INTRODUCTION

Factor IX is a vitamin K-dependent plasma protein that plays a crucial role in blood coagulation since the absence of its activity results in an X-linked bleeding disorder known as hemophilia B. The human protein is synthesized in the liver as a precursor molecule of 461 amino acids (1). The first 46 amino acids constitute the prepro leader sequence that is removed before secretion of the molecule. Also during biosynthesis, the protein undergoes several posttranslational modifications that include -carboxylation of first 12 Glu residues, partial hydroxylation of Asp⁶⁴, and glycosylation at residues Ser⁵³, Ser⁶¹, Asn¹⁵⁷, Asn¹⁶⁷, Thr¹⁵⁹, and Thr¹⁶⁹ (1-5). The resulting mature protein of 415 amino acids (M_r 57,000) is a zymogen of serine protease factor IXa and contains 17% carbohydrate by weight (6).

Gene arrangement, amino acid sequence, and the x-ray structure of the protein strongly suggest that factor IX is organized into several distinct domains (1, 7). Circulating factor IX consists of an amino-terminal -carboxyglutamic acid (Gla)¹ domain (residues 1-40), a short hydrophobic segment (residues 41-46), two epidermal growth factor (EGF)-like domains (EGF1 residues 47-84 and EGF2 residues 85-127), an activation peptide region (residues 146-180), and the carboxyl-terminal serine protease domain (residues 181-415). Based upon the crystal structure of the Gla domain of factor VIIa (8), the Ca²⁺ binding properties of factor X,² and the NMR structure of the Gla domain of factor IX (10), it would appear that this domain in factor IX possesses several low to intermediate affinity Ca²⁺ binding sites. In addition, the EGF1 and protease domain each possess one high affinity Ca²⁺ binding site (11, 12).

During blood coagulation, factor IX can be activated by factor VIIa·Ca²⁺·tissue factor (TF) and by factor XIa·Ca²⁺ (13). Activation by either enzyme occurs in two steps (6, 13). In the first step, the Arg¹⁴⁵-Ala¹⁴⁶ bond is cleaved which yields a two-chain disulfide-linked inactive intermediate called factor IX.³ In the second step, which is also the rate-limiting step (6, 13, 14), the Arg¹⁸⁰-Val¹⁸¹ bond is cleaved giving rise to factor IXa (or simply factor IXa) and an activation peptide (AP). Factor IXa thus formed activates factor X in the clotting cascade. For maximal activation rate, this reaction requires Ca²⁺, phospholipid (PL) and factor VIIIa (15).

Existing evidence suggests that the Gla domain of factor IX binds to PL vesicles in the presence of Ca²⁺ (16). The EGF1 domain of factor IX is required for its activation by factor VIIa·TF; in factor IXa, it may also interact with factor VIIIa (17, 18). Moreover, the Ca²⁺ binding site in the EGF1 domain appears to be necessary for its interaction with factor VIIIa (17, 18). The role of the EGF2 domain is not clear, but it may be involved in protein-protein and protein-cofactor interactions (19). Finally, the protease domain is thought to play a primary role in binding to factor VIIIa (20, 21). Although attempts have been made to investigate the role of the protease domain Ca²⁺ binding site by mutational analysis, the data do not provide mechanistic details as to the inability of these mutants to function in clotting (22, 23). In this report, we have conducted a series of experiments to investigate the significance of the protease domain Ca²⁺ binding site in factor IX function. Our data indicate that occupancy of this site in factor IXa results in maximal catalytic efficiency and factor VIIIa binding. Further, proteolysis at Arg³¹⁸-Ser³¹⁹[150-151]⁴ in the autolysis loop of factor IXa leads to a reduction in its affinity for factor VIIIa binding, and this proteolysis is prevented by binding of Ca²⁺ to the protease domain. An account of this work has been presented in abstract form (24).

EXPERIMENTAL PROCEDURES

Benzoyl-Ile-Glu-Gly-Arg-p-nitroanilide (S-2222) was purchased from Helena Laboratories. Dansyl-Glu-Gly-Arg-chloromethyl ketone (DEGR-ck) was obtained from Calbiochem. Phosphatidylcholine, phosphatidylserine, recombinant hirudin, fatty acid free bovine serum albumin (BSA), polyethylene glycol 8000, and p-aminobenzamidine (p-AB) were obtained from Sigma. Factor IX- and factor VIII-deficient plasmas were purchased from George King Biomedicals, and activated partial thromboplastin time reagent was obtained from Diagnostica Stago. Normal human plasma factor IX (IX_NP) and factor X were isolated as described (25), and factor Xa was prepared as outlined (26). Purified human factor XI and -thrombin (IIa) were purchased from Enzyme Research Laboratories (South Bend, IN). Factor XIa was prepared as described (17). Recombinant human TF of amino acids 1-243 containing the transmembrane domain was generously provided by Genentech Inc. (South San Francisco) and reconstituted as described (27). Phosphatidylcholine-phosphatidylserine vesicles (75% phosphatidylcholine, 25% phosphatidylserine) were prepared by the method of Husten et al. (28) as outlined (27). Recombinant human factor VIIa was a generous gift of Novo-Nordisk (Copenhagen). A monoclonal antibody-purified human factor VIII was obtained from Dr. Leon Hoyer (American Red Cross, Rockville, MD). The preparation was free of all other coagulation factors and contained human albumin as a stabilizing agent.

Factor IX and factor IXa activities were measured in a one-stage assay with automated activated partial thromboplastin time reagent as described (29).

SDS-gel electrophoresis was performed using the Laemmli buffer system (30). The acrylamide concentration was 15%, and the gels were stained with Commassie Brilliant Blue.

Automated Edman degradation of each protein component was performed using an Applied Biosystems 477A gas phase Sequencer. Approximately 0.2-0.5 nmol of protein was loaded on the filter cartridge. The proteins from SDS-gels were transferred to polyvinylidene difluoride membranes as described by Rosenberg (31).

Wild type factor IX (IX_WT), IX_E235K (IX in which Glu²³⁵ has been replaced by lysine), and IX_E245V (IX in which Glu²⁴⁵ has been replaced by valine) were constructed, expressed, and purified as described (17, 32).

The putative model of the protease domain of human factor IXa was constructed using a homology model building approach described earlier (25). Crystallographic structure of the protease domain of porcine factor IXa in the absence of Ca²⁺ was used as the starting template (7). The structure of trypsin and elastase provided the templates for the factor IXa region near the putative Ca²⁺ binding site (33, 34).

Binding of p-AB was measured by an increase in its intrinsic fluorescence upon binding to the active site of each factor IXa protein using a Perkin-Elmer 650-10S fluorescence spectrophotometer. Details are given in a previous paper² and in the legend to Fig. 3.

Each factor IX protein (200 µg/ml) was activated for 6 h by VIIa·TF complex (2 µg/ml) in the presence of 1 mM PL vesicles in TBS, pH 7.4 (0.05 M Tris, 0.15 M NaCl, pH 7.4) containing 5 mM Ca²⁺. SDS-gel electrophoretic analysis revealed full activation to IXa without degradation to IXa forms. DEGR-IXa_NP, DEGR-IXa_WT, DEGR-IXa_E235K, and DEGR-IXa_E245V were prepared by adding 20-fold molar excess of DEGR-ck to each reaction tube. The pH was adjusted to 7.4, and each tube was incubated at 37 °C for 2 h. At this time, an additional 20-fold molar excess of the inhibitor was added, pH adjusted to 7.4, and the tubes incubated for an additional 2-h period at 37 °C. Next, each tube again received 20-fold excess of the inhibitor; the samples were then incubated overnight at 4 °C, and the excess inhibitor was removed as follows. The samples were made 10 mM in EDTA and passed through Centricon 100 to remove the PL vesicles and relipidated TF. Free DEGR-ck was removed as described earlier (27, 35).

DEGR-IXa_E245V was prepared as follows. Factor IX_E245V was activated to IXa_E245V as described in the legend to Fig. 6. DEGR-IXa_E245V was prepared as above except four successive additions of the DEGR-ck were made instead of the three earlier; after the third addition, the tube was incubated for 2 h at 37 °C before the last addition and incubation overnight. Free DEGR-ck was removed as described earlier (27, 35). The absence of free DEGR-ck in our DEGR-IXa preparations was confirmed by the lack of their abilities to inhibit S-2222 hydrolysis by purified factor Xa. Moreover, when a known extinction coefficient (3,940 M¹ at 340 nm) of the dansyl probe was used (36), we obtained stoichiometric (1.1 ± 0.05) incorporation of the inhibitor into each factor IXa protein.

Except for reverse titration experiments (see Fig. 4), factor VIII at 40 units/ml was activated with 0.2 nM IIa in TBS/BSA, pH 8.0 (0.05 M Tris, 0.15 M NaCl, pH 8, containing 1 mg/ml BSA) and 5 mM Ca²⁺. 10-µl aliquots were removed at 30-s intervals, diluted in cold TBS/BSA, 5 mM CaCl₂, pH 8.0, and assayed for factor VIII activity in a modified activated partial thromboplastin time assay. In this assay, 50 µl of hereditary factor VIII-deficient plasma was incubated with 50 µl of automated activated partial thromboplastin time reagent for 5 min at 37 °C. At this time, 50 µl each of 25 mM CaCl₂ and the test sample were added simultaneously and the clotting time noted. In initial experiments, it was found that factor VIII activity increased ~10-fold at 2 min, after which it declined steadily (32). Based upon these observations, we activated factor VIII with IIa for 2 min at which time IIa was inhibited by recombinant hirudin (2.2 nM, final concentration). The factor VIII sample was used immediately in the factor X activation experiments. In all experiments described in this paper, factor VIII activity units refer to those before activation with IIa. The functional molar concentration of IIa-activated factor VIII was determined by a technique described previously (37) for TF. The details are given in the legend to Fig. 4.

These experiments were performed with IXa_NP and IXa_WT in both the absence and presence of PL vesicles. 50-µl reaction mixtures (in TBS/BSA, 5 mM CaCl₂, pH 8.0) in the absence of PL vesicles were prepared containing various concentrations of factor IXa_NP or factor IXa_WT (0-10 µM), a fixed concentration of factor VIIIa (20 units/ml prior to IIa activation),⁵ and a constant concentration of factor X, which was added last to initiate the reaction. The activation was carried out at 37 °C for 30-120 s at which time 1 µl of 0.5 M EDTA was added to stop further generation of factor Xa. A 40-µl aliquot was then added to a 0.1-ml quartz cuvette containing S-2222 in 75 µl of TBS/BSA, pH 8.0. The final concentration of S-2222 was 100 µM. The p-nitroaniline release was measured continuously (A₄₀₅/min) for up to 20 min (27).² Factor Xa generated was calculated from a standard curve constructed using factor Xa prepared by insolubilized Russell's viper venom. In control experiments, at each factor X concentration used, the rate of factor Xa generation was also measured at various concentrations of factor IXa (NP or WT) in the absence of factor VIIIa; these control values were ~5% of the experimental values in the presence of factor VIIIa and were subtracted before analysis of the data. The EC₅₀ (functional K_d) was calculated as the free concentration of factor IXa which provided 50% of the V_max using the enzyme kinetics program from Erithacus Software (GraFit). To investigate the dependence of EC₅₀ on factor X concentration, a series of such experiments was performed using several concentrations of factor X ranging from 15 nM to 5 µM. To obtain initial rates of factor Xa generation, less than 5% of factor X was allowed to activate in these experiments. Further, to prevent activation of factor X by the generated factor Xa (38), reactions were stopped before the formation of ~8 nM factor Xa in these experiments. This is based upon our observation that in these experiments factor Xa generation is linear with time only up to 10 nM, after which it increases with an upward slope.

The above experiments were also carried out in which the reaction mixtures contained 10 µM PL vesicles; this concentration of PL vesicles was chosen because it gave optimal rates of factor X activation (32). The concentration of factor VIIIa in these experiments was fixed at 0.1 unit/ml (before IIa activation), and the concentration of factor IXa (NP or WT) ranged from 0 to 20 nM for each concentration of factor X. To obtain EC₅₀ (functional K_d) values as a function of substrate concentration, a series of experiments was performed in which factor X was varied from 7.5 nM to 5 µM.

K_d,app values of factor VIIIa binding to various active site-blocked factor IXa proteins were determined from their abilities to inhibit factor X activation in the Tenase system in the absence or presence of PL vesicles. Reaction mixtures (50 µl) in the absence of PL contained 0.1 µM factor IXa_WT, 2 µM factor X, 20 units/ml factor VIII, 5 mM Ca²⁺, and varying concentrations of DEGR-IXa proteins. Reaction mixtures in the presence of PL vesicles contained 0.2 nM factor IXa_WT, 0.48 µM factor X, 0.1 unit/ml factor VIII, 5 mM Ca²⁺, 10 µM PL, and varying concentrations of DEGR-IXa proteins. The protocols for carrying out factor X activation experiments were as described above for EC₅₀ determinations.

Binding of each DEGR-IXa protein to factor VIIIa was determined from the above competition experiments that may be represented by the scheme presented in Equation 1. In this scheme, Ca²⁺ and/or PL are omitted for simplicity, and an assumption is made that each factor IXa protein binds reversibly to factor VIIIa with a stoichiometry of 1 mol of VIIIa/mol of IXa (39).

(Eq. 1)

The experimental conditions in both the absence and presence of PL are those in which factor IXa_WT concentrations are below the EC₅₀ values, and in each case <10% of it is bound to factor VIIIa in the absence of the competitor. Further, no measurable rates of activation of factor X in the absence of added factor VIIIa ± PL were observed under the conditions of these experiments. Thus, a decrease in the rate of factor X activation at a given concentration of DEGR-IXa protein is directly proportional to the reduced formation of the IXa_WT·VIIIa complex in the reaction mixtures. The steady-state inhibition curves generated under these conditions were analyzed to obtain the IC₅₀ values (concentration of the competitor yielding 50% inhibition) using IC₅₀-4 parameter logistic equation of Halfman (40) given below,

(Eq. 2)

where y is the rate of Xa formation in the presence of a given concentration of DEGR-IXa protein represented by x, a is the maximum rate of Xa formation in the absence of DEGR-IXa, and s is the slope factor. Each point was weighted equally, and the data were fitted to Equation 2 using the nonlinear regression analysis program obtained from Erithacus Software (GraFit). The value of the slope factor s was between 0.9 and 1.1 in all experiments indicating competition for a single binding site. The background value represented <5% of the maximum rate of Xa formation in the absence of DEGR-IXa.

To obtain the K_d,app values for the interaction of DEGR-IXa proteins with factor VIIIa, we used the following equation as described by Cheng and Prusoff (41) and discussed further by Craig (42).

(Eq. 3)

where A is the concentration of IXa_WT, and EC₅₀ is the concentration of factor IXa_WT which gives a 50% maximum response in the absence of the competitor at a specified concentration of factor X used in the experiment.

RESULTS AND DISCUSSION

SDS electrophoretic analysis of factor IX proteins using the Laemmli system (30) is shown in Fig. 1. Each protein is effectively homogeneous in this system (see zero time sample, Fig. 1). Plasma factor IX and each recombinant protein had similar Gla content (10.7-11.4 residues) as measured by the technique of Przysiecki et al. (43). The amino-terminal sequence of each mutant protein was also determined. All proteins revealed a major and a minor sequence. The major sequence in each case was Tyr-Asn-Ser-Gly-Lys-Leu, and the minor sequence in each case was Thr-Val-Phe-Leu. The major sequence corresponds to the sequence of mature protein in plasma, and the minor sequence corresponds to the protein in which the prosequence has not been cleaved (1). The minor sequence amounted to ~5% in IX_WT, ~4% in IX_E235K, and ~5% in IX_E245V. The minor sequence was not detected in plasma factor IX. The relative coagulant activity of each protein was: IX_NP, 100% (180 ± 10 units/mg); IX_WT, ~90%; IX_E235K, ~30%; IX_E245V, ~4%. The coagulant activity of our factor IX_E235K preparation was consistently less than half of that reported by Hamaguchi and Stafford (23), who found that IX_E235K has 70-80% activity of plasma factor IX. The reason(s) for this discrepancy are not clear; however, it should be noted that all of our purified recombinant proteins, including IX_E235K, are fully carboxylated, and ~95% of the molecules have the amino-terminal sequence of the mature protein.

Activation of each recombinant factor IX protein by VIIa·TF·Ca²⁺ was similar to that of IX_NP (data not shown) as analyzed by SDS-gel electrophoresis (Fig. 1). In contrast, activation by factor XIa·Ca²⁺ led to proteolysis in the protease domain autolysis loop at Arg³¹⁸-Ser³¹⁹[150-151] in IXa_E245V with a concomitant loss of procoagulant activity; this proteolysis was moderate in IXa_E235K and minimal in IXa_WT (Fig. 2) and IXa_NP (data not shown). Because IXa_E235K and IXa_E245V were not cleaved in the autolysis loop during activation by VIIa·TF but were cleaved during activation by factor XIa, it would indicate that factor XIa catalyzes the cleavage of the Arg³¹⁸-Ser³¹⁹[150-151] peptide bond. Moreover, it allowed us to obtain via the VIIa·TF system IXa_E235K and IXa_E245V proteins that were not degraded for further studies. After removal of the PL and TF (see "Experimental Procedures"), factor IXa preparations were assayed for coagulant activity. These preparations contained <0.01% VIIa·TF complex as judged by factor X activation assays (27). Although these preparations are not suitable for rigorous kinetic analysis (K_m and V_max determinations), we were able to measure their coagulant activities by serial dilutions. In control experiments, at increasingly higher dilutions of factor IXa samples, the contaminating VIIa·TF concentrations (4-20 pM) did not appear to influence the coagulant activity of purified IXa_NP or IXa_WT prepared by factor XIa. The coagulant activity of VIIa·TF-activated IXa_WT was ~93%, of IXa_E235K was ~27%, and of IXa_E245V was ~4% compared with IXa_NP (~9,800 units/mg).

From the coagulant activity data, we conclude that the loss of the Ca²⁺ binding site in the protease domain leads to ~25-fold reduction in the biologic activity of factor IXa and that lysine at position Glu²³⁵[70] only partially substitutes in maintaining the Ca²⁺-bound conformer of this domain. Our data on IX_E235K are consistent with the factor X studies of Rezaie and Esmon (44), who found that lysine at position 250 (equivalent to 235 in factor IX) does not totally mimic the influence of Ca²⁺ on factor X structure and function. Further, our data indicate that occupancy of the Ca²⁺ site in the protease domain protects factor XIa-mediated proteolysis at the Arg³¹⁸-Ser³¹⁹[150-151] peptide bond in the so-called autolysis loop (Fig. 2). Similarly, Ca²⁺ has been reported to inhibit significantly IIa-mediated proteolysis in the autolysis loop of plasma factor IX (45). Moreover, the autolysis loop-cleaved factor IXa_NP (45) or factor IXa_E245V (present study) results in a complete loss of coagulant activity. To further these studies, we have used IXa_NP, IXa_WT, IXa_E235K, IXa_E245V, and IXa_E245V to investigate their binding to p-AB (availability of S1 site) and their abilities to compete for factor VIIIa in the Tenase complex (IXa·VIIIa·Ca²⁺ ± PL).

The data for p-AB binding are presented in Fig. 3 and summarized in Table I. The fluorescence curve was observed to shift dramatically to the right in the absence of Ca²⁺ for IXa_WT and IXa_NP (data not shown) with the K_d shifting from ~80 µM to ~230 µM. Our data for IXa_NP are consistent with a previous report presented in abstract form (46). In contrast, the curves were identical for the mutants in the presence and absence of Ca²⁺. Moreover, the K_d of IXa_E245V was the same as IXa_WT in the absence of Ca²⁺, whereas it was ~150 µM for the IXa_E235K. This again indicates that lysine at position 235 can only partially substitute for the Ca²⁺ site to maintain a native conformation. p-AB is known to bind to the S1 site of serine proteases. Thus, our data would indicate that binding of Ca²⁺ to the protease domain increases the reactivity of S1 site by ~3-fold.

Table I. Dissociation constants for interaction of p-AB with various factor IXa mutants Protein Kda Ca2+ (5 mM) EDTA (1 mM) µM Factor IXaWT 78  ± 4 230  ± 12 Factor IXaE235K 150  ± 5 145  ± 6 Factor IXaE245V 225  ± 13 241  ± 15 Factor IXaE245V 330  ± 5 350  ± 6 a Kd values were calculated using the data in Fig. 3. The Kd value for factor IXaNP was 81 ± 3 µM in the presence of Ca2+, and it was 227 ± 10 µM in the presence of EDTA.

Because IXa_E245V could be prepared easily by incubating IX_E245V with XIa·Ca²⁺ (see gel data presented later in Fig. 6A, inset), we also investigated its binding to p-AB. These data are presented in Fig. 3D. IXa_E245V bound p-AB with slightly reduced affinity (K_d ~340 µM ± Ca²⁺); however, the enhancement of the intrinsic fluorescence of p-AB upon binding to the S1 site was only ~25% of that observed with IXa_E245V. It should be noted that the enhancement of intrinsic fluorescence of p-AB was not observed when DEGR-IXa_E245V or DEGR-IXa_E245V was used in the p-AB titration experiments; this strongly indicates that the increase in intrinsic fluorescence observed with IXa_E245V is the result of the binding of p-AB at the active site. Because a reduced fluorescence increase was observed with IXa_E245V, it would indicate that the environment of the p-AB bound to the S1 site in IXa_E245V is less nonpolar compared with that in the IXa_E245V molecule. Based upon the work with other serine proteases (47), it would appear that Trp³⁸⁵[215] in factor IXa contributes to the nonpolar environment and therefore enhancement of p-AB intrinsic fluorescence upon binding at the active site; if so, then this region in factor IXa_E245V is perturbed without the loss of S1 binding site. Such is also the case with factor Xa.²

Recently, p-AB binding to factor IXa mutants lacking the protease domain Ca²⁺ binding site has been reported (23); these authors performed all experiments in the presence of 5 mM Ca²⁺. In their study, binding of p-AB to factor IXa_E235K caused an increase in the fluorescence intensity which was similar to the IXa_WT. Moreover, only a small increase in the fluorescence intensity was observed in the case of either factor IXa_E245K or factor IXa_E235K&E245K (23). These observations can be rationalized based upon the data presented in Fig. 3. Previous measurements were made at a single concentration (150 µM) of p-AB and in the presence of 5 mM Ca²⁺ (23). It is evident from Fig. 3 that this concentration is higher than the K_d (~80 µM) for p-AB binding to factor IXa_WT in the presence of Ca²⁺ and is equal to the K_d (~150 µM) for binding of p-AB to factor IXa_E235K in the presence or absence of Ca²⁺. Thus, under these conditions (23), we estimate that 60 ± 10% of both factor IXa_WT and factor IXa_E235K molecules will have p-AB bound at their active sites. As observed (23), this will result in a similar increase in fluorescence intensity for both of these proteins. In the case of factor IX_E245K (or IX_E235K&E245K), because factor XIa was used for activation, a significant proportion of each IXa protein could be proteolyzed in the autolysis loop and exhibit reduced fluorescence enhancement. Under the conditions of their experiments (23) and using the data of Fig. 3D, we calculate that for these mutants a fluorescence intensity change of 15-20% will be observed, an estimate that is close to the value reported.

Influence of Factor X on the EC₅₀ of Interaction of IXa_NP and IXa_WT with Factor VIIIa

For these experiments, we first wished to determine the concentration of active factor IXa binding sites in our IIa-activated factor VIII preparation. Two sets of experiments were performed for this purpose. Data were obtained with both IXa_WT and IXa_NP. Because similar results were obtained with both proteins, only the data with the IXa_WT are given. In one set of experiments, the factor IXa concentration was varied, and the factor VIII concentration was kept constant; before IIa activation, it was 0.1 unit/ml in the presence and 20 units/ml in the absence of PL. The rates of activation of factor X (2 µM in the presence and 3 µM in the absence of PL) were measured. These data for the IXa_WT are presented in Fig. 4, A and B, in the presence and absence of PL, respectively. In a second set of experiments, the factor VIII concentration was varied, and the factor IXa concentration was kept constant; it was 0.025 nM in the presence and 15 nM in the absence of PL. Again, the rates of activation of factor X (2 µM in the presence and 3 µM in the absence of PL) were measured. These data for IXa_WT are presented in Fig. 4, C and D, in the presence and absence of PL, respectively. Values of V_max were calculated using the enzyme kinetics program from Erithacus software. The concentrations of factor VIII in molar terms were calculated using the following equation,

(Eq. 4)

where V_max1 is the rate of factor X activation at a constant concentration of IIa-activated factor VIII (Fig. 4, A or B) and V_max2 is the rate of factor X activation at a constant concentration of IXa_WT (Fig. 4, C or D). Using Equation 4 and the data of Fig. 4, A and C, we calculate that 1 unit/ml factor VIII clotting activity (before IIa activation) corresponds to 0.65 nM. Similarly, from the data of Fig. 4, B and D, 1 unit/ml factor VIII corresponds to 0.75 nM. Thus, 1 unit/ml factor VIII after IIa activation contains ~0.7 nM factor IXa binding sites as measured in the Tenase assay system with or without PL.

The nanomolar concentrations of factor VIIIa when indicated in this paper are based upon the above calculations. Here, we wish to point out that factor VIII after cleavage by thrombin is not a stable protein, and therefore all studies of factor IXa-factor VIIIa interaction are complicated by this inherent instability of factor VIIIa. This is complicated further by the fact that the unstable factor VIIIa molecule is, in part, stabilized by complexation with factor IXa and PL vesicles (48). Moreover, potential also exists that factor IXa at high concentrations may slowly proteolyze factor VIIIa, leading to further losses in factor VIIIa activity (49). Therefore, determinations of EC₅₀ values (functional K_ds) and K_d,app values of IXa-VIIIa interaction presented below should be interpreted with this caveat in mind. However, one should note that the knowledge of absolute concentrations of factor VIIIa is not critical to the conclusions drawn from this paper.

Next, we measured the EC₅₀ of interaction of IXa with factor VIIIa. These data are presented in Fig. 5 and summarized in Table II. Clearly the EC₅₀ value is influenced by the amount of factor X present in the reaction mixture in a PL-free system as well as in a PL-containing system. In the absence of PL, the EC₅₀ of IXa_NP or IXa_WT-factor VIIIa interaction at saturating concentrations of factor X is ~0.25 µM, and an extrapolated value in the absence of factor X is ~2.8 µM. In the presence of PL, the EC₅₀ of IXa_NP or IXa_WT-factor VIIIa interaction at saturating concentrations of factor X is ~0.09 nM, and an extrapolated value in the absence of factor X is ~1.6 nM. Thus, factor X decreases the EC₅₀ (functional K_d) by an order of magnitude both in the presence or absence of PL, whereas PL decreases this value by ~2,000-fold in either the presence or absence of factor X. 

) of IXa_NP or IXa_WT-factor VIIIa interaction.

Table II. Dependence of EC50 (functional Kd) of factor IXa-factor VIIIa interaction on PL and factor X concentration in the Tenase system Factor X EC50a PL-plus PL-minus Zero (extrapolated) 1.6  ± 0.07 nM 2.8  ± 0.1 µM Saturating 0.09  ± 0.01 nM 0.25  ± 0.01 µM a EC50 values were calculated using the data in Fig. 5 and are the same for both for IXaNP and IXaWT.

In an earlier study (39), it was reported that the factor IXa-factor VIIIa interaction is not influenced by the presence of factor X. However, in that study the total concentration of factor IXa used in the factor VIIIa titration experiment was ~19 nM, which is substantially higher than the functional K_d of ~2 nM (39) or EC₅₀ value of ~0.09 nM (present study). Use of such high concentrations of factor IXa ligand can lead to large inaccuracies in measuring K_d values in the picomolar range and obscure the effect of factor X. Our data using a wide range of factor X concentrations clearly demonstrate that the EC₅₀ (functional K_d) of the factor IXa-factor VIIIa interaction is dependent upon the concentration of the substrate, factor X. The functional K_d of the interaction of factor VIIa with TF has also been reported to be dependent upon the substrate concentration (50). However, it should be noted that the EC₅₀ (functional K_d) value for the factor Xa-factor Va interaction is not dependent upon the prothrombin concentration in the prothrombinase system (51).² An explanation for these observations might be that factor X (or another substrate) binding to factor IXa or factor VIIa locks each enzyme into a favorable conformation for interaction with factor VIIIa or tissue factor, respectively. In contrast, factor Xa might exist in a conformation, which is optimal for interaction with factor Va even in the absence of prothrombin.

Our data on the effect of PL in increasing the interaction of factor IXa with factor VIIIa are consistent with the observations made with other PL-dependent enzyme-cofactor interactions (27, 36, 50, 52, 53).² In the presence of PL vesicles, the K_d of the interaction of factor VIIa with TF decreases from ~1-10 nM to ~5-10 pM (27, 36, 50), and the K_d of the interaction of factor Xa with factor Va decreases from ~1 µM (52, 53)² to ~30 pM (52).² The effect of PL in all cases may be attributable to restricting the rotational and translational diffusion of the PL-bound proteins involved in interaction with each other (54). Overall, the data indicate that in the presence of PL, all three enzymes (factors VIIa, IXa, and Xa) bind to their respective cofactors (tissue factor for VIIa, VIIIa for IXa, and Va for Xa) with apparent dissociation constants in the low picomolar range.

Factor XIa-mediated activation of factor IX mutants lacking the protease domain Ca⁺ binding site results in proteolysis in the autolysis loop with concomitant loss of coagulant activity (Fig. 2). Furthermore, although these mutants could be activated normally without proteolysis in the autolysis loop by VIIa·TF (Fig. 1), we were unable to remove VIIa·TF completely from the activation mixtures. Since even minuscule amounts of contaminating VIIa·TF complex can contribute significantly to factor X activation, especially in the case of mutants, it precluded our measurements of EC₅₀ values (functional K_d) for the interaction of each activated mutant with factor VIIIa. However, each preparation of factor IXa protein after inhibiting with DEGR-ck was found to have no VIIa·TF activity as measured by its ability to activate factor X. To determine the K_d values of interaction of these active site-blocked mutants with factor VIIIa, we evaluated their abilities to compete with factor IXa_WT (prepared by factor XIa activation) in binding to factor VIIIa in the Tenase complex. Because a mutation in factor IXa could alter factor VIIIa binding by at least two mechanisms (by perturbation of the contact site and by spatial misalignment of the otherwise normal contact site above the PL surface) we investigated the inhibition of Tenase activity by these mutants in the presence and absence of PL vesicles.

The steady-state inhibition curves (55) obtained in the presence (Fig. 6A) and absence (Fig. 6B) of PL were analyzed as outlined under "Experimental Procedures." The K_d,app values for the interaction of active site-blocked mutants with factor VIIIa are listed in Table III. DEGR-IXa_WT and DEGR-IXa_NP interacted with factor VIIIa with a K_d ~70 pM; this value of K_d is close to the estimated EC₅₀ value (~90 pM) obtained at saturating concentrations of factor X. Because the DEGR moiety in factor IXa is expected not to participate in direct binding to factor VIIIa, it supports the concept that the increase in the affinity of factor IXa is the result of a conformational change induced by occupancy of the active site, either by DEGR-ck or by factor X. Our K_d,app value (~70 pM) of DEGR-IXa_WT interaction with factor VIIIa is much lower than the value (~2 nM) obtained by fluorescence anisotropy measurements (39). As pointed out earlier, this difference may be attributable to the ~300-fold higher concentration (above the K_d,app value) of active site-blocked factor IXa used in previous experiments.

Table III. Apparent Kd values for the interaction of various factor IXa proteins with factor VIIIa Protein Apparent Kda PL-plus PL-minus nM µM DEGR-IXaWT or DEGR-IXaNP 0.07  ± 0.01 0.19  ± 0.04 DEGR-IXaE235K 0.26  ± 0.02 0.68  ± 0.08 DEGR-IXaE245V 1.35  ± 0.21 2.49  ± 0.19 DEGR-IXaE245V 14.3  ± 0.5 15.6  ± 1.9 a The data of Fig. 6 were used to calculate the apparent Kd values.

The binding of factor VIIIa to each active site-blocked mutant was considerably weaker both in the presence and absence of PL vesicles (Table III). Compared with DEGR-IXa_WT, DEGR-IXa_E235K had similarly (~4-fold) reduced affinity for factor VIIIa in the presence or absence of PL. However, DEGR-IXa_E245V had ~20-fold reduced affinity in the presence of PL and ~13-fold reduced affinity in the absence of PL, whereas DEGR-IXa_E245V had ~200-fold reduced affinity in the presence of PL and ~80-fold reduced affinity in the absence of PL. These data indicate that in the case of DEGR-IXa_E245V or DEGR-IXa_E245V, in addition to the perturbation of factor VIIIa binding site, a further reduced affinity in the presence of PL could, in part, be due to the misalignment of the factor VIIIa contact site. Importantly, these data indicate that the protease domain Ca²⁺ binding site is essential for stabilizing the native conformation of this domain needed for factor VIIIa binding.

In this paper we have investigated the structural and functional significance of the Ca²⁺ binding site in the protease domain of human factor IXa. Based upon the three-dimensional structures of several serine protease (8, 33, 34), it is almost certain that the side chain carboxyl groups of Glu²³⁵[70] and Glu²⁴⁵[80] in factor IXa participate in binding to Ca²⁺ (12). In the current study, this Ca²⁺ binding site was abolished by replacing separately Glu²³⁵ by lysine and Glu²⁴⁵ by valine. The rationale for constructing E235K mutant was that thrombin and guinea pig factor IX each have lysine at this position (25) and that thrombin does not bind Ca²⁺ at this site (56). The E245V mutant was constructed based upon a naturally occurring variant in a hemophilia B patient (22). From our data, it is clear that occupancy of this Ca²⁺ binding site increases the binding of factor VIIIa to factor IXa by ~15-fold and reactivity of the S1 site by ~3-fold. Our data also indicate that lysine at position 235 cannot fully mimic the function of this Ca²⁺ binding site. However, compared with factor IX_WT, ~30% coagulant activity of factor IX_E235K would be adequate for normal hemostasis. Based upon the results with factor IXa_E245V, factor IXa_WT compared with IXa_WT is expected to have ~8-fold reduced affinity for factor VIIIa. This is consistent with our finding that compared with native factor Xa, autolysis loop-cleaved factor Xa has ~10-fold reduced affinity for factor Va.² Moreover, consistent with previous observations (20, 21, 57), our data provide further evidence that the protease domain of factor IXa constitutes a part of the binding site for factor VIIIa. However, it should be noted that the Ca²⁺ binding loop may not directly participate in factor VIIIa binding (58). The spatial arrangement of Ca²⁺ binding site, autolysis loop, and S1 site is depicted in Fig. 7. An examination of this figure would indicate that binding of Ca²⁺ in this domain allosterically affects the factor VIIIa as well as the S1 binding site. Further, binding of factor VIIIa to this domain appears to alter the conformation of the active site (9) with a resultant increase in the k_cat for factor X activation (15).