567th MEETING, DURHAM

397

of aortic cells, the pinocytosis proceeded with high efficiency. The uptake remained linear with time for at most 24h and then reached a plateau, probably owing to the competition for uptake by newly secreted unlabelled proteoglycans. A considerable amount of added proteoglycans, however, became associated with the cell surface (pericellular pool of glycosaminoglycans) and are not removed by repeated washing with Hanks solution. The association of 35S-labelled proteoglycans with the pericellular pool is time-dependent and requires about 4 h until the steady state is reached, but only a small part of this pool represents material required for immediate pinocytosis. The incorporation of proteoglycans into the pericellular pool cannot be explained by adsorption, since this process is in part temperature-dependent. From the kinetics of adsorption and uptake processes it has to be concluded that binding for integration into the pericellular pool and for pinocytotic uptake appear as separate processes. This is suggested by the observation that the proteoglycan concentration that is required for a half-maximal saturation of the pericellular pool is lower than that for the half-maximal rate of pinocytosis. Adsorptive pinocytosis exhibits specificity, the individual proteoglycans being internalized at different rates. The highest rate of uptake was measured for a dermatan sulphate-rich proteoglycan. N o competition of uptake between dermatan sulphate-rich and heparan sulphate-rich proteoglycans was observed. Competition experiments suggest that the individual proteoglycans are recognized by different receptors on the cell surface. Structural integrity is one of the prerequisites for optimal pinocytosis. Proteolytic digestion of the protein core or elimination of the polysaccharide chain by alkali results in marked decrease in uptake by arterial fibroblasts.

Macromolecular Interactions and Connective-TissueMetabolism HELEN MUIR Biochemistry Division, Kennedy Institute of Rheumatology, Bute Gardens, Hammersmith, London W6 7 D W, U.K.

The particular characteristics of each type of connective tissue result from differences in the relative proportions of fibrillar proteins and other constituents, principally in the relative proportions of collagen and proteoglycan. The types of collagen and proteoglycan vary in different connective tissues and interactions between them will also affect the properties of the tissue. The situation in cartilage has perhaps been studied most extensively, where the contribution of collagen and proteoglycan to its mechanical properties has been established. Cartilage is a highly specialized form of connective tissue that can withstand compressive forces to a remarkable degree. Articular cartilage is able to distribute stress under impact loading (Freeman & Kempson, 1973) because it is not rigid but stiff, although it contains about 70-75% water (Muir et al., 1969). The fluid pressure within cartilage rises immediately a load is applied, but water is driven out only slowly because proteoglycans, which are highly hydrated compounds, are entrapped in the collagen network and impede the flow of interstitial water. Proteoglycans remain within the cartilage, however (Linn & Sokoloff, 1965), and when the load is released the water is reimbibed (Maroudas, 1973; Freeman & Kempson, 1973). The resilience or compressive stiffness of cartilage is therefore directly correlated with the proteoglycan content (Kempson et al., 1970), whereas tensile stiffness and tensile strength depend on the collagen content (Kempson et al., 1973). The molecular interactions of collagen and proteoglycan are likely to affect this situation. The decreased tensile stiffness of articular cartilage from areas adjacent to osteoarthritic lesions compared with cartilage from unaffected joints (Kempson et al., 1973) may be due to some

VOl. 5

BIOCHEMICAL SOCIETY TRANSACTIONS

398

Collagen

Protein-link

Hyaluronate

Proteogl wan aggregates

Disaggregated proteoglycan

Fig. 1. Equilibrium density-gradient centrifugation in CsCl (a) Associative conditions: starting density usually 1.6g/ml, centrifuged at 1000OOga,. for 48 h at 20°C. (6) Dissociative conditions: lower fraction from associative gradient mixed with an equal volume of 7.5~-guanidiniumchloride at pH5.8. Starting density was adjusted to 1.5g/ml with CsCI.

change in collagen-proteoglycan interactions, which in turn may alter the behaviour of interstitial water. Indeed osteoarthritic cartilage is more hydrated than normal cartilage, even in the earliest phase of the disease, and the proteoglycans are more readilyextracted (McDevitt et al., 1974, 1977; McDevitt & Muir, 1976), which suggests that some change in interaction has occurred. Cartilage proteoglycans are distinct from those of other connective tissues in containing keratan sulphate. They are of large molecular size with mol.wts. of & Phelps, 1967; Eyring & Yang, 1968; about 1x106 to 5 . 8 ~ 1 0(Luscombe ~ Rosenberg et al., 1970). They tend to be rather polydisperse and variable in chemical composition, particularly in the relative proportion of chondroitin sulphate to keratan sulphate and in protein content. They appear to consist of a continuous spectrum of similar compounds, rather than of distinct species or subfractions. An example of a remarkable type of macromolecular interaction is the ability of cartilage proteoglycans to form multimolecular aggregates of very high mol.wt. of the order of 50 x 106-100 x lo6 (reviewed by Muir & Hardingham, 1975). The possibility of aggregation of proteoglycans was first recognized by Mathews & Lozaityte (1958). However, the features of this unique type of aggregation were not identified until new techniques had been introduced to extract and purify cartilage proteoglycans (Sajdera & Hascall, 1969; Hascall & Sajdera, 1969). Dissociative procedures were used to extract proteoglycans without disruptive homogenization of the cartilage, and the proteoglycans were then purified by equilibrium density-gradient centrifugation in CsCI. Proteoglycans are separated at the bottom of the gradient from collagen and other constituents of lower carbohydrate content and hence of lower buoyant density. Hascall & Sajdera (1969) showed that proteoglycans prepared in this way contained both aggregated and non-aggregated species and that aggregatescould be dissociated in 4~-guanidiniumchloride. Under these conditions constituents of aggregates separate at different buoyant densities during a second CsC1-densitygradient centrifugation (Fig. 1). These procedures, which are now in general use in this field, have enabled considerable advances to be made in the understanding of cartilage proteoglycans. Proteoglycan aggregates have been obtained from many kinds of adult cartilage and also from rat chondrosarcoma (Oegema et al., 1975) and cultures of embryonic-chick chondrocytes (Hascall et al., 1976). Although it was originally thought that proteoglycan aggregation involved a protein that separated at the top of the second density gradient (Hascall & Sajdera, 1969), it is now recognized that aggregation depends on the interaction of proteoglycans with hyduronic acid. Hardingham & Muir (1972a) found that

1977

567th MEETING, DURHAM

399

dissociated proteoglycans interacted with hyaluronic acid in a unique manner. Hyaluronic acid was subsequently identified in the middle of the second dissociative ) accounted for rather less than 1 % density gradient (Hardingham & Muir, 1 9 7 4 ~ and of the total uronic acid in the aggregate. Since this interaction leads to a large increase in molecular size, viscometry and gel chromatography were used to study the stoicheiometry of the interaction (Hardingham & Muir, 1972~).It was concluded that many proteoglycan molecules became bound to a single chain of hyaluronic acid and that each possessed only a single binding site for hyaluronate, because a gel was not formed at higher proportions of hyaluronate. With the use of published molecular weights for proteoglycan and hyaluronic acid, a model was proposed (Hardingham & Muir, 19746) whose dimensions agreed reasonably well with those calculated from electron micrographs of aggregates published by Rosenberg et al. (1970) before the participation of hyaluronic acid in aggregation was known. Although the interaction is not covalent, it is extremely specific to hyaluronic acid and does not occur with any other glycosaminoglycan, even one as closely related as chondroitin (desulphated chondroitin sulphate; Hascall & Heineglrd, 19746), which differs from hyaluronate only in the conformation of the hydroxyl group on C-4 of the hexosamine residues. The chondroitin sulphate chains are not necessary for binding, which occurs even when the chains have been largely removed by prior digestion of proteoglycans with chondroitinase (Hascall & Heineglrd, 19746). The minimum length of hyaluronate that binds strongly is ten sugar residues, since oligosaccharides of hyaluronate compete strongly with hyaluronic acid and inhibit the interaction, whereas octasaccharides and smaller oligosaccharides have little effect (Hardingham & Muir, 1973 ;Hascall & Heinegird, 19746). The hyaluronate-binding region, however, is quite large. Sequential chondroitinase and trypsin digestion of the proteoglycanhyaluronate complex indicates that this region has a mol.wt. of about 90000 (HeinegArd & Hascall, 1974). Proteoglycan aggregation is prevented by reduction of cystine residues (Hascall & Sajdera, 1969). Although reduction and alkylation of proteoglycan monomers abolished hyaluronate interaction, there was no loss of protein or change in molecular size, and it is therefore concluded that the tertiary structure of the binding region depends on intramolecular disulphide bridges, of which there are about five to seven (Hardingham et al., 1976). The structure of the binding site is fairly sensitive to chemical modification, and amino groups appear to be important and may take part in direct subsite interactions with carboxyl groups of hyaluronate. Fluorescence measurements suggest, however, that intact tryptophan residues are probably not directly involved in binding, but are needed to maintain the conformation of the binding site (Hardingham et af., 1976). It appears that many co-operative subsite interactionstake part somewhat analogous to those involved in the binding of lysozyme with its hexasaccharidesubstrate (GlcNAc-MurNAc)a(Chipman & Sharon, 1969). Proteoglycan aggregates contain, in addition, a third constituent known as 'protein-link', which separates at the top of the dissociative density gradient (Hascall & Sajdera, 1969). It has yet to be fully characterized and may comprise two proteins with mol.wts. of about 45000 and 65000 (Hascall & HeinegArd, 1974u), although in chondrosarcoma (Oegema et af., 1975) and embryonic-chick cartilage (Hascall et af., 1976) only the smaller is present. 'Protein-link' binds to hyaluronate (Hascall & Heineghd, 1 9 7 4 ~ )and its function in the aggregate is to stabilize the proteoglycan-hyaluronate complex. The complex, unlike the aggregate, behaves as an equilibrium, although the equilibrium is far on the side of interaction (Hardingham & Muir, 1975). Oligosaccharides of hyaluronate that interact with proteoglycan (Hardingham & Muir, 1973 ; Hascall & HeinegArd, 19746) dissociate the proteoglycan-hyaluronate complex, which results in a decrease in viscosity, whereas aggregates that contain the 'protein-link' are unaffected by the oligosaccharides (Hascall & Heinegtird, 19746). The function of aggregation is unknown, but it must help to immobilize

VOl. 5

BIOCHEMICAL SOCIETY TRANSACTIONS

400 Keratan rulphate

I

Chondroitin rulphate

I

-

Hyaluronic acid C o r e protein binding site

Fig. 2. Diagram of cartilage proteoglycan molecule

proteoglycans in the collagen network and it may protect proteoglycans from the effects of proteinases, since controlled partial degradation of proteoglycan was possible only with aggregated proteoglycans, whereas monomers were rapidly degraded (Hascall & Heinegard, 1974a; Heinegard & Hascall, 1974). Although cartilage proteoglycans consist of a population of molecules that vary in molecular size, chemical composition and buoyant density, most are able to interact with hyaluronate, and it is therefore concluded that all molecules that are capable of interaction (Fig. 2) contain the same hyaluronate-binding site, which is an invariant region devoid of carbohydrate, together with a region bearing glycosaminoglycan chains, which is variable in length (Heinegard & Hascall, 1974; Hardingham et al., 1976). This conclusion is consistent with the results of electron-microscopic studies of proteoglycan monomers (Thyberg et al., 1975) and with chemical analyses of proteoglycans of different buoyant density (Hardingham et al., 1976; HeinegArd, 1975, 1976), where molecular size, carbohydrate content and buoyant density decreased together. Besides proteoglycans that interact with hyaluronate there are some that do not do so. These are not present as aggregates and may be preferentially extracted in 0.15M-NaCI. In several respects they appear to represent a different population of proteoglycans and no hyaluronate may be separated from them on dissociative densitygradient centrifugation (Hardingham & Muir, 1 9 7 4 ~ ) .Such proteoglycans are of relatively small molecular size and contain less protein and little keratan sulphate (Tsiganos & Muir, 1969) compared with aggregating proteoglycans. In pulse-chase experiments in vitro the incorporation of [35S]sulphateinto different proteoglycan fractions suggested that the smaller non-aggregating proteoglycans were neither precursors nor degradation products of aggregating proteoglycans (Hardingham & Muir, 19726). Proteoglycan synthesis is greatly stimulated by depletion of the matrix of embryonic-chick cartilage (Hardingham et al., 1972), and the possibility that one or more constituents of the matrix may influence proteoglycan synthesis has been investigated by several laboratories (Nevo & Dorfman, 1972; Toole, 1973; Kosher et al., 1973). Only hyaluronic acid appears to be specific, however. When present in the medium in low concentration it inhibited proteoglycan synthesis by suspensions of chondrocytes from adult cartilage (Wiebkin & Muir, 1973) and by cultures of embryonic chondrocytes (Toole, 1973; Solursh et al., 1974). This effect of hyaluronate appeared to resemble and was as specific as its interaction with proteoglycan, and likewise oligosaccharides of the size of decasaccharides or above also inhibited proteoglycan synthesis (Wiebkin et al., 1975).The inhibitory effect on proteoglycan synthesis involved the attachment of hyaluronate to the cell surface, and hence when already bound to proteoglycan as in the complex or in aggregates it had no effect (Wiebkin et af., 1975). It has been suggested (Handley & Lowther, 1976) that hyaluronate may decrease the synthesis of core protein, or the synthesis or activity of the first glycosyltransferase involved in chondroitin sulphate-chain synthesis. 1977

567th MEETING, DURHAM

401

Cartilage cells synthesize both collagen and proteoglycans, but the processes are not closely interrelated either in embryonic (Bhatnagar & Prockop, 1966) or in adult cartilage in vitro (Dondi & Muir, 1976), since interference in the synthesis of one does not affect the other. Besides proteoglycans that aggregate, type II collagen is phenotypic of cartilage (Miller & Matukas, 1974) and is found in mature hyaline, articularandelasticmammalian cartilages (Eyre & Muir, 1975) and also in nucleus pulposus and annulus fibrosus of intervertebral discs. The latter contains both type I and type I1 collagens (Eyre & Muir, 1974) distributed in smoothly interchangingproportions(Eyre &Muir, 1976). Type 11collagen contains more hydrolysine residues than do type I or type I11 collagens, a large proportion of which are substituted by glucosylgalactosyl residues (Nimni, 1974). In osteoarthrosis it has been suggested that there is a change in synthesis to type I collagen (Nimni & Deshmukh, 1973). Nevertheless, no type I collagen was detected as CNBr peptides in osteoarthrosis of human articular cartilage with severe fibrillation (D. R. Eyre, unpublished work). Moreover, in experimental osteoarthrosis the newly synthesized radioactively labelled collagen that was deposited in the matrix was the normal type I1 collagen (Eyre et al., 1975). It therefore seems improbable that there is a significant change in the type of collagen in this disease that leads to the functional defects of articular cartilage. Bhatnagar, R. S. & Prockop, D. J. (1966) Biochim. Biophys. Acta 130, 383-392 Chipman, D. M. & Sharon, N. (1969) Science 165,454-465 Dondi, P. & Muir, H. (1976) Biochem. J. 160,117-121 Eyre, D. R. & Muir, H. (1974) FEBS Lett. 42, 192-196 Eyre, D. R. & Muir, H. (1975) Biochem. J. 151, 595-602 Eye, D. R. & Muir, H. (1976) Biochem. J. 157, 267-270 Eyre, D. R., McDevitt, C. A. & Muir, H. (1975) Ann. Rheum. Dis. 34, Suppl. 2,137-140 Eyring, E. J. & Yang, J. T. (1968) J. Biol. Chem. 243, 130&1311 Freeman, M. A. R. & Kenipson, E. G. (1973) in Adult Articular Cartilage (Freeman, M. A. R., ed.), pp. 228-246, Pitman Medical, London Handley, C. J. & Lowther, D. A. (1976) Biochim. Biophys. Acta 444,69-74 Hardingham, T. E. & Muir, H. (1972~)Biochim. Biophys. Acta 279,401-405 Hardingham, T. E. & Muir, H. (19726) Biochem. J. 126, 791-803 Hardingham, T. E. & Muir, H. (1973) Biochem. J. 135,905-908 Hardingham, T. E. & Muir, H. (1974~)Biochem. J . 139, 565-581 Hardingham, T. E. & Muir, H. (19746) in Normal and Osteoarthrotic Articular Cartilage (Ali, S. Y., Elves, M. W. & Leaback, D. H., eds.), pp. 51-63, Institute of Orthopaedics, London Hardingham, T. E. & Muir, H. (1975) Ann. Rheum. Dis.34, Suppl. 2,26-28 Hardingham, T. E., Fitton Jackson, S. & Muir, H. (1972) Biochem. J. 129, 101-112 Hardingham, T. E., Ewins, R. J. F. & Muir, H. (1976) Biochem. J. 156, 127-143 Hascall, V. C. & Heinegird, D. (1974~)J. Biol. Chem. 249,42324241 Hascall, V . C. & Heinegird, D. (19746) J. BioI. Chem. 249,4242-4249 Hascall. V . C . & Sajdera, S. W. (1969) J. Biol. Chem. 244,2384-2396 Hascall, V . C., Oegema, T. R. &Brown, M. (1976) J. Biol. Chem. in the press Heineghd, D. (1975) Ann. Rheum. Dis. 34, suppl. 2,29-31 Heineghd, D. (1976) J. Bid. Chem. in the press Heineg&rd,D. & Hascall, V. C. (1974) J. B i d . Chem. 249,4250-4256 Kempson, G . E., Muir, H., Swanson, S. A. V. & Freeman, M. A. R. (1970) Biochim. Biophys. Acta 215, 70-77

Kempson, G. E., Muir, H., Pollard, C. & Tuke, M. (1973) Biochim. Biophys. Acta 297,454-472 Kosher, R. A., Lash, J. W. & Minor, R. R. (1973) Deo. Biol. 35, 210-220 Linn, F. C. & Sokoloff, L. (1965) Arthritis Rheum. 8,481-494 Luscombe, M. & Phelps, C. F. (1967) Biochem. J. 102, 110-119 Maroudas, A. (1973) in Adult Articular Cartilage (Freeman, M. A. R., ed.), pp. 131-170, Pitman Medical, London Mathews, M. B. & Lozaityte, I. (1958) Arch. Biochem. Biophys. 74, 158-174 McDevitt, C. A. & Muir, H. (1976) J. Bone Jt. Surg. 58-B,94-101

Vol. 5

402

BIOCHEMICAL

SOCIETY TRANSACIIONS

McDevitt, C. A., Muir, H. & Pond, M. J. (1974) in Normal and Osteoarthrotic Articular Cartilage (Ali, S. T., Elves, M. W. & Leaback, D. H., eds.), pp. 207-217, Institute of Orthopaedics,London McDevitt, C. A., Gilbertson, E. & Muir, H. (1977) J. BoneJt. Surg. 59-B,24-35 Miller, E. J. & Matukas, V. J. (1974) Fed. Proc. Fed. Am. SOC.Exp. Biol. 33,1197 Muir, H. & Hardingham, T. E. (1975) in Biochemistry of Carbohydrates: International Review of Science (Whelan, W. J., ed.), pp. 153-222, Butterworths, London and University Park Press, Baltimore Muir, H., Maroudas, A. & Wingham, J. (1969) Biochim. Biophys. Acta 177, 492-500 Nevo, Z. & Dorfman, A. (1972) Proc. Natl. Acad. Sci. U.S.A. 69, 2069-2072 N i b , M. E. (1974) Semin. Arthritis Rheum. 4, 95-150 N i b , M. E. & Deshmukh, K. (1973) Science 181,751-752 Oegema, T. R., Hascall, V. C. & Dziewiatkowski, D. D. (1975)J. Biol. Chem. 250,6151-6159 Rosenberg, L., Hellman, W. & Kleinschmidt, A. K. (1970) J. Biol. Chem. 245,4123-4130 Sajdera, S . W. & Hascall, V. C. (1969) J. Biol. Chem. 244,77-87 Solursh, M., Vaerewyck, S. A. & Reiter, R. S. (1974) Deu. Biol. 41, 233-244 Thyberg, J., Lohmander, S. & HeinegArd, D. (1975) Biochem. J. 151, 157-166 Toole, B. P. (1973) Am. Zool. 13, 1061-1065 Tsiganos, C. P. & Muir, H. (1969) Biochem. J. 113,885-894 Wiebkin, 0. W. & Muir, H. (1973) FEBSLett. 37,4246 Wiebkin, 0. W., Hardingham, T. E. & Muir, H. (1975) in Dynamics of Conneetioe Tissue Macromolecules (Burleigh, P. M. C. & Poole, A. R., eds.), pp. 81-104, North-Holland Publishing Co.,Amsterdam

Concepts of Metabolic Pools in the Metabolism of Proteoglycans and Hyaluronic Acid JOHN T. GALLAGHER* Clycoprotein Research Unit, University Science Laboratories, University of Durham, South Road, Durham DH1 3LE, U.K.

( I ) Proteoglycans Aggregated proteoglycans are the major extracellular macromolecules of the polysaccharide-rich ground substance of cartilage. The so-called proteoglycan subunit consists of a single core protein, to which chondroitin sulphate and keratan sulphate chains are covalently linked (Muir, 1958; Partridge et al., 1961;Anderson et al., 1965; Tsiganos & Muir, 1967; Heinegkrd, 1972);formation of proteoglycan aggregates depends on a specific interaction in which many proteoglycan subunit molecules bind to a single molecule of hyaluronic acid (Hardingham & Muir, 1974), the resulting complex beind stabilized by ‘protein-link components’ (Hascall & HeinegBrd, 1974). Proteoglycan biosynthesis. Proteoglycan biosynthesis occurs through the concerted action of multi-glycosyltransferase systems responsible for the covalent attachment and assembly of chondroitin sulphate and keratan sulphate to precursor ‘core’ proteins. Metabolic heterogeneity of proteoglycans is implied from results showing that they contain reveral ‘core’ proteins differing in molecular size, amino acid composition and antigenic properties (Tsiganos et al., 1971; Heinegkrd, 1972; Brandt et al., 1973; Baxter & Muir, 1975). It has been generally assumed that superimposed on differences in the structure of the protein core is a wide variation in the number of attached chondroitin sulphate chains (Hascall & Sajdera, 1970). Proteoglycan heterogeneity may be exaggerated, however, from the use of pooled tissue preparations.Inastudyofbovinenasalcartilageobtainedfromasingleanimal,Hopwood

* Present address: Department of Medical Oncology, Christie Hospital, Wilmslow Road, Manchester M20 9BX, U.K. 1977