Factor Va is inactivated by activated protein C in the absence of cleavage sites at Arg-306, Arg-506, and Arg-679.

Activated protein C (APC) exerts its anticoagulant activity via proteolytic degradation of the heavy chains of activated factor VIII (FVIIIa) and activated factor V (FVa). So far, three APC cleavage sites have been identified in the heavy chain of FVa: Arg-306, Arg-506, and Arg-679. To obtain more insight in the structural and functional implications of each individual cleavage, recombinant factor V (rFV) mutants were constructed in which two or three of the APC cleavage sites were mutated. After expression in COS-1 cells, rFV mutants were purified, activated with thrombin, and inactivated by APC. During this study we observed that activated rFV-GQA (rFVa-GQA), in which the arginines at positions 306, 506, and 679 were replaced by glycine, glutamine, and alanine, respectively, was still inactivated by APC. Further analysis showed that the inactivation of rFVa-GQA by APC was phospholipid-dependent and sensitive to an inhibitory monoclonal antibody against protein C. Inactivation proceeded via a rapid phase (kx1=5.4 x 10(4) M(-1) s(-1)) and a slow phase (kx2=3.2 x 10(3) M(-1) s(-1)). Analysis of the inactivation curves showed that the rapid phase yielded a reaction intermediate that retained approximately 80% of the original FVa activity, whereas the slow cleavage resulted in formation of a completely inactive reaction product. Inactivation of rFVa-GQA was accelerated by protein S, most likely via stimulation of the slow phase. Immunoblot analysis using a monoclonal antibody recognizing an epitope between Arg-306 and Arg-506 indicated that during the rapid phase of inactivation a fragment of 80 kDa was generated that resulted from cleavage at a residue very close to Arg-506. The slow phase was associated with the formation of fragments resulting from cleavage at a residue 1.5-2 kDa carboxyl-terminal to Arg-306. Our observations may explain the unexpectedly mild APC resistance associated with mutations at Arg-306 (FV HongKong and FV Cambridge) in the heavy chain of FV.

Human blood coagulation factor V (FV) 1 is the inactive precursor of activated factor V (FVa), the non-enzymatic cofactor of the serine protease factor Xa (FXa) (1,2). FVa is formed via limited proteolysis of its single chain precursor FV (ϳ330 kDa) by thrombin. It consists of a heavy chain (ϳ105 kDa) and a light chain (71-74 kDa), which are held together via a Ca 2ϩ ion. FVa is an essential part of the prothrombinase complex, which composes FVa, FXa, calcium ions, and a negatively charged membrane surface. In this complex FVa acts as a non-enzymatic cofactor that accelerates FXa-catalyzed prothrombin activation 10 3 -10 5 -fold (3)(4)(5).
Proteolytic inactivation of FVa by activated protein C (APC) is an essential reaction in the anticoagulant protein C pathway (6), which is important in the regulation of thrombin formation and in the inhibition of acute inflammation triggered by coagulation (7). Congenital deficiencies in this pathway are associated with thrombotic disease (8,9). Inactivation of FVa activity by APC is associated with cleavage of three peptide bonds in the heavy chain of FVa, which have been localized at Arg-306, Arg-506, and Arg-679 (10).
Several mutations have been reported in the APC cleavage sites of human FV. The most important one is FV Leiden (FVL), in which arginine 506 has been replaced by glutamine (11)(12)(13). This mutation results in the laboratory phenotype of APC resistance and is associated with an increased risk of deep vein thrombosis (11,14,15). The high allele frequency of FVL (2-7% in Caucasian populations (16)) makes this mutation the most important heritable risk factor for venous thrombosis. Besides FVL, three other mutations have been found in the APC cleavage sites of FV. All three affect the Arg-306 position and are much less frequent than FV Leiden (17)(18)(19)(20). There are no indications that the mutations at Arg-306 are associated with an increased risk of venous thrombosis (18,21). Moreover, the effect of these mutations on the sensitivity to APC is not entirely clear. Studies using recombinant FV have shown that mutations at Arg-306 result in an unexpectedly mild APCresistant phenotype (22)(23)(24).
Several studies have been performed to obtain more insight in the complex mechanism of APC-catalyzed FVa inactivation. These investigations demonstrated that cleavage of the heavy chain of FVa by APC is stimulated by negatively charged phospholipids (25)(26)(27) and protein S (27)(28)(29), which is a non-enzymatic cofactor of APC. In the presence of negatively charged phospholipids, the inactivation of FVa by APC proceeds via a rapid and a slow phase, which are associated with cleavages at Arg-506 and Arg-306, respectively (10,30). Cleavage at Arg-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The original objective of the present study was to investigate the functional and structural implications of each individual APC cleavage in more detail. Recombinant factor V mutants were constructed in which a major part of the B-domain (rFV⌬B) was removed and two or three of the APC cleavage sites were mutated. After expression in COS-1 cells, rFV mutants were purified, activated with thrombin, and inactivated by APC. Interestingly, rFVa-GQA (rFVa in which the arginines at positions 306, 506, and 679 were replaced by glycine, glutamine, and alanine, respectively) was still inactivated by APC. In this paper we report on the mechanism of APC-catalyzed inactivation of rFVa-GQA.
Proteins-Human APC, protein S, prothrombin, and thrombin were purchased from Enzyme Research Laboratories (South Bend, IN). The APC concentration was determined from APC antigen and activity measurements. APC antigen was measured by ELISA using IgG fractions of rabbit anti-human protein C (DakoCytomation Denmark A/S, Glostrup, Denmark), whereas APC activity was measured with S2366 using the kinetic parameters reported by Sala et al. (32). Human factor Xa and monoclonal antibodies AHV-5146 and AHV-5112 were from Hematologic Technologies Inc., Essex, VT. Monoclonal antibody 3B1 directed against the heavy chain of human FV was a kind gift from Prof. B. N. Bouma. Goat anti-mouse IgG conjugated with horseradish peroxidase was from Bio-Rad. Neutralite TM avidin-horseradish peroxidase conjugate was from Southern Biotechnology Associated, Inc., Birmingham, AL. Human FVL was isolated from human plasma according to Nicolaes et al. (30).
FV Antigen Assay-FV light chain antigen was measured by ELISA using two different monoclonal antibodies directed against the light chain of FV (33). In this ELISA, mAb V-6 was used as coating antibody and biotinylated mAb V-9 as tagging antibody.
Mutagenesis-In this study expression vectors (pMT2) were used containing the cDNA of FV in which two or three APC cleavage sites were mutated (23,34). In these mutants the arginines at positions 306, 506, or 679 were replaced by glycine, glutamine, and alanine, respec-tively (see Table I). The vectors were used to construct FV expression vectors lacking a major part of the B-domain (deletion from amino acids 827-1499) as described before (35). B-domainless constructs, labeled pMT2FV⌬B, were checked by direct DNA sequencing using the CEQ2000 TM Dye Terminator Cycle Sequencing kit from Beckman Coulter Inc., Fullerton, CA.
Transient Expression and Purification of rFV⌬B Mutants-B-domainless recombinant FV (rFV⌬B) mutants were transiently expressed in COS-1 cells (175-cm 2 culture flasks) using FuGENE 6 Transfection Reagent from Roche Diagnostics. Twenty-four hours after transfection, cells were washed with phosphate-buffered saline and incubated with serum-free medium (Opti-MEM Glutamax, Invitrogen). Conditioned medium was harvested after 72 h, centrifuged for 20 min at 3000 rpm (4°C), and frozen at Ϫ20°C. FV expression was measured by functional FV assay and ELISA.
Recombinant FV⌬B mutants were isolated in a two-step procedure essentially as described previously (35). Briefly, conditioned medium was thawed, supplemented with 10 mM benzamidine, and loaded on an ion exchange column (SP-Sepharose fast flow), which was subsequently washed with 25 mM Hepes, 100 mM NH 4 Cl, 5 mM CaCl 2 , 10 mM benzamidine, pH 7.5, until the fall-through was protein-free. Recombinant FV⌬B was eluted from the column with 25 mM Hepes, 1.5 M NH 4 Cl, 5 mM CaCl 2 , 10 mM benzamidine, pH 7.5. Fractions containing FVa activity were supplemented with 2 mg/ml ovalbumin, dialyzed against a buffer containing 25 mM Hepes, 50 mM NaCl, 5 mM CaCl 2, 10 mM benzamidine, pH 7.3, pooled, and applied at a speed of 3 ml/h to an affinity column, consisting of 1 mg mAb-3B1/ml Sepharose. The column was washed with 25 mM Hepes, 50 mM NaCl, 5 mM CaCl 2 , pH 7.3, and eluted with 25 mM Hepes, 1.8 M NaCl, 5 mM CaCl 2 , pH 7.3. Eluted fractions were screened for FV (activity/antigen) and analyzed by SDS-PAGE and immunoblotting. FV containing fractions were supplemented with 2 mg/ml ovalbumin, dialyzed against 25 mM Hepes, 50 mM NaCl, pH 7.3, and stored at Ϫ80°C.
Factor Va Activity Assay-FVa activity was measured in a two-step procedure as described previously (35). This assay was calibrated using dilutions of pooled normal plasma corresponding with 0 -3 pM FV.
Curve Fitting of Time Courses of FVa Inactivation-Time courses of APC-catalyzed FVa inactivation were obtained by measuring loss of FVa activity as a function of time. Data reported in this paper indicate that inactivation of the activated rFV⌬B-GQA mutant by APC proceeds via two pathways (Pathways 1 and 2), This model is similar to the inactivation of plasma FVa via cleavages at Arg-506 and Arg-306 (30). k 1 , k 2 , and k 3 are pseudo first-order rate constants for cleavage at the as yet unidentified cleavage sites Rx1 and Rx2. Under first-order conditions, i.e. conditions at which the inactivation rate is directly proportional to the factor Va and APC concentration, the loss of FVa cofactor activity is described by Equation 1 (30).
In this equation Va t is the FVa activity at time t; Va 0 is the FVa activity at time 0; B is the activity of the FVa intermediate generated after cleavage at Rx1. k 1 , k 2 , and k 3 are the observed pseudo first-order rate constants, from which the second-order rate constants were calculated as follows: SDS-PAGE and Immunoblot Analysis-PAGE was performed on SDS-PAGE (Tris/HCl prefab gels, Gradipore, French Forest, Australia) under reducing conditions according to Laemmli (36). Protein transfer to polyvinylidene difluoride membranes (Millipore Corp., Bedford, MA) was carried out semi-dry on a blot system from Amersham Biosciences. To detect FV, murine monoclonal antibodies against the heavy and light chain (AHV-5146 and AHV-5112, respectively) were used. Goat antimouse IgG conjugated with horseradish peroxidase in combination with the peroxidase blotting substrate was used to visualize the immobilized antibodies. Both antibodies were checked for any cross-reaction with BSA present in the FV fractions and were found negative. Visualization with AHV5146 was optimized to detect low molecular degradation fragments from the heavy chain (see also Fig. 1). Due to the high sensitivity of this procedure, small amounts of intact heavy chains still show significant staining (see also Ref. 41).

Expression and Purification of Recombinant Factor V Mutants-Recombinant
FV⌬B mutants were constructed with mutations either in two or in three of the known APC cleavage sites (Arg-306, Arg-506, and Arg-679). In analogy to the naturally occurring FV HongKong (R306G) and FV Leiden (R506Q) mutations, Arg-306 was replaced by glycine and Arg-506 by glutamine, whereas Arg-679 was replaced by alanine (Table I).
After transfection of COS-1 cells, conditioned media containing the rFV⌬B molecules were collected, centrifuged, and stored at Ϫ20°C. Factor V expression levels (activity and antigen) of rFV⌬B-wt and its mutants were similar and ranged (depending on the transfection efficiency) between 0.2 and 4 nM. After thawing and pooling of conditioned media, rFV⌬B proteins were purified by ion exchange and immunoaffinity chromatography. The rFV⌬B mutants were activated with thrombin and analyzed by SDS-PAGE followed by silver staining or immunoblotting. Both detection methods showed two dominant bands with a molecular mass of 105 and 71 kDa, which corresponded to the expected molecular mass of the heavy and light chain of factor Va, respectively (data not shown). Apart from these two bands, also variable amounts of 90-kDa fragment were observed, which was most likely the result of thrombin cleavage at Arg-643 (37). The specific activities (activity/antigen) of the purified rFV mutants were slightly lower than in conditioned medium, being around 0.7.
Stability of Activated rFV⌬B Mutants-Recombinant FV⌬B proteins were activated with 9 nM thrombin for 20 min at 37°C and incubated in the absence or presence of phospholipids. The stability of the activated rFV⌬B mutants was assessed by following the loss of FVa activity as a function of time in a reaction mixture that did not contain APC. The first-order rate constants for spontaneous inactivation of activated rFV⌬B-wt and the activated rFV⌬B mutants were approximately similar, being around ϳ3.0 ϫ 10 Ϫ5 s Ϫ1 , which corresponds to a spontaneous loss of FVa activity of ϳ0.2% per min. The presence of phospholipids did not influence the stability of the activated rFV molecules.
APC-catalyzed Inactivation of Activated rFV⌬B-GQA-Preliminary experiments showed that apart from the activated rFV⌬B double mutants (rFV⌬B-RQA, rFV⌬B-GRA, rFV⌬B-GQR) also the activated triple mutant rFVa-GQA was inactivated by APC and that this inactivation was strongly dependent on the presence of negatively charged phospholipids (Fig.  2). In the absence of phospholipids, inactivation of rFVa-GQA by 20 nM APC was only slightly faster than the spontaneous loss of activity.
Preincubation of APC with a monoclonal antibody against protein C (C12), which blocks the active site of APC, or with 4-amidinophenylmethanesulfonyl fluoride (a nonspecific inhibitor of serine proteases) inhibited the activation of rFVa-GQA by APC, whereas preincubation with a monoclonal antibody against protein S (S18) did not affect the rate of inactivation (Table III). This indicates that the loss of FVa activity observed in the presence of APC and phospholipids (Fig. 2) was indeed mediated by APC and not a contaminating protease.
Inactivation of Activated rFV⌬B-GQA with Increasing APC Concentrations-The results presented above indicate that the inactivation of rFVa-GQA by APC results from one or more proteolytic cleavages at still unidentified cleavage sites. To learn more about the mechanism of inactivation, rFVa-GQA was incubated with increasing concentrations of APC under pseudo first-order conditions. Increasing the APC concentration resulted in an acceleration of FVa inactivation (Fig. 3a). The FVa inactivation curves could not be fitted with a single exponential (data not shown), indicating that inactivation involves at least two different reactions, a rapid reaction resulting in a partially active FVa molecule that is subsequently completely inactivated via a slow reaction. Because the rate constant for the second reaction was much lower than that for the first reaction, it was possible to calculate the pseudo firstorder rate constant (k 2 ) for the second reaction and the activity of the FVa intermediate (FVa B ) from semi-logarithmic plots (Fig. 3b). The slope of the lines yields k 2 at different APC concentrations, and the intercept at the y axis yields an estimate for the FVa activity remaining in the intermediate (FVa B ), which was found to be ϳ80%. Fig. 3c shows that the pseudo first-order rate constant k 2 obtained for the second slow reaction is a linear function of the APC concentration, which enabled calculation of the second-order rate constant (k x2 ϭ 3.2 ϫ 10 3 M Ϫ1 s Ϫ1 ) for the second cleavage reaction. The line extrapolated to a rate constant of 5.6 ϫ 10 Ϫ5 s Ϫ1 at [APC] ϭ 0 (intercept of the line with the y-axis), which most likely reflects the spontaneous loss of activity of the reaction intermediate FVa B (represented as kЈ s ). This rate constant is close to that determined for native rFVa (k s ϳ3.0 ϫ 10 Ϫ5 s Ϫ1 ).
The time courses of inactivation of rFVa-GQA by APC (Fig.  3a) were further analyzed in a random ordered two-step model (Equation 1 under "Experimental Procedures"), in which rFVa-GQA is rapidly cleaved by APC resulting in an intermediate (FVa B ) with slightly lower FVa activity, whereas a much slower cleavage in FVa and FVa B results in the formation of a FVa derivate that is completely inactive. This model is similar to that proposed by Nicolaes et al. (30) for the inactivation of plasma FVa via cleavages at Arg-506 (fast reaction) and Arg-306 (slow reaction). By using the pseudo first-order rate constants (k 2 ) and B calculated from Fig. 3, b and c, the pseudo first-order rate constants k 1 and k 3 were calculated by fitting the experimental data (Fig. 3a) to Equation 1. Like k 2 , k 1 and k 3 also increased with higher APC concentrations (Fig. 3, d and  e), which indicates that the rapid and slow inactivation phases both are associated with APC-mediated cleavages at two as yet unidentified cleavage sites, designated Rx1 (k x1 ) and Rx2 (k x2 and kЈ x2 ). From these plots, the second-order rate constants were calculated (see "Experimental Procedures"). The secondorder rate constant of the first reaction (k x1 ϭ 5.4 ϫ 10 4 M Ϫ1 s Ϫ1 ) appeared to be ϳ20-fold higher than that of the second reaction (compare with k x2 and kЈ x2 in Table IV). These rate constants are much lower than those reported for APC-catalyzed cleavage at Arg-306 (k 306 ϭ 2.3 ϫ 10 6 M Ϫ1 s Ϫ1 ) and Arg-506 (k 506 ϭ 4.3 ϫ 10 7 M Ϫ1 s Ϫ1 ) (30).
Effect of Protein S on the APC-catalyzed Inactivation of Activated rFV⌬B-GQA-Previous studies have shown that in the presence of saturating concentrations of protein S (490 nM), the inactivation of FVa by APC is accelerated by selective stimulation of the cleavage at Arg-306 (31). The inactivation of rFVa-GQA by APC was also strongly stimulated in the presence of 490 nM protein S (Fig. 4). The time course of inactivation of rFVa-GQA by APC in the presence of protein S suggests that the acceleration of FVa inactivation was mainly due to an increased rate of the slow cleavage reaction (cleavage at Rx2). Acceleration of the rapid reaction (cleavage at Rx1) will hardly contribute to the final inactivation curve, because the intermediate formed after cleavage at Rx1 (FVa B ) still has a high FVa activity. Assuming that k x1 (5.4 ϫ 10 4 M Ϫ1 s Ϫ1 ) and B (0.82) were not affected by the addition of protein S, k 2 and k 3 values were determined by fitting the data to Equation 1 (Table IV). The second-order rate constant kЈ x2 (calculated from k 3 ) was ϳ1.5 ϫ 10 5 M Ϫ1 s Ϫ1 , which indicates that protein S stimulates the APC-catalyzed cleavage at Rx2 about 50-fold.
SDS-PAGE Analysis of APC-catalyzed Inactivation of Activated rFV⌬B-GQA-APC-mediated proteolysis of the heavy and light chain of rFVa-GQA was monitored by SDS-PAGE and immunoblotting. In this procedure, heavy chain fragments were detected by monoclonal antibody AHV-5146, which recognizes an epitope between Arg-306 and Arg-506. Experiments with synthetic peptides identified the epitope for AHV-5146 between residues 427 and 440. 2 FVa light chains were detected by mAb AHV-5112. During the inactivation of rFVa-GQA with 20 nM APC, no degradation of the light chain was observed (data not shown). Analysis of the heavy chain, however, revealed the generation of several fragments, which could be associated with the loss of FVa activity (Fig. 5). The initial phase of inactivation (0 -30 min) was associated with generation of a fragment of ϳ80 kDa, whereas the second slow phase was associated with the generation of a doublet of 55/60 kDa "Experimental Procedures." Data are presented as means Ϯ S.D. The solid lines in a represent exponential curves obtained by fitting the data to Equation 1 (using k 2 and B obtained from b) with nonlinear least squares regression of the data. The pseudo first-order rate constants of the second slow phase (k 2 ) and the activity of the reaction intermediate B were estimated from the plots in b. The slope of the lines from 90 to 180 min represents k 2 , whereas the intercept of these lines with the y axis represents the activity of the reaction intermediate B. c, the pseudo first-order rate constants of k 2 (s Ϫ1 ) have been plotted versus the APC concentration. The slope of this line corresponds to the second-order rate constant k x2 . The pseudo first-order rate constants for k 1 (s Ϫ1 ) and k 3 (s Ϫ1 ) were obtained by fitting the experimental data to Equation 1, in which k 2 (s Ϫ1 ) and B were fixed. d, the pseudo first-order rate constants of k 1 (s Ϫ1 ) have been plotted versus the APC concentration. From the slope of this line, the second-order rate constant k x1 was determined. e, the pseudo first-order rate constants of k 3 (s Ϫ1 ) have been plotted versus the APC concentration. The slope of this line represents the second-order rate constant kЈ x2 . and fragments of 45 and 30 kDa (Fig. 5). Note that besides the heavy chain fragment of 105 kDa (intact heavy chain from position 1 to 709), a fragment of ϳ90 kDa was also detected, which was already present before the addition of APC. Most likely, this fragment was formed during the activation with thrombin, which can cleave the heavy chain of FVa at Arg-643 (37).
In the presence of protein S, the loss of FVa activity was essentially associated with the generation of a 60-kDa fragment (Fig. 5c). After 15 min 45-and faint 30-kDa fragments were also generated.
Localization of Potential APC Cleavage Sites in rFV⌬B-GQA-The experimental data and the amino acid sequence of FV were used to localize potential APC cleavage sites in the heavy chain of rFVa-GQA. The rapid cleavage at Rx1 in the absence of protein S was associated with formation of a 80-kDa fragment (Fig. 5, a and b), which still contained ϳ80% FVa activity. Such a fragment may result from a cleavage in a region 150 -200 amino acids from either the amino or the carboxyl terminus of the heavy chain. Because the aminoterminal region (residue 150 -200) contains no arginines and only a few lysines, whereas the region between residues 499 and 513 contains a surface-exposed loop with several arginines and lysines (38), it is likely that Rx1 is positioned in the region surrounding Gln-506. Cleavage in this region will result in the formation of an 80-kDa fragment from both the 105-and 90-kDa heavy chains and is also compatible with the high activity of the intermediate (FVa B ϳ80%) after cleavage at Rx1. Consequently, the slow cleavage (Rx2), which results in complete inactivation, should be positioned in the region of Gly-306. This would explain the generation of the observed 60-kDa fragment (cleavage in the intact heavy chain of 105 kDa) and the 30-kDa fragment (cleavage in the intermediate fragment of 80 kDa). A cleavage close to position 306 can also explain the generation of a 45-kDa fragment (cleavage in heavy chain fragment of 90 kDa), which has been described before (37). Furthermore, a location of Rx2 close to Arg-306 is compatible with the stimulation of kЈ x2 by protein S (formation of 60-kDa fragment in Fig.  5c), because cleavage at Arg-306 is also stimulated by protein S (31).
More information about the position of Rx1 and Rx2 was obtained from experiments in which the activated double mutants rFV⌬B-GRA (Arg-506 present) and rFV⌬B-RQA (Arg-306 present) were incubated with APC. Inactivation of activated rFV⌬B-GRA with APC (20 nM) generated an expected fragment of 75 kDa (amino acids 1-506), which was gradually cleaved into a 30-kDa fragment (data not shown), suggesting that in the region of Gly-306 indeed another APC cleavage site (Rx2) is present. The generation of this 30-kDa fragment was strongly accelerated in the presence of protein S (Fig. 6a). Inactivation of activated rFV⌬B-RQA with 20 nM APC generated fragments with the expected sizes of 60 (fragment 307-709) and 45 kDa (fragment 307-643), which were both accelerated in the presence of protein S (Fig. 6b). These fragments were no longer susceptible to APC cleavage in the Gln-506 region, which may be due to dissociation of the 307-643 and the 307-709 fragments after cleavage at Arg-306 (39).
Finally, the position of Rx2 could be estimated more precisely by comparing the mobility of the 30-kDa fragments, which are the final products formed after inactivation of activated rFV⌬B-wt, rFV⌬B-GRA, and rFV⌬B-GQA by APC (Fig. 7). The immunoblot analysis. Fragments from the heavy chain were visualized with mAb AHV-5146 (epitope between Arg-306 and Arg-506). The same procedure was also performed in the presence of 490 nM protein S, after which the samples were subjected to 8% SDS-PAGE and immunoblot analysis (c). final fragment generated during the inactivation of rFV⌬B-wt has an expected molecular mass of 30 kDa, corresponding to fragment 307-506. The fragment generated from the inactivation of activated rFV⌬B-GRA, most likely corresponding to position Rx2-506, was about 28 kDa suggesting that the position of Rx2 is about 10 -20 amino acids carboxyl-terminal to residue 306. Consequently, the position of Rx1 should be very close to residue 506, because the final product in the inactivation of rFV⌬B-GQA (Rx2-Rx1) also has a molecular mass of 28 kDa.
Cleavage of Rx1 in Activated FVL-Theoretically, cleavage at Rx1 (located in the region of residue 506) may serve as an alternative for a cleavage at Arg-506. To check this, purified plasma FVL was activated and inactivated using the same conditions as for the rFV⌬B mutants and subsequently subjected to analysis by SDS-PAGE and immunoblotting (Fig. 8). The immunoblot showed a very similar cleavage pattern as observed for the activated rFV⌬B-RQA mutant starting with a heavy chain doublet of 105 and 90 kDa. Initially a 50/60-kDa doublet was generated, which was most likely the result of cleavage at Arg-306 in the 1-679 and 1-709 fragments. Also a fragment of 45 kDa was generated, which was probably produced by cleavage of the 90-kDa fragment resulting in a fragment from residues 307-643. The fragments were not further cleaved at Rx1, analogous to what was observed during inactivation of activated rFV⌬B-RQA. This might be the result of dissociation of the domain carboxyl-terminal to Arg-306 (A2-domain).

DISCUSSION
To obtain more insight in the functional and structural implications of the individual cleavages at Arg-306, Arg-506, or Arg-679 in the heavy chain of FVa by APC, recombinant FV mutants were generated in which two or three of these APC cleavage sites had been mutated. Because the B-domain of FV has no crucial role in the APC-catalyzed inactivation of FVa, we used mutants lacking a major part of the B-domain in order to increase the yield of rFV mutants. The expression of these B-domainless mutants was about 10 times higher than that of the full-length constructs, which is in line with previous studies (35,40). Initial inactivation experiments with the purified re- was incubated by using the same conditions with 6 nM APC. At the indicated time points samples were drawn from the inactivation mixture and subjected to 4 -15% SDS-PAGE under reducing conditions followed by immunoblot analysis. Fragments from the heavy chain were visualized with mAb AHV-5146 (epitope between Arg-306 and Arg-506). In the blot of rFV⌬B-GRA (a) a small band with a molecular mass of around 50 kDa was observed, which appeared to be present in all time samples. Most likely, purified samples of this FV mutant were enriched with some IgG during the purification procedure, which could also be stained. FIG. 7. Comparison of 30-kDa fragments generated during the APC-catalyzed inactivation of activated rFV⌬B-GQA, rFV⌬B-wt, and rFV⌬B-GRA. The sample containing rFVa-GQA was the same as used in Fig. 5b (t ϭ 90 min). The sample containing activated rFV⌬B-wt was drawn after 30 min from an inactivation mixture containing 1 nM rFV⌬B-wt, 0.15 nM APC, 490 nM protein S, and 24.5 M phospholipid vesicles (DOPS/DOPC, 10:90) in 25 mM Hepes, pH 7.5, 175 mM NaCl, 3 mM CaCl 2 , and 5 mg/ml BSA at 37°C. The sample containing rFV⌬B-GRA was the same as used in Fig. 6a (t ϭ 30 min). Samples were subjected to 12-20% SDS-PAGE under reducing conditions followed by immunoblot analysis. Fragments from the heavy chain were visualized with mAb AHV-5146 (epitope between Arg-306 and Arg-506). The 30-kDa fragments represent the following cleavage products: rFV⌬B-GQA, Rx2-Rx1; rFV⌬B-wt, 307-506; and rFV⌬B-GRA, Rx2-506. combinant proteins showed that rFVa-GQA (in which the known APC cleavage sites Arg-306, Arg-506, and Arg-679 have been mutated) was still inactivated by APC, although at a much lower rate than normal FVa. The main objective of this study was to learn more about the mechanism underlying this unexpected observation.
All activated B-domainless rFV mutants used in this investigation were stable under the conditions chosen to study FVa inactivation by APC (Table II). Also the presence of a heavy chain fragment of 90 kDa (most likely representing fragment 1-643 (37) generated during the activation with thrombin) did not affect the stability of the activated rFV molecules. The inactivation of rFVa-GQA by APC is dependent on negatively charged phospholipids (Fig. 2) and is inhibited by a monoclonal antibody against protein C (Table III), indicating that the observed inactivation is mediated by APC.
Kinetic analysis of the inactivation of rFVa-GQA by APC revealed that under pseudo first-order conditions (FVa concentration ranging from 1 to 5 nM (30)) rFVa-GQA was inactivated in a biphasic reaction, indicating that analogous to the inactivation of normal FVa also the inactivation of rFVa-GQA was associated with at least two cleavages (designated Rx1 and Rx2). This conclusion is supported by SDS-PAGE analysis of the reaction products (Fig. 5). During inactivation of rFVa-GQA, an FVa intermediate (FVa B ) with ϳ80% of the original FVa activity was formed via rapid cleavage at Rx1. This intermediate was completely inactivated via a second slow cleavage at Rx2. The calculated second-order rate constant of the rapid reaction (k x1 ϭ 5.4 ϫ 10 4 M Ϫ1 s Ϫ1 ) was ϳ20-fold higher than that of the slow reaction (k x2 ϭ 3.2 ϫ 10 3 M Ϫ1 s Ϫ1 and kЈ x2 ϭ 2.6 ϫ 10 3 M Ϫ1 s Ϫ1 ). The rate constant of the rapid reaction is 50and 1000-fold lower than those reported for cleavage at Arg-306 and Arg-506, respectively (30), but is somewhat higher than the second-order rate constant for cleavage at Arg-679, which has been estimated at 7.6 ϫ 10 3 M Ϫ1 s Ϫ1 (30). Therefore, the knowledge of the existence of additional cleavage sites for APC, one close to residue 306 and one close to residue 506, is important for the determination of the kinetic parameters for cleavage at the individual sites Arg-306, Arg-506, or Arg-679 and need to be taken into account (especially in the case of analysis of cleavage at Arg-679).
Protein S strongly stimulated the inactivation of rFVa-GQA by APC (Fig. 4). Kinetic analysis revealed that the slow cleavage at Rx2 (kЈ x2 ) that fully inactivates rFVa-GQA was stimulated about 50-fold by protein S (Table IV). In this analysis we assumed that cleavage at Rx1 was not affected by protein S. This is supported by the observation that in the presence of protein S (Fig. 5c) mainly a 60-kDa fragment (cleavage at Rx2) was generated by APC instead of the 80-kDa fragment observed in the absence of protein S. Moreover, under the experimental conditions used, acceleration of the cleavage at Rx1 by protein S would hardly contribute to the final inactivation curve due to the high FVa activity associated with the intermediate formed after cleavage at Rx1.
To localize the position of the cleavage sites Rx1 and Rx2, the fragments generated during the APC-catalyzed degradation of rFVa-GQA were analyzed by SDS-PAGE and immunoblotting using an antibody with an epitope between residues 306 and 506 on the heavy chain of FVa. At time 0 of the inactivation curves, some 90-kDa fragment was also present in the activated rFV⌬B mutants (Figs. 5 and 6) and activated FVL (Fig.  8) apart from the expected 105-kDa heavy chain fragment. SDS-PAGE analysis of the time courses of thrombin activation revealed that this fragment was the result of cleavage by thrombin (data not shown), most likely at Arg-643 (37). Due to the high sensitivity of the detection method, it was not possible to quantify the relative amounts of 105-and 90-kDa fragments. However, under the conditions used, no effect of the cleavage at Arg-643 on the FVa activity was observed (data not shown).
Positioning of Rx1 and Rx2 close to residue 506 and 306, respectively, would best fit the experimental data obtained via immunoblot and kinetic analysis. A position of Rx1 near residue 506 would explain the generation of a fragment of ϳ80 kDa during the initial phase of the inactivation of rFVa-GQA by APC (Fig. 5a). It also would explain the relatively high FVa activity of the intermediate formed after cleavage at Rx1, because cleavage of factor Va at Arg-506 also results in the formation of an intermediate with relatively high factor Va activity (30,41). Location of Rx2 near residue 306 would explain the formation of peptides of 60 and 45 kDa (from the heavy chains fragments of 105 and 90 kDa, respectively) and 30 kDa (from the 80-kDa intermediate) during the second slow inactivating phase. A position of Rx2 close to residue 306 is also compatible with the observation that the cleavage at Rx2 is stimulated by protein S (Table IV), because protein S exerts its stimulatory effect on the inactivation of normal FVa by selec-  FVa activity of activated rFV⌬B-GQA Activated rFV⌬B-GQA was incubated for 180 min at 37°C in 25 mM Hepes, pH 7.5, 175 mM NaCl, 3 mM CaCl 2 , 5 mg/ml BSA with 20 nM APC, which had been preincubated with anti-protein C antibody C12 ‫06ف(‬ g/ml), anti-protein S antibody S18 ‫06ف(‬ g/ml), or 4-amidinophenylmethanesulfonyl fluoride (80 M). The remaining FVa activity is shown.

TABLE IV
Second-order rate constants for inactivation of activated rFV⌬B-GQA by APC via peptide bond cleavages at unidentified cleavage sites (Rx1 and Rx2) The second-order rate constants (M Ϫ1 s Ϫ1 ) in the absence of protein S were calculated from the slopes of the plots of the pseudo first-order rate constants versus the APC concentrations (Fig. 3, c-e). The second-order rate constants in the presence of protein S were measured at a single APC concentration of 20 nM. The pseudo first-order rate constants k 2 and k 3 were determined by fitting the data to Equation 1 (Fig. 4), assuming that k x1 (5.4 ϫ 10 4 M Ϫ1 s Ϫ1 ) and B (0.82) were not affected by the presence of protein S (see text). The second-order rate constants k x2 and kЈ x2 were calculated from the pseudo first-order rate constants as described under "Experimental Procedures."

Rate constants
ϪProtein tively stimulating the cleavage at Arg-306 (31). A more detailed localization of the position of Rx2 was obtained by comparing the 30-kDa fragment (residues 307-506) generated in the inactivation of rFV⌬B-wt with the fragment (Rx2-506) generated during the inactivation of activated rFV⌬B-GRA by APC, when cleavage at Arg-506 is followed by cleavage at Rx2 (Fig. 7). The fragment of Rx2-506 was about 1.5 kDa smaller than the fragment of 307-506, indicating that the position of Rx2 is localized 10 -20 residues carboxyl-terminal to residue 306. This confines the candidates for the position of Rx2 to the residues Arg-316, Arg-317, Lys-320, and Arg-321. Consequently, the position of Rx1 should be very close to residue 506, because the Rx2-Rx1 fragment produced during the inactivation of rFVa-GQA has the same molecular weight as the Rx2-506 fragment (Fig. 7). This would make residues Arg-505 and Arg-510 strong candidates for the location of Rx1. However, the localization of Rx1 at Arg-505 seems less likely, because this residue is positioned in a putative Xa-binding site (amino acids 493-506) (42).
Hence, cleavage at Arg-505 would likely result in an FVa intermediate with a lower FVa activity than the 80% activity calculated for the intermediate after cleavage at Rx1. Currently, attempts are made to produce sufficient amounts of rFV⌬B-GQA to identify the precise location of Rx1 and Rx2 by amino-terminal sequencing. Interestingly, the degradation patterns of the activated rFV⌬B-RQA and rFV⌬B-GRA mutants (Fig. 6) confirmed some of the structural implications of cleavage at Arg-306 or Arg-506. The final fragments in the APC-mediated degradation of activated rFV⌬B-RQA were 60 and 45 kDa (Fig. 6b), which indicated that after cleavage at Arg-306, Rx1 was no longer susceptible to APC cleavage. This might be explained by the dissociation of the 60-and 45-kDa fragments from the FVa heterotrimer, a mechanism that has been proposed as explanation for the low FVa activity after cleavage at Arg-306 (39). In contrast, the final fragment in the degradation of activated rFV⌬B-GRA was 30 kDa, which indicates that Rx2 can still be cleaved by APC after cleavage at Arg-506. This is the first time that APC cleavage sites are reported in the heavy chain of FVa different from the sites at Arg-306, Arg-506, and Arg-679. During the inactivation of plasma FVa by APC, cleavage at Rx1 or Rx2 will hardly occur, because the second-order rate constants are much lower than those reported for cleavage at Arg-506 and Arg-306, even in the presence of protein S. This might explain why the sites of Rx1 and Rx2 have not been identified before in studies using normal FVa (10,30). Also, no novel APC cleavage sites have been observed in studies using rFV molecules in which Arg-306 and Arg-506 had been mutated (22,24,41,43). However, the conditions used in these studies were such (much lower APC concentrations and shorter incubation time) that cleavages at Rx1 and Rx2 could not be detected.
During the present study we observed that Rx1 is not cleaved by APC as an alternative for Arg-506 during the inactivation of activated FVL (Fig. 7), most likely because of dissociation of the A2-domain after cleavage at Arg-306. Inactivation of activated rFV⌬B-GRA on the other hand clearly demonstrated that cleavage at Rx2 may occur as an alternative for cleavage at Arg-306. Therefore, cleavage at Rx2 may be relevant in the APC-catalyzed inactivation of activated FV-Hongkong or FV-Cambridge and may explain the unexpected mild APC-resistant phenotype associated with the carriership of these FV mutants.