Abstract
The objective of our study was to determine the tissue distribution and localization of ADAM-10 protein in human and bovine cartilage and the changes it undergoes with cartilage degeneration seen in osteoarthritis (OA) and under the influence of interleukin-1 (IL-1). Human normal and OA articular cartilage and bovine nasal cartilage cultured in the presence of IL-1α were processed for histology and immunohistochemistry. ADAM-10 protein was extracted from human cartilage and analyzed by Western blotting using anti-ADAM-10 antibodies. Fluor S Image analyzer and Quantity One software program were applied to quantify the total amount of ADAM-10. ADAM-10 protein was detected in both human and bovine cartilage. The strongest immunostaining was found in the cytoplasm and/or cell membranes of the superficial and upper middle layer of normal adult human cartilage, in the clusters and fibrillated areas of OA cartilage, and in IL-1α-stimulated bovine nasal cartilage. The distribution of ADAM-10 protein in bovine nasal cartilage was dependent on the length of exposure to IL-1 α and corresponded to the areas of proteoglycan depletion. By Western blotting analysis of human cartilage, ADAM-10 was primarily detected in the membrane-enriched fraction and its levels were increased in degenerated and OA cartilage compared to normal cartilage. The results of this study suggest that ADAM-10 might be an important factor associated with cartilage degenerative processes. (
Keywords
R
ADAMs have been described as membrane-anchored cell surface proteins structurally related to reprolysins (Wolfsberg et al. 1995). This family consists of at least 29 known members, with many containing a hydrophobic transmembrane sequence at their C-termini and some possessing a zinc-binding catalytic site similar to that of matrix metalloproteinases (Fox and Bjarnason 1996; Nagase 1996). Although the first members described in this family have been implicated as the proteins involved in cell–cell interaction (Blobel et al. 1992; Wolfsberg and White 1996; Blobel 1997), other ADAMs (ADAM 1, 9, 10, 12, 13, and 15) were predicted to have a functional metalloproteinase domain on the basis of their deduced amino acid sequences. To date, four ADAMs are known to be catalytically active (Black et al. 1997; Blobel 1997; Kuno et al. 1997; Moss et al. 1997; Tortorella et al. 1998, 1999). Two members of this family are involved in the degradation of ECM molecules (Arner et al. 1997, 1999; Kuno et al. 1999; Tortorella et al. 1999). One proteinase (ADAM-17) was shown to control the processing of other precursors (Black et al. 1997; Moss et al. 1997), and a fourth member (ADAM-10) was shown to be involved in the processing of TNF-α (Rosendahl et al. 1997), proteolytic cleavage and activation of Notch receptor (Schlöndorff and Blobel 1999), and cleavage of myelin basic protein (Howard et al. 1996). Taken together these observations indicate that some ADAMs could function as membrane-anchored metalloproteinases associated with the shedding of cell surface proteins, and others, as soluble proteinases, could be related to the cleavage of matrix molecules.
Recent studies have demonstrated that porcine, bovine, and human articular chondrocytes express mRNA for at least five members of this family: ADAM-9, -10, -12, -15, and -17 (McKie et al. 1997; Chubinskaya et al. 1998; Flannery et al. 1999). Previously, our laboratory showed an upregulation of ADAM-10 gene expression in newborn and OA cartilages (tissues in which a process of remodeling takes place) compared to normal adult cartilage. Interestingly, the highest level of ADAM-10 mRNA expression in OA cartilage was found in large clones of cell clusters and in fibrillated areas of the remaining upper layer. Moreover, we were able to document that ADAM-10 gene expression in OA cartilage was higher than in the age-matched normal tissue (Chubinskaya et al. 1998).
In addition to its predicted proteolytic activity (McKie et al. 1997) ADAM-10 also exhibits a high sequence homology with tumor necrosis factor-α (TNF-α) converting enzyme (TACE or ADAM-17) and an ability to cleave TNF-α substrates (Moss et al. 1997; Rosendahl et al. 1997). The amino acid sequence at the C-terminus of the ADAM-10 propeptide suggested that this proteinase could be activated/cleaved by furin-like proteases in the Golgi (Rosendahl et al. 1997). Alternatively, most of the members of this family are implicated as factors capable of cleaving different cell surface proteins (receptors, growth factors, cytokines, matrix metalloproteinases) by the same furin-like mechanism, i.e., being sheddases (Rosendahl et al. 1997). These findings suggest the possible involvement of the members of ADAM family (particularly ADAM-10) in the process of cartilage metabolism, degradation, and/ or remodeling.
The purposes of the present study were (a) to detect cellular localization, tissue distribution and the levels of ADAM-10 protein in human and bovine cartilage and (b) to investigate the changes in ADAM-10 protein in response to degeneration of human articular cartilage and to catabolic stimuli of bovine nasal cartilage.
Materials and Methods
Reagents
Human recombinant ADAM-10 and anti-ADAM-10 monoclonal antibodies were provided by Procter & Gamble Pharmaceuticals (Mason, OH). Electrophoresis grade reagents were purchased from Bio-Rad (Hercules, CA). Chemicals, either reagent or molecular biology grade, were purchased from Sigma Chemical (St Louis, MO) unless otherwise noted. Keratanase (Pseudomonas sp.; EC 3.2.1.103), keratanase II (Bacillus sp. Ks 36), and chondroitinase ABC (Proteus vulgaris; EC 4.2.2.2) were obtained from Seikagaku, (Tokyo, Japan).
Tissue Acquisition
Donor Cartilage. Full-thickness articular cartilage was dissected from load-bearing regions of the femoral condyles and talus of donors with no history of joint disease within 24 hr of death. Samples from males and females ranging from newborn to 94 years old were obtained with institutional approval through the Regional Organ Bank of Illinois according to their protocol. After opening of the joint, the surface of the cartilage was grossly examined. The appearance of each joint was given a grade based on the Collins scale (1949) as modified by Muehleman et al. (1997).
Osteoarthritic Cartilage. Human OA cartilage was obtained through collaboration with Dr. Gabriella Cs-Szabo and with the consent of the Department of Orthopedic Surgery (Rush–Presbyterian–St Luke's Medical Center, Chicago, IL) from patients (aged 50–86 years, both men and women) who underwent knee arthroplasty due to advanced OA. Within 3 hr after surgery, full-thickness non-calcified cartilage was removed from the femoral condyle.
Cartilage from all donors or patients was processed either for histology and immunohistochemistry (frozen or paraffin-embedded sections) or protein extraction for Western blotting analysis.
Bovine Cartilage. Nasal septum cartilage was harvested from 18-month-old steers and 2 mm diameter × 1 mm thick plugs were prepared for culture under aseptic conditions. Plugs were cultured in DMEM medium (Gibco; Grand Island, NY) supplemented with 1% insulin transferrin selenium A (ITS; Sigma), 10 mM HEPES (Gibco), 1% nonessential amino acids (Gibco), 0.01% fungizone (Gibco), 1% penicillin, streptomycin, glutamine supplement (Gibco), 50 μg/ml ascorbate (Sigma) with and without 50 ηg/ml human recombinant IL-1α (Genzyme; Cambridge, MA) for 0, 2, 7 or 14 days. IL-1α was chosen for this study because this isoform of IL-1 has been shown to be more effective in shutting down aggrecan synthesis in cultured bovine cartilage, whereas IL-1β is more effective in human cartilage (Häuselmann et al. 1996).
Histological Processing
Human cartilage was fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin or directly embedded in Tissue Freezing Medium for frozen tissue specimens (Triangle Biomedical Sciences; Durham, NC). Bovine cartilage was fixed in 10% neutral buffered formalin and embedded in paraffin. Sections (6 μm) were processed for histology and immunostaining. Histology sections of human cartilage were stained with Safranin O (Rosenberg 1971) to evaluate the grade of the cartilage (Mankin et al. 1971). Based on the two scores [Collins and Mankin (Collins 1949; Mankin et al. 1971; Muehleman et al. 1997)], we defined normal as cartilage with Collins scores of 0–1 (no visible signs of gross anatomical changes, or only slight, negligible fibrillation in articular cartilage for grade 1) and Mankin scores of 5 or less. Degenerative cartilage (donor cartilage with degenerative morphological changes) was defined as cartilage with a Collins score of 2–4 and a Mankin score greater than 5 (Chubinskaya et al. 1999a, b). OA cartilage was obtained from patients diagnosed with OA (Table 1). Histology sections of bovine cartilage were stained with toluidine blue to determine proteoglycan content.
Antibody Specificity
Anti-ADAM-10 monoclonal antibodies were raised against the synthetic peptides within either the disintegrin (IgG) or metalloproteinase domain (IgM) of recombinant ADAM-10 (rtADAM-10). The sequence of the synthetic peptide for the anti-metalloproteinase antibody was derived from the metalloproteinase domain of ADAM-10 and was C-H-E-V-G-H-N-F-G-S-P-H-D-S-G-T. The peptide sequence for the anti-disintegrin antibody was derived from the disintegrin domain of ADAM-10 and was C-R-D-D-S-D-C-A-K-F-G-I. The specificity of these peptides was tested through the BLAST search. The antibodies were purified from tissue culture media from hybridomas first by dialysis (PBS; 25,000 MWCO) and then EZEP media (Pharmacia; Piscataway, NJ). The specificity of these antibodies was tested against the rtADAM-10 and various matrix metalloproteinases (MMP-1, 2, 3, 7, 8, 9, and 13). RtADAM-10 was prepared as described previously (Tindal et al. 1999). For comparison of antibody selectivity, full-length and truncated recombinant MMPs (1, 2, 3, 7, 8, 9, and 13) and rtADAM-10 were run on a 4–20% Trisglycine polyacrylamide gel (Novex; Carlsbad, CA) and electroblotted onto PVDF membrane (Figure 1). Equal protein (10 μg) was loaded onto each lane. The membrane was incubated with a mouse anti-ADAM-10 IgM antibody raised to the metalloproteinase domain, 10 μg/ml for 2 hr at room temperature (RT) in Blotto (Tris pH 7.5, 5% w/v non-fat dry milk, 150 mM NaCl, 1 mM EDTA, 1 mM DTT). The membrane was washed in 1 × PBS, 0.1% Tween, and then incubated with a secondary antibody, goat anti-mouse IgM-horseradish peroxidase diluted 1:3000 in Blotto. Binding was visualized using enhanced chemiluminescence (Amersham Life Science; Poole, UK).
List of human cartilage samples used for the present study a
NB, newborn; M, male; F, female; N/A, for OA samples in which Collins score was not determined.
Immunohistochemistry
To improve the access of antibodies and increase the intensity of immunostaining, human articular cartilage sections were digested with a keratanase (0.1 U/ml), keratanase II (0.001 U/ml), and chondroitinase ABC (0.1 U/ml) cocktail in 0.1 M Tris, 0.05 M sodium acetate, pH 6.5, for 90 min at 37C. Sections of bovine nasal cartilage were digested with 0.1 U/ml chondroitinase ABC in 0.1 M Tris acetate, pH 8.0, for 1 hr at 37C. Alkaline phosphatase-based immunostaining was performed by using ImmunoPure ABC Alkaline Phosphatase Mouse IgG or IgM Staining Kits (Pierce; Rockford, IL) for human tissue and Histostain-SP kit (Zymed; San Francisco, CA) for bovine tissue according to the manufacturers' protocols. Nonspecific binding of antibodies was blocked with serum for 20 min (human) or 10 min (bovine) at RT. Cartilage sections were incubated with primary antibody at 1:200 dilution overnight at 4C. Biotinylated secondary antibody at 1:200 dilution was used for 2 hr (human) or 20 min (bovine) at RT. The color reaction was developed with ABC reagent and NBT/BCIP (human) or AEC (bovine) substrate. To inhibit endogenous alkaline phosphatase activity in human samples, ImmunoPure Phosphatase Suppressor (levamisole) was added to NBT/BCIP substrate at a 1:100 dilution. Sections of bovine cartilage were counterstained with hematoxylin.
Western blotting with anti-ADAM-10 antibody under reduced conditions. Lane 1, recombinant MMP-1; Lane 2, recombinant MMP-2; Lane 3, recombinant MMP-3; Lane 4, recombinant MMP-7; Lane 5, recombinant MMP-8; Lane 6, recombinant MMP-9; Lane 7, recombinant MMP-13; Lane 8, recombinant ADAM-10.
Negative controls contained no primary antibody (Figure 3A) or, for human sections, a blocking control when the synthetic peptide was preabsorbed with primary antibody before incubation with the tissue (Figure 3B). The synthetic peptide and the antibody were mixed at a 1:1 ratio, incubated overnight at 4C, centrifuged, and the supernatant was applied as a primary antibody.
Protein Extraction
Donor and OA cartilage was extracted in 1 M GuHCl (pH 7.4), 0.05 M NaAc, 0.005 M EDTA, 0.1 M ∊-amino caproin acid, 0.0005 M PMSF, 0.005 M benzamidine HCl, and 0.01 M N-ethylmaleamide for 3 hr on a shaker at 4C, dialyzed against water for 48 hr with several changes, and lyophilized.
Isolation of the Membrane-enriched Fraction
Membranes were isolated from donor and OA cartilage as described (Böhm et al. 1994). Briefly, human articular cartilage was homogenized and extracted with 8.5% sucrose, 10 mM Tris-HCl, pH 7.2, in the presence of proteinase inhibitors (0.0005 M PMSF, 0.005 M benzamide HCl, 0.01 M N-ethylmaleamide). After the removal of coarse ECM fragments by low-speed centrifugation at 1500 × g, the cell membranes were collected by high-speed centrifugation at 40,000 × g. An additional centrifugation of the resuspended pellet in a stepwise sucrose gradient at 30,000 × g resulted in the enriched membrane fraction at the 17%/40% and 8.5%/17% sucrose interfaces. The supernatant fraction consisted of sucrose solutions and contained all soluble matrix and cell components. The pellet fraction was formed during the separation by a sucrose gradient and contained all insoluble cell and matrix components. Supernatant and pellet fractions were dialyzed against 0.5 mM Tris buffer, pH 7.2, 10% SDS overnight at 4C in water.
Aliquots of samples were analyzed with or without digestion with keratanase (0.01 U/10 μg proteoglycan), keratanase II (0.001 U/10 μg proteoglycan), and chondroitinase ABC (0.1 U/10 μg proteoglycan) overnight at 37C, dialyzed against water, and lyophilized before electrophoresis.
Western Blotting Analyses
Immunoblotting analyses were performed with both anti-ADAM 10 antibodies. The lyophilized samples were solubilized in a buffer containing 10 mM Tris, pH 6.5, 1% SDS, 10% glycerol, 0.016% bromphenol blue. The samples were reduced with 10 mM dithiothreitol. Western blots were performed after SDS-PAGE. Additional binding sites were blocked with blocking solution containing 3% non-fat powdered milk (Carnation). The blots were incubated with primary antibody at the suggested dilution (1:200) and a goat anti-mouse IgG or IgM secondary antibody conjugated to horseradish peroxidase at 1:2000 dilution for IgG and 1:5000 dilution for IgM. The blots were developed with the ECL kit for Western blotting (Amersham). Specificity of the binding was compared with the binding of the antibodies to recombinant ADAM-10.
For quantitative purposes, the same amount of protein (30 μg) was loaded for each cartilage sample. Protein concentration was quantified by the Micro BSA Protein Assay Reagent Kit (Pierce).
Results
Antibody Specificity
On the Western blot, the antibody recognized only the recombinant ADAM-10 molecule; no crossreactivity with other MMPs was found (Figure 1). The band at ∼50 MW is consistent with the size expected for this rtADAM-10 (minus transmembrane and cytoplasmic domains) raised in E. coli (Tindal et al. 1999). Sequence analysis of this immunoreactive band confirmed that the antibodies indeed detect only ADAM-10 and not other cell proteins.
Immunohistochemistry of Human Cartilage
Antibodies were designed either to the disintegrin domain or the zinc-binding site of the metalloproteinase domain of ADAM-10. Both types of antibodies revealed a similar pattern in tissue distribution. Pretreatment of cartilage sections with an enzyme cocktail (chondroitinase ABC, keratanase and keratanase II) helped to unmask the epitopes.
By immunohistochemistry, we documented the expression of ADAM-10 protein in human normal (newborn and adult) and OA cartilage. However, different patterns of staining were found. In newborn cartilage, anti-ADAM-10 staining was detected throughout the entire tissue section and was localized in chondrocytes from the proliferative and hypertrophic regions. The strongest cell-associated staining was found in the upper epiphyses (developing articular surface), in the areas surrounding cartilage canals, and in the lower hypertrophic region of the tissue. In all these areas, staining was primarily cytoplasmic/cellular and only very little matrix staining was evident in the lower hypertrophic zone of newborn cartilage (Figures 2A and 2B).
In normal adult cartilage (either from femoral condyle or from talus) the intensity of specific anti-ADAM-10 staining was gradually decreased from the superficial to the deep layer of tissue. The strongest staining was detected primarily in the chondrocytes localized in the superficial and upper/middle zone of cartilage. The protein was mostly associated with cells/cell membrane and with the pericellular matrix or lacunae wall (Figures 2C and 2D). The interterritorial matrix of the superficial zone of normal cartilage was also slightly stained. The chondrocytes of the deep layer revealed some but very little immunoreactivity with anti-ADAM-10 antibodies, and the ECM of the deep layer showed no detectable staining.
Immunohistochemistry of human cartilage sections with anti-ADAM-10 antibodies. (
Negative controls. (
ADAM-10 protein expression was highly upregulated in donor degenerative cartilage and in OA cartilage. The histological appearance of the degenerative cartilage was very similar to that of cartilage obtained from OA patients representing fibrillated surface, fissures into the deep layer, and large clusters of chondrocytes. In this degenerative cartilage, cell-associated staining was evident in chondrocytes from all layers, with the most remarkable cellular (cytoplasmic) staining in the cell clusters from the remaining upper layer of the cartilage. There was no detectable immunostaining either in the cell membrane/lacunar wall or in the pericellular matrix. Conversely, in the middle and deep layers of the degenerative cartilage strong anti-ADAM-10 staining was found in the territorial matrix and to a lesser extent in the interterritorial matrix (Figures 2E and 2F).
In OA cartilage, very intense anti-ADAM-10 staining was detected in the fibrillated areas of the remaining upper part of the tissue. ADAM-10 protein was predominantly localized on the cell membrane/lacunar wall of cell clusters consisting of large clones of many chondrocytes. In the deep layer, only single chondrocytes showed some positive staining. Light matrix staining was also noted in OA cartilage in the interterritorial matrix of the upper part and in the territorial matrix of the deep zone of tissue (Figures 2G and 2H). Tissue sections used as negative controls were subjected to the antibody preabsorbed with antigenic peptide (Figure 3B) or no primary antibodies (Figure 3A) and revealed no immunoreactivity.
Western Blotting Analysis
By Western blotting, immunoreactive ADAM-10 bands were detected in all cartilage samples utilized in the current studies. In 1 M GuHCl extracts under non-reduced conditions, the majority of ADAM-10 protein was represented by a high molecular weight band at approximately 220 kD. However, lower molecular weight bands at 75–90, 60, 50, and under 20 kD were also evident (Figure 4A). Under reduced conditions, the highest molecular weight bands (over 220 kD) of ADAM-10 shifted to the lower-sized band at around 85–90 kD. The recombinant ADAM-10 under the same conditions showed the major immunopositivity at about 50–55 kD (Figure 4B, Lane 3), while the major band in cartilage extracts was found at 60 kD along with some additional bands of higher and lower molecular weight (Figure 4B). The difference in molecular weight between the positive control, recombinant ADAM-10, and protein extracted from the tissue might be explained by two facts. First, bacterial lysate recombinant protein is truncated and does not contain the transmembrane and cytoplasmic domains. Second, it is not a glycosylated protein.
Each Western blot was scanned with a Fluor S Image Analyzer (Bio-Rad) and analyzed with the Quantity One software program (Bio-Rad) for the density of the bands detected by the anti-ADAM-10 antibody. Importantly, the same amount of total protein (determined by the microprotein assay) for each cartilage sample was loaded onto the gel and each cartilage extract was treated with chondroitinase ABC to unmask the epitope. The effect of the background staining on each Western blot was also taken into consideration. Each group consisted of at least five cartilage extracts. Quantification of the content of total ADAM-10 (data were normalized to the total protein content) showed that extracts from OA cartilage contained more extractable ADAM-10 than extracts from normal tissue (p < 0.05 under non-reduced conditions and p < 0.001 under reduced conditions) (Figures 4C and 4D). There also was a statistical difference in total ADAM-10 protein content between normal cartilage and cartilage with degenerative changes (p < 0.05 under non-reduced conditions; Figure 4C).
Representative Western blots of knee cartilage extracts with a monoclonal antibody against the active site of the metalloproteinase domain of ADAM-10 molecule (
Because ADAM-10 is a membrane-anchored protein, the logical approach was to isolate cell membranes and determine whether ADAM-10 was localized in the cells, cell membrane, or ECM. Cell membranes were isolated from human cartilage as previously described (Böhm et al. 1994) and were subjected to SDS-PAGE. In all cartilage tested, the immunoreactive ADAM-10 was found primarily in the cell membrane fraction (Figure 5A), whereas in the pellet and supernatant fractions ADAM-10 was barely noticeable. The high molecular weight band in the membrane preparations might represent the insoluble membrane-bound form of ADAM-10. The lower molecular weight band (at about 60 kD), which is hardly detectable in the pellet and supernatant fractions (Figure 5A), might be a contaminant from the membrane fraction and/or a representative of a soluble form of ADAM-10. Primary association of ADAM-10 protein to cell membranes was evident for all types of cartilage: normal, degenerated, and OA. The highest level of ADAM-10 protein in cell membrane-enriched preparations was detected in OA cartilage compared to normal (p < 0.0001) and degenerated (p < 0.002) cartilage (Figures 5B and 5C).
Representative Western blots of cartilage cell membrane-enriched preparation with anti-ADAM-10 antibody under non-reduced conditions. (
Immunohistochemistry of bovine nasal cartilage sections stained with an anti-ADAM-10 antibody. (
Immunohistochemistry of Bovine Cartilage
Experiments were designed to investigate the expression of ADAM-10 protein under the controlled environment of IL-1α stimulation. Bovine cartilage was divided into four groups: control, cartilage cultured in the absence of IL-1α for up to 14 days, and three groups in which cartilage was cultured in the presence of IL-1α for 2, 7, or 14 days. Immunohistochemistry was performed on sections from all these groups using the anti-ADAM 10 antibody to the metalloproteinase domain of the molecule. Clear differences in the intensity of ADAM-10 staining and its distribution were observed among the experimental groups. In cultured control, ADAM-10 was detected at the very low levels only in the interterritorial matrix at either Day 2, 7, or 14 of culture (Figure 6 A). Conversely, cartilage cultured for 2 days with IL-1α showed a dramatic increase in ADAM-10 staining in the territorial and interterritorial matrix throughout the tissue section (Figure 6B); little cellular staining was noticed. In cartilage cultured for 7 days in the presence of IL-1α, strong ADAM-10 staining was primarily associated with the cytoplasm of the cells; little staining was found in the interterritorial matrix (Figure 6C). When bovine nasal cartilage was treated with IL-1α for 14 days, ADAM-10 was also detected in the cytoplasm, but some ADAM-10 staining was also found in the territorial and interterritorial matrices. However, the overall intensity of the matrix staining at this time point was lower than that on Day 2 of culture (Figure 6D). Negative control for immunostaining (no primary antibody was added) showed no anti-ADAM-10 binding (data not shown).
Comparison of toluidine blue staining (
Toluidine blue staining of bovine nasal cartilage plugs was performed to evaluate the changes in the distribution of proteoglycans in response to IL-1α treatment. Staining was strong in uncultured plugs and cultured control plugs (data not shown). However, by Day 2 of culture in the presence of IL-1α the content of proteoglycans was considerably reduced (Figure 7A), especially on the periphery of the tissue plug. The changes in the proteoglycan distribution correlated with the distribution of anti-ADAM-10 staining (Figure 7B). The area rich in proteoglycans showed no noticeable ADAM-10 staining, whereas areas with the depleted proteoglycans displayed strong matrix staining with the anti-ADAM-10 antibody.
Discussion
The purposes of this study were to determine whether chondrocytes from human and bovine cartilage of different histopathological status express ADAM-10 protein and to identify its cellular and/or matrix localization. An important segment of this project was investigation of changes in ADAM-10 protein distribution in bovine chondrocytes under the catabolic effects of IL-1α. This study was a continuation of previously published work by our group (Chubinskaya et al. 1998), in which we documented ADAM-10 mRNA expression in various human cartilage and the changes this mRNA expression underwent with age and cartilage degeneration. To our knowledge, this is the first paper that addressed the question of ADAM-10 protein localization and distribution in human articular cartilage as well as attempts to understand the mechanisms of its processing and function under degenerative processes naturally occurring or mimicked by the influence of the catabolic mediator IL-1α.
The present immunolocalization of ADAM-10 protein in normal and OA cartilage indicated its preferential expression in the vicinity of cell membranes and the lacunar walls, with minor levels in the cytoplasm and ECM. Fixation artifacts, such as rupture of the plasma membranes and shrinkage of the cells, do not allow a more precise localization at present. The identification of ADAM-10 in the territorial and interterritorial matrices of the degenerated cartilage might be explained as follows. The process of tissue degeneration is accompanied by the upregulated activity of different proteinases including matrix metalloproteinases (Nagase 1996; Billinghurst et al. 1997; Chubinskaya et al. 1999a, b). This enzymatic activity could lead to the processing of ADAM-10 and then eventually result in the shedding of this protein from the cell membrane. This supports the hypothesis that, via its metalloproteinase domain, ADAM-10 might be involved in the degradation and/or remodeling of the cartilage pericellular matrix or, when shed, could be proteolytically active against the constituents of the territorial and/or interterritorial matrices. However, substrates for ADAM-10 and mechanisms regulating its function have yet to be elucidated. The proteolytic activity of ADAM-10 has been demonstrated thus far only for the myelin basic protein. Nevertheless, it is unlikely that this specific cytoplasmic protein would be the only physiological substrate for this widely distributed proteinase (Howard et al. 1996).
Importantly, the localization of ADAM-10 protein in newborn, degenerated, and OA cartilage was identical to the localization of ADAM-10 message described previously by our laboratory (Chubinskaya et al. 1998), suggesting relatively rapid turnover of this molecule in human cartilage. Moreover, the highest expression of ADAM-10 in the tissues with more active metabolic activity (growth plate, cell clusters of OA cartilage) probably implies the importance of this protein in the processes of cell differentiation and tissue remodeling.
In all cartilage extracts tested, the band at 60 kD was detectable. This size of the protein has been reported for the ADAM-10 molecule (Rosendahl et al. 1997). Under non-reduced conditions, additional higher molecular weight bands (at about 70–80 kD and 220 kD and higher) could be visualized, suggesting a possible oligomeric structure for ADAM-10. In addition, higher molecular weight bands could indicate the potential binding of ADAM-10 protein to different cartilage matrix components extracted from the ECM. Lower molecular weight bands could either represent partially degraded products of ADAM-10 or indicate a proteolytically processed form of ADAM-10. The extraction of the cell membrane-enriched fraction from the tissue (Böhm et al. 1994) and Western blotting of these extracts with the specific antibodies confirmed the preferential cell membrane localization of ADAM-10 protein in adult human cartilage. These findings are supported by two facts: first, the domain structure of this protein, with the transmembrane domain providing the anchorage of ADAM-10 to the cell membrane (Blobel 1997; Wolfsberg et al. 1995; Wolfsberg and White 1996) and, second, ADAM-10 possesses TNF-α processing activity (Black et al. 1997; Moss et al. 1997; Rosendahl et al. 1997) that most likely occurs on the cell surface by the membrane-bound proteinase.
To address the question of whether shedding of ADAM-10 from the cell membrane occurs with tissue degeneration, we cultured plugs of bovine nasal cartilage in the presence of IL-1α. Proteoglycans are lost during the first 10 days [measured as glycosaminoglycan release (Farndale et al. 1982)], reaching a maximum by Day 4. Collagen loss as measured by hydroxyproline release (Ellis et al. 1994) begins at Day 11 and continues until the cartilage is completely degraded, by approximately Day 21. The appearance of activated matrix metalloproteinases in the culture media is parallel to hydroxyproline release (data not shown). The results of immunohistochemistry indicated the increased synthesis and/or elevated cellular release of ADAM-10 in response to IL-1α treatment as well as redistribution of this protein with time in culture. By Day 2 of IL-1α culture the majority of ADAM-10 was detected in the ECM. By Days 7 and 14 of culture ADAM-10 was primarily present in the cytoplasm of the chondrocytes. A possible explanation for the temporal pattern of ADAM-10 distribution in the bovine nasal cartilage described here is that the metalloproteinase domain of ADAM-10 (to which the antibody was raised) is likely to be shed by Day 2. This shedding could lead either to the inactivation of the enzyme or to the facilitation of the ECM degradation outside the pericellular region. Because we were not able to detect ADAM-10 released into the medium, it is possible that the protein is rapidly broken down in the cartilage or in the medum. Cellular/cytoplasmic ADAM-10 immunostaining at Days 7 and 14 is likely to represent new synthesis of this protein. The strong correlation that was observed between the loss of proteoglycans and the presence of ADAM-10 in the interterritorial matrix might be indicative of an involvement of ADAM-10 in cartilage matrix breakdown and especially in the catabolism of proteoglycans. However, it remains to be elucidated whether ADAM-10 has a direct effect on proteoglycan degradation or an indirect effect via activation of IL-1-induced TNF-α, thereby amplifying the cytokine-induced proteinase activities. The only ADAMs that have been shown thus far to be directly involved in the cleavage of proteoglycans are aggrecanases (Arner et al. 1997, 1999; Tortorella et al. 1999).
The full-length ADAM-10 molecule is membrane bound. Nevertheless, because there are two trypsin-like cleavage sites at the C-terminal of the metalloproteinase domain, it is hypothesized that the active metalloproteinase domain is shed as the result of serine-dependent protease activity. It is also possible that alternative splicing of ADAM-10 mRNA gives rise to a secreted isoform of the protein, although data from our laboratory indicate that only a single 4.5-kb ADAM-10 transcript can be detected by Northern blotting of mRNA from bovine nasal chondrocytes stimulated with IL-1α (data not shown).
In summary, we have identified ADAM-10 in human articular and bovine nasal cartilage primarily as a cell membrane-associated protein, whose expression correlates with the degree of cartilage damage and/or degeneration. Naturally occurring or IL-1-induced degeneration of either human articular or bovine nasal cartilage might lead to the shedding of this proteinase.
Footnotes
Acknowledgments
Supported by a grant from Procter & Gamble Pharmaceuticals and in part by NIH grant 2P50-AR-39239.
We thank Dr Arcady Margulis for the procurement of the human cartilage and Drs Gabriela Cs-Szabo and Richard Berger for providing OA cartilage. Collaboration with Dr Allan Valdellon (Regional Organ Bank of Illinois) and his staff is gratefully acknowledged. We also thank Mr Rocco Rotello for testing the specificity of the antibody.