagulant function. The function of APC can be found in several reviews.^14,71,72,73

By far, the best known function of TM is its role in protein C activation. When thrombin is bound to TM, it is no longer able to clot fibrinogen, activate platelets, activate factors V and VIII,⁷⁴ or interact with the protease-activated receptors.^75,76 Instead, thrombin-TM acts as a direct anticoagulant. TM also promotes the activation by thrombin of the plasma thrombin-activatable fibrinolysis inhibitor (TAFI).⁷⁷ TAFI inhibits plasmin-mediated fibrinolysis by removing carboxy-terminal lysine residues from fibrin, thereby reducing available binding sites for plasminogen and t-PA. In addition, TAFI is the major enzyme responsible for the removal of a C-terminal arginine from complement factor 5a (C5a),^78,79 leading to the inactivation of this potent anaphylotoxin generated during complement activation. Other vasoactive substances may also be inactivated by this enzyme. TM also accelerates the proteolytic inactivation of prourokinase (also called single-chain urokinase-type plasminogen activator [scu-PA]) by thrombin,^80,81 which may affect both fibrinolysis and tissue remodeling.⁸² Despite these antifibrinolytic effects of TM, many in vivo experiments have demonstrated that soluble TM infusion results in a net antithrombotic and/or antiinflammatory effect.⁸³

Independent of its effect on hemostasis, TM is essential to normal fetal development. When the TM gene is deleted by homologous recombination in mice, embryos die on day 8.5, prior to the development of a functional cardiovascular system,⁸⁴ implying that TM has functions in addition to its anticoagulant and antifibrinolytic properties. Both TM⁸⁵ and EPCR⁸⁶ are highly expressed on the giant trophoblast cells of the placenta. If TM expression is maintained on these cells, the TM null embryos survive past this blockade point.^87,88

The EPCR is a 220-amino-acid, type 1 transmembrane protein.^89,90,91,92 EPCR has two extracellular domains that show structural homology with the α and β domains of major histocompatibility complex (MHC) class 1 molecules, most notably the CD1d family. Because there are three Cys residues in the extracellular domain, the possibility of crosslinking with another protein exists. The cytoplasmic domain of human EPCR is only three amino acids long, Arg-Arg-Cys. The terminal Cys can be acylated with palmitate, which may have functional consequences.⁹³ Both protein C and APC bind to EPCR with similar affinity, approximately 30 nM.⁸⁹ Binding requires the presence of calcium and is enhanced in the presence of magnesium ions. In addition, a soluble form of EPCR found normally in plasma⁹⁴ is also capable of binding both protein C and APC with equivalent affinity.

EPCR augments protein C activation by the thrombin–TM complex in vitro and in vivo, primarily by decreasing the K_m (Michaelis-Menten dissociation constant) for protein C.^70,95,96 Just as thrombin changes its function from procoagulant to anticoagulant when it binds to TM, it appears that APC bound to EPCR undergoes a similar switch from anticoagulant to antiinflammatory molecule.^97,98 Unfortunately, however, early studies that suggested a possible therapeutic role for APC in human sepsis have not been borne out in clinical trials.⁹⁹ Deletion of the EPCR gene by homologous recombination leads to early embryonic lethality around day 9.5,¹⁰⁰ at which time EPCR is highly expressed in the giant trophoblasts of the placenta, but not in the embryo itself.⁸⁶ In contrast to TM knockout animals,¹⁰¹ the placentas of EPCR knockout embryos show significant fibrin deposition at the fetal maternal interface.

Plasmin, the major clot-dissolving protease in humans, is formed upon the cleavage of a single peptide bond within the zymogen plasminogen (Chap. 135). This tightly regulated reaction is strongly influenced by cells of the blood vessel wall, including endothelial cells, smooth muscles cells, and macrophages, which express plasminogen activators, plasminogen activator inhibitors, and fibrinolytic receptors.

ENDOTHELIAL CELL PRODUCTION OF FIBRINOLYTIC PROTEINS

In 1958, Todd demonstrated that human blood vessels possess fibrinolytic activity that is dependent upon an intact endothelium.^102,103 We now know that the endothelium is the principal source of t-PA in vivo where it appears to be highly restricted to small blood vessels in specific anatomic locations, a pattern that likely reflects the heterogeneity of endothelial cells as they respond to a myriad of tissue-specific cues.^104,105 In the baboon, for example, sites of t-PA production include 7 to 30 μm precapillary arterioles and postcapillary venules, but not large arteries and veins.¹⁰⁶ In the mouse lung, similarly, bronchial, but not pulmonary, endothelial cells express t-PA.¹⁰⁷ Moreover, enhanced expression of t-PA at branch points of pulmonary blood vessels may reflect stimulation by laminar shear stress.¹⁰⁸ In addition, peripheral sympathetic neurons that invest the walls of small arteries may represent a significant source of circulating t-PA.¹⁰⁹

Although in vitro studies suggest that t-PA expression in cultured endothelial cells is regulated by a wide array of factors, only a few of these pathways have been confirmed in vivo. Thrombin,¹¹⁰ histamine,^111,112 oxygen radicals,¹¹³ phorbol myristate acetate,¹¹⁴ DDAVP (deamino D-arginine vasopressin),¹¹⁵ and butyric acid liberated from dibutyryl cAMP¹¹⁶ all increase t-PA mRNA in cultured endothelial cells. Both thrombin and histamine appear to act via receptor-mediated activation of the protein kinase C pathway.¹⁰⁵ Laminar shear stress stimulates both t-PA secretion¹¹⁷ and steady-state mRNA levels.¹¹⁸ Hyperosmotic stress and repetitive stretch also enhance t-PA expression.^119,120 In addition, differentiating agents, such as retinoids,^121,122 stimulate transcription of t-PA in endothelial cells in vitro.

In vivo, the circulating half-life of t-PA is approximately 5 minutes. Infusion of DDAVP, bradykinin, platelet-activating factor (PAF), endothelin, or thrombin is associated with an acute release of t-PA, and a burst of fibrinolytic activity can be detected within minutes.¹²³ In the mouse lung, exposure to hyperoxia leads to 4.5-fold upregulation of t-PA mRNA in small-vessel endothelial cells.¹⁰⁷ In humans, infusion of TNF into patients with malignancy is associated with an increase in plasma t-PA.¹²³ Deficient release of t-PA in response to venous occlusion in humans is associated with deep venous thrombosis,¹²⁴ as well as atrophie blanche and other cutaneous vasculitides.¹²⁵

In vivo, urokinase plasminogen activator (u-PA) is not a product of resting endothelium,¹²⁶ but is produced primarily by renal tubular epithelium.¹²⁷ Expression of u-PA mRNA in endothelium, however, is strongly stimulated during wound repair and physiologic angiogenesis within ovarian follicles, corpus luteum, and maternal decidua.¹²⁸ Endothelial cells passaged in culture do synthesize u-PA,¹²⁹ and expression of its mRNA is stimulated by TNF-α by 5- to 30-fold.¹³⁰ Small increases in u-PA have also been observed in vitro in response to IL-1 and LPS.^131,132,133

The association of u-PA with the blood vessel wall appears to reflect its association with the u-PA receptor (uPAR) which may fulfill a variety of nonproteolytic functions ranging from directed cell migration to cellular adhesion, differentiation, and proliferation (Fig. 115–6).¹³⁴ In the adult mouse, uPAR mRNA is not normally detected by in situ hybridization in the endothelium of either large or small blood vessels.¹³⁵ However, upon stimulation with endotoxin, expression is detected in endothelium lining aorta, as well as arteries, veins, and capillaries in heart, kidney, brain, and liver,¹³⁵ and in renal tubular epithelial cells.¹²⁷

Plasminogen activator inhibitor (PAI)-1 is likely to function as a major regulator of plasmin generation in the vicinity of the endothelial cell. Thrombin, IL-1, transforming growth factor β, TNF, lipoprotein(a) (Lp[a]), and LPS all induce dramatic increases in steady state PAI-1 message levels.^{110,131,132,136,137} Heparin-binding growth factor 1 reduces PAI-1 mRNA production by cultured endothelial cells, but has no effect on t-PA.¹³⁸ Thus, synthesis and secretion of PAI-1 by the endothelial cell in vitro appears to be regulated independently of t-PA.

In vivo, elevated levels of circulating PAI-1 have been linked epidemiologically to risk for myocardial infarction.¹²⁴ Although the liver is the major source of plasma PAI-1, endothelial expression of PAI-1 is detected near neovascular sprouts during decidual neovascularization in the ovary.¹²⁸ In addition, inflammatory cytokines are powerful stimuli for induction of PAI-1 in a variety of tissues including liver, as injection of TNF in both rats and humans with active malignancy results in a striking increase plasma concentrations of PAI-1.^105,123

The endothelial cell coreceptor for t-PA and plasminogen, the annexin A2/S100A10 complex (see Fig. 115–6), appears to be expressed constitutively in vivo by endothelial cells in a wide variety of tissues in the chicken,¹³⁹ mouse,¹⁴⁰ rat,¹⁴¹ and human.¹⁴² Annexin A2 is upregulated transcriptionally by hypoxia both in vivo and in endothelial cells in vitro,¹⁴³ and by nerve growth factor in neuronal-like PC12 cells.¹⁴⁴ In addition, the in vitro transition of human monocyte to macrophage is associated with a severalfold increase in both annexin A2 protein and steady state mRNA expression.¹⁴⁵

The evidence that the annexin A2 system plays a role in maintaining vascular patency includes the findings that (1) overexpression of annexin A2 in blast cells in acute promyelocytic leukemia blast cells increases plasmin production and contributes to hyperfibrinolytic bleeding,^{146,147,148,149} (2) systemic injection of annexin A2 diminishes thrombotic vascular occlusion resulting from vascular injury in experimental animals,¹⁵⁰ (3) annexin A2–deficient mice display fibrin deposition on microvessels and impaired clearance of arterial thrombi following vascular injury,¹⁵¹ (4) high titer antibodies directed against annexin A2 are associated with thrombosis in antiphospholipid syndrome and in individuals with cerebral venous thrombosis,^152,153 and (5) that polymorphisms in the ANXA2 gene are associated with cerebral vascular occlusion and osteonecrosis of bone in patients with sickle cell disease.^154,155,156 Whether defects in S100A10, which could serve either as a chaperone for annexin A2 or as a direct binding site for plasminogen,¹⁵⁷ might also be associated with these clinical entities remains to be determined.

NONFIBRINOLYTIC VASCULAR FUNCTIONS OFPLASMIN

Although not yet demonstrated in vivo, plasmin may inactivate factor Va in vitro by cleaving both the heavy and light chains of this 168-kDa protein, in a manner that is distinct from the action of activated protein C.^158,159 Plasmin can also inactivate factor VIIIa, a procoagulant cofactor that is structurally related to factor Va.¹⁶⁰ In addition, platelet GPIIb/IIIa and GPIb, the cell surface receptors for fibrinogen and VWF, respectively, are both plasmin substrates.^161,162 Thus, plasmin formation in the vicinity of a hemostatic plug could lead to impaired adhesion and poor aggregation in response to agonists. In vivo, prolonged bleeding times were found in patients 90 minutes after t-PA infusion for thrombolysis, suggesting early impairment of platelet function upon plasmin generation.¹⁶³ However, there is also evidence that platelets may promote thrombotic reocclusion following successful thrombolytic therapy.¹⁶⁴

FIBRINOLYTIC FUNCTION IN VASCULARINJURY

Transgenic mouse models of vascular disease have helped to elucidate the complex role of the fibrinolytic system in atherosclerosis (Table 115–3).^165,166 In mice, the general effects of plasminogen deficiency include runting, fibrin deposition in intra- and extravascular locations, and premature death.^167,168 In addition, the mice display impaired healing of cutaneous wounds,¹⁶⁹ a response that appears to depend largely on the fibrinolytic action of plasmin as loss of fibrinogen eliminates these defects.¹⁷⁰ Mice doubly deficient in plasminogen and apolipoprotein E (ApoE) showed an increased predisposition to atherosclerosis compared to animals deficient in ApoE alone (Fig. 115–7A).¹⁷¹ Mice with ApoE deficiency combined with deficiency of either u-PA or t-PA showed the same predilection for early fatty streaks and advanced plaques as was observed in mice with isolated ApoE deficiency, suggesting that complete elimination of plasmin generating activity is required to exacerbate the proatherogenic state.¹⁷² Finally, mice doubly deficient in ApoE and PAI-1 exhibit no change in early plaque size at the aortic root,^173,174 decreased early plaque size at the carotid bifurcation,^173,174 but increased advanced plaque size with accelerated deposition of matrix.¹⁷⁵

Once the atherosclerotic plaque is established, plasmin may affect its evolution by mediating invasion of leukocytes (see Table 115–3).¹⁷⁶ In the peritoneal cavity, recruitment of inflammatory cells is profoundly influenced by the presence or absence of plasminogen.¹⁷⁷ In transplant-associated arteriosclerosis, the extent of disease is significantly reduced in plasminogen-deficient mice, reflecting, at least in part, reduced influx of macrophages, with an associated reduction in medial necrosis, fragmentation of elastic laminae, and remodeling of the adventitia.¹⁷⁸ Thus, the role of plasmin in degrading fibrin and other matrix constituents in the early lesion limits atherosclerosis, whereas its ability to promote cellular invasion later on appears to promote atherogenesis.

During aortic aneurysm formation in mice, deficiency of u-PA, but not t-PA, was associated with reduced medial destruction and impaired activation of downstream plasmin-dependent matrix metalloproteinases (Fig. 115–7B and Table 115–3).¹⁷² Similarly, u-PA–, but not t-PA–, deficient mice were protected from cardiac rupture secondary to ventricular aneurysm. In this study, temporary administration of PAI-1 or the general matrix metalloproteinase inhibitor, tissue inhibitor of metalloproteinase (TIMP)-1, completely protected wild-type mice from aortic rupture, reinforcing the concept that plasmin-based protease activity promotes aneurysm progression.¹⁷⁹

Vascular remodeling may occur following acute arterial injury induced by interventions for vascular compromise, leading to vascular restenosis (Fig. 115–7C and Table 115–3). This process reflects leukocyte invasion, proliferation and migration of smooth muscle cells, deposition of extracellular matrix, and reendothelialization. Electrical or mechanical injury studies in gene-targeted mice indicate that neointima formation, an initial step in restenosis, requires intact expression of plasminogen and u-PA, but not t-PA.^180,181,182 Interestingly, loss of uPAR has no effect on neointima formation,¹⁸³ whereas loss of PAI-1 is associated with increased neointimal stenosis.^184,185 In these injury models, which do not induce severe thrombosis, it is thought that vascular occlusion, reflecting migration of smooth muscle cells and leukocytes, is impaired when fibrinolytic potential is attenuated.¹⁸⁶

In the ferric chloride, Rose Bengal, and copper cuff models, on the other hand, thrombosis is observed within minutes of arterial injury (see Fig. 115–7 and Table 115–3). In these systems, deficiency of PAI-1 is associated with later and less-extensive thrombotic occlusion of the injured artery,^187,188 while loss of u-PA is associated with more rapid and more significant thrombotic occlusion.¹⁸⁹ At the same time, the absence of PAI-1 led to reduced vascular stenosis, regardless of whether ApoE was absent^190,191 or present.^192,193 In balloon-injured rat carotid arteries, finally, transduction of a PAI-1–expressing gene led to increased restenosis of the vessel, again suggesting that clearance of the initial thrombus may have longterm effects on vessel patency and neointima formation.¹⁹⁴ In these models, the predominant effect of the fibrinolytic system may be to clear the initial thrombus, which may provide a provisional scaffolding for later restenosis.

Endothelial cells use receptors, primarily uPAR and the annexin A2/S100A10 system, to assemble the fibrinolytic system on their surface (Chap. 135; Fig. 115–6). Recent evidence suggests that impairment of receptor-mediated fibrinolytic assembly may lead to vascular compromise.

LIPOPROTEIN(a)

Lp(a) is a low-density lipoprotein (LDL)-like particle that is an independent risk factor for atherosclerosis.^195,196,197 In addition to the apolipoprotein B-100 found on LDL, Lp(a) contains a disulfide-linked moiety called apolipoprotein(a) [apo(a)]. Apo(a) shares a remarkable degree of homology with plasminogen, as it possesses multiple tandem repeats of a kringle IV-like domain, a single domain resembling kringle V, and a pseudoprotease domain.^198,199 Plasminogen and apo(a), furthermore, are closely linked on chromosome 6 and appear to have arisen from a common ancestral gene.²⁰⁰

Whereas Lp(a) levels are only transiently responsive to diet,^201,202 plasma levels appear to be subject to mendelian inheritance.^203,204,205 Plasma Lp(a) concentrations correlate inversely with the ratio of kringle IV to kringle V encoding domains within the apo(a) gene,^206,207 such that larger apo(a) gene products are associated with lower plasma concentrations of apo(a). In addition, Lp(a) is an acute phase reactant in the postsurgical and post–myocardial infarction setting,²⁰⁴ and in patients with cancer,²⁰⁵ suggesting a role for soluble inflammatory mediators in regulating its synthesis or assembly. Apo(a) possesses a high-affinity lysine binding site within kringle 4 that closely resembles that of kringle 1 of plasminogen,²⁰⁸ and kringle 37 of the originally cloned apo(a) resembles the lysine-binding plasminogen kringle 4 of plasminogen.²⁰⁹ In vivo, Lp(a) colocalizes histologically with fibrin in atheromatous tissue.²¹⁰

When apo(a) is overexpressed in transgenic mice,²¹¹ cell-associated plasmin activity is reduced such that the animals are resistant to t-PA thrombolysis.²¹² Three potential explanations for the prothrombotic, proatherogenic effect of Lp(a) include (1) the observation that both Lp(a) and apo(a) inhibit Lys-plasminogen binding to endothelial cells (half-maximal inhibitory dose [ID₅₀] = 36-fold excess),²¹³ and to annexin A2,²¹⁴ with affinity similar to that for plasminogen^215,216,217; (2) endothelial cell exposure to Lp(a) in vitro enhances expression of PAI-1¹³⁷; and (3) Lp(a) may act as a competitive inhibitor of t-PA in the presence of fibrinogen,²¹⁸ or as an uncompetitive inhibitor of the fibrin-dependent enhancement of t-PA–induced plasmin generation.²¹⁹ Overexpression of Lp(a) in mice receiving a high-fat diet results in atherosclerosis-like lesions containing anti-apo(a) cross-reactive material.²²⁰ Deposition of both lipid and apo(a) was reduced in mice expressing apo(a) in which lysine binding sites had been mutated.²²¹ Thus, lysine binding sites of apo(a) appear to allow it to compete with plasminogen for cell surface receptors, thereby increasing atherogenicity.

HOMOCYSTEINE

Homocysteine is a thiol-containing amino acid that accumulates in nutritional deficiencies of vitamin B₆, vitamin B₁₂, or folic acid, or in inherited abnormalities of cystathionine β-synthase, methylene tetrahydrofolate reductase, or methionine synthase.²²² Multiple studies have shown homocysteine to be an independent risk factor for atherosclerosis,²²³ venous thromboembolism, and death.²²⁴ Homocysteine lowering in patients with inborn errors of homocysteine metabolism leads to a striking reduction in cardiovascular morbidity, but supplementation with B vitamins in patients with established cardiovascular disease is of no benefit.²²⁵ In vitro, homocysteine-treated endothelial cells bound approximately 50 percent less t-PA than untreated cells, and activated approximately 50 percent less plasminogen.²²⁶ Mass spectrometry studies indicate that homocysteine directly disables the t-PA binding domain of annexin A2 by forming a covalent adduction product with cysteine 9 within the tail domain of purified annexin A2, thus inhibiting its ability to bind t-PA.²²⁷ Mice on a high homocysteine-generating diet, moreover, displayed dysfunctional annexin A2, and loss of both fibrinolytic and angiogenic potential. This phenotype mimicked that of annexin A2-deficient mice, and was overcome upon intravenous infusion of unmodified annexin A2.²²⁸

ANTIPHOSPHOLIPID SYNDROME

Antiphospholipid syndrome (APS) is an autoimmune disorder characterized by thrombosis, recurrent pregnancy loss, and persistently positive antiphospholipid antibodies.^229,230 Compared to patients with lupus erythematosus, nonimmune thrombosis, or healthy controls, a relatively high proportion of patients with APS and severe thrombosis (22 percent) has antibodies directed against annexin A2. Antiannexin A2 antibodies may block t-PA–dependent cell-surface plasmin generation, and also induce expression of procoagulant molecules, such as TF.¹⁵² These events may require crosslinking of β₂-GPI bound to closely associated cell-surface A2,^231,232 and signaling via myeloid differentiation protein 88 (MyD88) and nuclear factor kappa B (NFκB)-dependent pathway.²³³ Additional evidence shows that A2 is required for the pathogenic effects of antiphospholipid antibodies in mice.²³⁴ High-titer antiannexin A2 antibodies are also associated with cerebral vein thrombosis in patients without the full diagnostic criteria for APS.¹⁵³

A proinflammatory environment is also prothrombotic. Endothelial cells express molecules that regulate binding of leukocytes to their surface during inflammation. These interactions have both direct and indirect roles in hemostasis and thrombosis, and, in some cases, interactions of leukocytes and platelets with inflamed endothelial cells feed forward to promote thrombosis.²³⁵ Moreover, the inflammatory response itself results in the expression of adhesion molecules and mediators that secondarily promote hemostasis. In addition, membrane microparticles derived from platelets, leukocytes, and perhaps endothelium, provide circulating sources of TF, proinflammatory lipids, and other molecules that have the potential to regulate thrombosis and inflammation at a distance from the primary site.^{236,237,238,239,240}

IMMEDIATE CHANGES

Either histamine or thrombin produced locally at the site of inflammation by degranulation of resident tissue mast cells stimulates the overlying endothelial cells to express P-selectin on their surfaces.²⁴¹ This change occurs within minutes and is caused by the rapid fusion of Weibel-Palade bodies, with the plasma membrane bringing P-selectin to the surface. Along with P-selectin expression, fusion of the Weibel-Palade bodies also results in the release of VWF into the local environment.

P-selectin serves as a leukocyte receptor for P-selectin glycoprotein ligand (PSGL)-1, L-selectin, and other ligands.²⁴² PSGL-1 is a specific sialomucin containing sialylated, fucosylated O-linked oligosaccharides as well as an unusual sulfated tyrosine residue motif.²⁴³ Dimerization of PSGL-1 is required for optimal recognition of P-selectin.²⁴⁴ Adhesive interactions between P-selectin and its ligands result in the tethering of passing leukocytes to, and rolling on, the surface of the endothelial cell as the first step in leukocyte emigration. PSGL-1 also interacts with P-selectin expressed on platelets that have become activated and adherent to endothelium.^45,245 L-selectin, another member of the selectin family, is constitutively expressed on most leukocytes. It binds to sialylated, fucosylated GP ligands expressed by endothelial cells in response to inflammation, as well as to CD34 constitutively expressed by cells of the high endothelial venules.

Adhesion of leukocytes to the endothelium at the site of inflammation results in their rolling along the luminal surface, which slows their movement and brings them into contact with a wide variety of chemical mediators that trigger the next stage of leukocyte emigration—tight adhesion to the endothelial surface. These mediators include surface-bound chemokines,²⁴⁶ new adhesion molecules expressed by the endothelium in response to inflammatory cytokines,²⁴⁷ PAF,²⁴⁸ soluble chemokines,²⁴⁹ and ligands that crosslink leukocyte CD31^250,251,252 and seem to work by stimulating the activation of leukocyte integrin adhesion molecules by so-called inside-out signaling. This process involves a conformational change and/or clustering of the two chains of these heterodimeric surface molecules such that the affinity or avidity, respectively, for their ligands on the surfaces of endothelial cells is increased.²⁵³ The ligands identified are members of a third family of adhesion molecules, the immunoglobulin gene superfamily.²⁵⁴

Table 115–4 lists some of the more common leukocyte/endothelial CAM pairs participating in the inflammatory response. It is interesting to note that the mucosal addressin MAdCAM-1, a unique molecule expressed by endothelial cells of high endothelial venules of mesenteric lymph nodes and Peyer patches, has structural features of both a mucin and an immunoglobulin superfamily molecule. It can bind both L-selectin and the leukocyte integrin α₄β₇, expressed by a subset of memory T cells. It is believed to interact with L-selectin through its mucin (carbohydrate) domain and with α₄β₇ through its immunoglobulin domains. Until recently, identified protein ligands for L-selectin (MAdCAM-1 and CD34) have been demonstrated to bind to L-selectin only in the context of lymphocyte homing, although recent evidence now suggests that they may play a role in rolling during inflammation.²⁵⁵ Interestingly, intravital microscopy shows that leukocytes may roll on already adherent leukocytes and platelets through the interactions of L-selectin and PSGL-1 to amplify the inflammatory process.^256,257

PAF is made and secreted acutely by leukocytes, mast cells, and endothelial cells at the site of inflammation. PAF (1-alkyl-2-acetyl-sn-glycero-3-phosphocholine) is produced enzymatically from phosphatidyl choline in the plasma membrane. Although its role as an activator of neutrophils in this environment has been established,²⁴⁸ it appears to be a relatively weak agonist of platelet activation in this location.

Adherent leukocytes migrate to nearby interendothelial junctions by repeated cycles of adhesion in the front and disadhesion in the rear.^254,258 At the junction, additional distinct molecular interactions between leukocytes and endothelial cells regulate transendothelial migration for the vast majority of neutrophils, monocytes, and natural killer (NK) cells. Transendothelial migration and the nomenclature of the junctional adhesion molecule (JAM) family are reviewed elsewhere.^17,259,260 PECAM/CD31 on the leukocyte contacts the same molecule concentrated at the endothelial junctions in a homophilic manner.^261,262,263 Although the relevant signal(s) transduced by this interaction remain unclear, a transient rise in endothelial cell intracellular calcium is required for transmigration.²⁶⁴ Blocking the function of either leukocyte PECAM or endothelial cell PECAM arrests the leukocyte poised over the junction,^263,265,266 a phenotype very similar to that seen when the rise in intracellular calcium is blocked by the chelator, bis(2-amino-5-methylphenoxy)ethane-N,N,N’,N’-tetraacetic acid tetraacetoxymethyl ester (MAPTAM).²⁶⁴

Because anti-PECAM reagents never block diapedesis completely, PECAM-independent pathways of transendothelial migration must exist. The leukocyte integrins α₄β₁ (very-late antigen [VLA]-4) and α_Lβ₂/α_Mβ₂ (lymphocyte function-associated antigen [LFA]-1/macrophage [Mac]-1) and their endothelial counter-receptors VCAM-1 and ICAM-1, have been implicated in transmigration.²⁵⁴ Interaction of leukocyte LFA-1 with JAM-A on endothelial cells has also been implicated in leukocyte recruitment.²⁶⁷ Antibodies directed against JAM-C also blocked migration of lymphocytes across endothelial cell monolayers, implicating its role in lymphocyte migration.²⁶⁸ In addition, under certain specialized conditions, there appear to be pathways across the endothelial cell that bypass the intercellular junction.^269,270

CD99 is a GP expressed on leukocytes, platelets, and erythrocytes, and concentrated at the endothelial cell borders. CD99 controls a step in diapedesis distal to the step controlled by PECAM both in vitro and in vivo,^271,272,273 by interfering with homophilic interaction between leukocyte CD99 and endothelial cell CD99-arrested monocytes. Their leading edges were below the endothelial cell monolayer, while their trailing uropods remained on the apical surface of the endothelial cell.

At the onset of most acute inflammatory responses, vascular permeability transiently increases as a result of histamine release. The endothelial junctions are soon re-established, and the junctions are closed to the leukocytes that arrive at the scene over the next hour. Studies performed both in vivo and in vitro indicate that, during subsequent diapedesis, leukocytes penetrate the vessel wall without further compromising the vascular permeability barrier.^264,274 Cortactin-deficient mice, for example, have constitutively leaky vascular junctions in the postcapillary venule circulation, and an exaggerated response to histamine, yet leukocyte recruitment is diminished because of inefficient clustering of ICAM-1,²⁷⁵ which prevents exposure of subendothelial collagen and VWF deposits to circulating platelets. Although PECAM-1 has no known role in binding platelets to endothelial cells, it has been hypothesized to maintain the tight apposition of endothelial cells and leukocytes during diapedesis.²⁶³

ACUTE CHANGES

In addition to stimulating endothelial cell immediate responses, cytokines and inflammatory mediators released at the site of inflammation activate new endothelial cell genetic programs. Within several hours of exposure to mediator, de novo synthesis of mRNA and protein establishes an inflammatory endothelial cell phenotype that is both procoagulant and proadhesive.

Inflammatory cytokines such as TNF-α and IL-1 induce endothelial cell surface expression of several important CAMs. Expression of E-selectin peaks at 4 to 6 hours in vitro, but may be maintained by interferon (IFN)-γ for several days in vivo.^276,277 E-selectin mediates the slow rolling of leukocytes bearing sialylated, fucosylated carbohydrate receptors similar to sialylated Lewis X antigen.²⁷⁸ Endothelial cell P-selectin stimulated by thrombin or histamine is transient, but can be prolonged to hours or days by IL-3, IL-4, or oncostatin M stimulation of human endothelium, and by TNF-α stimulation of murine, but not human endothelium.^{279,280,281,282}

In general, expression of the immunoglobulin superfamily members ICAM-1 and VCAM-1 is induced by the same stimuli that induce E-selectin. Some specializations exist, at least in vitro. For example, IL-4 induces VCAM-1 but not E-selectin or ICAM-1 in microvascular endothelial cells.^283,284 These molecules serve as counter-receptors for the leukocyte integrins in the tight adhesion step.

CHRONIC CHANGES

Stimulation of endothelial cells over several days with IFN-γ leads to surface expression of MHC class II molecules (human leukocyte antigen [HLA]-DR and -DQ). In human tissues such as skin and gut, class II is commonly seen even in the absence of overt inflammation, and is thought to be a result of chronic exposure of these sites to subclinical inflammation and antigenic stimulation. When costimulatory molecules such as CD40, ICAM-1, or LFA-3 are induced by inflammatory stimuli, the endothelial cell becomes capable (at least in vitro) of acting as an antigen presenting cell that can stimulate CD4+ memory T cells. This mechanism may stimulate graft rejection by the host when the endothelium belongs to an organ graft with foreign MHC class II.^285,286,287

In contrast, the expression of the adhesion molecule ICAM-2 does not change in response to inflammatory mediators. PECAM-1 shows a unique expression pattern in response to IFN-γ in vitro²⁸⁸ and in vivo,²⁸⁹ as its distribution becomes diffuse over the surface of the cell, rather than being concentrated at intercellular borders. In vitro chronic exposure of human umbilical vein endothelial cells to a combination of IFN-γ and TNF-α at relatively high doses leads to a decrease in total PECAM-1 expression.²⁹⁰ Such a response has not been described to date in vivo.

ADHESION MOLECULES IN A THROMBOTICMILIEU

Activation of the hemostatic system exposes leukocytes to ligands that promote their adhesion and recruitment to the vessel wall. For example, in vitro thrombin induces E-selectin expression and IL-8 secretion by human umbilical vein endothelial cells.²⁹¹ These changes are classically induced by inflammatory cytokines such as IL-1 and TNF-α. Table 115–5 lists some mediators that could have dual roles in inflammation and hemostasis/thrombosis.

LEUKOCYTE–PLATELET AND ENDOTHELIAL CELL–PLATELET INTERACTIONS

Activated platelets bind to circulating lymphocytes in a P-selectin–dependent manner. This interaction can facilitate leukocyte rolling on the endothelium²⁹² and also allows homing of lymphocytes to peripheral lymph nodes in the absence of L-selectin, because P-selectin on the adherent platelets will interact with the peripheral lymph node addressin.²⁹³ In vitro, neutrophils are capable of rolling on immobilized platelets via PSGL-1 on the leukocyte interacting with degranulated P-selectin on the platelet membranes.²⁹⁴ Moreover, α_Mβ₂ (CD11b/CD18)-dependent arrest and tight adhesion of neutrophils to bound platelets following P-selectin–dependent rolling has been described.^294,295 The endothelial ligand for this is not known. ICAM-2 has been found on the surface of activated platelets, but it is not a ligand for α_Mβ₂. In fact, antibodies against neither ICAM-2 nor its neutrophil receptor α_L (CD11) blocked this adhesion.^295,296 On the other hand, neutrophil α_Mβ₂ reportedly binds to fibrinogen, which may be present on the surfaces of activated platelets bound to α_IIbβ₃ (GPIIb/IIIa). Two additional platelet surface molecules, GPIbα and JAM-C, have been demonstrated as ligands for leukocyte CD11b/CD18. GPIbα is part of the GP1b–IX–V complex,^297,298 and JAM-C was originally described as a component of epithelial and endothelial cell tight junctions.

Platelets can interact with activated endothelial cells. Platelets express PSGL-1 and can use this expression to interact with P-selectin on the surfaces of activated endothelial cells.²⁹⁹ Activated platelets can also bind to endothelial cells via fibrinogen, fibrin, or VWF, forming a molecular bridge between platelet GPIIb/IIIa and endothelial cell integrin α_vβ₃ and ICAM-1.

Ultralarge VWF molecules are stored in Weibel-Palade bodies and released upon inflammatory endothelial cell activation. Ultralarge VWF molecules are normally cleaved by endothelial cell-surface proteases, notably the metalloprotease ADAMTS13 (a disintegrin and metalloprotease with thrombospondin repeats 13). Data in mice suggest that ADAMTS13 plays a homeostatic role in dampening inflammation.³⁰⁰ In ADAMTS13-deficient mice, platelets bound to ultralarge VWF molecules on the endothelial surface and supported slower leukocyte rolling on venules at rest and greater leukocyte extravasation in models of inflammation.³⁰⁰

LEUKOCYTE–ENDOTHELIAL CELL MATRIX INTERACTIONS THAT PROMOTECOAGULATION

The same proinflammatory stimuli that stimulate de novo expression of E-selectin and VCAM-1, and augment expression of ICAM-1 for the recruitment of leukocytes, may stimulate synthesis and expression of TF by endothelial cells.³⁰¹ Furthermore, adhesion of monocytic cell lines to cytokine-activated endothelial cells in culture leads to rapidly increased TF-related procoagulant activity. This effect is partially blocked by a monoclonal antibody directed against E-selectin on endothelium and is mimicked by crosslinking Le^X on the monocyte cell lines.³⁰² A similar increase in TF gene expression can be induced by crosslinking α₄ or β₁ integrin chains, the components of VLA-4 on monocytic cell lines.³⁰³

During prolonged interaction of peripheral blood monocytes with human endothelial cells, monocytes that migrated across endothelial cell monolayers expressed functional cell surface TF.³⁰⁴ Over the next several days, approximately half of these monocytes differentiated into immature dendritic cells bearing even higher levels of TF, and migrated back across the intact endothelial cell monolayer. This migration could be blocked by soluble fragments of TF. Therefore, in this system, TF was hypothesized to support both adhesion and a possible procoagulant role.³⁰⁴

Leukocytes that bind to P-selectin exposed on the surfaces of platelets on adherent thrombi promote the conversion of fibrinogen to fibrin.³⁰⁵ The leukocyte integrin CD11b/CD18 has been shown to bind fibrinogen.³⁰⁶ The same integrin has a conformational form that binds coagulation factor X.³⁰⁷ Monocytic cells are capable of activating the bound factor X to factor Xa when activated,³⁰⁸ defining a pathway for activation of factor X that is independent of TF.