Review The regulation of the ADAMTS4 and ADAMTS5 in : a review J. Bondeson1, S. Wainwright2, C. Hughes2, B. Caterson2

1Department of Rheumatology, Cardiff ABSTRACT on monomers, with the asso- University and 2Connective Tissue Destruction of articular is a ciated water molecules, is essential for Biology Laboratories, Cardiff School of key feature of a number of arthritides, the ability of articular cartilage to with- Biosciences, Cardiff University, Cardiff, osteoarthritis prominent among them. stand compressive deformation during UK. Aggrecan degradation, caused by joint articulation (1). The chondrocytes Jan Bondeson, MD PhD; increased activity of proteolytic en- synthetize and catabolize ECM mac- Shane Wainwright, PhD; Clare Hughes, PhD; zymes that degrade macromolecules romolecules, which in turn serve to Bruce Caterson PhD. in the cartilage , maintain the homeostasis of the cel- This work has been supported by the is followed by irreversible collagen lular environment and the cartilage Arthritis Research Campaign (UK), degradation. The degradation of ag- structure. In diseases like rheumatoid grants no. W0596, 13172 and 14570. grecan is mediated by various matrix arthritis (RA) and osteoarthritis (OA), Please address correspondence to: proteinases, mainly the aggrecanases, degradation of ECM macromolecules Prof. Bruce Caterson, Connective Tissue multidomain be- exceeds their synthesis, resulting in a Biology Laboratories, Cardiff School of longing to the ADAMTS (a net decrease in the amount of cartilage Biosciences, Cardiff University, Museum and with throm- matrix, eventually leading to total or Avenue, Cardiff CF10 3US, UK. bospondin motifs) family. There has partial erosion of the cartilage. The de- E-mail: [email protected] been much interest in the possible role pletion of aggrecan from articular car- Reprints will not be available from the of these aggrecanases, mainly AD- tilage, as evidenced by the release of authors. AMTS4 and ADAMTS5, as therapeutic aggrecan catabolites into the synovial Received on June 25, 2007; accepted in targets in osteoarthritis. There is still fl uid, is an essential early pathophysi- revised form on November 30, 2007. debate which of them is the major ag- ological event in OA (1-2). The aspect Clin Exp Rheumatol 2008; 26: 139-145. grecanase in osteoarthritis, however, most studied has been the of © CopyrightCopyright CLINICAL AND as well as major issues concerning the interglobular domain of aggrecan EXPERIMENTAL RHEUMATOLOGY 2007.2007. how they are regulated, with possible with release of the glycosaminoglycan discrepancies between murine models (GAG)-attachment regions, since it ap- Key words: , and results obtained using human oste- pears to be the most destructive to tissue interleukin-1, , oarthritis tissue. This review discusses function. NF-kappaB, osteoarthritis, tumour some recent data regarding the regula- It has been debated whether the prote- necrosis factor alpha. tion of ADAMTS4 and ADAMTS5 olysis of the interglobular domain of expression in osteoarthritis, with em- aggrecan is mediated by matrix met- phasis on the role of proinfl ammatory alloproteases (MMPs) or by aggreca- cytokines in driving these , and nases. The aggrecanases are members of the transcription factor NFκB i inn of the family of disintegrin and metal- mediating their expression. loproteases with mo- tifs (ADAMTS) that were fi rst char- Introduction acterised for their ability to cleave the Cartilage consists of a relatively small Glu373-Ala374 bond in the interglobular amount of chondrocytes, embedded in domain of aggrecan (1). These enzymes abundant extracellular matrix (ECM) are regulated at multiple levels through that contains numerous macromol- control of , mRNA ecules, major constituents being col- splicing and processing, as lagen fi brils and the large aggregating well as regulation of the expression of aggrecan. Aggrecan fi lls various naturally occurring inhibitors the interstices of the collagen meshwork (3). Several ADAMTS aggrecanases by forming large aggregated complexes have been identifi ed, among them ag- interacting with hyaluronan and link grecanase-1 (ADAMTS4) and aggre- . The high negative charge den- canase-2 (ADAMTS5). Recent studies Competing interests: none declared. sity of the glycosaminoglycan chains suggest that one or more aggrecanases

139 REVIEW Regulation of aggrecanases in osteoarthritis / J. Bondeson et al. are responsible for cleavage of the interglobular domain with destructive loss of GAG (4-5). ADAMTS4 and ADAMTS5 are multi-domain metallo- secreted from the cell into the extracellular space as furin active pro- teases. They both consist of a catalytic metalloprotease domain and a series of other ancillary domains that play a role in regulating their activity and substrate specifi city (Fig. 1). For example, AD- AMTS4 activity for the interglobular domain cleavage site is increased when the C-terminal spacer domain is re- moved, and additional C-terminal trun- cation also leads to a preferred substrate specifi city for small leucine-rich prote- oglycans and other proteins (6). Another recent study suggests that blocking ag- grecanase cleavage in the interglobular domain of aggrecan diminished aggre- can loss and cartilage erosion in murine models of surgically induced osteoar- Fig. 1. The structure of ADAMTS4, its only known splice variant, and ADAMTS5. thritis and infl ammatory arthritis, and appeared to stimulate repair following AMTS4 gene expression could be up- We used a model to effectively and acute infl ammation (7). regulated through treatment with either specifi cally neutralise the endogenous It has long been debated which of the IL-1β, TNF-α or oncostatin M, there production of these cytokines from is the main aggrecanase in was little effect on ADAMTS5 in ei- the OA synovial macrophages (18). OA. Due to observations of ADAMTS4 ther human chondrocytes or cultured Cultures were either left untreated, mRNA being inducible through in- human cartilage explants. In contrast, incubated with the p75 TNF soluble terleukin (IL)-1 in chondrocytes, this there was an additive effect of combi- receptor Ig fusion protein etanercept has attracted a good deal of at- nation treatment with oncostatin M and (Enbrel), incubated with a neutraliz- tention (8-10). But in models of murine either IL-1β or TNF-α in these sys- ing anti-IL-1β antibody, or incubated OA induced by antigen or surgical joint tems, leading to marked induction of with a combination of Enbrel and anti- destabilisation, ADAMTS4-null mice ADAMTS4 gene expression and also IL-1β. There was no effect of either did not show any protective effect on some induction of ADAMTS5 (15). In Enbrel or the neutralizing anti-IL-1β cartilage aggrecan loss compared with OA synovium or cartilage, aggrecanase antibody on ADAMTS5 expression, wild-type mice, whereas there was a activity and expression of ADAMTS4 nor was it at all affected by a combina- marked protective effect in ADAMTS5- and ADAMTS5 is present constitu- tion of these treatments (Fig. 2). Thus null mice (11-13). These studies sug- tively, without any requirement for any ADAMTS5 appears to be constitutive gest that in this model of murine degen- catabolic stimulation (15-16). in OA synovial cells. In contrast, AD- erative arthritis, ADAMTS5 plays a key To investigate the role of TNF-α, and AMTS4 was signifi cantly (p < 0.05) in- role in aggrecan degradation. The sig- IL-1 in driving ADAMTS4 and AD- hibited by Enbrel, and more potently (p nifi cance of this important fi nding for AMTS5 expression in the human OA < 0.01) inhibited by a combination of idiopathic human OA is not yet known. synovium, we used a model of cultures Enbrel and the neutralizing anti-IL-1β of synovial cells from digested OA syn- antibody (Fig. 2). This would indicate The role of proinfl ammatory ovium (17). These cells have the advan- that in the human OA synovium, the cytokines in the regulation of tage of spontaneously producing a vari- upregulation of ADAMTS4 is depend- ADAMTS4 and ADAMTS5 ety of both pro- and anti-infl ammatory ent on TNF-α and IL-1 produced by In models of cultured bovine and cytokines, including TNF-α, IL-1 and the synovial macrophages, whereas the porcine chondrocytes or cartilage IL-10, as well as the major MMPs and level of ADAMTS5 is not changed by explants, ADAMTS4 is induced fol- TIMPs. By means of specifi c neutrali- these cytokines (18). lowing stimulation with IL-1, tumour zation of macrophage-produced TNF-α In contrast to this wealth of data from necrosis factor (TNF)α, oncostatin M and IL-1, it was possible to assess the human, porcine and bovine models or transforming growth factor β, but contribution of these two proinfl amma- indicating that ADAMTS4 mRNA re- ADAMTS5 is not (8-10,14). A recent tory cytokines on ADAMTS4 and AD- sponds to IL-1, there are two recent study (15) indicated that although AD- AMTS5 gene expression. papers indicating that this is not the

140 Regulation of aggrecanases in osteoarthritis / J. Bondeson et al. REVIEW

A ADAMTS5 gene expression has also ADAMTS4 become clearer. ADAMTS4, but not ADAMTS5, has several NFκB binding sites on its 5’ fl anking region that are ADAMTS5 conserved between species (22). In bo- Patient 1 vine chondrocytes, ADAMTS4, but not GAPDH ADAMTS5, could be upregulated by IL- 1 stimulation (Fig. 3A). Using a model of transfecting bovine chondrocytes with the 5’ fl anking region of the AD- ADAMTS4 AMTS4 or ADAMTS5 gene luciferase receptor vector, it was observed that the IL-1-induced upregulation of the AD- Patient 2 ADAMTS5 AMTS4 gene involved its 5’ fl anking GAPDH region, whereas the 5’ fl anking region of the ADAMTS5 gene played no part (Fig. 3B). Mutation of any one of the three identifi ed NFκB binding sites re- sulted in the loss of the IL-1 response to the ADAMTS4 gene luciferase reporter B vector (Fig. 3C), indicating that the IL- 1-stimulated increase in ADAMTS4 gene transcription depends on two or more NFκB binding sites located in the 5’ fl anking region of this gene (22). In bovine nucleus pulposus tissue, TNF-α treatment induced upregulation of aggrecanase activity, ADAMTS4 in particular, in a NFκB dependent man- ner, although the specifi city of the small molecule NFκB inhibitor used in this study remains unproven (23). In human OA synovial fi broblasts, treatment with IL-1 or TNF-α, but not treatment with phorbol ester, resulted in upregulation of ADAMTS4, whereas ADAMTS5 was unaffected (Fig. 4A). In this model, it was possible to use adenoviral gene transfer of the endogenous inhibitor IκBα to specifi - Fig. 2. Effect of neutralization of TNF-α and/or IL-1 on the expression of ADAMTS4 and ADAMTS5 in the OA synovium. cally inhibit NFκB without affecting other The cells were either left untreated, incubated with a neutralizing anti-IL-1β antibody, incubated with signalling pathways or causing apoptosis the p75 TNF soluble receptor Ig fusion protein etanercept (Enbrel), or incubated with a combination (24). Whereas ADAMTS5 gene expres- of etanercept and anti-IL-1β. After incubation for 48 h the cells were washed with PBS and the RNA extracted using Tri-reagent for RT-PCR analysis using oligonucleotide primers specifi c for ADAMTS4 sion was not changed by gene transfer and ADAMTS5 in two patients, with analyis of GAPDH used for comparison of gene expression (A). of IκBα, the ADAMTS4 induction by In (B), the PCR data is expressed as percentage of the gene expression in untreated cells, as standard- IL-1 or TNF-α was potently inhibited by ised for GAPDH, with the SEM given (n = 4-5). NFκB downregulation (Fig. 4). These three papers (22-24) strongly case in mouse cells. In murine femo- The role of NFκB in the regulation suggest that the upregulation of AD- ral head explant cultures, ADAMTS4 of ADAMTS4 and ADAMTS5 AMTS4 induced by IL-1 or TNF-α is mRNA levels were unaffected by IL-1 It has been shown that both in RA and in NFκB dependent. There is no evidence (19). Monocytes from wild-type mice, OA, the transcription factor NFκB plays that NFκB plays any part in regulating but not monocytes from IL-1 defi cient an important part in regulating various ADAMTS5 expression, however. mice, upregulated ADAMTS5 mRNA proinfl ammatory and destructive media- in chondrocytes without affecting AD- tors, including several matrix metallo- An ADAMTS4 splice variant in AMTS4, again suggesting that murine proteases, i.e. MMP-1, 3, 9 and 13 (17, human OA synovium ADAMTS4 is unresponsive to IL-1 21). Recently, the role of this transcrip- Post-transcriptional regulation through (20). tion factor in regulating ADAMTS4 and alternative splicing has been recognised

141 REVIEW Regulation of aggrecanases in osteoarthritis / J. Bondeson et al.

A

GAPDH

ADAMTS4

ADAMTS5

Fig. 3. IL-1 upregulates ADAMTS4, but not ADAMTS5, in an NFκB dependent manner. Bovine chondrocytes were either left untreated or stimulated with IL-1α (10 ng/ml) before RNA was extracted and RT-PCR analysis carried out using oligonucleotide primers specifi c for bovine GAPDH, ADAMTS4 and ADAMTS5 (A). In (B), a luciferase assay was performed using pGL3 basic plasmids containing the 5’ fl anking region of the ADAMTS4 gene (pGL(TS4) transfected into bovine chondrocytes treated with the solvent DMSO, DMSO and IL-1α (10 ng/ml), or IL-1α and various concentrations of the NFκB inhibitor weldolactone. In (C), a luciferase assay was performed using either the wild-type 5’ fl anking region of the ADAMTS4 gene (pGL (TS4)), a construct in which this region had all four kB sites inactivated by mutation (pGL(TS4kB 1, 2, 3, 4) or a series of constructs in which each member had a different kB site unmutated. for several of the ADAMTS proteins, in- this alternative splice variant produces production of a low molecular weight cluding ADAMTS-6, -7 and -9 (3, 25). an ADAMTS4 protein that is secreted (37 kD) isoform of ADAMTS4, which We recently described the fi rst known in an active form that can cleave the was capable of degrading exogenous ag- splice variant of ADAMTS4 (Fig. 1) Glu373-Ala374 bond in the interglob- grecan at the interglobular domain site, in human OA synovium, using an oli- ular domain of aggrecan (6, 28). This and required de novo protein synthesis gonucleotide primer pair designed to splice variant has hitherto only been for its generation. In porcine chondro- amplify across the exon 8/9 junction of detected in human OA synovium, and cyte-agarose cultures, this 37kD AD- human ADAMTS4 (26). This alterna- not in other human tissues like , AMTS4 appears to be the major tively spliced transcript of ADAMTS4 cervix or ; nor has it been detect- responsible for IL-1-induced interglobu- is missing 161 base pairs from the 5’ ed in normal bovine synovium (26). It lar domain aggrenanase activity. Due to end of exon 9. The protein produced may be speculated that the release of its low molecular weight, it is obviously would lack the spacer domain and have low levels of this fully active variant of not the same protein as the high-molec- a different C-terminus lacking homolo- ADAMTS4 may be a factor in the slow ular weight ADAMTS4 splice variant gies with the normal human ADAMTS4 progress of superfi cial zone aggrecan described above, but it might represent C-terminal spacer domain (Fig. 1). loss in OA. another yet uncathergorised splice vari- This protein would lose functions de- Another recent study has demonstrated ant or protein cleavage product. pendent on its spacer domain, like sub- the presence of several C-terminally strate and matrix binding and inhibition truncated isoforms of ADAMTS4 and Which is the main aggrecanase in through fi bronectin (27). It is known ADAMTS5 in porcine articular carti- osteoarthritis? that removal of the spacer domain lage explants and porcine chondrocyte- There is good evidence that in OA syn- from ADAMTS4 increases its ability to agarose cultures exposed to IL-1 (29). ovium and cartilage, ADAMTS4 is the cleave aggrecan, and it may well be that In particular, IL-1 treatment induced aggrecanase induced by proinfl amma-

142 Regulation of aggrecanases in osteoarthritis / J. Bondeson et al. REVIEW

of a murine model for the study of hu- man aggrecanolysis. Studies on the effect of ADAMTS5 defi ciency on aggrecanolysis in mouse cartilage have provided contradictory results: in one study, spontaneous ag- grecan degradation was also ablated in murine epiphyseal chondrocyte cul- tures from ADAMTS5-null mice as compared to cultures from wild-type mice (34), in another, ADAMTS5 defi - ciency did not block aggrecanolysis at preferred cleavage points (19). Another recent study used a small in- terfering RNA approach to assess the effect of the inhibition of ADAMTS4 and ADAMTS5 in human chondro- cytes and cartilage explants. Sup- pression of either ADAMTS4 or AD- AMTS5 led to siginifi cant inhibition of the degradation of aggrecan induced by a combination of TNF-α and oncosta- tin M (15). Studies of the mRNA levels of ADAMTS4 and ADAMTS5 in nor- mal and osteoarthritic human cartilage have as yet failed to provide consistent results. In some studies, ADAMTS5 expression is higher than that of AD- AMTS4, in others the opposite is true (9, 15, 35-36). In a study of the effect of salt concentration on aggrecanoly- sis, recombinant ADAMTS4 and AD- AMTS5 had similar general proteolyt- ic effects, but the aggrecanase effect of Fig. 4. The role of NFκB in the regulation of ADAMTS4 and ADAMTS5 expression in the OA ADAMTS5 was markedly higher (37). synovium. Importantly, a reduction of aggrecan OA synovial fi broblasts were either left uninfected or infected with 30:1 of either AdvIκBα or Adv0. breakdown was observed after the sup- After 24 hrs, cells were stimulated with either PMA, IL-1β or TNF-α. RT-PCR analysis was carried out using oligonucleotide primers specifi c for ADAMTS4 and ADAMTS5, with analysis of GAPDH pression of either ADAMTS4 or AD- gene expression used for comparison (A). In (B), the PCR data is expressed as percentage of the gene AMTS5 through a small interfering expression in untreated cells, as standardised for GAPDH, with the SEM given (n = 3-5). RNA approach in unstimulated human OA cartilage. This would indicate that, tory cytokines, whereas ADAMTS5 ap- is the pathologically induced aggre- in contrast to the situation in genetical- pears to be constitutive (9-10, 18, 24). canase. Mice lacking ADAMTS4 de- ly modifi ed mice, both ADAMTS4 and If it is accepted that OA is a cytokine velop normally and develop surgically ADAMTS5 contribute to the structural driven disease, as indicated by some induced OA in a similar manner to damage in human OA (15). recent papers suggesting that macro- wild-type mice, but deletion of AD- phage-produced IL-1 and TNF plays a AMTS5 protects mice from develop- Aggrecanases as potential role in driving destructive responses in ing OA (11-13). These results suggest therapeutic targets in OA OA, this fi nding would render it like- that at least in murine models of OA, The recent data presented in this re- ly that ADAMTS4 is the aggrecanase ADAMTS5 is the major aggrecanase. view suggests that although both AD- responsible for aggrecanolysis in OA The only caveat to this conclusion is AMTS4 and ADAMTS5 cleave aggre- (18, 30-33). the apparent discrepancy between hu- can, they are two very different enzymes In marked contrast to this in vitro data man and murine cells with regard to with regard to their regulation. At least derived from studying human tissue, the regulation of ADAMTS4 (19-20). in human cells, ADAMTS4 responds recent studies using transgenic mice If the human, but not murine, AD- to IL-1 and TNF-α, but ADAMTS5 suggest that in murine models of de- AMTS4 gene responds to IL-1 stimu- does not (9-10, 18, 24). Another differ- generative joint disease, ADAMTS5 lation, this brings into question the use ence is that whereas the upregulation of

143 REVIEW Regulation of aggrecanases in osteoarthritis / J. Bondeson et al.

ADAMTS4 depends on the transcrip- erosion and promotes cartilage repair. J Clin nisms in rheumatoid synovium, but spares tion factor NFκB, ADAMTS5 is NFκB Invest 22007;007; 1117:17: 1627-36.1627-36. anti-infl ammatory mediators. Proc Natl Acad 8. TORTORELLA MD, MALFAIT AM, DECCICO Sci USA 1999; 96: 5668-73. independent and lacks kB elements on C, ARNER E: The role of ADAM-TS4 (aggre- 22. WAINWRIGHT SD, HEMING M, HUGHES CE: its promoter (22-24). With this in mind, canase-1) and ADAM-TS5 (aggrecanase-2) Evidence suggesting a role for NFκB in the it is interesting to note that treatment of in a model of cartilage degradation. Osteo- IL-1-induced transcriptional regulation of bovine cartilage explants with a small arthritis Cartilage 2001; 9: 539-52. the ADAMTS4 gene. Manuscript submitted 9. BAU B, GEBHARD PM, HAAG J, KNORR for publication. molecule IκB kinase inhibitor led to T, BARTNIK E, AIGNER T: Relative mes- 23. SEGUIN CA, BOJARSKI M, PILLIAR RM, prevention of IL-1-induced aggrecan senger RNA expression profi ling of colla- ROUGHLEY PJ, KANDEL RA: Differential depletion, suggesting that this process genases and aggrecanases in human articular regulation of matrix degrading enzymes in κ chondrocytes in vivo and in vitro. Arthritis TNF-alpha-induced model of nucleus pulpo- occurred in a NF B dependent man- Rheum 2002; 46: 2648-57. sus tissue degradation. Matrix Biol 22006;006; 225:5: ner (38). This differential regulation of 10. PRATTA MA, SCHERLE PA, YANG G, LIU RQ, 409-18. ADAMTS4 and ADAMTS5 has impli- NEWTON RC: Induction of aggrecanase- 24. BONDESON J, LAUDER S, WAINWRIGHT SD cations for the potential development 1 (ADAM-TS4) by interleukin-1 occurs et al.: Adenoviral gene transfer of the endog- through activation of constitutively produced enous inhibitor IkappaBalpha into human of disease-modifying osteoarthritis protein. Arthritis Rheum 2003; 48: 119-33. osteoarthritis synovial fi broblasts demon- drugs (39). A therapeutic strategy that 11. GLASSON SS, ASKEW R, SHEPPARD B et al.: strates that several matrix metalloproteinases would inhibit the cytokine-driven in- Characterization of and osteoarthritis sus- and aggrecanases are NFkappaB dependent. fl ammatory response would be likely ceptibility in ADAMTS-4-knockout mice. J Rheumatol 22007;007; 334:4: 5523-33.23-33. Arthritis Rheum 2004; 50: 2547-58. 25. BEVITT DJ, LI Z, LINDROP JL, BARKER MD, to downregulate ADAMTS4, as would 12. GLASSON SS, ASKEW R, SHEPPARD B et al.: CLARKE MP, MCKIE N: Analysis of full an inhibitor of NFκB. However, neither Deletion of active ADAMTS5 prevents carti- length ADAMTS6 transcript reveals alterna- strategy would be likely to infl uence lage degradation in a murine model of oste- tive splicing and a role for the 5’ untranslated ADAMTS5. oarthritis. Nature 2005; 434: 644-8. region in translational control. Gene 2005; 13. STANTON H, ROGERSON FM, EAST CJ et 359: 99-110. The design of small molecule aggreca- al.: ADAMTS5 is the major aggrecanase in 26. WAINWRIGHT SD, BONDESON J, HUGHES nase inhibitors is an area of considerable mouse cartilage in vivo and in vitro. Nature CE: An alternatively spliced transcript of AD- interest for the pharmaceutical industry 2005; 434: 648-52. AMTS4 is present in human synovium from (40-44). For such approaches to meet 14. YAMANISHI Y, BOYLE DL, CLARK M et al.: osteoarthritis patients. Matrix Biol 22006;006; 225:5: Expression and regulation of aggrecanase 317-20. with success, there is a need to appreci- in arthritis: the role of TGF-beta. J Immunol 27. HASHIMOTO G, SHIMODA M, OKADA Y: ate that ADAMTS4 and ADAMTS5 are 2002; 168: 1405-12. ADAMTS4 (aggrecanase-1) interaction with differentially regulated. The primary ag- 15. SONG R-H, TORTORELLA MD, MALFAIT A- the C-terminal domain of fi bronectin inhibits M et al.: Aggrecan degradation in human proteolysis of aggrecan. J Biol Chem 2004; grecanase (ADAMTS4 or ADAMTS5) articular cartilage explants is mediated by 279: 32483-91. involved in human OA also needs to be both ADAMTS4 and ADAMTS5. Arthritis 28. GAO G, PLAAS A, THOMPSON VP, JIN S, ZUO conclusively identifi ed. Rheum 2007; 56: 575-85. F, SANDY JD: ADAMTS4 (aggrecanase-1) 16. ILIC MZ, VANKEMMELBEKE MN, HOLEN I, activation on the cell surface involves C- BUTTLE DJ, CLEM ROBINSON H, HANDLEY terminal cleavage by glycosylphosphatidyl References CJ: Bovine joint capsule and fi broblasts de- inositol-anchored membrane type 4-matrix 1. CATERSON B, FLANNERY CR, HUGHES CE, rived from joint capsule express aggrecanase metalloproteinase and binding of the acti- LITTLE CB: Mechanisms involved in carti- activity. Matrix Biol 22000;000; 119:9: 2257-65.57-65. vated proteinase to and lage proteoglycan catabolism. Matrix Biol 17. AMOS N, LAUDER S, EVANS A, FELDMANN heparan sulfate on syndecan-1. J Biol Chem 2000; 19: 333-44. M, BONDESON J: Adenoviral gene transfer 2004; 279: 10042-51. 2. NAGASE H, KASHIWAGI M: Aggrecanases into osteoarthritis synovial cells using the 29. POWELL AJ, LITTLE CB, HUGHES CE: Low and cartilage matrix degradation. Arthritis endogenous inhibitor IκBα reveals that most, molecular weight isoforms of the aggreca- Res Ther 2003; 5: 94-103. but not all, inhibitory and destructive me- nases are responsible for the cytokine-in- 3. JONES GC, RILEY GP: ADAMTS proteinases: diators, are NFκB dependent. Rheumatology duced proteolysis of aggrecan in a porcine a multi-domain, multi-functional family with 2006; 45: 1201-9. chondrocyte culture system. Arthritis Rheum roles in extracellular matrix turnover and ar- 18. BONDESON J, WAINWRIGHT SD, LAUDER 2007; 56: 3010-9. thritis. Arthritis Res Ther 2005; 7: 160-9. S, AMOS N, HUGHES CE: The role of syno- 30. PELLETIER J-P, MARTEL-PELLETIER J, 4. STRUGLICZ A, LARSSON S, PRATTA M, vial macrophages and macrophage-produced ABRAMSON SB: Osteoarthritis: an infl amma- KUMAR S, LARK M, LOHMANDER S: Human cytokines in driving aggrecanases, matrix tory disease. Arthritis Rheum 2001; 44: 1237- osteoarthritis synovial fl uid and joint carti- metalloproteinases and other destructive and 47. lage contain both aggrecanase and matrix infl ammatory responses in osteoarthritis. 31. BLOM AB, VAN LENT PLEM, HOLTHUYSEN metalloproteinase generated aggrecan frag- Arthritis Res Ther 2006; 8: R187. AEM et al.: Synovial lining macrophages ments. Osteoarthritis Cartilage 2006; 14: 19. EAST CJ, STANTON H, GOLUB SB, ROGER- mediate osteophyte formation during experi- 101-13. SON FM, FOSANG AJ: ADAMTS-5 defi cien- mental osteoarthritis. Osteoarthritis Carti- 5. SANDY JD: A contentious issue fi nds some cy does not block aggrecanolysis at preferred lage 2004; 12: 627-35. clarity: on the independent and comple- cleavage sites in the chondroitin sulfate rich 32. VAN LENT PLEM, BLOM AB, VAN DER KRAAN mentary roles of aggrecanase activity and region of aggrecan. J Biol Chem 2007; 282: P et al.: Cricial role of synovial lining mac- MMP activity in human joint aggrecanolysis. 8632-40. rophages in the promotion of transforming Osteoarthritis Cartilage 2006; 14: 95-100. 20. ZWERINA J, REDLICH K, POLZER K et al.: growth factor beta-mediated osteophyte for- 6. KASHIWAGI M, ENGHILD JJ, GENDRON C TNF-induced structural joint damage is mation. Arthritis Rheum 2004; 50: 103-11. et al.: Altered proteolytic activities of AD- mediated by IL-1. Proc Natl Acad Sci USA 33. BENITO MJ, VEALE DJ, FITZGERALD O, VAN AMTS-4 expressed by C-terminal process- 2007; 104: 11742-7. DEN BERG WB, BRESNIHAN B: Synovial tis- ing. J Biol Chem 2004; 279: 10109-19. 21. BONDESON J, FOXWELL BMJ, BRENNAN FM, sue infl ammation in early and late osteoar- 7. LITTLE CB, MEEKER CT, GOLUB SB et al.: FELDMANN M: A new approach to defi ning thritis. Ann Rheum Dis 2005; 64: 1263-7. Blocking aggrecanase cleavage in the aggre- therapeutic targets: blocking NF-κB inhibits 34. STEWART MC, FOSANG AJ, BAI Y, OSBORN B, can interglobular domain abrogates cartilage both infl ammatory and destructive mecha- PLAAS A, SANDY JD: ADAMTS5-mediated

144 Regulation of aggrecanases in osteoarthritis / J. Bondeson et al. REVIEW

aggrecanolysis in murine epiphyseal chondro- BURKE JR: Collagen and aggrecan degrada- ADAMTS-4 inhibitors. Matrix Biol 2 2006;006; cyte cultures. Osteoarthritis Cartilage 2006; tion is blocked in interleukin-1-treated carti- 25: 261-7. 14: 392-402. lage explants by an inhibitor of IkappaB ki- 42. BURSAVICH MG, GILBERT AM, LOMBARDI 35. WACHSMUTH L, BAU B, FAN Z, PECHT A, nase through suppression of metalloprotein- S et al.: Synthesis and evaluation of aryl thi- GERWIN N, AIGNER T: ADAMTS-1, a gene ase expression. J Pharmacol Exp Ther 2005; oxothiazolidinone inhibitors of ADAMTS-5 product of articular chondrocytes in vivo 315: 382-8. (Aggrecanase-2). Bioorg Med Chem Lett and in vitro, is downregulated by interleukin 39. PELLETIER J-P, MARTEL-PELLETIER J, RAY- 2007; 17: 1185-8. 1beta. J Rheumatol 2004;2004; 331:1: 3315-20.15-20. NAULD J-P: Most recent developments in 43. GILBERT AM, BURSAVICH MG, LOMBARDI 36. KEVORKIAN L, YOUNG DA, DARRAH C et strategies to reduce the progression of struc- S et al.: 5-((1H-Pyrazol-4-yl)methylene)-2- al.: Expression profi ling of metalloprotein- tural changes in osteoarthritis: today and to- thioxothiazolidin-4-one inhibitors of AD- ases and their inhibitors in cartilage. Arthritis morrow. Arthritis Res Ther 2006; 8: 206-20. AMTS-5. Bioorg Med Chem Lett 2007; 17: Rheum 2004; 50: 131-41. 40. CHERNEY RJ, MO R, MEYER DT et al.: Potent 1189-92. 37. GENDRON C, KASHIWAGI M, LIM NH et al.: and selective aggrecanase inhibitors contain- 44. WITTWER AJ, HILLS RL, KEITH RH et al.: Proteolytic activities of human ADAMTS- ing cyclic P1 substituents. Bioorg Med Chem Substrate-dependent inhibition kinetics of an 5. Comparative studies with ADAMTS-4. Lett 2003;2003; 113:3: 11297-300.297-300. -directed inhibitor of ADAMTS- J Biol Chem 2007; 282: 18294-306. 41. THOMAS M, SABATINI M, BENSAUDE F et 4 (Aggrecanase 1). Biochemistry 2007; 46: 38. PATTOLI MA, MACMASTER JF, GREGOR KR, al.: A microplate assay for the screening of 6393-401.

145