Binding of substrate in two conformations to human prothrombinase drives consecutive cleavage at two sites in prothrombin.

Thrombin formation results from cleavage of prothrombin following Arg(271) and Arg(320). Both bonds are accessible for cleavage, yet the sequential action of prothrombinase on Arg(320) followed by Arg(271) is implied by the intermediate observed during prothrombin activation. We have studied the individual cleavage reactions catalyzed by prothrombinase by using a series of recombinant derivatives: wild type prothrombin (II(WT)) contained both cleavage sites; II(Q271) contained a single cleavable site at Arg(320); II(Q320) and II(A320) contained a single cleavable site at Arg(271); and II(QQ) was resistant to cleavage. Cleavage at Arg(320) in II(Q271) could account for the initial cleavage reaction leading to the consumption of either plasma prothrombin or II(WT), whereas cleavage at Arg(271) in either II(Q320) or II(A320) was found to be approximately 30-fold slower. Equivalent kinetic constants were obtained for three of the four possible half-reactions. Slow cleavage at Arg(271) in intact prothrombin resulted from an approximately 30-fold reduction in V(max). Thus, the observed pathway of bond cleavage by prothrombinase can be explained by the kinetic constants for the four possible individual cleavage reactions. II(Q320) was a competitive inhibitor of II(Q271) cleavage, and II(QQ) was a competitive inhibitor for each reaction with K(i) approximately K(m). The data are inconsistent with previous proposals and suggest a model in which substrates for each of the four possible half-reactions bind in a mutually exclusive manner and with equal affinity to prothrombinase in a cleavage site-independent way. Despite equivalent exosite binding interactions between all four possible substrates and the enzyme, we propose that ordered bond cleavage results from the constraints associated with the binding of substrates in one of two conformations to a single form of prothrombinase.

The formation of thrombin, a key reaction of the blood coagulation cascade, arises from specific and limited proteolysis of prothrombin (1). Although the serine proteinase, factor Xa, can catalyze this reaction, the rate of thrombin formation is greatly increased following its assembly into prothrombinase through interactions with membranes and factor Va (1)(2)(3). Prothrombinase is considered the physiologically relevant catalyst for rapid thrombin formation following the initiation of coagulation (2,4).
Thrombin formation results from cleavage of human prothrombin 1 following Arg 271 and Arg 320 (5,6). Initial cleavage at Arg 271 followed by cleavage at Arg 320 (Scheme I, Reactions 3 and 4) yields thrombin via the formation of prethrombin 2 and fragment 1.2 (P2 plus F1.2) 2 as intermediates. This cleavage pathway is evident in the action of factor Xa on prothrombin (7,8). Cleavage at Arg 320 followed by cleavage at Arg 271 (Scheme I, Reactions 1 and 2) results in thrombin formation via production of meizothrombin (mIIa) 3 as an intermediate. Within detection limits, bond cleavage in this order appears to quantitatively account for thrombin formation catalyzed by prothrombinase (9 -11). Prothrombinase cleaves the substrate in an apparently ordered fashion even though both Arg 320 and Arg 271 appear accessible to cleavage in prothrombin (9). The kinetic and molecular bases for these observations remain obscure.
Kinetic explanations for bond selectivity in prothrombin have been sought from studies using P2 plus F1.2 and mIIa as substrates (9 -13). The individual bonds in both intermediates are cleaved by prothrombinase (Scheme I, Reactions 2 or 4) with approximately equal catalytic efficiency (9 -12). Consequently, an explanation for ordered bond cleavage by prothrombinase requires that Arg 271 and Arg 320 in intact prothrombin are cleaved with different catalytic efficiencies. Therefore, formal consideration of the reactions of prothrombin activation requires a distinction to be made between cleavage at Arg 271 before and after Arg 320 cleavage (Arg 271 and Arg 271 *, Scheme I) or at Arg 320 before and after Arg 271 cleavage (Arg 320 and Arg 320 *, Scheme I). As previous measurements have established that recognition and cleavage at Arg 320 are independent of prior cleavage at Arg 271 (11,12), apparently ordered bond cleavage by prothrombinase can only result if cleavage at Arg 271 is slow in comparison to cleavage at Arg 271 * (11,12).
Despite general agreement with this logical construct, comparable kinetic constants have been reported for the action of prothrombinase on each of the four individual cleavage reactions assessed using recombinant derivatives of prothrombin (12). Kinetic discrepancies have also led to the suggestion that a significant or large fraction of thrombin is produced by channeling without intermediate release (14 -16). Studies with recombinant human prothrombin derivatives have yielded the novel suggestion that the two bonds in the substrate are recognized and cleaved by kinetically distinct and slowly interconverting conformers of prothrombinase (12). Yet binding studies indicate that all possible substrates and product bind competitively through exosite interactions to prothrombinase with affinities that are independent of the active site of the enzyme (17). These contradictory findings point to difficulties in providing a valid explanation for the action of prothrombinase on prothrombin by using kinetic models such as Scheme I. They are also inconsistent with models implicating a major role for exosite binding in substrate recognition (18 -21).
We have studied the action of human prothrombinase on a series of recombinant derivatives of human prothrombin to investigate these contradictory ideas. We present a model that adequately explains the pathway for prothrombin cleavage on the basis of the kinetic constants for the four possible enzymecatalyzed reactions illustrated in Scheme I. Our findings are inconsistent with the previous proposal (12), and instead suggest equivalent exosite binding interactions between all four possible substrate species and prothrombinase. We propose that ordered cleavage of the two bonds is driven by interactions between a single form of prothrombinase and two distinct conformations of substrate generated in the pathway for cleavage.
Recombinant Proteins-Prothrombin variants were cloned into pcDNA 3.1(ϩ) by using the Gateway Cloning System (Invitrogen) (23). PCR amplification of the cDNA encoding human prothrombin (6), followed by unidirectional cloning into the TOPO-adapted pENTR entry vector, yielded a cassette containing a Kozak sequence, translation start site, signal, propeptide, and mature protein sequence followed by a 3Ј-untranslated sequence (97 bases) extending to the polyadenylation site. This cassette was used as a template for further mutagenesis using the QuickChange mutagenesis kit (Stratagene). Mutagenic primers were used to introduce a codon for Gln in place of Arg 271 for the expression of II Q271 , Gln in place of Arg 320 for the expression of II Q320 , or Ala in place of Arg 320 for preparation of II A320 . Mutagenesis using the cDNA encoding II Q271 as template was used to generate an expression cassette encoding a prothrombin variant containing codons for Gln in place of both Arg 271 and Arg 320 (II QQ ). The integrity of each construct was established by DNA sequencing. Each variant cassette was subjected to phage integrase-mediated recombination into an adapted pCDNA 3.1(ϩ) destination vector. Final expression constructs were sequenced again before transfection.
HEK 293 cells in Opti-MEM (Invitrogen) were transfected by treating with 7.5 g of plasmid DNA and 30 l of LipofectAMINE 2000 (Invitrogen)/3 ϫ 10 5 cells for 4 -5 h followed by the addition of Dulbecco's modified Eagle's medium/F-12 containing 15 mM Hepes, 5% (v/v) fetal bovine serum, and 1 mM L-glutamine. Stable cell lines, generated by selection in media containing G418 (0.5 mg/ml), were screened for protein production using an immunoassay (Enzyme Research Laboratories, South Bend, IN) and by a functional assay for thrombin produced following the addition of prothrombinase or ecarin. Typical production levels ranged from 4 to 7 g/ml prothrombin/24 h in a confluent T-25 flask with 10 ml of serum-free medium. Stable cell lines were expanded into cell factories (Nunclon, Nunc), and large scale protein production was conducted in Dulbecco's modified Eagle's medium/F-12 media without phenol red containing 15 mM Hepes, 5 g/ml insulin/transferrin/ sodium selenite supplement (Invitrogen), 1 mM L-glutamine, and 10 g/ml reduced vitamin K (Abbott). Conditioned media, harvested daily, were treated with 5 mM benzamidine and stored at Ϫ20°C.
Conditioned media (10 liters) were thawed, pooled, and applied at room temperature to a 4.8 ϫ 6 cm column of Q-Sepharose (Amersham Biosciences) equilibrated with 20 mM Hepes, 1 mM benzamidine, pH 7.5. After washing with 20 mM Hepes, pH 7.5, bound protein was eluted with 20 mM Hepes, 0.6 M NaCl, pH 7.5. Fractions containing protein were pooled, cooled to 4°C, and treated with 11 mM Na 3 citrate followed by the addition of 1 M BaCl 2 over 15 min to a final concentration of 74.1 mM. The precipitate was collected by centrifugation, dissolved with 0.2 M EDTA, pH 8.0, dialyzed overnight against 20 mM Hepes, 1 mM EDTA, 1 mM benzamidine, pH 7.5, and applied to a HQ POROS column (10 ϫ 100 mm) (Applied Biosystems) equilibrated in 20 mM Hepes, pH 7.5. Bound protein was eluted with a gradient of increasing NaCl (0 -1.0 M, 15 ml/min, 100 min) in the same buffer. Fractions containing prothrombin were pooled, dialyzed against 1 mM NaP i , pH 6.8, applied to a ceramic hydroxyapatite matrix, CHT5-I (10 ϫ 64 mm) (Bio-Rad) equilibrated in the same buffer, and eluted with a gradient of increasing NaP i , pH 6.8 (1-500 mM, 3.0 ml/min, 17 min). Under-or un-carboxylated prothrombin elutes early in this gradient separated from fully carboxylated material that elutes at higher ionic strength. 4 Fractions were pooled to minimize contamination with under-carboxylated prothrombin, precipitated with solid (NH 4 ) 2 SO 4 (80% saturation), collected by centrifugation (56,000 ϫ g, 30 min), dissolved in 50% (v/v) glycerol, and stored at Ϫ20°C. Final yields of all prothrombin derivatives were typically ϳ2 mg per liter of conditioned media.
Plasma Proteins-Procedures for the purification of factor X, factor V, and prothrombin from plasmapheresis plasma have been described (21,24). Human factors Xa and Va were prepared and characterized as described previously (9,25,26). Kinetic titration of Xa preparations with p-nitrophenol-pЈ-guanidinobenzoate (27) yielded 0.96 -1.22 mol of active sites/mol of factor Xa. Further quality control of factor Va preparations was performed by fluorescence binding measurements assessing its ability to assemble into prothrombinase as described previously (26). Typical results from this approach yielded K d ϭ 1.74 Ϯ 0.45 nM for the assembly of prothrombinase and n ϭ 1.37 Ϯ 0.05 mol of Va bound/ mol of Xa at saturation. Proteolytic derivatives of human prothrombin (II PL ), fragment 1.2 (F1.2), prethrombin 2 (P2), and thrombin were purified and characterized by established procedures (5). Ecarin was purified from the venom of Echis carinatus pyramidum (Latoxan, Valence, France) (19). Modifications to procedures developed with bovine prothrombin were employed to prepare and purify human mIIa covalently inactivated with FPR-CH 2 Cl (mIIa i ) and mIIa inactivated with ATA-FPR-CH 2 Cl and labeled with 6-(iodoacetamido)fluorescein (mIIa F ) (17,19). Following purification, both mIIa i and mIIa F were dialyzed into 20 mM Hepes, 0.15 M NaCl, pH 7.5, concentrated by ultrafiltration, and stored at Ϫ20°C. Protein concentrations were determined using the following extinction coefficients (E 280 mg Ϫ1 ⅐cm 2 ) and molecular weights: human Xa, 1.16, 45,300 (28) (5). All prothrombin derivatives were exchanged into Assay Buffer either by dialysis or by centrifugal gel filtration before use.
Characterization of Prothrombin Variants-N-terminal sequence analysis of prothrombin species was performed by automated Edman degradation at the Emory University Microchemical Facility. Intact prothrombin species and their cleavage products were characterized by sequencing bands excised following SDS-PAGE and electroblotting as described (24). Analysis of 4-carboxyglutamic acid content was performed by base hydrolysis and quantitative determination of Gla and Asx separated by high pressure liquid chromatography and detected following post-column derivatization (31,32). Molecular weights were determined by mass spectrometry using SELDI/TOF/MS (Ciphergen, Fremont, CA). (v/v) glycerol, 0.01% (w/v) bromphenol blue, pH 6.8, and heated at 90°C for 2.5 min. SDS-PAGE of quenched samples (25 l, ϳ2.8 g of protein) was performed using 4 -12% NOVEX BisTris gradient gels run with MES buffer (Invitrogen). Bands were visualized following staining with 0.25% Kinetics of Bond Cleavage in Prothrombin Variants-Reaction mixtures containing 5.0 M prothrombin variant, 50 M PCPS, 20 M DAPA, and 50 nM Va in Assay Buffer and maintained at 25°C were initiated by the addition of 1 nM Xa. Samples (15 l) withdrawn at the indicated times were quenched by mixing with an equal volume of 125 mM Tris, 20% (v/v) glycerol, 2% (w/v) SDS, 0.02% (w/v) bromphenol blue, 50 mM EDTA, pH 6.8. Samples were treated with 62 mM dithiothreitol, heated at 90°C for 2.5 min, and subjected to electrophoresis (3.6 g of protein/lane) using 10% NOVEX Tris-glycine gels (Invitrogen). Protein bands visualized by staining with Coomassie Brilliant Blue R-250 and destaining were imaged in transmitted light using an EDAS290 digital camera system (Eastman Kodak).
Kinetics of Bond Cleavage in Meizothrombin Variants-Each prothrombin variant (5.8 M, 3 ml) was cleaved with ecarin (7 g/ml) for 18 min at 25°C in 20 mM Hepes, 0.15 M NaCl, 0.1% PEG 8000, pH 7.5, containing 50 M FPR-CH 2 Cl to yield mIIa i . Reaction mixtures were quenched with 10 mM EDTA, diluted with 3 ml of 20 mM Hepes, pH 7.5, maintained on ice, and treated with two sequential additions of 50 M p-amidinophenylmethanesulfonyl fluoride. Ecarin was rapidly removed by application of each reaction mixture to a column (0.25 ml) of S-Sepharose (Amersham Biosciences) equilibrated in 50 mM Hepes, 50 mM NaCl, pH 7.5. The unbound fraction was then applied to a column (0.25 ml) of Q-Sepharose (Amersham Biosciences) equilibrated in the same buffer to adsorb the mIIa i variants. Bound protein was eluted with 20 mM Hepes, 0.8 M NaCl, pH 7.5 (2.0 ml), and desalted by chromatography on a 10-ml column of Sephadex G-25 (Amersham Biosciences) equilibrated in Assay Buffer lacking Ca 2ϩ . Peak fractions were treated with 60 M DAPA and maintained on ice for the total of ϳ45 min required to prepare and process all prothrombin variants. Prior to use, mIIa i solutions were warmed to 25°C, brought to 2 mM CaCl 2 , and used to prepare reaction mixtures containing 3.2 M mIIa i , 50 M PCPS, 60 M DAPA, and 70 nM Va in Assay Buffer. Following initiation with 0.8 nM Xa, aliquots were withdrawn at the indicated times, quenched, and analyzed by SDS-PAGE as described above for prothrombin cleavage but without disulfide bond reduction.
Initial Velocity Measurements of Cleavage of II Q271 and P2 Plus F1.2-Initial, steady state velocities of the action of prothrombinase on II Q271 or P2 plus F1.2 were determined by discontinuous measurements of product formation as described previously (20,33). Reaction mixtures in Assay Buffer at 25°C contained increasing concentrations of either substrate, 28 -30 M PCPS LUV and 30 -35 nM Va. Following initiation with factor Xa (20 -40 pM), six serially quenched samples were further diluted and used in initial velocity measurements of S2238 hydrolysis (20). The linear dependence of the rate of S2238 hydrolysis on known concentrations of thrombin was used to infer the concentration of product produced as a function of time, and initial velocities were determined from the linear appearance of product with time (20).
For studies with P2 plus F1.2, F1.2 was present at 1.5 eq of the concentration of P2, and the concentration of substrate was considered to be equal to the concentration of P2 (33). The product produced by the action of prothrombinase on II Q271 was assumed to behave equivalently to thrombin in measurements of S2238 hydrolysis.
Alternative substrate studies were conducted with substrate derivatives that did not yield a product with activity toward S2238. Inhibition studies were conducted as above but in the presence of 3-5 different fixed concentrations of the indicated substrate or inhibitor derivative. Control experiments established that these alternative substrates had no effect on the rate of S2238 hydrolysis by thrombin in the quenched samples.
Initial Velocity Measurements of mIIa Cleavage-Cleavage of mIIa F by prothrombinase was accompanied by ϳ25% enhancement in fluorescence intensity of the fluorescein moiety. The kinetics of cleavage at Arg 271 * in mIIa was inferred from continuous measurements of this fluorescence change using a strategy previously established in the bovine system (19). Measurements were performed in a SpectraMax Gemini (Molecular Devices, Sunnyvale, CA) kinetic plate reader using black polystyrene 96-well plates that had been pretreated with 0.1% (v/v) Tween 20 in Assay Buffer and air-dried before use. Reaction mixtures (200 l) containing 70 nM mIIa F , increasing concentrations of mIIa i , 30 M PCPS LUV , and 35 nM Va in Assay Buffer were initiated with 30 pM Xa. Following brief mixing by vibration, fluorescence intensity was continuously monitored at 25°C using EX ϭ 470 nm and EM ϭ 525 nm with a 495 nm long pass filter in the emission beam. The total concentration of substrate (mIIa) was given by the sum of concentrations of mIIa i and mIIa F (19). The initial, steady state, rate of fluorescence change was converted to concentration terms using the total concentration of mIIa and the limits of the fluorescence signal signifying 0 and ϳ100% conversion of substrate to product (19). Inhibition studies were performed using the same approach but in the presence of the indicated fixed concentrations of II QQ .
Data Analysis-Initial velocity data were analyzed according to the indicated rate expressions using the Levenberg-Marquardt algorithm (34). Fitted constants are presented at Ϯ 95% confidence limits. For inhibition studies, alternative kinetic models were tested for and excluded on the basis of established criteria (35).
The concentrations of Va and PCPS were chosen to be in excess of the concentration of Xa and well above the measured equilibrium dissociation constants for the individual interactions within prothrombinase (36). Therefore, the concentration of prothrombinase (E) was considered equal to the limiting concentration of factor Xa present in the reaction mixture (36). As a result of this experimental design, only a vanishingly small fraction (ϳ0.2%) of all small unilamellar PCPS vesicles would be expected to be populated with prothrombinase at the low concentrations of Xa (20 -40 picomolar) needed for reliable initial velocity measurements. Rate-limiting mass transfer of membrane-binding reactants between nonproductive and productive vesicles in this system is known to lead to the slow assembly of factor Xa into prothrombinase and dominate the rate of substrate delivery (37)(38)(39)(40). Kinetic complexities arising from these physical limitations were avoided by decreasing the vesicle concentration at saturating phospholipids by the use of PCPS LUV for the initial velocity studies. Calculations based on measured vesicle size, a head group surface area of 74 Å 2 (41), and a 50:50 distribution of lipids between the two leaflets indicated that Ͼ54% of the PCPS LUV vesicles would be expected to be populated with enzyme at the lowest concentration of prothrombinase used.
Densitometry analysis of digital camera images was performed using the program ImageQuant TL (Amersham Biosciences). Staining intensity of each band was determined by volume integration (pixel intensity integrated over area of band image), corrected for background staining and normalized to the total staining intensity in each lane. For each experiment, images acquired at different shutter speeds were used to test for saturation-related artifacts. Reaction profiles for the action of prothrombinase on II PL or II WT were generated from band staining intensities using the considerations and simultaneous equations detailed previously (11). Molar fractional staining intensities for all species evident on the gel (II, F1.2-A, F1.2, IIa B ) were determined relative to prothrombin. These values were either inferred from band intensities observed at long times of digestion from the time course itself or from companion gels in which II WT and II PL were analyzed before and after prolonged digestion with prothrombinase or ecarin. Mean values were determined when fractional staining was calculated in multiple ways. Typical fractional molar staining intensities obtained for 10% Trisglycine gels processed with a consistent staining and destaining protocol were f II ϭ 1, f F1.

Characterization of Recombinant Prothrombin Variants-
The N-terminal sequence for all prothrombin variants (Table I) was consistent with appropriate processing of the leader and propeptide during secretion by the stable HEK293 cell lines. Molecular weights obtained by SELDI/TOF/MS were in agreement with the molecular weight of II PL (Table I). The slightly lower molecular weights observed for all recombinant prothrombin species might reflect differences in carbohydrate content with II PL . Gla content was generally consistent with the isolation of fully carboxylated recombinant prothrombin variants (Table I). However, II WT and II Q271 consistently yielded a Gla content greater than the expected value of 10 (Table I). SDS-PAGE implicated trace contamination with co-purified fragment 1 in this phenomenon, and calculations revealed that minor contamination with fragment 1 (Յ7 mol %) could skew the measured Gla/Asx ratio sufficiently to account for the findings. This suggestion is consistent with the expected Gla content observed for fragment 1 purified following quantitative proteolysis of II WT (Table I). We assert that the purified recombinant prothrombin preparations contain the appropriate complement of key post-translational modifications known to be necessary for function.
Equivalence between II PL and II WT was further documented by comparing full progress curves for product formation obtained by discontinuous measurements of S2238 cleavage following the addition of prothrombinase (not shown). II PL and II WT yielded product at the same rate and to a comparable extent. Accordingly, SDS-PAGE analysis following prolonged digestion of II PL or II WT with prothrombinase yielded bands indicating the production of thrombin and F1.2 arising from cleavage at both Arg 320 and Arg 271 ( Fig. 1 and Scheme I). The action of prothrombinase on II Q271 yielded bands consistent with the limiting formation of mIIa following cleavage only at Arg 320 , whereas cleavage of II Q320 produced bands consistent with the formation of P2 and F1.2 ( Fig. 1 and Scheme I). II QQ was resistant to digestion by prothrombinase (Fig. 1). The identities of bands produced from the recombinant species were also unambiguously established by N-terminal sequence analysis. In agreement with previous studies (12), the data establish the utility of the recombinant prothrombin derivatives as reagents for measurements of the individual cleavage reactions in intact prothrombin.
Kinetics of Bond Cleavage Catalyzed by Prothrombinase-Analyses by SDS-PAGE indicated that that the two bonds in II PL and II WT were cleaved in an equivalent manner by prothrombinase (Fig. 2, A and B). In each case, the disappearance of prothrombin was accompanied by the transient appearance of a band corresponding to F1.2-A. Bands corresponding to F1.2 and IIa B predominated at longer reaction times (Fig. 2, A and  B). The results agree with published findings in the bovine and human systems, implicating the sequential action of prothrombinase at Arg 320 and at Arg 271 * (9 -11, 42).
In parallel experiments, cleavage of II Q271 by prothrombinase at Arg 320 also proceeded rapidly (Fig. 2C). Although residual substrate was obvious following prolonged digestion, the disappearance of the zymogen band at the early time points approximated the disappearance of II PL or II WT (see below). In contrast, II Q320 was cleaved slowly at Arg 271 (Fig. 2D), and product bands were only evident at longer time points. Comparable results were obtained with II A320 as a substrate (Fig.  2E). Thus, despite the presence of a single cleavable site at Arg 271 , both II Q320 and II A320 are cleaved slowly by prothrombinase. The observations suggest that Arg 271 is poorly accessible to cleavage by prothrombinase in the otherwise intact zymogen. This effect is independent whether Arg 320 is rendered uncleavable by substitution with either Gln or Ala.
Analysis of Prothrombin Cleavage-Quantitative densitometry yielded equivalent profiles for the cleavage of II PL or II WT (Fig. 3). In either case, product profiles could be accounted for by the consideration of substrates and products relevant to Reactions 1 and 2 in Scheme I. Prothrombin disappearance was accompanied by the transient accumulation of mIIa and the  delayed appearance of thrombin, with features typical of an A 3 B 3 C reaction series (Fig. 3) (43). For both II PL and II WT , progress curves for the appearance of F1.2 were indistinguishable from those describing the appearance of thrombin (not shown). This finding indicates that flux toward thrombin formation via the formation of P2 plus F1.2 (Scheme I, Reactions 3 and 4) is below the limit of detection. In agreement with this conclusion, the initial rate of disappearance of prothrombin was matched by the initial rate of appearance of mIIa, and thrombin accumulated with an obvious lag phase (Fig. 3). Thus, within experimental error, cleavage of prothrombin at Arg 320 to yield mIIa (Scheme I, Reaction 1) is sufficient to account for the action of prothrombinase on the intact zymogen. Thrombin formation can seemingly be explained by the stepwise action of prothrombinase on Arg 320 followed by Arg 271 * in successive enzyme catalyzed reactions (Scheme I, Reactions 1 and 2). Quantitative densitometry of prothrombin consumption was used to compare the action of prothrombinase on prothrombin derivatives containing both cleavage sites (II PL and II WT ) with its action on the individual sites in II Q271 and II Q320 (Fig. 4). II Q271 was consumed, as a result of cleavage at Arg 320 , with an initial rate that was indistinguishable from the rate of consumption of either II PL or II WT (Fig. 4). Residual amounts of II Q271 (ϳ15%) remaining after even 30 min probably reflect the results of product inhibition. In contrast, both II Q320 and II A320 were consumed at a rate that was ϳ30-fold lower than that observed with either II WT or II Q271 (Fig. 4).
The action of prothrombinase at the Arg 320 site in II Q271 can quantitatively account for the consumption of prothrombin even when both Arg 271 and Arg 320 (in II WT or II PL ) are available for cleavage. In contrast, the Arg 271 site in the intact zymogen is acted upon by prothrombinase with a greatly reduced rate. Consequently, the bulk of thrombin produced would be predicted to result from initial cleavage at Arg 320 in prothrombin. These data provide an adequate explanation for the observed pathway for the action of prothrombinase on II PL or II WT (Fig. 3). However, more kinetic information is necessary to establish whether such rate differences can explain selectivity of bond cleavage at other reactant concentrations.
Kinetics of the Individual Cleavage Reactions-Initial velocity studies were pursued with large unilamellar PCPS vesicles (PCPS LUV ) to eliminate interpretation problems arising from rate-limiting mass transfer of reactants at saturating concentrations of phospholipid and picomolar concentrations of prothrombinase (see "Data Analysis"). To aid in comparisons with the results of SDS-PAGE, principal kinetic findings were confirmed with small unilamellar vesicles as well (not shown). However, at picomolar enzyme and saturating phospholipid, reactions conducted with PCPS LUV were characterized by a 2-3-fold lower K m and an 3-fold higher V max . Substrate inhibition, a pronounced problem with small unilamellar vesicles (44), was not obvious with PCPS LUV , and the initial rate was proportional to [E] from the picomolar to the nanomolar range (not shown). On these bases, we have chosen to present findings obtained with PCPS LUV by assuming that they more amenable to meaningful kinetic interpretation.
Initial velocity studies with II Q271 established kinetic constants for the action of prothrombinase at the Arg 320 site in intact prothrombin (Table II). The use of II Q320 (cleavable at Arg 271 ) as an alternative substrate yielded data that could adequately be described by the rate expression for classical competitive inhibition (Fig. 5A). The K i value for inhibition by II Q320 was comparable to the K m value for II Q271 (Table II). II Q271 and II Q320 , each with a different cleavable site, bind to prothrombinase in a mutually exclusive manner with approximately equal affinity. The perceived affinity of prothrombinase for the substrate is therefore the same regardless of SCHEME 1. Possible pathways for the activation of prothrombin. Schematic illustration of the species produced from the two possible cleavage pathways for prothrombin activation. The fully formed proteinase domain is shaded red. R 271 and R 271* refer to the 271 cleavage site either before or after cleavage at Arg 320 . Similarly, R 320 and R 320* denote the 320 cleavage site either before or after cleavage at Arg 271 .
whether the enzyme acts on Arg 320 or Arg 271 in the intact zymogen. Most surprisingly, it follows that the ϳ30-fold lower rate of cleavage at Arg 271 in II Q320 (Fig. 4) arises from a decreased V max and not a compromised K m .
II QQ was also a classical competitive inhibitor of II Q271 cleavage by prothrombinase with K i Ϸ K m (Fig. 5B and Table II). As both sites are uncleavable in II QQ , the data confirm the conclusion that substrate affinity is independent of the cleavage site in the intact zymogen that is acted upon by prothrombinase. Competitive inhibition indicates mutually exclusive interactions between II QQ or II Q271 , and prothrombinase, completely independent of the availability of cleavage sites in the substrate. Thus, increasing concentrations of II QQ are expected to decrease the rate of cleavage of II Q271 to zero at any concentration of II Q271 . Taken together with the fact that these findings were made in initial velocities determined following initiation with factor Xa, our findings obviate the need to invoke the differential recognition of the cleavage sites in the substrate by two slowly equilibrating but kinetically distinct enzyme conformers (12).
Kinetic constants for cleavage at the Arg 320 * site, assessed using P2 plus F1.2 as a substrate, were indistinguishable from those determined for cleavage at Arg 320 using II Q271 (Table II). Both II Q320 and II QQ acted as classical competitive inhibitors of P2 plus F1.2 cleavage with K i Ϸ K m (Table II). In agreement with previous suggestions (11,12), these findings document the kinetic equivalence of the action of prothrombinase on Arg 320 and on Arg 320 *.
Such equivalence was not evident for cleavage at Arg 271 versus Arg 271 *. Kinetic constants for the action of prothrombinase on Arg 271 * in mIIa were comparable with those observed for cleavage at Arg 320 and Arg 271 * (Table II). These findings imply that Arg 271 * in mIIa is cleaved with a significantly greater V max than is Arg 271 in II Q320 . Thus, the V max for cleav- age at Arg 271 is greatly influenced by the cleavage status of the Arg 320 bond. Classical competitive inhibition of mIIa by II QQ (Fig. 6) with K i Ϸ K m (Table II) provides the full cycle of documentation that all substrate derivatives bind in a mutually exclusive manner to prothrombinase with approximately equal affinity regardless of the site that is available for cleavage or is cleaved in the substrate.
Quantitative Accounting for the Pathway of Prothrombin Cleavage-Findings from initial velocity studies of the individual half-reactions permit an expanded description for the action of prothrombinase on prothrombin (Scheme II).
Each half-reaction of prothrombin activation is illustrated to result from the binding of the substrate to kinetically indistinguishable forms of prothrombinase (Scheme II). This representation explains competitive inhibition of each half-reaction by any possible substrate derivative regardless of the status of the individual cleavage sites. The K m value for each substrate derivative is approximately the same (Scheme II). As a result, all possible forms of substrate and product bind in a mutually exclusive fashion to prothrombinase with comparable affinity.
Extensive kinetic studies of intermediate cleavage in both the bovine and human systems have established that the ob-served V max is determined by equilibrium and rate constants for at least two steps that follow the initial binding of substrate (18 -21). Whereas three of the four half-reactions exhibit comparable values for V/E (Scheme II), cleavage of Arg 271 in intact prothrombin proceeds with a value for V/E that is ϳ30-fold lower. Thus, at any prothrombin concentration, Ͼ96% of the reaction flux toward thrombin formation is expected to occur via initial cleavage at Arg 320 followed by cleavage at Arg 271 * (Scheme I, Reactions 1 and 2). Thrombin or intermediates produced as a result of bond cleavage in the opposite order would be experimentally undetectable. These results now provide a kinetic explanation for the apparently sequential action of prothrombinase on prothrombin (9 -11).
Influence of Prior Cleavage at Arg 320 on Recognition and Cleavage at Arg 271 -Implicit in Scheme II is the idea that prior cleavage at Arg 320 greatly enhances cleavage at Arg 271 by affecting the conversion of Arg 271 to Arg 271 *. This interpretation predicts that the low rate of cleavage at Arg 271 in II Q320 (Fig.  2D) would be rectified if II Q320 could be converted to mIIa.
Ecarin cleaved all recombinant prothrombin variants to mIIa-like species regardless of the identity of the side chain at residue 320 (Fig. 7, inset). N-terminal sequence analysis con-   Table II. firmed the production of the thrombin B-chain following cleavage of each variant by ecarin. The findings indicate that ecarin is capable of cleaving at residue 320 despite substitution of the P 1 Arg with Gln.
The action of prothrombinase on Arg 271 * was assessed by SDS-PAGE and quantitative densitometry using covalently inactivated meizothrombin forms (mIIa i ) produced from the prothrombin variants (Fig. 7). mIIa i produced from II PL or II WT was rapidly cleaved by prothrombinase at Arg 271 * to yield thrombin and F1.2. In keeping with the lack of a cleavable site at 271, mIIa i produced from II Q271 was stable and not detectably consumed by prothrombinase (Fig. 7). In contrast to the slow cleavage at Arg 271 in intact II Q320 , the mIIa i derivative of II Q320 was cleaved at Arg 271 * at the same rate as mIIa i produced from II WT (Fig. 7). The findings verify a key kinetic prediction from Scheme II by documenting the rate enhancing effects of prior cleavage at Arg 320 on the action of prothrombinase on Arg 271 . DISCUSSION Since the initial demonstration of the apparently ordered action of prothrombinase on the two scissile bonds in prothrombin (9,10,42), major gaps and contradictory ideas have prevailed, affecting the understanding of the kinetic and molecular bases for bond selectivity in this process. Our results are consistent with a model that accounts for prothrombin activation by prothrombinase based on the measured kinetic properties of each of the four possible cleavage reactions with mutually exclusive binding interactions between each of the four possible substrate species and the enzyme (Scheme II). Our findings also provide new insights into the mechanisms underlying the action of prothrombinase on the two scissile bonds in prothrombin.
Three of the four possible cleavage reactions are characterized by equivalent steady state kinetic constants (Scheme II). The reduced V max for cleavage at Arg 271 in intact prothrombin lies at the heart of providing an explanation for the cleavage pathway observed with prothrombinase. As a result of the reduced V max for this step, Ͼ96% of the thrombin formed at any substrate concentration will occur via cleavage at Arg 320 followed by cleavage at Arg 271 *. Thus, whereas cleavage of the two bonds in prothrombin can indeed occur in either order, by far the predominant pathway for thrombin production is via the formation of mIIa. Flux via P2 plus F1.2 is expected to play a minor role in thrombin formation that is below the limits of detection. These conclusions explain observations in a series of studies of prothrombin activation catalyzed by prothrombinase using saturating concentrations of membranes and factor Va (9 -12). Factor Va and the substrate-membrane interaction are established to differentially affect the individual cleavage reactions illustrated in Scheme II (9 -13). Thus, changes in the contributions of the two possible pathways for prothrombin cleavage are expected in model systems in which factor Va and membranes are not saturating, cofactor function is altered, or membrane binding by the substrate is affected.
Slow cleavage at Arg 271 relative to Arg 320 in intact prothrom-  Table II. bin has been considered the most likely logical explanation for the seemingly ordered action of prothrombinase on prothrombin considered within the constraints of Scheme I (11,12,19). However, experimental verification of this key prediction has awaited the application of recombinant prothrombin derivatives suited for an assessment of the individual cleavage reactions in the otherwise intact zymogen. In contrast to our findings with such an approach, recent studies have reported approximately equal catalytic efficiencies for the action of prothrombinase on Arg 271 and Arg 320 in intact prothrombin (12). The inability of these observations to explain the action of prothrombinase on prothrombin was attributed to an inappropriate property of the recombinant mutant containing Ala in place or Arg at 320, rather than a failure of simplified reaction schemes (e.g. Scheme I) to adequately account for prothrombin activation or unforeseen effects of two other substitutions in that prothrombin derivative (12). We have documented equivalently slow cleavage at Arg 271 regardless of whether the 320 site is rendered uncleavable by substitution with Gln or Ala. Full rescue of slow cleavage at Arg 271 by prior cleavage at the 320 site further reduces the likelihood of deleterious effects on cleavage at Arg 271 by substitutions at 320. Thus, although we presently cannot provide an explanation, unexpected effects arising from replacing Arg at 320 with Ala cannot account for the discrepancy between this and the previous study (12).
Thrombin formation arising from cleavage at Arg 271 followed by cleavage at Arg 320 * (Scheme I, Reactions 3 and 4) is predicted to contribute in an experimentally insignificant way to prothrombin activation by prothrombinase. The intermediate, mIIa, accumulates at concentrations in vast excess of the concentration of prothrombinase, and the initial rate of prothrombin disappearance agrees with the rate of mIIa production. These are the principal arguments against an obvious contribution from channeling to thrombin formation (12, 14 -16). Although the overall process can adequately be described to result from two consecutive enzyme-catalyzed reactions separated by a product release step, some fraction of mIIa must be cleaved to thrombin without dissociating from the enzyme. The data indicate that any contribution from such a pathway is small and within experimental error.
The recent proposal that the cleavage sites in prothrombin are differentially recognized and cleaved by two slowly interconverting forms of prothrombinase requires that the binding of each of the four possible substrate species to the enzyme is cleavage site-dependent and that the enzyme-substrate interactions that govern cleavage at Arg 271 versus Arg 320 are not mutually exclusive (12). This idea cannot readily co-exist with the alternative suggestion of equivalent, exosite-dependent interactions between bovine prothrombinase and the substrate regardless of the site that is cleaved (19). The latter suggestion is further supported by equilibrium binding studies showing that all possible substrate derivatives and product bind to prothrombinase in a mutually exclusive manner, independent of the covalent occlusion of the active site of the enzyme (17).
Classical competitive inhibition of the cleavage at Arg 320 in II Q271 by II Q320 or by II QQ indicates mutually exclusive binding of these substrate derivatives to prothrombinase regardless of the site available for cleavage. Further support for this idea is provided by the ability of II QQ to act as a classical competitive inhibitor of three of the possible four half-reactions that could be investigated (Table II). These data now discount the need to incorporate two kinetically distinct enzyme isoforms in the kinetic scheme accounting for the two cleavage reactions (12). Instead, they are more consistent with the idea that each cleavage reaction arises from mutually exclusive interactions between the four possible substrate species and kinetically identical forms of the enzyme (Scheme II).
Competitive inhibition of the cleavage reactions by II QQ , with K i Ϸ K m , agrees with the idea that perceived affinity of prothrombinase for the various substrate derivatives is dominated by contributions from binding to an exosite on the enzyme (19,20). Docking of structural elements surrounding the scissile bond of the substrate with the active site of the enzyme occurs in a following step leading to catalysis (19,20). Whereas exosite binding is a primary determinant of K m , the unimolecular active site docking step contributes to the V max (19,20). Along this line of reasoning, mutually exclusive binding interactions between II Q320 and II Q271 with approximately equal affinity indicate that both substrate analogs, each with a different cleavable site, bind to prothrombinase through equivalent exosite interactions. Structural data places the Arg 320 and Arg 271 sites on opposite faces of the proteinase domain, separated by 36 Å (45,46). These geometric constraints imply that only one of the two cleavage sites will be appropriately positioned to engage effectively the active site of the enzyme. Optimized active site docking of structures surrounding the Arg 320 site and the reduced accessibility of those surrounding Arg 271 could explain the ϳ30-fold lower V max for the cleavage at Arg 271 in II Q320 in comparison to cleavage at Arg 320 in II Q271 . Flexibility in the conformation of prothrombin tethered through exosite binding to prothrombinase is implied by the fact that the V max for cleavage at Arg 271 in II Q320 is not zero.
The increased V max for cleavage at Arg 271 * in comparison to Arg 271 could arise from a change in the substrate following cleavage at Arg 320 that now facilitates enhanced docking of structures surrounding this scissile bond with the active site of the enzyme. Cleavage at Arg 320 leads to the conversion of the zymogen to proteinase and is associated with major conformational changes in the activation domains (45)(46)(47)(48)(49). Conformational changes associated with the transition of zymogen to proteinase represent candidate explanations for changes in the substrate that follow cleavage at Arg 320 and now permit the efficient docking of structures flanking Arg 271 * with the active site. Geometric constraints associated with the exosite-dependent tethering of the substrate in either the zymogen or the proteinase forms could determine accessibility and drive presentation of the individual cleavage sites to the active site of the enzyme. Such ideas provide a potential and experimentally testable explanation for the equivalent kinetic constants observed for cleavage at Arg 320 and Arg 320 *, large differences in the V max for cleavage at Arg 271 and Arg 271 *, and the observed order of bond cleavage catalyzed by prothrombinase.
In contrast to the previous kinetic model requiring two forms of prothrombinase to account for the two cleavage reactions in prothrombin (12), we suggest instead that a more comprehensive kinetic explanation can be provided by a simpler model comprising the four individual cleavage reactions (Scheme II) and a conformational change in the substrate following initial cleavage at Arg 320 . We propose that the ordered action of prothrombinase on the two sites in prothrombin arises from the constraining effects of substrate bound in two distinct conformations through equivalent exosite binding interactions to a single form of prothrombinase.