A kinetic analysis of the tissue plasminogen activator and DSPAalpha1 cofactor activities of untreated and TAFIa-treated soluble fibrin degradation products of varying size.

The kinetics of tissue plasminogen activator (t-PA) and DSPAalpha1-catalyzed plasminogen activation using untreated and TAFIa-treated fibrin degradation products (FDPs), ranging in weight average molecular weight (M(w)) from 0.48 x 10(6) to 4.94 x 10(6) g/mol, were modeled according to the steady-state template model. The FDPs served as effective cofactors for both activators. The intrinsic catalytic efficiencies of both t-PA (17.4 x 10(5) m(-1) s(-1)) and DSPAalpha1 (6.0 x 10(5) m(-1) s(-1)) were independent of FDP M(w). The intrinsic catalytic efficiency of t-PA was 12-fold higher than that measured under identical conditions with intact fibrin as the cofactor. At sub-saturating levels of cofactor and substrate, rates were strongly dependent on FDP M(w) with DSPAalpha1 but not t-PA. Loss of activity with decreasing FDP M(w) correlated with loss of finger-dependent binding of the activators to the FDPs. TAFIa treatment of the FDPs resulted in 90- and 215-fold decreases in the catalytic efficiencies of t-PA (0.20 x 10(5) m(-)(1) s(-1)) and DSPAalpha1 (0.028 x 10(5) m(-1) s(-1)), yielding cofactors that were still 30- and 50-fold better than fibrinogen with t-PA and DSPAalpha1, respectively. Our results show that for both activators the products released during fibrinolysis are very effective cofactors for plasminogen activation, and both t-PA and DSPAalpha1 cofactor activity are strongly down-regulated by TAFIa.

Hemeostasis requires a proper balance between the coagulation and fibrinolytic systems. In response to vascular injury, a hemostatic plug is generated by converting fibrinogen to an insoluble fibrin clot through the action of thrombin, the terminal enzyme of the coagulation cascade. Fibrinolysis, the breakdown of the fibrin clot, is achieved primarily by the activation of plasminogen to the serine protease plasmin, which catalyzes degradation of the insoluble fibrin clot to soluble fibrin degradation products (FDPs). 1 The activation of plasminogen can be catalyzed by both endogenous activators such as tissue-type plasminogen activator (t-PA) and urokinase or exogenous acti-vators such as streptokinase, staphylokinase, and Desmodus rotundus salivary plasminogen activators (DSPAs). These enzymes have all been used as thrombolytic agents for the dissolution of pathological thrombi, which can cause both myocardial infarction and stroke.
The fibrin clot is not only the substrate for plasmin but also a cofactor for plasmin generation by the various plasminogen activators. Both t-PA and DSPA␣1 are known as fibrin-selective plasminogen activators, because the rate of plasminogen activation with both activators is increased several orders of magnitude in the presence of fibrin, as compared with fibrinogen (1). Extensive plasminogen activation in the plasma, mediated via the cofactor effect of fibrinogen, results in systemic, plasmin-mediated fibrinogenolysis and consumption of ␣ 2 -antiplasmin, severely compromising the coagulation potential of the plasma (1,2). Fibrin selectivity is thus highly desirable for systemically administered thrombolytic agents. DSPA␣1 is considerably more fibrin-selective than t-PA, as the catalytic efficiency of DSPA␣1 is stimulated 13,000-fold, compared with only 820-fold for t-PA, when fibrin is the cofactor instead of fibrinogen (1). Furthermore, DSPA␣1 is intrinsically less fibrinogenolytic than t-PA because the catalytic efficiency of DSPA␣1 is 13-fold lower than t-PA when fibrinogen is the cofactor (50 versus 640 M Ϫ1 s Ϫ1 ) (1).
The stimulation of plasminogen activation by fibrin is mediated by interactions of both the activator and plasminogen with fibrin (3). Structures within t-PA by which it interacts with fibrin are its fibronectin finger-like domain and its kringle-2 domain. The interaction of DSPA␣1 with fibrin is presumably mediated solely by its finger domain, since it lacks a kringle-2 domain, although at least one other low affinity interaction is likely, because DSPA␤ and DSPA␥, highly homologous relatives of DSPA␣1 (89 and 91% identity, respectively) lacking the finger domain, are also stimulated by fibrin, albeit to a much lesser extent (1). The interaction of plasminogen with fibrin occurs by its lysine-binding kringle domains. Two forms of plasminogen exist. Glu-plasminogen, the full-length form found circulating in plasma, interacts only weakly with intact fibrin but strongly with partially degraded fibrin possessing carboxyl-terminal lysine and/or arginine residues (4). Lys-plasminogen, a truncated version of plasminogen produced by the plasmin-catalyzed removal of a 77-residue peptide from the amino terminus, binds to both native and partially degraded fibrin tightly (4). Lys-plasminogen is a considerably better substrate for t-PA, and its formation during t-PA-mediated fibrinolysis confers positive feedback on the process (3,4). Since the cleavage of fibrin by plasmin exposes carboxyl-terminal lysine and arginine residues, producing a fibrin surface containing high affinity plasminogen-binding sites, partial degradation of fibrin by plasmin results in the recruitment of plasminogen to the partially degraded fibrin surface (5-7). The partially de-graded fibrin is a superior cofactor than intact fibrin for plasminogen activation (8). This effect can be eliminated by the basic plasma carboxypeptidase, TAFIa, which removes the carboxyl-terminal lysine and arginine residues from the fibrin surface (8). Plasmin production, therefore, results in a positive feedback loop that can be down-regulated through the generation of TAFIa from its precursor TAFI.
Plasmin-catalyzed digestion of fibrin produces soluble FDPs which, owing to their structural similarity to fibrin and/or fibrinogen, likely act as cofactors for plasminogen activation. We have recently demonstrated that FDPs released from a perfused clot are composed of noncovalently associated products whose masses range from 250 kDa (the mass of DD/E) to ϳ10,000 kDa (9). Furthermore, our work showed that the majority of the FDPs compose structures much larger than DD/E. Since the relationship between the sizes of FDPs and their cofactor effects has not been studied, we isolated FDPs with different masses to study the relationship between FDP mass and cofactor activity in reactions with t-PA and DSPA␣1. Furthermore, since the FDPs contain carboxyl-terminal lysine and arginine residues as a result of plasmin degradation, we also investigated the effect of the TAFIa-catalyzed removal of the carboxyl-terminal lysine and arginine residues on the cofactor activity of FDPs. The work described in this paper represents the first extensive study of the cofactor effect of FDPs on plasminogen activation catalyzed by the fibrin-specific plasminogen activators, t-PA and DSPA␣1.
Preparation of FXIIIa Cross-linked Soluble Fibrin Degradation Products-Soluble FDPs were prepared as described previously (9), except that 1) clots were formed in columns with 9 ml of available volume, 2) the clots were perfused with 0.2 nM plasmin instead of 0.05 nM plasmin, and 3) the FDPs from four perfusions were pooled prior to gel filtration. The pooled FDPs (ϳ25 mg) were subjected to gel filtration on Sephacryl S-1000 and the weight average molecular weight (M w ) of the FDPs in the eluate was determined on-line using multiangle laser light scattering (9). Samples from the gel filtration column having M w ϭ 0.48 ϫ 10 6 , 1.08 ϫ 10 6 , 1.93 ϫ 10 6 , 3.08 ϫ 10 6 , 3.97 ϫ 10 6 , and 4.94 ϫ 10 6 g/mol were prepared by pooling appropriate fractions, and the samples were concentrated to Ͼ6.5 M by centrifugal concentration as described (9). The concentration of the FDPs refers to the concentration of fragment X equivalents present in the sample. Samples were in stored 0.02 M Hepes, 0.5 M NaCl, 0.001% Tween 80, pH 7.4 at 4°C.
Activation of 5AF-Pgn by t-PA and DSPA␣1 in the Presence of Soluble FDP Cofactors of Varying Sizes-The activation of 5AF-Pgn (to 5AF-Pn) by t-PA and DSPA␣1 in the presence of soluble FDP cofactors, both native and TAFIa-treated (see below), was monitored by fluorescence spectroscopy as described previously (3). Aliquots (90 l) of 5AF-Pgn, in 0.02 M Hepes, 0.053 M NaCl, 0.0256% Tween 80, 2.56 mM CaCl 2 , pH 7.4, were added to the wells of a fluorescence microtiter plate (Microfluor, Dynatech, Chantilly, VA), and the fluorescence intensities of the samples were determined over 10 -30 min using a PerkinElmer Life Sciences LS-50B fluorescence spectrophotometer equipped with the plate reader accessory. The samples were excited at 495 nm (5 nm slit), and the fluorescence was measured at 535 nm (3 nm slit) using a 515 nm cut-off filter. FDPs (25 l) were then added to the wells, and the fluorescence of the 5AF-Pgn/FDP samples was determined over 10 -30 min. The combination of the 5AF-Pgn and FDPs gave a solution containing 0.02 M Hepes, 0.15 M NaCl, 2 mM CaCl 2 , 0.02% Tween 80, pH 7.4. Plasminogen activation reactions containing FDPs (33.3-500 nM final) and 5AF-Pgn (33.3-500 nM final) were then initiated by the addition of 15 l of either t-PA (1-4 nM final) or DSPA␣1 (4 -10 nM final), and the fluorescence of the reactions was monitored every 80 s for 80 min. Both t-PA and DSPA␣1 were in 0.02 M Hepes, 0.15 M NaCl, 2 mM CaCl 2 , 0.02% Tween 80, pH 7.4. The fluorescence of 5AF-Pgn in reactions without FDPs was determined identically to the reactions described above, except that 0.02 M Hepes, 0.5 M NaCl, 0.001% Tween 80, pH 7.4, was used in place of the FDPs. All experiments were performed at ambient temperature (ϳ20°C).
Binding of 5AF-Pgn to Soluble FDP Cofactors of Varying Sizes-The binding of 5AF-Pgn to the FDPs was measured based on the decrease in fluorescence of the 5AF-Pgn in the presence of FDPs. Binding was determined by measuring the fluorescence of the reactions containing 5AF-Pgn and FDPs prior to the addition of the activator (I) and subtracting it from the measured fluorescence of the 5AF-Pgn in each reaction prior to the addition of FDP, corrected for dilution (I o ). The dilution factors accounting for the fluorescence change upon addition of FDPs and activator were determined from the experiments described above in which no FDPs were used. The difference in fluorescence ⌬I ϭ I o Ϫ I was measured for all concentrations of Pgn and FDP, and the binding was analyzed by nonlinear regression of the data according to Equation 1, where [Pgn-FDP] is the concentration of bound 5AF-Pgn, and ⌬I bound is the difference in fluorescence coefficients (fluorescence units/5AF-Pgn) between free and bound 5AF-Pgn. These coefficients are referred to as IC free and IC bound , respectively. The concentration of bound 5AF-Pgn is found from the quadratic binding Equation 2 Treatment of FDPs with TAFIa-TAFI was activated to TAFIa by thrombin in the presence of Solulin essentially as described previously (10). Briefly, 1.0 M TAFI was reacted with 20 nM thrombin, 80 nM Solulin in 0.02 M Hepes, 0.15 M NaCl, 5 mM CaCl 2 , 0.001% Tween 80, pH 7.4, for 10 min at ϳ22°C. The TAFIa was then stored on wet ice until used. A TAFIa titration was performed to determine the amount of TAFIa required to achieve maximal "deactivation" of the FDPs. The titration was based on the activation of 5AF-Pgn in the presence of FDPs treated with or without varying amounts of TAFIa. The 1.93 ϫ 10 6 g/mol FDP was used as the cofactor and 5 nM DSPA␣1 as the plasminogen activator. FDPs (2.6 M) were treated with varying concentrations of TAFIa (0.1-30 nM) for 60 min at room temperature. The TAFIa/FDP solutions were diluted in half with 0.02 M Hepes, 0.5 M NaCl, 0.001% Tween 80, pH 7.4, and a 25-l aliquot was added to 90 l of 180 nM 5AF-Pgn in the wells of a fluorescence microtiter plate. DSPA␣1 (15 l of 43.3 nM) was added to initiate cofactor-dependent activation, and the fluorescence of the reaction was monitored over time (see above). Based on the results of the TAFIa titration, the FDP samples were treated with 10 nM TAFIa. The FDPs (2.6 M in 0.02 M Hepes, 0.5 M NaCl, 0.001% Tween 80, pH 7.4) were treated with 10 nM TAFIa for 60 min at room temperature followed by a 2-h incubation at 37°C which served to both maximally inactivate the FDPs and inactivate the TAFIa. The TAFIa-treated FDPs were stored at 4°C.
Data Analysis-The rates of 5AF-Pgn activation were determined from the initial slopes of the activation reactions, using the fact that conversion of 5AF-Pgn to 5AF-Pn results in a 50% decrease in the fluorescence of the active site fluorescein label (11). The fluorescence of the 5AF-Pgn, prior to the addition of activator, was found to decrease as a result of 5AF-Pgn binding to the native but not TAFIa-treated FDPs. The rate data were corrected for this effect using the fact that the final fluorescence, i.e. the fluorescence of a reaction taken to completion, was dependent only on the initial 5AF-Pgn concentration and not on the initial fluorescence of the 5AF-Pgn/FDP solutions (data not shown (11)). Since the end point fluorescence of the reactions was independent of the FDP concentration, the difference between the fluorescence of the 5AF-Pgn/FDP solutions and the end point fluorescence for the 5AF-Pgn concentration defined the full-scale fluorescence change upon complete conversion of 5AF-Pgn to 5AF-Pn.
The data from the reactions were modeled according to the steadystate template model for plasminogen activation as described by Horrevoets et al. (3), with the terms for fibrin substituted by FDPs. The rate Equation 4 is given by where rate is the velocity of the reaction per nominal activator concentration; k cat is the turnover number for the reaction; [Pgn] free is the concentration of free plasminogen (calculated from the above binding Equation 4); [FDP] o is the total FDP concentration; K m is the Michaelis constant for the reaction; K A is the dissociation constant of the activator for the FDPs, and K is a constant whose value is equal to the concentration of FDPs required to give a rate equal to half k cat at saturating Pgn. The data from the experiments were initially analyzed using direct plots. The reaction rates at each [FDP] by using the relationships (Equations 7 and 8) to obtain values for the true k cat /K m ratio and K A . Finally, the data from all reactions for each activator were analyzed globally, according to Equation 4.
The k cat (app)/K m (app) was then plotted against [FDP] o and analyzed by nonlinear regression according to Equation 10, a modified form of Equation 6.
The Effect of TAFIa on t-PA-and DSPA␣1-mediated Fibrinolysis in Plasma-Clots (200 l), made from 66.6 l of TAFIa-deficient human plasma, 56 l of 0.02 M Hepes, 0.15 M NaCl, 0.02% Tween 80, pH 7.4, 4 l of 500 nM Solulin, 20 l of 6 nM t-PA or DSPA␣1, 20 l of 0 -200 nM TAFI, and 33.3 l of 36 nM thrombin, 60 mM CaCl 2 were formed in the wells of a microtiter plate. The clotting and subsequent lysis of the clots were monitored by turbidity at 405 nm at 37°C. The lysis time, the time at which the turbidity has decreased to one-half the maximal plateau value, was determined for each sample, and the results are presented as relative lysis times, which are the lysis times for each reaction divided by the lysis time for the reaction in the absence of TAFI.

RESULTS
Isolation of FDPs with M w Ranging from 0.48 ϫ 10 6 to 4.94 ϫ 10 6 g/mol-FDPs were made using a perfused clot system and subjected to gel filtration on Sephacryl S-1000. The M w of the FDPs in the eluate was determined on-line using multiangle laser light scattering (9). Fig. 1 shows a plot of the protein concentration and corresponding M w of the FDPs versus the volume of the eluate. The fractions that were pooled to give FDP samples having M w of 0.48 ϫ 10 6 , 1.08 ϫ 10 6 , 1.93 ϫ 10 6 , 3.08 ϫ 10 6 , 3.97 ϫ 10 6 and 4.94 ϫ 10 6 g/mol are indicated by shading.
Binding of 5AF-Pgn to FDPs-During the course of the t-PA/ DSPA␣1 cofactor activity experiments (see below), it was found that the fluorescent plasminogen derivative, 5AF-Pgn, displayed a reduced fluorescence in the presence of the FDPs relative to 5AF-Pgn alone. We used this property to investigate the binding of the 5AF-Pgn to the FDPs. The fluorescence changes versus the concentration of FDP were collected for all FDP sizes at all 5AF-Pgn concentrations. The data from each Pooled FDPs from perfusion fibrinolysis experiments were concentrated and passed over a Sephacryl-S1000 gel filtration column. The eluate from the column was passed through an absorbance monitor and a multiangle laser light scattering detector arranged in tandem. The figure shows the concentration of FDPs and the corresponding M w versus the volume of the eluate. The shaded boxes show the fractions that were pooled and concentrated to obtain the different samples. The M w (g/mol) values of the different samples are as follows: I ϭ 4.94 ϫ 10 6 , II ϭ 3.97 ϫ 10 6 , III ϭ 3.08 ϫ 10 6 , IV ϭ 1.93 ϫ 10 6 , V ϭ 1.08 ϫ 10 6 , and VI ϭ 0.48 ϫ 10 6 .
FDP sample of a particular M w were analyzed by nonlinear regression. This analysis showed no dependence of the K d on the FDP M w . Therefore, the data were fit globally using a single value for K d and letting ⌬I bound , the change in 5AF-Pgn fluorescence upon binding FDPs, vary for each of the FDP samples. Fig. 2 presents the observed data for the binding of 500 nM 5AF-Pgn to increasing concentrations of the different FDP samples (Fig. 2, symbols) as well as the calculated fit lines for each FDP sample from the global fit of the data at all concentrations of FDP and 5AF-Pgn. The percent decrease in fluorescence for the 5AF-Pgn bound to the different FDP samples was 5.8 Ϯ 0.8, 11.6 Ϯ 1.2, 19.4 Ϯ 1.8, 20.6 Ϯ 1.9, 20.5 Ϯ 1.9, and 21.7 Ϯ 2.0% for the FDPs having M w of 0.48 ϫ 10 6 , 1.08 ϫ 10 6 , 1.93 ϫ 10 6 , 3.08 ϫ 10 6 , 3.97 ϫ 10 6 , and 4.94 ϫ 10 6 g/mol, respectively. Despite the different extents of quenching of fluorescence, all FDPs bound with the same affinity (K d ϭ 225 Ϯ 60 nM). Although the differences in intensity changes suggest subtle size-dependent differences in the environment of bound 5AF-Pgn, the relevant parameter for modeling of kinetics is the K d . No binding was detected after treating the FDP samples with TAFIa (data not shown). The K d value and the apparent dependence of binding on carboxyl-terminal lysines and/or arginines are consistent with results reported by others (5-7) with partially degraded fibrin.
FDPs of M w Ranging from 0.48 ϫ 10 6 to 4.94 ϫ 10 6 g/mol Display Differential Cofactor Activity with Respect to t-PA and DSPA␣1-The different FDP samples were tested for their ability to serve as cofactors in the t-PA and DSPA␣1-catalyzed conversion of 5AF-Pgn to 5AF-Pn. The data from both the t-PA- (Fig. 3) and DSPA␣1 (Fig. 4)-catalyzed reactions were initially analyzed by regression of the data to the Michaelis-Menten equation according to the steady-state template model for Pgn activation as described by Horrevoets et al. (3).
The k cat (app), K m (app), and k cat (app)/K m (app) values for the reactions with t-PA, derived from the direct Michaelis-Menten plots, are shown in Table I. The data show that the reactions with t-PA exhibit Michaelis-Menten kinetics at any fixed FDP concentration for each of the different FDP samples when the substrate concentration is expressed as free 5AF-Pgn (Equation 5). The k cat (app) was found to increase with increasing FDP concentration, as expected in a cofactor-mediated, templatedependent reaction. No significant differences in k cat (app) between the different FDP samples, at any particular FDP concentration, were found. With the exception of the smallest FDP (M w ϭ 0.48 ϫ 10 6 g/mol), the K m (app) value for the reactions was fairly insensitive to the FDP concentration. This is consistent with reactions where K A ϳ K (Equation 8). For the FDP of M w ϭ 0.48 ϫ 10 6 g/mol, the K m (app) was found to decrease with increasing FDP concentration, consistent with K A Ͼ K for this FDP. The data indicate that the true K m value for reactions with t-PA (obtained at saturating FDP concentration, Equation 8) is the same for all FDPs, regardless of M w . The k cat (app)/ K m (app) ratios for all FDPs increase with increasing FDP concentration, indicative of a template mechanism.
The k cat (app), K m (app), and k cat (app)/K m (app) values for the reactions with DSPA␣1, derived from the direct Michaelis-Menten plots, are shown in Table II. The values for k cat (app) and K m (app) for the small FDPs of M w ϭ 0.48 ϫ 10 6 and 1.08 ϫ 10 6 g/mol could not be determined individually, since the plots did not exhibit saturable kinetics over the concentrations of 5AF-Pgn and FDPs used (Fig. 4). For the FDPs of higher M w the plots exhibited an approach to saturation with respect to substrate concentration and the k cat (app) increased with increasing FDP concentration. The data show that for FDPs with M w Ն 1.93 ϫ 10 6 g/mol, the k cat (app) with DSPA␣1 was insensitive to the M w of the FDPs. In contrast to that seen with t-PA, the K m (app) for the larger FDPs was found to increase with increasing FDP concentration, indicating that the K A Ͻ K for these reactions (Equation 8). Although the lack of saturation in the Michaelis-Menten plots of the reactions with the smaller FDPs (0.48 ϫ 10 6 and 1.08 ϫ 10 6 g/mol) precluded separate determinations of the k cat (app) and K m (app), and therefore any conclusions regarding the relationship between K A and K, the behavior of the k cat (app)/K m (app) with respect to FDP concentration was found from the slopes of the direct plots. Since the apparent catalytic efficiency was found to be a function of the FDP concentration, these data indicate that the smaller FDPs act as cofactors. The larger FDPs were found to influence plasminogen activation in a manner fairly independent of FDP size, as indicated by the behavior of the k cat (app)/K m (app) ratios with respect to FDP concentration (Table II).
By using the results from the direct plots as a guide, we fit the experimental data from each of the FDP samples with t-PA as the activator to the steady-state template model (Equation 4). When the data from all 5AF-Pgn and FDP concentrations were regressed together, we were unable to assign values simultaneously for all four parameters (k cat , K m , K A , and K) for each FDP size. We could, however, determine values for k cat , K A , and K, using a single value of K m , or for k cat , K m , and K, using a single value for K A , for all FDP sizes. From these two fits of the data, we found that the values for k cat and K were essentially invariant, with respect to FDP M w , when either the K m was fixed for all FDPs (k cat ϭ 0.065-0.081 s Ϫ1 , K ϭ 122-198 nM) or when the K A was fixed for all FDPs (k cat ϭ 0.076 -0.091 s Ϫ1 , K ϭ 140 -300 nM). Since the results from the direct Michaelis-Menten plots (Table I)  indicated by the solid lines in Fig. 3. The kinetic parameters obtained from the regression analysis for t-PA with each of the different FDP samples are shown in Table III. Consistent with the observations from the individual Michaelis-Menten plots, the value of K A is greater than the value of K for t-PA with 0.48 ϫ 10 6 g/mol FDPs, whereas K A ϳK with FDPs Ն 1.08 ϫ the smallest (M w ϭ 0.48 ϫ 10 6 g/mol) and the other (M w Ն 1.08 ϫ 10 6 g/mol) FDPs is minimal when t-PA is the activator and is attributable to an approximate 5-fold difference in the value of K A . The increasing FDP M w coincides with a decrease in the K A of the reactions, implying that larger FDPs have either more sites or higher affinity sites for t-PA binding.
The data from the DSPA␣1 experiments were also analyzed according to the steady-state template model based on differences in K A . When the data were modeled with either a single K A or K m , a fixed k cat was required for a solution for the small FDPs of M w ϭ 0.48 ϫ 10 6 and 1.08 ϫ 10 6 g/mol, due to the lack of curvature of the Michaelis-Menten curves for each FDP. For the FDPs of M w Ն 1.93 ϫ 10 6 g/mol, the k cat values were essentially invariant (k cat ϭ 0.327-0.384 s Ϫ1 ) and thus the value for k cat for the FDPs having M w ϭ 0.48 ϫ 10 6 g/mol and 1.08 ϫ 10 6 g/mol was fixed at the average k cat for FDPs M w Ն 1.93 ϫ 10 6 g/mol (0.358 s Ϫ1 ). Repeating the analysis showed that FDP of M w ϭ 0.48 ϫ 10 6 g/mol also required a fixed K for a solution to the rates observed at all FDP concentrations. The value of K for FDP with M w Ն 1.08 ϫ 10 6 g/mol was essentially invariant (1660 -2450 nM), and thus the value of K for FDP of M w ϭ 0.48 ϫ 10 6 g/mol was fixed at the average value of K for the FDPs of M w Ն 1.08 ϫ 10 6 g/mol (1990 nM). Fixing the parameters at the average values as an approximation was supported by the fact that a global fit of the data for all FDP sizes, using single k cat , K m , and K for all FDPs, yielded values of 0.35 Ϯ 0.15 s Ϫ1 , 670 Ϯ 340 and 2090 Ϯ 970 nM for k cat , K m , and K, respectively. The results of the regression analysis are indicated by the solid lines in Fig. 4. The kinetic parameters for DSPA␣1 with the different FDP samples are shown in Table  III. Consistent with the individual Michaelis-Menten plots using DSPA␣1 as the activator, the data show that K A Ͻ K (239 -588 versus 1660 -2450 nM) for FDPs with M w Ն 1.93 ϫ 10 6 g/mol. Furthermore, the data show that the k cat and K values are independent of the FDP M w when DSPA␣1 is the activator. Finally, the data show that the large decrease in the DSPA␣1 cofactor activity with the small FDPs is attributable to a large increase (ϳ43-fold) in the K A , indicating that the high affinity sites for DSPA␣1 disappear as the M w of the FDP approaches 0.48 ϫ 10 6 g/mol. This is consistent with the observations of Stewart et al. (2), who showed that DSPA␣1 binds to intact fibrin with high affinity (K A ϭ 150 Ϯ 40 nM) but not to fragment DD/E (K A Ͼ Ͼ 3000 nM), the smallest possible FDP (M w ϭ 0.25 ϫ 10 6 g/mol).

TAFIa Treatment of FDPs Eliminates the High Affinity 5AF-Pgn:FDP Binding, Markedly Reduces Both t-PA and DSPA␣1
Cofactor Activity, and Abrogates the FDP Size Dependence of DSPA␣1 Cofactor Activity-The rate constant for 5AF-Pgn activation is increased 2.5-fold by Pn-mediated exposure of carboxyl-terminal lysine and/or arginine residues on the fibrin surface during fibrinolysis (8). This feedback activation is down-regulated by the removal of the carboxyl-terminal lysines and arginines from the degraded fibrin surface by activated TAFIa, resulting in an attenuation of fibrinolysis (8). Since the soluble FDPs contain carboxyl-terminal lysine and arginine residues, we investigated the effect of TAFIa treatment of the FDPs on both t-PA-and DSPA␣1-catalyzed plasminogen activation. The FDPs from M w ϭ 0.48 ϫ 10 6 to 3.08 ϫ 10 6 g/mol were treated with TAFIa and then analyzed for their cofactor activity as described for the native FDPs.
The reaction rates were strictly linear with respect to the 5AF-Pgn concentrations for all TAFIa-treated FDP samples. Thus, we were unable to obtain individual k cat (app) and K m (app) values. The k cat (app)/K m (app) ratio for each reaction  (Equation 9). The data obtained with DSPA␣1 are shown in Fig. 5. Although not shown graphically, the results with t-PA were also linear, and no FDP M w -dependent differences in the rates of the reactions were observed. The k cat (app)/K m (app) ratios as functions of the FDP concentration for each FDP M w were analyzed by regression according to Equation 10 to obtain estimates of the true k cat /K m and for each activator with the TAFIa-treated FDPs. Table IV shows the dependence of the k cat (app)/K m (app) on FDP concentration for both t-PA and DSPA␣1. All TAFIa-treated FDPs exhibited cofactor behavior as seen by the increasing k cat (app)/K m (app) as a function of FDP concentration. Since these values showed at best a modest dependence on FDP M w , a global analysis of the data, using Equation 10 and a single true k cat /K m value for all TAFIa-treated FDPs, was performed. These yielded estimates of 0.20 Ϯ 0.01 ϫ 10 5 and 0.028 Ϯ 0.004 ϫ 10 5 M Ϫ1 s Ϫ1 for t-PA and DSPA␣1, respectively. The rate of 5AF-Pn formation at each FDP concentration was plotted against the 5AF-Pgn concentration, and the k cat (app) and K m (app) values were determined by regression to the Michaelis-Menten equation. The data show that the reactions with FDPs of M w Ն 1.93 ϫ 10 6 g/mol exhibit Michaelis-Menten kinetics with k cat (app) values increasing with FDP concentration, as expected in a cofactor-mediated reaction. The K m (app) values for these reactions increased with FDP concentration, indicating that K A Ͻ K. The reactions using FDPs of M w Յ 1.08 ϫ 10 6 g/mol did not show saturation over the 5AF-Pgn concentration range used. Thus, the k cat (app) and K m (app) values for these reactions could not be determined; however, the k cat (app)/K m (app) ratio was found from the slope of the rate versus 5AF-Pgn plot. The k cat (app)/K m (app) ratio for all FDP samples increased with increasing FDP concentration, indicative of a template mechanism. All data are shown Ϯ S.E. [

Summary of the kinetic parameters of t-PA-and DSPA␣1-catalyzed 5AF-Pgn activation in the presence of FDPs of different M w
The parameters from the regression of the data to the steady-state template model are presented Ϯ S.E. The reactions were modeled with k cat , K A , and K being dependent on FDP M w and a single K m for all FDPs. FDP cofactors yielded lower K m values with t-PA than with DSPA␣1. k cat and K were insensitive to the FDP M w with both activators. The data show that whereas DSPA␣1 has a higher turnover number, reactions with t-PA are more sensitive to FDP and 5AF-Pgn concentrations because of their respective lower K and K m values. DSPA␣1 activity displayed a profound size dependence on FDP M w , whereas that seen with t-PA was modest. These differences are reflected by the ranges of K A values observed for the two activators. A summary of the intrinsic k cat , K m , and k cat /K m (catalytic efficiency) values for t-PA and DSPA␣1-stimulated 5AF-Pgn activation using FDPs as cofactors, before and after treatment of the FDPs with TAFIa, is presented in Table V. The k cat and K m values presented for t-PA and DSPA␣1 with native FDPs were determined for each activator by fitting the data for all FDPs to a global steady-state template model using single values for k cat , K m , and K and values that were dependent on the FDP M w for K A . For comparison, Table V also shows values obtained by others, using intact fibrin (1,3) or fibrinogen (1) as the cofactor and either t-PA (1,3) or DSPA␣1 (1) as the activator with Glu-plasminogen (1, 3) and Lys-plasminogen (3) as substrates.
The Effect of TAFIa on t-PA-and DSPA␣1-mediated Fibrin-olysis in Plasma-The fibrinolysis of clots formed in human plasma, induced by t-PA or DSPA␣1, was prolonged by TAFIa in a concentration-dependent, saturable manner. As shown in Fig. 6, fibrinolysis induced with t-PA and DSPA␣1 was equally prolonged at low (Ͻ2 nM) concentrations of TAFIa, and the extent of prolongation with saturating TAFIa was marginally higher with t-PA (3-fold) than with DSPA (2.5-fold). When t-PA is the activator, TAFIa can prolong fibrinolysis by three separate mechanisms (8). As a basic carboxypeptidase, TAFIa removes exposed carboxyl-terminal lysine and arginine residues, thus preventing the plasmin-mediated up-regulation of the fibrin cofactor activity. In addition, TAFIa suppresses the plasmin-catalyzed conversion of Glu-plasminogen to Lys-plasminogen, a much better substrate for t-PA, thereby eliminating  9 under "Experimental Procedures"). The figure shows that the DSPA␣1 cofactor activity of the TAFIa-treated FDPs displayed only a modest dependence on M w . Furthermore, whereas all of the TAFIa-treated FDPs served as cofactors for DSPA␣1, they were all substantially less effective cofactors than their nontreated FDP counterparts (see Fig. 4).

TABLE IV Catalytic efficiency of t-PA-and DSPA␣1-catalyzed 5AF-Pgn activation in the presence of TAFIa-treated FDPs of different M w
The k cat (app)/K m (app) ratios were determined at each FDP concentration by linear regression of the reaction rate versus 5AF-Pgn concentration. The k cat (app)/K m (app) ratio for all FDPs samples increased with increasing FDP concentration, indicative of a template mechanism. TAFIa treatment yielded FDPs that were less active cofactors for both t-PA and DSPA␣1 than were their non-treated counterparts. t-PA was equally responsive to all FDPs, and the size dependence seen with DSPA␣1 was markedly attenuated by treatment of the FDPs with TAFIa. All data are shown Ϯ S.E. [ up-regulation through this mode. Finally, TAFIa can directly inhibit the activity of plasmin, although high TAFIa concentrations are required to achieve inhibition (8). The data in Fig.  6 show that fibrinolysis by DSPA␣1 is also down-regulated by TAFIa, most likely by the same mechanisms involved with t-PA.

DISCUSSION
The Catalytic Properties of t-PA and DSPA␣1 with FDPs as Cofactors-The intrinsic catalytic efficiency (k cat /K m ) of t-PA exceeds that of DSPA␣1 by ϳ3-fold when FDPs are used as cofactors and 5AF-Pgn is the substrate. t-PA and DSPA␣1 are qualitatively similar, however, in that their intrinsic k cat and K m values, and thus their intrinsic catalytic efficiencies are not dependent on the molecular weight of the FDP (Table III). The similarity in the k cat values with the different cofactors shows that the conversion of fibrin into FDPs and, by analogy, to partially degraded fibrin does not substantially alter the influence of the cofactor on the turnover of the ternary complex with either activator. Nonetheless, because three components are involved in the reactions, the reaction rates with both activators are sensitive to the FDP size when the reactions are not saturated with respect to the concentrations of FDP and 5AF-Pgn. Both activators showed decreased rates with decreasing FDP M w . The dependence is modest with t-PA but substantial with DSPA␣1 (Figs. 3 and 4). With both activators, the decrease in reaction rate with decreasing FDP M w can be accounted for by a decrease in the binding affinity of the activator for the smaller FDPs. With t-PA the decrease in affinity is modest, whereas with DSPA␣1 the decrease is large. For example, the K A values for the binding of t-PA to the two smallest FDPs (M w ϭ 0.48 and 1.08 ϫ 10 6 g/mol) are only 4.5-and 1.5-fold greater, respectively, than the average K A value for FDPs with M w Ն 1.93 ϫ 10 6 g/mol (182 nM). In contrast, with DSPA␣1, the corresponding K A values are 43-and 4-fold higher, respectively, than the average value for the FDPs with M w Ն 1.93 ϫ 10 6 g/mol (432 nM).
These differences can be rationalized on the basis of differences in structure between t-PA and DSPA␣1. t-PA interacts with fibrin by both its finger domain (K d ϭ 260 nM (4)) and its kringle-2 domain (K d ϭ 690 nM, (4)). DSPA␣1, which lacks the kringle-2 domain, interacts with intact fibrin with high affinity via the finger domain (K d ϭ 150 nM (1, 2)). Light scattering measurements by Stewart et al. (2) of the binding of t-PA to fragment DD/E, the terminal product of fibrin degradation by plasmin, indicated high affinity binding (K d ϭ 20 nM). The authors concluded that the binding was mediated by kringle-2 since the lysine analogue ⑀-amino caproic acid abolished the binding, and no binding of DSPA␣1 to the fragment was detected (K d Ͼ Ͼ 3000 nM). Thus, we infer that for the smallest FDPs used in our studies the predominant mode of binding of t-PA involves the kringle-2 domain and the lack of DSPA␣1 binding reflects its lack of a kringle-2 domain. This inference is supported by the quantitative agreement between the K A value for the binding of t-PA to the smallest FDP sample (895 nM), as determined by kinetics, and the direct measurement of the binding of a fingerless t-PA mutant to intact fibrin (690 nM (4)). The interactions between DSPA␣1 and the larger FDPs must be mediated, therefore, by its finger domain, and the modest increase seen in the affinity of t-PA for the larger FDPs is consistent with the ability of the larger FDPs to support both finger domain and kringle-2 domain-dependent binding. As a corollary, the smallest FDP does not contain the elements of structure necessary for finger-mediated binding, whereas the large FDPs do. fibrin, and fibrinogen as cofactors The data from the reactions with t-PA and DSPA␣1 were globally regressed to the steady-state template model using single values for k cat , K m , and K for all FDPs. The data show that the intrinsic catalytic efficiency of t-PA is higher than that of DSPA␣1 when the cofactor is either FDPs or TAFIa-treated FDPs. With FDPs as the cofactor, the intrinsic catalytic efficiency of t-PA is increased by a factor of 3(1) to 10(3) compared with fibrin and approaches that seen with fibrin when the substrate is Lys-plasminogen (3). Although the intrinsic catalytic efficiency of DSPA␣1 with FDPs was the same as that found by others (1) with fibrin, the effect of TAFIa on DSPA␣1-mediated fibrinolysis (Fig. 6) shows that the intrinsic catalytic efficiency of DSPA␣1 with FDPs is higher than that seen with fibrin (see "Discussion"). Although TAFIa treatment of the FDPs markedly decreases the catalytic efficiencies of t-PA (90-fold) and DSPA␣1 (210-fold), to values below those for intact fibrin, TAFIa-treated FDPs are still superior cofactors for both t-PA (30-fold) and DSPA␣1 (50-fold) in comparison to fibrinogen.  6. Effect of TAFIa on t-PA and DSPA␣1-mediated fibrinolysis in plasma. Clots were made in TAFI-deficient human plasma by the addition of purified human thrombin in the presence of varying concentrations of TAFI. Either t-PA or DSPA␣1 was included as the plasminogen activator. Solulin was included to promote the rapid formation of TAFIa. Fibrinolysis was followed by monitoring the turbidity of the clots at 405 nm. The figure shows the effect of TAFIa on the time required to achieve the Lysis Time, the time required to reach 50% lysis, expressed relative to the Lysis Time observed in the absence of TAFIa. The figure shows that TAFIa increases the lysis time of both t-PA and DSPA␣1 mediated reactions in a saturable manner and that TAFIa maximally prolongs DSPA␣1-mediated reactions marginally less (2.5fold) than it does reactions with t-PA (3-fold). contained in fragments larger than DD/E (9). This suggests that FDPs much larger than DD/E are released into circulation during fibrinolysis, and indeed FDP complexes of at least 2.0 ϫ 10 6 Da have been observed in the plasma of patients with disseminated intravascular coagulation (13,14) and chronic subdural hematoma (15). Whereas a large difference in activity between t-PA and DSPA␣1 would be expected with DD/E, this difference disappears with larger FDPs. Therefore, the relative fibrin specificity of DSPA␣1 compared with t-PA would be attenuated in the presence of larger FDPs.
Considerations on the Use of Plasminogen Activation to Measure Soluble Fibrin and Fibrin Degradation Products in Plasma-The t-PA and DSPA␣1 cofactor activity of FDPs larger than DD/E can be attenuated by the action of TAFIa. Our studies have implications regarding the use of plasminogen activation assays for the measurement of soluble fibrin and/or FDPs in plasma. The rate of plasminogen activation in plasma will be determined by the relative concentrations of soluble fibrin, native FDPs, FDPs exposed to TAFIa, and fibrinogen. Since the K m of the reactions with the FDPs is 10-fold lower and the intrinsic catalytic efficiency is 10-fold greater than with fibrin, and therefore presumably with soluble fibrin, small amounts of soluble FDPs could compromise the measurement of soluble fibrin. Furthermore, TAFIa treatment attenuates the cofactor activity of FDPs by ϳ100-fold. In determinations of FDPs, the rate of plasmin generation in the assay will essentially be a measure of the concentration of the native FDPs, which may or may not reflect the total concentration, the value of which will be underestimated according to the extent of FDP exposure to TAFIa. Thus, measurements using plasminogen activation represent the composite cofactor activity of plasma, which may or may not be a good measure of the species in question. The complexity of the cofactor mixture and differences in the kinetics associated with each component in the mixture may explain the differences observed with antibodybased and cofactor activity-based assays in the determination of the concentrations of both soluble fibrin and FDPs (16 -18).