ARTHRITIS & RHEUMATISMVol. 64, No. 12, December 2012, pp 3972–3981DOI 10.1002/art.34678© 2012, American College of Rheumatology

Stress-Induced Cartilage DegradationDoes Not Depend on the NLRP3 Inflammasome in

Human Osteoarthritis and Mouse Models

Carole Bougault,1 Marjolaine Gosset,1 Xavier Houard,1 Colette Salvat,1 Lars Godmann,2

Thomas Pap,2 Claire Jacques,1 and Francis Berenbaum3

Objective. The main feature of osteoarthritis(OA) is degradation and loss of articular cartilage.Interleukin-1� (IL-1�) is thought to have a prominentrole in shifting the metabolic balance toward degrada-tion. IL-1� is first synthesized as an inactive precursorthat is cleaved to the secreted active form mainly inthe “inflammasome,” a complex of initiators (includingNLRP3), adaptor molecule ASC, and caspase 1. The aimof this study was to clarify the roles of IL-1� and theinflammasome in cartilage breakdown.

Methods. We assessed IL-1� release by cartilageexplants from 18 patients with OA. We also evaluatedthe lipopolysaccharide (LPS)–, IL-1�–, and tumornecrosis factor � (TNF�)–induced activity of matrixmetalloproteinase 3 (MMP-3), MMP-9, and MMP-13in NLRP3-knockout mice and wild-type mice and the in-hibition of caspase 1 with Z-YVAD-FMK and the block-ade of IL-1� with IL-1 receptor antagonist (IL-1Ra).Cartilage explants from NLRP3-knockout mice andIL-1R type I (IL-1RI)–knockout mice were subjectedto excessive dynamic compression (0.5 Hz, 1 MPa) totrigger degradation, followed by assessment of load-

induced glycosaminoglycan (GAG) release and MMPenzymatic activity.

Results. Despite the expression of NLRP3, ASC,and caspase 1, OA cartilage was not able to produceactive IL-1�. LPS, IL-1�, and TNF� dose-dependentlyincreased MMP-3, MMP-9, and MMP-13 activity incultured chondrocytes and in NLRP3�/� chondrocytes,and this effect was not changed by inhibiting caspase 1or IL-1�. The load-induced increase in GAG releaseand MMP activity was not affected by knockout ofNLRP3 or IL-1RI in cartilage explants.

Conclusion. OA cartilage may be degraded inde-pendently of any inflammasome activity, which mayexplain, at least in part, the lack of effect of IL-1�inhibitors observed in previous trials.

Osteoarthritis (OA) is the most prevalent diseaseof articular joints and is the major cause of disability inolder adults in industrialized countries (1). The mainfeatures of OA are degeneration and loss of articularcartilage, which occur concomitantly with changes in theunderlying bone and some degree of synovial inflam-mation (2,3). Cartilage breakdown is attributable tononspecific cleavage of matrix molecules in response toabnormal biomechanical stress (3) and to specific cata-bolic processes involving matrix-degrading enzymes. Asa result of this degeneration, glycosaminoglycan (GAG)and collagen fragments (mostly type II collagen) arereleased from OA cartilage. Along with aggrecanases(ADAMTS-4 and ADAMTS-5), numerous matrix met-alloproteinases (MMPs) contribute to this degradation.In particular, MMP-3, MMP-9, and MMP-13 togethercan cleave proteoglycans and type II and type XIcollagens (4,5).

In addition to its crucial role in the context ofinfections and immune-mediated disease, the inflam-matory cytokine interleukin-1� (IL-1�) is considered to

Supported by the Agence Nationale de la Recherche (grantANR-06-PHYSIO-020-01) and the French Society of Rheumatology.Dr. Bougault’s work was supported by a Fondation pour la RechercheMedicale Postdoctoral Fellowship. Dr. Gosset’s work was supported byAgence Nationale de la Recherche Postdoctoral Fellowships.

1Carole Bougault, PhD, Marjolaine Gosset, DMD, PhD,Xavier Houard, PhD, Colette Salvat, Claire Jacques, PhD: UniversityPierre and Marie Curie Paris VI, Paris, France; 2Lars Godmann,Thomas Pap, MD, PhD: University Hospital Muenster, Muenster,Germany; 3Francis Berenbaum, MD, PhD: University Pierre andMarie Curie Paris VI and St. Antoine Hospital, AP-HP, Paris, France.

Drs. Jacques and Berenbaum contributed equally to thiswork.

Address correspondence to Francis Berenbaum, MD, PhD,UR4, University Pierre and Marie Curie Paris VI, 7 Quai St. Bernard,Paris 75252, Cedex 5, France. E-mail: francis.berenbaum@sat.aphp.fr.

Submitted for publication December 21, 2011; accepted inrevised form August 14, 2012.

3972

be involved in joint diseases, including OA (6). Notably,IL-1� is a potent regulator of MMP expression andactivity (6,7). IL-1� is first synthesized as an inactiveprecursor to pro–IL-1�, which requires cleavage of itsamino-terminal region by caspase 1 to change into thesecreted active form. Caspase 1 itself needs to beconverted from pro–caspase 1 to active caspase 1 via amolecular scaffold called the “inflammasome.” The in-flammasome is also involved in the processing of IL-18and IL-1F7 (for review, see refs. 8–10) but not IL-1�,which is processed independently by calpain proteases.

Four inflammasome subtypes have been definedon the basis of the initiator proteins. Three of theseinitiator proteins, NLRP1, NLRP3, and NLRC4/IPAF,belong to NLR family. The fourth possible initiatorprotein is AIM2. These initiators recruit the adaptormolecule ASC, which interacts with pro–caspase 1.Formation of the complex initiates proximity-inducedautocleavage of pro–caspase 1 into caspase 1. TheNLRP3 inflammasome subtype has been studied mostoften. Because the inflammasome is a master regulatorof inflammation, research is now focusing on the mech-anisms leading to its activation (for review, see refs.8–12). Known activators of NLRP3 are danger-associated molecular patterns and pathogen-associatedmolecular patterns, including ATP and particulate orcrystalline agonists such as fibrillar amyloid � peptide,monosodium urate, or silica (9).

In addition to its intracellular maturation bycaspase 1, pro–IL-1� can be activated by extracellularproteases such as trypsin, chymotrypsin, cathepsin G,elastase, or some MMPs (13). Specifically, MMP-9, andto a lesser extent MMP-3 and MMP-2, can process theIL-1� precursor into biologically active forms (14). Suchactivity from MMP family members is of particularinterest in the context of OA cartilage, because MMPlevels are particularly high in OA joints (4).

Given the primary role of the inflammasome inIL-1� maturation and the putative role of IL-1� in OApathology, we sought to clarify the role of both theinflammasome and IL-1� in cartilage breakdown. Wefirst investigated the expression of inflammasome com-ponents and the capacity of human OA knee jointcartilage explants to release active IL-1�. Second, wetriggered a prodegradative phenotype in primary mousearticular chondrocytes, using the following proinflam-matory treatments: lipopolysaccharide (LPS), IL-1�,and tumor necrosis factor � (TNF�). We studied theeffect of NLRP3 knockout, caspase 1 inhibition, andIL-1 blockade in order to assess the involvement of theinflammasome and IL-1� in the inflammatory stress–induced prodegradative responses of chondrocytes. Fi-

nally, we used dynamic compression to induce biome-chanical degradation in mouse cartilage explants inorder to investigate the role of NLRP3 and IL-1� inmechanical stress–induced cartilage breakdown.

MATERIALS AND METHODS

Effectors and inhibitors. LPS and TNF� were ob-tained from Sigma-Aldrich. Z-YVAD-FMK peptide, a cell-permeable inhibitor of caspase 1, was obtained from AlexisBiochemicals. IL-1�, IL-1�, and IL-1 receptor antagonist(IL-1Ra) were obtained from PeproTech.

Human OA cartilage and synovium. Human jointsamples were obtained from the Department of Bone andJoint Diseases at AP-HP Saint-Antoine Hospital as surgicalwaste, in the absence of patient opposition and in accordancewith French ethics laws (L. 1211-2 to L. 1211-7, L. 1235-2, andL. 1245.2). Both cartilage and synovium samples were obtainedfrom 18 patients undergoing total joint replacement for OA.The OA diagnosis was based on clinical and radiographicevaluation according to the American College of Rheumatol-ogy criteria (15). The investigation conformed to the principlesoutlined in the Declaration of Helsinki.

Animals. All experiments involving pharmacologic in-hibitor treatments were performed on primary chondrocytesextracted from 6-day-old Swiss mice (Janvier). Some experi-ments were performed using NLRP3-knockout mice (a giftfrom J. Tschopp, Lausanne University, Switzerland [16]) orIL-1R type I (IL-1RI)–knockout mice (a gift from Dr. Pap).All of the mice were housed in a pathogen-free facility. Allprocedures were performed in accordance with the EuropeanDirective N886/609 and the local committees for animal useand care.

Genomic DNA from mice tail fragments was pre-pared according to the HotSHOT technique (17) and wasused for genotyping. The screening strategy allowed for poly-merase chain reaction (PCR) amplification of both mutantand wild-type (WT) allele fragments in the same tube. ForNLRP3 genotyping, the mutant fragment was 500 bp, the WTfragment was 250 bp, and the primer sequences were asfollows: common sense TCAAGCTAAGAGAACTTTCTG,mutant antisense AAGTCGTGCTGCTTCATGT, and WTantisense ACACTCGTCATCTTCAGCA. For IL-1RI geno-typing, the mutant fragment was 170 bp, the WT fragment was350 bp, and the primer sequences were as follows: mutantsense CTGAATGAACTGCAGGACGA, mutant antisenseATACTTTCTCGGCAGGAGCA, WT sense CCACATATT-CTCCATCATCTCTGCTGGTA, and WT antisense TTTCG-AATCTCAGTTGTCAAGTGTGTCCC.

Primary culture of articular chondrocytes. Primarychondrocytes were isolated from the articular cartilage of mice(ages 4–6 days), as previously described (18). After 1 week ofamplification, the cells were placed in serum-free conditions(0.1% bovine serum albumin; Sigma-Aldrich) for 24 hoursand then treated in serum-free medium supplemented withup to 1 �g/ml LPS, 10 ng/ml IL-1�, or 100 ng/ml TNF� with orwithout pharmacologic inhibitors (10 �M Z-YVAD-FMK and100 ng/ml IL-1Ra). Human chondrocytes were isolated andamplified for 1 week, using identical protocols. Chondrocyteswere extracted from the articular cartilage from a patient withOA who was undergoing total joint replacement surgery.

ROLE OF INFLAMMASOME IN OA CARTILAGE DEGRADATION 3973

Mouse cartilage explants and compression experi-ments. The procedure for compression of mouse costal carti-lage explants was performed as previously described (19).Briefly, explants were harvested from the rib cages of mice(ages 4–6 days). Once cleaned, divided into segments, pooled,and weighed for further normalization, the explants wereallowed to rest for �20 hours in 3 ml of serum-free mediumand then washed, and 1.5 ml of fresh medium was added. Thesamples then underwent 6-hour dynamic compression (sinu-soidal waveform 0–1 MPa at 0.5 Hz), using a Flexcell compres-sion system.

Real-time PCR analysis. Total RNA was extractedfrom monolayer-cultured cells, using an RNeasy Mini Kit(Qiagen). Complementary DNA samples were obtained byreverse transcription of 1 �g RNA, using an Omniscript kit(Qiagen). Relative quantification of genes was performed usinga LightCycler 480 Real-Time PCR System (Roche AppliedScience) and GoTaq qPCR Master Mix (Promega). MessengerRNA (mRNA) levels were normalized to that of hypoxanthineguanine phosphoribosyltransferase (HPRT), which was usedas an internal standard. Hybridization at 60°C was performedfor the following murine primer sequences: for MMP-2, senseGATGCTGCCTTTAACTGGAGTA and antisense GGAGT-CTGCGATGAGCTT; for MMP-3, sense TGAAAATGAAG-GGTCTTCCGG and antisense GCAGAAGCTCCATACCA-GCA; for MMP-9, sense AACTACGGTCGCGTCCACT andantisense CCACAGCCAACTATGACCAG; for MMP-13,sense GATGGCACTGCTGACATCAT and antisense TGT-AGCCTTTGGAACTGCTT; for ADAMTS-4, sense CTTCC-TGGACAATGGTTATGG and antisense GAAAAGTCGC-TGGTAGATGGA; for ADAMTS-5, sense CAGCCACC-ATCACAGAA and antisense CCAGGGCACACCGAGTA;for IL-1�, sense GGGCCTCAAAGGAAAGAATC and anti-sense CCACTTTGCTCTTGACTTCTATC; for IL-18, senseTCTGCAACCTCCAGCAT and antisense TTTCTTCAGGT-ATAAAGTAAAGCGTG; for HPRT, sense AGGACCTCT-CGAAGTGT and antisense ATTCAAATCCCTGAAGTAC-TCAT.

Soluble IL-1� and IL-18 assay in OA cartilage andsynovium–conditioned medium. Explants of cartilage and sy-novium from the same patient with OA were cut into smallpieces (�1 mm3), weighed, and incubated for 24 hours inserum-free RPMI medium (with or without 1 �g/ml of LPS or100 ng/ml of TNF�). The incubation volume was normalized tothe wet weight of the explants (6 ml/gm). The levels of solubleIL-1� and IL-18 were measured in the conditioned medium,using a high-sensitivity enzyme-linked immunosorbent assay(ELISA) kit for human IL-1� with a detection threshold of0.4 pg/ml (Sanquin), and a kit for human IL-18 with a detectionthreshold of 12.5 pg/ml (Life Technologies).

Assay of MMP-3, MMP-9, and MMP-13 secretion.Total mouse MMP-3 secretion was assayed using a commer-cially available ELISA kit with a detection threshold of 0.2 ng/ml (R&D Systems). Total mouse MMP-13 secretion was de-termined by Western blot analysis, as previously described(20), with rabbit polyclonal antibody for MMP-13 (H-230;Santa Cruz Biotechnology). Densitometric analysis of immuno-blots was performed using Multi-Gauge software (FujifilmMedical Systems). MMP-9 secretion was analyzed by zymog-raphy with 8% acrylamide/bis-acrylamide separation gel con-taining 1.2 mg/ml gelatin (21). Briefly, after electrophoresis

under nonreducing conditions and sodium dodecyl sulfateremoval, gels were incubated overnight at 37°C in hydrolysisbuffer (50 mM Tris HCl, pH 7.4, 100 mM NaCl, 5 mM CaCl2).Lysis bands were detected by negative staining with Coomassiebrilliant blue R250. Samples were concentrated using cellulosemembrane columns with a weight cutoff of 3 kd (Ultracel-3K;Millipore); a bicinchoninic acid quantification kit (Interchim)was used to standardize measurements to the total proteinconcentration.

GAG assays. The release of sulfated GAG by cartilageexplants into culture media was examined by determining theamount of polyanionic material reacting with dimethylmethyl-ene blue (22), with shark chondroitin sulfate used as a stan-dard. Results were normalized to the milligram wet weight of

Figure 1. Release of interleukin-1� (IL-1�) by cartilage (C) andsynovium (S) from 18 patients with osteoarthritis (OA). A, SolubleIL-1� concentrations in conditioned medium after 24-hour incubationwith cartilage explants, as determined by enzyme-linked immunosor-bent assay (ELISA). Corresponding synovium explants were used aspositive controls. � � P � 0.05. B, Fold induction of IL-1� in responseto treatment with proinflammatory agents, as determined by ELISA.Cartilage and synovium explants from 4 patients with OA werestimulated with 1 �g/ml lipopolysaccharide (LPS) or 100 ng/ml tumornecrosis factor � (TNF�). Results for treated explants were normal-ized to those for untreated explants. Symbols represent individual datapoints. Boxed areas show the mean. The broken line represents noinduction. �� � P � 0.01 versus no induction.

3974 BOUGAULT ET AL

cartilage and the concentration of proteins in the culturesupernatant.

MMP enzymatic activity assays. Global MMP activitywas measured using Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2synthetic fluorogenic substrate (Bachem) in continuous assays(21,22). When 1% EDTA was added, enzymatic activity wascompletely inhibited; this demonstrated that metal cationswere needed as coactivators as expected, because all MMPsdepend on Zn2�. Results were normalized to the milligram wetweight of cartilage and the concentration of proteins in theculture supernatant.

Statistical analysis. Data are expressed as the mean �SEM and were analyzed by analysis of variance, using InStatsoftware (GraphPad). P values less than 0.05 were consideredsignificant.

RESULTS

Slight amounts of IL-1� released by human OAcartilage. We assessed IL-1� release by cartilage ex-plants and synovial tissue from 18 patients with OA.Soluble IL-1� release by OA cartilage samples waspoorly detectable (Figure 1A); it was not detected in 6of 18 patients. The release of IL-1� by synovium ex-plants was greater than that from cartilage (mean �SEM 854 � 533 versus 8 � 2 pg/gm) (Figure 1A). TheIL-18 concentration in cartilage and synovium from 9 of18 patients was measured; IL-18 was not detected inconditioned media from cartilage explants (detectionthreshold 75 pg/gm) but was detected in media from

synovium (mean � SEM 2,064 � 545 pg/gm) (data notshown). The concentrations obtained using this protocolmay represent the release of newly synthesized cytokinesand former cytokines trapped in the tissue.

Inability of human OA cartilage to produceIL-1� despite the presence of NLRP3–inflammasomecomponents. To evaluate the putative capacity of OAchondrocytes to produce IL-1� and IL-18, we investi-gated the presence of inflammasome components in-volved in pro–IL-1� and pro–IL-18 maturation. Westernblot analysis revealed the expression of caspase 1, ASC,and NLRP3 protein in lysates of primary chondrocytesfrom OA knee joint explants (data not shown). ThemRNA expression of inflammasome components inchondrocytes was confirmed by quantitative reversetranscription–polymerase chain reaction (data notshown). Therefore, OA chondrocytes expressed all ofthe components needed for the maturation of IL-1� andIL-18.

To determine the ability of OA tissue to secretenewly synthesized IL-1� and IL-18, we stimulated car-tilage and synovium explants from 4 patients with OAwith 1 �g/ml LPS for 24 hours. LPS stimulation ofcartilage samples did not increase the expression ofsoluble IL-1� in conditioned media (Figure 1B), andIL-18 expression remained undetectable (results notshown). However, the same LPS treatment increased

Figure 2. Dose-dependent stimulation of protease expression by primary cultures of mouse articular chondrocytes in response to inflammatorystress. Chondrocytes were left untreated (open bars) or were treated for 24 hours with LPS (10 ng/ml [shaded bars] or 1,000 ng/ml [solid bars]), IL-1�(1 ng/ml [shaded bars] or 10 ng/ml [solid bars]), or TNF� (10 ng/ml [shaded bars] or 100 ng/ml [solid bars]). The expression of matrixmetalloproteinase 2 (MMP-2), ADAMTS-4, ADAMTS-5, MMP-3, MMP-13, and MMP-9 was determined by real-time polymerase chain reactionanalysis. Values are the mean � SEM. � � P � 0.05; �� � P � 0.01 versus untreated. See Figure 1 for other definitions.

ROLE OF INFLAMMASOME IN OA CARTILAGE DEGRADATION 3975

IL-1� release from synovium explants (from 28-fold to74-fold) (Figure 1B); no change in IL-18 release wasobserved (results not shown). Cartilage and synoviumexplants from 4 other patients with OA were stimulatedfor 24 hours with 100 ng/ml TNF�. Similarly, TNF�treatment increased IL-1� release by synovium explants(from 2-fold to 32-fold) but not from cartilage explants

(Figure 1B). More precisely, IL-1� was detected in only1 of the 4 untreated cartilage explants, and its level wasnot further increased by TNF� stimulation. In 2 of the 3other samples, IL-1� was detected following TNF�treatment but only at low levels (17 pg/gm and 2 pg/gm,respectively). Incubation for 24 hours did not alter cellviability in the explants (data not shown).

Figure 3. Dose-dependent stimulation of protease release by primary cultures of mouse articular chondrocytes in response to inflammatory stress.Chondrocytes were left untreated (open bars) or were treated for 24 hours with LPS (10 ng/ml [shaded bars] or 1,000 ng/ml [solid bars]), IL-1�(1 ng/ml [shaded bars] or 10 ng/ml [solid bars]), or TNF� (10 ng/ml [shaded bars] or 100 ng/ml [solid bars]). A, Total matrix metalloproteinase 3(MMP-3) secretion in culture media, as determined by ELISA. B, Quantification of MMP-13 protein expression, as determined by Western blotanalysis; the immunoblots were analyzed by densitometry. C, MMP-9 activity, as determined using gelatin zymography. The results in B and C arerepresentative of 4 independent experiments. Values are the mean � SEM. � � P � 0.05 versus untreated; �� � P � 0.01 versus untreated. SeeFigure 1 for other definitions.

Figure 4. Involvement of NLRP3 in the inflammatory stress–induced prodegradative response of mouse primary articular chondrocytes. NLRP3�/�

and wild-type (WT) mouse chondrocytes were treated for 24 hours with LPS (1 �g/ml), IL-1� (10 ng/ml), or TNF� (100 ng/ml). A, Expression ofmatrix metalloproteinase 3 (MMP-3), MMP-13, and MMP-9 mRNA, as determined by real-time polymerase chain reaction analysis. B, TotalMMP-3 release, as determined by ELISA. C, MMP-13 protein expression, as determined by Western blot analysis. D, MMP-9 activity, as determinedusing gelatin zymography. Results in D are representative of 3 independent experiments. Values in A–C are the mean � SEM. � � P � 0.05;�� � P � 0.01 versus untreated. NS � not significant (see Figure 1 for other definitions).

3976 BOUGAULT ET AL

Inflammatory stress–induced prodegradative re-sponse of mouse articular chondrocytes. LPS, IL-1�,and TNF� treatments activated a prodegradative phe-notype in mouse articular chondrocytes, as evaluatedby induction of cartilage matrix degradation enzymes.The expression of MMP-3, MMP-13, and MMP-9mRNA was up-regulated, but the expression of MMP-2,ADAMTS-4, and ADAMTS-5 mRNA was not up-regulated (Figure 2A). A dose-dependent induction ofthese 3 major cartilage MMPs at the protein level wasconfirmed by the observation that proinflammatorytreatments significantly increased MMP-3 release intochondrocyte culture medium. The amount of MMP-3released was 447 ng/ml with 1,000 ng/ml LPS, 302 ng/mlwith 10 ng/ml IL-1�, and 51 ng/ml with 100 ng/ml ofTNF� (Figure 3A). In addition, MMP-13 release wasdose-dependently increased in response to LPS andIL-1� (Figure 3B). Finally, inflammatory stress–inducedMMP-9 secretion was assessed by gelatin zymography(Figure 3C). The 3 MMPs were used as prodegradativemarkers for the experiments with cultured mouse chon-drocytes.

NLRP3- and caspase 1–independent prodegra-dative response of mouse articular chondrocytes. Tocharacterize the role of the NLRP3 inflammasome inthe prodegradative response of chondrocytes, we usedNLRP3�/� mouse chondrocytes. The mRNA and pro-tein levels of MMP-3, MMP-13, and MMP-9 induced byinflammatory stress was similar in WT and NLRP3�/�

chondrocytes (Figure 4). The response may involveother inflammasome subtypes. Because caspase 1 is arequired partner of all inflammasomes, we inhibited itsenzymatic activity with Z-YVAD-FMK (10 �M). Thenontoxicity, efficiency, and specificity of the inhibitor atthis concentration have been shown in primary chondro-cytes (23). MMP-3, MMP-9, or MMP-13 stimulation atthe mRNA level (data not shown) or the protein level(Figures 5A–C) was similar with and without caspase 1inhibition. These results strongly suggest that chondro-cytes can acquire a catabolic phenotype in inflammasome-independent pathways.

IL-1�–independent prodegradative response ofmouse articular chondrocytes. Because proteolyticmechanisms other than inflammasomes may be impli-

Figure 5. Involvement of caspase 1 and IL-1� in the inflammatory stress–induced prodegradative response of mouse primary articularchondrocytes. Chondrocytes were treated for 24 hours with LPS (1 �g/ml), IL-1� (10 ng/ml), or TNF� (100 ng/ml), with or without 10 �MZ-YVAD-FMK to inhibit caspase 1 enzymatic activity (A–C) and with or without 100 ng/ml IL-1 receptor antagonist (IL-1Ra) to block IL-1� (D–F).A and D, Total matrix metalloproteinase 3 (MMP-3) release, as determined by ELISA. B and E, Protein levels of MMP-13, as determined by Westernblot analysis. C and F, MMP-9 activity, as determined using gelatin zymography. Values in A, B, D, and E are the mean � SEM. Results in C andD are representative of 3 independent experiments. � � P � 0.05; �� � P � 0.01; ��� � P � 0.001 versus untreated. Ctrl � control; NS � notsignificant (see Figure 1 for other definitions).

ROLE OF INFLAMMASOME IN OA CARTILAGE DEGRADATION 3977

cated in IL-1� maturation, we sought to determinewhether IL-1� could be released by mouse articularchondrocytes in response to proinflammatory treat-ments. Freshly produced IL-1� could act as an autocrineprodegradative inducer. Thus, we added a high concen-tration of IL-1Ra (100 ng/ml), together with LPS orTNF�, to the chondrocyte culture and observed that theinduction of MMP-3, MMP-9, or MMP-13 mRNA (datanot shown) and protein in response to LPS or TNF�was similar with and without IL-1Ra treatment (Fig-ures 5D–F). Thus, LPS- and TNF�-induced matrixprotease synthesis by chondrocytes does not involveIL-1�. IL-1Ra, at a dose of 100 ng/ml, could inhibit the

increased MMP-3, MMP-9, and MMP-13 release in-duced by 10 ng/ml of IL-1� (Figures 5D–F).

NLRP3- and IL-1�–independent prodegradativeresponse of mouse cartilage explants to compression.To ex vivo test our hypothesis that IL-1� is not crucialfor a prodegradative phenotype in chondrocytes, weused biomechanical load stimulation to trigger a pro-degradative response in mouse cartilage explants. Theprodegradative phenotype in the model was defined by2 criteria: increase in GAG release (a marker of carti-lage matrix degeneration) and increase in MMP enzy-matic activity in the culture medium. We verified thatcompression did not alter cell viability in our system(data not shown). Dynamic compression for 6 hoursincreased GAG release by 3-fold (Figure 6A) and MMPactivity by 3.7-fold (Figure 6B).

We analyzed the effect of compression on theprodegradative response of chondrocytes in NLRP3�/�

mouse cartilage explants. Similar to in vitro results,the prodegradative response of chondrocytes to com-pression did not differ between the WT and NLRP3-knockout mice in terms of GAG release (Figure 6A) orMMP activity (Figure 6B). Thus, load-induced GAGrelease did not depend on NLRP3. Furthermore, me-chanical stress–activated chondrocytes could producefunctional enzyme activity devoted to cartilage matrixdegradation independently of NLRP3.

To exclude the possibility that the prodegradativeresponse to compression was mediated by IL-1�, weused cartilage from mice lacking IL-1RI, which mediatesIL-1–dependent signaling. The prodegradative responseof these chondrocytes to compression did not differbetween the WT and IL-1RI–knockout mice in terms ofGAG release (Figure 6A) or MMP enzymatic activity(Figure 6B). Thus, load-induced cartilage matrix degra-dation did not depend on IL-1�.

DISCUSSION

The question of whether active IL-1� is producedin OA cartilage remains highly controversial. Our resultsshowing slight concentrations of soluble IL-1� releasedfrom OA cartilage explants at 24 hours suggest that OAchondrocytes may have poor exposure to IL-1�. Theseresults are consistent with those of previous studiesshowing low IL-1� concentrations (�0.4 pg/ml) in syno-vial fluid from OA knee joints (24–26). However, chon-drocytes may have greater exposure to IL-18, the con-centration of which in OA synovial fluid is �130 pg/ml(25). Even if IL-1� is present in OA joints, we believethat chondrocytes embedded in cartilage matrix may be

Figure 6. Involvement of NLRP3 and interleukin-1� (IL-1�) in theload-induced prodegradative response of mouse cartilage explants.Cartilage explants from IL-1 receptor type I–knockout (IL-1RI�/�)mice (n � 3), NLRP3�/� mice (n � 4), or wild-type (WT) mice (n �4) were subjected to dynamic compression (0.5 Hz, 1 Mpa) for 6 hours.The results for loaded cartilage explants were normalized to those forcorresponding nonloaded explants. A, Amount of glycosaminoglycan(GAG) released from cartilage explants into culture medium. B,Matrix metalloproteinase (MMP) enzymatic activity in culture me-dium. Values are the mean � SEM. � � P � 0.05; �� � P � 0.01;��� � P � 0.001 versus control (Ctrl). NS � not significant.

3978 BOUGAULT ET AL

barely exposed to this cytokine and may not have anyexposure in the deep zone.

The capacity of chondrocytes to produce activeIL-1� also remains highly controversial. In situ, IL-1�gene expression occurs at low levels in normal cartilageand is not significantly up-regulated in OA cartilage(27). However, local production of IL-1� in OA carti-lage may be possible. Colocalization of caspase 1, IL-1�,and IL-18 expression has been demonstrated in OAchondrocytes (28,29). Moreover, the production ofIL-1� and IL-18 in cartilage explants can be blocked bya specific caspase 1 inhibitor (29).

To our knowledge, no in vitro study has demon-strated the release of active IL-1� from cultured chon-drocytes, despite up-regulated mRNA or intracellularpro–IL-1� levels (30,31). In contrast, mature IL-18appears to be secreted in small amounts by humancultured chondrocytes (32). We observed that LPS,IL-1�, and TNF� dose-dependently increased IL-1�mRNA expression (which was not detected in controlchondrocytes). IL-18 mRNA expression was detected atthe same level in control and stimulated chondrocytes(data not shown). However, neither IL-1� expressionnor IL-18 expression was detected in culture superna-tants (detection thresholds 0.4 pg/ml and 12.5 pg/ml,respectively).

To explain why IL-1� does not seem to beproduced by activated chondrocytes, we investigatedexpression of components of the inflammasome, themain cytosolic complex known to convert pro–IL-1� intoactive IL-1�. We confirmed the protein expression ofNLRP3 (also known as NALP3 or cryopyrin), ASC, andcaspase 1 (also known as ICE [IL-1� converting en-zyme]) in articular chondrocytes from patients with OA.The expression of caspase 1 in human OA chondrocyteshas been studied in detail (28,29). Thus, chondrocytespossess the complete molecular machinery required forformation of the complex needed for IL-1� maturation.However, active IL-1� was barely detectable in OAchondrocyte lysates after in vitro incubation with pro–IL-1� (29), which suggests that the complex may not befunctional.

Apart from chondrocytes, other types of jointcells have shown expression of inflammasome compo-nents. For instance, the NLRP3/ASC complex is func-tional in osteoblasts, because NLRP3 was required forcaspase 1 activation, which led to osteoblast apoptosis(33). Of note, osteoblasts do not release the caspase1–dependent cytokines IL-1� and IL-18 (33). NLRP3,ASC, and caspase 1 levels are also detected, at least atthe mRNA level, in synovial fibroblasts (34,35) andwhole synovium extracts from patients with OA (36).

Whether IL-1� and IL-18 diffuse from the sur-rounding synovial fluid into the cartilage matrix is un-known, but the question is relevant in light of increasedlevels of IL-1� and IL-18 detected in the upper zone ofOA articular cartilage (28,29,37). Consistently, our re-sults showed that release of IL-1� and IL-18 fromsynovial membranes was greater than that from carti-lage. In addition, activated IL-1� signaling pathwayswere localized in the uppermost zones of both normaland OA cartilage (27). The secretion of IL-18 has beenstudied less, but IL-18 production may also occur in thesynovium (38).

The active IL-1� detected at slight concentra-tions in OA cartilage may be synthesized by cells fromthe synovium and diffused through synovial fluid intothe superficial zone of cartilage. More precisely, activeIL-1� is probably produced by immune cells present inthe synovium under pathologic conditions, even in pa-tients with early OA (34,39). Indeed, catabolic andproinflammatory mediators, including IL-1�, producedby the inflamed synovium may exacerbate the prodegra-dative mechanisms responsible for cartilage breakdown,which has led to increasing interest in synovium-targetedtherapy in OA (40).

We showed that chondrocytes can acquire aprodegradative phenotype without any contribution bythe NLRP3 inflammasome and IL-1�, in 2 differentmodels: release of cartilage catabolic enzymes (MMP-3,MMP-9, and MMP-13) from cultured articular chondro-cytes and enhanced matrix degradation of cartilageexplants. Neither blockade of the inflammasome com-plex by NLRP3 deficiency, caspase 1 inhibition, norinactivation of IL-1� pathways had any effect on suchprodegradative responses. Thus, inflammasome activitymay not be crucial for OA-like cartilage breakdown.

The role of inflammasomes has been clearlyrevealed in inflammatory arthritis, which is highly de-pendent on IL-1�. In gout, acute inflammation is at-tributable to the presence of monosodium urate crystalsthat trigger the formation of the NLRP3 inflammasome(16,41). Similarly, the NLRP3 inflammasome plays acritical role in arthritis; this role is associated with thedeposition of hydroxyapatite crystals (42). However,such inflammatory pathways rely on activation of theinflammasome of monocytes or macrophages ratherthan resident joint cells. ASC�/� mice are protectedagainst arthritis via inflammasome-independent path-ways, because mice deficient in NLRP3 or caspase 1 arestill susceptible to arthritis (35,43). These results suggestan effect of ASC in joint pathology through cell-mediated immune responses; to date, however, no studyhas examined its potential role in chondrocytes them-

ROLE OF INFLAMMASOME IN OA CARTILAGE DEGRADATION 3979

selves. Although some studies have demonstrated thatcaspase 1 inhibitors could reduce OA scores in animalmodels (44,45), we observed that caspase 1 inhibitionhad no effect on the chondrocyte prodegradative phe-notype. These results may not be discordant, becausethe favorable effect of caspase 1 inhibitors on OA-likelesions observed in vivo may not be attributable todecreased inflammasome activity but rather to dimin-ished apoptotic behavior (23,44).

More broadly, relatively few studies havebroached the role of IL-1� in OA. In agreement withour hypothesis, intraperitoneal injections of IL-1Ra didnot affect OA scores in an experimental murine modelof OA based on knee meniscectomy (46). Interestingly,a recent clinical study corroborated these results. Che-valier et al evaluated the clinical response to a singleintraarticular injection of anakinra, a recombinantform of IL-1Ra, in patients with knee OA and observedno improvement in OA symptoms (knee pain, func-tion, stiffness, cartilage turnover) compared with pla-cebo (47).

Richette et al reported that intraarticular injec-tion of IL-1Ra “might be self-limited” in patients withknee OA (48). Indeed, a high endogenous IL-1Ra–to–IL-1� ratio was observed in knee synovial fluid frompatients with OA. This ratio was not associated with painbut may explain the reduced efficiency of intraarticularsupplementation with IL-1Ra (48). A limitation of theseclinical studies may be timing, because all approachesinvolving anti–IL-1� have addressed established OA,whereas IL-1� is more likely to be involved in early-onset OA (49).

Moreover, according to Fan et al, OA articularchondrocytes are less responsive than normal chondro-cytes to stimulation with IL-1� (50). In vitro, OAchondrocytes showed increased levels of catabolic en-zymes and mediators, but this prodegradative phenotypewas not further exacerbated by IL-1� treatment (50).Likewise, in a study determining susceptibility to IL-1�in cartilage at different anatomic locations on humanOA knee joints, Barakat et al observed that cartilagebiopsy specimens obtained from 4 of 12 patients werenot susceptible to the effects of IL-1� (51). However, theliterature contains some discrepancies relating to thishypothesis (52).

Taken together, the results of this study suggestthat OA cartilage can be degraded independently ofinflammasome activity, and that the IL-1� that is in-volved in OA cartilage degradation may not be producedby cartilage itself but rather by synovial tissue.

ACKNOWLEDGMENTS

We thank Christine Lamouroux and colleagues formaintaining and breeding the genetically modified mice. Wealso thank Prof. Jean-Michel Dayer for critically reviewing themanuscript.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising itcritically for important intellectual content, and all authors approvedthe final version to be published. Dr. Berenbaum had full access to allof the data in the study and takes responsibility for the integrity of thedata and the accuracy of the data analysis.Study conception and design. Bougault, Houard, Jacques, Berenbaum.Acquisition of data. Bougault, Gosset, Salvat, Godmann, Pap.Analysis and interpretation of data. Bougault.

REFERENCES

1. Zhang Y, Jordan JM. Epidemiology of osteoarthritis. Clin GeriatrMed 2010;26:355–69.

2. Martel-Pelletier J, Pelletier JP. Is osteoarthritis a disease involvingonly cartilage or other articular tissues? Eklem Hastalik Cerrahisi2010;21:2–14.

3. Goldring MB, Goldring SR. Osteoarthritis. J Cell Physiol 2007;213:626–34.

4. Murphy G, Nagase H. Reappraising metalloproteinases in rheu-matoid arthritis and osteoarthritis: destruction or repair? Nat ClinPract Rheumatol 2008;4:128–35.

5. Yoshida M, Tsuji M, Funasaki H, Kan I, Fujii K. Analysis forthe major contributor of collagenase to the primary cleavage oftype II collagens in cartilage degradation. Mod Rheumatol 2005;15:180–6.

6. Jacques C, Gosset M, Berenbaum F, Gabay C. The role of IL-1and IL-1Ra in joint inflammation and cartilage degradation.Vitam Horm 2006;74:371–403.

7. Kapoor M, Martel-Pelletier J, Lajeunesse D, Pelletier JP, FahmiH. Role of proinflammatory cytokines in the pathophysiology ofosteoarthritis. Nat Rev Rheumatol 2011;7:33–42.

8. Bauernfeind F, Ablasser A, Bartok E, Kim S, Schmid-Burgk J,Cavlar T, et al. Inflammasomes: current understanding and openquestions. Cell Mol Life Sci 2011;68:765–83.

9. Schroder K, Tschopp J. The inflammasomes. Cell 2010;140:821–32.

10. Davis BK, Wen H, Ting JP. The inflammasome NLRs in immu-nity, inflammation, and associated diseases. Annu Rev Immunol2011;29:707–35.

11. Schroder K, Zhou R, Tschopp J. The NLRP3 inflammasome: asensor for metabolic danger? Science 2010;327:296–300.

12. Bryant C, Fitzgerald KA. Molecular mechanisms involved ininflammasome activation. Trends Cell Biol 2009;19:455–64.

13. Stehlik C. Multiple interleukin-1�–converting enzymes contributeto inflammatory arthritis [editorial]. Arthritis Rheum 2009;60:3524–30.

14. Schonbeck U, Mach F, Libby P. Generation of biologically activeIL-1� by matrix metalloproteinases: a novel caspase-1-independentpathway of IL-1� processing. J Immunol 1998;161:3340–6.

15. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K,et al. Development of criteria for the classification and reportingof osteoarthritis: classification of osteoarthritis of the knee.Arthritis Rheum 1986;29:1039–49.

16. Martinon F, Petrilli V, Mayor A, Tardivel A, Tschopp J. Gout-associated uric acid crystals activate the NALP3 inflammasome.Nature 2006;440:237–41.

17. Gosset M, Pigenet A, Salvat C, Berenbaum F, Jacques C. Inhibi-tion of matrix metalloproteinase-3 and -13 synthesis induced by

3980 BOUGAULT ET AL

IL-1� in chondrocytes from mice lacking microsomal prostaglan-din E synthase-1. J Immunol 2010;185:6244–52.

18. Gosset M, Berenbaum F, Thirion S, Jacques C. Primary cultureand phenotyping of murine chondrocytes. Nat Protoc 2008;3:1253–60.

19. Gosset M, Berenbaum F, Levy A, Pigenet A, Thirion S, Saffar JL,et al. Prostaglandin E2 synthesis in cartilage explants undercompression: mPGES-1 is a mechanosensitive gene. Arthritis ResTher 2006;8:R135.

20. Masuko-Hongo K, Berenbaum F, Humbert L, Salvat C, GoldringMB, Thirion S. Up-regulation of microsomal prostaglandin Esynthase 1 in osteoarthritic human cartilage: critical roles of theERK-1/2 and p38 signaling pathways. Arthritis Rheum 2004;50:2829–38.

21. Houard X, Monnot C, Dive V, Corvol P, Pagano M. Vascularsmooth muscle cells efficiently activate a new proteinase cascadeinvolving plasminogen and fibronectin. J Cell Biochem 2003;88:1188–201.

22. Farndale RW, Sayers CA, Barrett AJ. A direct spectrophotometricmicroassay for sulfated glycosaminoglycans in cartilage cultures.Connect Tissue Res 1982;9:247–8.

23. Lo MY, Kim HT. Chondrocyte apoptosis induced by hydrogenperoxide requires caspase activation but not mitochondrial poretransition. J Orthop Res 2004;22:1120–5.

24. Westacott CI, Whicher JT, Barnes IC, Thompson D, Swan AJ,Dieppe PA. Synovial fluid concentration of five different cytokinesin rheumatic diseases. Ann Rheum Dis 1990;49:676–81.

25. Denoble AE, Huffman KM, Stabler TV, Kelly SJ, Hershfield MS,McDaniel GE, et al. Uric acid is a danger signal of increasing riskfor osteoarthritis through inflammasome activation. Proc NatlAcad Sci U S A 2011;108:2088–93.

26. Bertazzolo N, Punzi L, Stefani MP, Cesaro G, Pianon M, Finco B,et al. Interrelationships between interleukin (IL)-1, IL-6 and IL-8in synovial fluid of various arthropathies. Agents Actions 1994;41:90–2.

27. Fan Z, Soder S, Oehler S, Fundel K, Aigner T. Activation ofinterleukin-1 signaling cascades in normal and osteoarthritic artic-ular cartilage. Am J Pathol 2007;171:938–46.

28. Moldovan F, Pelletier JP, Jolicoeur FC, Cloutier JM, Martel-Pelletier J. Diacerhein and rhein reduce the ICE-induced IL-1�and IL-18 activation in human osteoarthritic cartilage. Osteo-arthritis Cartilage 2000;8:186–96.

29. Saha N, Moldovan F, Tardif G, Pelletier JP, Cloutier JM, Martel-Pelletier J. Interleukin-1�–converting enzyme/caspase-1 in humanosteoarthritic tissues: localization and role in the maturationof interleukin-1� and interleukin-18. Arthritis Rheum 1999;42:1577–87.

30. Campo GM, Avenoso A, Campo S, D’Ascola A, Traina P, SamaD, et al. Purified human plasma glycosaminoglycans reducedNF-�B activation, pro-inflammatory cytokine production and ap-optosis in LPS-treated chondrocytes. Innate Immun 2008;14:233–46.

31. Mathy-Hartert M, Deby-Dupont GP, Reginster JY, Ayache N,Pujol JP, Henrotin YE. Regulation by reactive oxygen species ofinterleukin-1�, nitric oxide and prostaglandin E2 production byhuman chondrocytes. Osteoarthritis Cartilage 2002;10:547–55.

32. Olee T, Hashimoto S, Quach J, Lotz M. IL-18 is produced byarticular chondrocytes and induces proinflammatory and catabolicresponses. J Immunol 1999;162:1096–100.

33. McCall SH, Sahraei M, Young AB, Worley CS, Duncan JA,Ting JP, et al. Osteoblasts express NLRP3, a nucleotide-bindingdomain and leucine-rich repeat region containing receptor impli-cated in bacterially induced cell death. J Bone Miner Res 2008;23:30–40.

34. Kolly L, Busso N, Palmer G, Talabot-Ayer D, Chobaz V, So A.Expression and function of the NALP3 inflammasome in rheuma-toid synovium. Immunology 2010;129:178–85.

35. Kolly L, Karababa M, Joosten LA, Narayan S, Salvi R, Petrilli V,et al. Inflammatory role of ASC in antigen-induced arthritis isindependent of caspase-1, NALP-3, and IPAF. J Immunol 2009;183:4003–12.

36. Rosengren S, Hoffman HM, Bugbee W, Boyle DL. Expression andregulation of cryopyrin and related proteins in rheumatoid arthri-tis synovium. Ann Rheum Dis 2005;64:708–14.

37. Tetlow LC, Adlam DJ, Woolley DE. Matrix metalloproteinase andproinflammatory cytokine production by chondrocytes of humanosteoarthritic cartilage: associations with degenerative changes.Arthritis Rheum 2001;44:585–94.

38. Gracie JA, Forsey RJ, Chan WL, Gilmour A, Leung BP,Greer MR, et al. A proinflammatory role for IL-18 in rheumatoidarthritis. J Clin Invest 1999;104:1393–401.

39. Smith MD, Triantafillou S, Parker A, Youssef PP, Coleman M.Synovial membrane inflammation and cytokine production inpatients with early osteoarthritis. J Rheumatol 1997;24:365–71.

40. Sellam J, Berenbaum F. The role of synovitis in pathophysiologyand clinical symptoms of osteoarthritis. Nat Rev Rheumatol2010;6:625–35.

41. Kingsbury SR, Conaghan PG, McDermott MF. The role of theNLRP3 inflammasome in gout. J Inflamm Res 2011;4:39–49.

42. Jin C, Frayssinet P, Pelker R, Cwirka D, Hu B, Vignery A, et al.NLRP3 inflammasome plays a critical role in the pathogenesis ofhydroxyapatite-associated arthropathy. Proc Natl Acad Sci U S A2011;108:14867–72.

43. Ippagunta SK, Brand DD, Luo J, Boyd KL, Calabrese C, StienstraR, et al. Inflammasome-independent role of apoptosis-associatedspeck-like protein containing a CARD (ASC) in T cell primingis critical for collagen-induced arthritis. J Biol Chem 2010;285:12454–62.

44. D’Lima D, Hermida J, Hashimoto S, Colwell C, Lotz M. Caspaseinhibitors reduce severity of cartilage lesions in experimentalosteoarthritis. Arthritis Rheum 2006;54:1814–21.

45. Rudolphi K, Gerwin N, Verzijl N, van der Kraan P, van denBerg W. Pralnacasan, an inhibitor of interleukin-1� convertingenzyme, reduces joint damage in two murine models of osteo-arthritis. Osteoarthritis Cartilage 2003;11:738–46.

46. Kadri A, Ea HK, Bazille C, Hannouche D, Liote F, Cohen-SolalME. Osteoprotegerin inhibits cartilage degradation through aneffect on trabecular bone in murine experimental osteoarthritis.Arthritis Rheum 2008;58:2379–86.

47. Chevalier X, Goupille P, Beaulieu AD, Burch FX, Bensen WG,Conrozier T, et al. Intraarticular injection of anakinra in osteo-arthritis of the knee: a multicenter, randomized, double-blind,placebo-controlled study. Arthritis Rheum 2009;61:344–52.

48. Richette P, Francois M, Vicaut E, Fitting C, Bardin T, Corvol M,et al. A high interleukin 1 receptor antagonist/IL-1� ratio occursnaturally in knee osteoarthritis. J Rheumatol 2008;35:1650–4.

49. Santangelo KS, Pieczarka EM, Nuovo GJ, Weisbrode SE, BertoneAL. Temporal expression and tissue distribution of interleukin-1�in two strains of guinea pigs with varying propensity for sponta-neous knee osteoarthritis. Osteoarthritis Cartilage 2011;19:439–48.

50. Fan Z, Bau B, Yang H, Soeder S, Aigner T. Freshly isolatedosteoarthritic chondrocytes are catabolically more active thannormal chondrocytes, but less responsive to catabolic stimulationwith interleukin-1�. Arthritis Rheum 2005;52:136–43.

51. Barakat AF, Elson CJ, Westacott CI. Susceptibility to physiolog-ical concentrations of IL-1� varies in cartilage at different ana-tomical locations on human osteoarthritic knee joints. Osteo-arthritis Cartilage 2002;10:264–9.

52. Martel-Pelletier J, McCollum R, DiBattista J, Faure MP, Chin JA,Fournier S, et al. The interleukin-1 receptor in normal andosteoarthritic human articular chondrocytes: identification as thetype I receptor and analysis of binding kinetics and biologicfunction. Arthritis Rheum 1992;35:530–40.

ROLE OF INFLAMMASOME IN OA CARTILAGE DEGRADATION 3981