Prothrombinase Assembly and S1 Site Occupation Restore the Catalytic Activity of FXa Impaired by Mutation at the Sodium-binding Site*

Two loop segments (183–189 and 221–225) in the protease domain of factor Xa contribute to the formation of a Na+-binding site. Studies with factor Xa indicate that binding of a single Na+ ion to this site influences its activity by altering the S1 specificity site, and substitution of Tyr225 with Pro diminishes sensitivity to Na+. Using full-length factor XaY225P, the allosteric relationship between the Na+ site and other structural determinants in factor Xa and prothrombinase was investigated. Direct binding and kinetic measurements with probes that target the S1 specificity pocket indicate that assembly of the mutant in prothrombinase corrected the impaired binding of these probes observed with free factor XaY225P. This appears to result from the apparent allosteric linkage between the factor Va, S1, and Na+-binding sites, since binding of the cofactor to membrane-bound factor XaY225P enhances binding at the S1 site and vice versa. Additional studies revealed that the internal salt bridge (Ile16–Asp194) of factor XaY225P is partially destabilized, a process that is reversible upon occupation of the S1 site. The data establish that alterations at the factor Xa Na+-binding site shift the zymogen-protease equilibrium to a more zymogen-like state, and as a consequence binding of S1-directed probes and factor Va are adversely affected. Therefore, the zymogen-like characteristics of factor XaY225P have allowed for the apparent allosteric linkage between the S1, factor Va, and Na+ sites to become evident and has provided insight into the structural transitions which accompany the conversion of factor X to factor Xa.

Factor X (FX) 1 is a vitamin K-dependent two-chain glycoprotein that plays a central role in blood coagulation. This serine protease zymogen is a substrate for both the extrinsic (tissue factor/FVIIa) and intrinsic (FVIIIa/FIXa) tenase enzyme complexes, which cleave the Arg 15 -Ile 16 scissile bond 2 in FX, releasing a 52-amino acid activation peptide generating FXa (1). Factor Xa reversibly associates with its cofactor FVa on a membrane surface in the presence of Ca 2ϩ ions to form prothrombinase (2). While FXa catalyzes prothrombin cleavage, the macromolecular interactions that stabilize prothrombinase lead to a profound enhancement in catalytic efficiency (2), indicating that prothrombinase, not FXa, is the physiologically relevant enzyme leading to explosive thrombin generation.
Several studies have established that small ligands such as Na ϩ and Ca 2ϩ can allosterically modulate the protease domain function of FXa (3)(4)(5)(6)(7)(8)(9). Orthner and Kosow (4) demonstrated that FXa is capable of discriminating among monovalent and divalent cations, and both Na ϩ and Ca 2ϩ were found to stimulate the cleavage of oligopeptidyl substrates. The location of the Na ϩ -and Ca 2ϩ -binding sites in the protease domain has been identified through structural studies (10,11). The Ca 2ϩ site in the protease domain is essentially the same as that of trypsin (12), consisting of the Asp 70 -Glu 80 loop. Whereas Na ϩ occupies a similar site in FXa as compared with thrombin, the coordinating ligands to Na ϩ differ, employing two loop segments (183-189 and 221-225; residues that are part of the so-called activation domain of serine proteases), involving four carbonyl oxygen atoms and two water molecules (11).
It has become apparent over the past several years that metal ion binding sites in the catalytic domain of several serine proteases are allosterically linked to various structural determinants (8,(13)(14)(15)(16)(17). For example, two recent studies have shown that the Ca 2ϩ -and Na ϩ -binding sites in FXa appear to be allosterically linked (e.g. changes at one site influence the other and vice versa) (9,18). In addition to linkage between the Na ϩ -and Ca 2ϩ -binding sites, the S1 specificity site is also allosterically linked to the Na ϩ -binding site but not the Ca 2ϩ site (9). Similar results were also recently reported for the anticoagulant protein, activated protein C (19).
Whether Na ϩ binding to the catalytic domain of serine proteases has physiological relevance is open to debate, considering that the plasma concentration of Na ϩ is tightly regulated at ϳ150 mM. It is important to consider, however, that for certain enzymes such as thrombin, the dissociation constant for Na ϩ is close to 150 mM (13), indicating that small changes in the cation concentration or, more likely, changes in the availability of allosteric effector molecules that are linked to the Na ϩ -binding site can dramatically influence enzyme activity (20). The re-ported dissociation constant of Na ϩ for calcium-bound APC is ϳ23 mM, indicating that in vivo most of the APC would be bound to Na ϩ (19). The results with FXa and prothrombinase, however, are less clear. For example, the reported dissociation constant of Na ϩ for calcium-bound FXa ranges from 43 to 280 mM (9,18,21). Results with prothrombinase indicate that the activity of wild-type FXa incorporated into prothrombinase using various monovalent cations was minimally (Ͻ1.5-fold) affected by the presence of Na ϩ in the reaction buffer. However, this apparently was not the case using a Gla domainless form of FXa incorporated into prothrombinase or a Gla domainless variant (GDFXa Y225P ), since the rate of prethrombin-1 activation in the absence of Na ϩ was dramatically decreased (18). The Tyr 225 to Pro variant in GDFXa and thrombin was shown to be essentially insensitive to Na ϩ , and it behaves in a similar fashion to the wild-type enzymes assayed in the absence of Na ϩ (15,18,22). Reasons for the discrepancy between GDFXa Y225P and GDFXa incorporated into prothrombinase in the absence of Na ϩ compared with full-length FXa are currently not clear, but they could relate to unanticipated effects of various monovalent cations on other constituents of prothrombinase or to the nature of GDFXa compared with full-length FXa in the assembly and function within prothrombinase under these conditions.
In the current study, three unresolved questions with respect to the FXa Na ϩ -binding site were investigated: 1) does alteration of the Na ϩ -binding site affect prothrombinase complex assembly; 2) what is the molecular basis for the reduced affinity of active site-directed probes for FXa in the absence of Na ϩ ; and 3) is the FXa Na ϩ binding site allosterically linked to other structural determinants in the protease domain? These questions were approached using a full-length, fully ␥-carboxylated derivative of FXa, rFXa Y225P . This particular variant of FXa is a useful tool to study the Na ϩ -free state of the enzyme, since it has a markedly reduced sensitivity toward Na ϩ (as does thrombin Y225P (22)), thus allowing for assessment of prothrombinase assembly in the presence of physiologically relevant concentrations of this monovalent cation.
Mutagenesis, Expression, and Purification of Recombinant FX-The recombinant FX mutant Tyr 225 3 Pro (FXa Y225P ) was generated with the QuikChange site-directed mutagenesis kit (Stratagene) using two complementary oligonucleotides containing the desired mutation, the sense strand being 5Ј-CGTAAGGGGAAGCCCGGGATCTACACC-3Ј. The entire FX cDNA was sequenced in order to confirm the presence of the desired mutation and to ensure that there were no polymeraseinduced errors. Wild-type or mutant rFX in the mammalian expression plasmid pCMV4 were stably expressed in HEK 293 cells and purified as previously described (39,40).
Activation of FX to FXa and Purification on Benzamidine-Sepharose-Plasma-derived FX (pdFX) and rFX were activated using RVV X-CP and subsequently purified using benzamidine-Sepharose or Sephacryl S-200 as described (39,41). Following purification, plasmaderived and recombinant FXa molecules were precipitated, collected by centrifugation, dissolved in 50% glycerol, and stored at Ϫ20°C.
Characterization of Plasma-derived and rFX/Xa-Protein purity was assessed using precast NuPAGE 4 -12% Bis-Tris gels (Invitrogen) using the MES buffer system under reducing (50 mM dithiothreitol, final concentration) and nonreducing conditions followed by staining with Coomassie Brilliant Blue R-250. Gla analysis was carried out according to the modified method of Price (42) for alkaline hydrolysis. Separation of amino acids was accomplished using a DC-4A cation exchange column on a Waters LC-1 Plus high pressure liquid chromatograph (Milford, MA), and quantitation was done by postcolumn derivatization as described by Przysiecki (43). This analysis indicates that rwtFX and FX Y225P have essentially the full complement of Gla residues (10.5-10.8 mol of Gla/mol of FX) compared with pdFX (10.7 mol of Gla/mol of FX; theoretical ϭ 11 mol of Gla/mol of FX).
Determination of Kinetic Parameters for Peptidyl Substrate Hydrolysis-All kinetic measurements were performed in 20 mM Hepes, 0.15 M NaCl, 0.1% (w/v) polyethylene glycol 8000, 2 mM CaCl 2 , pH 7.5 (assay buffer) unless otherwise indicated. The kinetics of peptidyl substrate hydrolysis (SpecXa, S-2765, and S-2222) was measured using increasing concentrations of substrate and initiated with either FXa or FXa assembled into prothrombinase.
Inhibition of FXa and Prothrombinase by PAB-The ability of PAB to bind to FXa or prothrombinase was assessed by two independent methods. In the first method, the inhibitory constant (K i ) of PAB for FXa or prothrombinase was assessed assuming classical competitive inhibition by initial velocity measurements of SpecXa hydrolysis by either enzyme using increasing concentrations of substrate at different fixed concentrations of PAB as previously described (44). In the second method, the binding of PAB to FXa or prothrombinase was directly assessed by fluorescence measurements essentially as described (45).
Inhibition of FXa and Prothrombinase by rTAP and Determination of K i *-Overall dissociation constants for rTAP binding to FXa and prothrombinase were inferred from measurements of residual enzyme amidolytic activity following incubation of the protease with increasing concentrations of the inhibitor essentially as described (46).
Inhibition of FXa and Prothrombinase by Antithrombin III-The rate of inactivation of FXa or prothrombinase by human antithrombin III was measured under pseudo-first order rate conditions. For the inhibition of FXa, human antithrombin III (0.16, 0.32, and 0.64 M) was incubated with FXa (5 nM; FXa Y225P , 10 nM) in assay buffer for up to 75 min (12 time points). At the end of the time course, 50 l of SpecXa was added (100 M final concentration) to monitor residual enzyme activity. For the inhibition of prothrombinase, human antithrombin III (1.6, 3.2, and 6.4 M) was incubated with FXa (5 nM), FVa (50 nM), and PCPS (60 M) in assay buffer for up to 30 min (12 points). Residual enzyme activity was monitored as described above. In both assays, each measurement was made in duplicate.
Functional Binding Studies: The Effect of FVa on Peptidyl Substrate Cleavage by FXa Y225P -Reaction mixtures (200 l) containing SpecXa (10 -850 M) and PCPS (60 M) with different fixed concentrations of FVa (0.5, 1.0, 2.0, 3.0, 5.0, 10, and 20 nM) were prepared in the wells of a 96-well plate and allowed to incubate for 5 min at room temperature. Because the substrate stock solution was prepared in water, an appropriate volume of 10ϫ assay buffer solution adjusted to pH 7.75 was added to each mixture to ensure that the final pH and concentration of buffer solutes were invariant. The reaction was initiated with FXa Y225P (3 nM final concentration).
Carbamylation of Ile 16 by Reaction with NaNCO-Mixtures containing FXa Y225P or pdFXa (1-2 M) in assay buffer were incubated in the presence or absence of different fixed concentrations of PAB (50 -2200 M; 10 concentrations) and were reacted with either 0.2 or 0.6 M NaNCO. The final pH of the reaction mixture upon the addition of NaNCO was pH 7.45. At selected time intervals (5-300 min), 5 l of the reaction mixture was placed in 95 l of assay buffer, and the residual enzymatic activity was determined from initial steady state rates of SpecXa hydrolysis, or aliquots were mixed with 1.5 M hydroxylamine, pH 8.0, and then frozen at Ϫ80°C for N-terminal sequence analysis. Amino-terminal sequence analysis was performed in the laboratory of Dr. Jan Pohl (Emory University Microchemical Facility). Quenched protein samples at various time points (40 pmol) were mixed with rTAP (20 pmol; internal protein control) and spotted onto polyvinylidene difluoride membranes, and N-terminal sequence analysis was performed using an automated PE-Biosystems 491A pulsed liquid sequencer on-line with a PE-Biosystems 140S PTH analyzer. The peak area (in pmol) of PTH-Ile (first cycle of the heavy chain of FXa) and PHT-Tyr (first cycle of rTAP) were determined simultaneously. Control experiments included incubating the mutant protein as described above but in the presence of 0.2 M NaCl.
Data Analysis-Data were analyzed according to the indicated equations by nonlinear least squares regression analysis using the Marquardt algorithm (48). The qualities of the fits were assessed by the criteria described (49). Fitted parameters are reported Ϯ 95% confidence limits.
Determination of Steady State Kinetic Constants-Initial velocity measurements of peptidyl substrate or macromolecular substrate (prethrombin-2) hydrolysis by FXa or prothrombinase were analyzed by fitting the data to the Henri-Michaelis-Menten equation (50) to yield fitted values for K m and V max .
Inhibition Studies-Initial velocity measurements of SpecXa hydrolysis by FXa or prothrombinase using increasing concentrations of substrate at different fixed concentrations of PAB were analyzed according to the rate expression for linear competitive inhibition (50) to yield the fitted values for K m , V max , and K i . Initial velocity data obtained following incubation of increasing concentrations of rTAP with two fixed concentrations of FXa or prothrombinase were analyzed as described (45). Data derived from direct fluorescence binding measurements between PAB and FXa and PAB and prothrombinase were analyzed as described previously (45) taking into account the inner filter effect (51). The rate of inhibition of FXa or prothrombinase by antithrombin III was measured under pseudo-first order rate conditions, and the second order rate constant was calculated by dividing the pseudo-first order rate constant by the concentration of antithrombin III.
Equilibrium Constant for Prothrombinase Assembly-Dissociation constants and stoichiometries for the interaction between FXa and PCPS-bound FVa were obtained from the dependence of the initial rate on the concentrations of factor Va (52).
Global Analysis of Initial Velocity Data: Effect of FVa on Peptidyl Substrate Cleavage by FXa Y225P -The equilibrium dissociation constants for the binding of FVa to membrane-bound FXa in the absence (K d1 ) and presence of peptidyl substrate (K d2 ) as well as the substrate dissociation constant for FXa (K s1 ) and membrane-bound FXa saturated with FVa (K s2 ) were calculated from initial velocity measurements of SpecXa hydrolysis at different fixed concentrations of FVa according to the system of ordinary differential equations describing Scheme I and using the rapid equilibrium assumption. The entire data set was globally fit using the program Dynafit (53) to extract K s1 , k cat1 , K d1 , K s2 , K d2 , and k cat2 .
Carbamylation of FXa in the Presence of PAB: Global Analysis of First Order Inactivation Rate Constants-The first order rate inactivation constant for FXa by NaNCO in the absence of PAB (k 0 ) and in the presence of saturating concentrations of PAB (k 1 ) as well as the equilibrium dissociation constant (K d ) for PAB binding to FXa were calcu-lated by globally fitting activity data as a function of time to the ordinary differential equations describing the reaction mechanism shown in Scheme II using the program Dynafit (53) to extract K d , k 0 , and k 1 .

Expression and Purification of Recombinant Proteins-
rwtFX and FX Y225P were expressed in HEK 293 cells and purified to homogeneity. Following activation with RVV X-CP , each protein was applied to benzamidine-Sepharose. Whereas native FXa bound and was eluted from the column with 4 mM benzamidine, FXa Y225P did not bind to the resin (data not shown). These findings suggest that FXa Y225P has a reduced ability to bind benzamidine at the S1 specificity pocket. The mutant protein was subsequently purified by gel filtration. SDS-PAGE analysis of the purified zymogens (lanes 1-3) and purified proteases (lanes 4 and 5) before and after disulfide bond reduction are shown in Fig. 1.
Assessment of Binding at the Active Site-As detailed previously, the ability of GDFXa Y225P to cleave small peptidyl substrates is impaired, as is that of wild-type FXa in the absence of Na ϩ (9,18). Consistent with these observations, an increase in the K m for peptidyl substrates and a decrease in the k cat compared with wild-type FXa was observed for full-length FXa Y225P (Table I). These data indicate that the mutant protein has altered activity and suggest that the substrate binding cleft (S1-S3 site) is changed in some fashion. Similar results were obtained with rFXa Y225P bound to PCPS vesicles (data not shown; also see Fig. 3). In contrast to these results, the assembly of FXa Y225P into prothrombinase restored the K m for peptidyl substrates to that seen with wild-type proteins, whereas the k cat values were not significantly altered (Table I). At present, it is unclear why the k cat was reduced by a factor of 1.5-5 for both FXa Y225P and this mutant assembled in prothrombinase with the various peptidyl substrates. The data are consistent with the conclusion that prothrombinase complex assembly appears to correct defective binding of peptidyl substrates to the mutant protein.
Since changes in the K m for peptidyl substrates could arise from a variety of effects, the binding of a well defined S1 probe, PAB, to FXa was also investigated. Whereas the wild-type proteins bound PAB with essentially identical affinities, FXa Y225P had a ϳ10-fold reduced affinity for PAB, indicating the S1 specificity pocket of the mutant is altered (Table II). These data are consistent with the reduced affinity of PAB observed previously for wild-type FXa in the absence of Na ϩ (9). In contrast, assembly of the mutant with saturating concentrations of FVa and membranes restored PAB binding to that seen with wild-type FXa assembled in prothrombinase (Table II). Similar results were also obtained with both rTAP and antithrombin III. Together, these results indicate that prothrombinase complex assembly, in the context of an altered Na ϩ -binding site, appears to modify the affinity of active sitedirected probes.
Cleavage of the Macromolecular Substrate Prethrombin-2-Assuming that the substrate exosite is not changed in any fashion, the above data would suggest that rates of cleavage of the macromolecular substrate, prethrombin-2, by FXa Y225P assembled in prothrombinase would be similar to wild-type prothrombinase. Initial velocity measurements with saturating concentrations of FVa and membranes indicate that this is indeed the case (Fig. 2). Fitting the data to the Michaelis-Menten equation indicates that the mutant assembled in prothrombinase has similar kinetic parameters for prethrombin-2 . Consistent with results obtained with peptidyl substrates, the k cat for macromolecular substrate cleavage by the mutant assembled in prothrombinase was mod-erately reduced. These data are in contrast to those using GDFXa in the absence of Na ϩ or GDFXa Y225P , where the rate of prethrombin-1 activation was dramatically impaired (18).
Assessment of Thermodynamic Linkage-The data indicate that the binding of FVa to membrane-bound FXa Y225P appears to enhance binding of molecules that target the S1 site (Tables  I and II). This implies that a thermodynamic linkage exists between the FVa, S1, and Na ϩ -binding sites. Experiments were designed such that all relevant equilibrium binding constants depicted in Scheme I could be simultaneously evaluated.
In this model, FXa Y225P and FVa are membrane-bound, and S represents SpecXa. A signal to monitor the various binding interactions was provided by the difference in chromogenic activity between FXa Y225P and the mutant saturated with FVa and membranes (Table I). Initial velocity measurements of SpecXa cleavage by membrane-bound FXa Y225P at different fixed concentrations of FVa (Fig. 3) were made, followed by global analysis of all relevant equations describing the binding interactions depicted in Scheme I. The data indicate that membrane-bound FXa Y225P binds with a decreased affinity to FVa (K d1 ϭ 26.0 Ϯ 5.4 nM) and peptidyl substrates (K s1 ϭ 695 Ϯ 49 M) compared with wild-type FXa; however, occupation of the S1 site restored FVa binding (K d2 ϭ 2.1 Ϯ 0.26 nM), and saturating membrane-bound FXa Y225P with FVa restored peptidyl substrate binding (K s2 Ϸ 56 M). These data imply that there is allosteric linkage between the Na ϩ , FVa, and S1 sites (i.e. the binding of FVa to FXa Y225P enhances the binding of molecules that target the S1 site and vice versa). The rates of catalysis of FXa Y225P in the absence of FVa (k cat1 ϭ 43 Ϯ 1.6 s Ϫ1 ) or in the presence of saturating concentrations of FVa (k cat2 ϭ 41 Ϯ 0.7 s Ϫ1 ) are essentially the same, indicating that   Additional experiments were also performed aimed at assessing the FXa Y225P -FVa interaction using the macromolecular substrate, prethrombin-2. Measurements of the conversion of a single fixed concentration of prethrombin-2 to thrombin were conducted using increasing concentrations of FVa at a single, fixed concentration of FXa and PCPS (Fig. 4). Because of the experimental conditions chosen and the very high K m of membrane-bound FXa for prethrombin-2 (47), the equilibrium dissociation constant measured in the following experiment is expected to be analogous to K d1 depicted in Scheme I. The inferred equilibrium dissociation constant for membranebound FVa binding to FXa Y225P (K d1 ϭ 17.3 Ϯ 1.1 nM) was ϳ3-fold greater than that observed with the wild-type proteins (pdFXa, K d1 ϭ 6.2 Ϯ 0.4 nM; rwtFXa, K d1 ϭ 6.3 Ϯ 0.5 nM, assuming a stoichiometry of 1). These results are in reasonable agreement with K d1 obtained using peptidyl substrates.
Modification of the N-terminal Ile 16 -A possible explanation for the apparent altered S1-and FVa-binding sites on FXa Y225P is that mutation at the Na ϩ -binding site destabilizes the salt bridge between Ile 16 and Asp 194 , thereby favoring a zymogenlike conformation. In order to test this idea, FXa Y225P was reacted with NaNCO, which preferentially modifies the N terminus on proteins and to a lesser extent ␣-amino groups on lysines (54,55). The results indicate that the rate of inactivation by NaNCO of FXa Y225P was 6-fold faster compared with wild-type FXa (second-order rate constants, 45.0 mM Ϫ1 min Ϫ1 versus 7.50 mM Ϫ1 min Ϫ1 ; Fig. 5, closed symbols). Whereas cyanate has been used to investigate the susceptibility of the N terminus of FVIIa (Ile 16 ) to be chemically modified by activity and sequencing methods (56,57), it is possible that the reduction in activity of FXa Y225P may be unrelated to modification of the N terminus. To address this issue, FXa Y225P was incubated with cyanate, and the rate of disappearance of PTH-Ile 16 derived from the N terminus of the heavy chain was monitored over time. To control for internal inconsistencies in the sequencing set-up and reaction, an internal protein standard (rTAP) was included, and the values of PTH-Ile 16 (N terminus of heavy chain of FXa) to PTH-Tyr (N terminus of rTAP) was expressed as a ratio. Consistent with activity measurements, the rate of carbamylation of FXa Y225P was 6.8-fold faster compared with wild-type FXa using the sequencing method (second-order rate constants, 41.0 mM Ϫ1 min Ϫ1 versus 6.01 mM Ϫ1 min Ϫ1 ; Fig. 5, open symbols), indicating that the loss in activity correlates very well with modification of the N-terminal Ile 16 .
It is well documented for trypsin that formation of the Ile 16 -Asp 194 internal salt-bridge is allosterically linked to the S1 specificity site (58). Since FXa Y225P appears to have a partially destabilized N-terminal insertion, occupation of the S1 site of FXa Y225P should stabilize the Ile 16  was reacted with NaNCO over an extended time course (Fig.  6A). Global analysis of all data sets to equations describing all relevant interactions depicted in Scheme II (where P represents PAB) permitted the determination of the equilibrium dissociation constant (K d ) for PAB-FXa Y225P and the second order rate constants for the modification of FXa Y225P by NaNCO in the absence (k 0 ) and presence (k 1 ) of saturating PAB (*FXa Y225P indicates that it has been covalently modified at the N terminus with NaNCO). Inspection of the observed rate constants (k obs , derived from Fig. 6A) as a function of the PAB concentration (Fig. 6B) indicates that the interaction of FXa Y225P with PAB is relatively weak (K d ϭ 726 Ϯ 53 M), results that are in agreement with direct binding fluorescence and kinetic measurements (Table II). Additionally, saturation of FXa Y225P by PAB afforded complete protection of the enzyme from carbamylation, as evident by an extremely small second order rate constant (k 1 ϭ Ͻ1 ϫ 10 Ϫ5 mM Ϫ1 min Ϫ1 ) compared with FXa Y225P in the absence of PAB (k 0 ϭ 39.0 Ϯ 1.28 mM Ϫ1 min Ϫ1 ). These data indicate that occupation of the S1 site of the variant results in stabilization of the N-terminal insertion and resistance of the enzyme to carbamylation. These results also support the idea that the reduction in activity observed with FXa Y225P in the presence of cyanate is related to modification of the N terminus, since the loss in activity could be reversed upon occupation of the S1 site (Fig. 6). DISCUSSION The zymogen to protease transition in the trypsin-like serine protease family of proteins is initiated following proteolytic liberation of a highly conserved N terminus and removal of an activation peptide. As shown for the trypsinogen/trypsin system, a new free ␣-amino group (Ile 16 ) forms a salt bridge with the Asp 194 carboxylate group following activation that either results in or is associated with a conformational change in the so-called "activation domain" (residues comprising positions 16 -19, 142-152, 184 -194, and 216 -223) (59,60). Since N-terminal insertion is energetically linked to the maturation of the activation domain, alteration of any of these structural elements including the S1 specificity pocket should positively or negatively influence the others.
The results of the current study indicate that alteration of the FXa Na ϩ -binding site by changing residue 225 from a Tyr to a Pro results in the transformation of this serine protease from a protease-like state to a zymogen-like state, providing a reasonable explanation for the observations of reduced activity (this study and Ref. 18). This shift in favor to the zymogen-like state results from destabilization of the N terminus (Ile 16 ) of FXa Y225P and adversely affects FVa binding and binding of substrates to the S1 specificity pocket. Consistent with thermodynamic principles, protection of the N-terminal Ile 16 from modification was achieved by saturation of the S1 specificity site with PAB, thereby favoring the active or protease-like conformation (Fig. 6). Additionally, saturation of membranebound FXa Y225P with FVa also assisted the variant to adopt a more protease-like state, suggesting that the FVa and Na ϩbinding sites are allosterically linked (Figs. 2-4). These findings are consistent with the well documented zymogen to protease transition in trypsinogen/trypsin (61) (i.e. the zymogen state and the protease state exist in an equilibrium that can be shifted depending on environmental conditions (e.g. various ligands)).
At present, it is not clear whether the activity of free FXa is modulated by Na ϩ in vivo, since there are conflicting reports on the dissociation constant of Na ϩ for calcium-bound FXa (9,18,21). Additionally, it is unlikely that the levels of Na ϩ would change significantly enough to influence the activity of FXa, since the Na ϩ concentration is tightly regulated. However, if Na ϩ -bound and Na ϩ -free forms of FXa are equally populated in vivo, then these forms of the enzyme would be sensitive to changes in the availability of allosteric effector molecules that are linked to the Na ϩ -binding site. These changes could influence a variety of reactions in which FXa participates (e.g. activation of FV, FVII, cellular receptors, and interactions with TFPI and ATIII). With respect to prothrombin activation, the current data indicate that once membrane-bound FXa Y225P is saturated with FVa, the rate of macromolecular substrate cleavage is similar to native prothrombinase, calling into question the physiological significance of Na ϩ in influencing this reaction. However, as the catalytic activity of prothrombinase appears to be insensitive to changes at the FXa Na ϩ -binding site (Fig. 2), the binding of membrane-bound rFXa Y225P to FVa is adversely affected (Figs. 3 and 4). Thus, whereas the overall rate of thrombin generation by prothrombinase may be influenced by changes at the FXa Na ϩ -binding site, the degree of this change would depend upon the amount of FVa available and possibly other factors that could influence the equilibrium between the two forms of the enzyme.
The FXa Na ϩ and S1 specificity sites are part of the so-called activation domain and would presumably be disordered in the zymogen and become ordered upon N-terminal insertion. A recently described model of the zymogen FX suggests that this may be the case (62). Comparison of the zymogen model with the active enzyme reveals that residues making up the calcium (Asp 70 -Glu 80 ), sodium (Ala 183 -Asp 194 ; Gly 219 -Gly 226 ) and autolysis loops (Thr 144 -Arg 150 ) undergo major changes in their backbone positions upon the zymogen to protease transition. These results suggest that these regions in FX are at least influenced by this transition. Since it is already well documented, at least for trypsinogen/trypsin, that the S1 specificity site and formation of Ile 16 -Asp 194 are allosterically linked, it is reasonable to hypothesize that other elements of the activation domain are also linked to the zymogen to protease transition. By definition, this means that alterations at the Na ϩ binding site, for example, will influence the formation of the Ile 16 -Asp 194 salt bridge, consistent with observations of the current study (Figs. 5 and 6).
The idea that changes at the Na ϩ -binding site can influence ion pair formation is in agreement with recent findings indicating that an increase in the pH results in a decrease in the Na ϩ affinity for several serine proteases, including FXa (63). It was hypothesized that deprotonation of an ion pair (pK a Ϸ 9.2; assumed, but not proven to be Ile 16 -Asp 194 ) by changes in pH will result in a break of the salt bridge and yield an internal disruption of the Na ϩ -binding site. Whereas it is clear that large changes in the pH capable of disrupting ion pair formation cannot be a mechanism by which the FXa Na ϩ -binding site is modulated in vivo, the study nevertheless provides evidence that formation of the Ile 16 -Asp 194 salt bridge appears crucial in the stabilization of the architecture of the Na ϩ -binding site.
Thus, a key aspect of the zymogen to protease transition for serine proteases that are sensitive to Na ϩ may be the formation of a functional Na ϩ -binding site. This model would predict that the Na ϩ site of the zymogen binds the cation with a very weak affinity; high affinity binding would follow activation of the zymogen and ordering of the Na ϩ site. Support for this idea comes from the current study as well as other studies. For example, inspection of the crystal structure of the homologous zymogen prethrombin-2 in complex with hirugen reveals that, in contrast to thrombin, there are several key movements of loop segments and critical groups in the vicinity of the Na ϩ site that would prevent Na ϩ binding to the zymogen, consistent with its apparent absence in the prethrombin-2 structure (11,64). The same also appears to be true with the recently described FVII zymogen structure, which demonstrates that loop regions making up the Na ϩ -binding site (184 -193 and 215-224) are shifted up to 12 Å compared with the tissue factor/ FVIIa structure (65). Additionally, using a direct fluorescence binding assay, De Cristofaro et al. (16) were not able to detect Na ϩ binding to protein C, suggesting that the Na ϩ site in this zymogen is exposed only after its activation.
In addition to disruption of the N-terminal insertion and S1 specificity pocket, mutation at the Na ϩ site also results in decreased cofactor binding, suggesting that the FVa binding site is allosterically linked to elements in the activation domain. The data indicate that occupation of the S1 specificity site of membrane-bound FXa Y225P enhances binding at the FVa binding site and vice versa, explaining well why these two enzymes (FXa Y225P and mutant prothrombinase) have different reactivities toward active site-directed probes (Tables I and  II; see Scheme I). In contrast to these observations, the binding of membrane-bound FVa to wild-type FXa is accompanied by a minor change in active site function (66,67). A possible explanation for this is that linkage between the FVa and S1 binding sites is obscured in wild-type FXa, because the protease-like conformation is highly favored. In contrast, FXa Y225P , with an altered Na ϩ -binding site, exists in a zymogen-like conformation, thereby making linkage between the FVa and S1 sites more discernible. This explanation would appear consistent with the observation that the zymogen FX does not detectably bind FVa or S1-directed probes.
A possible alternative explanation to allosteric linkage is that steric factors are influencing the affinity of FVa for membrane-bound FXa Y225P . For example, disruption of the Ile 16 -Asp 194 salt bridge in FXa Y225P could result in movement of the N-terminal segment such that it sterically hinders the binding of FVa to membrane-bound FXa Y225P . Implicit in this argument is that the enzyme exists in two states, one in which the N terminus blocks the FVa binding site and one in which the site is accessible. Any molecule that facilitates the formation of the Ile 16 -Asp 194 salt bridge (e.g. binding of S1 probes) would shift the equilibrium between the two states, removing this steric factor and restoring FVa binding. At present, this scenario cannot be formally ruled out, and future studies are needed to distinguish between the two possibilities.
Examination of the FXa crystal structure in the vicinity of the putative FVa binding helix (68), Na ϩ site, and S1 specificity site may provide, however, some insight into how these elements could be allosterically linked at the structural level. As previously noted (9,11), the Na ϩ site in FXa is linked to the S1 specificity site (specifically Asp 189 ) via two water molecules. The FVa binding helix (residues 162-169) is connected to the Na ϩ -binding site through van der Waals' contacts between Tyr 225 and Val 163 and between the OH group of Tyr 225 and (via two water molecules) Ser 167 . When the FXa Tyr 225 to Pro mutation was modeled (9) the results indicate that its carbonyl oxygen atom would point away from the Na ϩ -binding site and that its side chain would be unable to make contacts with the FVa binding helix. The data of the current study are consistent with this model. Future experiments aimed at preventing internal salt bridge formation should provide further evidence that the FVa, S1, and Na ϩ -binding sites consist of a structural network allosterically linked to the zymogen to protease transition.