The role of the endothelium in regulating blood fluidity and trafficking of circulating hematopoietic cells has been the subject of several reviews (Cines et al., 1998; Aird, 2003). ECs express a variety of factors that inhibit coagulation, including soluble substances, such as nitric oxide and prostacyclin (acting to inhibit platelet activation), and tissue-type plasminogen activator (t-PA, acting to promote fibrinolysis), among many others. EC surface-bound molecules with anticoagulant activity include heparan sulfate-containing proteoglycans (see below), thrombomo-dulin (TM), complement regulatory proteins, as well as receptors for activated protein C (APC), urokinase, and plasminogen.
Unperturbed ECs also do not express several moieties that promote platelet and leukocyte adhesion, such as endothelial leukocyte adhesion molecule (ELAM), P-selectin, and platelet-activating factor (PAF). These can be induced, however, when the cells are stimulated by agonists, such as cytokines, thrombin (Drake et al., 1993; Kaplanski et al., 1998), or when the cells are injured by immune factors, atherosclerosis, or shear stress (Yu et al., 2005). Additionally, ECs exposed to such factors express a reduced content of heparan sulfate, internalize and degrade APC, elaborate tissue factor, and secrete abundant plasminogen activator inhibitor-1 (PAI-1), each of which may promote thrombus formation (Cines et al., 1998). Histochemical studies of the endothelium in murine models of inflammation have confirmed many of these observations, predicated in cell culture (Fries et al., 1993), affirming the notion that the endothelium undergoes multifaceted changes from an antithrombotic to a procoagulant phenotype in response to injury.
Also relevant to the pathogenesis of HIT is the remarkable heterogeneity of ECs, within and among different vascular beds, owing to genetic differences and acquired changes in phenotype (for reviews: see Cines et al., 1998; Aird, 2003). For example, only a small fraction of ECs constitutively expresses t-PA or urokinase-type plasminogen activator (u-PA) in vivo (Levin et al., 1994), whereas a different subset expresses tissue factor when exposed to endotoxin (Drake et al., 1993). ECs from different organs express tissue-specific promoters that regulate the expression of von Willebrand factor (vWF) in vivo (Aird et al., 1997). ECs also show regional variation in the synthesis of prostacyclin and expression of leukocyte adhesion molecules and Fcg receptors, among many other phenotypic differences.
There is also evidence to indicate that protein C activation on macrovascular ECs is mediated predominantly through the protein C receptor, whereas TM may dominate in the microvasculature (Laszik et al., 1997; Van de Wouwer et al., 2004). TM changes thrombin from a procoagulant to an anticoagulant enzyme (i.e., TM-bound thrombin activates the natural anticoagulant zymogen, protein C) (Esmon, 2001). Targeted disruption of the endothelial TM gene leads to juvenile onset of thrombosis (Isermann et al., 2001). The anticoagulant function of TM in the microvasculature may contribute to the pathogenesis of warfarin-associated venous limb gangrene that can complicate HIT. This syndrome has been attributed to the coincidence of persistent thrombin generation and acquired protein C deficiency that may occur during the first few days of anticoagulation with warfarin (Warkentin et al., 1997; Srinivasan et al., 2004) (see Chapters 2 and 12).
The behavior of ECs can also be modified during the evolution of vascular disease. For example, atherosclerotic vessels produce less nitric oxide in response to a variety of stimuli than do healthy vessels (Shaul, 2003). Atherosclerotic vessels may also undergo alterations in their expression of glycosaminoglycans (GAGs)
(Talusan et al., 2005) and an increase in expression of various cell adhesion molecules (for review: see Fuster et al., 1998). The binding of advanced glycation end products to specific EC receptors during normal aging and diabetes mellitus increases vascular permeability, exposing the subendothelial matrix to lipoproteins and other injurious substances (Basta et al., 2004). It is also likely that genetic variation in EC behavior contributes to the host response to antibody- and platelet-mediated EC injury, although the methods to identify or monitor such risk factors remain to be developed. Thus, any inquiry into the reason why only a subset of patients who develop anti-PF4-heparin antibodies develop thrombosis, or why thrombi occur at restricted vascular sites, must take into consideration the specific attributes of the affected endothelial vascular bed.
III. HEPARAN SULFATE-CONTAINING PROTEOGLYCANS, HEPARIN, AND THE ENDOTHELIUM
The expression and anticoagulant function of heparan sulfate-type proteoglycans (HSPGs) by ECs may be central to the pathogenesis of vascular thrombosis in patients with HIT. The biochemistry and function of these GAGs and the proteo-glycans to which they bind have been the subject of extensive study (for review: see Rosenberg et al., 1997; Esko and Lindahl, 2001; Forsberg and Kjellen, 2001). The involvement of heparan sulfate in the development of HIT is considered elsewhere (see Chapter 7). HSPGs expressed by ECs bind antithrombin (AT) in vitro and in vivo, and accelerate the inactivation of thrombin and factor Xa approximately 20-fold, an effect that is biologically equivalent to 0.1-0.5 U/mL of heparin (Marcum and Rosenberg, 1984). Yet less than 1% of the HSPGs isolated from cultured ECs express anticoagulant activity (Marcum and Rosenberg, 1984). Active species are characterized by an approximately twofold enrichment in glucuronyl 3-O-sulfated glucosamine residues, compared with inactive species (Marcum and Rosenberg, 1984). Interestingly, targeted deletion of the murine 3-O-sulfo-transferase-1 enzyme (the enzyme responsible for generating this anticoagulant modification of HSPGs) does not lead to a prothrombotic phenotype (HajMohammadi et al., 2003). The physiological mechanisms that control the synthesis and postsynthetic modifications of HSPG remain an active area of investigation (Forsberg and Kjellen, 2001).
Microheterogeneity in the composition of HSPG in arteries, veins, and capillaries has been noted (Lowe-Krentz and Joyce, 1991), but the significance of these differences is unknown. Expression of HSPG by ECs undergoes developmental changes (David et al., 1992), and its composition varies after the cells are exposed to thrombin (Benezra et al., 1993), homocysteine (Nishinaga et al., 1993), heparin (Nader et al., 1989), wounding and migration (Kinsella and Wight, 1986), and after induction by activated platelets (Yahalom et al., 1984), among other stimuli. ECs also bind heparin (for review: see Patton et al., 1995), which alters their proliferation, matrix composition, and many other vascular functions. It has also been reported that AT is displaced from ECs by heparin, and its binding is inhibited by PF4 (Stern et al., 1985). Whether HIT antibodies promote the capacity of PF4 to neutralize AT activity has not been reported.
The biochemistry of PF4 and its involvement in HIT is reviewed elsewhere (see Chapter 5). The metabolism of the protein is regulated by its interactions with the endothelium. PF4 is stored in the a-granules of platelets as a tetramer bound to chondroitin sulfate (Barber et al., 1972). The tetramer may dissociate from the GAG as the platelets are activated, but more likely, dissociation occurs subsequent to binding to EC HSPG, which contains a higher charge density. [125I]PF4 is cleared from the circulation with an a-elimination phase approximating 2 min, which primarily represents binding to the endothelium, and a (-elimination phase approximating 40 min, corresponding to uptake and degradation, predominantly by hepatocytes (Rucinski et al., 1986, 1990).
The endothelium binds approximately 50 pmol PF4/105 cells (Rybak et al., 1989). Several classes of binding sites have been identified, including a high-capacity, low-affinity site on HSPG, as well as higher-affinity binding sites involving specific chemokine receptors and certain coagulant proteins (see below). Binding of PF4 to the endothelium is attenuated by pretreatment with heparinase (Marcum et al., 1984), and plasma concentrations are increased 10- to 20-fold after heparin is infused intravenously (Dawes et al., 1982). Binding of PF4 to EC GAGs is electrostatic (Wu et al., 1984) and is independent of the pentasaccharide involved in the binding of AT (Loscalzo et al., 1985). The affinity of PF4 binding to ECs is lower than to purified heparin (Kd = 2-3 mmol/L vs. 2nmol/L, respectively) (Rybak et al., 1989), consistent with the biochemical heterogeneity of vascular matrix. PF4 has a 10- to 100-fold greater affinity for EC HSPG than does AT ( Jordan et al., 1982) and thus markedly attenuates the antiprotease cofactor activity of AT on intact vessels (Busch et al., 1980; Stern et al., 1985).
The involvement of PF4 in hemostasis is mediated in part by charge-dependent interactions with EC proteoglycans (Eslin et al., 2004). Recent in vivo studies utilizing murine PF4 knock-out (mPF4_/~) and human PF4 (hPF4+) transgenic animals show that PF4 helps stabilize clots formed in response to EC injury. Both mice lacking and those that overexpress PF4 show delayed and unstable clot formation, indicating that a narrow range of PF4 concentrations is needed for efficient clot formation (Fig. 1). The defect in thrombus formation in mPF4_/~ can be corrected by infusion of human PF4 or protamine sulfate. On the other hand, the overexpression of PF4 seen with hPF4+ animals as well as infusions of protamine into wild-type animals is associated with impaired thrombus formation, which can be reversed through charge neutralization using heparin (Eslin et al., 2004). These studies suggest that PF4 facilitates clot formation by neutralizing negatively charged surfaces on platelets and ECs, perhaps allowing closer approximation of platelets to each other and to the endothelial lining. In settings where insufficient or excessive PF4 is released, cell surfaces may retain a net negative or positive charge, respectively, which prevents optimal approximation of cellular elements (Eslin et al., 2004).
The existence of an "optimal" concentration range of PF4 for hemostasis mirrors the binding of HIT antibody to complexes formed at various molar ratios of PF4 to heparin (Bock et al., 1980; Greinacher et al., 1994; Rauova et al., 2005). PF4 and heparin form ultra-large complexes (ULCs, MW > 670 kDa) over a narrow range of molar ratios, approximating 1 mole of PF4 tetramer to 1 mole of unfractionated heparin, the ratio at which HIT antibody binding is optimal (Rauova et al., 2005) (see Chapter 5). These ULCs are preferentially recognized by KKO, a murine HIT-like antibody. Recent in vivo studies affirmed the importance of these findings. When KKO is administered to double transgenic mice expressing platelet hFcgRIIA and varying levels of hPF4 (hPF4high/hFcgRIIA, hPF4mid/ hFcgRIIA, or hPF4low/hFcgRIIA), the severity of thrombocytopenia is proportionate
Was this article helpful?