Expression and activity of ADAMTS-5 in synovium


M. N. Vankemmelbeke, University of Sheffield Medical School, Division of Genomic Medicine, Beech Hill Road, Sheffield S10 2RX, UK. Fax: + 44 114 2713781, Tel.: + 44 114 2712073, E-mail:


ADAMTS proteinases, belonging to the adamalysin subfamily of metalloproteinases, have been implicated in a variety of cellular events such as morphogenesis, cell migration, angiogenesis, ovulation and extracellular matrix breakdown. Aggrecanase-1 (ADAMTS-4) and aggrecanase-2 (ADAMTS-5) have been identified in cartilage and are largely responsible for cartilage aggrecan breakdown. We have shown previously that synovium, the membrane lining diarthrodial joints, generates soluble aggrecanase activity. We report here the expression, localization and activity of ADAMTS-5 from human arthritic and bovine synovium. ADAMTS-5 was expressed constitutively in synovium with little or no transcriptional regulation by recombinant human interleukin-1α or all-trans-retinoate, factors previously shown to upregulate aggrecanase activity in cartilage. Aggrecanase activity generated by synovium in vitro and recombinant ADAMTS-5 cleaved aggrecan extensively, resulting in aggrecan fragments similar to those generated by chondrocyte-derived aggrecanases, and the activity was inhibited by heparin. ADAMTS-5 was immunolocalized in human arthritic synovium, where staining was mostly pericellular, particularly in the synovial lining and around blood vessels; some matrix staining was also seen. The possibility that synovium-derived ADAMTS-5 may play a role in cartilage aggrecan breakdown is discussed.


a disintegrin and metalloproteinase


a disintegrin and metalloproteinase with thrombospondin motifs


1,9-dimethylmethylene blue


extracellular matrix


glyceraldehyde-3-phosphate dehydrogenase




poly(vinylidene difluoride)


recombinant human interleukin 1 α


rheumatoid arthritis




serum-free Dulbecco’s modification of Eagle’s medium


sulfated glycosaminoglycan

The ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs) proteinases are a growing group of recently identified metallopeptidases related to the ADAMs (a disintegrin and metalloproteinase) [1]. They are classified in the Merops database of peptidases as members of subfamily B (the adamalysin subfamily) of family M12 in clan MA [2]. So far, nine mammalian ADAMTS proteinases have been identified [3]. They, unlike most ADAMs, do not possess an epidermal growth factor repeat domain or transmembrane domain but instead have a thrombospondin type I motif, a cysteine-rich spacer region and usually one or more C-terminal thrombospondin submotifs [4]. The thrombospondin type I repeats in thrombospondins 1 and 2 possess affinity for heparin and other sulfated glycosaminoglycans (sGAGs) as well as fibronectin and also contain cell adhesion sequences [5–7]. ADAMTS enzymes are therefore soluble proteins that may become incorporated into the extracellular matrix (ECM) after secretion, as is the case for ADAMTS-1 [8]. They may be activated during secretion through furin cleavage and release of the propeptide [9]. ADAMTS proteinases have been implicated in a wide range of ECM-degrading events [1,10–17]. For many of the ADAMTS proteinases however, biological functions remain to be assigned [3,18]. Two members, ADAMTS-1 and -8 have also been shown to possess potent antiangiogenic properties [19].

Aggrecan, a large aggregating proteoglycan, is together with type II collagen the major constituent of articular cartilage. The type II collagen fibres provide the cartilage with tensile strength and aggrecan, highly hydrated due to the negative charge density of its sGAG chains, provides cartilage with the ability to resist compressive loads and is also important in providing a virtually friction-free articulating surface [20]. Diarthrodial joints are lined with a thin, highly cellular membrane called the synovium. The cells of the healthy synovium can be divided into type A, or macrophage-like cells and type B, or fibroblast-like cells [21]. The synovium controls the volume and composition of the synovial fluid and is an effective route of nutrition for the chondrocytes.

Degradation of cartilage aggrecan has been attributed mainly to the action of a glutamyl endopeptidase, termed ‘aggrecanase’, and aggrecan degradation products resulting from its action have been found in in vitro cultures of cartilage treated with proinflammatory cytokines or Ret, as well as in synovial fluid of osteoarthritis (OA) and rheumatoid arthritis (RA) patients [22–26]. The early work by Fell and Jubb showed that the synovium was capable of accelerating the degradation of articular cartilage [27]. Most of the subsequent attention has been focused on a cytokine-driven pathway whereby chondrocytes, activated by synovium-derived cytokines, degrade their surrounding matrix. In contrast, we have shown recently that synovium is capable of generating soluble aggrecanase activity which degrades articular cartilage aggrecan in a coculture model [28]. At the time the nature of this aggrecanase activity was largely unknown. Recently, two cartilage-derived aggrecanases have been identified [29,30]. Aggrecanase-1 or ADAMTS-4 {also described as ADMP-1, ADAMTS-2 (Tang) and KIAA0688 protein [2]} and aggrecanase-2 or ADAMTS-5 (also described as ADAMTS-11 and ADMP-2 [2]) show a large degree of sequence identity with ADAMTS-1, an inflammation-associated mouse gene product cloned from a cachexigenic tumour cell line and inducible by rhIL1 in vitro[31] and which exhibited metalloproteinase activity [32]. It has recently been shown that ADAMTS-1 is also capable of cleaving aggrecan in its C-terminal chondroitin sulfate attachment domain [33].

We describe here the expression, activity and localization of synovium-derived ADAMTS-5, and compare its activity with that of aggrecanase activity generated by synovium in vitro.

Experimental procedures


Bovine tracheal chondroitin sulfate A, Chaps, SYBR® Green, 1,9-dimethylmethylene blue (DMB), Mayer’s haematoxylin, the anti-FLAG monoclonal antibody M2 and M2 agarose affinity resin were from Sigma-Aldrich. Keratan sulfate endo-1,4-β-galactosidase (keratanase, endo-β-galactosidase; EC, chondroitin ABC lyase (EC and antibody 5/6/3-B-3 were from ICN Flow. Nitrocellulose membranes and the alkaline phosphatase colour detection kit were from Bio-Rad. Polyvinylidene difluoride (PVDF) membranes (Immobilon-P) were from Millipore Corporation. The goat anti-(rabbit IgG) Ig-alkaline phosphatase conjugate was from Promega. Q-Sepharose was from Pharmacia Biotech. Culture medium, TRIzol, and Moloney murine leukemia virus reverse transcriptase (MMLV-RT) were from Gibco Life Technologies Ltd. PCR primers were from Bioline Ltd (London, UK) and Amplitaq Gold was from PerkinElmer Applied Biosystems. The gel purification and plasmid miniprep kits were from Qiagen Ltd. The pCRII-TOPO vector and TA cloning kit were from Invitrogen. EcoRI was from Promega and Glycergel water-soluble mounting medium was from Dako Corp. The baculovirus BaculoGold and the vector pVL 1392 were from Pharmingen.

Recombinant ADAMTS-5

Human ADAMTS-5 from the furin cleavage site to the stop codon was generated by PCR, modified to contain an N-terminal signal sequence from the agouti-related protein gene followed by the FLAG sequence and cloned into the baculovirus vector pVL1392. Sf 9 cells were cotransfected with the BaculoGold™ baculovirus according to the manufacturer’s recommendations. Secreted protein expression was checked on Western blots using the anti-FLAG M2 monoclonal antibody, and recombinant human ADAMTS-5 was purified using the anti-(FLAG M2) agarose affinity resin with elution in 0.1 m glycine/HCl, pH 3.0. The pH was adjusted by adding 1 m Tris/HCl, pH 8.0 followed by overnight dialysis against phosphate-buffered isotonic saline, pH 7.2.

Synovium explant cultures

Bovine metacarpophalangeal joints were obtained from freshly slaughtered cattle. The synovial membrane not attached to capsular tissue, tendon or ligament was removed, washed in NaCl/Pi (0.154 m NaCl, 2.71 mm Na2HPO4, 1.54 mm KH2PO4, pH 7.4) and cultured in sterile tissue culture dishes. For every 1 g (wet weight) of synovium 10 mL serum-free Dulbecco’s modified Eagle’s medium (SF DMEM) supplemented with 2 mm glutamine, 25 µg·mL−1 gentamicin, 0.2 µg·mL−1 amphotericin B, 100 U·mL−1 penicillin G and 0.1 mg·mL−1 streptomycin sulfate solution was added and the cultures were maintained for 24 h (RNA extraction) or up to 3 days (immunolocalization and Western blotting). In some cases 0.3 nm recombinant human interleukin 1α (rhIL1α) or 1 µm all-trans-retinoate (Ret) were added to the culture medium. The conditioned media were stored at −40 °C until further use, the tissue was snap-frozen in liquid nitrogen, ground, dissolved in TRIzol [1 mL per 100 mg (dry weight) tissue] and stored at −80 °C. Tissue for immunolocalization was put into 4% (v/v) formaldehyde in NaCl/Pi overnight at 4 °C after which it was paraffin-embedded and 5 µm sections were cut.

Assays of aggrecan-degrading activity

Aggrecan-polyacrylamide particles. Aggrecan, purified from bovine nasal cartilage by precipitation with DMB [34] entrapped in polyacrylamide was used as a substrate to determine aggrecan-degrading activity [35,36]. Samples, in triplicate, of the aggrecan/polyacrylamide particles (≈ 4 mg dry weight) were placed in a minifuge tube to which was added 450 µL synovium-conditioned medium and 50 µL 1 m Tris/HCl, 1 m NaCl, 100 mm CaCl2, 1% (w/v) Chaps, pH 7.5 (10 × assay buffer). In some instances heparin (50–100 µg·mL−1) was added to the incubation mixture. The heparin was first boiled for 4 min to remove traces of proteoglycanolytic activity. The tubes were incubated at 37 °C for up to 5 h. At the end of incubation the particles were subjected to brief centrifugation and the sGAG content in the supernatant was assayed using DMB [35]. The amount of sGAG released was calculated following subtraction of background sGAG release occurring in the absence of conditioned medium and the amount of sGAG originally present in the conditioned medium. Results are expressed as µg sGAG released per 100 mg of tissue (wet weight) and are the mean ± SEM of three independent experiments.

Western blotting. Aggrecan monomer was purified from bovine articular cartilage as described previously [37]. Synovium-conditioned medium (0–3 days) (1 mL) or recombinant human ADAMTS-5 (≈ 10 µg) was added to 2.5 mg aggrecan monomer dissolved in 1 mL assay buffer, and incubated for 16 h at 37 °C. The resulting digestion mixture was treated with keratanase and chondroitinase ABC lyase (10 U·mL−1 each) and the deglycosylated samples were subjected to electrophoresis on 4–10% gradient polyacrylamide gels in the presence of SDS. Proteins were electro-transferred onto nitrocellulose membranes using 10 mm Caps, 10% (v/v) methanol, pH 11.0 as transfer buffer, at 4 °C. The membranes were then blocked with 5% (w/v) skimmed milk powder in NaCl/Tris (20 mm Tris, 0.15 m NaCl, pH 7.5) and incubated with the appropriate antibodies for 1 h at ambient temperature.

Antibody R663, which recognizes the new N terminus produced by aggrecanase cleavage at the Glu373–Ala bond in the aggrecan interglobular domain [28] was used at a 1 : 500 dilution in 2.5% (w/v) skimmed milk powder in NaCl/Tris containing 0.1% (v/v) Tween 20. Antibody 5/6/3-B-3, which recognizes terminal unsaturated chondroitin 6-sulfated disaccharides [38] was diluted as above to 1 : 256. The membranes were washed three times in NaCl/Tris containing 0.1% (v/v) Tween 20 and then incubated with the second antibody, goat anti-(rabbit IgG) Ig–alkaline phosphatase conjugate (1 : 1000). Finally, the blots were developed using an alkaline phosphatase colour detection kit, rinsed in water and air-dried.

N-Terminal amino acid sequencing

Aggrecan fragments were generated from bovine aggrecan by incubation with bovine synovium-conditioned medium or recombinant human ADAMTS-5, as described above. They were then isolated by anion-exchange chromatography on Q-Sepharose and CsCl density gradient centrifugation [35]. They were recovered in the bottom fraction of the gradient and were deglycosylated by treatment with chondroitin ABC lyase and keratanase. Following reduction and denaturation they were subjected to electrophoresis on 4–10% gradient polyacrylamide gels in the presence of SDS. Following electrophoresis proteins were electro-transferred to PVDF membranes at pH 11.0, as described above. The membranes were washed in water, stained with Coomassie blue, and the polypeptide bands were excised and applied to a Hewlett-Packard G1005A Protein Sequencer.

Western blotting of ADAMTS-5 in synovium-conditioned medium

Conditioned medium samples of human and bovine synovium were boiled in 4 × Laemmli sample buffer containing mercaptoethanol [39] and electrophoresed on 10% polyacrylamide gels containing SDS. Proteins were transferred onto nitrocellulose membranes and immunoblots were developed as described above. The antibody used (AB3612) was raised in rabbits by Alpha Diagnostic International (San Antonio, TX, USA). An 18-amino acid oligopeptide [Val-Asp-Lys-Thr-Lys-Lys-Lys-Tyr-Tyr-Ser-Thr-Ser-Ser-His-Gly-Asn-Trp-Gly(554–571)] corresponding to a region in the disintegrin-like domain of human [29] and bovine ADAMTS-5 (AAF07177) was coupled to keyhole limpet haemocyanin and injected into rabbits. Bleeds were tested by ELISA and the titre of the antibody used in these studies was 1 : 100 000 against the original peptide. AB3612 and preimmune serum were used at a 1 : 1000 dilution, goat anti-(rabbit IgG) Ig–alkaline phosphatase at 1 : 10 000 dilution.


Human arthritic synovial membranes were obtained from knee joints of adult patients undergoing total joint arthroplasty. Specimens were fixed in 4% (v/v) formaldehyde in NaCl/Pi overnight at 4 °C after which they were parafin-embedded and sectioned (5 µm). Sections were deparaffinized, blocked with 10% (v/v) goat serum, 1% (w/v) BSA in NaCl/Tris and incubated for 1 h at ambient temperature with AB3612 or preimmune serum diluted 1 : 100 in 2% (v/v) goat serum, 1% (w/v) BSA in NaCl/Tris, followed by three washes of 10 min with NaCl/Tris. A biotin–streptavidin detection system (Sigma) consisting of a biotinylated anti-(rabbit IgG) Ig [1 : 100 in 4% (v/v) normal human serum, 2% (v/v) goat serum, 1% (w/v) BSA in NaCl/Tris] and ExtrAvidin alkaline phosphatase [1 : 20 in 1% (w/v) BSA in NaCl/Tris] was used according to the manufacturer’s recommendations. Positive reaction was seen as a pink staining; counterstaining was with Mayer’s haematoxylin (blue). An additional negative control consisted of adsorption of the antiserum overnight at 4 °C with recombinant human ADAMTS-5 prior to addition to the tissue section.

RT/PCR analysis and sequencing

Total cellular RNA was extracted from the TRIzol samples according to the supplier’s guidelines. RNA (2 µg) was reverse transcribed using MMLV-RT (Gibco) following the supplier’s specifications. Primers were designed based on the published sequences for human ADAMTS-5 [40]: 5′-GCTAGGCGACAAGGACAAGAG-3′ (forward) and 5′-CGCAGGCAGCTTCTTGGTCAG-3′ (reverse). cDNA (< 1 µg) was subjected to the following PCR cycles in a PerkinElmer Gene Amp 9700 machine: 12 min at 95 °C followed by 45 cycles of (30 s at 95 °C, 1 min at 50 °C, 2 min at 72 °C) and 7 min at 72 °C. Each reaction contained 5 ng primer, 2 U Ampli Taq Gold and 1 mm dNTPs. PCR products were analysed on 0.5% agarose gels. Products of the correct size were excised from the gels and purified using the Qiagen gel purification kit.

The purified products were cloned into the pCRII-TOPO vector using the TA cloning kit following the manufacturer’s instructions, and positive colonies were picked and expanded. Plasmid preparations from positive clones were performed using a Qiagen mini-prep kit and the plasmids were sequenced to confirm their identity. Several clones containing the correct sequences for ADAMTS-5 were identified.

Real-time PCR analysis of ADAMTS-5

Expression levels of ADAMTS-5 in bovine synovium were analysed via real-time PCR using the LightCycler™ system from Roche Molecular Biochemicals. This experiment was performed twice generating similar results and for each experiment bovine synovium from at least six different metacarpophalangeal joints was used. Synovia were isolated and pooled followed by a 24-h culture period in SF DMEM (control), 1 µm Ret or 0.3 nm rhIL1α. Total RNA was isolated and cDNA prepared as described above. cDNA was subjected to nested PCR (in duplicate) as follows. Primary PCR amplification with the outer primers (Table 1) for ADAMTS-5 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (2 µL cDNA in a 50 µL reaction vol.) consisted of 20 cycles of 20 s incubations at 94 °C, 50 °C and 72 °C using a PE Applied Biosystems 2400 PCR machine. This was then followed by real-time PCR analysis using inner primers for ADAMTS-5 and GAPDH (Table 1) and SYBR Green technology in the LightCycler. The primary PCR amplification reaction mixture (2 µL) was added to 18 µL reaction buffer containing 1 mm dNTPs, 3 mm MgCl2, SYBR Green, 1 mg·mL−1 BSA, inner primers and Taq polymerase in glass capillary tubes. PCR consisted of an initial denaturation step at 95 °C (5 s) followed by 40 amplification cycles of 5 s at 50 °C, 15 s at 72 °C and data for each cycle were acquired at 82 °C for 5 s. Finally all samples were subjected to melting curve analysis in order to confirm specificity of the PCR products. Data analysis consisted of calculating the ratios of the crossing point (cycle number corresponding to the log–linear phase of product amplification) for ADAMTS-5 amplification in each sample to the GAPDH crossing point in the same sample. The data obtained were relative expression levels of ADAMTS-5 standardized by comparison with the levels of GAPDH mRNA determined under identical conditions.

Table 1.  Oligonucleotide primers used for real-time PCR analysis of ADAMTS-5 and GAPDH expression. Primer sequences correspond to the sequences from human ADAMTS-5 and GAPDH.
TemplatePrimer orientationPrimer sequenceProduct size (bp)

Results and Discussion

RT/PCR analysis of ADAMTS-5 mRNA from synovium

We have shown previously that synovium expresses aggrecanase activity [28]. The publication of the sequence of ADAMTS-5 enabled us to investigate its expression in bovine and human synovial tissue. We found that bovine synovium (Fig. 1, lane 1) and human OA synovium (Fig. 1, lane 2) expressed mRNA for ADAMTS-5. mRNA for ADAMTS-4 was also present in both bovine and human arthritic synovium (data not shown). Abbaszade et al. [29] had reported previously the expression of aggrecanases-1 and -2 in arthritic joint capsule. In addition, ADAMTS-1, recently shown to possess aggrecanase activity, is also expressed in arthritic joint capsule and fibrous tissue [29].

Figure 1.

RT/PCR analysis of bovine and human OA synovium RNA. Total cellular RNA (2 µg) was reverse-transcribed and < 1 µg cDNA was subjected to PCR analysis as described. Lane 1, ADAMTS-5 in bovine synovium; lane 2, ADAMTS-5 in human OA synovium; M, DNA size markers.

Semi-quantitative assessment of ADAMTS-5 expression by bovine synovium

Having shown the presence of ADAMTS-5 mRNA in bovine synovium, we next investigated its potential transcriptional upregulation by agents that have been shown to upregulate aggrecanase activity in cartilage [41,42].

We analysed ADAMTS-5 expression by real-time PCR using the LightCycler™ technology as described earlier. The ratio of ADAMTS-5 mRNA to GAPDH mRNA for untreated bovine synovium was 0.25. Treatment of bovine synovium with 0.3 nm rhIL1α or 1 µm Ret for 24 h did not lead to a substantial increase in the relative (standardized against GAPDH expression) ADAMTS-5 expression levels: 0.21 and 0.26, respectively. Similar results were obtained for ADAMTS-4 expression (data not shown). Thus, factors such as Ret and rhIL1α which, when added to cartilage explant cultures cause a dramatic up-regulation of aggrecanase activity, failed to alter expression levels of ADAMTS-4 and -5 mRNA in synovium.

It remained a possibility that merely placing synovium in culture was sufficient to ‘activate’ it in terms of aggrecanase expression, so that lack of a further increase in expression following treatment with rhIL1α or Ret was an artefact of the experimental system. However, when RNA from freshly isolated synovium was subjected to real-time PCR, no significant differences in ADAMTS-5 expression levels were detected in comparison to synovium maintained in culture for 24 h (not shown). Flannery et al. [43] demonstrated that human articular cartilage explant cultures treated with rhIL1 or Ret also showed no increase in mRNA levels for aggrecanase-1 and -2 despite a marked increase in release of aggrecanase-generated aggrecan fragments. Overall, the results suggest that control of synovial aggrecanase activity is not exerted at the transcriptional level and points to the potential importance of post-transcriptional controlling mechanisms. ADAMTS-1, by contrast, is an IL1-inducible gene upregulated under inflammatory conditions [31].

Western blotting of synovium-conditioned medium for ADAMTS-5

Conditioned medium from bovine synovium explant cultures was probed for the presence of ADAMTS-5 by Western blotting, using an antibody directed against a sequence in the disintegrin-like domain of human and bovine ADAMTS-5 (AB3612). Analysis of bovine synovium-conditioned medium consistently showed an intense band of approximately 60 kDa (Fig. 2A, lane 1) corresponding in size to active recombinant human ADAMTS-5 (Fig. 2A, lane 4) as well as two fainter bands of approximately 61 kDa and 63 kDa. Arner et al. [40] and Abbaszade et al. [29] have described purified active aggrecanase-2 as four different forms of the same gene product with molecular mass of ≈ 50, ≈ 54, ≈ 62 and ≈ 64 kDa, all containing the same N-terminal sequence. Full-length mature ADAMTS-5 however, has a calculated molecular mass of 74 kDa. Reasons for the apparent discrepancy between the calculated molecular mass and that deduced from SDS/PAGE are currently unclear. It has been shown that recombinant human ADAMTS-1 and -8 are both expressed as full-length and two processed forms [19]. One of these processed forms (≈ 80 kDa) corresponds to a single cleavage at the furin cleavage site in the prodomain, the other (≈ 65 kDa) however, must involve an additional C-terminal cleavage. A similar processing event may exist for ADAMTS-5. It has been demonstrated recently that certain matrix metalloproteinases are capable of performing a C-terminal cleavage releasing two thrombospondin submotifs from ADAMTS-1 [44]. The significance of this processing event in vivo and the enzymes involved remain to be established.

Figure 2.

Western blotting of ADAMTS-5 in synovium-conditioned medium. Antibody AB3612 was used to identify ADAMTS-5 in bovine and human arthritic synovium-conditioned medium. (A) Bovine synovium-conditioned medium (0–3 days); lane 1, control; lane 2, synovium treated with Ret; lane 3, synovium treated with rhIL-1α.; lane 4, recombinant human ADAMTS-5. (B) Human synovium-conditioned medium (0–3 days): lanes 1–6, synovium-conditioned medium from different OA patients; lane 7, synovium-conditioned medium from one RA patient.

We found no evidence for the presence of the pro-enzyme in the conditioned medium, the predicted mass of the unglycosylated human orthologue being ≈ 100 kDa.

Treatment of the synovium with 1 µm Ret or with 0.3 nm rhIL1α for 3 days did not dramatically change ADAMTS-5 protein levels in the conditioned medium (Fig. 2A, lanes 2 and 3, respectively).

Conditioned medium from human OA and RA synovium showed a more varied pattern of bands (Fig. 2B). As with bovine synovium, the majority of samples showed an intense band of ≈ 60 kDa, but fainter bands of higher and lower mobility were also detected. In some OA samples and in one RA sample, additional bands of ≈ 44 kDa (Fig. 2B, lanes 2, 4 and 7) and 25 kDa (Fig. 2B, lanes 2 and 4) not seen in medium from bovine synovium were detected. These smaller bands could represent truncated forms of ADAMTS-5 and cleavage of ADAMTS-5 could be a means of regulating its activity.

Quantification of aggrecan-degrading activity in synovium-conditioned medium

As ADAMTS-5 is one of the recently described aggrecanases, we evaluated the amount of aggrecan-degrading activity present in synovium-conditioned media with the aid of an assay, described in the Experimental procedures, in which proteolytic activity hydrolyses and releases aggrecan entrapped in polyacrylamide [28]. We were unable to detect a statistically significant increase in aggrecan-degrading activity in the conditioned medium of rhIL1α- or Ret-treated bovine synovium. Medium conditioned by 100 mg of tissue released 45 ± 6 (rhIL1α) and 39 ± 9 µg sGAG (Ret) in the assay compared to 37 ± 7 µg sGAG for medium from untreated synovium. This is in contrast with the situation in bovine cartilage where these two reagents are known to up-regulate aggrecanase activity [37,41]. The mechanisms regulating aggrecanase activity in synovium and cartilage may therefore be different. Aggrecan-degrading activity in human OA synovium-conditioned medium (52 ± 15 µg sGAG released in the assay) was generally higher than that in bovine synovium-conditioned medium but this result did not reach statistical significance due to large variation between human tissue samples.

Heparin showed a dose-dependent inhibition (16% inhibition at 50 µg·mL−1 and 70% inhibition at 100 µg·mL−1) of aggrecanase activity in synovium-conditioned medium. This is in line with a recent report showing a dose-dependent inhibition of aggrecanase activity in the presence of 10–1000 µg·mL−1 heparin [45]. Heparin has also been shown to interfere with binding of ADAMTS-1 to the extracellular matrix [8]. Taken together these results suggest that the interaction of ADAMTS proteinases with sGAGs may be extremely important in terms of their localization and activity.

Western blotting and N-terminal sequences of aggrecan fragments generated by aggrecanase activity in bovine synovium-conditioned medium and by recombinant human ADAMTS-5

Aggrecan fragments were generated by incubation of bovine synovium-conditioned medium with purified aggrecan monomer. The fragments were subjected to Western blotting with antibodies 5/6/3-B-3, which recognizes terminal unsaturated chondroitin 6-sulfate disaccharides, and R663 which recognizes the new N-terminus produced by aggrecanase cleavage within the interglobular domain of aggrecan, as described in the Experimental procedures. The results are shown in Fig. 3A. The digestion resulted in at least six aggrecan core protein fragments, as determined by staining with Coomassie blue. N-terminal sequences of these fragments (Table 2) showed the presence of typical aggrecanase cleavages between glutamate and small uncharged aliphatic amino acids in nonglycosylated regions of the aggrecan core protein: in the interglobular domain (Glu373–Ala), between the chondroitin sulfate 1 and 2 attachment regions (Glu1480–Gly) and within the chondroitin sulfate 2 attachment region (Glu1666–Gly, Glu1771–Ala, and Glu1871–Leu). Recombinant human ADAMTS-5 digested aggrecan monomer in a similar way as judged by the migration distance of the aggrecan fragments (Fig. 3B). Lack of the band corresponding to the intact aggrecan monomer and the presence of an additional small molecular mass band designated ‘G’ of 60 kDa, not detected in the digest of aggrecan monomer by synovium-conditioned medium suggested that digestion by the recombinant enzyme was extensive. N-terminal sequences of some of the fragments generated by incubation with recombinant human ADAMTS-5 (Table 3) revealed cleavage of glutamyl bonds in nonglycosylated regions of the aggrecan core protein: in the interglobular domain (Glu373–Ala) and between the chondroitin sulfate 1 and 2 attachment regions (Glu1480–Gly). These sequences are identical to two of those shown in Table 2 and to those released from cartilage explant cultures stimulated with Ret or rhIL1α[25,41,46]. Aggrecanase activity derived from synovium is therefore similar to the activity derived from cartilage. ADAMTS-5 could be generating some or all of the cleavages observed in synovium-conditioned medium although other ADAMTS proteinases such as ADAMTS-1 or -4 may also play a role. Immunoprecipitation of synovium-conditioned medium with AB3612 directed against ADAMTS-5 specifically removed ≈ 20% of the aggrecanase activity as assessed by the aggrecan-containing particle assay (results not shown).

Figure 3.

SDS/PAGE and Western blotting of aggrecan core protein fragments generated by bovine synovium-conditioned medium and recombinant human ADAMTS-5. (A) Aggrecan core protein fragments generated by incubation with synovium-conditioned medium: 1, Western blot with antibody 5/6/3-B-3; 2, Western blot with antibody R663; 3, Coomassie blue-stained gel of purified aggrecan monomer; 4, Coomassie blue-stained gel of aggrecan core protein fragments generated by incubation with synovium-conditioned medium. (B) Coomassie blue-stained gel of aggrecan core protein fragments generated by incubation with recombinant human ADAMTS-5.

Table 2.  N-terminal sequences of aggrecan core protein fragments generated by bovine synovium-conditioned medium. Fragments were generated, separated and sequenced as described. Amino acid numbering is according to the published sequence of bovine aggrecan [49]. The bands are designated as shown in Fig. 3.
BandMolecular mass (kDa)N-terminal sequenceYield (pmol)
C200G1481RGTIDI 8
Table 3.  N-terminal sequences of aggrecan core protein fragments generated by recombinant human ADAMTS-5. Fragments were generated, separated and sequenced as described. Amino acid numbering is according to the published sequence of bovine aggrecan. The bands are designated as shown in Fig. 3 [49].
BandMolecular mass (kDa)N-terminal sequenceYield (pmol)
G 60G1481RGTIDI3
A374XXXV< 2

Immunolocalization of aggrecanase-2 in human OA and RA synovium

ADAMTS-5 was immunolocalized in paraffin sections of human OA and RA synovium with AB3612. In samples from OA patients staining was observed mainly pericellularly at the lining of the synovial membrane (Fig. 4B). Strong positive staining was also associated with cells surrounding blood vessels (arrowhead). Jubb et al. [47] have previously described cartilage proteoglycan breakdown upon coculture of porcine articular cartilage with vascular tissue. In samples of synovium from RA patients pericellular staining around cells lining the pannus as well as staining of some areas of the ECM was observed (Fig. 5A). Fresh bovine synovium (Fig. 5C) showed weak staining for aggrecanase-2 in the synovial intima and subintima. We have observed ADAMTS-5 associated with the ECM of arthritic cartilage, particularly at the articular surface (not shown). The possibility exists that at least some of this cartilage-associated ADAMTS-5 is derived from soft articular tissues such as the synovium and capsule [48] and has traversed the synovial fluid before becoming sequestered in the cartilage ECM.

Figure 4.

Immunolocalization of ADAMTS-5 in human OA synovium. (A) and (B) are representative fields from sections from four OA patients. (A) Nonimmune control. (B) Antibody AB3612. Sections were counterstained with Mayer’s haematoxylin. Original magnifications 10–40×.

Figure 5.

Immunolocalization of ADAMTS-5 in human RA synovium and fresh bovine synovium. (A) and (B) are fields from sections from one representative RA patient. (A) Antibody AB3612. (B) Preadsorption control. Sections were counterstained with Mayer’s haematoxylin. Original magnifications 10–20×. (C) and (D) are representative fields from sections from fresh bovine synovium. (C) Antibody AB3612. (D) Nonimmune control. Original magnification 20×.


We have shown that human diseased and bovine synovium synthesize ADAMTS-5 and secrete ADAMTS-5 and aggrecanase activity into culture medium. Agents leading to increased aggrecanase activity in cartilage explants, rhIL1α and Ret, do not upregulate the expression of bovine synovial ADAMTS-5 transcriptionally, nor do they lead to a major increase in the amount of secreted ADAMTS-5 or an increase in synovium-derived aggrecanase activity. Synovium-derived aggrecanases have very similar specificity to the enzymes from cartilage, cleaving between glutamate residues and small hydrophobic amino acids in the aggrecan core protein, and they are inhibited by heparin. It is possible that ADAMTS-5 and other ADAMTS proteinases from soft articular tissues reach and bind to cartilage ECM components, the binding perhaps being mediated through the thrombospondin motifs. The enzyme may then contribute to the loss of cartilage aggrecan that occurs in arthritic diseases. The relative importance of the different ADAMTS proteinases in synovium-mediated aggrecan breakdown remains unknown but our data point to ADAMTS-5 as an enzyme contributing to this process.


The authors thank K. Corke for the preparation of paraffin sections, and P. Bryant, Barnsley District General Hospital, for provision of human tissue samples. Expert technical advice on the LightCycler™ by S. Read (Micropathology Ltd) is also gratefully acknowledged. M. N. V. and I. H. are supported by grants from the Arthritis Research Campaign (UK). This work was also supported by the Wellcome Trust (UK) and by grants from the Australian National Health and Research Council and the Arthritis Foundation of Australia.


  1. Enzymes: keratan sulfate endo-1,4-β-galactosidase (keratanase, endo-β-galactosidase; EC; chondroitin ABC lyase (EC