Prothrombin 487 to Factor Va Binding in the Prothrombinase Complex*

To identify sequences in prothrombin (fII) involved in prothrombinase complex (fXa (cid:1) fVa (cid:1) fII (cid:1) phospholipids) assembly, synthetic peptides based on fII sequences were prepared and screened for their ability to inhibit factor Xa (fXa)-induced clotting of normal plasma. The fII peptide (PT473–487, homologous to chymotrypsin residues 149D-163) potently inhibited plasma clotting assays and prothrombinase activity, with 50% inhibition of 12 and 10 (cid:1) M peptide, respectively. Prothrombinase inhibition by PT473–487 was factor Va (fVa)-dependent and se-quence-specific, because the peptide did not inhibit fII activation in the absence of fVa, and a scrambled sequence peptide, PT473–487SCR, was not inhibitory. Peptide PT473–487 did not inhibit the amidolytic activities of fXa and thrombin, suggesting that the peptide did not alter the integrity of their active sites. To deter-mine whether PT473–487 interacted directly with fVa, fluorescein-labeled fVa (Fl-fVa) was prepared. When PT473–487 was titrated into samples containing phos-pholipid-bound Fl-fVa, the peptide increased fluorescein anisotropy (EC 50 at 3 (cid:1) M peptide), whereas the con- trol peptide PT473–487SCR containing 50 m M m M and 5 m M 2 min room Thereafter, the hydrolysis of Fl-mfIIa i at Arg 271 was triggered by the addition of fXa (4 n M ) for 10 min at 37 °C. Subse- quently, the reaction was stopped by the addition of 20 (cid:4) l of hot SDS sample buffer. The samples were incubated on a boiling water bath and analyzed by SDS-PAGE, and then fluorescent protein bands were visualized using UV light on an EAGLE-EYE II (Stratagene) imager. The fluorescence image was digitalized, and cleavage at Arg 271 was quantified using UN-SCAN-IT version 5.1 (Silk Scientific Corp., Orem, UT). Spectral Measurements— Steady-state fluorescence intensity and anisotropy measurements were made using a SLM AB2 Luminescence spectrometer (SLM Aminco, Rochester, NY) or SLM 8100 fluorometer equipped with a 150-W xenon arc lamp, two holographic gratings in the excitation light path, and an IBM computer for data analysis. Fluorescein fluorescence was detected at its excitation and emission maxima wavelengths of 490 and 521 nm, respectively, using a 345-nm cut-off filter (Schott Glass Technologies, Duryea, PA) in the excitation beam path. Shutters were closed, except during measurements, to minimize photodegradation of the samples. A circulating water bath was used to maintain the sample compartment at 25 °C. All fluorescence experiments the solvent accessibility. The effect of increasing ionic strength because of NaI addition on fluorescence of Fl-fVa was corrected using a NaCl titration.

Prothrombin (fII) 1 is activated to ␣-thrombin by proteolytic cleavages between Arg 271 -Thr 272 and Arg 320 -Ile 321 in the pe-nultimate step in coagulation (1)(2)(3). This activation is accomplished at a physiologically relevant rate by the prothrombinase complex that consists of the serine protease coagulation factor Xa (fXa) associated with its nonenzymatic cofactor, factor Va (fVa), on a negatively charged phospholipid surface in the presence of Ca 2ϩ (4 -7). The prothrombinase complex has been kinetically well characterized and is considered an archetype for other blood clotting complexes. Assembly of this complex results in a 278,000-fold augmentation in the rate of fII activation compared with fXa alone (5). fVa functions both as a K-type and a V-type effector in this reaction, decreasing the K m and increasing the k cat , respectively. Additionally, in the presence of fVa and phospholipids, fII activation proceeds almost exclusively via a meizothrombin intermediate (initial cleavage at Arg 320 ), whereas in the absence of these two components, the activation proceeds via a prethrombin 2 intermediate (initial cleavage at Arg 271 ) (2,3,8). Therefore, the substrate specificity of fXa is altered in the presence of fVa and phospholipids. fII can be classified as a vitamin K-dependent blood coagulation protein, because it requires vitamin K for its normal biosynthesis (4). It bears extensive structural and sequential homology to other vitamin K-dependent proteins, such as activated protein C (APC) and fXa (9 -14). The N-terminal region (residues 1-40) of the mature protein is rich in post-translationally modified ␥-carboxyglutamic acid (Gla) residues and hence is termed the Gla domain. The Gla domain is followed by two domains that bear structural similarity to Danish kringles and are called "kringle domains" (residues 41-155 and 156 -271, respectively). These domains are followed by the serine protease domain (residues 272-579). The serine protease domain of fII is highly homologous to other chymotrypsin-like proteases, such as fXa, fIXa, and APC, and six variable regions and seven conserved regions have been identified (15).
A fundamental yet daunting question concerning multiprotein complexes is: what are the various contact points between the different proteins? Several fVa-binding loci seem likely to exist on fII, although the picture is still incomplete. Both fII kringle domains have been reported to interact with fVa (16,17). Additionally, the Gla domain of fII has also been reported to participate in fVa binding (18). Recently, it was reported that a cluster of cationic residues on the protease domain of fII termed "anion-binding proexosite-I" is involved in the interaction of fII with fVa (19,20). Mutations of these basic residues to acidic Glu have resulted in a fII with impaired fVa binding, suggesting the participation of this region in fVa recognition (21).
Synthetic peptides may provide useful information about protein-protein interactions. For example, certain sequences in the vitamin K-dependent enzyme APC were essential for interactions with its substrate, fVa (22)(23)(24). Subsequent site-directed mutagenesis of the APC autolysis loop (chymotrypsin (CHT) residues 149 -163) supported these implications (25). Based on the sequence homology of the serine protease domain of protein C with that of fII and fXa, we used synthetic peptides comprising related sequences in fII and fXa to test their ability to inhibit assembly of the prothrombinase complex (13). Here, we identified a distinct sequence in fII that is essential for its interaction with fVa, namely amino acids 473-487 (CHT 149D-163).

EXPERIMENTAL PROCEDURES
Materials-Human fVa and Echis carinatus venom were purchased from Hematologic Technologies Inc. (Essex Junction, VT). Human fII, fXa, and thrombin were obtained from Enzyme Research Laboratories (South Bend, IN). The chromogenic substrates S-2222 and S-2238 were purchased from DiaPharma (West Chester, OH) and CBS 34 -47 from Diagnostica Stago Inc. (Parsippany, NJ). Rabbit brain cephalin was obtained from Sigma. Bovine serum albumin, human fibrinogen, and D-phenylalalyl-L-prolyl-L-arginyl chloromethyl ketone were purchased from Calbiochem. Normal human citrate-anticoagulated pooled plasma and fVa-deficient plasma were obtained from George King Bio-Medical, Inc. (Overland Park, KS). Alexin, the aPTT reagent, was purchased from Sigma. Fluorescein-labeled fXa (Fl-fXa i ) was prepared as described (26). Small unilamellar vesicles were made as described (27) using synthetic dioleolyphosphatidylcholine (PC) and dioleolyphosphatidylserine (PS) purchased from Avanti Polar Lipids (Alabaster, AL), and percent recovery of lipids in vesicles was estimated post centrifugation using a L-3-phosphatidylcholine-1, 2-di[1-14 C]oleoyl PC (Amersham Biosciences) as a tracer. All other reagents were of the highest quality available.
Peptide Synthesis and Characterization-Fifteen amino acids long C-terminal amide peptides were synthesized using standard Boc chemistry, then N-acetylated and purified to homogeneity using reversephase high performance liquid chromatography as described earlier (22). For some experiments, peptides were obtained from Bio Synthesis Inc. (Lewisville, TX). Mass spectroscopic analyses of the purified peptides yielded the exact expected molecular weight for each individual peptide. Solutions of each peptide were prepared and the concentrations determined as described (22).
Fluorescein Labeling of fVa-fVa was labeled as described previously (28,29) with Fl (26) to yield fluorescein-labeled fVa, designated Fl-fVa. Briefly, fVa (4.3 mg/ml) in buffer containing 50 mM Hepes (pH 7.4), 150 mM NaCl, and 5 mM CaCl 2 was reacted with a 10-fold molar excess of the fluorescent reagent in the dark at 4°C for 8 h. The reactants were mixed throughout the reaction by a gentle tumbling motion. Subsequently, the labeled protein was separated from the excess unreacted dye by passing the reaction mixture over a Sephadex G-25 (Amersham Biosciences) column (40 ϫ 0.5 cm). The labeled protein eluted in the excluded volume of the column, whereas the free unreacted dye eluted later in the included volume of the column. Fl-fVa was aliquoted into convenient volumes, frozen in liquid nitrogen, and stored at Ϫ80°C for future use.
Fluorescein Labeling of Meizothrombin-Fluorescein-labeled meizothrombin was prepared as described previously (30). Briefly, human fII (1 mg/ml) was incubated for 10 min at 37°C with E. carinatus venom (5 M final) in the presence of fluorescein-Phe-Pro-Arg chloromethyl ketone (200 M final) in 50 mM Hepes (pH 7.4), 150 mM NaCl, and 2 mM CaCl 2 . The reaction was stopped with 5 mM EDTA and immediately chromatographed at room temperature on an anion exchange column (Mono-Q, Amersham Biosciences) using a 20-ml linear gradient from 0 to 600 mM NaCl. Two peaks were resolved. The second peak eluting at ϳ450 mM NaCl corresponded to fluorescein-labeled meizothrombin (Fl-mfIIa i ) based on SDS-PAGE analysis. Fractions containing Fl-mfIIa were pooled and dialyzed in 50 mM Hepes, 150 mM NaCl, and 5 mM CaCl 2 , aliquoted, quick frozen, and then stored at Ϫ80°C for future use.
fXa-1-stage Coagulation Assay-Peptides were screened for effects on fXa coagulant activity using a fXa-1-stage coagulation assay and normal human plasma described previously (22). Briefly, rabbit brain cephalin (100 l at 200 g/ml) was preincubated with 100 l of normal human plasma or factor Va-deficient plasma for 200 s at 37°C. Coagulation was then initiated by the addition of fXa (200 l at 0.31 nM in 50 mM Tris/HCl (pH 7.4), 100 mM NaCl, 0.5% BSA, 0.02% NaN 3 (TBS-BSA) plus 15 mM CaCl 2 ) that had been preincubated for 10 min with the peptide of interest at 37°C. The residual fXa clotting activity in the presence of each different peptide concentration was calculated from a standard curve derived from the double logarithmic plot of clotting time versus fXa concentration.
fXa and Thrombin Amidolytic Assays-Amidolytic activities of fXa and thrombin toward their respective peptide chromogenic substrates, S-2222 and CBS 34 -47 (or S-2238), were assayed (22,31), and the effect of synthetic peptides on amidolytic activity was determined. The peptide (0 -350 M) was incubated with the enzyme (5-10 nM) in buffer containing 50 mM Hepes (pH 7.4), 150 mM NaCl, 5 mM CaCl 2 , and 0.5% BSA for 10 min at room temperature. Thereafter, the activity of the enzyme toward its respective chromogenic substrate was assayed at six different substrate concentrations (ranging between 0 and 2 mM) using an OPTIMAX tunable microplate reader (Molecular Devices, Sunnyvale, CA).
Fibrinogen Clotting-Human thrombin (0.5 units/ml) was preincubated with peptide (0 -1 mM) at five different concentrations in 200 l of buffer containing TBS-BSA plus 2.5 mM CaCl 2 for 10 min at 37°C in polystyrene cuvettes. Clotting was initiated by the addition of 200 l of human fibrinogen (5 mg/ml) in TBS-BSA plus 2.5 mM CaCl 2 , and the time from the addition of fibrinogen to clot formation was measured using an Electra 700 Automatic Coagulation Timer (Medical Laboratory Automation, Inc., Mount Vernon, NY).
Prothrombinase Assay-fII activation was measured in a two-stage assay (22,31). The first step involved the generation of thrombin, which was subsequently measured in the second step using a chromogenic substrate S-2238 or CBS 34 -47. Amounts of fXa, phospholipids, and fII present in the assay were such that the rate of thrombin generation was linearly proportional to the amount of fVa present in the reaction mixture. Unless otherwise indicated, the peptide was preincubated with fVa (0.4 nM final) in TBS-BSA plus 2 mM CaCl 2 for 20 min at 22°C. Following preincubation, an aliquot of this solution was added to a solution containing fII (1.2 M final) and PC/PS vesicles (50 M final) in TBS-BSA plus 2 mM CaCl 2 . Thrombin generation was initiated by the addition of fXa (1 nM). To measure the amount of thrombin generated, a 20-l aliquot of the activation mixture was withdrawn every minute and quenched into 80 l of buffer containing TBS-BSA plus 10 mM EDTA. Finally, the amount of thrombin present was quantified by adding 50 l of 2 mM S-2238, and the rate of substrate hydrolysis was monitored at 405 nm for 2.5 min. The change in A 405 was directly proportional to the thrombin generation. For the assays performed in the absence of fVa, the protocols were the same except that the synthetic peptides were preincubated with fII before thrombin generation was initiated by the addition of fXa.
For some kinetic studies, a modified prothrombinase assay was used to conserve reagents. Briefly, different concentrations of fII (ranging from 0 to 6.5 M) in 50 mM Hepes (pH 7.4), 150 mM NaCl, 5 mM CaCl 2 , 0.1 mM MnCl 2 , 0.5% BSA and 0.02% NaN 3 were incubated with PC/PS vesicles (25 M) and fVa (114 pM) in the presence or absence of different concentrations of the peptide of interest at room temperature for 10 min in a total volume of 35 l. fII activation was initiated by adding 10 l of fXa (320 pM) in the same buffer. After 2.5 min, the reaction was quenched with 50 l of buffer containing 50 mM Hepes (pH 7.4), 150 mM NaCl, and 10 mM EDTA, and the thrombin that was generated was measured by using CBS 34 -47 (50 l, 1.3 mM), and then the rate of substrate hydrolysis was monitored by the rate of change in A 405 that was directly proportional to the thrombin concentration.
Activated Partial Thromboplastin Time (aPTT) Assays-The procoagulant activities of fVa and Fl-fVa were compared using an activated partial thromboplastin time assay. Briefly, fVa-deficient plasma (45 l) was incubated with Alexin, the aPTT reagent (50 l), and various concentrations of fVa or Fl-fVa (5 l) in buffer containing 50 mM Hepes (pH 7.4), 150 mM NaCl, and 0.5% BSA for 3 min at 37°C. Clotting was initiated by the addition of 30 mM CaCl 2 in the same buffer. Clotting times were recorded using an Amelung KC4A coagulometer (Sigma).
Western Blots-Samples containing Fl-fVa or unlabeled fVa (5 l, 0.8 M) were analyzed using 6% Tris/glycine SDS-PAGE to achieve maximum separation between heavy and light chains of fVa. The protein bands were subsequently transferred for 40 min onto an Immobilon polyvinylidene difluoride (Millipore) membrane at 24 V with a semidry transfer apparatus (Bio-Rad). To improve the quality of the blot, protein-binding sites on the membrane not occupied by protein bands were then blocked with 1% casein in 50 mM Tris/HCl (pH 7.4), 100 mM NaCl, and 5 mM CaCl 2 for 1 h. Thereafter, the membrane was incubated with the primary antibody (5 g/ml) against either the heavy chain of fVa (32) or light chain (clone V59) (33) or the fluorescein dye (Molecular Probes, Eugene OR) for 1 h at room temperature, followed by a 30-min incubation with the biotin-conjugated secondary antibody (1 g/ml) directed against the primary antibody. Finally, the membrane was incubated with a solution containing alkaline phosphatase-conjugated streptavidin (1 g/ml) for 30 min. This step was followed by the addition of an alkaline phosphatase substrate, a mixture of 5-bromo-4-chloro-3Јindolylphosphate and nitro blue tetrazolium chloride (Pierce). Color development was terminated after bands were visualized by washing the substrates off of the membrane with water.
Concentration Determination by Enzyme-linked Immunosorbent Assays (ELISA)-Sandwich ELISAs were performed on Maxisorp (Nalge Nunc, Intl.) 96-well plates. The plates were thoroughly washed with TBS for a minimum of three rinses between each step, as described below, to remove excess reagents. Incubations were for a time period of 1 h at room temperature unless mentioned otherwise. First, the wells were coated with 20 g/ml of rabbit anti-fVa antibody (75 l/well) for 1 h at 37°C (32). Next, other potentially unoccupied protein-binding sites on the wells were blocked nonspecifically with Superblock (Pierce) buffer. Afterward, various dilutions (400 -51,200-fold dilution of stock) of fVa (stock 4.3 mg/ml) or unknown concentrations of Fl-fVa (40 -2560fold dilution) were incubated in the wells for 2 h at room temperature. Subsequently, the plates were incubated with monoclonal anti-fVa detecting antibody (3B1, 5 g/ml) directed against the light chain of fVa (32,33). Thereafter, biotinylated anti-mouse IgG (1 g/ml) was added, followed by incubation with streptavidin/alkaline phosphatase (1 g/ml, 75 l/well). Finally, the amount of fVa was estimated from the alkaline phosphatase activity detected after the addition of p-nitrophenylphosphate substrate (100 l/well) in which hydrolysis was monitored at 405 nm.
Inhibition of Arg 271 Cleavage by PT473-487-Fl-mfIIa i (1.2 M) was incubated with PC/PS vesicles (50 M), fVa (4.2 nM), and varying concentrations of peptide PT473-487 (0 -50 M) in 75 l of buffer containing 50 mM Hepes (pH 7.4), 150 mM NaCl, and 5 mM CaCl 2 for 10 min at room temperature. Thereafter, the hydrolysis of Fl-mfIIa i at Arg 271 was triggered by the addition of fXa (4 nM) for 10 min at 37°C. Subsequently, the reaction was stopped by the addition of 20 l of hot SDS sample buffer. The samples were incubated on a boiling water bath and analyzed by SDS-PAGE, and then fluorescent protein bands were visualized using UV light on an EAGLE-EYE II (Stratagene) imager. The fluorescence image was digitalized, and cleavage at Arg 271 was quantified using UN-SCAN-IT version 5.1 (Silk Scientific Corp., Orem, UT).
Spectral Measurements-Steady-state fluorescence intensity and anisotropy measurements were made using a SLM AB2 Luminescence spectrometer (SLM Aminco, Rochester, NY) or SLM 8100 fluorometer equipped with a 150-W xenon arc lamp, two holographic gratings in the excitation light path, and an IBM computer for data analysis. Fluorescein fluorescence was detected at its excitation and emission maxima wavelengths of 490 and 521 nm, respectively, using a 345-nm cut-off filter (Schott Glass Technologies, Duryea, PA) in the excitation beam path. Shutters were closed, except during measurements, to minimize photodegradation of the samples. A circulating water bath was used to maintain the sample compartment at 25°C. All fluorescence experiments were performed in 5 mm ϫ 5 mm quartz cuvettes. Samples were mixed using a small teflon-coated 2 mm ϫ 2 mm magnetic spinbar as described previously (34,26). Absorption of protein to the cuvette walls was minimized by coating the cuvettes with 100% PC vesicles as described previously (35).
For a typical fluorescence experiment, two samples were prepared in parallel. The sample cuvette (S) received fluorescent protein in 50 mM Hepes (pH 7.4), 150 mM NaCl, and 2 mM CaCl 2 , whereas the blank cuvette (B) received an equal concentration of unlabeled protein in the same buffer. The initial net fluorescence intensity of the sample, designated F o , was obtained by the subtraction of the B cuvette signal from the S cuvette signal. For the titrations involving Fl-fVa, equimolar concentrations of peptides were added to two cuvettes in parallel: one (S) containing 50 nM Fl-fVa and 25 M PC/PS vesicles and the other (B) containing equimolar concentrations of unlabeled fVa and phospholipid vesicles. The signal of the B cuvette was never Ͼ1% of the S cuvette. After the signal observed in the absence of the dye (sample B) was subtracted from that in the presence of the dye (sample S), the net volume-corrected signal was designated F at that point in the titration. The interaction of peptide PT473-487 with lipid-bound Fl-fVa was measured by monitoring the change in the steady-state anisotropy of Fl-fVa.
Steady-state anisotropy was measured using Glan-Thompson prism polarizers on both the excitation and emission beams. The net emission intensity measured when the sample was excited by vertically (V) plane-polarized light and the emission detected through a horizontal (H) polarizer is termed F VH . F HH , F HV , and F VV are defined analogously. The component intensities of a dye-free blank were subtracted from the component sample intensities to give the net emission intensities. Anisotropy (r) was then calculated from the net intensities using r ϭ (F VV Ϫ GF HV )/(F VV ϩ 2GF VH ) where the grating factor G equals F HV /F HH .
The anisotropy of Fl⅐fVa⅐PC/PS complex in the absence of peptide was designated r o , whereas the anisotropy in the presence of peptide was designated r. To obtain binding constants, (r/r o ) Ϫ 1 was plotted against the peptide concentration and the data fit to a hyperbolic profile. The interaction of PC/PS-bound Fl-fVa with prothrombin was monitored analogously. The curves obtained from the fluorescence assays were fit using GraphPad Prism version 3.03 as described previously (31).
Dynamic Quenching-Collision of fluorophores with certain solutes, such as d-orbital containing I Ϫ ions, results in the dissipation of the exited state energy of the fluorophore to the solute molecules (36). This results in a loss of fluorescence signal or quenching. The solvent accessibility of the fluorescein dye in Fl-fVa was ascertained by testing the effect of various concentrations of I Ϫ ion on the fluorescence emission of Fl-fVa. Samples containing Fl-fVa (50 nM initial) were titrated in parallel with increasing concentrations of either NaI or NaCl. Appropriate blanks were subtracted as detailed in the section above. According to Stern-Volmer law, a plot of (F o /F)-1 against quencher concentration [Q] yields a straight line in the case of collisional quenching, where F o is the emission intensity of the dye before the addition of the quencher, and F is the net dilution-corrected emission intensity of the sample at a given point in the titration (36). The slope of the straight line equals the collisional constant (K sv ). The steeper the slope, the higher the quenching, and thus greater the solvent accessibility. The effect of increasing ionic strength because of NaI addition on fluorescence of Fl-fVa was corrected using a NaCl titration. Table I illustrates the homology between the fVabinding sequence in APC comprising residues 311-325 (23) and sequences in fXa and fII. Synthetic peptides of fII and fXa derived from these regions of homology were screened for their ability to inhibit fXa clotting activity in normal human plasma. Table II shows the amino acid sequences that were assayed and the residual fXa activity observed in a fXa-1-stage clotting assay in the presence of 500 M each peptide. The fII-derived sequence, namely PT473-487 (CHT 149D-163), inhibited Ͼ85% of fXa clotting activity at 500 M peptide, whereas the fXa-derived peptide, X329 -344, inhibited ϳ42% of fXa activity. Flanking sequences of PT473-487, namely PT463-477 and PT481-495, did not significantly inhibit clotting. PT473-487 dose-dependently inhibited fXa clotting activity with an EC 50 of ϳ12 M (Fig. 1). In contrast, the scrambled sequence of the peptide PT473-487SCR did not inhibit fXa-induced clotting, showing that the inhibition of fXa activity by PT473-487 is both dose-dependent and sequence-specific.

Inhibition of fXa-induced Clotting Activity by Peptide PT473-487-
Effect of Peptide PT473-487 on the Amidolytic Activity of fXa and fIIa-To test whether peptide PT473-487 inhibited fXainduced clotting because the peptide directly inhibited the active sites of fXa or thrombin, the activities of fXa and thrombin toward their respective small chromogenic substrates S-2222 and CBS 34 -47 (or S-2238) were measured in the presence or absence of PT473-487. PT473-487 did not significantly inhibit

fVa-binding Exosite in Prothrombin
the activities of fXa and thrombin toward their respective oligopeptide substrates over a wide range of substrate (0 -2 mM) and peptide concentration (0 -350 M) (data not shown). Additionally, PT473-487 also did not inhibit the clotting of fibrinogen by thrombin (not shown). These data indicate that PT473-487 does not inhibit the active sites of either fXa or thrombin.
fVa-dependent Inhibition of Prothrombinase by Peptide PT473-487-The inhibitory effect of PT473-487 in the clotting assay was further characterized using prothrombinase assays performed in the presence and absence of the cofactor fVa. PT473-487 was a potent inhibitor of fII activation in the presence of fVa, inhibiting the reaction with an EC 50 of 12 M (Fig.  2). However, the peptide did not significantly inhibit fII activation in the absence of fVa (Fig. 2). Thus, the effect of PT473-487 was fVa-specific. A control peptide with a scrambled sequence of PT473-487, namely PT473-487SCR, did not inhibit prothrombin activation in this purified system in the absence or presence of fVa (Fig. 2). Thus, the inhibition of PT473-487 was sequence-specific.
When the prothrombinase assays were performed at different fII and peptide concentrations and in the presence of fVa, PT473-487 showed an inhibition pattern similar to that of a classical noncompetitive inhibitor (Fig. 3) with all lines arising from different peptide concentrations having the same y intercept but different x intercept (Fig. 3) in a Lineweaver-Burk plot analysis. An inhibition constant (K i ) of ϳ20 M was determined from a replot of different slopes obtained for the Lineweaver-Burk plots against peptide concentration.
Inhibition of Arg 271 Cleavage by Peptide PT473-487-To test if peptide PT473-487 inhibited the hydrolysis of Arg 271 -Thr 272 bond in fII by the prothrombinase complex, an active site specifically labeled meizothrombin, namely fluorescein⅐Phe⅐ Pro⅐Arg⅐meizothrombin (Fl-mfIIa i ) was prepared. Cleavage at Arg 271 was quantitated using SDS-PAGE and densitometry as described under "Experimental Procedures." When Fl-mfIIa i was incubated with 50 M PC/PS vesicles, fVa, and fXa in the presence or absence of varying concentrations of peptide PT473-487, dose-dependent inhibition of Arg 271 -Thr 272 cleavage was observed with an EC 50 of ϳ4 Ϯ 2 M (data not shown), showing that peptide PT473-487 inhibited this bond cleavage as well as cleavage at Arg 320 .
Sulfhydryl-directed Labeling of fVa and Characterization of Fl-fVa-To test the hypothesis that peptide PT473-487 represents a fVa-binding site on fII, fVa was labeled with a fluorescent dye, namely Fl, to yield Fl-fVa. Fl-fVa was then used as a probe for the interaction of the protein with the peptide.
The fluorescein analog reagent labels at thiol groups and thus would label at free cysteines on fVa (26). The dye to protein ratio was determined to be ϳ0.75 using a molar extinction coefficient of 41,870 M Ϫ1 cm Ϫ1 at 484 nm, determined for a known concentration of the free dye in the same buffer. Fig. 4 shows a nonreduced SDS-PAGE analysis of Fl-fVa and the

fVa-binding Exosite in Prothrombin
unlabeled control fVa. Following gel electrophoresis, the protein bands from the SDS gel were transferred onto a membrane and analyzed by Western blot using anti-fluorescein (Fig. 4,  lanes 1-4), anti-fVa light chain (lanes 5-7), and anti-fVa heavy chain (lanes 8 -10) antibodies, respectively. Comparing lanes 2 and 3 in the immunoblot, only lane 3 is visualized by the anti-fluorescein antibody, indicating a covalent attachment of fluorescein to the protein. Two distinct bands were observed in Fig. 4, lane 3, for the Fl-fVa, an intense high molecular weight band corresponding to the heavy chain of fVa (compare with Fig. 4, lanes 8 and 9) and a less intense band corresponding to the light chain of fVa (compare the mobility of bands in lane 3 with that of bands in lanes 6 and 7). Based on the Western blot data, it can be concluded that, although the fluorescent probe primarily labels the heavy chain of fVa, there is measurable labeling in the light chain of fVa. Similar results have been observed for bovine fVa by two other groups using other cysteine-specific labeling reactions (28,29,37).
Next the clotting activity of Fl-fVa and unlabeled fVa were compared in an aPTT clotting assay. Concentrations of fVa were determined by ELISA, and the percentage of active fVa was determined from a fVa aPTT standard curve constructed using unlabeled fVa. Fl-fVa was estimated to be ϳ90% active (data not shown). In prothrombinase assays, Fl-fVa was ϳ75% as active as compared with unlabeled fVa (data not shown). From these experiments, we conclude that the labeling of fVa in the heavy and light chains with a fluorescein moiety did not significantly alter the functional properties of fVa. Our findings for derivatized human fVa are similar to previous reports that labeling bovine fVa with sulfhydryl-specific dyes, such as pyrene maleimides or iodoacetamido-fluorescein, does not lead to significant loss of fVa activity (37,28).
Fl-fVa exhibited a corrected fluorescence emission and excitation maxima at 521 and 490 nm, respectively, in 50 mM Hepes (pH 7.4), 150 mM NaCl, and 5 mM CaCl 2 . The anisotropy (r) of the fluorescein dye on fVa averaged 0.247 Ϯ 0.005 compared with 0.0152 Ϯ 0.001 for the same concentration of free dye, indicating that the fluorescent moiety in Fl-fVa has restricted rotation, presumably because the dye is conjugated to the protein.
The fluorescence intensity of the Fl-fVa signal was quenched by increasing concentrations of NaI but not equivalent concentrations of control NaCl. Thus, the decrease in the Fl-fVa signal was because of collisional quenching, and the quenching constants (K sv ) for NaI quenching at room temperature were determined to be 5.3 Ϯ 0.02 and 8.71 Ϯ 0.01 M Ϫ1 for Fl-fVa and free Fl, respectively. The slopes obtained for the NaCl titrations in a Stern-Volmer plot were essentially parallel to the x-axis, and a K sv of 0.2 M Ϫ1 was determined for these experiments. These results were very similar to earlier reports for fluorescein-labeled bovine fVa (28) and suggest that the fluorescein dye(s) on fVa were accessible to collisional quenchers. A decrease in K sv value for Fl-fVa compared with free Fl implies reduced accessibility for the dye coupled to the protein. However, this result can be reasonably explained because of the loss of some collisional accessibility for the iodide ion because of the conjugation of a phase of the exposed dye to a macromolecular structure (28).
Binding of Peptide PT473-487 to Fl-fVa-To test the hypothesis that PT473-487 binds to membrane-bound fVa, PC/PS vesicle-bound Fl-fVa was titrated with peptide PT473-487 or the scrambled control peptide PT473-487SCR. Initially, samples containing Fl-fVa in a cuvette were titrated with PC/PS vesicles. A small (ϳ5% Ϯ 2%) saturable increase of Fl-fVa anisotropy was observed, which reached a plateau at ϳ8 M lipid. To ensure that Fl-fVa was lipid-bound, an excess of PC/PS vesicles (25 M) were added in all studies before the peptides PT473-487 or PT473-487SCR were titrated into the sample cuvettes. Upon the addition of PT473-487, the anisotropy of Fl-fVa increased by ϳ20% and reached a plateau at ϳ10 M peptide with an apparent K d of 3.9 M (Fig. 5). However, in the case of the functionally inactive scrambled peptide PT473-487SCR, the fluorescence anisotropy of vesicle-bound Fl-fVa did not significantly increase. When fII was titrated into a cuvette containing PC/PS-bound Fl-fVa instead of the peptides, the anisotropy of the fluorescent moiety rose by ϳ21% and reached a plateau at ϳ1.5 M fII (K d app ϭ 0.6 M). These data strongly suggest that membrane-bound Fl-fVa interacts with PT473-487 and fII in a similar manner. The fact that the change in the fluorescein reporter on Fl-fVa was essentially the same made it impossible to use the Fl-fVa to monitor the competition between fII and PT473-487 for binding fVa. DISCUSSION Amino acids 311-325 (CHT 149D-163), which partially represent the autolysis loop (301-316) of protein C, were earlier identified to be critical for the interaction of activated protein C with fVa (23). Here we tested the hypothesis that homologous sequences in fII and fXa contribute functionally important exosites that mediate protein-protein interactions in the prothrombinase complex. To test this hypothesis, one fXa-derived

fVa-binding Exosite in Prothrombin
peptide (X329 -344) and three fII-derived peptides (PT463-477, PT473-487, and PT481-495) representing these regions of homology in fXa and fII were synthesized, purified to homogeneity, and tested in a fXa-induced clotting assay. Of the four peptides tested, only PT473-487 significantly and potently inhibited fXa-induced clotting (Table II and Fig. 1). Furthermore, a scrambled sequence of PT473-487 did not affect the clotting assay, indicating that the inhibition by PT473-487 was sequence-specific (Fig. 1). The inhibition of clotting by PT473-487 was not due to the peptide directly inhibiting the amidolytic activities of fXa or thrombin, because PT473-487 did not significantly alter the activity of either enzyme toward their respective chromogenic substrates (data not shown). Additionally, PT473-487 also did not affect the ability of thrombin to clot fibrinogen. Taken together, these data strongly suggest that PT473-487 does not alter the structural integrity of the active sites of thrombin and fXa. Also noteworthy is the fact that PT473-487 did not alter the ability of Fl-fXa i to bind PC/PS vesicles (data not shown), suggesting that the peptide inhibition was due to interference of the peptide with proteinprotein interactions in the prothrombinase complex.
In purified systems, PT473-487 potently and dose-dependently inhibited prothrombin activation by the prothrombinase complex in the presence, but not in the absence, of fVa (Fig. 2). Thus, the inhibition of prothrombinase activity by PT473-487 was completely fVa-dependent. The inhibition was of a classical noncompetitive nature, thus excluding the involvement of the active sites of thrombin and meizothrombin as a mode of inhibition and suggesting a more complicated mechanism, such as the involvement of exosites distal to the active site (Fig. 3). Thus, PT473-487 inhibits cleavage at Arg 320 . Although amidolytic assays that were used to monitor fII activation detect only the cleavage at Arg 320 in fII, clotting assays detect ␣-thrombin generation due to cleavages at both Arg 320 and Arg 271 . To test whether peptide PT473-487 also inhibits the hydrolysis of Arg 271 -Thr 272 bond cleavage in fII, inhibition of cleavage of Fl-mfIIa i by the prothrombinase complex was determined using SDS-PAGE analysis. Peptide PT473-487 inhibited the conversion of Fl-mfIIa i to Fl-fIIa i with an EC 50 of ϳ4 Ϯ 2 M. Thus, peptide PT473-487 inhibited Arg 271 -Thr 272 cleavage and Arg 320 -Ile 321 cleavage with similar potency.
Because of the fVa dependence on the inhibition of prothrombinase by the peptide, we tested the hypothesis that PT473-487 is a fVa-binding exosite on fII. To do this, fVa was labeled in a sulfhydryl-specific manner with a fluorescein analog to be used as a reporter to probe the interaction between Fl-fVa and PT473-487. Fl-fVa was normally functional in both clotting and prothrombinase assays, indicating that incorporation of the dye did not interfere with the functional properties of fVa. The sulfhydryl-specific fluorescein analog predominantly labeled fVa in the heavy chain, similar to other sulfhydryl-specific dyes reported earlier (28,29). But some labeling was also observed in the light chain subunit of fVa (Fig. 4). Our fluorescence quenching experiments suggest that the dye(s) on fVa are readily available to collisional quenchers and thus are solventaccessible. There are two cysteines that are not involved in disulfide linkages in the A2 domain of the heavy chain of human fVa (38). Reports from earlier studies suggest that sulfhydryl-specific reagents label only one cysteine in the heavy chain of fVa and do so without the loss of activity of the protein. Thus, it was concluded from those studies that the cysteine that reacts with sulfhydryl-specific agents must be solvent-exposed. According to a computer-generated homology model of fVa based on the crystal structure of ceruloplasmin, only one of the cysteines in the A2 domain, namely Cys 539 , is solvent-accessible (39). Therefore, we speculate that the label on the heavy chain is on Cys 539 . The light chain of human fVa has a free cysteine at position 1960 in spike 3 of the C1 domain (38). This cysteine appears partially exposed in the computer model and could be accessible to the labeling reagent (39). Additionally, Tyr 1947 in bovine fVa, which replaces Cys 1960 in human fVa, appears to be surface-exposed in the APC-inactivated crystal structure of bovine fVa (40). Therefore, this cysteine in human fVa could be the site at which the light chain is labeled. More work, such as labeling point mutants, needs to be done to identify which cysteines get labeled.
PT473-487 altered the fluorescence anisotropy of membrane-bound Fl-fVa, whereas the functionally inactive peptide with a scrambled sequence failed to produce an equivalent change in signal, suggesting that PT473-487 binds in a sequence-specific manner to Fl-fVa. The K d app obtained from the biophysical experiments was comparable with the kinetic data.  (47). Shown in purple are the catalytic triad residues CHT His 57 , Asp 102 , and Ser 195 and in red is the region of PT473-487 (CHT 149D-163). Also shown in green are the sites of proexosite-I mutants deficient in fVa interaction (21) in close proximity to PT473-487 residues. In yellow and blue are two previously characterized fXa-binding regions on fII, namely Tyr 557 -Gln 571 on the serine protease domain (31) and His 205 -Arg 220 (48) on the Kringle 2 domain.
Interestingly, fII also perturbed the fluorescence anisotropy of membrane-bound Fl-fVa to approximately the same magnitude as that of the peptide PT473-487. However, fII exhibited a 6.5-fold greater affinity for Fl-fVa compared with the peptide, presumably because of the existence of other exosites on fII that bind fVa (16 -21) or because of conformational differences between peptide and fII.
Recent studies using fluorescently labeled hirudin-derived peptides suggest that the proexosite I of fII might be involved in fVa binding (19). Mutations of the basic residues in proexosite I to acidic glutamine resulted in a fII with impaired fVa binding, suggesting the participation of this fII region in fVa recognition (21). Our data, that PT473-487 inhibits prothrombinase in a fVa-specific manner and binds in a sequence-specific manner to fVa, suggest that the region represented by the peptide could represent a fVa-binding exosite on fII. In the crystal structure of the serine protease domain of fII, PT473-487 is located immediately adjacent to the positive proexosite I of fII (Fig. 6). Therefore, our peptide inhibition data are consistent with the fII site-directed mutagenesis results. Our data support elements of a prothrombinase model in which a contiguous fVa-binding surface on the serine protease domain of fII is diametrically opposed to a fXa-binding surface identified earlier (Fig. 6). These hypothesized binding surfaces are represented in Fig. 6, where the amino acids representing PT473-487 are mapped alongside the proexosite I mutations, and the two fXa-binding sites reported earlier are shown on the opposite side of the fII protease domain.
Biophysical studies of phospholipid membrane-bound prothrombin and meizothrombin indicated that the prothrombin molecule extends radially ϳ110 Å above the membrane surface and that the catalytic triad in meizothrombin is located ϳ67 Å above the membrane surface, when complexed with fVa (41,30). Based on a recent crystal structure of APC-inactivated fVa (40) on a homology model of fVa (39) and on low resolution membrane-bound structures of homologous fVIIIa (42), we speculate that the prothrombin exosite on the serine protease domain identified by PT473-487 would bind to the A1-A2-A3 heterotrimer of fVa in the prothrombinase complex.
In conclusion, using peptide inhibition studies and fluorescence spectroscopy, we have identified new fVa-binding sites on fII that are adjacent to the fVa-binding proexosite 1 residues on fII (19 -21).