Binding of Fibrinogen Aα1–50-β-Galactosidase Fusion Protein to Thrombin Stabilizes the Slow Form

The interaction of fibrinogen Aα1-50-β-galactosidase fusion protein with the slow and fast forms of thrombin was studied and compared to thrombin-fibrinogen interaction under identical solution conditions. At equilibrium, the affinity of the fusion protein for the slow form of thrombin is 3 times higher than its affinity for the fast form. The fusion protein and fibrinogen have the same affinity for the fast form. On the other hand, the affinity of the fusion protein for the slow form of thrombin is 40 times tighter than that of fibrinogen. In the transition state, binding of the fusion protein has the same properties as fibrinogen, with the fast form showing higher specificity. The N-terminal fragment of the fibrinogen Aα chain thus contains residues that are responsible for the preferential binding of the fusion protein to the slow form at equilibrium and to the fast form in the transition state. If this fragment binds to thrombin in a similar way for fibrinogen and the fusion protein, then the N-terminal domains of the Bβ and γ chains of fibrinogen, that are not present in the fusion protein, must play a key role in the binding of fibrinogen to thrombin at equilibrium. These chains may destabilize binding to the slow form by nearly 2.4 kcal/mol, thereby favoring binding of fibrinogen to the fast form. We propose that the three chains of fibrinogen play different roles in the thrombin-fibrinogen interaction, with the Aα chain containing residues for preferential binding to the fast form in the transition state and the Bβ and γ chains containing residues that destabilize binding to the slow form at equilibrium.

Thrombin is an allosteric serine protease involved in blood coagulation. The enzyme exists in two forms, slow and fast (1), that have been targeted toward anticoagulant and procoagulant activities (2). The fast form preferentially cleaves fibrinogen, while the slow form cleaves protein C with higher specificity. The molecular basis of preferential binding of fibrinogen to the fast form (3) has been discussed in connection with the interaction of thrombin with the potent natural inhibitor hirudin (4). Hirudin and fibrinogen bind to thrombin by making contacts with residues located not only in and around the catalytic pocket, but also in the fibrinogen binding loop. This loop is separate from the catalytic pocket and is homologous to the Ca 2ϩ binding loop of trypsin and chymotrypsin (5)(6)(7)(8). Either molecule binds preferentially to the fast form with a free energy difference of about 1.7 kcal/mol (3,4). Energetic mapping of hirudin binding (4) and recent site-directed mutagenesis studies 1 have led to the conclusion that most of the preferential binding of hirudin to the fast form arises from interaction with residues in and around the catalytic pocket. The fibrinogen binding loop provides only a small contribution to the coupling with the allosteric transition. Whether a similar conclusion can be drawn in the case of the structural origin of preferential binding of fibrinogen to the fast form of thrombin remains to be established.
A useful model substrate for thrombin is the tripartite protein consisting of residues 1-50 of the fibrinogen A␣ chain linked by a 59-residue segment of collagen to Escherichia coli ␤-galactosidase (10). The central domain of fibrinogen, represented by the CNBr fragment containing A␣ chain residues 1-51, B␤ chain residues 1-118, and ␥ residues 1-78, makes contact with both the catalytic pocket and the fibrinogen binding loop in thrombin (11). Because the tripartite fusion protein lacks the N-terminal domains of the B␤ and ␥ chains of fibrinogen, it represents a simplified model for studying molecular recognition events involved in the thrombin-fibrinogen interaction. Specifically, comparative studies of the fusion protein and fibrinogen may shed light on the relative contribution of the three chains of fibrinogen. Functional studies indicate that the FpA 2 consisting of residues 1-16 of the A␣ chain is cleaved from the fusion protein by thrombin at a rate comparable to that of fibrinogen (12). This result suggests that the B␤ and ␥ chains may have little influence on the interaction of the A␣ chain with thrombin. However, perturbation of the N-terminal portion of the B␤ chain of fibrinogen often results in impaired clotting activity (13)(14)(15), and interaction of all three chains with the fibrinogen binding loop seems to be crucial for the correct hydrolysis of fibrinogen by thrombin (11,16). In an attempt to dissect the contribution of A␣1-50 from the Nterminal portion of the three fibrinogen chains, we have decided to explore the interaction of the fusion protein with the slow and fast forms of thrombin at equilibrium and in the transition state. Comparison of the results with those obtained with fibrinogen under identical solution conditions indicates that the B␤ and ␥ chains play an important role in the prefer-* This work was supported in part by National Institutes of Health Research Grants HL45100 (to S. T. L.) and HL49413 (to E. D. C.) and by grants from the American Heart Association (to E. D. C.) and the Monsanto-Searle Company (to E. D. C.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ Established Investigator of the American Heart Association and Genentech.
* ential binding of fibrinogen to the fast form by destabilizing binding to the slow form. The fibrinogen A␣ chain, on the other hand, contains all the structural epitopes for preferential cleavage by the fast form. Therefore, the N-terminal portions of the three chains of fibrinogen seem to have been targeted toward different roles in molecular recognition of this substrate by the slow and fast forms of thrombin.

MATERIALS AND METHODS
Human ␣-thrombin was purified and tested for activity as described (1,10,17). The chromogenic substrate S2238 was purchased from Chromogenix (Molndal, Sweden). The fusion protein was expressed, purified, and tested for activity as described (10,12). The release of FpA from the fusion protein was quantified by reverse-phase HPLC (18) using a Vydac C 18 column. Elution was carried out at a flow rate of 1 ml/min, with a gradient containing 25 mM Na 2 HPO 4 /NaH 2 PO 4 buffer at pH 6.0 (solvent A) and 50% acetonitrile in solvent A (solvent B). Optimal separation was obtained using a 30-min linear gradient to 40% of solvent B. The effluent was monitored at 206 nm. A molar absorption coefficient for FpA was determined by calibration curves. The slow and fast forms of thrombin were studied under experimental conditions of 5 mM Tris, 0.1% PEG, pH 8.0, at 25°C, in the presence of 200 mM ChCl (slow form) or NaCl (fast form). Progress curves for the release of FpA were analyzed using the expression for first-order kinetics shown in Equation 1, where e T is the active thrombin concentration, t is time, k cat and K m refer to the hydrolysis of FpA, [F] is the concentration of FpA at time t, and [F] ϱ is the asymptotic concentration of FpA.
This value was consistent with the concentration of fusion protein estimated from SDS-polyacrylamide gel electrophoresis analysis of the purified protein.
The equilibrium constant for the binding of the fusion protein to thrombin was measured using the viscogenic method introduced for the study of fibrinogen binding and described in detail elsewhere (19). In the viscogenic method, the K m for the hydrolysis of FpA from fibrinogen is measured as a result of the competition of the hydrolysis of S2238 by thrombin as a function of fibrinogen concentration. Values of K m determined as a function of the relative viscosity of the solution are analyzed using the expression shown in Equation 2, where ␣ is the ratio between the acylation and dissociation rates, ␤ is the ratio between acylation and deacylation, and rel is the relative viscosity of the medium.
The value of the equilibrium dissociation constant K d is obtained in a plot of K m versus rel as the extrapolation of K m for rel 30. The viscogenic method yields information on the equilibrium components of the binding interaction (K d ), as well as on the rate-limiting events during the catalytic conversion of the substrate leading to the release of FpA (␣ and ␤). The value of K m for the release of FpA from the fusion protein was derived from the competitive effect on the hydrolysis of S2238. Measurements of K m were then carried out as a function of relative viscosity, with thrombin either in slow or fast form. The results were expressed in units of [FpA] to allow a direct comparison between the fusion protein, containing 4 FpA fragments/molecule, and fibrinogen, containing 2 FpA fragments/molecule.

RESULTS
The results of the fusion protein binding to the slow and fast forms of thrombin are shown in Fig. 1 as a plot of K m versus rel . The data obey a straight line in the plot, over the range of relative viscosity values examined, suggesting that ␤ Ϸ 0 in Equation 2. This implies that deacylation and diffusion of the FpA away from the catalytic pocket of thrombin occur on a time scale much faster than acylation, in either the slow or fast forms. The value of ␣ in either form indicates that the fusion protein behaves as a "sticky" substrate, with the dissociation rate being comparable to acylation. This situation is seen for the cleavage of FpA from fibrinogen by the fast form, but not by the slow form (3). The extrapolation of K m for rel 3 0 in Fig. 1 gives the value of K d for the fusion protein binding to the slow and fast forms. Unlike the case seen for fibrinogen (3), the fusion protein preferentially binds to the slow form with an affinity nearly 3 times higher. The affinity for the fast form is about the same as that seen for fibrinogen. In the slow form, on the other hand, the fusion protein binds to thrombin with an affinity 40 times higher compared to fibrinogen (see Table I).
The release of FpA from the fusion protein upon interaction with the slow and fast forms of thrombin is documented in Fig.  2. The fast form cleaves FpA at a rate 9 times faster. Cleavage obeys first-order kinetics in both the fast and slow forms. This behavior parallels the results seen for the release of FpA from fibrinogen under identical solution conditions (2). The release of FpA is in this case nearly 7 times faster for the fast form of thrombin. In either thrombin form, fibrinogen is cleaved at a rate that is about 6 times faster compared to the fusion protein.
Important details on the kinetic mechanism of recognition of fibrinogen and the fusion protein by the slow and fast forms of thrombin are revealed by the individual rate constants for substrate binding (k 1 ), dissociation (k Ϫ1 ), acylation (k 2 ), and deacylation (k 3 ). The constants can be estimated from the data in Table I and Figs. 1 and 2. The values for fibrinogen can be derived from data published previously (2,3). The results are given in Table II. The origin of preferential binding of fibrinogen to the fast form is mostly due to the faster dissociation rate constant of this substrate from the slow form. In the case of the fusion protein, dissociation is much faster from the fast form leading to tighter binding to the slow form. Hence, the B␤ and ␥ chains destabilize binding to the slow form by enhancing the dissociation rate constant. In the absence of these chains, dissociation is faster from the fast form. The rate constants for fibrinogen binding to the slow and fast forms are significantly slower than that found for chromogenic substrates (1) and hirudin (4), but are nearly 1 order of magnitude faster than  Table I, ␤ ϭ 0 and ␣ ϭ 2.2 Ϯ 0.9 (E), or ␤ ϭ 0 and ␣ ϭ 1.4 Ϯ 0.3 (q). Note the higher affinity when thrombin is in the slow form.
those of the fusion protein. Acylation of fibrinogen occurs with comparable rate constants in the slow and fast forms, while acylation of the fusion protein is 20 times faster in the fast form. Hence, a significant conversion of the enzyme from the slow to the fast form may be expected before reaching the transition state for the acylation step in the case of fibrinogen, but not for the fusion protein.
The preferential interaction with either thrombin form is quantified by the coupling free energy, ⌬G c . This terms measures the difference in binding affinity between the fast and slow forms and provides a measure of the coupling between the binding process and the allosteric slow 3 fast transition (4). 1 If binding of a ligand involves residues that contribute equally in the slow and fast forms, then ⌬G c ϭ 0. If binding involves residues that contribute differently in the slow and fast forms, then ⌬G c 0. Hence, the coupling free energy is a very useful parameter to relate structural components to binding energetics of the slow3fast transition. The value of coupling free energy for fibrinogen is Ϫ1.7 Ϯ 0.1 kcal/mol (3). Fibrinogen binds to the fast form with higher affinity and with a free energy difference of Ϫ1.7 Ϯ 0.1 kcal/mol. The value of coupling free energy for the fusion protein is 0.7 Ϯ 0.1 kcal/mol (see Table I). Unlike fibrinogen, the fusion protein binds to the slow form with higher affinity and the value of ⌬G c is positive. The difference in coupling free energy between fibrinogen and the fusion protein is 2.4 Ϯ 0.1 kcal/mol. This quantity must be accounted for by the structural domains of fibrinogen that interact with thrombin and are not present in the fusion protein. A quantity analogous to ⌬G c can be defined in the transition state as a measure of the change in the specificity constant between the slow and fast forms (see Table I). If binding of a substrate in the transition state involves residues that contribute equally in the slow and fast forms, then ⌬G c ϭ 0, otherwise ⌬G c 0. Since FpA is cleaved from fibrinogen or the fusion protein with higher specificity if thrombin is in the fast form, the value of ⌬G c is negative for both substrates. The similarity of coupling free energies suggests that the structural components responsible for the stabilization of the transition state in the fast form relative to the slow form are common to fibrinogen and the fusion protein. DISCUSSION The fusion protein is the first substrate to be found to bind with higher affinity to the slow form of thrombin. All substrates, effectors, and inhibitors studied previously have been reported to bind to the fast form with higher affinity. Such is the case of fibrinogen (3), hirudin and its C-terminal fragment (4), thrombomodulin (2), and a variety of synthetic substrates and inhibitors (1,4). Although the fusion protein binds preferentially to the slow form at equilibrium, FpA is cleaved by the fast form with higher specificity. This result bears directly on thrombin-fibrinogen interaction. Fibrinogen binds preferentially to the fast form (3) and is cleaved by the fast form with higher specificity (2). This observation suggests that the structural components responsible for preferential binding to the fast form at equilibrium are also involved in the preferential stabilization of the transition state in this form. The results for the fusion protein, however, demonstrate that different structural domains of fibrinogen may control molecular recognition and k cat ϭ k 2 k 3 k 2 ϩ k 3 (1,19).  I Properties of the slow and fast forms of thrombin in their interaction with fibrinogen and the fusion protein All values are in units of FpA concentration. The third column reflects the ratio of affinity or specificity between the fast and slow forms. The last column gives the coupling free energy (in kcal/mol), ⌬G c ϭ ϪRTlnr, where R is the gas constant and T is the absolute temperature (2).  at equilibrium and at the transition state. In addition, the results reinforce the notion that the N-terminal domains of the B␤ and ␥ chains of fibrinogen play a key role in molecular recognition of the natural substrate. These chains presumably destabilize binding to the slow form. The binding affinities of the fusion protein and fibrinogen are the same when thrombin is in the fast form, but differ by a factor of 40 when thrombin is in the slow form. If the sequence 1-50 of the A␣ chain present in the fusion protein makes contacts with thrombin as the analogous sequence in the fibrinogen molecule, then this sequence must bind to the slow form with higher affinity. Preferential binding to the fast form, as seen in the case of fibrinogen, would result from contacts made by the N-terminal domains of the B␤ and ␥ chains with thrombin. These contacts would provide 2.4 kcal/mol toward the stabilization of the fast form. Alternatively, the B␤ and ␥ chains may constrain the A␣ chain of fibrinogen such that the contacts with thrombin are not analogous to those in the fusion protein.
The lack of structural information on the thrombin-fibrinogen complex makes assignment of residues involved in recognition very difficult. A structure of thrombin covalently bound to the fragment 1-16 of the fibrinogen A␣ chain has documented the expected contacts with primary recognition subsites in the catalytic pocket, along with hydrophobic contacts with the aryl binding site of thrombin and Gly-216 (8). A seemingly important salt bridge between Arg-173 of thrombin and Glu-11 of fibrinogen has also been reported in the crystal structure. This assignment, however, may be questionable since the side chain of Arg-173 is disordered. The E11A replacement in the fusion protein is without effect (11,12), but the E11G mutation in fibrinogen Mitaka II impairs thrombin binding (20). The mutation R173E of thrombin decreases the release of FpA and FpB from fibrinogen by a factor of 3 and 2, respectively, 1 which is an effect too small to be assigned to the lack of an important salt bridge interaction. The fibrinogen binding loop provides a significant portion of the binding free energy (19), but contributes very little to the value of ⌬G c (4). 1 This is because residues of the fibrinogen binding loop contribute almost equally to binding in the slow and fast forms. Mutagenesis studies of thrombin have indicated that Lys-60f, located strategically in between the catalytic pocket and the fibrinogen binding loop, may play a significant role in the recognition of fibrinogen since the mutant K60fE has a reduced clotting activity (9). This residue is also important in the binding of hirudin (5). The portion of the NSDK of fibrinogen interacting with the region of thrombin surrounding Lys-60f may hold the key to unravel the contribution of the B␤ and ␥ chains to the destabilization of binding to the slow form. The significant differences seen at equilibrium between fibrinogen and the fusion protein disappear in the transition state. Preferential binding to the fast form in the transition state must originate from contacts made with residues within the catalytic moiety and the recognition subsite Asp-189 of thrombin. All the molecular components responsible for the preferential interaction are contained in the A␣1-50, with no apparent contribution from the N-terminal domains of the B␤ and ␥ chains. The three chains in the NSDK of fibrinogen appear to have different roles in the recognition mechanism. The A␣ chain contains the structures required for recognition by the fast form in the transition state, while the B␤ and ␥ chains contain the structures that destabilize binding to the slow form at equilibrium. In this model the B␤ and ␥ chains act as intramolecular allosteric effectors of the A␣ chain. They destabilize binding to the slow form, inducing the slow 3 fast transition, which in turn facilitates binding of the A␣ chain in the transition state and the release of FpA by the fast form.
The observation that the fusion protein binds to the slow form with higher affinity is intriguing and represents a significant step toward our understanding of the molecular basis of the slow 3 fast transition of thrombin and the structural epitopes important for fibrinogen recognition. An important implication of our results is that synthetic inhibitors tailored after the 1-50 segment of the A␣ chain of fibrinogen may work as effective stabilizers of the anticoagulant slow form of thrombin. Such inhibitors would also be effective in enhancing the enzyme specificity toward protein C and may reveal key details involved in the interaction of thrombin with thrombomodulin.