Distinguishing Aggrecan Loss from Aggrecan Proteolysis in ADAMTS-4 and ADAMTS-5 Single and Double Deficient Mice*

Aggrecan loss from mouse cartilage is predominantly because of ADAMTS-5 activity; however, the relative contribution of other proteolytic and nonproteolytic processes to this loss is not clear. This is the first study to compare aggrecan loss with aggrecan processing in mice with single and double deletions of ADAMTS-4 and -5 activity (Δcat). Cartilage explants harvested from single and double ADAMTS-4 and -5 Δcat mice were cultured with or without interleukin (IL)-1α or retinoic acid and analyzed for (i) the kinetics of 35S-labeled aggrecan loss, (ii) the pattern of 35S-labeled aggrecan fragments released into the media and retained in the matrix, (iii) the pattern of total aggrecan fragments released into the media and retained in the matrix, and (iv) specific cleavage sites within the interglobular and chondroitin sulfate-2 domains. The loss of radiolabeled aggrecan from ADAMTS-4/-5 Δcat cartilage was less than that from ADAMTS-4, ADAMTS-5, or wild-type cartilage under nonstimulated conditions. IL-1α and retinoic acid stimulated radiolabeled aggrecan loss from wild-type and ADAMTS-4 Δcat cartilage, but there was little effect on ADAMTS-5 cartilage. Proteolysis of aggrecan contributed most to its loss in wild-type, ADAMTS-4, and ADAMTS-5 Δcat cartilage explants. The pattern of proteolytic processing of aggrecan in these cultures was consistent with that occurring in cartilage pathologies. Retinoic acid, but not IL-1α, stimulated radiolabeled aggrecan loss from ADAMTS-4/-5 Δcat cartilage explants. Even though there was a 300% increase in aggrecan loss from ADAMTS-4/-5 Δcat cartilage stimulated with retinoic acid, the loss was not associated with aggrecanase cleavage but with the release of predominantly intact aggrecan consistent with the phenotype of the ADAMTS-4/-5 Δcat mouse. Our results show that chondrocytes have additional mechanism for the turnover of aggrecan and that when proteolytic mechanisms are blocked by ablation of aggrecanase activity, nonproteolytic mechanisms compensate to maintain cartilage homeostasis.

Aggrecan catabolism has been studied extensively in the past 30 -40 years as part of the metabolic processes important in normal functional cartilage and in cartilage pathology. The major enzymes involved in degradation of aggrecan in normal and pathological cartilage are aggrecanases members of the ADAMTS 2 subfamily of adamalysin (M12) metalloproteinases (1,2). These enzymes cleave the aggrecan core protein at aggrecanase-specific cleavage sites between glutamate and a small hydrophobic amino acid residue (3,4). Most studies suggest that in vitro and in vivo ADAMTS-4 and -5 are the main aggrecanases (5)(6)(7)(8)(9)(10)(11); however, ADAMTS-1, -8, -9, -15, -16, and -18 also exhibit aggrecanase activity (12)(13)(14)(15). Recombinant ADAMTS-4 and -5 have been suggested to be the most potent aggrecanases (15)(16)(17)(18)(19), although the relative potency in aggrecan degrading activity has yet to be clarified (11,20,21). In addition, recombinant ADAMTS-1, -4, and -5 cleave aggrecan at 5 sites: one within the interglobular domain and four in the CS attachment region generating aggrecan fragments that match those characterized previously in bovine cartilage and synovial fluid (3,18,19). The processing of aggrecan by aggrecanases does not appear to be species-specific because the same cleavages have been reported in bovine, equine, porcine, murine, and human aggrecan (3,6,9,(22)(23)(24). The aggrecanases are initially synthesized as latent enzymes that require proteolytic modification by autolysis or the action of other proteinases as well as interaction with other matrix macromolecules to gain activity (15,(25)(26)(27)(28). Furthermore it appears that the degree of proteolytic processing of these proteinases may influence the kinetics of loss of aggrecan and the mechanism of aggrecan degradation (15,26). The glycosylation pattern of aggrecan has been reported to have an effect on aggrecanase activity as well (21, 29 -32).
The generation of mice with targeted deletions in ADAMTS genes to ablate catalytic activity in the expressed proteins (⌬cat) has provided an opportunity to investigate which aggrecanase activity is responsible for aggrecan degradation in normal and pathological cartilage. In mouse models of osteoarthritis and inflammatory arthritis, the absence of ADAMTS-5 activity confers partial protection against aggrecan loss and cartilage erosion (5,9). Furthermore the analysis of aggrecan degradation in ADAMTS-1-, -4-, and -5-deficient mice showed that ADAMTS-5 is the main aggrecanase (5,9,(33)(34)(35). In pathology, the increased loss of aggrecan from the matrix is associated with aggrecanase cleavage resulting in the loss of fragments lacking the G1 domain from the matrix (36). The use of wild type and ADAMTS-4-, -5-, and -4/-5-deficient mice together for the first time has allowed us to investigate the contribution of aggrecanase activity to aggrecan loss. This was done by determining the relationship between the kinetics of loss of newly synthesized 35 S-labeled aggrecan from the matrix and the pattern and degree of proteolytic processing of radiolabeled and total aggrecan core protein in cartilage explant cultures maintained in the presence or absence of interleukin-1␣ (IL-1␣) or retinoic acid. The radiolabeling techniques in conjunction with the use of anti-CS, anti-G3, and neoepitope antibodies have been used to overcome the limitations that arise from the low amounts of tissue available from mouse joints. This has also made it possible to obtain an overall picture of aggrecan degradation not restricted to the immunoanalysis of specific cleavage sites.
Cartilage Cultures-Femoral head (hip) cartilage free of bone and adhering fibrous connective tissues (9,22,34) was isolated from 3-week-old mice and cultured in Hepes-, Bes-, and Tesbuffered Dulbecco's modified Eagle's medium (DMEM) containing 20% newborn calf serum for 24 h at 37°C in screwcapped tubes (37). The tissue was then incubated with [ 35 S]sulfate (200 Ci/ml) (PerkinElmer Life Sciences) under the same conditions for 6 h at 37°C. At the end of the incubation period the tissue was washed with DMEM to remove unincorporated radiolabeled sulfate and cultured (2-4 femoral heads/ml medium) in serum-free DMEM in the presence or absence of 10 ng/ml human recombinant IL-1␣ (Peprotech) or 1 M retinoic acid (Sigma) for 6 days with a change of medium every two days. The spent medium was stored at Ϫ20°C in the presence of proteinase inhibitors (38). At the end of the culture period the tissue was extracted with 1 ml of 4 M guanidinium chloride buffered at pH 5.8 in the presence of proteinase inhibitors (38) for 48 h at 4°C and then extracted with 1 ml of 0.5 M NaOH for 24 h.
Measurement of Loss of 35 S-Labeled Aggrecan from Cartilage Explants-The rate of loss of 35 S-labeled aggrecan from the matrix of explant cultures was calculated from the amount of radiolabeled macromolecules appearing in the medium and remaining in the matrix at the end of the culture period (3). The experiments were repeated twice using duplicate cultures. The variation in the rate of aggrecan loss from the two experiments for ADAMTS-4, -5, and -4/-5 ⌬cat mice was Յ5%. The wildtype control cultures were analyzed in parallel with ADAMTS-4-, -5-, and -4/-5-deficient cartilage, and the kinetic data show the outcome from three experiments. Wild-type cultures showed greater variation in the rate of aggrecan loss and represent the natural variation between individual animals on a mixed Ser-129/C57BL6 background.
Analysis of Aggrecan Core Proteins-Aggrecan was isolated from pooled spent medium and guanidinium chloride tissue extracts by ion exchange chromatography as described previ-ously (3). Purified samples were concentrated and exchanged into H 2 O containing proteinase inhibitors using Amicon Ultra-4 centrifugal filter devices with molecular weight cut-off of 10,000 (Millipore Corp., Bedford, MA) as described by the manufacturer, lyophilized and reconstituted in 0.  (38). Digested samples were exchanged into H 2 O containing proteinase inhibitors using the filter devices and lyophilized. Samples were subjected to electrophoresis on 4 -10% gradient polyacrylamide/SDS slab gels. Some gels were fixed and soaked in Amplify (Amersham Biosciences) for 20 min, dried, and exposed to Kodak BioMax Light film at Ϫ80°C for 6 -8 weeks. Other gels were electrotransferred onto polyvinylidene difluoride membranes (Immobilon P, Millipore Corp.) and probed with monoclonal antibody 2B6 (IgG) (kindly donated by Prof. B. Caterson, University of Wales, Cardiff, UK), which recognizes terminal unsaturated chondroitin 4-sulfated disaccharides (41), polyclonal antibodies anti-374 ALGS, anti-SELE 1279 , and anti-FREEE 1467 , which react with the neoepitopes generated by cleavage of mouse aggrecan (42) as described previously (22), and polyclonal antibody anti-G3 (kindly provided by Dr. Jayesh Dudhia, Royal Veterinary College, Hatfield, UK). The primary antibody was detected with mouse or rabbit horseradish peroxidase-conjugated secondary antibodies (Chemicon International) using enhanced chemiluminescence (Chemicon International). All Western blot membranes were probed first by aggrecanase-specific neoepitope antibodies. The membranes were then stripped using mild antibody stripping solution (Re-Blot Plus, Chemicon International) according to the manufacturer's instructions and reprobed with the antibody 2B6.

Rate of Loss of 35 S-Labeled Aggrecan from Cartilage Explant
Cultures-The kinetics of loss of 35 S-labeled aggrecan in cartilage explant cultures examined in the presence or absence of IL-1␣ or retinoic acid is shown in Fig. 1. Approximately 18% of radiolabeled aggrecan was lost from the matrix of ADAMTS-4/-5 ⌬cat cartilage cultured with or without IL-1␣ after a 6-day culture period (Fig. 1A). In contrast, retinoic acid stimulated the loss of radiolabeled aggrecan to 56% from ADAMTS-4/-5 ⌬cat cartilage (Fig. 1B). Compared with ADAMTS-4/-5-deficient cartilage, a higher rate of loss of radiolabeled aggrecan was observed in unstimulated wild-type cartilage (41%) and in wildtype cartilage stimulated with IL-1␣ (65%) and retinoic acid (77%) (Fig. 1, A and B) after a 6-day culture period.
The loss of radiolabeled aggrecan from unstimulated ADAMTS-4 ⌬cat and ADAMTS-5 ⌬cat cartilage cultures was similar to that in the wild type ( Fig. 1, C and D, open symbols). In ADAMTS-4 ⌬cat cartilage, the loss of radiolabeled aggrecan was increased to 75% with Il-1␣ and to 73% with retinoic acid (Fig. 1, C and D). IL-1␣ did not promote loss of radiolabeled aggrecan from ADAMTS-5 ⌬cat cartilage, but there was a small stimulation to 53% by retinoic acid (Fig. 1, C and D). These results reporting the loss of newly synthesized (radiolabeled) aggrecan agree with those reported previously for the loss of total aggrecan from ADAMTS-4/-5 ⌬cat cartilage (35) and ADAMTS-4 ⌬cat and ADAMTS-5 ⌬cat cartilage (9, 22) treated with catabolic agents.
Analysis of 35 S-Labeled Aggrecan Core Proteins-The radiolabeled aggrecan isolated from the matrix and medium of wildtype cartilage explants analyzed by SDS-PAGE and fluorography revealed a number of distinct bands ( Fig. 2A). The banding pattern of radiolabeled and total aggrecan core proteins that results from aggrecanase activity is very similar in bovine and human cartilage explants and in synovial fluid in these species (3,23,36,43,44). This work was based on identification of aggrecan fragments by amino-terminal amino acid sequence analysis (3,23,36,43,(45)(46)(47). The pattern of aggrecan degradation in mouse cartilage is a close match to that reported for bovine cartilage. This is not surprising given the similarity of susceptible sites along the core protein and the carboxyl terminally driven process of aggrecan degradation in cartilage (43). The homologous cleavage sites in mouse aggrecan (42) are located at Glu 373 -Ala within the interglobular domain, at Glu 1279 -Gly between the CS-1 and CS-2 domains, and at Glu 1467 -Gly, Glu 1572 -Ala, and Glu 1672 -Leu bonds within the CS-2 domain (Fig. 3A). These aggrecanase-specific cleavage sites have been confirmed in mouse aggrecan with antibodies to neoepitopes that result from the cleavage of the core protein at Glu 373 -Ala, Glu 1279 -Gly, Glu 1467 -Gly, and Glu 1572 -Ala bonds (5,9,22,34,35,42).
The radiolabeled aggrecan peptides 1, 2a, and 2b present in the matrix and peptides 1-7 present in the medium of wild-type cartilage cultures are shown in Fig. 2A and represented schematically in Fig.  3A. The fragment numbers are the same as those published previously for stimulated and unstimulated cartilage cultures (3,23,43).
Compared with unstimulated cultures, a more extensive degradation of radiolabeled aggrecan was observed in cultures treated with IL-1␣; the cartilage extracts contained reduced levels of intact aggrecan ( Fig. 2A, a and b). However, there was no change in the migration pattern of aggrecan fragments present in the medium of cultures with or without IL-1␣ ( Fig. 2A, c and d). This lack of effect of catabolic stimulus on the pattern of radiolabeled aggrecan degradation has been observed in bovine cartilage explant cultures, where there was no difference in the degradation pattern under any conditions of culture including DMEM alone or in the presence of 20% newborn calf serum or retinoic acid (3).
In Fig. 2B, a, b, e, and f show that in the presence of retinoic acid the majority of intact aggrecan in wild-type cultures has been degraded and that the pattern of degradation of radiolabeled aggrecan is similar to that observed in the presence of IL-1␣ ( Fig. 2A, a-d). In contrast, in ADAMTS-4/-5 ⌬cat cartilage cultures intact aggrecan was the major aggrecan species present in the matrix and medium even in the presence of retinoic acid (Fig. 2B, c, d, g, and h). Also present in these cultures were two large aggrecan fragments, and one of them migrated to the same position as peptide 2a seen in the matrix and medium of wild-type cultures. The presence of these fragments indicated a limited proteolytic processing of the core protein within the CS-2 domain (Fig. 2B, c and d). Peptides 3 and 6 resulting from aggrecanase activity in the interglobular domain and in the CS-2 domain were not detected in the medium of ADAMTS-4/-5 ⌬cat cartilage explants. Peptides 2a, 2b, 4, 5, and 7 were weakly detected in the medium (Fig. 2B, g and h), and there was weak stimulation of peptide 5 by retinoic acid in the ADAMTS-4/-5 ⌬cat cultures (Fig. 2B, g and h). In separate experiments, the addition of IL-1␣ to ADAMTS-4/-5 ⌬cat car-

Effect of ADAMTS-4 and/or -5 Deficiency on Aggrecan Loss
tilage cultures did not change the pattern of cleavage or the levels of aggrecan fragments compared with unstimulated cultures (data not shown). Fig. 2, C and D, shows the pattern of radiolabeled fragments from ADAMTS-4 ⌬cat and ADAMTS-5 ⌬cat cartilage cultures, respectively, with or without IL-1␣ treatment. The pattern of fragments and the response to IL-1␣ did not differ between wild-type and ADAMTS-4 ⌬cat cultures (Fig. 2, A and  C). A similar pattern of radiolabeled fragments was also seen in ADAMTS-5 ⌬cat cartilage stimulated with IL-1␣ (Fig. 2D, b  and d). However, unstimulated ADAMTS-5 ⌬cat cartilage cultures contained peptide 2a in the matrix and peptides 5 and 7 in the medium. In IL-1␣-treated tissue, significant levels of peptide 2b and peptides 3, 4, and 6 appeared in the matrix and medium, respectively, whereas there was only a marginal effect on the levels of peptides 5 and 7 (Fig. 2D). These results show that in the absence of ADAMTS-5 activity mouse aggrecan is still cleaved within Glu 1467 -Gly and Glu 1672 -Leu bonds in unstimulated conditions and that IL-1␣ stimulates cleavage at Glu 373 -Ala, Glu 1279 -Gly, and Glu 1572 -Ala bonds.
Western Blot Analysis of Aggrecan Fragments-To monitor the degree of proteolytic processing of total aggrecan and to correlate this with the radiolabeled fragments, aliquots of media and extracts were fractionated on SDS gels and analyzed by Western blotting with neoepitope antibodies (9,22,35,42). The immunopositive bands were matched to bands detected on the same membrane with antibody 2B6 (41) to show the overall pattern of degradation products.
The 2B6-positive fragments present in wild-type cartilage cultures, with and without treatment with retinoic acid, were peptides 1, 2a, and 2b in the matrix (Fig. 4A, a and  b) and peptides 1-7 in the medium (Fig. 4A, e and f). This pattern of 2B6-positive fragments was almost identical to the pattern of radiolabeled fragments seen in Fig. 2B. More aggrecan was degraded after treatment with retinoic acid (Fig.  4A, a, b, e, and f). In contrast, the majority of 2B6-positive aggrecan in ADAMTS-4/-5 ⌬cat cartilage cultures was intact (Fig. 4A, c, d, g, and  h). Aggrecan peptides 2a and 2b were present in the matrix, and in the medium peptides 2a, 2b, and 5 were also present. These and a number of additional weak bands in the matrix and medium samples of ADAMTS-4/-5 ⌬cat indicated a low level of proteolytic processing of aggrecan that matched the pattern of radiolabeled peptides shown in Fig. 2B, c, d, g, and h.
Cleavage at the Glu 1467 -Gly bond (Fig. 3A) that generates the G1-containing peptide 2a and peptide 5 was confirmed by analysis with the anti-FREEE 1467 antibody which showed the presence of peptide 2a in the matrix (Figs. 4B, a-d, and 3A). Cleavage at this site was increased by retinoic acid treatment in ADAMTS-4/-5 ⌬cat explant cultures (Fig. 4B, c, d, g, and h). Another smaller FREEE 1467 -positive fragment was also present in the medium of wild-type cultures and may correspond to a minor fragment shown in Fig. 3B and discussed below. Analysis with the anti-SELE 1279 antibody detected strong bands corresponding to peptide 2b in wild-type matrix and peptide 3 in wild-type medium (Fig. 4C, a, b, e, and f)   2b in the matrix of ADAMTS-4/-5 ⌬cat cartilage explants (Fig.  4C, c, d, g, and h). Thus, aggrecanase cleavage at the Glu 1279 -Gly bond was not up-regulated by retinoic acid in the double mutant mouse cartilage. A weak increase in the cleavage at this site reported previously may have been because of the higher concentration of retinoic acid used in that study (35). Cleavage at the Glu 373 -Ala bond was not detected in ADAMTS-4/-5 ⌬cat explants shown by the absence of peptide 3 positive band using anti-374 ALGS and anti-SELE 1279 antibodies (Fig. 4C, g  and h, and D, c and d). The analysis of aggrecan core proteins from ADAMTS-4/-5 ⌬cat explants with the anti-G3 antibody confirmed the presence of intact aggrecan in the culture media (Fig. 4E, c and d). This is further supported by the fact that there are no reports in the literature of high molecular weight aggrecan fragments in the media of explant cultures or in synovial fluid matching the electrophoretic mobility of intact aggrecan (44,48,49). Peptides 5 and 7 in media from double deficient cultures and peptides 4 -7 in media from wild-type cultures also contained the G3 domain (Fig. 4E, c, d, and g).
The pattern of 2B6-positive fragments detected in the matrix and media of wild-type, ADAMTS-4 ⌬cat, and ADAMTS-5 ⌬cat cartilage explants after treatment with IL-1␣ (Fig. 5A) correlated well with the pattern of radiolabeled fragments from these cultures (Fig. 2, A, C, and D). ADAMTS-5 ⌬cat cultures contained more intact aggrecan in matrix and fewer fragments in the medium than either wild-type or ADAMTS-4 ⌬cat cultures (Fig. 5A). Similarly reduced levels of FREEE (Fig. 5B, f), SELE (Fig. 5C, f), and ALGS (Fig. 5D, f) neoepitopes were observed in ADAMTS-5 ⌬cat cartilage culture indicating lower aggrecanase activity. Lower levels of ALGS neoepitopes in ADAMTS-5 ⌬cat cartilage culture compared with wild-type and ADAMTS-4 ⌬cat cultures were more apparent after a shorter exposure time (data not shown).
The analysis of medium samples from wild-type, ADAMTS-4 ⌬cat, and ADAMTS-5 ⌬cat explants with anti-body anti-FREEE 1467 showed the presence of G1-containing peptide 2a as well as another peptide comigrating with peptide 2b (Fig. 5B). A weak band comigrating with peptide 2b was also detected with antibody anti-374 ALGS. This antibody also reacted with another fragment slightly smaller than peptide 1 indicated by the arrow (Fig. 5D). It is likely that this is not because of the nonspecific binding of antibodies anti-374 ALGS and anti-FREEE 1467 but because of a small proportion of aggrecan core protein that is cleaved directly in the interglobular domain by aggrecanase activity without any prior or with limited processing within the CS-2 domain (Fig. 3B). The origin of this weak aggrecanolytic activity is not known, although it might be at least partly because of an activity other than ADAMTS-4 or ADAMTS-5. Indeed, in the medium of ADAMTS-4/-5 ⌬cat cartilage cultures a radiolabeled fragment corresponding in size to this aggrecan fragment was also observed (Fig. 2B, g and h). In unstimulated cultures the cleavage within the interglobular domain (peptide 3) was more prominent in wild-type than in ADAMTS-4 ⌬cat cartilage cultures and was not detected in ADAMTS-5 ⌬cat cartilage cultures (Fig. 5E, e).
The unidentified bands of radiolabeled and total proteoglycans at 50 kDa and below are likely to contain the intact core proteins and fragments of small proteoglycans, decorin, and biglycan shown to represent ϳ5-10% of total proteoglycans in cartilage matrix (50). The bands between 200 and 50 kDa are likely to be aggrecan catabolites that result from further processing of fragments described above that have not been identified at this stage.

DISCUSSION
This study investigated the kinetics of loss of radiolabeled aggrecan and proteolytic processing of radiolabeled and total aggrecan in wild-type, ADAMTS-4 ⌬cat, ADAMTS-5 ⌬cat, and ADAMTS-4/-5 ⌬cat cartilage cultures. This study shows that the process of aggrecan loss can be delineated from proteolytic processing and that the stimulated loss of aggrecan from the matrix need not necessarily be a consequence of proteolytic processing. Indeed, the elucidation of the pattern of aggrecan processing in ADAMTS-4/-5 ⌬cat cartilage cultures has shown that the major species lost to medium was intact aggrecan even in cultures stimulated with retinoic acid (Figs.  2B, g and h, and 4A, g and h, and E, c and d). The loss of intact aggrecan and G1-containing aggrecan fragments from the cartilage matrix occurs physiologically, and these aggrecan species can be found in synovial fluid from normal and pathologic joints (3,36,44).
The results suggest that loss of intact or poorly processed aggrecan from unstimulated ADAMTS-4/-5 ⌬cat cartilage explants occurs at a level that is sufficient to maintain steady state aggrecan loss required for normal cartilage function because such deficiency in proteolytic processing of aggrecan does not affect the integrity or function of what appears to be normal cartilage in ADAMTS-4/-5 ⌬cat mice (35). A similar low rate of radiolabeled aggrecan loss to that observed in ADAMTS-4/-5-deficient cultures was also observed in bovine cartilage explants in the presence of calf serum where intact aggrecan was the main aggrecan species lost from the matrix (3). This is the first study to report a significant increase in aggrecan loss from tissue that did not involve its degradation. The 300% increase in the loss of radiolabeled aggrecan from the matrix of ADAMTS-4/-5 ⌬cat cartilage cultures in the presence of retinoic acid, but not IL-1␣, was because of the release of predominantly intact aggrecan. The mechanism responsible for the accelerated loss of intact aggrecan from culture treated with retinoic acid is not known. It is also not known whether it is lost as a monomer or as an aggregate. Hyaluronan degradation has been suggested to contribute to the enhanced loss of aggrecan aggregates in retinoic acid-stimulated fetal bovine cartilage explants (51). Such a mechanism might also have a role in the retinoic acid-stimulated release of G1-containing aggrecan in ADAMTS-4/-5 ⌬cat cartilage explants; however, high levels of intact and high molecular weight aggrecan in the tissue and limited processing of this macromolecule would be expected to hinder the hyaluronidase access to hyaluronan and would remain a barrier to radical scavengers. Release of G1-containing aggrecan might also be facilitated by degradation of link proteins, disruption of the fibrillar networks that organize cartilage matrix, and maturation of the G1 domain that is required for its function (51)(52)(53)(54)(55)(56). As to the reason for the absence of significant loss of G1-containing aggrecan in the retinoic acid-stimulated cultures of other genotypes, it is likely because aggrecanase activation is a more rapid process. The aggrecan loss from wild-type,  DECEMBER 28, 2007 • VOLUME 282 • NUMBER 52

JOURNAL OF BIOLOGICAL CHEMISTRY 37425
ADAMTS-4 ⌬cat, and ADAMTS-5 ⌬cat cultures clearly involved aggrecanase activity but to varying degrees. Accordingly, one significant finding of this study is that the overall contribution of proteolytic processing to aggrecan loss occurs in the following order under all conditions of culture: wild type Ն ADAMTS-4 ⌬cat Ͼ ADAMTS-5 ⌬cat Ͼ ADAMTS-4/-5 ⌬cat cartilage explants. Furthermore, this study shows that although ADAMTS-5 is the major aggrecan-degrading activity in mouse cartilage, similar but weaker activity can be generated by ADAMTS-4 in mouse cartilage. In stimulated ADAMTS-5 ⌬cat cartilage explants, a typical pattern of aggrecan processing was generated showing increased levels of peptides 2b in the matrix (Fig. 2D, b) and peptides 3, 4, 6, and to a lesser extent peptides 5 and 7 in the medium of explant cultures (Figs. 2D, d; 3A; and 5, A-D). Based on densitometric measurements of 2B6-positive peptides (Fig.  5A), our data indicate that the contribution of ADAMTS-5 to cleavages within the interglobular domain and CS-2 domains is approximately twice that of ADAMTS-4 (peptides 3 and 4). Western blotting with anti-374 ALGS and anti-SELE 1279 antibodies confirms the difference in activity between the two enzymes is 2-3-fold (Fig. 5, C and D). The difference in activity between the two proteinases at other sites was less pronounced (see bands 5-7 in Fig.  5A, e and f).
The ADAMTS-4 cleavage at the Glu 373 -Ala bond within the interglobular domain has not been reported previously in ADAMTS-5 ⌬cat cartilage cultures (9,22,35). The difference is likely to be because peptide detection by Western blotting in the present study was made more sensitive by concentrating samples on gradient gels with small wells. In addition, the culture conditions used in this present work were different from those used previously. In the present work, cartilage was cultured for 6 days with a change of medium with or without 10 ng/ml IL-1␣ or 1 M retinoic acid every 2 days. The previous studies used a 3-day culture period with or without 10 ng/ml IL-1␣ or 10 M retinoic and without medium change. The results in this study were confirmed by additional analysis of aggrecan peptides isolated from separate cartilage cultures maintained in the presence of IL-1␣ and retinoic acid (data not shown).
ADAMTS-4 has been reported to be a potent aggrecan-degrading enzyme in articular cartilage (2,7,8,10,11,39,57), yet we detected only a moderate increase in this activity compared with that of ADAMTS-5 in mouse cartilage explants stimulated with IL-1␣ or retinoic acid. There are two main reasons for this result. First, in stimulated cultures the protein levels of active ADAMTS-4 may be lower than that of ADAMTS-5. This may be because ADAMTS-4, unlike ADAMTS-5, is not up-regulated with IL-1␣ or retinoic acid in mouse cartilage (22), and active ADAMTS-4 is generated mainly from the low levels of ADAMTS-4 protein already present in the matrix. The activation process itself is not likely to be at fault because the production of active ADAMTS-4 isoforms from the constitutively produced ADAMTS-4 protein was readily stimulated in cartilage explants in the presence of catabolic stimulators including IL-1 (8,11,57). Second, there might be a difference in the activities between mouse ADAMTS-4 and -5. The comparative studies on the aggrecan-degrading activity of these enzymes using recombinant human enzymes have reported  varying results showing ADAMTS-5 to be less than and up to 1000 times more active than ADAMTS-4 (10,20,21). This study used cartilage derived from young animals, and future studies are needed to determine the differential impact of ADAMTS-4 and -5 on aggrecan degradation in aged cartilage. It has been reported that cleavage by ADAMTS-4, but not ADAMTS-5, in the interglobular domain was affected by the age of animal and human cartilage (21). It has also been suggested that the glycosylation pattern of aggrecan, which varies with age, might also affect the activity of aggrecanases (29,30).
This study shows that chondrocytes in explant culture can use more than one mechanism for the normal turnover of aggrecan and that when proteolytic mechanisms are blocked nonproteolytic mechanisms compensate to maintain homeostasis in the cartilage matrix. It is likely that proteolytic processing of aggrecan predominates in cartilage pathology because significant levels of aggrecan fragments are observed in synovial fluids from human joints with cartilage pathology and joint trauma (36). Because the pattern of fragments in these conditions is consistent with the activity of ADAMTS-4 and -5, the physiological relevance of these two proteinases in articular cartilage may relate to their role in responding to tissue injury.