Transient kinetics of heparin-catalyzed protease inactivation by antithrombin III. Characterization of assembly, product formation, and heparin dissociation steps in the factor Xa reaction.

The kinetics of alpha-factor Xa inhibition by antithrombin III (AT) were studied in the absence and presence of heparin (H) with high affinity for antithrombin by stopped-flow fluorometry at I 0.3, pH 7.4 and 25 degrees C, using the fluorescence probe p-aminobenzamidine (P) and intrinsic protein fluorescence to monitor the reactions. Active site binding of p-aminobenzamidine to factor Xa was characterized by a 200-fold enhancement and 4-nm blue shift of the probe fluorescence emission spectrum (lambda max 372 nm), 29-nm red shift of the excitation spectrum (lambda max 322 nm), and dissociation constant (KD) of about 80 microM. Under pseudo-first order conditions [( AT]0, [H]0, [P]0 much greater than [Xa]0), the observed factor Xa inactivation rate constant (kobs) measured by p-aminobenzamidine displacement or residual enzymatic activity increased linearly with the "effective" antithrombin concentration (i.e. corrected for probe competition) up to 300 microM in the absence of heparin, indicating a simple bimolecular process with a rate constant of 2.1 x 10(3) M-1 s-1. In the presence of heparin, a similar linear dependence of kobs on effective AT.H complex concentration was found up to 25 microM whether the reaction was followed by probe displacement or the quenching of AT.H complex protein fluorescence due to heparin dissociation, consistent with a bimolecular reaction between AT.H complex and free factor Xa with a 300-fold enhanced rate constant of 7 x 10(5) M-1 s-1. Above 25 microM AT.H complex, an increasing dead time displacement of p-aminobenzamidine and a downward deviation of kobs from the initial linear dependence on AT.H complex concentration were found, reflecting the saturation of an intermediate Xa.AT.H complex with a KD of 200 microM and a limiting rate of Xa-AT product complex formation of 140 s-1. Kinetic studies at catalytic heparin concentrations yielded a kcat/Km for factor Xa at saturating antithrombin of 7 x 10(5) M-1 s-1 in agreement with the bimolecular rate constant obtained in single heparin turnover experiments. These results demonstrate that 1) the accelerating effect of heparin on the AT/Xa reaction is at least partly due to heparin promoting the ordered assembly of antithrombin and factor Xa in an intermediate ternary complex and that 2) heparin catalytic turnover is limited by the rate of conversion of the ternary complex intermediate to the product Xa-AT complex with heparin dissociation occurring either concomitant with this step or in a subsequent faster step.

4 To whom correspondence should be addressed. complex and that 2) heparin catalytic turnover is limited by the rate of conversion of the ternary complex intermediate to the product X,-AT complex with heparin dissociation occurring either concomitant with this step or in a subsequent faster step.
Heparin, a highly sulfated glycosaminoglycan, possesses a potent anticoagulant activity which derives from its ability to catalyze the inactivation of blood coagulation proteases by their primary inhibitor, antithrombin I11 (AT)' (1). Studies of the mechanism by which heparin accelerates the inhibition of these proteases by antithrombin have suggested that the mode of heparin action may not be the same for all proteases (1,2). Thus, in the case of thrombin, substantial evidence has accumulated in favor of a surface catalytic or "approximation" mechanism in which binding of both antithrombin and protease to the heparin polysaccharide surface promotes an initial interaction between the two proteins prior to their reaction to form a stable inhibitor-protease complex. This evidence includes: 1) the requirement for catalytic activity of a heparin chain length corresponding to the minimum size which can accommodate both proteins simultaneously (ie. z 18 saccharide residues) (3-5); 2) the ability to markedly reduce heparin's accelerating effect at high heparin concentrations where binding of antithrombin and protease to separate heparin molecules is favored (6)(7)(8)(9); and 3) the selective loss of heparin accelerating activity by natural mutation or chemical modification of amino acid residues in either antithrombin or protease which are presumably essential for heparin binding (10)(11)(12)(13)(14). Rapid kinetic studies have provided direct evidence for the assembly of thrombin and antithrombin on the heparin surface prior to their reaction to form the product thrombin-AT complex. Such studies have demonstrated that heparin acceleration is due to heparin promoting the initial encounter of thrombin and antithrombin in the initial assembly step rather than enhancing the rate of the subsequent product formation step (15, 16). Later studies showed that heparin is released concomitant with product complex formation and does not limit the rate of heparin catalytic turnover (17).
In the case of factor X,, several observations suggest that the surface catalytic mechanism elucidated for the heparinenhanced thrombin/AT reaction cannot explain heparin's accelerating effect on the inactivation of this protease by antithrombin. Thus, heparin oligosaccharides containing the unique pentasaccharide necessary for high affinity binding of The abbreviations used are: AT, antithrombin 111; X,, factor X.; P, p-aminobenzamidine; H, heparin with high affinity for antithrombin 111; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
the inhibitor and as small as this pentasaccharide, effectively accelerate antithrombin inactivation of factor X., despite their lack of activity with thrombin as the protease (3-5,18). Moreover, high heparin concentrations do not appear to decrease heparin's accelerating effect on the AT/X. reaction as they do with the AT/thrombin reaction (3, 6). Such observations are consistent with factor X. binding to heparin not being essential for heparin rate enhancement and have suggested that the conformational change induced in antithrombin upon binding to the specific pentasaccharide site on heparin may make antithrombin a better inhibitor of factor X. (1,2,6). That this allosteric mechanism is insufficient to completely account for heparin's accelerating effect is suggested, however, from the increased effectiveness of heparins of increasing chain length in accelerating the AT/X, reaction (5, 19). It is thus possible that a surface catalytic mechanism may contribute to the rate enhancement produced by larger heparin chains.
Because of the success of the rapid kinetic approach in resolving the elementary mechanistic steps in the heparincatalyzed AT/thrombin reaction (e.g. intermediate complex assembly, conformational change, product formation, and heparin dissociation steps), we have examined the heparincatalyzed AT/X, reaction using a similar approach as a basis for delineating mechanistic differences between these two protease reactions. The present studies were thus undertaken to determine: 1) whether the initial assembly of an intermediate AT. X, complex and subsequent formation of the product inhibitor-protease complex could be resolved in the presence and absence of heparin; 2) whether heparin accelerated the AT/X, reaction by promoting the assembly step or the product formation step; and 3) whether antithrombin conformational change or heparin dissociation steps limited the rate at which heparin could recycle as a catalyst. The results of these studies indicate that the heparin-catalyzed AT/X, reaction can be described by the same sequence of reaction steps previously found to characterize the heparin-enhanced AT/ thrombin reaction although clear quantitative differences in binding and rate parameters as well as heparin effects on individual reaction steps are evident in these two reactions. The implications of our results for the different hypothesized modes of heparin action are discussed.

RESULTS
Kinetics of Factor X, Inhibition by Antithrombin-The kinetics of antithrombin I11 (AT) inhibition of factor X, were monitored continuously in the presence of p-aminobenzamidine (P) from the decrease in fluorescence accompanying the displacement of the probe from the factor X, active site by the inhibitor. Assuming that antithrombin inhibits factor X. in a two-step process involving an initial reversible formation of a noncovalent X.. AT complex followed by the irreversible formation of a covalently linked X,-AT complex (Scheme l), and that equilibration of X,. P and intermediate X,. AT complexes is rapid relative to the rate of covalent X,-AT complex formation, then displacement of p-aminobenzamidine is predicted to occur in two phases: 1) an initial rapid displacement of a fraction of the bound probe due to equilibrium formation of the intermediate X,. AT complex, followed by 2) a slower * Portions of this paper (including "Materials and Methods," part of "Results," and Figs. 1 and 3 ) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. displacement of the remaining bound probe due to quantitative conversion of factor X. to the covalent X,-AT complex (15). Kid KX,,,, X..P e P + X . + A T where hF, , , . , represents the maximum fluoresence change achieved when factor X, is saturated with p-aminobenzamidine. Fig. 2 (left panel) shows factor X, reactions with three different concentrations of antithrombin monitored by paminobenzamidine displacement under pseudo-first order conditions. A simple exponential decay of bound probe fluoresence was observed for all reactions. Pseudo-first order rate constants determined from the data of antithrombin concentration (Fig. 4). In accordance with the predicted behavior of Equation well with the bimolecular rate constant of 2.5 k 0.1 X IO3 M-' s" determined for the AT/X, reaction either in the presence or absence of Polybrene by discontinuous assay of residual factor X. enzymatic activity under the same solution conditions (see "Materials and Methods," in Miniprint), while the latter value is indistinguishable from the K D determined in Fig. 1B by direct binding measurements. These results thus justified the use of p-aminobenzamidine as a reporter of the AT/X. reaction.

X . A T " * X -A T k SCHEME
To determine whether the saturable formation of an intermediate X,. AT encounter complex could be observed on the X./AT reaction pathway (Scheme l), kinetic measurements of the reaction were made in the stopped-flow fluorimeter using the p-aminobenzamidine displacement technique over an effective antithrombin concentration range extending as high as -300 p~ and under pseudo-first order conditions. In all cases, single exponential reactions were observed which showed no significant loss in reaction amplitude due to displacement of the probe in the instrument dead time and which yielded pseudo-first order rate constants that increased linearly with the effective antithrombin concentration (Fig. 4).
The slope of the linear kinetic plot provided a bimolecular rate constant, ~/ K x . , A T ,

Kinetics of Factor X , Inhibition by Antithrombin-Heparin
Complex- Fig. 2 compares reactions of factor X, , with either antithrombin or antithrombin-heparin complex at identical concentrations. Pseudo-first order conditions resulting in only a single heparin turnover were achieved for the latter reac- Dependence of pseudo-first order rate constants and amplitudes for the AT/factor X. reaction on antithrombin concentration. Pseudo-first order rate constants and fluorescence amplitude changes were measured for reactions of antithrombin with 0.4-1 p~ factor X. and 40 or 80 p~ p-aminobenzamidine by fitting probe displacement curves to a single exponential decay function, as described under "Materials and Methods" and plotted uersus the antithrombin concentration, corrected for probe competition, as shown. Average values for reaction amplitudes (in volts) were corrected (<16%) for the small inner filter effect due to end absorbance of antithrombin at higher inhibitor concentrations, as described previously (15). Rate constant data were fit by linear regression analysis while an average value was calculated for the experimentally indistinguishable amplitudes (solid lines). tions by employing a large molar excess of AT.H complex over the factor X, concentration (15)(16)(17). Moreover, a 1.1-2fold molar excess of antithrombin over heparin insured that nearly all heparin molecules were complexed with the inhibitor, based on the dissociation constant of 0.23 k 0.02 p~ determined for this interaction under these reaction conditions (16, 17). As can be seen in Fig. 2, these conditions resulted in a single exponential decay of bound p-aminobenzamidine fluorescence at all AT.H complex concentrations with reaction amplitudes equal to the initial bound probe fluorescence. Indistinguishable reaction curves were obtained whether heparin was premixed with antithrombin or with factor X.. In contrast to the reaction with antithrombin, antithrombin-heparin complex inhibited factor X. over a time frame that was about 300-fold shorter at comparable inhibitor concentrations. Like the reaction with antithrombin, however, the pseudo-first order rate constant increased linearly with the effective AT.H complex concentration at least as high as 25 p~ (Fig. 5). Over this concentration range, the reaction thus appeared to be a simple bimolecular process with antithrombin-heparin complex rather than antithrombin acting as the inhibitor in Scheme 1. A bimolecular rate constant 300-fold greater than that found in the absence of  Fig. 4. The contribution of the uncatalyzed reaction to k& due to the excess free antithrombin was calculated to be negligible (<0.5%). Solid lines represent nonlinear least squares fits to Equations 1 and 2 (with AT. H complex replacing AT). The inset compares kobs values measured by p-aminobenzamidine displacement (circles) at lower AT.H complex concentrations with those measured by quenching of intrinsic protein fluorescence (triangles). The latter data were obtained from reactions of 0.1-0.5 p~ factor X. with 1-6 p~ A T and a 1.3-3-fold molar excess of heparin as described under "Materials and Methods." calculate [AT. HI) up to 6 p~ with a similar bimolecular rate To confirm the proportionality between koba and the AT. H complex concentration, the heparin concentration was fixed at 3 p~ and the antithrombin concentration varied, again under conditions where the calculated AT. H complex concentration was always in large molar excess over the X. concentration. This resulted in a saturable dependence of kob. on antithrombin concentration (not shown) which, when fit to the appropriate form of Equation 1 (16), indicated the titration of a protein-heparin complex having a KO of 0.26 k 0.09 p~ that was indistinguishable from the value determined for the AT. H interaction by direct binding experiments (16,17).
At effective antithrombin-heparin complex concentrations greater than 25 p~, a significant downward curvature from the initial proportional increase in k&s with AT-H complex concentration was found up to the highest experimentally feasible concentrations of about 150 p~ (Fig. 5 ) , suggesting the progressive saturation of an intermediate X,. AT. H ternary complex. Consistent with this conclusion, a concomitant increasing loss of reaction amplitude was observed over this range of A T . H complex concentration due to a fraction of the bound p-aminobenzamidine being displaced in the 2-ms dead time of the stopped-flow instrument (Fig. 5). These results were thus consistent with antithrombin-heparin complex inhibition of factor X, proceeding by the two-step reaction of Scheme 2.

SCHEME 2
From the nonlinear least squares fit of rate constant data to the predicted hyperbolic dependence of Equation 1 (with AT. H replacing AT), a dissociation constant for the X. interaction with AT. H complex, KX.,ATH, of 210 f 60 pM was determined, and a unimolecular rate constant for conversion of the intermediate ternary complex to the covalent complex, KH, of 140 f 30 s-l was obtained. A reciprocal plot of the amplitudes, according to Equation 2, is given in Fig. 5 which shows the predicted linear dependence on effective antithrombin-heparin complex concentration (15). From the intercept/slope of the weighted linear regression line, a dissociation constant of 140 f 40 pM was calculated, in reasonable agreement with the kinetic value, considering the experimental error.
Characterization of Protein Fluorescence Changes Accompanying the Reaction of Antithrombin with Factor X , in the Absence and Presence of Heparin-Although the previous single turnover experiments were consistent with the irreversible formation of the covalent X,-AT complex from an intermediate X,.AT. H ternary complex, they provided no indication whether heparin was released concomitant with the formation of the covalent complex or in a subsequent slower step (Scheme 3). SCHEME 3 The latter possibility would result in heparin dissociation limiting the rate at which heparin can complete a catalytic cycle. To investigate these alternatives, we monitored the loss of the 40% enhanced protein fluorescence of the antithrombin-heparin complex as a signal for heparin dissociation during the reaction with factor X, (17).
Before we could do this, it was necessary to determine whether the heparin-independent reaction of antithrombin :tor X , Transient Kinetics 5455 with factor X. was accompanied by any changes in protein fluorescence as were previously found to accompany the AT/ thrombin reaction (17,33). Fig. 6 compares the protein fluorescence changes accompanying the reaction of a 2-fold molar excess of antithrombin with factor X. in the absence and presence of heparin concentrations which saturated the inhibitor. In the absence of heparin, the total protein fluorescence following complete conversion to the product X.-AT complex was unchanged from the summed fluorescence contributions due to factor X. and antithrombin, indicating that covalent complex formation did not alter the protein fluorescence of either the inhibitor or the protease. In the presence of heparin levels which saturated antithrombin, the summed protein fluorescence of antithrombin-heparin complex and factor X. was greater than that of equivalent concentrations of antithrombin and factor X, by an amount equal to the approximately 40% enhanced fluorescence of the AT.H complex. A stoichiometric quenching of this enhanced fluorescence occurred upon reaction with factor X. resulting in a final level of protein fluorescence due to the X.-AT complex that was indistinguishable from that found for the X,-AT complex produced in the absence of heparin (after correction for the excess AT.H complex employed in these experiments). Heparin at these levels was found to have no detectable effect on the protein fluorescence of either factor X. or the covalent X.-AT complex. Moreover, the amounts of reactive sitecleaved antithrombin (AT") formed in this reaction (15-20% as judged from inhibitor/protease titrations; see "Materials and Methods" and Ref. 5), together with the similar fluorescence yields of antithrombin and ATM (17), indicated that ATM made no significant contribution to these protein fluorescence changes. Analysis of reaction products formed in these experiments by sodium dodecyl sulfate-polyacrylamide gel electrophoresis confirmed the quantitative conversion of factor X, to the covalent X.-AT complex with insignificant complex degradation. These results thus indicated that protein fluorescence changes could be used to follow the rate of heparin dissociation during a single turnover reaction of antithrombin-heparin complex with factor X,. When the reaction between AT. H complex and factor X. plus AT (-) and factor X. plus AT.H complex (------). The error bur represents the range of variation of the three lower spectra observed in experiments conducted at these and double the concentration of all components after normalization. Kinetics was examined in the stopped-flow fluorimeter using protein fluorescence detection under pseudo-first order conditions, protein fluorescence was found to decay exponentially with no evidence for lags as would be expected if heparin dissociation occurred as a distinct rate-limiting step subsequent to the covalent complex formation step (i.e. hB8 << kH in Scheme 3) (17). Observedpseudo-first order rate constants determined for these reactions increased linearly with antithrombin-heparin complex concentration over the concentration range examined and were indistinguishable from factor X, inactivation rate constants obtained using thep-aminobenzamidine signal (when compared at the same effective AT. H complex concentrations) (Fig. 5, inset). A bimolecular rate constant of 6.6 f 0.1 X 10' M" s" indistinguishable from that determined by probe displacement was thus obtained from the slope of this linear plot. These results indicated that heparin was released either concomitant with covalent complex formation or in a subsequent more rapid step (kdiss >> kH in Scheme 3).

Correlation between Rate Constants for Single and Multiple
Heparin Turnover Reactions-To confirm that the reaction model for heparin catalysis deduced from single heparin turnover kinetic studies accurately predicted the behavior of heparin under conditions where multiple heparin turnovers occurred, the kinetics of the heparin-catalyzed AT/X, reaction were investigated at heparin concentrations well below those of antithrombin and factor X. . Under these conditions the reaction can be treated by classical Michaelis-Menten kinetics since it is formally equivalent to a two-substrate/enzyme reaction where factor X, and antithrombin are the substrates and heparin is the enzyme (i.e. catalyst) (17,34,35). Thus, at . This equation indicates that the initial velocity of factor X. inactivation will increase linearly with [HI,, but exhibit a hyperbolic dependence on the initial factor X, concentration, [X.],, when the concentration of heparin or X,, respectively, is varied at fixed concentrations of the other reaction components. Fig. 7A shows that initial velocities measured by p-aminobenzamidine displacement (and corrected for probe competition) were linearly dependent on [HI0 up to a heparin concentration of 100 nM when fixed AT, X,, and P concentrations of 5, 0.5, and 80 p~, respectively, were used. The linearity of this plot at heparin concentrations only 5-fold less than [X.], is consistent with the negligible concentrations of ternary complex produced at these X., AT, and heparin concentrations, as predicted from the weak affinity of X. for the AT.H complex (Fig. 5). The ordinate intercept  (36). In these cases, initial velocities could be more accurately determined from the exponential rate constant (see "Materials and Methods"). Because the antithrombin concentrations employed in the experiments of Fig. 7, A and B, were saturating, the true &/K, for factor X. could be obtained from the slopes of the linear plots of these figures after dividing by the fixed X. or heparin concentrations employed, respectively (Equation 3) (17). The respective values obtained, 6.7 f 0.2 X 10' and 6.8 f 0.4 X lo6 M" s-l, were indistinguishable from the bimolecular rate constant determined for antithrombinheparin complex inactivation of factor X. in a single heparin turnover (Fig. 5).
Under conditions where the factor X. concentration is well below its K,, the two-substrate/enzyme reaction model predicts that the apparent K,,, for antithrombin will be identical to the KD for its interaction with the catalyst, assuming that binding steps are rapidly equilibrated (36). Fig. 7C shows initial velocity data obtained as a function of [ATl0 at fixed X. and heparin concentrations of 0.3 pM and 20 nM, respectively, to determine the apparent K,,, for this substrate. A saturable dependence of initial velocities on [ATIo with a limiting value similar to that measured in Fig. 7B at 0.3 p~ factor X. was observed, after correction for the uncatalyzed rate. The data were satisfactorily fit by the Michaelis-Menten equation with antithrombin as the varied substrate (solid line) which provided a KiTnp,, of 0.23 k 0.09 pM that was in direct agreement with the KD determined for the antithrombinheparin interaction (see above).

DISCUSSION
A combination of rapid kinetic and steady-state kinetic approaches employing both intrinsic and extrinsic fluorescence probes has been used to characterize the heparincatalyzed reaction of antithrombin with factor X. as a basis for comparison with previous studies conducted with thrombin as the protease. The 200-fold fluoresence enhancement of p-aminobenzamidine when bound at the active-site of factor X. provided one signal for continuously monitoring the progress of factor X. inhibition by antithrombin in the absence and presence of heparin from the decreased fluorescence accompanying the competitive displacement of the bound probe. The KD of -80 p~ determined for the X..P interaction in this study is somewhat lower than the value of 115 p~ recently reported (37). This may be due to the failure to correct for the weaker non-active site binding component we observed in this study which could have the effect of increasing the apparent KD. This weak binding component could possibly reflect an interaction of the positively charged paminobenzamidine molecule with the anionic y-carboxyglutamic acid region of factor X, , , since in the case of trypsin and thrombin, only active site binding of p-aminobenzamidine was observed (28). Because this secondary binding interaction does not involve the active site and contributes negligibly to the enhanced fluorescence of p-aminobenzamidine at concentrations required to saturate the active site interaction, it is not likely to interfere with the use of p-aminobenzamidine as a probe of factor X. active site interactions.
Examination of the reaction of antithrombin with factor X. in the absence of heparin over a wide range of effective inhibitor concentrations (i.e. corrected for probe competition) extending as high as 300 PM has shown no evidence for an intermediate X.. AT noncovalent complex. That an interme-diate complex exists in this reaction, however, is suggested from the slow bimolecular association rate constant which is several orders of magnitude below that expected for a diffusion-limited association (38). The linear dependence of k b s and independence of reaction amplitudes on antithrombin concentration contrasts with the hyperbolic dependence of rate constants and amplitudes on inhibitor concentration previously observed over the same effective antithrombin concentration range for the thrombin reaction (15). This would imply that the K D for an intermediate X, -AT complex must greatly exceed that of the intermediate thrombin .AT complex (KO -1 mM) ( Table I). Likewise, the unimolecular rate constant for conversion of the intermediate X.. AT complex to the product X.-AT complex must be considerably greater than the highest measured rate constant for this reaction of 0.6 s-'. The failure to observe saturation behavior in this reaction means that only a bimolecular rate constant equal to the ratio, k/KX.,AT (Scheme 1)) can be determined from our data (15). The value we obtained of 2.1 X lo3 M-' s-' compares favorably with other reported values for this rate constant in the literature measured under somewhat different experimental conditions (6, 37, 39, 40) and is -4-5 fold lower than the value previously measured for the thrombin/AT reaction (Table I).
A linear dependence of kob on the antithrombin-heparin complex concentration was also observed at effective concentrations up to 25 p~, consistent with the heparin-accelerated reaction proceeding as a simple bimolecular association between free factor X. and AT. H complex under these conditions. Such kinetic behavior does not preclude a reaction pathway in which factor X. initially binds heparin followed by reaction with free antithrombin as would occur in the random addition model of Griffith (35). Thus, the latter pathway would not be evident if AT-H binary complexes predominated over X.. H binary complexes under the conditions of our experiments. This would be in keeping with 1) reported binding constants for the factor X.-heparin binary complex interaction which are about 2 orders of magnitude weaker than the antithrombin-heparin binary complex interaction (41), 2) the considerably weaker affinity of factor X. as compared to antithrombin for heparin-Sepharose (22,42), and 3) the anionic charge of factor X. (43) and heparin. It follows, however, that a reaction pathway in which factor X. binds to heparin before antithrombin can make only a minor contribution to the reaction mechanism over a wide range of antithrombin, factor X., and heparin concentrations, including those likely to occur physiologically (16).

TABLE I Comparison of kinetic parameters for factor X , and thrombin inhibition by antithrombin or antithrombin-heparin complex
Deviations from a linear dependence of kObs on antithrombin-heparin complex concentration were observed at effective AT.H complex concentrations above 25 PM. That these deviations were due to the progressive saturation of a ternary X.. AT. H complex rather than some other effect of high AT. H complex concentrations (e.g. AT self-association3) was indicated from the dead time displacement of p-aminobenzamidine from factor X. which accompanied these deviations. The latter observations thus independently confirmed that an active site-dependent interaction between factor X. and AT. H complex with a K D similar to that determined from the rate constant data was rapidly established prior to product X,-AT complex formation. The insignificant binding of paminobenzamidine to heparin or antithrombin documented in previous studies (15) (Table I). Since the faster limiting rate constant for the heparin-enhanced reaction of factor X, relative to thrombin compensates for its weaker initial binding interaction, the bimolecular rate constants for these two protease reactions are similar under these experimental conditions (Table I). 4 The evidence for an intermediate binding step in the presence, but not the absence, of heparin over a comparable range of inhibitor concentrations indicates that at least part (ie. minimally 10-fold) of the 300-fold heparin enhancement of the XJAT bimolecular rate constant is due to heparin increasing the binding affinity of factor X. for antithrombin in the intermediate complex. The full extent to which heparin enhances this binding affinity and whether the 140 s" limiting rate constant represents a heparin-enhanced value cannot, however, be determined from our data due to our inability to quantitate the two reaction steps in the absence of heparin.
Assuming a rapid equilibrium addition of antithrombin and factor X, to the heparin catalyst, K , and kcat values for factor X. obtained in multiple heparin turnover studies should correspond to the ternary complex KO and the limiting product formation rate constant, respectively, determined in single turnover studies (17,36). K , values for factor X. of 160 and 100 nM and kcat values of -1 s and 0.7 s" at saturating antithrombin concentrations were reported by Griffith (35) and Pletcher and Nelsestuen (34), respectively, for the hepa-Chromatography of antithrombin on Sephacryl S-200 (2.5 X 100) at initial loading concentrations of 1, 0.1, and 0.01 mM (in -5 ml) resulted in indistinguishable symmetrical elution bands at equivalent elution volumes, consistent with no detectable self-association occurring in the absence of heparin at the antithrombin concentrations employed in this study. This is not true at physiological ionic strength ( I 0.15) where a further decrease in KO for the binding step of the heparin-enhanced thrombin/AT reaction (unpublished data) results in thrombin being inhibited by AT.H complex with a 30-fold faster bimolecular rate constant than factor X. (45). rin-catalyzed XJAT reaction. These values are clearly several orders of magnitude lower than those predicted by our rapid kinetic studies. Since this discrepancy could be due to a slow release of heparin from the product of the reaction which would not have been detected by p-aminobenzamidine displacement in a single heparin turnover, we used protein fluorescence changes to follow the rate of heparin dissociation during the reaction.
Quenching of the enhanced protein fluorescence of the antithrombin-heparin complex was found to accompany its reaction with factor X. as was previously found with the thrombin reaction (17), indicating that the antithrombin conformation responsible for tight binding of heparin (44) is lost upon formation of the covalent X.-AT complex, thereby facilitating rapid heparin dissociation from the reaction product (2,6,17). Indistinguishable bimolecular rate constants were measured when single turnover reactions of AT. H complex with factor X. were monitored by protein fluorescence quenching or p-aminobenzamidine displacement, implying that heparin must dissociate either concomitant with the 140 s" step or in a much faster subsequent reaction step (Scheme 3). Heparin dissociation thus cannot be rate-limiting during catalytic turnover. To confirm the conclusions of our rapid kinetic studies, we examined the reaction kinetics under conditions similar to past kinetic studies (34,35), i.e. at catalytic heparin concentrations where multiple heparin turnovers would occur. These kinetic data were consistent with the prediction of our rapid reaction data of an extremely weak K , for factor X. and showed no evidence for saturation of the reaction at factor X, concentrations previously reported to be saturating. The higher ionic strength conditions employed in our experiments cannot account for the discrepancies with past studies since the bimolecular rate constant for this reaction, kH/KX,,ATH, was found to be largely independent of ionic strength. The saturation behavior with respect to factor X. reported by Pletcher and Nelsestuen (34) may be due to experimental error since the highest factor X. concentration they employed was less than one-fifth of K,. In contrast, Griffith's study (35) employed factor X. concentrations nearly equivalent to his reported K , for factor X. . It is possible that systematic errors in measuring true initial velocities at higher factor X, concentrations could have contributed to the apparent saturation behavior observed in the latter study.
In summary, our results indicate that the heparin-catalyzed AT/X. reaction can be described by the same sequence of reactions previously demonstrated for the heparin-catalyzed AT/thrombin reaction (Scheme 4). Thus, heparin initially binds to antithrombin and induces a favorable conformational change in the inhibitor which shifts the binding equilibrium in favor of complex formation (17,29,44). A ternary X.. AT. H complex is then formed in which the reactive site bond of antithrombin is reversibly bound at the protease active site. This binding interaction, although weak relative to the analogous interaction with thrombin, is nevertheless enhanced by X8 SCHEME 4 heparin by at least an order of magnitude. Factor X, and antithrombin subsequently react in the ternary complex in an irreversible reaction step at 140 s-' to form the covalent complex. This reaction induces antithrombin to undergo an additional conformational change either concomitant with this step or in a subsequent faster reaction step that results in the X.-AT product complex binding heparin with an affinity several orders of magnitude weaker than that of antithrombin (6,17). Rapid heparin dissociation from the product and rebinding to unreacted antithrombin to begin another catalytic cycle is thus favored, thereby preventing the accumulation of nonproductive product-heparin complexes.
While the reaction pathways for heparin-catalyzed reactions of antithrombin with factor X. and thrombin are similar, the mechanism by which heparin accelerates these two reactions may still differ significantly. Thus, the ability of heparin oligosaccharides as small as the unique pentasaccharide, which binds antithrombin to produce an accelerating effect on the AT/X, reaction comparable to that of larger heparin chains, whereas heparin chains of at least 18 saccharide units are required to accelerate the AT/thrombin reaction, has implied a fundamental mechanistic difference between these reactions (1, 2). These observations have suggested that the antithrombin conformational change plays a primary role in heparin's ability to accelerate the AT/X. reaction whereas the binding of both antithrombin and protease to the same heparin molecule is largely responsible for heparin's ability to accelerate the AT/thrombin reaction. In the context of these hypotheses, our current results would suggest that the heparin-induced antithrombin conformational change functions in part by making antithrombin more complementary to the factor X. active site. The role of this conformational change in enhancing the rate of covalent complex formation as well as the contribution of a factor X.-heparin interaction to heparin's accelerating effects, however, remain to be elucidated. hexylglycyl-glycyl-L-arginine-p-nnroanilide (Specuozyme F X ; , was prchased from American Diagnostics Gel finration media and heparin-Sepharose were from Pharmacia. Anernatively, heparin-Sepharose was Prepared by linking the Soma heparh prcduci at 40 mg/ml to an equal volume d Sepharose 48 following actlvatlon d me gel wth CNBr essentially aaprding to March, at el. (21). All other chemicals were of the highest grade commercially available. PWE!NS Factor X was Prepared from expired plasma essentially aaxlrding to Miletiih. e1 al. (22). except that heparin-Sepharose was used in place d suMed Sephadex 0-50 and a find gel filtration step was Performed on Sephacryl S-XCI.~ Factor X was activated using me purin$c wagulm protein from  Semimicro-or micro-fluorescence cuvenes with reduced excitation pathlengths were used in most cases to minimize inner finer Btteco. p-Aminobemamidine excitation spectra Were obtained using Aem 370 nm, emlss1on spectra were taken with .Iex 330 nm. and fixed wavelength titrations were performed with .Iex 345 nm and Aem 368 nm. Sandpasses were 581 as in previous sludles (17.28). Concentrations empbyed lor emlation spectra w e 20 pMpaminobemmidlne and 1 pM factor X , end for amission spectra were 50 or 2W pMp-aminobenramidine and 1 pM factor %. Spectra d the factor X. -P complex were calculated by Subtraction of beep-aminobemamldine and protein flmescence contributions Obtained from spectra of the indiwdual components at identical Concantratims. Fluorescenut trations of factor Xa wilh paminobenramidine were mnected for free probe flumescence and inner flbr Matts due to absorbance ol the probe by performing control inrations dp-aminobenzamldire into buffer as dasuibed previously (18.28) with conact-for dilution (<5%). Titrations dnetii~factor Xawithp-aminobemamidine were also corrected for lhe non-active-rife binding wnlribvtion by suMraction of lhe enhanced probe fluorescence ObSeNed in tnrations of an equal concantration d Bctiw-SIB blocked factor X , withpamlnobenzamidine. Corrmed Inration dale ware analyzed wim mS quadratic binding function (29) assummg a 1 :1 binding sloidlomelry.
Protein RuoreSCsnce emission spectra Were obtained using )Lex 280 nm with bandpasses set as in previous work (17). Inner finer correct-(6%) were made based on me 280 nm absofbance and the formed in the absence or preseme 01 heparin were calculated by subtraction d the fluorascenca due to calibrated pethlengm d the semimbro-arvene (0.3 c m ) (16). spectra of ma slab18 factor Xa-AT complex excess antinhrombin OT amhrombin-heparin complex, present lo awld protechpis d Ma protease-inhibitor complex (17). Stoichiometric titrabons d factor Xa wim antnhrombin in the absence or presence of heparln corrections for ail fluorescence speclre were made by suMracting a butfer spectrum. Continuous slirrmg of (see below) ware used lo determine axcess inhibitor levels in the laner experiments. Background samples durlng aCqUisWOn d protein fiUOreSCenCe Spectra eliminated any Signincam problems due to photodecompoSRion.  absence and presence of heparin were found to be independent of the factor X , concentration, confirming a pseudo-llrsl-order process.
Reaction d 0.5-1.5 pM MULTIPLE ULRNOVER KINETIC STILDLES presence of catawlc concentrations of heparon were monitored byp-amlnobenramdine dlspiacsment (1 7) The kinetics of factor X, reactions wnh antnhrombin the 370 nm (8 nm bandpass), respectively Factor X , , heparln andp-ammobemamodine were added tn uslng the SLM spectrofluorometer at excitatkon and emlsston wavelengths of 325 nm (4 nm bandpass) and react~on buffer to the cuvette and piemarbated at 25OC. Alter recordinp me initial fluorescence, the leactlon was innialed by addltlon of temperature equilibrated antnhrombin. Ccmplele progress curves were acquired through an iBM XT intefiace (SLM instruments) end Stored lor later analysis. For most reactions.
the m a l 40% of the reactm was fit to the second order polynomial, Ft = at2 + bl + c, where a. b and c are flned constants. and b was taken as the lnltlal rate of fluorescence decay (AFt/At) (30). The rate was convened to the lnltial velcmty of factor X, inactivation (q) using tha equation (17) W a l t AFt [X& ",=" The fluorescence emission spectrum ofp-aminobenzamldlne (P) was enhanced and blue-%te:nk$esence of factor X , (Fig. 1A). (Xlaiitatively similar changes inp-aminobenzamldine fluorescence Were previously found in the presence of trypsm and *thrombin and Shown to be due to the blndlng of th0 probe in the hydrophobic specficny pocket d the active-SRe of these proteases (28) TO demonstrate that the factor Xa-inducad flUmeSCence changes in p-amlnobenzamidine were also due to active-sne bndfng, we first determined whether these spemal changes corresponded to a saturable bmdlng process. Initial expements in whchp-aminobenramidine was tltrated into a solution of factor X, 10 a final concentration of about 1 mM and the fluorescance enhancement d the probe monitored, yielded btphasic titration CUNBS wlth a failure to reach Saturation, suggesting the presence of a "an-actlve-site blndlng component. This was confirmed from the Salurable fluoresmnce enhancement Observed when higher concentrat#ons ofp-aminobemmidine were tnrated into a dution d factw X , W e d a1 Rs active-5818 wnh p-~,dinophenylmelhyls"Uonyin"~de (not shown). Nmlinear least squares analysis of this bmding data indicated a non-active-sire blnding component wim a relalively weak affinity (KO 6-1 1 mM) and a fluorescence anhancement less than 10% that due Io the &e-se binding component alp-aminobenzamidine concentrations where the letter was 90% saturated (Fp. 10). Conmion d native factor X , bmdlng curves for mis "on-actre-siie bmdmg component resolved a saturabk aniva-snedependem increase mp-aminobBcLLamidine nuoramnce (Fp. tB) which was sanstactorily m by a single binding sne model. Ths prwided a KO of 81 + 2.2 pM for Me X, -P interaCLm. mese binding data were used 10 correctp.aminobenzamtdine excitation and emission spurs taken in lhs presence of factor X, for the free probe fluarescam contribution. Comparison of free (28) and a&sL hndp-eminobemamidine spectra at equd concentrations revealed that bindlng olp-aminobemamldine 10 factor Xa poduced a redshe Of lhe exCiMlon maximum from 293 nm Io 322 nm and a blubshifl d t h e emission maximum from 376 nm to 372 nm. Integration d me emission spectra funher indicated a 2Mfold enhancamant inp-ammobenzamidine fluorescence up^ binding d the probe to the factor X , active-siie (Fig 1A).
probe inhlbnion d factor X , hydrdysis of lha chrmogenic substrate. Spearoryme FX, . Fig. 1C shows a Dixon plot of the inhibitory enact of varyingp-aminobenzamidine mncentrations on me Miial VeloCnleS 01 substrate hydroiysis at three fixed substrate MmenuatiOns (32). me inlersBctim d these three CUNES at a common poim with an Ordtnate vdua indininguishabla from Vmax indicated lhatp-aminobenzamidine inhibited factor X. hydrolysis d substrate by a competM interaction a1 me polease ective-site. The abscissa of this ImersBcticn point obtained from a nonlmear lean squares analysis provided a KI of 63 f 5 pM, in reasonable agreement with the KO oblained in direct binding studies.
To confirm thalp-aminobenzamidine was binding to the protease mive-S?e. we examined me pattern Of