Chapter 113: Molecular Biology and Biochemistry of the Coagulation Factors and Pathways of Hemostasis

THE VITAMIN K–DEPENDENT ZYMOGENS: PROTHROMBIN, FACTOR VII, FACTOR IX, FACTOR X, AND PROTEIN C

The vitamin K-dependent zymogens circulate in an inactive state and require proteolytic activation to function as a serine protease. All share a similar domain structure of a C-terminal serine protease domain and an N-terminal γ-carboxy glutamic acid (Gla) domain, which are connected by two epidermal growth factor (EGF)-like domains or kringle domains (Fig. 113–1). Each protein domain has a well-defined function and facilitates substrate recognition, interaction with protein cofactors, or binding to a negatively charged lipid surface, such as that of activated platelets or endothelial cells, thereby restricting coagulation to the site of injury. The latter is mediated via the Gla domain, a domain that is characteristic to the vitamin K–dependent proteins.

The high level of protein and gene homology suggests that the vitamin K–dependent zymogens originate from a common ancestral gene as a result of gene duplications.¹ Exon shuffling and tandem duplication may account for the generation of the ancestral gene, in which the functional domains that are encoded by a single exon each were combined and duplicated.² This process may also account for the presence of the kringle domains as opposed to EGF-like domains in prothrombin.

The Gla domain refers to the 42-residue region located in the N-terminus of the mature protein that comprises 9 to 12 glutamic acid residues that are posttranslationally γ-carboxylated into Gla residues by a specific γ-glutamyl carboxylase in the endoplasmatic reticulum of hepatocytes.³ This γ-carboxylase requires oxygen, carbon dioxide, and the reduced form of vitamin K for its action, hence the name vitamin K–dependent proteins. For each Glu residue that is carboxylated, one molecule of reduced vitamin K is converted to the epoxide form (Fig. 113–2). Vitamin K epoxide reductase converts the epoxide form of vitamin K back to the reduced form.⁴ Warfarin and related 4-hydroxycoumarin–containing molecules inhibit the activity of vitamin K epoxide reductase, thereby preventing vitamin K recycling and inhibiting γ-carboxylation. This results in a heterogeneous population of circulating undercarboxylated forms of the vitamin K–dependent proteins with reduced activity. Because warfarin blocks the reductase and not the carboxylase, the inhibitory effect of warfarin can be (temporarily) reversed by administration of vitamin K. Recognition by and interaction with γ-carboxylase is facilitated by the propeptide sequence that is located C-terminal to the signal peptide. The propeptide is highly conserved among the vitamin K–dependent proteins, and amino acids at positions –18, –17, –16, –15, and –10 are critical to recognition by the γ-carboxylase.^5,6 Following γ-carboxylation, the propeptide is removed through limited proteolysis prior to secretion of the mature protein.

A correctly γ-carboxylated Gla domain is essential for interaction of the vitamin K–dependent proteins with phosphatidylserine, a negatively charged phospholipid. Under normal conditions, phosphatidylserine is not exposed on the outer membrane leaflet of cells. However, in activated endothelial cells or platelets, phosphatidylserine is part of the extracellular cell surface where it supports blood coagulation reactions. The Gla domain interacts with the anionic cell surface in a calcium-dependent manner. These calcium ions are coordinated by Gla residues and induce a conformational change in the Gla domain that is characterized by the appearance of a hydrophobic surface loop (Fig. 113–3). Membrane binding by the Gla domain occurs when this hydrophobic surface loop penetrates into the hydrophobic portion of the phospholipid bilayer, which is facilitated by the interaction of the Gla-bound calcium ions with the phosphate head groups of phosphatidylserine.^7,8 It has been shown that the phosphate head groups of exposed phosphatidylethanolamine are also capable of coordinating calcium ions, thereby contributing to the interaction of the Gla domains with the negatively charged membrane surface.⁹

The serine protease domains of the vitamin K–dependent proteins are highly homologous, as they bear a chymotrypsin-like fold and display trypsin-like activity.¹⁰ Once activated, they cleave peptide bonds following a positively charged amino acid (Lys or Arg). Activation proceeds through proteolysis at one or more sites N-terminal to the serine protease domain (see Fig. 113–1). Subsequently, the newly formed N-terminus inserts into the serine protease domain to form a salt bridge with an Asp residue, which is associated with conformational changes in the serine protease domain. These lead to an optimal configuration of the active site through alignment of the active site residues His, Ser, and Asp, and to formation of the substrate-binding exosites, allowing for substrate conversion. The substrate-binding exosites are unique to each vitamin K–dependent protease and are responsible for their highly specific substrate recognition and associated function in coagulation.

Interaction of the vitamin K–dependent proteases with specific cofactors on a (anionic) membrane surface (Table 113–2) further enhances substrate recognition, as the cofactors interact with both the protease and the substrate, bridging the two together. This results in a dramatic enhancement of the catalytic activity (Table 113–3), thereby making the cofactor–protease complex the physiologic relevant enzyme. The increase in catalytic rate has also been attributed to a cofactor-induced conformational change in the protease.¹¹ However, whether this molecular mechanism holds true for all cofactor–protease complexes remains to be determined. Tissue factor is the cofactor for factor VIIa, factor VIIIa is the cofactor for factor IXa, and factor Va is the cofactor for factor Xa, while thrombin does not require a cofactor for its procoagulant activity. However, upon association with the cofactor thrombomodulin, thrombin’s specificity is changed from procoagulant to anticoagulant (cleaving and activating protein C). The complexes are also named for their physiologic substrate: the factor VIIIa–factor IXa complex is termed the “tenase” or “intrinsic tenase” complex; the tissue factor–factor VIIa complex is termed the “extrinsic tenase” complex; and the factor Va–factor Xa complex is termed the “prothrombinase” complex.

PROTHROMBIN (FACTOR II)

Prothrombin, or factor II, which was discovered by Pekelharing in 1894, is one of the four coagulation factors that were described by Paul Morawitz in 1905, in addition to fibrinogen (factor I), thromboplastin (thrombokinase, factor III, now tissue factor), and calcium (factor IV).^12,13 The zymogen prothrombin is primarily synthesized in the liver and circulates in plasma as a single-chain protein of 579 amino acids (Mr ≈72,000) at a concentration of 1.4 μM with a plasma half-life of 60 hours (see Table 113–1).

Protein Structure

Prothrombin is composed of fragment 1 (F1), fragment 2 (F2), and the serine protease domain. F1 consists of the Gla domain, which comprises 10 Gla residues, and the kringle 1 domain; F2 contains the kringle 2 domain (see Fig. 113–1). The two kringle domains, which replace the EGF-like domains present in most vitamin K–dependent zymogens, are conserved secondary protein structures that fold into large loops that are stabilized by three disulfide bonds and schematically resemble a Danish pastry called a “kringle.” Their primary function is to bind other proteins such as the cofactor Va and serine protease factor Xa that activate prothrombin.

Other than γ-carboxylation of Glu residues, prothrombin is posttranslationally modified via N-glycosylation in the kringle 1 (Asn78, Asn143) and serine protease domains (Asn373), which contributes to the stability of the prothrombin precursor during processing in the endoplasmatic reticulum.^14,15

Prothrombin Activation and Thrombin Activity

Prothrombin is proteolytically activated by the prothrombinase complex (i.e., factor Va, factor Xa, calcium, and anionic phospholipids) that cleaves at Arg271 and Arg320 (see Fig. 113–1). Cleavage at Arg320 opens the active site of the protease domain, while cleavage at Arg271 removes the activation fragment F1.2 (F1.2). Both cleavages are necessary to generate procoagulant α-thrombin (IIα) (Fig. 113–4). The composition of the membrane surface directs the cleavage order in prothrombin and the formation of either the zymogen prethrombin 2 (initial cleavage at Arg271) or the proteolytically active intermediate meizothrombin (initial cleavage at Arg320).^16,17 Meizothrombin has impaired procoagulant activity as compared to α-thrombin, but superior anticoagulant activity as it displays increased thrombomodulin-dependent protein C activation, which is likely facilitated by membrane binding of meizothrombin through its Gla domain.¹⁸ The snake venom protease Ecarin is capable of generating meizothrombin specifically through proteolysis at Arg323 only. However, this meizothrombin is instable as a result of autocatalysis at Arg155, thereby removing the Gla domain containing F1. The so formed meizo-des-F1 can be converted to thrombin by prothrombinase, but at a slower rate as it is incapable of membrane binding. Assessment of F1.2 levels reflects prothrombin activation and is commonly used as a marker for thrombin generation.

Thrombin (IIα) is a two-chain serine protease (Mr ≈37,000) comprising a light chain of 49 residues (A chain; Mr ≈6000) that is covalently linked to the catalytic heavy chain of 259 residues (B chain; Mr ≈31,000). Thrombin’s main function is to induce the formation of a fibrin clot by removing fibrinopeptides A and B from fibrinogen to form fibrin monomers, which then spontaneously polymerize. In addition, thrombin is able to cleave a wide variety of substrates with high specificity, which is mediated via its negatively charged, deep active site cleft and via the anion binding exosites I and II that specifically interact with cofactors and/or substrates.¹⁹ The dynamic structural conformation of thrombin allows for binding to diverse ligands, and the subsequent ligand-induced conformational stabilization, known as thrombin allostery, regulates and controls thrombin activity.^20,21

Thrombin initiates important procoagulant pathways by proteolytic activation of the cofactors V and VIII and zymogen factor XI that collectively amplify thrombin and fibrin formation, and by activating factor XIII that crosslinks and stabilizes the fibrin polymers. Another procoagulant function of thrombin is to inhibit fibrinolysis by proteolytic activation of the thrombin-activatable fibrinolysis inhibitor (TAFI), a reaction enhanced by the endothelial-bound cofactor thrombomodulin. Thrombin also has an anticoagulant function and upon binding to the cofactor thrombomodulin, it is capable of proteolytically activating protein C, which inactivates the cofactors Va and VIIIa.

Thrombin activates the seven-transmembrane domain, G-protein–coupled protease-activated receptors (PARs) PAR1, PAR3, and PAR4 that are expressed on a wide range of cell types in the vasculature by proteolytic cleavage of their N-terminal extracellular domains.^22,23,24,25 Thrombin is one of the strongest platelet activators in vivo and activates platelet-expressed PAR1 and PAR4.²⁵ The platelet glycoprotein (GP) Ib (GPIb) serves as a cofactor for thrombin in PAR1 cleavage (Chap. 112). Thrombin-mediated activation of endothelial-PAR1 triggers release of von Willebrand factor (VWF) and P-selectin, which promote rolling and adhesion of platelets and leukocytes. In addition, this stimulates the endothelial production of platelet-activating factor, a potent platelet and leukocyte activator, as well as the production of chemokines, cyclooxygenase (COX)-2, and prostaglandins.²⁵ Thrombin-mediated PAR activation is not only critical for coagulation, but also plays an important role in inflammatory and proliferative responses associated with vascular injury, such as in atherosclerosis and cancer.²⁶

The physiologic inhibitors of thrombin are the serine protease inhibitors (serpins) antithrombin, heparin cofactor II, protein C inhibitor, and protease nexin 1, with antithrombin being the primary plasma inhibitor. For all four serpins, the rate of thrombin inhibition can be accelerated by glycosaminoglycans, such as heparin (Table 113–4), through mutual binding to the serpin and thrombin (see Fig. 113–2), which ensures rapid inhibition of thrombin at the intact endothelial cell surface where heparin-like glycosaminoglycans are found.

Heparin and heparin-derivatives are clinically used as anticoagulants to inhibit thrombin via antithrombin. Hirudin, which originates from the salivary glands of medicinal leeches, and its recombinant and synthetic derivatives are potent and highly specific inhibitors that directly target the active site and exosite I of thrombin.²⁷ The target-specific oral anticoagulant dabigatran also inhibits thrombin directly with high specificity and reversibly binds the active site of thrombin.^27,28

Gene Structure and Variations

Prothrombin is encoded by a gene (F2) on chromosome 11p11.2 that is approximately 20 kb long.²⁹ The coding sequence is divided over 14 exons that range in size from 25 to 315 bp (Fig. 113–5). The reference sequence of prothrombin mRNA comprises 2018 bases. There are no common, well characterized, splicing variants with known biology.

Homozygosity or compound heterozygosity for loss of function mutations in the prothrombin gene leads to a bleeding tendency. This condition is quite rare with perhaps one case per 2,000,000 newborns.³⁰ Heterozygous carriers of loss-of-function mutations are without a bleeding phenotype. Mutations have been characterized in a relatively small number of cases with homozygous or compound heterozygous prothrombin deficiency (consult the human gene mutation database at http://www.hgmd.org for details). The majority of mutations underlying prothrombin deficiency are missense mutations, but several small deletions/insertions have also been reported.

Gain-of-function mutations in the prothrombin gene increase thrombotic risk. The best known variation is G20210A.³¹ This variation of the last nucleotide preceding the poly(A)-tail of the mature mRNA has an effect on 3′-end mRNA processing and increases the level of prothrombin in plasma by approximately 10 to 20 percent in heterozygous individuals.³² This relatively small increase in the level of prothrombin results in a two- to threefold enhanced risk for venous thrombosis. Homozygotes for the G20210A variation are quite rare, and the risk associated with homozygosity has not been measured with certainty. The G20210A variation is relatively common in whites, with a strong south-north gradient in that the variation is most common in southern Europe.³³

FACTOR VII

Factor VII, which was discovered around 1950,³⁴ is synthesized in the liver and circulates in plasma as a single-chain zymogen of 406 amino acids (Mr ≈50,000) at a concentration of 10 nM with a short plasma half-life of 3 to 6 hours (see Table 113–1).

Protein Structure

Factor VII consists of a Gla domain with 10 Gla residues, two EGF-like domains, a connecting region, and the serine protease domain (see Fig. 113–1). Calcium coordination in EGF-1 is mediated by partial hydroxylation of Asn63 and O-linked fucosylation of Ser60.³⁵ Further posttranslational modifications of factor VII consist of O-linked (Ser52 in EGF-1) and N-linked (Asn154 in the connecting region, Asn322 in the serine protease domain) glycosylation.

Factor VII Activation and Factor VIIa Activity

Factor VII is proteolytically activated once it has formed a high-affinity complex with its cofactor tissue factor. A number of coagulation proteases including thrombin and factors IXa and XIIa are capable of cleaving factor VII at Arg152 to generate factor VIIa (see Fig. 113–1), with factor Xa being considered the most potent and physiologically relevant activator of factor VII.³⁶ Autoactivation can also occur, which is initiated by minute amounts (approximately 0.1 nM) of preexisting factor VIIa.³⁷

Factor VIIa is a two-chain serine protease composed of a light chain (Mr ≈20,000) comprising the Gla and EGF domains and the catalytic heavy chain (Mr ≈30,000), which are covalently linked via a disulfide bond. Factor VIIa activity is only expressed when bound to tissue factor, which induces an active conformation of the factor VIIa serine protease domain (Fig. 113–6).¹¹ Factor VIIa interacts with tissue factor via its Gla and EGF domains.

The tissue factor–factor VIIa complex activates both coagulation factors IX and X, which is considered to be the main initiating step of the extrinsic pathway of coagulation. In addition, the tissue factor–factor VIIa (–factor Xa) complex is not only critical to processes in coagulation, but also to wound healing, angiogenesis, tissue remodeling, and inflammation through proteolytic activation of PAR2.^38,39,40

The ternary tissue factor–factor VIIa–factor Xa complex is inhibited by the tissue factor pathway inhibitor (TFPI). Tissue factor–factor VIIa is also inhibited by antithrombin, but only in the presence of heparin (see Table 113–4).

Gene Structure and Variations

The gene encoding factor VII (F7) is located on chromosome 13q34, is almost 15 kb in length, and comprises 9 exons (Fig. 113–7). The canonical mRNA encoding factor VII comprises 3000 bases.⁴¹ Alternatively spliced transcript variants encoding multiple isoforms have been observed, but their biology is not well characterized.⁴²

Inherited factor VII deficiency is a rare autosomal recessive disorder that affects approximately one in 500,000 newborns.³⁰ Factor VII deficiency is the most common of the inherited rare bleeding disorders, although the reported prevalences vary between countries. Homozygotes and compound heterozygotes develop a hemorrhagic diathesis that may vary from mild to severe.

The human gene mutation database (http://www.hgmd.org) lists 258 mutations in the factor VII gene. The majority of these are missense mutations, but splicing and regulatory mutations also occur. Small deletions account for almost 10 percent of the documented mutations. Other gross gene abnormalities appear to be uncommon.

The factor VII gene harbors many common polymorphisms of which three are notable: Arg353Gln in the catalytic domain, a 10-bp insertion in the promotor region, and a variable number of 37 bp repeats in intron 7.⁴³ The minor alleles of these polymorphisms are associated with decreased levels of factor VII and explain up to 30 percent of the variation in activated factor VII levels. Furthermore, the minor alleles have been claimed to lower the risk of myocardial infarction. However, this finding has not led to routine genotyping in the management of this disorder. The relationship between factor VII levels, factor VII polymorphisms, and venous thrombosis has not been established with certainty.

FACTOR IX

Factor IX was originally reported in 1952 as “Christmas factor,” named after one of the first identified hemophilia B patients.^34,44 Factor IX is synthesized in the liver and circulates in plasma as a single-chain zymogen of 415 amino acids (Mr ≈55,000) at a concentration of 90 nM with a half-life of 18 to 24 hours (see Table 113–1).

Protein Structure

Factor IX consists of a Gla domain, two EGF-like domains, a 35-residue activation peptide, and the serine protease domain (see Fig. 113–1). The Gla domain contains 12 Gla residues, of which the 11th and 12th Gla (Glu36 and Glu40) are not evolutionary conserved in other vitamin K–dependent proteins and are not essential for normal factor IX function.⁴⁵

Factor IX comprises several posttranslational modifications that are not only important for its structure and function, but are also involved in the plasma clearance and distribution of factor IX.³⁵ Factor IX is sulfated at Tyr155 and phosphorylated at Ser158 in the activation peptide. Hydroxylation of Asp64 in EGF 1 mediates calcium binding, and while only approximately 40 percent of total plasma factor IX carries this modification, complete absence because of a point mutation at this position dramatically reduces factor IX activity resulting in hemophilia B.^46,47 An O-linked fucose (Ser61) and glucose (Ser63) are found in the EGF 1 domain, in addition to several O-linked glycans in the activation peptide (Thr159, Thr169, Thr172, and Thr179). Further modification of the activation peptide includes N-linked glycosylation of Asn residues 157 and 167, which modulates the circulating levels of factor IX.^48,49,50

Factor IX binds with high affinity to the extracellular matrix component collagen IV via residue Lys5 in the Gla domain.^51,52 Although factor IX variants incapable of collagen IV binding exhibit a greater recovery, collagen IV association generates an extravascular reservoir of factor IX that enables prolonged action of factor IX at a hemostatic relevant region.

Factor IX Activation and Factor IXa Activity

Limited proteolysis of factor IX at both Arg145 and Arg180 by either the tissue factor–factor VIIa complex or factor XIa results in the release of the activation peptide and generation of factor IXa (see Fig. 113–1). Cleavage at Arg180 generates factor IXaα, which displays catalytic activity toward synthetic substrates only, whereas fully active factor IXaβ is formed following cleavage at Arg145.^53,54

Factor IXa is a two-chain serine protease (Mr ≈45,000) that is composed of a light chain of 145 residues (Mr ≈17,000) and the catalytic heavy chain of 235 residues (Mr ≈28,000), which are covalently linked via a disulfide bond.

Factor IXa has a low catalytic efficiency as a result of impaired access of substrates to the active site that results from steric and electrostatic repulsion.⁵⁵ Reversible interaction with the cofactor VIIIa on anionic membranes and subsequent factor X binding leads to rearrangement of the regions surrounding the active site and proteolytic factor X activation.

The primary plasma inhibitor of factor IXa is the serpin anti-thrombin, and this inhibition is enhanced by heparin (see Table 113–4), which induces a conformational change in antithrombin that is required for simultaneous active site and exosite interactions with factor IXa.⁵⁶

Gene Structure and Variations

The gene encoding factor IX (F9) is located on chromosome Xq27.1 and covers nearly 25 kb.⁵⁷ It is divided into eight exons from which a mature mRNA molecule is transcribed with an ultimate length of 2802 bases (Fig. 113–8).

A defect or deficiency in factor IX leads to hemophilia B. Chapter 123 discusses the prevalence, clinical characteristics, and molecular genetics of hemophilia B in detail.

Conversely, increased levels of factor IX are a strong risk factor for venous thrombosis.⁵⁸ This is in agreement with a rare gain of function mutation (Arg335Leu; factor IX Padua), which renders the protein hyperfunctional and is associated with familial early-onset thrombophilia.⁵⁹

FACTOR X

Factor X was originally reported in the late 1950s as the “Stuart-Prower factor,” named after the first two identified factor X–deficient patients.^60,61,62 Factor X is primarily synthesized in the liver and circulates in plasma as a two-chain zymogen of 445 amino acids (Mr ≈59,000) at a concentration of 170 nM with a half-life of 34 to 40 hours (see Table 113–1).

Protein Structure

Factor X is synthesized as a single-chain precursor and during intracellular processing, the three-amino acid peptide Arg140-Lys141-Arg142 is excised. The resulting two-chain zymogen consists of a light chain (Mr ≈16,000), comprising the Gla domain with 11 Gla residues and the EGF domains, which is linked via a disulfide bond to the heavy chain (Mr ≈42,000) that consists of a 52-residue activation peptide and the serine protease domain (see Fig. 113–1).

Hydroxylation of Asp63 mediates calcium binding to the EGF 1 domain and orients the Gla domain, which is essential for factor X clotting activity.³⁵ N-linked glycosylation of the activation peptide residues Asn181 and Asn191 has been implicated in prolonging the factor X half-life.⁶³ Further posttranslational modification of factor X consists of O-linked glycosylation at Thr159 and Thr171 in the activation peptide and Thr443 in the serine protease domain. There is some evidence that glycosylation of the human factor X activation peptide may also contribute to substrate recognition by the intrinsic or extrinsic factor X-activating complex.^64,65

Factor X Activation and Factor Xa Activity

Factor X is proteolytically activated by either the factor VIIIa–factor IXa (“intrinsic tenase”) or the tissue factor–factor VIIa (“extrinsic tenase”) enzyme complexes following cleavage at Arg194 in the heavy chain (see Fig. 113–1). This results in the release of the activation peptide and generation of factor Xa, also known as factor Xaα. A snake venom protease from Russell’s viper venom (RVV-X) is capable of generating factor Xa in a similar manner.

Factor Xa consists of the Gla and EGF domains-comprising light chain (Mr ≈16,000) and the catalytic heavy chain (Mr ≈29,000) that are covalently linked via a disulfide bond. Factor Xa reversibly associates with its cofactor factor Va on an anionic membrane surface in the presence of calcium ions to form prothrombinase, the physiologic activator of prothrombin. Factor Xa is also involved in the proteolytic activation of factors V, VII, and VIII.^66,67,68

Similar to thrombin, factor Xa plays a role in other biologic and pathophysiologic processes that are not directly related to coagulation. Factor Xa triggers intracellular signaling via activation of PAR1 and/or PAR2. Factor Xa cleaves PAR2 by itself as well as in complex with tissue factor–factor VIIa. These direct cellular effects of factor Xa contribute to wound healing, tissue remodeling, inflammation, angiogenesis, and atherosclerosis, among others.^26,69

Further autocatalytic cleavage at Arg429 near the C-terminus of the factor Xa heavy chain leads to release of a 19-residue peptide, yielding the enzymatically active factor Xaβ.^70,71,72 Plasmin-mediated cleavage of factor Xa at adjacent C-terminal Arg or Lys residues also results in the generation of factor Xaβ and factor Xaβ derivatives.^73,74 While the coagulation activity is eliminated in the factor Xaβ derivatives, they are capable of interacting with the zymogen plasminogen and enhance its tissue plasminogen activator-mediated conversion to plasmin, thereby promoting fibrinolysis.⁷⁵

A primary plasma inhibitor of factor Xa is the serpin anti-thrombin, and this inhibition is enhanced by heparin (see Table 113–4), which induces a conformational change in antithrombin that is required for simultaneous active site and exosite interactions with factor Xa.⁷⁶ Another potent factor Xa inhibitor is TFPI, which inhibits both the ternary tissue factor–factor VIIa–factor Xa complex as well as free factor Xa, for which protein S functions as a cofactor.^77,78 Free factor Xa is also inhibited by the protein Z/protein Z–dependent protease inhibitor (ZPI) complex on membranes.⁷⁹

Low-molecular-weight heparin and synthetic derivatives (e.g.fondaparinux) are clinically used as anticoagulants to enhance factor Xa inhibition by antithrombin specifically. The target-specific oral anticoagulants rivaroxaban, apixaban, edoxaban, and analogues directly inhibit both free factor Xa and prothrombinase complex-assembled factor Xa with high specificity through a high-affinity, reversible interaction with the factor Xa active site.^80,81,82,83

Gene Structure and Variations

The gene encoding factor X (F10) is located on chromosome 13q34 and spans almost 27 kb.⁸⁴ The 8 exons in the factor X gene give rise to a mature mRNA of 1560 bases (Fig. 113–9). There are no common alternative splice variants with known biology.

Loss of function mutations in the factor X gene lead to a rare bleeding disorder with a recessive mode of inheritance. Factor X deficiency occurs in approximately one in every 1,000,000 newborns. Most cases of documented factor X deficiency experience serious bleeding problems. In fact, factor X deficiency may be the most severe of the rare congenital bleeding disorders.³⁰ Well over 100 mutations have been documented in cases with factor X deficiency (http://www.hgmd.org). The majority of these mutations are missense and nonsense mutations.

Gain-of-function mutations in factor X could potentially increase thrombotic risk, but such mutations have not been documented. There is uncertainty about whether common gene variations influence the level of factor X in plasma.⁸⁵

PROTEIN C

Protein C, which plays a central role in the anticoagulant pathway, was discovered in 1960, and being the third protein peak (“peak C”) observed in a vitamin K–dependent plasma protein purification, it was named protein C.^86,87 Protein C is synthesized in the liver and circulates in plasma as a two-chain zymogen of 417 amino acids (Mr ≈62,000) at a concentration of 65 nM with a half-life of 6 to 8 hours (see Table 113–1).

Protein Structure

Protein C is synthesized as a single-chain precursor and during intracellular processing amino acids Lys146-Arg147 are excised. The resulting two-chain zymogen consists of a light chain (Mr ≈21,000) comprising the Gla domain with nine Gla residues and the EGF domains, which is linked via a disulfide bond to the heavy chain (Mr ≈41,000) that consists of the 12-residue activation peptide and the serine protease domain (see Fig. 113–1).

In addition to γ-carboxylation, protein C is hydroxylated at Asp71 in the EGF-1 domain, which coordinates calcium binding.³⁵ N-linked glycosylation of Asn97 in EGF-1 and Asn248, Asn313, and Asn329 in the serine protease domain are important for efficient protein secretion, proteolytic processing of Lys146-Arg147, and proteolytic activation.^88,89,90 Some of the total plasma protein C is not glycosylated at either Asn329 (β-protein C) or at both Asn329 and Asn248 (γ-protein C), of which the impact on protein function remains unclear.⁹¹

Protein C Activation and Activated Protein C Activity

Protein C is proteolytically activated by α-thrombin in complex with the endothelial cell surface protein thrombomodulin following cleavage at Arg169 (see Fig. 113–1). The activation peptide is released and the mature serine protease activated protein C (APC) is formed. Activation of protein C is enhanced by its localization on the endothelial surface through association with the endothelial cell protein C receptor (EPCR).⁹² Several snake venom proteases (RVV-X and Protac) are also capable of activating protein C.

APC consists of the disulfide-linked light chain comprising the Gla and EGF domains (Mr ≈21,000) and the catalytic heavy chain (Mr ≈32,000). In complex with its cofactor protein S, APC proteolytically inactivates factors Va and VIIIa in a calcium- and membrane-dependent manner. Intact factor V has been reported to function as a cofactor for the inactivation of factor VIIIa in the presence of protein S.⁹³

Downregulation of thrombin formation through inactivation of these cofactors seems to occur preferentially on the endothelial cell surface as opposed to that of platelets,⁹⁴ where it prevents coagulation and potential thrombosis. However, protein C activation is also accelerated by platelet factor 4 (PF4), which is secreted by activated platelets. Upon interaction with the Gla domain of protein C, PF4 modifies the conformation of protein C, thereby enhancing its affinity for the thrombomodulin-thrombin complex.⁹⁵ This ensures APC generation in close proximity of the injury site where platelets are activated, which serves to impede dissemination of coagulation.

APC also plays a major role in the cytoprotective pathway to prevent vascular damage and stress.⁹⁶ These activities include antiapoptotic activity, antiinflammatory activity, alterations of gene-expression profiles, and endothelial barrier stabilization. Most of these functions require binding to EPCR and PAR1 cleavage.

APC is primarily inhibited by the heparin-dependent serpin protein C inhibitor and by plasminogen activator inhibitor-1 (PAI-1). Because PAI-1 is the major inhibitor of tissue plasminogen activator, inhibition through complex formation with APC contributes to enhanced fibrinolysis. Chapter 114 discusses these and other factors that attenuate the anticoagulant activity of APC.

Gene Structure and Variations

The protein C gene (PROC) is located on chromosome 2q14.3 and spans almost 11 kb.⁹⁷ The gene is divided into nine exons and the mature mRNA has a length of 1790 bases (Fig. 113–10). There are no alternative mRNA species with known biology.

Loss-of-function mutations cause protein C deficiency. In homozygous or compound heterozygous form this leads to life-threatening purpura fulminans at birth which, if left untreated, is fatal.⁹⁸ In cases where there is still some protein C activity detectable, symptoms may be much milder.

Heterozygous protein C deficiency increases the risk of venous thrombosis. This is true for most deficiencies of natural anticoagulants and sets them apart from rare bleeding disorders where heterozygosity for loss of function mutations is mostly asymptomatic. The risk for venous thrombosis is increased approximately 10-fold in heterozygotes for protein C deficiency, albeit that the risk estimates vary considerably between studies.⁹⁹ Family studies in particular suggest a high risk, whereas case-control studies may show markedly lower estimates.¹⁰⁰

Heterozygous protein C deficiency can be categorized as type I or type II. In type I deficiency, antigen levels are approximately 50 percent of normal, whereas in type II deficiency, antigen levels are (near) normal but activity levels are decreased by 50 percent.

The genetic basis of protein C deficiency, consistent with what is observed in general for congenital loss of function disorders, is heterogeneous. In line with this, more than 300 mutations have been documented and are tracked in the human gene mutation database (http://www.hgmd.org). Two-thirds of these documented mutations are missense or nonsense.

Several common polymorphisms, in particular in the promotor region of the protein C gene, are known to have a small but measurable effect on plasma protein C levels. Alleles of these polymorphisms that are associated with lower protein C levels are also associated with an increased thrombotic risk, albeit that the effect is small.¹⁰¹ Therefore, it is not surprising that measurement of these polymorphisms have not found any clinical application.

Factors V and VIII both function as cofactors in coagulation and dramatically enhance the catalytic rate of their macromolecular enzyme complexes, resulting in the generation of thrombin and factor Xa, respectively. Apart from their functional equivalence, they also share similar gene structures, amino acid sequences, and protein domain structures, which is not surprising considering that factors V and VIII are assumed to descend from the common ancestral A1-A2-A3 domain-containing copper-binding plasma protein ceruloplasmin through a gene duplication event.¹⁰² After acquiring C-type domains as well as the central B domain, a second gene duplication ultimately separated the ancestral genes of factors V and VIII.

Factors V and VIII undergo similar mechanisms of intracellular processing in the endoplasmic reticulum (ER) and Golgi apparatus. Trafficking through this early secretory pathway involves interaction of factors V and VIII with a receptor complex that consists of the mannose-binding lectin-1 gene product LMAN1 (also called ER-Golgi intermediate compartment (ERGIC)-53) and multiple coagulation deficiency protein 2 (MCFD2).¹⁰³ Defects or deficiencies in one of the two subunits of the receptor complex can result in a combined deficiency of factors V and VIII (Chap. 124).

FACTOR V

In 1943, Norwegian physician Paul Owren discovered the fifth coagulation factor thus far known and named it factor V.^104,105,106 Factor V is synthesized in the liver and circulates in plasma as a large single-chain procofactor of 2196 amino acids (Mr ≈330,000) at a concentration of 20 nM with a half-life of 12 to 36 hours (see Table 113–1).

Approximately 20 percent of the total factor V in blood is stored in the α-granules of platelets. Although it was originally thought that megakaryocytes synthesize factor V, studies in humans indicate that platelet factor V originates from plasma through endocytic uptake.^107,108,109 Platelet factor V is modified intracellularly such that it is functionally unique compared to its plasma-derived counterpart. It is partially activated, more resistant to inactivation by APC, and has several different posttranslational modifications.¹¹⁰

Platelet factor V is associated with the large multimeric protein multimerin.¹¹¹ Multimerin has a massive repeating structure, with some of the multimers having molecular weights of several million daltons. Although the function of this platelet factor V-specific multimeric chaperon protein is similar to that of VWF, the multimeric chaperon protein of factor VIII in plasma, multimerin and VWF share no structural homology.

Following platelet activation, platelet factor V becomes available at the site of injury and can reach local concentrations that exceed the factor V plasma concentration by more than 100-fold.¹¹² Interestingly, the origin of factor V in mouse platelets differs from humans in that it is synthesized in megakaryocytes and stored into the α-granules before platelets are released from the marrow.^113,114

Protein Structure

Factor V has an A1-A2-B-A3-C1-C2 domain structure (Fig. 113–11). The three A-type domains share significant homology with those of ancestral ceruloplasmin as well as with the factor VIII A domains (approximately 50 percent sequence identity). The two C-type domains belong to the family of discoidin domains, which are generally involved in cell adhesion, and share approximately 55 percent sequence identity with the factor VIII C domains. The C domains mediate binding to the anionic phospholipid surface, thereby localizing factor V to the site of injury and facilitating interaction with factor Xa and prothrombin.^{115,116,117,118} In contrast, the large central B domain of factor V shows weak homology to the factor VIII B domain or to any other known protein domain. However, this domain comprises so-called basic and acidic regions that are highly conserved throughout evolution and serve to negatively regulate factor V function and prevent activity of the procofactor.^119,120

Factor V undergoes extensive posttranslational modifications, including sulfation, phosphorylation, and N-linked glycosylation.^35,121 Sulfation at sites in the A2 and B domain are involved in the thrombin-mediated proteolytic activation of factor V.¹²² Phosphorylation at Ser692 in the A2 domain enhances the APC-dependent inactivation of the cofactor Va.¹²³ N-linked glycosylation occurs throughout the whole protein; however, the majority of carbohydrates are linked to Asn residues within the B domain and play a role in the LMAN1-MCDF2 receptor complex-mediated trafficking of factor V from the ER to the Golgi in the early secretory pathway.¹⁰³ Partial glycosylation at Asn2181 in the C2 domain of factor V results in a lower binding affinity for negatively charged membranes of the glycosylated form, thereby reducing the factor V procoagulant activity, particularly at low phospholipid concentrations.^124,125 Furthermore, factor V comprises several disulfide bonds that are important for the three-dimensional structure of the A and C domains.¹²¹

Factor V Procofactor Activation and Factor Va Cofactor Function

Sequential proteolytic cleavage of the procofactor factor V at Arg709, Arg1018, and Arg1545 in the B domain results in release of the inhibitory constraints exerted by the B domain and in the generation of the heterodimeric cofactor Va (see Fig. 113–11).¹²⁶ Maximal cofactor activity correlates with cleavage at Arg1545, which is consistent with the observation that a snake venom protease from RVV-V, which cleaves only at Arg1545, results in full activation. Thrombin has generally been recognized to be the principal activator of factor V. However, recent findings suggest that in the initiation phase of coagulation factor V is primarily activated by factor Xa.¹²⁷ Factor Xa initially cleaves factor V at Arg1018, followed by proteolysis at Arg709 and Arg1545.¹²⁸

Factor Va is composed of a heavy chain (Mr ≈105,000) comprising the A1-A2 domains and the A3-C1-C2 light chain (Mr ≈74,000), which are noncovalently associated via calcium ions. Factor Va is a nonenzymatic cofactor within the prothrombinase complex that greatly accelerates the ability of factor Xa to rapidly convert prothrombin to thrombin.¹²⁹ APC catalyzes the inactivation of factor Va by cleavage at the main sites Arg306 and Arg506, upon which the cleaved A2 fragment dissociates and factor Va can no longer associate with factor Xa.¹³⁰ A common Arg506Gln mutation in factor V leads to resistance to inactivation by APC (factor V Leiden) and is associated with an increased risk of venous thromboembolism (Chap. 133).¹³¹

Both factor V and an alternatively spliced isoform of factor V (factor V-short), which lacks the major part of the B domain (residues 756 to 1458) and normally circulates in low abundance, interact with full-length TFPI (TFPIα), most likely through the acidic B domain region.^132,133 The linkage of factor V and TFPIα is considered to attenuate the bleeding phenotype in factor V–deficient patients, as the low TFPIα levels in these patients allow the residual platelet factor V to be sufficient for coagulation.^132,134 Conversely, increased factor V–short expression caused by an A2440G mutation in the factor V gene leads to a dramatic increase in plasma TFPIα, resulting in a bleeding disorder.¹³³

Gene Structure and Variations

The gene for factor V (F5) is located on chromosome 1q23. It is located very close to the genes for the selectin family of leukocyte adhesion molecules. The factor V gene spans approximately 70 kb and consists of 25 exons (Fig. 113–12). The gene structure is very similar to that of the factor VIII gene, with exon–intron boundaries occurring at exactly the same location in 21 out of 24 cases.¹³⁵

Homozygosity or compound heterozygosity for loss-of-function mutations in the factor V gene lead to a bleeding disorder (termed parahemophilia or Owren parahemophilia).¹³⁶ At the time of writing, 152 mutations in the factor V gene have been collected in the human gene mutation database (www.hgmd.org).

Gain-of-function mutations in the factor V gene increase the risk of thrombosis. This is particularly the case for venous thrombosis and not so much for arterial thrombosis. In whites, the most common gain-of-function mutation in the factor V gene is factor V Leiden (Arg506Gln), which leads to a plasma abnormality known as APC resistance (Chap. 133).^137,138

FACTOR VIII

Factor VIII (antihemophilic factor) was first discovered in 1937, but it was not until 1979 that its purification by Tuddenham and coworkers led to the molecular identification of the protein.^139,140 Factor VIII is synthesized as a single-chain preprocofactor of 2351 amino acids and, subsequent to intracellular processing, is secreted as a series of metal ion-linked heterodimers due to proteolysis at the A3-B junction and differential processing in the central B domain (Fig. 113–13). The mature factor VIII procofactor comprises 2332 amino acids (Mr ≈300,000) and circulates in a high-affinity complex with its carrier protein VWF at a concentration of approximately 0.7 nM and a circulatory half-life of 8 to 12 hours (see Table 113–1). Complex formation with VWF protects factor VIII from proteolytic degradation, premature ligand binding, and rapid clearance from the circulation.

The primary source of factor VIII is the liver,^141,142 but extrahepatic synthesis of factor VIII also occurs.^143,144 While contradictory evidence exists on the cellular origin of both hepatic and extrahepatic factor VIII synthesis, recent studies in mice support that endothelial cells from many tissues and vascular beds synthesize factor VIII, with a large contribution from hepatic sinusoidal endothelial cells.^145,146,147 This is consistent with observations on factor VIII expression in human endothelial cells from the liver and lung.^148,149

Factor VIII is less-efficiently secreted from the cell as compared to factor V, because it interacts with the ER-chaperon proteins calnexin and calreticulin, whereas factor V interacts with calreticulin only.¹⁵⁰ Both chaperons preferentially interact with GPs comprising mono-glucosylated N-linked oligosaccharides and promote correct folding of proteins that enter the secretory pathway and target misfolded proteins for degradation. Factor VIII, but not factor V, also interacts with the ER-chaperon immunoglobulin-binding protein (BiP/GRP78), which appears to enhance the stability of factor VIII, but also retards its secretion.¹⁵¹ Factor VIII trafficking from the ER to the Golgi is mediated via the LMAN1-MCDF2 receptor complex, similar to factor V.¹⁰³

Several clearance receptors are responsible for actively removing factor VIII from the circulation, which include the low-density lipoprotein (LDL) receptor-related protein 1 (LRP1), the LDL receptor, and receptors that specifically interact with carbohydrate structures on factor VIII.^{152,153,154,155,156}

Protein Structure

The A1-A2-B-A3-C1-C2 domain structure of factor VIII shares significant homology with factor V except in the B domain region (see Fig. 113–13). In contrast to factor V, the factor VIII B domain is dispensable for procoagulant activity. The mature factor VIII procofactor comprises a variably sized heavy chain (A1-A2-B; Mr ≈200,000 to 90,000 depending on the extent of proteolysis) and a light chain (A3-C1-C2; Mr ≈80,000). The C-terminal regions of the A1 and A2 domains and the N-terminal portion of the A3 domain contain short segments of 30 to 40 negatively charged residues known as the a1, a2, and a3 regions. Interaction with VWF is facilitated by the a3 region and C1 domain.^157,158 The C domains mediate binding to the anionic phospholipid surface, thereby localizing factor VIII to the site of injury and facilitating interaction with factor IXa and factor X.^159,160,161

Factor VIII is heavily glycosylated and the majority of the N-linked glycosylation sites are found in the B domain, which mediate interaction with the chaperons calnexin and calreticulin and, in part, with the LMAN1–MCDF2 receptor complex.^103,150,162 Sulfation of tyrosine residues is required for optimal activation by thrombin, maximal activity in complex with factor IXa, and maximal affinity of factor VIIIa for VWF.^35,163 Factor VIII comprises two phosphorylation sites that are located in the A1 (Thr351) and B (Ser1657) domains.

Factor VIII Procofactor Activation and Factor VIIIa Cofactor Function

Thrombin and factor Xa are the principal activators of the procofactor VIII and generate the cofactor VIIIa through sequential proteolysis at Arg740, Arg372, and Arg1689.^{126,164,165,166} The heterotrimeric factor VIIIa is composed of the A1 (Mr ≈50,000), A2 (Mr ≈43,000), and the A3-C1-C2 light chain (Mr ≈73,000) subunits (see Fig. 113–13). The A1 and A3-C1-C2 subunits are noncovalently linked through calcium ions, whereas A2 is associated with weak affinity primarily by electrostatic interactions.^167,168 Once activated, factor VIIIa functions as a cofactor for factor IXa in the phospholipid-dependent conversion of factor X to factor Xa. The rapid and spontaneous loss of factor VIIIa cofactor activity is attributed to A2 domain dissociation from the heterotrimer.^167,168 Additional proteolysis by APC, factor Xa, or factor IXa also results in the downregulation of factor VIIIa cofactor activity.¹⁶⁹

Gene Structure and Variations

The factor VIII encoding gene (F8) is situated at chromosome Xq28. The factor VIII gene contains 26 exons (Fig. 113–14), one more than factor V, because exon 5 of factor V corresponds to exons 5 and 6 of the factor VIII gene.¹⁷⁰ In addition, the gene for factor VIII is much larger than that of factor V, spanning approximately 190 kb. This is largely because six of the introns in the factor VIII gene are much larger than the corresponding F5 introns. The mRNA for factor VIII is also much larger than the factor V mRNA because of a 1.8 kb 3′-untranslated region.

A defect or deficiency in factor VIII leads to hemophilia A. Chapter 123 discusses the prevalence, clinical characteristics, and molecular genetics of hemophilia A in detail.

High levels of factor VIII are a common and strong risk factor for venous thrombosis. It has been suspected that certain genetic variations in the factor VIII gene might play a role in determining the level of factor VIII; however, such variations have not been identified.¹⁷¹ The ABO blood group does play a role in determining the level of factor VIII, but probably indirectly through an effect on the level of VWF.^172,173

PROTEIN S

Protein S, which is named after the city (Seattle) where it was discovered by the group of Earl Davie in 1977, is a vitamin K–dependent single-chain GP of 635 amino acids (Mr ≈75,000) that circulates with a plasma half-life of 42 hours (see Table 113–1). Part of the total protein S pool circulates in a free form at a concentration of 150 nM, whereas the majority (approximately 60 percent; 200 nM) circulates bound to the complement regulatory protein C4b–binding protein (C4BP). Protein S is primarily synthesized in the liver by hepatocytes, in addition to endothelial cells, megakaryocytes, testicular Leydig cells, and osteoblasts.^{174,175,176,177,178}

Protein Structure

The protein structure of protein S differs from the other vitamin K–dependent proteins as it lacks a serine protease domain and, consequently, is not capable of catalytic activity. Protein S is composed of a Gla domain comprising 11 Gla residues, a thrombin-sensitive region (TSR), four EGF domains, and a C-terminal sex hormone–binding globulin (SHBG)-like region that consists of two laminin G-type domains (Fig. 113–15). The SHBG-like domain is involved in the interaction with the β-subunit of C4BP.

Apart from γ-carboxylation of Glu residues, protein S is posttranslationally modified via N-glycosylation in the second laminin G-type domain of the SHBG-like region (Asn458, Asn468, Asn489). β-Hydroxylation of Asp95 or Asn residues (Asn136, Asn178, Asn217) in each EGF domain allows for calcium binding that orients the four EGF domains relative to each other.³⁵

Protein S Cofactor Function

Free protein S serves as a cofactor for APC in the proteolytic inactivation of factors Va and VIIIa.^179,180 Interaction of protein S with APC on a negatively charged membrane surface alters the location of the APC active site relative to factor Va,¹⁸¹ which accounts for the selective protein S-dependent rate enhancement of APC cleavage at Arg306 in factor Va.¹⁸² C4BP-bound protein S also exerts a similar stimulatory effect on Arg306 cleavage, albeit to lower extent, whereas it inhibits the initial APC-mediated factor Va cleavage at Arg506, resulting in an overall inhibition of factor Va inactivation.¹⁸³ Cleavage of the TSR by thrombin and/or factor Xa results in a loss of APC-cofactor activity.¹⁸⁴ Protein S also functions as a cofactor for TFPIα in the inhibition of factor Xa, which is mediated by the SHBG-like region in protein S.^77,185

Protein S has been implied to play a role in phagocytosis of apoptotic cells, cell survival, activation of innate immunity, vessel integrity and angiogenesis, and local invasion and metastasis through interaction with a family of protein tyrosine kinase receptors referred to as Tyro-3, Axl and Mer (TAM) receptors.^186,187

Gene Structure and Variations

The gene encoding protein S (PROS1) is located on the long arm of chromosome 3 (3q11.1), very close to the centromere. A highly homologous protein S pseudogene (PROSP) is located on the other side of the centromere. This pseudogene is inactive, as it is not transcribed into mRNA.¹⁸⁸ The active protein S gene encompasses 15 exons and covers a little more than 100 kb (Fig. 113–16). The mRNA sequence consists of 3560 bases. Several alternative transcripts have been identified, but none of these have known biology.

Loss-of-function mutations in PROS1 lead to protein S deficiency. Several cases of homozygous and compound heterozygous protein S deficiency have been described with extremely low protein S levels. These very rare cases suffer from life-threatening purpura fulminans at birth.¹⁸⁹

Much more common are heterozygous deficiencies of protein S, which can be categorized into three types of deficiency. Type I deficiency is characterized by antigen levels that are approximately 50 percent of normal. In type II deficiency, antigen levels are (near) normal while activity levels are decreased by 50 percent. Type III deficiency is defined by a low level of free protein S. In keeping with this classification, clinical chemistry laboratories may offer a protein S activity assay, free antigen assay, or total antigen assay (or a combination thereof). These assays are not without problems and the evaluation of protein S levels is fraught with complications that need careful attention before a final diagnosis can be made.¹⁹⁰

The genetic basis of protein S deficiency is highly heterogeneous and there are more than 200 entries in the human gene mutation database (www.hgmd.org). Most of these are missense mutations. However, protein S deficiency is often characterized by gross gene deletions, that sometimes even involve neighboring genes.¹⁹¹ The reason for this preponderance of gross gene abnormalities remains unknown.

It is commonly assumed that protein S deficiency increases venous thrombotic risk by 10-fold.¹⁰⁰ This assertion is mainly based on studies in thrombophilic families. In population-based case-control studies, however, the risk increase appears to be much more modest, if present at all.¹⁹² The reason for this discrepancy between family and population-based studies remains enigmatic. The findings argue against including tests for protein S deficiency in a thrombophilia workup of venous thrombosis cases with a negative family history.

VON WILLEBRAND FACTOR

Chapter 126 discusses the structure, function, and molecular biology of VWF in detail. VWF is a large multimeric GP that is required for normal platelet adhesion to components of the vessel wall and that serves as a carrier for factor VIII. It is exclusively synthesized in megakaryocytes and endothelial cells and stored in specialized organelles in platelets and endothelial cells. Release of VWF multimers from these organelles follows upon a stimulus or via unstimulated basal secretion from endothelial cells.¹⁹³ VWF multimers circulate at a concentration of 10 nM with a half-life of 8 to 12 hours (see Table 113–1). Clearance of VWF multimers is mainly mediated by macrophages from the liver and spleen.¹⁹⁴

Large VWF multimers are cleaved by the plasma protease ADAMTS-13 (a disintegrin and metalloproteinase with thrombospondin motifs 13).¹⁹⁵ This cleavage produces the smaller size VWF multimers that circulate in plasma. Reduced ADAMTS-13 activity is linked to various microangiopathies with increased platelet activity.

Protein Structure

The precursor protein of VWF is composed of a 22-residue signal peptide and of a proVWF protein comprising 2791 amino acids that has 14 distinct domains in the order of D1-D2-D′-D3-A1-A2-A3-D4-B1-B2-B3-C1-C2-CK.¹⁹⁶ Upon translocation to the ER, the signal peptide is cleaved off, and the proVWF dimerizes in a tail-to-tail fashion through cysteines in its cysteine knot (CK) domain. During transit through the Golgi apparatus, proVWF dimers multimerize in a head-to-head fashion through the formation of disulfide bonds between cysteine residues in the D3 domain. At the same time, D1 and D2 domains are cleaved off as a single fragment to form the VWF propeptide (741 amino acids), while the remaining domains comprising 2050 amino acid residues and up to 22 carbohydrate chains form mature VWF. In the trans-Golgi network, the VWF propeptide promotes mature VWF to assemble into high-molecular-weight multimers (Mr ≈500,000 to 20,000,000). These multimers subsequently aggregate into tubular structures that are packaged into α-granules in megakaryocytes and into Weibel-Palade bodies in endothelial cells.

von Willebrand Factor Function

Upon exocytosis from Weibel-Palade bodies and at high shear rates, multimeric VWF unrolls from a globular to a filamental conformation (often called VWF strings), up to many microns long, which becomes a high-affinity surface for the platelet GPIb–V–IX complex. Large VWF multimers are more active than smaller multimers, which is explained by the fact that the former contain multiple domains that support the interactions between platelets, endothelial cells, and subendothelial collagen.

VWF binds to matrix collagens via its A1 and A3 domains. The A1 domain also mediates binding to platelet GPIb, which is required for the fast capture of platelets.¹⁹⁷ Platelet adhesion to VWF is further supported by VWF immobilization on a surface (collagen, other platelets) and by high shear stress.

VWF complexes with factor VIII through the first 272 residues in the N-terminal region of the mature VWF protein subunit,¹⁹⁸ thereby protecting factor VIII from proteolytic degradation, premature ligand binding, and rapid clearance from the circulation.

Gene Structure and Variations

The VWF gene (VWF) is located on chromosome 12p13.3, spans approximately 180 kb, and contains 52 exons.¹⁹⁹ The VWF mRNA is 8.7 kb long. There are no alternative transcripts with known biology. A partially inactive pseudogene that includes exons 23 to 34 is located on chromosome 22p11–13.¹⁹⁹ The VWF gene is very polymorphic, which makes it sometimes difficult to distinguish between disease causing mutations and neutral gene variations.

Qualitative or quantitative deficiencies in VWF cause von Willebrand disease (VWD), a mild to severe bleeding disorder. Quantitative deficiency of VWF leads to type 1 or type 3 VWD, whereas functional defects lead to type 2 VWD. Type 1 VWD is the most common form, but type 3 VWD is the most severe. Chapter 126 discusses VWD in detail.

High levels of VWF are a risk factor for venous and arterial thrombosis. Genome-wide association studies led to the identification of several genomic loci that influence the level of VWF, including the VWF gene itself, the ABO blood group, STXB5, and SCARA5.²⁰⁰ Polymorphisms in several of these loci are also associated with thrombotic risk.²⁰¹

FACTOR XI

Factor XI, which was discovered in the early 1950s,^202,203 is synthesized in the liver and secreted as a single-chain zymogen of 607 amino acids (Mr ≈80,000). In the circulation, factor XI is found as a homodimer (Mr ≈160,000) at a concentration of 30 nM with a plasma half-life of 60 to 80 hours (see Table 113–1). All factor XI homodimers circulate in complex with high-molecular-weight kininogen (HK).²⁰⁴ HK is thought to mediate binding of factor XI to negatively charged surfaces, thereby facilitating factor XI activation.²⁰⁵ There is conflicting evidence suggesting that HK may be also involved in the interaction of factor XI with the activated platelet surface via GPIb.²⁰⁶

Protein Structure

Each factor XI subunit comprises four apple domains and a serine protease domain (see Fig. 113–15). The apple domains are structured by three disulfide bonds and form a disk-like platform on which the serine protease domain rests.²⁰⁷ The dimerization of two factor XI subunits is mediated by interactions between the two apple 4 domains that involve a disulfide bond between the Cys321 residues, hydrophobic interactions, and a salt bridge, of which only the latter two are required for dimerization.²⁰⁶ The domain structure of factor XI is highly similar to that of the monomer prekallikrein (PK), the zymogen of the protease kallikrein, which also circulates in complex with HK.²⁰⁶

Factor XI does not bear a Gla domain and thus does not require γ-carboxylation to exert its procoagulant activity. N-linked glycosylation occurs at three sites in the apple 1, 2, and 4 domains (Asn82, Asn114, Asn335) and at two sites in the serine protease domain (Asn432, Asn473).

Factor XI Activation and Activity

Activation of a factor XI subunit to factor XIa proceeds through proteolysis at Arg369 in the N-terminal region of the serine protease domain and yields two-chain activated factor XIa. There are several catalysts capable of factor XI activation, which include the contact factor XIIa, thrombin, or factor XIa itself in the presence of negatively charged surfaces.^208,209 However, their mechanisms differ as factor XI must be a dimer to be activated by factor XIIa, whereas thrombin and factor XIa lack this requirement.²¹⁰ An activated factor XI dimer may comprise either one (1/2-factor XIa) or two factor XIa subunits.²¹¹

Following activation of factor XI, binding sites for the substrate factor IX become available in the apple 3 domain and serine protease domain of factor XIa.^212,213 Factor XIa proteolytically activates factor IX to factor IXa in a calcium-dependent but phospholipid-independent manner. Both forms of the factor XIa dimer as well as monomeric factor XIa activate factor IX in a similar manner.²¹¹

Accumulating evidence supports the notion that factor XIIa–dependent activation of factor XI is not essential to normal hemostasis, but is important in pathologic thrombus formation.^214,215,216 Thrombin-mediated activation of factor XI, on the other hand, seems most significant under conditions of low tissue factor and is assumed to enhance clot stability through thrombin-activation of TAFI.^217,218

Factor XI has been reported to interact with platelet GPIb, which is mediated through a site within the apple 3 domain, and to platelet apolipoprotein E receptor 2 (ApoER2).^219,220 It has been proposed that the dimeric structure allows for simultaneous interaction with the platelet by one subunit, thereby localizing factor XI to the site of clot formation, while binding to factor IX with the other subunit.²²¹

Factor XIa function is regulated by the serpins protease nexin 1, antithrombin, C1-inhibitor, α₁-protease inhibitor, protein Z–dependent protease inhibitor, and α₂-antiplasmin.^216,222 Platelets also contain a factor XIa inhibitor, the Kunitz-type inhibitor protease nexin 2.²²³

Gene Structure and Variations

The human factor XI gene (F11) is 23 kb in length and is localized to chromosome 4q35. It consists of 15 exons and 14 introns (Fig. 113–17). Each of the four apple domains is encoded by two exons. The serine protease domain is encoded by five exons, with an organization similar to the homologous protein PK.

Deficiencies of factor XI in humans can lead to a bleeding tendency, although not as severe as in hemophilia A or B.²²⁴ Deficiency of factor XI is relatively common among Ashkenazi Jews in Israel.²²⁵ The human gene mutation database lists 232 mutations in the factor XI gene (www.hgmd.org).

Increased levels of factor XI are a risk factor for venous thrombosis.²²⁶ Genetic variations in the form of common single nucleotide polymorphisms (SNPs) seem to play a role in determining the level of factor XI and contribute to thrombotic risk.²²⁷

Factor XII, HK, and PK are part of the contact system in blood coagulation, which is triggered following contact activation of factor XII mediated via a negatively charged surface. PK is synthesized in the liver, circulates as a zymogen, and is highly homologous to factor XI (see Table 113–1). Conversion into the serine protease proceeds through limited proteolysis by activated factor XII, and the generated kallikrein reciprocally activates more factor XII. HK, which is also synthesized in the liver, is a nonenzymatic cofactor that circulates in complex with factor XI or PK (see Table 113–1). HK is cleaved at two sites by kallikrein to release the bioactive nonapeptide bradykinin, a potent vasodilator.

The contact system is at the basis of the activated partial thromboplastin time (APTT) assay that is widely used in clinical practice. In this clinical laboratory test, the negatively charged surface is provided by reagents such as glass, kaolin, celite, or ellagic acid. Factor XIIa activates factor XI, which then activates factor IX. Despite HK and PK being required for a normal APTT, they appear to be dispensable for coagulation in vivo.²²⁸ Individuals who are deficient in any of these factors do not have a bleeding tendency, even after significant trauma or surgery. However, factor XII, HK, and PK do participate in bacteremia or inflammatory responses in acute-phase reactions that do not involve the coagulation, but the classical complement system.²²⁸

FACTOR XII

Factor XII was originally reported in 1955 as the “Hageman factor,” named after the first identified factor XII–deficient patient.²²⁹ Factor XII is synthesized in the liver and circulates in plasma as a single-chain zymogen of 596 amino acids (Mr ≈80,000) at a concentration of 500 nM with a half-life of 50 to 70 hours (see Table 113–1).

Protein Structure

Factor XII, which is homologous to plasminogen activators, consists of an N-terminal fibronectin type I domain, an EGF-like domain, a fibronectin type II domain, a second EGF-like domain, a kringle domain, a proline-rich region, and a C-terminal serine protease domain (see Fig. 113–15). The proline-rich region is unique to factor XII, as it is not found in any of the other serine proteases.

Factor XII comprises an O-linked fucose in EGF 1 (Thr90), N-linked glycosylation sites in the kringle domain (Asn230) and the serine protease domain (Asn414), and several O-linked glycosylation sites in the kringle domain and proline-rich region.^230,231

Factor XII Activation and Activity

Limited proteolysis by kallikrein at Arg353 in factor XII yields the activated two-chain α-factor XIIa, in which the heavy chain (the fibronectin types I and II domains, both EGF domains, the kringle domain, and proline-rich region; Mr ≈52,000) and light chain (serine protease domain; Mr ≈28,000) are linked via a Cys340–Cys467 disulfide bond (see Fig. 113–15). Once activated, α-factor XIIa activates factor XI to factor XIa. Furthermore, α-factor XIIa activates PK, thereby contributing to its own feedback activation.²³²

Factor XII is also known to acquire α-factor XIIa activity upon contact with a negatively charged surface, the latter inducing a conformational change in factor XII.²³³ This conformational change induces a limited amount of proteolytic activity in factor XII, known as autoactivation.^234,235 Furthermore, the surface-induced active conformation of factor XII is suggested to enhance the proteolytic conversion to α-factor XIIa.²³⁶ The fibronectin types I and II domains, EGF-2, the kringle domain, and the proline-rich region are reported to contribute to interaction with a negatively charged surface.^{237,238,239,240} These naturally occurring surfaces include platelet polyphosphate (poly-P), microparticles derived from platelets and erythrocytes, RNA, and collagen.^{241,242,243,244}

Further cleavage of α-factor XIIa by kallikrein at Arg334 and Arg343 in the light chain (proline-rich region) results in the generation of β-factor XIIa, which comprises a nine-residue heavy-chain fragment that is disulfide-linked to the light chain.²³⁰ Given the absence of the heavy chain, β-factor XIIa does not interact with anionic surfaces. Even though β-factor XIIa is still capable of activating PK, it no longer activates factor XI.²⁴⁵

Despite its contribution to fibrin formation in vitro, factor XII has long been considered to be dispensable for coagulation in vivo, because factor XII deficiency is not associated with a bleeding.^229,246 However, newer in vivo studies indicate that factor XII contributes to surface-induced pathologic thrombosis via activation of factor XI.^{215,242,247,248}

The serpin C1 inhibitor is the main plasma inhibitor of α-factor XIIa and β-factor XIIa. In addition, antithrombin (AT) and PAI-1 also inhibit factor XIIa activity. Conditions in which the factor XIIa activity is not properly controlled, such as in C1 inhibitor deficiency states or in case of a constitutively active form of factor XIIa, can result in the disorder hereditary angioedema.²⁴⁹

Gene Structure and Variations

The gene for factor XII is located on chromosome 5q35.3, spans approximately 12 kb, and contains 14 exons.²⁵⁰ The intron–exon structure of the gene is similar to the plasminogen activator family of serine proteases. Portions of the gene are homologous to domains found in fibronectin and tissue-type plasminogen activator.

Loss-of-function mutations in the factor XII gene do not cause clinical symptoms in the form of a bleeding tendency in homozygous or compound heterozygous individuals, although they have a prolonged APTT.

Several common allelic variations in the factor XII gene have been examined to determine whether these variations influence plasma factor XII levels and whether these are associated with thrombotic risk. Best studied is a 46C>T transition four nucleotides upstream of the start codon. TT homozygotes have lower plasma factor XII levels than CC homozygotes, but there was no relationship with risk for venous thrombosis or myocardial infarction.²⁵¹

TISSUE FACTOR

Tissue factor, also known as thromboplastin or CD142, is the cellular receptor and cofactor for factors VII and VIIa (see Fig. 113–6) and was first described in 1905.¹² Tissue factor is expressed in extravascular tissue, particularly fibroblasts and smooth muscle cells, where it serves as a hemostatic “envelope,” poised to activate coagulation upon vascular damage. Generally, tissue factor is not exposed to the blood, but endothelial cells and adhered leukocytes may express tissue factor in response to injury or stimuli such as endotoxin or cytokines.

Protein Structure

Although many of the coagulation factors share some degree of homology, the structure of tissue factor is unique. It is the only procoagulant protein that is an integral membrane protein and shares structural homology with class II interferon receptors. Tissue factor consists of 263 amino acids (Mr ≈47,000) and comprises a 219-residue extracellular domain, a 23-residue hydrophobic transmembrane portion, and a short 21-residue intracellular tail.²⁵² The extracellular domain is made up of two fibronectin type III domains, which each comprise a disulfide bond (Cys49–Cys57, Cys186–Cys209). Elimination of the second disulfide link distorts the coagulant activity of tissue factor.

Tissue Factor Activation and Cofactor Function

The tissue factor–factor VIIa complex is generally acknowledged to be the major physiologic initiator of blood coagulation. The process of coagulation is initiated when an injury ruptures a vessel and allows blood to come into contact with extravascular tissue factor. Escape of blood from the vessel allows factor VII to bind to extravascular tissue factor and initiate coagulation. However, it is very likely that in the absence of injury, tissue factor located in close proximity of the vessels is already associated with factor VIIa.²⁵³ An injury allows the extravascular tissue factor–factor VIIa complexes to come into contact with blood and initiate thrombin generation on activated platelet surfaces. Interaction of tissue factor with factor VII induces conformational changes in the serine protease domain of factor VIIa (see Fig. 113–6), thereby allowing the latter to proteolytically activate factors IX and X.¹¹

Tissue factor does not require proteolytic activation to express its activity. However, it appears that tissue factor can occur in an inactive or “encrypted” state, and procoagulant activity follows after an appropriate stimulus. Even though the exact nature of the molecular mechanism remains to be identified, several models explaining tissue factor decryption have been put forward.

Originally, it was assumed that tissue factor encryption–decryption depends on the phospholipid environment, with decryption following upon expression of negatively charged phosphatidylserine on the membrane surface. Interaction of tissue factor with phosphatidylserine restricts the orientation of the tissue factor–factor VIIa complex, thereby ensuring correct alignment of the factor VIIa active site with the membrane-bound substrates factors X and IX.²⁵⁴ Encryption of tissue factor has been proposed to occur upon localization into lipid rafts, which are known to be poor in phosphatidylserine. In endothelial cells, assembly of the ternary tissue factor–factor VIIa–factor X complex does result in tissue factor translocation to caveolae, which renders tissue factor inactive.²⁵⁵ In addition, cell-membrane anchoring of tissue factor via acylation of palmitic and stearic acids may serve to target tissue factor to specific lipid domains.²⁵⁶

In a second model, the tissue factor–dependent procoagulant activity is explained by oxidation and reduction of the Cys186–Cys209 bond. This disulfide bond is less stable because of its strained conformation, and disruption of this link may cause conformational changes that alter the affinity of tissue factor for factor VIIa.^257,258 The breaking and formation of this disulfide link is suggested to be modulated by protein disulfide isomerases.²⁵⁵

A final model assumes that decryption relies on the dimerization of tissue factor. Like other members of the class II interferon receptors, tissue factor is capable of dimerization in a manner determined by the redox (oxidation-reduction) environment and the exposure of phosphatidylserine. However, both monomeric and dimeric forms of tissue factor appear to possess procoagulant activity.²⁵⁵

Tissue factor is not only the primary initiator of the extrinsic pathway of coagulation, it also supports activation of PAR2 on endothelial cells and smooth muscle cells. Activation of PAR2 by the tissue factor–factor VIIa(–factor Xa) complex is not necessarily directly relevant for coagulation, but it is currently speculated that this event is important for wound healing, angiogenesis, tissue remodeling, and inflammation.^38,39,40

Gene Structure and Variations

The human tissue factor gene is located on chromosome 1p21-p22. The DNA sequence of the tissue factor gene has been determined and consists of six exons and five introns that span approximately 12 kb (Fig. 113–18).

The primary transcript encoding full-length tissue factor contains six exons, but an alternatively spliced form of tissue factor (asTF) also exists in which exon 5 is spliced out. Because of a 3′ frameshift mutation, the full-length tissue factor transmembrane and cytoplasmic tail are replaced with a hydrophobic C-terminal domain, which renders the asTF soluble. asTF is expressed in lung, pancreas, placenta, heart, endothelium, and monocytes.^259,260,261 Although the level of asTF in human plasma may be substantial and amounts to 10 to 30 percent of total tissue factor,²⁶² it remains a matter of debate whether asTF contributes to coagulation.

In theory, variations in the tissue factor gene could influence thrombotic and bleeding risk. There are claims that polymorphisms in the tissue factor gene influence thrombotic risk but these claims have not been sufficiently confirmed.²⁶³

No relationship between loss-of-function mutations and bleeding has been described. This is perhaps not surprising in view of the fact that mice lacking tissue factor die early in gestation.

THROMBOMODULIN

Thrombomodulin, which was first identified by Esmon and colleagues in the early 1980s,^264,265 is a predominantly endothelial transmembrane protein and functions as an endothelial receptor for thrombin. In addition to endothelium, thrombomodulin has also been detected on a number of other cell types, including megakaryocytes, monocytes, and neuthrophils.²⁶⁶

Protein Structure

Mature single-chain thrombomodulin comprises 557 residues (Mr ≈78,000) and is composed of a lectin-like domain, a hydrophobic region, six EGF-like domains, a serine- and threonine-rich region, a transmembrane domain, and a 23-residue cytoplasmic tail. The highly charged lectin-like domains bear homology to the C-type lectins. Posttranslational modifications include five N-linked glycosylation sites that are located in the lectin-like and EGF-4 and -5 domains. O-linked glycosylation in the serine- and threonine-rich region (Ser474) supports attachment of a glycosaminoglycan, a chondroitin sulfate moiety, which forms a low-affinity binding site for thrombin.

Thrombomodulin Cofactor Function

Thrombomodulin interacts with thrombin through its EGF-5 and -6 domains in a calcium-dependent manner.²⁶⁷ As a result, thrombin’s procoagulant exosite I is shielded, which causes thrombin’s specificity to switch to the anticoagulant substrate protein C, requiring EGF domains 4 to 6 of thrombomodulin, and to TAFI, requiring EGF-3 to -6.²⁶⁸ Thrombomodulin enhances the thrombin-dependent activation of protein C more than 1000-fold.

As a result of the relatively large endothelial surface area in capillary beds, the thrombomodulin-dependent activation of protein C proceeds efficiently in the microcirculation, which serves a major role in preventing thrombosis from occurring on intact endothelium.²⁶⁹ In larger vessels where the endothelial surface area-blood volume ration is low, the presence of EPCR aids in the interaction with and presentation of protein C to the thrombomodulin-thrombin complex.²⁷⁰

Thrombomodulin also enhances the thrombin-mediated conversion of single-chain urokinase-type plasminogen activator to thrombin-cleaved two-chain urokinase-type plasminogen activator, which interferes with the generation of plasmin. Furthermore, thrombomodulin is a negative regulator of PAR signaling, as thrombomodulin-bound thrombin is incapable of PAR activation.²⁷¹ Based on this and because thrombomodulin is the cofactor responsible for APC generation, thrombomodulin plays an important role, albeit indirect, as an antiinflammatory protein. A direct contribution to suppress inflammation has been attributed to the lectin-like domains and hydrophobic region of thrombomodulin, independent of its anticoagulant activity.²⁷²

Protein C inhibitor is an effective inhibitor of the thrombomodulin-thrombin complex.²⁷³

Proteolysis of thrombomodulin by neutrophil-derived metalloproteinases and possibly rhomboids results in the generation of soluble thrombomodulin.²⁷⁴ Normal plasma levels of soluble thrombomodulin are 3 to 50 ng/mL, but may increase as a result of vascular damage associated with infection, sepsis, or inflammation.²⁷⁴

Gene Structure and Variations

The human thrombomodulin gene (THBD) is located on chromosome 20p11.2, spans approximately 3.5 kb, and consists of a single exon (Fig. 113–19). Intronless genes are uncommon in eukaryotes and include rhodopsin, angiogenin, mitochondrial genes, interferons α- and β-adrenergic receptors. Intronless genes represent recent additions to the genome, created mostly by retroposition of processed mRNAs with retained functionality. Genetic variation in thrombomodulin has been studied in conjunction with venous thrombosis, bleeding and atypical hemolytic uremic syndrome (aHUS).

There are early reports that mutations in thrombomodulin are present in patients with venous thrombosis, but it was difficult to prove causality.²⁷⁵ More recent work that made use of thrombomodulin sequencing in relatively large studies support the putative relationship between thrombomodulin function and venous thrombosis.²⁷⁶ Such mutations, however, do not explain a large proportion of the heritability of venous thrombosis as they seem to be quite rare.

Recently a novel thrombomodulin mutation, p.Cys537Stop, was described in a family with a history of posttraumatic bleeding.²⁷⁷ The endogenous thrombin potential was markedly reduced at low tissue factor concentrations in heterozygous carriers. Plasma thrombomodulin levels were elevated (433 to 845 ng/mL, normal range 2 to 8 ng/mL), and the addition of exogenous protein C further decreased thrombin generation. It was surmised that as a consequence of the premature stop codon, the truncated thrombomodulin is shed from the endothelial surface into the blood plasma, which would promote systemic protein C activation, thereby explaining the bleeding phenotype.

Missense mutations in thrombomodulin were also reported in patients with aHUS and this involvement in aHUS is probably related to the role of thrombomodulin in the complement system.²⁷⁸ Thrombomodulin binds to C3b and factor H and negatively regulates complement by accelerating factor I-mediated inactivation of C3b. In addition, by promoting activation of TAFI, thrombomodulin also accelerates the inactivation of C3a and C5a. Thrombomodulin variants associated with aHUS had diminished capacity to inactivate C3b and to activate TAFI and were thus less protected from activated complement, thereby providing an explanation for their involvement in aHUS.

ENDOTHELIAL PROTEIN C RECEPTOR

The EPCR, a single-chain transmembrane receptor discovered in 1995 by Fukodome and Esmon,²⁷⁹ binds both protein C and APC. EPCR increases the rate of activation of protein C⁹² and alters the function of APC from anticoagulant to cytoprotective.²⁸⁰ EPCR is mainly expressed by endothelial cells but also by leukocytes and other cell types.

Protein Structure

EPCR is homologous to CD1 and major histocompatibility class I proteins and folds with a β-sheet platform supporting two α-helical regions that form the potential binding pocket for protein C and APC. The mature protein (Mr ≈49,000) consists of 223 amino acids and is glycosylated through four N-linked glycosylation sites (Asn30, Asn47, Asn119, Asn155). EPCR contains a 25-residue long C-terminal transmembrane region with a short 3-residue cytoplasmic tail.

Endothelial Protein C Receptor Function

EPCR enhances the activation of membrane-bound protein C by the thrombomodulin–thrombin complex,⁹² thereby enhancing the APC-mediated anticoagulant pathway.

APC bound to membrane-associated or soluble EPCR is disabled in its anticoagulant capacity. Instead, EPCR-bound APC activates PAR1 in an alternative manner by noncanonical cleavage at a Arg46,²⁸¹ resulting in an increased barrier function of endothelial cells mediated via the β-Arrestin/PI3K (phosphatidylinositide 3′-kinase)/AKT/Rac1 pathway. This is in contrast to the barrier-disruptive Arg41 cleavage of PAR1 by thrombin that activates the G-protein/ERK (extracellular regulated kinase) 1.2/RhoA pathway.²⁸¹

EPCR is essential at the maternal–embryonic interface on trophoblast giant cells where it prevents fibrin formation. Consequently, complete EPCR deficiency leads to embryonic lethality. EPCR-deficient embryos rescued by the presence of EPCR in the trophoblast are viable and thrive, which seems to indicate that EPCR is not essential to blood circulation, at least in mice.²⁸² Additional ligands for EPCR have been discovered such as factor VIIa, Plasmodium falciparum erythrocyte membrane protein, and the V(γ)4V(δ)5 T-cell receptor.²⁸³ These additional ligands indicate potential involvement of EPCR in the therapeutic effect of factor VIIa in hemophilia patients, and roles for EPCR in malaria, cytomegalovirus infection, and cancer.

Gene Structure and Variations

The chromosomal location of the EPCR gene (PROCR) is 20q11.2 and it contains four exons and spans 6 kb. Exon 1 encodes for the 5′-untranslated region and the signal peptide; exons 2 and 3 encode for almost the entire extracellular region; and exon 4 encodes for the transmembrane domain and cytoplasmic tail. One single mRNA encodes the centrosomal protein CCD41 and EPCR. Deletion of the signal sequence confers the centrosomal location of CCD41, while the unprocessed protein is incorporated into cell membranes as EPCR.

Variants of EPCR with reduced protein C affinity or increased cellular shedding are reported to be associated with unprovoked venous thromboembolism.²⁸⁴

FIBRINOGEN

Fibrinogen, when converted to fibrin, forms the structural meshwork that consolidates an initial platelet plug into a solid hemostatic clot. Fibrinogen is synthesized in the liver and circulates in a concentration of approximately 7.4 μM. The plasma half-life of fibrinogen is 3 to 5 days, with only a small proportion of the catabolism caused by consumption.²⁸⁵ Fibrinogen is also found in the α-granules of platelets. It was initially assumed that megakaryocytes synthesized fibrinogen. However, although some γ-chain transcripts are present in marrow precursors, it appears that most of the fibrinogen found within platelets is taken up from the plasma by endocytosis.^286,287

Protein Structure

Chapter 135 provides a detailed description of the biochemistry of fibrinogen and of fibrin formation and degradation. Fibrinogen is a dimeric GP (Mr ≈340,000) and each of the two subunits contains three disulfide-linked polypeptide chains that are referred to as the Aα (Mr ≈66,500), Bβ (Mr ≈52,000), and γ (Mr ≈46,500) chains. A trinodular model of fibrinogen structure has been established from the crystal structure of fibrinogen (Fig. 113–20).²⁸⁸

Because human fibrinogen is subject to modification at a number of different sites both during and after biosynthesis, the fibrinogen present in the circulation is a heterogeneous mixture of molecules. These normal variants are caused by alternative splicing, modification of certain amino acids by sulfation, phosphorylation, and hydroxylation, different degrees of glycosylation, and proteolysis. It has been estimated that the number of nonidentical fibrinogen molecules that can be produced by these mechanisms is in excess of 1 million.²⁸⁹ Some of these variations may have significant functional consequences. For example, the level of one variant of fibrinogen with an alternatively spliced γ chain (fibrinogen-γ′) is associated with a risk of venous thrombosis.²⁹⁰

Fibrinogen Activation and Fibrin Function

Thrombin binds to the central domain of fibrinogen and proteolytically releases two fibrinopeptides A (Aα, residues 1 to 16) and two fibrinopeptides B (Bβ, residues 1 to 14) from each fibrinogen molecule.²⁹¹ Release of the fibrinopeptides exposes binding sites in the E domain that have complementary sites in the D domains of other fibrin monomers.^292,293 These complementary binding sites lead to the initial formation of two-stranded protofibrils with a half-staggered overlap configuration (Fig. 113–21). Protofibrils then aggregate into thick fibers that branch into a meshwork of interconnected thick fibers.²⁹⁴ The half-staggered overlap of the fibrin monomers gives a characteristic cross-banded pattern on electron micrographs.²⁹⁵

During fibrin monomer polymerization, other plasma proteins also bind to the surface of the developing meshwork. These include elements of the fibrinolytic system and a variety of adhesive proteins, such as fibronectin, thrombospondin, and VWF. These surface proteins influence the generation, crosslinking, and lysis of fibrin. Fibrin(ogen) also has specific integrin-binding sites that are essential for platelet binding. The thrombin that initiates fibrin polymerization also activates factor XIII, which stabilizes the fibrin polymer by crosslinking. Factor XIIIa also crosslinks other bound proteins, for example, PAI-1, vitronectin, fibronectin, and α₂-antiplasmin, to the fibrin network.

Once formed, the fibrin mesh can be degraded by the fibrinolytic system. Plasmin cleaves fibrin and fibrinogen in an ordered sequence at arginyl and lysyl bonds, giving rise to a series of soluble degradation products.²⁹⁶ In this process, the crosslink between two D fragments remains intact, resulting in the formation of a fragment consisting of two D domains and one E domain, called D-dimer. Circulating D-dimer concentrations are often measured as a surrogate marker of activated coagulation.

In addition to its obvious procoagulant role in stabilizing the initial platelet hemostatic plug, fibrin can also act as an important inhibitor of thrombin generation. Fibrin functions as “antithrombin I” by sequestering thrombin in the developing fibrin clot, and also by reducing the catalytic activity of fibrin-bound thrombin.²⁹⁷

Gene Structure and Variations

The genes for the three chains of fibrinogen are found within a 50-kb region on chromosome 4 at q23-q32 (Fig. 113–22). The genomic sequences show a high degree of homology, suggesting they were derived through duplication of a common ancestral gene. The homology extends to sites upstream of the gene, suggesting that common regulatory elements may reside in these areas, thus helping to coordinate synthesis of the three chains.

The physiologic importance of fibrinogen is underscored by the bleeding diathesis associated with afibrinogenemia and some dysfibrinogenemias (Chap. 125). Other dysfibrinogenemias are associated with thromboembolic disease. Although afibrinogenemia is associated with a bleeding tendency, it is usually not as severe as classical hemophilia.

FACTOR XIII

The GP factor XIII is a protransglutaminase that, upon activation, crosslinks and stabilizes fibrin clots.²⁹⁸ Plasma factor XIII is a heterotetramer consisting of two factor XIIIA subunits (731 amino acids; Mr ≈83,000) bound to two factor XIIIB subunits (641 amino acids; Mr ≈76,500) that circulates in plasma as an A2B2 complex (Mr ≈320,000) at a concentration of 94 nM with a plasma half-life of 10 days (Fig. 113–23; see Table 113–1). Factor XIII circulates in plasma associated to fibrinogen via interaction of the factor XIIIB subunit with the fibrinogen γ′ chain.

The A and B subunits of factor XIII are synthesized and expressed separately and assemble in the circulation to the heterotetramer factor XIII-A2B2.²⁹⁹ Factor XIIIA is synthesized in monocytes/macrophages, megakaryocytes, and hepatocytes, whereas factor XIIIB is synthesized exclusively in the liver and kidney.

Protein Structure

The factor XIIIA subunit consists of an activation peptide, a β-sandwich, a catalytic transglutaminase, and two β-barrel domains.²⁹⁸ Factor XIIIB acts as a carrier protein providing the long plasma half-life of factor XIII. Factor XIIIB consists of 10 Sushi domains in tandem, of which the first two Sushi domains are crucial for the binding to factor XIIIA.

Factor XIII Activation and Factor XIIIa Activity

Factor XIIIA is a proenzyme that is proteolytically activated by thrombin via cleavage at Arg37 (see Fig. 113–23), resulting in release of the activation peptide and dissociation of the factor XIIIB subunit from factor XIII-A2B2, thereby exposing the active site Cys314 of factor XIIIA. Cofactors for the activation of factor XIIIA by thrombin are calcium and fibrin(ogen). Platelets only contain the factor XIIIA (cfactor XIII-A2) dimer that is activated intracellularly in a proteolytic-independent manner through a rise in cytosolic calcium and subsequent conformational change before secretion by the stimulated platelet (see Fig. 113–23).³⁰⁰

Activation of factor XIII by thrombin is not a late event in blood coagulation, as it is activated with the same velocity by thrombin as the cleavage of the fibrinopeptides of fibrinogen.³⁰¹ Factor XIII can also be alternatively activated and inactivated by neutrophil elastase.³⁰²

Factor XIIIa consists of a factor XIII-A2 dimer comprising two activated factor XIIIA subunits that result from either thrombin-activation (factor XIIIa^*) or conformational-activation (factor XIIIa°). The transglutaminase activity of factor XIIIa crosslinks a γ-carbon of glutamine in one protein chain to the ε-amino group of lysine in another protein chain in a reaction named transamidation. The specificity of factor XIIIa crosslinking of proteins stems mainly from the recognition by factor XIIIa of specific glutamines, while the crosslink to lysine appears to be random and limited to the ones that are in the vicinity.

When thrombin cleaves the fibrinopeptides A and B of fibrinogen molecules, the binding site on the central E domain for the D domain of other fibrin molecules is uncovered. This initiates lateral aggregation of protofibrils and fiber formation. Factor XIIIa stabilizes the forming of protofibrils by linking two γ chains in adjacent D domains in fibrin polymers. Binding of Arg158 in one subunit of the factor XIIIa dimer to the αC region residue AαGlu396 of fibrin facilitates crosslinking via αC chains to a next fibrin molecule by the other activated factor XIIIA subunit in factor XIIIa.³⁰³

The crosslinking of fibrin γ-chains or α-chains by factor XIIIa has independent and specific effects on clot formation and structure. Crosslinks stabilize a clot by incorporation of the plasmin inhibitor α₂-antiplasmin which makes it resistant to fibrinolytic attack by plasmin.³⁰⁴

Several other processes during clotting are factor XIII-dependent, among which red blood cell incorporation in clots,³⁰⁵ complement factor 3 (C3) crosslinking to fibrin,³⁰⁶ and clot retraction by platelets.³⁰⁷ Factor XIII also plays a role in the functioning of multiple adhesive and contractile proteins and in angiotensin type I receptor crosslinking.³⁰⁸ Fetal specific crosslinking of Fas by factor XIII dampens apoptosis, suggesting that factor XIII may play a role in cell survival prenatally.³⁰⁹ Related to its expression by monocytes/macrophages, factor XIII levels may drop after inhibition of IL-6 receptor signalling by tocilizumab.³¹⁰ Besides its crucial role in hemostasis, factor XIII has important functions during tissue regeneration and infection.³¹¹ Factor XIII is necessary to prevent bleeding/stroke, maintain pregnancy, and aids in wound healing.²⁹⁸

Gene Structure and Function

The factor XIIIA chain gene (F13A1) has been localized to chromosome 6 p25.1.³¹² It contains 15 exons, and is larger than 160 kb (Fig. 113–24). The fibrin-binding domain is encoded by exons 2 to 12. The active site, with its reactive thiol at Cys314, is present in exon 7. The gene encoding the factor XIIIB (F13B) chain has been localized to chromosome 1q31.3. It has 12 exons and is approximately 28 kb (Fig. 113–25).³¹³ Each Sushi domain is encoded by a single exon. The regulation of factor XIIIB expression is poorly understood. A total of 30 potential start sites are located upstream of the initial methionine.

Homozygosity or compound heterozygosity for loss-of-function mutations in the factor XIIIA or XIIIB genes leads to a severe bleeding disorder that is rare (1 in 2,000,000 of the population).³¹⁴ Factor XIII–deficient newborns often present with bleeding from the umbilical cord. The natural course is characterized by a life-long bleeding tendency and spontaneous miscarriages in affected women. Acquired factor XIII deficiency by development of an inhibitory antibody may lead to fatal bleeding if not treated.³¹⁵

Mutations underlying factor XIII deficiency are more commonly found in the gene encoding factor XIIIA than the one encoding factor XIIIB. In accordance with this, the human gene mutation database lists 107 mutations in the factor XIIIA gene and 19 mutations in the factor XIIIB gene. Mutations in the gene encoding factor XIIIB often lead low levels of both factor XIII subunits. This is probably because free factor XIIIA has a short plasma half-life.

A Val34Leu polymorphism associated with fatal atherothrombotic ischemic stroke results in a faster factor XIIIA activation rate by thrombin, which affects clot structure.³¹⁶

THROMBIN-ACTIVATABLE FIBRINOLYSIS INHIBITOR

TAFI is the zymogen of a zinc-bound metalloprotease, and is also known as carboxypeptidase B, R, or U. TAFI is synthesized in the liver. Most TAFI present in the blood is in the plasma compartment, which circulates at 70 to 275 nM (see Table 113–1).

Protein Structure

TAFI is a 401-amino-acid proenzyme (Mr ≈60,000) and consists of an N-terminal activation peptide (residues 1 to 76), a linker region (residues 77 to 92), and a catalytic domain (residues 93 to 401). Twenty percent of the protein mass of TAFI is accounted for by carbohydrate side chains that are attached to four sites within the activation peptide (Asn22, Asn51, Asn63) and linker region (Asn86). The active site residues (Glu271, Arg125) and zinc-binding residues (His67, Glu70, His196) in TAFI are conserved between other members of the carboxypeptidase A family.

Thrombin-Activatable Fibrinolysis Inhibitor Activation and Thrombin-Activatable Fibrinolysis Inhibitor-a Activity

TAFI is proteolytically activated by plasmin or thrombin, reactions that are accelerated 1000-fold when thrombin is bound to thrombomodulin. Both enzymes cleave TAFI at Arg92 to give rise to activated TAFI (TAFIa; Mr ≈37,000) upon release of the activation peptide. TAFIa catalyzes removal of C-terminal lysine and arginine residues from fibrin and fibrin cleavage products. These residues are important for binding and activation of plasminogen, and removal of these residues by TAFIa reduces formation of plasmin on clots resulting in decreased clot lysis.

TAFIa may also have an antiinflammatory role as it can efficiently cleave C-terminal arginines of anaphylatoxins such as bradykinin and the complement activation peptides C3a and C5a.

Inhibitors of TAFIa have not been identified. Rather, the primary regulatory mechanism of TAFI activity involves its intrinsic thermal instability with a half-life of less than 15 minutes at 37°C.

Gene Structure and Variations

The gene for TAFI (CPB2) has been localized to 13q14.13. The gene contains 11 exons with 10 introns and spans 48 kb. Homozygosity or compound heterozygosity for mutations in the gene encoding TAFI have not been described.

In total, 19 SNPs have been identified in the gene encoding TAFI, of which six are in the coding region. Of the latter SNPs, two lead to an amino acid substitution: an Ala/Thr substitution at position 147 and a Thr/Ile substitution at position 325.³¹⁷ There appears to be a strong correlation between plasma levels of TAFI and polymorphisms in the promoter and 3′-region, but their clinical significance is unclear.

Epidemiologic studies have indicated that elevated TAFI levels are correlated with an increased risk of venous thrombosis, albeit that methods to quantify TAFI or TAFIa have limitations.³¹⁸

ANTITHROMBIN

AT was previously known as AT III as a result of a classification of several AT activities in plasma discovered in the 1950s.³¹⁹ AT is mainly synthesized in the liver and circulates in plasma as a single-chain GP of 432 amino acids (Mr ≈58,000) at a concentration of 2.5 μM with a half-life of 60 to 70 hours (see Table 113–1). AT is a member of the large serpin family and is known as SERPINC1 in the systematic nomenclature.

Protein Structure

AT consists of an N-terminal heparin-binding domain, a carbohydrate-rich domain, and a C-terminal serine protease-binding region that comprises the long, flexible, and surface-exposed reactive center loop. Structural stability is provided by three disulfide bonds, two of which are located in the N-terminal region and one in the serine protease-binding region. Posttranslational modifications comprise four N-glycosylation sites, with three in the carbohydrate-rich domain (Asn96, Asn135, Asn155) and one in the serine protease-binding region (Asn192).

Antithrombin Function

The primary proteases targeted by AT are thrombin, factor Xa, and factor IXa. In addition, AT also inhibits factors XIa and XIIa, as well as tissue factor–factor VIIa; however, the latter is only inhibited in the presence of heparin.

Similar to other serpins, AT acts as a “suicide” substrate for its target proteases. These cleave at site Arg393 in the reactive center loop of AT, upon which AT is, unlike normal substrates, not released, but forms a 1:1 covalent complex with the protease, thereby blocking the active site. This complex is facilitated by a conformational change of the reactive center loop that folds into the N-terminal region of AT. By doing so, the covalently attached protease is dragged along, resulting in distortion of its serine protease domain and effectively converting the protease back into a zymogen-like state.³²⁰

Heparin and related molecules, such as endothelial-bound glycosaminoglycans, dramatically accelerate the rate of protease inhibition by AT (see Table 113–4) through two distinct mechanisms that characterize inhibition of either factors IXa and Xa or thrombin (Fig. 113–26). In case of the former, binding of a specific pentasaccharide sequence in heparin results in a conformational change in the reactive center loop of AT, which allows for enhanced access by the target protease and is known as allosteric activation of AT (Fig. 113–26, middle panel).³²⁰ This results in a 500-fold acceleration of inhibition of factors IXa and Xa.³²¹ The pentasaccharide sequence is present in all forms of heparin including low-molecular-weight heparin and fondaparinux, a synthetic pentasaccharide. Heparin-accelerated thrombin inhibition involves bridging of AT to the protease, which serves to align the two molecules and enhances the rate of complex formation (Fig. 113–26, right panel).^322,323 This mechanism, also known as the template mechanism, requires longer heparin molecules found in unfractionated heparin (UFH). Heparin-bridging of AT also contributes to some extent to the AT-mediated inhibition of factors IXa and Xa, but the majority of rate enhancement is provided by the allosteric activation of AT.

Protease–AT complexes are cleared from the circulation by lipoprotein receptor-related protein (LRP)-1–mediated endocytosis in the liver.^324,325

Gene Structure and Variations

The 13.5-kb AT gene (SERPINC1) is localized on chromosome 1q25.1 and consists of seven exons. The cDNA is 1395 bp long, whereas the mRNA is approximately 1.4 kb.

Because of its essential role as an inhibitor of coagulation, individuals who are heterozygous for loss-of-function mutations are at increased risk for thrombosis. The prevalence of this condition in the general population is approximately one in 5000 individuals,³²⁶ while it occurs in approximately 5 percent of patients with a history of thromboembolic disease.³²⁷ AT deficiency can be categorized into type I and type II deficiencies.³²⁸ Type I deficiency is characterized by reduced plasma levels of AT; however, homozygous type I deficiency is not compatible with life. Type II deficiency covers all functional AT defects.

The human gene mutation database (www.hgmd.org) lists 274 mutations. Mutations resulting in type I deficiency consist of large deletions, frameshift mutations, premature stop codons, splice-site mutations, and missense mutations. Mutations observed in type II deficiency impair heparin binding or affect the overall protein structure. Chapters 130 and 133 provide a more detailed description of the clinical significance of AT deficiency.

TISSUE FACTOR PATHWAY INHIBITOR

TFPI is a Kunitz-type protease inhibitor that inhibits factor Xa and tissue factor–factor VIIa activity and was discovered by Broze and Miletich in 1987.³²⁹ TFPI circulates in plasma at 2.5 nM in multiple forms of which the majority is either truncated at the C-terminus or lipoprotein-associated. Only 10 percent of the circulating TFPI is the full-length TFPIα form of 276 amino acids (Mr ≈40,000; see Table 113–1).³³⁰ The half-life of TFPIα in the circulation is only 2 minutes because it readily associates with the vessel wall endothelium.

Protein Structure

Full-length TFPIα consists of three tandem Kunitz domains and a C-terminus that contains a basic region (see Fig. 113–15).³³¹ However, TFPI is very heterogeneous as a result of proteolysis and alternative splicing. The latter gives rise to TFPIβ that lacks the third Kunitz domain and C-terminus, but instead includes a sequence that facilitates anchorage to the endothelial cell membrane via GPI linkage.³³²

Endothelial cells and platelets are the main producers of TFPI, with endothelial cells expressing both TFPIα and β, while platelets only produce TFPIα that is secreted upon platelet activation. Although a significant fraction of TFPIβ is GPI-linked to the endothelial cells, it is also found in plasma. In vivo, most of the full-length TFPIα appears to be bound to endothelial heparan sulphate proteoglycans through its positively charged C-terminus. This because total plasma TFPI levels rise by approximately threefold upon heparin treatment, which is completely attributable to an increase in TFPIα. In addition, TFPIα also circulates in complex with factor V.

Tissue Factor Pathway Inhibitor Function

The physiologic relevance of TFPI stems from its ability to regulate tissue factor–dependent coagulation as well as its direct inhibition of factor factor Xa. TFPI inhibits the tissue factor–factor VIIa complex in a two-step mechanism. TFPI will bind via its second Kunitz domain to the active site of factor Xa, thereby inhibiting the proteolytic capacity of factor Xa.³³¹ This step is accelerated profoundly via protein S through interactions with the third Kunitz domain of TFPI.^333,334 The following step is the inhibition of the catalytic activity of tissue factor–factor VIIa complexes by formation of the quaternary tissue factor–factor VIIa–factor Xa–TFPI complex. This complex formation depends on the binding of Kunitz 1 to the factor VIIa active site. Overall, the effects of TFPI as regulator of tissue factor–initiated thrombin generation appear to depend on the fast protein S-dependent TFPI interaction with factor Xa.³³³

TFPI that is truncated at the C-terminus is effective in inhibiting tissue factor–factor VIIa activity; however, this seems to occur too slow to control thrombin generation at least in vitro. In contrast, inhibition of tissue factor–factor VIIa activity by GPI-anchored TFPIβ is effective and independent of protein S.³³² GPI-anchored TFPIβ also acts as an inhibitor of tissue factor–factor VIIa signaling by PARs, a function that TFPIα seems to lack.

TFPI may prevent prothrombinase formation by factor Xa in the presence of the procofactor factor V, factor V that is partially activated by factor Xa, or platelet factor V.³³⁵ However, the factor Va–factor Xa prothrombinase complex is not inhibited by TFPI as a result of competition by prothrombin.

The heterogeneity and different activities of the multiple forms of TFPI have frustrated the measurement of TFPI for clinical purposes. However, tests that estimate the free full-length form in plasma indicate an association of low TFPIα levels with venous thrombosis.³³⁶ Low levels of TFPIα are observed in protein S–deficient patients, which may be the result of a lack of association of TFPI with protein S in the circulation and faster clearance of free TFPIα.³³⁷ High TFPIα levels have been observed in patients with increased expression of a splice variant of factor V, known as factor V-short. Factor V-short, which lacks the major part of the B domain, interacts with the basic C-terminus of TFPIα, most likely through the acidic B domain region in factor V.^132,133 Increased factor V-short levels lead to a dramatic increase in plasma TFPIα, resulting in a bleeding disorder.¹³³

TFPI activity is downregulated by proteolysis at the C-terminus, upon which the basic C-terminal region or the third Kunitz domain are removed, thereby impairing inhibition of factor Xa and tissue factor–factor VIIa. Complete inactivation of TFPI is observed after proteolysis by the neutrophil derived proteases elastase and cathepsin G, which also cleave in between Kunitz 1 and 2. In this way, tissue factor–factor VIIa activity may be protected or reactivated during inflammatory processes.³³⁸

The in vivo relevance of TFPI was shown by the sensitization of rabbits to tissue factor–triggered disseminated intravascular coagulation after immunodepletion of TFPI.³³⁹ Furthermore, mice lacking the first Kunitz domain of TFPI are not viable.³⁴⁰

Gene Structure and Variations

The human TFPI gene (TFPI) is located on chromosome 2q31-q32.1 and has nine exons that span 70 kb. TFPI is synthesized in two alternatively spliced forms, α and β.³³² TFPIβ is formed by an alternative splice event after exon 7 such that TFPIβ lacks the third Kunitz domain and instead has a unique C-terminus. Exon 2 appears to downregulate translation of the TFPIβ splice variant by a unique interaction with a sequence in the 3′-end of the TFPIβ mRNA.

Homozygosity or compound heterozygosity for loss of function mutations in the gene encoding TFPI has not been described. Several genetic polymorphisms have been identified and their relationship with venous thrombosis has been investigated. There is one report describing that a T33C polymorphism in intron 7 is highly associated with total TFPI antigen and protects against venous thrombosis,³⁴¹ but this relationship with thrombosis was not confirmed in a subsequent study.³⁴²

PROTEIN Z/PROTEIN Z–DEPENDENT PROTEASE INHIBITOR

ZPI is a serine protease inhibitor (Mr ≈72,000; SERPINA10 in the systematic nomenclature) that inhibits coagulation factors Xa and XIa. ZPI circulates in plasma at 60 nM with a half-life of 60 hours (see Table 113–1).³⁴³ The ZPI-dependent inhibition of factor Xa is enhanced in the presence of protein Z.⁷⁹ Protein Z is a vitamin K–dependent plasma GP (Mr ≈62,000) that circulates at 40 nM (see Table 113–1). In normal plasma, which has a molar excess of ZPI over protein Z, all protein Z circulates in complex with ZPI.³⁴⁴

Protein Structure

ZPI displays 25 to 30 percent homology with other serpins such as AT. Based on this homology, Tyr387 was predicted and confirmed as P1 residue in the reactive center loop of ZPI and shown pivotal for inhibition of factor Xa.³⁴⁵ Unlike other serpins, the N-terminal region of ZPI contains a very acidic domain.

Protein Z consists of a Gla domain, a hydrophobic region, and two EGF-like domains. Even though the C-terminal region of protein Z contains a domain that is homologous to the serine protease domains of the other Gla-containing proteins, it lacks the His and Ser active site residues characteristic for trypsin-like serine proteases.³⁴⁶ Thus, protein Z has no protease activity.

Protein Z/Protein Z–Dependent Protease Inhibitor Function

The protein Z–ZPI complex associates with anionic phospholipid membranes in a calcium-dependent manner mediated by the Gla domain of protein Z, which facilitates formation of the ternary protein Z–ZPI–factor Xa complex. Furthermore, protein Z has also been suggested to induce conformational changes in ZPI, resulting in alignment of the reactive center loop and P1 site of ZPI with the factor Xa active site.³⁴⁷ Together, these effects of protein Z enhance the inhibitory activity of ZPI to factor Xa by 1000-fold. In contrast, ZPI inactivation of factor XIa is protein Z–independent.⁷⁹ Similar to other serpins, ZPI acts as a “suicide” substrate. However, factors Xa and XIa eventually cleave ZPI at the P1 residue Tyr387, which results in release of a 4.2-kDa C-terminal ZPI peptide.⁷⁹

The combination of protein Z and ZPI dramatically delays the initiation and reduces the ultimate rate of thrombin generation in mixtures containing prothrombin, factor V, phospholipids, and calcium. However, in similar mixtures containing factor Va, protein Z and ZPI do not inhibit thrombin generation.⁷⁹ Thus, the major effect of protein Z and ZPI is to dampen the coagulation response prior to the formation of the prothrombinase complex.

In mice, protein Z and ZPI deficiency is associated with a prothrombotic phenotype and both deficiencies dramatically increase mortality in animals with the factor V Leiden mutation. This indicates that protein Z and ZPI deficiency may be risk factors for thrombotic disease in humans.³⁴⁸ Indeed, low protein Z levels appear weakly associated with thrombosis and ischemic stroke in subgroups of some small studies. However, these studies lack power to show a definite interaction with vascular disease.³⁴⁹ Other studies demonstrate a possible association of protein Z and ZPI mutations with thrombosis and pregnancy complications.³⁵⁰

Gene Structure and Variations

The chromosomal location of the ZPI gene (SERPINA10) is 14q32.13. The gene encodes six exons and spans almost 10 kb. The gene for protein Z (PROZ) is on the long arm of chromosome 13 (q34) in close proximity to the genes for factor X and factor VII. The protein Z gene spans 14 kb and consists of nine exons. The intron/exon boundaries are identical to the other Gla-containing coagulation proteins. There is an alternative exon that codes for a unique peptide of 22 amino acids in the prepro leader sequence. The gene is transcribed into a 1.6-kb mRNA.

Several mutations and polymorphisms have been described for the gene encoding ZPI,³⁵¹ but the association between such gene variations and risk of venous thrombosis has not been established with certainty.³⁵²

The human gene mutation database lists nine loss-of-function mutations in the protein Z gene. The relationship between these mutations and disease is uncertain at best, but a relationship with ischemic stroke and recurrent fetal loss cannot be excluded.^353,354,355

PATHWAYS OF HEMOSTASIS

Early Coagulation Schemes

With the accumulated knowledge of the biochemistry of hemophilia it was recognized in the 1960s that blood coagulation was regulated by a sequential series of steps in which activation of one clotting factor led to the activation of another, finally leading to a burst of thrombin generation.^356,357 Each clotting factor was thought to exist as a proenzyme that could be converted to an active enzyme.

Since then the original waterfall reaction scheme of enzymes has been modified extensively. Factors V and VIII were identified as nonenzymatic procofactors for factors Xa and IXa, respectively, and the subsequent clotting events were divided into so-called extrinsic and intrinsic systems (Fig. 113–27). The extrinsic system was shown to consist of factor VIIa and tissue factor, the latter being viewed as extrinsic to the circulating blood. The tissue factor pathway could be activated in clotting tests by a brain tissue extract. The factor XII–dependent intrinsic system could even be activated by china clay (kaolin) and was viewed as being intravascular. Both pathways activate factor X, which, in complex with the activated cofactor Va, converts prothrombin to thrombin. Although these earlier concepts of coagulation were extremely valuable, investigators recognized that the intrinsic and extrinsic systems could not operate independently. The pivotal role of factor XII for the initiation of intrinsic thrombin generation and lack of bleeding tendency of factor XII deficient individuals was inconsistent with the extreme bleeding associated with a deficiency in factor VIII or IX, participants of the same intrinsic pathway, which urged investigators to search for additional links in the cascades.

Revision of the Coagulation Scheme

Key to the revision of the coagulation model was the purification and characterization of the transmembrane protein tissue factor.^358,359 A crucial observation was that the tissue factor–factor VIIa complex not only activates factor X, but also factor IX.³⁶⁰ This showed that factor VIII or IX deficiencies, which result in hemophilia A or B, respectively, are in fact abnormalities of the tissue factor–factor VIIa pathway, even though factors VIII and IX are components of the intrinsic system. With the notion that traces of tissue factor–factor VIIa are rapidly inactivated by TFPI, it became clear that factor VIIIa–factor IXa activity is necessary to sustain hemostatic factor Xa and thrombin generation.^{361,362,363,364} The embryonic lethality caused by both tissue factor as well as TFPI deficiency in mice underscores the importance of tissue factor–mediated thrombin generation and the control thereof.^340,365 Finally, it was observed that thrombin could directly activate factor XI,³⁶⁶ thus providing an amplification loop once thrombin generation has been initiated: thrombin → factor XIa → factor IXa → FXa → thrombin. This model explains the mild bleeding tendency observed in factor XI–deficient patients and the absence of bleeding in factor XII–deficient individuals. A revised coagulation scheme is depicted in Fig. 113–28. This scheme builds on the conclusion that the major initiating event in hemostasis in vivo is the formation of the tissue factor–factor VIIa complex at the site of injury.³⁶⁷

It was also recognized that in vivo coagulation is regulated by control mechanisms, one of which is the localization of the coagulation reactions to cell surfaces. In addition, earlier and more recent observations emphasized the importance of plasma inhibitors targeting each step of the coagulation process. These include (1) TFPI, which controls tissue factor–factor VIIa and factor Xa activity in cooperation with protein S,^331,333 (2) thrombomodulin and APC, which inactivate thrombin and factors Va and VIIIa, the latter also in cooperation with protein S,³⁶⁸ (3) AT, which inhibits thrombin and other coagulation proteases,³⁶⁹ and (4) ZPI, which inhibits factor Xa on phospholipid surfaces in cooperation with protein Z prior to incorporation of factor Xa into the prothrombinase complex.⁷⁹ See also Fig. 113–28 for an overview of the inhibitory mechanisms of coagulation.

Most of the essential pathways of tissue factor–dependent thrombin generation and inhibitory control have been captured in a mathematical model based on the empirically derived rate constants of the individual reactions.³⁷⁰ Completion and refinement of such models will aid our understanding of the biochemistry of the coagulation reactions on artificial membranes, given that the enzymatic events on cellular membranes under (reduced) flow are still complicated, which hampers their incorporation into a mathematical model.³⁷⁰

A Cell-Based Scheme of Coagulation

The goal of coagulation is to produce a fibrin clot that seals the site of injury in the vessel wall. This process is initiated when tissue factor–bearing cells are exposed to blood at the damaged site. Tissue factor is anchored to cells via a transmembrane domain and acts as a receptor for plasma factor VII. Both trace amounts of factor VIIa as well as zymogen factor VII that is rapidly converted to factor VIIa by factor Xa and/or autoactivation bind to tissue factor. Tissue factor is expressed around vessels and in the epithelium, where it forms a “hemostatic envelope.” The tissue factor surrounding the vessels may already be in complex with factor VIIa, even in the absence of an injury.²⁵³

The tissue factor–factor VIIa complex catalyzes two very important reactions: (1) activation of factor X to factor Xa and (2) activation of factor IX to IXa. The initial factors Xa and IXa formed on tissue factor–bearing cells may have distinct functions in initiating the process of blood coagulation.³⁷¹ When a vessel is damaged, the blood delivers platelets to the site of injury. These bind to extravascular matrix components to produce the primary hemostatic plug and become partially activated in the process. The platelets are consequently localized in close proximity to active tissue factor–factor VIIa complexes.

The factor Xa formed on the tissue factor–bearing cell interacts with factor Va to form prothrombinase complexes that generate small amounts of thrombin (Fig. 113–29). Although this amount of thrombin may not be sufficient to clot fibrinogen, it is sufficient to initiate events that “prime” the clotting system for a subsequent burst of thrombin generation. Experiments using tissue factor–activated whole blood and cell-based systems have shown that platelets can be activated by thrombin that is generated by direct tissue factor–factor VIIa activation of factor Xa.^371,372,373 The small amounts of factor Va required for prothrombinase assembly are likely provided by activated platelets, by factor Xa activation, or potentially by noncoagulation proteases secreted by the tissue factor–bearing cells.^335,374,375

The small amounts of thrombin generated are capable of accomplishing the following: (1) activating platelets; (2) activating factor V; (3) activating factor VIII and dissociating factor VIII from VWF; and (4) activating factor XI (see Fig. 113–29).^366,372,373 The activity of the factor Xa formed by the tissue factor–factor VIIa complex will be mostly restricted to the tissue factor–bearing surface because free factor Xa that diffuses off the cell surface is rapidly inhibited by TFPI, AT, and/or the protein Z–ZPI complex. Factor IXa, on the other hand, will most likely act on activated platelets in close proximity to the tissue factor–bearing cell. This is because factor IXa can diffuse to adjacent cell surfaces as it is not inhibited by TFPI and ZPI, while the rate of factor IXa inhibition by AT is much lower than that of factor Xa (see Table 113–4).

The Role of Activated Platelets

Platelets also play a major role in localizing clotting reactions to the site of injury, as they adhere and aggregate at the same location where tissue factor is exposed to blood. Platelet localization and activation are mediated by VWF, thrombin, platelet receptors, and vessel wall components such as collagen (Chap. 112). Once platelets are activated, the cofactors Va and VIIIa are rapidly localized to the platelet membrane surface (see Fig. 113–29). Cofactor binding is mediated in part by the exposure of phosphatidylserine on the platelet membrane, a process resulting from a flip-flop mechanism whereby phosphatidylserine on the inner leaflet of the membrane bilayer flips to the outer membrane leaflet.³⁷⁶ Endothelial cells, platelets, and leukocytes also generate procoagulant microvesicles that sustain thrombin generation. While the procoagulant characteristics of microvesicles have been studied in detail in vitro, their relative contribution to coagulation in vivo is still subject of debate.

Factor Xa generation is amplified on platelets by localization of factors IXa and XIa through specific binding sites,^377,378 and thrombin-mediated factor XI activation is enhanced by poly-P that is released by activated platelets (see Fig. 113–29).³⁷⁹ Once formed, factor Xa associates with factor Va on the platelet surface to generate a burst of thrombin that is sufficient to clot fibrinogen and form a hemostatic plug. Subsequently, thrombin-activated factor XIII crosslinks fibrin and stabilizes the hemostatic plug, thereby rendering it impermeable. Thrombin also activates TAFI, which helps to stabilize the fibrin clot. The factor XIa-mediated feedback loop has been implicated to generate ample thrombin required for TAFI activation.³⁸⁰

It should be noted that the balance between the pro- and anticoagulant reactions, the pro- and antifibrinolytic potential, as well as stress on the local vasculature vary greatly in the different the organs, muscles, joints, and other sites in the body. This is probably fundamental to the variation in bleeding phenotypes observed in various coagulation factor deficiencies.³⁸¹ The notion that factors XI and XII are not crucial to hemostasis, but are involved in thrombosis, has led to their identification as new targets to improve the safety of anticoagulant therapy by reducing the risk of bleeding complications.^382,383

The Role of Immune Cells

It has become clear that thrombi may have a major physiologic role in immune defense. This so-called immunothrombosis may aid in the recognition, containment, and killing of pathogens.³⁸⁴ However, if not controlled, it may contribute to thrombosis.³⁸⁵ Figure 113–30 provides an overview of immunothrombosis.

The tight link between immune host defense and thrombosis is further demonstrated by the fact that immune cells play an active role in clot formation through several processes. First, monocytes express tissue factor upon activation by pathogens, which leads to thrombin generation by the extrinsic pathway.³⁸⁶ Second, activated endothelium recruits neutrophils and, as a result of platelet-neutrophil interplay, neutrophil extracellular traps (NETs) can be formed,³⁸⁷ which consist of decondensated chromatin fibers and DNA released by neutrophils.³⁸⁸ NET-related proteins, such as histones and neutrophil elastase, activate platelets and inactivate TFPI, respectively, with both processes driving clot formation.^338,389 Moreover, NET DNA activates factor XII, resulting in enhanced thrombus formation mediated by the intrinsic pathway.³⁸⁵ NETs also interact with VWF, which facilitates platelet binding to NETs.

The decondensation of chromatin that contributes to NET formation is caused by PAD4 relocalization to the neutrophil nucleus where it citrullinates histones. This results in unwinding of the DNA, which subsequently bursts out of the neutrophil.³⁸⁸ The crucial role that PAD4-mediated NET formation may play in thrombosis has been shown in a physiologically relevant inferior vena cava thrombosis model in mice.^385,390 Consistent with this, NETs have been shown to be an integral component of human thrombi.³⁹¹ The notion that DNA is involved in pathologic thrombus formation opens doors to alternative ways of anticoagulation.

The Role of Endothelial Cells

Once a blood clot is formed to seal the injury, the clotting process must be terminated to avoid thrombotic occlusion in the adjacent, nonperturbed vascular bed. If the coagulation reactions are not controlled, clotting could occur throughout the entire vasculature, even after a modest procoagulant stimulus.

Endothelial cells play a major role in confining the coagulation reactions to the site of injury and preventing clot extension to areas where the endothelium is intact (Chap. 115). Endothelial cells have two major types of anticoagulant/antithrombotic activities, as illustrated in Fig. 113–31. First, the protein C/protein S/thrombomodulin/EPCR system is activated in response to thrombin generation. This was demonstrated by Hanson and colleagues who showed that thrombin infusion in vivo is anticoagulant in a protein C–dependent manner.³⁹² This indicates that thrombin present at sites other than that of vascular damage is able to generate APC, which subsequently inactivates the cofactors Va and VIIIa, thereby terminating or preventing downstream thrombin generation.

Second, the protease inhibitors AT and TFPI are bound to heparan sulfates expressed on the endothelial surface, where they can inactivate proteases.³⁹³ GPI-anchored TFPIβ may also play a role in controlling intravascular thrombin generation.³³² Furthermore, endothelial cells inhibit platelet activation by releasing the inhibitors prostacyclin (PGI₂) and nitric oxide (NO), as well as degrading adenosine diphosphate (ADP) by their membrane ecto-ADPase, CD39.³⁹⁴

Role of Plasma Protease Inhibitors

Circulating protease inhibitors are also critical for localizing the coagulation reactions to specific cell surfaces by directly inhibiting proteases that diffuse away from the site of clot formation. Not only are the plasma protease inhibitors key players in confining a clot to the proper location, their synergistic action imposes a threshold effect on the coagulation process.^363,395,396 Thus, in the presence of inhibitors, coagulation does not proceed unless procoagulant factors are generated in sufficient amounts to overcome the inhibitory mechanisms. In case the triggering event is inadequate, the system returns to baseline rather than continuing through the coagulation process. Under pathologic conditions, the trigger for clotting may be so strong as to overwhelm the control mechanisms, leading to disseminated intravascular coagulation or thrombosis (Chaps. 129 and 133).

ROLE OF FIBRINOLYSIS

Once a hemostatic clot has been formed and protected from fibrinolysis by the action of factor XIII and TAFI, some provision must be made for its eventual removal as wound healing takes place. Dissolution of clots is accomplished by the fibrinolytic system, as discussed in detail in Chap. 135.

The Concept of Basal Coagulation and Anticoagulation

The coagulation process only proceeds when enough thrombin is generated on or near the tissue factor–bearing cell to trigger activation of platelets and cofactors. One may wonder, however, if minute hemostatic plugs are not constantly formed throughout the body to maintain the integrity of the vasculature. Indeed, a low level of coagulation factor activation probably occurs at all times.³⁹⁷ More than 30 years ago, it was shown that fibrinopeptides are continuously cleaved from fibrinogen at low levels in normal individuals.³⁹⁸ Minute amounts of factor VIIa as well as the activation peptides of factors IX and X have also been demonstrated to circulate in the bloodstream of normal individuals.^399,400,401 This is known as basal coagulation or idling. Basal activation of coagulation factors likely results from minor injuries that occur during normal daily activities. The basal coagulation must be balanced by basal activity of the anticoagulation and fibrinolytic systems. This is evidenced by the presence of low levels of the protein C activation peptide and tissue plasminogen activator activity in normal individuals.⁴⁰²