Exosite-dependent regulation of factor VIIIa by activated protein C

Protein C is a multidomain vitamin K–dependent serine protease zymogen in plasma that, upon activation by the complex of thrombin and thrombomodulin, down-regulates the coagulation cascade by degrading cofactors Va and VIIIa by limited proteolysis.^1-3  Activated protein C (APC) circulates in plasma as a light and heavy chain molecule held together by a single disulfide bond.⁴  Similar to other vitamin K–dependent coagulation proteases, the N-terminal light chain of APC contains the noncatalytic γ-carboxyglutamic domain and 2 epidermal growth factor–like domains.⁵  The catalytic domain of APC with a trypsin-like primary specificity pocket is located on the C-terminal heavy chain of the molecule.⁶  Structural data have indicated that the catalytic domain of APC contains several positively charged residues that are clustered at a region on the right side of the active site pocket.⁶  These residues are conserved at the same 3-dimensional location as those on the fibrinogen-binding exosite of thrombin.^6,7  The basic residues of this region in APC are clustered on 3 conserved surface loops, known in the chymotrypsinogen numbering system⁷  as the 39-loop (Lys³⁷-Lys³⁹), 60-loop (Lys⁶² and Lys⁶³), and 70-80–loop (Arg⁷⁴, Arg⁷⁵, and Lys⁷⁸). Earlier structure-function studies have demonstrated that basic residues of all 3 loops are involved in the interaction of APC with heparin⁸ ; however, only basic residues of the 39- and 70-80–loops are required for the anticoagulant function of the protease.^9,10  In addition to these loops, the autolysis loop (148-loop) of APC is also highly basic, with 5 Lys/Arg between residues 143 to 154. Previous mutagenesis studies have indicated that these residues are also critical for the anticoagulant function of APC.¹⁰ 

Factors Va and VIIIa are 2 essential procoagulant cofactors of the extrinsic and intrinsic pathways of the coagulation cascade, respectively.^11-13  Both cofactors are homologous glycoproteins, synthesized as single chain precursors with domain structures A1-A2-B-A3-C1-C2.^14-17  During proteolytic activation by either thrombin or factor Xa, the B domain from both cofactors is released and the resulting A1-A2 heavy chain remains noncovalently associated with the A3-C1-C2 light chain in a divalent cation-dependent manner.^17-20  While A1-A2 domains in factor Va remain contiguous, the cleavage at Arg³⁷² of factor VIII by thrombin transforms the heavy chain of the cofactor into 2 separate subunits.²¹  The A1 subunit of factor VIIIa retains a stable metal-ion–dependent linkage with the A3-C1-C2 subunit, whereas the A2 subunit remains weakly associated with the dimer through electrostatic interactions.²²  The inactivation of human factor Va by APC involves 2 cleavages at Arg³⁰⁶ and Arg⁵⁰⁶ sites, located on the A1 and A2 domains of the cofactor, respectively.^3,23  It has been demonstrated that APC cleaves these 2 sites in a sequential kinetic order, with an initial cleavage occurring rapidly at the Arg⁵⁰⁶ site followed by a slower cleavage occurring at the Arg³⁰⁶ site.³  While the cleavage of the Arg⁵⁰⁶ site is independent of a cofactor, the cleavage of the Arg³⁰⁶ site is accelerated approximately 20-fold by protein S on negatively charged membrane surfaces.²⁴  In the case of factor VIIIa, APC also cleaves the cofactor at the homologous sites Arg³³⁶ and Arg⁵⁶², located on the A1 and A2 subunits, respectively.^21,25  Unlike factor Va, however, in which a single cleavage at Arg⁵⁰⁶ site only partially inactivates the cofactor,^3,24  the APC cleavage of either site in factor VIIIa leads to a near-complete inactivation of the cofactor.²⁶  In the case of factor VIIIa, protein S has a modest cofactor effect toward the cleavage of the Arg³³⁶ site (∼ 2-fold), however, it enhances the cleavage rate of the Arg⁵⁶² site approximately 5-fold.²⁷ 

APC is a highly specific enzyme with cofactors Va and VIIIa being the only known substrates for the protease in plasma.²  The exact mechanism by which APC interacts with these cofactors is not known in detail. In the case of factor Va, results from several laboratories have indicated that basic residues of the 39-loop (Lys³⁷-Lys³⁹) and 70-80–loop (particularly Arg⁷⁴ and Arg⁷⁵) play critical roles in determination of the cleavage specificity of the Arg⁵⁰⁶ site, without influencing the cleavage of the Arg³⁰⁶ site.^8-10  In the case of factor VIIIa, however, the role of these basic residues in determination of the cleavage specificity of APC has not been studied. Thus, we addressed this question in this study by examining the time course of inactivation of human factor VIIIa by the exosite mutants of APC in both the absence and presence of protein S. Inactivation kinetics (IXa-mediated factor Xa generation in the presence of factor VIIIa) and sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) analysis of cleavage reactions both indicated that the same basic residues of APC, which are involved in a protein S–independent interaction with factor Va, are also involved in interaction of APC with factor VIIIa. However, unlike the critical role of these residues in the selective cleavage of the A2 domain of factor Va, these residues were found to be involved in determination of the specificity of cleavage at both Arg³³⁶ and Arg⁵⁶² scissile bonds of the cofactor.

Construction, expression, and purification of wild-type and mutant protein C derivatives in which the 3 basic residues of the 39-loop (Lys³⁷-Lys³⁹) were substituted with the identical sequence of thrombin (Pro-Gln-Glu) in a triple mutant (Lys-Lys-Lys/Pro-Gln-Glu) and basic residues of the 60-loop (Lys⁶², Lys⁶³, and Arg⁶⁷) and 70-80–loop (Arg⁷⁴, Arg⁷⁵, and Lys⁷⁸) were individually substituted with Ala (Lys62Ala, Lys63Ala, Arg76Ala, Arg74Ala, Arg75Ala, and Lys78Ala) have been described previously.^8,28  The wild-type and mutant protein C derivatives were all expressed in HEK293 cells using the pRC/RSV expression vector (Invitrogen, San Diego, CA) as described.⁸  All protein C derivatives were purified by immunoaffinity chromatography using the Ca²⁺-dependent monoclonal antibody, HPC4, linked to Affi-Gel 10 (Bio-Rad, Hercules, CA) as described.⁸ 

Human plasma proteins factor Va, protein S, and factor IXa were purchased from Haematologic Technologies (Essex Junction, VT). Recombinant factor VIII preparations were gifts from Bayer (Berkeley, CA). Factor VIII was activated with thrombin and resulting factor VIIIa was purified on CM-Sepharose column (Pharmacia, Piscataway, NJ) as described.²⁹  The A1/A3-C1-C2 dimer was isolated from thrombin-treated factor VIIIa using a Mono S column (Pharmacia) and residual A2 subunit was removed by an anti-A2 subunit monoclonal antibody coupled to Affi-Gel 10 as described.³⁰  The A1 subunit was isolated from thrombin-treated factor VIIIa heavy chain by fast-protein liquid chromatography using a combination of Hi-Trap Heparin (Pharmacia) and Mono Q columns (Pharmacia) as described.³¹  Recombinant proteins factor X,³²  thrombin,³³  and protein C inhibitor⁸  were prepared by cited methods. Phospholipid vesicles containing 80% phosphatidylcholine and 20% phosphatidylserine (PC/PS) were prepared as described.³⁴  The chromogenic substrates, Spectrozyme PCa (SpPCa) and Spectrozyme FXa (SpFXa) were purchased from American Diagnostica (Greenwich, CT).

Following purification on the HPC4 immunoaffinity column, 1 mg of each protein C derivative was incubated with thrombin (100 μg) in 0.1 M NaCl, 0.02 M Tris (tris(hydroxymethyl)aminomethane)–HCl (pH 7.4; Tris-buffered saline [TBS]) containing 5 mM EDTA (ethylenediaminetetraacetic acid) for 2 hours at 37°C. Activated protein C derivatives were separated from thrombin by fast-protein liquid chromatography using a Mono Q column developed with 40-mL linear gradient from 0.1 to 1.0 M NaCl, 0.02 M Tris-HCl (pH 7.4) as described.⁸  Partially and fully γ-carboxylated APC derivatives were eluted from the Mono Q column as 2 distinct peaks at approximately 0.3 M and approximately 0.4 M NaCl, respectively, as described.⁸  The APC concentrations were determined from the absorbance at 280 nm assuming a molecular weight of 56 200 and extinction coefficient (E_1cm^1%) of 14.5,³⁵  by an amidolytic activity assay using SpPCa, and by stoichiometric titration of enzymes with known concentrations of protein C inhibitor as described previously.⁸ 

Factor VIIIa inactivation by APC derivatives was evaluated using a 3-stage coupled assay as described.³⁶  Briefly, in the first stage, factor VIIIa (1 nM) was incubated with wild-type or mutant APC (2.5 nM) on PC/PS vesicles (50 μM) at room temperature in TBS containing 5 mM Ca²⁺, 1 mg/mL bovine serum albumin (BSA), and 0.1% PEG 8000. In the second stage, at different time intervals (0-20 minutes), the remaining factor VIIIa activity was determined in an intrinsic Xase assay from the factor VIIIa–dependent factor X activation by factor IXa as described.^27,36  Factor X activation with the intrinsic Xase complex was carried out for one minute with excess factor X (0.5 μM) and saturating factor IXa (20 nM) on PC/PS (25 μM) vesicles at room temperature. In the third stage, the remaining activity of factor VIIIa was determined from a decrease in the rate of factor Xa generation as monitored by an amidolytic activity assay using 200 μM SpFXa.

SDS-PAGE analysis of the cleavage reactions was carried out for both factor VIIIa (A1/A2/A3-C1-C2) and the A1/A3-C1-C2 dimer as described.²⁷  Factor VIIIa (250 nM) or the A1/A3-C1-C2 dimer (250 nM) was incubated with each APC derivative (40 nM) on PC/PS vesicles (100 μM) in the absence or presence of saturating protein S (250 nM) for 5 minutes at 37°C in TBS containing 5 mM Ca²⁺ and 0.1% PEG 8000. The cleavage reactions were stopped by addition of 5 × reducing sample buffer followed by boiling for 5 minutes. In a Bio-Rad minigel system, 10-μL samples of the cleavage reactions were analyzed by a 10% SDS-PAGE using the methods of Laemmli. Following electrophoresis for one hour at 200 V, proteins were transferred to nitrocellulose membranes using a semidry Bio-Rad minute-transblot apparatus at 15 V for 30 minutes in a buffer containing 48 mM Tris, 39 mM glycine, 20% MetOH, and 0.0375% SDS (pH 9.2). The membranes were blocked with TBS containing 2 mg/mL BSA for 2 hours. They were then incubated with either one of the 2 monoclonal antibodies, the 58.12 anti-A1 subunit (a gift from Bayer) or the R8B12 anti-A2 subunit (Green Mountain Antibodies, Burlington, VT), followed by goat antimouse horseradish peroxidase–conjugated secondary antibody as described.²⁷  The signals on blots were developed by enhanced chemiluminescence (ECL) system using luminol as the substrate. Blots were exposed to films and the band densities were scanned by Scan Analysis software (Scion Image Beta 4.02; Scion, Frederick, MD).

SDS-PAGE analysis of the A1 subunit cleavage reaction was carried out as described in the previous paragraph with the exception that the A1 subunit (2.5 μM) was incubated with APC derivatives (250 nM) on PC/PS vesicles (100 μM) for 2 to 4 hours at 37°C.

Expression, purification, and activation of wild-type and mutant protein C derivatives by thrombin have been described previously.^8,28  The amidolytic activity and the anticoagulant function of these APC mutants have been evaluated by both clotting and factor Va degradation assays in several previous studies.^8-10  Results have demonstrated that, with the exception of the Arg67Ala mutant, all other APC mutants exhibited normal amidolytic activities.¹⁷  In the case of Arg67Ala, in addition to the proteolytic activity, the amidolytic activity of the mutant has also been impaired.¹⁷  Thus, it is possible that mutagenesis of this residue has lead to a global alteration of the APC structure. Although we have included results with this mutant in the figures, we cannot draw any firm conclusion as to whether or not Arg⁶⁷ contributes to the specificity of APC interaction with factor VIIIa. Since all other mutants are activated normally by thrombin and have normal amidolytic activities, they have most likely folded properly, and, therefore, changes in the specificity of these mutants in factor VIIIa recognition can be attributed to mutant residues with a relatively high degree of certainty. The rationale for substituting Lys³⁷-Lys³⁹ of APC with corresponding residues of thrombin was based on the observation that these 2 proteases have the greatest structural similarities among the coagulation proteases and that such substitution may minimally affect the folding of the mutant proteins. However, the presence of a proline (Pro) in this sequence could lead to local structural constraints on the loop that may not be present in the wild-type protease. Thus, results with this mutant should be interpreted with this limitation in mind.

Previously, plasma-based activated partial thromboplastin time (aPTT) assays have established that basic residues of the 39-loop (Lys³⁷-Lys³⁹) and 70-80–loop (particularly Arg⁷⁴ and Arg⁷⁵) are important for the anticoagulant function of APC.^8-10  Factor Va degradation assays in purified systems have indicated that these residues are involved in recognition and cleavage of the Arg⁵⁰⁶ scissile bond, which is located on the A2 subunit of the cofactor.^9,10  As shown in Figure 1, factor VIIIa degradation assays revealed that the same basic residues of APC are also involved in recognition and efficient inactivation of factor VIIIa in both the absence and presence of protein S. In the absence of protein S, wild-type APC and the Lys62Ala, Lys63Ala, and Lys78Ala mutants inactivated 50% of the cofactor function of factor VIIIa in approximately 15 minutes (Figure 1A). Under the same experimental conditions, however, the time required to inactivate 50% of the cofactor activity of factor VIIIa was prolonged 2-fold (∼ 30 minutes) for both Arg74Ala and Arg75Ala mutants, and 3- to 4-fold for the Arg67Ala and Lys-Lys-Lys/Pro-Gln-Glu mutants (45-60 minutes). The catalytic efficiency of APC derivatives toward factor VIIIa in the presence of protein S was improved (Figure 1B). Protein S, however, did not restore the catalytic defects of the APC mutants since the rate of factor VIIIa inactivation by the mutants was impaired nearly to the same extent. APC inactivated 50% of the cofactor activity of factor VIIIa in approximately 1.5 to 2 minutes in the presence of protein S. Similar to results in the absence of protein S, minimum impairments in the catalytic efficiency of Lys62Ala, Lys63Ala, and Lys78Ala mutants were observed in the presence of protein S. However, the catalytic efficiencies of Arg74Ala and Arg75Ala mutants were moderately and those of Lys-Lys-Lys/Pro-Gln-Glu and Arg67Ala mutants were severely impaired (50% inactivation of the cofactor activity occurring in 4-5 minutes and 20 minutes, respectively; Figure 1B). These results indicate that protein S can promote the rate of factor VIIIa inactivation approximately 5- to 10-fold with all APC derivatives, which is consistent with the literature.²⁷  The results further suggest that, similar to inactivation of factor Va, the basic residues of the 39-loop (Lys³⁷-Lys³⁹) and 70-80–loop (Arg⁷⁴ and Arg⁷⁵) are recognition sites for interaction with factor VIIIa independent of protein S.

The APC cleavage of factor VIIIa at either Arg³³⁶ or Arg⁵⁶², located on the A1 and A2 subunits, respectively, can result in the inactivation of the cofactor.²⁷  To determine whether either cleavage was selectively affected by the APC mutagenesis, samples of the factor VIIIa cleavage reactions (5-minute time points) were run on 10% SDS polyacrylamide gels and transferred to nitrocellulose membranes for Western blotting. Detection used monoclonal antibodies specific for either the A1 (58.12) or the A2 (R8B12) subunit of factor VIIIa. Moreover, in addition to intact factor VIIIa trimer (A1/A2/A3-C1-C2), the APC cleavage reactions of the A1/A3-C1-C2 dimer and isolated A1 subunit were also analyzed by these procedures. Wild-type APC was capable of cleaving both the A1 (Figure 2A) and A2 (Figure 2B) subunits of factor VIIIa, though the cleavage of the A1 subunit occurred with a significantly higher rate. In agreement with the kinetic data, catalytic activities of Lys62Ala, Lys63Ala, and Lys78Ala mutants were minimally affected as they cleaved both the A1 and A2 subunits with catalytic efficiencies that appeared nearly comparable with those of the wild-type APC. However, the catalytic efficiencies of both Arg74Ala and Arg75Ala mutants were moderately impaired, while that of the Lys-Lys-Lys/Pro-Gln-Glu mutant was severely impaired. Scans of the band intensities of cleavages by these mutants are presented in Figure 3. Apparent rate values were estimated using the experimental conditions described in the legend of Figure 2. Wild-type APC cleaved the A1 subunit with an apparent rate (0.46 minutes^–1) that was 46-fold higher than that for cleavage by the Lys-Lys-Lys/Pro-Gln-Glu mutant (0.01 minutes^–1). It should be noted that these values may be underestimated since the rate of cleavage by the wild-type APC was determined from a cleavage reaction in which approximately 37% of the cofactor was cleaved in 5 minutes (Figure 3A). The apparent rate of A1 subunit cleavage by the Arg74Ala (0.11 minutes^–1) and Arg75Ala (0.25 minutes^–1) mutants was impaired approximately 4- and approximately 2-fold, respectively. The apparent rate of cleavage by the Lys62Ala (0.32 minutes^–1) and Lys63Ala (0.40 minutes^–1) mutants was minimally affected and the rate of cleavage by Lys78Ala mutant was slightly improved (0.50 minutes^–1). The catalytic properties of APC mutants toward the cleavage of the A2 subunit were nearly a mirror image of the cleavage reactions observed with the A1 subunit, with the exception that all APC forms cleaved the A2 subunit approximately 2-fold slower (Figures 2B,3B).

The Western blot analysis and the scan of the band intensities for cleavage reactions using the A1/A3-C1-C2 dimer yielded essentially identical results (Figure 4A, Western blot only). In this case, however, the cleavage of the A1 subunit by the wild-type APC was slightly improved, the cleavage by the Lys62Ala mutant was slightly impaired, and no cleavage reaction was observed for either the Arg67Ala or the Lys-Lys-Lys/Pro-Gln-Glu mutant. These observations suggest that the presence of the A2 subunit may have some conformational effect on the A1 subunit in the trimeric cofactor, and thus its absence in the A1/A3-C1-C2 dimer could lead to subtle alterations in the specificity of APC derivatives to cleave this substrate. This hypothesis is consistent with earlier observations.²⁷  The Western blot of the isolated A1 subunit is shown in Figure 4B. Although APC cleaved A1 in complex with A3-C1-C2 at a rate that was comparable with the cleavage rate of the intact trimeric cofactor (compare Figures 2A and 4A), it cleaved the A1 subunit in the absence of A3-C1-C2 with a dramatically impaired efficiency. As indicated in “Materials and methods” and in the Figure 4B legend, in order to observe a quantitatively comparable cleavage product, the concentration of the APC and the A1 subunit had to be increased 6- to 10-fold and the incubation time had to be increased substantially (4 hours at 37°C). These conditions were due in large part to kinetic parameter (Km) considerations, since the isolated A1 subunit does not associate with phospholipid surfaces. Similar to cleavage of the trimeric and dimeric substrates, the catalytic efficiency of the Lys-Lys-Lys/Pro-Gln-Glu and Arg67Ala mutants toward the Arg³³⁶ scissile bond was dramatically impaired (Figure 4B). The rate of cleavage of this bond by other mutants was impaired less than 2-fold. An interesting observation, however, was the finding that, in addition to the Arg³³⁶ cleavage site, APC slowly cleaved the A1 subunit at an alternative site. The size of the new cleavage product appeared to be approximately 30 to 35 kDa. Noting that the A1 subunit contains 372 residues with an apparent molecular weight of 54 kDa, it is likely that APC mutants cleave after one of several Arg residues located between residues 200 and 250. Cleavage at this site was not detected using the dimeric and trimeric substrates, suggesting the association of A1 with A3-C1-C2 may alter the presentation of this scissile bond to the enzyme active site. With the exception of Arg75Ala and Lys78Ala, the mutants recognized the secondary site of the A1 subunit with relatively high efficiency. This was evidenced by the observation that the band intensity of the alternative cleavage product was nearly comparable with the band intensity of the main product generated from cleavage at Arg³³⁶. This was particularly true with the Lys62Ala and Lys63Ala mutants (Figure 4B, lanes 4-5). Thus, the interaction of basic residues of this exosite with a complementary site of the A1 subunit not only participates in the efficient cleavage of the Arg³³⁶ scissile bond but also appears to confer a restricted specificity for APC hindering its interaction with other nonspecific sites on the cofactor.

Basic residues of the 3 surface loops including 39-loop (Lys³⁷-Lys³⁹), 60-loop (Lys⁶², Lys⁶³, and Arg⁶⁷), and 70-80–loop (Arg⁷⁴, Arg⁷⁵, and Lys⁷⁸) form a basic exosite on the catalytic domain of APC that is 3-dimensionally homologous to exosite 1 of thrombin.^6,7  The location and relative orientation of these residues in the crystal structure of the catalytic domain of APC are shown in Figure 5.⁶  Previous studies have indicated that, similar to thrombin, the interaction of basic residues of this exosite with natural substrates, cofactors, and inhibitors plays an important role in determination of specificity and the catalytic function of APC.^8-10  Both plasma-based clotting and factor Va degradation assays in the purified system have indicated that all of these residues with the exception of the 2 basic residues of the 60-loop (Lys⁶² and Lys⁶³) are required for the anticoagulant function of APC.^8-10  APC inactivates factor Va by a sequential cleavage first at the Arg⁵⁰⁶ and then at the Arg³⁰⁶ scissile bonds.³  Studies in the purified system have demonstrated that basic residues of APC are primarily critical for the cleavage specificity of the Arg⁵⁰⁶ scissile bond on the A2 domain of factor Va without any effect on the cleavage specificity of the Arg³⁰⁶ scissile bond that is located on a loop in the boundary of the A1 and A2 domains of the cofactor.³⁷  The results of the current study suggest that the same basic residues of APC are also involved in interaction and degradation of factor VIIIa. However, unlike the degradation of factor Va, these residues appear to be critical for the cleavage specificity of both the Arg³³⁶ and Arg⁵⁶² scissile bonds located on the A1 and A2 subunits of factor VIIIa, respectively.

The sites on the cofactor subunits that interact with the basic exosite of APC have not been identified. However, an interactive site for APC has been reported on the light chain of both factors Va and VIIIa.³⁸  This site has been mapped to the C-terminal end of the A3 domain in both cofactors.³⁸  This was evidenced by the observation that the peptide HAGMSTLFIV corresponding to residues 2009-2018 of factor VIII inhibited the anticoagulant activity of APC and its inactivation of factor VIIIa in the purified system.³⁸  A similar sequence derived from the C-terminal end of theA3 domain of factor V exhibited similar properties and inhibited the APC inactivation of factor Va. It is known that the C2 domain of the light chain of both factor Va and VIIIa binds on the phospholipid surface. Noting that the putative inhibitory peptide is located on the membrane proximal A3 domain of the cofactor,³⁹  it is likely that it interacts with the light chain of APC, but not the catalytic domain of the protease on the heavy chain. The fact that this peptide is not acidic and contains several hydrophobic residues is also consistent with this hypothesis. On the other hand, both of the A1 and A2 subunits of factor VIIIa contain acidic C-terminal sequences that might potentially provide contact sites for the basic exosite of APC.¹⁷  In support of this hypothesis, the catalytic function of the APC mutants in degradation of the trimeric factor VIIIa, dimeric A1/A3-C1-C2, and the isolated A1 subunit was similarly affected. This indicates that an interactive site on the cofactor is located within the A1 domain. It is interesting to note that previous studies have indicated that the same C-terminal sequence of the A1 subunit may also provide an interactive site for factor X.³⁰  It has been shown that the association of the substrate factor X with the intrinsic Xase complex confers a protective effect for factor VIIIa against inactivation by APC, specifically blocking the cleavage at the A1 site.²⁷  Thus, the A1 subunit of the cofactor may have overlapping binding sites for both factor X and APC on the C-terminal acidic domain. Furthermore, recent data also support the association of factor Xa with this region as a step leading to cleavages within the A1 subunit.⁴⁰  On the other hand, the A2 subunit of factor VIIIa shows an extended interactive surface for factor IXa.⁴¹  Factor IXa also demonstrates protection of factor VIIIa from APC, specifically blocking cleavage at the A2 site (Arg⁵⁶²).²⁷  Blocking of the cleavage at this site likely results from factor IXa interacting at the 558-565–loop,⁴²  thereby masking the susceptible bond. It is not known whether additional protective effects are derived from overlapping interactive sites on A2 for the 2 proteases.

The observation that APC mutants cleaved the A1 subunit at an alternative site with similar efficiency to that of the Arg³³⁶ site suggests that in addition to mediating the effective cleavages of the Arg³³⁶ and Arg⁵⁶² scissile bonds, the basic residues of APC impede the protease interaction with a nonspecific site on the A1 subunit of the cofactor. The observed cleavage at a secondary site on theA1 subunit by the Lys62Ala and Lys63Ala mutants was of particular interest since these mutants exhibited nearly normal activities in both clotting and factors Va and VIIIa degradation assays, apparently suggesting that these residues may be dispensable for the anticoagulant function of APC.^9,10  However, the results of this study demonstrate that all basic residues of this exosite are critical for determination of the specificity of factor VIIIa degradation, albeit with a lack of impairment in the catalytic function of some mutants in functional assays. This is a good example of how coagulation proteases in general and APC in particular select their macromolecular substrates and inhibitors with a high degree of specificity.

APC also reacts very slowly with plasma inhibitors, and thus has a long half-life of 27 minutes in circulation.⁴³  Protein C inhibitor, α₁-antitrypsin, and the plasminogen activator inhibitor-1 (PAI-1)–vitronectin complex are thought to be primarily responsible for regulation of the APC activity in circulation.^28,44  We recently demonstrated that the basic residues of the 39-loop (Lys³⁷-Lys³⁹) are also required for the protease interaction with PAI-1.²⁸  PAI-1 has 2 acidic glutamic acid (Glu) residues at the P4′ and P5′ sites of the reactive site loop. Previous results have indicated that the interaction of basic residues of the 39-loops of the plasminogen activators with the P4′ and/or P5′ residues of PAI-1 is responsible for the rapid neutralization of the profibrinolytic enzymes by the serpin.⁴⁵  The same appears to be true for the APC–PAI-1 reaction since it has been shown that the reactivity of the Lys-Lys-Lys/Pro-Gln-Glu mutant of APC with the serpin is dramatically impaired.²⁸  Interestingly, examination of the acidic C-terminal domain of the A1 subunit of factor VIIIa at the primed site of the Arg³³⁶ scissile bond indicated that both P5′ and P6′ residues are also Glu,¹⁵  one or both of which can interact with the basic 39-loop of APC. Unlike the Arg³³⁶ site on the A1 subunit, the sequence surrounding the Arg⁵⁶² scissile bond from the P3-P6′ site (DQRGNQIMS) on the A2 subunit contains no acidic residues at the primed site, but instead a nonoptimal aspartic acid (Asp) residue at the P3 site of the cleavage site. We previously demonstrated that the presence of an acidic residue at the P3 site of substrates is inhibitory for interaction with APC because of its possible collision with an acidic Glu residue at position 192 of the catalytic pocket.⁴⁶  The scan of band intensities of the factor VIIIa cleavage reactions showed that APC cleaves the Arg³³⁶ scissile bond approximately 2-fold faster than the Arg⁵⁶² scissile bonds. Thus, differences in the structure of residues surrounding the 2 cleavage sites on the A1 and A2 subunits of factor VIIIa may account for their differential reactivity with APC.

In summary, results of this study suggest that the basic residues of all 3 surface loops (39, 60, and 70-80) in APC are critical for determination of factor VIIIa specificity. The interactions of these residues with factor VIIIa are critical for the cleavage specificity of both Arg³³⁶ and Arg⁵⁶² scissile bonds, located on the A1 and A2 subunits, respectively. It is known that the binding of both procofactors V and VIII on exosite 1 of thrombin is also required for their rapid activation by thrombin.⁴⁷  Further studies are needed to determine whether the sites on procofactors that interact with exosite 1 of thrombin are the same sites on activated cofactors that interact with the homologous exosite of APC.

We thank Dr Lisa Regan at Bayer for the generous gifts of recombinant factor VIII. We also acknowledge Julie Dill and Jan Freas for excellent technical assistance in preparation of factor VIIIa substrates.