Internalization of hyaluronan by chondrocytes occurs via receptor-mediated endocytosis

The integrity of cartilage extracellular matrix is essential for the load-bearing function of the tissue. Matrix turnover, even in minor amounts, must be carefully controlled and coordinated to ensure this integrity. Cartilage extracellular matrices are composed predominately of collagen and proteoglycans. Often more than 50 high molecular mass proteoglycans, termed “aggrecan”, become bound to a single filament of hyaluronan (HA) (Heinegard and Paulsson, 1984; Buckwalter and Rosenberg, 1982; Buckwalter et al., 1985; Rosenberg and Buckwalter, 1986). HA is itself a high molecular mass, nonsulfated glycosaminoglycan devoid of a covalent core protein (Laurent and Fraser, 1992). These HA/proteoglycan aggregates, with molecular masses between 10⁷ and 10⁸ Da (Rosenberg and Buckwalter, 1986; Buckwalter and Rosenberg, 1982; Buckwalter et al., 1985; Pita et al., 1979), are maintained at high concentrations within the cartilage matrix in order to create the osmotic swelling forces that resist compression.

The turnover of HA and proteoglycan have been shown to be coordinately regulated, as they have similar half-lives in adult tissues (Morales and Hascall, 1988; Ng et al., 1992). On the other hand, whereas a large portion of partially degraded proteoglycan is lost from the tissue in the process of matrix turnover, little intact HA or small degraded fragments of HA are shed from the tissue (Morales and Hascall, 1988; Ng et al., 1992). Also, small degraded fragments of HA are not found within the extracellular matrix of the tissue. Rather, the HA is removed by a different, unknown mechanism (Morales and Hascall, 1988; Ng et al., 1992). Most studies on cartilage catabolism have centered around extracellular processing of matrix components (Roughley et al., 1992; Tyler et al., 1992; Ilic et al., 1992), i.e. expression of collagenases, gelatinases, stromolysins as well as their corresponding protease inhibitors. An “aggrecanase” has been implicated in the proteolytic dissolution of proteoglycans from HA/proteoglycan aggregates (Sandy et al., 1991; Ilic et al., 1992). Nevertheless, no enzymes for the extracellular processing of cartilage HA have been documented (Kresse and Glossl, 1987). Except for specialized tissues such as the testes (Meyer, 1971) and possibly the embryonic salivary gland (Bernfield et al., 1984), most mammalian tissues, including cartilage, exhibit only a low pH, lysosomal form of hyaluronidase (McGuire et al., 1992; Toole, 1972; Orkin et al., 1982; Bernanke and Orkin, 1984). Therefore, except for extracellular free-radical mechanisms that have been proposed (Ng et al., 1992; Roughley et al., 1992; Baker et al., 1989), catabolism of HA must occur within the chondrocyte.

Recent evidence from our laboratory has demonstrated that HA as well as HA/proteoglycan complexes are anchored to the surface of chondrocytes via interaction with HA receptors (Knudson, 1993), which appear related to the lymphocyte homing receptor CD44 (Knudson, 1993; Knudson and Knudson, 1991; Knudson et al., 1993). One function of HA receptors on chondrocytes appears to be directing the assembly of a pericellular matrix around the cells (Knudson, 1993; Knudson and Knudson, 1991; Knudson et al., 1993). However, the interaction of HA with these HA receptors may also play a role in the endocytotic turnover of cell surface-bound HA. Receptor-mediated endocytosis of HA via CD44-like cell surface HA receptors has recently been demonstrated on hamster alveolar macrophages and simian virus-transformed 3T3 cells (Culty et al., 1992).

The CD44-like cell surface HA receptors present on macrophages, tumor cells and chondrocytes have similar physical characteristics and can be grouped into a family of non-integrin, hydrophobic membrane glycoproteins (Toole, 1990). A variety of molecular masses have been reported; 85 kDa (Underhill et al., 1985; Underhill, 1989; Sy et al., 1991; Culty et al., 1990); 69, 90 and 93 kDa (Banerjee and Toole, 1991) and 150 kDa (Sy et al., 1991), depending on the cell type studied. The physical and functional properties that are common to this group of receptors include: (1) a high binding affinity for HA (Kd > 10⁻⁹ M); (2) a high degree of specificity for HA (when receptors are assayed on nonextracted, intact membranes); (3) binding affinity for HA increases with increase in ionic strength; (4) binding is stable to mild fixation of the receptor with glutaraldehyde; and, most importantly, (5) binding can be competed for by HA oligosaccharides with a minimum size of six monosaccharides. These and other properties help to distinguish this family of HA binding proteins/receptors from other HA binding proteins including aggregating proteoglycans (aggrecan), link proteins, glial-HA-binding protein, hyaluronectin and the HA clearance receptor present on liver endothelial cells (Goldberg and Toole, 1984; Toole, 1991).

In this study we have used cultures of rat chondrosarcoma chondrocytes as a highly reproducible model with which to study the role of cell surface HA receptors in pericellular matrix catabolism. Similar experiments were also performed on freshly isolated bovine articular chondrocytes. Here, we present evidence that fluorescein-labeled HA as well as ³H-labeled HA was endocytosed by chondrocytes and, this endocytosis was mediated via cell surface HA receptors. The endocytosis was blocked by HA hexasaccharides or unlabeled high molecular mass HA. Endocytosis was also blocked by addition of anti-CD44 monoclonal antibodies. Further results show that the internalized HA was destined for degradation.

All reagents, unless otherwise specified, were reagent grade and purchased from Sigma Chemical Co. (St. Louis, MO). Tissue culture supplies were purchased from GIBCO-BRL (Grand Island, NY).

Rat chondrosarcoma chondrocytes selected for continuous longterm cell culture were obtained from Dr James H. Kimura (Henry Ford Hospital, Detroit, MI), and grown in culture flasks (Falcon) as monolayer cultures. The morphology of these cells and that of their pericellular matrix have been recently described (Knudson, 1993). Cells were maintained in Dulbecco’s modified Eagle’s medium containing 4.5 g/l glucose supplemented with 10% fetal bovine serum (Hyclone, Logan, UT), 1% penicillin/streptomycin solution, 50 μg/ml ascorbic acid and 2 mM glutamine, at 37°C in a humidified incubator supplied with 5% CO₂/95% air. Cells from near-confluent monolayers were dissociated with 0.25% trypsin/0.05 M EDTA solution (GIBCO-BRL) and 1×10⁴ cells/ml were plated into 2-well tissue culture chamber slides (Nunc, Inc.) in complete medium for further studies. Bovine articular chondrocytes were obtained from Dr Margaret B. Aydelotte (Department of Biochemistry, Rush-Presbyterian-St. Luke’s Medical Center, Chicago, IL). Chondrocytes were isolated from full-thickness articular surfaces of young bovine (18-24 month steers) metacarpophalangeal joints in Dr Aydelotte’s laboratory as described previously (Aydelotte and Kuettner, 1988). After 5 days in primary culture the chondrocytes were plated into 2-well tissue culture chamber slides as the rat chondrosarcoma chondrocytes.

Fluorescein-hyaluronan (fl-HA) was prepared following a modification of the method described by de Belder and Wik (1975). Briefly, 50 mg of hyaluronan (HA) (Sigma, grade I) was dissolved in 40 ml of water, and mixed with 20 ml of DMSO. Fluoresceinamine, isotype I (25 mg, Aldrich) in DMSO (0.5 ml) containing acetaldehyde (25 μl, Fluka) and cyclohexyl isocyanide (25 μl, Fluka) was added into the mixture. The pH value of the solution was adjusted to 4.5. The mixture was incubated at 22°C for 5 h, and then poured into 800 ml of 95% ethanol. The precipitated fl-HA was collected by centrifugation at 1,000 g and air-dried in a chemical hood. The fl-HA was then dissolved in water and re-precipitated by 95% ethanol 2 times. The fl-HA was further purified by exhaustive dialysis against distilled water followed by lyophilization. The fl-HA was then characterized by Sepharose CL-2B chromatography and a proteoglycan-dependent ELISA.

³H-labeled hyaluronan ([³H]HA) was purified from the conditioned medium of rat fibrosarcoma cells labeled with [³H]acetate (New England Nuclear) as described previously (Knudson and Toole, 1987; Nemec et al., 1987). Following purification, the [³H]HA has been shown to be free of contaminating glycosaminoglycans (as determined by DEAE chromatography (Knudson and Toole, 1987; Nemec et al., 1987) and has a specific activity of 6.8×10⁴ cpm/μg.

The fl-HA was characterized for its capacity to bind aggregating proteoglycan using a modification of an enzyme-linked immunosorbent assay (ELISA) for HA developed in our laboratory (Li et al., 1989). Briefly, ELISA plates were coated with chon-droitinase-digested rat chondrosarcoma D1D1 proteoglycan as described previously (Li et al., 1989). After washing, various concentrations of fl-HA in PBS or FITC-dextran (M_r 2×10⁶, Sigma) were added to the plate and incubated for 1 h at room temperature. Bound fl-HA or FITC-dextran was then detected using a biotinylated anti-fluorescein antibody (1:1500 in 1% BSA/PBS, Molecular Probes, Inc.), followed by streptavidin-peroxidase (1:500 in 1% BSA/PBS, Vector Labs) and o-phenylenediamine peroxidase substrate. In order to determine the specificity of fl-HA interaction in the ELISA, plates were incubated with fl-HA in the presence or absence of excess unlabeled HA.

Characterization of fl-HA and [³H]HA was performed by sizeexclusion chromatography on a 0.8 cm × 50 cm Sepharose CL-2B column eluted with PBS buffer. Fractions were collected at a rate of 5 ml/h and monitored by absorbance at 490 nm (for fl-HA) or by liquid scintillation counting (for [³H]HA).

[³H]HA degradation products were characterized by Sephadex G50 chromatography. Samples of [³H]HA isolated from various cellular pools were applied to a 0.8 cm × 50 cm Sephadex G50 column and eluted with PBS. Fractions of 0.5 ml were collected at 0.4 ml/min and assayed by liquid scintillation counting.

Rat chondrosarcoma or bovine articular chondrocytes, in tissue culture chamber slides, were digested with 12.0 units/ml of testicular hyaluronidase (type I-S, Sigma) in DME containing 10% FBS for 1 hour at 37°C to remove their pericellular matrix and expose all the hyaluronan receptors. The cells remained attached to the substratum following this treatment and, following 3 washes with PBS, no residual enzymatic activity remained (Knudson, 1993). The cells were then cultured in the presence or in the absence of 100 μg/ml fl-HA for various times at 37°C. In one set of control experiments chondrocytes were incubated with 100 μg/ml of FITC-dextran (Sigma, M_r= 2×10⁶) for 2 h at 37°C. Following incubation with labeled ligand, the medium was removed, the cells were washed 3 times with PBS and then fixed with 4% paraformaldehyde in PBS, pH 7.4, for 30 min. In some experiments, cells were digested with 0.25% trypsin (GIBCO-BRL, 1:250), for 30 min at 37°C or, with 3 units/ml of Streptomyces hyaluronidase (Sigma Type IX) for 20 min at 37°C, before fixation. Cells were mounted in glycerol/PBS and photographed with an Olympus Microscope equipped with epifluorescence optics using a fluorescein filter set.

In order to determine the role of hyaluronan receptors in the endocytosis of fl-HA, competition experiments with unlabeled HA or HA hexasaccharides were performed. Cells were pre-incubated with or without unlabeled HA (1 mg/ml, Sigma grade I) or unlabeled HA hexasaccharides (1 mg/ml), for 30 min and then incubated with 100 μg/ml fl-HA, with the corresponding presence or absence of unlabeled HA or HA hexasaccharides. HA hexasaccharides were prepared from testicular hyaluronidase digests of HA as described previously (Knudson, 1993; Knudson and Knudson, 1991).

Rat chondrosarcoma chondrocytes plated into 4-well tissue culture chamber slides at 1×10⁴ cells/ml, were digested with testicular hyaluronidase to remove their pericellular matrix. Cells were cultured in the presence or absence of 8.0 μg/ml [³H]HA for 2 h at 37°C. After the medium was removed, some cells were again digested with 0.25% trypsin for 20 min at 37°C. Cells were then fixed with 4% paraformaldehyde/0.2% glutaraldehyde in 0.1 M cacodylate buffer containing 1 mM CaCl, pH 7.3, for 1 h at 4°C. The slides were then dipped in B-max Hyperfilm emulsion (Amer-sham) and stored at 4°C for two weeks before developing in D-19 developer (Kodak). Three PBS washes were used between each of the steps described above. In some experiments cells were incubated with [³H]HA in the presence of unlabeled HA (1 mg/ml) or HA hexasaccharides (1 mg/ml), following a 30 min pre-incuba-tion with unlabeled HA or HA hexasaccharides. Cells were photographed in bright-field as well as dark-field on a Nikon Microphot-FXA microscope.

Rat chondrosarcoma and bovine articular chondrocytes were plated in 35 mm tissue culture dishes at 6×10⁶ cells/ml. After allowing the cells to plate for 24 h, the cells were digested with 12 units/ml testicular hyaluronidase for 60 min at 37°C to remove their endogenous pericellular matrix. Following 3 washes with HBSS, 26.5 μg/ml [³H]HA was added to cells in complete culture medium. After 24 h of incubation the conditioned medium was collected and cells were washed 3 times with HBSS. The cell surface pool was collected by incubation of the cells with 0.25% trypsin at 37°C for 30 min followed by a 5 min centrifugation at 400 g. This supernatant was defined as the cell surface pool. The cell pellet was again washed 3 times with HBSS and then resuspended in 0.1 M Tris-HCl containing 5.6 units/ml of protease (type XIV, Sigma). The cell pellet was incubated at 37°C for 12 h as described by Culty et al. (1992) to solubilize the cells and release the total intracellular material. In some experiments the intracellular pool was isolated by direct extractions of the cell pellet with 4.0 M guanidine-HCl in the presence of 0.5% CHAPS at 4°C for 12 h. Similar profiles were obtained using either method. The [³H]HA within the medium, cell surface and intracellular pools was further characterized, without additional concentration, on Sepharose CL-2B or Sephadex G50 columns eluted in PBS.

In another set of experiments, bovine chondrocytes, following testicular hyaluronidase treatment, were incubated with 200 μM chloroquine in complete medium for 1 h prior to addition of 26.5 μg/ml [³H]HA in the presence of 200 μM chloroquine. Cells were incubated with HA and chloroquine for only 8 h instead of 24 h, to avoid prolonged exposure to chloroquine. The medium was removed, cells washed and the intracellular pool isolated as described above.

In order to determine the role of CD44 receptors as the class of HA receptors involved in the binding and endocytosis of [³H]HA by chondrocytes, bovine articular chondrocytes were treated with testicular hyaluronidase and then pre-incubated for 30 min with or without KM201 or IM7.8.1 anti-CD44 monoclonal antibodies (anti-mouse Pgp-1; American Type Culture Collection) before addition of 26.5 μg/ml [³H]HA. The [³H]HA was added in the presence or absence of the same concentration of KM201 or IM7.8.1 antibodies. The cells were then incubated for 8 h at 37°C in order to insure sufficient intracellular accumulation of [³H]HA in control cultures, but to avoid prolonged exposure of cells to antibody. Following this incubation, the cell surface and intracellular pools were collected enzymatically as described above and assayed directly by scintillation counting. In another set of experiments, following testicular hyaluronidase treatment, the bovine articular chondrocytes were fixed for 5 min at room temperature with a 0.5% solution of glutaraldehyde in PBS as described previously (Knudson and Knudson, 1991). Following three washes in PBS containing 1% BSA, the cells were pre-incubated with or without KM201 or IM7.8.1 antibodies followed by addition of [³H]HA as before. After 8 h of incubation, the cultures were washed three times with PBS and bound [³H]HA was recovered by extraction with 1% SDS, mixed with scintillation cocktail and counted.

Fluorescein-amine was conjugated to high molecular mass hyaluronan (HA) and characterized by gel filtration chromatography on Sepharose CL-2B. As shown in Fig.1A, the majority of the fluorescein-labeled material, which exhibits an absorbance at 490 nm, eluted in the void volume of the column. No free fluorescein or lower molecular mass contaminants were detected. The fluorescein-hyaluronan (fl-HA) was then applied to a modified HA-ELISA. As shown in Fig.1B, the fl-HA bound to the proteoglycan-coated plates and the binding was proportional to the fl-HA concentration. Interestingly, the linear range of concentrations detected by this assay were identical to non-conjugated HA assayed by the standard HA-ELISA procedures (Li et al., 1989). The fl-HA was therefore considered as fully functional HA. Addition of high molecular mass native HA competed with the binding of fl-HA to the ELISA plates (data not shown) and no absorbance was observed in assays minus fl-HA (Fig.1B). Fluorescein-labeled dextran (FITC-dextran), of equivalent molecular mass, also showed no binding to the proteoglycan-coated ELISA plates (Fig.1B).

The [³H]HA used in this study has been extensively characterized previously (Knudson and Knudson, 1991; Knudson, 1993), is also of high molecular mass (see Fig.6), fully functional for binding to proteoglycans or receptors and contains no free, unincorporated label.

Cultures of rat chondrosarcoma chondrocytes were treated with testicular hyaluronidase to remove their hyaluronan-containing pericellular matrix (Knudson and Knudson, 1991; Knudson, 1993). We have shown previously that this treatment removes greater than 98% of the hyaluronan, and 93% of the proteoglycan bound at the chondrocyte cell surface (Knudson, 1993). This treatment was used to expose the total number of cell surface hyaluronan receptors, a large proportion of which are normally occupied by endogenous pericellular hyaluronan (Knudson and Toole, 1987). The cells remain attached to the substratum and residual hyaluronidase activity is readily removed by washing steps. fl-HA was then added to the cultures, which were incubated at 37°C for varying periods of time. Time points at 10 and 30 min showed little fluorescent material bound to the cell surface (data not shown). However, following 60 to 120 min of incubation, the rat chondrosarcoma cells displayed considerable staining (Fig.2A,B, respectively), representing cell surface plus intracellular accumulation of fl-HA. In order to distinguish endocytosed fluorescent material, one set of cells was digested with trypsin for 30 min at 37°C to remove pericellular, cell surface bound fl-HA. As can be seen in Fig.2C, this treatment revealed fl-HA localized into clusters or patches indicative of intracellular vesicles. A faint halo of staining was sometimes observed just beneath the cell surface. No fluorescence was observed in the absence of added fl-HA, i.e. little autofluorescence was observed on these cell cultures.

As shown in Fig.2D,E, pericellular as well as intracellular staining with fl-HA was blocked by incubation of the cells in the presence of excess high molecular mass HA or HA hexasaccharides. This suggests that the fl-HA becomes bound to the cell surface via interaction with HA receptors. In addition, this suggests that no internalization of fl-HA occurred via other non-receptor-mediated mechanisms such as fluid-phase pinocytosis or bulk phagocytosis. To extend this conclusion further, the cells were incubated with FITC-labeled dextran, a polysaccharide of similar molecular mass to the fl-HA. As shown in Fig.2F, in the presence of equivalent concentrations of FITC-dextran, no cell surface binding or intracellular accumulation was observed.

In order to demonstrate that the endocytosis of fl-HA was not restricted to transformed chondrocytes, first passage cultures of normal bovine articular chondrocytes were incubated with fl-HA. As shown in Fig.3A, bovine chondrocytes also bound significant quantities of fl-HA to their cell surface after 2 h of incubation. A portion of the bound fl-HA was also internalized as shown after trypsin treatment, removing pericellular fl-HA and revealing the intracellular vesicular clusters of fl-HA shown in Fig.3B. As discussed above, another method to remove pericellular HA is via treatment with hyaluronidase. Fig.3C depicts bovine chondrocytes post-treated with Streptomyces hyaluronidase to distinguish intracellular fl-HA, which is apparently present within small vesicles. As with the rat chondrosarcoma cells, binding and endocytosis was inhibited by high molecular mass HA as well as HA hexasaccharides (data not shown).

In order to determine that the endocytosis of fl-HA was not due to its conjugation to fluorescein, the binding and endo-cytosis of another form of HA, namely [³H]HA, was characterized. Binding and uptake of [³H]HA was followed by autoradiography using radiographic development techniques designed for in situ hybridization studies. As shown in bright-field (Fig.4A) as well as dark-field (Fig.4B) microscopy, after 2 h of incubation of [³H]HA with rat chondrosarcoma chondrocytes, intense cell surface and intracellular development of grains was observed. When these cells were trypsinized to remove pericellular-localized material, intracellularly localized [³H]HA was displayed as intense grain development within small vesicles (Fig.4C,D) similar to the patterns observed with the fl-HA. No pericellular or intracellular grain development occurred when the cells were incubated with [³H]HA in the presence of high molecular mass HA (Fig.4E,F) or HA hexasaccharides (Fig.4G,H). Thus, blocking the interaction of [³H]HA to HA receptors blocks cell surface binding as well as endo-cytosis.

Confluent cultures of rat chondrosarcoma as well as bovine articular chondrocytes were incubated for various times with [³H]HA. In initial experiments, 2 h of incubation at 37°C resulted in insufficient radioactivity accumulated within the intracellular compartment for accurate quantification by subsequent column chromatography (data not shown). The incubation time was therefore extended to 24 h. Following this incubation, the cells were washed three times with HBSS and then treated with trypsin to remove pericellular-associated [³H]HA. The cell pellet following trypsinization was washed extensively with HBSS and then subjected to exhaustive protease digestion to solubilize the cells totally and liberate intracellular [³H]HA. Fig.5 illustrates the Sephadex G50 chromatographic profiles of various [³H]HA fractions isolated from bovine articular (Fig.5A,C,E) as well as from rat chondrosarcoma chondrocytes (Fig.5B,D,F). Both the medium [³H]HA (Fig.5A,B) and [³H]HA recovered from the trypsinate fractions (Fig.5C,D; trypsinate fraction represents the cell surface-associated pool) were eluted totally within the void volume of the G50 column. Even when the cell-surface trypsinate fractions are displayed at an expanded scale, no small molecular mass included fractions are observed. However, the ³H-labeled material localized within the intracellular compartment (Fig.5E,F) contained a high proportion of small molecular mass degradation products (44% and 43% of the total counts, Fig.5E and F, respectively). In separate control studies, no small [³H]HA degradation products were generated independently by the protease treatment itself (Fig.5F, inset). The G50 column was used to demonstrate that hyaluronan degradation products such as mono- and disaccharides, and smaller, were being generated. However, when the rat chondrosarcoma or bovine articular chondrocyte intracellular fractions were applied to a Sepharose CL-2B column (Fig. 6A,B, respectively), while 7% to 29% of the [³H]HA remained in the void volume, the majority of the radioactivity eluted with a partition coefficient (K_av)>0.68. This indicates that a larger proportion of the endocytosed HA is in the process of being degraded within the cell - more than can be appreciated by Sephadex G50 chromatography. Thus, the [³H]HA endocytosed by two different types of chondrocytes was destined for degradation.

The total counts within each of these cellular fractions depicted in Fig.5 were quantified. Of the total [³H]HA in the system, 1.4% was associated with the cell surface of the rat chondrosarcoma chondrocytes; 0.2% with the cell surface of bovine articular chondrocytes (30,745 and 5,060 cpm, respectively); that is, when given the same amount of labeled ligand to the same number of cells, approximately 7 times more [³H]HA bound to the cell surface of the chondrosarcoma cells. However, the intracellular fraction represented 1.5% of the cell surface-bound [³H]HA for the rat chondrosarcoma chondrocytes yet, 14.0% for bovine articular chondrocytes (i.e. 480 and 820 cpm, respectively). Although less [³H]HA was bound to the bovine articular chondrocytes, 2 times more [³H]HA accumulated within the intracellular compartment. Since these values only characterize the 24 h time point they do not represent a rate, but rather the relative distribution of counts after 24 h of incubation. However, given the distribution of counts within the intracellular pools (i.e.% of [³H]label present as small degradation products; Fig.5E,F), the rates of intracellular degradation by both of the cell types are similar. Therefore, the increased accumulation of intracellular counts in the normal bovine articular chondrocytes may, in fact, reflect a faster rate of endocytosis as compared to that of the rat chondrosarcoma chondrocytes.

In order to verify that the production of extensively degraded [³H]HA fragments within the intracellular pool was due to acid hydrolase activity, chondrocytes were next incubated with exogenous [³H]HA in the presence of the lysosomotropic agent, chloroquine. Bovine articular chondrocytes were treated with chloroquine 1 h before and during an 8 h incubation with [³H]HA. The intracellular pool was collected and applied to Sephadex G50. As shown in Fig.7, compared to control, chloroquine treatment reduced by 85% the amount of extensively degraded HA eluting in the total volume. Therefore, the intracellular degradation of HA following endocytosis occurs via lysosomal enzymes.

Bovine articular chondrocytes were incubated with [³H]HA in the presence or absence of anti-mouse CD44 antibodies. One of the antibodies (KM201) has recently been shown to inhibit the binding of hyaluronan to human cells (Thomas et al., 1992) and inhibit the formation of hyaluronan-dependent pericellular matrices on cells transfected with human CD44 cDNA-containing plasmids (Knudson et al., 1993). As shown in Table 1, 46% less [³H]HA bound to the cell surface of bovine chondrocytes in the presence of the 10 μg/ml KM201 anti-CD44 antibody as compared to control cultures. This percentage was mirrored by the intracellular compartment in which 44% less [³H]HA accumulated within chondrocytes incubated in the presence of the anti-CD44 antibody. Similarly, in another set of experiments (Table 1, experiment B) incubation of bovine chondrocytes in the presence of 10 μg/ml of another anti-CD44 antibody, IM.7.8.1, resulted in approximately 40% less [³H]HA bound to the cell surface as compared to control cultures. As with the KM201 antibody, the reduction of cell surface-associated [³H]HA was mirrored by a similar reduction in [³H]HA accumulating intracellularly. Since a saturable dose of the IM7.8.1 antibody had never been determined from previous experiments, inhibition with 5.0 and 7.5 μg/ml antibody was attempted. As shown in Table 1, only 15% inhibition of cell surface-associated [³H]HA was obtained at 5.0 μg/ml whereas nearly 40% inhibition was obtained with 7.5 μg/ml antibody. Thus it is unlikely that more inhibition will occur with higher concentrations of antibody.

Therefore, a portion of the binding and endocytosis capacity of bovine chondrocytes is due to CD44-like cell surface receptors. The remaining non-inhibited portion may reflect the inability of the anti-mouse/human CD44 antibodies to efficiently recognize the bovine antigen, or the presence of another class of HA receptors unrelated to CD44.

In order to determine whether anti-CD44 antibody inhibition of cell surface [³H]HA binding was due to receptor clustering and removal from the cell surface during the 8 h time course, inhibition-of-binding experiments were performed on glutaraldehyde-fixed bovine articular chondrocytes. As shown in Table 1, experiment C, incubation of fixed cells with either antibody still resulted in significant inhibition of binding of [³H]HA to the cell surface. This does not rule out the possibility that antibody-induced receptor clustering may occur in living cells, but demonstrates that this type of mechanism is not necessary to explain our results - these anti-CD44 antibodies can inhibit cell surface accumulation of [³H]HA by directly inhibiting the binding to fixed, non-mobile receptors.

Several cell types such as human synovial cells (Truppe et al., 1977), SV40-transformed 3T3 fibroblasts (Culty et al., 1992), embryonic myocardial cells (Bernanke and Orkin, 1984), and macrophages (Culty et al., 1992; Gustafson and Forsberg, 1991) have been shown to bind, internalize and degrade extracellular HA. It has been suggested that chondrocytes must also possess a mechanism for the endocytosis and degradation of HA, particularly HA present within the pericellular matrix (Morales and Hascall, 1988; Ng et al., 1992). Compared to the turnover of the other major extracellular matrix component of cartilage, type II collagen, with a turnover half-life measured in years (Maroudas and Urban, 1980), HA and proteoglycan turnover in cartilage could be considered extremely rapid. For example, Morales and Hascall (1988) found that newly synthesized ³H-labeled HA was lost from cartilage explants with a halflife roughly equivalent to the half-life of newly synthesized proteoglycan ⁠. In more recent studies Ng et al. (1992) reported lower half-life values, yet the turnover of HA and proteoglycan was again nearly equivalent (⁠⁠, proteoglycan = 12 days). However, while greater than 90% of the proteoglycan lost from these tissues was due to release into the culture medium, this mechanism accounted for less than 9% of newly synthesized [³H]HA (Morales and Hascall, 1988; Ng et al., 1992). Also, small degradation products of [³H]HA were not recovered within the extracellular matrix of the tissue itself (Ng et al., 1992). Taken together with the fact that enzymes capable of degrading HA extracellularly have never been found in cartilage or synovial fluid (Kresse and Glossl, 1987), these data suggest that endocytosis of HA by chondrocytes must be occurring in order to effect the observed rates of HA turnover and the disappearance of HA from the tissue (Morales and Hascall, 1988; Ng et al., 1992). In the present study we show direct evidence that HA is, in fact, endocytosed by chondrocytes. Both bovine articular and rat chondrosarcoma chondrocytes bound fl-HA as well as ³H-labeled HA onto their cell surface and internalized these HA molecules into intracellular compartments in as little as 60 min. The ³H-labeled HA internalized by chondrocytes was degraded by both of the cell types and recovered as small fragments, which would be the predicted products of lysosomal hydrolases (Culty et al., 1992; Orkin et al., 1982; Gustafson and Forsberg, 1991). The use of exogenously labeled fl-HA or [³H]HA in this study also avoided problems associated with following the catabolic fate of macromolecules labeled metabolically in situ, where careful distinctions must be made between the small percentage of [³H]HA present and the significantly larger percentage of ³H-labeled chondroitin sulfate and [³H]chondroitin sulfate degradation products.

The binding and internalization of exogenously added HA by chondrocytes also appeared to be receptor mediated. No cell surface binding or internalization occurred in the presence of excess unlabeled high molecular mass HA or HA hexasaccharides. The latter act as competitive inhibitors of HA/HA receptor interactions but do not interfere with the binding of HA to other HA-binding proteins such as proteoglycans, link proteins, hyaluronectin or liver endothelial cell-like HA clearance receptors (Knudson, 1993; Knudson and Knudson, 1991; Bernanke and Orkin, 1984). The binding and endocytosis of [³H]HA was also inhibited by approximately 40-50% by anti-CD44 antibodies. Thus, the endocytosis of HA by chondrocytes requires that the HA must first be bound to cell surface HA receptors. Similar CD44/hyaluronan receptor-mediated endocytosis was recently demonstrated by Culty et al. (1992) on mouse SV40-transformed 3T3 cells and hamster alveolar macrophages. As in these studies the antibodies effected only a partial inhibition of [³H]HA binding and degradation of [³H]HA on SV40-transformed 3T3 cells. In the current study, since antibodies specifically directed against bovine or rat CD44 antigens were not available, we cannot definitively conclude that only CD44-like HA receptors are present on chondrocytes. The anti-mouse CD44 (KM201 and IM7.8.1) antibodies may only partially inhibit binding of [³H]HA to bovine CD44/HA receptors or, another entirely different class of HA receptors, unrelated to CD44, may also be present on these chondrocytes.

The inhibition of binding and endocytosis of HA in the presence of excess HA hexasaccharides also suggests that HA internalization did not occur via fluid-phase pinocytosis or phagocytosis. These processes would not be affected by the presence of small unlabeled oligosaccharides. In fact, all methods that retarded binding of HA to cell surface HA receptors, decreased or eliminated the internalization of HA. This suggestion was further corroborated by addition of FITC-labeled dextran to chondrocytes. FITC-dextran, of similar size and specific activity of labeling, showed no binding to the cell surface or accumulation within intracellular compartments.

Another role of CD44/HA receptors on cells, chondrocytes in particular, is to participate in the assembly of pericellular matrices around the cells (Knudson, 1993; Knudson et al., 1993; Knudson and Knudson, 1991). We have shown previously that chondrocytes exhibit prominent pericellular matrices in culture - matrices that are dependent upon the interaction of cell surface HA receptors with extracellular HA and aggregating proteoglycans (Knudson, 1993). These pericellular matrices can be completely displaced via incubation of the cells with HA hexasaccharides (Knudson, 1993; Knudson et al., 1993; Knudson and Knudson, 1991) or inhibited from re-assembling by addition of HA hexasaccharides, or in some cases anti-CD44 antibodies (Knudson et al., 1993) or antibodies to other cell surface HA binding proteins (Yu et al., 1992). It would thus appear that, on chondrocytes at least, the same receptors are responsible for matrix assembly as well as endocytosis. Matrix assembly and endocytosis may indeed be tightly linked. For example, we have found that bovine articular chondrocytes in monolayer culture have relatively small pericellular matrices; with morphometric matrix area-to-cell area ratios of approximately 1.7 ± 0.1. Rat chondrosarcoma chondrocytes, on the other hand, have larger matrices with matrix area-to-cell area ratios of 3.3 ± 0.3 (Knudson, 1993). Although there may clearly be differences in the rates of synthesis of extracellular matrix components by these cells, it is interesting to consider that these matrix sizes are also inversely proportional to the apparent rate of HA endocytosis by these two cell types. In other words, the reduced rate of HA endocytosis by the rat chondrosarcoma cells may serve to promote the increased size of the pericellular matrix observed on these cells.

Maintenance of cartilage matrix integrity, composition and organization is critical to the normal functioning of cartilage as a primary load-bearing tissue. Alterations in the controlling mechanisms of matrix metabolism may lead to loss of function and arthritic disease. HA plays a key role in the organization of cartilage extracellular matrix; serving as a backbone structure for the fixation of chondroitin sulfate-rich proteoglycans within the tissue. Mechanisms governing HA turnover are therefore crucial to an understanding of cartilage extracellular matrix metabolism. For example, it as been suggested that nonfunctional HA, that is, HA bound up with remnants of proteoglycan and link protein binding regions, accumulates within aging cartilage (Roughley et al., 1985). This accumulation may be due to an inability of the resident chondrocytes to bind and internalize these molecules, or because of reduced expression of cell surface HA receptors on the aged chondrocytes. However, the whole process of HA degradation may not occur intracellularly. It has been shown that HA within the cartilage extracellular matrix does become partially depolymerized, most likely due to the action of free radicals (Ng et al., 1992; Roughley et al., 1992; Baker et al., 1989). The size of the HA obtained from papain digests of articular cartilage also decreased with age (Holmes et al., 1988). This initial depolymerization may represent part of the mechanism for HA catabolism. For instance, McGuire et al. (1987) demonstrated that myocardial cells bind, internalize and degrade partially depolymerized HA more efficiently than high molecular mass native HA. Thus initial depolymerization of HA, as well as proteolytic removal of attached proteoglycan and link proteins, may occur extracellularly, followed by the binding and endocytosis of somewhat shorter HA chains to chondrocytes, resulting in the complete degradation of the HA by intracellular enzymes.

This work was supported by research grants AR39507 and AR39239 from the National Institutes of Health. This paper is part of the fulfillment of the requirements for a Ph.D. degree, for Qiang Hua.