Calcium ion modulation of meizothrombin autolysis at Arg55-Asp56 and catalytic activity.

When a recombinant variant of prothrombin with the cleavage site mutations R155A, R271A, and R284A (rMZ) is exposed to either prothrombinase or ecarin, a form of meizothrombin (rMZa) is generated that is stable for weeks in the presence of Ca2+ (Côté, H. C. F., Stevens, W. K., Bajzar, L., Banfield, D. K., Nesheim, M. E., and MacGillivray, R. T. A. (1994) J. Biol. Chem. 269, 11374-11380). In the absence of Ca2+ however, rMZa is rapidly cleaved within a disulfide bonded loop in the F1 domain at Arg55 in the sequence RTPR downward arrowDKL, yielding a molecule with 3 chains joined by two disulfide bonds (rMZa*). Cleavage kinetics are first order regardless of the rMZa concentration, indicating an intramolecular cleavage. This cleavage does not occur at Ca2+ concentrations in excess of 1.0 mM. To assess the role of the F1 domain in rMZa activity, another variant lacking the R155A mutation (rMZdesF1) was expressed, which when activated yields meizothrombin lacking the F1 domain (rMZdesF1a). Rates of hydrolysis of the tripeptide substrate S2238 by rMZa or rMZa* increase from 60% to 90% that of recombinant thrombin as Ca2+, Mg2+, or Mn2+ concentrations are varied from 0 to 10 mM. Km and kcat values for rMZa in the absence and presence of 5 mM Ca2+ are 1.9 and 2.2 microM and 65 and 105 s-1. TAME esterase activity of rMZa also increases with 5 mM Ca2+. No such metal ion-dependent effects are obtained with either thrombin or rMZdesF1a. Fibrinogen clotting activities, relative to that of thrombin, increase in a manner analogous to those obtained with small substrates, for rMZa and rMZa* but not rMZdesF1a. Complexes of the active site probe dansylarginine N-(3-ethyl-1,5-pentanediyl)amide with rMZa and rMZa*, but not thrombin or rMZdesF1a exhibit large cation-dependent decreases in fluorescence intensity, suggesting that metal ion binding in the F1 domain alters the environment of the probe at the active site. These results indicate that in the absence of divalent cations, the activity of rMZa is inhibited, perhaps by obstruction of the active site by the F1 domain, and that Ca2+ binding to the F1 domain modulates the properties of not only the F1 domain but also the protease domain.

Prothrombin is one of a group of plasma proteins involved in blood coagulation that require vitamin K-dependent carboxyl-ation of several amino-terminal glutamic acid residues for full biological function (1). In prothrombin, 10 of the first 33 residues are ␥-carboxyglutamic acid (Gla) 1 residues and define the Gla domain. This domain mediates the metal ion-dependent interaction of prothrombin and the other vitamin K-dependent coagulation factors with negatively charged membrane surfaces (2). Unlike other Gla-containing coagulation factors such as factor IX, factor X, and protein C, prothrombin loses its Gla domain following activation to the serine protease thrombin.
Prothrombin activation is catalyzed by the prothrombinase complex (3), composed of factor Xa (a serine protease), factor Va (an essential protein cofactor), negatively charged phospholipids, and calcium ion. Prothrombinase-catalyzed activation of prothrombin to thrombin results in peptide bond cleavage at two sites in the molecule (Fig. 1). Cleavage occurs initially at Arg 320 , within a disulfide loop, forming the active intermediate meizothrombin (4). Further cleavage at Arg 271 produces thrombin and the amino-terminal activation peptide referred to as fragment 1.2. Thrombin itself can cleave prothrombin. Two major sites are recognized, one at Arg 155 , between the kringle domains of prothrombin, and a second at Arg 284 , within the A chain of the protease domain. Thrombin has also been reported to cleave prothrombin at a third site, at Arg 55 within a disulfide loop between the Gla and first kringle domains (5). Two of the intermediates of prothrombin activation, meizothrombin and meizothrombin des-fragment 1, may also catalyze the cleavage of prothrombin at these sites.
In contrast to thrombin, the prothrombin activation intermediate meizothrombin retains an intact Gla domain. Although naturally unstable, this intermediate was recently obtained as a stable product by introducing three mutations (R155A, R271A, R284A) which remove susceptibility to cleavage by prothrombinase at Arg 271 and by thrombin at the two feedback cleavage sites (6). If Arg 155 is retained, cleavage by either thrombin or meizothrombin produces a species without the Gla domain and the first kringle, known as meizothrombin desfragment 1 (Fig. 1).
Recent studies of the epidermal growth factor-like and Gla domains of coagulation factors IXa and X suggest that calcium ion binding in the Gla domain may affect properties of the other domains of these proteases. For example, calcium ion binding in the Gla domain of factor IXa alters the conformation of the serine protease domain (7), and a study of amino-terminal fragments of factor X (8) revealed that Ca 2ϩ binding in the Gla domain of factor X influences Ca 2ϩ binding in the amino-terminal of the epidermal growth factor-like domain. Previous studies of meizothrombin activity (6,9,10)  To understand the consequences of retaining the aminoterminal portion of the prothrombin molecule on meizothrombin activity, recombinant prothrombin and two site-directed mutants were prepared. One mutant was prothrombin (R155A,R271A,R284A) designated rMZ, which upon activation with either the prothrombinase complex or ecarin yields fulllength meizothrombin (6). A second mutant prothrombin (R271A,R284A) designated rMZdesF1 yields meizothrombin des-fragment 1 upon activation. These and recombinant prothrombin (rII) were expressed, isolated, and activated, and their activities against small substrates and fibrinogen were compared in the presence of varying amounts of Ca 2ϩ .

EXPERIMENTAL PROCEDURES
Expression Vectors-For expression in mammalian cells, the cDNAs for rII and rMZ were ligated into the pNUT expression vector (11) downstream of the zinc-inducible mouse metallothionein promoter and upstream of the human growth hormone polyadenylation signal, as described previously for rMZ (6). The expression vector for rMZdesF1 was prepared by digesting the pNUT rMZ vector with the restriction enzyme BstEII, liberating a 1-kilobase fragment corresponding to a small portion of the vector backbone and the first 690 base pairs of the rMZ cDNA. This fragment, containing the R155A mutation, was then replaced with the analogous fragment obtained from pNUT containing the wild-type prothrombin cDNA, yielding a construct coding for a mutant prothrombin (R271A,R284A), with the thrombin cleavage site at Arg 155 restored.
Proteins-Human plasma factor V (12), factor X (13), and fibrinogen (13) were isolated, and factors V and X were activated as described previously. The prothrombin activator of Echis carinatus venom, ecarin, was purified from crude venom (Sigma) by anion exchange chromatography and polyacrylamide preparative gel electrophoresis as described previously (14). The recombinant proteins rMZ and rMZdesF1 were expressed and isolated using the methods described for rMZ by Côté et al. (6). Conditioned medium (Opti-Mem (Life Technologies Inc.), supplemented with 50 M ZnCl 2 ) was collected from stably transfected lines of BHK cells cultured at 37°C in roller bottles. Sodium citrate was added to a final concentration of 0.025 M, and the prothrombin was adsorbed with the addition of 1 M BaCl 2 (80 ml/liter) and subsequent formation of a barium citrate precipitate. The barium citrate pellet was eluted with 0.2 M EDTA, and the eluate was concentrated and dialyzed against 0.02 M Tris, 0.15 M NaCl, pH 7.4. Subsequent ion exchange using a salt gradient and pseudo-affinity chromatography with a calcium ion gradient on a Pharmacia Mono Q column yielded homogenous species (6). Samples of the recombinant proteins rMZ and rMZdesF1 (1 mg at 0.1 mg/ml) were activated by incubation with ecarin (1.5 g/ml) at 22°C in the presence of 5 mM CaCl 2 for approximately 30 min. The activation mixtures were passed at 22°C over a 2-ml column of benzamidine Sepharose, previously equilibrated in 0.02 M HEPES, 0.15 M NaCl, 5 mM CaCl 2 , pH 7.4. The column was washed with 20 ml of the equilibration buffer, and the enzymes were eluted with the same buffer supplemented with 10 mM benzamidine. Fractions containing rMZa or rMZdesF1a were identified using a Bio-Rad protein assay, mixing 90 l of each fraction with 20 l of the dye reagent, and monitoring color development at 595 nm. Fractions containing protein were pooled and dialyzed against 0.02 M HEPES, 0.15 M NaCl, 5 mM CaCl 2 , pH 7.4 and subsequently stored at 4°C. Recombinant thrombin was isolated by activating rII directly in the conditioned media. Conditioned medium (40 ml of Opti-Mem) was treated with 5 nM factor Va, 5 nM factor Xa in the presence of 10 M phosphatidylcholine/phosphatidylserine (3:1) vesicles and 5 mM CaCl 2 . The activation mixture was passed over a 2 ml column of SP-C50 cation exchange resin, previously equilibrated with 0.02 M HEPES, 0.15 M NaCl pH 7.4, and the column was subsequently washed with 20 ml of the equilibration buffer. The rIIa was eluted from the column with 0.02 M HEPES, 0.5 M NaCl, pH 7.4. Fractions containing recombinant thrombin were identified by amidolytic activity against S2238. Aliquots of each fraction (10 l) were incubated with 90 l of 0.5 mM S2238, and color development at 405 nm was monitored. Fractions containing S2238 activity were pooled and stored at 4°C until used (within 5 days). The isolated rIIa migrated as a single band (nonreduced) on NaDodSO 4 -polyacrylamide minigels, which comigrated with thrombin prepared from human plasma.
Active Site Titration-Concentrations of rIIa, rMZa, and rMZdesF1a were determined by titrating an aliquot of the protein (1.6 ml, ϳ100 nM) with 5 M PPACk in the presence of 200 nM DAPA, while measuring the fluorescence energy transfer from the proteases to DAPA ( ex 280 nm, em 545 nm). Reaction of the protease with PPACk displaced DAPA from the active site resulting in a decrease in fluorescence intensity. The decrease was linear with respect to the concentration of PPACk, and extrapolation to the baseline fluorescence value yielded the active site concentration. The PPACk stock was dissolved shortly prior to use, and the concentration provided by the manufacturer was assumed to be correct. Titration of wild-type thrombin using this method yielded equivalent results to a p-nitrophenyl-p-guanidinobenzoate determination of active site titer (15).
Time Course of Proteolysis at Arg 55 -Stock solutions of rMZa at 25,50, and 100 g/ml were prepared in 0.02 M HEPES, 0.15 M NaCl, 2 mM CaCl 2 . Immediately prior to, and 4, 8, 12, 16, 20, and 36 min after addition of a small aliquot of 0.5 M EDTA, pH 7.4, to chelate the Ca 2ϩ present in the solution, aliquots equivalent to 2.5 g of protein were withdrawn and quenched with the addition of 2 volumes of 0.2 M acetic acid. The samples were reduced to dryness using a Savant Speed Vac, dissolved in gel loading buffer, and analyzed on NaDodSO 4 -polyacrylamide gradient (5-15%) minigels under reducing conditions (16). After staining with Coomassie Blue, the gels were destained and scanned with an LKB 2202 laser densitometer, and the traces were subjected to gravimetric analysis. For this experiment and subsequent experiments where EDTA was added to buffered solutions containing Ca 2ϩ , the pH of an equivalent mixture was monitored to ensure that the pH of the solution did not significantly change upon chelation of Ca 2ϩ .
Calcium The variant rMZdesF1 has the cleavage site mutations R271A and R284A and, upon activation, proceeds through meizothrombin to meizothrombin des-fragment 1 (rMZdesF1a). A stable recombinant form of meizothrombin (rMZ) was also produced with the additional mutation R155A. This species, when activated and stored in the presence of calcium ion, remains as meizothrombin (rMZa), but rapidly autocatalyzes cleavage at Arg 55 in the absence of calcium ion to form a three-chain form of meizothrombin (rMZa*).
at 22°C, quenched with acetic acid, reduced to dryness, run on gels, and analyzed as described above.
Amidolytic Assays-Aliquots of rIIa, rMZa, and rMZdesF1a (50 l) to yield final concentrations of 0.25 nM were incubated with 150 l of 0.02 M HEPES, 0.15 M NaCl, 0.01% Tween 80 with varying amounts (0 -10 mM) of either CaCl 2 , MgCl 2 , or MnCl 2 for 5 min at 37°C in microtiter wells in the sample compartment of a TiterTek Twinreader. The substrate S2238 (0.5 mM in 0.02 M HEPES, 0.15 M NaCl, 0.01% Tween 80, pH 7.4) was added, and initial rates of hydrolysis were determined by measuring the absorbance at 405 nm at 1-min intervals.
Esterase Activity-Samples (30 l) of rIIa or rMZa were diluted with 0.05 M Tris, pH 8.1, to a final concentration of 40 nM, and placed in quartz cuvettes in the sample compartment of a Perkin-Elmer Lambda 4B spectrophotometer. The reaction was initiated by the addition of 100 l of 0.01 M TAME in water, and the absorbance at 247 nm was monitored continuously. Initial slopes of substrate consumption were used to determine reaction rates, either in the absence or presence of 5 mM CaCl 2 .
Fibrinogen Clotting-Concentrations of rIIa (1 nM), rMZa/rMZa* (15 nM), or rMZdesF1a (3.6 nM) that yielded approximately equal clotting times in the presence of 5 mM CaCl 2 were established. Samples of concentrated stocks of these proteins (50 l in 0.02 M HEPES, 0.15 M NaCl, 5 mM CaCl 2 , 0.01% Tween 80, pH 7.4) were added to wells of a microtiter plate, containing 50 l of buffer supplemented with varying levels of EDTA to yield Ca 2ϩ concentrations between 0 and 10 mM. The mixture was allowed to incubate for 1 min, and then 100 l of 2 mg/ml fibrinogen in 0.02 M HEPES, 0.15 M NaCl, 0.01% Tween 80, pH 7.4, was added, and the time to onset of clotting was determined by measuring turbidity at 340 nm, at 10-s intervals. Results were analyzed by comparing times of onset of clotting for rMZa, rMZa*, or rMZdesF1a to the time obtained with rIIa at the equivalent calcium ion concentration.

Isolation and Characterization of Recombinant Proteins-
Recombinant prothrombin and the variants rMZ and rMZdesF1 were expressed in stably transfected lines of BHK cells. The prothrombin variants rMZ and rMZdesF1 were isolated and were homogeneous as judged by analysis of Coomassie Blue-stained NaDodSO 4 -polyacrylamide gradient minigels. These proteins were activated by the snake venom protease ecarin, and the active species was captured on benzamidine Sepharose. Migration on 5-15% NaDodSO 4 -polyacrylamide gradient gels of these proteins (reduced and nonreduced) was consistent with a single cleavage at Arg 320 for rMZa and additional loss of the F1 domain for rMZdesF1a (data not shown).
Stability of rMZa-Although rMZa was shown previously to be stable for up to 28 days in the presence of the components of the prothrombinase complex at 4°C (6) and is stable when purified for 6 months at Ϫ20°C in 50% glycerol, storage of the purified protein in the absence of Ca 2ϩ yielded rapid conversion to a species with a cleavage in the F1.2A chain visible on reduced gels (Fig. 2), but co-migrating with active meizothrombin under nonreducing conditions. Following NaDodSO 4 -polyacrylamide gel electrophoresis and blotting onto a polyvinylidene difluoride membrane, amino-terminal sequence analysis of the cleaved F1.2A chain revealed a new amino terminus with the sequence DKLAA, consistent with cleavage within a disulfide bonded loop, between the Gla domain and the kringle of Fragment 1. Cleavage of rMZa at Arg 55 , within the sequence RTPR2DKLA yields a species consisting of three peptide chains held together by two disulfide bonds; this species is referred to as rMZa*.
Further analysis revealed that the cleavage giving rise to rMZa* occurred only in the absence of Ca 2ϩ and was prevented by storage of rMZa in 5 mM CaCl 2 . Analysis of the time course of cleavage indicated that rMZa has a half-life in excess of 28 days in the presence of Ca 2ϩ , but only 8 min in the absence of Ca 2ϩ . It is highly unlikely that this cleavage is a result of a trace contaminant in the ecarin preparation, as identical results were obtained when the prothrombinase complex was used as the activator. The cleavage follows first order kinetics, with the percent conversion following Ca 2ϩ removal identical at three concentrations of the enzyme between 25 and 100 g/ml (Fig. 3). The rate of cleavage is not affected by equimolar concentrations of bovine F1 or a recombinant human prothrombin variant lacking factor Xa and thrombin cleavage sites (data not shown) nor are these species cleaved during the time course of the experiment. Although the concentration range studied was limited for technical reasons, modelling the meizothrombin proteolysis as an intermolecular reaction occurring with either high (Ն10 M), intermediate (1 M), or low (Յ0.1 M) affinity using Kinsim (17), failed to produce a reasonable fit to the data, whereas modelling the reaction as an intramolecular event (first order) produced an excellent global fit to the data (Fig. 3). These results indicate that the cleavage at Arg 55 is an intramolecular rather than intermolecular event.
Calcium Ion Requirement for Maintenance of Intact rMZa-Samples of rMZa were incubated in buffer containing varying levels of Ca 2ϩ for 60 min, and the extent of cleavage at Arg 55 was analyzed by NaDodSO 4 -polyacrylamide gel electrophoresis. Since the data in Fig. 3 (Fig. 4) revealed that the minimal Ca 2ϩ concentration required for stability was approximately 1 mM, with cleavage occurring rapidly below this concentration. To account for the possibility that rMZa could be proteolyzed in subsequent experiments to rMZa* during incubation at low calcium ion concentrations, rMZa* was isolated and included as a control.
Metal Ion Dependence of rIIa, rMZa, rMZa*, and rMZdesF1a Activity against the Amidolytic Substrate S2238 -Initial rates of S2238 cleavage for rIIa, rMZa, rMZa*, and rMZdesF1a in 0.5 mM EDTA and varying concentrations of CaCl 2 are displayed in Fig. 5. Relative to rIIa, both rMZa and rMZa* had initial activities of approximately 60%, while rMZdesF1a had a slightly higher activity (ϳ70%). While the activity of rMZdesF1a paralleled that of rIIa as the concentration of Ca 2ϩ increased, the activity of both rMZa and rMZa* increased from ϳ60% to ϳ90% that of rIIa over the range of Ca 2ϩ concentrations from 0 to 10 mM. Similar results were obtained when Mn 2ϩ and Mg 2ϩ were used in place of Ca 2ϩ (data not shown). The EC 50 for this change varied with the metal ion, with the pattern Mg 2ϩ (2 mM) Ͼ Ca 2ϩ (1 mM) Ͼ Mn 2ϩ (0.1 mM). For all species, a decrease in rate was observed at low concentrations (0 -0.35 mM) of divalent ion. This may be due to a direct metal ion effect on the protease domain, since the effects are not limited to rMZa or rMzdesF1a. The increases in activity ob-served for rMZa and its cleaved variant (rMZa*), however, suggest that either the Gla domain or the F1 kringle interacts with divalent metal ions and thus affects the catalytic efficiency of the protease domain. The change in rates of S2238 hydrolysis by rMZa and rMZa* in the presence of divalent metal ions can be attributed to alteration of the k cat toward S2238. K m values for rMZa toward S2238 remain unchanged (1.9 Ϯ 0.2 versus 2.2 Ϯ 0.2 M) in the presence or absence of 5 mM CaCl 2 , whereas the k cat increases from 65 Ϯ 3 s Ϫ1 in the absence of Ca 2ϩ to 105 Ϯ 5 s Ϫ1 in the presence of Ca 2ϩ .
Hydrolysis of the Esterolytic Substrate TAME by rMZa in the Presence and Absence of Ca 2ϩ -Effects similar to those observed with S2238 hydrolysis are observed with TAME hydrolysis. When monitored by absorbance of the product at 247 nm, cleavage of this ester substrate is enhanced in the presence of Ca 2ϩ with the turnover number increasing from 29 Ϯ 0.9 s Ϫ1 to 37 Ϯ 1.2 s Ϫ1 in the presence of 5 mM CaCl 2 .
Effect of Ca 2ϩ on Clotting of Fibrinogen by rMZa and rMZdesF1a Compared to rIIa-To determine whether Ca 2ϩ has similar effects on the activity of rMZa toward macromolecular substrates, the effect of varying Ca 2ϩ levels on clotting times of purified fibrinogen was examined. Since Ca 2ϩ has a direct effect on the rate of fibrin polymerization (18), the effects of Ca 2ϩ on clotting times obtained with rMZa and rMZdesF1a were examined relative to those obtained with rIIa. Concentrations of the three proteases required to achieve similar clotting times in 5 mM CaCl 2 were established, and the times to clot at these concentrations were determined at varying Ca 2ϩ levels. The results are presented in Fig. 6. The activity of rMZdesF1a toward fibrinogen was parallel to that of rIIa at all Ca 2ϩ concentrations, with a rate to onset of clot formation that was 27% that of rIIa. In contrast, both rMZa and rMZa* demonstrated a Ca 2ϩ -dependent doubling in rate, from 3.5% that of rIIa in the absence of Ca 2ϩ to 7% that of rIIa in the presence of Ca 2ϩ at levels greater than 2 mM, although the transition appears to require higher levels of Ca 2ϩ for rMZa*. The Ca 2ϩ concentration dependence of the increase in activity of rMZa toward fibrinogen closely parallels that observed in rates of S2238 hydrolysis.
Effect of Ca 2ϩ Concentration on Energy Transfer to DAPA by rIIa, rMZa, rMZa*, and rMZdesF1a-Changes in the active site environment upon exposure to Ca 2ϩ were inferred by tryptophan energy transfer to the fluorescent active site inhibitor DAPA. Tryptophan residues of rIIa, rMZa, rMZa*, or rMZdesF1a were continuously excited at 280 nm, and the fluorescence from DAPA was monitored at 545 nm. The traces presented in Fig. 7 represent changes in fluorescence with incremental additions of CaCl 2 . Whereas Ca 2ϩ has no effect on the DAPA fluorescence of either rIIa or rMZdesF1a, both rMZa and rMZa* demonstrate a considerable decrease (30%) in DAPA fluorescence intensity in response to Ca 2ϩ addition. Similar to the results obtained for fibrinogen clotting, the transition for the cleaved form of meizothrombin (rMZa*) appears to require higher levels of calcium. Because the DAPA concentration used in these studies was sufficient to nearly saturate the active sites of these enzymes, the change of fluorescence reflects a change in efficiency of energy transfer to the dansyl moiety of DAPA rather than binding affinity and thus is indic-ative of a change in active site environment. The initial fluorescence signals, obtained in the absence of Ca 2ϩ , from the DAPA-rMZa complex and the DAPA-rMZa* complex, are much higher (2.5-3-fold) than those of either the DAPA-rIIa or DAPA-rMZdesF1a complexes. DISCUSSION Previous studies of the binding of prothrombin and its activation intermediates to hirudin tail peptides (19), thrombomodulin, and exosite antibodies (20), have suggested that the fragment 1 domain may in some way block access to the primary anion binding exosite. Through measurements of the Ca 2ϩ dependence of meizothrombin autoproteolysis at Arg 55 , activity toward a variety of substrates, and energy transfer to the active site probe DAPA, this study provides further evidence for interactions between the fragment 1 and protease domains of meizothrombin.
Cleavage of prothrombin at Arg 55 has previously been reported following lengthy incubation of prothrombin with thrombin in the absence of Ca 2ϩ (5). In the present studies, however, the cleavage of rMZa at Arg 55 follows first order kinetics, with the time course of the reaction being identical at three different concentrations of rMZA, indicative of an intramolecular cleavage event. These results imply a direct interaction between the amino-terminal portion of the fragment 1 domain with the active site of meizothrombin. Cleavage of rMZa at Arg 55 occurs rapidly in the absence of Ca 2ϩ with a half-life under 10 min, but is profoundly attenuated in the presence of Ca 2ϩ .
Structural changes in the fragment 1 domain of prothrombin have been shown to accompany divalent metal ion binding by a number of methods including sedimentation velocity (21), intrinsic fluorescence (21), circular dichroism (22), and immunologic techniques (23). Due to the sensitivity of cleavage at Arg 55 to levels of Ca 2ϩ , and the location of the cleavage site within the fragment 1 domain, this event is likely modulated by metal ion binding in the Gla domain. Indeed, cleavage of prothrombin by thrombin at Arg 155 is also inhibited in the presence of Ca 2ϩ , and the protective effect has been attributed to conformational changes mediated by Ca 2ϩ binding to the Gla domain (24).
Although the three-dimensional structures for the apo-and Ca 2ϩ -bound forms of human fragment 1 are not available, those of bovine fragment 1 in the absence and presence of Ca 2ϩ have been determined (25,26). The structure of the Gla domain is disordered in the absence of Ca 2ϩ , but the disulfide loop containing the cleavage site at Arg 55 is defined in both structures. Interestingly the region around Arg 55 changes significantly upon Ca 2ϩ binding, with a trans to cis conformational change at Pro 54 and a 90°rotation of Arg 55 relative to its neighboring residues, enabling the formation of new salt links between Arg 55 and Gla 15 , Leu 19 , and Gla 20 . These changes at the Arg 55 cleavage site likely explain the insensitivity of Ca 2ϩbound meizothrombin to autoproteolysis. Another contributing factor to the stability of rMZa in the presence of Ca 2ϩ may be the disruption of interactions between the negatively charged residues of the Gla domain with positively charged portions of the protease domain (possibly the primary anion binding exosite) upon Ca 2ϩ binding. In any case, autocatalytic cleavage of rMZa at Arg 55 necessitates direct interaction between the fragment 1 and protease domains.
In addition to preventing autoproteolysis of rMZa, Ca 2ϩ increases the activity of rMZa toward small ester (TAME) and amide (S2238) substrates as well as increasing its fibrinogen clotting activity. Equivalent increases in activity are obtained with rMZa*, a species cleaved at Arg 55 , so the increase in activity cannot be simply attributed to a Ca 2ϩ -mediated protection from autoproteolysis. Neither rMZdesF1a, lacking the give approximately equal clotting times in the presence of 5 mM CaCl 2 were added to the wells of a microtiter plate containing varying concentrations of EDTA to give a range of Ca 2ϩ concentrations (0 -5 mM). Fibrinogen cleavage was initiated with the addition of 2 mg/ml fibrinogen in 0.02 M HEPES, 0.15 M NaCl, 0.01% Tween 80, pH 7.4, and the time to onset of clotting was determined by monitoring turbidity at 320 nm. Rates of fibrin formation relative to rIIa are plotted for rMZA (E) and rMZa* (q). The clot times for all species (rIIa, rMZa, rMZa*, and rMZdesF1 are ç, E, q, and É, respectively) are inset. fragment 1 domain, nor thrombin demonstrate this Ca 2ϩ -dependent increase in activity, suggesting that interactions of Ca 2ϩ with the fragment 1 domain are responsible for the altered activity of rMZa.
Similar effects on the activity of rMZa are observed with both small substrates (amide and ester) and fibrinogen, although fibrinogen binding involves considerably more complex interactions with the enzyme than do the small substrates. Thus, obstruction of the active site or reduced turnover for some other reason, rather than a change in substrate binding, is responsible for the reduced activity of rMZa in the absence of Ca 2ϩ . This interpretation is supported by the observation that values obtained for the K m for S2238 hydrolysis remain constant (1.9 versus 2.2 M) in the absence or presence of 5 mM Ca 2ϩ , while k cat values change (65 versus 105).
Titration of rMZa with Ca 2ϩ in the presence of the reversible fluorescent inhibitor DAPA results in a 30% decrease in fluorescence energy transfer, with a half-maximal effect occurring at 0.5 mM Ca 2ϩ . This result is similar to that obtained with meizothrombin prepared by activation of prothrombin with ecarin in the presence of dansyl-glutamylglycylarginyl chloromethyl ketone, yielding an active site blocked species with a covalently attached dansyl reporter group (27). In both cases, a change in the active site environment is revealed by a substantial decrease in probe fluorescence upon the addition of Ca 2ϩ . It is unlikely that Ca 2ϩ has a direct effect on the fluorescence of DAPA, as direct excitation of DAPA in the absence of protein yields less than 2% of the signal obtained in the presence of rMZa, and the fluorescence decrement observed with rMZa or rMZa* does not occur with either rMZdesF1a or rIIa. In addition, both rMZa and rMZa* display much greater fluorescence than either rIIa or rMZdesF1a in the absence of Ca 2ϩ (2.5-3fold). The decreases in DAPA fluorescence of rMZa and rMZa*, but not rMZdesF1a, upon the addition of Ca 2ϩ are consistent with a Ca 2ϩ binding event in the fragment 1 domain affecting the probe environment at the active site.
The Ca 2ϩ concentrations required for half-maximal increase in rMZa activity, protection from autoproteolysis, and decreases in DAPA fluorescence (0.5-1 mM) are significantly higher than those reported necessary for conformational change in the isolated fragment 1 domain (0.2-0.4 mM) (21,22). These levels of Ca 2ϩ are, however, consistent with those reported necessary to prevent the proteolytic cleavage of prothrombin at Arg 155 by thrombin (0.6 mM) (24,28). Divalent metal ions have been widely reported to cause selfassociation of prothrombin fragment 1 (21,29) and, under some conditions, prothrombin (29,30). Metal ion-dependent rMZa dimer formation is unlikely to be responsible for any of the effects observed in these studies, due to the high protein and metal ion concentrations necessary for dimerization to occur. In addition, Mg 2ϩ at the concentrations used in this study does not seem to be capable of inducing fragment 1 or prothrombin dimerization (21,29,30), but will support the increase in S2238 hydrolysis observed with rMZa.
Cleavage of rMZa at Arg 55 has little effect on the rate of S2238 hydrolysis at varying Ca 2ϩ concentrations. There is, however, a marked effect on the concentration necessary to produce half-maximal fibrinogen clotting activity, and decrease in DAPA fluorescence, with higher (1-1.5 mM versus 0.5 mM) levels necessary to produce these changes in the cleaved species. These differences may be indicative of a reduced ability of the cleaved form to undergo the necessary conformational changes to relieve the enzyme from inhibition at a given Ca 2ϩ concentration, with both fibrinogen clotting and DAPA fluorescence being more sensitive to interdomain interactions distant from the active site than S2238 hydrolysis.
These results together provide evidence for interaction between the fragment 1 and protease domains of rMZa at Ca 2ϩ levels below 1 mM. The consequences of this interaction include meizothrombin autoproteolysis at Arg 55 , as well as decreases in catalytic activity toward small ester (TAME) and small and large amide substrates (S2238, fibrinogen). Interestingly, the transition for these effects borders on the range of free Ca 2ϩ in the plasma (1-1.3 mM) (31). Finally, these results are indicative of flexibility in the structure of prothrombin, allowing for association between the amino-terminal fragment 1 and carboxyl-terminal serine protease domains.