Mechanisms for dissociating proteoglycan aggregates.

Preformed Cs2SO4 zonal gradients were used to purify aggregates from proteoglycan preparations derived from associative extracts of the Swarm rat chondrosarcoma. Zonal gradients were used in solvents with different concentrations of guanidine HCl and different solvent pH values to study the mechanisms for dissociating the aggregates. Aggregates are stable in concentrations of guanidine HCl up to 1.5 M at pH 6.6. At 2 M guanidine HCl, partial dissociation occurs over 20 h in which a link protein is completely dissociated for every monomer proteoglycan dissociated from the aggregate structure. This suggests that in this solvent disaggregation occurs concurrent with complete separation of link protein from monomer. At solvent pH 2.7 to 3.3 in ionic conditions which normally promote aggregation, dissociation occurs by a mechanism in which the link protein remains associated with monomer. Thus, link protein-monomer complexes dissociate as bimolecular units from hyaluronic acid; such complexes then exhibit physical properties indistinguishable from pure monomers. The link protein-monomer complexes reassociate with hyaluronic acid to form link-stabilized aggregates when the solvent pH is raised to pH 7, i.e. to associative conditions. The study provides additional evidence for the role that link protein-monomer interactions have in proteoglycan aggregate structures.

The abbreviations used are: aAl, a proteoglycan preparation isolated by using an associative (a) extraction solvent (2) followed by an associative CsCl isopycnic gradient isolating proteoglycan from the bottom (AI) fraction (3); aAI-Dl and aAl-D4, the monomer preparations and link protein-enriched preparations isolated from a A l in a subsequent dissociative density gradient either in the bottom (Dl) or top (D4) fractions. HA+0 indicates a population of oligomers prepared from partial digests of hyaluronic acid with testicular hyaluronidase and having an average length of 50 monosaccharides, proteoglycan monomers, link proteins, and hyaluronic acid (5-9). Dissociation of aggregates also occurs when the solvent pH is lowered below pH 4 in electrolyte solutions which are associative a t higher pH (5, 10). The experiments described in this paper utilized the velocity zonal centrifugation procedure to study the mechanisms for dissociating aggregates in guanidine HCI solutions and at low solvent pH. The results indicate that, while increasing concentrations of guanidine HC1 cause separation of all three components of the aggregate, low solvent pH dissociates a stable complex of proteoglycan monomer-link protein as a unit from hyaluronic acid.

EXPERIMENTAL PROCEDURES
Materials-Hyaluronidase from Streptomyces hyalurolyticus was obtained from Calbiochem. Papain (twice crystallized) and hyaluronic acid (type I) were from Sigma. Oligosaccharides of hyaluronic acid with an average of about 50 monosaccharides, HA-50, were isolated from partial digests of hyaluronic acid with testicular hyaluronidase by chromatography on Bio-Gel P-30 as described elsewhere (11). All other materials used in the experiments are described in the preceding paper (1).
Methods: Cesium Sulfate Zonal Gradients-The details for preparing preformed Cs2SO4 gradients for zonal centrifugation are given in the preceding paper (1). For some of the experiments described in this paper, the gradients were prepared in the presence of different guanidine HC1 concentrations and different solvent pH values. In each of these cases, the two Cs2S04 solutions used to prepare the gradient contained the appropriate solvent, for example, 2 M guanidine HCl, such that the solvent was the same throughout the gradient. Solute samples in each case were dialyzed or diluted into the same solvent in the absence of C&304 and maintained at 4 "C for 16-20 h prior to application onto the gradients. For low pH solvents, the 0.1 M sodium acetate, 0.1 M Tris solvent normally present in the Cs2.904 solutions was replaced with 0.2 M sodium acetate/acetic acid to give the final pH for the experiments. The final solvent conditions are described under "Results." Analysis for Hyaluronic Acid-A modification of the sensitive and specific assay described by Jourdian et al. (12) was used to measure hyaluronic acid contents of fractions. The procedure estimates the amount of the unsaturated hexuronic acid formed during digestion with the hyaluronic acid-specific eliminase from Streptomyces using the thiobarbituric acid colorimetric procedure (13). The presence of protein in the samples introduced turbidity in the 1-butanol extracts of the chromophore when the method was used as described (12). Therefore, samples from gradient fractions were treated as follows: (a) fractions were dialyzed against 0.1 M sodium acetate, pH 7.0, and then lyophilized. Otherwise, proteolgycans and hyaluronic acid in samples were quantitatively precipitated by the addition of 3 volumes of ice-cold absolute ethanol containing 1.3% (w/v) potassium acetate and the precipitates were collected by centriguation as described elsewhere (14); (b) the precipitates were dissolved in 0.3 ml of 0.1 M sodium acetate, pH 7.0,5 m~ EDTA, 5 mM cysteine HCl and digested at 60 "C for 5 h with 30 pg of papain: (c) ethanol precipitation was done as described above to remove cysteine. The precipitate was dissolved in 0.3 ml Hz0 and then the papain was inactivated by heating at 100 "C for 3 min, and the glycosaminoglycans were then precipitated with ethanol as described above; (d) the final precipitates were dissolved in 50 of 0.1 M sodium acetate buffer, pH 5.0, and digested with 5 turbidity-reducing units (nominal) of Streptomyces hyaluronidase for 5 h at 37 "C; (e) the subsequent development of the 3828 Dissociation of Proteoglycan Aggregates chromophore was done at one-fourth the final volume described by Jourdian et al. (12). Fig. 1 shows a standard curve for the assay with a sample of hyaluronic acid, indicating the linearity and sensitivity of the method (approximately 0.04 absorbance units/pg of hyaluronic acid). The results given in the inset show that the recovery of a standard amount of hyaluronic acid (8 pg) which was carried through the papain digestion and ethanol precipitation steps either alone (bur B ) or with 300 pg of purified monomer proteoglycan (bar C) was quantitative. The presence of excess proteoglycan did not interfere with the assay, and controls with no added hyaluronic acid had absorbance values less than 0.005. Analysis for Link Protein-Lyophilized fractions or ethanol precipitates of fractions were prepared as described above. The samples were then dissolved in 100 pl of 0.1 M sodium acetate, 0.1 M Tris-HCI, pH 8.0, and digested with chondroitinase ABC (0.2 units) at 37 "C for 90 min. A 35-p1 aliquot of 6% sodium dodecyl sulfate and 40 mM dithiothreitol was added to each. After heating at 90 "C for 5 min, the samples were analyzed on 7% polyacrylamide slab gels and stained with Coomassie blue as described elsewhere (15). After destaining, bands corresponding to link protein and proteoglycan core were sliced from the gel in rectangular strips of nearly the same size. Slices from adjacent areas which contained no bands were used as blanks. The dye was extracted from each gel slice by incubation in 0.8 ml of 1% sodium dodecyl sulfate containing 0.02 M NaOH a t 37 "C for 24 h with continuous gentle mixing. Each extract was then acidified with 10 pl of concentrated acetic acid to restore the color of the dye, and the absorbance was measured at 590 nm. Fig. 2 shows a standard curve for the link protein band in analyses of a series of different amounts of a sample of aAl-D4 isolated from the rat chondrosarcoma. The increase in absorbance is nearly linear with concentration at low amounts of link protein but the color response falls off slightly in the 5-to 10-pg range. The assay is sensitive to about 1 pg.
Analysis for Proteoglycan-For aliquots from gradients which did not contain guanidine HC1, hexuronic acid concentrations were measured directly with an automated carbazole procedure as described in the preceding paper (1) to estimate proteoglycan contents. In gradients with guanidine HCl, the proteoglycans were precipitated with ethanol as described above and redissolved in 0.2 M sodium acetate, pH 7.2, before analysis for hexuronic acid content.  respective solvents were layered on the gradients followed by centrifugation for the indicated times a t 25,000 rpm, 11 "C. Fig. 3 shows the hexuronic acid profiles obtained from the gradients. Since each of the solvents has a different density and viscosity (16), different centrifugal times are required for the proteoglycan peak fraction to sediment to the middle fractions; about 12 h for 2 M guanidine HCI, 16 h for 4 M guanidine HC1, and 8 h for the pH 3.3 solvent. Subsequent experiments utilized centrifugal times such that monomers, if present, would migrate to the middle fractions of the gradient. This permits maximum resolution of monomers from intact aggregates, which migrate to the bottom fractions, and from unbound link protein molecules and hyaluronic acid, which remain in the upper fractions of the gradients.

Centrifugation
Dissociation of Aggregate-Aggregate was separated from nonaggregated proteoglycans by centrifuging aAl samples on standard preparative preformed CsnSO, gradients as described in the preceding paper (1). Aliquots of purified aggregate (about 5 mg in 1 m l ) were dialyzed in the cold for 20 h against the different solvents used in the gradients. After standing at room temperature for 4 h, the samples were layered on the appropriate gradients and centrifuged for the lengths of time as indicated in Fig. 5. Fractions of about 2.4 ml were collected (7/gradient) and each was dialyzed against 1 mM sodium acetate, 1 mM Tris-HC1, pH 7.2, and lyophilized. Each was dissolved in 1 ml of H20 and aliquots were analyzed for contents of proteoglycan, link protein, and hyaluronic acid as described under "Experimental Procedures." Fig. 4 shows photographs of sodium dodecyl sulfate-polyacrylamide slab gel analyses for some of the gradients, and Fig. 5 summarizes the overall results for the experiments. The positions of the link protein band and of the two bands which contain chondroitinase-digested proteoglycan core molecules are indicated in Fig. 4. The other bands in the gels are derived from proteins in the chondroitinase enzyme preparation used for the digestions.
In the pH 6.6 standard gradient, the aggregate remained intact and all of the link protein, proteoglycan, and hyaluronic acid sedimented to the bottom two fractions, with the large majority of each in the bottom fraction (Figs. 4 and 5). In the presence of 4 M guanidine HCl, the three components of aggregate were completely dissociated; the link protein was observed primarily in fraction 1 at the top of the gradient, hyaluronic acid peaked in fraction 2, and the proteoglycan monomer yielded a broad profie with a peak in fraction 4 (Figs. 4 and 5). In gradients with 1 and 1.5 M guanidine HCl, the aggregate remained intact, indicating that the aggregate is stable to these solvent conditions over an exposure time of at least 24 h (Fig. 5). However, the results for the gradient in 2 M guanidine HCI indicate that the aggregate is partially dissociated in this solvent. About 79% of the proteoglycan, 60% of the link, and at least 93% of the hyaluronic acid were recovered in the aggregate fraction. The remaining proteoglycan was recovered in the middle, monomer region while most of the remaining link protein was found in the top fractions ( Figs. 4 and 5). These results indicate that 2 M guanidine HCl is sufficiently concentrated to begin to dissociate some of the aggregate during the time of the experiment, about 24 h. Additionally, they suggest that dissociation of a monomer from the aggregate structure in this solvent is accompanied by the release of a link protein as well.
Dissociation of aggregate at lower solvent pH occurs by a different mechanism, however. At pH 2.9 and 3.3, aggregate was not observed the proteoglycans sedimented to the middle, monomer region of the gradient while hyaluronic acid was recovered primarily in the top two fractions. In contrast with the results in guanidine HCl solutions, however, the link protein was not recovered in the top of the gradient; rather it sedimented with a nearly identical profiie as that of the proteoglycans (Figs. 4 and 5). This indicates that link proteins are still associated with the monomers and suggests that dissociation of aggregate at low solvent pH occurs by releasing monomer-link protein complexes from hyaluronic acid without dissociating link protein from monomer.
At pH 3.6, aggregate was not present. However, unlike the results at lower solvent pH, the hyaluronic acid was observed in the middle, monomer region rather than in the top fractions, suggesting that at this pH the hyaluronic acid may interact weakly with the monomer-link protein complex (Fig. 5). When 1 M guanidine HCl was included in the pH 3.6 solvent, the hyaluronic acid was again observed in the top fractions well resolved from the middle, monomer region. In this solvent, the link protein still remained associated with the monomer.
The sodium dodecyl sulfate-polyacrylamide gel analyses in Fig. 4 reveal an additional feature about the proteoglycan monomers. Two bands were observed for the core after chondroitinase digestion, with the more rapidly migrating band 2 accounting for about 20% of the Coomassie blue staining relative to the major, slower migrating band 1. In the pH 3.3 solvent where dissociation was complete, the relative proportion of the band 2 proteoglycan was greater in the upper  fractions of the gradient, Fig. 6a, indicating that it represents a smaller proteoglycan, consistent with its behavior on the sodium dodecyl sulfate-polyacrylamide gels following chondroitinase digestion. Both proteoglycan species contain functional hyaluronic acid-binding regions since they are present in purified aggregate, and both appear to dissociate to the same extent in solvents in which partial dissociation occws (Fig. 6b). It is likely that the minor, smaller proteoglycan is a specific proteolytic breakdown product which either accumulates with time in the matrix of the tumor (2,17) or occurs to a limited extent during processing steps while isolating aggregate.
Reaggregation after Exposure to Low Solvent pH-Aliquots of purified aggregate were exposed to pH 3.3 solvent for 24 h before adjusting the pH to 7.0 for treatment as described in Fig. 7. In one case, aliquots were then kept at room temperature for 2 or 4 h before centrifugation on standard Cs2S04 zonal gradients (Fig. 7, a and b). Approximately 85 and 91% of the proteoglycans, respectively, were observed in the bottom fractions of the gradients, suggesting that reaggregation was occurring with time at pH 7.
Oligomers, HA-50, of hyaluronic acid were used to test the specificity of the reaggregation process. In previous experiments (11,17,18), it was shown that HA-m oligomers are long enough to accommodate both the hyaluronic acid-binding site of a monomer and the binding site of an adjacent link protein to form link-stabilized ternary complexes with physical properties similar to monomer (11, 18). In one case, HAISo oligomers were added to an aliquot of aggregate immediately after the pH was adjusted to 7. The amount of oligomer was about 5 to 6 times the amount of macromolecular hyaluronic acid already present in the aggregate sample. After 4 h at room Reaggregation at pH 7.0 of aggregate which was first dissociated at pH 3.3. Aggregate was dialyzed against a 0.2 M sodium acetate/acetic acid, pH 3.3, at 4 "C for 24 h. Aliquots (1.25 mg) were centrifuged after the following: a, 2 h after the pH was raised to 7.0; b, 4 h after the pH was raised to 7.0; c, 4 h after the pH was raised, with HA-50 oligomers (100 pg) added 2 h after the pH was raised; d, 4 h after the pH was raised to 7.0 followed by the addition of HA-5o oligomers just prior to centrifugation; e, 4 h after the pH was raised to 7 with H A + I (100 pg) added just after the pH was raised; f, an aliquot of the original aggregate sample maintained at pH 7.0; g, an aliquot of the original aggregate in pH 7.0 to which HA-m (100 pg) was added followed by incubation for 2 h. The per cent above each peak indicates the proportion of hexuronic acid in fractions from tube 1 to tube 16 to total hexuronic acid. The per cent in parentheses indicates the proportion in these fractions excluding hexuronic acid due to HA-50 oligomers.
by guest on March 23, 2020 http://www.jbc.org/ Downloaded from temperature, the aliquot was centrifuged and the analysis is shown in Fig. 7e. The unbound oligomers remain in the top of the gradient as indicated by the hexuronic acid contents of the top 4-5 fractions. Of the remaining hexuronic acid (i.e. that due to proteoglycan), approximately 80% was recovered in the middle, monomer region and 20% in the bottom, aggregate fractions. The presence of the HA-, prevented the formation of aggregates during the 4-h incubation when 90% of the proteoglycans aggregated in the absence of oligomers (Fig. 7b).
The interaction of monomer with hyaluronic acid is reversible (11, [19][20][21][22] whereas link protein-stabilized aggregates are much more stable (11, 18, 19, 23, 24). Thus, while the experiments shown in Fig. 7 , a, b, and e, show that monomer is able to reassociate with hyaluronic acid after raising the solvent pH to 7 , they do not test for stabilization of the aggregate by functional link protein. The results of the experiments shown in Fig. 7 , c and d , however, provide evidence that the link protein also retains its functionality after exposure to low solvent pH. HA-, oligomers were added to aliquots which had been adusted to pH 7 either 2 or 4 h previously. After an additional 2 h in the fist case or immediately in the second, the mixtures were centrifuged. About 22 and 7% of the proteoglycan-related hexuronic acid, respectively, was observed in the middle, monomer regions of the gradients. These values were close to those obtained in Figs. 7 , a and b. The results indicate that aggregates formed prior to the addition of the oligomers remained intact whereas the presence of the oligomers prevented additional aggregation. Since monomers do not remain associated with macromolecular hyaluronic acid in the absence of link protein when HA-, oligomers are present (see Fig. Be below), the experiments in Fig. 7 , c and d , indicate the presence of functional link protein in the reassociated aggregates. Fig. 7f indicates that the original sample which was not exposed to low solvent pH contained only proteoglycan aggregates. The addition of HA-SO oligomers to another aliquot of the original sample, Fig. 7g, did not lead to any appreciable disaggregation since the hexuronic acid was observed essentially only in the top fractions (the oligomers) and the bottom fractions (aggregate).
Interaction of Aggregating Monomer with Hyaluronic Acid-A series of experiments was designed to determine whether the interaction of monomers with hyaluronic acid under the conditions described in Fig. 8 is stable in standard Cs2S04 zonal gradients. Monomer was isolated from purified aggregate using a 4 M guanidine HC1 dissociative CsCl isopycnic centrifugation. A sample of the monomer was dialyzed into 0.2 M sodium acetate, pH 7 . An aliquot was run directly on a standard zonal gradient (Fig. 8a). Essentially a l l of the proteoglycan was recovered in the middle, monomer region with less than 8% of the hexuronic acid in the bottom fraction (fractions [21][22][23][24][25][26]. When 0.5% (w/v) of macromolecular hyaluronic acid was added followed by incubation a t room temperature, about 65% of the proteoglycans sedimented to the bottom fractions with 35% (shaded area) sedimenting in the monomer region.
The proportion in the bottom fraction increased to about 80% for a 1% mixture of hyaluronic acid with monomer, which has been shown previously ( 7 ) to be nearly optimal for proteoglycan-hyaluronic acid mixtures. At 2% hyaluronic acid, the amount of proteoglycan sedimenting further into the gradient than the monomer region was still about 60% but more was found in the fractions immediately above the bottom, indicative of smaller complexes, i.e. fewer monomers per hyaluronic acid molecule. These results indicate that the proteoglycan hyaluronic acid interaction is stable to a large room temperature for 4 h before centrifugation. Another aliquot of Agg-Dl was mixed with 2% hyaluronic acid. After 2 h at room temperature, HA-w (100 pg) was added and the sample was kept at the same temperature for 2 h before centrifugation (e). Samples were applied to standard rate zonal Cs&04 density gradients as described under "Methods." The shaded areas approximate the amount of unbound monomer in each of the mixtures with HA, about 36, 18, and 11% for b, c, and d, respectively. extent to the conditions of the zonal gradients. In the experiment shown in Fig. 8e, hyaluronic acid (2%) was added to aggregating monomer. After incubation at room temperature for 2 h, HA-, oligomers (10%) were added and incubation continued for 2 additional h before centrifugation. The proteoglycans in this case were recovered almost solely in the monomer region of the gradient. This demonstrates the reversibility of the proteolgycan-hyaluronic acid interaction since equilibrium in the mixture would favor proteoglycanoligomer interactions with each macromolecular hyaluronic acid molecule having only a few monomers associated with it. This result is in sharp contrast with the experiment in Fig. 7g where the oligomers were added to link protein-stabilized aggregates.

DISCUSSION
The experiments described in this report utilized preformed Cs2S04 zonal gradients in different solvents to study dissociation of purified proteoglycan aggregates. At pH 6.6 in guanidine HCl solutions up to 1.5 M, the aggregate remained associated whereas at 2.0 M, partial dissociation over 20 h was observed. In this latter solvent, for each monomer dissociated from the aggregate, a link protein was also dissociated from both monomer and hyaluronic acid. In this case, the ratelimiting step may well be destabilization of link protein function and its release from the aggregate, allowing concurrent dissociation of monomer from hyaluronic acid. In solvents with pH 2.9-3.3, the link protein remains tightly bound to monomer, and both dissociate from hyaluronic acid. In this case, the solvent appears to reverse the interaction of hyaluronic acid with both the binding sites in the hyaluronic acidbinding region of monomer and in the link protein, perhaps by protonation of carboxyl groups on the glucuronic acid by guest on March 23, 2020 http://www.jbc.org/ Downloaded from residues in the hyaluronic acid, since these groups are major determinants in the interaction with monomer (25).
The link protein-monomer association appears to remain stable at the low solvent pH, clearly separating from the hyaluronic acid in the zonal gradients. Additionally, the link protein-monomer complex is able to reassociate with hyaluronic acid to form link protein-stabilized aggregates. This was demonstrated by the stability of reformed aggregates in the presence of HA-a oligomers in contrast to the instability of complexes of monomer with hyaluronic acid in the same conditions. The demonstration that monomer-link protein complexes can exist independent of hyaluronic acid provides additional support for the recently reported evidence (18) that such a complex is probably an intermediate in the formation of proteoglycan aggregates from newly synthesized and secreted proteoglycans in chondrocyte cultures. The Cs2S04 zonal gradients, then, can be used at low solvent pH to isolate the monomer-link protein complexes for further study.
The monomer fraction isolated from purified aggregate contained two populations of proteoglycans, both able to aggregate. The minor population (-20% of the total) was slightly smaller in hydrodynamic size, sedimented slightly slower in dissociative zonal gradients, and yielded a core preparation after chondroitinase digestion which migrated further into 7% sodium dodecyl sulfate-polyacrylamide gels. It is likely that this minor population represents a selective breakdown product of the larger proteoglycans in the major population which either accumulates in the tumor matrix with time or is generated during the extraction and isolation steps.' It is unlikely to be a separate biosynthetic product since in studies reported elsewhere (26), it is shown that the aggregating proteoglycan population synthesized in culture by chondrocytes derived from the chondrosarcoma appears to have a uniform core protein.
Acknowledgments-We wish to thank Dr. S. DeLuca for the hexuronic acid analyses and Dr. C. B. Caputo for the sodium dodecyl sulfate-polyacrylamide gel electrophoresis. We are also grateful to Thelma Prather for excellent technical assistance. K. K. is very indebted to Dr. George R. Martin, National Institute of Dental Research, for continuing support.
Note Added in Proof-Since this paper was submitted, Franzen et al. (28) described an experiment in which purified link protein was Oike et al. (27) have shown that commercial preparations of chondroitinase ABC contain some proteolytic activity. However, it is unlikely in this case that enzyme digestion yielded the smaller and minor band on the gels because the relative content of this band to the larger, major band in each fraction is reflective of the distribution of smaller proteoglycans on the gradient (see Fig. 6a). It is likely that proteolytic activity in the chondroitinase ABC would produce similar proportions of the minor component throughout. mixed with purified monomer under associative conditions and the mixtures centrifuged in sucrose-stabilized velocity gradients. Link protein, as estimated by immunological procedures, sedimented with the monomer.