Skip to main content
NCBI home page
Protein Science : A Publication of the Protein Society logoLink to Protein Science : A Publication of the Protein Society
. 2014 Oct 14;23(12):1717–1727. doi: 10.1002/pro.2553

The length of the linker between the epidermal growth factor-like domains in factor VIIa is critical for a productive interaction with tissue factor

Egon Persson 1,*, Jesper J Madsen 2, Ole H Olsen 2
PMCID: PMC4253812  PMID: 25234571

Abstract

Formation of the factor VIIa (FVIIa)-tissue factor (TF) complex triggers the blood coagulation cascade. Using a structure-based rationale, we investigated how the length of the linker region between the two epidermal growth factor (EGF)-like domains in FVIIa influences TF binding and the allosteric activity enhancement, as well as the interplay between the γ-carboxyglutamic acid (Gla)-containing and protease domains. Removal of two residues from the native linker was compatible with normal cofactor binding and accompanying stimulation of the enzymatic activity, as was extension by two (Gly-Ser) residues. In sharp contrast, truncation by three or four residues abolished the TF-mediated stabilization of the active conformation of FVIIa and abrogated TF-induced activity enhancement. In addition, FVIIa variants with short linkers associated 80-fold slower with soluble TF (sTF) as compared with wild-type FVIIa, resulting in a corresponding increase in the equilibrium dissociation constant. Molecular modeling suggested that the shortest FVIIa variants would have to be forced into a tense and energetically unfavorable conformation in order to be able to interact productively with TF, explaining our experimental observations. We also found a correlation between linker length and the residual intrinsic enzymatic activity of Ca2+-free FVIIa; stepwise truncation resulting in gradually higher activity with des(83–86)-FVIIa reaching the level of Gla-domainless FVIIa. The linker appears to determine the average distance between the negatively charged Gla domain and a structural element in the protease domain, presumably of opposite charge, and proximity has a negative impact on apo-FVIIa activity.

Keywords: factor VIIa, tissue factor, allostery, structural complementarity, linker region

Introduction

Coagulation factor VIIa (FVIIa) is a serine protease of pivotal importance to the blood coagulation cascade, forming the initiating complex with its cognate cofactor, TF.1 The precursor FVII is synthesized as a single-chain, inactive zymogen, and FVIIa results from activation by endoproteolytic cleavage after Arg-152. Trace amounts FVIIa patrol the vasculature and respond instantly to the exposure of TF as a sign of vascular damage. Free FVIIa is an enzyme in a quiescent state lacking significant biological activity, but upon association with TF it becomes an efficient catalyst capable of rapidly initiating the blood coagulation cascade by converting more FVII to FVIIa and by activating factors IX and X.1 The TF-interactive interface on FVIIa2 can be divided into two parts, the first involving the light chain and providing the majority of the binding energy3 and the second primarily being composed of contact points on the protease domain, including Met-306, which is the origin of the allosteric effect on FVIIa.4 FVIIa is homologous to other vitamin K-dependent coagulation factors and contains the same four domains as, for instance, factors IX and X, namely a γ-carboxyglutamic acid (Gla)-containing domain, two epidermal growth factor (EGF)-like domains and a serine protease domain.5 The first three domains make up the light chain, whereas the protease domain constitutes the heavy chain. The last Cys residue of the first EGF-like domain and the first Cys residue of the second EGF-like domain are separated by a 9-residue linker sequence. The linker comprises a hinge region (one residue in particular) which permits different relative orientations of the domain pairs on each side (Gla plus first EGF-like domains and second EGF-like plus protease domains, respectively).6 In the complex with TF, FVIIa has an extended conformation with a stretched linker region dictated by the interactions with the protein cofactor.2

Even though the amino acid sequence of the linker in FVII(a) is not conserved between species,7,8 its length and flank regions, including the most pronounced hinge residue Gln-88, are. This suggests that interdomain distance and flexibility are of functional importance. Factors IX(a) and X(a) on the other hand have a shorter linker consisting of five and seven amino acid residues, respectively. Only the first glutamic acid residue is conserved between the three human factors.9,10 Rather than directly participating in protein–protein interactions, the linker regions may serve to define the distance between distant parts of the proteins and allow them to interact optimally with for instance protein cofactors and substrates.

The aim of this study was to determine what FVIIa can tolerate in terms of truncation of the linker region before the interaction with TF is compromised and whether the effect occurs abruptly or gradually as an increasing number of amino acid residues are removed. With this purpose, we executed a structure-based truncation strategy and demonstrated that the length of the linker between the EGF-like domains in FVIIa indeed is crucial for the optimal physical interaction with TF. This is in turn necessary for stabilization of the active conformation of FVIIa and a productive outcome in terms of enhanced FVIIa enzymatic activity. Too short a linker cannot, at least not in any low-energy conformation with allowed backbone dihedral angles, support an appropriate distance between the two EGF-like domains nor between distant patches of the TF-interactive interface on FVIIa. We also propose that the linker length governs the spatial distance between the Gla domain and an element in the protease domain and thereby influences the residual enzymatic activity of the apo form of free FVIIa.

Results

Molecular design rationale behind linker truncations

Alignment of FVIIa sequences from a wide range of mammals reveals that linker residues Glu-82, Gln-88, Leu-89, and Ile-90 are invariably conserved, whereas positions 83–87 are promiscuous with respect to amino acid occupancy. This strongly suggests that deletions within residues 83–87 will affect the linker length without disrupting important secondary structural elements. Compared with the X-ray crystal structure of FVIIa bound to TF,2 structures of free FVIIa with visible EGF-like domains6,11 (PDB entry codes 1qfk and 1dva) confirm the presence of the short β-strand at residues Thr-83 to His-84, but reveal no consensus on the position of the helical fragment observed. This suggests that the flexible nature of the linker allows the formation of half or full turns but only transiently, consistent with previous computational studies of free and TF-bound FVIIa.12 We conclude that only the length of the linker is likely to be affected in the des86- and des(85–86)-FVIIa variants, whereas in des(84–86)-FVIIa, Asp-87 will move to the position occupied by the native residue 84. This creates a sequence (Glu-82, Thr-83, and Asp-84) present in bovine and canine FVIIa making retention of the β-sheet in free des(84–86)-FVIIa plausible. However, in the structural minimizations of the des(84–86)-FVIIa-TF complex, no β-sheet is observed. This is not due to the relative β-sheet propensity of aspartic acid, but because the linker length no longer allows it. The β-sheet might be entirely missing in free des(83–86)-FVIIa. To evaluate the physical extension of various linker lengths, molecular dynamics simulations were performed on native FVIIa and the four truncated variants. Representative snapshots from the simulations are shown in Figure 1. The distance between the last disulfide in the first EGF-like domain and the first disulfide of the second EGF-like domain decreases with the number of residues truncated while, indeed, the helical element observed in the X-ray crystal structure of FVIIa-TF2 is missing in all the simulated free FVIIa variants and, instead, residues 85–91 are engaged in varying degrees of intermittent turn/coil transitioning (Fig. 1). In addition, the short β-sheet in which strand residues Thr-83 and His-84 participate is preserved in the simulation of FVIIa. The strand property is retained in simulations of the truncated des86-, des(85–86)-, and des(84–86)-FVIIa variants, but is absent in des(83–86)-FVIIa. These results suggest that the linker region in all the free FVIIa variants is devoid of helical elements and that a short strand is present when up to three residues are removed. Hence truncation of the linker, at least by up to three residues, affects its length but not secondary structural content.

Figure 1.

Figure 1

Open in a new tab

Structures and secondary structure assignments of the linker region in free FVIIa variants based on molecular dynamics simulations. Top, representative structures (30 ns snapshots) showing the distance in Å (illustrated by a gray line) between the Cα atoms of Cys-81 and Cys-91 (full-length numbering). The backbone traces start at residue Phe-71 in the first EGF-like domain and the side chains are shown for all residues present in the linker region and the flanking Cys residues. Bottom, secondary structure assignments of the linker region based on the simulations shown as weblogos.13 The one-letter secondary structure codes are: H, α-helix; G, 310-helix; I, PI-helix; E, extended conformation (sheet); B, isolated bridge; T, turn; C, random coil (none of the previous).

Effects of sTF on activity and conformational stability of FVIIa linker variants

One purpose of this study was to investigate whether the length of the linker between the two EGF-like domains in FVIIa is critical for the interaction with, and cofactor-induced effects of, TF. In the initial characterization, all the FVIIa variants, both those with an extended and those with a truncated linker, were found to possess virtually the same intrinsic amidolytic and proteolytic activity as wild-type FVIIa (data not shown). The latter was measured using FX as the substrate both in the presence and absence of phospholipids. In the next step, the amidolytic activities were measured in sTF titration experiments. The FVIIa variants with the smallest truncations, lacking one (des86-FVIIa) or two residues (des(85–86)-FVIIa) in the linker, responded normally to sTF with a fold activity enhancement at saturating concentrations of sTF and an EC50 value indistinguishable from those of wild-type FVIIa [Fig. 2(A)]. In sharp contrast, those with longer truncations, that is, des(84–86)- and des(83–86)-FVIIa, were virtually nonresponsive to sTF (even at 5 µM cofactor). An increased incubation time with sTF did not help to activate the two shortest FVIIa variants (data not shown). The variants containing extensions of the linker, 86G87- and 86GS87-FVIIa, behaved like FVIIa [Fig. 2(B)].

Figure 2.

Figure 2

Open in a new tab

Functional response of FVIIa linker variants upon sTF association. (A) 10 nM FVIIa (green circles), des86-FVIIa (□), des(85–86)-FVIIa (▪), des(84–86)-FVIIa (○), or des(83–86)-FVIIa (•) was incubated with the indicated concentrations of sTF and 1 mM S-2288. The ratio of the amidolytic activities of a FVIIa variant in the presence and absence of sTF is shown as a function of [sTF]. A ratio value of 1 is assigned to the situation without sTF. The results obtained with des(83–86)- and des(84–86)-FVIIa are virtually identical and the open-circle data points representing des(84–86)-FVIIa are therefore not visible. (B) The corresponding data shown for 86G87-FVIIa (▵) and 86GS87-FVIIa (▴) with the FVIIa data from A included for comparison. The data shown are mean ± standard deviation of all experiments (n = 4).

Carbamylation, a chemical modification of the N-terminal amino group of the protease domain and a measure of its accessibility to solvent,14 results in loss of enzymatic activity. It was used to investigate whether the ability (or inability) of sTF to enhance the activity of a particular FVIIa variant was reflected in an increased (or unaltered) degree of burial of the N-terminus in the presence of the cofactor. In the absence of sTF, all the studied FVIIa variants, whether truncated or extended in the linker, Gla-domainless or full-length, were modified at indistinguishable rates and lost 13–14% activity per 10 min. When bound to sTF, FVIIa, 86G87-FVIIa, 86GS87-FVIIa, des86-FVIIa, and des(85–86)-FVIIa were all modified at a significantly slower rate (2–3%/10 min). In sharp contrast, the rates of carbamylation of des(84–86)- and des(83–86)-FVIIa were not affected by complex formation with sTF. Thus there appears to be an absolute correlation amongst the FVIIa linker variants between sTF-induced activity stimulation and burial of the N-terminus into the activation pocket.

Binding of FVIIa linker variants to sTF

The physical binding of FVIIa variants to sTF was examined with the main purpose to investigate whether the failure of sTF to stimulate certain FVIIa variants (those with the largest truncations of the linker region) was due to an abolished protein–protein interaction. The kinetics of sTF binding were very similar for FVIIa, 86GS87-FVIIa, des86-FVIIa, and des(85–86)-FVIIa, albeit with slightly slower association rate and faster dissociation rate for des(85–86)-FVIIa (Fig. 3 and Table1). These FVIIa variants also exhibited indistinguishable degrees of sTF-induced activity enhancement (Fig. 2). In contrast, des(83–86)- and des(84–86)-FVIIa bound sTF with dramatically different kinetics because of the association rates being ∼80 times slower than those of the longer variants. However, once formed, the complexes of des(83–86)- and des(84–86)-FVIIa with sTF were virtually as stable, as judged by the dissociation rate constants, as those containing FVIIa variants with a longer linker region. This suggests that the shortest FVIIa variants eventually form an extended interactive interface with TF, but with much lower success rate and without a concurrent stimulation of the catalytic machinery located in the protease domain.

Figure 3.

Figure 3

Open in a new tab

Kinetics of the binding of FVIIa linker variants to sTF measured by surface plasmon resonance. Corrected sensorgrams are shown for the interactions between sTF and des(84–86)-FVIIa (A), des(85–86)-FVIIa (B), FVIIa (C), and 86GS87-FVIIa (D). Des(84–86)-FVIIa was injected in a 2-fold dilution series in concentrations from 40 to 640 nM (represented by the curves from bottom to top), whereas the other three variants were injected in a 2-fold dilution series in concentrations from 4 to 64 nM. Experimental data are in black and fitted data in red. Binding curves for des(83–86)-FVIIa and des86-FVIIa are not shown here, but the derived constants are included in Table1.

Table I.

Kinetic Parameters of the Binding of FVIIa and FVIIa Linker Variants to sTF Measured by Surface Plasmon Resonance

Ligand kon (×105 M−1 s−1) koff (×10−3 s−1) Kd (nM)
86GS87-FVIIa 3.7 ± 0.3 1.2 ± 0.1 3.3 ± 0.3
FVIIa 4.5 ± 0.3 1.3 ± 0.2 3.0 ± 0.4
des86-FVIIa 3.7 ± 0.2 1.2 ± 0.0 3.3 ± 0.1
des(85–86)-FVIIa 2.7 ± 0.4 1.8 ± 0.1 6.8 ± 0.6
des(84–86)-FVIIa 0.054 ± 0.003 1.7 ± 0.2 310 ± 30
des(83–86)-FVIIa 0.052 ± 0.003 1.4 ± 0.0 270 ± 10
Open in a new tab

Biotinylated sTF was captured on a streptavidin-coated chip and the binding of serial dilutions of the FVIIa variants was measured. The kinetic parameters (mean ± SEM) were determined using the Biacore 3000 evaluation software (Biaevaluation 4.1) and a model based on 1:1 binding stoichiometry.

Modeling of FVIIa linker variants in complex with TF

For a functional FVIIa-TF complex to be formed, FVIIa must attain a conformation which is complementary to TF and accommodate specific exosite interactions with TF. Our striking experimental finding that the entire change of the functional properties of FVIIa occurred upon removal of the third residue from the linker asked for a plausible explanation. To investigate how the length of the linker connecting the EGF-like domains in FVIIa affects the ability of the molecule to form a fully functional complex, the linker was gradually truncated by one to four residues and an attempt to structurally explain the slow, nonproductive TF binding of des(83–86)- and des(84–86)-FVIIa was made by generating structural ensembles for all of the Ca2+-loaded FVIIa variants in the TF-bound form. The nature of the linker in the X-ray crystallographic structure of FVIIa bound to TF2 encompasses both a β-sheet, pairing the two-residue β-strand composed of Thr-83 and His-84 with Phe-76 and Glu-77, and a short 310 helical fragment from Lys-85 to Asp-87. This suggests that partial unfolding of the helical element may make it possible for the FVIIa-TF complex to tolerate small deletions in the linker region with retained functionality, which is indeed what was experimentally observed.

The structural characterization of the phenomenon was done in two manners. First, the strain in the X-ray crystal structure introduced by truncations in the linker was gauged. For this purpose, an ensemble of possible linker conformations were initially generated using Rosetta's loop modeling application on the otherwise fixed structure of the FVIIa-TF complex. This approach revealed that in the case of a four-residue deletion, no combination of allowed backbone dihedral angles and atomic bond lengths could yield a linker conformation connecting the two flanking Cys residues without a chain break (not shown). For this simple reason, we conclude that it is very unlikely that one can truncate the linker by four residues and retain the interactions with TF (at least those seen in the crystal structure) required for a functional FVIIa-TF complex. Therefore, a second approach to evaluate the consequences of linker truncation on the tertiary structure of the FVIIa-TF complex by forcing ideal geometry (and hence loop closure) was employed. If four residues were deleted and the bond lengths idealized, the FVIIa-TF complex adopts a warped conformation in which FVIIa clamps TF causing it to bend significantly and deviate from the known structure. Interestingly, the structural ensembles using the native linker or a linker after deletion of one or two residues scored virtually identically (Fig. 4). Inspection of representative structures selected by clustering methods showed that for a one- or two-residue deletion the loop straightens to some extent and the characteristic helical fragment of the linker indeed unfolds, but it does so without greatly penalizing the overall score of the complex or the linker region. On the other hand, structures for the three- and four-residue deletions were energetically punished, in part, by their unfavorable backbone dihedral angles which could be confirmed by Ramachandran plots for the linker residues (Fig. 4).

Figure 4.

Figure 4

Open in a new tab

Molecular modeling of FVIIa-TF complexes. Average scores and standard deviations in negative Rosetta energy units (REUs) of the best scoring 10% of the models of complexes between FVIIa linker variants and TF are shown. The Ramachandran plots visualize the backbone dihedral angles (phi and psi) of the amino acid residues in the linker region in these complexes. wt, wild-type FVIIa; desX-FVIIa, FVIIa variant with residue(s) X deleted.

Influence of linker length and Gla domain on the activity of apo-FVIIa

In the presence of 5 mM Ca2+, FVIIa and des(1–44)-FVIIa (Gla-domainless FVIIa) had similar amidolytic activities (Km and kcat) as measured with the substrate S-2288 (data not shown). NaCl in the range 0.1–0.3M had a slightly stimulatory effect on the activity of both forms of FVIIa, an effect, which gradually decreased with increasing concentration (no net effect at 0.5M) and turned inhibitory at higher concentrations.15

In the absence of Ca2+, full-length and des(1–44)-FVIIa exhibited much lower amidolytic activity, but des(1–44)-FVIIa retained significantly higher activity than FVIIa as published.16 The residual activities of the apo (Ca2+-free) forms of FVIIa and des(1–44)-FVIIa were higher when the ionic strength was maintained with sodium chloride than with choline chloride (Fig. 5). Moreover, increasing the NaCl concentration (at least up to 0.6M) gradually enhanced the activity of full-length and des(1–44)-FVIIa in the absence of Ca2+ (data not shown), whereas increasing the choline concentration had a marginal effect. This suggests that NaCl shields the charged Gla domain and further stimulates the amidolytic activity of FVIIa by saturation of a putative sodium site in the protease domain.17,18

Figure 5.

Figure 5

Open in a new tab

Residual amidolytic activity of the apo (Ca2+-free) forms of FVIIa linker variants. The residual activity after the removal of Ca2+ from FVIIa, FVIIa linker variants and des(1–44)-FVIIa was measured in the presence of 0.1M NaCl (black bars) or choline chloride (red bars). All data are mean ± S.D., n = 3.

With this background knowledge about the effects of various ions on and the properties of full-length and Gla-domainless FVIIa in the absence of Ca2+, we studied the influence of the linker length. FVIIa retained about 7% amidolytic activity upon removal of Ca2+ from assay buffer containing 0.1M choline chloride (Fig. 5). Similar residual activity was measured with 86GS87- and 86G87-FVIIa. As the linker length decreased, the FVIIa variants gradually retained more activity and des(83–86)-FVIIa possessed the same residual activity as des(1–44)-FVIIa (17–18%). The pattern was very similar when using NaCl as the salt, but the level of residual activity was generally twice as high and ranged from 13 to 35%, conceivably because of Na+-specific stimulation (Fig. 5).

Carbamylation, even when using a relatively low concentration of potassium cyanate (50 mM) to achieve a decreased but still measurable rate of activity loss with FVIIa in the presence of Ca2+, can apparently not distinguish between the degrees of N-terminus exposure in Ca2+-loaded and -free FVIIa (data not shown).

Discussion

Complex formation with TF is energetically driven by interactions involving the light chain of FVIIa, whereas establishment of the interactions involving the protease domain leads to allosteric stabilization of the active conformation of FVIIa.2–4 Together, this shapes a high-affinity, enzymatically active complex with FVIIa positioned for efficient substrate processing. In free FVIIa, the linker between the EGF-like domains in the light chain appears to be a hinge for interdomain flexibility.6,12,19 FVIIa bound to TF is motionally restricted and adopts an extended conformation which intuitively does not permit much reduction in linker length if the tight and productive interaction with TF, with participation of the light chain and protease domain, is to be maintained.2,12,19 Measurement of FVIIa amidolytic activity enhancement provides an unambiguous assessment of the stimulatory effect of TF looking exclusively at the catalytic machinery and the state of the S1 and proximal subsites without influence from exosites. We reduced the length of the linker by truncations at positions which, according to molecular dynamics simulations, minimally affected secondary structural elements. Importantly, the simulations of free FVIIa variants showed that the helical structure observed in FVIIa-TF is missing throughout. We were able to show that the TF-induced stimulation of FVIIa activity was abrogated when three amino acid residues are removed from the linker, whereas truncation of two residues, or an extension, were perfectly tolerated. The same, very dramatic pattern was observed for the physical interaction between FVIIa and sTF. Binding kinetics and affinity were indistinguishable from those of native FVIIa unless at least three residues were removed from the linker, in which case the association rate plummeted. These observations clearly demonstrated that a linker lacking three residues does not allow the physical separation of the light chain and protease domain parts of the TF-interactive interface to render allosteric activity enhancement and native affinity possible. The weakest part of the interface, that between the protease domain and TF, apparently capitulated and this resulted in loss of TF-induced activity increase. Finally, we found that a truncation of the linker by four residues eliminated the negative impact of the Gla domain on the intrinsic activity of FVIIa in the absence of Ca2+.

The slower sTF association rates of des(83–86)- and des(84–86)-FVIIa might be explained by their preferential existence in energetically favorable conformations encompassing allowed dihedral angles. However, because of the relatively short linkers in these variants, these conformations do not allow a productive interaction with sTF and establishment of the contacts seen in the crystal structure of the complex. The conformations which would fit to TF and form all the interactions known from the structure are, if at all possible, extremely unfavorable and scarce. This would represent a particular problem if FVIIa initially docks with the protease domain, in itself a low-affinity interaction with sTF. In this case, an imminent clash or a suboptimal distance between the two parts of the TF-interactive interface on FVIIa (protease domain and Gla/first EGF-like domains) would prevent the ensuing tethering of the complex by the contacts between the light chain of FVIIa and TF. This binding pathway has been inferred from transition state analysis and association rate studies of the FVIIa:sTF complex formation.20,21 There is evidence in support of the notion that Gln-88 acts as hinge residue in the linker between the EGF-like domains in FVIIa.6 The first EGF-like domain has been resolved by X-ray crystallography in several structures of FVIIa in complex with TF with a similar domain configuration, well represented by the structure of Banner et al.2 To our knowledge, only two structures are available of the free form of FVIIa with the first EGF-like domain resolved.6,11 The first EGF-like domains, as represented in the structures of TF-bound and free FVIIa, cannot be overlaid without a clash between TF and the protease domain of FVIIa when oriented as in free FVIIa. An overlay of the protease domain structures does not cause a corresponding clash of the first EGF-like domain with TF. To further substantiate this interesting observation, we explored the configurational space around Gln-88. We found that the population of models which does not clash with TF is larger when the protease domain interacts with TF than when the light chain does (data not shown). This would indicate that it is entropically favorable for the FVIIa protease domain to initiate complex formation with TF and be in agreement with the experimentally supported binding pathway.20,21

Of the two parts of the TF-interactive interface on FVIIa located on opposite sides of the linker region, the one involving the light chain contributes the majority of the binding energy.3 The virtually normal rate of dissociation of the established complex between sTF and des(83–86)- or des(84–86)-FVIIa, combined with the inability of sTF to stimulate the activity of these variants, thus suggests that the light chain is engaged in its native interactions with sTF with retained affinity and that the protease domain simultaneously interacts with sTF in an alternative, non-native mode which is capable of maintaining the normal stability but not forming the allosterically important contacts between enzyme and cofactor. These crucial contacts are retained in the complexes between sTF and FVIIa variants lacking only one or two residues in the linker. The slow rate of association between des(83–86)- or des(84–86)-FVIIa and sTF suggests that a large proportion of the docking events mediated by the protease domain does not result in the formation of a stable complex because the suboptimal linker length forces the protease domain to slide relative to sTF to allow the light chain to bind. This is compatible with and a consequence of the mentioned preference of the FVIIa variants to exist in an energetically favorable conformation. At some stage during the transition process from encounter to stable complex, neither the initial nor the final interactions between FVIIa and sTF are established making the complex vulnerable and prone to dissociation. An alternative explanation for the binding kinetics (slow association and normal dissociation) of for instance des(84–86)-FVIIa could be that only a small fraction of the molecules are in a conformation which is at all compatible with sTF binding but do form the complex as we know it. However, this situation is less plausible because of the lack of detectable stimulation by sTF. Interestingly, the interaction of the isolated light chain of FVIIa with sTF, which like those of des(83–86)- and des(84–86)-FVIIa is of considerably lower affinity than that involving FVIIa, is primarily characterized by a much faster dissociation rate with a marginally affected association rate.3 Altogether, the experimental and molecular modeling data obtained with the FVIIa variants with the shortest linkers support the hypothesis that FVIIa forms an encounter complex with sTF involving interactions between the protease domain and the cofactor, after which the light chain engages in binding to tether the proteins to each other. Our findings regarding FVIIa binding to TF are almost certainly pertinent also for the binding of FVII and for its ensuing activation. Enzyme and zymogen employ the same interactive interfaces on both sides of the linker in their very similar, perhaps identical, set of contacts with TF.22

The linker region between the two EGF-like domains varies in length and sequence between the homologous factors VII(a), IX(a), and X(a). This may represent adaption to their respective cofactor (TF, factor VIIIa, and factor Va, respectively) and serve as one of several players in the optimization of the intermolecular interplay. Factor IXa has the shortest linker and an extension has been shown to result in a diminished activity enhancement upon assembly with factor VIIIa.23 We have shown that the linker length is critical also in FVIIa and that a limited truncation results in a TF binding defect and lack of activity enhancement upon TF association. An extension will, at some point, most likely also become unfavorable and have functional consequences. Thus the linker length appears to be important for an optimal interaction, plausibly serving a spacer function between key cofactor-interactive interfaces on the enzyme surface. Moreover, data with factor Xa indicate that the linker sequence becomes more exposed and available for protein-protein interactions upon activation of the zymogen.24 This could be part of the explanation for the higher affinity of factor Va for factor Xa as compared with X,25 definitely also involving rearrangements in the protease domain ensuing zymogen cleavage. An optimal combination of linker length and availability might be an important interaction determinant in this family of coagulation factors.

Materials and Methods

Proteins and reagents

FVIIa, des(1–44)-FVIIa and sTF were prepared as previously described.26–28 Their concentrations were determined by absorbance measurements at 280 nm using absorption coefficients of 1.32, 1.26, and 1.5, respectively, for a 1 mg/ml solution and molecular masses of 50, 46, and 25 kDa, respectively. Biotin-E219C-sTF, a variant of sTF biotinylated on the introduced unpaired Cys residue, was a gift from Dr. Henrik Østergaard, Haemophilia Biochemistry, Novo Nordisk A/S. Human factor X (FX) was purchased from Enzyme Research Laboratories (South Bend, IN). The chromogenic substrates S-2288 and S-2765 were from Chromogenix (Milan, Italy), phospholipid-TGT from Rossix (Mölndal, Sweden), and potassium cyanate from Fluka.

Mutagenesis, expression, and purification of FVIIa variants

Residues in FVII were deleted or inserted using the QuikChange kit (Stratagene, La Jolla, CA) and the human FVII expression plasmid pLN174.29 The following sense (and complementary reverse) primers were used to generate the FVIIa linker variants: des86-FVIIa, 5′-CGGAACTGTGAGACGCACAAGGACCAGCTGATCTGTGTG-3′; des(85–86)-FVIIa, 5′-CGGAACTGTGAGACGCACGACCAGCTGATCTGTG TG-3′; des(84–86)-FVIIa, 5′-CGGAACTGTGAGAC GGACCAGCTGATCTGTGTG-3′; des(83–86)-FVIIa, 5′-GGCCGGAACTGTGAGGACCAGCTGATCTGTGT GAACG-3′; 86G87-FVIIa, 5′-GAGACGCACAAGG AT GGCGACCAGCTGATCTGTGTG-3′; 86GS87-FVIIa, 5′-GAGACGCACAAGGATGGCAGCGACCAGCTGAT CTGTG-3′. The insertions between wild-type residues 86 and 87 in the two latter variants are underlined. Plasmids were prepared using a QIAfilter plasmid midi kit (Qiagen, Valencia, CA) and the introductions of the desired changes confirmed by sequencing. Baby hamster kidney cell transfection and selection, as well as protein expression, purification, and autoactivation were performed as described.4,30

FVIIa activity assays

All assays were routinely performed at ambient temperature in 50 mM Hepes, pH 7.4, containing 0.1M NaCl, 5 mM CaCl2, 0.1% (w/v) PEG 8000 and 0.01% (v/v) Tween 80 (assay buffer). The amidolytic activity was measured using 300 nM FVIIa, des(1–44)-FVIIa or FVIIa linker variants in assay buffer. When studying the Ca2+ dependence of the activity, 2 mM EDTA replaced CaCl2 in the assay buffer. Either fixed (1 mM) or varying (0.2–12.8 mM) concentration of S-2288 was used. The amidolytic activity dependence on ionic strength and Na+ was also measured using 1 mM S-2288 in 50 mM Hepes, pH 7.4, containing 5 mM CaCl2 or 2 mM EDTA, 0.1% (w/v) PEG 8000 and 0.01% (v/v) Tween 80, by including NaCl or choline chloride at concentrations ranging from 0.1 to 0.6 M. The effect of the cofactor TF on the amidolytic activity was assessed by mixing 10 nM FVIIa variant, no or 2–200 nM sTF, and 1 mM S-2288. Those FVIIa variants that were not stimulated at an [sTF] of 200 nM were also tested at 2 and 5 µM sTF. For proteolytic activity assessment, 1 µM FX was incubated with 50 nM FVIIa variant in a total volume of 100 µL for 20 min, where after FX activation was terminated by the addition of 50 µL assay buffer containing 20 mM EDTA instead of CaCl2. The formed FXa was measured by the addition of S-2765 substrate (final concentration 0.5 mM). Alternatively, FX was used at a concentration of 150 nM in the presence of 25 µM phospholipids. Enzymatic activities were continuously monitored for 20 min (FVIIa or FVIIa:sTF amidolytic activity) or 5 min (FXa amidolytic activity) using a SpectraMax 190 kinetic microplate reader (Molecular Devices, Sunnyvale, CA).

Carbamylation

The FVIIa variants alone (2 µM) or FVIIa variants (0.5 µM) in the presence of 10 µM sTF were incubated with 0.2M potassium cyanate in 50 mM Hepes, pH 7.4, containing 0.1M NaCl, 5 mM CaCl2, 0.1% (w/v) PEG 8000, and 0.01% (v/v) Tween 80. At time zero and after 20 and 60 min, 20 µL samples were transferred to a well containing 160 µL buffer and the amidolytic FVIIa activity was measured after the addition of 20 µL S-2288 substrate (final concentration 1 mM).

Surface plasmon resonance analyses

Biotin-E219C-sTF was injected at a concentration of 150 nM (in the assay buffer described above) and a flow rate of 20 µL/min for 100 s over a streptavidin (SA) chip (GE Healthcare Bio-Sciences AB, Uppsala, Sweden), resulting in an immobilization level of 90–120 resonance units. FVIIa, des86-FVIIa, des(85–86)-FVIIa, and 86GS87-FVIIa were injected in two-fold dilutions in assay buffer from 64 to 4 nM, whereas des(84–86)- and des(83–86)-FVIIa were injected in 2-fold dilutions from 640 to 40 nM, all at a flow rate of 20 µL/min. Association and dissociation phases lasted for 4 and 10 min, respectively. Regeneration between runs was done with a 30 s pulse of 50 mM EDTA, pH 7.4, at 20 µL/min. All experiments were conducted using a Biacore 3000 instrument and evaluated using a 1:1 model in Biaevaluation 4.1 supplied by the manufacturer.

Molecular dynamics simulations

Free FVIIa and variants thereof with a deleted linker region between the EGF-like domains were analyzed by molecular dynamics. The models were based on the crystal structure of FVIIa in the FVIIa-TF complex2 (PDB entry code 1dan) after removal of the co-crystallized, covalently bound active-site inhibitor and TF. In FVIIa, the last 11 amino acid residues of the light chain are missing, but no attempts have been made to model the missing part to avoid perturbation of the X-ray data. The truncated structures were generated by deleting linker residues and rebuilding the remaining linker region. A final relaxation was conducted using 250 steps of conjugated gradient (CONJ) energy minimization in CHARMm (Accelrys, San Diego, CA) using the CHARMM27 force field.31

Molecular dynamics simulations were then performed using the program NAMD 2.532 and the CHARMM27 force field for protein. The 9 calcium ions embedded in the X-ray crystal structure were preserved. The proteins were solvated (with TIP3P water molecules) in a periodic, truncated octahedron with box boundaries at least 6 Å from any given protein atom, and water molecules were treated as rigid. Sodium and chloride ions were added to obtain an ionic strength of ∼0.15. The resulting systems were composed of ∼16,000 water molecules, ∼42 sodium ions, ∼43 chloride ions, and 9 calcium ions (in total ∼54,000 atoms). A cut-off of 12 Å (switching function starting at 10 Å) for van der Waals interactions was applied. The particle mesh Ewald (PME) method was used to compute long-range electrostatic forces in all simulations.33 An integration time step of 1 fs was used. Langevin dynamics was applied to enforce constant temperature (T = 300 K) conditions. The Langevin dampening coefficient was set to 5 ps−1. The pressure was maintained at 1 atm using the hybrid Nosé–Hoover Langevin piston method with a decay period of 100 ps and a dampening time constant of 50 ps.34,35 Each system was equilibrated in the constant number, pressure and temperature (NpT) ensemble and simulated for at least 100 ns. The secondary structure of the linker was assigned using STRIDE36 and the corresponding logos were generated by WebLogo.13

Molecular modeling

The X-ray crystallographic structure of the FVIIa-TF complex2 (PDB entry code 1dan) was subject to structural modeling using the Rosetta modeling suite.37 It was processed by a Rosetta high-resolution protein structure refinement protocol consisting of three steps. First, stripping the structure of nonprotein matter and patching the γ-carboxylated glutamic acid residues, then idealization of the geometry by IdealizeMover, and, finally, structural refinement with backbone perturbations and sidechain repacking by FastRelaxMover. This protocol was set up via the RosettaScripts interface.38 Using this protocol, 1,000 decoy models were generated for the wild-type FVIIa-TF complex and each of the complexes between TF and FVIIa with a truncated linker [des86-, des(85–86)-, des(84–86)-, and des(83–86)-FVIIa, respectively]. The decoy models which scored best (lowest 10% evaluated by the Rosetta score 12 energy function) were chosen as the representative ensemble of the most probable structures. In this way, possible linker conformations were sampled while allowing for some flexibility in other parts of the structure including, of particular interest, the interface between FVIIa and TF. Finally, the generated decoy complexes were structurally analyzed with attention on the FVIIa-TF interface (using InterfaceAnalyzer39) and the linker residues in FVIIa (the sequence flanked by the last Cys residue of the first EGF-like domain and the first Cys residue of the second EGF-like domain).

The configurational (φ,ψ)-space of backbone dihedral angles around the hinge residue Gln-88 was sampled in the X-ray crystallographic structure of FVIIa-TF2 to quantify the available freedom of motion in FVIIa not causing a clash with TF. Two scenarios were investigated; one with the FVIIa protease domain aligned with TF as in the crystal structure and the FVIIa light chain rotated (roaming), and another with the FVIIa light chain aligned with TF and the protease domain roaming. The (φ,ψ)-space was exhaustively searched using 2° increments and all 32,400 generated decoys were scored without any further processing by the full-atom force field in Rosetta (score12).

Acknowledgments

The authors thank Anette Østergaard for excellent technical assistance and Dr. Henrik Østergaard, Novo Nordisk A/S, for the gift of biotin-E219C-sTF.

Glossary

EGF

epidermal growth factor

FVII(a)

(activated) coagulation factor VII

Gla

γ-carboxyglutamic acid

sTF

soluble TF (residue 1-219)

TF

tissue factor.

References

  1. Davie EW, Fujikawa K, Kisiel W. The coagulation cascade. Initiation, maintenance, and regulation. Biochemistry. 1991;30:10363–10370. doi: 10.1021/bi00107a001. [DOI] [PubMed] [Google Scholar]
  2. Banner DW, D'Arcy A, Chène C, Winkler FK, Guha A, Konigsberg WH, Nemerson Y, Kirchhofer D. The crystal structure of the complex of blood coagulation factor VIIa with soluble tissue factor. Nature. 1996;380:41–46. doi: 10.1038/380041a0. [DOI] [PubMed] [Google Scholar]
  3. Persson E. Characterization of the interaction between the light chain of factor VIIa and tissue factor. FEBS Lett. 1997;413:359–363. doi: 10.1016/s0014-5793(97)00941-1. [DOI] [PubMed] [Google Scholar]
  4. Persson E, Nielsen LS, Olsen OH. Substitution of aspartic acid for methionine-306 in factor VIIa abolishes the allosteric linkage between the active site and the binding interface with tissue factor. Biochemistry. 2001;40:3251–3256. doi: 10.1021/bi001612z. [DOI] [PubMed] [Google Scholar]
  5. Furie B, Furie BC. The molecular basis of blood coagulation. Cell. 1988;53:505–518. doi: 10.1016/0092-8674(88)90567-3. [DOI] [PubMed] [Google Scholar]
  6. Pike ACW, Brzozowski AM, Roberts SM, Olsen OH, Persson E. Structure of human factor VIIa and its implications for the triggering of blood coagulation. Proc Natl Acad Sci U S A. 1999;96:8925–8930. doi: 10.1073/pnas.96.16.8925. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Hagen FS, Gray CL, O'Hara P, Grant FJ, Saari GC, Woodbury RG, Hart CE, Insley M, Kisiel W, Kurachi K, Davie EW. Characterization of a cDNA coding for human factor VII. Proc Natl Acad Sci U S A. 1986;83:2412–2416. doi: 10.1073/pnas.83.8.2412. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Idusogie E, Rosen E, Geng J-P, Carmeliet P, Collen D, Castellino FJ. Characterization of a cDNA encoding murine coagulation factor VII. Thromb Res. 1996;75:481–487. [PubMed] [Google Scholar]
  9. Kurachi K, Davie EW. Isolation and characterization of a cDNA coding for human factor IX. Proc Natl Acad Sci U S A. 1982;79:6461–6464. doi: 10.1073/pnas.79.21.6461. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Leytus S, Chung DW, Kisiel W, Kurachi K, Davie EW. Characterization of a cDNA coding for human factor X. Proc Natl Acad Sci U S A. 1984;81:3699–3702. doi: 10.1073/pnas.81.12.3699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Dennis MS, Eigenbrot C, Skelton NJ, Ultsch MH, Santell L, Dwyer MA, O'Connell MP, Lazarus RA. Peptide exosite inhibitors of factor VIIa as anticoagulants. Nature. 2000;404:465–470. doi: 10.1038/35006574. [DOI] [PubMed] [Google Scholar]
  12. Ohkubo YZ, Morrissey JH, Tajkhorshid E. Dynamical view of membrane binding and complex formation of human factor VIIa and tissue factor. J Thromb Haemost. 2010;8:1044–1053. doi: 10.1111/j.1538-7836.2010.03826.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Crooks G, Hon G, Chandonia J-M, Brenner SE. WebLogo: a sequence logo generator. Genome Res. 2004;14:1188–1190. doi: 10.1101/gr.849004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Higashi S, Nishimura H, Aita K, Iwanaga S. Identification of regions of bovine factor VII essential for binding to tissue factor. J Biol Chem. 1994;269:18891–18898. [PubMed] [Google Scholar]
  15. Neuenschwander PF, Branam DE, Morrissey JH. Importance of substrate composition, pH and other variables on tissue factor enhancement of factor VIIa activity. Thromb Haemost. 1993;70:970–977. [PubMed] [Google Scholar]
  16. Neuenschwander PF, Morrissey JH. Roles of the membrane-interactive regions of factor VIIa and tissue factor. J Biol Chem. 1994;269:8007–8013. [PubMed] [Google Scholar]
  17. Schmidt AE, Shah P, Gauthier EM, Bajaj SP. Protease domain sodium, calcium, and zinc sites in factor VIIa: crystal structures and kinetic studies. Blood. 2004;104:122a. [Google Scholar]
  18. Bajaj SP, Schmidt AE, Agah S, Bajaj MS, Padmanabhan K. High resolution structures of p-aminobenzamidine- and benzamidine-VIIa/soluble tissue factor. J Biol Chem. 2006;281:24873–24888. doi: 10.1074/jbc.M509971200. [DOI] [PubMed] [Google Scholar]
  19. Mosbæk CR, Nolan D, Persson E, Svergun DI, Bukrinsky JT, Vestergaard B. Extensive small-angle X-ray scattering studies of blood coagulation factor VIIa reveal interdomain flexibility. Biochemistry. 2010;49:9739–9745. doi: 10.1021/bi1011207. [DOI] [PubMed] [Google Scholar]
  20. Österlund M, Persson E, Svensson M, Carlsson U, Freskgård P-O. Transition state analysis of the complex between coagulation factor VIIa and tissue factor: suggesting a sequential domain-binding pathway. Biochem Biophys Res Commun. 2005;327:789–793. doi: 10.1016/j.bbrc.2004.12.058. [DOI] [PubMed] [Google Scholar]
  21. Österlund M, Persson E, Carlsson U, Freskgård P-O, Svensson M. Sequential coagulation factor VIIa domain binding to tissue factor. Biochem Biophys Res Commun. 2005;337:1276–1282. doi: 10.1016/j.bbrc.2005.09.190. [DOI] [PubMed] [Google Scholar]
  22. Kelley RF, Yang J, Eigenbrot C, Moran P, Peek M, Lipari MT, Kirchhofer D. Similar molecular interactions of factor VII and factor VIIa with the tissue factor region that allosterically regulates enzyme activity. Biochemistry. 2004;43:1223–1229. doi: 10.1021/bi035738i. [DOI] [PubMed] [Google Scholar]
  23. Celie PHN, van Stempvoort G, Fribourg C, Schurgers LJ, Lenting PJ, Mertens K. The connecting segment between both epidermal growth factor-like domains in blood coagulation factor IX contributes to stimulation by factor VIIIa and its isolated A2 domain. J Biol Chem. 2002;277:20214–20220. doi: 10.1074/jbc.M108446200. [DOI] [PubMed] [Google Scholar]
  24. Ambrosini G, Plescia J, Chu KC, High KA, Altieri DC. Activation-dependent exposure of the inter-EGF sequence Leu83-Leu88 in factor Xa mediates ligand binding to effector cell protease receptor-1. J Biol Chem. 1997;272:8340–8345. doi: 10.1074/jbc.272.13.8340. [DOI] [PubMed] [Google Scholar]
  25. Persson E, Hogg PJ, Stenflo J. Effects of Ca2+ binding on the protease module of factor Xa and its interaction with factor Va. J Biol Chem. 1993;268:22531–22539. [PubMed] [Google Scholar]
  26. Thim L, Bjoern S, Christensen M, Nicolaisen EM, Lund-Hansen T, Pedersen A, Hedner U. Amino acid sequence and posttranslational modifications of human factor VIIa from plasma and transfected baby hamster kidney cells. Biochemistry. 1988;27:7785–7793. doi: 10.1021/bi00420a030. [DOI] [PubMed] [Google Scholar]
  27. Nicolaisen EM, Petersen LC, Thim L, Jacobsen JK, Christensen M, Hedner U. Generation of Gla-domainless FVIIa by cathepsin G-mediated cleavage. FEBS Lett. 1992;306:157–160. doi: 10.1016/0014-5793(92)80989-t. [DOI] [PubMed] [Google Scholar]
  28. Freskgård P-O, Olsen OH, Persson E. Structural changes in factor VIIa induced by Ca2+ and tissue factor studied using circular dichroism spectroscopy. Protein Sci. 1996;5:1531–1540. doi: 10.1002/pro.5560050809. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Persson E, Nielsen LS. Site-directed mutagenesis but not γ-carboxylation of Glu-35 in factor VIIa affects the association with tissue factor. FEBS Lett. 1996;385:241–243. doi: 10.1016/0014-5793(96)00400-0. [DOI] [PubMed] [Google Scholar]
  30. Persson E, Kjalke M, Olsen OH. Rational design of coagulation factor VIIa variants with substantially increased intrinsic activity. Proc Natl Acad Sci U S A. 2001;98:13583–13588. doi: 10.1073/pnas.241339498. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. McKerell AD, Jr, Bashford D, Bellott M, Dunbrack RL, Jr, Evanseck J, Field MJ, Fisher S, Gao J, Guo H, Ha S, Joseph D, Kuchnir L, Kuczera K, Lau FTK, Mattos C, Michnick S, Ngo T, Nguyen DT, Prodhom B, Reiher IWE, Roux B, Schlenkrich M, Smith J, Stote R, Straub J, Watanabe M, Wiorkiewicz-Kuczera J, Yin D, Karplus M. All-hydrogen empirical potential for molecular modelling and dynamics studies of proteins using the CHARMM22 force field. J Phys Chem. 1998;102:3586–3616. doi: 10.1021/jp973084f. [DOI] [PubMed] [Google Scholar]
  32. Phillips JC, Braun R, Wang W, Gumbart J, Tajkhorshid E, Villa E, Chipot C, Skeel RD, Kale L, Schulten K. Scalable molecular dynamics with NAMD. J Comp Chem. 2005;26:1781–1802. doi: 10.1002/jcc.20289. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Essman U, Perera L, Berkowitz T, Darden T, Lee H, Pedersen LG. A smooth particle mesh method. J Chem Phys. 1995;103:8577–8593. [Google Scholar]
  34. Martyna GJ, Tobias DJ, Klein ML. Constant pressure molecular dynamics algorithms. J Chem Phys. 1994;101:4177–4189. [Google Scholar]
  35. Feller SE, Zhang Y, Pastor RW, Brooks BR. Constant pressure molecular dynamics simulation: the Langevin piston method. J Chem Phys. 1995;103:4613–4621. [Google Scholar]
  36. Frishman D, Argos P. Knowledge-based protein secondary structure assignment. Proteins. 1995;23:566–579. doi: 10.1002/prot.340230412. [DOI] [PubMed] [Google Scholar]
  37. Leaver-Fay A, Tyka M, Lewis SM, Lange OF, Thompson J, Jacak R, Kaufman KW, Renfrew D, Smith CA, Sheffler W, Davis IW, Cooper S, Treuille A, Mandell DJ, Richter F, Ban Y-EA, Fleishman SJ, Corn JE, Kim DE, Lyskov S, Berrondo M, Mentzer S, Popovic Z, Havranek JJ, Karanicolas J, Das R, Meiler J, Kortemme T, Gray JJ, Kuhlman B, Baker D, Bradley P. ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol. 2011;487:545–574. doi: 10.1016/B978-0-12-381270-4.00019-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Fleishman SJ, Leaver-Fay A, Corn JE, Strauch E-M, Khare SD, Koga N, Ashworth J, Murphy P, Richter F, Lemmon G, Meiler J, Baker D. RosettaScripts: a scripting language interface to the Rosetta macromolecular modeling suite. PLoS ONE. 2011;6:e20161. doi: 10.1371/journal.pone.0020161. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Lewis SM, Kuhlman BA. Anchored design of protein-protein interfaces. PLoS ONE. 2011;6:e20872. doi: 10.1371/journal.pone.0020872. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of The Protein Society

RESOURCES