Sulfated Proteoglycans and Sulfated Proteins in Guinea Pig Megakaryocytes and Platelets in Vivo RELEVANCE TO MEGAKARYOCYTE MATURATION AND PLATELET ACTIVATION*

This study has examined changes in proteoglycan synthesis during megakaryocyte maturation in vivo. Guinea pigs were injected with Na,”SO,, and megakaryocytes and platelets were isolated from 3 h to 5 days later. The proteoglycans and other sulfated molecules in both cells were characterized at each time point by gel filtration, ion-exchange chromatography, gel electrophoresis, and chemical and enzymatic digestions. Two populations of chondroitin 6-sulfate proteo- glycans were found by DEAE-Sephacel chromatography. The major fraction was eluted with 4 M guanidine hydrochloride and the minor fraction with 4 M guanidine HCl, 2% Triton X-100. The K,, of the major proteoglycan peak in the platelets at 1 day after injection was 0.18-0.20 on Sepharose CL-GB and decreased gradually to 0.12 by 3 days, when proteoglycan raqio-activity per cell was maximal. The peak for mega- karyocyte proteoglycans at 3 h was broad, with K,, = 0.1-0.2. The appearance of different portions of the proteoglycan peak in platelets coincided with their disappearance from megakaryocytes. Proteoglycan size was a function of glycosaminoglycan chain length. The proteoglycans eluted with Triton X-100 from DEAE-Sephacel Chemical were released from proteoglycans by digestion M NaOH, 1 M NaBH, for 18 h at 42 "C (25). Papain and trypsin digests were done as described previously (24) except that a 10-fold greater enzyme concentration was used. Pronase digestion was per- formed as described (26). Chondroitinase ABC and AC-I1 and chon-dro-4- and -6-sulfatase digestions were performed as previously de- scribed (24, 27), with chondroitin 4-sulfate or chondroitin 6-sulfate as carrier. Digestion products were separated by thin-layer chroma- tography using 20 X 20-cm plates (28), bands were visualized under UV light, and the lanes were scraped in 0.5-cm portions for radioac- tivity determinations. Alternatively, digestion products were separated by chromatography on 0.9 X 60-cm columns of Sephadex G-25, eluted with 50 mM NaAc, 0.2 M NaC1, or Bio-Gel P-2, eluted with 0.4 M ammonium acetate (29). Uronic acid was quantitated as described (30). Core Protein Analysis-The 35S-labeled proteoglycans obtained by DEAE-Sephacel chromatography were dialyzed and lyophilized. They were digested with chondroitinase ABC as described by Oike et al. (31), except that the digestion was stopped by adding 3 volumes of 1.3% KOH in 95% ethanol. The digest was allowed to precipitate at -20 "C overnight. The precipitate was collected by centrifugation at 15,000 X g for 5 min and then was analyzed by polyacrylamide gel electrophoresis in the Laemmli system (32) as modified by Rosenberg et al. (33) using 10% acrylamide gels. The gels were stained with Coomassie were separated by centrifugation. The super- natant was analyzed by gel filtration on Sepharose CL-GB in 4 M guanidine HCl without further treatment, and the pellet was solubi- lized in the Zwittergent/guanidine mixture for gel filtration.

through an extraordinary process of cytoplasmic and membrane formation and differentiation. Platelets are devoid of nuclei, Golgi, and endoplasmic reticulum and therefore have virtually no capacity for protein or proteoglycan synthesis. The platelet proteins and proteoglycans appear to be derived by their synthesis by the megakaryocyte during platelet production (1).
The proteoglycans of megakaryocytes and platelets are of interest to study for several reasons. First, they may be important for megakaryocyte development. Changes in proteoglycans with cell maturation have been reported in several types of cells, both in terms of glycosaminoglycan composition and proteoglycan size (2)(3)(4)(5)(6). Proteoglycans have been shown to be important for tissue differentiation (7,8). Megakaryocytes appear to synthesize proteoglycans or other large sulfated molecules at all stages of development, according to autoradiographic studies (9,10); and it is possible that the nature of the molecules synthesized could change as the cells mature. This might be important for migration of the cells within the marrow stroma, as has been shown for marrow granulocytes (11); for the sorting out of the developing platelet plasma membranes into appropriate territories for the individual platelets, analogous to tissue differentiation (7,8); and for granule formation since the CY granules of the platelets contain proteoglycans (9,(12)(13)(14)(15). In addition, proteoglycans appear to be involved in platelet activation. The platelet proteoglycans appear to be found predominantly in the CY granules and to a lesser extent in the plasma membrane. They are found in the analogous structures in megakaryocytes, i.e. the a granules and the demarcation membrane system, which is presumably the source of the surface membrane of the platelet (9,12,13). The proteoglycans are released from the CY granules by thrombin (14,15) and presumably from the platelet surface by ADP treatment (15,16). Release of surface proteoglycans could affect platelet-platelet or platelet-vessel wall interactions. Understanding the timing of synthesis of proteoglycans in the megakaryocyte could help us to understand the processes of a granule and membrane synthesis in these cells that are critical for platelet function.
Two major difficulties in studying the biochemistry of developing megakaryocytes are the inability to isolate pure populations of very young cells from the marrow mononuclear cells of other lineages and the inability to separate the complex array of the isolatable, relatively mature "recognizable" megakaryocytes into their several ploidy and maturational classes (17). The isolatable megakaryocytes represent predominantly the more mature spectrum of megakaryocytes (18). We have tried to circumvent these problems by in vivo studies. Since the product cell, the blood platelet, is completely separated from the precursor cell compartment, a rationale could be established for interpretation of the radiolabeling of molecules found in both cells at various times after administration of a single injection of sodium [35S]sulfate to guinea pigs. The assumptions were that the single injection would serve as a pulse label, with the [35S]sulfate taken up at high specific activity during a brief period by megakaryocytes at all stages of development, and that some or all of the molecules synthesized during this time would be found eventually in the platelets. Thus, molecules synthesized by relatively mature megakaryocytes would appear in the platelets at early time points after the pulse injection, and molecules synthesized by the younger megakaryocytes would appear in the platelets at later times.
This approach has permitted us to study proteoglycan metabolism over the time required for megakaryocyte maturation and platelet production and has, in addition, provided evidence for the presence of other types of sulfated molecules in megakaryocytes and platelets. The in vivo labeling has also enabled us to explore the role of these different molecules in platelet activation.
A preliminary account of this work has been presented (19).
Animals-Male guinea pigs of the Hartley strain were obtained at 350 g from Hazleton Research Products (Denver, PA) and killed at 350-500 g. Animals were housed individually in wire-bottomed cages and fed and watered daily.
Injection of Animals with [35S]Sulfate-The radiosulfate was obtained as a lyophilized salt, and dissolved in saline, and sterilized by filtration through a 0.2-pm Millipore filter. The material was administered intraperitoneally. The amount of radiolabel incorporated into megakaryocytes and platelets in different animals had a variability of +20% at any given time point, but duplicate determinations on platelets obtained from separate syringes of blood from the same animal and duplicate determinations on megakaryocytes from the same preparation were always identical. The dose response was linear over a range of 1-5 pCi/g; most experiments were performed with 3 pCi/g, which provided sufficient radioactivity for analysis of molecules in the megakaryocytes and platelets of individual animals.
Cell Preparations-Guinea pig platelets were prepared as described previously (20). Blood was drawn by cardiac puncture into a 10-ml syringe containing 1 ml of NIH acid/citrate/dextrose anticoagulant. Anesthesia was either light ether or Vetalar/acepromazine. Plateletrich plasma was prepared by centrifuging the blood at room temperature at 250 X g for 15 min. Platelets were pelleted from the plateletrich plasma at 750 X g at 4 "C for 15 min and washed twice under these conditions with calcium-and magnesium-free Hanks' balanced salt solution, pH 6.5. This washing was sufficient to remove the plasma-associated [35S]sulfate from the cells since the second wash contained only 2-4% of the total radioactivity of the platelet pellet. Aliquots of the platelet suspension were removed before the final centrifugation for platelet counts and for radioactivity determinations. Platelets were counted on a Baker MK-4/HC platelet counter. Radioactivity determinations were made by mixing 100 p1 of the ' The abbreviations used are: CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; GdnHC1, guanidine hydrochloride. platelet suspension with 5 ml of Scintiverse 11. The data were in excellent agreement with radioactivity determinations of aliquots of the platelet extracts. The platelets were free of red and white cell contamination as determined by phase-contrast microscopy. Approximately 2 X 10' platelets were obtained per animal.
Human platelets were prepared as described previously (21). Blood was collected in one-tenth volume of 3.8% citrate and washed with TrisPDTA. Contamination with white cells was less than l/lO,OOO.
Guinea pig megakaryocytes were prepared by the method of Levine and Fedorko (18) with subsequent modifications (22). Marrow was obtained from the femora, humeri, and tibiae of each animal. The marrow cells were disaggregated in a medium containing 1 mM adenosine, 2 mM theophylline, and 0.32% sodium citrate in calciumand magnesium-free Hanks' balanced salt solution, washed in this same solution but without citrate and with 5% bovine serum albumin, and passed through a density gradient centrifugation and two successive velocity sedimentations on bovine albumin gradients. Adenosine and theophylline were included in all gradient solutions to optimize cell morphology. The yield of megakaryocytes was 4-8 X 105/animal, at a purity of 80-90% by cell number. At this degree of purity, the megakaryocyte mass accounts for at least 97% of the cell suspension because of the large size of the megakaryocytes relative to the contaminating cells (22). The viability as judged by trypan blue exclusion is at least 90%.
Other Cells-In some experiments, the radiolabeling of cells from the upper layers of the velocity sedimentation gradients, which are normally discarded during the megakaryocyte isolation, was analyzed for comparison to the megakaryocytes.
Extraction of Cellular Proteoglycans-The megakaryocytes, platelets, and other marrow cells were completely solubilized by the procedure of Kimura et al. (23). The cells were treated first with 8% Zwittergent 3-12 in 50 mM sodium acetate, pH 5.8, containing the protease inhibitors 2 mM phenylmethylsulfonyl fluoride, 20 mM EDTA, 5 mM benzamidine, and 0.1 M 6-aminocaproic acid for 30 min at 4 "C. An equal volume of 8 M guanidine HCl in the same buffer was then added. The radiolabeled material was stable in these extracts for at least 1 month at 4 "C. Alternatively, the cells were dissolved completely in 8 M urea, 50 mM Tris-HC1, 0.1 M NaCI, 0.2% Triton X-100, and this was the method of choice for analyses beginning with ion-exchange chromatography.
Gel Filtration of Megakaryocyte and Platelet Extracts-The extracts were applied directly to Sepharose CL-GB columns (60 X 0.9 cm) equilibrated and eluted with 50 mM sodium acetate, 4 M guanidine HCI, 0.2% Triton X-100, pH 7.0 (24). Elution was performed at about 3.5 ml/h. Fractions of about 0.5 ml were collected directly into 7-ml scintillation vials for radioactivity determinations. The void and total volumes of each column were determined with blue dextran and phenol red or sodium [35S]sulfate. Inclusion of the dyes did not alter the migration of the %-labeled molecules on the columns. Platelets which were solubilized directly in the 8 M urea solvent behaved identically to the Zwittergent/guanidine HCI extracts on the Sepharose columns.
Zon-exchange Chromatography-The initial experiments were performed using DEAE-Sephacel columns (8 X 0.9 cm), equilibrated with 8 M urea, 50 mM Tris-HC1, 0.1 M NaC1, 0.2% Triton X-100, pH 8.0, and eluted with this solvent system, followed by a gradient of 0.1-0.8 M NaCl (24). The Zwittergent/guanidine extracts were either diluted to a final concentration of 0.1 M chloride ion and applied directly to the column or first chromatographed on a PD-10 column with 4 M guanidine HCI, 50 mM sodium acetate, 0.1 M sodium sulfate, 0.2% Triton X-100, after which the excluded fractions were concentrated by Centricon-10 filtration and resuspension in the urea solution. A major problem was precipitation of %-labeled material when any of these preparations was exposed to 8 M urea. This problem was circumvented by solubilizing the platelets directly in the DEAE-Sephacel column buffer, avoiding use of Zwittergent 3-12. Then, since most of the proteoglycan eluted in a sharp peak at 0.5 M NaCl during the gradient elution, the protocol was modified to a batch procedure. The DEAE-Sephacel was equilibrated in the 8 M urea buffer, but without Triton X-100. The sample was applied to the column and eluted with 3 bed volumes each of 8 M urea, 50 mM Tris-HCI, 0.1 M NaCI, pH 8.0; 0.23 M NaCl in the same buffer; 4 M guanidine HCI, 50 mM sodium acetate, pH 8.0; and then 4 M guanidine HCI, 50 mM sodium acetate, 2% Triton X-100 or 2% CHAPS. After the first fraction of the 2% Triton (or CHAPS) solvent was collected, the column flow was stopped for several hours before elution of the final fractions. Nearly quantitative recovery (at least 97%) of radioactivity was achieved by this procedure.
Chemical and Enzymatic Digestion of Proteoglycans-Glycosaminoglycans were released from the proteoglycans by digestion in 0.05 M NaOH, 1 M NaBH, for 18 h at 42 "C (25). Papain and trypsin digests were done as described previously (24) except that a 10-fold greater enzyme concentration was used. Pronase digestion was performed as described (26). Chondroitinase ABC and AC-I1 and chondro-4-and -6-sulfatase digestions were performed as previously described (24,27), with chondroitin 4-sulfate or chondroitin 6-sulfate as carrier. Digestion products were separated by thin-layer chromatography using 20 X 20-cm plates (28), bands were visualized under UV light, and the lanes were scraped in 0.5-cm portions for radioactivity determinations. Alternatively, digestion products were separated by chromatography on 0.9 X 60-cm columns of Sephadex G-25, eluted with 50 mM NaAc, 0.2 M NaC1, or Bio-Gel P-2, eluted with 0.4 M ammonium acetate (29). Uronic acid was quantitated as described (30).
Core Protein Analysis-The 35S-labeled proteoglycans obtained by DEAE-Sephacel chromatography were dialyzed and lyophilized. They were digested with chondroitinase ABC as described by Oike et al. (31), except that the digestion was stopped by adding 3 volumes of 1.3% KOH in 95% ethanol. The digest was allowed to precipitate at -20 "C overnight. The precipitate was collected by centrifugation at 15,000 X g for 5 min and then was analyzed by polyacrylamide gel electrophoresis in the Laemmli system (32) as modified by Rosenberg et al. (33) using 10% acrylamide gels. The gels were stained with Coomassie Blue to identify proteins, and the radioactive bands were identified by fluorography. The gels were treated with EN3HANCE (Du Pont-New England Nuclear) and kept at -70 "C in the presence of Kodak XAR film. The film was developed by an automated X-Omat processor.
Electrophoresis of Intact Proteoglycans and Glycosaminoglycan.-Intact proteoglycans were electrophoresed as described by Rosenberg et al. (33) except that a polyacrylamide gradient of 4-8 or 4-10% was used. Glycosaminoglycan electrophoresis was performed as described by Min and Cowman (34) using either 32-or 20-cm gels of 10% polyacrylamide run in a continuous buffer system. The chondroitin sulfates obtained from Miles Laboratories were run as standards. The gels were stained with Alcian blue, and the radioactive bands were visualized by fluorography.
Analysis of %-Labeled Molecules Released from Platelets by Thrombin and ADP Treatment-Platelets were washed as described above. Platelets were treated with 2 units/ml thrombin at 37 "C in the presence of 1 mM CaC12 in magnesium-free Hanks' balanced salt solution, pH 7.4, for 5 min with stirring. The suspension was cooled to 4 "C and centrifuged for 15 min at 750 X g. The supernatant was removed for subsequent analysis by gel filtration or ion-exchange chromatography, and the pellets were redissolved either in the Zwittergent/guanidine HC1 solutions for gel filtration or in the 8 M urea solution for ion-exchange chromatography.
For ADP treatment, the cells were incubated in Hanks' balanced salt solution containing 1 mM MgC12, 0.5 mM CaCl,, and 1 mg/ml fibrinogen, pH 7.4, for 10 min at 37 "C. ADP was then added to a final concentration of 8-60 I.IM, and the suspension was stirred in an aggregometer (Payton Scientific Inc.) to determine when aggregation was maximal. The suspension was cooled to 4 'C, and the supernatant and platelet fragments were separated by centrifugation. The supernatant was analyzed by gel filtration on Sepharose CL-GB in 4 M guanidine HCl without further treatment, and the pellet was solubilized in the Zwittergent/guanidine mixture for gel filtration.

Time Course of Incorporation of P5SlSulfate into Megakaryocytes and Platelets following Injection of Sodium P"S1Sulfate
The time courses of total incorporation of radioactivity per cell for megakaryocytes and platelets and the incorporation of radioactivity into macromolecules and proteoglycans per cell for platelets are shown in Fig. 1. These and other data in this paper have been normalized to compare one megakaryocyte to 1000 platelets, reflecting the average relative protein content of the populations of the two cells (20). The earliest time point obtained was 3 h following the injection. Maximal radioactivity was observed at this time in the megakaryocytes, with a rapid fall through the first 3 days and a . Each point has a standard deviation of f15-20%. The radioactivity in macromolecules in platelets (lower) was determined by subtracting the material at the V, of the Sepharose CL-GB columns in Fig. 2, and the radioactivity in proteoglycans in platelets was calculated from the percent of total cell radioactivity recovered in the 4 M GdnHCl and 4 M GdnHCl -k 2% Triton X-100 eluates from DEAE-Sephacel columns (standard deviation = 4-8% at each point, 6-12 animals at each point). Megakaryocyte proteoglycan radioactivity was >90% of total cell radioactivity at 3 h and about 70% at 1 day after injection. much slower decline thereafter. More than 90% of the whole cell radioactivity at all times was in macromolecules (see Fig.  3). Platelet radioactivity was very high at 3 h, but was predominantly due to material at the V, which was probably adherent free [35S]sulfate; this has been reported previously in mouse platelets (35). In contrast, in platelets at all other time points, most of the radioactivity was contained in macromolecules (see also Figs. 2 and 3). The radioactivity in platelets increased from 1 to 3 days after injection and then declined slowly. The decrease in radioactivity is most likely due to degranulation of the platelets during their time in the circulation and to the removal of platelets from the circulation (10,36). The time course of appearance and disappearance of total cell radioactivity in guinea pig platelets is the same as that reported by others for rat (36) and rabbit (37) platelets. The labeling of megakaryocytes has not been quantitated previously. peak in platelets until 3 days after injection could be accounted for by the disappearance of similar material from megakaryocytes over a given time period. At 3 h, the megakaryocyte extract had a broad asymmetrical peak ranging from K., = 0.1 to 0.2. At 1 day, the lower molecular weight portion of this peak was greatly diminished; and the KaV at this point was less than 0.1. Over the following 48 h, the material under the Kay = 0.1 peak was lost, corresponding to the increased labeling of the platelets and the shift of the Kav to higher molecular weight in the platelet extract.

CL-GB Chromatography
A second region of the eluates that is of interest is the portion from KaV = 0.3 to 0.6, which was also seen in megakaryocytes and platelets, but followed a different time course from that of the major peak. At 3 h after injection, very little was found in the isolated megakaryocytes. However, this portion of the eluate increased in megakaryocytes for at least 2 days and gradually increased in platelets over 3 and, in some experiments, 4 days. There was a net loss of radioactivity in these molecules from the megakaryocytes only after 3 or 4 days after labeling. This peak increased gradually in platelets and was usually maximal at 4 days. Considerable labeling of molecules in this size range was found 3 h after injection in the cells remaining in the upper layers of the velocity sedimentation gradients (Fig. 4), which are discarded during megakaryocyte purification and which contain many of the small, young megakaryocytes, along with the erythroid and myeloid cells. One possible interpretation of the time course of labeling of these molecules in megakaryocytes and platelets is that they are synthesized primarily by megakaryocytes that are too small to be isolated by our methodology at the early time points following sulfate injection, but that after 1 or 2 days, these cells have matured to the point where they can be isolated, and then after 2-3 days, to the point where they have released platelets. Thus, the amount of labeling of the molecules between Kay = 0.3 and 0.6 in megakaryocytes at any point from 2 to 5 days after injection would be the net result of 35S-labeled molecules in cells entering and leaving the isolatable cell population.
A third peak running just before the V, of the Sepharose column was also seen in all megakaryocyte and platelet extracts. The changes in height of this peak in the platelet extracts generally paralleled the increase and decrease of the height of the major peak, and this peak declined with time in the megakaryocytes. As discussed below, about one-third to one-half of this material is released when platelets are stimulated with ADP or thrombin. We thus suggest that this material may not be free sulfate, but may include sulfated amines such as serotonin sulfate or dopamine sulfate, which have been found by others in platelets (38,39). The pool of these molecules in the platelets also appears to be derived from the megakaryocytes.
Thus, all the labeling of molecules observed in platelets by means of Sepharose CL-GB chromatography of whole cell extracts can be accounted for by their appearance first in megakaryocytes.
The order in which the radiolabeled molecules are lost from the platelets appears to follow the order in which they appear in the cells: the portion of the major peak at K,, = 0.18 disappears first, followed by the higher molecular weight material. We cannot determine from our data whether this represents a partial slow release of a granule contents by a last-in-first-out mechanism, complete release of granule contents, or removal of the whole platelet from the circulation.

Characterization of Sulfated Molecules Synthesized by Guinea Pig Megakaryocytes i n Vivo
The sulfated molecules found in the megakaryocytes and platelets in vivo were characterized by ion-exchange chromatography, alkaline borohydride digestion, and enzymatic treatments as described below.

Characterization of the 35S-Labeled Molecules Separated by Ion-exchange Chromatography
The Zwittergentlguanidine HC1 extract of platelets labeled i n vivo for 4 days was passed through a PD-10 column, concentrated by Centricon-10 filtration, and diluted to 0.1 M C1-with 8 M urea, 50 mM Tris HC1,0.2% Triton X-100 before application to the DEAE-Sephacel column. The very low molecular weight 35S-labeled molecules were lost during Centricon filtration and so were not applied to the column. About 25% of the remaining radioactivity was eluted in the washthrough with 0.1 M NaCl in the 8 M urea buffer, a few percent in the early fractions of the gradient, and about 50% at 0.5 M NaC1. The remainder could not be eluted from the column even with 2 M NaCl.
A similar distribution of radioactivity was observed when platelets were solubilized directly in the 8 M urea, 50 mM Tris HC1, 0.1 M NaCl, 0.2% Triton X-100 buffer and eluted by the batch procedure described above. A typical elution pattern is shown in Fig. 5   injection. About 30% of the radiolabel in the extract eluted in the wash-through. The Sepharose CL-GB profile of each fraction of the DEAE-Sephacel column is shown in Fig. 6. The material in the four fractions differed significantly from each Platelet Proteoglycans and Sulfoproteins 1057 other, and all fractions are described below. 0.1 M NaCl Eluate-The material in this fraction had a broad distribution of radioactivity from the V,, to about KaV = 0.5. This material was completely resistant to chondroitinase digestion. Hydrolysis of this fraction with mild alkaline borohydride, Pronase, or papain resulted in nearly complete release of "'S radioactivity as fragments which eluted a t K,, -0.9 on Sepharose CL-GB and were found mostly in the included volume of Bio-Gel P-2 columns. On this basis, the 0.1 M NaCl eluate from the DEAE-Sephacel column appeared to contain some sulfated proteins, and the small fragments released by alkaline borohydride or Pronase digestion may be small sulfated oligosaccharides. The very low molecular weight material near the V, on the Sepharose columns of the whole cell extracts also eluted in this fraction and could be seen when the Sepharose column was run without prior dialysis (data not shown).
0.23 M NaCl Eluate-This fraction contained a high molecular weight material as well as some molecules in the size range found in the 0.1 M NaCl eluate. These molecules appear to be sulfoproteins by the same criteria described for the 0.1 M NaCl eluate.
4 M Guanidine HCl Eluate-The material which eluted with 4 M guanidine HCl in the absence of Triton X-100 produced a single broad peak on Sepharose CL-GB. The K,, and the shape of the peak varied in the same direction as did the major peak in the whole cell extracts (cf. Fig. 2) depending upon the time after ["'S]sulfate injection that the cells were isolated. This difference could also be shown by SDS-PAGE of the isolated proteoglycans (Fig. 7). About 96% of the material in this fraction could be degraded with chondroiti- nase ARC, and further characterization using chondro-4sulfatase, chondro-6-sulfatase, and chondroitinase AC-I1 demonstrated that the glycosaminoglycans consisted entirely of chondroitin 6-sulfate. This finding was obtained by analysis of the digestion products by Sephadex G-25 and Bio-Gel P-2 column chromatography and by thin-layer chromatography (data not shown).
Mild alkaline borohydride treatment of the proteoglycans produced large fragments migrating with Kay consistent with glycosaminoglycans ( Fig. 8), which were digested to 99% with chondroitinase AC-11. No degradation was detected with nitrous acid. Subdivision of the Sepharose CL-GB eluate of the proteoglycans as shown in Fig. 8 and digestion of each subfraction with mild alkaline borohydride demonstrated that the size of the proteoglycans was a function of the glycosaminoglycan chain length. This same relationship was observed when proteoglycans from megakaryocytes obtained 3 h after sulfate injection were analyzed by the same protocol. Electrophoresis of the glycosaminoglycans demonstrated that the chain length was in the same range as the chondroitin 4sulfate standard obtained from Miles Laboratories, approximate M, 40,000-80,000 (Fig. 9). Comparison of the size of the radiolabeled glycosaminoglycans found in platelets a t 1 and 3 days after sulfate injection demonstrates that only the lower molecular weight glycosaminoglycans are labeled in the platelets a t 1 day after injection, but the whole spectrum of glycosaminoglycans seen with the Alcian blue stain is labeled a t 2 and 3 days (see also Fig. 12). These data, together with Fig. 7, demonstrate that the most mature megakaryocytes (i.e. those cells which will produce platelets within 24 h after the sulfate injection) synthesize smaller proteoglycans with smaller glycosaminoglycan chains than do the younger megakaryocytes.
In addition to the glycosaminoglycans, borohydride digestion produced small fragments migrating a t K., -0.85 on Sepharose CL-GB which may be small sulfated oligosaccharides. The proteoglycans were totally resistant to trypsin digestion. Papain digestion produced a broad spectrum of fragments with a peak at K., = 0.30-35 (data not shown), and Pronase digestion also produced fragments similar in size to those of the alkaline borohydride digest. The peak at KBv = 0.85 was generated by both papain and Pronase.  Fig. 6 were subfractionated as shown (a), and each fraction was digested with NaOH/NaBH, to release the glycosaminoglycans

3
v    39,000 and 36,000, with several minor bands. None of these bands are present in the undigested material. Fig. 10 also compares the core proteins from platelet proteoglycans obtained 1 and 3 days after ["S]sulfate injection. The higher molecular weight band is more prominent in the 3-day than in the 1day sample, suggesting that this band is associated preferentially with higher molecular weight proteoglycans and the smaller core protein with the smaller proteoglycans. A similar experiment in which the proteoglycans were obtained from the three portions of the Sepharose CL-GB eluate as shown in Fig. 8 demonstrated the same core protein and proteoglycan size relationship. We cannot conclude unambiguously from these experiments that the different proteoglycan size components have different core proteins, but the data suggest that there are some small differences in molecular weight. We speculate that these differences may be due to different degrees of glycosylation, possibly due to the small sulfated oligosaccharide. The multiple bands do not appear to be an artifact of the length of digestion since a 24-h incubation produced the same banding pattern. Triton X-100 (or 2% CHAPS) Eluate-The molecules eluting in this fraction were larger than those eluting with 4 M guanidine HCl without detergent. This finding did not appear to be an artifact of the presence of detergent since the same profile on Sepharose CL-GB was obtained when the material was chromatographed with 0.2% Triton X-100, when the proteoglycans were precipitated from the 2% Triton solution to remove most of the detergent, when the sample was dialyzed extensively against 4 M guanidine HCl to remove the Triton (confirmed by monitoring or when Triton was replaced by 2% CHAPS for the DEAE-Sephacel column. This material migrated a t a higher molecular weight on SDS-PAGE than did the proteoglycans from the 4 M guanidine HCl fraction (Fig. 7), and no radiolabel was present in this molecular weight range in platelets obtained at 1 day after injection according to Sepharose CL-GB and DEAE-Sephacel chromatography and SDS-PAGE. Thus, this large proteoglycan appears to be synthesized considerably more than 24 h before platelet production. The glycosaminoglycans obtained from this fraction migrated as a very broad peak with Kay -0.32 on Sepharose CL-GB (Fig. 6, lower) and thus did not fit into the general pattern relating proteoglycan size to glycosaminoglycan chain length in the 4 M guanidine eluate described above, which would have predicted K., < 0.25 for these glycosaminoglycans. These glycosaminoglycans also were entirely chondroitin 6-sulfate. Thus, by several criteria, the proteoglycans in this fraction appear to be different from those described above. This material would probably correspond to the uronic acid-containing material from human platelets which could only be eluted from DEAE-Sephacel with 0.5 N NaOH and which contained 15% of the platelet uronic acid (40).

M Guanidine HCl + 2%
The core proteins of this proteoglycan fraction were not studied because insufficient material was available.

Control Experiment for the Effect of ["S]Sulfate Radioactivity on Megakaryocytes
In order to determine whether the effects of radiation on the marrow might be responsible for the changes in the sulfated proteoglycan size which we observed over time, several guinea pigs were injected with ["S]sulfate twice before isolation of platelets and megakaryocytes. The injections were made 5 days and again 1 day before death. The first injection was 3 mCi/kg, and the second was 2 mCi/kg. The pattern of labeling was what would have been anticipated by adding together the eluates of Sepharose CL-GB columns from animals injected once a t each time point. Therefore, the radioactivity did not appear to alter proteoglycan synthesis in the megakaryocytes.

Labeling of Megakaryocyte-poor Fractions Obtained during
Megakaryocyte Purification Fig. 4 shows the labeling profile on Sepharose CL-GB of megakaryocytes obtained at 3 h after sulfate injection in comparison with the cells in the upper layers of the two velocity sedimentation gradients. All three samples were obtained from the same animal. The cells from the first velocity sedimentation contain some erythrocytic, but mostly myelocytic cells, and the cells from the second gradient appear to be mostly myelocytic. Each of these fractions contains a small percentage, but significant number of recognizable megakaryocytes ( Table I), most of which are small, i.e. 15-20 pm in diameter, or quite large and apparently at terminal stages of development with pycnotic nuclei. The megakaryocyte number in these cell populations is probably underestimated because of the difficulty in distinguishing immature 4N and 8N cells from the large precursor cells of other hematopoietic cell lineages by phase-contrast microscopy (17). The amount of radiolabel per 5 X IO5 purified megakaryocytes is compared to 25 X 10' other marrow cells. The purified megakaryocytes contain much more total radioactivity per cell than do the other cell populations. There are striking differences in the profiles of the molecules synthesized by the megakaryocytes compared to the other cell fractions. The other cells show a much more disperse pattern of labeling than the megakaryocytes, with a much greater amount between K., = 0.3 and 0.6.
The peak near the V, is a smaller percentage of the total labeling of the purified megakaryocytes than of the extracts from the other cell populations. The radioactivity per cell in the upper layers of the velocity gradient decreased steadily with time over 5 days (data not shown). Whereas our data do not permit us to conclude with certainty that the young megakaryocytes contribute to the synthesis of the molecules migrating at Kay = 0.3-0.6 in the cells from the upper layers of the velocity gradients, the disappearance of these molecules from this cell population follows a time course consistent with their appearance in the isolatable mature megakaryocytes.

Labeling of Plasma after P'SlSulfate Injection
The Sepharose CL-GB elution profile of plasma after sulfate injection showed that a small amount of the labeling coincides with the material found in platelets, with peaks near K., = 0.2 and a shoulder at K. , = 0.3-0.4 leading to a major peak at TABLE I Comparison of ~5S]sulfate incorporation in vivo into megakaryocytepoor fractions of marrow from velocity sedimentation compared to purified megakaryocytes Cells were isolated from guinea pig marrow 3 h following injection of [36S]sulfate. Cells were solubilized in Zwittergent 3-12 and guanidine HC1 as described in the text, and these extracts were applied to Sepharose CL-GB columns equilibrated and eluted with 4 M guanidine HC1, 50 mM NaAc, 0.2% Triton X-100.

CL-GB"
Data are from the experiment shown in Fig. 4. Kay = 0.55 on columns run without guanidine HCl and detergent (data not shown). Whereas one must consider the possibility that some of the radioactivity in the platelet extracts was due to contamination with plasma, we consider this unlikely for several reasons. First, the second wash of the platelets contained no more than 2-4% of the total radioactivity found in the final platelet pellet, consistent with a previous report (35). Second, molecules in the same size range were labeled in megakaryocytes, and the time course of their disappearance from megakaryocytes could be correlated with their appearance in platelets. It is unlikely that the megakaryocytes were contaminated with plasma since the cells are washed or passed through gradients at least seven times before they are extracted. The labeling profile of plasma was the same from 1 to 5 days after injection for molecules migrating with K., up to 0.55 on Sepharose CL-GB. The lower molecular weight peaks at K,, = 0.85 and the V , predominated only at the very early time points, accounting for most of the radioactivity in the plasma. At 24 h, the peak at K., = 0.85 had only 28% of the total radioactivity, and this declined rapidly thereafter. The total sulfate radioactivity in plasma at 3 h was about 6 pCi/ml, at 48 h 0.12 &i/ml, and at 5 days 0.06 pCi/ml. This is consistent with other reports of the decay of plasma sulfate-specific activity in man (41) and mice (10).

Release of 36S-Labeled Molecules by Thrombin and ADP
From 55 to 70% of the radioactivity in the cells was released to the medium by thrombin treatment, in accord with the findings of Ward and Packham (15) with rabbit platelets and Riddell and Bier (42) with pig platelets. Similar data were obtained with guinea pig platelets obtained 1, 2, or 3 days after [35S]sulfate injection. Fig. 11 shows the difference between the labeling of molecules that are released from 4-day labeled thrombin-treated platelets and the labeling pattern of molecules that are retained in the cell. Both the supernatant and pellet were fractionated by DEAE-Sephacel chromatography, and the proteoglycans were chromatographed on Sepharose CL-GB. Almost all the radiolabeled material which was found in the supernatant was retained on the DEAE-Sephacel column and could be eluted with 4 M guanidine HC1. The small peak in the wash-through fraction contains mostly the low molecular weight material found at the Vt of the Sepharose column shown in Fig. 2. When the 4 M guanidine HCl eluate was chromatographed on the Sepharose CL-GB column, a single symmetrical peak with Kay -0.12 was obtained. In contrast, most of the radiolabel from the resolubilized platelet pellet eluted in the wash-through fraction of the DEAE-Sephacel column. The pellet itself could be completely resolubilized after thrombin treatment; however, if the 0.1 M NaCl eluate from the DEAE-Sephacel column was dialyzed, 40-60% of the radiolabeled material could not be resolubilized subsequently, This problem was not encountered using material from unstimulated cells and may be the result of formation of a cytoskeleton-like material. The resolubilized material eluted from Sepharose CL-GB primarily between Kay 0.3 and 0.5. The proteoglycan in the 4 M guanidine HC1 eluate from the pellet was larger than that which was released to the supernatant. The K,, of this molecule ranged from 0.04 to 0.07 in four experiments. It should be noted that all solvents used in this experiment contained Triton X-100, in contrast to the detergent-free solvent systems used for the unstimulated whole cells. This molecule appears to be the same as the one described above which was eluted from the DEAE-Sephacel column only with high detergent in the whole cell experiments. These data suggest that the sulfated macromolecules released from stimulated platelets are primarily proteoglycans, and the material retained consists of mostly sulfoproteins and a small amount of a proteoglycan that is larger than the released proteoglycans.
It should be noted that in all experiments about half the radiolabeled material near the V, of the Sepharose CL-GB column was released from the thrombin-treated platelets. Studies are in progress to determine the identity of these molecules; they are likely to include sulfated amines such as serotonin sulfate and dopamine sulfate, which have been reported to be present in platelets (38, 39). Fig. 12 compares the proteoglycans released by thrombin from platelets labeled for 1 and 4 days and the glucosaminoglycans generated from both. The relationship between proteoglycan size and glycosaminoglycan chain length observed in the proteoglycans purified from whole cell extracts of 3day labeled platelets (Fig. 8) is also seen in the released proteoglycans. Also seen is the small peak at Kay = 0.85 which was found in the borohydride digests of the DEAE-purified proteoglycans from whole cells and the peak at the V, which represents released small molecules. ADP treatment of platelets caused release of 10-14% of the sulfated material in platelets (data not shown), again in agreement with Ward et al. (15,16) with rabbit platelets. The proteoglycans and the material near the V, were the only molecules released from the platelets after treatment with 8, 20, or 60 p~ ADP. The K,, of the released material was representative of the average size of the proteoglycans in the whole cell extract and thus did not appear to differ from that of the proteoglycans released by thrombin, although the amount released was much less. This was true of platelets obtained 1-4 days after [35S]sulfate injection. About 20-30% of the low molecular weight material was released in all experiments.

Comparison of the Molecular Weight of Human and Guinea
Pig Platelets Proteoglycans were isolated from human and guinea pig platelets by DEAE-Sephacel chromatography and were then  1 0.2 0.3 0.4 0.5 0.6 0.7 0.8  cochromatographed on Sepharose CL-GB. Platelets were taken from animals 3 days after [36S]sulfate injection because this was the time of maximal labeling and was expected to be most representative of the overall proteoglycan population. The number of human platelets was 10 times that of guinea pig platelets. Aliquots of each column fraction were taken for uronic acid determinations (30) and radioactivity determinations. The molecular weight distribution of the guinea pig platelet proteoglycans was virtually identical to that of the human platelets.

DISCUSSION
Our experimental approach has differed from those of previous studies of platelet proteoglycans by analyzing metabolic aspects of proteoglycans in both megakaryocytes and platelets. By monitoring the pattern of incorporation of [35S]sulfate into specific molecules in both megakaryocytes and platelets in vivo over the 5-day maturation period of megakaryocytes, a metabolic diversity has been demonstrated in what otherwise appears to be a homogeneous or heterodisperse population of proteoglycan molecules. The time course of appearance and disappearance of radiolabel in specific molecules in megakaryocytes and platelets has provided the first biochemical evidence that the platelet proteoglycans are derived from the megakaryocytes. Unfortunately, our protocol was unable to monitor proteoglycans which might be specific for very young megakaryocytes and which would be lost during the early stages of development. The chondroitin sulfate proteoglycans produced by the most mature megakaryocytes are smaller than those produced by the less mature cells, suggesting a relationship between the degree of megakaryocyte maturation and the size of the proteoglycans and glycosaminoglycan chains synthesized by the cells. This pattern is consistent with reports of synthesis of proteoglycans with shorter chondroitin sulfate chains by mature cells in studies with chondrocytes (3) and bone cells (43). The importance of this finding with regard to cell development is not understood.
The significance to platelet physiology of a diverse popu-lation of sulfated proteoglycans also is not understood. Most of the [35S]sulfate-labeled proteoglycans are released from the platelets by thrombin, including both the molecules at Kay = 0.18-0.20 in the 1-day labeled platelets and the larger molecules near K,, = 0.1-0.13 which are found at 3 and 4 days after labeling. These proteoglycans are most likely stored in the a granules. One role for the platelet proteoglycan is thought to be as a carrier for platelet factor 4 (14, 44). However, it is likely that other basic proteins in the a granules could also bind to the proteoglycans, and specific proteoglycans could bind preferentially to specific proteins within a given mixture. The data of Huang et al. (44) would argue against this possibility since their proteoglycan did not bind to platelet-derived growth factor or to other molecules in platelet lysates and the glycosaminoglycans did not bind to platelet factor 4. However, it is possible that the binding properties of their proteoglycan were altered by long exposure to plasma proteases. Barber et al. (14) reported that several glycosaminoglycans could bind to platelet factor 4 as well as or better than chondroitin 4-sulfate. It is intriguing to consider the findings of Cramer et al. (45) that several a granule proteins are arranged in an orderly fashion within the granule and the apparent location of mucopolysaccharide in a "nucleoid" within the a granule as described by Behnke (12). The diversity of proteoglycan size may contribute toward developing and maintaining an orderly arrangement of the a granule constituents.
It is of interest that the thrombin-releasable platelet proteoglycans, which are probably derived from the a granule, are degraded by Pronase and papain, in contrast to the mast cell lysosomal proteoglycans which are resistant to these proteases (46). This suggests that the core proteins of the platelet proteoglycans are quite different from those of the mast cells and may reflect the different nature and function of the proteins within the granules.
The proteoglycan that is found associated with the platelet pellet after thrombin treatment is larger than the proteoglycans secreted from the cells. It is probably the same molecule which requires high detergent concentration for elution from DEAE-Sephacel in our studies and which did not elute from DEAE-Sephacel with high salt in a study with human platelets (40). We suggest that this may be the membrane proteoglycan. Ward et al. (15,16) proposed that the proteoglycan released by ADP is a membrane proteoglycan, but only the nature of the glycosaminoglycan chain rather than the intact molecule was identified in their studies. However, we found that the proteoglycans released by ADP and thrombin appear to be the same molecules. This large proteoglycan appears to be synthesized 2-3 days before platelet release.
The analysis of whole cell extracts has demonstrated the presence of molecules which appear t o be sulfated proteins or sulfated glycoproteins in both megakaryocytes and platelets in uiuo. To our knowledge, sulfation of platelet proteins has not been reported previously. The function of these proteins remains to be established. It is possible that they form a portion of the glycocalyx. The platelet surface and the demarcation membrane system of the megakaryocytes can be labeled with reagents which are known to complex with proteoglycans but can also react with other negatively charged substances (13,47), and these membranes are also labeled metabolically with [35S]sulfate (13). However, Barber et al. (14) found only 2% of platelet uronic acid associated with the plasma membrane. The rather low amount of uronic acid detected in membranes by the biochemical study could result from loss of the proteoglycans during membrane purification; however, the apparent discrepancy between the biochemical and morphological studies could also represent an alternative source of sulfate labeling in the plasma membrane, possibly the sulfoproteins. Since each protein would contain only one or several sulfate residues, in contrast to 100-200/proteoglycan molecule, these proteins may be more abundant on a molar basis. The characterization of these molecules is in progress in this laboratory and will be reported in a future publication.' Sulfoproteins have been identified in several types of cells (e.g. Refs. 47-50).
Activated platelets are known to release substantial amounts of protease activity. Thus, it was of interest to note that the sulfate-labeled molecules, both the proteoglycans and the putative sulfoproteins, appeared not to be degraded as a result of platelet activation or by the thrombin in the medium. This conclusion is based on our ability to account for all portions of the whole cell eluate in the sum of the elution curves for the pellet and supernatant and the absence of detectable new peaks which would represent extensive degradation products. Therefore, any degradation of these molecules i n uiuo would result from the action of enzymes in plasma or tissues in which the platelets become activated.
We have found that the molecular weight range of the guinea pig platelet proteoglycans is the same as that of human platelet proteoglycans, although the former contain chondroitin 6-sulfate and the latter chondroitin 4-sulfate (14,44). The glycosaminoglycan chains have molecular weights of 40,000-80,000 based on electrophoresis and on elution position on Sepharose CL-GB (51), and the core proteins are M, 36,000-39,000. These data suggest a higher molecular weight than those of previous studies. The calibration of Heinegard and Hascall (52) for cartilage proteoglycan digests would suggest that the proteoglycans in our study have an average M , of about 200,000. However, two previous studies have characterized a human platelet proteoglycan as having M, 56,000 with four chondroitin 4-sulfate chains of M, 12,000.
These molecules were derived from the releasate of thrombintreated platelets (14) or from outdated frozen-thawed plateletrich plasma (44). In contrast, a recent study (40) which used methods similar to ours described the human platelet proteoglycan as a molecule of M , 136,000 with chondroitin sulfate chains of M , > 22,000 and presented evidence for the existence of other proteoglycans of different size or charge density; however, no reason for the diversity was established. The similarity of the guinea pig and human platelet proteoglycans suggests that our work is relevant to human platelets as well.
In conclusion, our studies suggest that megakaryocyte and platelet sulfate metabolism is more complex than previous studies have suggested. It will now be of interest to determine the specific roles of the sulfated molecules in megakaryocyte development and platelet function.