Recombinant Fibrinogen Studies Reveal That Thrombin Specificity Dictates Order of Fibrinopeptide Release*

During cleavage of fibrinogen by thrombin, fibrinopeptide A (FpA) release precedes fibrinopeptide B (FpB) release. To examine the basis for this ordered release, we synthesized A′β fibrinogen, replacing FpB with a fibrinopeptide A-like peptide, FpA′ (G14V). Analyses of fibrinopeptide release from A′β fibrinogen showed that FpA release and FpA′ release were similar; the release of either peptide followed simple first-order kinetics. Specificity constants for FpA and FpA′ were similar, demonstrating that these peptides are equally competitive substrates for thrombin. In the presence of Gly-Pro-Arg-Pro, an inhibitor of fibrin polymerization, the rate of FpB release from normal fibrinogen was reduced 3-fold, consistent with previous data; in contrast, the rate of FpA′ release from A′β fibrinogen was unaffected. Thus, with A′β fibrinogen, fibrinopeptide release from the β chain is similar to fibrinopeptide release from the α chain. We conclude that the ordered release of fibrinopeptides is dictated by the specificity of thrombin for its substrates. We analyzed polymerization, following changes in turbidity, and found that polymerization of A′β fibrinogen was similar to that of normal fibrinogen. We analyzed clot structure by scanning electron microscopy and found that clots from A′β fibrinogen were similar to clots from normal fibrinogen. We conclude that premature release of the fibrinopeptide from the N terminus of the β chain does not affect polymerization of fibrinogen.

During cleavage of fibrinogen by thrombin, fibrinopeptide A (FpA) release precedes fibrinopeptide B (FpB) release. To examine the basis for this ordered release, we synthesized A␤ fibrinogen, replacing FpB with a fibrinopeptide A-like peptide, FpA (G14V). Analyses of fibrinopeptide release from A␤ fibrinogen showed that FpA release and FpA release were similar; the release of either peptide followed simple first-order kinetics. Specificity constants for FpA and FpA were similar, demonstrating that these peptides are equally competitive substrates for thrombin. In the presence of Gly-Pro-Arg-Pro, an inhibitor of fibrin polymerization, the rate of FpB release from normal fibrinogen was reduced 3-fold, consistent with previous data; in contrast, the rate of FpA release from A␤ fibrinogen was unaffected. Thus, with A␤ fibrinogen, fibrinopeptide release from the ␤ chain is similar to fibrinopeptide release from the ␣ chain. We conclude that the ordered release of fibrinopeptides is dictated by the specificity of thrombin for its substrates. We analyzed polymerization, following changes in turbidity, and found that polymerization of A␤ fibrinogen was similar to that of normal fibrinogen. We analyzed clot structure by scanning electron microscopy and found that clots from A␤ fibrinogen were similar to clots from normal fibrinogen. We conclude that premature release of the fibrinopeptide from the N terminus of the ␤ chain does not affect polymerization of fibrinogen.
Fibrinogen is a 340-kDa plasma protein that is involved in the final phase of the coagulation cascade. Fibrinogen consists of two pairs of three polypeptide chains (A␣, B␤, and ␥) that fold to produce a trinodular protein with two distal (D) nodules connected to a central nodule (E) by coiled-coil regions. The central nodule of the molecule consists of the N termini of all six polypeptide chains, and the D nodules consist predominantly of the C termini of the ␤ and ␥ chains, each folded into a globular domain. To initiate polymerization, the serine protease thrombin cleaves four specific Arg-Gly bonds at the N termini of both the A␣ and B␤ chains, releasing fibrinopeptides A (FpA) 1 and B (FpB), respectively. The release of FpA, a 16-residue peptide, exposes the "A" site, which noncovalently interacts with the "a" site in the ␥ chain of the D nodule of another molecule. This A:a interaction results in the linear arrangement of half-staggered, double-stranded protofibrils (1). The release of FpB, a 14-residue peptide, exposes the "B" site (2)(3)(4), which presumably interacts with a "b" site in the ␤ chain of the D nodule of another molecule (5). This B:b interaction is thought to be responsible for lateral aggregation of protofibrils to form fibers (2) and to be analogous to the A:a interaction; however, the mechanism of this interaction is not yet well understood. The final product of this polymerization is a complex, branching network of fibers.
The interactions of thrombin with fibrinogen have been extensively studied (Refs. 6 and 7; for a review, see Ref. 8). Thrombin contains three domains that interact with fibrinogen: the active site, an apolar specificity pocket, and a fibrinogen-binding exosite. The exosite, also called the fibrinogen recognition site, confers the specificity with which thrombin binds to fibrinogen. Upon binding, thrombin cleaves FpA and initiates polymerization. FpA release from fibrinogen follows first-order kinetics, described by the kinetic constant k 1 . In contrast, FpB is released from fibrinogen at a slow initial rate (3,9,10), which is then increased upon polymerization (2, 9 -11) and the depletion of FpA as a substrate for thrombin. This efficient release of FpB, subsequent to FpA release, follows first-order kinetics and is described by the kinetic constant k 2 , assuming that the release of fibrinopeptides A and B from thrombin occurs through two successive first-order processes.
Crystallographic data have depicted the contacts between FpA and thrombin (12); but, to date, similar data are not available for thrombin and FpB. The sequence of FpB is different from that of FpA (13) such that FpB should require different contacts with thrombin (12). Because we do not have structural information, our understanding of FpB release is indirect and based on kinetic studies that examined the timing of FpB release. These studies have shown that the majority of FpB is released from fibrin after FpA has been removed (3,14) and that the rate of FpB release is enhanced upon polymerization of des-A polymers (2)(3)(4)15). To date, however, the mechanism responsible for the delay in FpB release remains undetermined.
To extend our understanding of the mechanism of thrombin on fibrinopeptide B and its resulting effect on polymerization, we have designed a variant recombinant fibrinogen (AЈ␤) that contains a fibrinopeptide A-like substrate on the N termini of the ␤ chains. This fibrinogen was designed to probe the mechanism of fibrinopeptide release and more specifically to deter-mine whether the delayed release of FpB is a consequence of its affinity for thrombin or of its location on the N terminus of the ␤ chain. Our studies have revealed that it is the specificity of thrombin for FpB that is responsible for the order of fibrinopeptide release during fibrin polymerization.

EXPERIMENTAL PROCEDURES
Materials-HEPES, ⑀-aminocaproic acid, sodium phosphate, and all other reagents were obtained from Sigma unless otherwise noted. Human ␣-thrombin was purchased from Enzyme Research Labs (South Bend, IN); two different lot numbers were used for this work: HT1330, 1.85 mg/ml equivalent to 5655 units/ml; and HT1540PA, 1.96 mg/ml equivalent to 6392 units/ml. Human ␣-thrombin (4300 units/ml, 1.71 mg/ml) was a generous gift from Dr. Frank Church (University of North Carolina, Chapel Hill). Recombinant hirudin was purchased from Calbiochem-Novabiochem. All restriction enzymes were obtained from New England Biolabs Inc. (Beverly, MA). Glutaraldehyde, osmium tetroxide, and sodium cacodylate were all electron microscopy-grade materials obtained from Electron Microscopy Sciences (Fort Washington, PA). Eight-well strip polymerase chain reaction tubes were obtained from NalgeNunc (Naperville, IL). UV-transparent 96-well microtiter plates (catalog no. 3635) were purchased from Corning Costar. All plasmid vectors, Chinese hamster ovary cells, bacteria, and culture medium have been previously described (16). Monoclonal antibody IF-1 (17) was obtained from Iatron Lab, Inc. (Tokyo).
Expression Vector Construction-Cloning procedures and vectors were as described previously (16). The expression plasmid for the variant AЈ␣ chain (pMLP-AЈ␣) was assembled from two encoding fragments and the expression vector p284. A Gly-to-Val substitution at position 14 was introduced by oligonucleotide-directed mutagenesis of the human A␣ chain cDNA cloned in a single-stranded phage using the T7-GEN in vitro mutagenesis kit (U. S. Biochemical Corp.) and the oligonucleotide 5Ј-AAC GCG TGG GCC ACG CAC GAC TC. The resulting clone encoded Val at position 14 of the A␣ chain and contained a new MluI site (underlined). The altered segment was isolated from this clone as an 870-bp PvuII/BglII fragment (40 bp of 5Ј-untranslated sequence, the signal sequence, and codons for AЈ␣ residues 1-174) and ligated to a 1360-bp BglII/SspI fragment (codons for AЈ␣ residues 174 -625 plus 10 bp of 3Ј-untranslated sequence) isolated from a plasmid containing the A␣ cDNA; the product fragment (2230 bp) was ligated into expression vector p284 cleaved with SmaI, and the products were transformed into competent DH5␣ cells. Plasmid DNAs from individual colonies were screened by restriction enzyme analysis, including the newly introduced MluI site, and candidate clones were sequenced as described (16). The complete AЈ␣ coding segment was sequenced to demonstrate that the desired codon change was present, and no unanticipated changes were introduced.
The expression plasmid for the variant AЈ␤ chain (pMLP-AЈ␤) was prepared by ligation of a polymerase chain reaction fragment encoding the ␣ chain signal sequence and FpAЈ and a fragment encoding B␤ residues 15-461 into expression vector p284. The polymerase chain reaction fragment was prepared by amplification of a segment from pMLP-AЈ␣ with oligonucleotides 5Ј-TCCACAACCCTTGGGCCACG-GACG (which introduced a DsaI site, underlined) and 5Ј-TTCAGATCT-GGCCATACACTT; the product was cleaved with SalI and DsaI, and the 156-bp fragment was isolated by electroelution following agarose gel electrophoresis. The fragment encoding B␤ residues 15-461, along with the stop codon and 10 bp of 3Ј-untranslated sequence, was isolated from pMLP-B␤ as a 1350-bp fragment following cleavage with DsaI and SnaBI. The p284 vector was cleaved with SmaI and SalI. Following ligation into the vector, individual colonies were screened by restriction enzyme analysis, and candidate clones were sequenced. The sequence demonstrated that the junction between FpAЈ and B␤ residue 14 was as intended and that no unanticipated changes were introduced.
Recombinant Fibrinogen Synthesis and Purification-The normal, AЈ␣, and AЈ␤ clones secreting the highest fibrinogen concentration were selected, and fibrinogens were synthesized as described previously (18). Briefly, Chinese hamster ovary cell lines expressing fibrinogen were grown in roller bottles in serum-free medium, and the medium was harvested and stored at Ϫ70°C with protease inhibitors (16). Of note, the expression of AЈ␣ fibrinogen was approximately one-tenth of the expression of normal or AЈ␤ fibrinogen. These proteins were purified as described (19). In brief, the medium was thawed at 37°C, and fibrinogen was precipitated with a 40% ammonium sulfate cut. The precipitate was redissolved in buffer containing 10 mM CaCl 2 and applied to a Sepharose 4B column coupled with monoclonal antibody IF-1, specific for fibrinogen. Fibrinogen was eluted with buffer containing 5 mM EDTA; dialyzed against 20 mM HEPES (pH 7.4), 150 mM NaCl, and 1 mM CaCl 2 for one change and then extensively against 20 mM HEPES (pH 7.4) and 150 mM NaCl; and stored at Ϫ70°C. The fibrinogen concentration was determined at A 280 using the extinction coefficient ⑀ 280 ϭ 1.506 for a 1 mg/ml fibrinogen solution (20). Purity of the proteins was analyzed by 9% reducing SDS-polyacrylamide gel electrophoresis following the method of Laemmli (21).
Thrombin-catalyzed Fibrinopeptide Release-The thrombin-catalyzed release of fibrinopeptides A, B, and AЈ from normal, AЈ␣, and AЈ␤ fibrinogens was performed essentially as described (18). Fibrinogen solutions were diluted to 0.3 M in 20 mM HEPES (pH 7.4), 150 mM NaCl, 5 mM ⑀-aminocaproic acid, and 1 mM CaCl 2 . ⑀-Aminocaproic acid was included to inhibit any possible plasmin contaminant activity. Thrombin was added to the fibrinogen solutions to a final concentration of 0.005 units/ml (0.054 nM). The tubes were mixed by inversion, and the digests were aliquoted into 440-l portions for the 1-, 2-, and 5-min time points and into 220-l aliquots for the 10-, 40-, 80-, 120-, and 180-min time points. All of the manipulations following thrombin addition were completed within 1 min. The aliquots were incubated at ambient temperature, and the reaction was halted by immersion in boiling water for 15 min. The samples were kept on ice for the remainder of the time course. At the end of the reaction, all samples were centrifuged for 15 min at 4°C, and the supernatants were removed and stored at Ϫ70°C prior to analysis by reversed-phase HPLC. Zero-time point controls were prepared by boiling 430 l of the fibrinogen solution in the absence of thrombin.
To increase the sensitivity at early time points in the reaction, the assays were also performed in larger volumes, with less thrombin. This macroscale assay on normal and AЈ␤ fibrinogens was performed as described above, with the following modifications. The thrombin concentration was 0.002 units/ml (0.018 nM), and the fibrinogen solution was diluted in 20 mM HEPES (pH 7.4) and 150 mM NaCl. For this assay, the aliquots at the 0-, 1-, 2-, and 5-min time points each contained 2.2 ml; those at the 10-, 15-, and 30-min time points each contained 1.1 ml; and those at the 45-, 60-, and 180-min time points each contained 220 l. These reactions were stopped by the addition of 1 l of 2000 units/ml hirudin. The samples were boiled and centrifuged, and the supernatants were stored for later HPLC analysis as described above.
Similar assays were performed in the presence of Gly-Pro-Arg-Pro to examine the role of polymerization in the kinetics of fibrinopeptide release from normal and AЈ␤ fibrinogens. Fibrinogen was diluted to a final concentration of 0.3 M in 20 mM HEPES (pH 7.4), 150 mM NaCl, and 1 mM Gly-Pro-Arg-Pro acetate salt peptide (GPRP); this concentration of GPRP was sufficient to abolish an increase in turbidity at 350 nm, due to polymerization, for 3 h. Thrombin was added to a final concentration of 0.002 units/ml (0.017 nM), and the samples were mixed by inversion and aliquoted into 440-l aliquots for 2-and 5-min time points and into 220-l aliquots for 10-, 15-, 30-, 45-, 60-, and 180-min time points. The samples were set in boiling water for 15 min, centrifuged, and stored for later HPLC analysis as described above.
Fibrinopeptide release was monitored by reversed-phase HPLC as described (18). Briefly, the samples were loaded onto an Isco HPLC system with a Vydac C 18 column equilibrated with buffer A (25 mM NaH 2 PO 4 /Na 2 HPO 4 (pH 6.0)). Samples containing 200 l, 400 l, 1 ml, or 2 ml were loaded onto this column, depending on the experiment and the time point. Fibrinopeptides were eluted with a linear gradient from 100% buffer A to 40% buffer B (25 mM NaH 2 PO 4 /Na 2 HPO 4 (pH 6.0) with 50% acetonitrile) and monitored by absorbance at 206 nm. The retention times for FpAЈ differed from those for FpA and FpB (22). The differences in molar absorption for FpA, FpAЈ, and FpB were not part of our analysis, even though the molar absorption of FpA and FpB is slightly different at A 206 (23). Fibrinopeptide peak area was determined using the accompanying Isco software (Chemresearch Version 2.4). Fibrinopeptide release curves were prepared by plotting the percent release versus time as described (18).
All FpA and FpAЈ data were fitted with a simple first-order equation. The FpB data from normal and AЈ␣ fibrinogens were fitted to a stand-ard equation describing two consecutive first-order processes as described (23). The curves described by these equations were plotted using Delta Graph (DeltaPoint, Inc., Monterey, CA). Specificity constants (k cat /K m ) were determined by dividing first-order rate constants by the thrombin concentration as described (18).
Polymerization of Recombinant Fibrinogens-Polymerization was monitored at 350 nm in a SpectraMax-340PC 96-well microtiter plate reader (Molecular Devices, Sunnyvale, CA) at ambient temperature. Three separate experiments were performed in quadruplicate for each polymerization condition. For each row used in the plate, four wells contained normal fibrinogen, and four wells contained AЈ␤ fibrinogen. To each reaction well was added 90 l of normal recombinant or AЈ␤ fibrinogen in 20 mM HEPES (pH 7.4), 150 mM NaCl, and 1 mM CaCl 2 . To initiate the polymerization reaction, 10 l of thrombin (4 or 1 units/ml) was added to all reaction wells with a multichannel pipette such that all reactions began simultaneously. Immediately after the addition of enzyme, the samples were automixed by the instrument for 5 s. Turbidity was monitored every 10 s for 1 h. All turbidity readings were normalized to a 1-cm path length by the PathCheck sensor within the instrument. Lag time and V max were determined for each polymerization reaction. Lag time was measured as the time elapsed until an increase in turbidity was seen, and V max was calculated as the slope of the steepest part of the polymerization curve.
Scanning Electron Microscopy-Clots were formed under the same conditions as described for polymerization. For each condition, scanning electron microscopy was performed on two clots each with at least two separate microscopy preparations. The scanning electron microscopy preparation was performed as described (24) with minor modifications. In short, clots were polymerized in caps of eight-well strip polymerase chain reaction tubes from which the bottoms had been cut. One side of each cap was sealed with Parafilm, and 45 l of fibrinogen solution was added. A 5 l-sample of enzyme was added to each well, and the samples were mixed with the pipette tip. Polymerization proceeded in a moist environment at ambient temperature for 4 h. The Parafilm was gently removed, and the caps were rinsed in 0.05 M sodium cacodylate buffer (pH 7.3) for 15 min with three changes. The clots were then fixed in 2% glutaraldehyde overnight, rinsed again in sodium cacodylate with three changes, and stained with 2% osmium tetroxide for 30 min. The clots were rinsed with distilled water and dehydrated with a series of increasing concentrations of ethanol for 10 min each, up to 100% ethanol. The samples were critical point-dried in a Balzers CPD020 for ϳ1 h, mounted, sputter-coated with ϳ20 nm of gold-palladium, and viewed on a Cambridge StereoScan S200 (LEO Electron Microscopy, Thornwood, NY). All images were taken at ϫ16,200 with a 17.0-mm working distance and 20.0-kV accelerating voltage. Fiber diameters were calculated using ScionImage (Scion Corp., Frederick, MD). Statistical analysis comparing fiber diameters was carried out by unpaired t test using StatView. A difference is significant when p Ͻ 0.05.

RESULTS
Characterization of Recombinant Fibrinogens-We synthesized two variant fibrinogens with altered fibrinopeptides AЈ where glycine at position 14 was changed to valine. We chose the G14V substitution based on the previous result that the specificity of thrombin for the A␣-(1-51) fusion substrate with this substitution was not significantly altered compared with normal FpA (22). In AЈ␣ fibrinogen, FpAЈ replaced FpA on the two ␣ chains, and in AЈ␤ fibrinogen, FpAЈ replaced FpB on the two ␤ chains. SDS-polyacrylamide gel electrophoresis analysis run under reducing conditions showed that each variant had the expected bands corresponding to the A␣, B␤, and ␥ chains of fibrinogen (Fig. 1) and that no contaminating proteins were detected. The expected fibrinopeptides were released following digestion with thrombin, confirming the presence of the intended sequence substitutions. The HPLC chromatograms shown in Fig. 2 demonstrated that the G14V substitution in FpA altered the HPLC retention time relative to that for FpA or FpB as described previously (22). As expected, the fibrinopeptides from AЈ␣ fibrinogen had retention times characteristic of FpAЈ and FpB (Fig. 2B), and the fibrinopeptides from AЈ␤ fibrinogen had retention times characteristic of FpA and FpAЈ (Fig. 2C).
Thrombin-catalyzed Fibrinopeptide Release-We examined the rate of thrombin-catalyzed fibrinopeptide release by measuring the peak areas of FpA, FpAЈ, and FpB as detected by HPLC and plotting the data as the percent fibrinopeptide release with time. Representative curves are shown in Figs. 3-5; specificity constants from multiple averaged experiments are presented in Table I. AЈ␣ fibrinogen was synthesized to test the effect of introduc-

FIG. 2. HPLC chromatograms depict retention times of fibrinopeptides released from normal (A), A␣ (B), and A␤ fibrinogens (C).
ing the G14V substitution on FpA release. When comparing AЈ␣ fibrinogen with normal fibrinogen, FpA and FpAЈ release data were fit to a first-order rate equation, whereas FpB release data were fit to a standard equation describing two consecutive first-order processes (23). Specificity constants (k cat / K m ) were calculated as the rate constant (k 1 for FpA and FpAЈ and k 2 for FpB) divided by the thrombin concentration as described previously (23). Representative data, presented in Fig. 3A, showed that the release of FpAЈ from AЈ␣ fibrinogen was modestly slower than FpA release from normal fibrinogen. Nevertheless, the specificity constants, presented in Table I (part A), were similar (k cat /K m for normal FpA ϭ (14.7 Ϯ 4.5) ϫ 10 6 M Ϫ1 s Ϫ1 and for FpAЈ ϭ (9.7 Ϯ 1.3) ϫ 10 6 M Ϫ1 s Ϫ1 ). These results demonstrated that FpAЈ was comparable to normal FpA as a substrate for thrombin. We found that the rate of FpB release from AЈ␣ fibrinogen was faster than FpB release from normal fibrinogen (Fig. 3A). The specificity constant for FpB  The significance of this finding remains unresolved. AЈ␤ fibrinogen was synthesized to determine the effect of substituting FpAЈ for FpB on the kinetics of fibrinopeptide release. The fibrinopeptide release data showed that the shape of the curve describing FpAЈ release from AЈ␤ fibrinogen more closely resembled an FpA release curve than a FpB release curve (Fig. 3B). Residual analyses of curves fit (25) to either a first-order rate equation or two consecutive first-order equations revealed that the release of FpAЈ from AЈ␤ fibrinogen was best described by a first-order rate equation. Thus, all specificity constants were calculated with first-order equations for FpA and FpAЈ and with two consecutive first-order equations for FpB. Comparing the specificity constants (Table I, part A) for the release of FpA ((8.1 Ϯ 2.4) ϫ 10 6 M Ϫ1 s Ϫ1 ) and FpAЈ ((6.9 Ϯ 2.1) ϫ 10 6 M Ϫ1 s Ϫ1 ) from AЈ␤ fibrinogen showed that the release of these fibrinopeptides was similar. Representative curves describing FpA and FpAЈ release were similar (Fig. 3B); however, the amount of FpAЈ release was always less than the amount of FpA released. In comparison with normal fibrinogen, the specificity constant for FpA cleavage from AЈ␤ fibrinogen ((8.1 Ϯ 2.4) ϫ 10 6 M Ϫ1 s Ϫ1 ) was about half that of FpA release from normal fibrinogen ((14.7 Ϯ 4.5) ϫ 10 6 M Ϫ1 s Ϫ1 ) ( Table I, part A). FpAЈ cleavage from AЈ␤ fibrinogen ((6.9 Ϯ 2.1) ϫ 10 6 M Ϫ1 s Ϫ1 ) was similar to FpB cleavage from normal fibrinogen ((4.7 Ϯ 1.4) ϫ 10 6 M Ϫ1 s Ϫ1 ). Representative curves describing the release of FpAЈ from AЈ␤ fibrinogen and FpB from normal fibrinogen are qualitatively different from one another, as depicted in Fig. 3B, despite their similar specificity constants.
To better examine the early time points, we increased the sensitivity of the assay by 1) slowing down the reaction with lower thrombin concentrations and 2) performing the assay on a macroscale, by increasing the reaction volume and thus increasing the amount of fibrinopeptides measured. The results of this assay (Fig. 4) revealed that fibrinopeptides from both fibrinogens were released from the beginning of the reaction, although the amount of FpB released was much lower than the amount of each of the other fibrinopeptides (Fig. 4B). Specifically, we noted that there was not a delayed enhancement of the rate of FpAЈ release from AЈ␤ fibrinogen, as seen in FpB release from normal fibrinogen. Comparing the specificity constants (Table I, part B), the rate of FpA release from AЈ␤ fibrinogen ((16.5 Ϯ 6.7) ϫ 10 6 M Ϫ1 s Ϫ1 ) was less than the rate of FpA release from normal fibrinogen ((20.2 Ϯ 2.9) ϫ 10 6 M Ϫ1 s Ϫ1 ), but greater than the rate of FpAЈ release from AЈ␤ fibrin-ogen ((8.6 Ϯ 0.1) ϫ 10 6 M Ϫ1 s Ϫ1 ). The rate of FpAЈ release from AЈ␤ fibrinogen was similar to the rate of FpB release from normal fibrinogen ((9.3 Ϯ 0.2) ϫ 10 6 M Ϫ1 s Ϫ1 ), although the curve describing FpAЈ release from AЈ␤ fibrinogen was qualitatively different from the curve describing FpB release from normal fibrinogen.
To examine whether polymerization influences the release of fibrinopeptides from AЈ␤ fibrinogen, as it does from normal fibrinogen (3,14), we monitored fibrinopeptide release in the presence of GPRP, a known inhibitor of fibrin polymerization (26). In agreement with previous reports (3,14), we found that the release of FpA from normal fibrinogen was unchanged, but the release of FpB was markedly slower in the presence of GPRP (Fig. 5). In contrast, with AЈ␤ fibrinogen, neither FpA release nor FpAЈ release was changed in the presence of GPRP. Comparing the specificity constants from the averaged data (Table I, part C), we saw that with normal fibrinogen, the release of FpB was 3-fold slower (compare (1.9 Ϯ 0.4) ϫ 10 6 M Ϫ1 s Ϫ1 with (4.6 Ϯ 1.4) ϫ 10 6 M Ϫ1 s Ϫ1 ) in the presence of GPRP, whereas FpA was unchanged (compare (13.5 Ϯ 1.4) ϫ 10 6 M Ϫ1 s Ϫ1 with (14.7 Ϯ 4.5) ϫ 10 6 M Ϫ1 s Ϫ1 ). In contrast, with AЈ␤ fibrinogen, the release of FpAЈ was similar in the presence of GPRP (compare (4.7 Ϯ 1.3) ϫ 10 6 M Ϫ1 s Ϫ1 with (6.9 Ϯ 2.1) ϫ 10 6 M Ϫ1 s Ϫ1 ), as was the release of FpA (compare (7.2 Ϯ 3.5) ϫ 10 6 M Ϫ1 s Ϫ1 with (8.1 Ϯ 2.4) ϫ 10 6 M Ϫ1 s Ϫ1 ). We conclude that unlike the release of FpB, the release of FpAЈ from the N terminus of the ␤ chain was not influenced by polymerization.
Thrombin-catalyzed Fibrin Polymerization-To determine whether the altered kinetics of fibrinopeptide release were associated with altered polymerization, we monitored the polymerization of normal and AЈ␤ fibrinogens. The polymerization of AЈ␤ fibrinogen as measured by turbidity at 350 nm was similar to the polymerization of normal fibrinogen, as shown in Fig. 6. We measured the polymerization parameters lag time, V max , and final turbidity for both proteins and found no significant differences between these parameters for AЈ␤ and normal fibrinogens (Table II).
Scanning Electron Microscopy Results-We examined the final fibrin clot structures for both normal and AЈ␤ fibrinogens by scanning electron microscopy. The two clots were similar in appearance (Fig. 7), and measurement of the fiber diameters showed that there was not a significant difference between the diameters of AЈ␤ fibers (136 Ϯ 26 nm) and normal fibrin fibers (148 Ϯ 27 nm; p ϭ 0.16). This finding was consistent with the polymerization curves for normal and AЈ␤ fibrinogens, which had similar final A 350 values, indicative of similar fiber diameters (27).  4), 150 mM NaCl, and 1 mM GPRP. All FpA and FpAЈ release curves were fit to a first-order rate equation, whereas FpB release curves were fit to an equation describing two consecutive firstorder equations. f and q, FpA and FpB release, respectively, from normal fibrinogen; Ⅺ and E, FpA and FpAЈ release, respectively, from AЈ␤ fibrinogen.

DISCUSSION
We have synthesized a recombinant fibrinogen (AЈ␤) to determine what effect the substitution of FpAЈ on the N terminus of the ␤ chain has on the kinetics of fibrinopeptide release. The release of FpA and FpAЈ from AЈ␤ fibrinogen occurred from the beginning of the reaction and followed first-order kinetics, indicating that both FpA and FpAЈ are equally competitive substrates for thrombin and have cleavage sites that are accessible to thrombin at the beginning of the reaction. Thus, the assumption used to characterize FpB release, that FpB release depends on prior FpA release, does not accurately describe the release of FpAЈ from AЈ␤ fibrinogen. In addition, the similar rate of release of either FpAЈ or FpB from the ␤ chain indicates that despite the fibrinopeptide placed on the N terminus of the ␤ chain, the rate of cleavage of the Arg-Gly bond by thrombin remains constant. Taken together, we conclude that it is thrombin, in its interaction with FpB, that is responsible for the delay in efficient FpB release during fibrin polymerization, i.e. the specificity of thrombin for FpB, and not its location on the N terminus of the ␤ chain, accounts for the kinetics of fibrinopeptide release from fibrinogen.
Additional experiments support the conclusion that FpAЈ release from AЈ␤ fibrinogen is dictated by thrombin specificity and not fibrin polymerization. We found that FpAЈ release is qualitatively different from FpB release from normal fibrinogen, indicating that the release of FpAЈ from the ␤ chain is not dependent on prior polymerization of des-A monomers. This conclusion is supported by our studies in the presence or absence of GPRP, an inhibitor of fibrin polymerization. Our results showed that FpB release was impaired in the presence of GPRP, whereas FpAЈ release was not affected. Together, these results indicate that FpAЈ release from AЈ␤ fibrinogen is independent of fibrin polymerization. Thus, we have created a fibrinogen in which the delay in the rate enhancement of fibrinopeptide release from the ␤ chain has been eliminated.
We present the following models of the interaction of thrombin with fibrinogen, which accommodate our findings with AЈ␤ and normal fibrinogens. Initially, thrombin is oriented such that all three fibrinogen interaction sites (the apolar specificity pocket, fibrinogen-binding exosite, and active site) are aligned to accommodate FpA as the preferred substrate. Upon binding to AЈ␤ fibrinogen, thrombin is best suited for FpA or FpAЈ  cleavage, regardless of the placement of the peptide on the ␣ or ␤ chain; thus, all four fibrinopeptides are cleaved simultaneously and with similar affinity. In normal fibrinogen, the thrombin active site is specific for FpA, cleaving it efficiently to initiate polymerization, whereas thrombin cleavage of FpB is much less efficient. During polymerization, however, efficient cleavage of FpB becomes favorable, i.e. FpB release becomes more efficient upon the appearance of des-A polymers (2)(3)(4)15). We propose two possible mechanisms, as depicted in Fig. 8, for this shift to efficient cleavage of FpB. Either 1) upon the release of FpA, thrombin changes its conformation such that FpB can be accommodated and cleaved efficiently, or 2) the local conformation of FpB is altered over the course of polymerization, which correctly orients the Arg 14 -Gly 15 bond for efficient cleavage within the active site of thrombin.
Although our results do not give us reason to favor one possibility over the other, previous studies lend credence to both. Studies on thrombin binding to exosite-binding fragments of hirudin, heparin cofactor II, and the thrombin receptor have shown that exosite interaction can allosterically modify the active site of thrombin (28 -32). Because thrombin binds to fibrinogen in the exosite, it is possible that conformational changes that occur in fibrinogen during polymerization can indirectly affect the thrombin active site. Alternatively, conformational changes in the fibrinogen molecule have been proposed (2,3,14,23,(33)(34)(35) and measured (36) during fibrin polymerization, and these changes could reposition the E domain such that the scissile bond in FpB is properly oriented for efficient cleavage by thrombin. Whether it is conformational changes in fibrinopeptide B, thrombin, or both, these studies suggest that either model is possible and reemphasize that we can only conclude from our studies that it is the specificity of thrombin for the fibrinopeptides that dictates the rate and timing of their release.
Production of AЈ␤ fibrinogen also allowed us to evaluate the effect of early exposure of the B site on fibrin polymerization. If early exposure of the B site did affect polymerization, we would expect a polymerization curve similar to that for fibrin monomer polymerization, when the A and B sites are exposed from the start of the reaction. Our results did not follow this pattern, thus suggesting that exposure of the B site alone does not directly influence polymerization. Rather, the participation of the B site in polymerization likely depends on certain polymerization events. Previous studies (33,37) suggest that this event is the polymerization of des-A monomers to form protofibrils of a critical length.
In summary, by making this fibrinogen with essentially four fibrinopeptides A, we have synthesized a model substrate that eliminated the delay in fibrinopeptide release from the ␤ chain. Thus, the normal delay in fibrinopeptide B release likely arises from a specific interaction between thrombin and FpB. We therefore conclude that the kinetics of fibrinopeptide release are dictated by the affinity of thrombin for its substrates. In addition, our work suggests that early exposure of the B site does not affect the polymerization process.
FIG. 8. Schematic representation of the possible mechanisms for the enhanced cleavage of FpB during fibrin polymerization. A, initially, the conformation of thrombin preferentially accommodates FpA (arrow) and initiates cleavage. FpB maintains a conformation that is different from FpA and does not fit into the thrombin active site. Upon cleavage of FpA, the conformation of thrombin changes (represented by a conversion from an arrow to a circle) during polymerization such that it then accommodates FpB (circle) and initiates efficient cleavage. B, initially, the conformation of thrombin preferentially accommodates FpA (arrow). FpB (skewed arrow) is unavailable for cleavage by thrombin because its conformation is not compatible with the thrombin active site. Upon cleavage of FpA, the conformation of FpB changes (arrow) and is capable of being cleaved by thrombin.