Interaction of Factor IXa with Factor VIIIa

We previously identified a high affinity Ca2+ binding site in the protease domain of factor IXa involving Glu235 (Glu70 in chymotrypsinogen numbering; hereafter, the numbers in brackets refer to the chymotrypsin equivalents) and Glu245[80] as putative ligands. To delineate the function of this Ca2+ binding site, we expressed IXwild type (IXWT), IXE235K, and IXE245V in 293 kidney cells and compared their properties with those of factor IX isolated from normal plasma (IXNP); each protein had the sameM r and γ-carboxyglutamic acid content. Activation of each factor IX protein by factor VIIa·Ca2+·tissue factor was normal as analyzed by sodium dodecyl sulfate-gel electrophoresis. The coagulant activity of IXaWT was ∼93%, of IXaE235K was ∼27%, and of IXaE245V was ∼4% compared with that of IXaNP. In contrast, activation by factor XIa·Ca2+ led to proteolysis at Arg318-Ser319[150–151] in the protease domain autolysis loop of IXaE245V with a concomitant loss of coagulant activity; this proteolysis was moderate in IXaE235K and minimal in IXaWT or IXaNP. Interaction of each activated mutant with an active site probe, p-aminobenzamidine, was also examined; theK d of interaction in the absence and presence (in parentheses) of Ca2+ was: IXaNP or IXaWT 230 μm (78 μm), IXaE235K 150 μm (145 μm), IXaE245V 225 μm (240 μm), and autolysis loop cleaved IXaE245V 330 μm (350 μm). Next, we evaluated the apparentK d (K d,app) of interaction of each activated mutant with factor VIIIa. We first investigated the EC50 of interaction of IXaNPas well as of IXaWT with factor VIIIa in thepresence and absence of phospholipid (PL) and varying concentrations of factor X. At each factor X concentration and constant factor VIIIa, EC50 was the free IXaNPor IXaWT concentration that yielded a half-maximal rate of factor Xa generation. EC50 values for IXaNP and IXaWT were similar and are as follows: PL-minus/X-minus (extrapolated), 2.8 μm; PL-minus/X-saturating, 0.25 μm; PLplus/X-minus, 1.6 nm; and PL-plus/X-saturating, 0.09 nm. Further,Kd ,app of binding of active site-blocked factor IXa to factor VIIIa was calculated from its ability to inhibit IXaWT in the Tenase assay.Kd ,app values in the absence and presence (in parentheses) of PL were: IXaNPor IXaWT, 0.19 μm (0.07 nm); IXaE235K, 0.68 μm (0.26 nm); IXaE245V, 2.5 μm (1.35 nm); and autolysis loop-cleaved IXaE245V, 15.6 μm(14.3 nm). We conclude that (a) PL increases the apparent affinity of factor IXa for factor VIIIa ∼2,000-fold, and the substrate, factor X, increases this affinity ∼10–15-fold; (b) the protease domain Ca2+ binding site increases this affinity ∼15-fold, and lysine at position 235 only partly substitutes for Ca2+; (c) Ca2+ binding to the protease domain increases the S1 reactivity ∼3-fold and prevents proteolysis in the autolysis loop; and (d) proteolysis in the autolysis loop leads to a loss of catalytic efficiency with retention of S1 binding site and a further ∼8-fold reduction in affinity of factor IXa for factor VIIIa.

We conclude that (a) PL increases the apparent affinity of factor IXa for factor VIIIa ϳ2,000-fold, and the substrate, factor X, increases this affinity ϳ10 -15-fold; (b) the protease domain Ca 2؉ binding site increases this affinity ϳ15-fold, and lysine at position 235 only partly substitutes for Ca 2؉ ; (c) Ca 2؉ binding to the protease domain increases the S1 reactivity ϳ3-fold and prevents proteolysis in the autolysis loop; and (d) proteolysis in the autolysis loop leads to a loss of catalytic efficiency with retention of S1 binding site and a further ϳ8-fold reduction in affinity of factor IXa for factor VIIIa.
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 Asp 64 , and glycosylation at residues Ser 53 , Ser 61 , Asn 157 , Asn 167 , Thr 159 , and Thr 169 (1)(2)(3)(4)(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][42][43][44][45][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 Ca 2ϩ 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 Ca 2ϩ binding sites. In addition, the EGF1 and protease domain each possess one high affinity Ca 2ϩ binding site (11,12).
During blood coagulation, factor IX can be activated by factor VIIa⅐Ca 2ϩ ⅐tissue factor (TF) and by factor XIa⅐Ca 2ϩ (13). Activation by either enzyme occurs in two steps (6,13). In the first step, the Arg 145 -Ala 146 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 Arg 180 -Val 181 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 Ca 2ϩ , phospholipid (PL) and factor VIIIa (15).
Existing evidence suggests that the Gla domain of factor IX binds to PL vesicles in the presence of Ca 2ϩ (16). The EGF1 domain of factor IX is required for its activation by factor VIIa⅐TF; in factor IXa, it may also interact with factor VIIIa (17,18). Moreover, the Ca 2ϩ 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 Ca 2ϩ 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 Ca 2ϩ 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 Arg 318 -Ser 319 [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 Ca 2ϩ to the protease domain. An account of this work has been presented in abstract form (24).

EXPERIMENTAL PROCEDURES
Proteins and Reagents-Benzoyl-Ile-Glu-Gly-Arg-p-nitroanilide (S-2222) was purchased from Helena Laboratories. Dansyl-Glu-Gly-Argchloromethyl ketone (DEGR-ck) was obtained from Calbiochem. Phosphatidylcholine, phosphatidylserine, recombinant hirudin, fatty acid free bovine serum albumin (BSA), polyethylene glycol 8000, and paminobenzamidine (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.
Coagulation Assay of Factor IX and Factor IXa-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-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.
Amino Acid Sequence Analysis-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).
Construction, Expression, and Purification of Recombinant Factor IX Proteins-Wild type factor IX (IX WT ), IX E235K (IX in which Glu 235 has been replaced by lysine), and IX E245V (IX in which Glu 245 has been replaced by valine) were constructed, expressed, and purified as described (17,32).
Molecular Modeling-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 Ca 2ϩ was used as the starting template (7). The structure of trypsin and elastase provided the templates for the factor IXa region near the putative Ca 2ϩ binding site (33,34).
p-AB Binding-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 paper 2 and in the legend to Fig. 3.
Preparation of DEGR-ck Inhibited Various Factor IXa Proteins-Each factor IX protein (200 g/ml) was activated for 6 h by VIIa⅐TF complex (2 g/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 Ca 2ϩ . 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 M Ϫ1 at 340 nm) of the dansyl probe was used (36), we obtained stoichiometric (1.1 Ϯ 0.05) incorporation of the inhibitor into each factor IXa protein.
Activation of Factor VIII by IIa-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 Ca 2ϩ . 10-l aliquots were removed at 30-s intervals, diluted in cold TBS/BSA, 5 mM CaCl 2 , pH 8.0, and assayed for factor VIII activity in a modified activated partial thromboplastin time assay. In this assay, 50 l of hereditary factor VIII-deficient plasma was incubated with 50 l of automated activated partial thromboplastin time reagent for 5 min at 37°C. At this time, 50 l each of 25 mM CaCl 2 and the test sample were added simultaneously and the clotting time noted. In initial experiments, it was found that factor VIII activity increased ϳ10-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.
Determination of EC 50 of Factor IXa-Factor VIIIa Interaction in the Tenase Complex-These experiments were performed with IXa NP and IXa WT in both the absence and presence of PL vesicles. 50-l reaction mixtures (in TBS/BSA, 5 mM CaCl 2 , pH 8.0) in the absence of PL vesicles were prepared containing various concentrations of factor IXa␤ NP or factor IXa␤ WT (0 -10 M), 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 l of 0.5 M EDTA was added to stop further generation of factor Xa. A 40-l aliquot was then added to a 0.1-ml quartz cuvette containing S-2222 in 75 l of TBS/BSA, pH 8.0. The final concentration of S-2222 was 100 M. The p-nitroaniline release was measured continuously (⌬A 405 /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 ϳ5% of the experimental values in the presence of factor VIIIa and were subtracted before analysis of the data. The EC 50 (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 EC 50 on factor X concentration, a series of such experiments was performed using several concentrations of factor X ranging from 15 nM to 5 M. 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 ϳ8 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 M 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 EC 50 (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 M.
Determinations of K d,app Values for the Interaction of Factor VIIIa with Various Active Site-blocked Factor IXa Proteins-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 l) in the absence of PL contained 0.1 M factor IXa␤ WT , 2 M factor X, 20 units/ml factor VIII, 5 mM Ca 2ϩ , and varying concentrations of DEGR-IXa proteins. Reaction mixtures in the presence of PL vesicles contained 0.2 nM factor IXa␤ WT , 0.48 M factor X, 0.1 unit/ml factor VIII, 5 mM Ca 2ϩ , 10 M PL, and varying concentrations of DEGR-IXa proteins. The protocols for carrying out factor X activation experiments were as described above for EC 50 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, Ca 2ϩ 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).
The experimental conditions in both the absence and presence of PL are those in which factor IXa WT concentrations are below the EC 50 values, and in each case Ͻ10% 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 Ϯ 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 ⅐VIIIa complex in the reaction mixtures. The steady-state inhibition curves generated under these conditions were analyzed to obtain the IC 50 values (concentration of the competitor yielding 50% inhibition) using IC 50 -4 parameter logistic equation of Halfman (40) given below, 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 Ͻ5% 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).
where A is the concentration of IXa WT , and EC 50 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
Purification, Gla Content, NH 2 -terminal Sequence, and Activity of Factor IX Proteins-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 ϳ5% in IX WT , ϳ4% in IX E235K , and ϳ5% 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 Ϯ 10 units/mg); IX WT , ϳ90%; IX E235K , ϳ30%; IX E245V , ϳ4%. 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 ϳ95% of the molecules have the amino-terminal sequence of the mature protein.
Activation of Factor IX Proteins by VIIa⅐TF and by Factor XIa⅐Ca 2ϩ -Activation of each recombinant factor IX protein by VIIa⅐TF⅐Ca 2ϩ was similar to that of IX NP (data not shown) as analyzed by SDS-gel electrophoresis (Fig. 1). In contrast, activation by factor XIa⅐Ca 2ϩ led to proteolysis in the protease domain autolysis loop at Arg 318 -Ser 319 [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 VIIa⅐TF but were cleaved during activation by factor XIa, it would indicate that factor XIa catalyzes the cleavage of the Arg 318 -Ser 319 [150 -151] peptide bond. Moreover, it allowed us to obtain via the VIIa⅐TF 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 Ͻ0.01% VIIa⅐TF 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 VIIa⅐TF 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 VIIa⅐TF-activated IXa␤ WT was ϳ93%, of IXa␤ E235K was ϳ27%, and of IXa␤ E245V was ϳ4% compared with IXa␤ NP (ϳ9,800 units/mg).
From the coagulant activity data, we conclude that the loss of the Ca 2ϩ binding site in the protease domain leads to ϳ25-fold reduction in the biologic activity of factor IXa␤ and that lysine at position Glu 235 [70] only partially substitutes in maintaining the Ca 2ϩ -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 Ca 2ϩ on factor X structure and function. Further, our data indicate that occupancy of the Ca 2ϩ site in the protease domain protects factor XIa-mediated proteolysis at the Arg 318 -Ser 319 [150 -151] peptide bond in the so-called autolysis loop (Fig. 2). Similarly, Ca 2ϩ 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 (IXa⅐VIIIa⅐Ca 2ϩ Ϯ PL).
p-AB Binding-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 Ca 2ϩ for IXa␤ WT and IXa␤ NP (data not shown) with the K d shifting from ϳ80 M to ϳ230 M. 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 Ca 2ϩ . Moreover, the K d of IXa␤ E245V was the same as IXa␤ WT in the absence of Ca 2ϩ , whereas it was ϳ150 M for the IXa␤ E235K . This again indicates that lysine at position 235 can only partially substitute for the Ca 2ϩ 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 Ca 2ϩ to the protease domain increases the reactivity of S1 site by ϳ3-fold.
Because IXa␥ E245V could be prepared easily by incubating IX E245V with XIa⅐Ca 2ϩ (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 ϳ340 M Ϯ Ca 2ϩ ); however, the enhancement of the intrinsic fluorescence of p-AB upon binding to the S1 site was only ϳ25% 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 Trp 385 [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 Ca 2ϩ binding site has been reported (23); these authors performed all experiments in the presence of 5 mM Ca 2ϩ . 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 M) of p-AB and in the presence of 5 mM Ca 2ϩ (23). It is evident from Fig. 3 that this concentration is higher than the K d (ϳ80 M) for p-AB binding to factor IXa␤ WT in the presence of Ca 2ϩ and is equal to the K d (ϳ150 M) for binding of p-AB to factor IXa␤ E235K in the presence or absence of Ca 2ϩ . Thus, under these conditions (23), we estimate that 60 Ϯ 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 EC 50 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 M in the presence and 3 M 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 M in the presence and 3 M 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, 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 ϳ0.7 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 IXafactor 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 Factor IXa␤ preparations in each case were made by incubation of factor IX proteins with factor VIIa⅐TF using a 1:100 enzyme:substrate ratio for 6 h; analysis by SDS-gel electrophoresis indicated that Ͼ95% of each protein was converted to factor IXa␤. Factor IXa␥ E245V was prepared by incubation of factor IX E245V with factor XIa using a 1:50 enzyme:substrate ratio for 4 h; SDS-gel electrophoretic analysis indicated that the protein is completely converted to factor IXa␥ (gel is shown later in Fig. 6A,  inset, lane 3).   Fig. 3. The K d value for factor IXa␤ NP was 81 Ϯ 3 M in the presence of Ca 2ϩ , and it was 227 Ϯ 10 M in the presence of EDTA. slowly proteolyze factor VIIIa, leading to further losses in factor VIIIa activity (49). Therefore, determinations of EC 50 values (functional K d s) 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 EC 50 of interaction of IXa␤ with factor VIIIa. These data are presented in Fig. 5 and summarized in Table II. Clearly the EC 50 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 EC 50 of IXa␤ NP or IXa␤ WT -factor VIIIa interaction at saturating concentrations of factor X is ϳ0.25 M, and an extrapolated value in the absence of factor X is ϳ2.8 M. In the presence of PL, the EC 50 of IXa␤ NP or IXa␤ WT -factor VIIIa interaction at saturating concentrations of factor X is ϳ0.09 nM, and an extrapolated value in the absence of factor X is ϳ1.6 nM. Thus, factor X decreases the EC 50 (functional K d ) by an order of magnitude both in the presence or absence of PL, whereas PL decreases this value by ϳ2,000-fold in either the presence or absence of factor X.
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 ϳ19 nM, which is substantially higher than the functional K d of ϳ2 nM (39) or EC 50 value of ϳ0.09 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 EC 50 (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 EC 50 (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 The concentration of factor X in the system containing PL (panels A and C) was 2 M and in the system without PL (panels B and D) was 3 M. Factor Xa activity was measured by S-2222 hydrolysis as outlined under "Experimental Procedures." Factor VIII for these studies was activated at 4,880 units/ml with 2 nM IIa; this resulted consistently in a ϳ10-fold increase in clotting activity at 2 min, at which time hirudin (20 nM, final concentration) was added to inhibit IIa. Factor VIII was used immediately after activation. Analysis of the data to obtain factor IXa binding sites in our factor VIII preparation was carried out using Eq. 4.  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 ϳ1-10 nM to ϳ5-10 pM (27,36,50), and the K d of the interaction of factor Xa with factor Va decreases from ϳ1 M (52, 53) 2 to ϳ30 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.

Measurements of Apparent K d Values for the Interaction of Each Factor IXa Species with Factor VIIIa in the Tenase Complex-Factor
XIa-mediated activation of factor IX mutants lacking the protease domain Ca ϩ 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 VIIa⅐TF (Fig. 1), we were unable to remove VIIa⅐TF completely from the activation mixtures. Since even minuscule amounts of contaminating VIIa⅐TF complex can contribute significantly to factor X activation, especially in the case of mutants, it precluded our measurements of EC 50 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 VIIa⅐TF 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 ϳ70 pM; this value of K d is close to the estimated EC 50 value (ϳ90 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 (ϳ70 pM) of DEGR-IXa␤ WT interaction with factor VIIIa is much lower than the value (ϳ2 nM) obtained by fluorescence anisotropy measurements (39). As pointed out earlier, this difference may be attributable to the ϳ300-fold higher concentration (above the K d,app value) of active site-blocked factor IXa used in previous experiments.
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 (ϳ4-fold) reduced affinity for factor VIIIa in the presence or absence of PL. However, DEGR-IXa␤ E245V had ϳ20-fold reduced affinity in the presence of PL and ϳ13-fold reduced affinity in the absence of PL, whereas DEGR-IXa␥ E245V had ϳ200-fold reduced affinity in the presence of PL and ϳ80-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 Ca 2ϩ binding site is essential for stabilizing the native conformation of this domain needed for factor VIIIa binding.  Concluding Remarks-In this paper we have investigated the structural and functional significance of the Ca 2ϩ 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 Glu 235 [70] and Glu 245 [80] in factor IXa participate in binding to Ca 2ϩ (12). In the current study, this Ca 2ϩ binding site was abolished by replacing separately Glu 235 by lysine and Glu 245 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 Ca 2ϩ 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 Ca 2ϩ binding site increases the binding of factor VIIIa to factor IXa by ϳ15-fold and reactivity of the S1 site by ϳ3-fold. Our data also indicate that lysine at position 235 cannot fully mimic the function of this Ca 2ϩ binding site. However, compared with factor IX WT , ϳ30% 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 ϳ8-fold reduced affinity for factor VIIIa. This is consistent with our finding that compared with native factor Xa, autolysis loop-cleaved factor Xa has ϳ10-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 Ca 2ϩ binding loop may not directly participate in factor VIIIa binding (58). The spatial arrangement of Ca 2ϩ binding site, autolysis loop, and S1 site is depicted in Fig. 7. An examination of this figure would indicate that binding of Ca 2ϩ 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). FIG. 7. Schematic representation of the polypeptide backbone of the protease domain of human factor IXa depicting the Ca 2؉ binding site, autolysis loop, and the S1 binding site. The amino and carboxyl termini of the polypeptide are marked with N and C, respectively. The putative binding site for an extended substrate is shown as an arrow running from left to right. The branch extending from this arrow marks the position of P1 side chain, bound in the substrate binding pocket. The latter is marked with the stippled shadow and contains D359[189] (S1 site), which binds to the positive charge of the P1 substrate side chain. The side chains of the catalytic triad residues, D269[102], H221 [57], and S365 [195], and the residue D359[189] located in the substrate binding pocket are shown as thick, dark tubes. The peptide region depicting the autolysis loop is shown by stippled shadow, and the location of Arg 318 [150] residue in this loop is indicated. The Ca 2ϩ is marked with a circle, and the two glutamic acid ligands E235[70] and E245[80] are shown as stick models. The average root mean square deviation of C␣ atoms of residues 70 -80 between the modeled loop (plus Ca 2ϩ ) and the observed structure (in the absence of Ca 2ϩ ) of the porcine IXa loop is 0.78 Å. Occupancy of calcium at this site prevents proteolysis at the Arg 318 -Ser 319 [150 -151] peptide bond, increases catalytic efficiency, and potentiates factor VIIIa binding.