The Link Protein in Proteoglycan Aggregates from the Swarm Rat Chondrosarcoma*

Aggregated proteoglycans aggregated), isolated from the Swarm rat chondrosarcoma by extraction with 4 M guanidinium chloride in the presence of protease inhibitors and purified by centrifugation in an associative cesium chloride gradient, were separated into the component parts by centrifugation in a dissociative cesium chloride gradient. The gradient was cut into five equal fractions. The bottom fraction contained 98% of the chondroitin sulfate and 84% of the protein of the aggregate preparation. Sedimentation equilibrium studies on the protein core of this fraction, isolated by column chromatography from chondroitinase suggest that its molecular The intermediate of contained


Aggregated
proteoglycans (70% aggregated), isolated from the Swarm rat chondrosarcoma by extraction with 4 M guanidinium chloride in the presence of protease inhibitors and purified by centrifugation in an associative cesium chloride gradient, were separated into the component parts by centrifugation in a dissociative cesium chloride gradient. The gradient was cut into five equal fractions. The bottom fraction contained 98% of the chondroitin sulfate and 84% of the protein of the aggregate preparation. Sedimentation equilibrium studies on the protein core of this fraction, isolated by column chromatography from chondroitinase ABC digests, suggest that its molecular weight is 2.0 x lo" to 2.2 x 1W. The intermediate fractions of lower buoyant densities contained hyaluronic acid (0.8% of the total weight of the aggregated preparations) and proteoglycan monomers with fewer chondroitin sulfate chains relative to the protein core than in the bottom fraction. The top fraction, in addition to proteoglycans, contained a link protein. The link protein was separated from the proteoglycans by chromatography on Sephadex G-200 in the presence of 4 M guanidinium chloride.
Its molecular weight was estimated to be 40,000. It stabilized complexes of proteoglycan monomers and hyaluronic acid so that they could be seen in the analytical ultracentrifuge at pH 5.8. The amino acid composition of the link protein differs significantly from that of the protein core of the proteoglycan monomers and from that of the hyaluronic acid binding region of the protein core of the proteoglycans.
In hyaline cartilages, proteoglycans (glycosaminoglycans covalently linked to protein) are the structural components primarily responsible for the characteristic elasticity and reversible resistance to compressive forces (1). These molecules exist as aggregate structures in the extracellular matrix (2). * This work was suuoorted bv National Institute of Dental Re-. .
search Grant DE-02731 to the University of Michigan and American Cancer Society Institutional Grant 13N 13. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734solely to indicate this fact $ To whom reprint requests should be sent.
The aggregates are effectively dissociated and their noncovalent interactions are minimized in solvents with high concentrations of electrolytes; they are readily extracted from hyaline cartilages by such solvents (3,4). When the salt concentration in the extracts is lowered to about 0.5 M by dilution or dialysis, the proteoglycans reaggregate. Subsequently, isopycnit cesium chloride gradient procedures can be used to separate the proteoglycan aggregates from other materials in the extracts. In turn, the components of the proteoglycan aggregates can be separated from each other by centrifugation in a dissociative cesium chloride gradient. The chemical and physical properties of proteoglycan aggregates and proteoglycan monomers from a variety of cartilages, including the Swarm rat chondrosarcoma, have been described recently (5-16). Hascall and Sajdera (6) showed that aggregated proteoglycans could be dissociated and separated into at least two fractions, a proteoglycan monomer fraction and a link protein(s) fraction, by gradient density centrifugation in the presence of 4 M guanidinium chloride. Hardingham and Muir (17) found that proteoglycan monomers formed complexes with hyaluronic acid which could be detected by viscosimetry and gel filtration. However, these complexes were not demonstrable in the analytical ultracentrifuge under conditions used for the demonstration of aggregates. The seeming discrepancy was resolved by Gregory (181, who showed that both hyaluronic acid and the protein-rich top fraction of a dissociative gradient were required for a demonstration of proteoglycan aggregates with an analytical ultracentrifuge at pH 5.8. The protein-rich top fraction contains link protein(s) and proteoglycans of low buoyant density.
On the basis of work in several laboratories, a model for the structure of proteoglycan aggregates in hyaline cartilages has been proposed (19,20). The proteoglycan monomer is polydisperse, average molecular weight of 2.5 x 10" and a range of 1.0 to 4.0 x 10" (8,9). It contains about 100 chondroitin sulfate chains, each with an average molecular weight of 2 x 104, and 35 to 50 keratan sulfate chains, each with an average molecular weight of about 5 x lo" (11-l>). The polysaccharide chains are covalently linked to a protein core, which has a weight average molecular weight of 1.8 x 10" to 2.0 x lo" (21). One end of the protein core interacts with hyaluronic acid (22). This end is relatively free of polysaccharide chains. The remainder of the protein core provides the attachment points through the hydroxyls of serine and threonine, for the chondroitin sulfate and keratan sulfate chains (22)(23)(24)(25). These polysaccharide chains are not uniformly distributed over the protein core. On the protein core of the proteoglycans of bovine nasal septum and trachea, about 60 to 70% of the keratan sulfate chains and less than 10% of the chondroitin sulfate chains are attached to a segment of the protein core immediately adjacent to the segment of the protein core involved in binding to hyaluronic acid (19).
Such proteoglycan monomers can complex with hyaluronic acid (17). The complexes are stabilized by link proteins, which can interact noncovalently with both the protein core of the proteoglycan monomers and with hyaluronic acid (24). Hoffman,et al. (26,27) have suggested an alternative model for proteoglycan aggregates, in which a specific link protein is not required. They suggest that formation of proteoglycan aggregates is an expression of the polydispersity of the proteoglycans. In their model, proteoglycans of low buoyant density (high protein content) help stabilize aggregate structures of high buoyant density proteoglycans and hyaluronic acid.
In general, the proteoglycan aggregate preparations from the Swarm rat chondrosarcoma have properties very similar to those reported for proteoglycan aggregates from bovine nasal septum and trachea (7) with respect to the following: (a) monomer and aggregate size; (b) the presence of a large proportion of aggregate; (c) the ability of the proteoglycan monomer to interact with hyaluronic acid; and (d) the presence of a hyaluronic acid.protein complex which is resistant to digestion with chondroitinase and trypsin. They are different in that (a) keratan sulfate is not present in the proteoglycan molecules and (6) only one link protein is present.
This report describes a simple methodology for the isolation of the link protein in proteoglycan aggregate preparations from Swarm rat chondrosarcoma.
Moreover, the amino acid composition of the link protein is compared with the amino acid compositions of the protein cores of proteoglycan monomer fractions prepared from preparations of proteoglycan aggregates by centrifugation in a dissociative cesium chloride gradient. It is suggested that in aggregates for each proteoglycan molecule there is a molecule of link protein. was added to increase the concentration of GdmCl to 4 M. CsCl was then added to adjust the density of the solution to 1.50." Dissociative gradients were established by centrifugation in a Beckman Ti-50 rotor at 40,000 'pm for 40 to 48 h at 10". The gradients were cut into 4 parts, labeled Al-D1 to Al-D4, from bottom to top (5). In most of the experiments, to obtain a finer fractionation, the gradient was cut into 5 equal parts, labeled Al-Dla, Al-Dlb to Al-D4.

Comparison of Size of Proteins i n Al and i n Al-D Fractions
Chondroitin sulfate chains were removed from the core proteins by incubation with chondroitinase ABC for 60 min at 37", as previously described (7). The digests were reduced with ,6-mercaptoetha-no1 and subjected to electrophoresis on 3% polyacrylamide gels in the presence of 0.1% sodium dodecyl sulfate.

Preparation of Protein
Core of Al -Dla The protein core of the proteoglycan monomer (Al-Dla) was prepared by chondroitinase ABC digestion and it was purified as previously described (7). The material eluted from the Sepharose 4B column at K,, = 0.60 was examined in the analytical centrifuge. To this end, this fraction was dialyzed three times, for 12 h each time, against 100 volumes of 0.5 M sodium acetate, pH 7.0, at 4". The final dialysate was used for the preparation of reagent blanks and for dilutions of the retentate for analysis in the analytical ultracentrifuge.

Isolation of Link Protein
Pooled Al-D4 fractions were concentrated in an Amicon concentrator cell with an UM 05 membrane at 4". In example, a volume of 80 ml was reduced to approximately 2 ml. The concentrate was diluted with 4 M GdmCl, 0.05 M sodium acetate buffer, pH 5.8, to 20 ml and reconcentrated to 2 ml. This was repeated three times. Onemilliliter aliquots of the concentrate were placed on a column of Sephadex G-200, prepared, and used as described above. Absorbance at 280 nm and hexuronic acid, protein, and neutral sugar contents of the effluent fractions were determined.

Preparation of High Molecular Weight Hyaluronic Acid
The preparation of the hyaluronic acid from umbilical cords (2 mg in 1 ml of 0.5 M sodium acetate buffer, pH 7.0) was fractionated by the use of a Sepharose 2B column. Elution was with 0.5 M sodium acetate buffer, pH 7.0. The fraction with K,, of 0.30 was designated as the high molecular weight fraction (HA,). The other hyaluronic acid sample (HA,,), estimated average molecular weight of 120,000, was more polydisperse than the preparation from umbilical cords. On the same column its K,, was 0.67. An AnJ rotor was used at speeds of 9,000, 10,000, and 11,000 rpm and at a temperature of 20". A partial specific volume of 0.67 g/ml was calculated from a dry weight protein content of 51.2%, assuming a partial specific volume of 0.72 g/ml for protein and 0.52 g/ml for the carbohydrate portion. The data were collected and calculated as described by Hascall and Riolo (21). Apparent weight average molecular weights were obtained by extrapolation to zero protein concentration.

Characteristics
of Al-D Fractions-A typical distribution of protein, hexuronic acid, and galactosamine in a dissociative cesium chloride gradient of Al preparations from the Swarm rat chondrosarcoma is shown in Table I. About 84% of the protein and about 98% of the galactosamine and hexuronic acid are present in the bottom fifth of the gradient. In this fraction (Al-Dla) the ratio of galactosamine to protein is higher than in any of the other fractions; the ratio of galactosamine to protein decreases as the buoyant density of the fractions decreases. On the other hand, the ratio of glucosamine to galactosamine in the proteoglycans increases as the buoyant density of these decreases. Realizing that hyaluronic acid was present in fractions of intermediate buoyancy, the glucosamine of the hyaluronic acid was eliminated from the estimate of the ratios of glucosamine to galactosamine by the use of the procedure of Antonopoulos et al. (28) for the estimation of hyaluronic acid in the fractions.
As the buoyant densities of the proteoglycans in a dissociative gradient decrease, their molecular weights apparently decrease; they are retarded to a progressively greater extent on a column of Sepharose 2B (Table II). This, however, is not a reflection of shorter chondroitin sulfate chains. It appears, Table II, that the average length of the chondroitin sulfate chains prepared by either digestion with papain or treatment with 0.5 M sodium hydroxide, is the same in the proteoglycans at all levels of the dissociative gradient. These data in conjunction with the data in Table I, which clearly indicate a decreas- in 3% polyacrylamide gels in the presence of sodium dodecyl sulfate, it was found (Fig. 1) that the mobility of the protein cores increased as the buoyant density of the proteoglycan fractions decreased. This decrease in the size of the protein cores occurred in discrete steps, since as seen in the electrophoretograms, the protein cores in any given fraction contained one or more clearly separated protein entities. In Al-Dla only a single protein component was seen. Two components were readily discerned in Al-Dlb; a slower migrating component was in higher concentration than a faster migrating component (Fig. 1B). The reverse was seen in the case of Al-D2 (Fig.  1C). A third protein component was additionally present in the Al-D3 fraction; its mobility ( Fig. 1D) was even greater than that of the two protein components seen in Al-Dlb and Al-D2. The protein core of the proteoglycans in the Al-D4 is primarily of the size of the fastest migrating protein component in the Al-D3 fraction. The Al-D4 fraction, however, also contains two other protein components, whose mobility is less than that of the major protein component, but decidedly faster than that of the two protein components seen in Al-Dlb and Al-D2. Each of the fractions was recentrifuged in a dissociative isopycnic cesium chloride gradient at the same starting density. The buoyant density of the fractions was not altered. Nor were the protein/uranic acid ratios changed thereby. Moreover, the electrophoretograms of the recentrifuged fractions were indistinguishable from the electrophoretograms of the fractions before recentrifugation.
For comparison, the protein components in Al preparations are shown in Fig. 18'. It is apparent from such electrophoretograms that in Al preparations there is a mixture of proteoglycans, since their protein cores vary in size. It is further apparent that in such preparations a protein core of lowest mobility is present in highest concentration.
To obtain an estimate of the molecular weight of the protein rA . core in the Al-Dla fraction (Fig. lA), the Al-Dla fraction was treated with chondroitinase ABC, as above. The resultant protein core was examined in the analytical ultracentrifuge, using the techniques of sedimentation to equilibrium. On extrapolation of the apparent molecular weights to zero protein concentration, it was found that the molecular weight of the protein core was in the range of 2.0 x l@ to 2.2 x lo".
The amino acid profiles of the proteoglycan fractions, expressed in residues per 1000 amino acid residues are given in Table III. Certain progressions are apparent. As the buoyant density of the fractions decreases, the protein core contains progressively more lysine and arginine and less threonine and serine. There is also an increase in the ratio of glucosamine of oligosaccharides (1% cetylpyridinium chloride fraction) to galactosamine of chondroitin sulfate as the buoyant density of the proteoglycan fractions decreases (Table I). Since the proteoglycan of the Swarm rat chondrosarcoma lacks keratan  (19,25,34,35). The amino acid composition of the hyaluronic acid binding region of the protein core of the proteoglycan monomer, designated in this report as Al-DB-T-2B,-G150,, approaches that of the protein core of the proteoglycans in the Al-D4 fraction (Table III). This region of the protein core, in terms of residues per 1000 amino acid residues, also has more lysine and arginine and less threonine and serine than the protein core of the proteoglycans in the Al-Dla fraction.
Additionally, it has more aspartic acid, alanine, tyrosine, and phenylalanine and less proline and glycine than the protein core in the Al-Dla fraction.
Examination of the papain digests of the AI-D fractions for the glycosaminoglycans in them by the method of Antonopou-10s et al. (28) indicated that most of the hyaluronic acid was present in Fractions Al-D2 and Al-D3 and accounted for 0.8% of the weight of the Al preparations. It was previously reported (7) that when hyaluronic acid is at about this concentration in a mix of hyaluronic acid and proteoglycan monomer (AI-Dla) the complex formation of these two macromolecules is maximal, as indicated by the exclusion of the complexes from columns of Sepharose 2B. Nature of Link Protein -The separation of the link protein (Al-D4-G200,,) from the proteoglycan monomers in the Al-D4 fraction of a dissociative gradient on a column of Sephadex G-200 with 4 M GdmCl as the eluent was as shown in Fig. 2. The fractions indicated by the bar above Peak II were pooled. When such pools of the fractions were examined by electrophoresis in polyacrylamide gels, one protein component was found (Fig. 3, A and B). Its mobility was greater than that of the hyaluronic acid-binding peptide derived from the protein core of the Al-Dla fraction (Fig. 3C). The amino acid composition of the link protein was different than that of the hyaluronic acid-binding peptide and that of the protein cores of the proteoglycans in the Al-D fractions (Table  III). In terms of residues per 1000 amino acid residues, there was significantly more lysine, arginine, aspartic acid, glycine, half-cystine, tyrosine, and phenylalanine and less threonine, serine, glutamic acid, alanine, valine, and methionine in the link protein than in the hyaluronic acid-binding peptide. Significantly more aspartic acid, alanine, half-cystine, tyrosine, and phenylalanine and less threonine, serine, and methionine were found in the link protein than in the protein cores of the proteoglycans in all of the Al-D fractions. Small amounts of glucosamine and galactosamine were found in preparations of the link protein.
In view of the identity of the ratio of these hexosamines in the preparations of the link protein and the ratio of these hexosamines in the proteoglycans (Al-D4-G200,) in the Al-D4 fraction, it is possible that the link protein preparations (Al-D4-G200,,) may still contain a small amount of proteoglycan. Caterson and Baker (36) also found these amino sugars to be present in link protein(s) isolated from bovine nasal septa. Attempts to determine whether other sugars were present in the link protein preparations have been unsuccessful, possibly because of the small amounts of material used. If there are other sugars, the sum of these is probably less than 2% of the weight of the link protein, as suggested by the use of phenolsulfuric acid assay (7) and by the use of the electrophoretic procedure of Segrest and Jackson (31). That the carbohydrate content may be very low is also suggested by the virtual absence of periodic acid-Schiff-positive staining of the link protein in acrylamide gels after electrophoresis.
In line with this is the virtual absence of neutral sugars in the efIluent fractions in which the link protein appeared when columns of Sephadex G-200 were used for separating it from proteoglycans in the Al-D4 fraction (Fig. 2) 3. Disc electrophoresis of link protein (Al-D4-G20OJ and of the hyaluronic acid-binding peptide of the protein core of an Al preparation.
The arrow indicates the tracking dye. B, 10 kg of Al-D4-GZOO,, in a 7% polyacrylamide gel in the presence of sodium dodecyl sulfate. C, densitometric tracing of a 7% polyacrylamide gel when 10 pg of the hyaluronic acid-binding peptide was electrophoresed under the same conditions used in B. column (Fig. 4). When the link protein was mixed with hyaluronic acid, the K,, for the hyaluronic acid was lowered, protein was present in association with the hyaluronic acid, and there was a decreased amount of the link protein in the V, of the column (Fig. 4). The elution profile of the Al-Dla preparation of proteoglycan monomers from a column of Sepharose 2B was as shown in  58'). On the addition of link protein to the Al-Dla, neither the schlieren pattern nor the elution profile (Fig. 5C) from the Sepharose 2B column were altered. However, most, if not all of the link protein was in association with the proteoglycan in that none of the link protein was detected at V, of the Sepharose 2B column. When a mixture of hyaluronic acid and proteoglycan was used the schlieren pattern was again as that seen with the proteoglycan alone ( Fig. Q'), but the elution profile was markedly altered (Fig. 5B). The elution profiles of a mixture of link protein, hyaluronic acid, and proteoglycan monomers were comparable to the elution patterns of a mixture of hyaluronic acid and proteoglycan (Fig. 5,D and El,but now the schlieren patterns indicated that some of the proteoglycan was in the form of aggregates (Fig. 5,G and H).
In this series of experiments, Fractions Al-Dla through Al-D3 were recombined in the original proportions and then examined in the ultracentrifuge.
The aggregated form of the proteoglycans was not seen. The aggregated form of the proteoglycans also was not seen when to such mixtures of the Al-Dla through Al-D3 fractions the proteoglycans (Al-D4-G20OJ of the Al-D4 fraction were added in amounts proportional to that originally present in the Al-D4 fraction. These data strongly suggest that the proteoglycans of low molecular weight do not stabilize hyaluronic acid. proteoglycan complexes.
Link protein in sufficient amounts was not available to extend these experiments so that a stoichiometric relation of link protein to proteoglycan in aggregate preparations could be precisely established. However, as shown in Table I, all of the protein in an Al-D4 fraction was subfractionated into proteoglycans of low buoyant density (Al-D4-G200,) and link protein (Al-D4-G200,J.
The weight of the latter, 0.75 mg, is equivalent to 0.020 PM, assuming a molecular weight of 40,000 for the link protein. The amount of protein core in the Al preparation from which this amount of the link protein was isolated can be estimated by summing the amounts of protein in the Al-Dla through Al-D4 fractions minus the protein in the Al-D4-G200,, subfraction. Supposedly, only 70% of this protein in the form of proteoglycan monomers was in the aggregate form, as suggested by the areas of the peaks seen in schlieren patterns of this Al preparation.
The link protein differs in yet another way from the proteoglycan monomers; it is insoluble in water.
In u&-o, the differences between complexes of proteoglycans and hyaluronic acid and proteoglycan aggregates are 2-fold. The latter is more stable and it is less sensitive to treatment with trypsin (22). The noncovalent interactions of the link protein with both the hyaluronic acid and the protein core probably bring these three components into a spatial relationship with each other such that the link protein and a portion of the protein core of the proteoglycans are less accessible to trypsin.
The hyaluronic acid of the proteoglycan aggregates is also protected. Faltz et al. (30)