Proteoglycan Carrier of Human Platelet Factor 4 ISOLATION AND CHARACTERIZATION*

A large scale purification procedure for the human platelet factor 4 proteoglycan carrier molecule has been developed. A yield of 46% and a 33,000-fold puri- fication have been achieved, using poly-L-lysine-Seph-arose affinity column chromatography, PF4-agarose affinity column chromatography, and Bio-Gel A-0.5m gel filtration. The purified proteoglycan migrates as a single band during electrophoresis on cellulose acetate strips. A single symmetric peak was observed in sedi- mentation velocity analysis with an s value of 2.85. The molecular weight of the proteoglycan was determined to be 53,000 by sedimentation equilibrium. The purified proteoglycan contains 32% uronic acid, 31% galactosa- mine, 6.1% sulfate, and 9.9% protein. Aspartic acid, glutamic acid, leucine, glycine, and serine account for 55% of the total amino acids. The chondroitinase AC digest of the proteoglycan is sensitive to hydrolysis by chondro-4-sulfatase but not by chondro-6-sulfatase, in- dicating the presence of chondroitin 4-sulfate but not chondroitin 6-sulfate in the proteoglycan molecule. The interaction between this proteoglycan carrier of human PF4 and PF4 is strongly ionic strength-dependent. 0.3 M NaCl is required to dissociate the proteoglycan PF4 complex.

A large scale purification procedure for the human platelet factor 4 proteoglycan carrier molecule has been developed. A yield of 46% and a 33,000-fold purification have been achieved, using poly-L-lysine-Sepharose affinity column chromatography, PF4-agarose affinity column chromatography, and Bio-Gel A-0.5m gel filtration. The purified proteoglycan migrates as a single band during electrophoresis on cellulose acetate strips. A single symmetric peak was observed in sedimentation velocity analysis with an s value of 2.85. The molecular weight of the proteoglycan was determined to be 53,000 by sedimentation equilibrium. The purified proteoglycan contains 32% uronic acid, 31% galactosamine, 6.1% sulfate, and 9.9% protein. Aspartic acid, glutamic acid, leucine, glycine, and serine account for 55% of the total amino acids. The chondroitinase AC digest of the proteoglycan is sensitive to hydrolysis by chondro-4-sulfatase but not by chondro-6-sulfatase, indicating the presence of chondroitin 4-sulfate but not chondroitin 6-sulfate in the proteoglycan molecule. The interaction between this proteoglycan carrier of human PF4 and PF4 is strongly ionic strength-dependent. 0.3 M NaCl is required to dissociate the proteoglycan PF4 complex.
Platelet factor 4 is released from human platelets during normal blood coagulation or when platelets are exposed to damaged blood vessel walls (1-7). Early in coagulation, there also occurs a rise in the glycosaminoglycan content in human serum (8). Platelets exposed to thrombin release a specific glycosaminoglycan (9); subsequently, platelets were shown to contain at least two distinct glycosaminoglycans (10). Barber et al. (11) showed that platelet factor 4 is released from platelets as a high molecular weight, proteoglycan-platelet factor 4 complex. The complex was dissociated by elution in high salt and the proteoglycan was purified and partidy characterized.
We recently have purified platelet factor 4 and determined the complete amino acid sequence of this protein (12). The COOH-terminal is unusual and seemed a likely binding site for glycosaminoglycans. The COOH-terminal region of PF4' * This research was supported by Grants CA22409, HL14147, and HL22119 from the United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. + Recipient of Grant 5-T32-HL07081 from the United States Public Health Service and the American Heart Association, Missouri Affdiate, Inc.
1 The abbreviations used are: PF4, platelet factor 4; SDS, sodium dodecyl sulfate. contains 4 lysine residues, occurring in pairs (-Lys-Lys-Ile-Ile-Lys-Lys-Leu-Leu-Glu-Ser-COOH). We further observed that the COOH-terminal tryptic peptide in high concentration partidy reversed the prolonged thrombin time induced by heparin, suggesting heparin binding was localized to the COOH-terminal region of PF4. Support for the involvement of these lysines in the PF4-heparin binding activity was provided by the work of Handin and Cohen (13) who showed that guanidination of lysines in PF4 decreased heparin neutralizing activity, but that modification of arginines was without effect.
In an effort to investigate further the interactions of PF4 with glycosaminoglycans, we have purified the PF4-proteoglycan carrier molecule using affinity chromatography. This manuscript describes a novel procedure for the purification of the carrier molecule by PF4-agarose affinity chromatography, its characterization by physical and chemical criteria, and the interaction of the proteoglycan carrier molecule and PF4.

Methods
Human platelet factor 4 was purified by previously described methods (12), including heparin-agarose chromatography and gel filtration with Sephadex G-100. The purified PF4 appeared as a single band in SDS-polyacrylamide gel electrophoresis as reported previously ( 12).
Preparation of Poly-L-lysine-Sepharose 4B-One g of poly-L-lysine was dissolved in 100 ml of 0.1 M NaHCO3 (pH 8.0) and added to 900 g (wet weight) of cyanogen bromide-activated Sepharose 4B in 900 ml of 0.1 M NaHC03 (pH 8.0). After mixing at 4 "C overnight, the gel was sequentially washed with 2 liters of 0.1 M NaHC03 (pH 8.0), 2 liters of 1.0 M NaC1, and 4 liters of distdled water.
Preparation of PF4-~Aminobutyl Agarose-Ten mmol of 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide was added to a gel suspension of w-aminobutyl-agarose (50 g wet weight of gel in 50 d of water, pH 4.8). Twenty mg of PF4 in 4 ml of 1 M NaCl was added drop by drop. The gel suspension was stirred for 4 h and the pH was maintained at 4.8, incubated at room temperature overnight, and washed with 2 liters of 1 M NaCl and extensively with water. More than 95% of the PF4 added was coupled under these conditions. Cellulose Acetate Electrophoresis-Proteoglycan samples (1 mg/T ml) were applied to cellulose acetate plates for electrophoresis with zinc acetate buffer (pH 3.5) at a constant current 0.5 mA/cm for 2 h (11) or with Tris:barbital:sodium barbital (pH 8.8) for electrophoresis of PF4 and PF4-proteoglycan complex at 160 V for 15 min. Duplicate plates were analyzed. The proteoglycan was stained with 0.05% alcian blue in 1% acetic acid and proteins were stained with ponceau S staining solution and destained in 1% acetic acid. SDS-polyacrylamide gel electrophoresis was performed according to the method of Laemmli (14) using 5 8 gels, and the gels were stained with Coomassie brilliant R-250 for protein or with alcian blue for proteoglycans.
Chondroitinase AC Digestion of Proteoglycan-Proteoglycan (0.6 mg) in 1 ml of 0.1 M Tris:sodium acetate buffer (pH 7.3) was digested with chondroitinase AC (0.25 unit/mg) at 37 "C for 3 h. Aliquots were spotted on cellulose thin layer plates (IO X 3 cm) and developed (ascending chromatography) using butanolacetic acid:l M ammonia, 2:3:1 (v/v/v), for 30 min. The disaccharide product was visualized using a short wavelength mineral UV light (15) or stained with alkaline silver nitrate reagent (16). One mg of chondroitin sulfate types A and C was digested by chondroitinase AC as a control.
Chondroitin Sulfatase Digestion-The chondroitinase AC digests of proteoglycan and chondroitin sulfate types A and C were further digested with chondro-+sulfatase or chondro-6-sulfatase at 37 "C for 30 min (15). An aliquot of the digest was spotted on cellulose thin layer plates (Kodak) and subjected to electrophoresis with 0.05 M sodium citrate:citric acid buffer (pH 5.0) a t 45 V/cm for 30 min. The digested product was visualized under a short UV light.
Analytical Methods-Protein was measured by the method of Lowry et al. (17). Galactosamine was determined in a Beckman amino acid analyzer after hydrolysis in 4 M HCl under vacuum at 100 "C for 8 h. Hexuronic acid was determined by the modified carbazole reaction (18) using D-glUCUrOniC acid as a standard. Total sulfate was determined according to the method described by Dodgson and Price (19).
Amino Acid Analysis-The amino acid analyses were performed with a Beckman amino acid analyzer according to the method of Spackman et al. (20). The proteoglycan sample was hydrolyzed with 5.7 N HCI for 24, 48, and 72 h at 108 "C in evacuated sealed tubes. Half-cystine residues were determined as cysteic acid from separate analyses after performic acid oxidation and acid hydrolysis (21).
Analytical Ultracentrifugation-Sedimentation equilibrium studies were performed using a Beckman model E ultracentrifuge at a rotor temperature of 12 O C and a speed of 12,500 rpm. A value of 0.54 cm/g was assumed for the partial specifk volume of proteoglycan (22). The proteoglycan concentration was 0.89 mg/ml in 1 M NaC1. Sedimentation velocity studies were performed at a rotor temperature of 18.5 "C and a speed of 52,000 rpm. The proteoglycan concentration was 4 mg/ml, in 1 M NaCI.

RESULTS
The large scale purification of the proteoglycan carrier of PF4 involves two affinity columns, poly-L-lysine-Sepharose 4B and PF4-o-aminobutyl agarose, and gel filtration on Bio-Gel A-0.5m. A purification scheme was summarized in Table   I. About 100 units (volume -4000 ml) of outdated, frozen human platelet-rich plasma were thawed and centrifuged at 13,000 X g for 30 min to remove cell debris and other insoluble materials. Protease inhibitors including 1 mM diisopropyl fluorophosphate, 0.05 rrm phenylmethylsulfonyl fluoride, 10 mM benzamidine-HC1, 0.03 mM ~-tosylamido-2-phenylethyl chloromethyl ketone, 10 rn EDTA, and 10 mg/liter of soybean trypsin inhibitor were added during the thawing procedure. The clear plasma solution was then mixed at a ratio of 20 g (wet weight) of poly-L-lysine-Sepharose 4B gel per 500 ml of plasma solution. After mixing overnight at 4 "C, the gel SUSpension was packed onto a column (5 X 45 cm). The column was extensively washed with 0.15 M NaCl (absorbance at 280 nm < 0.1). The proteoglycan carrier eluted at an ionic strength of -0.45 M NaCl in a linear gradient from 0.15 M NaCl to 1 M NaCl (Fig. 1). About a 560-fold purification with 85% yield was achieved. The starting large volume of plasma (4000 d) was reduced to 100 ml of column eluent. The proteoglycancontaining fractions were collected, dialyzed against 0.15 M NaCl, and applied onto a column of PF4-w-aminobutyl-aga- Since human platelet-rich plasma contains a substantial quantity of proteins/glycoproteins which produce an intense brown color during sulfuric acid treatment in the carbazole reaction, interfering with the assay of uronic acid, it was not possible to measure directly the PF4 proteoglycan carrier-specific uronic acid content in human platelet-rich plasma. To estimate the concentration of the PF4 proteoglycan carrier-specific uronic acid, an internal standard of the purified PF4 proteoglycan carrier was added to the plasma for measuring the recovery after poly-L-lysine column chromatography. Based on the 85% recovery from the poly-L-lysine column, the concentration of PF4-proteoglycan carrier-specific uronic acid in human platelet-rich plasma was calculated at 117.2 pmo1/4,000 ml of plasma. rose. The proteoglycan elutes at -0.3 M NaC1; most contaminating proteins do not bind to PF4-w-aminobutyl-agarose gel (Fig. 2). PF4 coupled to Sepharose activated with cyanogen bromide was not effective in binding the proteoglycan to the column, suggesting that the lysine groups of PF4 appear essential for the interaction of PF4 and the proteoglycan. This PF4-a-aminobutyl-agarose affinity column results in a further 96-fold purification, with 83% yield. The proteoglycan preparation obtained from the PF4-w-aminobutyl-agarose column was further purified by gel filtration on Bio-Gel A-0.5m. The proteoglycan was recovered in the fractions close to the void volume, whereas contaminating proteins appeared in later fractions (Fig. 3). About 38 mg of proteoglycan is readily

FRACTION NUMBER
FIG. 2. Affinity column chromatography of PF4-w-aminobutyl-agarose. The proteoglycan-containing fractions obtained from poly-L-lysine-Sepharose column were collected, dialyzed against 0.15 M NaCI, and then applied to the PF4-w-aminobutyl-agarose column (volume, 50 ml) which was previously equilibrated with 0.15 M NaC1. The proteoglycan was eluted with a linear gradient (300 ml) of 0.15 M to 1 M NaCI. The fraction volumes were 2.6 ml. The proteoglycan was pooled as indicated, dialyzed against water, and lyophilized.

FRACTION NUMBER
FIG. 3. Gel filtration on Bio-Gel A-0.5m of the proteoglycan from PF4-w-aminobutyl-agarose column chromatography. The lyophilized proteoglycan from I'F4-w-aminobutyl-agarose column was dissolved in 5 ml in 30 mM Tris-HCI, pH 7.4, containing 0.12 M NaCl and applied to a column of Bio-Gel A-0.5m (2.5 X 90 cm) previously equilibrated with the same buffer. Fractions of 5.8 ml were collected and analyzed for A?,, ( M ) and uronic acid content was measured a t As:,,, "", (M) by carbazole reaction.
obtained from 4 liters of platelet-rich plasma with a recovery of 46%. The purity of the purified proteoglycan has been assessed by cellulose acetate electrophoresis, SDS-gel electrophoresis, and ultracentrifugation. The proteoglycan appears as a single positively alcian blue-stained band after electrophoresis on cellulose acetate. No protein bands were observed when the strip was stained for protein. SDS-polyacrylamide gel electrophoresis was also used to analyze the purity of the proteoglycan. One hundred pg of the proteoglycan was subjected to SDS-polyacrylamide gel electrophoresis. No contaminating protein bands were detected by Coomassie blue staining although less than 1 pg of protein is detectable in our gel system. Alcian blue staining demonstrated a single stained band of the proteoglycan at the junction of stacking gel and separating gels. It is likely that aggregation of the proteoglycan accounted for the failure of the material to enter the 5% gel.
Further analyses of purity was attempted using sedimentation velocity. The proteoglycan carrier molecule sediments as a single sharp symmetrical peak in the analytical ultracentrifuge (Fig. 4) and has an uncorrected sedimentation coefficient cf 2.85 S. No evidence of impurities was found. The molecular weight of the proteoglycan was measured by sedimentation equilibrium analysis. The calculated molecular weight in this experiment is 53,000. High concentrations of salt were needed to prevent self-association during centrifugation; 1 M NaCl was used in all centrifugation analyses. Gel filtration (Sepharose CL-2B, 90 X 0.9 cm, 1 M NaCI) analysis was done; the platelet factor 4 proteoglycan carrier molecule eluted with a K., of 0.76. By comparison with the Kz,v obtained with three other reference proteoglycans used by Silvestri et al. (23), the platelet factor 4 proteoglycan carrier had a M , -56,000 estimated in this analysis.
Chemical analyses were also done. Two mg of the purified proteoglycan were subjected to NH2-terminal amino acid analysis using Edman degradation in a Beckman Sequencer. No phenylthiohydantoin-amino acid residues were released after Edman degradation of the purified proteoglycan, consistent with previous results indicating no significant contamination by other peptides or proteins. It is likely that the high carbohydrate content of the proteoglycan molecule hinders detection by Edman degradation of NHp-terminal amino acid residues in the peptide moiety.
The chemical composition of the proteoglycan is shown in Table 11. The proteoglycan contains 10% protein, 32% uronic acid, 31% galactosamine, and 6.1% sulfate. No glucosamine was found in the proteoglycan molecule. The amino acid composition of the peptide moiety in the proteoglycan is shown in Table 111. Since the proteoglycan contains -10% protein, the molecular weight of the peptide moiety was calculated as -5300. A total of 43 amino acid residues were present in 1 mol of the proteoglycan. Aspartic acid, glutamic acid, leucine, glycine, and serine account for -55% of the total amino acid content of the peptide portion of the PF4-proteoglycan carrier molecule. No methionine residues were found. The proteoglycan shows the characteristics of chondroitin sulfate containing glucuronic acid, galactosamine, and sulfate.
The identity of chondroitin 4-sulfate or 6-sulfate in the proteoglycan was established by a combination of digestion with chondroitinase AC and chondro-4-or -6-sulfatase (15). L \ y w . " I . u . , L, -" FIG. 4. Sedimentation velocity pattern of proteoglycan at 4 mg/ml in 1 M NaCl. The direction of sedimentation is from left to right. Centrifugation was performed at 18.5 "C and 52,000 rpm. The picture was taken 30 min after reaching full speed.

TABLE I1
Chemical composition of proteoglycan from platelet-rich plasma The chondroitinase AC digest of proteoglycan yielded one product, with the same mobility as that of the standard chondroitin 4-sulfate treated with chondroitinase on paper chromatography (data not shown). For confirmation of the chondroitin 4-sulfate configuration, the chondroitinase AC digest was further hydrolyzed with chondro-4-or -6-sulfatase (Fig. 5). A marked decrease in electrophoretic mobility of the product was found when the product was digested with chon- "The data were the average values from the analyses of three * The data shown are the nearest integers.
'The values of serine and threonine residues were corrected for separate samples. dro-4-sulfatase but not with chondro-&sulfatase, It is concluded therefore that the glycosaminoglycan chains of proteoglycan carrier consist of chondroitin 4-sulfate. The data (Fig. 5) also provide evidence that the isolated chains are homogeneous and, thus, additional evidence that the intact isolated proteoglycan has been purified to homogeneity as well.
The use of PF4-o-aminobutyl-agarose to purify proteoglycan from platelet-rich plasma clearly demonstrates the affinity of the proteoglycan for PF4. The dissociation of proteoglycan molecule from PF4 affinity gel at high ionic strength (-0.3 M NaC1) suggested that the interaction between PF4 and the proteoglycan molecule is mainly through ionic interactions. The complex formation between PF4 and proteoglycan was further confirmed by cellulose acetate electrophoresis studies. Under the conditions shown (Fig. 6), PF4 remained at the origin and showed a positive reaction to ponceau S staining while proteoglycan migrated rapidly toward the anode and showed a positive reaction with alcian blue staining. The proteoglycan-PF4 complex migrated more slowly toward anode than the proteoglycan alone and showed a positive reaction to both protein and proteoglycan staining. The data clearly demonstrated that a complex of PF4 and proteoglycan was formed which has different electrophoretic mobility from PF4 and the proteoglycan alone. The binding of t.he proteoglycan to PF4 has not been studied in further detail.
We have used purified human platelet-derived growth factor (24) and concentrated human platelet lysates to seek other proteins binding to the proteoglycan carrier molecule. In preliminary experiments, no evidence of significant binding of other proteins has been found, suggesting the proteoglycan has a high degree of specificity for PF4. PF4 is a chemotactic protein for human monocytes and for human neutrophils (25). Cellulose acetate electrophoresis patterns of proteoglycan, PF4, and the proteoglycan-PF4 complex. Five p1 of the complex ( a . 1 mg/ml of proteoglycan and PF4). l'F4 ( b , 1 rng/ml), and proteoglycan (c and d, 1 mg/ml), was applied to two separate cellulose acetate plates. After electrophoresis at pH 8.8 at 160 V for 15 min, one plate was stained for proteoglycan with alcian blue and the other were stained for protein (PF4) with ponceau S. The two plates were superimposed and photographs were taken to de.nonstrate the precise relationships of the alcian blue/ponceau S staining materials.
Addition of the PF4 proteoglycan carrier molecule with PF4 in chemotactic assays failed to reduce the chemotactic activity of PF4, suggesting that PF4 bound to the carrier proteoglycan remains able to subsequently effect a physiological response with human monocytes.' The proteoglycan after chondroitinase digestion or after desulfation fails to bind PF4; the intact glycosaminoglycan chain and sulfate group of the proteoglycan molecule appear to be required for the interaction between proteoglycan and PF4.

DISCUSSION
The purification procedure for the proteoglycan carrier of PF4 described in this communication is relatively simple and results in a high recovery (46%), with an -33,000-fold purification achieved. PF4 coupled to cyanogen bromide-activated agarose through the amino groups of PF4 was not effective in binding the proteoglycan to the affinity gel, suggesting the E amino groups of the lysine residues in PF4 are important in the interaction between PF4 and proteoglycan. It is believed that the two pairs of lysine in the carboxyl-terminal end of the PF4 molecule are the binding sites for heparin (12,13). It is likely that the proteoglycan binds poly-L-lysine gels through a similar interaction with the multiple charged groups of the polylysine ligand.
The proteoglycan contains 10% protein and is rich in uronic acid and galactosamine. The lack of glucosamine in the proteoglycan rules out contamination with heparin. Recently, Silvestri et al. (23) isolated a proteoglycan inhibitor for Clq from human serum. Although it was shown to contain chondroitin 4-sulfate, its chemical composition and molecular size are quite different from our PF4 carrier proteoglycan, suggesting that these molecules are distinct proteoglycan molecules. The molecular weight (53,000) of the PF4-proteoglycan carrier is smaller than that of the Clq proteoglycan inhibitor and proteoglycans isolated from other sources (26,27). A question arises of whether the M, = 53,000 proteoglycan is the proteolytic product of a larger proteoglycan. In order to minimize proteolysis during isolation and purification, we included multiple protease inhibitors (diisopropyl fluorophosphate, phenylmethylsulfonyl fluoride, ~-tosylamido-2-phenylethyl chloromethyl ketone, beczamidine-HC1, EDTA, and soybean trypsin inhibitor) in preparations of human platelet-rich plasma as they were thawed. The separate isolations of the PF4-proteoglycan carrier consistently resulted in a M , = 53,000 product. In the preparation of Silvestri et al. (23), protease inhibitors were not used, calcium was present, and two 37 "C incubations were included in the purification. If the two molecules are related, it seems likely that cleavage of proteoglycan of Clq inhibitor to a M , = 53,000 species would have occurred and been detected. It is not clear why the proteoglycan does not enter 5% SDS-polyacrylamide gels, but, under the conditions of electrophoresis, it seems likely that the proteoglycan molecule aggregates at the top of the gel.
Several experiments suggest that the proteoglycan isolated is the specific proteoglycan released by platelets in association with PF4 (11). The proteoglycan is present in high concentration in platelet-rich plasma. It binds to PF4-agarose and to PF4 during cellulose acetate electrophoresis. No other proteoglycan was found to bind to PF4-agarose; high ionic strength (0.3 M NaCl) is required to release the proteoglycan from the column. No other proteins in platelet lysates were found to bind to the proteoglycan. The role of the proteoglycan carrier molecule in the platelet is not clear, however. One possible role for the proteoglycan carrier molecule is in the "packaging" S. S. Huang, J. S. Huang, and T. F. Deuel, unpublished results. of PF4 in platelet a-granules. PF4 is relatively insoluble in physiological ionic strength (1 I), whereas the PF4-proteoglycan complex is highly soluble at this ionic strength. PF4 binds heparin with high affinity. PF4 recently has been shown to be a powerful chemoattractant protein for human monocytes and for human polymorphonuclear leucocytes (25). The proteoglycan carrier molecule may serve to maintain PF4 in soluble form for interactions in the circulation or for delivery to sites of blood vessel injury. Investigations presently are in progress to further characterize the structure and function of this potentially important molecule.