Purification and characterization of novel heparan sulfate proteoglycans produced by murine erythroleukemia cells in the growing phase.

Murine erythroleukemia cells (Friend erythroleukemia cells of a C-10-6 line) synthesized sulfated glycosaminoglycans consisting mainly of heparan sulfate (more than 95%) with a small amount of chondroitin 4-sulfate. The heparan sulfate occurred as proteoglycans, of which the cell-associated component was separated into urea-insoluble (UI) and urea-soluble (US) fractions. The UI proteoglycan consisted of a single homogeneous molecular species with an estimated Mr of 360,000 (C(UI)PG), whereas the US component was composed of two subfractions: a homogeneous species with an Mr of 280,000 (C(US)PGI) and a mixture of compounds with Mr values of less than 80,000 (C(US)PGII), which were isolated in yields of about 110, 340, and 80 micrograms of hexuronate (HexUA), respectively, from 1.37 g of an acetone powder prepared from 5.7 x 10(9) cells in the logarithmic phase of growth. The proteoglycan released into the medium (12 liters) was a single homogeneous species with an Mr of 320,000 (MPG) which was purified in a yield of 500 micrograms of hexuronate. The major, cell-associated proteoglycan, C(US)PGI, had very high contents of serine and glycine, accounting for approximately 80% of the total amino acids. This proteoglycan as well as the other two large proteoglycans, C(UI)PG and MPG, were highly resistant to degradation by various proteinases. These three proteoglycans, C(UI)PG, C(US)PGI, and MPG, had heparan sulfates with estimated Mr values of 32,000, 27,000, and 30,000. On the other hand, the Mr of the smaller proteoglycan, C(UI)PGII, was not significantly different before and after beta-elimination, indicating that it contains only a small peptide, if any. The heparan sulfate of this proteoglycan consisted of smaller and heterogeneous molecular species with Mr values of 26,000, 20,000, and 4,000. Digestion of these heparan sulfates with heparitinase I plus II resulted in almost complete depolymerization and gave six unsaturated disaccharides, delta HexUA-GlcNAc, delta HexUA-Glc-NAc(6-SO4), delta HexUA-GlcNSO3, delta HexUA-GlcNSO3 (6-SO4), delta HexUA(2-SO4)-GlcNSO3, and delta HexUA(2-SO4)-GlcNSO3(6-SO4). The relative amounts of these disaccharides generated from the individual heparan sulfates showed that an average ratio of sulfate residues to repeating disaccharide units of the C(US)PGII-derived heparan sulfate (0.97) was significantly higher than those of the other three large proteoglycan-derived glycosaminoglycans (0.54-0.70).

Murine erythroleukemia cells (Friend erythroleukemia cells of a (2-10-6 line) synthesized sulfated glycosaminoglycans consisting mainly of heparan sulfate (more than 95%) with a small amount of chondroitin 4-sulfate. The heparan sulfate occurred as proteoglycans, of which the cell-associated component was separated into urea-insoluble (UI) and urea-soluble (US) fractions. The UI proteoglycan consisted of a single homogeneous molecular species with an estimated M, of 360,000 (C(UI)PG), whereas the US component was composed of two subfractions: a homogeneous species with an M, of 280,000 (C(US)PGI) and a mixture of compounds with M, values of less than 80,000 (C(US)PGII), which were isolated in yields of about 110, 340, and 80 pg of hexuronate (HexUA), respectively, from 1.37 g of an acetone powder prepared from 5.7 % los cells in the logarithmic phase of growth.
The proteoglycan released into the medium (12 liters) was a single homogeneous species with an M, of 320,000 (MPG) which was purified in a yield of 500 pg of hexuronate. The major, cell-associated proteoglycan, C(US)PGI, had very high contents of serine and glycine, accounting for approximately 80% of the total amino acids. This proteoglycan as well as the other two large proteoglycans, C(U1)PG and MPG, were highly resistant to degradation by various proteinases. These three proteoglycans, C(UI)PG, C(US)PGI, and MPG, had heparan sulfates with estimated M, values of 32,000, 27,000, and 30,000. On the other hand, the M, of the smaller proteoglycan, C(UI)PGII, was not significantly different before and after &elimination, indicating that it contains only a small peptide, if any. The heparan sulfate of this proteoglycan consisted of smaller and heterogeneous molecular species with M , values of 26,000,20,000, and 4,000. Digestion of these heparan sulfates with heparitinase I plus I1 resulted in almost complete depolymerization and gave six unsaturated disaccharides, AHexUA-GlcNAc, AHexUA-Glc-NAc(6-S04), AHexUA-GlcNS03, AHexUA-GlcNSO3 (6-S04), AHexUA(2-S04)-GlcNS03, and AHexUA(2-S04)-GlcNS03(6-S04). The relative amounts of these disaccharides generated from the individual heparan sulfates showed that an average ratio of sulfate resi- Recent structural studies on proteoglycans of hemopoietic cells have demonstrated some common characteristics of their glycosaminoglycan and core peptide portions. The glycosaminoglycans of hemopoietic cells can be classified into three types on the basis of their degrees of sulfation. The most widespread type is fully sulfated chondroitin 4-sulfate (1-12). The second most common type is oversulfated chondroitin 4sulfate (or dermatan sulfate) species with an additional sulfate residue a t position 6 of their N-acetylgalactosamine residues, named chondroitin sulfate E, or at position 2 of their iduronic acid residues, named chondroitin sulfate Di-B (2, [13][14][15][16]. The third type is the most highly sulfated glycosaminoglycan, heparin, which is the major glycosaminoglycan component of connective tissue mast cells (17)(18)(19)(20)(21). All these glycosaminoglycans are present as proteoglycans, and these proteoglycans have been demonstrated or supposed to exist in storage granules of these hemopoietic cells, forming complexes with various granule proteins, such as with platelet factor 4 (22, 23), basically charged exo-and endopeptidases (24), and lysosomal enzymes (25,26). Nearly equimolar or excess sulfate groups per repeating disaccharide unit of these glycosaminoglycans are believed to be important for formation of complexes of the proteoglycans with the respective proteins (24).
Other common features of proteoglycans or their core peptides of hemopoietic cells are that they are highly resistant to various proteinases and that the M , values of their core peptides are relatively small (<20,000) (24). The property of proteinase resistance has been attributed to the highly acidic nature of glycosaminoglycans described above, the high degree of substitution of the glycosaminoglycan chains along the peptide cores, and the existence in these core peptides of clusters of serine-glycine repeats as glycosaminoglycan attachment domains (24). The last possibility is supported by the finding of such a cluster in a core peptide of heparin proteoglycan of connective tissue mast cells (18) and the recent findings of similar clusters in the amino acid sequences deduced from the cDNAs of core peptides of storage granule proteoglycans of other hemopoietic cells such as rat basophilic leukemia cells (27) and human promyelocytic leukemia cells of an HL-60 line (28).
Cell Culture and Radiolabeling with ~5S]Sulfate-Friend erythroleukemia cells of a C-10-6 line (32) were inoculated at an initial density of 3 X lo4 cells/ml of Ham's F-12 supplemented with 8% fetal bovine serum and cultured in suspension in a humidified incubator under 5% CO, in air at 37 "C. For obtaining sufficient amounts of proteoglycans for chemical analyses from cells in the logarithmic growth phase, culture was carried out on a large scale (12 liters of cell suspension), and cells were harvested on day 3, just before they reached a saturation density. We have demonstrated that fetal bovine serum contains chondroitin sulfate proteoglycans equivalent to 4.5 pg of hexuronate/ml (38), so 12 liters of culture medium was calculated to contain proteoglycans equivalent to 4.46 mg of hexuronate. For distinguishing proteoglycans produced by the cells from exogenous proteoglycans, a small scale culture (60 ml) was labeled with [3sS]sulfate (50 pCi/ml) during the period of cell growth (Fig. l), and then radiolabeled macromolecular fractions were combined with the unlabeled fractions, as shown in Fig. 2, and were used as tracers for purification of these fractions.
Extraction and Fractionation of Proteoglycans-As summarized in were washed three times with the medium without fetal bovine serum and the isotope and extracted with 15 ml of 4 M guanidine HC1, 50 mM Tris-HC1, pH 7.3, containing 10 mM EDTA, 10 mM N-ethylmaleimide, 1 mM phenylmethylsulfonyl fluoride, and 0.036 mM pepstatin A as proteinase inhibitors for 24 h on ice with continuous stirring. After centrifugation at 50,000 X g for 40 min, the residue was reextracted with 5 ml of the same buffer for another 12 h. Unlabeled cells (5.67 X lo9 cells) were also washed with medium without fetal bovine serum. The cells were then suspended in 400 ml of chilled (-20 "C) acetone containing the proteinase inhibitors and stood for 2 h at -20 "C with occasional stirring. Insoluble materials were collected by centrifugation at 7,000 X g for 10 min. This treatment was repeated three times, and the defatted material was then dried at room temperature. The acetone powder (1.37 g) was extracted by suspension in 900 ml of extraction buffer in the same way as described for extraction of radiolabeled materials. The suspension was centrifuged, and the residue was reextracted with 200 ml of the same extraction buffer. The radiolabeled and unlabeled residues were combined, washed with cold water several times, suspended in 50 mM Tris-HC1, pH 8.0, and digested extensively with Pronase. The solubilized material collected by centrifugation is referred to as fraction GI. The radiolabeled and unlabeled extracts were combined and dialyzed against five changes of 4 liters of 4 M urea, 20 mM NaC1, 20 mM Tris-HC1, pH 7.3, on ice. The insoluble material appearing during dialysis was collected by centrifugation at 8,000 X g for 20 min, washed twice with dialyzing buffer, and dissolved in the extraction buffer. This fraction is referred to as fraction UI. The dialyzate and the washing solution were combined and referred to as fraction US.
The 35S-labeled medium (60 ml) and washing medium (15 ml) were combined and applied to a column (2.6 X 90 cm) of Bio-Gel P-6 in 0.2 M NaCI, 50 mM Tris-HC1, pH 7.3, at 4 "C to remove radiolabeled low molecular materials. The unlabeled medium (12 liters) and the washing medium (500 ml) were combined, dialyzed against water in ice, and lyophilized. The freeze-dried material was dissolved in 2 liters of 0.2 M NaC1, 50 mM Tris-HC1, pH 7.3. This sample was combined with the 35S-labeled macromolecular fraction eluted in the void volume from the Bio-Gel P-6 column and referred to as fraction M.
Chromatographies-Column chromatographies of samples on gels and DEAE-Sephacel were performed at 4 "C unless otherwise indicated. HPLC for analysis of unsaturated disaccharides was performed using a Hitachi model 638-30 with a column (2.6 X 250 mm) of chemically modified silica gel, LiChrosorb NH2 (Merck), as described previously (33).
Analytical Methods-Hexuronate was measured by the method of Bitter and Muir (39) with glucuronolactone as a standard. Removal of materials interfering with the reaction from the samples was carried out as described previously (37). Hexosamine was determined by the Elson-Morgan method (40) as modified by Boas (41) with galactosamine as a standard, after hydrolysis of samples with 3 M HC1 and separation of the hydrolysates by chromatography on a column (0.7 X 15 cm) of AG-50W X4 (H' form) into glucosamine and galactosamine as described previously (42). The protein content was measured by the method of Lowry et al. (43), with bovine serum albumin as a standard, after desalting the samples.
Amino Acid and Hexosamine Analysis-Purified proteoglycans were hydrolyzed in sealed evaporated tubes with 6 M HCl at 110 "C for 16 h for amino acid analysis and with 3 M HCl at 100 "C for 16 h for hexosamine analysis. Analyses were carried out with an automated amino acid analyzer (Hitachi model 835) by the method of Spackman et al. (44).
Enzymatic and Chemical Treatments-Samples were digested with Pronase (1 mg/ml) in 2 mM CaC12, 50 mM Tris-HC1, pH 8.0, at 50 "C for 72 h under toluene with further addition of 0.5 mg of enzyme at 24-h intervals. Digestion of glycosaminoglycan samples with chondroitinase ABC (45), heparitinase and heparinase (46), or heparitinase I plus I1 (33) was carried out as described previously. Digestions of purified proteoglycans with Pronase, trypsin, pepsin, and papain were carried out as described previously (37). Deaminative degradation of glycosaminoglycans by nitrous acid treatment was carried out as described previously (47).

RESULTS
Cell Culture-When cells of the C-10-6 cell line were inoculated at an initial density of 3 x IO4 cells/ml and cultured in suspension in Ham's F-12 medium supplemented with 8% fetal bovine serum, they grew with a doubling time of 16.5 h and reached a saturation density of 6-7 X IO5 cells/ml on day 4 ( Fig. 1).
Extraction and Fractionation of Proteoglycans-Macromolecules were extracted and fractionated as shown in Fig. 2. A sample of 5.67 X lo9 cells yielded 1.37 g of acetone powder. Almost all the radiolabeled and hexuronate-containing macromolecules associated with the cells could be extracted with extraction buffer. When the extract was dialyzed against 4 M urea, 20 mM NaCl, 20 mM Tris-HC1, pH 7.3, a visible amount of aggregates was precipitated. This material was readily soluble in the extraction buffer. The macromolecules produced by the cells were separated into four fractions, GI, UI, US, and M. The amounts of radioactivity and hexuronate recovered in these four fractions are shown in Table I. Fraction GI did not contain a significant amount of radioactivity or hexuronate and was not studied further. Fraction M contained a large amount of proteoglycans, derived from fetal bovine serum, so the radioactivity of 35S relative to the hexuronate content was low.
For examination of the overall molecular species of macromolecules incorporating [35S]sulfate, small aliquots of fractions UI, US, and M were analyzed chemically and enzymatically. When samples treated with 0.3 M NaOH in 1 M NaBH4 were applied to a column of Sephadex G-50, their 35S radioactivity was eluted near the void volume of the column (data not shown). The samples thus obtained were digested with chondroitinase ABC and applied to the Sephadex G-50 column (Fig. 3). Fraction UI contained no chondroitinase-susceptible material (Fig. 3A), and fractions US and M contained materials corresponding to only 2% (Fig. 3C) and 5% (Fig.  3E), respectively, of the total 35S-labeled materials. Fractions containing depolymerized materials were combined, desalted, concentrated, and subjected to paper chromatography. Only one spot co-migrating with authentic ADiCS-IS was detected in both samples (data not shown), indicating that the cells synthesized a small amount of chondroitin 4-sulfate. On specific deamination of the N-sulfate group with nitrous acid, the enzyme-resistant materials eluted near the void volume were completely depolymerized (Fig. 3, B , D, and F), indicating that all the 35S-labeled materials were heparan sulfateand/or heparin-like molecules.
The 35S-labeled heparan sulfate-like material associated with cells could be separated into two fractions, UI and US, as shown above. This fractionation was quantitatively repeatable even with tracer levels of materials without a large amount of unlabeled materials. Thus, for examination of the TABLE I Distribution of 35S-labeled and he*uronate-containing macromolecules in the four fractions Small portions of the individual fractions were treated with alkaliborohydride and then digested with Pronase. The digests were brought to 7% trichloroacetic acid, and the insoluble material was removed by centrifugation. The supernatant was dialyzed against water and lyophilized. The resulting residue was dissolved in 0.5 M ammonium acetate, pH 5.5, and applied to a column (1 X 110 cm) of Bio-Gel P-6. The excluded fraction was lyophilized. The residue was dissolved in water, and its "S-radioactivity and hexuronate content were measured.    reason for this separation, the heparan sulfate-like materials in small portions of the two fractions and fraction M were subjected to domain analysis with heparitinase, which is specific for hexosaminidic linkages between GlcNAc/GlcNSO:3 and ClcUA, and heparinase, which is specific for those between GIcNSO~(6-S0*) and IdUA (48). Heparitinase treatment of the chondroitinase-resistant materials in the samples derived from UI, US, and M resulted in their almost complete conversion to oligosaccharides that were retarded on a column of Sephadex G-50 (Fig. 4, A , D, and G). The elution profile of the UI-derived oligosaccharide (eluted in tubes 50-85) was different from those of the US-and "derived oligosaecharides, which were eluted from the column in similar positions (tubes 61-88) although with some variations in their relative amounts. A large portion of the UI-derived material was resistant to heparinase and was excluded from the column ( Fig. 4 B ) whereas oligosaccharides released from the USand "derived materials were eluted in similar positions as two major peaks with partition coefficients, Kb values, of 0.15 and 0.70 (Fig. 4, E and H). After digestion with a combination of the two enzymes, the oligosaccharide generated from the UIderived material was eluted from the column as broad peaks centered at K n values of 0.37 and 0.66 (Fig. 4C) whereas the oligosaccharides released from the US-and "derived materials were eluted as sharp peaks with Kb values of 0.67 and 0.79, respectively (Fig. 4, F and I). These results indicated that the heparan sulfate-like material in fraction UI differed from those in fractions US and M, and that the materials in the latter two fractions were similar.
Purification of Heparan Sulfate Proteoglycam-All purification steps were carried out at 4 "C except dialysis, which was done on ice. When fraction US (containing 3 X lo6 cpm of 35S radioactivity, 1.3 mg of hexuronate, and 530 mg of protein) was chromatographed on DEAE-Sephacel (Fig. 5A), the flow-through fraction did not contain any radioactivity or ----72 1 1 hexuronate and was discarded. 35S-Labeled material was eluted mainly with hexuronate-containing material as a single peak with 0.45-0.6 M NaC1, separated from most nucleic acids (eluted as three peaks with maximum absorbance at 260 nm at 0.27, 0.34, and 0.44 M NaC1) and protein (mainly eluted in tubes . The position of the radioactive peak was slightly different from that of hexuronate. This was mainly due to heterogeneity in the sulfation patterns of the proteoglycans contained in fraction US, as shown later. Significant amounts of both radioactivity and hexuronate were eluted on both sides of the main peak so the three peaks indicated in Fig. 5A were pooled separately and their radioactivity and hexuronate contents were determined. Fraction M (containing 2.6 X lo6 cpm of 35S, 6.3 mg of hexuronate, and 33 g of protein) was also subjected to DEAE-Sephacel chromatography in the absence of urea (Fig. 5B). A single peak of radioactivity was eluted from the column with 0.45-0.6 M NaC1, together with about one-quarter of the hexuronate applied. The remainder of the materials containing hexuronate were eluted as two peaks with small amounts of radioactivity, one with 0.3-0.4 M NaCI, and the other with 0.6-0.7 M NaCl. The recoveries of 35S-labeled-and hexuronate-containing materials from these columns are summarized in Table 11.

CfUS1-b C(US)-c
Large amounts of proteoglycans were introduced into the culture in fetal bovine serum (38), as seen from the elution profiles of hexuronate-containing materials in Fig. 5B, so the characteristics of these exogenous proteoglycans should be summarized. Notable features of these proteoglycans (38) were as follows.  (Fig. 6). The 35S-labeled material in fraction UI was eluted " Amounts of radioactivity and hexuronate applied to the column. from the column as a single peak at KD 0.35 (referred to as C(U1)PG) (Fig. 6). The hexuronate content was not monitored during this chromatography to save material. The radiolabeled material as well as that containing hexuronate in the C(US)-b sample was separated into two fractions, C(US)PGI, eluted as a sharp, symmetrical peak at KD 0.40, and C(US)PGII eluted as a broad peak between KD 0.60 and 0.86 (Fig. 6B). The main radioactive peak (KO 0.64) of C(US)PGII was shifted significantly from the main hexuronate peak ( K D 0.72), indicating that this fraction was composed of heterogeneous molecules differing in their ratios of radioactivity of 35S to hexuronate. About 68% of the radioactivity and 81% of the hexuronate in the C(US)-b sample were recovered in fraction C(US)PGI and the rest in fraction C(US)PGII. The radiolabeled material in fraction M was eluted as a single sharp peak at K D 0.37 (referred to as MPG) whereas the material containing hexuronate was eluted as two peaks, one of which, accounting for 62% of the applied hexuronate, was eluted as a sharp peak in a slightly different position from the radioactive peak, and with a hexuronate-positive shoulder (Fig. 6C), suggesting some contamination with a fetal bovine serum-derived proteoglycan. The other material without radioactivity, which was thus derived from fetal bovine serum, was eluted as a somewhat broad peak in retarded fractions (tubes 64-77). From a calibration curve with standard proteoglycans (Fig. 6A, inset), the M, values of the radiolabeled proteoglycans with K O values of 0.35 (C(UI)PG), 0.40 (C(US)PGI), and 0.60-0.86 (C(US)PGII) from the cells and the proteoglycan from the medium with a KO of 0.37 (MPG) were estimated to be 360,000, 280,000, less than 80,000, and 320,000, respectively. The smallest proteoglycan was found only in cells but not in the medium.
Since the four fractions from Sepharose CL-4B were contaminated with a considerable amount of protein as indicated by the elution profiles of absorbance at 280 nm, they were subjected to CsCl density gradient centrifugation in 4 M guanidine HCl solution to remove protein (Fig. 7). The 35Slabeled proteoglycans in the three samples from cells cosedimented with hexuronate-containing materials to the bottom three fractions with densities of more than 1.53 g/ml, free from most contaminating proteins, which were recovered in the top three fractions (Fig. 7, A-C). Nucleic acid in the samples, with a buoyant density of 1.38-1.50 g/ml, was also separated from the bottom fractions. Most of the radiolabeled materials in the MPG sample were collected in the bottom three fractions and completely separated from protein (Fig.  70). The hexuronate-containing materials in this sample were, however, distributed in the bottom, middle, and top fractions, the latter two being serum-derived chondroitin sulfate proteoglycans. As the bottom three fractions of this sample were considered still to be contaminated with a significant amount of serum-derived proteoglycans, they were purified further. For this, these bottom fractions of the MPG sample were pooled and digested with proteinase-free chondroitinase ABC. On Sepharose CL-4B (Fig. 8A), 97% of the radioactivity and 87% of the hexuronate of the digest were co-eluted from the column as a sharp, symmetrical peak at KO 0.37, which was the same KO value as that of the proteoglycan without enzyme digestion (see Fig. 6C). The remaining radioactive and hexuronate-containing materials were converted to small molecules by chondroitinase digestion and eluted near the total volume of the column, indicating that  35S-labeled chondroitin sulfate proteoglycans (3%) and serumderived chondroitin sulfate proteoglycans (about 10% of the hexuronate-containing materials in this sample) were removed from the main proteoglycan fraction. The main fraction was then subjected to DEAE-Sephacel chromatography (Fig. 8B). The radioactivity, hexuronate, and protein in the sample were co-eluted, indicating that the proteoglycan was reasonably purified. Thus we concluded that these four molecular species of heparan sulfate proteoglycans, C(UI)PG, C(US)PGI, C(US)PGII, and MPG, were purified to almost homogeneity.
Characterization of the Purified Heparan Sulfate Proteoglycam-The recoveries and chemical compositions of the heparan sulfate proteoglycans, C(UI)PG, C(US)PGI, C(US)PGII, and MPG, were determined (Table 111). About 80% of the radioactivity in fraction UI was recovered in fraction C(U1)PG whereas 36 and 14% of those in fraction US were recovered in fractions C(US)PGI and C(US)PGII, respectively, and 57% of that in fraction M was recovered in fraction MPG. Moreover, 94, 26, 6 , and 8% of the hexuronate in individual original fractions were recovered in fractions C(UI)PG, C(US)PGI, C(US)PGII, and MPG, respectively. The ratios of radioactivity of 35S to hexuronate content were significantly different in these four proteoglycans. Glucosamine was the only hexosamine in all the cell-derived proteoglycan samples. On the other hand, the MPG sample contained a significant amount of galactosamine (about 6% of the total hexosamine). In all these samples, the molar ratio of glucosamine to hexuronate was approximately 1, indicating that these glucosamines were derived from glycosaminoglycan chains of these purified proteoglycans. This result also indicated that the glycosaminoglycan side chains of all the purified proteoglycans were exclusively heparan sulfates. The amino acid composition of the C(US)PGI proteoglycan, which could be obtained in relatively large amount, was analyzed after hydrolysis of an aliquot with 6 M HCl (Table IV). Serine and glycine accounted for 80% of the total amino acids.
The C(US)PGI proteoglycan had a very unbalanced amino acid composition and its susceptibilities to four proteinases, Pronase, trypsin, papain, and pepsin, which are often used to test the susceptibilities of various proteoglycans to proteinases, were examined and compared with those of the two other large proteoglycans, C(U1)PG and MPG. Small aliquots of the purified samples were digested with these proteinases, the digests were chromatographed on Sepharose CL-4B, and the radioactivities of the eluates were monitored (Fig. 9). Chondroitin sulfate proteoglycan H from chick embryo cartilage (50) was run as a positive control that is susceptible to these proteinases. Intact proteoglycan H was eluted in the void volume, and its chondroitin sulfate released by p-elimination was eluted as a broad but symmetrical peak centered at KD 0.61 (Fig, 9A). Pronase digestion gave a very similar elution profile to that obtained after p-elimination (Fig. 9B). The other three proteinases also degraded proteoglycan H, but to

Comparison of amino acid composition of the purified proteoglycans C(US)PGI with those of heparinproteoglycan of rat serosal mast cells (la), chondroitin sulfate E proteoglycan of mast cells from mouse bone marrow (49), and hybrid proteoglycan containing chondroitin sulfate
Di-B and heparin from rat basophilic leukemia cells (16) Residues"/1,000 amino acids  different degrees (Fig. 9, C, D, and E ) . In the p-elimination reaction, the proteoglycans C(UI)PG, C(US)PGI, and MPG released glycosaminoglycans, which were eluted from the column as sharp, symmetrical peaks a t KD values of 0.64, 0.68, and 0.66, respectively (Fig. 9, F, K , and P ) . On the other hand, treatments with the three proteinases, Pronase, trypsin and pepsin, did not cause any detectable degradation of any of these proteoglycans (Fig. 9, G-I, L-N, and Q-S) although papain caused their partial degradation (Fig. 9, J , 0, and T ) .
These results showed unexpectedly that not only the C(US)PGI proteoglycan but also the other two proteoglycans, C(U1)PG and MPG, were highly resistant to these proteinases. The elution profiles of 35S-labeled C(US)PGII proteoglycan on the same column of Sepharose CL-4B before and after the p-elimination reaction were not significantly different (Fig. lo), indicating that this proteoglycan has only a small peptide, if any.
For characterization of glycosaminoglycan moieties of the purified proteoglycans, the glycosaminoglycan side chains were released by p-elimination, and aliquots (1 pg of hexuronate) of the released samples were subjected to two-dimensional electrophoresis on a cellulose-acetate membrane (Fig. 11). Alcian blue-reactive materials in all the samples co-migrated with "S-labeled materials. The glycosaminoglycans derived from the three proteoglycans, C(UI)PG, C(US)PGI, and MPG, gave single compact spots without any tailing in a position coinciding with that of bovine kidney heparan sulfate used as an external reference, indicating that these glycosaminoglycans were heparan sulfates with very homogeneous molecular sizes and charge densities. On the contrary, the heparan sulfate of the C(US)PGII proteoglycan migrated as a single but rather broad spot with tailing. The glycosaminoglycans in the tailing portion seemed to have higher specific radioactivity than that of the portion in the head of the spot, staining strongly with Alcian blue, suggesting that the heparan sulfate was heterogeneous not only in molecular size but also in charge density. These results were consistent with those obtained by gel filtration of the heparan sulfates on Bio-Gel A-0.5m and by HPLC analysis of their disaccharide units as described below.
The heparan sulfates from proteoglycans C(UI)PG, C(US)PGI, and MPG were eluted from a Bio-Gel A-0.5m column as sharp symmetrical peaks of both radioactivity and hexuronate at KD values of 0.22, 0.26, and 0.23, respectively (Fig. 12, A , B, and D, respectively), indicating that the chain sizes of these heparan sulfates were all very homogeneous. From a calibration curve (inset in Fig. 12A), their apparent M , values were estimated to be 32,000, 27,000, and 30,000, respectively. On the contrary, the heparan sulfate of the small proteoglycan C(US)PGII was distributed over the wide M , range and separated into three peaks with KD values of 0.27, 0.42, and 0.63 (estimated M , values of 26,000, 11,000, and 4,000, respectively; Fig. 12C). Moreover, the ratios of radioactivity of 3sS to the hexuronate content of these three glycosaminoglycans appeared to differ as suggested by the results of electrophoresis. These results showed that the heparan sulfate of the C(US)PGII proteoglycan was composed of components with heterogeneous sizes and charge densities.
For characterization of repeating disaccharide units consisting of heparan sulfate chains, the heparan sulfate samples were digested with a combination of heparitinases I and 11, and the degrees of depolymerization of the four heparan sulfates in the digests were estimated by HPLC on a TSKgel column with monitoring of the refractive index (33). Results showed that all the heparan sulfates were almost completely converted to unsaturated disaccharides (data not shown). The digests were then subjected to HPLC on an amine-bound silica column with a linear gradient of 16 mM to 800 mM NaH2P04 (Fig. 13). Almost all the disaccharides (86-94% of the individual samples) were eluted with the same retention times as those of the six authentic disaccharides (Fig. 13E). The remaining minor unidentified compounds, probably with unsaturated hexuronic acid at their nonreducing ends, which were released significantly from the C(US)PGII sample, were eluted as four unidentified peaks, X1-X4 (Fig. 13C), and some of these compounds were also released from the other samples (Fig. 13, A , B, and D). The molar amounts of the six disaccharides of the individual samples identified were calculated to be shown in Table V

DiSCUSSION
The present study demonstrated that Friend erythroleukemia cells of a C-10-6 cell line, which can be differentiated to produce hemoglobin by dimethyl sulfoxide treatment (32), synthesized sulfated glycosaminoglycans consisting mainly of heparan sulfate (more than 95%) with a small amount of chondroitin 4-sulfate. Examination of other cell lines of Friend erythroleukemia cells including the dimethyl sulfoxide-and hexamethylenebisacetamide-sensitive cell lines, TSFAT-3 (32) and DS19SC4 (51), and the insensitive cell lines, C-9-9 (32) and DRlO (52), showed that all these cell lines synthesized these two types of sulfated glycosaminoglycans although the relative ratios of the two types varied significantly in different lines of cells (data not shown). Thus, regardless of their difference in sensitivity to chemical inducers, Friend erythroleukemia cells clearly synthesized heparan sulfate as a major sulfated glycosaminoglycan and chondroitin sulfate as a minor component.
The heparan sulfate produced by C-10-6 cells occurred as proteoglycans. During the process of isolation of the proteoglycans, the cell-associated component could be separated into  urea-insoluble (UI) and urea-soluble (US) fractions. This separation was quantitatively repeatable even with tracer levels of material without addition of a large amount of unlabeled material. The heparan sulfate of the UI proteoglycan differed from that of the US component in susceptibilities to heparitinase, heparinase, and a combination of the two enzymes. This finding, together with the facts that the C(U1)PG proteoglycan in the crude fraction readily redissolved in 4 M guanidine HC1 and also after purification readily dissolved in 4 M urea solution, suggests strongly that this proteoglycan is associated, probably by specific interaction of its heparan sulfate chains, with some material(s) in low salt solution and forms an aggregate(s).
On the basis of the ratios of the radioactivity of 35S to hexuronate content of the purified proteoglycans, the amounts of the four heparan sulfate proteoglycans, C(UI)PG, C(US)PGI, C(US)PGII, and MPG, present in cultures of 5.7 X lo9 cells/l2 liter were calculated to be 114, 597, 191, and 841 pg of hexuronate, respectively. The three former values represent the pool sizes of the respective proteoglycans in cells in the logarithmic growth phase whereas the fourth represents the amount of the proteoglycan secreted and accumulated in the medium during the period of cell growth (80 h) because no further metabolism of MPG proteoglycan was observed in the present culture system.' The content of total cell-associated heparan sulfates in C-10-6 cells (16 pg of hexuronate/lO' cells) is comparable to those of sulfated glycosaminoglycans in other hemopoietic cells such as rat basophilic leukemia cells (21.7 pg of hexuronate/lO' cells) (53) and mouse bone marrow-derived mast cells (32-48 pg of hexuronate/lO' cells) (15) but much lower than that of connective tissue mast cells (1,800 pg of hexuronate/lO' cells) (17). In comparison with the amount of protein in other glycoconjugates of cells of the erythroid lineage, the content of the C(US)PGI proteoglycan in C-10-6 cells (1.9 pg of core protein/lO' cells, calculated from the data shown in Table  VII), is comparable to that of glycophorin A in human erythrocytes (3.5 pg of glycophorin A protein/lO' cells) or K-562 cells (3.9 pg of glycophorin A protein/lO' cells) (54).
The C(US)PGI proteoglycan was demonstrated to have a characteristic amino acid composition with very high contents of serine and glycine and to be highly resistant to treatments with various proteinases. The core peptide was calculated to be composed of 189 amino acids with an M, of about 17,000. Core peptides having such unbalanced amino acid compositions and small M , values have been found in heparin proteoglycan of rat serosal mast cells (17)(18)(19), chondroitin sulfate E proteoglycan of mouse bone marrow-derived mast cells (49), a hybrid type proteoglycan with chondroitin sulfate Di-B and heparin-like glycosaminoglycan of rat basophilic leukemia cells (16), and other proteoglycans in several peripheral blood cells (55)(56)(57). These proteoglycans are present in storage granules of these hemopoietic cells and are highly resistant to degradation by a number of proteinases. The latter property is attributed to a high degree of sulfation (or an acidic nature) of their glycosaminoglycans, a high degree of substitution of these glycosaminoglycans along the core peptides, and the existence in their core peptides of clusters of serine-glycine M. Okayama, K. Oguri, K. Yoshida, and T. Ohkita, unpublished observation.
repeats as glycosaminoglycan attachment domains (8, 18, 19,  24). Similar unique clusters have been found recently in the amino acid sequences deduced from the cDNAs of core peptides of storage granule proteoglycans of hemopoietic cells such as rat basophilic leukemia cells (27) and human promyelocytic leukemia cells of an HL-60 cell line (28). Taken together, the results obtained in the present study suggest strongly that the core peptide of the C(US)PGI proteoglycan contains a cluster of serine-glycine repeats. Heparan sulfate proteoglycan containing this kind of core peptide has not been reported previously. Moreover, the present study demonstrated that the core peptides of two other large heparan sulfate proteoglycans, C(U1)PG and MPG, were very similar, if not identical, to that of the C(US)PGI proteoglycan.
The properties of core peptides described above appear to be common to storage granule proteoglycans of hemopoietic cells (24). However, it should be noted that such core peptides are not limited to the storage granule proteoglycans but are also found in the proteoglycans present in the extracellular matrices of a rat yolk sac tumor (58) and rabbit bone marrow, a hemopoietic organ (37). Proteoglycans in the extracellular matrix of hemopoietic organs are considered to be produced and secreted into the matrix by nonhemopoietic cells, i.e. hemopoiesis-supporting stromal cells (37, 59-61). In this regard, it is noteworthy that the yolk sac is the first hemopoietic environment in embryonic development, where the blood islands are formed. Thus, it is more likely that the above structural characteristics of core peptides, regardless of the molecular species of glycosaminoglycans linked to them, are common to hemopoiesis-related proteoglycans rather than to the storage granule proteoglycans of hemopoietic cells. This concept may be extended to cell surface proteoglycans because our preliminary results indicated that some heparan sulfate proteoglycans produced by C-10-6 cells are located on their cell surface (62).
The small C(US)PGII proteoglycan, which had a very small, degraded peptide, was found only in the cellular fraction. The heparan sulfate released by p-elimination of this proteoglycan consisted of smaller, heterogeneous molecular species, which was shown previously to be formed by degradation of the heparan sulfates of the cell-associated large proteoglycan, C(U1)PG and/or C(US)PGI (63). Thus, the C(US)PGII proteoglycan is a degradation product formed by degradation of both a core protein and heparan sulfate chains. Moreover, the results indicate that degradation of the heparan sulfate proteoglycan(s) occurs in the cellular compartment in the present culture system as shown for that of heparin proteoglycan in cultured mouse mastocytoma cells (64).
On digestion with a combination of heparitinases I and 11, all the heparan sulfates of the four proteoglycans were depolymerized almost completely and converted to identified six unsaturated disaccharides. The total recoveries of individual samples as these unsaturated disaccharides (85.5-94.0%) were almost the same as those of 35S-labeled samples (83.2-91.0%). The disaccharide with an N-sulfated glucosamine residue accounted for 31-47% of the total disaccharides produced, indicating that the four glycosaminoglycans can be classified as heparan sulfate, not heparin (48). The average molar ratios of sulfate residues to repeating disaccharides of the heparan sulfates (0.54-0.70) except the C(US)PGII-heparan sulfate (0.97) exhibit much lower degrees of sulfation than those of storage granule-type glycosaminoglycans; the glycosaminoglycans of storage granule proteoglycans usually have nearly equal or higher ratios of sulfate groups to repeating disaccharide units, as seen in fully sulfated chondroitin 4-sulfate, chondroitin sulfate E, chondroitin sulfate Di-B, and heparin.
These high degrees of sulfation are considered to be important for the interactions of these proteoglycans with other granule proteins to form complexes (24). The sulfation degree of the degraded smaller C(US)PGII-heparan sulfate was significantly higher than those of three other heparan sulfates. This finding leads us to speculate the occurrence of two possible degradation ways to produce such highly sulfated heparan sulfate fragments: one is a degradation coupled with a further sulfation reaction, and the other is a degradation coupled with a specific cleavage or elimination reaction of nonsulfated and/ or lower sulfated portions of the heparan sulfate chains. In regard to the former possibility, it is noteworthy to refer to the internalization of the cell surface heparan sulfate proteoglycan observed in the cultured rat hepatocytes, which proceeds in association with a degradation of the heparan sulfate, coupling with a further sulfation specific to a position 2 of glucuronic acid residues (65). This processing is implicated in the control of cell growth (66, 67).
Our previous studies indicated that some cell-associated heparan sulfates are located on the cell surface and disappear during differentiation of C-10-6 cells into erythroblasts induced by treatment with dimethyl sulfoxide (62). This finding is consistent with a histochemical observation that early proliferating cells of the erythroid lineage were enveloped by sulfated glycosaminoglycan, whereas cells in later stages are not (68) because Friend erythroleukemia cells are virus-transformed cells whose differentiation is believed to be blocked in an early stage of the erythroid lineage (29). These findings are also consistent with a fact that no sulfated proteoglycans have yet been found in erythrocytes (69). Taken together, the above results suggest that all heparan sulfate proteoglycans described in the present study are expressed only transiently in an early stage(s) of erythroid differentiation and are involved in regulation of cellular activities such as differentiation, proliferation, and adhesion of cells, as shown for other types of heparan sulfate proteoglycans produced by nonhemopoietic cells (48).