Protein disulfide isomerase has no stimulatory chaperone effect on factor X activation by factor VIIa-soluble tissue factor

Protein disulfide isomerase has no stimulatory chaperone effect on factor X activation by factor VIIa-soluble tissue factor

Thrombosis Research (2008) 123, 171–176 www.elsevier.com/locate/thromres REGULAR ARTICLE Protein disulfide isomerase has no stimulatory chaperone e...

241KB Sizes 0 Downloads 43 Views

Thrombosis Research (2008) 123, 171–176

www.elsevier.com/locate/thromres

REGULAR ARTICLE

Protein disulfide isomerase has no stimulatory chaperone effect on factor X activation by factor VIIa-soluble tissue factor Egon Persson ⁎ Haemostasis Biochemistry, Novo Nordisk A/S, Novo Nordisk Park (G8.2.76), DK-2760 Måløv, Denmark Received 5 February 2008; received in revised form 27 March 2008; accepted 21 April 2008 Available online 11 June 2008

KEYWORDS Factor X activation; Protein disulfide isomerise; Factor VIIa; Tissue factor; Gla domain

Abstract Introduction: It was recently reported that protein disulfide isomerase (PDI) stimulates factor X (FX) activation by factor VIIa (FVIIa) bound to soluble tissue factor (sTF) in a purified system and that PDI may be responsible for activating cellular tissue factor (TF) and switching it between its roles in blood coagulation and cellular signalling. This study further investigates the former effect of PDI. Method: FX activations by FVIIa-sTF1–219 were carried out in the presence of different forms of PDI, with annexin V or detergent present in the system and using various forms of FVIIa and FX. In addition, FVIIa-lipidated TF was used as the FX activator. Results: Recombinant human PDI did not influence FX activation by FVIIa-sTF1–219, whereas PDI purified from bovine liver enhanced the activation rate in a dosedependent manner. The inclusion of annexin V or detergent abolished the stimulatory effect. Removal of the phospholipd-interactive γ-carboxyglutamic acid (Gla)containing domain from either FVIIa or FX obliterated the bovine PDI-induced enhancement of FX activation, as did the introduction of F4A or L8A mutation in FVIIa. The presence of 25 nM bovine PDI lowered the apparent Km for FX from far above 10 μM to 1–2 μM. No PDI effect was seen when FVIIa-lipidated TF was the FX activator. Conclusions: FX activation is insensitive to PDI per se and a phospholipid contaminant in the bovine PDI preparation acts stimulatory when sTF, but not lipidated TF, is the cofactor. Strong support is provided by the lacking effect of bovine PDI after removal or modification of the Gla domain in either FVIIa or FX as well as by the effects of annexin V and detergents and the decreased Km value. © 2008 Elsevier Ltd. All rights reserved.

Abbreviations: FVIIa, activated factor VII; FX, factor X; FXa, activated factor X; Gla, γ-carboxyglutamic acid; PDI, protein disulfide isomerase; sTF, soluble tissue factor; TF, tissue factor. ⁎ Tel.: +45 4443 4351; fax: +45 4466 3450. E-mail address: [email protected]. 0049-3848/$ - see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.thromres.2008.04.012

172

Introduction Tissue factor (TF) is the integral membrane cofactor and receptor for the enzyme factor VIIa (FVIIa). The binary complex initiates blood coagulation via activation of factors IX and X and mediates cell signalling events through cleavage of protease-activated receptor 2 [1]. Protein disulfide isomerase (PDI) is a protein normally present in the endoplasmic reticulum where it catalyses protein disulfide rearrangements and assists protein folding [2,3]. It was reported that cryptic, i.e. signalling but non-coagulant, TF contains unpaired Cys186 and Cys209 residues, whereas these cysteines are engaged in a disulfide bond in procoagulant TF and that extracellular PDI, by employing its oxidoreductase activity, regulates the thiol status of TF and the distribution between signalling (and coagulantly dormant) and procoagulant forms [4–6]. However, conflicting data exist which question the involvement of PDI and favour the phospholipid composition as the principal TF switch [7]. Independent of this, PDI has been shown to enhance factor X (FX) activation by FVIIa bound to soluble TF (sTF) by a chaperone mechanism not related to its oxidoreductase activity [8]. The studies that suggest a regulatory [5,6] and stimulatory [8] role of PDI employed endogenous cellular PDI and PDI purified from bovine liver, respectively. Two studies [5,8] presented circumstantial evidence of a direct association between PDI and TF using immunoprecipitation visualised by Western blotting. In addition, data from the purified system indicated that PDI facilitated formation of the ternary FVIIa-sTF-FX complex and the subsequent release of the product factor Xa (FXa) [8], but the molecular details pertinent to the PDI-induced stimulation of FX activation remain unknown. The ostensible biological importance of PDI in the context of TF certainly warrants efforts aiming at elucidating the underlying mechanism. I have obtained data suggesting that PDI does in fact not influence FX activation by means of a chaperone mechanism. Recombinant human PDI is unable to enhance FX activation by FVIIa-sTF, whereas liver-derived bovine PDI increases the rate as previously observed, but not when lipidated TF is the cofactor [8]. The enhancement in the presence of the bovine PDI preparation is abolished by annexin V and detergents, by removal of the γ-carboxyglutamic acid (Gla) domain of either FVIIa or FX or by mutations of the hydrophobic side chains in positions 4 and 8 of FVIIa critical for phospholipid binding. Moreover, the inclusion of bovine PDI reduces the Km of FVIIa-sTF for FX. Thus, PDI itself appears to be without any effect on FX activation. Instead, a contaminant in the bovine PDI preparation with the properties of phospholipid

E. Persson appears to cause the increase in the rate of FX activation. Materials and methods Proteins and reagents Recombinant human PDI was purchased from ProSpec Tany TechnoGene (Rehovot, Israel). Bovine PDI (purified from liver) and annexin V were from Sigma (St Louis, MO, USA). Triton X-100, CHAPS and Tween 80 were from Sigma or Merck. FX and FXa were from Enzyme Research Laboratories (South Bend, IN, USA), whereas relipidated TF (Innovin) was from Dade Behring (Marburg, Germany). FVIIa was obtained as described [9] and Gla-domainless FVIIa (des(1–38) and des(1–44)) was isolated after autodigestion and cathepsin G-mediated cleavage, respectively [10,11]. sTF1–209 and sTF1–219 were prepared from E. coli inclusion bodies, reduced and refolded as described [12]. Gla-domainless FX (des(1–44)) was isolated after limited proteolysis with chymotrypsin [13]. The F4A, L8A and F4A/L8A mutations were introduced into FVII using the QuikChange kit (Stratagene, La Jolla, CA, USA) and the human FVII expression plasmid pLN174 [14]. The following sense primers (and complementary reverse primers) were used, with base substitutions in italic and changed codons underlined: F4A, 5’CGCGCCAACGCGGCCCTGGAGGAGCTGC-3’; L8A, 5’-CCTGGAGGAGGCGCGGCCGGGCTCCC-3’; F4A/L8A, 5’-GCCAACGCGGCCCTGGAGGAGGCGCGGCCGGGC-3’. Plasmid preparations, baby hamster kidney cell transfections and selection and protein expression were performed as described [15,16]. The mutants were purified by chromatography on Q Sepharose Fast Flow (Ca2+ elution) and sTFSepharose (elution by Ca2+ chelation). All three FVII mutants autoactivated to completion in 5–6 days at room temperature without signs of degradation and displayed the same amidolytic activity as the wild-type enzyme.

FX activation assays All proteins were normally diluted in and the assays run in 20 mM HEPES, pH 7.4, containing 130 mM NaCl, 1.5 mM CaCl2, 0.5 mM MgCl2, 0.1% (w/v) PEG 8000 and 0.01% (v/v) Tween 80. To assess the effect of PDI on FX activation by FVIIa bound to sTF, 5 nM FVIIa, 25 nM sTF1–219 and 150 nM FX were incubated for 10 minutes at room temperature in the absence or presence (25 nM) of PDI (total volume 100 μl), optionally in the presence of 500 nM annexin V, 0.1 or 0.5% (v/v) Triton X-100, 1% (w/v) CHAPS or 1% (v/v) Tween 80. F4A-, L8A- and F4A/L8A-FVIIa were tested under the same conditions but without optional agent. In some experiments, the concentration of sTF1–219 (25-400 nM) or PDI (0.8–100 nM (and only 2 nM FVIIa)) was varied, sTF1–209 was used instead of sTF1–219, or Mg2+ was omitted from the buffer. FX activation by FVIIa bound to relipidated TF was measured by incubating 150 nM FX with 1 nM FVIIa and 1 pM TF for 10 min with or without 25 nM PDI, optionally with annexin V or detergent present. A decreased function of F4A-, L8A- and F4A/L8A-FVIIa (at 1 and/or 0.25 nM) due to an expected diminished phospholipid affinity was verified using lipidated TF. To determine the Km of FVIIa-sTF1–219 for FX, 50 pM FVIIa and 25 nM sTF1–219 were incubated with 150 nM-10 μM FX for 30 minutes. The incubation time was reduced to 10 minutes when the experiment was performed in the presence of 25 nM bovine PDI. All activation reactions were terminated by adding 50 μl buffer containing 20 mM EDTA, followed by the addition of 50 μl of a 2-mM solution of S-2765 (Chromogenix, Instrumentation Laboratory, Milan, Italy) to quantify the formed FXa by continuous absorbance measurement in a SpectraMax 190 microplate spectrophotometer (Molecular Devices, Sunnyvale, CA, USA).

PDI does not promote FX activation FX activation was also performed with des (1–38)-FVIIa as the enzyme; 20 nM des(1–38)-FVIIa and 100 nM sTF1–219 were incubated with 150 nM FX, and with des(1–44)-FX as the substrate; 5 nM FVIIa and 25 nM sTF1–219 were incubated with 0.5 μM des(1–44)-FX. The incubation time was 60 minutes in both cases and the assays were performed with and without 25 nM bovine PDI and optionally without Mg2+ in the buffer.

173 total volume was 200 μl and the samples were read as described above. PDI activity was measured by an insulin denaturation assay. 800 nM PDI was mixed with 75 µM insulin (B30, Novo Nordisk A/S) in 100 mM K2HPO4, 2 mM EDTA, pH 7, containing 0.5 mM reduced glutathione (Sigma), and the turbidity was monitored continuously at 650 nm.

Other activity assays The amidolytic activity of FVIIa-sTF was measured by incubating 10 nM FVIIa and 25 nM sTF1–219 with 1 mM S-2288 (Chromogenix, Instrumentation Laboratory, Milan, Italy), the amidolytic activity of FVIIa and mutants thereof was measured by incubating 100 nM enzyme with 1 mM S-2288 and the amidolytic activity of FXa was measured by incubating 5 nM FXa with 0.5 mM S-2765. All assays were performed in the absence and presence of 25 nM bovine PDI. The

Results and discussion Purified bovine, but not recombinant human, PDI enhances FX activation by FVIIa-sTF The addition of 25 nM recombinant human PDI had no effect on FX activation by FVIIa-sTF1–219 (Fig. 1A). However, the

Figure 1 Influence of human and bovine PDI on FX activation by FVIIa-sTF1–219. (A) The effect of 25 nM recombinant human (hu) or purified bovine PDI (bo) on the rate of FX activation (5 nM FVIIa/25 nM sTF1–219 or 1 nM FVIIa/1 pM lipidated TF and 150 nM FX in buffer containing 1.5 mM Ca2+ and 0.5 mM Mg2+). The same results were obtained using 25 nM FX in the latter system, excluding that the lacking enhancement by bovine PDI was due to saturating concentration of FX. (B) The effect on FX activation as a function of [PDI]. The graph shows the total amount of FXa formed during a 10-min incubation of 2 nM FVIIa and 25 nM sTF1–219 with 150 nM FX in the presence of the indicated concentrations of human (●) or bovine PDI (○). (C) The effect of the addition of 25 nM bovine PDI and/ or 0.5 mM Mg2+ on the rate of FXa generation. The system is the same as that described in (A), but with Ca2+ as the only divalent metal ion in the starting buffer. (D) The effect of annexin V and detergents on the enhancing effect of bovine PDI. Annexin V (Ann, 500 nM), Triton X-100 (Tri, 0.1% (v/v)), CHAPS (CHA, 1% (w/v)) or Tween 80 (Twe, 1% (v/v)) was added to the FVIIa-sTF1–219 system described in (A) containing 25 nM bovine PDI. The two left columns are based on the same data as the two right columns in (C).

174 finding by Versteeg and Ruf [8] that bovine PDI is capable of enhancing FX activation was confirmed (Fig. 1A), whereas there was no effect of bovine PDI on the amidolytic activity of FVIIa-sTF or FXa (not shown). Both forms of PDI were active in an insulin denaturation assay (not shown) but neither of them affected FX activation by FVIIa bound to lipidated TF (Fig. 1A). A titration experiment revealed a linear relationship between the concentration of bovine PDI, at least up to 100 nM, and the degree of enhancement of FX activation by FVIIa-sTF1-219, whereas human PDI was without any effect in the whole concentration range (Fig. 1B). The effect of bovine PDI was significant already at a concentration of 3 nM. Under the conditions closest resembling those in the published study, i.e. using sTF1–219 and omitting Mg2+ from the buffer, the addition of 25 nM bovine PDI increased the rate of FXa formation (from 150 nM FX) by 5 nM FVIIa in the presence of 25 nM sTF1–219 from around 0.05 nM/min to 0.7–0.8 nM/min (Fig. 1C). This corresponds approximately to a 14-fold rate enhancement, slightly more than previously observed [8]. As shown before [8], the PDI effect was not influenced by the sTF1–219 concentration, at least in the range from 25 to 400 nM (not shown). There was a modest (10%) increase in FXa generation going from 25 to 50 nM sTF1–219, but considering the affinity of FVIIa for sTF1–219 this is presumably just a manifestation of an increased degree of saturation of FVIIa. Above 50 nM sTF1–219, the FXa generation in the presence of PDI was constant, and not declining, indicating that a higher concentration of sTF1–219 does not increasingly compete with the small, virtually constant amount of FVIIa-sTF1–219 complexes for PDI. This indicates that there is no direct interaction between sTF1–219 and PDI. Previous findings with bovine PDI come from experiments in which Ca2+ is the sole metal ion present, but we and others have shown that the presence of Mg2+ enhances the rate of FX activation by FVIIa bound to sTF [17,18]. It was therefore of some interest to see whether the effect of bovine PDI on this system was influenced by the presence of 0.5 mM Mg2+. Mg2+ itself enhanced the rate of FX activation by almost 5-fold (to 0.25 nM/min), and the addition of PDI now merely resulted in a further 4- to 5-fold enhancement (≈ 1 nM/min) (Fig. 1C). However, the absolute, bovine PDI-induced increase in FX activation was similar with and without Mg2+. For some reason, the magnitudes of the effects of bovine PDI and Mg2+ depend on the length of the sTF used. In the presence of Ca2+ and Mg2+, 25 nM bovine PDI increased the rate of FX activation by 5 nM FVIIa in the presence of 25 nM sTF1–209 approximately 25-fold (from about 0.1 nM/min to 2.5 nM/min), whereas the rate was increased about 50-fold with Ca2+ alone (from 0.03–0.04 nM/ min to around 2 nM/min). Thus the effects when using sTF1–209 were more dramatic than those seen with sTF1–219 mainly because FX was activated faster with sTF1–209 than with sTF1–219 in the presence of bovine PDI. In addition, Mg2+ only stimulated FX activation by FVIIa-sTF1–209 by 2.5-fold, or half as much as when sTF1–219 was the cofactor. It should be mentioned that all experiments except for those represented by the two first columns in Fig. 1C were performed in the presence of 0.5 mM Mg2+ representing the most physiological conditions. Both annexin V and detergents abolished the stimulatory effect of bovine PDI. The inclusion of either 500 nM annexin V, 0.1% (v/v) Triton X-100 or 1% (w/v) CHAPS in the assay system returned the rate of FX activation by FVIIa-sTF1–219 in the presence of 25 nM bovine PDI to the level observed in the presence of additive but without PDI (Fig. 1D). This level was very similar to or slightly lower than that observed with neither

E. Persson additive nor PDI. At these additive concentrations, the ability of relipidated TF to support FX activation by FVIIa was also lost (not shown). PDI appeared to retain its oxidoreductase activity in the presence of Triton X-100 as measured in the insulin denaturation assay, indicative of that the effects of the detergents is not due to inhibition or destruction of PDI itself (not shown). Tween 80 (1% (v/v)) almost, but not completely, abolished the bovine PDI-induced enhancement and the cofactor activity of lipidated TF. These findings, together with the fact that the human PDI is recombinant and the bovine PDI is purified from liver, suggest that the different effects of human and bovine PDI are due to contaminating phospholipid in the bovine preparation and not to species differences.

The effect of bovine PDI requires the Gla domain of FVIIa and FX To support the hypothesis that phospholipid in the bovine PDI preparation is responsible for the promotion of FX activation by FVIIa-sTF1–219, reactions were carried out using N-terminally truncated forms of the enzyme and the substrate. The Gla domain mediates the binding of FVIIa and FX to phospholipid surfaces and FX activations performed with Gla-domainless FVIIa or Gla-domainless FX should not be stimulated by added phospholipid. Indeed, the removal of one Gla domain or the other had identical impact, namely a total abolishment of the bovine PDI-induced increase (Fig. 2). Even with sTF1–209 as the cofactor, which gave a more dramatic PDI effect on FX activation by full-length FVIIa, no PDI-induced stimulation was observed when using the Gla-domainless form of FVIIa or FX (not shown). While investigating the importance of the Gla domains for the bovine PDI-induced stimulation of FX activation, I found, as expected, a Mg2+-induced 5-fold enhancement of FX activation by FVIIa-sTF1–219 (Fig. 1C) but no effect of Mg2+ when des(1–44)FX was the substrate as reported [18]. However, in contrast to a previous report, there was only a marginal (≤20%) Mg2+-induced

Figure 2 Effect of bovine PDI on FX activation using variants of FVIIa and FX. The following enzyme/substrate pairs were used; FVIIa/FX (A, n = 6), des(1–38)-FVIIa/FX (B, n = 4), FVIIa/ des(1–44)-FX (C, n = 3), F4A-FVIIa/FX (D, n = 3), L8A-FVIIa/FX (E, n = 3) and F4A/L8A-FVIIa/FX (F, n = 3). The relative rate, shown as mean± standard deviation, is defined as the rate of FX activation in the presence of 25 nM bovine PDI divided by the rate observed in its absence. The buffer contained Ca2+ and Mg2+. See Materials and methods for more experimental details.

PDI does not promote FX activation

175

increase in FX activation by des(1–38)-FVIIa-sTF1–219. This experiment was repeated several times, using different batches of des(1–38)-FVIIa and des(1–44)-FVIIa, with very similar results. Consequently, I now favour a model where the Gla domain of both FVIIa and FX is required in order for Mg2+ to speed up FX activation by FVIIa-sTF1–219. Whether Mg2+ affects the interactions of both Gla domains with sTF1–219 or facilitates a direct interaction between the two Gla domains (or both) is not known. It is still possible that Mg2+ somehow improves the alignment between FX and FVIIa-sTF1–219 or facilitates FXa release and thereby makes the activation process more efficient. However, the concentrations and excess of sTF1–219 used, both in the present and previous study [18], and the affinity of FVIIa for sTF1–219 in the presence of Ca2+ (Kd =3–5 nM), as well as the fact that 1 mM Ca2+ suffices for optimal binding between the two proteins [19], rule out that the stimulation by Mg2+ is explained by an increased affinity of FVIIa for sTF1–219.

In order to obtain an additional supportive piece of evidence for a phospholipid-mediated enhancement of FVIIa-sTF1–219catalyzed FX activation by the bovine PDI preparation, FX activation was carried out at different substrate concentrations. The addition of phospholipid to the incubation mixture should result in a lowering of the Km for FX. It was obvious that the bovine PDI preparation reduced the Km value. Two separate experiments gave values of 1.4 and 2.2 μM, respectively, in the presence of 25 nM bovine PDI, whereas values far above 10 μM were obtained without PDI (Fig. 3). The FX concentration dependence of the bovine PDI-induced enhancement explains why the effect at 150 nM FX was greater than that at 1 μM, and the observed enhancement by bovine PDI at 1 μM FX is in good agreement with that seen by Versteeg and Ruf also using 1 μM FX [8].

The effect of bovine PDI requires an intact ω loop in FVIIa

Concluding remarks

The ω loop is a structural element in the Gla domain containing hydrophobic side chains instrumental for phospholipid membrane binding. In FVIIa, Phe4 and Leu8 together with Leu5 form the membrane attachment site, as do the corresponding residues in the Gla domain of homologous vitamin K-dependent coagulation proteins [20–23]. Although we could demonstrate that higher concentrations of F4A-, L8Aand F4A/L8A-FVIIa than of FVIIa were required to saturate lipidated TF, the substitution of alanine for phenylalanine in position 4 or leucine in position 8 did not affect the ability of FVIIa when bound to lipidated TF or sTF1–219 to activate FX. However, either change sufficed to prevent the bovine PDI preparation from enhancing FX activation by FVIIa-sTF1–219 (Fig. 2). The requirement for full-length FVIIa and FX, and even for an intact ω loop, strongly support that contaminating phospholipid in the bovine PDI preparation is responsible for the stimulatory effect.

Bovine PDI reduces the Km of FVIIa-sTF1–219 for FX

Here, I present convincing evidence for a phospholipid contaminant in the bovine PDI preparation which, rather than bovine PDI itself, mediates the enhancement of FX activation by FVIIa-sTF1–219. The enhancer being phospholipid is supported by a) the neutralizing effect of annexin V and detergents, b) the need for the Gla domain/intact ω loop, and c) the reduction of the Km for FX. Indirect evidence is also provided by the inability of the bovine PDI preparation to promote FX activation by FVIIa bound to lipidated TF and, very importantly, by the inability of recombinant human PDI to stimulate FX activation by FVIIa-sTF1–219. The fact that a larger excess of sTF does not lead to a decreased FX activation also speaks against a direct interaction between PDI and TF. During the revision of this manuscript, a publication appeared which supports the present conclusions [24]. Whether extracellular PDI plays a role as an oxidoreductase in the injury response and regulation of cryptic TF in the hemostatic and thrombotic processes is another story, a field of intense research [4–7,25] and a topic of debate.

Acknowledgements The author thanks Anette Østergaard for excellent technical assistance and Dr. Henrik Østergaard for discussions and constructive suggestions.

References Figure 3 Effect of bovine PDI as a function of FX concentration. The graph shows the total amount of FXa formed during a 10-min incubation of 50 pM FVIIa and 25 nM sTF1–219 with various concentrations of FX in the absence (○) and presence of 25 nM bovine PDI (●). The data were fitted to Michaelis-Menten kinetics. The inset shows data from a second experiment to illustrate the consistency of the obtained results.

[1] Monroe DM, Key NS. The tissue factor-factor VIIa complex: procoagulant activity, regulation, and multitasking. J Thromb Haemost 2007;5:1097—105. [2] Gruber CW, Čemažar M, Heras B, Martin JL, Craik DJ. Protein disulfide isomerase: the structure of oxidative folding. Trends Biochem Sci 2006;31:455—64. [3] Hatahet F, Ruddock LW. Substrate recognition by the protein disulfide isomerases. FEBS J 2007;274:5223—34.

176 [4] Chen VM, Ahamed J, Versteeg HH, Berndt MC, Ruf W, Hogg PJ. Evidence for activation of tissue factor by an allosteric disulfide bond. Biochemistry 2006;45:12020—8. [5] Ahamed J, Versteeg HH, Kerver M, Chen VM, Mueller BM, Hogg PJ, et al. Disulfide isomerization switches tissue factor from coagulation to cell signaling. Proc Natl Acad Sci USA 2006;103:13932—7. [6] Reinhardt C, von Brühl ML, Manukyan D, Grahl L, Lorenz M, Altmann B, et al. Protein disulfide isomerase acts as an injury response signal that enhances fibrin generation via tissue factor activation. J Clin Invest 2008;118:1110—22. [7] Pendurthi UR, Ghosh S, Mandal SK, Rao LVM. Tissue factor activation: is disulfide bond switching a regulatory mechanism? Blood 2007;110:3900—8. [8] Versteeg HH, Ruf W. Tissue factor coagulant function is enhanced by protein-disulfide isomerase independent of oxidoreductase activity. J Biol Chem 2007;282:25416—24. [9] Thim L, Bjoern S, Christensen M, Nicolaisen EM, LundHansen T, Pedersen AH, et al. Amino acid sequence and posttranslational modifications of human factor VIIa from plasma and transfected baby hamster kidney cells. Biochemistry 1988;27:7785—93. [10] Sakai T, Lund-Hansen T, Thim L, Kisiel W. The γ-carboxyglutamic acid domain of human factor VIIa is essential for its interaction with cell surface tissue factor. J Biol Chem 1990;265:1890—4. [11] 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—60. [12] 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—40. [13] Morita T, Jackson CM. Preparation and properties of derivatives of bovine factor X and factor Xa from which the γ-carboxyglutamic acid containing domain has been removed. J Biol Chem 1986;261:4015—23. [14] 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—3.

E. Persson [15] 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—6. [16] Persson E, Kjalke M, Olsen OH. Rational design of coagulation factor VIIa variants with substantially increased intrinsic activity. Proc Natl Acad Sci USA 2001;98:13583—8. [17] 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—88. [18] Persson E, Østergaard A. Mg2+ binding to the Gla domain of factor X influences the interaction with tissue factor. J Thromb Haemost 2007;5:1977—8. [19] Persson E. Characterization of the interaction between the light chain of factor VIIa and tissue factor. FEBS Lett 1997;413: 359—63. [20] Sunnerhagen M, Forsén S, Hoffrén AM, Drakenberg T, Teleman O, Stenflo J. Structure of the Ca2+—free Gla domain sheds light on membrane binding of blood coagulation proteins. Nat Struct Biol 1995;2:504—9. [21] Freedman SJ, Blostein MD, Baleja JD, Jacobs M, Furie BC, Furie B. Identification of the phospholipid binding site in the vitamin K-dependent blood coagulation protein factor IX. J Biol Chem 1996;271:16227—36. [22] Falls LA, Furie BC, Jacobs M, Furie B, Rigby AC. The ω-loop region of the human prothrombin γ-carboxyglutamic acid domain penetrates anionic phospholipid membranes. J Biol Chem 2001;276:23895—902. [23] Grant MA, Baikeev RF, Gilbert GE, Rigby AC. Lysine 5 and phenylalanine 9 of the factor IX ω-loop interact with phosphatidylserine in a membrane-mimetic environment. Biochemistry 2004;43:15367—78. [24] Kothari H, Sen P, Pendurthi U, Rao LVM. Bovine protein disulfide isomerase-enhanced tissue factor coagulant function: is phospholipid contaminant in it the real culprit? Blood 2008;111:3295—6. [25] Cho J, Furie BC, Coughlin SR, Furie B. A critical role for extracellular protein disulfide isomerase during thrombus formation in mice. J Clin Invest 2008;118:1123—31.