Isolation and Characterization of Proteochondroitin Sulfate from Pig Skin*

Proteoglycans extracted from pig guanidine 4°C in the presence of protease inhibitors and were precipitated with ethanolic KSCN to separate them from collagen. They were fractionated into two major components on the basis of buoyant density in a CsCl gradient under associative Proteodermatan was in the lighter fractions, it to 60% The proteochondroitin in a dissociative M guanidine HCl) density gradient. a protein content 4 to 5%, proteoglycan hyaline chromatogra- phy

(1). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC. Section 1734 solely to indicate this fact.
are polydisperse and have average molecular weights of a few million, and most of the molecules exist in the form of very large aggregates as a result of their ability to bind specifically to hyaluronic acid, both in the tissue (5) and after isolation. PG of tissues that contain predominantly dermatan sulfate, such as dermis, aorta, and heart valves, have been less thoroughly studied, partly because of the greater difficulty of extraction.
In aorta, there is evidence for a PG in which dermatan sulfate and chondroitin sulfate are attached to the same protein core (6, 7), and another component of such tissues, heparan sulfate, also appears to be in PG form. Toole and Lowther (8) (14) or by the sum of the amino acid analysis. Hydroxyproline was determined as described by Woessner (15).
Extraction of Proteoglycans-Pig skin was freed of fat and hair, cut into small pieces, frozen in liquid Nz, and pulverized in a Wiley Mill with a 20-mesh screen, cooled with liquid Ns. The powder was immediately lyophilized and extracted three times for 3 h each at 4°C with petroleum ether (16), freed of organic solvent, and stored at -15OC.
An initial extraction was made with 0.1 M NaCl in 0.05 M sodium acetate, pH 5.8 (15 ml/g of powder), stirred for 3 h at 4°C. This solvent and all further solutions, as far as associative and dissociative density gradients, also contained proteolytic inhibitors: 0.1 M 6-aminohexanoic acid, 0.01 M ethylenediaminetetraacetic acid, and 0.005 M benzamidine hydrochloride (17). The suspension was centrifuged at 23,500 X g for 30 min at 4"C, and the residue was further extracted, for exploratory purposes, with various concentrations of salts (30 ml/ g of powder) at 4°C for 48 h with gentle rocking. The pH varied from 4 to 5.8 depending upon the salt present. Centrifugation at 23,500 x g for 45 min at 4°C gave a clear solution that was dialyzed against water and analyzed for uranic acid.
The procedure chosen (summarized in Fig. 1) was extraction with 0.1 M NaCl as above, then extraction of PG with 4 M guanidine HCl, 0.05 M sodium acetate, pH 5.8, for 48 h at 4°C (30 ml/g of powder), centrifugation, re-extraction with the same volume of solvent for 3 h at 4"C, and centrifugation.
The solutions were combined, and PG was precipitated from the guanidine solution by addition of 2.3 volumes of ethanolic KSCN as described by Toole and Lowther (8) for urea extracts. The washed precipitate was suspended in water (6 to 7 ml/ g skin powder) and dialyzed at 4°C against 0.05 M sodium acetate, pH 5.8, with proteolytic inhibitors. Addition of guanidine HCl, to a concentration of 4 M, then completed the dissolution of the precipitate.
As a control preparation, a pig skin sample that had been observed to contain little or no proteochondroitin sulfate (50 g) was mixed with 2 g of bovine nasal septum and pulverized and extracted exactly as described. This sample was carried through all subsequent steps for the isolation of proteochondroitin sulfate. Associative Density Gradient-PG solution was dialyzed against 9 volumes of buffer with inhibitors to reduce the guanidine concentration to associative conditions (18). Cesium chloride was added to a density of 1.65 g/ml (some cloudiness usually appeared). After centrifugation at 84,100 X g,, at 16°C for 64 h in a Beckman-Spinco type 42.1 rotor, a thin gelatinous layer was lifted from the surface, and the gradient was divided into six nearly equal fractions by aspiration from the top. These were labeled A6 to Al from top to bottom (19). The fractions were thoroughly dialyzed against 0.05 M sodium acetate, pH 5.8, analyzed for uranic acid, hexosamine, and protein, and digested with papain for electrophoresis of glycosaminoglycans.
Dissociative Density Gradient-The Al fractions from a density gradient were combined and diluted with approximately 0.5 volumes of water, and guanidine HCl was added to 4 M and CsCl to a density of 1.5 g/ml (20). The final volume was one-fourth to one-third of the volume in the associative gradient. Centrifugation was repeated as before, and gradients were divided into six fractions (Al-D6 to Al-Dl) and analyzed as above.
Column Chromatography-Analytical columns (0.9 X 136 cm) of Sepharose 2B or Sepharose 6B were equilibrated and eluted with 0.5 M sodium acetate, pH 7.0, or 4 M guanidine HCl, 0.05 M Tris-HCl, pH 7.0. Flow rates were 6 ml/h, and fractions of 1 ml were collected. V, was determined with highly polymerized calf thymus DNA or tritiumlabeled h phage DNA, and V, with glucuronolactone.
Sepharose 2B was calibrated with freshly prepared bovine nasal PG monomers and aggregates (20) and 6B with standard glycosaminoglycans. Samples contained 0.5 to 1 mg of uranic acid in 1 ml. K,, values were determined by the equation of Laurent and Killander (21).
Characterization of Glycosaminoglycans-Samples of PG in 0.5 M sodium acetate, 0.1 M potassium phosphate, 0.01 M sodium EDTA, and 0.01 M cysteine, pH 6.5, were added to papain suspension that had been activated for 60 min at 37°C in 2 volumes of the same solvent (papain/PG 1:125). After 18 h at 64"C, the digests were dialyzed against 50 to 100 volumes of 0.2 M sodium chloride and then against large volumes of distilled water with stirring. If necessary, samples were clarified by centrifugation for cellulose-acetate electrophoresis or chromatography on Sepharose 6B. Cellulose-acetate electrophoresis of 0.6~,a1 samples of glycosaminoglycan (1 to 2 mg/ml) was carried out on the PhoroSlide medium (Millipore Biomedica, Acton, MA) in the cadmium acetate system of  (24), which attacks only hyaluronic acid. PG (24 pg of uranic acid) was treated with 20 turbidity reducing units of enzyme in 250 al of 0.1 M sodium acetate, pH 5.0, at 60°C for 6 h, and for a further 16 h with another 20 turbidity reducing units of enzyme in 100 ~1. The digest was exhaustively dialyzed against the same buffer, and HCl was added for hydrolysis and hexosamine determination. For comparison, 42 pg of hyaluronic acid was similarly digested. All digested samples were related to controls in which enzyme was omitted.
For the determination of susceptibility to chondroitinases (25), digestions were done in 0.1 M sodium acetate, 0.1 M Tris-HCl, pH 7.3 (26). Samples of PG containing 50 pg of uranic acid in 200 ,ul were dialyzed against the buffer and digested with 0. I unit of chondroitinase AC or ABC at 37°C for 1.5 h. An additional 0.1 unit of enzyme and a drop of toluene were added, and digestion was continued for 20 h. Digests were exhaustively dialyzed against 0.1 M sodium acetate, and HCl was added for hydrolysis and hexosamine determination for comparison against undigested controls.
The ratio of chondroitin 4-sulfate to B-sulfate in the proteochondroitin sulfate was determined by Method II of Saito et al. (27). Since there was no evidence of dermatan sulfate in this PG, chondroitinase ABC was not used. Digestion mixtures contained chondroitinase AC and 6sulfatase, with or without 4-sulfatase. A sample of the final disaccharide product, ADi-OS, was used as a standard in the Morgan-Elson calorimetric measurement. Nitrous Acid Treatment-The low pH nitrous acid procedure of Shively and Conrad (28)  Acid-Binding studies were carried out essentially by the procedure of Hardingham et al. (30,31). Samples of proteoglycan (Al or Al-Dl) containing 0.8 to 1.2 mg of uranic acid in 0.8 to 1.5 ml of associative buffer (0.5 N sodium acetate, pH 70) were mixed with solutions of hyaluronic acid (molecular weight in excess of 0.5 x 10G) in ratios of 25:l to 4OO:l (PG/hyaluronic acid). Mixtures were kept at 4°C for 60 min and then chromatographed on Sepharose 2B in the same buffer. Experiments were also carried out at 22'C, but no difference was found. As a control, PG monomer (Al-Dl) prepared from bovine nasal septum in the presence of skin was combined with hyaluronic acid in the ratio 125:l (31), and chromatographed on Sepharose 2B. The uranic acid solubilized, including the amount extracted initially with 0.1 M NaCl (---), is given as per cent of total tissue hexuronic acid. All solutions contained protease inhibitors. Guanidine HCl (0) and urea (U) were in 0.05 M sodium acetate, pH 5.8. Solutions of LiCl (A), CaC12 (a), MgCla (0) and LaClj (0) were not buffered (pH between 4 and 5.5). seen previously with calcium extraction of cartilage (18). There were also differences in the ratio of galactosamine to glucosamine in the extracts. With guanidine HCl, urea, and most of the salts, the ratios were 1.3 to 1.5, whereas with LiCl the ratio was 0.27. A 48-h extraction period was optimal in each case, longer extractions solubilizing more collagen with little increase in PG. With guanidine HCl, extraction was independent of pH between 5 and 8. The observations, as a whole, are similar to those of Sajdera and Hascall (18) on extraction of cartilage, except for the somewhat lower extractability of skin PG.

Extraction of
In the procedure finally used (Fig. l), a preliminary extraction with 0.1 M NaCl removed some of the easily extractable glucosamine and some noncollagenous protein (Fraction a, Table I). Cellulose acetate electrophoresis (Fig. 3) showed that this glucosamine was mainly in hyaluronic acid, but not all of the hyaluronic acid is easily extracted, as will be seen below. Subsequent extraction with 4 M guanidine HCl removed 71% of the total galactosamine, with substantial amounts of glucosamine and collagen (Fraction b, Table I sulfates, and these were used for the preparation of proteochondroitin sulfate described here. The presence of hyaluronic acid and heparan sulfate in the extract may account largely far the glucosamine content. The residue after extraction was characterized following proteolytic digestion. It contained much of the total glucosamine (Fraction c, Table I), mostly as hyaluronic acid and heparan sulfate (Fig. 3). Hyaluronic acid resistant to extraction has been reported also in bovine aorta (7), and this may   Sample e, 59% of the galactosamine was recovered, and there was a substantial reduction of protein and especially of collagen. The yield of PG was rather variable in the step, but the procedure was included because it greatly improved the following density gradient fractionation.
PG prepared as above was centrifuged in a CsCl density gradient, first under associative conditions so that aggregate forms of PG, if present, might be preserved and detected by gel chromatography. Fig. 4.4 shows the distribution of several components in the resulting six fractions. These values, with those in Table II, indicated that 30% of the uranic acid was recovered in Fraction Al together with more than half of the galactosamine.
Only chondroitin sulfate could be seen on cellulose acetate electrophoresis of a papain digest of Al (Fig.  3). Very little of the total protein was present here, as would be expected if this fraction contained a proteochondroitin sulfate of high polysaccharide content and consequently of high buoyant density.
Dermatan sulfate was found in all fractions of the gradient (Fig. 3) except the densest, although mainly in the upper three. This suggests a protein-rich structure consistent with the 60% protein previously reported for proteodermatan sulfate (9). Heparan sulfate appeared in Fractions A2, A3, and A4, and might represent a separate PG as observed by others in aorta (6) and in lung (32). Hyaluronic acid was spread among all fractions except Al. Although these rather diffuse distributions might suggest various degrees of interaction among some of the PG and hyaluronic acid under associative conditions, no specific binding of proteochondroitin sulfate to hyaluronic acid could be demonstrated, as shown below. Measurements of DNA by absorbance at 260 nm, by the diphenylamine reaction (12), and by the carbazole method (lo), indicated that enough DNA was present in Fractions A3 and A4 (Table II) to account quantitatively for the apparent high ratio of uranic acid to hexosamine.
Fraction Al was further purified in a dissociative density gradient ( Fig. 4B and Table II). About 82% of the uranic acid of Fraction Al was recovered in the bottom two fractions, but only the densest fraction (Al-Dl) was analyzed in detail. The ratio of glucosamine to galactosamine was lower than that of Al, possibly because of the removal of traces of hyaluronic acid or glycoprotein.
The nature of the remaining glucosamine is considered later. A small amount of protein, present in Al, appeared in the upper fractions of the dissociative gradient (Fig. 4B), where there was very little uranic acid. From the amino acid analysis of the PG (Al-Dl) and on the assumption that it contained 30% uranic acid, a protein content of 4 to 5% was calculated for proteochondroitin sulfate. On Sepharose 2B (Fig. 5A), the PG moved with a K,, of 0.53, and PG of bovine nasal septum, with a M, of 2.5 X 10" (33), had a K,, of 0.18. On the assumption of similar hydrodynamic properties of the two PG, the skin product would Initial density in the associative gradient was 1.65 g/ml and in the dissociative gradient 1.52 g/ ml; densities of fractions are given in Table II. have an M, of approximately 10". Evidence for its PG structure is found in a comparison of its gel chromatographic behavior before and after papain digestion.
Undigested Al-D1 was totally excluded from a column of Sepharose 6B, but after digestion the resulting chondroitin sulfate chains could be compared with a series of standard glycosaminoglycans ( Fig  6). The logarithms of the M, of the standards were plotted against their elution volumes, and the M, of the skin chondroitin sulfate was graphically found to be 20,000, the same as that of cartilage chondroitin sulfate (34). In addition, dermatan sulfate chains resulting from papain digestion of Fraction A6 were found to have an M, of 26,000 (Fig. 6).
Aggregation of Proteoglycan-In PG extracted from hyaline cartilage (20), Fraction Al contains largely aggregates, and Fraction Al-D1 consists of monomers from which hyaluronic acid and link proteins have been dissociated and removed. In the control experiment PG of bovine nasal septum was isolated in the presence of skin in an associative density gradient and was chromatographed on Sepharose 2B (Fig. 5   bovine nasal septum Al). Most of the PG appeared in the excluded volume as expected of an aggregated preparation (30, 31), indicating that the isolation procedure had neither removed bound hyaluronic acid nor damaged the binding region of the PG. A monomer, prepared from this Al in a dissociative gradient, was retarded with a K,, of 0.18 (Fig. 5, bovine nasal septum Al-Dl).
Addition of hyaluronic acid to this material largely restored the aggregated form and moved the elution position back to the excluded volume.
In contrast, in the skin proteochondroitin sulfate, the molecular sizes of Al and Al-D1 were essentially the same (Fig.  5), whether chromatographed under associative or dissociative conditions, and there was no evidence of aggregate. The addition of various proportions of hyaluronic acid to Al or Al-D1 from skin had no effect on the position of their peaks. In this regard the skin proteochondroitin sulfate differs from that of cartilage. Chemical and Enzymatic Analyses-The amino acid composition of proteochondroitin sulfate (Al-Dl) is given in Table  III together with analyses of other PG for comparison. There is a general resemblance to the composition of PG from bovine nasal cartilage, with the major difference that no sulfur-containing amino acids were detected in the skin product. Proteodermatan sulfate of skin, however, as analyzed by Gbrink (9) (Table IV), and the electrophoretic pattern (Fig. 3), constitute good evidence for the identity of the chondroitin sulfate, although the total absence of iduronic acid cannot be claimed. In the proteochondroitin sulfate, about 8% of the hexosamine is glucosamine (Table  III). If the PG contains only conventional glycosaminoglycans, this could indicate the presence of hyaluronic acid, heparan sulfate, keratan sulfate, or heparin, either as a contaminant or as part of the PG structure. Attempts were made to define the nature of the glucosaminecontaining component by specific enzymatic and chemical treatments.
Digestion with hyaluronidase caused little change in the hexosamine content, whereas under the same conditions a hyaluronic acid control was almost entirely destroyed (Table  IV). Digestion with chondroitinase AC or ABC had no effect on the glucosamine, although galactosamine, as a part of chondroitin sulfate, was lost as expected.
Compounds that contain N-sulfate groups, such as heparan sulfate and heparin, would be split by nitrous acid (28), and some material of this nature may be present since some of the glucosamine was lost under conditions that caused complete destruction of a heparin control (Table IV). At present no clear decision can be made about the nature of the glucosamine, but heparan sulfate and keratan sulfate are possible sources, the former because it is seen in the tissue (Fig. 3), and the latter because it would be resistant to the treatments listed in Table IV. Linkage of Polysaccharide to Protein-The ,l3 elimination reaction provides evidence for glycosidic linkage of carbohydrates to proteins (35). A preliminary treatment of Fraction Al-D1 with 0.5 N NaOH at 23°C for 24 h caused destruction of 53% of the threonine and 66% of the serine. The loss of serine is consistent with the presence of a xylosyl-serine linkage of chondroitin sulfate to protein as described in cartilage PG by Lindahl and Rod& (36). The loss of threonine was unexpected because attachment of chondroitin sulfate chains to protein has generally been thought to be exclusively through serine (37). The elimination was therefore repeated with the inclusion of borohydride and PdCla as described by Downs et al. (29), who reported that 2-aminopropenoic and 2amino-2-butenoic acid residues produced from serine and threonine were quantitatively reduced to alanine and oc-aminobutyric acid. On treatment of Fraction Al-D& 55% of the threonine and 78% of the serine residues were destroyed ( Table V). The reduction of unsaturated amino acid residues was, however, not quantitative, since only 72% of the threonine destroyed was converted to a-aminobutyric acid, and 53% of  HCl, the most effective, removed 71% of the galactosamine of the whole tissue (Table I). Two distinct types of PG were obtained by sedimentation in density gradients, one of low buoyant density and high protein content (proteodermatan sulfate) and one with a density greater than 1.7 g/ml (proteochondroitin sulfate). The yield of the latter (Fraction Al-Dl) was 27% of the galactosamine of the dry skin powder, or about 0.75 mg of PG/g.
The proteochondroitin sulfate contained 4 to 5% of protein, and after papain digestion only chondroitin sulfate could be seen on cellulose-acetate electrophoresis (Fig. 3). This glycosaminoglycan had an M, of 20,000 and a content of 4-sulfate of 84%, like that of bovine nasal cartilage (38). About 8% of the hexosamine was glucosamine, not derived from hyaluronic acid, some of which might have come from traces of heparan sulfate since this compound was apparent in adjacent gradient fractions (Fig. 3) and since part of the glucosamine was lost on treatment with nitrous acid (Table IV). Some might also have arisen from keratan sulfate, as it does in cartilage PG (35)) or from oligosaccharides resembling keratan sulfate ( 17). In cartilage, keratan sulfate is linked to protein partly through threonine residues, and this could explain the loss of threonine as well as serine on /3 elimination of skin proteochondroitin sulfate (Table V). Other types of glucosamine-containing oligosaccharides, like those found in cornea1 PG (39), might also be present.
Skin proteochondroitin sulfate does not bind to hyaluronic acid, as shown by the absence of aggregate in Fraction Al from the associative gradient (Fig. 5) and by the lack of interaction of Fraction Al or Al-D1 with added hyaluronic acid. It is unlikely that this is the result of damage to a binding region, since cartilage PG prepared by the same route in the presence of skin shows normal aggregation.
The skin product, although smaller, resembles cartilage PG in having a low protein content and in its general amino acid composition (Table III). The absence of detectable cystine, however, is a striking difference and may be related to the lack of binding by guest on March 24, 2020 http://www.jbc.org/ Downloaded from to hyaluronic acid, since cystine disulfide linkages are essential to the functional integrity of the binding region of cartilage PG (4).
A speculative model of the proteochondroitin sulfate can be offered. A molecule of 10" daltons would have about 50 chains of chondroitin sulfate attached to serine residues. More than enough serine was lost on p elimination to account for these (62 mol/mol, Table V). The loss of threonine (16 mol/mol) might have resulted from the presence of keratan sulfate-like oligosaccharides as in PG of chondrosarcoma (17). At present there is no evidence for binding of PG to hyaluronic acid in any tissue except for cartilage and bovine aorta (40), and other macromolecular interactions must account for the very variable difficulty of extraction of PG from various types of connective tissue. Attempts have been made to explain the phenomena in terms of simple dissociation of PG aggregates (18) and in more complex physicochemical terms (41), but extractability may simply depend on the relative amounts of PG and collagen in the tissue and on whether the PG are entrapped in the structure of collagen fibers as well as in the interstices of the collagen network. With relatively mild extraction of the skin sample used here, more proteochondroitin sulfate (Al , Table  II) than proteodermatan sulfate (A6) was extracted. The latter had a protein content of 50 to 60% and had an M, of about 1.3 x lo" as determined by chromatography on Sepharose 6B,2 in contrast to the value of 2.9 x 10" reported by Gbrink (8) for proteodermatan sulfate extracted under much more drastic conditions.
The large size of his product may have resulted from irreversible changes during extraction. Toole and Lowther (7) and Gbrink (8) also noted some chondroitin sulfate in dense gradient fractions, but did not characterize it as PG. The present observation in skin of two PG with very different compositions and sizes raises questions of their roles in the molecular organization of the tissue.