Estrogen receptor–related receptor α regulation by interleukin-1β in prostaglandin E2– and cAMP-dependent pathways in osteoarthritic chondrocytes

Authors

  • Edith Bonnelye,

    Corresponding author
    1. INSERM U664, Lyon, France
    • INSERM U664, Faculté de Médecine RTH Laennec, Rue Guillaume Paradin, 69372 Lyon Cedex 08, France
    Search for more papers by this author
    • Drs. Bonnelye and Reboul contributed equally to this work.

  • Pascal Reboul,

    1. Centre de Recherche du Centre Hospitalier de l'Université de Montréal, Montréal, Québec, Canada
    Current affiliation:
    1. UMR 7561, CNRS–Université Henri Poincaré-Nancy 1, Vandoeuvre lès Nancy, France
    Search for more papers by this author
    • Drs. Bonnelye and Reboul contributed equally to this work.

  • Nicolas Duval,

    1. Pavillon des Charmilles, Vimont, Québec, Canada
    Search for more papers by this author
  • Marco Cardelli,

    1. University of Toronto and Princess Margaret Hospital, Toronto, Ontario, Canada
    Search for more papers by this author
  • Jane E. Aubin

    Corresponding author
    1. University of Toronto and Princess Margaret Hospital, Toronto, Ontario, Canada
    • Department of Molecular Genetics, Faculty of Medicine, University of Toronto, Room 4245 Medical Sciences Building, 1 King's College Circle, Toronto, Ontario M5S 1A8, Canada
    Search for more papers by this author

Abstract

Objective

We reported previously that the orphan nuclear receptor, estrogen receptor–related receptor α (ERRα), is expressed in articular chondrocytes and is dysregulated in a mouse model of inflammatory arthritis. The aim of this study, therefore, was to determine whether ERRα is also dysregulated in patients with osteoarthritis (OA).

Methods

ERRα messenger RNA (mRNA) and protein were quantified in normal and OA cartilage samples and in OA chondrocytes in vitro, with and without short-term treatment with a variety of OA-associated factors and signaling pathway agonists and inhibitors.

Results

ERRα expression was lower in OA than in normal articular cartilage. Interleukin-1β (IL-1β) markedly up-regulated ERRα expression in OA chondrocytes in vitro, and agonist or inhibitor treatment indicated that the up-regulation was dependent on cyclooxygenase 2 (COX-2; NS398), prostaglandin E2, cAMP (8-bromo-cAMP), and protein kinase A (PKA; KT5720). Treatment with the ERRα inverse agonist XCT790 decreased the expression of SOX9 and the up-regulation of ERRα by IL-1β, suggesting autoregulation of ERRα in the IL-1β pathway. Matrix metalloproteinase 13 (MMP-13) expression was also decreased by treatment with XCT790 plus IL-1β versus IL-1β alone, and the down-regulation of MMP-13 mRNA and protein observed with XCT790 alone suggests that the up-regulation of MMP-13 by IL-1β is ERRα-dependent.

Conclusion

We report the first evidence that ERRα expression is regulated by IL-1β in COX-2–, cAMP-, and PKA-dependent pathways in OA chondrocytes. We confirmed that SOX9 is an ERRα target gene in human, as in rodent, chondrocytes and identified MMP-13 as a potential new target gene, which suggests that ERRα may both respond to the healing signal and contribute to extracellular degradation in OA cartilage.

Osteoarthritis (OA) is a chronic disease characterized by slowly progressive destruction of the articular cartilage, combined with changes in the synovium and subchondral bone (1). Various estimates suggest that more than 40% of 70-year-old people currently have the disease, ranking OA as the most common arthritic condition. The prevalence and the pain and disability associated with progression of the disease have led to intense interest in defining markers for OA as well as the mechanisms underlying the disease.

The cellular component of cartilage is the chondrocyte, and adult articular chondrocytes, although considered quiescent in normal cartilage, can respond to mechanical injury and biologic stimuli such as cytokines and growth factors. Interleukin-1β (IL-1β) is a well-known proinflammatory cytokine that is implicated in cartilage degradation, in part via its ability to stimulate the expression of metalloproteinases and reduce cartilage-specific molecules, such as type II collagen (2). IL-1β also stimulates prostaglandin E2 (PGE2) production by up-regulation of cyclooxygenase 2 (COX-2), which may worsen symptoms of OA (3, 4). On the other hand, stimulation of PGE2 production by IL-1β is also reported to up-regulate type II collagen transcription (5), making the net effect of changes in PGE2 production complex to determine.

Nuclear steroid receptors are transcription factors that comprise both ligand-dependent molecules, such as estrogen receptors (ERs), and a large number of so-called orphan receptors, for which no ligands have yet been determined (6). Three orphan receptors, estrogen receptor–related receptor α (ERRα), ERRβ, and ERRγ (or, NR3B1, NR3B2, and NR3B3, respectively, according to the Nuclear Receptors Nomenclature Committee, 1999) share similarities with ERα and ERβ (or, NR3A1 and NR3A2, respectively) (7, 8), but they do not bind estrogen (9). Sequence alignment of ERRα and the ERs reveals a high similarity (68%) in the 66 amino acids of the DNA-binding domain, but only a moderate similarity (36%) in the ligand-binding domain, which may explain the fact that ERRα recognizes the same DNA-binding elements as the ERs, but yet it does not bind estrogen (10).

Although not regulated by natural ligands, ERRα can be deactivated by the synthetic molecule XCT790 (11, 12). ERRα regulates fatty acid oxidation and the adaptive bioenergetic response (13, 14). ERRα is highly expressed in skeletal (bone and cartilage) tissues (15, 16) and has been reported to regulate osteoblast development and bone formation both in vitro (15, 16) and in vivo (17, 18). ERRα expression in fetal and adult articular chondrocytes suggests that it may be involved not only in cartilage formation, but also in its maintenance and integrity throughout the lifetime of the tissue. Consistent with this hypothesis, we showed previously that ERRα appears to have an anabolic function in chondrogenesis; that is, it positively regulates SOX9 expression in chondrocytes in vitro (16). This is consistent with our findings that ERRα expression is decreased in both the joints and subchondral bone in a murine model of inflammatory arthritis (19).

Given the data summarized above, we sought to determine whether ERRα is also dysregulated in chondrocytes in human OA and whether known proinflammatory cytokines and their associated signaling pathways are involved.

MATERIALS AND METHODS

Reagents.

Recombinant human IL-1β was obtained from R&D Systems. PGE2 was obtained from EMD. Wortmannin (phosphatidylinositol 3-kinase [PI3K] inhibitor) and the selective COX-2 inhibitor NS398 were from Calbiochem. KT5720 (PKA inhibitor) was from Upstate Biotechnology and XCT790 from Sigma.

Collection of human cartilage samples.

Femoral condyles and tibial plateaus were obtained from 46 OA patients (32 women and 14 men with a mean ± SD age of 60 ± 27 years) following total knee arthroplasty and were used for analysis of messenger RNA (mRNA) and cell cultures after chondrocyte extraction. All patients were evaluated by a certified rheumatologist and, based on the criteria developed by the American College of Rheumatology (20) were diagnosed as having OA. Ten normal cartilage samples from cadaver donors (5 women and 5 men with a mean ± SD age of 52 ± 18 years) and 10 OA cartilage samples (7 women and 3 men with a mean ± SD age of 68 ± 11 years) taken from the knee joints were also obtained for direct RNA extraction from cartilage. Specimen collection and all procedures were approved by the Ethics Committee of the University of Montreal Hospital Center and by the Office of Research Ethics, University of Toronto.

Human chondrocyte cultures.

Chondrocytes were released from the articular cartilage by sequential enzymatic digestion at 37°C, as previously described (21, 22). OA chondrocytes obtained at first passage were seeded at 1 × 105 cells/cm2 in 24- or 12-well plates in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) containing 10% fetal bovine serum (FBS; Invitrogen) and an antibiotic mixture (100 units/ml of penicillin, 100 μg/ml of streptomycin; Invitrogen) and were cultured for 48 hours at 37°C in a humidified atmosphere of 5% CO2/95% air. The medium was then replaced with DMEM containing 0.5% FBS and culture continued for 24 hours. After this period, cells were incubated for 24–96 hours in fresh medium containing 0.5% FBS, in the presence or absence of recombinant human IL-1β (100 pg/ml), with or without other factors as designated below (PGE2 [50 nM, 500 nM], wortmannin [250 nM], NS398 [10 μM], KT5720 [2 μM], or XCT790 [0.5–5 μM]).

Immunoblotting.

Cell proteins were extracted in 0.5% sodium dodecyl sulfate (SDS), separated in 4–12% SDS–polyacrylamide gel electrophoresis gels, then transferred to nitrocellulose membranes (Bio-Rad) according to established methods (23). Immunodetection was performed as previously described (21), with a rabbit monoclonal antibody against ERRα (Epitomics) at a dilution of 1:1,000 and the secondary antibody (horseradish peroxidase [HRP]–conjugated goat anti-rabbit IgG) at a dilution of 1:20,000 (Pierce). For evaluating protein loading, a rabbit polyclonal antibody against GAPDH (Sigma-Aldrich) was used at a dilution of 1:20,000; secondary antibody was used at a dilution of 1:20,000 (Pierce). An enhanced chemiluminescence kit (Santa Cruz Biotechnology) was used for detection.

Immunocytochemistry.

Human OA cells were fixed in culture wells with 3.7% paraformaldehyde (Sigma) in PBS for 10 minutes, permeabilized with 0.2% Triton X-100 in PBS for 7 minutes, and then treated with a peroxidase blocking reagent (Dako). Cells were incubated overnight at 4°C with goat polyclonal antibody against human ERRα (Santa Cruz Biotechnology) at a dilution of 1:60, washed with PBS, and then incubated for 1 hour at room temperature with HRP-conjugated donkey anti-goat IgG (Santa Cruz Biotechnology) at a dilution of 1:300. Secondary antibody was used as control. After washing, cells were treated with 3,3′-diaminobenzidine (Dako). Counterstaining was performed with Mayer's hematoxylin (Merck).

Enzyme-linked immunosorbent assay (ELISA).

Cell culture media were collected from first-passage cultured OA chondrocytes. Levels of matrix metalloproteinase 13 (MMP-13) were determined with a specific MMP-13 ELISA (Quantikine human proMMP-13 immunoassay; R&D Systems) according to the manufacturer's protocol. This assay has a mean minimum detection limit of 7.7 pg/ml for proMMP-13.

Real time reverse transcription-polymerase chain reaction (RT-PCR).

RNA extraction from cartilage.

Human articular cartilage was homogenized on ice in TRIzol reagent (Invitrogen), followed by the addition of chloroform, as described previously (24). After being shaken and cooled at 4°C for 1 hour, the solution was centrifuged at 12,000g for 2 hours at 4°C. The aqueous phase was transferred to a new tube and precipitated overnight at –20°C with 1 volume of isopropyl alcohol. After centrifugation (12,000g for 30 minutes at 4°C), the pellet was resuspended in the Qiagen lysis buffer from the RNeasy Plant Minikit and extracted according to the manufacturer's protocol. RNA was quantified using RiboGreen RNA reagent (Molecular Probes).

RNA extraction from chondrocytes.

Total RNA was extracted using TRIzol reagent according to the manufacturer's protocol and treated with DNA-free DNase (Ambion). RNA was quantified using RiboGreen RNA reagent.

Samples of total RNA (1 μg) were reverse-transcribed using random hexamer (Amersham) and a Superscript II first-strand synthesis kit (Invitrogen). Real-time PCR was performed on a MyiQ Optical Module (Bio-Rad) with primers specific for human L32 (101 bp; 5′-CAAGGAGCTGGAAGTGCTGC-3′ and 5′-CAGCTCTTTCCACGATGGCT-3′), S29 (62 bp; 5′-TTCAAACCGGCACGGTCTGA-3′ and 5′-GACGGAAACACTGGCGGCAC-3′), ERRα (101 bp; 5′-ACCGAGAGATTGTGGTCACCA-3′ and 5′-CATCCACACGCTCTGCAGTACT-3′), ERα (330 bp; 5′-TCCTCTCCCACATCAGGC-3′ and 5′-TTGGCTAAAGTGGTGCATGATGAGG-3′), ERβ (125 bp; 5′-GAATATCTCTGTGTCGTCAAGGC-3′ and 5′-CCAAAGCATCGGTCACGGCG-3′), ERRβ (110 bp; 5′-GGCCGTCAGAAATACAAGCG-3′ and 5′-GCCACCAGTAGGTATGAGACAATCTT-3′), type II collagen α-chain (116 bp; 5′-GATCCCCTTCGGAGAGTGCT-3′ and 5′-GTCCTACAATATCCTTGATGTCTCCA-3′), SOX9 (101 bp; 5′-ACTTGTAATCCGGGTGGTCCT-3′ and 5′-CAAGCTCTGGAGACTTCTGAACG-3′), MMP-13 (314 bp; 5′-CTTAGAGGTGACTGGCAAAC-3′ and 5′-GCCCATCAAATGGGTAGAAG-3′), and COX-2 (98 bp; 5′-GCTGGAACATGGAATTACCCA-3′ and 5′-CTTTCTGTACTGCGGGTGGAA-3′).

PCR was performed using iQ SYBR Green Supermix according to the manufacturer's instructions (Bio-Rad), with an initial step of 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C and 1 minute at 59°C. Using the iQ5 software, we verified that a single peak was obtained for each product. Data analysis was carried out with the iQ5 software using the comparative Ct method. Expression of mRNA for the markers of interest was normalized to that of ribosomal protein L32 or S29.

Statistical analysis.

Data are expressed as the mean ± SEM and were analyzed statistically by one-way analysis of variance, followed by post hoc t-tests to assess the differences between groups. P values less than 0.05 were considered significant. The total number of specimens examined varied from 5 to 10, depending on the experiment.

RESULTS

Expression of ERRα mRNA in human chondrocytes and dysregulation in OA.

Expression of ERRα mRNA was lower in human OA articular cartilage than in normal articular cartilage (Figure 1A). Interestingly, the reduction was statistically significant in women, but not in men (Figure 1A), probably because of the low number of male patients sampled. ERRα protein was also expressed in OA chondrocytes and was localized primarily to the cytoplasm, with a subset of cells also showing prominent nuclear labeling (Figure 1B; see also Western blots illustrated below). Notably, ERRα, SOX9, COX-2, and MMP-13 expression levels were not significantly different in OA cartilage versus OA chondrocytes in vitro (Figure 1C), suggesting that the OA cartilage samples were not in an inflammatory phase at the time they were obtained.

Figure 1.

A, Expression of estrogen receptor–related receptor α (ERRα) mRNA in articular cartilage obtained from 10 normal cadaver donors and 10 patients with osteoarthritis (OA). Real-time reverse transcription–polymerase chain reaction (RT-PCR) analysis showed lower expression of ERRα in OA patients compared to normal donors. Results are also shown separately for male and female subjects. B, Localization of ERRα by immunocytochemistry in both the nucleus (arrows) and the cytoplasm of OA chondrocytes. Inset shows secondary antibody control (Ct). Original magnification × 40. C, Comparison of expression levels (ΔCt as determined by RT-PCR) of ERRα, SOX9, matrix metalloproteinase 13 (MMP-13), and cyclooxygenase 2 (COX-2) between OA cartilage (ca) (in vivo) and OA chondrocytes (ch) (untreated in vitro). Each gene in each case was normalized against the expression of mRNA for S29. Expression levels of ERRα, SOX9, MMP-13, and COX-2 were not significantly different in vitro and in vivo, suggesting a good correlation between the two models. The Ct values for ERRα, SOX9, MMP-13, COX-2, and S29 before normalization were 25, 19, 21, 29, and 23, respectively. D, ERRα and SOX9 expression in OA chondrocytes. OA chondrocytes were left untreated (control) or were treated for 24 hours with interleukin-1β (IL-1β) (n = 8), hepatocyte growth factor (HGF) (n = 7), or transforming growth factor β (TGFβ) (n = 3), and expression was determined by RT-PCR. ERRα and SOX9 expression was up-regulated by IL-1β (mean ± SEM 4.07 ± 0.9 and 1.7 ± 0.6, respectively); treatment with TGFβ or HGF had no effect. In A, C, and D, expression was normalized to ribosomal protein L32. Each data point represents a single subject; horizontal lines show the mean. P values in D are versus control.

Up-regulation of ERRα mRNA expression by IL-1β in human OA chondrocytes.

To determine whether factors implicated in OA regulate ERRα expression in OA cells, human OA chondrocytes were treated for 24 hours with IL-1β, hepatocyte growth factor (HGF), or transforming growth factor β (TGFβ). ERRα expression was increased by IL-1β treatment, but was not affected by HGF or TGFβ treatment (Figure 1D). Similar to the findings for ERRα, SOX9 was also up-regulated by IL-1β, but was not affected by HGF or TGFβ (Figure 1D).

ERRα up-regulation by IL-1β was also seen after 96 hours, when chondrocytes were stimulated twice (at time 0 and at 48 hours), but not after only 1 treatment (at time 0) (Figure 2B). The lack of ERRα up-regulation by IL-1β before 24 hours suggests a transcriptional mechanism for up-regulation (data available upon request from the author). ERRβ mRNA was only slightly up-regulated as compared to ERRα (Figures 2A and B). ERα and ERβ expression were not regulated by IL-1β treatment (Figure 2C). As controls for IL-1β efficacy, the expression of MMP-13 and Col2a1 was also assessed. As expected, MMP-13 expression was dramatically increased and Col2a1 decreased following all IL-1β treatments (Figure 2D).

Figure 2.

IL-1β–induced up-regulation of ERRα in OA chondrocytes. Chondrocytes from OA patients (n = 8) were cultured and treated with IL-1β for 96 hours (1 treatment [at time 0] or 2 treatments [at time 0 and at 48 hours]). A, Western blots show representative amplimers in agarose gels. B, RNA was extracted after 96 hours, and real-time RT-PCR was performed as described in Figure 1D. ERRα expression was regulated by IL-1β (mean ± SEM 4.06 ± 1.07 after 2 treatments). ERRβ was also regulated by IL-1β, but to a lesser extent than ERRα (mean ± SEM 1.3 ± 0.6 after 2 treatments). C, The expression of estrogen receptor α (ERα) and ERβ was not affected by IL-1β. D, As expected, the expression of Col2a1 was decreased (mean ± SEM 0.099 ± 0.07 and 0.025 ± 0.32, after 1 and 2 treatments, respectively) and MMP-13 was increased (mean ± SEM 16.07 ± 8 and 16.48 ± 6.9, after 1 and 2 treatments, respectively) by treatment with IL-1β. In B–D, expression was normalized to ribosomal protein L32. Each data point represents a single subject; horizontal lines show the mean. P values are versus control (Ct). See Figure 1 for other definitions.

COX-2–dependent up-regulation of ERRα mRNA by IL-1β.

Given the robust regulation of ERRα by IL-1β, we next sought to determine which IL-1β–induced signaling pathways regulate ERRα expression. Human OA chondrocytes from several patients were first treated with NS398, an inhibitor of COX-2. While no regulation was observed with NS398 alone, the up-regulation of ERRα induced by IL-1β was blocked by NS398, suggesting that IL-1β–induced ERRα up-regulation is COX-2 dependent (Figure 3A). Consistent with this idea, analysis over a time course from 30 minutes to 24 hours showed that COX-2 expression peaked at 4 hours after IL-1β treatment, while ERRα expression changed later, peaking at 20 hours (data available upon request from the author). Immunoblotting for ERRα protein confirmed ERRα mRNA regulation after IL-1β and NS398 treatments in cells from 2 representative OA patients (see below). On the other hand, the slight up-regulation of ERRβ induced by IL-1β was NS398-independent (Figure 3A). Neither ERα nor ERβ expression was affected by NS398 treatment (Figure 3B).

Figure 3.

Dependence of IL-1β–induced up-regulation of ERRα on COX-2. Chondrocytes from OA patients were treated for 24 hours with IL-1β (n = 8 patients), with or without the selective COX-2 inhibitor NS398 (n = 6 patients). RNA was extracted after 24 hours, and real-time RT-PCR was performed as described in Figure 1D. A, IL-1β increased ERRα in a COX-2–dependent manner (mean ± SEM 1.8 ± 0.6 after IL-1β treatment alone). NS398 treatment did not alter the effects of IL-1β on ERRβ. B, NS398 treatment also did not alter the effects of IL-1β on estrogen receptor α (ERα) or ERβ. Expression was normalized to ribosomal protein L32. Each data point represents a single subject; horizontal lines show the mean. P value is versus control (Ct). See Figure 1 for other definitions.

PGE2 regulation of ERRα in OA chondrocytes.

Given the role of COX-2 in prostaglandin synthesis, we next asked whether PGE2 up-regulated ERRα in OA chondrocytes in a concentration-dependent (50 or 500 nM) and/or a time-dependent (24 hours or 96 hours) manner. We found that PGE2 up-regulated ERRα at 24 hours, but not at 96 hours (Figure 4A), suggesting that the increase in ERRα induced by IL-1β at 24 hours (Figures 1 and 3) is mediated by PGE2. Immunoblotting confirmed that the ERRα protein level was also increased after 24 hours of PGE2 treatment (Figure 4C). Similar to the findings for ERRα, SOX9 was up-regulated by PGE2 treatment at 24 hours (Figure 4B), but it was also up-regulated at 96 hours by 2 treatments with high concentrations of PGE2, suggesting that this latter regulation is ERRα-independent (Figure 4B).

Figure 4.

A, Chondrocytes isolated from OA patients (n = 7) were cultured and treated with prostaglandin E2 (PGE2) for 24 hours (1 treatment at time 0) or 96 hours (1 treatment [at time 0] or 2 treatments [at time 0 and at 48 hours]). RNA was extracted, and real-time RT-PCR was performed as described in Figure 1D. A, ERRα was up-regulated by PGE2 after 24 hours (mean ± SEM 1.53 ± 0.4 at 50 nM and 1.53 ± 0.29 at 500 nM). B, SOX9 was also up-regulated by PGE2 after 24 hours (3.06 ± 0.88 at 50 nM) and after 96 hours (3.1 ± 1.2 at 500 nM). Expression was normalized to ribosomal protein L32. Each data point represents a single subject; horizontal lines show the mean. P values are versus control (Ct). C, Chondrocytes isolated from OA patients were cultured and treated with PGE2 for 24 hours. Western blots were performed with ERRα and GAPDH antibodies. Results are from 2 representative patients, confirming the regulation of ERRα expression by PGE2. See Figure 1 for other definitions.

PKA- and cAMP-dependent up-regulation of ERRα by IL-1β.

To determine the signaling pathways involved in IL-1β–induced ERRα expression, chondrocytes were preincubated for 1 hour with kinase inhibitors, such as wortmannin (PI3K inhibitor) and KT5720 (PKA inhibitor), or with cAMP followed by the addition of IL-1β for 24 hours. As expected, ERRα expression was increased by IL-1β treatment (Figure 5A, lane 1 versus lane 2), an up-regulation that was decreased by NS398, as expected (lane 1 versus lanes 2 and 3 in Figure 5A), but up-regulated when PGE2 was also added (lane 2 versus lanes 3 and 4 in Figure 5A).

Figure 5.

A, OA chondrocytes were treated for 24 hours with IL-1β (n = 7), NS398 (n = 8), prostaglandin E2 (PGE2) (n = 7), wortmannin (n = 8), 8-bromo-cAMP (n = 8), or KT5720 (n = 7). The up-regulation of ERRα by IL-1β (lane 2; mean ± SEM 3.16 ± 0.9) is PGE2- and COX-2–dependent (compare lane 2 with lanes 3 and 4; mean ± SEM 1.31 ± 1 and 2.1 ± 1.3, respectively) and cAMP-dependent (compare lane 2 with lane 5; mean ± SEM 3.4 ± 2). Inhibition of protein kinase A (PKA) and phosphatidylinositol 3-kinase decreased the up-regulation of ERRα induced by IL-1β (compare lane 2 with lanes 6 and 7), but only PKA reached statistical significance (1.36 ± 1). B, SOX9 was also up-regulated after IL-1β treatment (compare lane 1 with lane 2), an effect that was blocked by NS398 (compare lane 2 with lane 3) but was not restored by any treatment, including PGE2, wortmannin, 8-bromo-cAMP, or KT5720. Expression was normalized to ribosomal protein L32. Each data point represents a single subject; horizontal lines show the mean. C, Western blots of chondrocytes from 2 representative patients confirm the up-regulation of ERRα expression by IL-1β via COX-2– and cAMP/PKA-dependent pathways. GAPDH was used as a loading control. See Figure 1 for other definitions.

Since cAMP is one mediator of PGE2 that is able to activate PKA, we next added 8-bromo-cAMP to IL-1β plus NS398–treated cells, which rescued ERRα levels to those seen with IL-1β alone (Figure 5A, lane 2 versus lane 5), implicating cAMP in the COX-2/PGE2 pathway for IL-1β–induced ERRα expression. Moreover, treatment with KT5720 also induced a down-regulation of ERRα in the presence of IL-1β, which also implicates PKA in the IL-1β–induced expression of ERRα (lane 2 versus lane 6 in Figure 5A). The nonsignificant trend toward reduced ERRα expression after wortmannin treatment suggests that IL-1β–induced ERRα expression is PI3K-independent (lane 2 versus lane 7 in Figure 5A).

By comparison, SOX9 expression was stimulated as expected by IL-1β (Figure 5B, lane 1 versus lane 2), an effect that was blocked by NS398 and was not rescued by NS398 cotreatment with PGE2 or 8-bromo-cAMP (lane 3 versus lanes 4 and 5 in Figure 5B). KT5720 and wortmannin had no effect. Immunoblotting confirmed that not only ERRα mRNA expression, but also ERRα protein expression, was increased after IL-1β treatment, an effect that was blocked by NS398 and KT5720 (Figure 5C).

Repression of SOX9 and MMP-13 expression after XCT790 treatment in human OA chondrocytes.

XCT790 is a synthetic reverse agonist of ERRα. Consistent with our previous results identifying SOX9 as an ERRα target gene in rodent cells (17), SOX9 was dose-dependently down-regulated in OA chondrocytes after 24 hours of XCT790 treatment (Figure 6A). Interestingly, XCT790 dose-dependently decreased the IL-1β–induced ERRα expression at both the mRNA and protein levels (Figure 6B), suggesting an autoregulation of ERRα in the IL-1β pathway. IL-1β–induced MMP-13 mRNA and protein levels were also decreased by XCT790 treatment (Figure 6C), suggesting that ERRα directly regulates MMP-13. Quantification of MMP-13 mRNA by real-time PCR and of protein by ELISA confirmed the dose-dependent decrease in MMP-13 after XCT790 treatment (Figure 6D), suggesting that MMP-13 is a target gene of ERRα in human OA chondrocytes.

Figure 6.

A, OA chondrocytes were treated for 24 hours with XCT790 (n = 6). As expected, SOX9 expression was decreased (mean ± SEM 0.38 ± 0.17–fold) after XCT790 treatment at a concentration of 5 × 10–6M. B, IL-1β–treated chondrocytes were incubated with XCT790, and showed a decrease in ERRα by RT-PCR (mean ± SEM 4.03 ± 2.67–fold with IL-1β alone and 2.4 ± 0.8–fold with IL-1β plus 10–6M XCT790), as well as by Western blotting. GAPDH was used as a loading control in the Western blot. C, IL-1β–treated chondrocytes incubated with XCT790 also showed a decrease in MMP-13 mRNA expression by RT-PCR (mean ± SEM 39.52 ± 17–fold with IL-1β alone and 15 ± 6–fold with IL-1β plus 10–6M XCT790), as well as MMP-13 protein expression by enzyme-linked immunosorbent assay (ELISA) (mean ± SEM 65 ± 14 pg/μg of protein in controls, 724 ± 143 pg/μg with IL-1β alone, and 103 ± 37 pg/μg with IL-1β plus 5 × 10–6M XCT790). D, MMP-13 expression was also decreased after XCT790 treatment alone both at the mRNA level, as determined by RT-PCR (mean ± SEM 0.44 ± 0.2–fold with 10–6M XCT790 and 0.3 ± 0.25–fold with 5 × 10–6M XCT790), and at the protein level, as determined by ELISA (mean ± SEM 65 ± 14 pg/μg of protein in controls and 37 ± 7 pg/μg with 10–6M XCT790). Expression was normalized to ribosomal protein L32. Each data point represents a single subject; horizontal lines show the mean. A schematic representation summarizing the regulation of ERRα by IL-1β in human OA cells in vitro is also shown. PGE2 = prostaglandin E2; PKA = protein kinase A. See Figure 1 for other definitions.

DISCUSSION

We report here that ERRα mRNA is expressed in human chondrocytes, is down-regulated in OA versus normal human cartilage, and is up-regulated by IL-1β through PGE2-, cAMP-, and PKA-dependant pathways in OA chondrocytes in vitro. These results with human chondrocytes are consistent with our previous observations that ERRα expression in the joints of susceptible DBA/1 male mice with collagen-induced inflammatory arthritis is decreased (19). We also report that XCT790, the synthetic inverse agonist of ERRα, down-regulates SOX9 in OA chondrocytes in vitro, confirming our previous data describing SOX9, a transcription factor required in chondrogenesis, as an ERRα target gene (16).

Indeed, we have previously reported that ERRα is highly expressed in proliferative chondrocytes during cartilage formation and that modulating ERRα expression in a chondrogenic cell line accelerates the maturation of proliferative chondrocytes to hypertrophy, effects paralleling SOX9 function in cartilage (16, 25). The fact that the ERRα inverse agonist XCT790 decreases SOX9 expression in OA chondrocytes is also of interest because XCT790 is known to disrupt the interaction between ERRα and the transcriptional coactivator, peroxisome proliferator–activated receptor γ coactivator 1α (PGC-1α) (12). PGC-1α exhibits differential expression during chondrocyte differentiation and modulates SOX9 activity (26). Taken together, our results support the hypothesis that ERRα and PGC-1α (26) function together not only in the formation of cartilage, but also in its maintenance and integrity.

Our data also support the view that ERRα plays a role in cartilage homeostasis and may also play complex and multifactorial roles in the response of OA chondrocytes. Indeed, the fact that ERRα expression is up-regulated by the proinflammatory mediators IL-1β and PGE2, but not by HGF and TGFβ, the latter being associated with mild cartilage remodeling, suggests that ERRα plays a role in cartilage degradation (2, 27). IL-1β is a proinflammatory cytokine produced endogenously not only by articular chondrocytes (28), but also by the synovial membrane in OA (29). IL-1β stimulates several factors, including PGE2, through COX-2 regulation in human articular chondrocytes (30–32). Although 4 different PGE2 receptors (EPs 1–4) have been identified, human chondrocytes express primarily the EP2 and EP4 isoforms. Which of these plays a predominant role in OA is still a subject of controversy (33, 34), but both activate the cAMP/PKA pathway (35), stimulating a vicious circle in which PGE2 regulates the expression of cytokines, including IL-1β and IL-6 (3, 36).

We found that ERRα is also regulated by IL-1β through the PGE2/COX-2/cAMP/PKA pathway. Time course experiments confirmed that an IL-1β–induced increase in ERRα expression was first detectable 10 hours after the peak of COX-2 (4 hours) and reached its maximum at 20 hours, indicating that regulation of ERRα by IL-1β is a COX-2–dependent transcription process. Very recently, PGE2 was shown to regulate aromatase expression through ERRα in prostate stromal cells (37). The fact that the ERRα is regulated by cAMP is consistent with the modulation of ERRα observed in cells under adipogenic differentiation conditions, which include an inducer of cAMP, glucocorticoid, and insulin (38). Moreover, cAMP has been reported to increase ERRα protein and its transcriptional activity in lung type II cells on surfactant protein A expression, a factor that is also stimulated by IL-1β (39). Indeed, endogenously produced prostaglandins play an important role in promoting type II cell differentiation and induce surfactant protein A in cultured human fetal lung tissue (40).

ERRα regulation by the PGE2/cAMP/PKA axis may therefore be a common, if not ubiquitous, process in human tissues. In any case, taken together, the data suggest that ERRα is involved in an inflammation/cartilage degradation pathway in OA, at least as an early response to IL-1β.

The fact that XCT790, under both basal and IL-1β–induced conditions, decreases the expression of MMP-13, a metalloproteinase involved in cartilage destruction (41, 42), further supports the view that ERRα plays a role in a cartilage-degradation pathway in OA. To our knowledge, this is the first report of a metalloproteinase gene as an ERRα target gene. Nevertheless, it is important to note that ERRα expression is decreased in OA versus normal cartilage and that its up-regulation by IL-1β in OA chondrocytes is time-dependent and lost by 96 hours, suggesting an early-degradative pathway response of ERRα that may not be maintained later during the process.

The fact that ERRα expression level was the same in OA cartilage versus OA chondrocytes in vitro, as was expression of SOX9, MMP-13, and COX-2 relative to ERRα, suggests that the OA cartilage samples were not in an inflammatory phase or that the ERRα response to inflammation had returned to baseline at the time the samples were obtained. Of course, we cannot exclude the possibility that OA chondrocytes in vitro do not fully recapitulate the phenotype of OA cartilage chondrocytes and matrix. It is also worth noting that MMP-13 ablation has recently been shown to inhibit several principal regulators of chondrocyte differentiation, including 3 ERRα target genes SOX9, vascular endothelial growth factor, and runt-related transcription factor 2, suggesting that MMP-13 is not only implicated in cartilage matrix degradation but is also involved in primary human articular chondrocyte differentiation (43). Taken together, the data suggest a dual function of ERRα in OA: in cartilage matrix remodeling (via SOX9 and, potentially, PGE2 function in bone) and in cartilage degradation (via IL-1β and MMP-13) (Figure 6D).

ERRα is regulated by estrogen in many tissues, including bone, and ERα binds the ERRα promoter and activates its expression (44, 45). Accumulating evidence supports a role of estrogen in adult joint tissues but, similar to ERRα, estrogen may have dual effects (46, 47). For example, reports of the efficacy of estrogen replacement therapy in restoring cartilage in joint tissues in patients with OA are mixed (48, 49), possibly reflecting both the proinflammatory and antiinflammatory effects attributed to estrogen (47, 50). Furthermore, circulating IL-1β levels have been shown to be increased after menopause, and estrogen modulates IL-1β-induced proteoglycan degradation and de novo expression of MMPs 1, 3, and 13 in chondrocytes (47), activities that parallel ERRα regulation of MMP-13.

In summary, ERRα is expressed in human chondrocytes, is down-regulated in OA versus normal cartilage, and is up-regulated by 2 proinflammatory factors, IL-1β and PGE2, through cAMP- and PKA-dependent pathways. It seems likely, however, that ERRα plays dual roles in OA chondrocytes, as evidenced by the time-dependent effects of IL-1β on ERRα expression and the ability of IL-1β to concomitantly increase the levels of SOX9 and MMP-13 (prochondrogenic and cartilage-degradative factors, respectively), and ERRα target genes. Further understanding of the function of ERRα, either alone or in combination with ERs, in normal or OA adult chondrocytes is required.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Drs. Bonnelye and Aubin had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Bonnelye, Reboul, Aubin.

Acquisition of data. Bonnelye, Reboul, Duval, Cardelli.

Analysis and interpretation of data. Bonnelye, Reboul, Duval, Cardelli, Aubin.

Acknowledgements

We are very grateful to Usha Bhargava and Ruolin Guo for their technical help.

Ancillary