Swarm Rat Chondrosarcoma Proteoglycans PURIFICATION OF AGGREGATES BY ZONAL CENTRIFUGATION OF PREFORMED CESIUM SULFATE GRADIENTS*

A procedure utilizing zonal centrifugation on pre- formed cesium sulfate gradients (0.15 to 0.50 M) is described for purifying the aggregate fraction from car- tilage proteoglycan samples. The procedure permits rapid (6 h) separation of aggregates from nonaggre- gated proteoglycans at centrifugation temperatures of 2 to 11 “C with a high loading capacity (up to 100 mg of aggregate in one preparative run). Hexuronic acid analyses are used directly on gradient fractions without interference to monitor proteoglycan distributions. For the proteoglycan fraction obtained from the Swarm rat chondrosarcoma, using an associative extraction procedure and an associative isopycnic density gradient purification step, about 75% of the total hexuronic acid was recovered in aggregates and about 25% in nonag- gregated proteoglycans. Aggregating monomers were obtained from purified aggregate by centrifugation in an isopycnic dissociative CsCl density gradient. The nonaggregated proteoglycans were fractionated further on the Cs2S04 zonal gradients into slower and faster sedimenting populations. Each of the nonaggregated populations as well as the aggregating popula- tion was subfractionated on dissociative glycerol zonal

A procedure utilizing zonal centrifugation on preformed cesium sulfate gradients (0.15 to 0.50 M) is described for purifying the aggregate fraction from cartilage proteoglycan samples. The procedure permits rapid (6 h) separation of aggregates from nonaggregated proteoglycans at centrifugation temperatures of 2 to 11 "C with a high loading capacity (up to 100 mg of aggregate in one preparative run). Hexuronic acid analyses are used directly on gradient fractions without interference to monitor proteoglycan distributions. For the proteoglycan fraction obtained from the Swarm rat chondrosarcoma, using an associative extraction procedure and an associative isopycnic density gradient purification step, about 75% of the total hexuronic acid was recovered in aggregates and about 25% in nonaggregated proteoglycans. Aggregating monomers were obtained from purified aggregate by centrifugation in an isopycnic dissociative CsCl density gradient. The nonaggregated proteoglycans were fractionated further on the Cs2S04 zonal gradients into slower and faster sedimenting populations. Each of the nonaggregated populations as well as the aggregating population was subfractionated on dissociative glycerol zonal gradients into two or three groups according to differences in their average sedimentation rates. Analyses of the subfractions for (a) antigenicity with antibodies directed against either proteoglycan monomer or the purified hyaluronic acid-binding region, (b) average chondroitin sulfate chain sizes, and (c) chemical composition suggest: 1) polydispersity of aggregating monomers is primarily the result of differences in the average chondroitin sulfate chain sizes on different molecules; 2) about 15% of the aggregating monomers contain a discretely smaller core which was observed after chondroitinase digestion (Kimata, K., Hascall, V. C., and Kimura, J. H. (1982) J. Biol  possibly representing a selective breakdown product of the major population; 3) the nonaggregated proteoglycan fraction of smallest size (about 10% of the original sample is distinctly different from aggregating monomers; 4) the nonaggregated proteoglycans of larger size (about 15% of the original sample) may be related to aggregating monomers but still have some compositional and antigenic differences; and 5) the extent of polydispersity of the largest nonaggregated proteoglycan fraction is similar to that of the aggregating monomers and is likewise primarily related to differences in the average chondroitin sulfate chain sizes on different molecules.
Proteoglycans are present in cartilages primarily as high molecular weight aggregates (1) in which proteoglycan monomers interact specifically with hyaluronic acid and link proteins (2)(3)(4)(5). The monomers contain a central core protein having structural and functional polarity. Chondroitin sulfate chains are covalently attached a t one end (about 60% of the total), while the other end  contains the hyaluronic acid-binding region essential in the formation of the aggregates. Other oligosaccharide chains and usually keratan sulfate chains are also present on the core protein (6)(7)(8).
Aggregates are preferentially recovered in the bottom (Al) fraction after equilibrium sedimentation on associative CsCl density gradients (3,9).' However, usually about 25%, but occasionally 50% or more of the proteoglycans in A1 fractions, are not aggregated. Several biochemical and electron microscopic studies (6, 9, 12-17, among others) have shown that proteoglycan monomers derived from an A1 fraction are very heterogeneous with an average M , = 2.5 x lo6. On the basis of the analyses of lower buoyant density fractions of monomers derived from dissociative density gradients of Al, a model has been proposed suggesting that the population of proteoglycans that aggregate contains a constant hyaluronic acid-binding region and an adjacent keratan sulfate-rich re-' The abbreviations used for identifying proteoglycan fractions follow the systematic nomenclature of Heineglrd (10) where AI indicates the proteoglycan fraction which contains aggregates, isolated from the bottom of an associative density gradient; D l is the monomer fraction isolated from the bottom of a dissociative gradient. In addition, aAl are aggregate fractions isolated from the Swarm rat chondrosarcoma directly from associative extracts without a prior dissociation step (11); Agg and Agg-Dl are aggregates from the bottom of an associative CsBSOl zonal gradient and its monomer fraction, respectively; NA-S and NA-F are subfractions of nonaggregated proteoglycans from an associative Cs2SOI gradient which sedi-gion but that the chondroitin sulfate-rich regions of the core are variable in size (6,15,18). However, more recently, evidence has been presented suggesting that the aggregating proteoglycan population from most hyaline cartilages contains two discrete subpopulations (19).
Earlier studies using sequential extraction with low and subsequent high ionic strength solvents demonstrated that cartilages also contain proteoglycans which are not aggregated (18,20). However, the low ionic strength solvents without protease inhibitors may have yielded some degradation products derived from aggregating proteoglycans. A recent report (21) described the properties of nonaggregated proteoglycans separated from the A1 fraction of bovine nasal cartilage, using repetitive sedimentation velocity centrifugation and precautions to prevent proteolytic degradation. This fraction constituted about 10% of the total proteoglycan and the molecules appeared to lack the hyaluronic acid-binding region. More recently, Hoffman (22) obtained a similar nonaggregated monomer fraction from the A1 fraction of the same tissue using associative rate zonal sedimentaiion in a preformed NaCl density gradient similar to the method described originally by Franek and Dunstone (23). These studies demonstrate that A1 fractions as currently isolated contain nonaggregated proteoglycans. The fist purpose of this paper is to describe a rapid, convenient, and high capacity method for separating and purifying the proteoglycan aggregate population from nonaggregated proteoglycans using zonal centrifugation on a preformed Cs2S04 gradient. The second purpose is to describe the characteristics of aggregating and nonaggregating proteoglycans purified from the aAl fraction obtained from the Swarm rat chondrosarcoma using this method.

EXPERIMENTAL PROCEDURES
Materials-Ultrapure Cs2S04 was purchased from Sigma. Ultrapure guanidine HCl and CsCl were from Schwarz/Mann; Bio-Gel A-1.5m (200-400 mesh) was from Bio-Rad; polyvinyl and polystyrene microtiter plates were from Cook Laboratories; and horseradish peroxidase conjugated to goat anti-rabbit IgG and pig skin hyaluronic acid were from Miles Laboratory. High molecular weight hyaluronic acid from rooster comb was kindly provided by Seikagaku Kogyo, Japan. Molecular weight markers for chondroitin sulfate (M, values of 25,400, 19,200, and 12,400 with K,, values on Sepharose 6B of 0.45, 0.51, and 0.59 (24)) were a kind gift from Dr. A. Wasteson, Uppsala,

Sweden.
A proteoglycan fraction, d l , was isolated from the Swarm rat chondrosarcoma by associative extraction and direct associative density gradient centrifugation as described previously (11). Approximately 70% of the proteoglycans in this preparation are aggregated. The monomer, aA1-Dl, fraction was prepared from the aAl with dissociative density gradient centrifugation as described previously (11).
The complex of hyaluronic acid-binding region, link protein, and hyaluronic acid was purified after clostripain digestion of aAl as described elsewhere (25). The hyaluronic acid-binding region was then isolated from this complex by chromatography on Sepharose CL-GB in 4 M guanidine HCl (4,25).
Antibodies were raised against aAl-DI in rabbits and characterized as described in detail elsewhere (26,27). Antibodies were raised in rabbits against the complex of hyaluronic acid-binding region, link protein, and hyaluronic acid; the characteristics of this antiserum will be described in detail elsewhere (28). the gradients were centrifuged in a Beckman SW 27.1 rotor at 25,000 rpm for 6 h at 11 * 2 "C unless otherwise noted. The tubes were then fractionated into 0.6-ml fractions with an ISCO density gradient fractionator Model 640 using 2.5 M Cs2S04 in 0.1 M sodium acetate-0.1 M Tris-HC1, pH 7.2, as a displacing solution. For preparative runs, 32ml linear gradients were formed on 3-ml cushions in cellulose nitrate tubes for the SW 27.1 rotor. In these cases, 3.0-ml samples were applied.
Separation of Aggregate from Nonaggregated Proteoglycans in aA-Aliquots of an aAl fraction, about 8 mg/ml (2 mg/ml as hexuronic acid) were centrifuged on preparative Cs2S04 zonal gradients as described above. The weight of proteoglycan was calculated by assuming that 25% by weight is hexuronic acid. About 70% of the hexuronic acid was recovered as aggregate (Agg) in the fractions near the bottom cushion as described under "Results." The nonaggregated proteoglycans in the upper fractions were recovered and dialyzed, and high molecular weight hyaluronic acid from rooster comb w s then added to give a ratio of proteoglycan to hyaluronic acid ofab&t 50:l. After 3 h at 10 "C, the mixture was applied to Cs2S04 zonal density gradients and recentrifuged as described above. The hexuronic acid distributing in the upper fractions as nonaggregated proteoglycan was recovered for further study (see Fig. 3a, below).
Preparation of Aggregating Monomers from Purified Aggregate-The aggregate fraction recovered at the bottom of the preparative gradients, see Fig. 3e below, was desalted by concentration on an Amicon PM-10 membrane at 4 "C and dilution with 0.2 M Tris-HC1-0.2 M sodium acetate, pH 7.2. An equal volume of 8 M guanidine HC1 was then added, bringing the solution to 4 M guanidine HCl-0.1 M Tris-HC1-0.1 M sodium acetate, pH 7.2. Then 0.55 g of CsCl/g of solution was added to give an initial density of 1.5 g/cc. A dissociative isopycnic density gradient was prepared by centrifugation at 40,000 rpm for 50 h at 11 "C in a Beckman Ti-50 rotor. The tubes were then divided into four approximately equal fractions (Agg-Dl through Agg-D4).
Rate Zonal Sedimentation on Dissociative Glycerol Density Gradients-Samples were centrifuged on dissociative glycerol zonal density gradients essentially as described elsewhere (29,30). Briefly, I-ml samples in 4 M guanidine HC1-0.02 M Tris-HC1-0.01 M sodium EDTA, pH 8.0, were layered on 13-ml linear gradients of glycerol (IO-35%. w/ v) formed in the 4 M guanidine HC1 solvent on a cushion of 1 ml of 50% glycerol in the same solvent. The gradients were centrifuged in a Beckman SW 27.1 rotor at 25,000 rpm for 25 h at 19 * 2 "C.
Fractionation was done with an ISCO fractionator as described above for the associative zonal gradients. Each fraction was dialyzed against 0.1 M sodium acetate-0.1 M Tris-HC1, pH 7.2, to remove guanidine HCl and glycerol before further analyses.
Molecular Sieve Chromatography of Glycosaminoglycans-Proteoglycans in pooled subfractions (100-500 pg) were precipitated by adding ethanol and concentrated potassium acetate to final concentrations of 70% ethanol (v/v) and 1% potassium acetate (w/v). The precipitates were collected by centrifugation. Each of the precipitates was dissolved in 300 p1 of 0.1 M NaOH-0.4 M NaBH4 and kept at 37 "C for 24 h to release glycosaminoglycan chains (31). The solutions were slightly acidified with 30 pl of 1 M acetic acid. After the addition of 25 pl of 1% (w/v) phenol red to provide a marker for the total column volume, each sample was applied to a Bio-Gel A-1.5m column (0.5 X 47.6 cm) equilibrated with 0.4 M ammonium acetate, pH 5.0, and 0.57ml aliquots were collected. The standard molecular weight markers of chondroitin sulfate were eluted with peak fractions having K., = 0.27, 0.36, and 0.49, corresponding to M, = 25,400, 19,200, and 12,400, respectively.
Antigenic Activities of Proteoglycan Fractions-Enzyme-linked immunosorbent assays were used as described elsewhere (26) to quantitate the core protein and hyaluronic acid-binding region.' Purified aAI-DI (2 pg/well) or hyaluronic acid-binding region (200 ng/ well) was used to coat polyvinyl microtiter plates. Aliquots of 110 p1 for samples or standards were diluted with phosphate-buffered saline containing 0.05% (v/v) Tween 20, pH 7.0. The dilutions were mixed with 110 pl of a 1/3000 dilution of antibody solution (with antibodies directed against either aAl-DI or the hyaluronic acid-binding region). After incubation at 4 "C overnight in a humid atmosphere, 200-pl aliquots of the mixtures were transferred into the antigen-coated microtiter wells. After incubation for 30 min at room temperature on a rocking platform, the solutions were removed and the wells washed with phosphate-buffered saline-0.05% Tween 20, pH 7.0. Goat antirabbit IgG antiserum conjugated with horseradish peroxidase (200 pl, 1/ 1OOO dilution) was then added as a second antibody. After 90 mi n, the plates were washed three times with the above buffer, and by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 200 p1 of 0-phenylenediamine-0.03% Hz02 (v/v) were added as substrate. The color was allowed to develop at room temperature for 90 min, and the reaction was stopped by adding 50 pl of 8 M HzS04 to each well. The color was measured spectrophotometrically at 492 nm with a Titertek Multiskan plate reader. The data were analyzed using the method of Rodbard (32).
Analytical Methods-Amino acid and hexosamine analyses were done with a Durrum D-500 amino acid analyzer as described previously (25).
Hexuronic acid contents were determined by an automated procedure using the carbazole method (33). Sialic acid was measured by an automated procedure (7) using the resorcinol method of Jourdian et al. (34).

RESULTS
Parameters of Centrifugation-Standard conditions were developed for rate zonal sedimentation of proteoglycan preparations on preformed Cs2S04 linear gradients. The use of Cs2S04 for the gradient provides a satisfactory density stability over a relatively narrow electrolyte concentration range and permits direct analysis for hexuronic acid by both manual and automated procedures without interference, in contrast to other rate zonal sedimentation methods using sucrose (29), glycerol (30), or halide salts (22, 23) for the gradient.
Different times of centrifugation were tested in which identical samples of aAl were centrifuged at 25,000 rpm and 11 "C. The amount of hexuronic acid in the slower sedimenting fractions relative to that in the aggregate fractions decreased with time until about 5 h. After that, there was no significant decrease in the proportion of hexuronic acid in the slower sedimenting peak, although the peak position migrated further into the gradient with time (data not shown). The 6-h time point was selected for further experiments, since, as indicated in Fig. 1, bottom, base-line hexuronic acid values separate the two components.  The experiment shown in Fig. 2 tested both the effect of solute concentration and temperature on the sedimentation of aA1-Dl monomer in the zonal gradient under standard conditions. Concentrations of proteoglycan between 0.4 and 2.0 mg/ml as hexuronic acid gave similar profiles. At higher concentrations, the macromolecules sedimented somewhat more slowly, and there was an asymmetrical broadening of the profiies. At the lower temperature, 4 "C, the profiles were essentially the same as at 11 "C except that the peak fractions occurred about one fraction earlier at each concentration. The results indicate that the gradients have a high solute capacity, achieving satisfactory results with up to 8 mg of proteoglycan (2.0 mg of hexuronic acid)/ml in the applied sample, and that the method is relatively insensitive to differences in temperature. Usually, concentrations with 1.0-1.5 mg hexuronic acid/ ml were used in the experiments.
The small peak of hexuronic acid in the aggregate fractions at the bottom of the gradient for each profile in Fig. 2 represents proteoglycans bound to the small amount of hyaluronic acid often present in aA1-Dl monomer fractions (see next section).
Preparation ofAggregate and Aggregate-Dl-Preparative rate zonal centrifugation was used to isolate a large amount of aggregate from an aAl preparation from the Swarm rat chondrosarcoma. As for the samples described above (Fig. 1, bottom), about 70% of the proteoglycans were recovered in the aggregate fraction. The nonaggregated proteoglycan fraction is described below. Proteoglycan monomers in the aggregate fraction were separated from the other components of the aggregate, namely hyaluronic acid and link protein as well as from some aggregating monomers with lower buoyant densities, by centrifugation in an isopycnic dissociative (4 M guanidine HC1) gradient as described under "Methods." About 90% of the hexuronic acid was recovered in the bottom fraction which will be referred to as Agg-Dl. About 7.0, 2.5, and 1.2% Polydispersity of Aggregated and Nonaggregated Proteoglycans of the hexuronic acid were recovered in Agg-D2, Agg-D3, and Agg-D4, respectively. Aliquots of the purified aggregate and Agg-Dl were centrifuged on standard analytical zonal gradients, Fig. 3, d and  and the complex would sediment to the bottom.3 When dissociative solvents were used in the zonal gradients as described in Fig. 4 below, no proteoglycan sedimented into the bottom fractions.
Subfractionation of Nonaggregated Proteoglycans-On the preparative zonal gradients, the majority of the nonaggregated proteoglycans was broadly distributed in the fractions in the upper two-thirds, similar to Fig. 1, bottom. It is possible that the nonaggregated proteoglycan fraction could contain proteoglycans which are able to interact with hyaluronic acid but which were not recovered in the aggregate due to insufflcient amount of hyaluronic acid in the aA sample. Such potentially aggregating proteoglycans should be recovered in the bottom fraction of zonal gradients after adding exogenous high molecular weight hyaluronic acid.3 The nonaggregated proteoglycan fraction was incubated for 3 h at 10 "C with 2% hyaluronic acid, and the mixture was applied to the standard analytical zonal gradients and recentrifuged.
Less than 5% of the total hexuronic acid was recovered in the aggregate fraction, Fig. 3a, indicating that the proteoglycans in the nonaggregated fraction did not bind to hyaluronic acid to any significant extent. However, since an asymmetrical sedimentation profile was observed, the preparation was subdivided into two fractions: a slower sedimenting fraction corresponding to fractions 1-8 and a faster sedimenting fraction corresponding to fractions 9-16 in Fig. 3a standard analytical gradients gave the reproducible sedimentation profiles, Fig. 3, b and c, indicating that the two subfractions clearly differed in their sedimentation rates. A large proportion of the NA-F fraction sedimented in the same region as proteoglycans in the Agg-Dl preparation. However, there was marked broadening of the profile and some of the macromolecules sedimented slower and some faster than proteoglycans in Agg-Dl, indicating that NA-F contains proteoglycans with a broad range of sizes. The majority of the macromolecules in NA-S had slower sedimentation rates than those in the NA-F and the Agg-Dl fractions. Size Subfractionation of Proteoglycans in Dissociative Zonal Gradients-Aliquots of aggregate and Agg-Dl were dissociated in 4 M guanidine HCl and centrifuged in dissociative zonal gradients made of 15-35% glycerol in 4 M guanidine HCI-0.02 M Tris-HC1, pH 8.0 (30). Sedimentation rates of proteoglycans on this gradient correspond roughly to their hydrodynamic sizes (29). Both samples gave essentially the same broad sedimentation profile of hexuronic acid as shown in Fig. 4, indicating the polydispersity of aggregating monomers. Subfractions from the aggregate and the Agg-Dl profiles were then isolated as indicated in Fig. 4 and referred to as (I, /3, and y, according to their sues.
Aliquots of NA-S and NA-F were also centrifuged on dissociative zonal gradients in the same way. NA-S gave two peaks of hexuronic acid which were almost equivalent in amount, referred to as (Y and /3 as shown in Fig. 4. The NA-S-(Y was the slowest sedimenting proteoglycan fraction and contained approximately 11% of the total hexuronic acid in the original aAl fraction. The NA-S-/3 fraction also sedimented somewhat slower than the aggregating monomers. These results indicate that the hydrodynamic sizes of proteoglycans in NA-S are definitely smaller than those of aggregating monomers. The NA-F fraction gave a broad distribution which was subfractionated into three fractions. The slowest, NA-F-(I, was similar in profile to the NA-S-/3 fraction while the main peak, Na-F-/3, distributed almost in the same fractions as the main peak, Agg-Dl+, of the aggregating monomers. Immunological Differences of Subfractions-An antiserum raised against aA1-Dl from the rat chondrosarcoma and described in detail elsewhere (26) was used to compare the antigenicity of the different proteoglycan subfractions (Table   I and Fig. 5). The purified aggregate and the Agg-Dl monomer fractions gave nearly the same ratio of antigenicity to hexuronic acid content, Table I, although the value for Agg-Dl was reproducibly slightly higher, perhaps because some antigenic sites accessible in monomer are masked in aggregate. It was difficult to determine exact antigenic activity in these cases because dilutions of these fractions gave inhibition curves different from the standard, causing a nonlinear relationship between dilution and antigenic activity (see Fig. 5b). Therefore, the dilution which yielded about 50% inhibition was selected for the calculation. For a constant dilution of antiserum, equivalent dilutions on the basis of hexuronic acid contents of the original antigen aA1-Dl and of the Agg-Dl monomers gave almost identical absorption curves, Fig. 5a, indicating that almost all of the antigens recognized by the antiserum are present in the aggregating monomer population. The ratio of antigenicity to hexuronic acid content was much less for the NA-F proteoglycan fraction compared to Agg-Dl and even less for the NA-S fraction ( Table I). The absorption curves for these fractions, as shown in Fig. 5b, are displaced greatly toward solutions with more concentrated solutes but show a similar range of absorption. When a purified sample of hyaluronic acid-binding region derived from clostripain digests of aAl (25) was used to absorb the antiserum, the solid curve shown in Fig. 5b was observed. The range of absorption was only about 70% that of aA1-Dl, indicating that while most of the antigens recognized by the antiserum are directed against sites in the hyaluronic acid-binding region a significant proportion are not.
Further information was obtained when aliquots from the dissociative velocity gradients shown in Fig. 4 were tested for antigenicity (open circles, Fig. 4). There was greater antigenicity per hexuronic acid in slower sedimenting fractions of aggregate and Agg-Dl with the effect more pronounced in the fractions from aggregate. The larger nonaggregated proteoglycans in NA-S-/3 and the NA-F fractions had less antigenicity per hexuronic acid than did Agg-Dl proteoglycans, indicating some differences from aggregating monomers. The slowest sedimenting component, NA-S-a, did not exhibit any significant antigenicity in the absorption test. The results for peak fractions are summarized in Table 11.
Because the antiserum contains a large proportion of antibodies directed against the hyaluronic acid-binding region, the results suggest that the greater ratio of antigenicity to Absorption assay with anti-aAl-D1 as described in Table I.
Absorption assay with antiserum against the complex of hyaluronic acid-binding region-link protein and hyaluronic acid (l/8000) using purified hyaluronic acid-binding region (HABR) to coat the plate for ELISA.
Values are given as equivalents to the test antigen. Numbers in parentheses indicate fraction numbers from gradients shown in Fig. 4. hexuronic acid for the smaller proteoglycans from the aggregate profile (Agg-a) reflects a higher ratio of hyaluronic acidbinding region to chondroitin sulfate in these molecules. The effect would be less pronounced for the Agg-Dl profile since a proportion of the smallest, low buoyant density monomers is not recovered in the Agg-Dl fraction. More direct evidence for this was obtained by testing peak fractions in an ELISA4 for the purified hyaluronic acid-binding region. Again, there was greater antigenicity per hexuronic acid in the smaller aggregating monomer fractions, Agg-a and Agg-Dl-a (Table  11). Antigenicity per hexuronic acid was less for the NA-F fractions and much less for NA-S-P. NA-S-a had no detectable antigenicity for the hyaluronic acid-binding region.
Polydispersity of Agg-Dl-Fractions from the aggregate and Agg-Dl were analyzed for average chondroitin sulfate chain sizes after alkaline borohydride treatment by chromatography on Bio-Gel A-1.5m (Fig. 6). The starting aAl sample had a peak value of M , = 22,000 for the chondroitin sulfate chains. A small peak, 2.4% of the hexuronic acid, eluting in the excluded fraction was shown to be hyaluronic acid.5 The average chondroitin sulfate chain size increased with an increase in the sedimentation rates for both aggregate and Agg-D l fractions, i.e. a < fl< y, indicating that the polydispersity in hydrodynamic size for aggregating monomers correlates to a large extent with differences in average chondroitin sulfate chain sizes in the subpopulations. The peak chain size for Agg-a was slightly shorter than that for Agg-Dl-a, consistent with the removal of some aggregating monomer of smaller size and lower buoyant density in the dissociative CsCl gradients used to isolate Agg-Dl.
In the accompanying paper (35), the results of sodium dodecyl sulfate-polyacrylamide gel electrophoresis indicate that Agg-a would contain some link protein. Therefore, only The abbreviation used is: ELISA, enzyme-linked immunosorbent assay.
The peak was identified as hyaluronic acid by its susceptibility to Streptomyces hyaluronidase. Since about 70% of the hexuronic acid in aAl is in aggregate, the proportion of hexuronic acid in hyaluronic acid for purified aggregate is 2.4%/0.7 = -3.5%. subfractions of Agg-Dl were used for further analyses. Amino acid analyses for subfractions of Agg-Dl showed that the compositions are very similar except possibly for a small relative increase in serine and glycine for Agg-Dl-y (Table   111). The results of sodium dodecyl sulfate-polyacrylamide gel electrophoresis of chondroitinase-digested samples in the accompanying paper (35) show a predominant, single core band across a size fractionation of aggregating proteoglycans with a smaller amount of a smaller core band which is more pronounced in fractions with smaller proteoglycan molecules. If proteoglycan monomers are selectively degraded in the tumor and portions of the chondroitin sulfate attachment region are removed, this could account for the reduced relative serine and glycine contents in Agg-Dl-a and -P relative to Agg-Dl-y and for the presence of the smaller core band (see discussion in Ref.

35).
Sialic acid and glucosamine have been identified as constituents of oligosaccharides in proteoglycan monomers from the chondrosarcoma (7). The sialic acid and glucosamine contents decreased with an increase in hydrodynamic size for Agg-Dl fractions (Table IV)

Compositions of Nonaggregated Proteoglycan Fractions-
The smallest nonaggregated proteoglycan, NA-S-a, had chondroitin sulfate chains of similar size, 22,000, to those of the major aggregating monomers (Fig. 6). However, the distribution was skewed more toward smaller sizes as shown by the retardation of the hexuronic acid profiie and the high baseline for the carbazole reaction around fractions 50-60. The NA-S-a also showed a high ratio of sialic acid per chondroitin sulfate chain, Table IV, suggesting that a large proportion of oligosaccharides is present. Further, NA-S-a had a different amino acid composition from the aggregating proteoglycans, with much higher glycine and somewhat lower proline, alanine, and serine (Table 111). These results plus the absence of  Chain size analyses for NA-S-a, Fig. 6, showed hyaluronic acid in excluded fractions (&la), most, if not all, of which was derived from the exogenously added hyaluronic acid used to test for the ability of proteoglycans in the nonaggregated fraction to bind to hyaluronic acid (Fig. 3a). Therefore, the glucosamine content in NA-S-a, as well as in the other fractions, was corrected by subtracting the glucosamine due to hyaluronic acid. Calculated by multiplying the ratio (sialic acid/hexuronic acid) by the average molecular weight of chondroitin sulfate chains (see Fig. 6). e Calculated by multiplying the glucosamine/galactosmine ratio by the molecular weight of chondroitin sulfate chains as in Footnote b.
From the data of Oegema et al. (36).
antigenicity suggest that NA-S-a is a distinctly different proteoglycan species from the major proteoglycans. Subfractions from NA-F had almost identical amino acid compositions which appeared to be nearly the same as for aggregating proteoglycans, with slightly higher relative glycine content (Table 111). Sialic acid contents per hexuronic acid, Table IV, were nearly the same as those for aggregating monomers. There was also a similar increase in average chondroitin sulfate chain size with increase in hydrodynamic size although the NA-F fractions had somewhat longer chains (Fig. 6). The results suggest that the subpopulations of NA-F may differ from each other primarily in average chondroitin sulfate chain sizes on different molecules.
The NA-S-P fraction had similar molecular features to NA-F-a in chondroitin sulfate chain size, Fig. 6, in sialic acid content, Table IV, in amino acid compositions, Table 111, in size distribution on the glycerol gradients, Fig. 4, and in identical antigenic properties, Table 11. The NA-S-P fraction, then, probably contains the same subpopulation of molecules as the NA-F-a fraction.

DISCUSSION
The present study introduces a new method for the isolation of pure aggregate. Centrifugation of aAl from the rat chondrosarcoma for 6 h on the zonal gradients yielded effective separation of aggregate from nonaggregated proteoglycans. This allowed isolation from purified aggregate of an Agg-Dl fraction in which all the monomers contain a functional hyaluronic acid-binding region. The advantages of this procedure are: ( a ) greater purity is obtained than for the differential sedimentation used by Heinegird and Hascall (21), (6) more stable gradients are achieved at lower electrolyte concentrations than in the original procedure described by Franek and Dunstone (23) or the preparative procedure described by Hoffman (22), (c) hexuronic acid contents can be analyzed directly by automated methods which are not possible when halides are present, and ( d ) high loading capacities are possible (up to 100 mg of purified aggregate can be obtained in one preparative run). Other solvent conditions can be substituted for the 0.1 M Tris-HC1-0.1 M sodium acetate, pH 7.2, solvent used in this study as is demonstrated in the accompanying paper concerning mechanisms for aggregate formation (35).
Starting with the aAl fraction which contains intact aggregates isolated from the tissue without a dissociation step (111, 70-75% of the hexuronic acid was recovered in the purified aggregate. Monomers from the purified aggregate still showed a polydispersity in hydrodynamic size as shown by subfractionation on dissociative zonal gradients, clearly indicating that polydispersity is an intrinsic property of the aggregating monomers. The present study indicates that the polydispersity in hydrodynamic size is primarily related to a variation in the average size of chondroitin sulfate chains per core protein. However, we have shown in the accompanying paper (35) that monomers in the aggregate have two discrete core sizes, 80% large and 20% somewhat smaller. The latter population may represent a selective degradation product in which a piece near the outer end of the chondroitin sulfate attachment region has been trimmed off. Therefore, it is likely that two cores of different size will contribute secondarily to polydispersity in size and may account for the small differences in relative serine and glycine content for the Agg-Dl fractions.
Recent evidence from studies of biosynthesis of proteoglycans with chondrocytes from the rat chondrosarcoma in culture supports these conclusions. The core protein prior to adding chondroitin sulfate chains has been identified by double antibody immunoprecipitation using antibodies against the hyaluronic acid-binding region and has been shown to contain a functional hyaluronic acid-binding region (28). In this case, the core protein prior to adding chondroitin sulfate gives a single, uniform band of high apparent molecular weight (-370,000) on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. In a separate study using the same chondrocyte system (37), the size polydispersity of the newly synthesized, aggregating proteoglycans was shown to be primarily related to differences in average chondroitin sulfate chain size per core protein in agreement with the results described in this paper.
The smallest proteoglycan (NA-S-a), which is about 11% of the aAl fraction, clearly represents a different population from aggregating monomers. It has different hydrodynamic size, lacks cross-reacting antigenicity, has a different amino acid composition, and has a relatively higher sialic acid content. These proteoglycans may be related to the class of small proteoglycans observed by other investigators in middle fractions from density gradients used to purify proteoglycans (38, 39).
The second population of nonaggregated proteoglycans (NA-S-P and Na-F-a to NA-F-y), which are about 17% of the aAl fraction, is more difficult to define. NA-S-P and NA-F-a are somewhat smaller in hydrodynamic size than aggregating monomers, while NA-F-P and -y have a size distribution which overlaps Agg-Dl. While the proteoglycans in this population share some antigenicity with the hyaluronic acid-binding region, the proportion per hexuronic acid is much less than for aggregating monomers, and they do not appear to interact with hyaluronic acid. Their amino acid compositions are similar. This population has many similarities with the nonaggregated proteoglycans isolated from bovine nasal A1 preparations (21) and may represent a distinct class or classes of proteoglycan from the aggregating population.