knockout of adamts5 does not eliminate cartilage aggrecanase activity but abrogates joint fibrosis...

Knockout of ADAMTS5 Does Not Eliminate Cartilage AggrecanaseActivity but Abrogates Joint Fibrosis and Promotes Cartilage AggrecanDeposition in Murine Osteoarthritis Models

Jun Li,1 Wendy Anemaet,1 Michael A. Diaz,2 Samuel Buchanan,1 Micky Tortorella,3 Anne Marie Malfait,1,2 Katalin Mikecz,4

John D. Sandy,1 and Anna Plaas1,2

1Department of Biochemistry, Rush University Medical Center, Chicago, Illinois, 2Department of Internal Medicine (Rheumatology), Rush UniversityMedical Center, Chicago, Illinois, 3Guangzhou Institute for Biomedicine and Health, Guangzhou, China, 4Department of Immunology,Rush University Medical Center, Chicago, Illinois

Received 17 April 2010; accepted 24 June 2010

Published online 26 October 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.21215

ABSTRACT: To investigate the role of ADAMTS5 in murine osteoarthritis (OA), resulting from destabilization of the medial meniscus (DMMmodel) or from TGFb1 injection and enforced uphill treadmill running (TTR model). Wild-type (WT) and ADAMTS5�/�mice were subjectedto either DMM or TTR and joints were evaluated for meniscal damage, cartilage changes, and fibrotic ingrowths from the joint margins.Cartilage lesions were quantified on an 8-point scoring system. Cartilage chondroitin sulfate (CS) content was evaluated by SafraninOstaining and by quantitative electrophoresis (FACE). The abundance of aggrecan, versican, and specific aggrecanase-generated productswas determined by Western analysis. Joint changes were similar for WT mice taken through either the DMM or the TTR model. ADAMTS5ablation essentially eliminated cartilage erosion and fibrous overgrowth in both models. In the TTR model, ADAMTS5 ablation did noteliminate aggrecanase activity from the articular cartilage but blocked fibrosis and resulted in the accumulation of aggrecan in the articularcartilage. The cartilage protection provided by ADAMTS5 ablation in the mouse does not result from prevention of aggrecanase activity perse, but it appears to be due to a blockade of joint tissue fibrosis and a concomitant increase in cartilage aggrecan content. �2010 Orthopaedic

Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 29:516–522, 2011

Keywords: osteoarthritis; ADAMTS5; aggrecan; versican; transforming growth factor beta 1; chondrogenesis; fibrosis

Knowledge of the pathways by which joint injury leadsto osteoarthritis (OA) in the human knee remainsincomplete (reviewed in1); however, it is agreed thatdisease progression can be attributed to loss of homeo-stasis of the joint ‘‘organ.’’ This loss is associated withprocesses such as biomechanical overload, synovialfibrosis, and cartilage degeneration. Insight into apotentially central pathway in OA pathology wasprovided by the finding that ablation of murineADAMTS5 was cartilage-protective in both the destabi-lization of the medial meniscus (DMM) model of OA2 andantigen-induced inflammatory arthritis.3 Currently, themost prevalent interpretation of these seminal studies isthat ablation of ADAMTS5 directly eliminates thecapacity of murine chondrocytes to degrade cartilageaggrecan in vivo. The DMM widely used in murine OAstudies, requires severing of the menisco-tibial ligament,which necessarily results in injury to the meniscus, theadjacent synovium, and joint capsule. This model there-fore mimics many features of human post-injury OAincluding meniscal damage, synovial inflammation,fibrosis, and cartilage erosion. In addition, one interest-ing, but unexplained aspect of the DMM model is thatmale, but not female C57BL/6 mice develop OA-likesymptoms.4 Our need of a nonsurgical model withoutgender effects, led us to studies5 showing that multiple

injections of TGFb1 into the murine knee joint generatesOA-like symptoms. This suggested that TGFb1 injection,followed by a biomechanical challenge, could represent asimple approach to model development. We report herethat two intra-articular injections of TGFb1, followed by2 weeks of daily uphill treadmill running (named the‘‘TTR model’’ for TGFb1 injection and Treadmill Run-ning) causes reproducible OA in mature male and femaleC57Bl/6 mice. We also show that ADAMTS5 ablationprevents fibrosis and cartilage erosion in both this newmodel and the widely used DMM model.

MATERIALS AND METHODS

Induction of Joint Pathology in the TGFb1 and Treadmill Running(TTR) ModelAll protocols used were approved by Rush University IACUC.Joint injury was initiated by intra-articular TGFb1 injectionessentially as described.5 Briefly, C57BL/6 mice (12-week-oldmales) were given injections (on days 1 and 3) of 200 nghrTGFb1 (Active Form, PrepoTech Inc) in 6ml of sterile salinecontaining 0.1 % ultrapure BSA (Sigma). Injection, into theright knee only, was done with a Kendall Monoject U-100Insulin syringe and a 29G needle. Sham controls were injectedwith 6ml saline/BSA vehicle. Mice were then maintained eitherat cage activity, or subjected to enforced running on the flat oran uphill (178) gradient on a Stoelting/Panlab treadmill at32 cm/s for 20 min/day for 14 days. The generation andgenotyping of ADAMTS5�/� mice, and the DMM modelprocedure were as described.4

Macroscopic Joint Evaluation and Cartilage Erosion ScoringKnee joints were carefully opened, and after rinsing with PBS,India Ink was applied to all surfaces with a paint brush andrinsed again. Menisci and articular surfaces were photo-graphed under a Nikon dissecting microscope (SMZ1000, X6)

516 JOURNAL OF ORTHOPAEDIC RESEARCH APRIL 2011

Jun Li and Wendy Anemaet contributed equally to this study.Abbreviations: ADAMTS; A distintegrin and metalloproteinasewith thrombospondin-like motifs; CS; chondroitin sulfate; GAG;glycosaminoglycan; OA; osteoarthritis; WT; wild-type; TGFb1;transforming growth factor beta 1.Correspondence to: Anna Plaas (T: 312-942-7194; F: 312-942-3053;E-mail: [email protected])

� 2010 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.

and images processed with Spot Basic, Diagnostic Instru-ments, Inc. to X32 mag. Cartilage erosion scores (0–8) werebased on a visual estimate of the fraction (0/8, 1/8, 2/8, etc) ofeach of four areas (the anterior and posterior aspects of thelateral and medial compartments) affected by surface lesions(see Fig. 1 for lesion examples). The maximum possible scorefor each complete femoral or tibial surface was 32.

HistologyHind limbs were skinned and placed intact in 10% neutralbuffered formalin for 48 h, decalcified in 0.5 M EDTA for14 days, processed, and embedded in paraffin. Whole-jointsaggital sections (5 mm) were obtained from �10 locations onthe lateral and medial compartments separately, deparaffi-nized, and stained with SafraninO/Fast Green.

Quantitation of CS in CartilagesKnee joints were opened and the tibial and femoral cartilagesurfaces were separately immersed in 25 ml of 0.1 mg/mlproteinase K in 0.1 M ammonium acetate, pH 7.3, in a 500 mlEppendorf tube, and incubated at 508C for 1 h. Bone ends wererinsed twice with PBS and the proteinase K-solubilized GAGswere processed and quantitated by FACE analyses asdescribed.6 Undigested tissue was fixed for SafraninO staining

which showed that >90% of the articular cartilage has beendigested, whereas the growth plates remained essentiallyintact (data not shown).

Western Blots of Cartilage and Whole-Joint ExtractsHind legs were removed quickly, skinned, and immediatelyplaced in ice-cold PBS with proteinase inhibitors. Whole jointswere left closed and harvested by sharp cuts of the tibia andfemur to generate an approximate 5 mm� 3�mm� 2 mmsample. These were washed briefly in ice-cold PBS, andpulverized in a custom-made stainless steel impactor underliquid nitrogen. For cartilage-only analysis, joints were openedin ice-cold PBS with proteinase inhibitors, menisci and peri-articular tissues removed stringently, and cartilage-richsurfaces were removed by scalpel from both tibial and femoralsurfaces of any one joint. Pulverized whole joints and washedcartilages were each extracted at 48C for 24 h, in 4 M guanidineHCl, 50 mM sodium acetate, pH 7.3 (containing proteinaseinhibitors). Cartilage extracts were concentrated and desaltedby ethanol precipitation. Whole-joint extracts were water-dialyzed and adjusted to 7 M urea/50 mM Tris acetate, pH 8.0before isolation of proteoglycans on DEAE cellulose. Chon-droitinase ABC-digested portions were analyzed by Westernblot for high molecular weight aggrecan (anti-DLS),7 versican(Ab 1033, Millipore), aggrecan G1-NVTEGE, and aggrecan-TSSELE.8

RESULTS

Articular Surface Imaging Shows That ADAMTS5�/�MiceAre Protected from Cartilage Erosion and Joint Fibrosis inthe DMM ModelA common method for evaluating joint pathology inmurine OA is frontal-plane histological analysis of thecartilage across the entire tibio-femoral joint.2,3 Thisrequires the processing, staining, and visual scoring ofcartilage by two blinded individuals on at least15 sections per joint; however, it does not evaluate anyjoint structures other than cartilage. Since OA is nowwidely recognized to involve all tissues of the joint, wehave developed a method which provides a ‘‘global’’visual evaluation of joint pathology. It includes synovialand peri-articular fibrosis and the appearance of thecartilage on all tibial, femoral, and patello-femoralsurfaces. In addition, we developed an objective 8-pointscoring system for cartilage surface per se. Naı̈vefemoral condyles (Fig. 1, top row) and tibial plateaus(Fig. 1, fourth row), consistently appeared smooth andshiny with isolated cases of small rough patches.Patello-femoral surfaces were also undamaged in naı̈vemice (data not shown). Four weeks after meniscalsurgery, wild-type (WT) femoral condyles (panels a–d),and tibial plateaus (panels i–l) showed various pathol-ogies (see black arrows). These ranged from surfaceroughening (a, i, k, l) to deep grooves (c, d, l). Fibroticremodeling at the margins of WT surfaces was commonand is marked with an ‘‘F.’’ Similar pathologies wereseen on both patella and patellar groove cartilagesincluding erosions and fibrosis at the margins (data notshown). In contrast, ADAMTS5�/� surfaces exhibitedonly minor lesions (black arrows), on the femoral (e andf), and tibial (m, o, p) surfaces. Most striking was the

Figure 1. Typical appearance of medial articular surfaces fromWT and ADAMTS5�/� mice after 3 weeks in the DMM model.Four examples of each of WT femoral condyles (panels a–d),ADAMTS5�/� femoral condyles (panels e–h), WT tibial plateaus(panels i–l) and ADAMTS5�/� tibial plateaus (panels m–p) areshown. Two examples of the same areas from naı̈ve mice are shownat the top. Horizontal black bars in panels c and k¼500mm. FiveWT and five knockout mice were used.

ADAMTS5 AND MURINE OSTEOARTHRITIS MODELS 517

JOURNAL OF ORTHOPAEDIC RESEARCH APRIL 2011

absence of the fibrosis seen with high frequency inthe WTs. Close inspection showed that on someADAMTS5�/� surfaces there appeared to be a localizedcartilage deposition (see dashed lines). In summary, inthe DMM model the ablation of ADAMTS5 diminishedmajor fibrotic and erosive pathologies and resulted inthe appearance of presumptive cartilaginous deposits.

Articular Surface Imaging at High Magnification ProvidesDetail on the Effects of ADAMTS5 Ablation in the TTRModel of OATypical surface imaging in the TTR model is shown forthe medial femoral condyle (Fig. 2) and medial tibialplateau (Fig. 3) from WT male mice (top panels) andADAMTS5�/� mice (bottom panels). Adjacent lateralsurfaces (not shown) were generally similar to themedial. The femoral condyles from WT (Fig. 2, panels a–d) showed surface roughening (black arrows) adjacent todeep lesions (panel d). Much as seen for the DMMmodel, in each case this was accompanied by fibrosisat the joint margins (labeled as ‘‘F’’). In contrast,ADAMTS5�/� mice (lower panels) showed relativelyminor roughening (e and f), very limited fibrosis (panel f,labeled ‘‘F’’), and as for the DMM model, there wasevidence of cartilage thickening on ADAMTS5�/�surfaces (dashed lines, panels g and h). The tibialplateaus (Fig. 3) from WTs also showed pathologiesranging from surface erosion (black arrows, panels a–d)to cracking and collapse of the cartilage layer (panel d).These lesions were accompanied by fibrosis and remod-eling at the joint margins (labeled as ‘‘F’’). The interfaceof fibrous tissue and roughened cartilage surface is bestillustrated in panel c. Again, ADAMTS5�/� miceshowed no proliferative fibrosis and only minor cartilageroughening (black arrow, panel f). In addition, similar tothe femoral surfaces (Fig. 2), there was evidence forcartilage deposition on the tibial plateaus (dashed lines,panels g and h). When evaluated separately (data notshown), both menisci in the injected knees of TTR-treated WT mice showed extensive damage with afibrous in-growth from the anterior margins. In con-trast, ADAMTS5�/�mice showed essentially no menis-cal damage, indicating that macroscopic evaluation ofthe menisci might represent a simple and sensitivereadout of changes occurring throughout the joint.

Quantitative Scoring of Cartilage Changes in the TTR ModelThe application of the cartilage scoring system to theTTR model is illustrated in Figure 4. Partial TTR-treatments were designed to determine the effects ofincreasing treadmill severity with and without TGFb1injection. For each group (n¼ 10), the mean (�SD)scores for both femoral and tibial surfaces in TGFb1-injected (plus signs) and Sham-injected (minus signs)are shown. For the WT mice, significant differences(between TGFb1-injected and contra lateral joints) werefound only in groups receiving TGFb1 injection and themost marked differences (p¼ 0.0002) were found forboth the femoral and tibial surfaces in the TGFb1 plus

uphill treadmill group. For this group there wasexcellent reproducibility between mice as shown bythe very low standard deviations. In summary, itappears that TGFb1 injections were required to inducestatistically significant cartilage changes at any activitylevel. Importantly, uphill treadmill running alonedid not induce widespread cartilage changes but itmarkedly exacerbated the TGFb1-induced cartilagechanges seen with cage activity or flat treadmill. InADAMTS5�/�mice (Fig. 4, right), only the TGFb1 plus

Figure 2. Typical appearance of medial femoral surfaces fromWT and ADAMTS5�/� mice in the TTR model. Four examples ofWT surfaces (a–d) and ADAMTS5�/� surfaces (e–h) are shown.Dashed circles indicate areas of cartilage deposition onADAMTS5�/� surfaces. See the text for further detailed descrip-tion. Horizontal black bars in panels g and h¼500mm.

518 LI ET AL.


uphill treadmill treatment was evaluated and thisshowed essentially complete cartilage protection onboth the femoral and tibial surfaces of the right andleft knees.

Cartilage Protection in ADAMTS5�/� Mice Is Associatedwith Increased Total CS in the Articular CartilageSurface imaging (Figs. 2 and 3) indicated thatADAMTS5�/� mice in the TTR model showed not only

cartilage ‘‘protection,’’ but also what appeared to becartilage deposition on the surfaces. To investigatethis, we prepared SafraninO-stained sections of jointsfrom WT mice (Fig. 5, panels a–d) and ADAMTS5�/�mice (Fig. 5, panels e–h). Naı̈ve mice (WT: a, b;ADAMTS5�/�: e, f) showed no major differencesbetween genotypes, however, sections from mice ana-lyzed at the completion of the TTR model exhibitedmarked differences. WT joints showed thinning, loss ofstaining, cellular disorganization, and increased fibrousovergrowth of the femoral surface (compare d, b),consistent with cartilage aggrecan loss. In contrast,the ADAMTS5�/� mice showed cartilage thickeningand increased matrix staining intensity, also seen in thejoint capsule (compare panels h, f), consistent withproteoglycan deposition. To explore this further, wedetermined the total chondroitin sulfate (CS) per jointsurface in TTR-treated WT and ADAMTS5�/� mice(Fig. 5, bottom panels). The results confirmed that,relative to naı̈ve age-matched mice, the WT mice lost CSfrom both the femoral and tibial cartilages of TTR-treated joints, whereas the ADAMTS5�/� mice signifi-cantly increased the CS content of these cartilages.

Figure 3. Typical appearance of medial tibial surfaces from WTand ADAMTS5�/� mice in in the TTR model. Four examples ofwild-type surfaces (a–d) and ADAMTS5�/� surfaces (e–h) areshown. Dashed circles indicate areas of cartilage deposition onADAMTS5�/� surfaces. See the text for further detailed descrip-tion. Horizontal black bars in panels g and h¼ 500mm.

Figure 4. Cartilage erosion scores for femoral and tibial surfacesof WT and ADAMTS5�/� mice in the TTR model. Scores(mean�SD) for the three treatment groups in WT mice are shownwith femoral condyles above and tibial plateaus below. In each case,lateral and medial scores were combined for a maximum score of 32.For this experiment, in all cases (þ) refers to two TGFb1 injectionsinto the right knee (black columns) and (�) refers to two vehicle onlyinjections into the right knee. In each case, data from the uninjectedcontralateral left knee for all treatments served as a control (whitebars). Total scores were examined for statistically significantdifferences between injected and uninjected contralateral legsusing a nonparametric, two-tailed Mann–Whitney U-test. Each ofthe eight experimental groups consisted of 10 animals.



Aggrecan and Versican Accumulate in TTR-Treated Jointsof ADAMTS5�/� MiceThe accumulation of CS in the cartilage of TTR-treatedjoints of ADAMTS5�/� mice (Fig. 5) suggested that achondrogenic response occurs in these mice. To examinethis further, we analyzed cartilage extracts and whole-joint extracts for the abundance of aggrecan (Fig. 6,panel A) and versican (Fig. 6, panel B). Western blotsconfirmed the presence of both proteoglycans in allextracts and showed that for WT mice, both were inlower abundance in the articular cartilages from TTR-treated joints, thus supporting the data obtained oncartilage CS by FACE analysis (Fig. 5). In contrast,ADAMTS5�/� mice showed a marked increase in

aggrecan and versican abundance in TTR-treated joints,also consistent with the histological and FACE data inFigure 5. The Western data showed that the aggrecanpopulation which accumulates in both the cartilage andwhole-joint extracts of TTR-treated knockout mice,contains an abundant high-molecular weight specieswhich was not detected in extracts of WT tissues ornaive knockout joints. In addition, high-molecularweight versican core species also accumulated in thecartilages of only the knockout mice after TTR treat-ment.

Aggrecanase-Mediated Cleavage of Cartilage Aggrecan InVivo Is Not Eliminated by Ablation of ADAMTS5Analysis of cartilage samples for aggrecanase-generatedneo-epitopes (NVTEGE and TSSELE) showed that a64 kDa G1-NVTEGE doublet product was present, asexpected, in WT extracts (Fig. 6, panel C). However, weunexpectedly found a similar abundance of this doublet

Figure 5. Histological appearance and quantitation of CS con-tent of articular cartilages from WT and ADAMTS5�/�mice. Mice(n¼ 5 per group) were subjected to the TTR model or were age-matched naı̈ve. Whole-knee joints were sectioned and stained withSafraninO. The low power images (a, c, e, and g) show the femoro-patellar joint and the high power (b, d, f, and h) show femoro-tibialcontact areas. Naı̈ve joints are in panels a, b, e, and f, whereas jointstaken post-TTR are in panels c, d, g, and h. Thin black arrowsindicate proteoglycan deposition in the joint capsules of knockoutmice and wide black arrows show areas of aggrecan depletion inwild-types (panel d). White arrows highlight the thickening andintense staining of the cartilage layer in knockouts (panel h). Thequantitation of total CS per joint surface, post-TTM(bottom panels),showed statistically significant decreases in WT and increases inknockout mice. For each joint surface and genotype, the CS contentof naı̈ve mice was statistically compared to that of TGFb1-treatedmice using a nonparametric, two-tailed Mann–Whitney U-test.

Figure 6. Western analysis for aggrecan, versican, and aggrecanfragments in the cartilage and whole joints of WT and ADAMTS5�/�mice. Joints were harvested and proteoglycans analyzed from theright knees of WT (n¼8 total) and ADAMTS5�/� (n¼ 8 total) mice,either naive (minus columns) or at the completion of the TTR modelprotocol (plus columns). Whole-joint extracts (n¼2) were analyzedfor aggrecan (panel A) and versican (panel B). Cartilage-richsurface region extracts (n¼6) from each genotype were analyzedfor aggrecan (panel A), versican (panel B) and the aggrecanase-generated neo-epitopes, NVTEGE (panel C), and TSSELE (panelD). The Western blots shown for the anti-NVTEGE and anti-TSSELE probes are typical of those obtained with six different micefor each genotype. (see Ref. 8 for antibody specificities).

520 LI ET AL.


product in ADAMTS5�/� extracts, and further theabundance was similar in naı̈ve and TTR-treated jointsfrom both genotypes. In support of this finding, theintensity of immunostaining for G1-NVTEGE in carti-lages of ADAMTS5�/�mice was indistinguishable fromthat in WTs (data not shown). Taken together, thesedata persuasively demonstrate that cleavage of theinterglobular domain of aggrecan by aggrecanaseactivity in vivo is not diminished by ablation ofADAMTS5. In addition, analysis of extracts for thewell-characterized 250 kDa murine G1-TSSELE prod-uct8–10 showed it also to be present in both treatedand contra-lateral joints from both WT andADAMTS5�/� mice (Fig. 6, panel D). Indeed, thisproduct was most abundant in cartilage from TTR-treated joints of the knockout mice, suggesting thatADAMTS5 ablation concomittantly increases the abun-dance of cartilage aggrecan and the rate of aggrecanase-mediated cleavage of the CS-rich region. These dataargue that the rate of aggrecan synthesis exceeds that ofaggrecan degradation in ADAMTS5�/� mice underthese conditions.

DISCUSSIONThe data presented in this paper suggest thatADAMTS5 has unexpected and more far-reachingfunctions than simply cartilage aggrecan degradation.We come to this conclusion for a number of reasons.Firstly, detailed whole-joint analysis by surface photog-raphy (Figs. 1–3) showed that the effect of ADAMTS5ablation, in both the DMM model2 and the novel TTRmodel described here, was not on cartilage protectionalone. Ablation also prevented meniscal damage and itessentially eliminated fibrotic in-growth to the menisciand to articular cartilage surfaces from the peri-articular tissues (Figs. 1–3). Secondly, histological andbiochemical analysis of the articular cartilages fromADAMTS5�/� mice (Figs. 5 and 6) showed that thesemice were not only cartilage-protected but that theyactually increased (by �1.8-fold) the amount of CS inboth tibial plateaus and femoral condyles. Moreover,this increase was confirmed by Western analysis foraggrecan (Fig. 6A) and versican (Fig. 6B) in bothcartilage and whole-joint extracts. Perhaps most impor-tant was the observation (Fig. 6C) that the increase inaggrecan content of knockout tissues was accompaniedby the accumulation of a very high-molecular weight‘‘mesenchymal-like’’ aggrecan core species which wasnot found in any other samples. Lastly, and perhapsmost significantly, we found that aggrecanase-gener-ated fragments of aggrecan (G1-NVTEGE and G1-TSSELE) were present in both WT and ADAMTS5�/�cartilages (and ADAMTS4/5double�/�cartilages, datanot shown), illustrating that ADAMTS5 ablation doesnot eliminate aggrecanase-mediated degradation ofaggrecan in mouse cartilages. We conclude that aggre-canases other than ADAMTS5 are active in murineknee joint cartilages in vivo, supporting in vitro obser-vations on ADAMTS4/5 double knockout mice.10 It

should also be noted that the conclusions we make arebased on a series of studies between 5 and 10 mice pergroup, which is a relatively small number for individualstudies of this kind. On the other hand, we propose thatwhen the data from these related but independentstudies are combined the final conclusions are sound.Regarding the most unexpected finding (Fig. 6), we haveundertaken multiple experiments of this kind (notreported here) and the detection of aggrecanase-gen-erated fragments in the cartilage of ADAMTS5�/�micehas been a consistent finding.

It should be made clear that the data presented heredoes not establish which aggrecanases (ADAMTS1, 4, 5,8, 9, 15) are primarily responsible for cartilage aggrecandegradation in murine or human OA. Clearly, eventhough ADAMTS5 ablation does not eliminate cartilageaggrecanase activity, it remains possible that aggrecandegradation in WT mice results partly from ADAMTS5.In addition, in the present study the Western analyseswere done only at the end of the TTR model and theresults therefore describe only the ‘‘snap-shot’’ product ofsynthesis, degradation, and removal of aggrecan overthis period.

The findings do however differ from earlier reports2,3

suggesting that ADAMTS5 is the major, or only murinecartilage aggrecanase. This difference may be explainedby the fact that in the present study aggrecanase activitywas assessed with joint cartilage excised immediatelyafter sacrifice of mature mice, and it should unambigu-ously describe structures of the cartilage aggrecan invivo. In contrast, previous studies2,3,9–12 used ex vivoexplants of femoral head cartilage from 3-week-old mice.Since these products were generated ex vivo fromcytokine-stimulated cartilage, the results obtained mayprimarily reflect the response of epiphyseal chondrocytesto the culture conditions. It should also be noted that,since our data show the presence of aggrecanase-generated products in naı̈ve mice, the detection of theseproducts by Western or immunohistology (e.g., 13,14),cannot alone be used as proof of an ongoing destructiveaggrecanolysis. Indeed, the abundance of the G1-TSSELE product (Fig. 6) was highest not in degradingcartilage, but in intact cartilage which was undergoingrapid aggrecan deposition. This applies to both theNITEGE and TASELE neo-epitopes which can be foundin normal cartilages from a range of species, includingbovine, human, and mouse.

Another important finding in the present study wasthat ablation of ADAMTS5 resulted in an accumulationof CS-substituted cartilage aggrecan and versican inresponse to the TTR protocol (Fig. 5). Given theestablished role of versican in mesenchymal condensa-tion,15 this suggests that ablation of ADAMTS5 promotesconditions favorable to TGFb1-induced chondrogene-sis,16 and at the same time blocks TGFb1-inducedfibrosis.17 In this regard, it appears to be significant thatADAMTS5 ablation was accompanied by the accumu-lation of a ‘‘mesenchymal-like’’ aggrecan, similar to PG-M, first described as a pre-chondrogenic product of chick



mesenchyme,18 and later identified as versican V0. Themechanism by which ADAMTS5 might modulate theeffects of TGFb1 on chondrogenesis and fibrosis in vivois unknown, but it may be indicated by our recentfindings with chondrocytes subjected to scratch-woundmigration assays in the presence of TGFb1. At thescratch-wound edges, WT cells assume a fibroblasticmorphology and migrate rapidly as single cells, whereasADAMTS5�/� cells migrate poorly and form cellclusters, surrounded by an aggrecan/versican-rich peri-cellular matrix (J. Velasco and A. Plaas, unpublishedobservations). These effects suggest that ADAMTS5products can alter Smad2/3-dependent cellular events,pointing to a central role for ADAMTS5 in the phenotypicmodulation of mesenchymal progenitor cells required forsoft tissue repair.

ACKNOWLEDGMENTSThis study was supported by Seikagaku Inc (A.P.), Katz/Rubschlager Endowment for OA Research at RUMC (A.P.),and the Arthritis Foundation (J.D.S.). We gratefully acknowl-edge Dr. Vincent Wang (RUMC) for statistical analyses of data.

REFERENCES1. Brandt KD, Dieppe P, Radin EL. 2008. Etiopathogenesis of

osteoarthritis. Rheum Dis Clin North Am 34:531–559.2. Glasson SS, Askew R, Sheppard B, et al. 2005. Deletion of

active ADAMTS5 prevents cartilage degradation in a murinemodel of osteoarthritis. Nature 434:644–648.

3. Stanton H, Rogerson FM, East CJ, et al. 2005. ADAMTS5 isthe major aggrecanase in mouse cartilage in vivo and in vitro.Nature 434:648–652.

4. Malfait AM, Ritchie J, Gil AS, et al. 2010. ADAMTS-5deficient mice do not develop mechanical allodynia associatedwith osteoarthritis following medial meniscal destabilization.Osteoarthritis Cartilage 18:572–580.

5. van Beuningen HM, Glansbeek HL, van der Kraan PM, et al.2000. Osteoarthritis-like changes in the murine knee jointresulting from intra-articular transforming growth factor-beta injections. Osteoarthritis Cartilage 8:25–33.

6. Plaas AH, West LA, Thonar EJ, et al. 2001. Altered finestructures of corneal and skeletal keratan sulfate andchondroitin/dermatan sulfate in macular corneal dystrophy.J Biol Chem 276:39788–39796.

7. Plaas A, Osborn B, Yoshihara Y, et al. 2007. Aggrecanolysis inhuman osteoarthritis: confocal localization and biochemicalcharacterization of ADAMTS5-hyaluronan complexes inarticular cartilages. Osteoarthritis Cartilage 15:719–734.

8. Stewart MC, Fosang AJ, Bai Y, et al. 2006. ADAMTS5-mediated aggrecanolysis in murine epiphyseal chondrocytecultures. Osteoarthritis Cartilage 3:392–402.

9. East CJ, Stanton H, Golub SB, et al. 2007. ADAMTS-5deficiency does not block aggrecanolysis at preferred cleavagesites in the chondroitin sulfate-rich region of aggrecan. J BiolChem 282:8632–8640.

10. Rogerson FM, Stanton H, East CJ, et al. 2008. Evidence of anovel aggrecan-degrading activity in cartilage: studies of micedeficient in both ADAMTS-4 and ADAMTS-5. ArthritisRheum 58:1664–1673.

11. Glasson SS, Askew R, Sheppard B, et al. 2004. Character-ization of and osteoarthritis susceptibility in ADAMTS-4-knockout mice. Arthritis Rheum 50:2547–2558.

12. Ilic MZ, East CJ, Rogerson FM, et al. 2007. Distinguishingaggrecan loss from aggrecan proteolysis in ADAMTS-4 andADAMTS-5 single and double deficient mice. J Biol Chem282:37420–37428.

13. van Lent PL, Grevers LC, Blom AB, et al. 2008. Stimulation ofchondrocyte-mediated cartilage destruction by S100A8 inexperimental murine arthritis. Arthritis Rheum 58:3776–3787.

14. Cao J, Li S, Shi Z, et al. 2008. Articular cartilage metabolismin patients with Kashin–Beck disease: an endemic osteo-arthropathy in China. Osteoarthritis Cartilage 16:680–688.

15. Kamiya N, Watanabe H, Habuchi H, et al. 2006. Versican/PG-M regulates chondrogenesis as an extracellular matrixmolecule crucial for mesenchymal condensation. J Biol Chem281:2390–2400.

16. Shintani N, Hunziker EB. 2007. Chondrogenic differentiationof bovine synovium: bone morphogenetic proteins 2 and 7 andtransforming growth factor beta1 induce the formation ofdifferent types of cartilaginous tissue. Arthritis Rheum56:1869–1879.

17. Gauldie J, Bonniaud P, Sime P, et al. 2007. TGF-beta, Smad3and the process of progressive fibrosis. Biochem Soc Trans35:661–664.

18. Shinomura T, Nishida Y, Ito K, et al. 1993. cDNA cloning ofPG-M, a large chondroitin sulfate proteoglycan expressedduring chondrogenesis in chick limb buds. Alternative splicedmultiforms of PG-M and their relationshiops to versican.J Biol Chem 268:14461–14469.

522 LI ET AL.


knockout of adamts5 does not eliminate cartilage aggrecanase activity but abrogates joint fibrosis...

Documents