Characterization of the kinetic pathway for liberation of fibrinopeptides during assembly of fibrin.

The time dependence of the release of fibrinopeptides from fibrinogen was studied as a function of the concentration of fibrinogen, thrombin, and Gly-Pro-Arg-Pro, an inhibitor of fibrin polymerization. The release of fibrinopeptides during fibrin assembly was shown to be a highly ordered process. Rate constants for individual steps in the formation of fibrin were evaluated at pH 7.4, 37 degrees C, gamma/2 = 0.15. The initial event, thrombin-catalyzed proteolysis at Arg-A alpha 16 to release fibrinopeptide A (kcat/Km = 1.09 X 10(7) M-1s-1) was followed by association of the resulting fibrin I monomers. Association of fibrin I was found to be a reversible process with rate constants of 1 X 10(6) M-1s-1 and 0.064 s-1 for association and dissociation, respectively. Assuming random polymerization of fibrin I monomer, the equilibrium constant for fibrin I association (1.56 X 10(7) M-1) indicates that greater than 80% of the fibrin I protofibrils should contain more than 10 monomeric units at 37 degrees C, pH 7.4, when the fibrin I concentration is 1.0 mg/ml. Association of fibrin I monomers was shown to result in a 6.5-fold increase in the susceptibility of Arg-B beta 14 to thrombin-mediated proteolysis. The 6.5-fold increase in the observed specificity constant from 6.5 X 10(5) M-1s-1 to 4.2 X 10(6) M-1s-1 upon association of fibrin I monomers and the rate constant for fibrin association indicates that most of the fibrinopeptide B is released after association of fibrin I monomers. The interaction between a pair of polymerization sites in fibrin I dimer was found to be weaker than the interaction of fibrin I with Gly-Pro-Arg-Pro and weaker than the interaction of fibrin I with fibrinogen.

Characterization of the Kinetic Pathway for Liberation of Fibrinopeptides during Assembly of Fibrin* (Received for publication, November 2, 1984) Sidney D. Lewis, Paul P. Shields, and Jules A. ShaferS From the Department of Biological Chemistry, The University of Michigan, Ann Arbor, Michigan 48109 The time dependence of the release of fibrinopeptides from fibrinogen was studied as a function of the concentration of fibrinogen, thrombin, and Gly-Pro-Arg-Pro, an inhibitor of fibrin polymerization. The release of fibrinopeptides during fibrin assembly was shown to be a highly ordered process. Rate constants for individual steps in the formation of fibrin were evaluated at pH 7.4, 37 "C, l'/2 = 0. 15. The initial event, thrombin-catalyzed proteolysis at Arg-Aal6 to release fibrinopeptide A (kcat/Km = 1.09 X lo7 M"s") was followed by association of the resulting fibrin I monomers. Association of fibrin I was found to be a reversible process with rate constants of 1 X lo6 M"S" and 0.064 s" for association and dissociation, respectively. Assuming random polymerization of fibrin I monomer, the equilibrium constant for fibrin I association (1.56 X lo7 M-') indicates that >80% of the fibrin I protofibrils should contain more than 10 monomeric units at 37 "C, pH 7.4, when the fibrin I concentration is 1.0 mg/ml. Association of fibrin I monomers was shown to result in a 6.5-fold increase in the susceptibility of Arg-BB14 to thrombin-mediated proteolysis. The 6.5-fold increase in the observed specificity constant from 6.5 X IO" ~" s " to 4.2 X IO6 M"S" upon association of fibrin I monomers and the rate constant for fibrin association indicates that most of the fibrinopeptide B is released after association of fibrin I monomers. The interaction between a pair of polymerization sites in fibrin I dimer was found to be weaker than the interaction of fibrin I with Gly-Pro-Arg-Pro and weaker than the interaction of fibrin I with fibrinogen.
The plasma protein fibrinogen is converted to the insoluble fibrin matrix of blood clots by a multistep process (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14), wherein the fibrinogen molecule which is comprised of 2 Aa-, 2 BP-, and 2 y-polypeptide chains (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25) undergoes limited proteolysis and aggregation. Much evidence has accumulated to support the view that fibrin assembly is a highly ordered process which is initiated by thrombin-catalyzed hydrolysis at Arg-Aal6 to form fibrin I monomer and release a 16-aminoacyl residue peptide (FPA') from the N-terminus of each Aa-chain. This event is thought to be followed by end to end aggregation of the resulting fibrin I monomers to form fibrin I oligomers (protofibrils). Association of fibrin I appears to be followed by thrombin-mediated hydrolysis at Arg-BBl4 to form fibrin I1 and release a 14-aminoacyl residue peptide (FPB) from the amino terminus of each of the BP-chains. The resulting fibrin I1 protofibrils have a greater propensity than fibrin I protofibrils to undergo lateral interactions which result in fiber formation (9, 10). A reaction sequence for release of fibrinopeptides and fiber formation is shown in Scheme 1, wherein AaBPy and aBPy represent half-molecule units and (aB@y)* represents a pair of interacting half-molecules in dimers and higher oligomers of fibrin I.
The observation that FPA is released more rapidly from fibrinogen than is FPB and the observation that inhibitors of fibrin I aggregation inhibit release of FPB without inhibiting release of FPA suggests that the reactions denoted by the rate constants k3 and k4 are minor pathways for the release of FPB. Observations by others (26, 27) suggest, however, that premature release of FPB via the k4 and ky reactions occurs to some extent. In this regard, kinetic analysis at low concentrations of fibrinogen of the time dependence of the thrombincatalyzed release of FPA and FPB indicates that kl > 30k4 SO that at least 97% of the time FPA is released prior to FPB (14). Thus, release of FPB via the k4 step can often be neglected in kinetic studies of the release of fibrinopeptides. The value of k3 and the relative contribution of this step to the time-dependent release of FPB has never been determined, however.
More importantly, Scheme 1 has never been put to a rigorous quantitative test. For example, it has never been determined whether a single set of rate constants could account for the time dependence of the release of FPA and FPB at high thrombin concentrations where the rate of release of FPB should be limited by the rate of association of fibrin I as well as at low thrombin concentrations where the rate of proteolysis should limit the rate of release of FPB. In this work we report for the first time a set of rate constants ( kl, k3, k,, k,, and k2) evaluated at the physiological conditions of pH 7.4,37 "C, and r/2 = 0.15, which shows that Scheme 1 can quantitatively account for the time-dependent release of fibrinopeptides over a wide range of thrombin and fibrinogen concentrations.

EXPERIMENTAL PROCEDURES
Materiak-Fibrinogen was purified from outdated plasma by the method of Jakobsen and Kierulf (28), except that the fibrinogen solvent used to redissolve fibrinogen was the citrate saline solvent of Straughn and Wagner (29), and final dialysis was against 0.30 M NaC1. This fibrinogen preparation behaved identically with the fibrinogen preparation used previously in this laboratory (30). All fibrinogen solutions were stored at -70 "C prior to use. An E% of 15.1 and a molecular weight of 340,000 were used to calculate fibrinogen concentrations. Fibrin I was prepared by allowing 6 ml of a solution of fibrinogen (2.5 mg/ml) in 0.15 M NaC1, 5 mM phosphate buffer, pH 6.8, to react overnight with 0.01 ml of a solution of Reptilase-R, which was obtained by addition of 1 ml of water to a vial containing 34 mg of lyophilized Reptilase-R from Ahbott Laboratories. The resulting clot was collected on a glass rod, rinsed with water, and dissolved in 1 ml of 0.02 M acetic acid. An E% of 14.0 and an M, of 340,000 were used to calculate the concentration of fibrin I monomers in 0.02 M acetic acid. Pure human a-thrombin with a specific activity of 24 thrombin units/pg (using the method of Lewis and Shafer (31)) was generously supplied by John W. Fenton 11, New York State Department of Health. The thrombin which was greater than 93% active by active site titration (32) was stored in 0.3 M NaCl at -70 "C. a-Thrombin accounted for >99% of the thrombin. Values of kcat and $,,/K, reported in this work are based on the concentration of thrombin determined from its absorbance at 280 nm using an E$$, of  HPLC-Elution conditions were those described previously (14), except for analysis of reaction mixtures containing <0.1 p M fibrinogen. In these runs, 1.2-ml samples were injected and eluted from a Spherisorb ODS-2 3-p (0.46 X 10 cm) column isocratically using 16% CHBCN in 0.083 M phosphate buffer. The isocratic system produced flatter baselines and the 3-p column sharpened the FPB peaks considerably at the usual detection limit of 0.02 absorbance unit full scale at 205 nm. After FPB was eluted, a 30-min linear gradient to 40% acetonitrile was used to rinse the column. The column was reequilibrated at 16% CHsCN before injection of another sample.
Reaction Kinetics-The rates of release of FPA and FPB were measured at pH 7.4, 37 'C, in 137 mM NaCl, 2.5 mM KCl, and 9.47 mM sodium phosphate containing 0.001% or 0.1% PEG. At low thrombin concentrations (0.07-1.4 nM), the kinetics were carried out essentially as already described (9), except that at the lowest fibrininto the HPLC system. Additionally, at 1.4 nM thrombin, individual ogen concentration a 1.2-ml aliquot of reaction mixture was injected 1.5-ml microfuge tubes containing the reaction mixture (0.8-1.2 ml) were pre-equilibrated at 37 "C before an aliquot of thrombin was rapidly added with mixing to each tube. At appropriate times, 0.1 ml of 3 M HClO, was added wich vigorous mixing to one of the tubes and the volume was adjusted to a known final value (1.3-1.5 ml) with H20. The resulting solution was centrifuged and an appropriate volume (1.0-1.2 ml) was injected in the HPLC system. For the runs carried out in the presence of calcium chloride, 18.9 mM sodium phosphate, 5 mM KC1 was added with stirring to an equal volume of 274 mM NaCl, 0.2% PEG containing 2 mM CaCl. This procedure avoided precipitation of calcium phosphate. Rates of fibrinopeptide release at 40 nM thrombin were measured using a flow system consisting of a multichannel Reeve-Angel peristaltic pump which was placed in a 37 "C room. Two channels of the pump contained the fibrinogen and thrombin feed lines which led to a mixing chamber. The mixing chamber was similar to one which has been described previously (34). The peristaltic pump was fitted with tygon tubing to which was connected %e-inch outside diameter Teflon tubing equipped with appropriate fittings for connection to the mixing chamber. The solutions of fibrinogen and thrombin were pumped at equal flow rates into the mixing chamber. The concentrations of fibrinogen and thrombin were twice the value desired in the reaction. Upon exiting the mixing chamber, the reaction mixture was allowed to flow through a single piece of 0.8-mm inside diameter Teflon tubing of known volume to another mixing chamber. A third channel of the peristaltic pump contained a line through which 1 M HClO, was pumped at the same rate as the other solutions to the second mixing chamber to quench the thrombin-catalyzed reaction. The time of reaction was determined from the flow rate (0.10-0.15 ml/s) through the reaction tube and the volume of the 0.8-mm inside diameter tubing connecting the two flow chambers. The volume of the connecting tube was determined from the weight of distilled water it held. By varying the tube length from 5 cm to 4 m, reaction times from 0.17 to 20 s could be obtained at the flow rates used. Prior to each run, the precise flow rate for each line was determined (and adjusted) by measuring the volume of feed liquid utilized over a known time interval. The correct functioning of the flow system was indicated by observation of pseudo-first kinetics for the release of FPA wherein the quotient of the pseudo-first order rate constant and the thrombin concentration yielded a specificity constant for FPA release of 10.9 ~" s " which was within experimental error (7%) of the rate constant observed for this constant at lower thrombin concentrations in a static system. It is important to note that whereas a flow system might be designed to give lower reaction times by increasing the flow rate and decreasing the diameter of the reaction tube, a practical limit to this approach is reached wherein high shear forces cause fragmentation of fibrinogen. Fibrinogen fragmentation results in the appearance of several peptides during HPLC which are independent of the presence of thrombin.

RESULTS
In this study the concentration of fibrinogen was kept below 0.3 p~ to simplify the kinetic analysis. As shown previously (14) under these conditions, the fractional saturation of thrombin is low (i.e. [SI 5 10 K,) so that the Michaelis-Menten equation (1) reduces to for thrombin-catalyzed hydrolytic release of FPB from (~B P Y )~ ensembles, where the subscript "a" denotes self-associated fibrin I as the substrate. The time dependence for the release of FPB at 0.14 p~ fibrinogen and 0.28 nM thrombin is shown in Fig. 1A. The solid line was calculated from Equation 4 using the previously determined value (14) of 4.2 X IO6 M"s" for kcatBa/KmB. and a value of 1.09 X lo7 M"s" determined for kcatA/KmA from the fit of FPA release to Equation 3. This value is in reasonable agreement with the previously reported value of 1.16 X lo7 M"S" (141.
To verify that keatse/KmBe reflects the specificity constant for thrombin-catalyzed release of FPB from fibrin I polymer rather than from some other species, fibrin I polymer was The methods used to evaluate k3, k,, and kp and to calculate the dashed line are described in the text.
isolated and its interaction with thrombin was characterized.
To obtain fibrin I, fibrinogen was treated with Reptilase-R, an enzyme which catalyzes the release of FPA, but not FPB, from fibrinogen (5). The resulting clot was dissolved in dilute acetic acid and a sample of this solution was transferred to the reaction mixture which was adjusted to pH 7.4,37 "C, and 0.14 p~ fibrin I. After a 20-min incubation to ensure formation of fibrin I polymer, thrombin was added. The resultant release of FPB followed a pseudo-first order rate law wherein the ratio of the pseudo-first order rate constant and the thrombin concentration yielded a value for the specificity constant for the release of FPB from fibrin I polymer which was within 5% of the value of kcatBa/KmBa determined by fitting the time dependence of the release of FPB to Equation 4 when thrombin was reacted with fibrinogen. Thus, the kinetic properties of isolated fibrin I polymer are consistent with a reaction pathway for the conversion of fibrinogen to fibrin wherein essentially all of the FPB is released from fibrin I polymer.
Although Equation 4 with k a t A / K m A = 1.09 X lo7 M"s" and kcatBa/K,& = 4.2 X lo6 ~" s " fits the time dependence of release of FPB over a wide range of thrombin concentrations reasonably well, substantial deviations from Equation 4 become apparent at very high thrombin concentrations. These deviations are exemplified in Fig. 1B where the solid line is the fit of the data to Equation 4 with the values of kCatA/KmA and kcatB/KmB set at 1.09 X lo7 and 4.2 X IO6 "'s", respectively. The deviations from Equation 4 are consistent with Scheme 1. According to Scheme 1, a change in the ratedetermining step from proteolysis to polymerization should occur as the thrombin concentration is increased. When polymerization becomes rate-controlling, deviations from Equation 4 occur, because the rate of release of FPB will be limited by the rate of association of fibrin I rather than proteolysis, provided of course k3 < k z .
The rate constant ( k 3 ) for the thrombin-catalyzed release of FPB from unassociated aB& ensembles was determined from measurements of the time dependence of [FPB] in the presence of saturating concentrations of the tetrapeptide Gly-Pro-Arg-Pro. This peptide competes with the Gly-Pro-Arg-Val sequences at the N termini of the a-chains for the polymerization sites in the D-domain in fibrin I (12). As shown in where k3 is the product of the thrombin concentration and the specificity constant ( kcstBu/KmBu) for the release of FPB from an unassociated monomer. A nonlinear least squares fit of the data in Fig. 2 to Equation 5 yielded a value of 0.65 X lo6 "'s" for kcatBu/KmBu. Comparison of the values of the specificity constants kcatBa/KmBa and kcatBu/KmBu indicates that self-association of aB@y ensembles results in a 6.5-fold enhancement in the susceptibility of Arg-Bo14 to attack by thrombin.
To evaluate kp and k , in Scheme 1, the time dependence of the release of FPB was observed over a wide range of thrombin concentrations in the presence and absence of the polymerization inhibitor Gly-Pro-Arg-Pro. The tetrapeptide (T) by binding to aBPy but not to ( (~B p y )~ should shift the association equilibrium as shown in Equation 6 If one assumes that the binding of T to aBPy is rapid so that the equilibrium denoted by KO is maintained throughout the where To fit the experimental data, the values of k,, kl, and k3 were fixed at their predetermined values and k; and k , were systematically varied until the best fit to the experimental data was obtained. It can be shown that unless polymerization is at least partially rate-determining, the time dependence of release of [FPB] will not yield a unique value for k; or k, . Thus, high concentrations of thrombin were used to increase the rate of the kl and k2 steps so that polymerization would become partially rate-determining. Initial experiments done in the presence of high concentrations of thrombin, but in the absence of polymerization inhibitor, yielded data which allowed us to estimate kp with reasonable precision, but these data did not permit accurate evaluation of k-p. The difficulty in fixing k , arose because at the initial concentration of fibrinogen (20.07 PM) required to obtain accurate determinations of FPB, the extent of dissociation of (~B P T )~ was too small to permit accurate evaluation of k,. To evaluate k , accurately, the extent of association of (~B P Y )~ was decreased by using Gly-Pro-Arg-Pro (7') to decrease selectively the value of k; (see Equation 17). When the thrombin concentration is sufficiently low so that k , is substantially larger than k2, the aBPr half-molecule has sufficient time to fully equilibrate with (LYBPY)~. Under these conditions, one can apply the analytical methods used by Beaty  Time dependencies for the release of FPB at 0.071 nM thrombin ( k , = 7.7 X s-', kz = 3.0 X s-') and 1.4 nM thrombin ( k l = 1.5 X lo-' s-l, k2 = 5.9 X s-') are illustrated in Fig. 3. In both cases, the best fit of the data was obtained with k;/k, equal to 6.25 X lo5 M-'. When the thrombin concentration was increased to 40 nM, however, a good fit to the experimental data could no longer be obtained by integration of Equations 10, 20,21, and 13-16. This observation was taken to indicate that at 40 nM thrombin (where k2 = 0.168 s-'), k , was no longer much larger than kz and that the steady state value of [ ( C Y B P~)~] / [~B P~]~~ fell below its equilibrium value so that Equations 20 and 21 could no longer be used in place of Equations 11 and 12. To evaluate k; and k,, differential equations 10-16 were numerically integrated keeping k,, kz, k3, and k;/k, at their previously determined values while systematically varying k; until the best fit to the experimental data at 40 nM thrombin was obtained. As shown in Fig. 4, the best fit of the data yielded a value of 4 X lo4 ~" s " for k; and a value of 0.064 s" for k,. Since the value of k , should be independent of the presence of tetrapeptide, it was possible to use the values of kl, kz, k3, and k , to determine k; as a function of the tetrapeptide concentration. Additionally, these kinetic constants could be used to determine kp from the time dependence of the release of FPB in the absence of tetrapeptide. The dashed line in Fig. 1 B and Fig. 6, A and B. The data in Fig. 6C show that at 100 p~ tetrapeptide and

DISCUSSION
Scheme 1 depicts the simplest reaction pathway for the conversion of fibrinogen to fibrin I1 fibers which quantitatively accounts for our observations of thrombin-catalyzed release of fibrinopeptides from normal human fibrinogen at pH 7.4,37 "C, r/2 = 0. 15. It is important to note that previous studies (14) have demonstrated that release of FPB via the k4 step accounts for less than 3% of the FPB and thus release of FPB prior to FPA was neglected in our analysis. Interpretation of the dependence of the release of FPB on the concentration of thrombin, fibrinogen, and Gly-Pro-Arg-Pro in terms of the reaction pathway in Scheme 1 yielded ( a ) a rate constant for fibrin I association which is close to that obtained (10) by direct measurement of self-association of fibrin I to protofibrils using light scattering; ( b ) a specificity constant for thrombin-catalyzed release of FPB from fibrin I polymer which corresponds to that obtained from independent measurements of thrombin-catalyzed release of FPB from separately prepared fibrin I polymer. These independent verifications of rate constants for putative intermediates indicate that conversion of fibrinogen to fibrin is well represented by the reaction pathway depicted in Scheme 1. Additionally, interpretation of the thrombin-catalyzed release of FPB in terms of this reaction pathway yielded an equilibrium constant for the binding of Gly-Pro-Arg-Pro to fibrin I monomer which is similar to that previously reported (13) for the binding of this tetrapeptide to fibrinogen. This observation is consistent with previous proposals ( e g . Ref. 13) of the existence of an exposed polymerization site in the D-domain of fibrinogen and fibrin I monomer.
Although the time dependencies of the release of FPA and FPB fit a model wherein the reactivity of each half-molecule ensemble is independent of the state of the adjacent halfmolecule ensemble on the same molecule, the existence of intramolecular effects has not been ruled out. In this regard, studies of the polymerization of fibrin I subsequent to hydrolysis at Arg-Aa16 suggest that hydrolysis at the second Arg-Aal6 residue in a fibrinogen molecule may be more than an order of magnitude more rapid than the first (37-39). The existence of such an effect would indicate that the value of katA/K,,,A reported in this work is the specificity constant for the first rate-limiting cleavage at Arg-Aal6. Should other intramolecular interactions come to light, it should be possible to utilize the values of rate constants reported here to assign rate constants in amended versions of the reaction pathway.
Since fibrin I can associate with fibrinogen (e.g. Refs. 35,40,and 41) as well as undergo self-association, it is appropriate to consider whether the kinetics of FPB release might be complicated by formation of complexes between fibrinogen and fibrin I. Substitution of the specificity constants for the sequential release of FPA (1.09 X lo7 " ' s " ) and FPB (4.2 X lo6 " ' s " ) in Equations 3 and 4 indicate that at a time when 50% of the FPA has been released, the release of FPB is less than 7%. Consequently, when FPB release is greater than 7%, the concentration of aBPy units is greater than that of the intact AaBBy units. Since the affinity of an aBPy unit for another aBPy is -10-fold greater than that for an AaBpy Kinetic Pathway for Liberation of Fibrinopeptides unit (35), the concentration of fibrin0gen:fibrin I complexes is probably too low to have a detectable effect on the kinetics of FPB release after a few per cent of the FPB have been released.' These considerations together with the low value observed for k, relative to k, and k2 and the low value of k4 relative to k2 suggest that essentially all of the FPB is released from fibrin I polymer. It is important to note, however, that the relative magnitudes of rate and equilibrium constants in Scheme 1 may be functions of the reaction conditions and the structure of fibrinogen. Thus, the degree of sequentiality of the process may be altered by changes in pH, temperature, ionic strength, or by variations in fibrinogen structure such as those associated with many dysfibrinogenemias. It is interesting to consider further the dimerization of fibrin I as represented in Equation 23 kP 2aBPy since the equilibrium constant for binding of fibrinogen to fibrin I is -106-107 M-' (35,40). Further work is necessary to determine the structural basis for the higher affinity of fibrinogen for fibrin I and the reason why the D-E interactions in If aB@r binds 10 times more tightly to aB@y than to fibrinogen, equal concentrations of (aB0y)z and aB@r:AruB@r would be expected at a time when -10% of the FPA is released. The kinetic parameters for FPA and FPB release together with Equations 4 and 5 indicate, however, that the release of FPB is ~0 . 2 % when the release of FPA is 10%.
Standard unitary free energy changes do not contain the free energy change for bringing molecules together in a 1 M solution. For a bimolecular reaction, this corresponds to 7.98 e.u. units or 2.48 kcal/mol at 37 "C. As  fibrin I dimer are weaker than those found when fibrin I complexes with Gly-Pro-Arg-Pro and fibrinogen.
It is important to note that the value of 1.56 X lo7 M-' determined for the apparent equilibrium constant (k,/k,) indicates that aggregation of fibrin I monomers to protofibrils should be reversible and dependent on the concentration of fibrinogen. The size distribution of linear fibrin protofibrils should be given by the equation of Flory (43) for polymerization of a bifunctional monomer. At a 10-fold higher concentration of fibrinogen (3.0 PM -1 mg/ml), Equations 26-28 indicate that over 80% of the fibrin I monomers should be present in protofibrils more than 10 monomeric units in length (at 37 "C, p H 7.4, r/S = 0.15).
Since long protofibrils would be expected to have a greater propensity to participate in the side to side interactions involved in fiber growth, fiber formation should serve to drive the polymerization of fibrin I monomer toward completion. In this regard, it is interesting to note that Shainoff and Dardik (35) in their studies of the effect of fibrinogen on the elution position of fibrin I during gel filtration have estimated an equilibrium constant of -lo8 M" for incorporation of a fibrin I monomer into a fibrin I fiber. Side to side interactions which are present in fibrin I fibers, but absent in fibrin I protofibrils, may account for this equilibrium constant for self-association of fibrin I being larger than the equilibrium constant of 1.56 X lo7 M" reported in this work for fibrin I dimerization.
It is also important to realize that interactions between adjacent D-domains which exist in the trimer and higher oligomers, but not in the dimer (lo), might cause the value of the equilibrium constant for dimerization of fibrin I monomer to be less than that for addition of a fibrin I monomer to a growing protofibril. If such were the case, the measured value of k,/k, would be an average value, which could cause us to underestimate somewhat the size of protofibrils. Further work is necessary, however, to determine whether interactions between adjacent D-domains contribute significantly to the stability of protofibrils.