Isolation and Characterization of a Human Plasma Protein with Affinity for the Lysine Binding Sites in Plasminogen ROLE IN THE REGULATION OF FIBRINOLYSIS AND IDENTIFICATION AS HISTIDINE-RICH GLYCOPROTEIN*

During chromatography of plasminogen- and fibrin- ogen-depleted human plasma on an insolubilized fragment of plasminogen which contains the high affinity lysine-binding site (LBS I-Sepharose), both az-antiplas-min and another protein are bound and subsequently eluted with 10 m 6-aminohexanoic acid. This other protein was purified to homogeneity by DEAE-Sepha- dex chromatography followed by immunoadsorption and obtained with a yield of 2.5 mg/liter of plasma. Alternatively, this protein was partially purified (-80% pure) by chromatography on CM-cellulose followed by chromatography on LBS I-Sepharose and obtained with a yield of 40 mg/liter of plasma. The purified protein was found to interact with the high affinity lysine-binding site in plasmin with an apparent dissociation constant of 0.9 PM, thereby mark-edly reducing the reaction rate between plasmin and az-antiplasmin. The partially purified protein reacted similarly with an apparent dissociation constant of 2 PM. The purified protein caused only a limited increase of the activation rate of plasminogen by urokinase. In a ‘“1-fibrin system, the purified protein retarded fibrinolysis by a mixture of tissue plasminogen activator and plasminogen. This phenomenon may be explained by reduction of the apparent concentration of plasminogen with the high affinity lysine-binding site available for binding to hydrolyzed with 6 in a vacuum at llO°C for 20 h. Analysis was performed using a Locarte amino acid analyzer. NHZ-terminal and COOH-terminal Amino Acid Sequence Deter-mination-NHz-terminal sequence analysis was performed by manual Edman degradation (27). The phenylthiohydantoins were identified by thin layer chromatography (28) and quantitated spectrophotomet- rically (27). To these results, the dansyl method of Gray

has a dissociation constant of 1 to 2 PM, about 50 per cent of the circulating plasminogen in blood is expected to occur reversibly complexed with the protein, thereby reducing the "effective" plasminogen concentration and resulting in an antifibrinolytic effect. The purified protein thus seems to represent a physiological counterpart of antifibrinolytic amino acids such as 6-aminohexanoic acid.
The specific interaction between the lysine-binding sites of plasminogen and a*-antiplasmin has been used for the isolation of the inhibitor by affinity chromatography on the insolubilized proenzyme (8,9). An improved version of this affinity chromatography using an insolubilized fragment of plasminogen containing the high affinity lysine-binding site was recently proposed and found to yield a purified inhibitor in a single-step procedure (16).
Using this method, however, we observed that besides azantiplasmin, another plasma protein was also bound to the insolubilized high affinity lysine-binding site of plasminogen and could be dissociated with 10 mM 6-aminohexanoic acid. This protein was further purified to homogeneity by DEAE-Sephadex chromatography and immunoadsorption. In the present study, this protein will be referred to as "the purified protein." In view of its affinity for the lysine-binding site(s) of plasminogen, it might represent a physiological analogue of 6aminohexanoic acid and therefore, its potential role in fibrinolysis was investigated. Physicochemical characterization revealed that this protein is a single-chain glycoprotein with a molecular weight of -60,000 and an unusually high histidine content. Its amino acid composition was found to be very similar to that of a previously described plasma protein with unknown biological function which was called histidine-rich ' In the present study, az-antiplasmin designates the fast-acting plasmin inhibitor of human plasma, as suggested by the Subgroup on glycoprotein (17,18). The identity of these two proteins was eventually confirmed b y immunological cross-reactivity of an antiserum raised against histidine-rich glycoprotein with the protein purified in the present study.

MATERIALS AND METHODS
Plasminogen and Plasmin-Native plasminogen (NHZ-terminal glutamic acid) was prepared from human plasma by affinity chromatography on lysine-Sepharose (19) followed by gel filtration on Ultrogel AcA 44. Activation to plasmin was performed with streptokinase (1000 IU/mg of plasminogen) in 0.1 M phosphate, pH 7.30, containing 25% glycerol at 0°C. The concentration was determined by active site titration usingp-nitrophenyl-p'-guanidinobenzoate (20); at least 95% activation was obtained. The solution was stored in aliquots at -2OOC until used.
The plasminogen fragment LBS I' (triple-loop structures numbers 1 to 3 in the plasmin A-chain, M, = 35,000) and low molecular weight plasminogen (triple-loop number 5 of the A-chain and the intact Bchain, M, = 39,000) were obtained by elastase digestion of plasminogen followed by gel filtration on Sephadex G-75 and affinity chromatography on lysine-Sepharose essentially as described by Sottrup-Jensen et al. (21). Treatment of the low molecular weight plasminogen in the same way as the intact plasminogen yielded 55% activation to low molecular weight plasmin (low Mr-plasmin), as determined by active site titration with p-nitrophenyl-p'-guanidinobenzoate (20). Glu-plasminogen types I and I1 were obtained by chromatography on lysine-Sepharose, essentially as described by Brockway and CasteIlino (3). az-Antiplasmin-This was purified by affinity chromatography of human plasma on LBS I-Sepharose (16), followed by chromatography on DEAE-Sephadex A-50. For the kinetic experiments, az-antiplasmin was dissolved in 0.1 M sodium phosphate buffer, pH 7.30. Its activity was determined by titration against plasmin of known concentration (11) and amounted to 100%. The solution was stored in aliquots at -20°C until used.
Fibrinogen-Prepared from freshly frozen citrated plasma according to the method of Blomback and Blomback (22), the concentration was determined spectrophotometrically using A ;Fm = 15.1 at 280 nm (22). It was labeled with Nalz5I (IRE, F l e w s , Belgium) using a reduced volume version (23) of the chloramine-T method (24).
Thrombin-Human thrombin was purified essentially as described by Fenton et al. (25). The concentration was determined spectrophotometrically using A ; : m = 18.0 at 280 nm (25). Tissue Plasminogen Actzuator-Prepared from human uterine tissue as described (26), the activator was a kind gift from Dr. D. C. Rijken (Center for Thrombosis and Vascular Research, Leuven). Its activity was expressed in urokinase equivalent units by comparison of its fibrinolytic effect on plasminogen-enriched fibrin films.
Amino A c~d Analysis-The samples were hydrolyzed with 6 M HC1 in a vacuum at llO°C for 20 h. Analysis was performed using a Locarte amino acid analyzer.
NHZ-terminal and COOH-terminal Amino Acid Sequence Determination-NHz-terminal sequence analysis was performed by manual Edman degradation (27). The phenylthiohydantoins were identified by thin layer chromatography (28) and quantitated spectrophotometrically (27). To confirm these results, the dansyl method of Gray and Hartley (29) was used.
The abbreviations used are: LBS I, a fragment of plasminogen isolated after digestion with elastase, consisting of triple-loop structures ("kringles") 1 to 3  COOH-terminal amino acids were determined by digestion with carboxypeptidase Y and amino acid analysis (30). About 30 nmol of protein (determined by reaction with ninhydrin after alkaline hydrolysis) was dissolved in 0.05 M sodium acetate buffer, pH 5.5, to a protein concentration of -4 mg/ml. Digestion was performed at 25°C with an enzyme to substrate ratio of 1:200 (w/w) and sampling over a time period of 20 to 180 min. The sample was heated in a boiling waterbath for 5 min to inactivate the enzyme and diluted with 0.2 M sodium citrate buffer, pH 1.9, for amino acid analysis.
Polyacrylamide Gel Electrophoresis-This was performed in the presence of sodium dodecyl sulfate using 7% polyacrylamide gels as described by Weber and Osbom (31).
Immunochemical Analysis-Double immunodiffusion was performed according to Ouchterlony (32) and electroimmunoassay according to Laurell (33). Rabbit antisera against human az-antiplasmin and against the purified protein were prepared by injection of the purified antigens using the immunization scheme previously described (34). The antiserum against the histidine-rich glycoprotein of human plasma was a kind gift from Dr. Karges (Behringwerke AG, Marburg/ Lahn, Federal Republic of Germany).
For immunoadsorption, the immunospecific immunoglobulins isolated from antisera against az-antiplasmin were coupled to CNBractivated Sepharose 4B. A column (1.6 X 20 cm) with a binding capacity of over 15 mg of az-antiplasmin was used.
Determination of the Dissociation Constant of the Complex between Plasmin and the Purified Protein-This dissociation constant was determined as the protein concentration that caused a decrease of the apparent rate constant ( k~,~~~) of the reaction between plasmin and az-antiplasmin to 50%, as previously described (35).
The purified protein was mixed in different concentrations (0 to 5 p~) with the plasmin substrate S-2251 (final concentration, 0.6 mM) and plasmin was added to a final concentration of 8 nM. After a 30-5 incubation, the change in absorbance at 410 nm was recorded for -60 s, an-antiplasmin was added to a final concentration of 12.5 nM, and the change in absorbance was again recorded. The apparent rate constant, kl,app, was obtained from the residual plasmin activity at different time intervals using the classical second order rate equation. Plasmin activities were obtained from the initial slope of the curve (before 50% of the enzyme was inactivated). The influence of the purified protein on the reaction between plasmin (final concentration, 25 nM) and az-antiplasmin (final concentration, 115 I") in the presence of 1 mM 6-aminohexanoic acid and on the reaction between low M,-plasmin (final concentration, 15 nM) and az-antiplasmin (final concentration, 115 I") was also examined (pseudo-first order conditions) using the same procedure. Under these conditions, the rate of the reaction is decreased by about two orders of magnitude (11,12).
AU kinetic measurements were carried out with a Pye Unicam SP 1800 spectrophotometer in 0.1 M sodium phosphate buffer, pH 7.30, at 25T, using 1.0-ml volumes in cuvettes with a 1-cm path length.
Znfluence of the Purified Protein on the Activation of Glu-Plusminogen by Urokinase-Plasminogen was dissolved in phosphatebuffered saline containing 25% glycerol to a concentration of about 6 p~. Activation in the absence or presence of different concentrations of the purified protein was performed by incubation of the mixture with urokinase (4500 IU/mg of plasminogen) at 25°C. At different time intervals, 10-pl samples were removed from the incubation mixture and plasmin was measured in a total volume of 1 ml of 0.1 M phosphate buffer, pH 7.30, using 0.3 mM S-2251. The experiments with Glu-plasminogen type I and I1 were performed under the same conditions.
Znfluence of the Purified Protein on the Activation of Plasminogen by Tissue Plasminogen Activator in the Presence of Fibrin-These experiments were performed on tissue culture plates (Falcon) as described (36), however, with the following modifications. Into each well of the plate (6-mm inner diameter) 100 pl of 12SI-fibrinogen solution (4 X lo6 cpm/ml corresponding to 180 to 200 pg/ml) was added and the plate was dried overnight at 37°C. Then, 75 111 of phosphate-buffered saline containing 3 ng of puritied thrombin was added and the plates were incubated for 10 to 16 hours. Three subsequent washes with 100 pI of phosphate-buffered saline were performed to remove thrombin. Before use, the plates were washed twice again to remove small amounts of soluble radioactivity.
Plasminogen (400 n~) was premixed with different concentrations of the purified protein (0 to 40 p~) and 100 pl of the mixtures were transferred to each well. Then, 100 111 of human tissue plasminogen activator (with a fibrinolytic activity equivalent to 1 IU of urokinase/ ml) were added. The incubation mixture consisted of a volume of 200 pl in phosphate-buffered saline, pH 7.40, containing 0.02% sodium azide and 2.5 mg of calf skin gelatin/ml. The plates were shaken at 37°C and at fixed time intervals, IO-pl samples were removed and lZ5I counted in a Berthold scintillation counter BF 5300 (Benelux Analytic Instruments, Vilvoorde, Belgium). The cumulative-released radioactivity was plotted against time and the time required to solubilize half of the radioactive fibrin (S,) was determined. Plots of S, versus log plasminogen concentration yielded a straight line in the range of 10 to 200 n M .
The concentration of the purified protein which increased the S , to a value corresponding to 50% of the plasminogen concentration in the absence of the protein was considered to represent the digsociation constant of the complex between plasminogen and the purified protein in the presence of fibrin. Similar experiments in which plasminogen was premixed with 6-aminohexanoic acid were performed to determine the dissociation constant between plasminogen and 6-aminohexanoic acid.
Influence of the Purified Protein on the Binding of Plasminogen to Fibrin-These experiments were performed essentially as described by Rtikoczi et al. (7). Mixtures of 0.5 ml of normal plasma or of plasma which was depleted of the purified protein by chromatography on CM-cellulose as described below, containing a trace amount of '251-labeled Glu-plasminogen (about 6 0 , o O O cpm), were clotted by the addition of 0.4 NIH unit of thrombin. The fibrin threads were wound on a glass rod at 37OC during -30 min and squeezed against the wall of the tube to express as much fluid as possible. The fibrin film was then washed twice by soaking it in 1 ml of 0.075 M NaCl, 0.05 M Tris-HC1 buffer, pH 7.5, for 15 min at room temperature. Bound plasminogen was then eluted with the same buffer containing 20 m~ 6-aminohexanoic acid.
Similar experiments were performed in a purified system. The clotting solution consisted of fibrinogen (final concentration, 6.3 p~) , Glu-plasminogen containing 'l-labeled plasminogen (final concentration, 1.5 PM) or Glu-plasminogen preincubated with the purified protein (final concentration, 1.8 p f ) and 0.4 NIH unit of thrombin in a total volume of 0.5 ml .
The amount of lZ5I-labeled plasminogen in the supernatant after clotting, in the wash buffer, in the eluates, and on the fibrin, was quantitated in a scintillation counter and expressed as a percentage of the original amount added to the plasma.

RESULTS
Isolation of "the Purified Protein"-Purification was performed using fibrinogen and plasminogen-depleted human plasma by affinity chromatography on LBS I-Sepharose and elution with 6-aminohexanoic acid, essentially as described by Wiman (16). Plasminogen was removed by chromatography on lysine-Sepharose and fibrinogen by precipitation with 10% ethanol at -4°C. The materials obtained from 6 and 10 batches of 4 liters of human plasma each (200 to 300 mg of protein) were dialyzed against 0.05 M phosphate buffer, pH 7.0, and applied to a DEAE-Sephadex A-50 column (5 X 20 cm) at 4°C with a flow rate of 30 ml/h. Elution was performed with a linear gradient of 5 column volumes of 0.05 M phosphate, pH 7.0, to 5 volumes of the same buffer containing 0.4 M NaCl at 60 ml/h. The elution pattern (Fig. lA) shows 2 major protein peaks.
Total protein recovery from the ion exchange column was -70%. Peak I1 contains the pure and fuUy active az-antiplasmin as shown by Laurell electrophoresis, titration against plasmin of known concentration, amino acid analysis, and NHz-terminal sequence analysis. The same techniques indicated that Peak I contained an inactive form of a2-antiplasmin together with another component. The inactive a2-antiplasmin present in Peak I was removed by immunoadsorption. In one typical experiment, the protein (37 absorbance units at 280 nm) was applied to the antibody column (1.6 X 20 cm) in 0.01 M phosphate buffer, pH 7.40, containing 0.14 M NaCl and 0.04% azide at 4°C and at a flow rate of 20 ml/h. Elution of ae-antiplasmin was performed with 0.1 M glycine-HC1, pH 2.80 (Fig. 1B). The fitrate through this column (Peak IA) contains the purified protein (23.5 absorbance units), while the eluate (peak IB) contains inactive az-antiplasmin (7.5 absorbance units). Both were desalted by gel fitration on Sephadex G-25 in water. The protein recovery in this step was 84%. Assuming a plasma concentration for the purified protein of 100 mg/liter (see below), the total yield in this procedure is only 2%. SDSgel electrophoresis revealed a single band with an estimated molecular weight of -60,000.
The complete removal of 6-aminohexanoic acid during this purification procedure was demonstrated by amino acid analysis. Therefore, we added 6-aminohexanoic acid (10% by weight) to our purified protein (1 mg/ml), the protein was precipitated with 20% CCLCOOH, and a sample of the supernatant (containing about 70 nmol of 6-aminohexanoic acid) was applied to an amino acid analyzer. 6-Aminohexanoic acid was eluted in a position between histidine and arginine with a recovery of 50%. A control sample of the purified protein alone treated in the same way did not reveal any traces of 6aminohexanoic acid. The lower detection limit of our analyzer is about 0.2 nmol of amino acid, corresponding to less than 0.05% of 6-aminohexanoic acid in the purified protein.
Determination of the Dissociation Constant of the Complex between Plasmin and the Purified Protein-The influence of the purified protein on the apparent rate constant of the rapid reversible complex formation, kl,app, between plasmin and a2antiplasmin is illustrated in Fig. 2. The purified protein induces a concentration-dependent reduction of this reaction rate. The shape of this sigmoidal curve is compatible with a single association reaction between plasmin and the purified protein. The dissociation constant of this interaction, determined as the concentration of purified protein which decreases the rate of the reaction between plasmin and an-antiplasmin to 50% of its normal value, is 0.9 p~. Control experiments showed that addition of the purified protein to plasmin did not influence the rate of hydrolysis of S-2251.
Additional experiments showed no influence of the purified protein on the rate of the reaction between plasmin and a2antiplasmin in the presence of 1 6-aminohexanoic acid (which saturates the high affinity lysine-binding site), or on the rate of the reaction between low M,-plasmin (which lacks the lysine-binding sites) and az-antiplasmin (Fig. 2). A l l these findings indicate that the purified protein competes with atantiplasmin for the high affinity lysine-binding site in plasmin. Albumin did not interfere with the plasmin-a2-antiplasmin reaction in concentrations up to 0.1 mM.
Influence of the Purified Protein on the Activation of Glu-Plasminogen by Urokinase- Fig. 3 shows that the activation of Glu-plasminogen by urokinase is to some extent influenced by the presence of the purified protein. At a final concentration of 10 p~ purified protein, a significant enhancement of the activation rate is observed. This influence increases to a final concentration of -100 p~; a further increase to 400 pM only yields a minor additional change in the activation rate. The influence of the purified protein is not as pronounced as that obtained with 1 mM 6-aminohexanoic acid. A control experiment with 500 p~ albumin did not show any increase in the activation rate of Glu-plasminogen.
The activation of Glu-plasminogen types I and 11 by urokinase was studied under the same experimental condi:ions in the absence and the presence of 50 p~ purified protei.1. The activation rate of type I1 Glu-plasminogen was shown to be slightly slower than that of type I, but the effect of the purified protein was very similar to that observed with native Gluplasminogen.
Influence of the Purified Protein on the Activation of Plasminogen by Tissue Activator in the Presence of Fibrin-Using mixtures of tissue plasminogen activator (final concentration was equivalent to 0.5 IU of urokinase) and different

ELUTION VOLUME (ml)
. -1 2 plasminogen concentrations (0 to 500 nM), the rate of '"1 release from '251-fibrin plates increased with increasing plasminogen concentration (Fig. 4A). By plotting the time required to solubilize half of the radioactive fibrin (S,) versus the logarithm of the plasminogen concentration, a straight line is obtained in the range of 10 to 200 n~ (Fig. 4B).
Using this system, we studied the influence of 6-aminohexanoic acid (0 to 40 PM) and the purified protein (0 to 20 PM) on the digestion of '"1-fibrin by mixtures of human plasminogen (200 m) and tissue plasminogen activator. The rate of 1251 release as a function of the concentration of 6-aminohexanoic acid or of the purified protein is shown in Fig. 5, A and  B, respectively. The displacement in the curves resulting in higher Sm values in the presence of 6-aminohexanoic acid or the purified protein is compatible with the interpretation that complex formation between plasminogen and 6-aminohexa- noic acid or purified protein prevents binding of plasminogen to fibrin and subsequent activation by tissue plasminogen activator. This effect is already clearly observed at physiological concentrations of the purified protein (-1.8 p~) .
Using the calibration curve shown in Fig. 4B, the apparent plasminogen concentration corresponding to the Ss values obtained for each ligand concentration was determined. By plotting this apparent plasminogen concentration, in the percentage of the control value in the absence of ligand, against the logarithm of the concentration of ligand (Fig. 5C), a sigmoidal curve suggestive of a single association reaction is obtained. The concentration of 6-aminohexanoic acid or purified protein which reduces the apparent plasminogen concentration to 50% was determined. Since this reduction occurs at ligand concentrations which are at least an order of magnitude larger than the plasminogen concentration, these values would represent the dissociation constant of the complexes between plasminogen and 6-aminohexanoic acid (14 p~) and between plasminogen and purified protein (1.1 PM).
Influence of the Purified Protein on the Binding of Plasminogen to Fibrin- Table I summarizes the results obtained for binding of Glu-plasminogen to fibrin in a purified system, in normal plasma, and in plasma depleted in the purified protein as described below. The depleted plasma was obtained by adsorbing plasma twice with 50 g of CM-cellulose/liter at I51 A ,OOt 50 100 150 pH 6.0 after 2-fold dilution with distilled water. The plasma was then reconcentrated to its original volume by ultrafiltration and the pH readjusted to 7.4. The concentration of the purified protein, measured by electroimmunoassay, was less than 10% of the original concentration. The fibrinogen concentration in the depleted plasma was reconstituted to its original value of 2.2 mg/ml by addition of 0.5 mg of purified fibrinogen/ml. The plasminogen concentration was unaltered by the adsorption with CM-cellulose. In a purified system, the presence of the purified protein decreases the amount of Glu-plasminogen that specifically (eluted with 20 m~ 6-aminohexanoic acid) binds to fibrin from 9.8 to 6.1%. In the purified protein-depleted plasma, 3.6% of Glu-plasminogen binds specifically to the fibrin clot compared to 2% in normal plasma. These findings support the concept that the purified protein reduces the binding of plasminogen to fibrin by complex formation with the lysine-binding site(s) of the proenzyme. The apparent dissociation constant of this interaction would be -1 PM.   ' The data represent mean f S.D.
This amino acid composition is very similar to that of a previously described plasma protein with unknown biological function, called histidine-rich glycoprotein (17,18). In immunodiffusion, the purified protein reacted with the antiserum against histidine-rich glycoprotein, obtained from Behringwerke, but not with antiserum against an-antiplasmin (Fig.  6A).
In Laurell immunoelectrophoresis, the antiserum against histidine-rich glycoprotein reacted with the purified protein and with a single component in human plasma (Fig. 6 B ) . The plasma concentration, determined by electroimmunoassay, was about 100 mg/liter which is in accordance with the findings of Heimburger et al. (18).
Edman degradation revealed the NHn-terminal sequence Val-Ser-Pro-with a recovery of 0.78 mol of Val/mol of protein.
The sequence Val-Ser-was confirmed using the dansyl method. Digestion with carboxypeptidase Y revealed -Phe-Leu a s COOH-terminal sequence (Table 111).

Partial Purification of the Histidine-rich Glycoprotein from Human Plasma and its Influence on the Reaction between Plasmin and an-Antiplasmin-One liter of human
plasma was diluted 2-fold with distilled water, the pH was adjusted to 6.0, and the mixture was stirred with 50 g of CMcellulose 52 for 1 h at room temperature (17). The ion exchanger was removed, washed with water, and transferred to a column (2.5 X 20 cm). Elution with a linear gradient of 4 column volumes of water to 4 volumes of 0.5 M NH4HCOs yielded one major protein peak (370 absorbance units at 280 nm). This pool was devoid of an-antiplasmin, plasminogen, and fibrinogen. It was dialyzed against 0.05 M phosphate buffer, pH 7.0, and applied to a column (2.5 X 40 cm) of LBS I-Sepharose (300 mg of insolubilized LBS I). The major part of the protein (228 absorbance units at 280 nm) was recovered by washing with 2 column volumes of 0.05 M phosphate buffer, pH 7.0. Elution with 10 m~ 6-aminohexanoic acid yielded a single protein peak (35 absorbance units at 280 nm) containing 50% of the amount of histidine-rich glycoprotein applied to the column. After extensive dialysis and desalting on Sephadex G-25 to remove 6-aminohexanoic acid, -50 mg of protein was obtained. SDS-gel electrophoresis revealed two protein bands (Fig. 7A ). Analysis of the material by electroimmunoassay revealed that the histidine-rich glycoprotein accounted for about 80% of the total protein (Fig. 7B). Thus, the final recovery of histidine-rich glycoprotein in the partially purified preparation was 40%.
This partially purified protein decreased the rate of the reaction between plasmin and an-antiplasmin in a similar way  as the highly purified protein used in our previous experiments. The apparent dissociation constant of the interaction between the histidine-rich glycoprotein and plasmin was found to be -2 PM, assuming a purity of 80%. This value is comparable to that of 0.9 PM obtained for the highly purified protein. Control experiments showed no influence of the partially purified protein on the activity of an-antiplasmin or on the stability of plasmin.

DISCUSSION
During purification of an-antiplasmin from human plasma by affinity chromatography on an insolubilized fragment of plasminogen which contains the high affinity lysine-binding site (LBS I), another protein is also bound and elutes with 10 m~ 6-aminohexanoic acid. This protein was further purified to homogeneity by DEAE-Sephadex chromatography and immunoadsorption and was obtained in a yield of approximately 2.5 mg/liter of plasma.
The purified protein apparently interacts with one or more of the lysine-binding sites in plasminogen and in that respect could have similar properties to 6-aminohexanoic acid and related antifibrinolytic amino acids. Three properties of 6aminohexanoic acid have been clearly delineated. The first property is that in purified systems, saturation of the high affinity lysine-binding site in plasmin slows down the rate of its interaction with as-antiplasmin -50 times (11). In this respect, the purified protein resembles 6-aminohexanoic acid. It interacts with plasmin by an apparently single association reaction with a dissociation constant of 0.9 p~. That this interaction is indeed mediated through the lysine-binding site is substantiated by the finding that no effect was observed on the reaction between as-antiplasmin and low molecular weight plasmin which lacks the lysine-binding sites (12,21), or on the reaction between plasmin and a2-antiplasmin in the presence of 1 m~ 6-aminohexanoic acid, which saturates the high affinity lysine-binding site (11). The dissociation constant of the complex between Glu-plasminogen and as-antiplasmin is 4 p~ (35) and between plasmin and as-antiplasmin (with an additional interaction through the active site) is 0.2 nM (11).
From these data, we have previously calculated that at the concentrations of plasminogen (1.5 to 2 p~) and as-antiplasmin (total, 1 PM, of which -% with affinity for the lysine-binding sites in plasminogen (37)) which occur in plasma, about 30% of the az-antiplasmin would occur reversibly complexed with plasminogen. From the dissociation constant between plasminogen and the purified protein (1.1 p~) , which is in good agreement with that between plasmin and the purified protein (0.9 p~) we can estimate that at the concentrations of plasminogen (1.5 to 2 p~) and the purified protein (1.8 p~) which occur in plasma, about 50% of the proenzyme would circulate in association with this protein. This interaction between the purified protein and plasminogen would reduce the concentration of free plasminogen in blood and reduce the extent of complex formation between plasminogen and an-antiplasmin by a factor of 2, thereby increasing the concentration of free inhibitor by about 15%.
Another property of antifibrinolytic amino acids is that they induce a conformational change in the plasminogen molecule which results in an enhanced activation rate by urokinase (5, 38-39). It was recently shown that at concentrations which saturate the low affinity lysine-binding sites, a conformational change occurs (40). The purified protein in concentrations above 10 p~ and up to 500 PM had only a limited accelerating effect on the activation rate of plasminogen. This indicates either that the purified protein reacts only very weakly with the low affinity lysine-binding sites or that such interaction only leads to limited acceleration of the activation rate, possibly as a result of steric hindrance. From these findings, we extrapolate that the interaction between plasminogen and the purified protein in plasma would not lead to enhanced activation of plasminogen.
A third well delineated property of 6-aminohexanoic acid and its analogs is that they interfere with the binding of plasminogen to fibrin (6, 7). This interaction most probably constitutes the molecular basis of the antifibrinolytic effect of these compounds (13). In a fibrinolytic assay mixture cornposed of '251-labeled fibrin, plasminogen and tissue plasminogen activator, 6-aminohexanoic acid induced a concentrationdependent prolongation of the time required to solubilize half of the fibrin. This effect is probably due to complex formation between the high affinity lysine-binding site and the ligand, resulting in abolishment of the interaction with fibrin. Indeed, an apparent reduction of the free plasminogen concentration to 50% was obtained at a concentration of 14 p~ which is in good accordance with the previously determined dissociation constant of 9 p~ (14). In this assay, the purified protein also induced a concentration-dependent decrease of the apparent free plasminogen concentration compatible with a single association reaction with an apparent dissociation constant of 1.1 PM. This value is in good agreement with that obtained for the interaction between plasmin and the purified protein.
Thus, at the concentration of purified protein occurring in the blood, the effective plasminogen concentration would be reduced by 50%. It should be stressed that the measurements were performed with a plasminogen concentration of 200 nM which is -10 times lower than that occurring in plasma. At such a high plasminogen concentration, the effect of 6-aminohexanoic acid or the purified protein would not be observed in our system. Still, there is ample clinical evidence that 6aminohexanoic acid has antifibrinolytic properties and from the analogy in the results obtained with the purified protein, we extrapolate that it has a similar antifibrinolytic effect. This is further substantiated by the finding that the purified protein at a concentration of 1.8 p~ reduces the binding of plasminogen to a fibrin clot by -50%, as well in a purified system as in plasma. The observed phenomena cannot be explained by contamination of our purified protein with 6-aminohexanoic acid, since this would require a contamination exceeding 10% by weight. By amino acid analysis, we could not detect any 6aminohexanoic acid in our purified protein (detection limit lower than 0.05% by weight).
In summary, our findings suggest that the purified protein might play a role in the regulation of fibrinolysis, mainly by interference with the binding of plasminogen to fibrin. This would result in retardation of fibrinolysis. The limited enhancing effect on the activation of plasminogen at very high concentrations is probably physiologically irrelevant. Physicochemical analysis revealed that the purified protein is a single-chain glycoprotein with a molecular weight of -6O,OOO, with NHs-terminal sequence Val-Ser-Pro-and COOH-terminal sequence -Phe-Leu. Both the amino acid composition and immunochemical analysis indicated that it is identical with a previously described protein with unknown biological function called histidine-rich glycoprotein (17, 18).
The concentration of the purified protein (histidine-rich glycoprotein) in plasma is -100 mg/liter and the recovery in our first purification is only -2%. The alternative partial purification procedure, also using LBS I-Sepharose, but in the absence of as-antiplasmin and using less extensive washing, yielded the (partially purified) protein with the same properties in a yield of -40%. The low yield of the purified protein in our isolation procedure is, therefore, most likely not due to selection of a specific molecular form of this protein which would represent only a small fraction of the total. This indicates that both as-antiplasmin and the histidine-rich glycoprotein probably have a comparable affinity for LBS I-Sepharose. The greatly differing recovery of a2-antiplasmin (-25%) and the purified protein (-2%) in our first purification procedure could thus result from different rate constants in the dissociation reaction from the ligand.