The Yin–Yang of thrombin and activated protein C

Authors


Prof. C. H. Toh, Haematology Department, Royal Liverpool University Hospital, Prescot Street, Liverpool L7 8XP, UK.
E-mail: c.h.toh@liv.ac.uk

Summary

The concepts of Yin and Yang provided the intellectual framework of much of Chinese scientific thinking, especially in the fields of biology and medicine. The organs of the body were seen to be inter-related and their functions could be best appreciated through understanding connections and correlations, in the same way as in other naturally occurring phenomena. Although this principle is recognizable within our current understanding of pro- and anti-coagulant mechanisms, the literature in this field seldom conveys the unifying nature of its overall structure. Considering the coagulation cascade and the anticoagulant system as opposing but complementary forces would be central to the concept of Yin and Yang. This article is presented along such lines with a focus on thrombin and activated protein C (APC) as key examples of that balancing axis in maintaining haemostatic harmony. The emphasis will be on how understanding this relationship at the molecular and cellular level holds promise in the translation to improved clinical care.

Background

Essentials of the Yin-Yang School began in the early Han Dynasty (207 BC–9 AD). Underpinned by the central philosophy that the universe was governed by a single principle, all matters were felt to be shaped by two opposing forces; i.e. Yin and Yang. Each of these opposites produces the other and neither can be continually dominant (Gu, 2003). With this concept as contextual reference, the evolution of the coagulation system has balanced the needs of stemming fluid loss upon injury with maintaining a circulatory transport system. Its evolutionary history predates mammalian development itself. Data from biochemical, molecular cloning and comparative sequence analysis shows evidence of similar coagulation components in all jawed vertebrates and the human haemostatic system. This would therefore appear to be a highly conserved system that evolved before the divergence of teleosts and tetrapods over 430 million years ago (Davidson et al, 2003a). Phylogenetic analysis supports evolution of the network through several rounds of gene duplication, giving rise to novel sequences that result in new proteins with related characteristics but different functions (Davidson et al, 2003b).

This is exemplified by the vitamin K-dependent proteins involved in coagulation and anticoagulation; prothrombin, factors VII, IX, X, protein C (PC) and protein S (PS), which all share gamma carboxyglutamic acid domains (Fernlund & Stenflo, 1982). The functional consequence of this is in conserving an enzymatic system that assembles a vitamin K-dependent protein on to negatively charged phospholipid surfaces together with a catalytic co-factor in the presence of calcium. For example, the prothrombinase complex involves prothrombin as the vitamin K-dependent protein and activated factor X (factor Xa) as co-factor, the tenase complex brings factor X as the vitamin K-dependent protein to interact with activated factor IX (factor IXa) as its co-factor, and similarly, the anticoagulant APCase complex assembles vitamin K-dependent PC with PS as co-factor in the inactivation of activated factors V and VIII (factor Va and VIIIa respectively). When stably localized to an anionic phospholipid template, the enzymatic rate can be accelerated 250 000 fold (Boskovic et al, 1990). In addition, variations in lipid composition can directly influence the resulting coagulant activity. In circulation, procoagulant lipids, such as phosphatidylserine, are predominantly found on activated platelet surfaces (Cutsforth et al, 1989; Lentz, 2003) and triglyceride-rich particles in plasma. The latter is primarily in the form of very low density lipoprotein (Govers-Riemslag et al, 1994) but low density lipoprotein is also able to promote prothrombin activation when oxidized (Rota et al, 1998). Conversely, high density lipoprotein (HDL) exhibits anticoagulant properties (Griffin et al, 2001). Other anticoagulant lipids include phosphatidylethanolamine (Smirnov & Esmon, 1994), cardiolipin (Fernandez et al, 2000) and glucosylceramide (Deguchi et al, 2001) and some of these are specifically located within cell membrane microdomains to concentrate their anticoagulant activity (Simons & Ikonen, 1997; Brown & London, 2000).

Molecular and cellular aspects of thrombin and APC generation

Coagulation

Thrombin generation is characterized by the sequential activation of a series of serine proteases that can be broadly divided into initiation, propagation and amplification stages (Davie et al, 1991). Upon injury and consequent tissue factor exposure, the activation of factor VII leads to initial thrombin generation, which can then propagate further thrombin generation through activating components within the intrinsic pathway (Osterud & Rapaport, 1977). This leads to amplification in the coagulant process with the rapid and explosive conversion of fibrinogen to fibrin and the securing of haemostasis (Blombäck, 1996). Simultaneously, there is thrombin-mediated platelet activation through protease activated receptors (PARs) (Andersen et al, 1999) as well as thrombin cleavage of activated factor XIII (factor XIIIa) to stabilize fibrin formation (Sakata & Aoki, 1982; Shen & Lorand, 1983). Further clot consolidation is mediated by thrombin through activation of thrombin-activatable fibrinolysis inhibitor (TAFI), which retards clot lysis by removing binding sites for the attachment of tissue plasminogen activator (tPA) to fibrin for consequent lysis (Bajzar et al, 1996).

Whilst a burst in thrombin generation is crucial in achieving prompt haemostasis, uncontrolled thrombin generation may be harmful. Indeed, thrombin generation is kept in close check under physiological conditions with a short half-life of only 15 s (Jesty, 1986). It is rapidly inactivated by different inhibitors within the family of serine protease inhibitors (SERPINS), such as antithrombin (Rosenberg & Rosenberg, 1984), α1-antitrypsin (α1AT) and heparin co-factor II (Derechin et al, 1990), as well as by α2-macroglobulin (α2M) (Feinman et al, 1985). At the boundaries of tissue injury, any excess thrombin is then bound by the receptor thrombomodulin (TM) on intact endothelial surfaces. TM-bound thrombin can no longer express procoagulant activity (Esmon et al, 1982; Maruyama et al, 1985) but instead activates PC (Esmon & Owen, 1981). This rate of activation is increased significantly when PC is bound or complexed to the endothelial protein C receptor (EPCR) at the endothelial surface (Stearns-Kurosawa et al, 1996; Taylor et al, 2001). In the absence of EPCR, PC activation is reduced by 20-fold (Fukudome et al, 1998). The resulting APC can then express anticoagulant properties by cleaving and inactivating factors Va (Kisiel et al, 1977; Marlar et al, 1982) and VIIIa (Vehar & Davie, 1980) to reduce thrombin generation and procoagulant activity. As with other enzymes, APC has its own regulatory network consisting of PC inhibitor (Suzuki et al, 1987) and α1AT as first-line inactivators (Heeb & Griffin, 1988) with α2M as a secondary, calcium-dependent inhibitor (Hoogendoorn et al, 1991) (Scully et al, 1993).

Of interest is that APC anticoagulant activity in solution can be inhibited by soluble EPCR (sEPCR) (Fukudome et al, 1996). This is somewhat paradoxical as membrane-bound EPCR promotes anticoagulant activity. This apparent Janus type function is due to altered exposure of the APC active site upon conformational change in EPCR, as a result of cleavage at the cell surface (Esmon, 2003). sEPCR, when cleaved and released from the cell surface by tumour necrosis factor-α (TNF-α) converting enzyme (TACE) (Qu et al, 2007), is 4 kDa smaller than its membrane counterpart. Receptor shedding is upregulated by thrombin (Gu et al, 2000; Xu et al, 2000) and this apparent auto-regulatory mechanism of EPCR could theoretically modulate coagulant activity in the immediate vicinity of the cell surface. However, this alteration in APC within sEPCR could have other structure-function consequences given that it retains proteolytic activity towards small substrates (Regan et al, 1996).

Membrane-bound EPCR is homologous to the CD1/major histocompatibility class 1 family of molecules, most of which are involved in inflammation and antigen presentation (Fukudome & Esmon, 1995). Indeed the crystal structure of EPCR in complex with PC shares similarities with phospholipid binding within the antigen-presenting groove of the CD1 family (Oganesyan et al, 2002). Presentation and exchange of pathogen or cell-derived lipids may be involved in the processes by which these molecules modulate the inflammatory and immune response (Prigozy et al, 2001).

The relatively long half-life of APC, at approximately 20 min (Okajima et al, 1990), maintains a circulating concentration of 38 pmol/l (Gruber & Griffin, 1992). This contrasts with the much shorter half-life of thrombin and is likely to relate to the importance of maintaining vascular patency following localization of clot formation. Thrombin needs to be rapidly inactivated in solution to ensure that its potent effects are not further disseminated. Clot-based thrombin, however, can still organize and consolidate the fibrin structure, and also trigger the repair/healing mechanisms following injury (Mann et al, 2003). It is not known if there are functionally significant concentrations of clot-based APC although leucocytes bearing TM (Grey & Hancock, 1996) and EPCR (Galligan et al, 2001) are entrapped within developing thrombi. APC has been associated with pro-fibrinolytic properties that have been ascribed to its ability to complex plasminogen activator inhibitor-1 (de Fouw et al, 1987; Rezaie, 2001). However, the kinetics of this reaction do not appear to be physiologically relevant and an indirect role via reducing thrombin activation of TAFI would seem to be more likely in explaining this pro-fibrinolytic role.

Inflammation

Thrombin has a number of cellular effects. In addition to promoting platelet activation (Davey & Lüscher, 1967) and stimulating release of their mediators (Hamberg et al, 1975), it exerts influence over monocytes, macrophages (Bar-Shavit et al, 1986) and neutrophils in processes related to tissue repair at the site of injury (Frenette et al, 1996; Goldsack et al, 1998) In vitro, thrombin increases the expression of genes influencing vasomotor tone (Tesfamariam et al, 1993; Hamilton et al, 2001), cell proliferation, inflammation and leucocyte adhesion (Wu & Aird, 2005). Mediation of the thrombin pro-inflammatory response on the endothelium is thought to occur through various pathways including nuclear factor-κB (NFκB), early growth response factor-1 and GATA binding proteins (Minami et al, 2004). Primarily, this involves signalling through the family of 7-transmembrane G-protein-coupled PARs (Coughlin, 2000) and in humans, this is predominantly PAR-1 and PAR-3 (Ishihara et al, 1997).

With increasing recognition of the pleiotropic nature of APC, it is not surprising that areas of overlap with thrombin have been discovered. Whilst an anti-inflammatory role for APC can be indirect by nature of its ability to reduce thrombin generation, APC also has direct anti-inflammatory properties. APC can cleave and activate PAR1-dependent cellular pathways (Riewald et al, 2002). Proteolytic cleavage of the extracellular N-terminal tail forms a ligand within the G-protein coupled-PAR-1 receptor to trigger intracellular signalling (Coughlin, 2000). Unlike thrombin however, free APC is unable to cleave PAR-1 and requires membrane-bound EPCR as a co-factor for this effect (Guo et al, 2004). A further distinction from the thrombin–PAR-1 interaction is that EPCR–APC mediated signalling characteristically leads to anti-inflammatory and anti-apoptotic cellular effects. Examples of this include phosphorylation of mitogen-activated protein kinase (MAPK) (Riewald et al, 2002), suppression of NFκB expression (Joyce et al, 2001), as well as downregulation of p53 and thrombospondin 1 (Riewald & Ruf, 2005). An endothelial barrier protective effect is also observed through mechanisms involving sphingosine 1-phosphate (Feistritzer & Riewald, 2005; Finigan et al, 2005). In addition, the APC–PAR-1 signalling pathway leads to the generation of microparticles that can present and disseminate anticoagulant-active APC in complex with EPCR (Perez-Casal et al, 2005). These respective effects contrast with those mediated by high-dose thrombin in both enhancing endothelial permeability (van Nieuw Amerongen et al, 1998) and generating pro-apoptotic microparticles (Sapet et al, 2006).

Non-PAR-1-dependent mechanisms for the anti-inflammatory properties of APC also exist. Many of these are EPCR-dependent and relate to EPCR expression on cells other than at the endothelial surface. These include neutrophils (Kurosawa et al, 2000), eosinophils (Feistritzer et al, 2003) and monocytes (Feistritzer et al, 2006). In vitro, APC on human neutrophils can inhibit chemotaxis triggered by interleukin-8 (Sturn et al, 2003). The binding of sEPCR to CD11b/CD18 and to activated neutrophils via proteinase-3 can interfere with leucocyte adhesion and neutrophil-signalling events (Kurosawa et al, 2000). These findings may be relevant to the observation of reduced neutrophil accumulation in response to rhAPC in an in vivo human model of acute pulmonary inflammation (Nick et al, 2004).

Physiological and pathological aspects of thrombin and APC generation

The molecular link between thrombin and APC has also been demonstrated in vivo. Low-dose thrombin leads to an anticoagulant effect whilst excessive thrombin is pro-coagulant. In early canine studies, it was observed that extra corporeal circulation could be maintained in the absence of heparin as an anticoagulant (Fletcher et al, 1976). Subsequently, it was characterized that low-dose thrombin infusion was associated with an early anticoagulant and fibrinolytic response (Comp et al, 1982). Whilst there is no evidence that the low thrombin generating potential in haemophilia can exacerbate bleeding through enhancing PC activation, there is evidence of reduced TAFI activity within haemophilic plasma (Broze & Higuchi, 1996).

Thromboembolism

At the clinical level, deficiencies in the PC pathway have been known to cause thrombophilia. The clinical phenotype of heterozygous PC deficiency ranges from asymptomatic or superficial venous thrombophlebitis to venous thromboembolism (VTE) (Marciniak et al, 1985). Patients with homozygous deficiencies usually have severe thromboembolic consequences in the neonatal period (Marlar & Neumann, 1990).

The most significant illustration of the dynamics between thrombin and APC is that of APC resistance due to the factor V (FV) Leiden mutation. Dahlback et al (1993) observed that the clotting time of plasma from a thrombophilic patient could not be prolonged by the addition of exogenous APC. The patient’s plasma was described as APC-resistant and following further studies, including linkage analysis in two thrombophilic families, an association between the APC resistant phenotype and the F5 locus was identified. The principal cause of this was found to be a F5 R506Q mutation – FV Leiden – affecting one of the APC cleavage sites on FV (Bertina et al, 1994). FV Leiden is the most common inherited thrombophilic defect in Caucasian populations. In Europe, 3–8% of the general population are carriers for the FV Leiden mutation (Lee, 2001) and data from the Dutch cohort revealed its presence in 22% of patients with a first VTE episode (de Visser et al, 1999). It is estimated that the risk of a first VTE in heterozygous FV Leiden patients is approximately three- to sevenfold and this is increased to 80-fold if the individual is homozygous (Price & Ridker, 1997). The observation that VTE events in carriers of the FV Leiden mutation are often precipitated by use of the oral contraceptive pill points to a synergistic link with other metabolic changes. For women who are heterozygous for FV Leiden, the risk is probably 28 to 50 of 10 000 women-years compared to 2–5 of 10 000 years for those not known to be carriers (Waselenko et al, 1998). Women using combination oral contraceptives have higher total serum triglyceride and cholesterol concentrations (Powell et al, 1984). As elevated triglycerides have been causally related to increased thrombin generation (Moyer et al, 1998), this is likely to be further enhanced in the presence of a prothrombotic mutation. The steroid type and dose of the pill can lower HDL levels, which is also relevant as HDL can enhance the anticoagulant activity of APC (Griffin et al, 1999). Indeed, HDL deficiency has been associated with VTE in men aged less than 55 years (Deguchi et al, 2005a). Other contributing mechanisms include the induction of an APC resistant phenotype by the contraceptive pill (Rosing et al, 1999). Factor VIII levels are increased in users of the pill (Kadir et al, 1999) and a direct effect may occur because of reduced glucosylceramide synthesis, a glycolipid cofactor that enhances the activity of APC (Yegneswaran et al, 2003). Its deficiency in plasma has also been associated with an increased risk of VTE (Deguchi et al, 2005b).

Sepsis

Under resting physiological conditions, PC is continuously activated to maintain an anticoagulant endothelial surface. In severe sepsis, the host response leads to generalized dysfunction of the endothelium (Aird, 2003). The resting balance is shifted in favour of a prothrombotic state as tissue factor expression is increased in response to cytokine release (Drake et al, 1993). Acquired PC deficiency occurs as a result of mechanisms involving consumption (Esmon, 2005), increased degradation by proteolytic enzymes (Takano et al, 1990) and reduced hepatic synthesis (Dhainaut et al, 2001). At the cell surface, PC activation is compromised by a reduction in TM and EPCR expression levels. This is due to endotoxin and cytokine-mediated downregulation as well as enzymatic cleavage to release the soluble receptors (Moore et al, 1987; Conway & Rosenberg, 1988; Lentz et al, 1991). In this situation, the procoagulant and pro-inflammatory effects of thrombin predominate, resulting in microvascular thrombosis and vascular leakage (Taylor & Kinasewitz, 2004).

Experimental models of sepsis have helped delineate the relevance of the thrombin–APC axis. Firstly, a recently developed mouse model of the pro-thrombotic FV Leiden mutation has shown that heterozygote animals are protected against the lethal effects of sepsis, whilst homozygotes have an adverse outcome. This is borne out in humans, with a survival advantage in sepsis for patients who are heterozygous for the FV Leiden mutation (Kerlin et al, 2003). The implication is that enhanced, but not excessive, thrombin generation will optimize APC generation, which provides protection against the effects of sepsis. This is consistent with the low dose thrombin infusion experiments described above. Indeed, thrombin generation in mice models leads to an increase in PROCR mRNA levels (Gu et al, 2000) and EPCR overexpression will increase APC levels by eightfold in response to thrombin infusion (Li et al, 2005).

In baboon studies, infusion of an antibody that blocked PC activation in vivo during Escherichia coli-mediated endotoxaemia, led to a more severe response. The response to sublethal concentrations of E. coli was made lethal but this was prevented by infusion of exogenous PC (Taylor et al, 1987). Subsequent studies have since delineated the involvement of EPCR in this process. In further baboon studies, blocking EPCR could convert a sublethal septic challenge into lethality (Taylor et al, 2000). Histological changes from the EPCR-blocked animals showed increased microvascular thrombosis, widespread necrotic foci and white cell infiltration within the adrenals, kidneys and liver. In separate experiments, heart muscle damage was more severe in EPCR-deficient mice as compared to wild-type after endotoxin administration (Iwaki et al, 2005).

The interaction of EPCR–APC with PAR-1 appears to be relevant in vivo. Intravenous injection of APC into mice leads to PAR-1-dependent gene induction (Guo et al, 2004). In cytokine-stimulated endothelial cells, gene expression profiling also demonstrated marked differences between APC and thrombin signalling. APC downregulated transcripts for pro-apoptotic proteins including p53 and thrombospondin-1, but the latter was upregulated by thrombin (Riewald & Ruf, 2005). However, there is debate over how the APC effect might co-exist in vivo when thrombin generation is in excess. In vitro studies would suggest that APC is 103- to 104-fold less potent than thrombin in cleaving PAR-1 (Ludeman et al, 2005). The physiological relevance of the EPCR–APC–PAR-1 axis in sepsis may be related to topographical dynamics, both at the vascular site-specific level and at the level of the single cell. Firstly, the local in vivo environment of the vascular bed is difficult to reproduce experimentally. Although thrombin concentrations are initially much greater at sites of injury, removal by circulating inhibitors and blood flow gradients may contribute to a situation where APC activity prevails to rescue cells from the progressive effects of sepsis (Ruf, 2005). Secondly, EPCR and PAR-1 are not evenly distributed throughout the vasculature (Yin et al, 2003). EPCR expression is mainly in large diameter blood vessels (Laszik et al, 1997) and the APC–PAR-1 effect may be directed to these areas. In certain cell types, there may also be biophysical factors to consider. In endothelial cells, the critical receptors required for both PC activation and APC cellular signalling, i.e. TM, EPCR and PAR-1, are co-localized in lipid rafts (Bae et al, 2007). This implies that preferential association of thrombin with TM, within a PAR-1-containing raft, will facilitate EPCR–APC–PAR-1-dependent protective events. As with the ability of cell membranes to flip-over and expose negatively charged phospholipid surfaces to promote and localize coagulation reactions (Bevers et al, 1991), a mechanism that could assemble and re-route PAR-1 from pro to anti-inflammatory effects could be highly relevant in maintaining cellular homeostasis during periods of stress.

Clinical mechanisms and translational aspects of thrombin and APC generation

Thromboembolism

Neutralizing thrombin activity and reducing its generation form the cornerstones of modern thromboembolic disease management (Toh, 1997). This is in regard to both acute treatment and the prevention of recurrence. In terms of diagnosis, low thrombin activity as manifested through low D-dimer levels has been extremely useful in excluding VTE (Schutgens et al, 2003). From a positive prediction perspective, there is encouraging evidence that persistent increases in biomarkers of thrombin activity (thrombin-antithrombin III, D-dimer) may be of clinical value in identifying those patients at most risk of recurrent events (Palareti et al, 2003). Of interest is that sEPCR levels fall during warfarin administration, suggesting that those with exceedingly high sEPCR levels may have increased endogenous thrombin production (Stearns-Kurosawa et al, 2002). This prompts the suggestion that sEPCR levels might be another marker for indicating a hypercoagulable state. Higher levels of sEPCR have been correlated with the A3 EPCR haplotype and proposed as a candidate risk factor for thrombosis (Saposnik et al, 2004). Conversely, the 4678CC genotype (A1A1 homozygotes) has been associated with elevated APC levels and a reduced risk of VTE (Medina et al, 2004).

The mainstay of managing these conditions is in reducing thrombin procoagulant activity through the use of heparin and oral vitamin K antagonists, which are well reviewed elsewhere (Buller et al, 2005). However, the vast majority of patients who receive antithrombotic treatment are not anticoagulated to full efficacy due to haemostatic considerations. Despite such treatment, thrombotic occlusions resulting in myocardial infarction and ischaemic stroke continue to contribute to mortality statistics (Anderson & Smith, 2005). Future strategies may involve developing more thrombus-specific anticoagulants; e.g. by exploiting the allosteric nature of thrombin to create mutants that serve as PC activators at the endothelial lining (Cantwell & Di Cera, 2000). The epitope for thrombomodulin binding overlaps with that of fibrinogen (Esmon, 1995) and molecular engineering has created thrombin mutants that are defective in cleaving fibrinogen, but retain TM-dependent PC activation. A double mutant thrombin (Trp215Ala/GLu217Ala) has been shown to be a safe and potent anticoagulant without significantly impairing haemostasis in a primate model (Gruber et al, 2007).

Atherothrombosis sees a greater interplay between the processes of inflammation with coagulation as compared to VTE (Ridker, 2004). This is exemplified by the observation that the presence of prothrombotic mutations, i.e. the FV Leiden and prothrombin gene (F2) mutation G20210A only become significant risk factors in arterial disease when combined with metabolic risk factors (de Moerloose & Boehlen, 2007). It is perhaps in this context that therapeutic strategies involving the PC pathway, with its links between coagulation and inflammation, might have greater potential. At the forefront of potential translational relevance is the setting of thrombotic stroke. In experimental ischaemic stroke, APC administration restored cerebral blood flow and reduced the brain infarct volume. These effects of APC, likely to be related to improved maintenance of the blood–brain barrier to neutrophils, were not associated with intracerebral bleeding (Shibata et al, 2001). The mechanisms appear to be dependent on the interaction of APC with EPCR and PAR-1 to generate anti-inflammatory and cytoprotective signals (Cheng et al, 2003). Furthermore, APC in combination with tPA treatment in an experimental thrombotic stroke model reduced the neuronal toxicity of tPA alone without increasing bleeding (Liu et al, 2004; Cheng et al, 2006).

Sepsis

Acquired PC deficiency is a well-recognized consequence of sepsis and studies have demonstrated its value as a prognostic marker (Yan et al, 2001). There is a negative correlation between PC levels in septic patients with increasing morbidity and mortality (Shorr et al, 2006). In biopsies taken from patients with meningococcal septicaemia, there is downregulation in the expression of both TM and EPCR with evidence of dysfunctional PC activation (Faust et al, 2001). Therapeutically, administration of a recombinant form of human APC has been shown to significantly reduce the relative risk of death in patients with severe sepsis. A randomized, double blind, placebo controlled, multicentre trial [The Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis (PROWESS) trial] demonstrated a 19·4% reduction in the relative risk of death from any cause at 28 d for patients with a clinical diagnosis of severe sepsis receiving recombinant APC (Bernard et al, 2001). Patients who demonstrated most benefit from the drug were those with the highest Acute Physiology and Chronic Health Evaluation II (APACHE II) scores who had extensive organ dysfunction (Ely et al, 2003). This included patients with disseminated intravascular coagulation (DIC) although there was an increased trend for bleeding in this group (Dhainaut et al, 2004). Even in the absence of DIC, severe bleeding complications including intracranial haemorrhage have been associated with APC treatment. Whilst the Extended Evaluation of Recombinant Human Activated Protein C (ENHANCE) trial provided supportive evidence for the favourable benefit/risk ratio observed in the PROWESS trial, there were higher serious bleeding rates (Vincent et al, 2005).

To improve on the benefit-risk ratio of APC treatment in sepsis, knowledge of the mechanisms involved in vivo would be highly relevant. Neutralizing thrombin is likely to be an important aspect, but failures of treatment with high-dose antithrombin (Warren et al, 2001) or tissue factor pathway inhibitor (Abraham et al, 2003) suggest that the success of APC is due to properties other than anticoagulation. With a focus on the APC anti-inflammatory role involving PAR-1, mutants have now been created that are less anti-coagulant but retain anti-inflammatory effects (Mosnier et al, 2007). The clinical relevance of these molecules awaits further investigation.

Other clinical areas

Whilst pharmaceutical developments continue to refine established pro-haemostatic or anticoagulant approaches, there may be new disease frontiers where modulation of the thrombin–APC axis could prove beneficial. In gastroenterological disorders, gastric inflammation has been related to a reduction in PC activation (Oka et al, 2000) and similarly, in inflammatory bowel disease (Faioni et al, 2004). In both Crohn’s disease and ulcerative colitis, inflamed mucosal microvasculature exhibits a loss of endothelial EPCR and TM expression. Administration of recombinant APC onto isolated human intestinal endothelial cells has demonstrated potent anti-inflammatory effects. These include reducing cytokine-dependent cell adhesion molecule expression, chemokine production and leucocyte adhesion. The results have been consistent in vivo supporting the extended role of the PC pathway in rescuing inflamed intestinal microvasculature (Scaldaferri et al, 2007).

Activated protein C also appears to have relevance in wound healing. EPCR expressed by human keratinocytes facilitates APC–PAR-1-mediated cell proliferation and activation of ERK as well as p38 kinase signalling pathways (Xue et al, 2005). More recently, the anti-inflammatory and immune modulatory effects of APC have been linked to rheumatoid arthritis – an autoimmune condition characterized by synovial joint inflammation. Likely mechanisms in this setting are thought to relate to the inhibition of NFκB activation and production of TNF-α from monocytes. APC inhibits the migration and activation of monocytes from patients with rheumatoid arthritis, and could conceivably reduce synovial colonization (Xue et al, 2007). The relevance of the PC pathway in other auto-immune states remains to be explored. In the context of recurrent miscarriage, autoantibodies targeting EPCR have been found at higher levels (Hurtado et al, 2004). These have also been associated with acute myocardial infarction in young women (Montes et al, 2005). Whether these findings lead to strategies that promote the PC activation pathway in such conditions remains speculative at the present time but are plausible considerations for the future.

Conclusion

The taijitu symbol that we associate with yin and yang forces reflects the relative and inseparable relationship between thrombin and APC. Whilst each can be regarded as its opposite when viewed from the alternative perspective, there are general similarities in both. The essence is in their working together in achieving an overall balance. In this context, the thrombin–APC interactions at the cell surface in regulating clot formation are also similar for inflammatory modulation at the level of PAR-1 (Fig 1). For a phylogenetically ancient system, such as the haemostatic system, to have evolved to this state of refinement, it must have developed from a fundamental system of checks and balances. Although the tendency over the past 50 years has been on the discovery of factors involved in haemostatic control and has concentrated on distinctive, individual function, there is now a shift towards understanding the cohesion in which these factors cooperate and re-equilibrate responses. This article has provided examples of how thrombin and APC interrelate at the molecular, cellular, physiological and clinical levels. This axis appears to have permeated into areas not conventionally linked to haemostatic dysfunction and may well underscore the recurrent molecular themes in the pathogenesis of different diseases. The challenge is in whether new knowledge about thrombin and APC will unveil novel approaches that can translate into improved management of current and future public health priorities.

Figure 1.

 The homeostatic balance between thrombin and activated protein C in coagulation and inflammation: activated thrombin (IIa) promotes the generation of activated protein C (aPC) and the two molecules influence the extent of both fibrin (clot) formation and the inflammatory response. The calibre of fibrin formation is mediated via the activation of thrombin activatable fibrinolysis inhibitor (TAFIa) and the inflammatory response occurs through differential signalling of endothelial protease activated receptor-1 (PAR-1).

Ancillary