Immunochemical Characterization of a Low Affinity Lysine Binding Site within Plasminogen*

A degradation product corresponding to the fourth kringle has been isolated from porcine elastase digests of human plasminogen, and an antiserum has been utilized to immunochemically characterize this deriva- tive. This antiserum bound radioiodinated kringle 4 and plasminogen, but 0.2 M 6-aminohexanoic acid markedly reduced the binding of these ligands. Fur-thermore, introduction of three w-aminocarboxylic acids, trans-4 - aminomethylcyclohexanecarboxylic acid, 6-aminohexanoic acid, and lysine into the radioimmunoassay produced a concentration-dependent inhi- bition of the binding of either kringle 4 or plasminogen by the antiserum. The concentration of these deriva- tives producing 50% inhibition was reflective of the dissociation constant of the w-aminocarboxylic acids for the low affinity lysine binding within kringle 4. Modification of kringle 4 by reduction and alkylation, cyanogen bromide cleavage, or chymotrypsin degradation markedly decreased or abolished its capacity to interact with lysine-Sepharose and caused a Concomi-tant decrease in its capacity to be bound by the anti- serum. Taken together, these observations suggest that an available and functionally intact lysine binding site is required for kringle 4 and kringle 4-containing derivatives to be bound by the antiserum. Only kringle 4- containing derivatives of plasmin(ogen) interacted micro-Bonda-pak A 25-min gradient employed from 70%/30% to 50%/50% Solvent A/ Solvent B under 1500 p.s.i. Elution monitored at 212 Plasmin(ogen) Derivatives-Plasminogen was converted to plas- min with either urokinase or streptokinase, and complete conversion was established by SDS-polyacrylamide gel electrophoresis under reducing conditions. The plasmin. a*-antiplasmin complex was formed in a 10% excess of arantiplasmin and was purified by affinity chro- matography on lysine-Sepharose followed by molecular exclusion chromatography on Ultrogel AcA44 (22, 23). The plasmin light and heavy chains were isolated following mild reduction and alkylation of plasmin. The light chain was obtained as the unbound fraction from lysine-Sepharose, and the heavy chain was eluted with 0.2 M 6-aminohexanoic acid (13). Purity of all derivatives was assessed by SDS-polyacrylamide gel electrophoresis, and a homogeneity of >90% for each derivative was indicated by the staining patterns.

inhibitors as well as by structural features inherent to the plasmin(ogen) molecule itself. The capacity of plasmin(ogen) to reversibly bind w-aminocarboxylic acids such as lysine, 6aminohexanoic acid, and trans-4-aminomethylcyclohexanecarboxylic acid is mediated by specific "lysine binding sites" (1)(2)(3)(4)(5)(6)(7), and these sites appear to play an important role in regulating interaction of plasmin with its primary substrate, fibrin (8,9), and its primary inhibitor, a*-antiplasmin (10,11). Markus et al. (5) demonstrated that plasmin contains two classes of binding sites with respect to affinity for 6-aminohexanoic acid: one high affinity site with a Kd of 9 PM and approximately five low affinity sites with a Kd of 5 mM. In addition, in comparing the binding of 6-aminohexanoic acid to native Glu-plasminogen and modified Lys-plasminogen, which arises from limited degradation in the NHn-terminal region of Glu-plasminogen, significant differences were noted (12).
The lysine binding sites of plasminogen reside in the region from which the heavy chain of plasmin is derived (13). The entire plasminogen molecule has now been sequenced (14, 15), and the lysine binding sites are associated with the fmt four of the five "kringle-like" structures of plasminogen (15). These kringles are envisioned as looped disulfide structures of 80-90 amino acids and are highly homologous in sequence to one another and to the nonthrombin region of prothrombin (15,16). When plasminogen is subjected to limited digestion with porcine elastase, three major derivatives are obtained. Elastase degradation product I contains kringles 1, 2, and 3 and retains capacity to interact with lysine derivatives. EDP 11,' a fragment of M , = 10,000-12,000 corresponds to kringle 4 and contains a single low affiity lysine binding site. EDP I11 consists of kringle 5 and the remaining carboxyl-terminal region of plasminogen. EDP I11 does not bind to lysine-Sepharose and may be activated to form a low molecular weight plasmin (7,15). In the present study, we describe an antiserum to EDP I1 which appears to be entirely specific for the lysine binding site within this fragment. The antiserum has been utilized to characterize the interaction of a-aminocarboxylic acids with kringle 4 and to compare this lysine binding site as expressed by isolated EDP I1 or when contained in the plasminogen molecule.

MATERIALS AND METHODS
plasma or from Cohn's Fraction I11 by affinity chromatography on Plasminogen-Glu-plasminogen was isolated from fresh human lysine-Sepharose, followed by molecular exclusion chromatography on Bio-gel A1.5m or Ultrogel AcA44 (3,17). Lys-plasminogen was The abbreviations used are: EDP (I, 11, 111), elastase (porcine) degradation products of plasminogen; AMCA, trans-4-aminomethylcyclohexanecarboxylic acid; SDS, sodium dodecyl sulfate; ABC-33%, antigen binding capacity derived from the a n t i s e m dilution binding 33% of the ligand. isolated from Fraction 111, but the extraction and isolation were performed at 22 "C (12). By NH~terminal analysis (18), glutamic acid was the only detected NH~-tenninal residue in Glu-plasminogen preparations, and Lys-plasminogen contained predominantly lysine as well as traces of methionine, valine, and serine as NH2 termini. These NHz termini are consistent with the reported properties of Glu-and Lys-plakminogen (19)(20)(21). Elastase Degradation Products-Limited digestion of Clu-plasminogen by porcine elastase was performed according to the method of Sottrup-Jensen et al. (15). Briefly, 150 mg of Glu-plasminogen in 10 ml of 0.3 M NH.HCO3, pH 8.3, containing 1 mg of pancreatic trypsin inhibitor was incubated for 3.25 h in 22 "C with 0.45 mg of porcine elastase. Diisopropylfluorophosphate was added to a final concentration of 0.15 mM, and after 15 min, solid NH~HCOI was added to adjust the final bicarbonate concentration to 0.55 M. Following an overnight incubation, the digest was applied to a column (2.5 X 30 cm) of lysine-Sepharose equilibrated in 0.1 M NH~HCOJ. The unbound fraction contained EDP 111, which was further purified by chromatography on Sephadex G-75, and the bound fraction containing EDP I and EDP I1 was eluted with 0.2 M 6-aminohexanoic acid.
This eluate was concentrated by lyophilization and subjected to molecular exclusion chromatography on a column of Sephadex G-75 (2.5 X 90 cm) in 0.1 M NH4HC03, pH 8.3. The major high molecular weight derivative was EDP I (see Fig. lB), and the second major fraction containing EDP I1 was lyophilized, redissolved in a minimum volume of 10% acetic acid, and further purified by high pressure liquid chromatography on an analytical Cla (reverse-phase) micro-Bondapak column (Waters, Milford, MA). Solvent A was 0.1 M NaH2P04, 0.1 M &PO,, and Solvent B was 605% acetonitrile/40% Solvent A. A 25-min gradient was employed from 70%/30% to 50%/50% Solvent A/ Solvent B under 1500 p.s.i. Elution was monitored at 212 nm.
Plasmin(ogen) Derivatives-Plasminogen was converted to plasmin with either urokinase or streptokinase, and complete conversion was established by SDS-polyacrylamide gel electrophoresis under reducing conditions. The plasmin. a*-antiplasmin complex was formed in a 10% excess of arantiplasmin and was purified by affinity chromatography on lysine-Sepharose followed by molecular exclusion chromatography on Ultrogel AcA44 (22,23). The plasmin light and heavy chains were isolated following mild reduction and alkylation of plasmin. The light chain was obtained as the unbound fraction from lysine-Sepharose, and the heavy chain was eluted with 0.2 M 6aminohexanoic acid (13). Purity of all derivatives was assessed by SDS-polyacrylamide gel electrophoresis, and a homogeneity of >90% sample containing approximately m, oOO cpm of '%I-EDP II was treated in parallel and was utilized to assess binding of EDP I1 on columns of lysine-Sepharose (0.9 X 10 cm) on the basis of radioactivity (10-30 mg of lysine/ml of Sepharose 2B).
Radioimmunoassay for EDP ZZ-Rabbit antiserum to EDP I1 was prepared by injecting 100 pg of EDP I1 initially in complete Freund's adjuvant, followed by biweekly injections in incomplete Freund's adjuvant. An early bleeding following the fourth immunization was utilized for this study. EDP I1 was radioiodinated by a modified chloramine-T procedure (24). To 50 pg of EDP I1 in 50 pl of 0.1 M sodium phosphate, pH 7.2, 1 mCi of carrier-free '%I (17 Ci/mg, Amersham) and 10 pg of chloramine-T were added. After 5 min at 22 O , 10 pg of sodium metabisulfite were added. Bovine serum albumin (1%) was added to bring the final volume to 600 pl, and the sample was dialyzed extensively to remove free '%I. The capacity of T -E D P I1 to bind to lysine-Sepharose was 80-9076, and greater than 90% of the radioactivity was precipitable by 15% trichloroacetic acid.
The radioimmunoassay system utiliied was of the double antibody type, employing goat anti-rabbit immunoglobulin to achieve precipitation in a 0.04 M borate buffer, pH 8.3, containing 0.025 M NaCl and 2% normal rabbit serum (25). '=I-EDP I1 or '2sII-Glu-plasminogen at 0.7 nM, inhibitors, and antiserum were added in 0.5-ml additions, and after an 18-h incubation at 4 "C, 0.5 ml of optimally diluted second antibody was added. After 6 h at 4 "C, immunoprecipitates were removed by centrifugation and 1.0 ml of the supernatant was counted. Per cent inhibition was calculated relative to controls lacking inhibitors, and antigen binding capacities (ABC-33%) were calculated from the dilution of the antiserum required to achieve 33% binding of the ligand (26).
SDS-Polyacrylamide Gel Electrophoresis-The system of Swank and Munkres (27) with the samples in 6 M urea, 1% SDS was employed, utilizing 6-cm cylindrical gels of 9.6% acrylamide and 0.8% bisacrylamide. Electrophoresis was for 2.5 h at 5 mA/gel. Molecular weights were estimated from relative mobilities using myoglobin (17,000) and a cyanogen bromide digest of myoglobin containing fragments of 14,500, 8,400, and 6,800 for calibration. Purity of higher molecular weight derivatives was assessed on 7.5% polyacrylamide gels according to the method of Weber and Osborn (28). Reagents-Sephadex and Sepharose resins were from Pharmacia, Bio-Gel A1.5 from Bio-Rad, and Ultrogel AcA44 from LKB. Of the w-aminocarboxylic acids, lysine and 6-aminohexanoic acid were pur-chased from Sigma and AMCA was from Kabi, Stockholm, Sweden. Porcine elastase, chymotrypsin, and trypsin were from Worthington.

RESULTS
Plasminogen with glutamic acid as the NH2 terminus was isolated from human plasma or Cohn's Fraction I11 and subjected to limited elastase digestion. When the digest was chromatographed on lysine-Sepharose (Fig. lA), the unbound material contained EDP I11 and the bound protein was eluted with 0.2 M 6-aminohexanoic acid. This eluate was subjected to molecular exclusion chromatography on Sephadex G-75 (Fig. IB), and two major as well as two minor components were resolved. The second major peak, eluting from 350 to 380 m l , contained EDP I1 and was further purified by reverse phase chromatography on high pressure liquid chromatography (Fig. IC). The fiial product was homogeneous as judged by its high pressure liquid chromatography elution pattern and by polyacrylamide gel electrophoresis in the presence of urea and SDS (Fig. 1D). The molecular weight of the isolated derivative was estimated to be 11,000, consistent with the reported value of EDP I1 (7, 15). The identity of the isolated derivative as EDP I1 was based upon its amino acid composition (Table I), which is in reasonable agreement with that predicted from the published sequence of EDP 11, and by identification of valine as the only detectable NH2-terminal amino acid (7, 15). Antiserum to EDP I1 was raised in rabbits and utilized to develop a double antibody radioimmunoassay system. Both lz5I-EDP I1 and '251-Glu-plasminogen were effectively bound by the antiserum (Fig. 2). At a 1/50 dilution of the antiserum, greater than 85% of each ligand was bound, and the ABC-33% was 1.4 nmol of EDP I1 and 0.8 nmol of Glu-plasminogen bound/ml of antiserum. When 6-aminohexanoic acid at a final concentration of 0.2 M was included in the assays, binding of either ligand by the antiserum was markedly reduced. The reductions represented at least a 35-fold decrease in the antigen binding capacity of the antiserum for the ligands in the presence of 6-aminohexanoic acid.
With numerous other antisera to plasminogen derivatives and to unrelated proteins, 6-aminohexanoic acid had no effect on antibody binding. The interaction of 6-aminohexanoic acid with the lysine binding site of EDP I1 could result in the observed inhibition of lz5I-EDP I1 and 1251-Glu-plasminogen binding by anti-EDP 11, and this possibility was explored in detail. Serial dilutions of lysine, 6-aminohexanoic acid, and AMCA were introduced into the radioimmunoassay as potential inhibitors for the binding of lZ5I-EDP I1 by anti-EDP 11. As shown in Fig. 3, each derivative produced a concentrationdependent inhibition curve and caused complete inhibition at high concentration. The concentrations required for 50% competitive inhibition were 5.0, 0.24, and 0.047 mM for lysine, 6aminohexanoic acid, and AMCA, respectively; the series AMCA > 6-aminohexanoic acid > lysine is consistent with their relative potency as antifibrinolytic agents (4). Generically similar results were obtained utilizing '251-Glu-plasminogen as the ligand with anti-EDP 11, and the concentrations of the lysine analogues required for 50% inhibition are summarized in Table 11. In each case, a higher concentration of the w-aminocarboxylic acid was required for 50% inhibition of '251-Glu-plasminogen binding, but the relative inhibitory ca-  pacity of AMCA > 6-aminohexanoic acid > lysine was maintained with either ligand. The effect of radioimmunoassay conditions on the concentration of 6-aminohexanoic acid required for 50% inhibition was assessed. In six separate determinations, the concentrations of 6-aminohexanoic acid required for 50% inhibition of the binding of lZ5I-EDP I1 and '251-Glu-plasminogen by anti-EDP I1 were 2.8 f 1.4 X and 14.9 f 5.6 X

M,
respectively, indicating that the values were reproducible.
Varying the anti-EDP I1 dilution to produce either 80 or 25% binding of lz5I-EDP I1 in the assay in the absence of 6aminohexanoic acid did not alter the concentration of 6-aminohexanoic acid required for 50% relative inhibition. Furthermore, varying the initial incubation time of lz5I-EDP 11, anti-EDP 11, and 6-aminohexanoic acid from 18-72 h and the immunoprecipitation phase of the assay from 6-48 h also did not change the 50% relative inhibition value. The kringles of plasminogen exhibit a high degree of primary structural homology (15), and the reactivity of anti-EDP I1 with various plasminogen derivatives was examined. As shown in Fig. 4, only EDP 11-containing derivatives inhibited the binding of ls51-EDP I1 by anti-EDP 11. Thus, only EDP 11, the heavy (A) chain of plasmin, and Glu-plasminogen competitively inhibited, whereas EDP I, EDP 111, and the light (B) chain of plasmin did not inhibit. Higher concentrations of the heavy chain of plasmin (a 4.3-fold excess) and Glu-plasminogen (a 9.5-fold excess) were required to give similar inhibition as EDP 11, suggesting that regions within spatial proximity of EDP I1 may hinder the accessibility of antibody to the EDP I1 determinants in the larger derivatives.
Modulation of EDP I1 determinants during the conversion of plasminogen to plasmin and complexation of plasmin with az-antiplasmin was assessed (Fig. 5). The inhibition curves of Glu-plasminogen, Lys-plasminogen, and plasmin were virtually superimposable, indicating that the EDP I1 region was not significantly perturbed during the activation of the zymogen. A 3.3-fold lower concentration of the plasmin.az-antiplasmin complex was required to produce inhibition similar to the free plasmin(0gen) molecules. This is consistent with the interpretation that conformational modulations of the plasmin moiety during complexation with a2-antiplasmin increase the exposure of the EDP I1 region.
The capacity of chemical and enzymatic modification of EDP I1 to modulate its ability to interact with anti-EDP I1 and to bind to lysine-Sepharose was evaluated (Table 111). When EDP I1 was subjected to either 2-mercaptoethanol or iodoacetate, its binding to lysine-Sepharose and anti-EDP I1 was minimally altered. However, when subjected to reduction and alkylation, both activities were completely abolished.
Cyanogen bromide cleavage in 70% formic acid partially diminished the capacity of EDP I1 to bind to lysine-Sepharose and reduced its inhibitory capacity in radioimmunoassay 8fold, whereas exposure of EDP I1 to formic acid alone was without signlficant effect. The loss of lysine-binding activity following reduction and alkylation is consistent with the recent report of Lerch and Rickli (29) although they found that CNBr also completely destroyed this function. As they reported, we also observed cleavage at only one of the two methionyl residues based upon the polyacrylamide gel patterns of reduced and nonreduced samples. Chymotryptic digestion for 1 h at 37 "C completely inhibited the capacity of EDP I1 to bind to anti-EDP I1 and lysine-Sepharose, but tryptic digestion for 18 h at 37 "C did not alter either activity. Digestion of EDP I1 by chymotrypsin to derivatives of molecular weights of less than 4000 was verified by polyacrylamide gel electrophoresis in the presence of SDS and urea, whereas EDP I1 incubated with trypsin for 18 h was unchanged. (The resistance of EDP I1 to tryptic degradation was verified with four different trypsin preparations.) Thus, in all cases, modifications which reduced the capacity of EDP I1 to bind to lysine-Sepharose resulted in a parallel reduction in its capacity to be bound by anti-EDP 11.
The lysine binding sites participate in the interaction of plasmin(ogen) with fibrin(ogen) and as-antiplasmin (8)(9)(10)(11). The preceding data suggest that when the lysine binding site of EDP I1 is occupied, neither EDP I1 nor plasminogen will  react with anti-EDP 11. On this basis, the capacity of fibrinogen and az-antiplasmin to inhibit the binding of lZ5I-EDP I1 by anti-EDP I1 was assessed; but, at concentrations as high as M, these molecules were not inhibitory. Interaction of the lysine binding site of EDP I1 with other plasma components was also assessed (Fig. 6). Isolated Glu-plasminogen was diluted in buffer or in plasminogen-depleted plasma. If the lysine binding site in the EDP I1 region of plasminogen is occupied in the plasma milieu, serial dilutions of plasminogen in plasminogen-depleted plasma should be less inhibitory than those in buffer. In contrast, dilutions in plasminogen-depleted plasma were slightly more inhibitory. This may be due to small amounts of residual plasminogen in the plasminogendepleted plasma, or to interactions at other sites of plasminogen which secondarily increase the exposure of the EDP I1 region.

DISCUSSION
The w-aminocarboxylic acids exert profound effects upon the conformation of plasminogen (30-32) and dissociate the noncovalent complexes of plasminogen with fibrinogen (8,9) and a*-antiplasmin (10,11). In addition, the kinetics of plasminogen activation (33) and of the interaction of plasmin with an-antiplasmin (11) are markedly affected by the w-aminocarboxylic acids. Taken together, these data point to an important role of lysine binding sites in regulating the activation and subsequent interactions of plasmin with its primary substrate and inhibitor. The lysine binding sites reside in the heavy chain region of plasminogen (13) and have been further localized in a recent study to specific enzymatic degradation products (7,15). EDP I1 corresponds to the fourth kringle of plasminogen and contains a single low a f f i t y lysine binding site (7). In this study, we have characterized an antiserum to EDP I1 which appears to require an unoccupied and functionally intact lysine binding site within EDP I1 for recognition by anti-EDP 11. This conclusion is based on several lines of evidence. First, binding of lZ5I-EDP I1 or '251-Glu-plasminogen by anti-EDP I1 was markedly inhibited by 0.2 M 6-aminohexanoic acid, representing at least a 35-fold reduction in the antigen binding capacity of the antiserum. Second, three waminocarboxylic acids produced concentration-dependent inhibition binding of EDP I1 and Glu-plasminogen by anti-EDP 1 1 , and the relative inhibitory eapacity of lysine < 6-aminohexanoic acid < AMCA is consistent with the apparent affinity of these w-aminocarboxylic acids for the lysine binding sites (5, 6 ) and their potency as antifibrinolytic agents (4). Third, modifications, including reduction and alkylation, cyanogen bromide cleavage, and chymotryptic digestion, which reduced or abolished the capacity of EDP I1 to interact with lysine-Sepharose, markedly reduced the capacity of EDP I1 to be bound by anti-EDP 11. The sensitivity of antibody binding to unfolding of EDP I1 induced by reduction and alkylation suggests that the recognized antigenic determinant(s) is conformationally dependent. Interaction of w-aminocarboxylic acids with the lysine binding site of EDP I1 could directly block the capacity of antibody to bind to the key amino acids comprising this determinant, or could suffkiently perturb the conformation of the antigenic locus so that it could no longer be recognized by antibody.
An argument may be developed to suggest that the concentrations of the w-aminocarboxylic acids required for 50% relative inhibition of the binding of either EDP I1 or Glu-plasminogen by anti-EDP I1 are the values of the dissociation constants of the w-aminocarboxylic acids for the lysine binding site within EDP 11. The reactions occurring in the radioimmunoassay may be described as two equilibria in which LBS is the lysine binding site of EDP 11, I is the w-aminocarboxylic acid and Ab is anti-EDP 11:

K1
KZ LBS + I e LBS.1 (1) LBS + Ab + LBS.Ab (2) The dissociation constant ( K z ) of the antibody-antigen interaction (Reaction 2) can be anticipated to be quite low relative to the dissociation constant of Reaction 1 ( K l ) . Thus, the effect of Reaction 2 on the equilibrium of Reaction I is to reduce the effective concentration of LBS. Kl is defined by the equation: A t the concentration I producing 50% relative inhibition in the radioimmunoassay, [I]50%, the concentration of free LBS is equal to [LBS -11 independent of the effective concentration of LBS available in the reaction; i.e. uncomplexed with antibody. Thus.
By ultrafitration, Markus et al. (5) estimated the average dissociation constant of the low affinity lysine binding sites of Glu-plasminogen for 6-aminohexanoic acid to be 5 mM, and Lerch et al. (7) determined the dissociation constant of 6aminohexanoic acid for isolated EDP I1 to be 0.036 mM by equilibrium dialysis. By the immunochemical analysis in this study, the concentrations of 6-aminohexanoic acid required for 50% inhibition of binding were 1.49 m~ with lZ5I-Gluplasminogen and 0.28 mM with '251-EDP 11. Thus, these values are in reasonable agreement (in one case, 3.3-fold lower and in the other 7.8-fold higher) with those determined by more conventional approaches. Differences in the conditions of analysis, such as pH and ionic strength, could be responsible for the observed differences in values. Our immunochemical data suggest that dissociation constants for lysine relative to 6-aminohexanoic acid are 23.6-and 13.5-fold higher for EDP I1 and Glu-plasminogen, respectively. Markwardt (4) found that AMCA was 10-fold more potent than 6-aminohexanoic acid as an antifibrinolytic agent, and, with both lZ5I-EDP I1 and '251-Glu-plasminogen, an 8.8-fold lower concentration of AMCA was required to produce the same inhibition as 6aminohexanoic acid.
In further immunochemical analyses, it was found that only the plasminogen derivatives containing EDP I1 (the plasmin heavy chain, plasmin, and plasminogen) were reactive with anti-EDP 11, whereas derivatives lacking the EDP I1 region in their structure (EDP I, EDP 111, and the plasmin light chain) were unreactive. Thus, despite the extensive structural homologies between the five plasminogen kringles (15), they were effectively discriminated by the antibody. This implies that the low affiity lysine binding sites are nonidentical and differences in their functional activities may be considered. The conversion of plasminogen to plasmin did not alter the accessibility to the EDP I1 region to anti-EDP 11; however, complex formation between plasmin and a*-antiplasmin increased the exposure of the EDP I1 region. We have previously shown that the conformation of free a*-antiplasmin is modulated by complex formation with plasmin (34), and the present observation suggests that the conformation of plasmin is also altered. At high concentrations M), fibrinogen and a*antiplasmin did not inhibit the binding of EDP I1 by anti-EDP 11. This implies that the affinity of the lysine binding site of EDP I1 for these proteins is less than lo5 M-', and is consistent with kinetic analysis suggesting that the high affinity lysine binding site participates primarily in these intermolecular interactions (11). In addition, the lysine binding site within EDP I1 also appeared to be unoccupied in plasma, suggesting that potential interactions of plasminogen with other plasma proteins such as the histidine-rich glycoprotein (35) are not mediated by this low affinity site.