Expression of Allosteric Linkage between the Sodium Ion Binding Site and Exosite I of Thrombin during Prothrombin Activation*

The specificity of thrombin for procoagulant and anticoagulant substrates is regulated allosterically by Na+. Ordered cleavage of prothrombin (ProT) at Arg320 by the prothrombinase complex generates proteolytically active, meizothrombin (MzT), followed by cleavage at Arg271 to produce thrombin and fragment 1.2. The alternative pathway of initial cleavage at Arg271 produces the inactive zymogen form, the prethrombin 2 (Pre 2)·fragment 1.2 complex, which is cleaved subsequently at Arg320. Cleavage at Arg320 of ProT or prethrombin 1 (Pre 1) activates the catalytic site and the precursor form of exosite I (proexosite I). To determine the pathway of expression of Na+-(pro)exosite I linkage during ProT activation, the effects of Na+ on the affinity of fluorescein-labeled hirudin-(54–65) ([5F]Hir-(54–65)(SO–3)) for the zymogens, ProT, Pre 1, and Pre 2, and for the proteinases, MzT and MzT-desfragment 1 (MzT(–F1)) were quantitated. The zymogens showed no significant linkage between proexosite I and Na+, whereas cleavage at Arg320 caused the affinities of MzT and MzT(–F1) for [5F]Hir-(54–65)(SO–3) to be enhanced by Na+ 8- to 10-fold and 5- to 6-fold, respectively. MzT and MzT(–F1) showed kinetically different mechanisms of Na+ enhancement of chromogenic substrate hydrolysis. The results demonstrate for the first time that MzT is regulated allosterically by Na+. The results suggest that the distinctive procoagulant substrate specificity of MzT, in activating factor V and factor VIII on membranes, and the anticoagulant, membrane-modulated activation of protein C by MzT bound to thrombomodulin are regulated by Na+-induced allosteric transition. Further, the Na+ enhancement in MzT activity and exosite I affinity may function in directing the sequential ProT activation pathway by accelerating thrombin formation from the MzT fast form.

procoagulant activator of factor V and factor VIII in reactions that are dependent on phospholipid membranes (21,22). MzT bound to thrombomodulin activates the anticoagulant, protein C, at a rate equal to or greater than thrombin. Protein C activation by thrombomodulin-bound MzT is also stimulated by membranes (19,20).
Exosite I interactions are allosterically linked to binding of a single sodium ion to thrombin (23,24). Binding of Na ϩ converts thrombin from a "slow" to a "fast" form (25). The fast form has higher catalytic efficiency (k cat /K m ) for the procoagulant substrates, fibrinogen and protease-activated receptor-1, greater reactivity toward antithrombin, and increased amidolytic chromogenic substrate activity, notably toward D-Phe-Pro-Arg-pNA (15,24). Alanine substitution mutagenesis and analysis of high resolution crystal structures of the fast and slow forms in their free states have identified four structural features critical to the allosteric transition to the fast form (23, 26 -28): (a) alignment of residues stabilizing the Na ϩ binding site; (b) a change in orientation of Asp 189 in the primary substrate specificity site; (c) a shift in the Glu 192 side chain that links a network of water molecules to Ser 195 ; and (d) a change in Ser 195 orientation, all of which optimize catalysis. Comparison of these structures with those of thrombin, either inactivated with the transition state analog, D-Phe-Pro-Arg-CH 2 Cl (FPR-CH 2 Cl), or in which exosite I is occupied by the COOH-terminal hirudin peptide, hirugen (hirudin-(53-64)), shows that thrombin adopts the fast conformation in these complexes (27,29). Recent structural studies (30,31), consistent with kinetic studies of Na ϩ binding (31)(32)(33), suggest that the catalytically active Na ϩ -free slow form is in equilibrium with a low level of an inactive thrombin in which the catalytic site and Na ϩ site are occluded.
Previous studies of exosite I expression on ProT activation intermediates characterized a precursor form of exosite I on ProT (proexosite I), with a ϳ100-fold lower affinity for the model exosite I-specific ligand Hir-(54 -65)(SO 3 Ϫ ) (34 -36). Removal of fragment 1 from ProT to form prethrombin 1 (Pre 1) is accompanied by a ϳ7-fold increase in proexosite I affinity (35). The further 11-to 20-fold increase in affinity, representing full exosite I expression, occurs in concert with expression of catalytic activity during formation of MzT and thrombin (35,36). Quantitative analysis of binding of the ProT activation fragments 2 and 1.2 to exosite II on Pre 2 and thrombin indicates that this exosite becomes unmasked upon cleavage at Arg 271 and subsequent dissociation of the fragments (35,36). How linkage between the Na ϩ site and (pro)exosite I is expressed during ProT activation has not been investigated.
In the present studies, fluorescein-labeled Hir-(54 -65)(SO 3 Ϫ ) ([5F]Hir-(54 -65)(SO 3 Ϫ )) was used as a (pro)exosite I-specific probe to examine the linkage between Na ϩ binding and (pro)exosite I affinity (34 -37). The goal was to determine if Na ϩ regulates the affinity of proexosite I for ProT, Pre 1, and Pre 2 zymogen forms, and the catalytically active intermediates of ProT and Pre 1 activation, MzT and meizothrombin desfragment 1 (MzT(ϪF1)), respectively. The results demonstrate no significant effect of Na ϩ on ProT, Pre 1, or Pre 2 proexosite I affinity for [5F]Hir-(54 -65)(SO 3 Ϫ ). Allosteric linkage between Na ϩ binding and exosite I affinity occurred concomitantly with activation to MzT and MzT(ϪF1). The Na ϩ enhancement of the rate of chromogenic substrate hydrolysis for MzT compared with MzT(ϪF1) and thrombin, however, followed a kinetically different mechanism. The results demonstrate for the first time that MzT and MzT(ϪF1) are allosterically regulated by Na ϩ -exosite I interactions, suggesting that the unique substrate specificity of MzT may be Na ϩ -regulated.

EXPERIMENTAL PROCEDURES
Protein and Peptide Purification and Characterization-Human ProT, factor Xa, and thrombin were purified and characterized as described previously (38,39). Pre 1 was purified from products of ProT cleavage by thrombin (38). FPR-MzT(ϪF1) was prepared by activation of 20 M Pre 1 with 2 units/ml (based on the manufacturer's specifications) of purified ecarin from Echis carinatus venom (Sigma) in the presence of a 10-fold excess of FPR-CH 2 Cl (40). FPR-MzT was prepared from ProT at the same reactant concentrations followed by chromatography on Resource Q (Amersham Biosciences) in 20 mM sodium citrate buffer, pH 6.0. The column was washed with 10 ml of equilibration buffer and eluted with a 50-ml NaCl gradient in the same buffer up to 0.5 M to remove ecarin. Pre 2 was purified from factor Xa-ProT activation reactions as described (36). The R155A substitution mutant of ProT (ProT R155A ) was prepared as described (21). Protein concentrations were determined by absorbance at 280 nm using the following absorption coefficients ((mg/ml) Ϫ1 cm Ϫ1 ) and molecular weights (M r ) ( Ϫ ) was prepared as described previously (37).
Fluorescence Titrations-Binding of [5F]Hir-(54 -65)(SO 3 Ϫ ) was measured by titrating the labeled peptide with the protein of interest and monitoring the change in fluorescence at 520 nm with excitation at 491 nm (4 or 8 nm band passes), using acrylic cuvettes coated with polyethylene glycol (PEG) 20,000. Fluorescence changes expressed as (F obs Ϫ F o )/F o ϭ ⌬F/F o , as a function of total protein concentration were fit by the quadratic binding equation to obtain the maximum fluorescence change ((F max Ϫ F o )/F o ϭ ⌬F max /F o ), and the dissociation constant (K D ), with the stoichiometric factor fixed at 1 (37). Fluorescence measurements were corrected for background signal by subtraction of a buffer blank, and for dilution, which was not more than 10%.
Because of the relatively small differences in fluorescence between titrations in NaCl and ChCl, it was not possible to determine accurately the dissociation constant for Na ϩ binding from the Na ϩ concentration dependence at a fixed thrombin concentration. Titrations in 110 mM NaCl and 110 mM ChCl were fit independently by the quadratic binding equation to obtain the maximum fluorescence change ⌬F max, P⅐H /F o and K H, Ch from the titration in ChCl, and ⌬F max, P⅐Na⅐H /F o and K H, Na from the titration in the presence of NaCl.
Kinetic and End Point Fluorescence Titrations-To quantitate the effects of Na ϩ on binding of [5F]Hir-(54 -65)(SO 3 Ϫ ) to exosite I on active MzT(ϪF1) and MzT R155A , it was necessary to employ a kinetic approach to separate the ecarin-catalyzed formation of MzT(ϪF1) and MzT R155A from slower autocatalytic cleavages that process the products further (43). Experiments were performed for Pre 1 by measuring F o for 20 nM [5F]Hir-(54 -65)(SO 3 Ϫ ), the instantaneous decrease in fluorescence on addition of Pre 1 due to peptide binding to (pro)exosite I, and the time dependence of the subsequent fluorescence decrease initiated by addition of 2 units/ml of ecarin. The ecarin-initiated fluorescence time courses were fit by the sum of a single exponential decay and a straight line to determine the fluorescence end point corrected for any slow linear drift, and the results were expressed as ⌬F/F o . For analysis of the progress curves, the integrated Michaelis-Menten equation was used to represent the ecarin (E)-catalyzed conversion of Pre 1 or ProT R155A (C) into MzT(ϪF1) or MzT R155A (P), respectively (Scheme 2). Comparison of this analysis with the integrated rate equation including competitive product inhibition showed no evidence for product inhibition with MzT(ϪF1), whereas including product inhibition for MzT R155A improved the fit modestly.
Under the experimental conditions used, the fluorescence change accompanying peptide (H) binding is not a linear function of the concentration of P or C. Moreover, these species compete for the peptide, and binding is accompanied by unequal fluorescence changes for the P⅐H and C⅐H complexes (Scheme 3). To analyze the time courses, the cubic competitive binding equation derived previously for Scheme 3 (44) in combination with the integrated Michaelis-Menten equation was used for nonlinear least-squares fitting. Fluorescence versus time data, collected as a function of the concentration of C (Pre 1) in NaCl-or ChCl-containing buffers, were analyzed with the parameters determined independently for H ([5F]Hir-(54 -65)(SO 3 Ϫ )) binding to C (Pre 1) fixed. The fitted parameters were the maximum fluorescence change for the MzT(ϪF1)⅐[5F]Hir-(54 -65)(SO 3 Ϫ ) complex (⌬F max, P⅐H /F o ), K P⅐H , K m , and V max . The fluorescence and binding parameters determined from this analysis were compared with those obtained by fitting of the quadratic binding equation to the reaction end points as a function of C (Pre 1) concentration determined in a model-independent manner by the exponential fit described above. The same approaches were used in Ϫ ) in these reactions did not influence the results, because Յ6% of the total Pre 1 or ProT R155A was bound at any time, over the protein concentration range studied.
Pre 1 Activation Assay-Assays specific for MzT(ϪF1) were performed by modifications of the published methods (2,5). This approach enables differential measurement of thrombin and MzT(ϪF1) based on the much faster rate of thrombin inactivation by antithrombin⅐heparin complex compared with MzT(ϪF1). Aliquots of activation reactions of Pre 1 in the buffers containing NaCl or ChCl initiated at 25°C by addition of ecarin were quenched by dilution to 10 nM Pre 1 in 250 mM Hepes, 150 mM NaCl, 75 mM EDTA, 1 mg/ml PEG 8000, pH 7.70 (for NaCl reactions) or 250 mM Hepes, 590 mM NaCl, 75 mM EDTA, 1 mg/ml PEG 8000, pH 7.65 (for ChCl reactions) containing 50 nM antithrombin and 18 g/ml heparin (Sigma) for exactly 30 s. An aliquot of the quenched reaction was diluted 10-fold in 50 mM Hepes, 125 mM NaCl, 1 mM EDTA, 1 mg/ml PEG 8000, pH 7.4, containing 100 M D-Phe-Pip-Arg-pNA, and the initial rate of chromogenic substrate hydrolysis was measured from the linear increase in absorbance at 405 nm. Thrombin was confirmed to be Ͼ99.9% inactivated during the 30-s quenching time. To correct for the differences in kinetic parameters of thrombin and MzT(ϪF1) and the small loss of MzT(ϪF1) during quenching, the rates obtained for Pre 1 fully activated by incubation with 2 units/ml ecarin for 60 min were measured as a function of time after quenching, relative to an equivalent concentration of thrombin. Extrapolation of the slow, linear decrease in MzT(ϪF1) activity with time due to antithrombin⅐heparin inactivation and possible autocatalytic thrombin formation to zero time was taken to represent the total MzT(ϪF1) concentration. The extrapolated chromogenic substrate activity of MzT(ϪF1) relative to thrombin was determined to be 1.21 and 1.23 in the NaCl and ChCl experiments, respectively, and the correction for loss of MzT(ϪF1) activity was 10 and 6% in the NaCl and ChCl experiments, respectively.
Na ϩ Dependence of Chromogenic Substrate Hydrolysis-Initial rates of hydrolysis of 200 M (saturating) D-Phe-Pip-Arg-pNA were measured at 25°C as a function of Na ϩ concentration using mixtures of buffers containing 110 mM NaCl or ChCl. These experiments measured the Na ϩ dependence of k cat , reflecting Na ϩ binding to the enzyme⅐substrate complex. The initial rates at zero and saturating Na ϩ concentration, and the apparent K D for Na ϩ binding were obtained by least-squares SCHEME 1 SCHEME 2 SCHEME 3

Expression of Thrombin Sodium Site Linkage
fitting of the hyperbolic titrations. To estimate the affinities of Na ϩ for the free enzymes, k cat , K m , and K i for product were determined by full progress curve analysis of D-Phe-Pip-Arg-pNA hydrolysis as a function of Na ϩ concentration (45). Five progress curves at 5-15 M substrate were fit simultaneously to obtain the kinetic parameters. The K D for Na ϩ binding to thrombin and MzT(ϪF1) was obtained by fitting the hyperbolic dependence of k cat /K m on Na ϩ concentration (45). In the case of MzT R155A for which this was not possible, an estimate of the K D for Na ϩ binding to the free enzyme was obtained from the dependence of the observed K m (K m,obs ) on Na ϩ concentration, where K m was assumed to be equivalent to the dissociation constant for substrate binding. Equation 1, described by Segel (46) for the rapid equilibrium, nonessential activator mechanism, was fit to the data to obtain K m , K Na , and ␣, which represents the linkage factor by which K Na is changed by substrate binding.

RESULTS
Na ϩ -exosite I Linkage for ProT Zymogen Forms-To assess the linkage between Na ϩ binding and proexosite I, the effect of Na ϩ on the affinity of proexosite I for the 5-fluorescein-labeled Tyr 63 -sulfated hirudin-(54 -65)-peptide ([5F]Hir-(54 -65) (SO 3 Ϫ )) was determined in fluorescence titrations in buffers containing 110 mM NaCl or 110 mM ChCl at constant 0.17 M ionic strength. As shown in Fig. 1, ProT, Pre 1, and Pre 2 bound the peptide with virtually the same affinity in NaCl and ChCl. Although the titrations consistently showed a small increase in affinity by Na ϩ (1.3-to 1.6-fold), this was within the experimental error of the binding parameters (Table 1).
Na ϩ -exosite I Linkage for Thrombin and FPR-Thrombin-As expected, thrombin affinity for [5F]Hir-(54 -65)(SO 3 Ϫ ) increased 5.6-fold from dissociation constants of 270 Ϯ 20 nM in the absence of Na ϩ to 48 Ϯ 4 nM in NaCl ( Fig. 2 and Table 1). Inactivation of thrombin with FPR-CH 2 Cl virtually obliterated the effect of Na ϩ on affinity and normalized the dissociation constants in the presence and absence of Na ϩ to values indistinguishable from those of the fast form of native thrombin ( Fig.  2 and Table 1).
Analysis of Na ϩ -exosite I Linkage for MzT(ϪF1)-MzT and MzT(ϪF1) are unstable in their active forms as a result of autocatalytic cleavage (43). To characterize the effect of Na ϩ on active MzT(ϪF1), a kinetic approach was used to quantitate [5F]Hir-(54 -65)(SO 3 Ϫ ) binding to MzT(ϪF1) during its generation by ecarin activation of Pre 1 (see "Experimental Procedures"). The integrated Michaelis-Menten equation was used to calculate the full time course of ecarin-catalyzed MzT(ϪF1) formation. The fluorescence change accompanying peptide binding is not a linear function of MzT(ϪF1) or Pre 1 concentration, and it was also necessary to account for the fluorescence changes due to competitive [5F]Hir-(54 -65)(SO 3 Ϫ ) binding to Pre 1, because its concentration decreased during the reaction. A binding equation for a model in which two ligands (Pre 1 and MzT(ϪF1)) bind competitively to a common fluorescent acceptor ([5F]Hir-(54 -65)(SO 3 Ϫ )) with unequal fluo-rescence changes (44) was used to convert the fluorescence data into the concentrations of MzT(ϪF1) and Pre 1 with time.   Table 1. Titrations were performed and analyzed as described under "Experimental Procedures." in fluorescence corresponds to binding of the peptide to Pre 1 and the subsequent time-dependent decrease to generation of a product with a higher affinity for the peptide. To validate the experiments, a chromogenic substrate assay, which enables quantitation of MzT(ϪF1) formation in the presence of thrombin, was used to follow MzT(ϪF1) formation. Progress curves of MzT(ϪF1) generation in the absence and presence of Na ϩ were in good agreement with the calculated progress curves (Fig. 3).
To confirm that the product of Pre 1 activation was MzT(ϪF1), SDS-gel time courses were done under the same experimental conditions in the absence and presence of Na ϩ (Fig. 4). The reduced samples showed the disappearance of Pre 1 and formation of the MzT(ϪF1) B-chain and fragment 2-Achain. The nonreduced samples showed a single band representing Pre 1 and co-migrating MzT(ϪF1) for at least 20 min, corresponding to the longest activation reactions performed (Fig. 4). Traces of thrombin only appeared after 40-min incu-  Table 1. Titrations were performed and analyzed as described under "Experimental Procedures."  Table 1, and the calculated progress curve of MzT(ϪF1) formation using the same parameters. Fluorescence and activity measurements and data analysis were performed as described under "Experimental Procedures." bation in the reaction mixture containing Na ϩ as can be seen by the faint band at M r ϳ 36,000 in Fig. 4A. The results demonstrated that the observed fluorescence changes were well described by [5F]Hir-(54 -65)(SO 3 Ϫ ) binding with increased affinity for MzT(ϪF1) and that the results could be analyzed quantitatively.
Regulation of MzT(ϪF1) Exosite I Affinity by Na ϩ -Fluorescence-monitored time courses at varied Pre 1 concentration in NaCl or ChCl (Fig. 5, A and C) were fit simultaneously as described under "Experimental Procedures." The independently determined dissociation constant and maximum fluorescence change for [5F]Hir-(54 -65)(SO 3 Ϫ ) binding to Pre 1 (see Fig. 1) were fixed parameters, whereas K m and V max for the ecarin-catalyzed reaction, the maximum fluorescence change, and the dissociation constant for labeled peptide binding to MzT(ϪF1) were fitted parameters. The K m and V max for MzT(ϪF1) formation were 128 Ϯ 8 nM and 2.1 Ϯ 0.03 nM/s in NaCl and 470 Ϯ 60 nM and 2.9 Ϯ 0.2 nM/s in ChCl.
Model-independent values for the affinity and fluorescence change maxima for MzT(ϪF1) were obtained by fitting of the quadratic binding equation to the reaction end points, representing a titration of [5F]Hir-(54 -65)(SO 3 Ϫ ) with MzT(ϪF1) (Fig. 5, B and D). The parameters from this analysis were in excellent agreement with those obtained for the kinetic analysis and indicated that the affinity of exosite I for [5F]Hir-(54 -65)(SO 3 Ϫ ) increased 5.7-fold in the presence of Na ϩ (Table 1).
Effect of Na ϩ on Exosite I of Active MzT-Intact, native MzT generated from ProT could not be studied because of rapid autocatalytic conversion to MzT(ϪF1) during ecarin activation (results not shown). A ProT mutant in which the fragment 1 cleavage site at Arg 155 was substituted with Ala was activated to stable MzT R155A , which allowed Na ϩ -exosite I linkage to be   Table 1. C, kinetic titrations for activation of 116 -1450 nM Pre 1 in buffer containing ChCl (E) otherwise as described in A, along with the simultaneous fit with the parameters listed in Table 1. D, end point titration for the reactions in buffer containing ChCl as described in B. Reactions were performed and analyzed as described under "Experimental Procedures."  Fig. 7, A and C, respectively. SDS-gel electrophoresis confirmed that MzT R155A was the product of the reactions and was stable at concentrations and incubation times in excess of the experimental conditions (not shown). Full-length MzT R155A behaved similarly to MzT(ϪF1) but with a few distinguishing features. Unlike the results for MzT(ϪF1), the fit of the kinetics of MzT R155A formation was improved modestly by including competitive product inhibition of ecarin. The fitted K D for [5F]Hir-(54 -65)(SO 3 Ϫ ) binding to MzT R155A was 81 Ϯ 3 nM in NaCl and was 10-fold higher (830 Ϯ 40 nM) in ChCl. The kinetic parameters for ProT R155A activation by ecarin in NaCl were V max 2.1 Ϯ 0.2 nM/s, K m 110 Ϯ 30 nM, and K i 100 Ϯ 50 nM, and were indistinguishable for V max , K m , and K i , determined in ChCl at 2.0 Ϯ 0.3 nM/s, 80 Ϯ 40 nM, and 210 Ϯ 160 nM, respectively. Analysis of the reaction end points for [5F]Hir-(54 -65)(SO 3 Ϫ ) binding to MzT R155A in NaCl gave a K D of 100 Ϯ 30 nM (Fig. 7B). The end point results in ChCl could not be fit independently due to continued slow decreases in fluorescence at high ProT concentrations and long reaction times. Assuming an end point equivalent to that determined kinetically gave a K D of 770 Ϯ 120 nM (Fig. 7D). Together, the results demonstrated slightly larger effects of Na ϩ on the affinity of [5F]Hir-(54 -65)(SO 3 Ϫ ) for MzT R155A (8-to 10-fold) compared with MzT(ϪF1) (5-to 6-fold).
Na ϩ Binding Affinity Estimated from the Kinetics of Chromogenic Substrate Hydrolysis-To compare the apparent affinity of Na ϩ for thrombin, MzT(ϪF1), and MzT R155A , the initial rate of 200 M (saturating) D-Phe-Pip-Arg-pNA hydrolysis was determined as a function of Na ϩ concentration. Na ϩ produced hyperbolic increases in rate of 4.9-fold, 3.1-fold, and 3.3-fold, from which apparent K D values of 12 Ϯ 4, 9 Ϯ 2, and 14 Ϯ 4 mM were obtained for the substrate-bound forms of thrombin, MzT(ϪF1), and MzT R155A , respectively. The apparent affinities were the same and in reasonable agreement with the published K D of 10 mM for Na ϩ binding to the chromogenic substrate⅐thrombin complex at 25°C but at higher pH and ionic strength (25).
The K D values for Na ϩ binding to the free forms of thrombin and MzT(ϪF1) were estimated from the dependence of k cat /K m on Na ϩ concentration (45). Despite the analysis of five progress curves at each Na ϩ concentration, the error in the dissociation constants was large (Fig. 8A). This was due to the necessity of fitting three parameters (k cat , K m , and K i ) to the progress curves and three parameters to k cat /K m as a function of Na ϩ concentration (k cat /K m in the absence and at saturating Na ϩ and K D ).  Table 1. C, kinetic titrations for activation of 91-1198 nM ProT R155A in buffer containing ChCl (E) otherwise as described in A, along with the fit with the parameters listed in Table 1. 12 reactions out of a total of 21 analyzed are shown. D, end point titration for the reactions in buffer containing ChCl as described in B except that the line represents the fit of the binding equation with ⌬F max /F o fixed at Ϫ35%. Reactions were performed and analyzed as described under "Experimental Procedures." The K D values for thrombin and MzT(ϪF1) were both 26 Ϯ 20 mM, roughly consistent with the value of 22 mM, determined at higher ionic strength and pH (25,45). By contrast, k cat /K m for MzT R155A increased linearly up to 110 mM NaCl. This was accompanied by a ϳ12-fold decrease in K m , whereas this parameter decreased only ϳ2to 3-fold for thrombin and MzT(ϪF1) (Fig. 8B). Analysis of the dependence of K m on Na ϩ concentration according to the non-essential activation mechanism as described under "Experimental Procedures," gave an estimate of ϳ185 mM for the dissociation constant for Na ϩ binding to free MzT R155A and a linkage factor (␣) of ϳ0.085, corresponding to a ϳ12-fold increase in Na ϩ affinity at saturating substrate concentration. The value of ␣K Na of ϳ16 mM from this analysis was in agreement with 14 Ϯ 4 mM determined directly at saturating substrate concentration. The dependence of k cat on Na ϩ concentration for all of the enzymes was consistent with the directly measured dissociation constants for the substrate-saturated forms described above. The results demonstrated distinctly different kinetic behavior and a much higher K D for Na ϩ binding to MzT R155A , which contains the fragment 1 domain, compared with thrombin and MzT(ϪF1), which do not.

DISCUSSION
Examination of the appearance of Na ϩ -exosite I linkage on ProT and its activation products supports the conclusion that proexosite I on ProT, Pre 1, and Pre 2 is not linked to Na ϩ binding. Expression of linkage accompanies activation of the thrombin, MzT(ϪF1), and MzT catalytic sites by cleavage at Arg 320 . Exosite I affinity and catalytic activity of MzT and MzT(ϪF1) are shown for the first time to be allosterically regulated by Na ϩ . The well established coupling of protein substrate specificity to Na ϩ binding for thrombin suggests that the distinctly different specificity of MzT will be regulated by the MzT slow form to fast form transition. Increasing MzT exosite I affinity and catalytic activity in the allosteric transition by Na ϩ binding may also affect recognition of MzT as a substrate of the prothrombinase complex.
The affinity of [5F]Hir-(54 -65)(SO 3 Ϫ ) for (pro)exosite I on the zymogen forms, ProT, ProT R155A , Pre 1, and Pre 2 was consistently 1.3-to 1.7-fold higher in the presence of Na ϩ . This difference is well within the experimental error and may not be significant. As shown in the crystal structure of the Pre 2⅐hirugen complex, the Na ϩ site is not properly formed on Pre 2 (49). On this basis, the results demonstrate that the absence of link-age between Na ϩ and proexosite I is due to the inability of all zymogen forms to bind Na ϩ . The conformational changes that complete the folding of the catalytic domain and formation of the catalytic site are necessary for expression of allosteric linkage.
In a recent structural analysis of human thrombin, crystal packing interactions were identified that have influenced interpretation of previous structures of the fast, slow, and inactive forms (31). The structure of the D102N thrombin mutant in the absence of Na ϩ lacks crystal-packing interactions and is thought to represent more closely the inactive slow form in equilibrium with the active slow form and fast form. In the D102N thrombin structure, the catalytic site is blocked by large movements of Trp 215 and Arg 221a to occlude the catalytic and primary specificity sites, whereas the guanidinium group of Arg 187 occupies the Na ϩ site (31). Rapid-reaction kinetic studies of Na ϩ binding indicated that 23% of wild-type thrombin is in the inactive slow form at 15°C and that this form is not significantly populated at 37°C (31,33). Presumably this form was also present at some level in the studies done here at 25°C, suggesting that binding of [5F]Hir-(54 -65)(SO 3 Ϫ ) to exosite I contains some contribution from inactive slow forms. Whether such inactive slow forms exist for MzT(ϮF1) is not known.
Inactivation of thrombin, MzT(ϪF1), and MzT with FPR-CH 2 Cl virtually obliterated Na ϩ -exosite I linkage as probed by [5F]Hir-(54 -65)(SO 3 Ϫ ) binding affinity and normalized all of the affinities to values indistinguishable from that of the fast form of thrombin. This is fully consistent with an analysis of the structures of the thrombin slow and fast forms, free and FPRinactivated, which demonstrated that active site labeling locks thrombin in a fast form (26), and the essentially identical structures of FPR-thrombin and the Hir-(53-64)⅐thrombin complex (29). The Na ϩ independence of the exosite I affinities is explained by binding of the transition state analog to the S1-S4 4 specificity sites, formation of a hemiketal with Ser 195 , and subsequent alkylation of His 57 . Although FPR-thrombin does bind Na ϩ (26), the relaxation of Asp 189 in the S1 site, Ser 195 , and His 57 to their conformations in the slow form in the absence of Na ϩ cannot take place. It should be noted that this finding does not compromise the conclusions of our previous studies of exosite I expression using FPR-inactivated thrombin and MzT(ϪF1), because, as shown here, these inactivated enzymes have affinities for the hirudin peptide essentially the same as the active fast forms. The affinities of MzT R155A for the peptide, however, are 2-to 3-fold lower at 110 mM NaCl.
Kinetic analysis of Na ϩ binding to MzT R155A revealed a lower affinity compared with thrombin and MzT(ϪF1), which was associated with a larger decrease in K m for the chromogenic substrate, and not significantly larger increases in k cat . This unexpected behavior suggests that the presence of the fragment 1 domain attenuates affinity for Na ϩ binding. The large effect of the presence of fragment 1 on the Na ϩ -dependent apparent substrate binding affinity may be due to steric or conforma-tional differences affecting the active site or the Na ϩ binding site, and indirectly exosite I. Further studies are needed to define the kinetic and molecular mechanism of MzT regulation by Na ϩ .
Although the mechanism of Na ϩ regulation of MzT R155A is not resolved, the results demonstrate for the first time that MzT(ϪF1) and MzT R155A are Na ϩ -regulated proteinases exhibiting linkage to exosite I. Na ϩ regulates the specificity of thrombin for procoagulant and anticoagulant substrates and effectors (15,17,18,23,24). The fast form has higher specificity (k cat /K m ) than the slow form for the procoagulant substrates, fibrinogen and protease-activated receptors, whereas relative to the fast form, the slow form is more specific for protein C activation (48). Compared with thrombin, MzT(ϮF1) has Յ10% activity toward fibrinogen and Յ2% activity in platelet activation despite having an active catalytic site and exosite I (19,20). Thrombin activates factor V in solution and factor V bound to phospholipid membranes in exosite I-and II-dependent reactions (22, 44, 50 -52), whereas MzT activates factor V at a significant rate only when the substrate and enzyme are membrane-bound (21,22). It has been suggested that MzT may play an important procoagulant role as a specific physiological activator of membrane-bound factor V during initiation of blood coagulation (21,22). Based on the present results, and the observation that factor V activation is catalyzed more efficiently by the thrombin fast form (52), MzT-catalyzed factor V activation is predicted to be regulated by Na ϩ . The dissociation constant for Na ϩ binding to thrombin under physiological conditions is 110 mM, indicating an equilibrium of 60% fast and 40% slow forms (25,53,54). The K D for Na ϩ binding to MzT R155A at 37°C was not determined for the reasons cited above, but, depending on the temperature dependence of Na ϩ binding, MzT may be similarly poised for Na ϩ regulation physiologically.
In the thrombin⅐thrombomodulin complex, the slow form of thrombin becomes a potent anticoagulant proteinase due in part to the loss of specificity for fibrinogen and other procoagulant substrates, whereas the capacity for protein C activation is greatly increased (15-18, 48, 55). Activation of protein C by MzT and MzT(ϪF1) is also stimulated by thrombomodulin to rates comparable to thrombin and even faster rates when MzT is bound to phospholipid membranes (19,20). This has been suggested to represent an important physiological role for MzT as an anticoagulant (19,20). The present studies suggest that, like thrombin, Na ϩ -bound and -free forms of MzT may exhibit higher procoagulant and anticoagulant activities, respectively.
ProT activation is dependent on expression of exosites on factor Xa in the factor Xa⅐Va⅐membrane complex that mediate a substrate recognition mechanism consisting of initial exosite binding and a subsequent conformational change engaging the catalytic site in peptide bond cleavage (7)(8)(9)(10)(11)(12)(13)(14). Other studies support the hypothesis that (pro)exosite I on ProT also contributes to substrate recognition by mediating productive ProT binding and/or affecting the subsequent conformational change through interactions with the ProT binding site on factor Va in the prothrombinase complex (56 -59). ProT activation catalyzed by prothrombinase proceeds sequentially, with initial cleavage at Arg 320 to form MzT, followed by cleavage at Arg 271 to generate thrombin and fragment 1.2 (2)(3)(4)(5)(6)(7). Recent studies of the activation pathway support a mechanism in which substrates in the zymogen and proteinase states bind in alternate conformations that direct sequential cleavage at the two activation sites (8). The ProT zymogen is positioned for optimal cleavage at Arg 320 , whereas the product, MzT in the proteinase conformation, ratchets to a different bound conformation for cleavage at Arg 271 . On the basis of the evidence for the role of (pro)exosite I in substrate recognition through factor Va (56 -59), and the differential expression of the exosite on ProT zymogen and proteinase forms (35,36), expression of exosite I on MzT may contribute to ratcheting of the bound substrate for presentation of Arg 271 for cleavage. The present studies show that the chromogenic substrate activity and affinity of MzT exosite I are enhanced by Na ϩ , suggesting that coupling of Na ϩ binding and exosite I expression upon cleavage at Arg 320 may contribute to recognition of this substrate. Previous studies of the role of Na ϩ on the rate of Pre 1 activation in the absence of membranes by the Na ϩ -insensitive factor Xa mutant, Y225P, also lacking the ␥-carboxyglutamic acid domain, showed a 10-fold enhancement by Na ϩ , which was correlated with a modest increase in affinity for factor Va (60). On the other hand, with wild-type factor Xa assembled into prothrombinase the Na ϩ enhancement was only Ͻ1.5-fold (60). Whether Na ϩ affects the rate of MzT cleavage by prothrombinase remains an open question.