Protein disulfide-isomerase – a trigger of tissue factor-dependent thrombosis

1 The tissue factor pathway of coagulation

Tissue factor (TF, thromboplastin), the membrane receptor for coagulation factor VII (FVII/FVIIa), is well-recognized as the principal initiator of the extrinsic coagulation pathway, triggering blood clotting [5, 26, 35]. This type 2 cytokine receptor [21] forms a high affinity complex with plasmatic FVIIa and Ca^2 + [5], the membrane localized serine protease that converts coagulation factor X (FX) to FXa in the initial step of the extrinsic coagulation pathway [47]. Subsequently, FXa binds to activated FV to form the prothrombinase complex, catalysing the zymogen activation of prothrombin and resulting in quantitative formation of thrombin. Thrombin is the central coagulation protease that cleaves fibrinogen to fibrin, the final step of clot formation [16].

The prevailing view is that TF of blood vessel surrounding fibroblasts forms the so-called hemostatic envelope, ensuring hemostasis if vascular integrity is disrupted [12]. Under unperturbed conditions, TF-dependent coagulation initiation is tightly controlled by tissue factor pathway inhibitor (TFPI), a Kunitz type inhibitor binding coagulation factor Xa with its Kunitz domain 2 and associating in a high affinity complex with the binary TF-VIIa complex [7]. However, this control can be disturbed under infectious or inflammatory conditions [32]. The paradigm of the envelope theory has been challenged in 1999, when it was demonstrated that TF can initiate intravascular thrombosis arising from blood-borne TF sources, e.g. lipopolysaccharide-stimulated monocytes and procoagulant microvesicles(microparticles, MV) derived from activated and apoptotic cells [14, 36]. It has long been recognized that disrupted platelets can enhance the clot-promoting properties of leukocytes [36], but the exact TF-dependent mechanisms remained elusive. Beyond its recognized role in thrombosis, TF has crucial roles for embryonic blood vessel development [8], triggers postnatal vascular remodeling [44], and promotes tumor angiogenesis through activation of protease-activated receptors [6]. A detailed understanding of the regulatory mechanisms involved in TF-dependent coagulation initiation will enable development of novel targeted intervention strategies to combat the burden of vascular thrombosis.

2 The cellular sources of cryptic, blood-borne tissue factor

In the intact vasculature, plasmatic clotting factors cannot become activated by fibroblastic TF constituting the hemostatic envelope [12], but activated monocytes and monocyte-dereived microvesicles (MV, microparticles) represent relevant sources of TF inside the circulation [38]. TF is present in an encrypted, functional inactive state [45, 48]. In a laser injury mouse thrombosis model, the hematopoietic cell–derived MV have been demonstrated as the major source of TF accumulating in the developing thrombus during the initial phase of thrombus formation [20]. Moreover, elevated levels of monocyte TF expression have been linked to cardiovascular disease [29] and increased counts of procoagulant MV have been associated with acute coronary syndromes [51]. Whether platelets, under certain conditions, can express procoagulant TF [33] or whether TF procoagulant activity results from the transfer of TF-bearing microparticles onto platelets [42] was controversial [14, 39]. Similarly, and in contrast to monocytes, neutrophils were suggested not to express TF, but to acquire their procoagulant activity from other sources [37].

Although TF is expressed and entirely surface exposed by activated myeloid-monocytic cells [13], it is not fully active [3], a phenomenon termed TF encryption [19]. TF encryption is independent of TF de-novo synthesis and thus refers to the post-translational suppression of TF procoagulant activity. It was repeatedly shown that TF becomes fully activated when cells are lysed [19]. TFde-encryption was initially associated with apoptosis and described to depend on calmodulin [19]. It could be demonstrated, that caspase-1 activity is mandatory for a calpain-independent generation of thromboinflammatory TF-bearing MV [46]. This process is regulated by protein disulfide-isomerase (PDI) [17].

Importantly, during the past decade, PDIs have been implicated in the activation of cryptic TF on myeloid-monocytic MV in arterial [45] and venous thrombus formation [54] in vivo. In vitro experiments have uncovered that human TF contains an evolutionary conserved, redox-labile disulfide at Cys186/Cys209 [43, 52] that is prone to reduction and represents a molecular switch regulating TF procoagulant function [23] (Fig. 1). In biochemical analyses, the standard redox potential of this scissile – RH staple Cys186/Cys209 disulfide bond was determined by titration with dithiothreitol to –278 mV in the human recombinant TF extracellular domain and the sulfurs of these half-cystines were in the close proximity of 3 to 6 Å distance [30].

3 Mechanistic insights into tissue factor de-encryption

In addition to the inhibition of the binary activator complex (TF/FVIIa) by TFPI bound to coagulation factor Xa (FXa/TFPI), TF is present in an encrypted form in the plasma membrane of myeloid cells and in procoagulant MV in order to tightly control intravascular coagulation activation [45]. TF encryption/de-encryption has been investigated intensively and the involvement of several mechanisms has been proposed.

It has previously been suggested that TF dimers (changes in quarternary structure) might be responsible for the cryptic TF activity of myeloid-monocytic cells [3]. In cross-linking studies the production of homodimeric TF has been observed, which could be prevented by treatment with the calcium ionophore ionomycin [2]. Hence, calcium influx into the cytosol and calcium-mediated changes in the asymmetric distribution of phosphatidylserine in the plasma membrane were suggested to be mechanistically linked to TF de-encryption [4]. Inhibitors of calmodulin blocked the ionophore-dependent increase in TF procoagulant activity. However, the calcium ionophore-induced increase in TF procoagulant activity was not reduced by blockade of solvent exposed free thiol groups, excluding a mechanism, acting on redox regulation of TF [17].

In silico analysis of the crystal structure of TF by Hogg and co-workers identified that this receptor molecule contains a labile allosteric disulfide bond (–RH staple conformation) at Cys186/Cys209 that is prone to reduction into the free thiol state [49] (Fig. 1; referring to DOI: 10.2210/pdb1boy/PDB). Indeed, site-directed mutagenesis studies have clarified that the critical disulfide pair required for coagulation initiation is not the intact TF Cys49/Cys57 disulfide, but the solvent exposed Cys186/Cys209 (corresponding to Cys190/Cys213 in mouse), which is necessary for the binding of the FVIIa Gla domain to TF [43, 52].

In cell culture experiments, it was demonstrated that reduction of the Cys186/Cys209 TF disulfide (Fig. 1) can switch the procoagulant function of TF into a cryptic non-coagulant function [1]. Interestingly, the signaling function of TF was preserved in the C209A mutant that is capable of mediating TF/VIIa dependent PAR2 signaling, but was lost in the C186A adenoviral construct, showing that this allosteric disulfide also has a role in controlling TF signaling specificity via protease-activated receptor-2 (PAR2) [1]. In this study, extracellular PDI was found associated with TF, depending on the Ca^2 + concentration, and it was identified to regulate the redox state of the Cys186/Cys209 TF disulfide pair [1].

TF procoagulant function has been uncovered to depend on post-translational modifications of the redox-labile Cys186/Cys209 disulfide pair (Fig. 2). Cell lysis, leading to TF de-encryption, resulted in an increase of the oxidized form of the vicinal TF Cys186/Cys209 disulfide pair [45]. Also treatment of monocytes with the supernatant of activated platelets leads to an increase in the coagulant oxidized TF form. When studying TF de-encryption, the cell system was shown to be critical because in contrast to HL60 myeloid leukemia cells, monocytic U937 cells, or THP1 cells, the breast cancer cell line MDA-MB231 did not show thiol-dependent TF de-encryption [31]. Treatment of cells with thiol blocking agents, e.g. 5,5’-Dithiobis-2-nitrobenzoate (DTNB) or 3-(N-Meleimido-propionyl)biocytin (MPB) diminished cellular TF procoagulant activity, while oxidation of the Cys186/Cys209 dithiols with HgCl₂ or cross-linking of the vicinal thiols with phenylarsine oxide (PAO) increased cell-based TF-dependent procoagulant activity [1, 53]. In vitro binding studies uncovered that reduced TF retained the capacity to bind FVIIa, but lost its cofactor function [27].

In several cell-based studies, including HUVECs, CHO cells, LPS-stimulated monocytes and procoagulant MV, it was demonstrated that PDI increases TF procoagulant activity as measured by the FXa generation assay [45, 53]. Conversely, treatment of MV-platelet suspensions with anti-PDI antibody or inhibition of PDI with bacitracin effectively reduced TF procoagulant activity [45]. Antibody-mediated complement activation was uncovered as an additional trigger, switching monocyte TF via C5-dependent PDI mediated cell surface disulfide exchange to its procoagulant state [28].

The prothrombotic effect of PDI was also demonstrated by intravital imaging of thrombus formation in ligated carotid arteries by antibody blocking of PDI or infusion of recombinant PDI [45]. In cremaster arterioles of mice, the anti-thrombotic effect of the PDI inhibitor bacitracin and the anti-PDI antibody clone RL90 was shown in the laser injury model in vivo [11]. In this study, the presence of PDI was visualized in nascent thrombi. Moreover, PDI was also implicated in deep vein thrombosis [54]. In addition to the prototypic PDI (P4HB), specifically ERp5 from platelets and endothelial cells has been described as a PDI family member that supports thrombus formation [10, 40]. In histologic analyses, PDI was localized to the wounded area of the carotid vessel wall and platelet depletion reduced luminal PDI exposure [45]. However, in a subsequent in vivo study, it was demonstrated that endothelium-derived PDI rather than platelet-derived PDI supports thrombus formation in the laser injury model of mouse cremaster arterioles [24].

In the cell-based studies, it was suggested that the dithiol pair at Cys186/Cys209 of TF is kept in the encrypted state since it can be stabilized by protein s-glutathionylation, as determined by mass spectrometry analysis at Cys209 or with an antibody directed against glutathione [45, 53]. It was proposed that reduced PDI can react with the cysteine-bound glutathione and thus the dithiol can switch into the oxidized Cys186/Cys209 dithiol and TF is procoagulant (Fig. 2). Furthermore, it was proposed that the Cys186/Cys209 dithiol of TF could be S-nitrosylated and thus, the nitrosylase and transnitrosylase function of PDI [41, 50] could render TF procoagulant [1, 52]. NO reacts with reduced glutathione (GSH) to nitrosylated glutathione (GSNO), which was sufficient to suppress TF procoagulant activity [1]. It is still unclear and more detailed biochemical analyses are required to resolve whether nitrosylated glutathione results in protein s-nitrosylation, protein s-glutathionylation, or both (Fig. 2). Also the exact role of PDI on TF procoagulant function as an NO-donor/acceptor, either adding or abstracting NO from TF, needs to be resolved by future research. Furthermore,in biochemical settings, thioredoxin has been demonstrated to reduce the Cys186/Cys209 allosteric disulfide bond, both by treatment of the recombinant extracellular TF domain and TF expressed by cells [17, 45]. The exact mechanism how the encrypted TF is kept in its inactive conformation needs to be resolved by future investigations.

4 PDI enzymes as antithrombotic targets

In recent years, PDIs were recognized as an antithrombotic target [15]. By intravital imaging of thrombus formation it was demonstrated that quercetin-3-rutinoside (rutin), a flavonol abundant in many plants, blocks thrombus formation by selectively inhibiting PDI [25]. It is found in high concentrations in tea, fruits, berries, and buckwheat. Interestingly, it was shown to inhibit both, platelet aggregation and endothelial cell mediated fibrin generation. Animal studies and clinical trials using dietary flavonols proved the non-toxicity of quercetin-3-rutinoside at high concentrations, suggesting that inhibiting PDI is a safe strategy to prevent thrombosis [22].

5 Outlook

In the past decade, PDI was uncovered as an activator of cryptic TF procoagulant activity in response to injury, but the posttranslational mechanisms keeping TF in a functionally silent conformation remain to be elucidated. In spite of the recent advances in the mechanistic understanding of TF de-encryption, the functional role of PDI oxidoreductases for coagulation activation in ischemic and hypoxic tissues is still unresolved. Future studies should therefore investigate the role of PDI enzymes in ischemic disease states (e.g. in stroke or myocardial infarction). Also, the role of complement pathways in PDI-triggered TF activation awaits further investigation. While it is increasingly clear that PDIs are relevant triggers of TF-dependent thrombosis, the tissue specific expression profile of PDIs and the specific roles of various PDI enzymes in TF de-encryption remain poorly resolved. Since PDIs are interesting targets for the development of novel antithrombotic therapies, clinical studies need to explore, whether selective PDI inhibitors could be beneficial in the treatment of cardiovascular disease states.

Conflicts of interest

The authors declare that no conflicts of interest exist.