Immunohistochemical localization of articular cartilage proteoglycan and link protein in situ using monoclonal antibodies and lectin-binding methods

Lectins have specificity for certain carbohydrate structures in macromolecules. Lectins are, therefore, useful histochemical tools for demonstrating the composition and localization of components of connective tissue matrices, such as articular cartilage. In order to assess the significance of observed lectin-binding patterns, experiments were performed in which monoclonal antibodies against chondroitin sulphate- and keratan sulphate-containing proteolgycans and link proteins were applied to sections of bovine articular cartilage after enzymatic digestion with chondroitinase ABC and keratanase. The following conclusions were made: (1) Binding of peanut agglutinin (PNA) in the interterritorial matrix predominantly indicates the presence of keratan sulphate, but may also detectO-linked oligosaccharides of proteoglycans. (2) In normal cartilage wheat germ agglutinin (WGA) binds nearly exclusively to keratan sulphate. In cartilage degraded with chondroitinase ABC and keratanase this lectin may also detect carbohydrates in link protein due to enhanced accessibility. Binding of WGA toO-linked oligosaccharides may eventually occur. (3) In enzymatically digested cartilage matrix, staining with soybean agglutinin (SBA) may be due to link protein, but not to chondroitin sulphate, because specific breakdown of the glycosaminoglycan chain is required for binding of SBA. (4)Ulex europaeus agglutinin I (UEA I) binding sites are only detectable in digested cartilage matrix.     

Proteoglycans of the intervertebral disc. Homology of structure with laryngeal proteoglycans.

The structure of the proteoglycans from normal pig nucleus pulposus and relatively normal human annulus fibrosus and nucleus pulposus was investigated in detail and the results were compared with the current structural model of proteoglycans of hyaline cartilage. Like proteoglycans of cartilage, those of intervertebral disc contain keratan sulphate and chondroitin sulphate attached to a protein core; they are able to aggregate to hyaluronic acid; the protein core likewise has three regions, one lacking glycosaminoglycans, another rich in keratan sulphate and a third region rich in chondroitin sulphate. However, disc proteoglycans contain more keratan sulphate and protein and less chondroitin sulphate and are also considerably smaller than cartilage proteoglycans. In proteoglycans of human discs, these differences appeared to be due principally to a shorter region of the core protein bearing the chondroitin sulphate chains, whereas in proteoglycans of pig discs their smaller size and relatively low uronic acid content were due to shorter chondroitin sulphate chains. There were subtle differences between proteoglycans from the nucleus and annulus of human discs. In the latter a higher proportion of proteoglycans was capable of binding to hyaluronate.     

Articular-cartilage proteoglycans in aging and osteoarthritis.

The composition of macroscopically normal hip articular cartilage obtained from dogs of various ages was studied. Pieces of cartilage with signs of degeneration were studied separately. In normal aging, the extraction yield of proteoglycans decreased; the keratan sulphate content of extracted proteoglycans increased and the chondroitin sulphate content decreased. The extracted proteoglycans were smaller in the older cartilage, mainly owing to a decrease in the chondroitin sulphate-rich region of the proteoglycan monomers. The hyaluronic acid-binding region and the keratan sulphaterich region were increased and the molar concentration of proteoglycan probably increase with increasing age. The degenerated cartilage had higher water content and the proteoglycans, as well as other tissue components, gave higher yields. The proteoglycan monomers from the degenerated cartilage were smaller than those from normal cartilage of the same age, and hence had a smaller chondroitin sulphate-rich region and some of the molecules also appeared to lack the hyaluronic acid-binding region. Increased proteolytic activity may be involved in the process of cartilage degeneration.     

Monoclonal antibodies to different protein-related epitopes of human articular cartilage proteoglycans.

Monoclonal antibodies produced against chondroitinase-treated human adult cartilage proteoglycans were selected for their ability to recognize epitopes on native proteoglycans. Binding analyses revealed that four of these monoclonal antibodies (BCD-4, BCD-7, EFG-4 and KPC-190) each recognized a different epitope on the same proteoglycan molecule which represents a subpopulation of a high buoyant density (D1) fraction of human articular cartilage proteoglycans (10, 30, 50 and 60% in fetal-newborn, 1.5 years old, 15 years old and 52-56 years old cartilages, respectively). Analysis of epitope specificities revealed that BCD-7 and EFG-4 monoclonal antibodies recognized epitopes on proteoglycan monomer which are associated with the protein structure in that they are sensitive to cleavage by Pronase, papain and alkali treatment and do not include keratan sulphate, chondroitin sulphate or oligosaccharides. The BCD-4 and KPC-190 epitopes also proved to be sensitive to Pronase or papain digestion or to alkali treatment, but keratanase or endo-beta-galactosidase also reduced the immunoreactivity of these epitopes. These observations indicate that the BCD-4 and KPC-190 epitopes represent peptides substituted with keratan sulphate or keratan sulphate-like structures. The BCD-4 epitope is, however, absent from a keratan sulphate-rich fragment of human adult proteoglycan, while the other three epitopes were detected in this fragment. None of these four epitopes were detected in the link proteins of human cartilage, in the hyaluronic acid-binding region of human newborn cartilage proteoglycan, in Swarm rat chondrosarcoma proteoglycan, in chicken limb bud proteoglycan monomer and in the small dermatan sulphate-proteoglycan of bovine costal cartilage. EFG-4 and KPC-190 epitopes were not detected in human fetal cartilage proteoglycans, although fetal molecules contained trace amounts of epitopes reactive with BCD-4 and BCD-7 antibodies.     

Fractionation of proteoglycans from bovine corneal stroma.

Proteoglycans were extracted from bovine corneal stroma with 4M-guanidinum chloride, purified by DEAE-dellulose chromatography (Antonopoulos et al., 1974) and fractionated by precipitation with ethanol into three fractions of approximately equal weight. One of these fractions consisted of a proteoglycan that contained keratan sulphate as the only glycosaminoglycan. In the othertwo fractions proteoglycans that contained chondroitin sulphate, dermatan sulphate and keratan sulphate were present. Proteoglycans which had a more than tenfold excess of galactosaminoglycans over keratan sulphate could be obtianed by further subfractionation. The gel-chromatographic patterns of the glucosaminoglycans before and after digestion with chondroitinase AC differed for the fractions. The individual chondroitin sulphate chains seemed to be larger in cornea than in cartilage. Oligosaccharides, possibly covalently linked to the protein core of the proteoglycans, could be isolated from all fractions. The corneal proteoglycans were shown to have higher protein contents and to be of smaller molecular size than cartilage proteoglycans.     

Variations in the composition of bovine hip articular cartilage with distance from the articular surface.

Punch biopsies of bovine hip articular cartilage was sectioned according to depth and the proteoglycans were isolated. The mid-sections of the cartilage contained more proteoglycans than did either the superficial or the deepest portions of the cartilage proteoglycans than did either the superficial or the deepest portions of the cartilage. The most superficial 40 micrometer of the cartilage contained relatively more glucosaminoglycans compared with the remainder of the cartilage. The proteoglycans recovered from the surface 200 micrometer layer contained less chondroitin sulphate, were smaller and almost all of these molecules were able to interact with hyaluronic acid to form aggregates. From about 200 micrometer and down to 1040 micrometer from the surface, the proteoglycans became gradually somewhat smaller, probably owing to decreasing size of the chondroitin sulphate-rich region. The proportion of molecules that were able to interact with the hyaluronic acid was about 90% and remained constant with depth. The proteoglycans from the deepest layer near the cartilage-bone junction contained a large proportion of non-aggregating molecules, and the average size of the proteoglycans was somewhat larger. The alterations of proteoglycan structure observed with increasing depth of the articular cartilage beneath the surface layer (to 200 micrometer) are of the same nature as those observed with increasing age in full-thickness articular cartilage. The articular-cartilage proteoglycans were smaller and had much higher keratan sulphate and protein contents that did molecules isolated from bovine nasal or tracheal cartilage.     

The detection of substructures within proteoglycan molecules. Electron-microscopic immuno-localization with the use of Protein A-gold.

Proteoglycan monomers from pig laryngeal cartilage were examined by electron microscopy with benzyldimethylalkylammonium chloride as the spreading agent. The proteoglycans appeared as extended molecules with a beaded structure, representing the chondroitin sulphate chains collapsed around the protein core. Often a fine filamentous tail was present at one end. Substructures within proteoglycan molecules were localized by incubation with specific antibodies followed by Protein A-gold (diameter 4 nm). After the use of an anti-(binding region) serum the Protein A-gold (typically one to three particles) bound at the extreme end of the filamentous region. A small proportion of the labelled molecules (10-15%) showed the presence of gold particles at both ends. A monoclonal antibody specific for a keratan sulphate epitope (MZ15) localized a keratan sulphate-rich region at one end of the proteoglycan, but gold particles were not observed along the extended part of the protein core. This distribution was not changed by prior chondroitin AC lyase digestion of the proteoglycan. Localization with a different monoclonal antibody to keratan sulphate (5-D-4) caused a change in the spreading behaviour of a proportion (approx. 20%) of the proteoglycan monomers that lost their beaded structure and appeared with the chondroitin sulphate chains projecting from the protein core. In these molecules the Protein A-gold localized antibody (5-D-4) along the length of the protein core whereas in those molecules with a beaded appearance it labelled only at one end. Labelling with either of the monoclonal antibodies was specific, as it was inhibited by exogenously added keratan sulphate. The differential localization achieved may reflect structural differences within the proteoglycan population involving keratan sulphate and the protein core to which it is attached. The results showed that by this technique substructures within proteoglycan molecules can be identified by Protein A-gold labelling after the use of specific monoclonal or polyclonal antibodies.     

Articular cartilage cultured with interleukin 1. Increased release of link protein, hyaluronate-binding region and other proteoglycan fragments.

Pig articular cartilage was maintained in culture for 3 days with and without porcine interleukin 1. The proteoglycans remaining in the cartilage and those released into the medium were analysed by using radioimmunoassays for the hyaluronate-binding region, link protein and keratan sulphate. In interleukin 1-treated cultures after 3 days there was 38% release of total glycosaminoglycans into the medium, 18% release of binding region, 14% release of link protein and 20% release of keratan sulphate epitope, whereas in control cultures the proportions released were much less (16, 9, 10 and 7% respectively). Characterization of the proteoglycans in the media after 1.5 days and 3 days of culture showed that interleukin 1 promoted the release of proteoglycan of large average size and also the release of link protein and of low-Mr binding region which was unattached to proteoglycan. Both the link protein and binding region released were able to bind to exogenously added hyaluronate, whereas the proteoglycan in the medium was not. The proteoglycans extracted from cultured cartilage were similar to those from fresh cartilage: they contained a high proportion of aggregating proteoglycans and some low-Mr binding region. The proportion of this binding region extracted from the interleukin 1-treated cartilage was increased. The presence of interleukin 1 in the cultures therefore appeared to increase the rate of proteolytic degradation of proteoglycan in the matrix and to lead to a more rapid loss of intact binding region, of link protein and of large proteoglycan fragments into the medium.     

Comparisons between extracted and residual proteoglycans on the glycosaminoglycan level and changes with ageing

Sheep nasal cartilage from animals of five different ages were studied. Significant variations in the composition and on the extractability of the tissue occur with ageing. The ratio chondroitin sulphate to keratan sulphate in extracted and residual proteoglycans does not change in the same manner with ageing. The relative distribution of molecular sizes of keratan sulphate differs between extracted and residual proteoglycans. Hyaluronic acid and chondroitin sulphate appear homogeneous on the gel chromatography for all ages both in extracted and residual proteoglycans.     

Separation and characterization of two populations of aggregating proteoglycans from cartilage.

Intermediary gel immunoelectrophoresis was used to show that purified aggregating cartilage proteoglycans from 2-year-old steers contain two distinct populations of molecules and that only one of these is immunologically related to non-aggregating cartilage proteoglycans. The two types of aggregating proteoglycans were purified by density-gradient centrifugation in 3.5M-CsCl/4M-guanidinium chloride and separated by zonal rate centrifugation in sucrose gradients. The higher-buoyant-density faster-sedimenting proteoglycan represented 43% of the proteoglycans in the extract. It had a weight-average Mr of 3.5 X 10(6), did not contain a well-defined keratan sulphate-rich region, had a quantitatively dominant chondroitin sulphate-rich region and contained 5.9% protein and 23% hexosamine. The lower-buoyant-density, more slowly sedimenting, proteoglycan represented 15% of the proteoglycans in the extract. It had a weight-average Mr of 1.3 X 10(6), contained both the keratan sulphate-rich and the chondroitin sulphate-rich regions and contained 7.3% protein and 23% hexosamine. Each of the proteoglycan preparations showed only one band on agarose/polyacrylamide-gel electrophoresis. The larger proteoglycan had a lower mobility than the smaller. The distribution of chondroitin sulphate chains along the chondroitin sulphate-rich region was similar for the two types of proteoglycans. The somewhat larger chondroitin sulphate chains of the larger proteoglycan could not alone account for the larger size of the proteoglycan. Peptide patterns after trypsin digestion of the proteoglycans showed great similarities, although the presence of a few peptides not shared by both populations indicates that the core proteins are partially different.     

Age-related changes of proteoglycan subunits from sheep nasal cartilage

Proteoglycan subunits of sheep nasal cartilage from animals of five different ages were studied. There is a continuous reduction in the size and chondroitin sulphate content of the aggregable and non-aggregable subunits with ageing. For each age group, the non-aggregable are poorer in protein and keratan sulphate than the corresponding aggregable molecules. Irrespective of age, the amount of proteoglycan protein extracted from each gramme wet cartilage is the same. The amino acid composition and the proportion of the aggregable proteoglycans are also the same.     

Enhanced extractability of articular cartilage proteoglycans in osteoarthrosis (Short Communication)

Tissue contents of small, easily extracted, proteoglycans, relatively poor in keratan sulphate, were compared in normal and osteoarthrotic cartilage. Although the amounts of small proteoglycans were similar in each tissue, as were the collagen contents, some proteoglycans in the diseased cartilage were much more readily extracted than those in the normal tissue.     

Spatial and temporal changes in the distribution of proteoglycans during avian neural crest development

In this study, we describe the distribution of various classes of proteoglycans and their potential matrix ligand, hyaluronan, during neural crest development in the trunk region of the chicken embryo. Different types of chondroitin and keratan sulfate proteoglycans were recognized using a panel of monoclonal antibodies produced against specific epitopes on their glycosaminoglycan chains. A heparan sulfate proteoglycan was identified by an antibody against its core protein. The distribution of hyaluronan was mapped using a biotinylated fragment that corresponds to the hyaluronan-binding region of cartilage proteoglycans. Four major patterns of proteoglycan immunoreactivity were observed. (1) Chondroitin-6-sulfate-rich proteoglycans and certain keratin sulfate proteoglycans were absent from regions containing migrating neural crest cells, but were present in interstitial matrices and basement membranes along prospective migratory pathways such as the ventral portion of the sclerotome. Although initially distributed uniformly along the rostrocaudal extent of the sclerotome, these proteoglycans became rearranged to the caudal portion of the sclerotome with progressive migration of neural crest cells through the rostral sclerotome and their aggregation into peripheral ganglia. (2) A subset of chondroitin/keratan sulfate proteoglycans bearing primarily unsulfated chondroitin chains was observed exclusively in regions where neural crest cells were absent or delayed from entering, such as the perinotochordal and subepidermal spaces. (3) A subset of chondroitin/keratan sulfate proteoglycans was restricted to the perinotochordal region and, following gangliogenesis, was arranged in a metameric pattern corresponding to the sites where presumptive vertebral arches form. (4) Certain keratan sulfate proteoglycans and a heparan sulfate proteoglycan were observed in basement membranes and in an interstitial matrix uniformly distributed along the rostrocaudal extent of the sclerotome. After gangliogenesis, the neural crest-derived dorsal root and sympathetic ganglia contained both these proteoglycan types, but were essentially free of other chondroitin/keratan-proteoglycan subsets. Hyaluronan generally colocalized with the first set of proteoglycans, but also was concentrated around migrating neural crest cells and was reduced in neural crest-derived ganglia. These observations demonstrate that proteoglycans have diverse and dynamic distributions during times of neural crest development and chondrogenesis of the presumptive vertebrae. In general, chondroitin/keratan sulfate proteoglycans are abundant in regions where neural crest cells are absent, and their segmental distribution inversely correlates with that of neural crest-derived ganglia.     

Chondroitin sulphate and heparan sulphate sulphation motifs and their proteoglycans are involved in articular cartilage formation during human foetal knee joint development

Novel sulphation motifs within the glycosaminoglycan chain structure of chondroitin sulphate (CS) containing proteoglycans (PGs) are associated with sites of growth, differentiation and repair in many biological systems and there is compelling evidence that they function as molecular recognition sites that are involved in the binding, sequestration or presentation of soluble signalling molecules (e.g. morphogens, growth factors and cytokines). Here, using monoclonal antibodies 3B3(-), 4C3 and 7D4, we examine the distribution of native CS sulphation motifs within the developing connective tissues of the human foetal knee joint, both during and after joint cavitation. We show that the CS motifs have broad, overlapping distributions within the differentiating connective tissues before the joint has fully cavitated; however, after cavitation, they all localise very specifically to the presumptive articular cartilage tissue. Comparisons with the labelling patterns of heparan sulphate (HS), HS-PGs (perlecan, syndecan-4 and glypican-6) and FGF-2, molecules with known signalling roles in development, indicate that these also become localised to the future articular cartilage tissue after joint cavitation. Furthermore, they display interesting, overlapping distributions with the CS motifs, reflective of early tissue zonation. The overlapping expression patterns of these molecules at this site suggests they are involved, or co-participate, in early morphogenetic events underlying articular cartilage formation; thus having potential clinical relevance to mechanisms involved in its repair/regeneration. We propose that these CS sulphation motifs are involved in modulating the signalling gradients responsible for the cellular behaviours (proliferation, differentiation, matrix turnover) that shape the zonal tissue architecture present in mature articular cartilage.     

Structure of proteoglycans from different layers of human articular cartilage

Full-depth plugs of adult human articular cartilage were cut into serial slices from the articular surface and analysed for their glycosaminoglycan content. The amount of chondroitin sulphate was highest in the mid-zone, whereas keratan sulphate increased progressively through the depth. Proteoglycans were isolated from each layer by extraction with 4M-guanidinium chloride followed by centrifugation in 0.4M-guanidinium chloride/CsCl at a starting density of 1.5 g/ml. The efficiency with which proteoglycans were extracted depended on slice thickness, and extraction was complete only when cartilage from each zone was sectioned at 20 microns or less. When thick sections (250 microns) were extracted, hyaluronic acid was retained in the tissue. Most of the proteoglycans, extracted from each layer under optimum conditions, could interact with hyaluronic acid to form aggregates, although the extent of aggregation was less in the deeper layers. Two pools of proteoglycan were identified in all layers by gel chromatography (Kav. 0.33 and 0.58). The smaller of these was rich in keratan sulphate and protein, and gradually increased in proportion through the cartilage depth. Chondroitin sulphate chain size was constant in all regions. The changes in composition and structure observed were consistent with the current model for hyaline-cartilage proteoglycans and were similar to those observed with increasing age in human articular cartilage.     

Lack of chondroitin sulphate epitope in the proliferating zone of the growth plate of chicken tibia

Summary Monoclonal antibodies specific to chondroitin sulphate (CS-56) and keratan sulphate (AH12) were used to localize proteoglycans in the proximal tibial articular cartilage and growth plate of broiler chickens. There was no CS-56 labelling in the proliferative zone of the growth plate. In contrast, intense labelling with this antibody was observed in the transitional and hypertrophic zones of the growth plate and the articular cartilage. This was confirmed by extracting chondroitin sulphate fractions from different zones of the growth plate and articular cartilage, and examining their antigenicities to CS-56 by ELISA inhibition assay. It was suggested that the maturation of chondrocytes in the growth plate is related to the production of chondroitin sulphate with CS-56 epitope, which may be a prerequisite for normal endochondral bone formation in the chicken tibia. The role of chondroitin sulphate recognized by CS-56 in the articular cartilage is unknown.     

Studies on the in vitro incubation procedure for [35]sulphate labelling of articular cartilage glycosaminoglycans

Articular cartilage from cow and calf femoral condyles was incubated in Tyrodes solution containing [³⁵S]sulphate for different periods up to 80 min. Glycosaminoglycans from the cartilage tissue and incubation medium were fractionated on Cetylpyridinium chloride and ECTEOLA cellulose microcolumns.The incorporation of [³⁵S]sulphate into all individual fractions of chondroitin sulphate and keratan sulphate was found to be linear from 20 to 80 min incubation time. As a rule the total specific activities of keratan sulphate and chondroitin sulphate were similar for both calves and cows.The proteoglycan material recovered from the medium amounted to about 1% of the tissue dry weight and was found to have a higher chondroitin sulphate: keratan sulphate ratio than the corresponding cartilage tissue for both calf and cow.The solubility profiles for the newly synthesised glycosaminoglycans, obtained from determination of the radioactivity in the individual fractions, were compared with those of glycosaminoglycans already present. These curves indicated that newly synthesised chondroitin sulphate had a higher average molecular size than that present in the tissue whereas the newly synthesised keratan sulphate had a smaller average molecular size. These newly synthesised components were also detected in the proteoglycans recovered from the incubation medium.     

The core molecule from type H proteoglycan. Release of mannose-containing oligosaccharides by digestion with N-oligosaccharide glycopeptidase.

Chick-embryo cartilage contains a unique set of proteoglycans. Type H proteoglycan (PG-H) is the most abundant, constituting over 90% of the total cartilage hexuronate. We previously showed that treatment of PG-H with chondroitinase ACII and keratanase yields a protein-enriched core molecule [PG(-CS,KS)] with enzymically modified linkage oligosaccharides of the chondroitin sulphate and keratan sulphate chains. We report here that further treatment of PG(-CS,KS) with pepsin and N-oligosaccharide glycopeptidase (almond glycopeptidase) released four distinct types of mannose-containing oligosaccharide. Two of them were shown to be: (Formula: see text). Of the mannose-containing glycopeptides formed by pepsin digestion, about 40% (as mannose) were resistant to N-oligosaccharide glycopeptidase. Since the resistant fraction was enriched in keratan sulphate remnants, it is suggest that the mannose-containing oligosaccharides in this fraction represent those located in a keratan sulphate-enriched region of PG-H.     

Age-related changes in the composition and structure of human articular-cartilage proteoglycans.

1. Analysis of the purified proteoglycans extracted from normal human articular cartilage with 4M-guanidinium chloride showed that there was an age-related increase in their content of protein and keratan sulphate. 2. The hydrodynamic size of the dissociated proteoglycans also decreased with advancing age, but there was little change in the proportion that could aggregate. 3. Results suggested that some extracts of aged-human cartilage had an increased content of hyaluronic acid compared with specimens from younger patients. 4. Dissociated proteoglycans, from cartilage of all age groups, bind to hyaluronic acid and form aggregates in direct proportion to the hyaluronic acid concentration. 5. Electrophoretic heterogeneity of the dissociated proteoglycans was demonstrated on polyacrylamide/agarose gels. The number of proteoglycan species observed was also dependent on the age of the patient.     

Specific binding of peanut agglutinin and soybean agglutinin to chondroitinase ABC-digested cartilage proteoglycans: Histochemical, ultrastructural cytochemical, and biochemical characterization

Summary The binding of peanut agglutinin (PNA) and soybean agglutinin (SBA) to cartilage proteoglycans was investigated by histochemical, ultrastructural cytochemical, and biochemical methods. Following aldehyde fixation, specimens of rat epiphyseal cartilage were examined by horseradish peroxidase-labelled lectin cytochemistry with and without prior digestion in chondroitinase ABC. At the light microscope level neither PNA nor SBA exhibited any affinity for cartilage matrix, but became strongly bound following chondroitinase treatment. Similarly, at the ultrastructural level, extracellular matrix granules, presumed to be proteoglycan monomer(s), lacked PNA affinity in undigested specimens, and stained very weakly with SBA. Both PNA and SBA weakly to moderately stained thetrans cisternae of the Golgi-flattened cisternae in chondrocytes. The chondrocyte plasmalemma lacked PNA staining, but reacted weakly with SBA. Following chondroitinase digestion, PNA and SBA stained matrix granules, and the cell surface of chondrocytes intensely, whereas the Golgitrans cisternae, the Golgi-derived vacuoles, and multivesicular bodies demonstrated weak to moderate reactivity. Proteoglycan aggregates purified from rat chondrosarcoma and bovine nasal cartilage bound PNA and SBA avidly after digestion with chondroitinase. Undigested proteoglycans lacked affinity for PNA and reacted very weakly with SBA. These results indicate that both PNA and SBA specifically react with chondroitinase-modified oligosaccharide(s) bound to core proteins of cartilage proteoglycans. This provided a specific histochemical and ultrastructural cytochemical procedure for localizing chondroitin sulphate-containing proteoglycans.