Functional characterization of recombinant human meizothrombin and Meizothrombin(desF1). Thrombomodulin-dependent activation of protein C and thrombin-activatable fibrinolysis inhibitor (TAFI), platelet aggregation, antithrombin-III inhibition.

Recombinant human prothrombin (rII) and two mutant forms (R155A,R271A,R284A (rMZ) and R271A,R284A (rMZdesF1)) were expressed in mammalian cells. Following activation and purification, recombinant thrombin (rIIa) and stable analogues of meizothrombin (rMZa) and meizothrombin(desF1) (rMZdesF1a) were obtained. Studies of the activation of protein C in the presence of recombinant soluble thrombomodulin (TM) show TM-dependent stimulation of protein C activation by all three enzymes and, in the presence of phosphatidylserine/phosphatidylcholine phospholipid vesicles, rMZa is 6-fold more potent than rIIa. In the presence of TM, rMZa was also shown to be an effective activator of TAFI (thrombin-activatable fibrinolysis inhibitor) (Bajzar, L., Manuel, R., and Nesheim, M. E. (1995) J. Biol. Chem. 270, 14477-14484). All three enzymes were capable of inducing platelet aggregation, but 60-fold higher concentrations of rMZa and rMZdesF1a were required to achieve the effects obtained with rIIa. Second order rate constants (M−1·min−1) for inhibition by antithrombin III (AT-III) were 2.44 × 105 (rIIa), 6.10 × 104 (rMZa), and 1.05 × 105 (rMZdesF1a). The inhibition of rMZa and rMZdesF1a by AT-III is not affected by heparin. All three enzymes bound similarly to hirudin. The results of this and previous studies imply that full-length meizothrombin has marginal procoagulant properties compared to thrombin. However, meizothrombin has potent anticoagulant properties, expressed through TM-dependent activation of protein C, and can contribute to down-regulation of fibrinolysis through the TM-dependent activation of TAFI.

Prothrombin is the inactive precursor of thrombin, a multifunctional serine protease that plays a central role in hemostasis. During the activation of prothrombin to thrombin by the prothrombinase complex, the active intermediate meizothrombin is generated initially (1)(2)(3)(4). Meizothrombin is formed by the hydrolysis of Arg 320 -Thr 321 by factor Xa (2,5). Unlike thrombin, meizothrombin retains the two kringle domains and the ␥-carboxyglutamic acid (Gla) 1 domain which is involved in the calcium-dependent interaction of prothrombin with phospholipid surfaces (6,7). Cleavage of the peptide bond between Arg 155 -Ser 156 in meizothrombin by thrombin-like activity generates another enzymatically active intermediate called meizothrombin(desF1) (8).
When the properties of meizothrombin and meizothrombin(desF1) are compared to those of thrombin, numerous similarities and differences emerge. Bovine meizothrombin and meizothrombin(desF1) have activities comparable to thrombin toward small substrates such as S-2238 (2,9,10). However, their activities toward macromolecular substrates such as fibrinogen or platelets are reduced (9,10). As measured in Factor V-deficient plasma, bovine meizothrombin and meizothrombin(desF1), respectively, exhibit 1.4% and 12% of the activity of thrombin in the activation of Factor V (9). In the presence of negatively charged phospholipids, however, recombinant human meizothrombin (prothrombin R155A) is a potent Factor V activator (11). Although bovine meizothrombin (9) and meizothrombin(desF1) (12) can activate protein C (PC) in a thrombomodulin (TM)-dependent reaction, several studies have been unable to demonstrate significant binding between cell-surface recombinant TM and recombinant meizothrombin derived from the active site serine-to-alanine mutant human prothrombin (13,14). Therefore, the conclusion was reached that meizothrombin is unlikely to be an important TM-dependent protein C activator (13). The inhibition of purified meizothrombin by antithrombin III (AT-III) has not been studied; however, meizothrombin(desF1) (15)(16)(17) is inhibited by antithrombin III with a second order rate constant approximately 2-3-fold lower than that for ␣-thrombin, and the interaction is not promoted by heparin (16,17). Active-site blocked meizothrombin was also shown to have no heparin-neutralizing properties as opposed to active-site blocked thrombin (18).
Evidence obtained in vitro indicates that meizothrombin is a major intermediate during the clotting of whole blood (19). Similarly, a study ex vivo using human umbilical vein endothelial cells demonstrated the formation of meizothrombin and the accumulation of meizothrombin(desF1) during prothrombin activation (20). Meizothrombin was also shown to have vasoconstrictive activity 5-fold greater than that of ␣-thrombin (21).
Together, these studies suggest that meizothrombin and/or meizothrombin(desF1) produced in vivo have functions different from those of thrombin. Studies of these two intermediates in vitro, however, are complicated by their transient nature resulting in their conversion to thrombin. Previously, we reported the expression in mammalian cells of the recombinant human prothrombin derivative rMZ (R155A,R271A,R284A) (4), which can be activated to a stable form of meizothrombin (rMZa). A second prothrombin derivative rMZdesF1 (R271A,R284A) (10) which can be activated to a stable form of meizothrombin(desF1) (rMZdesF1a) was also produced. The molecules were stabilized via arginine to alanine mutations at one of the two factor Xa cleavage sites and either one or both of the thrombin cleavage sites. The enzymatic activities of rMZa and rMZdesF1a toward fibrinogen and small substrates (4,10) were similar to those described by others (2,9) in studies of native meizothrombin. In addition, the Ca 2ϩ and phospholipid binding properties of prothrombin are retained by rMZa (4). Thus, these activated mutants are useful surrogates for meizothrombin and meizothrombin(desF1). Consequently, the present studies were undertaken to analyze their functional properties and compare them to those of recombinant human thrombin (rIIa).

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium, Ham's F-12 (1:1), Opti-MEM, Ultroser G, and newborn calf serum were purchased from Life Technologies, Inc. Methotrexate was from David Bull Laboratories (Mulgrave, Victoria, Australia), and vitamin K 1 was purchased from Sabex (Boucherville, Quebec) or Abbott Laboratories (Montreal, Quebec). The BHK cell line and pNUT vector were gifts from Dr. Richard Palmiter (Howard Hughes Medical Institute, University of Washington). The chromogenic substrates S-2238 and S-2366 were from Helena Laboratories (Mississauga, Ontario). Phospholipid vesicles (75% PC/ 25% PS (w/w)) were prepared as described previously (22). Human Factor X (1), Factor V (23), fibrinogen (24), and protein C (25) were purified and activated as described previously. TAFI (thrombin-activatable fibrinolysis inhibitor) was isolated by anion exchange, gel filtration, and plasminogen-Sepharose affinity chromatography as described previously (26). The purity of all proteins was verified by SDS-PAGE analysis. Additionally, thrombin activity was absent from the protein C and TAFI preparations as assayed by S-2238. Antithrombin III was provided by Alpha Therapeutics and further purified by affinity chromatography on heparin-Sepharose (27). Recombinant human thrombomodulin (TM LEO ) was purified from Chinese hamster ovary cells and characterized as described (28). This protein is a soluble form of TM comprising amino acids 4 to 490, with the following mutations M388L, R456G, H457Q, and S474A, and does not contain the chondroitin sulfate side chain (29). Hirudin was purchased from Calbiochem (San Diego, CA). Heparin (grade I) (used in platelet preparation), prostaglandin E 1 , and apyrase (grade V) were from Sigma. Heparin PMH 146000 (used for the AT-III study) was from Fisher (Whitby, Ontario). DAPA was prepared as described (30). The prothrombin activator of Echis carinatus venom (Sigma) was purified by anion-exchange chromatography and preparative electrophoresis in polyacrylamide as described previously (3). The Bio-Rad protein assay dye was from Bio-Rad Laboratories.
Protein Expression-The construction of the new pNUT-hMZdesF1 expression vector was accomplished by isolation and ligation of restriction digest fragments from the wild type prothrombin construct and the rMZ construct (4). Thus, a BstEII fragment from pNUT-hII encoding part of the promoter region and amino acids Met Ϫ43 to Ala 183 of the human wild type prothrombin was ligated into the BstEII-digested pNUT-hMZ construct encoding amino acids Val 184 to Glu 579 of rMZ and the pNUT vector. The cDNAs encoding wild type human prothrombin, hMZ and hMZdesF1, cloned into the expression vector pNUT, were stably transfected into baby hamster kidney (BHK) cells as described previously (4). For protein production, the cells were seeded in roller bottles and cultured until confluence in Dulbecco's modified Eagle's medium-F12 (1:1) supplemented with 5% newborn calf serum and 440 M methotrexate. After 3-4 days, the cells were washed with Dulbecco's modified Eagle's medium-F12 and subsequently cultured in 0.5% Ultroser G or Opti-MEM (supplemented with 50 M ZnCl 2 ) containing 10 g/ml vitamin K 1 . With the exception of the 1st day, the medium was collected every day (150 -200 ml), pooled, and stored at 4°C. The conditioned medium was concentrated using an Amicon CH2 concentrator with a S1Y10 spiral cartridge, and rMZ was purified as described previously (4). rMZdesF1 was purified similarly except that the chromatography using a Ca 2ϩ gradient was omitted.
Activation of the Recombinant Proteins-Wild type recombinant human prothrombin (rII) was activated by addition of the prothrombinase complex (5 nM Factor Xa, 5 nM Factor Va, 10 M phospholipid vesicles, and 5 mM CaCl 2 ) directly to the conditioned medium (Opti-MEM). The activation mixture was then purified on an SPC-50 column (2 ml) equilibrated with 20 mM HEPES, 150 mM NaCl, pH 7.4 at 22°C. Recombinant thrombin (rIIa) was eluted from the column with 20 mM HEPES, 0.5 M NaCl, pH 7.4. Fractions containing rIIa were identified by amidolytic activity using S-2238. This procedure yielded pure rIIa as judged by SDS-PAGE, with 100% activity as measured by active-site titration. Both rMZ and rMZdesF1 were activated with ecarin. Briefly, ecarin (1.5 g/ml) was added to the pure protein in 20 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 at 22°C. After 45 min, the activation mixture was chromatographed at 22°C on a column of benzamidine-Sepharose (2 ml) equilibrated in 20 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 , pH 7.4. After a wash, the enzymes were eluted with the same buffer containing 10 mM benzamidine. Fractions containing the proteases were identified by the Bio-Rad protein assay. The pooled fractions containing the active proteins were dialyzed against 20 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 , pH 7.4, and then concentrated, if necessary, using a Centricon PM30 membrane (Amicon, Danvers, MA). For rMZdesF1a, the removal of fragment 1 was complete with no zymogen remaining, as indicated by SDS-PAGE analysis under reducing conditions. This suggests that the feedback cleavage at position Arg 155 -Ser 156 is carried out by meizothrombin/meizothrombin(desF1) and does not require the presence of endogenously formed ␣-thrombin. Both rMZa and rMZdesF1a were found to be 80 -85% active.
Active-site Titration and Displacement of DAPA by Hirudin-The concentration of active proteases (rIIa, rMZa, and rMZdesF1a) was determined by titration with PPA⅐CK. Small aliquots (3.2 l) of PPA⅐CK (5 M) were added to a protein sample (1.6 ml at 100 nM) containing 200 nM DAPA, and the decrease in fluorescence due to the displacement of DAPA from the active site was monitored ( ex ϭ 280 nm, em ϭ 545 nm) in a Perkin-Elmer LS-50B fluorescence spectrophotometer. Plots of fluorescence intensity versus the concentration of PPA⅐CK were constructed, and extrapolation to the baseline fluorescence values yielded the active-site concentrations of each protease. The interaction of the recombinant proteins with hirudin was measured similarly. Briefly, small aliquots (3.2 l) of hirudin (5 M) were added to a protein sample (1.6 ml at 100 nM) containing 200 nM DAPA, and the decrease in fluorescence was monitored ( ex ϭ 280 nm, em ϭ 545 nm). The displacement of DAPA by hirudin was considered to be the result of two competing equilibrium binding interactions, E ϩ H º E⅐H and E ϩ D º E⅐D, characterized by dissociation constants K H and K D (E, H, and D represent enzyme, hirudin, and DAPA, respectively). By utilizing the binding equations , the following equation was derived, which relates fluorescence intensity (proportional to [E⅐D] with proportionality constant C); the concentrations of hirudin, enzyme, and DAPA and the dissociation constants for the enzyme-DAPA and the enzyme-hirudin interactions.
ϭ 200 nM, K D ϭ 20 nM, and [E 0 ] ϭ 100 nM, the above equation can be written as follows: . The above equation with H 0 substituted by a ϫ H 0 to compensate for slight differences between the nominal and active concentrations of hirudin was fit by nonlinear regression to the intensity versus [hirudin] data with C, a, and K H as fit parameters, thereby yielding the dissociation constant for the enzyme, hirudin interactions.
Platelet Aggregation-Platelets were prepared from freshly drawn blood, using a modification of the method described by Mustard et al. (31). Platelets were washed once in TA buffer (137 mM NaCl, 2.7 mM KCl, 12 mM NaHCO 3 , 0.42 mM NaH 2 PO 4 , 1.0 mM MgCl 2 , 5.6 mM glucose, 2.5 mg/ml bovine serum albumin, pH 7.35) containing 30 g/ml apyrase, 5 M prostaglandin E 1 , 28 units/ml heparin, and washed again in the same buffer without the heparin. The platelets were finally resuspended in TA buffer containing 50 mM HEPES, 3 g/ml apyrase (HTA). The number of platelets in a small aliquot (10 l) was deter-mined in a Baker System 9000 Cell Counter. Aggregation studies were carried out in a single-chamber Payton Aggregation Module (Payton Scientific, Buffalo, NY) using siliconized glass cuvettes. Gain settings were adjusted to give 90% transmittance for HTA buffer and 10% for the starting platelet suspension. Varying amounts of rIIa, rMZa, and rMZdesF1a were added to platelets (1 ϫ 10 8 /ml) in HTA buffer containing 2 mM CaCl 2 . The aggregation of the platelets was followed continuously by measuring the percentage of the increment in light transmittance with time. The platelet preparation was tested for its response to rIIa (0.8 nM) at the beginning and the end of the experiment.
Protein C Activation-Human protein C, TM, and PCPS vesicles were incubated in various combinations and concentrations in 20 mM HEPES, 150 mM NaCl, 5 mM CaCl 2 , pH 7.4, for 10 min at 37°C in the wells of a microtiter plate. The reactions were initiated by the addition of rIIa, rMZa, or rMZdesF1a in the same buffer. Final concentrations were: protein C, 1.0 M; TM, 0 -80 nM, PCPS vesicles, 0 -400 M, and the enzymes, 0.5 or 2.0 nM. After 10 min, protein C activation was terminated by the addition of 150 l of a solution containing 0.33 mM S-2366, 10 mM EDTA, and 30 M DAPA in 20 mM HEPES, 150 mM NaCl, pH 7.4. The conversion of S-2366 by activated protein C (APC) was followed by determining the absorbance at 405 nm, at 1-min intervals for 30 min. A control experiment performed in the absence of recombinant protease showed no conversion of the substrate. APC concentrations were calculated from the rates of change of A 405 with time in conjunction with a standard curve generated with purified APC. Rates of protein C activation were calculated from the concentration of APC measured after 10 min of incubation. The data were subsequently fit by non-linear regression analyses to the saturation equation r ϭ R max ⅐[TM]/(K d ϩ [TM]) using SYSTAT software (Systat, Evanston, IL). In this equation, r ϭ rate of protein C activation, R max ϭ rate at saturating TM, and K d is the TM concentration when r ϭ R max /2. Similarly, rates versus PCPS concentrations were fit to the equation where R 0 is the rate in the absence of PCPS, ⌬R max is the maximum increment in rate achieved with PCPS, and K d indicates the PCPS level needed to obtain half the maximum increment.

RESULTS
As shown by Fig. 1 and the parameters derived from these experiments in Table I, TM stimulates the activation of protein C not only by rIIa, but also by rMZa and rMZdesF1a. The reactions were performed in the presence of Ca 2ϩ (5 mM) and in the absence of PCPS vesicles. Saturation of rates with respect to the concentration of TM was observed with all three enzymes, and the concentrations required to achieve one-half the maximum rate ranged from 24 -35 nM (for rIIa and rMZdesF1a) to 63-72 nM (for rMZa). The rates at saturation ranged from 5-6 mol of protein C converted per mol of enzyme per min with rIIa and rMZa and approximately 4 with rMZdesF1a. The K d values reported for the thrombin-TM interaction by others (28) are generally lower than those observed here. This is most likely a consequence of the higher salt concentration (0.15 M versus 0.10 M) used here. 2 The effects of PCPS vesicles on the rates of protein C activation by the TM complexes of the three enzymes are depicted 2 C. T. Esmon, personal communication.

FIG. 1. Activation of protein C by the recombinant proteases, in the presence of thrombomodulin.
The enzymes rIIa (E), rMZdesF1a (å), and rMZa (q) (0.5 nM) were incubated in 5 mM CaCl 2 with protein C (1.0 M) and soluble TM at the concentrations indicated on the horizontal axis for 10 min at 37°C. Following termination of protein C activation by addition of EDTA and DAPA, APC concentrations were determined from the rates of S-2366 hydrolysis. Rates are expressed as mol of APC formed per min per mol of enzyme. The lines were fit by non-linear regression analyses, and the results are tabulated in Table I. in Fig. 2 and Table II. Phospholipid vesicles had no effect on protein C activation by the rIIa-TM complex. In contrast, upon addition of PCPS vesicles, the rate approximately doubled with rMZdesF1a-TM and increased by approximately 13-fold with rMZa-TM. At saturating levels of PCPS, rMZa-TM thus catalyzes protein C activation at a rate which is ϳ6-fold greater than that for rIIa-TM. These data demonstrate not only that rMZa and rMZdesF1a can activate protein C in the presence of TM, but also that the rate achieved in the presence of phospholipid vesicles and rMZa appreciably exceeds that observed with rIIa.
A second substrate was recently identified for the thrombin-TM complex: TAFI (32). We therefore investigated the ability of rMZa to activate TAFI. The data displayed in Fig. 3 indicate the effect of TM on the rMZa-catalyzed proteolytic activation of TAFI. The experiments were performed at a range of initial concentrations of TAFI (0.26 to 1.75 M). As the data indicate, saturation of rates was observed with respect to the TAFI concentration at all TM concentrations. The rates also showed saturation with respect to the TM concentration at all of the TAFI concentrations. The K m values for TAFI did not vary with TM concentrations, but V max values showed saturation. All of the data therefore were fit to the single equation indicated under "Experimental Procedures." The lines in Fig. 3 were obtained from regression analyses of the data. The fit was such that the residuals were distributed randomly and no individual datum deviated by more than 15% of the value predicted from the rate equation, thus indicating that the equation and implied model represent the data well. The values of parameters that were returned by the non-linear regression analysis were: K m ϭ 0.7 M, k cat ϭ 0.08 s Ϫ1 , and K d ϭ 2.6 nM (for the rMZa-TM interaction). The K d value inferred from the kinetics of TAFI activation is lower than that for protein C activation. This may be due to the difference in temperature (37°C for protein C versus 22°C for TAFI) or may reflect differential contributions of the two proteins (TAFI and protein C) in the formation of the ternary complex. The model which fits these data (3) implies a mechanism whereby rMZa can interact with either TM or TAFI to form the respective binary complexes, and then the third component can be added to form a ternary rMZa-TM-TAFI complex, in which activation of TAFI occurs. The primary effect of TM is to increase the k cat of the reaction. This shows that rMZa is a potent activator of TAFI when complexed to TM.
Inhibition of thrombin by AT-III is a physiological control mechanism. Therefore, we investigated the kinetics of the inhibition of rIIa, rMZa, and rMZdesF1a by AT-III (Fig. 4). Under the conditions used in this experiment, inhibition was first order and a linear relationship was observed between the pseudo-first order rate constants and AT-III concentrations. The value of the second order rate constant was 2.44 ϫ 10 5 M Ϫ1 min Ϫ1 for the inhibition of rIIa by AT-III alone. The second order rate constants for inhibition of both rMZa and rMZdesF1a were lower (6.10 ϫ 10 4 and 1.05 ϫ 10 5 M Ϫ1 min Ϫ1 , respectively). rMZa (which retains the Gla domain) was twice as resistant as rMZdesF1a to inhibition by AT-III, suggesting that the fragment 1 domain may further reduce the interaction between the two molecules. The inhibition of the three recombinant enzymes also was monitored by measuring the decrease in fluorescence due to the displacement of the active-site fluorescent probe DAPA by AT-III (Fig. 5). Under these conditions, upon the addition of heparin, a rapid decrease in the concentration of the rIIa-DAPA complex was observed, while the concentrations of the DAPA complexes with rMZa and rMZdesF1a were unchanged. These data confirm reports by others that inhibition of thrombin by AT-III is promoted by heparin, whereas the inhibition of meizothrombin (2,9) or meizothrombin(desF1) (16,17) is not.
Hirudin is a highly specific thrombin inhibitor currently in clinical trials for the prevention of thrombosis. We investigated whether rMZa and rMZdesF1a would be sensitive to hirudin inhibition. The interactions of hirudin with the three recombinant enzymes was analyzed by measuring the decrement in fluorescence as hirudin displaces DAPA from the active site (Fig. 6). rIIa, rMZa, and rMZdesF1a were similarly affected by the leech inhibitor. Although the rMZa-DAPA complex exhibits greater fluorescence intensity than rIIa-DAPA or rMZdesF1a-DAPA, all three enzymes yielded baseline fluorescence at the same hirudin concentration. The dissociation constants (pM) determined by non-linear regression analyses for the interactions of hirudin with the three enzymes were: 134 Ϯ 36 (rIIa), 355 Ϯ 62 (rMZa), and 231 Ϯ 37 (rMZdesF1a). These results suggest that the hirudin binding sites are similarly or possibly equally accessible in meizothrombin, meizothrombin(desF1), and thrombin.
Thrombin is a potent physiological activator of platelet aggregation. We tested whether the three recombinant enzymes were capable of inducing platelet aggregation. rMZa and rMZdesF1a were both approximately 60-fold less effective than rIIa with respect to the concentration required to give maximal rates of aggregation. As shown in Fig. 7, 0.8 nM rIIa induced a maximal rate of platelet aggregation (22% T/min), and approximately 50 nM concentrations of either rMZa or rMZdesF1a   Table II. were required to generate aggregation rates of 12.5% T/min and 7.5% T/min, respectively. With each of the enzymes, a decrease in the rate of platelet aggregation was observed at concentrations 1 order of magnitude greater than those required to give maximal rates of aggregation. Comparatively, rMZa and rMZdesF1a are weak platelet activators and therefore less procoagulant than rIIa.

DISCUSSION
In this study, we investigated the activities of rMZa and rMZdesF1a to determine their relative effects on procoagulant and anticoagulant reactions. Table III summarizes these properties relative to rIIa. Both rMZa and rMZdesF1a show poor procoagulant function as demonstrated by their fibrinogen clotting and platelet aggregation activities. Although both enzymes were capable of initiating platelet aggregation, peak rates of aggregation required 60-fold higher concentrations of rMZa or rMZdesF1a compared to rIIa. In addition, these peak rates were lower than that of rIIa. These results suggest that rMZa and rMZdesF1a interact less favorably with the thrombin receptor described by Vu et al. (33) and/or act through a different receptor.
In contrast, both rMZa and rMZdesF1a exhibit a TM-dependent maximum rate of protein C activation that is similar to that of rIIa, in the absence of phospholipids. In the presence of phospholipid vesicles, the maximum rate for rMZa is 6-fold higher than that of rIIa. The TM-dependent activation of TAFI was catalyzed at a significant rate by rMZa, but still only 10% that of rIIa-TM (32). Both rMZa and rMZdesF1a were inhibited by AT-III although 2-to 4-fold less efficiently than rIIa.
Meizothrombin and thrombin display both procoagulant and anticoagulant activities. When rMZa is compared to rIIa, however, its anticoagulant properties predominate. One possible exception to this interpretation comes from the data reported by Tans et al. (11) which show that a recombinant form of human meizothrombin (prothrombin R155A) activates factor V at 4 times the rate of thrombin in the presence of 90% PC/10% PS phospholipid vesicles, but at a much lower rate (70-fold) in their absence. Furthermore, our data on protein C and TAFI activation indicate TM-dependent potentiation of rMZa activity. These observations appear inconsistent with the report by Wu et al. (13) who were unable to demonstrate significant binding between recombinant active-site serine-to-alanine mutant meizothrombin and cell surface recombinant TM. Differences in the TM species may account for some of the discrepancies between our results and those of others. Our study employs a recombinant soluble form of thrombomodulin containing four mutations and lacking chondroitin sulfate, whereas Wu et al. (13) utilized cellular thrombomodulin. Thus, results with soluble thrombomodulin and synthetic vesicles probably cannot be reliably extrapolated to infer characteristics of protein C activation on cellular surfaces by meizothrombin. Further studies with cell surface or membrane-bound thrombomodulin would be needed to rationalize the apparent discrepancy. A possible source of error in our experiment could be a contamination of both protein C and TAFI. We feel that this is not the case because the control experiments without enzyme showed no activation of protein C or TAFI or endogenous amidolytic activity. In addition, the potentiation of protein C activation by phospholipids with rMZa, but not rIIa, clearly indicates that catalysis is due to rMZa rather than contaminating thrombin.
Our data provide some insight into the functionality of the fragment 1 domain and the two potential anion binding exosites of meizothrombin. The marked stimulation of protein C activation by PCPS vesicles with rMZa, but not rMZdesF1a or rIIa, clearly indicates the importance of the Gla domain for potentiation of the reaction by phospholipids. Furthermore, assuming that rMZa, rMZdesF1a, and rIIa interact similarly with TM, our results imply that exosite 1 is available on rMZa and rMZdesF1a. This is consistent with the work by Ni et al. (34) who demonstrated by NMR spectroscopy the partial accessibility of exosite 1 on prothrombin, prethrombin-1, and meizothrombin(desF1).
The reduced second order rate constants for inhibition of rMZa and rMZdesF1a by AT-III, compared to that of rIIa, confirms previous observations (15) that the fragment 2 domain, either covalently attached to the protease domain or non-covalently associated with it, attenuates the interaction  between AT-III and the protease domain. As reported previously (2,16,17), the inhibition of meizothrombin(desF1) is not promoted by heparin. The present work shows that this is also true for rMZa. These observations imply that the covalently associated fragment 2 domain blocks in the protease domain the anion binding exosite 2 which has been shown in thrombin to interact with heparin (35)(36)(37). Thus, the fragment 2 domain of rMZa and rMZdesF1a prevents formation of the ternary AT-III-protease-heparin complex required for stimulated inhibition by AT-III (38). The reduced rate constant for inhibition of rMZa by AT-III, compared to that of thrombin, suggests that rMZa would have a longer half-life in circulation than thrombin, and inhibition of rMZa and rMZdesF1a would not be influenced by heparin. In contrast, rMZa, rMZdesF1a, and rIIa would likely be inhibited similarly by hirudin in vivo.
Our studies, along with those of others, show that both rMZa (and presumably native meizothrombin) and thrombin possess procoagulant, anticoagulant, and antifibrinolytic activities. The latter two activities are expressed through the TM-depend-ent activation of protein C and TAFI, respectively. Thrombin and meizothrombin clearly are not equipotent in these three activities. Meizothrombin is only 10% or less effective as thrombin in expressing procoagulant and antifibrinolytic activities, as defined by fibrinogen clotting, platelet activation, and TAFI activation. However, it expresses substantial anticoagulant activity, as defined by protein C activation, which approaches that of thrombin in the absence of phospholipids and exceeds it severalfold in the presence of phospholipids. These properties of meizothrombin suggest that it plays a significant role in the control of hemostasis, perhaps different from that of thrombin. This would be especially so in small vessels which are dense with TM. Thus, because prothrombin activation generates two enzymes, each with unique properties, two distinct states are conceivably possible. One state, extant when meizothrombin predominates, would be relatively anticoagulant and profibrinolytic. This state also would likely be self-limiting because of negative feedback through the anticoagulant pathway. The other state, extant when thrombin predominates, would be relatively procoagulant and antifibrinolytic. If mechanisms ex- FIG. 4. Inhibition by antithrombin-III. The proteases were incubated with AT-III (1.29 M in this instance) at 22°C for 0 to 30 min, and the remaining activity was measured at the indicated times by assays with S-2238 for rIIa (E), rMZdesF1a (å), and rMZa (q). Pseudofirst order rate constants are plotted versus the AT-III concentration in the inset. These plots yielded second order rate constants (M Ϫ1 ⅐min Ϫ1 ) of 2.44 ϫ 10 5 (rIIa), 1.05 ϫ 10 5 (rMZdesF1a), and 6.10 ϫ 10 4 (rMZa). ist in vivo which favor either meizothrombin or thrombin generation, then prothrombin activation could potentially lead to, or be switched toward, two distinct end points, one of which could be characterized as anticoagulant and profibrinolytic, and the other as procoagulant and antifibrinolytic.   Fig. 3, and plasmaderived thrombin (Bazjar et al. (32)).
f ND, not determined. g From second order rate constants, Fig. 4.