S. M. Kanse, Institute for Biochemistry, Justus-Liebig-University Giessen, Friedrichstrasse 24, 35392 Giessen, Germany Fax: +49 641 9947509 Tel: +49 641 9947521 E-mail: email@example.com
Factor VII-activating protease (FSAP) circulates as an inactive zymogen in the plasma. FSAP also regulates fibrinolysis by activating pro-urokinase or cellular activation via cleavage of platelet-derived growth factor BB (PDGF-BB). As the Marburg I polymorphism of FSAP, with reduced enzymatic activity, is a risk factor for atherosclerosis and liver fibrosis, the regulation of FSAP activity is of major importance. FSAP is activated by an auto-catalytic mechanism, which is amplified by heparin. To further investigate the structural requirements of polyanions for controlling FSAP activity, we performed binding, activation and inhibition studies using heparin and derivatives with altered size and charge, as well as other glycosaminoglycans. Heparin was effective in binding to and activating FSAP in a size- and charge density-dependent manner. Polyphosphate was more potent than heparin with regard to its interactions with FSAP. Heparin was also an effective co-factor for inhibition of FSAP by plasminogen activator inhibitor 1 (PAI-1) and antithrombin, whereas polyphosphate served as co-factor for the inhibition of FSAP by PAI-1 only. For FSAP-mediated inhibition of PDGF-BB-induced vascular smooth muscle cell proliferation, heparin as well as a polyphosphate served as efficient co-factors. Native mast cell-derived heparin exhibited identical properties to those of unfractionated heparin. Despite the strong effects of synthetic polyphosphate, the platelet-derived material was a weak activator of FSAP. Hence, negatively charged polymers with a high charge-to-size ratio are responsible for the activation of FSAP, and also act as co-factors for its inhibition by serine protease inhibitors.
Factor VII-activating protease (FSAP) is a serine protease that is predominantly expressed in the liver. It circulates as an inactive zymogen with a concentration of 12 μg·mL−1 in the plasma, and is known to activate factor VII and pro-urokinase [1,2]. It was first purified by its ability to bind to hyaluronic acid, and was therefore designated as hyaluronic acid binding protein 2 (HABP-2) . Activation of FSAP requires cleavage between residues R313 and I314, separating the light chain and the heavy chain .
Negatively charged polyanions such as heparin [4,5], nucleic acids [6,7] and dextran sulfate  bind to FSAP. This interaction leads to auto-catalytic activation [4,5], followed by auto-proteolysis. This propensity to partial proteolysis has been used to determine the domain responsible for binding to heparin and RNA. Multiple regions of FSAP contribute to polyanion binding, but the epidermal growth factor like-3 (EGF3) domain with a cluster of positively charged amino acids is particularly important .
Although FSAP was initially isolated on a hyaluronic acid column , no information is available as to how hyaluronic acid can bind to FSAP, nor whether it can activate FSAP . The concentration of hyaluronic acid, as well as its transition from the high- to low-molecular-weight form, is related to the regulation of angiogenesis, atherosclerosis, restenosis and inflammation . FSAP activation is also mediated by nucleic acids, with RNA having a stronger effect than DNA [6,7]. Heparin is the most extensively studied polyanion with respect to FSAP function. It has been shown that unfractionated heparin is a strong activator of FSAP, but low-molecular-weight heparin has not been systematically tested. The role of the more ubiquitous heparan sulfate and other glycosaminoglycans is also not known.
Polyphosphate (PolyP) is a linear polymer of orthophosphate (Pi) residues linked by high-energy phosphoanhydride bonds, found in many cell types . PolyP, with an approximate chain length from 70–75 phosphate units, is stored in platelet-dense granules  and released upon platelet activation. PolyP can amplify coagulation by activation of the contact factor pathway, as well as activation of factor V, inhibition of the anticoagulant function of tissue factor pathway inhibitor (TFPI), and enhancing the activity of thrombin-activated fibrinolysis inhibitor (TAFI) .
Once activated, FSAP can be rapidly inhibited by serine protease inhibitors (SERPINs), such as α1-antitrypsin, α2-antiplasmin, antithrombin (AT) and C1 inhibitor [4,12–14], as well as plasminogen activator inhibitor 1 (PAI-1)  and protease nexin 1 . AT and α2-antiplasmin were shown to be efficient inhibitors in the presence of heparin , whereas PAI-1 was shown to be an inhibitor only in the presence of RNA but not heparin .
The presence of a naturally occurring polymorphism in the FSAP gene leading to an amino acid exchange (G534E, or Marburg I polymorphism) results in diminished proteolytic activity towards factor VII, pro-urokinase  and PDGF-BB (platelet-derived growth factor BB) . The Marburg I polymorphism is associated with a higher risk for carotid stenosis , and, in comparison to wild-type FSAP, is not able to inhibit neointima formation in a mouse model . Similarly, Marburg I FSAP is associated with advanced liver fibrosis, which may be due to its inability to inhibit PDGF-BB-mediated proliferation of hepatic stellate cells . These findings indicate the importance of FSAP enzymatic activity with respect to its function in vivo. However, it is not clear which polyanions are relevant for the regulation of FSAP activity. This prompted us to investigate the requirements for FSAP interaction with polyanions known to be present in atherosclerotic arterial wall and/or fibrotic liver, and also to define the molecular basis of the binding, activation and regulation mechanisms.
FSAP binding to polyanions
Electrophoretic mobility shift assays were performed to characterize the interaction between FSAP and various polyanions. Preincubation of FSAP with unfractionated heparin, low-molecular-weight heparin, PolyP65 or PolyP35 induced a shift in the mobility of FSAP in polyacrylamide gels with or without urea. Other polyanions had no influence at all. When BSA was used as a control, none of the polyanions induced a shift in the BSA band (Fig. 1A). Concentration-dependent analysis indicated that the EC50 was 95 ± 7 nm for the shift with unfractionated heparin and 28 ± 3 nm for PolyP65 (Fig. 1B and Figs S1 and S2).
To examine whether the various polyanions use the same region in the FSAP molecule for binding, we performed competition binding assays in which binding of biotinylated unfractionated heparin to FSAP was measured (Fig. 1C). Unfractionated heparin competed with biotinylated heparin for binding to FSAP, whereas low-molecular-weight heparin showed low competition (Fig. 1C, upper panel). PolyP competed for this binding in a chain length-dependent manner. All other heparin derivatives, as well as chondroitin sulfate, dermatan sulfate, polysialic acid, heparan sulfate and hyaluronic acid, showed no competition, indicating no binding to FSAP (Fig. 1C, lower panel, and Fig. S3A). Thus, using gel-shift and competition binding assays, it was demonstrated that binding to FSAP depends on the size and charge density of the macromolecule.
Activation of FSAP by various polyanions
We next investigated all the polyanions described above with respect to their ability to activate FSAP. Unfractionated heparin was a strong activator, low-molecular-weight heparin activated FSAP to a smaller extent, and all other heparin-derivatives exhibited no activation (Fig. 2, upper panel). PolyP showed potent activation of FSAP in a chain length-dependent manner. There was a 4–6-fold increase in Vmax with unfractionated heparin and PolyP65, with no change in KM (Fig. S2). Heparan sulfate and dermatan sulfate showed weak activation of FSAP at high concentrations (Fig. 2, middle panel). Polysialic acid and hyaluronic acid did not activate FSAP (Fig. 2, lower panel). N-acetyl heparin, de-N-sulfated heparin, N-acetyl-de-O-sulfated heparin, polysialic acid and hyaluronic acid totally failed to increase FSAP activity.
To assess the specificity of the PolyP effect, it was degraded using calf intestinal phosphatase, which is also a highly active exopolyphosphatase . The accelerating effect of PolyP on FSAP activity was decreased by phosphatase pretreatment in a time- and dose-dependent manner (Fig. S4). As a control, we observed that phosphatase treatment did not influence unfractionated heparin-mediated activation of FSAP (Fig. S4). Hence, the effect of PolyP was not due to a contaminant. These studies show that the pattern of binding of polyanions to FSAP is identical to the pattern of their ability to activate FSAP.
Polyanions as co-factors for the inhibition of FSAP by PAI-1 and AT
SERPINs exhibit enhanced or altered substrate specificity in the presence of heparin or other co-factors . To examine the co-factor function of polyanions with respect to FSAP inhibition, active two-chain FSAP was preincubated with PAI-1 or AT with or without various concentrations of polyanions. Inhibition of FSAP by PAI-1 was increased by unfractionated heparin, low-molecular-weight heparin and to a lower extent by N-acetyl heparin (Fig. 3A, upper panel). PolyP exhibits strong co-factor function for the inhibition of FSAP by PAI-1 in a chain length-dependent manner. The IC50 of PAI-1 for the inhibition of FSAP was halved by unfractionated heparin and PolyP65 (Fig. S5). Heparan sulfate was a co-factor at high concentrations (Fig. 3A, lower panel), and dermatan sulfate and polysialic acid at even higher concentrations (Fig. S3B), but hyaluronic acid had no effect at all (Fig. 3A, lower panel).
In the case of FSAP inhibition by AT, only unfractionated heparin and heparan sulfate were able to serve as co-factors (Fig. 3B). PolyP and other tested polyanions showed no co-factor properties for the AT-dependent FSAP inhibition (Fig. 3B, lower panel and Fig. S3C). The activity of FSAP was increased by low-molecular-weight heparin, N-acetyl heparin (Fig. 3B, upper panel) and PolyP (Fig. 3B, lower panel) even in the presence of AT.
To consolidate these findings, real-time interaction studies were performed using surface plasmon resonance (SPR). These results confirm that FSAP interacts with AT only in the presence of unfractionated heparin (KA of ∼ 2.9 × 107 [1/M]) but not in the presence of PolyP. In contrast, FSAP interacts with PAI-1 without a co-factor (KA of ∼ 1.6 × 107 [1/M]) in the presence of unfractionated heparin (KA of ∼ 3.2 × 107 [1/M]) as well as in the presence of PolyP (KA of ∼ 97 × 107 [1/M]) (Fig. 3C). Hence, polyanions can selectively promote inhibition of the enzymatic activity of FSAP.
Polyanions as co-factors for the FSAP-dependent inhibition of VSMC proliferation
A major function of FSAP is the specific proteolytic cleavage and inactivation of PDGF-BB , and this process is enhanced by heparin and RNA . We observed that low-molecular-weight heparin and heparan sulfate also increase the inhibitory effect of FSAP on proliferation of vascular smooth muscle cells (VSMC), but to a lower extent compared to unfractionated heparin. PolyP also promoted the inhibitory effect of FSAP on VSMC proliferation, whereas de-N-sulfated heparin and hyaluronic acid were ineffective (Fig. 4). The ability of each polyanion to inhibit cell proliferation matched the respective pattern of FSAP binding and activation.
Assessment of mast cell heparin and platelet PolyP as co-factors for FSAP function
Mast cell-derived macromolecular heparin and platelet-derived PolyP were isolated as native substances and tested for their interaction with FSAP. The mast cell-derived heparin bound to FSAP, as indicated by a mobility shift in native polyacrylamide gels (Fig. 5A, upper panel). When compared to unfractionated heparin, mast cell heparin was even more efficient with respect to competition of biotinylated heparin binding to immobilized FSAP (Fig. 5A, middle panel) and FSAP activation (Fig. 5B, upper panel).
In mobility shift assays, platelet-derived PolyP bound to FSAP weakly (Fig. 5A, upper panel). However, it competed with biotinylated heparin for binding to immobilized FSAP more strongly than its synthetic analogue PolyP65 did (Fig. 5A, lower panel). Unexpectedly, activation of FSAP by native platelet-derived PolyP was much lower when compared to the synthetic material (Fig. 5B, lower panel). Thus, mast cell-derived heparin was identical to unfractionated heparin for all aspects investigated, but there were differences between platelet-derived and synthetic PolyP.
Genetic studies show that the presence of the Marburg I single-nucleotide polymorphism is a risk factor for carotid stenosis  and liver fibrosis . This isoform of FSAP exhibits reduced enzymatic activity , indicating that the local proteolytic activity of FSAP may play a crucial role in development of the disease state. Therefore, it is important to understand the regulation of FSAP activity in order to define its pathophysiological role. Polyanions have been shown to play a key role in regulating FSAP activity by promoting auto-catalytic activation. In the present study, we systematically characterized the effects of various polyanions on FSAP activity.
Heparin and other glycosaminoglycans
The binding to FSAP and the subsequent activation of FSAP by heparin depends on its size and overall negative charge. Low-molecular-weight heparin exhibits lower potential for binding to and activating FSAP. The heparin homologues N-acetyl heparin, de-N-sulfated heparin and N-acetyl-de-O-sulfated heparin, which have the same size but reduced negative charge, neither bind to nor activate FSAP (Fig. 6). The proteoglycan heparan sulfate has an even lower negative charge, compared to unfractionated heparin, and exhibits weak FSAP binding and activation. Mast cell-derived heparin has a higher charge than unfractionated heparin, and exhibits a stronger ability to bind to and activate FSAP .
Chondroitin sulfate, dermatan sulfate and polysialic acid also have a less negative charge density than unfractionated heparin and show no FSAP binding or activation potential (Fig. 6). FSAP was first purified based on its binding to hyaluronic acid . In the present study, we demonstrate that there is no tight interaction between hyaluronic acid and FSAP, most likely due to the relatively low negative charge density in the polyanion. Its isolation on hyaluronic acid columns could be due to altered physical properties of immobilized hyaluronic acid. The significance of these results is that the very ubiquitous heparan sulfate proteoglycans and other matrix-associated glycosaminoglycans play no role in the regulation of FSAP activity. This is rather related to the proximity and activation state of mast cells that secrete heparin, such as in atherosclerotic plaques .
PolyP was a more potent activator of FSAP than heparin. PolyP65 was the most active form of PolyP, with smaller forms showing diminished activity. Degradation by phosphatases decreased its properties with respect to FSAP binding and activation, and any influence on FSAP activity was completely neutralized. In order to put these findings in a pathophysiological context, we compared the activity of synthetic PolyP with that of native platelet-derived material. Platelet-derived PolyP exhibited quite anomalous properties compared to synthetic PolyP. In gel-shift assays, it demonstrated weak binding, but was as efficient as synthetic PolyP in competing for heparin binding to FSAP. Native PolyP was a very weak activator of FSAP compared to the synthetic version. One reason for this discrepancy between synthetic and native PolyP could be that synthetic PolyP65 is a heterogeneous mixture, with polymers up to 200 units, whereas native PolyP is extremely pure and has a more homogeneous size with 70–75 units [10,27]. In addition to their difference in size, we cannot exclude the possibility of a contaminant that has a confounding effect on the interaction of native PolyP with FSAP. No comparable data exist in the literature, as this is one of the first studies to compare the activities of synthetic with platelet-derived PolyP. Given the robust activity of synthetic PolyP, the role of endogenous platelet-derived material needs to be investigated further.
SERPINs such as protease nexin 1 and PAI-1 can efficiently inhibit FSAP. Whereas protease nexin 1 inhibits proteases independently of any co-factor , PAI-1 is known to require heparin as a co-factor for inhibition of some of its targets such as thrombin . The co-factor effect of heparin is due to a change in the conformation of the SERPIN as well as the ability of heparin to co-join the protease with the inhibitor. Previously published data showed that heparin was not a co-factor for PAI-1-dependent inhibition of FSAP . In this study, we demonstrate that both heparin and polyphosphate are potent co-factors for the inhibition of FSAP by PAI-1. A reduction in size and charge density in heparin led to lower inhibition of FSAP by PAI-1. AT inhibits FSAP only in the presence of heparin but not PolyP. The size and negative charge of heparin has an even greater importance for the interaction with AT, as indicated by the fact that low-molecular-weight heparin and N-acetyl heparin promote an increase in FSAP activity rather than inhibiting it. Thus, polyanion binding to SERPINs, over and above their binding to FSAP, plays a decisive role in mediating its inhibition.
PolyP increased the inhibition of FSAP by PAI-1 but not by AT. Whereas heparin changes the tertiary structure of AT , PolyP was shown to be unable to induce any conformational changes in AT, as determined by measurement of the intrinsic protein fluorescence of AT incubated with PolyP (F. A. Ruiz, unpublished results). Both polyanions decreased the IC50 for the inhibition of FSAP by PAI-1 twofold. SERPINs inhibit their target protease by a suicide substrate mechanism that involves a 1 : 1 formation of an irreversible covalent complex . Only protease-inactive mutants show reversible binding to SERPINs , and the FSAP–PAI-1 complex demonstrated some dissociation in our experiments (Fig. 3C), indicating some deviation from the classical model of protease–inhibitor interactions. Hence, the overall inhibition of FSAP depends not only on the inhibitor but also on the presence of an appropriate co-factor in the vicinity of FSAP.
The two major polyanions, heparin and PolyP, use the same binding region in the FSAP molecule, as revealed by the competition binding assay. Charge density, size and also conformational flexibility determine the affinity of this interaction. Other matrix-derived polyanions were not effective. Binding to polyanions was also observed in the presence of a strong denaturant, urea, indicating a strong charge interaction. The region of FSAP that is probably responsible for this binding is the EGF3 domain, which contains a positively charged cluster of amino acids, although other regions of FSAP promote this interaction . Using a recombinant EGF3 domain deletion mutant of FSAP, no activation of FSAP was obtained with either heparin or with PolyP , further confirming the involvement of this region in polyanion binding and activation. Polyanions strongly reduced the proliferative activity of PDGF-BB in the presence of FSAP. This could explain the influence of polyanions such as heparin on smooth muscle proliferation in vivo , and a similar function is expected for PolyP. As a lowering in FSAP activity is correlated with diseases [19,20], these new insights into the regulation of FSAP activity will lead to increased understanding of FSAP function under physiological und pathophysiological conditions. Identification of specific size, sequence and charge requirements may allow rational design of polyanions with higher specificity for the regulation of FSAP activity.
FSAP was isolated as described previously . PolyP 65-mer (molecular mass ∼ 6.6 × 103 Da) and PolyP 15-mer (molecular mass ∼ 1.5 × 103 Da) were obtained from Sigma (Munich, Germany), and PolyP 35-mer (molecular mass ∼ 3.5 × 103 Da) was obtained from Roth (Karlsruhe, Germany). Unfractionated heparin (molecular mass ∼ 15 × 103 Da), heparan sulfate, dermatan sulfate, chondroitin sulfate C, low-molecular-weight heparin (molecular mass ∼ 3 × 103 Da), N-acetyl heparin, de-N-sulfated heparin and N-acetyl-de-O-sulfated heparin (all molecular masses ∼ 15 × 103 Da), hyaluronic acid (molecular mass ∼ 1 × 105 Da) from human placenta or rooster comb and biotinylated heparin albumin were obtained from Sigma. Polysialic acid (molecular mass ≤ 38 × 103 Da) was separated from oligosialic acid as described previously . Calf intestinal alkaline phosphatase was obtained from Fermentas (St Leon-Rot, Germany). PAI-1 was generously provided by Dr Paul Declerck (Katholieke Universiteit, Leuven, Belgium). AT was obtained from CSL Behring (Marburg, Germany).
Isolation of platelet-derived PolyP and mast cell-derived macromolecular heparin
Platelet homogenates were prepared as described previously . After centrifugation at 19 000 g, the pellet was used to extract native PolyP using perchloric acid . PolyP was further purified on an OMIX C18 100 μL tip (Varian, Lake Forest, CA) before use. Native macromolecular heparin (molecular mass 75 × 104 Da; range 5 × 105–1 × 106) was purified from granule remnants of rat serosal mast cells, as described previously . Briefly, granule remnants were treated with 2 m KCl to release heparin-bound molecules (notably chymase and other proteases) from heparin proteoglycans and to disintegrate the granule remnants into heparin proteoglycan monomers . The incubation mixture was then applied to a Sephacryl S-200 column (GE Healthcare Life Sciences, Uppsala, Sweden) column for isolation and separation of heparin proteoglycans. The residual chymase activity in the heparin proteoglycan fraction was inhibited using phenylmethanesulfonyl fluoride.
Electrophoretic mobility shift assays to detect polyanion binding to FSAP
Polyacrylamide–bisacrylamide (37.5:1) native gels (6–10%) were poured with Tris/borate/EDTA (TBE) (90 mm Tris, 90 mm boric acid, 2 mm EDTA, pH 8.3), with or without 6.7 m urea, in a horizontal gel chamber. FSAP (5 μg) was preincubated for 30 min with or without respective polyanions (10 μg), native sample buffer (TBE with sucrose and bromphenol blue) was added, and samples were loaded onto the gel. After separation, the gel was stained either with toluidine blue to visualize polyanions (not shown) or with Coomassie brilliant blue to visualize proteins. Densiometric analysis was performed to determine the affinity of these interactions.
Competition of heparin binding to immobilized FSAP with various polyanions
Microtiter plates were coated with 50 μL of a 10 μg·mL−1 FSAP solution in 100 mm sodium carbonate (pH 9.5) overnight at 4°C. Wells were washed, and non-specific binding sites were blocked with NaCl/Tris (25 mm Tris/HCl, pH 7.5, 150 mm NaCl) containing 3% w/v BSA for 1 h. Biotinylated heparin albumin (0.5 ng·mL−1) mixed with dilutions of polyanions was allowed to bind for 1 h at room temperature in NaCl/Tris containing 0.1% w/v BSA, after which the plates were washed three times with NaCl/Tris containing 0.1% w/v Tween-20 (NaCl/Tris-T). Bound biotinylated heparin albumin was detected using peroxidase-conjugated streptavidin (DAKO, Glostrup, Denmark) and an immunopure TMB substrate kit (Thermo Fischer Scientific, Rockford, IL, USA).
FSAP enzyme activity assay
FSAP activity assays were performed as described previously . In brief, microtiter plates were blocked with NaCl/Tris containing 3% w/v BSA for 1 h, and washed with NaCl/Tris-T. The standard assay system consisted of NaCl/Tris, 1 μg·mL−1 FSAP and 0.2 mm of the chromogenic substrate S-2288 (H-d-isoleucyl-l-prolyl-l-arginine-p-nitroanilinedihydrochloride) (Haemochrome, Essen, Germany) and was followed over a period of 60 min at 37°C at 405 nm in an EL 808 microplate reader (BioTek Instruments, Winooski, VT, USA). If an inhibitor was used, this was added together with FSAP to the plates with and without polyanionic co-factor 30 min before adding the chromogenic substrate.
Characterization of FSAP–inhibitor interaction using surface plasmon resonance (SPR) technology
Immobilization on sensor chips, and association and dissociation of interacting biomolecules, were followed in real time by monitoring the change in SPR signal expressed in resonance units (RU). All experiments were performed at 25°C. To prepare the sensor chip surface, antibodies to FSAP or isotype controls were immobilized on a CM5 research-grade chip (Biacore/GE Healthcare, Freiburg, Germany) at 10 000 RU, via amino coupling (Biacore) and using HBS-N (20 mm Hepes, pH 7.4, 100 mm NaCl), as running buffer. Interaction analysis experiments were performed at a flow rate of 20 μL·min−1 using HBS-P [20 mm Hepes, pH 7.4, 100 mm NaCl, 0.05% Surfactant P20 (Biacore cat.nr.:BR-1000-54)] supplemented with 2 mm CaCl2 as running buffer. FSAP (25 μL, 10 μg·mL−1) was captured on the immobilized antibodies, and then AT or PAI-1 (25 μL, 0–5 μg·mL−1) were injected alone and in the presence of unfractionated heparin or PolyP (10 μg·mL−1). Sensorgrams were analyzed using BIAevaluation software version 3.2 RC1. Kinetic constants were obtained using the Langmuir binding model 1:1.
Mouse vascular smooth muscle cells (VSMC) were cultured in Iscove’s modified medium (Invitrogen, Karlsruhe, Germany) with 10% v/v fetal calf serum (HyClone, Logan, UT, USA), 10 U·mL−1 penicillin, 10 μg·mL−1 streptomycin, 2 mm l-glutamine and 1 mm sodium pyruvate (Invitrogen). Cells were growth-arrested in serum-free medium for 18 h prior to experiments.
DNA synthesis assays
VSMC were stimulated for 36 h with the test substances in medium containing 0.2% fetal calf serum. For the last 24 h, 5-bromo-2-deoxyuridine (BrdU) was added, and the cells were processed using a BrdU detection kit (Roche Diagnostics, Mannheim, Germany) as described by the manufacturer.
The assistance of Susanne Tannert-Otto is greatly appreciated. We are grateful to Dr Paul Declerck (Department of Pharmaceutical Sciences, Katholieke Universiteit, Leuven, Belgium) for providing PAI-1. This study was financed by a grant from the Deutsche Forschungsgemeinschaft to S.M.K. (SFB 547: C14). Wihuri Research Institute is maintained by the Jenny and Antti Wihuri Foundation (Helsinki, Finland).