Structural and Energetic Characteristics of the Heparin-binding Site in Antithrombotic Protein C*

Human activated protein C (APC) is a key component of a natural anticoagulant system that regulates blood coagulation. In vivo, the catalytic activity of APC is regulated by two serpins, α1-antitrypsin and the protein C inhibitor (PCI), the inhibition by the latter being stimulated by heparin. We have identified a heparin-binding site in the serine protease domain of APC and characterized the energetic basis of the interaction with heparin. According to the counter-ion condensation theory, the binding of heparin to APC is 66% ionic in nature and comprises four to six net ionic interactions. To localize the heparin-binding site, five recombinant APC variants containing amino acid exchanges in loops 37, 60, and 70 (chymotrypsinogen numbering) were created. As demonstrated by surface plasmon resonance, reduction of the electropositive character of loops 37 and 60 resulted in complete loss of heparin binding. The functional consequence was loss in heparin-induced stimulation of APC inhibition by PCI, whereas the PCI-induced APC inhibition in the absence of heparin was enhanced. Presumably, the former observations were due to the inability of heparin to bridge some APC mutants to PCI, whereas the increased inhibition of certain APC variants by PCI in the absence of heparin was due to reduced repulsion between the enzymes and the serpin. The heparin-binding site of APC was also shown to interact with heparan sulfate, albeit with lower affinity. In conclusion, we have characterized and spatially localized the functionally important heparin/heparan sulfate-binding site of APC.

The protein C pathway is a functionally important anticoagulant system that regulates blood coagulation in vivo. The key component of this pathway is the vitamin K-dependent protein C (1-3). Protein C circulates in plasma as a zymogen to a serine protease that has anticoagulant properties. Protein C is a multidomain molecule that is composed of two disulfide-linked chains. The light chain comprises a ␥-carboxyglutamatic acid domain and two epidermal growth factor (EGF)-like domains, whereas the heavy chain is composed of the activation peptide and a serine protease domain (4). Protein C is activated on endothelial cells by thrombin bound to thrombomodulin. Acti-vated protein C (APC) 1 regulates blood coagulation by cleaving and inhibiting two cofactors, activated factor V (FVa) and activated factor VIII (FVIIIa) (5), which serve as phospholipidmembrane-bound cofactors to factor Xa (FXa) and factor IXa (FIXa), respectively. FXa is the enzyme that activates prothrombin to thrombin, whereas FIXa converts FX to its active form (1,6).
In vivo, the proteolytic activity of APC is regulated by two serpins, namely ␣1-antitrypsin and protein C inhibitor (PCI) (7,8). Inhibition of APC by PCI is potentiated by the glycosaminoglycan heparin, whereas the inhibition by ␣1-antitrypsin is insensitive to the presence of heparin. Structurally, heparin is heterogeneous in nature and is composed of long, highly negatively charged, unbranched polysaccharide chains. It is hypothesized that heparin binds to both PCI and APC during PCImediated inhibition of APC, thus guiding the encounter of these proteins via a template mechanism (9,10). We therefore expect the formation of a ternary complex similar to the one suggested between antithrombin, heparin, and thrombin (11,12). Recently some residues in APC were implied to interact directly with heparin during the PCI-induced inhibition of APC (13,14), but a more complete definition of a heparin-binding site(s) in APC and energetic characteristics of the heparin interaction with APC were lacking.
Binding of heparin to proteins is usually ionic in nature, involving the side chains of clustered basic amino acids on the protein and negatively charged groups on the heparin molecule. Amino acid sequence patterns such as XBBXBX and XBBBXXBX (B denotes basic, and X denotes nonbasic residues) are potential heparin recognition sites (15,16). Alternatively, the basic residues may be located far apart in the linear sequence but are topological neighbors in the three-dimensional structure (17)(18)(19)(20). In the three-dimensional structure of APC, a basic cluster of amino acids is located on loops 37, 60, and 70 (chymotrypsinogen nomenclature) (see Fig. 1) (21)(22)(23). Heparin-binding sites sharing similar distribution of charged amino acids are present on the surface of many proteins that bind heparin, e.g. hepatocyte growth factor (24). Multiple-sequence and structural alignments of APC with other serine proteases formed the basis for our mutagenesis strategy, aimed at identification of the APC heparin-binding site.
In the present study, site-directed mutagenesis and recombinant human protein C expression were used to localize the heparin-binding site in APC. A cluster of lysines located on loops 37 and 60 was found to be crucial for heparin binding.
Characterization of the binding of heparin to APC under different salt concentrations provided evidence for the electrostatic signature of the interaction. Reduction of the electropositive character of these loops resulted in lost heparin binding and reduced heparin stimulation of APC inhibition by PCI.
Recombinant Human Protein C Variants-Three new recombinant human protein C variants, K37S/K38Q/K39Q, K37S/K38Q/K39Q/ K62N/K63D, and R74Q, were created by site-directed in vitro mutagenesis using the polymerase chain reaction technique. In brief, full-length human protein C WT cDNA (1425 base pairs) was inserted into the BcII site of the eukaryotic expression vector pGT-hyg (Eli Lilly). The following primers were used to introduce the mutations into the human protein C WT cDNA: sense primer, 5Ј-CTG CTG GAC TCA AGC CAG CAG CTG GCC TGC GGG for variant K37S/K38Q/K39Q; sense primer, 5Ј-GAG TAT GAC CTG CAG CGC TGG GAG AAG for variant R74Q. The construction of variants E60aS/S61R and K62N/K63D have been previously described (13,14). The human protein C variant K37S/K38Q/K39Q-cDNA and sense primer 5Ј-TGC ATG GAT GAG TCC AAC GAC CTC CTT GTC AGG CTT was used to construct mutant K37S/K38Q/K39Q/K62N/K63D. Exchanged nucleotides as listed above are marked in bold, and codons corresponding to the exchanged amino acids are underlined. The sequence of cDNAs coding for WT protein C and all variants were confirmed using automated DNA sequencing (PerkinElmer Life Sciences) before transfection.
Purification of Recombinant Human WT Protein C and Protein C Variants-Cells were grown in Optimem Glutamax (Life Technologies, Inc.) medium supplemented with 10 g/ml vitamin K 1 . After 72 h, the conditioned medium was collected, and the recombinant expressed human protein C was purified from the supernatant by anion-exchange chromatography and barium citrate adsorption as previously described (26). The purity and integrity of the isolated protein C was evaluated by SDS-polyacrylamide gel electrophoresis and Western blot analysis. Protein C concentrations were quantified by measurement of absorbance at 280 nm using an extinction coefficient of 14.5 (280 nm, 1%, 1 cm). In addition, activated protein C concentrations were determined by a chromogenic assay measuring the rate of S-2366 (Chromogenix) hydrolysis.
SDS-PAGE and Immunoblotting Analysis-Proteins subjected to 10% SDS-PAGE were either detected by silver staining or transferred to polyvinylidene difluoride membranes (Millipore) for Western blotting. After transfer of the proteins to the polyvinylidene difluoride, membranes were quenched with Tris-buffered saline containing 3% (v/v) fish gelatin and 0.1% (v/v) Tween 20 and then incubated with antibodies. The proteins were detected using a polyclonal antibody against human protein C (No. 370 DAKO) followed by an alkaline phosphatase-conjugated swine anti-rabbit antibody (No. 306 DAKO), and BCIP (5-bromo-4-chloro-3-indolyl phosphate)/nitro blue tetrazolium chloride (NBT) substrate was used for development.
Activation of Protein C and Catalytic Activity against Small Substrates-Purified WT protein C or protein C variants were incubated with human ␣-thrombin (10:1 mol/mol) in 20 mM Tris-HCl, 150 mM NaCl, pH 7.4, at 37°C for 2 h in the presence of 5 mM EDTA. A sulfopropyl-Sepharose column was used to remove thrombin from the reaction mixture. APC concentrations were estimated by measurement of absorbance at 280 nm. Amidolytic activities were determined using chromogenic substrate S-2366 (20 mM Tris-HCl, 150 mM NaCl, pH 7.4 at 37°C) in a microplate reader (ELX808IU from Bio-TEK instruments) and expressed in absorbance change at 405 nm (⌬A/min). The concentrations of chromogenic substrate S-2366 ranged between 0.015 and 3 mM, and the concentrations of APC ranged between 1 and 50 nM. K m and V max values were obtained from Lineweaver-Burk plots, and k cat was calculated from the Michaelis-Menten equation (27,28).
Biotinylation of Heparin and Heparan Sulfate-Heparin or heparan sulfate (0.5 mg of each) were dissolved in 200 l of 0.1 M MES, pH 5.6. EZ-link biotin hydrazide and N-ethyl-NЈ-(dimethylaminopropyl)carbodiimide were added to a final concentration of 2.5 and 0.5 mM, respectively. After a 6-h incubation at room temperature with constant shaking, the biotinylated glycosaminoglycans were desalted on a PD-10 column (Amersham Pharmacia Biotech) equilibrated in water, then freeze-dried, and dissolved in 0.3 M NaCl.
Binding of Activated Protein C to Immobilized Heparin and Heparan Sulfate-Binding of WT and variant APCs to heparin and heparan sulfate was analyzed by surface plasmon resonance using BIAcore 2000. Biotinylated, unfractionated heparin and heparan sulfate were immobilized in flow cells 2 and 3, respectively, of a streptavidin sensor chip. Functional integrity of the chip-bound heparan sulfate was shown by the fact that it bound bFGF with high affinity. APC (5-150 g/ml) was injected at a flow rate of 30 l/min at various NaCl concentrations (80 -300 mM) into flow cells containing heparin. In addition, APC (150 g/ml) was injected at 30 l/min into a flow cell containing immobilized heparan sulfate. Flow cell 1 without any immobilized glycosaminoglycan was used as control. To investigate whether fluid phase heparan sulfate could compete with immobilized heparin for the binding to APC, 0.5 mg/ml heparan sulfate was included in the flow buffer. The derived BIAcore sensograms were evaluated with BIAevaluation 3.0 software to calculate association (k on ) and dissociation (k off ) rate constants.
Affinity Chromatography on Heparin-Sepharose-WT and variant APCs in 20 mM Tris-HCl, 50 mM NaCl, pH 7.4 buffer were applied on heparin-Sepharose, and bound APC was eluted with a linear NaCl gradient from 0.05 to 0.4 M. APC and NaCl concentrations were measured by hydrolysis of chromogenic substrate S-2366 hydrolysis and flame photometry, respectively (13).
Electrostatic Potential-The side chains of variant APC molecules were replaced interactively using InsightII and Biopolymer, and all structures were energy-minimized using the simulation program Discover (Biosym-MSI). The electrostatic potential isosurfaces were computed with DelPhi as part of the Biosym-MSI-modeling suite (inner/ outer dielectrics 4/80; physiological ionic strength, formal charges) for APC molecules and for the highly negatively charged heparin molecule.

Kinetics of APC Inactivation by PCI in the Absence and Presence of Heparin or Heparan Sulfate-The rate of inhibition of WT and mutant
APCs by PCI in the absence of heparin or heparan sulfate was measured under pseudo-first-order conditions. In brief, WT and variant APCs were incubated (between 0 to 120 min) with a 10-fold molar excess of human plasma PCI at room temperature in 20 mM Tris-HCl, 0.10 M NaCl, 5 mM CaCl 2 , pH 7.5, containing 0.1% bovine serum albumin. Chromogenic substrate S-2366 was added to a final concentration of 0.2 mM, and the rate of substrate hydrolysis was measured with a V max kinetic microplate reader (Bio-TEK instruments). First-order rate constants (k) were calculated as Ϫln a/t [I], where a is the fractional protease activity remaining relative to the uninhibited control, t is the time of incubation, and [I] is the PCI concentration.
In the presence of unfractionated heparin (M r from 5,000 to 30,000; Leo) the rate of APC inactivation was too fast to be measured under the conditions that are described above. Therefore, PCI (7.5-200 nM at final concentration) was mixed with APC (30 nM) and 1 IU/ml (equal to 6.6 g/ml) unfractionated heparin in a final volume of 100 l. After defined time points, the reaction was stopped by the addition of 2 mg/ml Polybrene. The residual APC activities were determined by adding 50 l of S-2366 (final concentration was 0.2 mM), and the ⌬A at 405 nm was measured. The following equation was used to calculate the k 2 value.
[A] 0 and [B] 0 were the initial concentrations of APC and PCI, respectively, and was the molar concentration of APC⅐PCI complexes formed after time t; k 2 was calculated from data points demonstrating 15-85% proteinase inhibition. All experiments were performed in triplicates. To determine the effect of heparan sulfate on the rate of inhibition of APC by PCI, PCI (60 nM final concentration) was mixed with APC (30 nM final concentration) and increasing concentrations of heparan sulfate (0 -100 g/ml) in a final volume of 100 l. Under these conditions, maximum stimulation of inhibition was observed at concentrations Ն50 g/ml. At intervals, the reactions were terminated by the addition of 2 mg/ml Polybrene. Residual activity of APC was determined by the addition of 50 l of S-2366, and the ⌬A at 405 nm was measured. The k 2 values were calculated as described for heparin from data points demonstrating 15-85% proteinase inhibition. All experiments were performed in triplicate.

RESULTS AND DISCUSSION
Quantitative and Qualitative Analysis of APC Variants-To evaluate the involvement of positively charged amino acid residues in loops 37, 60, and 70 in the binding of heparin to APC, five protein C variants were constructed by site-directed mutagenesis (Fig. 1). According to our structural analysis, four of the variants were expected to have reduced affinity toward heparin (K37S/K38Q/K39Q, K62N/K63D, K37S/K38Q/K39Q/ K62N/K63D, R74Q), whereas one mutant was created with the intention of increasing the heparin affinity (E60aS/S61R). Since the target amino acid residues are solvent-exposed on loop structures and naturally occurring in related serine proteases, it should be expected that the above amino acid substitutions will not alter the folding of APC or damage its catalytic machinery. Furthermore, because the newly introduced residues are similar in size to the original ones, these mutations can be considered as conservative, and they should be better tolerated in the structure than, for instance, small hydrophobic alanine substitutions. Clustering of three to five alanine residues could create a destabilizing solvent-exposed hydrophobic patch no longer able to form hydrogen bonds with water molecules or could induce misfolding due to the high helical propensity of this amino acid.
The various protein C cDNA variants were used to transfect the eukaryotic cell line HU293, and stable cell lines were established. The different protein C variants demonstrated similar expression levels (2.7 to 6.7 mg/liter). The proteins were purified with overall recoveries of 35-40%, and the isolated proteins were more than 90% pure, as estimated by SDS-PAGE. On SDS-PAGE, the protein C variants migrated at similar positions as WT protein C (Fig. 2). Under nonreducing conditions, the different protein C variants migrated as single bands. After reduction of disulfide bridges, the 41-kDa heavy chains and the 21-kDa light chains were observed in addition to the 62-kDa single chain forms of protein C. Under reducing The residues that are mutated in the present study to probe the APC-heparin interaction are solvent-exposed and located on loop structures (blue spheres). Below the model, amino acid sequences of loops 37 and 60 of a few serine proteases are shown. The five lysine residues that form the heparin-binding site of APC are conserved in four different species (boxed) but absent in the other vitamin Kdependent enzymes.
conditions, for all recombinant proteins a small amount of single chain protein C remained. After activation, the different recombinant protein variants demonstrated similar amidolytic activities (K m and k cat ) ( Table I). The equally high expression levels observed for all the different protein C variants and their full amidolytic activities after activation together with the above structural considerations suggest that the recombinant proteins were correctly folded and could be used for further functional characterizations.
Binding of APC Variants to Immobilized Heparin-Binding of WT APC and APC variants to immobilized heparin was analyzed with surface plasmon resonance using a BIAcore 2000 (Table II). At physiological NaCl concentration, the dissociation constant (K d obs ) of WT APC for heparin was 0.32 M. The R74Q variant showed slightly lower affinity (K d obs ϭ 0.54 M), whereas E60aS/S61R bound more tightly (K d obs ϭ 0.19 M) to heparin than WT APC (Table II, Fig. 3). There was no detectable binding of K37S/K38Q/K39Q, K62N/K63D, and K37S/ K38Q/K39Q/K62N/K63D to immobilized heparin molecules. To qualitatively confirm the results obtained by BIAcore, WT APC and the five APC variants were applied on heparin-Sepharose, and bound protein was eluted with a linear NaCl gradient. The results agreed well with those obtained by the BIAcore analysis. Thus, removal of the positively charged residues at positions 37/38/39 or 62/63 or 37/38/39/62/63 yielded diminished or undetectable binding to the heparin-Sepharose (results not shown). In contrast, the E60aS/S61R variant bound tighter to the heparin column than WT APC and was eluted at higher NaCl concentrations. These results strongly suggest that the heparin-binding site of APC is entirely dependent on the presence of a few positively charged residues on loops 37 and 60 (Fig. 4).
Heparin is derived from mast cells, and it is unlikely that APC interacts with heparin under normal physiological conditions if heparin is not administered as a therapeutic. Heparan sulfate present on the surface of endothelial cells is the glycosaminoglycan that is more likely to interact with circulating APC. For this reason, we tested the binding of APC to immo-   bilized heparan sulfate in the BIAcore under similar conditions as for heparin. In contrast to the clear binding curves observed with heparin, no binding of APC to heparan sulfate could be detected. In this context, it should be considered that BIAcore is only able to detect interactions characterized by relatively high association rate constants. Therefore, many physiologically relevant interactions cannot be detected by this method. To further investigate whether heparan sulfate was able to interact with APC, 0.5 mg/ml fluid phase heparan sulfate was co-injected with APC in the flow cell containing immobilized heparin. Under these conditions, complete inhibition of binding of APC to heparin was observed, suggesting that APC may interact with heparan sulfate, albeit weaker than with heparin.
Energetic Characteristics of the WT APC-Heparin Interaction-BIAcore was used to determine the kinetic characteristics of the binding of WT APC to heparin at various salt concentrations, the purpose being to gain insights into the relative contributions of ionic versus nonionic components to the interaction (Table III). The results were evaluated using the counter-ion condensation theory, assuming that heparin behaves as a polyelectrolyte surrounded by a counter-ion condensation volume (29,30). Counter-ions (e.g. Na ϩ ) can interact with heparin by delocalized, long range electrostatic interactions and by direct binding to ionic sites. Displacement of counter-ions upon heparin binding to protein has been experimentally demonstrated (31). In cases where a plot of log K d obs against log [salt] is linear, the magnitude of the slope corresponds to the effective number of heparin counter-ions released upon its binding to a protein. This approach has been used successfully to study many interactions such as the ones between heparin and thrombin or bFGF (11,32). The bFGF predictions were remarkable, since it was reported that ϳ30% of the binding energy between bFGF and heparin resulted from purely ionic interactions. In this case, the plot of log K d obs against log [salt] was linear (slope ϭ 1.95), and it was estimated that only 2-3 net ionic interactions were involved between the two molecules. This result was fully supported at a later stage by x-ray analysis of the bFGF-heparin complex (33).
To evaluate the APC-heparin interaction, the following scheme was used.
APC Zϩ ϩ heparin Ͻ-------Ͼ K d obs APC-heparin ϩ ⌿ЈZ Na ϩ SCHEME I The observed K d (K d obs ) is related to the nonionic equilibrium dissociation constant (K d nonionic ) through the relationship log K d obs ϭ log K d nonionic ϩ Z [Na ϩ ]. The plot of log K d obs versus log [Na ϩ ] is linear, and the value of Z can be calculated from the slope. The formation of Z electrostatic interactions between WT APC and heparin is accompanied by the displacement of ⌿Z-condensed counter-ions from the heparin. ⌿ is the effective fraction of counter-ion bound per heparin negative charge, which has been estimated to be 0.8 (11). The slope (0.8 ϫ Z ϭ 4.86) provides the effective number of purely ionic interactions formed between WT APC and heparin. Thus, about 4 -6 ionic interactions seem to be involved in this process. At 1.0 M NaCl, ionic interactions were neutralized, and log K d obs ϭ log K d nonionic , which we estimate to be 5600 M. Because the K d for the APC-heparin interaction at physiological ionic strength was 0.32 M, 66% of the binding free energy (⌬G) was estimated to be ionic, with the remaining 34% nonionic. The ⌬G of WT APC binding to heparin at physiological salt concentration (150 mM) was estimated to be Ϫ8.8 kcal/mol, with the nonionic and ionic contributions to this interaction Ϫ3 kcal/mol and Ϫ5.7 kcal/mol, respectively. Thus, the WT APC binding to heparin is mainly electrostatic in nature just like the binding of heparin to thrombin (K d obs ϭ 6 -10 M, about 80% ionic) (11) or to mucus proteinase inhibitor (K d obs ϭ 0.05 M; Ͼ80% ionic) (34).
There is a considerable effect of salt on K d obs for the APCheparin interaction (Table III)  ues run from Ϫ9.6 kcal/mol at 80 mM NaCl to Ϫ6 kcal/mol at 300 mM NaCl (the K d obs increases 626-fold). Salt influences mostly k on , whereas the k off is not dramatically affected, suggesting that long range electrostatic interactions are important for the formation of an initial encounter complex. Once salt bridges are formed, NaCl does not significantly weaken the short range ionic interactions. We note also that although the APC-heparin interaction is mainly electrostatic, the rate of association is well below the maximum rate for collision of molecules in solution, which is around 10 9 -10 10 M Ϫ1 s Ϫ1 (35).
PCI-mediated APC Inactivation in the Absence and Presence of Heparin or Heparan Sulfate-During inhibition of APC by PCI, heparin or heparan sulfate may function as templates to guide PCI into the active site of APC during the formation of the Michaelis-like complex. Assuming the identified heparin/ heparan sulfate-binding site on APC to be functionally important, the recombinant APC modifications are expected to affect the heparin/heparan sulfate stimulation of inhibition by PCI. This was indeed found to be the case. Heparin (at 1 IU/ml, equal to 6.6 g/ml) accelerated the rate of PCI-mediated inhibition of WT APC 800-fold, whereas heparin only yielded an 11-fold increase in the rate of inhibition of the K37S/K38Q/ K39Q/K62N/K63D variant (Table IV). The E60aS/S61R variant, which bound heparin with higher affinity than WT APC, demonstrated even higher stimulation of inhibition by heparin (11,000-fold). Heparan sulfate (at 50 g/ml) was tested in the same experimental setup and was found to give a 350-fold increased rate of inhibition of WT APC and a 2200-fold increased inhibition of the E60aS/S61R variant but only a 40-fold increased rate of inhibition of K62N/K63D and a 6-fold increased rate of inhibition of the K37S/K38Q/K39Q/K62N/K63D variant (results not shown) . These results demonstrate that heparan sulfate is able to interact with the same site as heparin on APC and that heparin and heparan sulfate yield similar functional effects on the rate of inhibition by PCI. However, based on both BIAcore results and the PCI inhibition data, heparin appears to bind APC with higher affinity than does heparan sulfate.
In the absence of heparin/heparan sulfate, E60aS/S61R and R74Q variants were inhibited by PCI equally fast as WT APC, whereas the rate of PCI inhibition was increased for the K37S/ K38Q/K39Q, K62N/K63D, and K37S/K38Q/K39Q/K62N/K63D variants (Fig. 5A). The biggest effect was shown for the variant K37S/K38Q/K39Q/K62N/K63D. Already after 10 min more than 90% of the variant was inhibited, whereas WT APC activity was reduced only by 10% (Fig. 5A). This suggests that the lysine cluster in loops 37 and 60 repulses PCI in the absence of heparin. The poor heparin-mediated stimulation of PCI inhibition of the K37S/K38Q/K39Q, K62N/K63D and K37S/K38Q/ K39Q/K62N/K63D (Fig. 5B) is in agreement with the bridging theory involving heparin or heparan sulfate binding to the lysine cluster of APC as well as to PCI. In line with this theory, heparin and heparan sulfate were found to provide increased stimulation to the PCI inhibition of the E60aS/S61R variant, which demonstrated increased affinity for heparin (Fig. 5B).
It is noteworthy that the now demonstrated heparin/heparan sulfate-binding site in APC is located on another part of the protease domain as compared with the heparin binding exosite II of thrombin (Fig. 1). Similarly, PCI binds heparin on the other side (around helix H) as compared with the heparin binding D-helix of antithrombin (10,36,37). Therefore, the helix H area of PCI should be the topological neighbor of APC loops 37 and 60 in the Michaelis-like complex.
In conclusion, we have characterized the structural and energetic basis of a functionally important heparin-binding site in the serine protease domain of APC. This binding site is functionally important in the heparin/heparan sulfate-stimulated inhibition of APC by its serpin inhibitor PCI. In the absence of heparin, APC (20 nM) was incubated for 30 min at room temperature with a 10-fold molar excess of human plasma PCI in 20 mM Tris-HCl, 0.10 M NaCl, 5 mM CaCl 2 , pH 7.5, containing 0.1% bovine serum albumin. Chromogenic substrate S-2366 was added, and the rate of substrate hydrolysis was measured. In the presence of heparin, 7.5 nM PCI was mixed with APC (30 nM) and 1 IU/ml unfractionated heparin. E60aS/S61R (OE), K62N/K63D (E), and R74Q (f) were inhibited equally fast as WT APC (q) by PCI, whereas higher rates of inhibition were detected for K37S/K38Q/K39Q (‚) and K37S/K38Q/K39Q/K62N/K63D (Ⅺ). Each data point shows the mean value of three experiments. independently performed.