Quantitative determination of the binding of epsilon-aminocaproic acid to native plasminogen.

The binding of epsilon-amino[14C]caproic acid (6-aminohexanoic acid, EACA) to native human plasminogen was determined using the ultrafiltration technique of Paulus (Paulus, H. (1969) Anal. Biochem. 32, 91-100) at free ligand concentrations ranging from 2 micrometer to 16 mM. One strong binding site (Kd = 0.009 mM) and approximately five weaker ones (Kd = 5 mM) were found. The constants were obtained by fitting the experimental points to the simple assumption of two sets of noninteracting sites. The distinct separation of the two kinds of sites allowed the correlation of the well known epsilon-aminocaproic acid-induced conformational transition in plasminogen with the saturation of the weaker group of binding sites by this ligand. The conformational transition was monitored by measurements of the sedimentation coefficient, as was done by others earlier. The midpoint of the transition occurred at approximately 3.3 mM free ligand. A dissociation constant of 0.32 mM was also obtained for L-lysine by measurements of the competition between this compound and labeled epsilon-aminocaproic acid for the strong binding site. The correlation between epsilon-aminocaproic acid binding and effects of the compound on various physical and functional properties is discussed. A discussion of the possible sources of error encountered in the technique used is also included.

between r-aminocaproic acid binding and effects of the compound on various physical and functional properties is discussed. A discussion of the possible sources of error encountered in the technique used is also included.
The discovery that l -amino-n-caproic acid (6.aminohexanoic acid) inhibits the conversion of plasminogen to plasmin was made by Okamoto in 1954 (1). Since then, the literature on this substance has been steadily growing and by now includes a large volume of clinical material dealing with the use of this substance in a variety of hemorrhagic and clotting disorders. Its best known biochemical effect, the inhibition of plasminogen activation, is not fully understood even today and is probably the result of several independent effects of the compound. For example, the in vitro inhibition by Eaminocaproic acid of iibrinolysis by urokinase-activated plasminogen could be the net result of the effects of this compound on plasminogen, on urokinase (Z), on the plasmin formed, and on fibrinogen and fibrin (3,  present communication will be restricted to the effects of Eaminocaproic acid on plasminogen itself. The first clue as to the mode of action of l -aminocaproic acid was found by Alkjaersig (5) who demonstrated that the sedimentation coefficient of human plasminogen decreased substantially in the presence of the compound, and concluded that the change in the physical parameter reflected a change in the conformation of the protein molecule. She also noted that the change in the sedimentation velocity of plasminogen occurred at a higher concentration than did the earlier observed augmentation of solubility of acid-prepared plasminogen by the compound. The effects on the sedimentation rate were confirmed by Brockway and Castellino (6) and were extended to changes in circular dichroism (7) (see also Sjiiholm et al. (8)), and, in the same laboratory, to changes in rotational diffusion by measurements of polarization of fluorescence (9). The concentration dependence of these physical changes suggested that they are manifestations of the same conformational transition.
Recently, new insight was gained into the conformational properties of plasminogen when it was realized that this protein, unless prepared under conditions which preclude proteolytic attack on the molecule during purification, will suffer a cleavage near its NH,-terminal end (10). This "modified plasminogen" differs from its native counterpart by having an increased hydrodynamic volume, altered conformational parameters, and greater ease of activability by urokinase, properties which resemble those caused by interaction of the zymogen with l -aminocaproic acid. This realization led Sjoholm et ~2. (8) to recognize that the conformation of modified plasminogen, and of native plasminogen in the presence of l -aminocaproic acid, are closely related.
This phenomenon has been confirmed and extended by other groups of workers in the field (11)(12)(13)(14)(15). Recently, the important discovery was made by Rickli and Otavsky (16) that the site, or sites, on the plasminogen molecule which strongly binds the compound, are located on the portion of the peptide chain which upon activation becomes the heavy, or A chain of plasmin (17). They found that only the heavy chain of mildly reduced and carboxymethylated plasmin was adsorbed to a lysine/Sepharose affinity column; the light chain which carries the active site (17)  were placed into counting vials and were soaked overnight in 1 ml of 1 mM unlabeled c-aminocaproic acid. After addition of 10 ml of Bray's solution, the samples were counted for 10 min each in a Beckman LS-230 liquid scintillation system. The counts ranged from 1,000 to 30,000 cpm.
Calculations-Moles of bound ligand were evaluated from the radioactivity retained on the discs and the specific activity of the stock solution, a sample of which was counted with each run. The amount of the free ligand was obtained by subtraction of the bound from the known total ligand, present initially. The molar binding ratio, i, was obtained by dividing the moles of bound ligand by the moles of protein present in the sample (molecular weight of plasminogen was taken as 91,000). The correction for the amount of free ligand contained in the small amount of liquid retained by the discs was carried out by: (a) calculating the volume retained on the discs from the amount of radioactivity retained after filtration of solutions of known radioactivity but containing no protein (microliters retained = counts retained/counts/p1 of the same ligand solution); (b) obtaining the amount of free ligand retained by assuming that the concentration of free ligand is the same in that volume as it was in the bulk of the solution filtered through the discs; and (cl finally, subtracting this amount from the total amount retained. The volume retained by the discs was determined to be 1.00 ~1 + 0.18 (S.D.), i.e. 1/150th of the total volume filtered.
This correction is very small in the initial, tight binding region, not more than about 1% of the total counts retained; however, in the weaker binding region, where the concentration of free ligand is very high, as much as 70% of the total amount retained can be free ligand. Under the latter conditions, even a small error in the estimate of retained volume leads to a substantial error in i and it is this error which ultimately sets the limit to the usefulness of the method. It is essential that the correction be carried out in this manner, as the procedure recommended in the original publication of the method (22), subtraction of the counts retained by a disc after filtration of a protein-free sample (blank) from the counts retained by a disc after filtration of a similar, but protein-containing solution, leads to erroneous results. It is easy to see that, in the case of strong binding, the concentration of free ligand in the solution retained from the protein-containing sample will be much less than in the corresponding protein-free sample, and the correction therefore will result in an underestimation of the amount bound, and may, in effect, lead to negative amounts bound. These calculations of binding parameters, including the above correction, were carried out by using a computer program written for the Sharp Compet 363P desk calculator.

l -Aminocaproic
Fitting of Binding Curve-Fitting of the experimentally determined points by a calculated curve was carried out using a twocomponent binding function of the form i = (il + fZ) = k,n,C/(k,C + 1) + k,n,Cl(kC + 1) (33). A second program was written for the evaluation of i, and iZ as functions of C, the free ligand concentration, for selected trial values of k,, k,, n,, and n,, i.e. the two association constants and the number of binding sites in the two groups of sites, respectively.
Sedimentation -Measurements of the sedimentation coefficient of plasminogen were carried out in a Beckman-Spinco model E analytical ultracentrifuge at 56,100 rpm, at 20".

RESULTS
Binding of l -Aminocaproic Acid - Fig. 1 shows a Scatchard plot of the binding data obtained for this ligand using a single preparation of native human plasminogen at pH 7.8 and 20". The data were collected in four separate experiments and cover a range of free ligand concentrations from 2 PM to 16 mM. Two binding regions, a strong one and a very much weaker one, are easily distinguished.
In all fully activable preparations examined in the course of this study, a single strong binding site was found consistently, along with much weaker ones, the exact number of which is less certain. The best fit for the entire curve (see "Materials and Methods") was obtained with the assumption of 0.93 strong site and 4.93 weak sites, and the association constants 112 rnrv-', and 0.2 mM-1, respectively (corresponding dissociation constants are 0.009 and 5 mM, respectively). The broken line in Fig. 1 is calculated using these constants. The assumption of four, rather than five, sites for the second group gave inferior tits which could not be improved by changing k,; also, in none of the binding experiments in this study did r exceed a total of 6, even when the free ligand rose to 33 mM. Fig. 2 shows the same data on a semilogarithmic plot off versus the free ligand concentration.
This "titration curve" separates the two binding regions very clearly and expands the region of weaker binding. As can be seen, the calculated line, while compatible with the data, unfortunately does not establish directly the limiting value of 6, due to our inability to obtain reliable data at free ligand concentrations much higher than 10 mM (see "Materials and Methods").

Correlation of Sedimentation
Coefficient with Binding of c-Aminocaproic Acid -Having realized that there are two distinct sets of sites for the binding of l -aminocaproic acid, it became important to determine which of these was responsible for the conformational transition which occurs upon addition of increasing amounts of the ligand to plasminogen (see below). We chose to look at the changes in the sedimentation coefficient since this parameter is easily determined and because there already exists reliable data on the subject. The total ligand concentration range used for the experiment was from 1 pM to 0.1 M. The sedimentation was carried out in the same buffer and at the same protein concentration at which the rest of the binding experiments were done. In addition, samples identical in composition with those used for the ultracentrifugation, but containing radioactive c-aminocaproic acid, were also made up for the determination of the ligand binding. Both sets of these data are shown in Fig. 2. It is quite apparent that the conformational change encompasses the entire range of the binding data and is certainly not restricted to the saturation of the first site. In effect, only about 15% of the total change in the sedimentation coefficient coincides with the saturation of the single strong site. It should be noted, however, that the saturation of the first site Acid Binding by Plasminogen still appears to have an effect on the sedimentation coefficient, since a 15% decrease in this parameter at the level of? = 1 is considerably more than could be accounted for by the calculated saturation of the secondary sites, which should be filled only to 6% of their capacity at this point. Thus, the saturation of both kinds of sites may contribute to the conformational change; nevertheless, the bulk of that change must be ascribed to ligand binding to the secondary group of sites. Taking 5.52 and 4.40 as the initial and final values of the sedimentation coefficient, the midpoint of the transition, 4.96, is at 3.3 mM free l -aminocaproic acid. This value compares favorably with the dissociation constant of 5 mM, assumed for the fitting of ligand binding to the secondary sites (see below), but it is significantly higher than 0.45 mM, the value obtained by Brockway and Castellino (6) in their study of the sedimentation behavior of human plasminogen in l -aminocaproic acid.
Our value, 3.3 mM for the midpoint of the transition, is identical with the value obtained for freshly prepared rabbit plasminogen, determined more recently by Violand et al. (34). A plausible explanation for the discrepancy between the results obtained on the human material may be that Brockway and Castellino (6) eluted their plasminogen from the affinity column by use of an e-aminocaproic acid gradient and then used only the fractions with the higher affinity for the ligand, while we did not fractionate our plasminogen in this manner: this would certainly result in a somewhat stronger binding in their preparation.
A difference in the source of the plasminogen may also have contributed to the discrepancy; Brockway and Castellino, in the study cited (61, prepared their plasminogen from Cohn Fraction III, while we used fresh plasma. It is safe to assume, however, that whatever the exact value is, the change in sedimentation velocity must be correlated in both cases with the ligand saturation of the weaker set of sites, since the dissociation constant for the first site, 0.009 mM, would be far too small to fit the transition in either study. Competition by L-Lysine-If the binding of l -aminocaproic acid is of physiological significance, it is probably a reflection of affinity of plasminogen to the natural substance, L-lysine, or to a lysyl residue, and in effect, the affinity chromatographic purification of the proenzyme, developed by Deutsch and Mertz (301, utilizes lysine as the specific ligand for the column. The effect of this amino acid on the sedimentation velocity of human plasminogen was measured by Brockway and Castellino (6); lysine depressed the coefficient to the same value as did e-aminocaproic acid, but the midpoint of the transition for lysine was much higher than that for the other compound. From the discussion above, it seems probable, therefore, that the conformational effect of lysine, as manifested by the change in sedimentation coefficient, is exerted by binding to the secondary sites. Since attempts to measure the binding of L-[WIlysine directly and reliably failed due to its relatively weak affinity to plasminogen we estimated the dissociation constant by measuring the effect of unlabeled lysine on the binding of e-amino[Wlcaproic acid in the concentration range in which the first, strong site is saturated. Two kinds of inhibition experiments were done: (a) l -aminocaproic acid concentration was varied at a single, subsaturating (with respect to the first site) L-lysine concentration. An example of the first type is shown in the form of a Scatchard plot in Fig. 3. The experiment carried out in the presence of 0.185 mM L-lysine shows a decrease in the binding energy but no significant loss in the probable value of n,. Even though the best visually drawn lines intersect somewhat below the abscissa, the data were treated only assuming competition between the two ligands, and the association constant for n-lysine was evaluated from the intercept on the T/c axis at i = 0, using the formula for this limiting case: i/c = h/(1 + k,.I), where k, r', n, and c have their usual significance (see above) and all refer to l -aminocaproic acid; I is the concentration of L-lysine, and K, is its association constant2 For the case illustrated in Fig. 3, a dissociation constant of 0.38 mM was obtained for clysine, and in another experiment using a higher lysine concentration, a value of 0.31 was obtained for the same parameter.
(b) The second type of experiment is illustrated in the inset to Fig. 3 and shows a plot of r for r-aminocaproic acid against the concentration of lysine used, at a constant concentration of e-aminocaproic acid. The dissociation constant of a competitive inhibitor can be evaluated from the conditions at 50% inhibition by the formula derived by Cheng   We have not attempted a direct, or even a displacementbased measurement of the binding of lysine to the secondary sites, but according to the sedimentation data of Brockway and Castellino (6), lysine binds about 150 times weaker than l -aminocaproic acid to that group of sites. DISCUSSION The information acquired on the binding of l -aminocaproic acid to human plasminogen now enables us to draw correlations between the amount and strength of ligand binding and some of the known physical and biochemical effects of the compound. It seems that the effects of saturating the single strong binding site are more subtle than those associated with occupation of the weaker ones. We have mentioned above that the range of l -aminocaproic acid concentration in which the solubility of acid-prepared plasminogen was improved (5) preceded that in which the sedimentation constant was altered. It is very likely, therefore, that this effect was due to binding of the first ligand molecules.. A more significant conclusion that can now be drawn is that the inhibition of activation rate by urokinase is also associated with the occupation of the strong binding site. The rate of urokinaseactivation of both native and modified human plasminogen is significantly decreased even at the range of 1 pM to 0.1 mM Eaminocaproic acid, as shown by the work of Thorsen et al. (14,15). Between 0.1 and 1 mM this inhibition of activability is followed, in the case of native plasminogen, by a marked increase in activation rate. The region of this facilitation coincides with the saturation of the secondary sites, and it is also the region where the conformational transition discussed above takes place. There is no doubt that binding of e-aminocaproic acid to the secondary sites results in an "opening up" of the plasminogen structure; facilitation of limited proteolysis (hydrolysis of the activation bond), decrease in sedimentation rate (increased hydrodynamic volume), and reduction of the rotational relaxation time (increased intramolecular mobility) are all compatible with increased relative separation of segments within the molecule. On the other hand, the decrease in the activability at the low concentration range just mentioned is most easily explained by the phenomenon of "ligand stabilization," i.e. by a decrease in conformational fluctuations along the peptide chain, an effect which increases with increasing free energy of ligand binding, and results in decreased hydrolyzability of peptide bonds (36,37). The inhibitory effect of l -aminocaproic acid on the streptokinase activation of human plasminogen is a complicated one in that it depends heavily on the predominant activator form present; activation by the plasminogen-native . streptokinase complex is inhibited much more effectively than when the activator is the plasmin-modified.streptokinase complex.3 At present, no definite statement can be made on the effective concentration range of the compound in this case. Finally, it should be noted that while both native and modified plasminogen possess the one primary and the five secondary sites, the primary