Structural Biology Of Factor VIIa/Tissue Factor Initiated Coagulation

Abstract

Factor VII (FVII) consists of an N-terminal gamma-carboxyglutamic acid domain followed by two epidermal growth factor-like (EGF1 and EGF2) domains and the C-terminal protease domain. Activation of FVII results in a two-chain FVIIa molecule consisting of a light chain (Gla-EGF1-EGF2 domains) and a heavy chain (protease domain) held together by a single disulfide bond. During coagulation, the complex of tissue factor (TF, a transmembrane glycoprotein) and FVIIa activates factor IX (FIX) and factor X (FX). FVIIa is structurally “zymogen-like” and when bound to TF, it is more “active enzyme-like.” FIX and FX share structural homology with FVII. Three structural biology aspects of FVIIa/TF are presented in this review. One, regions in soluble TF (sTF) that interact with FVIIa as well as mapping of Ca²⁺, Mg²⁺, Na⁺ and Zn²⁺ sites in FVIIa and their functions; two, modeled interactive regions of Gla and EGF1 domains of FXa and FIXa with FVIIa/sTF; and three, incompletely formed oxyanion hole in FVIIa/sTF and its induction by substrate/inhibitor. Finally, an overview of the recognition elements in TF pathway inhibitor is provided.

2. Introduction

Human factor VII (FVII) is a vitamin K-dependent trace plasma protein; it is synthesized by hepatocytes and secreted into blood as a single chain molecule of Mr ∼50,000 (1,2). Starting at the N-terminus, FVII contains a gamma-carboxyglutamic acid-rich domain (Gla domain, residues 1-38), a short hydrophobic segment (residues 39-45) that often is considered as part of the Gla domain, two epidermal growth factor (EGF)-like domains (residues 47-84 and 85-131), and the C-terminal serine protease domain (residues 153-406) (3). FVII has negligible activity and participates in the extrinsic pathway of coagulation principally after its activation to FVIIa (4-7). A number of coagulation enzymes including FXa, FIXa, and FVIIa can activate FVII (8-11); however, FXa appears to be the most potent activator of FVII (12). Activation of FVII by each enzyme involves cleavage of a single peptide bond between Arg-c15{152} and Ile-c16{153}, located in the connecting region between the EGF2 and the protease domain. This results in the formation of a two-chain FVIIa molecule consisting of a light chain of 152 amino acids and a heavy chain of 254 amino acids held together by a single disulfide bond (between Cys-135 and Cys-262) (3).

Tissue factor (TF) is a membrane protein with structural homology to the class 2 cytokine receptor family (13). Human TF consists of an N-terminal cytoplasmic domain of 19 residues, a transmembrane domain of 23 residues and a C-terminal ectodomain domain of 219 residues. The ectodomain of TF consists of two fibronectin type-III repeats. The crystal structure of the C-terminal domain pair—soluble tissue factor (sTF)—is known (14-17); the two fibronectin type-III domains are connected end-to-end at an angle of 120 degrees (14-17). Similar to full-length TF, sTF binds FVIIa with high affinity and potentiates its enzymatic activity (18-20).

Binding of TF increases the amidolytic activity of FVIIa several-fold (21-23) by restructuring the active site region. In this regard, Higashi et al. (24) proposed that free FVIIa is structurally more zymogen-like and when bound to TF it is more active enzyme-like. Alanine scanning mutational studies of surface residues in the protease domain of FVIIa support this concept (25). Thus, TF binding induces the concerted structural rearrangements in FVIIa similar to those observed upon the proteolytic activation of trypsinogen to trypsin and related serine proteases (26). Further, these changes occur only when the free NH₂-group of the Ile-c16{153} is allowed to make a salt bridge with the carboxylate of Asp-c194{343} of FVIIa (24). TF cannot induce these changes in zymogen FVII in which Arg152-Ile153 bond has not been cleaved (27). Although TF binding to FVIIa increases its activity by several orders of magnitude, it does not appear to do so by forming the oxyanion hole; it is the substrate that induces formation of the oxyanion hole in FVIIa (28). In the absence of TF, the FVIIa regions that play an important role in preventing the transition from the zymogen-like to the active enzyme-like molecule, and how TF binding promotes this transition is an area of active investigation (29-33), which is discussed elsewhere in this issue (34).

Initiation of coagulation begins by exposure of blood to TF in the extravascular space at an injury site and formation of the Ca²⁺-dependent complex between TF and plasma FVIIa. The Ca²⁺/FVIIa/TF complex formed on the cell surfaces then activates both FX and FIX leading to thrombin generation and fibrin formation. In this reaction, FVIIa, FX and FIX (see below) anchor to the phospholipid (PL) bilayer through their Gla domains for optimal rates of FXa and FIXa formation. Consequently, in the absence of PL, the rates of activation of FX and FIX by FVIIa/sTF are relatively slow (18-20). TF pathway of coagulation is regulated by tissue factor pathway inhibitor (TFPI), a FXa-dependent inhibitor of the Ca²⁺/FVIIa/TF complex (35). The translated amino acid sequence of human TFPI (major form) cDNA reveals that it consists of a highly negative N-terminus, three tandemly repeated Kunitz-type domains with intervening linker regions, and a highly positive C-terminus (36). The mechanism of inhibition by TFPI involves its binding first to FXa through domain 2 and then to the TF/FVIIa complex through domain 1 (37, 38). The third Kunitz-type domain has no inhibitory activity and has other biologic functions that will not be discussed here.

Human FX and FIX share structural homology with FVII. FX circulates as a zymogen with a molecular weight of ∼59,000 and consists of a light chain (amino acids 1-139) and a heavy chain (amino acids 143-448) held together by a single disulfide bond between Cys-132 and Cys-c302 (3, 39). Upon activation, a single peptide bond in FX between residues Arg-c15{194} and Ile-c16{195} is cleaved with resultant formation of a serine protease, FXa, and release of a 52-residue activation peptide (3). The NH₂ terminus light chain of human FXa contains 11 Gla residues and represents the Gla domain (residues 1-39); the Gla domain is followed by a few aromatic residues (amino acids 40-45), and two EGF-like domains (EGF1 residues 46-84, EGF2 residues 85-128). The heavy chain contains the serine protease domain essential for catalysis and features the active site triad of His-c57{236}, Asp-c102{282}, and Ser-c195{379} (3).

FIX is a single chain protein and contains a Gla domain (residues 1-40), a short hydrophobic segment (residues 41-46), two EGF-like domains [residues 47-84 (EGF1) and 85-127 (EGF2)], an activation peptide region (residues 146-180), and a serine protease module (residues 181-415). Activation of FIX involves proteolytic cleavages at Arg145-Ala146 and Arg-c15{180} and Val-c16{181} bonds with a concomitant release of a 35-residue activation peptide (3, 40). The FIXa thus formed contains a light chain (residues 1-145) and a heavy chain (residues 181-415) held together by a single disulfide bond. The light chain consists of the Gla, EGF1, and EGF2 domains whereas the heavy chain contains the serine protease domain that features the catalytic triad of residues His-c57{221}, Asp-c102{269} and Ser-c195{365} .

In this review, we discuss the structural biology aspects of FVIIa/sTF interface and its interaction with the macromolecular substrates FIX and FX. The functional roles of Ca²⁺, Mg²⁺ and Na⁺, and the inhibitory role of Zn²⁺ in TF binding and catalysis are described. Unlike other serine proteases of the coagulation cascade, a unique feature of FVIIa bound to TF is the absence of oxyanion hole at the active site and its induction by the substrate. The importance of this mechanism in achieving selectivity of the physiologic substrate by FVIIa/TF is outlined. Lastly, an overview of the important recognition elements in the first and the second domain of TF pathway inhibitor (TFPI) is provided.

3. Molecular Recognition in Tf-Induced Coagulation

3.1. Structures of FVIIa/sTF in Ca²⁺ and in Ca²⁺/Mg²⁺

The formation of FVIIa/TF complex is the first step in the extrinsic pathway of blood coagulation. The crystal structures of human active-site inhibited FVIIa/sTF complex in the presence of Ca²⁺ only (41, PDB id 1DAN) and in the presence of Ca²⁺/Mg²⁺ (28, PDB id 2A2Q) are shown in Figure 1A and Figure 1B, respectively. In the 1DAN structure, the active site is occupied by D-Phe-Phe-Arg-chloromethylketone and the oxyanion hole is fully formed. In the 2A2Q structure, the active site is occupied by p-aminobenzamidine (pAB) and the oxyanion hole is incompletely formed; the significance of this observation is discussed later in section 3.8. In both structures, FVIIa molecule is in an extended form where the protease domain is at the top and the membrane binding Gla domain is at the bottom. In this complex, FVIIa is tightly bound to sTF through salt bridges, hydrogen bonds, ionic contacts and hydrophobic interactions. All domains of FVIIa make strong interactions with sTF. These are outlined in detail earlier (41). The observed interactions between the enzyme FVIIa and the cofactor sTF in the crystal structures (28,41) are consistent with the biochemical and experimental evidence (22,25,29,42-52).

Figure 1.

The cartoon representation of FVIIa/sTF complexes. The FVIIa/sTF structure of Banner et al., at 2.0 angstrom resolution (41) (PDB id 1DAN) in the presence of Ca²⁺ only is shown in A, and that of Bajaj et al., at 1.8 angstrom resolution (28) (PDB id 2A2Q) in the presence of Ca²⁺ and Mg²⁺ is shown in B. The locations of Na⁺ and Zn²⁺ sites are also shown in the 2A2Q structure. FVIIa consists of Gla (red), EGF1 (green), EGF2 (blue) and protease (cyan) domains, and the sTF (magenta) contains two fibronection type III domains. The different metal ions Ca²⁺(orange), Mg²⁺(green), Zn²⁺ (light blue) and Na⁺ (purple), bound to different domains of FVIIa are shown as spheres. Three Ca²⁺ ions in the GLA domain at positions 1, 4 and 7 (53) in the 1DAN structure are replaced by Mg²⁺ in the 2A2Q structure. The Ca²⁺ ion bound to the FVIIa EGF1 domain is labeled Ca8 and to the protease domain is labeled Ca9. In the 2A2Q structure Na⁺ ion is labeled as Na and the two Zn²⁺ ions labeled Zn1 and Zn2. The position of the active site Ser-195 is shown in ball and stick representation. The position of the C-terminal helix of the Gla domain (residues 34-45) of FVIIa is slightly tilted towards the TF2 domain in the 2A2Q structure resulting in a somewhat more favorable hydrophobic interactions between L13, F31 L39, and F40 of FVIIa with Y156, W158 and V207 of sTF. The residues interacting at the interface regions between FVIIa and sTF have been outlined in detail in an earlier report (41).

In the 1DAN structure (41), seven Ca²⁺ ions are bound to the N-terminal Gla domain in a linear fashion. In the 2A2Q structure (28), three of the seven Ca²⁺ ions at positions 1, 4 and 7 (53) are replaced by Mg²⁺. In both structures, a single Ca²⁺ ion is bound to the EGF1 domain as well as to the protease domain. In addition, two putative Zn²⁺ binding sites and one putative Na⁺ binding site have been identified in the protease domain of FVIIa in the 2A2Q structure. Further, significant flexibility exists in FVIIa domains in solution (54), which is lost upon binding to sTF. In contrast, sTF undergoes minimal conformational changes upon binding to FVIIa (55). This is consistent with the concept that cofactor TF stabilizes the conformation of enzyme FVIIa in a stable active form (24).

The fold of the Gla domain omega-loop in FVIIa is different in the Ca²⁺ only structure (41) verses the Ca²⁺/Mg²⁺ structure (28). The 2A2Q structure (28) was obtained under concentrations of Mg²⁺ that were supraphysiologic; however, folding of the omega-loop under 5 mM Ca²⁺/2.5 mM Mg²⁺ is similar to the 2A2Q structure, with identical positions for four Ca²⁺ and three Mg²⁺ (56, PDB id 3TH2). Further, the folding of the omega-loop under 45 mM Ca²⁺/5 mM Mg²⁺ is similar to that reported in the presence of Ca²⁺ only (41); however, positions 1 and 7 are occupied by Mg²⁺ (56, PDB id 3TH4). Thus, it would be appear that four Ca²⁺ and three Mg²⁺ ions are bound to the Gla domain in the circulating FVII/FVIIa. Moreover, circulating free FVII/FVIIa will need high Ca²⁺ concentrations to have its position four switched from Mg²⁺ to Ca²⁺.

Interestingly, Gla domain of activated protein C (APC) bound to endothelial protein C receptor under 10 mM Ca²⁺/10 mM Mg²⁺ has metal ion position four occupied by Ca²⁺ and only positions 1 and 7 occupied by Mg²⁺ (56,57; PDB id 3JTC and 1LQV). Similarly, the structure of the ligand bound Gla domain of FIX/FIXa under 5 mM Ca²⁺/2.5 mM Mg²⁺ has metal position four occupied by Ca²⁺ (58); and high concentrations of Ca²⁺ are needed to substitute Mg²⁺ at position four by Ca²⁺ in the ligand free FIX/FIXa molecule (59). Based upon these observations, we propose that vitamin K-dependent clotting and anticlotting proteins circulate in blood with central four Ca²⁺ ions bound to their Gla domains. The remaining three (or more in FIX and FX) external divalent metal binding sites in each Gla domain are occupied by Mg²⁺. The conformation of the omega-loop in each circulating vitamin K-dependent protein may not be optimal for binding to PL and is possibly achieved by switching Mg²⁺ at position four to Ca²⁺(60). Thus, the metal ion at position four regulates PL-dependent coagulation and anticoagulation reactions. In this context, Mg²⁺ potentiates PL-dependent physiologic reactions by occupying only the external divalent metal binding sites and none of the central Ca²⁺ sites (28,56,61,62). One should note that experimental evidence is needed to validate and substantiate this concept.

3.2. Modeled structures of FVIIa/sTF with FXa and with FIXa/FIX_alpha

The FVIIa/TF complex activates FIX (63) as well as FX (64), which leads to the activation of prothrombin to thrombin. In a reciprocal fashion, FXa and FIXa activate FVII/TF (2,8,9,12). Although primary sequences of FXa and FIXa are highly homologous, the interaction of FIXa with FVIIa/TF is weaker than FXa (65,66); further, as compared to FXa, FIXa is a poor activator of FVII/TF (12,67). Moreover, the affinities of FX-Gla and FX-EGF1 domains towards FVIIa/TF are stronger than the FIX-Gla and FIX-EGF1 domains (68). To explain these differences, computational models of the ternary complexes of FVIIa/sTF-FXa, FVIIa/sTF-FIXa and FVIIa/sTF-FIX_alpha have been proposed (67,69-72). In building these models, biochemical evidence was combined with newly available algorithms for protein-protein docking (73-75). In the absence of experimentally determined structures of FVIIa/sTF-FXa and FVIIa/sTF-FIXa or FVIIa/sTF-FIX_alpha, these models serve as reasonable basis for understanding how Gla and EGF1 domains of FIX/IXa and FX/Xa interact with FVIIa/TF.

In 2003, structural models of the FVIIa/sTF-FXa complex were proposed by two independent groups, namely, the Scripps group (70) and the Chapel Hill group (71). Based upon the methodology used to build these complexes, one is referred to as the Scripps static model (Ss model, 70) and the other as the original Chapel Hill solution-equilibrated model (CheA model, 71). Although different approaches were used, the two models have similar overall architecture and comparable residue interactions although with some significant variations. In general, each model was built in three stages using different protein-protein docking algorithms. In stage 1, the Protease-EGF2 domains of FXa were docked onto FVIIa/sTF. In stage 2 and stage 3, EGF1 domain and the Gla domain were docked onto FVIIa/sTF, respectively. At each stage the models were screened and those that corraborated the experimental evidence were selected for the next stage of modeling. The overall architecture of FVIIa/sTF-FXa is shown in Figure 2A. In both models, TF is sandwiched between the extended forms of FVIIa and FXa. Recently, both Ss and CheA model structures (70,71) were subjected to 10 ns all-atom molecular dynamic simulations to obtain solution-equilibrated models (72). The resulting models are referred to as Scripps solution-equilibrated (Se) and current Chapel Hill solution-equilibrated (CheB) models (72). In all models, the Gla domain of FXa interacts with TF domain as well as with Gla domain of FVIIa. The EGF1 domain of FXa interacts only with TF in the Ss and Se models (70,72), whereas it interacts with TF and FVIIa in the CheA and CheB models (71,72). Further, in all models, the EGF2 domain and the serine protease (SP) domain of FXa interact with the serine protease domain of FVIIa and no interaction is noted with TF.

Figure 2.

Cartoon representations of modeled structures of the ternary complexes. Structure of the FVIIa/sTF-FXa model is shown in A, structure of the FVIIa/sTF-FIXa model is shown in B, and structure of the FVIIa/sTF-FIX_alpha model is shown in C. FIX_alpha represents FIX cleaved at Arg145-Ala146 peptide bond, whereas FIXa represents FIX cleaved at Arg145-Ala146 and Arg-c15{180} -Val-c16{181} peptide bonds. The FVIIa/sTF-FXa model (PDB id 1NL8) is from Norledge et al (70) and the energy minimized FVIIa/sTF-FIXa model (67) is based upon 1NL8 as the template. The FVIIa/sTF-FIX_alpha model is from Chen et al (69). As in figure 1, FVIIa Gla, EGF1, EGF2, and the protease domains are colored red, green, blue and cyan, respectively. The sTF is shown in magenta. The Gla, EGF1, EGF2 and protease domains of FXa, FIXa and FIX_alpha are colored deep teal, salmon, orange, and yellow, respectively. The active site residue (Ser-195) of FVIIa protease domain is shown in the space filling representation. The Ca²⁺ ions bound to FVIIa are shown as orange spheres, whereas the Ca²⁺ ions bound to FXa, FIX_alpha and FIXa are shown as green spheres. Note that the Gla domains of FVIIa, FXa and FIXa are expected to bind Mg²⁺ at specific sites under physiologic conditions. These sites have been replaced by Ca²⁺ during model building.

We used FVIIa/sTF-FXa as a template (1NL8, Ss model) to build the FVIIa-sTF-FIXa model (66). Since the complete structure of FIXa is not available, we modeled the FIXa using the coordinates of FIXa Gla (58), EGF1 (76) and the protease-EGF2 domains (77). The FIXa domains were structurally aligned to the respective domains of FXa in the FVIIa/sTF-FXa complex using the program ‘O’ (78). To remove the steric clashes and the short contacts, the entire FVIIa/sTF-FIXa complex was subjected to minimization using CHARMM (79). The resulting FVIIa/sTF-FIXa model is shown in Figure 2B. The model of FVIIa/sTF-FIX_alpha complex (69) is shown in Figure 2C. Similar to FXa, the Gla domain of FIXa/IX_alpha in FVIIa/sTF-FIXa or FVIIa/sTF-FIX_alpha interacts with the Gla domain of FVIIa and TF, whereas the EGF1 domain interacts with TF only. However, the interactive residues in the FVIIa/sTF-FIXa and the FVIIa/sTF-FIX_alpha models are different. Based on these computational models, we outline the possible mode of interactions between FVIIa/sTF and FXa/FIXa/FIX_alpha in the next three sections.

3.3. Interactions of FVIIa/sTF with the Gla Domains of FX/Xa and FIX/IXa

The C-terminal region of TF interacts with the Gla domains of FX/Xa (22,80-83) and FIX/IXa (66,67,69). Further, the Gla domains of FX/Xa (68,70-73) and FIX/IXa (66,69) also interact with the Gla domain of FVIIa. The possible modes of interaction between FVIIa/sTF with the Gla domains of FXa (Ss model), FIXa and FX_alpha are shown in Figure 3 A, B and C, respectively. Although the structures of FXa and FIXa are similar, the variations in their sequences at the interacting regions with sTF and with the Gla domain of FVIIa make associations different from each other. The residues 9 to 12 (KKGH) in the Gla domain of FXa are basic, whereas residues 10 to 13 in FIXa are hydrophobic and neutral (VQGN). The residues K10 and H12 in FXa make salt bridge with D33 of FVIIa. In comparison, Q11 in FIXa makes weak hydrogen bond with D33 of FVIIa. However, K38 of FVIIa makes a salt bridge with Gla14 in FXa and similarly with Gla15 in FIXa. Overall, the sequence variations in this region make interactions weak for FIXa.

Figure 3.

Modeled interactions of FVIIa/sTF with the Gla domains of FXa, FIXaand FIX_alpha. The interacting residues of FVIIa-Gla (red), sTF (magenta) and FXa-Gla (deep teal) are shown in stick representation. The carbon atoms are green in FVIIa-Gla, magenta in sTF, and deep teal in the Gla domains of FXa, FIXa and FIX_alpha. The N and O atoms are blue and red, respectively. The hydrogen bonds are shown as black dashed lines. A) The modeled interaction of FXa-Gla with sTF and FVIIa-Gla. FXa-Gla:sTF—Gla32:K166, D35:R200, Gla39:K201 and K36:T167; FXa-Gla:FVIIa-Gla—K10:D33, H12:D33 and Gla14:K38. B) The modeled interaction of FIXa-Gla with sTF and FVIIa-Gla. FIXa-Gla:sTF—Gla33:K166, Gla36:R200, R37:D204/T167, and Gla40:R200/K201; FIXa-Gla:VIIa-Gla—Q11:D33 and Gla15:K38. C) The modeled interaction of FIX_alpha-Gla with sTF and FVIIa-Gla. FIX_alpha-Gla:sTF—Gla15:N184, C23:S162, Q44:D180/201 and Y45:S163; FIX_alpha-Gla:FVIIa-Gla—Gla7:K32; M19:R36, Gla20:F31/I30 and Gla21:R28/K32.

The C-terminal helix of the Gla domains in FX/Xa and FIX/IXa interact with the C-terminal domain of TF and make no contact with FVIIa. In the Ss model, FXa residues Gla32, D35, and Gla39 make salt bridges with TF residues K166, R200, and K201, respectively (Figure 3A). In addition, K36 of FXa makes hydrogen bond with T167 of TF. The C-terminal helix of FIXa Gla domain interacts with TF in a similar fashion. In FIXa, corresponding residues Gla33, Gla36, and Gla40 make salt bridges with TF residues K166, R200, and K201, respectively. In addition, R37 of FIXa can make a hydrogen bond with either T167 or D204 of TF (Figure 3B). Mutagenesis studies support the role of TF residues R200, K201, T167 and K165 in its interaction with the Gla domain of FXa/FIXa (80-83). Thus, comparable interactions exist between the C-terminal helix of the Gla domains of FXa or FIXa with TF.

Interactions of the Gla domain of FXa with TF and FVIIa in the CheA/CheB models (71,72) are somewhat different from the Ss/Se models (70,72). The important differences are—a) ionic interactions between K165 of TF and D35 of FXa present in the CheA/CheB models and absent in the Ss/Se models; and b) ionic interactions between R36 of FVIIa and Gla14 of FXa present in the Ss/Se models and absent in the CheA/CheB models. By mutagenesis, residue R36 in Gla domain of FVIIa has also been identified as one of the key residues in substrate recognition of FX (84). However, our recent FVIIa/sTF crystal structure at 1.72 angstrom reveals that R36 in FVIIa is involved in network of hydrogen bonds with TF (56). Further in this structure, the TF region containing residues K165 and K166 are well defined and it appears that only one of these residues side chain can interact with FXa. Moreover, hydrophobic interactions between the Gla domains of FVIIa and FXa are more prominent in the CheA/CheB models as compared to the Ss/Se models. Clearly, experimentally determined structure of FVIIa/sTF-FXa is needed to resolve these discrepancies.

As compared to FIXa, the mode of interaction between the Gla domain of FIX_alpha and the FVIIa/sTF is considerably different. The FIX_alpha Gla residues that interact with FVIIa/sTF are shown in Figure 3C. These interactions involve possible hydrogen bonds between residues Gla7, Gla15, M19, Gla20, and Gla21 of FIX_alpha with R28, I30, F31, K32, and R36 of FVIIa and N184 and S162 of TF. The Gla domain C-terminal helix residues Q44 and Y45 of FIX_alpha also interact with residues S163, D180 and K201 of TF via hydrogen bonds. In general, FIX_alpha versus FIXa Gla domain interacts strongly with FVIIa/sTF. Again, experimentally determined structures are needed to resolve these anomalies.

3.4. Interactions of FVIIa/sTF with the EGF1 Domains of FX/Xa and FIX/IXa

The experimental evidence indicates that EGF1 domain participates in the complex formation of FVIIa/sTF with FIX (66,85) and FX (66,70,86), and in the reciprocal activation of FVII/TF (67). In FIX/FIXa, side chains of residues E52, N54 and N58 interact with complimentary side chains of K15/N107, N199/E105 and E99 in sTF, respectively (Figure 4). Similarly in the Ss model, side chains of residues E51, S53 and N57 in FX/Xa interact with side chains of K15, E105 and E99 in sTF, respectively. In addition, side chains of T52, Q58 and K60 in FX/Xa interact with side chains of T13, T101 and T197 in sTF, respectively. Carbonyl oxygen of C50 in FXa may also interact with N199 side chain nitrogen of sTF (Figure 4). Thus, based on the Ss model, sTF interacts stronger with EGF1 domain of FX/Xa than of FIX/IXa.

Figure 4.

Modeled sTF interactions with the EGF1 domain of FXa, FIXaand FIX_alpha. The interacting residues between sTF (magenta) and EGF1 (wheat) are shown in stick representation. The carbon atoms are magenta in sTF, and wheat in FVIIa-EGF1. The N and O atoms are blue and red, respectively. The black dashed lines represent hydrogen bonds between the interacting residues. A) The modeled interaction between FXa-EGF1 domain and sTF. FXa-EGF1:sTFs—C50:N199, E51:K15, T52:T13, S53:E105, N57:E99, Q58:T101 and K60:T197 B) The modeled interaction between FIXa-EGF1 domain and sTF. FIXa-EGF1:sTFs—E52:K15/N107, N54:N199/E105 and N58:E99. C) The modeled interaction between FIX_alpha-EGF1 domain and sTF. FIX_alpha-EGF1:sTFs—D49:K165, N54:K169, K63:Y157 and D64:K166.

The mode of interactions of EGF1 domain of FX/Xa in the CheA/CheB models is distinct from the Ss/Se models. In the CheA/CheB models, the E51 of FXa interacts with K201 of sTF (72); the residue E51 of FXa interacts with K15/N107 of sTF in the Ss model. None of the other residues of FXa proposed in the Ss/Se models interact with sTF in the CheA/CheB models. In the CheB model, G66 of FXa interacts with E24 of sTF, E77 of FXa interacts with K41/E99 of sTF, and K79 of FXa interacts with E105 of sTF. In addition, E74 of FXa interacts with the K170D of the protease domain of FVIIa. Thus, EGF1 domain of FXa interacts with sTF and FVIIa in the CheA/CheB models.

In the FVIIa/sTF-FIX_alpha model, the mode of interaction is different from the FVIIa/sTF-FIXa model. The N54 in FIX_alpha interacts with K169 of sTF, whereas N54 in FIXa interacts with N199 of sTF. As compared to FIXa, all other residues in FIX_alpha that are proposed to interact with sTF are different (Figures 4B and 4C). In FIX_alpha, D49, D64, and K63 interact with K165, K166 and Y157 of sTF, respectively.

3.5. Interactions of FVIIa-protease Domain with the Protease Domains of FXa, FIXa and FIX_alpha

The possible mode of interactions between FVIIa protease domain with the protease domains of FXa, FIXa and FIX_alpha are shown in Figure 5 A, B and C, respectively. In FVIIa/sTF-FXa Ss model (Figure 5A), the side chains of Q-c20{200}, K-c23{203}, K-c134{318}, K-c186{371}, R-c202{387}, K-c204{389} and D-c205{390} from FXa interact with main and or side chains of L-c153{295}, T-c151{293}, D-c186{334}, E-c154{296}, S-c188A{336}, D-c186{334} and D-c146{289}/R-c147{290} from FVIIa, respectively. In addition, main chain oxygen of V-c160{343} from FXa interact with side chain nitrogen of K-c20{157} from FVIIa. Notably, R-c147{290} in FVIIa is also implicated to be important in FX activation (87).

Figure 5.

Modeled FVIIa-protease domain interactions with the protease domains of FXa, FIXa and FIX_alpha. The interacting residues between FVIIa-protease domain (Cyan) and the protease domains of FXa, FIXa and FIX_alpha (yellow) are shown in stick representation. The carbon atoms are cyan in FVIIa, and yellow in FXa, FIXa and FIX_alpha. The N and O atoms are blue and red, respectively. The hydrogen bonds are shown as dashed lines. A) The modeled interaction between FXa protease domain and FVIIa protease domain. FXa-protease:FVIIa-protease—Q20:L153/E154; K23:T151; K134:D186; V160:K20; K186:E154; R202:S188A; K204:D186; and D205:D146/R147. B) The modeled interaction between FIXa protease domain and FVIIa protease domain. FIXa-protease:FVIIa-protease—G19:E154; K23:T151; R159:G19; V160:K20 and E204:D186/A221A. C) The modeled interaction between FIX_alpha protease domain and FVIIa protease domain. FIX_alpha-protease:FVIIa-protease—Q26:R147; R159:G19; E186:K24; R188A:D72/E154. The P1 residue R15 in FIX_alpha is shown to interact with D189 and S190 of FVIIa. The typical hydrogen bond observed between the main chain nitrogen of P1 residue (R15 in FIX_alpha) and the main chain oxygen of residue S214 in FVIIa is also shown. The chymotrypsin numbering scheme is used to label the residues. The insertion residues are labeled with the number followed by an alphabet A, B, etc.

Predicted interactions of FVIIa-protease domain with FXa in the Se model are shown in Table 1 (72). Comparative interactions in the CheB are also shown in table 1 (72). Clearly, there are differences between the Se and CheB models. Moreover, only three interactions involving K20, D186, and S188A of FVIIa protease domain and V160, K134 and R202 of FXa protease domain are common between the Ss and the Se model.

Table 1. Predicted interactions of FVIIa protease domain with FXa in the Se and the CheB models (72). M, Main chain; S, Side chain; HI, Hydrophobic interaction; LC, light chain.

Residues in Fxa, Se Model	Interaction	Residues in VIIa	Interaction	Residues in Fxa, CHeB Model
E159	SS	K20	MS	Y162
Y185	HI	V21	HI	P161
		V21	HI	Y185
T185B	SS	E26
K186	Ionic	D72
		E75	Ionic	K223
I137	HI	L145
M157	HI	L145
Y207	HI	L145
R202	Ionic	D146
E138LC	Ionic	R147
E138LC	SS	R147
Q20	SM	L153
K186	Ionic	E154	SS	Y185
		E154	Ionic	K186
		E154	SS	K186
D92LC	Ionic	R170C
D95	SS	R170C
		K170D	SM	Glu74LC
		K170D	Ionic	D92LC
		K170D	Ionic	D95LC
		K170D
K134	Cation-Pi	Y184
K134	Ionic	D186
R202	SM	S188A	SM	K204

In FVIIa/sTF-FIXa complex, G-c19{184}, K-c23{188}, R-c159{327}, V-c160{328} and E-c204{374} from FIXa make hydrogen bonds with E-c154{296}, T-c151{293}, G-c19{156}, K-c20{157} and D-c186{334}/A-c221A{369} from FVIIa, respectively. In the FVIIa/sTF-FIX_alpha model, the exosite residues Q-c26{191}, R-c159{327}, E-c186{355} and R-c188A{358} from FIX_alpha interact with R-c147{290}, G-c19{156}, R-c147{290} and D-c72{212}/E-c154{296} of FVIIa. There are obvious differences between the FIXa and FIX_alpha interactions with FVIIa. In addition to the inherent accuracies of the models, the apparent differences also stem from the fact that FIXa is in the enzyme form and the FIX_alpha is in the zymogen form.

From the foregoing, it is clear that there are limitations to the accuracy and reliability of the modeled complexes. Despite these limitations, they provide useful information to design mutagenesis experiments. Experimentally determined structures are needed to precisely define the interactive regions in the ternary complexes. In fact, how protease domains of zymogen FX and FIX interact with FVIIa/sTF should be the primary defining interactions between the FVIIa/sTF and FX/FIX. Further, we know from the K_M data (65,66) that FX and FIX interact to FVIIa/TF with equal affinity whereas FXa and FIXa interact differently. Knowing how zymogens FX and FIX interact with FVIIa/TF will answer these lingering questions. One also needs to know, how FXa and FIXa interact with FVII/TF in the reciprocal activation.

3.6. Interface between EGF2/Protease domains of FVIIa with sTF

The protease domain of FVIIa has a fold similar to that of other serine proteases. It consists of two beta-barrel structures with three alpha-helical segments. Based upon the experimentally determined structures (PDB id 1DAN and 2A2Q), the interface between the EGF2 and the protease domains of FVIIa and sTF is shown in Figure 6. Residues from all three alpha-helices of FVIIa protease domain make hydrogen bond with N-terminal domain of TF. These involve residues F-c129F{276}, R-c134{277}, T-c165{307}, Q-c166{308}, D-c167{309}, and R-c230{379} of FVIIa with S39, D44, W45, E91, Y94 and N96 of TF. In addition K85 of EGF2 domain makes hydrogen bond with D61 of TF.

Figure 6.

FVIIa-protease, FVIIa-EGF2 and sTF interface interactions. Interactive residues at the interface region of FVIIa protease domain (cyan), EGF2 domain (salmon) and sTF (magenta and yellow) are shown in stick representation. For clarity, either the side chain or main chain atoms that are involved in hydrogen bonding interactions are shown. The hydrogen bonds are represented by black dashed lines. Blue and red represent nitrogen and oxygen atoms, respectively. The residues N95, Y101, C102, D104, R113, H115 and G136 from EGF2 domain interact with P120, L123, E125, R129B and T129C of the protease domain. The residues S39, D44, W45, E91, Y94 and N96 of sTF interact with R134, T165, D167, Q166 and R230 of FVIIa-protease domain. The residue K85 of FVIIa-EGF2 domain makes hydrogen bond with D61 of sTF. Between the FVIIa-EGF2 and sTF, the hydrophobic interactions are more dominant (not shown). The chymotrypsin numbering scheme is used to label the residues.

The loop residues on the either side of the c130 and c160 helical segments of FVIIa interact with TF as well as with the EGF2 domain of FVIIa. The key residues identified to be important in increasing the activity of FVIIa were found in this interface region. The side chain of residue M-c164{306} implicated in allosteric activation of FVIIa is snuggly fitted into the hydrophobic pocket comprised of F76, P92 and Y94 of sTF (28,41). This hydrophobic pocket is lined by sTF residues R74 and N96. Substitution of M-c164{306} by Asp essentially abolishes the TF induced allosteric activation of FVIIa (88). This may be, in part, due to interaction of Asp-c164{306} of FVIIa with side chain of N96 of sTF (88); alternatively Asp-c164{306} could make hydrogen bond with R74. In either case, this could alter the c160 loop in FVIIa and impair its allosteric activation. Moreover, substitution of R74 by Ala in sTF (89) could affect the F76/P92/Y94 hydrophobic pocket and disallow the proper fixation of M-c164{306} in FVIIa. Thus, R74A mutant of sTF appears to have an indirect effect on FVIIa allosteric activation rather than playing a direct role.

The EGF2 domain strongly interacts with the FVIIa-protease domain through the hydrogen bonds and hydrophobic interactions. The residues involved in hydrogen bond formation are shown in Figure 6. EGF2 domain residues C102, D104, N95, Y101, R113, H115, and G136 interact with T-c129C{272}, R-c129B{271}, E-c125{265}, P-c120{260}, and L-c123{263} in the protease domain. In addition, the EGF2 domain stabilizes the FVIIa protease domain by shielding the hydrophobic residues on the surface of the protease domain. In the absence of such contacts, the shielded hydrophobic surface of the protease domain will be exposed to the solvent, which could lead to destabilization of the protease domain. Hence, these two domains are considered to represent a single structural unit.

3.7. Na⁺-site and Zn²⁺-sites in the protease domain of FVIIa and their functional roles

In the FVIIa/sTF complex, Na⁺ ion was located in the region involving c184{332} -c191{340} and c220{368} -c225{374} loops; this region is homologous to the proposed site in FIXa, FXa and APC but not to thrombin (28). Na⁺ is coordinated to the backbone carbonyl oxygens of Tyr-c184{332}, Ser-c185{333}, His-c224{373}, Thr-c221{370} and two water molecules (Figure 7A). Further, one of the water molecules is hydrogen bonded to Asp-c189{338} . Since there is a three residue insertion in the c183 loop of thrombin (90), Na⁺ site in thrombin is coordinated to the carbonyl groups of c221 and c224, and four water molecules (91). Binding of Na⁺ to FVIIa modestly increases its affinity for TF (28). This is consistent with the observations that the Na⁺-site in FXa and FIXa increases the affinity for FVa and FVIIIa, respectively (92,93). Further, similar to FXa binding to FVa (92) and FIXa binding to FVIIIa (93), binding of FVIIa to TF diminishes the effect of Na⁺ on its activity (94,95). Notably, zymogen forms of thrombin, FXa, FIXa, and FVIIa each lack Na⁺-site and is developed during conversion to their respective enzymatic form. Thus, Na⁺ binding to these enzymes contributes to their structural stability.

Figure 7.

The Ca²⁺- and putative Na⁺- and the Zn²⁺-sites in the protease domain of FVIIa. A) The Na⁺-site in FVIIa protease domain. The backbone carbonyl oxygens of T221 and H224 from the c220 loop, and Y184 and S185 from the c184 loop serve as ligands for Na⁺. The side chain of D189 is linked to Na⁺ through a water molecule. Na⁺ and water molecules are shown as purple and red spheres, respectively. The Na⁺ ion coordination to its ligands is shown by dashed lines. B) The Ca²⁺ site and the Zn²⁺ sites in FVIIa-protease domain. The residues H76, S82 and E80 for the first Zn²⁺ (Zn1), and K24, G69, D79, and H117 for the second Zn²⁺ (Zn2) serve as the protein ligands. The residues E75, D72, E80 and E70 serve as protein ligands for Ca²⁺. Note that the Zn²⁺-sites (light gray spheres) are located on each side of the Ca²⁺-site (green sphere). The coordinated water molecules are shown as red spheres. The black dashed line represents the coordination of metal ions to their respective ligands.

Two putative Zn²⁺ sites have been identified in the protease domain of FVIIa (28,96). In an earlier report, Zn²⁺ increased the binding of FVIIa to TF expressed on the bladder carcinoma J82 cells (97). However, in subsequent direct binding studies, it was found that Zn²⁺ reduces the affinity of FVIIa for sTF (96). Importantly, Zn²⁺ inhibits the proteolytic activity of FVIIa more potently than FVIIa/TF (96,97). Mutagenesis experiments and modeling studies predicted that the residue His-c76{216} and His-c117{257} are involved in Zn²⁺ binding to the protease domain of FVIIa (96). The residues His-c76{216}, Glu-c80{220}, Ser-c82{222} and three water molecules provide ligands for the first Zn²⁺-site and residues His-c117{257}, Lys-c24{161}, Asp-c79{219}, Gly-c69{209} and two water molecules provides ligand for the second Zn²⁺-site (28). Notably, these interactions are unique for FVIIa. Further, the two Zn²⁺-sites are located on each side of the protease domain Ca²⁺-site (Figure 7B). The side chain of residue Glu-c80{220} provides coordination for Zn1 and Ca²⁺, whereas Zn2 and Ca²⁺ are interconnected through a network of hydrogen bonds involving two water molecules (Figure 7B). Overlapping Ca²⁺ and Zn²⁺ -sites, in part, could provide explanation for experimental observations that the inhibitory effect of Zn²⁺ is substantially reversed by Ca²⁺ (96). In this regard, inhibition by Zn²⁺ is partially overcome by the allosteric activation and stabilization of the FVIIa protease domain induced by binding of TF to residues c164{306} -c167{309} and Ca²⁺ binding to residues E-c70{210} -E80{220} (96).

3.8. Incompletely formed oxyanion hole in FVIIa/sTF and its induction by substrate/inhibitor

A defining feature of each serine protease, including those contained divergent folds from the chymotrypsin/trypsin, subtilisin, and the carboxypeptidase families, is the presence of an oxyanion-binding site, universally known as the oxyanion hole (98). In the chymotrypsin/trypsin like proteases, Gly-c193 and Ser-c195 backbone amide nitrogens form this binding site. The neculeophilic attack by the hydroxyl group of Ser-c195 on the carbonyl carbon atom of the substrate changes the geometry around this carbon from trigonal planar to tetrahedral. This tetrahedral transition state intermediate is intrinsically unstable because of the negative charge on the peptide carbonyl oxygen atom. However, in serine proteases this charge in the oxyanion hole is stabilized by H-bonds with the amide NH groups of Ser-c195 and Gly-c193 in a site usually referred to as the oxyanion hole. These interactions result in preferential binding of the substrate in the transition state, a necessary requirement for enzyme catalysis.

Although the presence of an oxyanion hole is an essential feature of serine proteases and its geometry is not disrupted by inclusion of benzamidine or pAB at the bottom of S1 site (26,98,99), it is absent in pAB-FVIIa/sTF (PDB id 2A2Q) or the benzamidine-FVIIa/sTF(PDB id 2AER) structures (28). This is shown in Figure 8A. The absence of an oxyanion hole in FVIIa/sTF was initially observed with an amidinophenylurea-based inhibitor (100). However, this was attributed to the unusual nature of the inhibitor, in which the amidino NH groups of the inhibitor induced ∼180 degree rotation of the c192-c193 peptide bond such that the interactions between the NH groups of the inhibitor could occur with the c192 main chain C=O group (100), Thus, a normal conformation of the c192-c193 peptide bond would be changed to the unexpected nonstandard one. Flipped peptide bond between Lys-c192 and Gly-c193 of FVIIa has also been observed by Olivero et al. (101) and by Zbinden et al. (102). In these structures, nitrogen at the para position of benzamidine moiety of the sulfonamide inhibitor (101) or the phenylglycine amide inhibitor (102) makes hydrogen bond with the main chain carbonyl group of Gly-c192 and the hydroxyl group of Ser-c195. Based upon the benzamidine-VIIa/sTF (PDB id 2AER) and the pAB-FVIIa/sTF (PDB id 2A2Q) crystal structures, one can safely conclude that the abnormal conformation of the c192-c193 peptide bond in FVIIa/Stf is not related to the presence of the unusual inhibitors in the active site; instead it is an inherent property of the FVIIa active site.

Figure 8.

Substrate induced formation of the oxyanion hole in FVIIa protease domain. A) Absence of oxyanion hole in the FVIIa/sTF structure with pAB (PDB id 2A2Q). The hydroxyl group of S195 makes hydrogen bond with the para amino group of pAB. The amide nitrogen of G193 makes hydrogen bond with O_delta of Q143, which prevents formation of the oxyanion hole. The residues 192-195, and Q143 are shown in stick representation. The carbon atoms of protease residues are colored salmon, whereas the carbon atoms in the inhibitor are colored yellow. The ‘N’ and ‘O’ atoms are colored blue and red, respectively. The hydrogen bonds are shown with black dashed lines. B) The fully formed oxyanion hole in FVIIa/sTF with d-FPR (PDB id 2FIR). Note that there is a 180° flip in the peptide bond between K192 and G193 residues as compared to the structure with pAB shown in ‘A’. Now, the carbonyl oxygen of K192 makes hydrogen bond with the backbone amide nitrogen of Q143, and the O_delta of Q143 makes hydrogen bond with R147 (not shown). C) The fully formed oxyanion hole in free FVIIa with benzamidine and sulfate ion in the active site (PDB id 1KLI). The negatively charged sulfate ion in the active site is hydrogen bonded to the amide nitrogen of G193; this interaction induces the formation of oxyanion hole in this structure. D) The K192-G193 peptide bond in standard conformation after removal of the sulfate ion from the FVIIa active site (PDB id 1KLJ). The crystals were obtained under the same condition as that in panel ‘C’; however benzamidine and the sulfate ions were removed by washing the crystal in the buffer lacking benzamidine and sulfate.

Importantly, TF does not increase the activity of FVIIa by inducing the formation of the competent oxyanion hole. It is the occupancy by an oxyanion of the substrate (or substrate analogue) at the active site that provides energy for 180 degree flipping of the c192-c193 peptide bond in FVIIa/sTF crystals. Thus, when the benzamidine-VIIa/sTF crystals were soaked with either EGRck (PDB id 2B8O) or with D-FPRck (PDB id 2FIR), the benzamidine was displaced by the EGR or the DFPR; consequently, the standard conformation of the c192-c193 peptide bond was observed. The D-FPR-FVIIa/sTF active site configuration is shown in Figure 8B. Furthermore, in the benzamidine-FVIIa/sTF crystals the 192-193 conformation is stabilized by the hydrogen bond between Gly-c193 N and the side chain of Gln-c143. In contrast, in the EGR-FVIIa/sTF, d-FPR-FVIIa/sTF, D-FFR-FVIIa/sTF, and the BPTI mutant- FVIIa/sTF (103) crystals, the c192-c193 conformation is stabilized by the hydrogen bond between main chain oxygen of c192 and the main chain N of Q-c143.

The structures (PDB id 1KLI and 1KLJ) of the EGF2/protease domain of FVIIa (104) obtained in the absence of TF support the concept that substrate/inhibitor transition state analog can induce the formation of oxyanion hole in FVIIa. In this structure, the bottom of the S1 site is occupied by benzamidine and top of the S1 site is occupied by a sulfate ion (104). Here, the sulfate anion occupies the position normally occupied by the oxyanion of the carbonyl group of the P1 substrate/inhibitor residue in the transition state. This is shown in Figure 8C. Similar to the negatively charged oxygen in the tetrahedral transition state intermediate during catalysis, oxygen of the sulfate ion in this structure makes a hydrogen bond with amide nitrogen of Gly-c193. Thus, it is quite conceivable that occupancy of sulfate ion in the S1 site of FVIIa induced 180° flip of the c192-c193 peptide bond during crystallization. Moreover, as in the structures of EGR-FVIIa/sTF, D-FPR-FVIIa/sTF, D-FFR-FVIIa/sTF, and mutant BPTI-FVIIa/sTF, carbonyl oxygen of Lys-c192 in the 1KLI structure makes hydrogen bond with the main chain nitrogen of Gln-c143. Upon removal of the sulfate ion by washing the crystal under sulfate free condition, the standard conformation of the oxyanion hole was maintained (Figure 8D). This could be due to stabilization by the hydrogen bond between carbonyl oxygen of c192 and the main chain nitrogen of c143. These observations lend support to a concept that the oxyanion hole in FVIIa is induced by the substrate/inhibitor and not by TF.

The Gla and EGF1 domains of FIX and FX represent the primary recognition determinants in binding to the FVIIa/TF and formation of the ternary complex (66,67,69-71,81-86,105). In the formed ternary complex, the scissile peptide bond sequence in FIX or FX then approaches the active site cleft in FVIIa and induces the formation of the oxyanion hole for efficient proteolysis. This exosite recognition mechanism between the enzyme and the substrate provides unique selectivity in activation of FIX and FX by FVIIa/TF under physiologic conditions in the presence of other plasma proteins.

3.9. Interactions of FVIIa with Kunitz domain-1 of TFPI and of FXa with Kunitz Domain-2 of TFPI

TF induced coagulation is primarily regulated by TFPI. The N-terminal Kunitz domain-1 of TFPI inhibits FVIIa and the Kunitz domain-2 inhibits FXa. The residues of FVIIa that interact with Kunitz domain-1, and of FXa that interact with Kunitz domain-2 have been detailed earlier (35,103,106). Among others, Kunitz domain-1 residues that are important for interaction with FVIIa appear to be D11, R20, and E46. Residue D11 interacts with R147 and K192 of FVIIa; R20 interacts with D60 of FVIIa, and E46 interacts with K60A/K60C of FVIIa. Similarly, among other residues, Kunitz domain-2 residues that are important for interaction with FXa appear to be Y17, R32, and E46. Residue Y17 fits into a hydrophobic cavity with possible interactions with R143 and Q151 of FXa; R32 interacts with E39 of FXa, and E46 interacts with K62 of FXa. The TF residues that interact with FXa in the FVIIa/sTF-FXa ternary complex appear to be also involved in the FVIIa/sTF-FXa-TFPI quaternary complex (107). The precise interactions between the Kunitz domain-1 and FVIIa protease domain and between the Kunitz domain-2 and FXa protease domain must await until the experimentally determined structures are available.

4. Perspective and Future Directions

At the time of writing this review, we know the experimentally determined structures of only FVIIa with or without sTF. Further all structures are with active site inhibited FVIIa. Thus how FVIIa become more active when bound to sTF remains difficult to define precisely. To obtain insight in to this observable fact, structures of the protease domain of FVIIa plus/minus sTF without occupancy of the active site with an inhibitor are needed. Experimentally determined structure of the zymogen form of FVII also remains to be determined. These are daunting tasks and challenges that need to be undertaken. Further, sodium site and the zinc sites in FVIIa need to be confirmed by anomalous data collection. Crystals of ternary complexes of FVIIa with FIX and FX are to be obtained in order to understand the differences in affinity of VIIa/TF for FIX/IXa and FX/Xa and their interactive sites. Again, this may be a difficult task since FVIIa will cleave FIX and FX when zymogens are used. One way is to use Ser-c195Ala mutant of FVIIa for such studies. Recently, in addition to calcium, magnesium has been implicated in promoting the TF-induced coagulation that has not been addressed structurally. Finally, to understand the molecular regulation in TF-induced coagulation, one needs to obtain crystals of FVIIa/FXa/TFPI/sTF complex and solve its structure.

Abbreviations used are

F

Factor

FVIIa

Factor VIIa

FIXa

Factor IXa

FIX_alpha

Factor IX_alpha

FXa

Factor Xa

TF

Tissue factor

sTF

soluble TF

PL

phospholipid

TFPI

Tissue factor pathway inhibitor

BPTI

bovine pancreatic trypsin inhibitor

Ss model

Scripps static model

Se model

Scripps solution-equilibrated model

CheA model

Original Chapel Hill solution-equilibrated model A

CheB model

Current Chapel Hill solution-equilibrated model B

pAB

p-aminobenzamidine

EGRck

Glu-Gly-Arg-chloromethylketone

D-FPRck

D-Phe-Pro-Arg-chloromethylketone

D-FFR

D-Phe-Phe-Arg-chloromethylketone