Selective inhibition of ADAMTS-1, -4 and -5 by catechin gallate esters


D. J. Buttle, Division of Genomic Medicine, University of Sheffield Medical School, Sheffield Children's Hospital, Stephenson Wing, D-Floor, Sheffield S10 2TH, UK. Fax: + 44 114 2755364, Tel.: + 44 114 2717556, E-mail:


Three mammalian ADAMTS enzymes, ADAMTS-1, -4 and -5, are known to cleave aggrecan at certain glutamyl bonds and are considered to be largely responsible for cartilage aggrecan catabolism observed during the development of arthritis. We have previously reported that certain catechins, polyphenolic compounds found in highest concentration in green tea (Camellia sinensis), are capable of inhibiting cartilage aggrecan breakdown in an in vitro model of cartilage degradation. We have now cloned and expressed recombinant human ADAMTS-1, -4 and -5 and report here that the catechin gallate esters found in green tea potently inhibit the aggrecan-degrading activity of these enzymes, with submicromolar IC50 values. Moreover, the concentration needed for total inhibition of these members of the ADAMTS group is approximately two orders of magnitude lower than that which is needed to partially inhibit collagenase or ADAM-10 activity. Catechin gallate esters therefore provide selective inhibition of certain members of the ADAMTS group of enzymes and could constitute an important nutritional aid in the prevention of arthritis as well as being part of an effective therapy in the treatment of joint disease and other pathologies involving the action of these enzymes.


l-α-aminobutyryl ((S)-2-amino-butanoyl)


a disintegrin and metalloproteinase with thrombospondin motifs










(–)-epicatechin gallate




(–)-epigallocatechin gallate




N-[(dimethylamino)-1H-1,2,3-triazolo[4,5-b]pyridin-1-ylmethylene]-N-hexamethylmethanaminium hexafluorophosphate N-oxide




keyhole limpet haemocyanin




metalloproteinase-1 with thrombospondin motifs


n-propyl gallate


recombinant human


sulfated glycosaminoglycan


tissue inhibitor of metalloproteinases

Green tea, made from the leaves of Camellia sinensis, contains catechins, a group of polyphenolic compounds with antioxidant properties that have been at the centre of investigations into the potential medical benefits of consuming green tea. The most abundant catechin in green tea is (–)-epigallocatechin gallate (EGCG) with others such as (–)-epicatechin (EC), (–)-epigallocatechin (EGC) and (–)-epicatechin gallate (ECG) also present. Anti-inflammatory and anti-mitotic properties have been attributed to these compounds [1–3] and they have also been reported to inhibit certain matrixins such as the gelatinases [4–6]. The beneficial effects on a range of clinical conditions including cancer growth and metastasis [7–11], cardiovascular and liver diseases [12] may therefore be due to one or a combination of these properties.

Aggrecan, a large aggregating proteoglycan, is together with type II collagen the major constituent of articular cartilage. Degradation of cartilage aggrecan has mainly been attributed to the action of glutamyl endopeptidases, termed ‘aggrecanases’. Aggrecan degradation products resulting from aggrecanase action have been found in in vitro cultures of cartilage treated with proinflammatory cytokines as well as in synovial fluid of arthritis patients [13–16]. To date three mammalian ‘aggrecanases’ have been identified: a disintegrin and metalloproteinase with thrombospondin motifs (ADAMTS)-1, -4 and -5 [17–19]. The ADAMTS enzymes belong to a subgroup of metallopeptidases in Family M12 of Clan MA in the Merops database [20] and are related to the ADAMs and matrixins [21]. So far, at least 18 mammalian ADAMTS enzymes have been identified, most of which remain to be fully characterized [21,22].

It has been shown recently that inhibition of ADAMTS-4 and -5 can prevent aggrecan breakdown in osteoarthritic cartilage [23]. An in vivo role for ADAMTS-1 in cartilage aggrecan turnover awaits confirmation following the finding that it cleaves aggrecan at glutamyl bonds in vitro[19,24]. ADAMTS-1 and -4 have also been shown to cleave other members of the large aggregating proteoglycan family such as versican and brevican [25–27].

Work in our laboratory has shown that catechin gallate esters are inhibitors of aggrecan and collagen degradation in an in vitro model of cartilage breakdown [28]. The aim of this study was to investigate if catechins and gallate esters directly inhibit the aggrecanases ADAMTS-1, -4 and -5. To this end we have expressed, purified and characterized recombinant forms of these ADAMTSs.

Experimental procedures

Construction of expression vectors

Human ADAMTS-1 (KIAA1346) and ADAMTS-4 (KIAA 0688) clones were kindly provided as inserts in the pBluescript II SK+ vector (SalI and NotI sites) by T. Nagase (Kazusa DNA Research Institute, Kisarazu, Chiba, Japan). Two sets of primers were designed to subclone the respective coding sequences including their signal peptides into the pIB/V5-His/TOPO® vector (Invitrogen™ Life Technologies, Paisley, UK). For ADAMTS-4, primer pair 5′-GCCATGTCCCAGACAGG-3′ (sense) and 5′-GGTT ATTTCCTGCCCGC-3′ (antisense) and for ADAMTS-1 primer pair 5′-GACATGGGGAACGCGGAG-3′ (sense) and 5′-CTT AACTGCATTCTGCCATTG-3′ (antisense) were used. The inserts were sequenced in both directions.

The recombinant baculovirus vector pVL 1392 (Pharmingen, San Diego, CA, USA) was a kind gift from R. Maki (Neurocrine Biosciences Inc., San Diego, CA, USA). It contained the coding sequence for human ADAMTS-5 (GenBank accession number AF142099), which had been modified to contain an N-terminal signal sequence from the agouti-related protein followed by the FLAG™ sequence replacing the first 60 nucleotides of the native coding sequence. External primers 5′-GAAGATCTGACTACAAGGACGACGATGAC-3′ (sense) and 5′-CCTCTAGATTACTAACATTTCTTCAACAAGCATTG-3′ (antisense) containing a BglII and XbaI restriction site, respectively, were designed to generate a PCR product which contained the FLAG™ sequence followed by the coding sequence for ADAMTS-5. This was subcloned into the pMT/BiP/V5-HisB expression vector (Invitrogen™ Life Technologies, Paisley, UK) and its nucleotide sequence was examined by sequencing in both directions. At this stage, two nonconservative point mutations were found in the first thrombospondin repeat G1851A and A1855G resulting in the following amino acid substitutions: G577S and Q579R. These were corrected using overlap extension PCR. The corrected insert including the Drosophila BiP signal sequence provided by the vector, was then subcloned into the pIB/V5-His/TOPO® vector using two primers, 5′-CCGATCTCAATATGAAGTTATGC-3′ (sense) and 5′-CCTCTAGATTACTAACATTTCTTCAACAAGCATTG-3′ (antisense) and the TOPO TA cloning® methodology according to the manufacturer's recommendations. The correct coding sequence was confirmed by sequencing in both directions.

Cell culture

High Five™ cells (Invitrogen™ Life Technologies, Paisley, UK) were maintained and propagated in HyQSFX® serum-free insect cell culture medium (Perbio Science UK, Ltd. Cheshire, UK) containing 10 µg·mL−1 gentamycin at 27 °C. The cells were transfected with the recombinant or empty expression vector using the lipid-based CellFectin® reagent (Invitrogen™ Life Technologies, Paisley, UK). Conditioned cell culture medium was harvested at days 2, 3, 4 and 5 post-transfection and assayed for aggrecanase activity. Stably transfected cell lines were generated in the presence of 80 µg·mL−1 Blasticidin (Invitrogen™ Life Technologies, Paisley, UK). After selection stably transfected cells were maintained in the presence of 10 µg·mL−1 blasticidin. Conditioned medium was stored at −40 °C.

Aggrecanase activity assay

Aggrecan, purified from bovine nasal cartilage by extraction with 4 m guanidinium (Gn) HCl and dissociative CsCl2 density gradient ultracentrifugation [29] was entrapped in polyacrylamide and used as a substrate to determine aggrecan-degrading activity as previously described [29–31]. Aliquots of the aggrecan/polyacrylamide particles (4 mg) were incubated with the respective enzymes in a total volume of 500 µL (assay buffer: 0.1 m Tris/HCl, 0.1 m NaCl, 10 mm CaCl2, 0.1% w/v Chaps, pH 7.5). The tubes were incubated at 37 °C for up to 3 h. At the end of the incubation the particles were subjected to brief centrifugation and the sulfated glycosaminoglycan (sGAG) content in the supernatant was measured using dimethylmethylene blue [32]. One unit of enzyme activity was defined as that which released 5 µg sGAGs per h at 37 °C in the assay.

Purification and characterization of recombinant ADAMTS enzymes

rhADAMTS-1, -4 and -5 were all purified using an identical protocol. Conditioned High Five™ cell culture medium was thawed and the proteinase inhibitors 3,4-dichloroisocoumarin (DCI) and l-trans-epoxysuccinyl-leucylamido-(4-guanidino)butane (E-64) were added to final concentrations of 50 µm and 10 µm, respectively. After an initial buffer exchange on a preparative Sephadex G-25 column (Amersham Pharmacia Biotech, Buckinghamshire UK), samples were assayed for aggrecanase activity (see above) and applied to a heparin-Sepharose Fast-Flow column (Amersham Pharmacia Biotech, Buckinghamshire UK) equilibrated with buffer A (50 mm Tris/HCl, 0.15 m NaCl, 0.1% w/v Chaps, pH 7.0). Activity was eluted using a step-wise gradient (0.15–1 m NaCl). Active fractions were pooled, desalted and loaded onto a Mono Q HR 5/5 (Amersham Pharmacia Biotech, Buckinghamshire UK) column, equilibrated with buffer B (50 mm Tris/HCl, 0.1% w/v Chaps, pH 7.5). Active fractions were eluted between 0.2 m and 0.45 m NaCl on a salt gradient and were pooled. The purity of the enzyme preparations was assessed by SDS/PAGE followed by silver staining of the gels and via Western blot analysis using ADAMTS-specific antibodies (see below).

We assayed conditioned medium from mock-transfected insect cells for aggrecanolytic activity using the assay described above. Proteinase contamination of the rhADAMTS enzyme preparations was examined in two ways. Firstly, the enzyme preparations were incubated with purified aggrecan monomer. Briefly, 10 units of rhADAMTS-1, and 5 units of rhADAMTS-4 or -5 were incubated with 5 mg purified aggrecan monomer (see above) in enzyme assay buffer: 50 mm Tris/HCl, 0.1 m NaCl, 10 mm CaCl2, 0.1% w/v Chaps, pH 7.5 at 37 °C for 16 h. Aggrecan fragments were detected by Western blotting using the monoclonal antibody 5/6/3-B-3 (ICN Flow) which recognizes terminal unsaturated chondroitin 6-sulfate disaccharides. The fragments were then isolated and subjected to N-terminal sequence analysis as previously described [31]. An additional control consisted of analysing the final enzyme preparations for potential contaminating MMP activity using a quenched fluorescence substrate which is cleaved by all MMPs, but which is not cleaved by aggrecanases (see below).

Antibodies to ADAMTS enzymes

Antibodies were raised to peptides prepared by standard solid-phase methods and purified by reverse-phase HPLC. The identity of the peptides was confirmed by MS. Peptides were coupled to KLH using N-succinimidyl bromoacetate [33] or to ovalbumin. The ADAMTS-1 antibody MV-8 was raised in rats against KLH-coupled DPLKKPKHFID-Abu-C (human ADAMTS-1 amino acids 932–942). The ADAMTS-4 antibody was raised in rabbits using a mixture of two peptides VMAHVDPEEPGGC and CGGYNHRTDLFKSFPGP (human ADAMTS-4 amino acids 394–403 and 590–603, respectively) ovalbumin conjugates. The rabbit ADAMTS-5 antibody (3235) was raised against ovalbumin-conjugated ILTSIDASKPGGC and CGGKNGYQSDAKGVKTFV (human ADAMTS-5 amino acids 442–451 and 636–650) and affinity-purified immunoglobulin was prepared using a Sulfolink™ column (Pierce Rockford, IL) substituted with the peptide CGGKNGYQSDAKGVKTFV. Animals were immunized by fortnighly injections of carrier-conjugated peptides emulsified in Freund's adjuvant. Test bleed titres were determined by ELISA. Briefly, plates were coated with immunizing peptide in carbonate buffer, pH 9.0, and blocked in 1% BSA, rinsed and treated with the primary antibody. After 1.5 h, the plates were washed and incubated with alkaline phosphatase conjugated secondary antibody, washed three times, and substrate was added (p-nitrophenyl phosphate tablets; Sigma-Aldrich). Absorbance was measured at 405 nm on a plate reader. The specificity of the antisera was determined via comparison with nonimmune control serum. Cross-reactivity of the antisera with the other ADAMTS enzymes was also examined.

Synthesis of the ADAM-10 substrate Mca-Leu-Ala-Gln-Ala-Val-Arg-Ser-Ser-Ser-Dpa-Arg-OH

This was made on Fmoc-Arg(Pbf)-NovaSyn® TGA resin (0.1 mmol) using standard Fmoc protocols in a PerSeptive Biosystems 9050 Plus PepSynthesiser. Briefly, Fmoc-amino acids (0.4 mmol) were activated with HATU (0.4 mmol) in the presence of diisopropylethylamine (0.8 mmol). HOAt (0.4 mmol) was added when coupling Gln. Fmoc deprotection was with a mixture of 2% (v/v) piperidine and 2% (v/v) 1,8-diazabicyclo[5,4,0]undec-7-ene in dimethylformamide. For the coupling of 7-methoxycoumarin-4-acetic acid (0.4 mmol), the resin was gently shaken with HOAt (0.4 mmol) and diisopropylcarbodiimide (0.5 mmol) in a minimal volume of dichloromethane containing 10% (v/v) N,N-dimethylpropyleneurea and the reaction was allowed to proceed to completion overnight. The peptide was released by treatment with trifluoroacetic acid/water/triisopropylsilane (92.5 : 5 : 2.5, v/v) for 2 h at 21 °C, applied to a column of Vydac 218TPB1520 and eluted with a gradient of 5–50% acetonitrile in 0.1% trifluoroacetic acid. Fractions containing homogeneous product were identified by analytical HPLC, pooled and freeze-dried. The identity of the purified peptide was confirmed by MALDI-TOF (expected mass 1542.6 Da, observed mass 1542.5 ± 0.7 Da).

Assays for matrixin and ADAM-10 activity

rhADAM-10, expressed as a soluble enzyme, was provided by Procter & Gamble, Cincinnati, OH, USA. Purified human collagenase-1 (EC and collagenase-3 were both from Biogenesis Ltd, Poole, UK. The substrate used for the assay of the matrixins was Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2[34,35]. Cleavage of the ADAM-10 quenched fluorescence substrate (see above) followed Michaelis–Menten kinetics with an approximate Km of 20 µm. Determination of a more accurate Km value was not possible due to the insolubility of the substrate. The methods for assaying the matrixins and ADAM-10 were essentially the same using a Perkin Elmer LS 50B luminescence spectrometer (excitation 328 nm, emission 393 nm) controlled by the Flusys software package [36]. Peptide substrates were used at 5 µm in 100 mm Tris/HCl, 0.1 m NaCl, 10 mm CaCl2, 0.2% v/v Triton X-100 pH 7.5, or in 50 mm Tris/HCl, 100 µm ZnCl2, 10% v/v MeOH, pH 7.5 for the matrixins and ADAM-10, respectively. In all assays substrate hydrolysis never exceeded 10% of total.

Determination of enzyme inhibition

For rhADAMTS assays, EC, ECG, EGC, EGCG and PG (all ≥ 95% pure from Sigma-Aldrich Company Ltd, Poole, Dorset, UK) were dissolved in Me2SO to 2 mm to provide the stock solution and diluted to the appropriate final concentration which ranged from 2 to 2000 nm in enzyme assay buffer. Enzymes were preincubated with the inhibitors or the appropriate concentration of Me2SO as control at 4 °C for 30 min prior to assaying their activity using the aggrecanase assay described above. Percentage inhibition was calculated by comparing the levels of activity with those of the enzyme and Me2SO controls. Log-linear plots of the dose–response curves in combination with regression analysis allowed us to determine approximate IC50 values for the inhibitory catechins. In the case of rhTIMP-3 (the full-length inhibitor was kindly provided by Immunex Corp., Seattle, WA, USA), the enzymes were preincubated with 100 nm of the inhibitor for 30 min at 4 °C before assaying.

For the determination of inhibition of collagenases and ADAM-10, quenched fluorescence substrate assays were employed (above). The catechins and PG were used at concentrations ranging from 0.2 µm to 50 µm. The hydroxamate inhibitor BB-94, a broad-spectrum MMP and ADAM inhibitor (provided by A. Galloway, British Biotech Ltd), was used as a positive control at 10 µm final concentration. The initial steady state of substrate cleavage (vo) was recorded, after which inhibitor or Me2SO control were added to the reaction mix in a minimal volume, and the new steady state (vi) was documented. The percentage inhibition was calculated as (1 − vi/vo) × 100.


Purification and characterization of rhADAMTS enzymes

Conditioned medium from mock-transfected insect cells contained no detectable aggrecanolytic activity (results not shown). rhADAMTS-1, -4 and -5 were partially purified by affinity chromatography on heparin-Sepharose followed by anion-exchange chromatography using a Mono Q column at pH 7.5. The purification process was monitored using the aggrecanase assay described in Experimental procedures. As shown in Table 1, each chromatography step was associated with an increase in specific activity of the respective enzymes, with final purification factors ranging from 190- to 370-fold. An increase in total rhADAMTS-1 activity following affinity purification may be indicative of separation from an inhibitor.

Table 1. Purification of rhADAMTS-1, -4 and -5 from conditioned insect cell medium. Conditioned medium was applied to a heparin-Sepharose Fast-Flow column after an initial buffer exchange on a preparative Sephadex G-25 column equilibrated with 50 mm Tris/HCl, 0.1 m NaCl, 0.1% w/v Chaps, pH 7.0. Proteins were eluted using a step-wise gradient up to 1 m NaCl, and fractions were assayed as described in the Experimental Procedures Section. Active fractions were pooled, desalted and loaded onto a Mono Q HR 5/5 column, equilibrated with 50 mm Tris/HCl, 0.1% w/v Chaps, pH 7.5. Active fractions were eluted using a linear (0–1 m NaCl) gradient and pooled. Total protein amounts were estimated via absorption at 280 nm, assuming A280,cm of 1.0 = 1 mg mL−1 protein.
StepProtein (mg)Total Activity (U)Specific activity (U·mg−1)Yield (%)Purification (-fold)
 Heparin F-F1.4473414749
 Heparin F-F1.1827575125
 Heparin F-F2.84101505723

Silver staining of SDS/PAGE gels of the partially purified enzyme preparations revealed that a substantial number of protein bands remained (Fig. 1, lanes 1, 3 and 5). Western blots using the MV-8 antibody, directed against the C-terminus of the rhADAMTS-1, revealed an intense band of approximately 89 kDa (Fig. 1, lane 2). This band probably corresponds to the active, mature enzyme (theoretical molecular mass 78.8 kDa). A band of similar size (p87) has previously been reported for mature ADAMTS-1 [37]. C-Terminal processing of recombinantly expressed ADAMTS-1 has been described previously [37,38]. The MV-8 antibody, directed against the C-terminus of the enzyme also detected a faint band of about 59 kDa (Fig. 1, lane 2), which could represent the C-terminus of a truncated form of ADAMTS-1. Antibody MV-8 did not cross-react with ADAMTS-4 or -5 (data not shown).

Figure 1.

SDS PAGE and Western blots of partially purified rhADAMTS-1, -4 and -5. Panels A, B and C represent preparations of rhADAMTS-1, -4 and -5, respectively. Mono-Q peaks of aggrecanase activity were concentrated 10- to 20-fold using Microcon® YM-30 devices (Millipore Corporation, Bedford, USA.) and about 2 µg total protein was loaded on to 7.5% polyacrylamide SDS-gels run under reducing conditions and subjected to silver staining (lanes 1, 3 and 5). For Western blots 2.5 µg total protein was used in each case. The blot of rhADAMTS-1 with antibody MV-8 (1 : 500) is shown in lane 2, of rhADAMTS-4 with antibody 3170 (1 : 500) in lane 4 and of rhADAMTS-5 with antibody 3235 (1 : 500) in lane 6. The arrows indicate possible forms of rhADAMTS-1, -4 and -5 corresponding to those identified by Western blotting.

Antibody 3170 detected three bands of ≈ 81 kDa, 76 kDa and 55 kDa in our partially purified rhADAMTS-4 (Fig. 1, lane 4). The 81- and 76-kDa bands are likely to represent different forms of mature ADAMTS-4 (theoretical molecular mass 68.3 kDa). The 75- and 55-kDa forms of rhADAMTS-4 (the 55-kDa form shows higher aggrecanase activity) have been described as the mature and secondarily processed forms of the enzyme upon expression in a human chondrosarcoma cell line [39]. More recently Flannery et al. described two autocatalytic processing events for recombinantly expressed ADAMTS-4 [40]. A cleavage in the spacer region generated a 53-kDa form of the enzyme, in agreement with our data. We did not detect a 40-kDa form generated by cleavage in the cysteine-rich region of the enzyme. This could be due to the difference in expression systems used. Comparison of Western blots with the corresponding silver-stained gels (Fig. 1, lane 3, arrows) suggests three potential ADAMTS-4 bands.

Western blot analysis of partially purified rhADAMTS-5 with antibody 3235 revealed two main bands of ≈ 64 and 37 kDa. The 64-kDa band is likely to represent a C-terminally processed form of mature ADAMTS-5 as its theoretical molecular mass (not accounting for carbohydrate) is 73.7 kDa. ADAMTS-5 purified from cartilage-conditioned medium has revealed multiple bands in the range of 40–65 kDa [17]. In addition, similarly processed forms of the enzyme have been detected in synovium-conditioned medium from arthritis patients [31] and in a GnHCl extract from cartilage of an arthritic patient [23].

Aggrecanolytic activity of rhADAMTS enzymes

Inhibitors of cysteine and serine proteinases were added to conditioned culture medium to abolish activity of any such contaminating proteinases. We analysed the partially purified enzymes for matrixin activity by use of a quenched fluorescence substrate known to be cleaved by these enzymes [35]. No such activity was detected in any of the three rhADAMTS preparations (data not shown).

ADAMTS-1, -4 and -5 are unusual in that they have strict specificity for cleavage between glutamatic acid residues and uncharged aliphatic amino acids in the core proteins of large aggregating proteoglycans. This specificity is unique among mammalian proteinases. The reported cleavage sites in aggrecan generated by ‘aggrecanase’ activity are Glu373→Ala in the interglobular domain, Glu1480→Gly between the chondroitin sulfate 1 and 2 attachment regions and Glu1666→Gly, Glu1771→Ala, and Glu1871→Leu within the chondroitin sulfate 2 attachment region [13]. We digested purified aggrecan monomer with the rhADAMTS enzyme preparations and isolated the fragments for N-terminal sequence analysis as described in Experimental procedures. The digestions resulted in all cases in at least five aggrecan core protein fragments, as determined by staining with colloidal Coomassie blue (Fig. 2, lanes 2, 4 and 6) and Western blot analysis with antibody 5/6/3-B-3 (lanes 3, 5 and 7) [41]. The fragments generated by the three different rhADAMTS enzyme preparations were very similar. N-Terminal sequence analysis of these fragments (Table 2 and Fig. 3) showed that they resulted from typical aggrecanase cleavages in poorly glycosylated regions of the aggrecan core protein. These cleavage sites have been reported for both ADAMTS-4 and -5 [17,31,42]. We describe here for the first time the N-terminal sequences of the major aggrecan fragments generated by ADAMTS-1. This enzyme was initially reported to cleave aggrecan only at the C-terminus [19]. Recently however, the use of cleavage site-specific antibodies has demonstrated that ADAMTS-1 is capable of generating similar aggrecan fragments to those produced by ADAMTS-4 and -5 [24]. The finding in a different laboratory that ADAMTS-1 failed to cleave aggrecan is not in line with this larger body of evidence [43].

Figure 2.

SDS/PAGE and Western blotting of aggrecan core protein fragments generated by rhADAMTS-1, -4 and -5. rhADAMTS-1 (10 units) and rhADAMTS-4 and -5 (5 units) were incubated with 5 mg of aggrecan monomer followed by deglycosylation of the generated fragments with chondroitin ABC lyase and electrophoresis on 4–10% gradient polyacrylamide gels. Lane 1 is undigested aggrecan monomer treated with chondroitin ABC lyase, 95 kDa. Panels A, B and C represent aggrecan digests produced by rhADAMTS-1, -4 and -5, respectively. Lanes 3, 5, and 7 are Western blots of aggrecan core protein fragments detected with antibody 5/6/3-B-3; lanes 2, 4 and 6 are the Coomassie blue-stained gels of these fragments. See Table 2 for the N-terminal sequences of the main aggrecan fragments.

Table 2. N-terminal sequences of aggrecan core protein fragments generated by rhADAMTS′. Fragments were generated, separated and sequenced as described under Experimental procedures, and their positions on 4–10% polyacrylamide gels are shown in Fig. 2. Amino acid numbering is according to the published sequence of bovine aggrecan [66].
Molecular mass (kDa)N-terminal sequenceYield (pmol)
Figure 3.

Schematic representation of aggrecanase cleavage sites in the aggrecan core protein. G1, G2 and G3 represent the three globular domains of the aggrecan core protein; KS, CS1 and CS2 represent the keratan sulfate, chondroitin sulfate 1 and chondroitin sulfate 2 attachment regions, respectively.

TIMP-3 has been shown to be a potent inhibitor of ADAMTS-4 and -5 [44,45]. We therefore assayed our rhADAMTS-1, -4 and -5 preparations with and without 100 nm TIMP-3. This concentration of TIMP-3 resulted in almost complete inhibition of the activity of these three ADAMTS enzymes: 93 ± 0.2%, 100 ± 0.1% and 95 ± 1.7% for ADAMTS-1, -4 and -5, respectively.

In summary, the addition of type-specific covalent inactivators of serine and cysteine proteinases, the lack of cleavage of a quenched fluorescence substrate sensitive to hydrolysis by matrixins, the detection of aggrecan fragments generated exclusively by cleavage of glutamyl bonds and inhibited by TIMP-3, and the lack of aggrecanolytic activity in medium from mock-transfected cells, are consistent with the presence of recombinant aggrecanase activities in our enzyme preparations, with no contaminating proteolytic activities being detected.

Inhibition of ADAMTS activity by catechin gallate esters

Work in our laboratory has previously shown that catechin gallate esters, found in abundance in green tea effusions, inhibit cartilage aggrecan breakdown in an in vitro model [28]. We therefore analysed these compounds for inhibition of the aggrecan-degrading activity of ADAMTS-1, -4 and -5. The catechin gallate esters EGCG and ECG potently inhibited ADAMTS-1, -4 and -5 in a dose-dependent manner over a 2 nm to 2 µm concentration range, whereas catechins lacking the gallate moiety, EC and EGC, and the gallate group in isolation represented by PG, showed very little inhibition even at the highest concentration tested of 2 µm(Fig. 4A–C). The presence of both the catechin and the gallate ester moiety as separate molecules in the same assay each at 500 nm was also not sufficient to inhibit the ADAMTS activities (data not shown).

Figure 4.

Inhibition of rhADAMTS-1, -4 and -5 by catechins and gallates. rhADAMTS-1 (A), -4 (B), and -5 (C) were assayed using aggrecan-containing polyacrylamide particles in the absence and presence of catechins and gallates over the concentration range 2 nm to 2 µm. EC, ○; ECG, □; EGC, ▴; EGCG, ◆; PG, X. The lines represent linear regression analyses. All assays were performed at least twice.

IC50 values for inhibition by EGCG and ECG deduced from regression analyses of the dose–response curves gave approximate values of 100–150 nm for rhADAMTS-4 and -5, and 200–250 nm for rhADAMTS-1. The correlation coefficients for the regression analyses were all equal to or larger than 0.9. The inhibition observed with the catechin gallate esters was not due to a Zn2+-chelating effect since similar levels of inhibition were achieved when the enzymes were assayed in the presence of 100 µm ZnCl2 (data not shown). The inhibition was reversible, as removal of the catechin gallates by buffer exchange (dialysis or gel filtration) produced a reappearance of enzyme activity (data not shown).

Inhibition of collagenase and ADAM-10 activity by catechin gallate esters

Catechins and gallates were also analysed for inhibitory potential against collagenase-1 (MMP-1) and collagenase-3 (MMP-13), as well as ADAM-10 as a representative of the ADAM group of metalloproteinases. Statistically significant inhibition of the collagenases by any of the catechins or gallates required a concentration of at least 20 µm. Even at the highest concentration tested of 50 µm, the maximum inhibition observed was still below 50% (EGCG; 29 ± 4% inhibition of collagenase-1 activity and 30 ± 9% inhibition of collagenase-3: ECG; 14 ± 5% inhibition of collagenase-1 and 20 ± 7% inhibition of collagenase-3). The inhibition of ADAM-10 was equally poor, less than 20% inhibition being achieved by any of the catechins or gallates at the highest concentration of 50 µm. As a positive control, the hydroxamate inhibitor BB-94, a broad-spectrum matrixin and ADAM inhibitor, completely inhibited collagenase-1 and -3 and ADAM-10 at 10 µm concentration.


No aggrecanolytic activity was detected in mock-transfected insect cell-conditioned medium. The expression of human ADAMTS-1, -4 and -5 using a constitutive system in insect cells produced relatively low amounts of recombinant protein. Partial purification led to increases in specific activity of between 190- and 370-fold, but SDS gel electrophoresis still demonstrated a number of contaminating proteins. Despite the lack of purity of our enzyme preparations, we were unable to detect any contaminating proteolytic activity. Possible serine and cysteine proteinase activity was abolished by the addition of enzyme inactivators, and no matrixin-like activity was detected by use of a broad-spectrum quenched fluorescence substrate. In addition, only fragments of aggrecan produced by the action of glutamyl endopeptidase activity were found following hydrolysis of the aggrecan core protein.

ADAMTS-1, -4 and -5 are thought to be the proteinases responsible for the breakdown of cartilage aggrecan, which is one of the events leading to joint failure in the arthritic diseases [46]. As such, they are candidate targets for novel therapeutic intervention strategies. The use of broad-spectrum matrixin inhibitors in clinical trials has so far proved unsuccessful due to unacceptable side-effects [47–49], and it is to be expected that more selective proteinase inhibitors will be required for successful chondroprotective intervention. A recent report has described a series of inhibitors that show good aggrecanase vs. matrixin selectivity, but no information for inhibition of ADAMs was given [50]. We present here the surprising finding that catechin gallate esters, abundant components of green tea effusions, provide selective inhibition of aggrecanases, even when compared to phylogenetically related proteinases such as an ADAM and two collagenases, with a difference in potency of approximately two orders of magnitude. The poor inhibition by catechins lacking the gallate ester group (EC and EGC), or by the gallate group in isolation (PG) and the fact that both modules as separate molecules in the same assay did not inhibit the rhADAMTS enzymes indicates a co-operative effect between the catechin and gallate moieties. We have previously demonstrated the inhibition of cartilage aggrecan breakdown by catechin gallate esters in a tissue culture model [28], and the data presented in this paper suggest that this is due to a direct inhibitory effect of these compounds on the activity of ADAMTS-1, -4 and -5. We have not attempted to define the mechanism of inhibition other than to demonstrate that inhibition was reversible and was not due to Zn sequestration. We hypothesize that these relatively small compounds are competing with substrate for the active site of the aggrecanases. However, other possibilities exist, such as allosteric inhibition, or even direct binding to substrate.

We have previously reported the inhibition by catechin gallates of type II collagen breakdown in an in vitro model of cartilage breakdown [28]. Our finding that two collagenases implicated in cartilage collagen hydrolysis, collagenase-1 and collagenase-3 [51–54] were not potently inhibited by EGCG and ECG, is evidence that this is not via direct inhibition of these collagen-degrading enzymes. Other mechanisms are possible, including effects on proinflammatory signalling pathways. For instance, it has been reported that EGCG inhibits the chymotrypsin-like activity of the proteasome, with IC50 values in the range 86–194 nm, and subsequent accumulation of IκB-α[55]. This would result in inhibition of NF-κB activation and in downstream inhibition of NF-κB-regulated genes such as the matrixins [56,57].

The catechin gallate esters are bioavailable following the consumption of green tea, with reported plasma concentrations in the range 0.1–5 µm[58], and a half-life of a few hours [59,60]. It is therefore possible that the drinking of green tea will have a prophylactic effect on cartilage integrity. Indeed, it has been reported that mice fed on a polyphenolic fraction of green tea have reduced signs of collagen-induced arthritis [61]. Alternatively these compounds could serve as lead compounds in the design of more potent inhibitors that will halt cartilage breakdown.

There are many reports in the literature of beneficial effects of green tea consumption, some of which relate to pathologies in which turnover of extracellular matrix proteins is a major component, such as stroke and cerebral haemorrhage [62] and cancer [63,64]. The anticancer effects of polyphenolic compounds from green tea have been attributed, at least in part, to their direct inhibition of matrixins such as the gelatinases [4–6]. However, the reported IC50 values for inhibition of these proteinases of about 20 µm, similar to our findings reported here for two collagenases and ADAM-10, are beyond the concentration attainable following green tea consumption. It is plausible that at least some of the beneficial effects are provided instead by direct inhibition of ADAMTS enzymes, some of which have been implicated in cancer [65], perhaps in combination with down-regulation of other proteinases at the mRNA level.


We wish to thank the Arthritis Research Campaign, UK for funding this research. This work was also supported by the Wellcome Trust UK, Australian National Health and Research Council and the Arthritis Foundation of Australia.