Objectives Basic calcium phosphate (BCP) crystal and interleukin 6 (IL-6) have been implicated in osteoarthritis (OA). We hypothesise that these two factors may be linked in a reciprocal amplification loop which leads to OA.
Methods Primary murine chondrocytes and human cartilage explants were incubated with hydroxyapatite (HA) crystals, a form of BCP, and the modulation of cytokines and matrix-degrading enzymes assayed. The ability of IL-6 to stimulate chondrocyte calcification was assessed in vitro. The mechanisms underlying the effects of HA on chondrocytes were investigated using chemical inhibitors, and the pathways mediating IL-6-induced calcification characterised by quantifying the expression of genes involved in chondrocyte mineralisation. The role of calcification in vivo was studied in the meniscectomy model of murine OA (MNX), and the link between IL-6 and cartilage degradation investigated by histology.
Results In chondrocytes, BCP crystals stimulated IL-6 secretion, further amplified in an autocrine loop, through signalling pathways involving Syk and PI3 kinases, Jak2 and Stat3 molecules. Exogenous IL-6 promoted calcium-containing crystal formation and upregulation of genes involved in calcification: the pyrophosphate channel Ank, the calcium channel Annexin5 and the sodium/phosphate cotransporter Pit-1. Treatment of chondrocytes with IL-6 inhibitors significantly inhibited IL-6-induced crystal formation. In meniscectomised mice, increasing deposits of BCP crystals were observed around the joint and correlated with cartilage degradation and IL-6 expression. Finally, BCP crystals induced proteoglycan loss and IL-6 expression in human cartilage explants, which were reduced by an IL-6 inhibitor.
Conclusions BCP crystals and IL-6 form a positive feedback loop leading to OA. Targeting calcium-containing crystal formation and/or IL-6 are promising therapeutic strategies in OA.
Statistics from Altmetric.com
Osteoarthritis (OA) is the most common form of chronic arthropathy and a leading cause of pain and disability. Biomechanical factors, joint trauma, age, gender and obesity have been identified as risk factors for OA development. OA is characterised by cartilage degradation, subchondral bone changes and mild synovitis. Although OA is not considered an inflammatory disease, several cytokines have catabolic effects on cartilage matrix.1 However, it remains unclear if there is a common link that unites mechanical and inflammatory mechanisms in OA pathogenesis.
A common denominator that potentially integrates these mechanisms could be articular calcium-containing crystals. Crystal deposits were identified in 50% of synovial fluids2 and in 100% of cartilage obtained during joint replacement3 and correlated with the severity of radiographic and histological OA.4 Calcium crystals were also found in some normal joints5 ,6 suggesting that calcifications are present before cartilage breakdown. Pathogenic crystals encompass calcium pyrophosphate dihydrate (CPPD) and basic calcium phosphate (BCP) crystals, (the latter including octacalcium phosphate (OCP), carbonated-apatite (CA) and hydroxyapatite (HA) crystals),7 and are generated by mineralising-competent cells and their cell membrane-derived matrix vesicles. In particular, high extracellular inorganic pyrophosphate (PPi) and inorganic phosphate (Pi) lead to CPPD and BCP crystal formation, respectively.8 The pathways which regulate this balance involve multiple ectoenzymes (Pc-1 that cleaves ATP in AMP and PPi, and Tnap that hydrolyses PPi in Pi) and transporters such as the PPi transporter Ank and the Pi transporters Pit-1 and Pit-2, and the Ca2+ transporter AnnexinV.9 Loss of function mutations in Pc-1 or Ank led to HA crystal deposition, calcifications and to OA-like changes in murine joints,10–12 and cartilage of patients with OA showed increased expression of Pc-1 and Ank.8 ,13 We demonstrated that intra-articular injection of BCP crystals in mice led to low grade inflammation and cartilage degradation.14 It has been reported that CPPD and BCP crystals can activate cells via different signalling pathways.15–21 In chondrocytes, it has been reported that BCP crystals induce inducible nitric oxide (iNOS) expression, nitric oxide (NO) production 20 and Mmp-13 production,22 but whether crystals induce additional inflammatory and catabolic response in this cell type, and the possible underlying signalling pathways, remains to be explored.
IL-6 is a pleiotropic cytokine increased in synovial fluids and sera of patients with OA, and increased IL-6 serum level correlates with radiographic knee OA.23 IL-6 induced Mmp-1, Mmp-3 and Mmp-13 production by chondrocytes and synoviocytes.24 ,25 The catabolic effects of injury together with tumour necrosis factor-alpha (TNF-α) on cartilage were mediated by endogenous IL-6.26 Using in vitro and in vivo approaches, Ryu et al27 demonstrated that IL-6 was a crucial mediator of hypoxia inducible factor-2-alpha (HIF-2α)-induced cartilage destruction via upregulation of Mmp-3 and Mmp-13 and found that IL-6 deficient mice were protected against OA. Although these studies support a key role of IL-6 in OA, further investigations are needed to clarify the triggers of IL-6 within the joint and IL-6 effects on chondrocytes.
In the present study we evaluated the effects of BCP crystals in primary murine chondrocytes, in the meniscectomy model of murine OA and in human cartilage explants. We demonstrated that BCP crystals trigger IL-6 secretion and IL-6-mediated cartilage degradation. Furthermore, we investigated the possible signalling pathway involved in BCP-induced IL-6 secretion in chondrocytes. Finally, we studied the mechanism by which IL-6 could be of a key importance in chondrocyte mineralisation. Our experimental data provide: (1) novel insights in the mechanisms of OA, suggesting interplay between BCP crystals and IL-6 and (2) novel therapeutic targets for this frequent and as yet poorly treated chronic condition.
Mice and induction of experimental OA
Female C57BL/6 mice (8–10 weeks old) were purchased from Charles River. Mice were anaesthetised and knee joint instability was induced surgically by partial medial meniscectomy (MNX) of the right knee, whereas the contralateral knee was sham-operated as control.28 Experiments were performed in accordance with the Swiss Federal Regulations. The protocol was approved by the “Service de la consommation et des affaires vétérinaires du Canton de Vaud”, Switzerland.
MicroCT scans were performed with a SkyScan 1076 X-ray µCT scanning system (SkyScan, Belgium) using the following parameters: 18 μm resolution, 60 kV, 167 μA, 0.4° rotation step over 360°, 0.5 mm aluminum filter, 1180 ms exposure time. In vivo or ex vivo acquisitions were made using anaesthetised mice or formol-fixed knees, respectively. Images were reconstructed using NRecon V.188.8.131.52 (Skyscan, Belgium) considering the following parameters: grey values=0.0000–0.105867, ring artefact reduction=3, beam hardening correction=40%. Quantitative analyses (bone mineral density (g/cm3), new formation volume (mm3) and new formation crystal content (µg)) were performed using CTAnalyzer V.1.10 (SkyScan, Belgium) for different volumes of interest.
Mouse knee histology and immunohistochemistry
Knee joints were processed and histological analysis performed as described.29 Immunohistochemical analysis of collagen 2, Mmp-induced neoepitope VDIPEN and IL-6 was performed using an anticollagen 2 biotinylated monoclonal antibody (MD Bioproduct), an affinity-purified anti-Val-Asp-Iso-Pro-Glu-Asn metalloproteinase generated neoepitope (VDIPEN) antibody30 and an anti IL-6 antibody (Abcam), respectively. IL-6 scoring was performed in the anteromedial part of the knee joint (following the same method as for Safranin-O score).
Calcium phosphate crystals
BCP and CPPD crystals were synthesised as previously published.31 BCP crystals were sterilised by γ-radiation and assessed as pyrogen-free. Prior to experimentation, crystals were resuspended in sterile phosphate buffered saline (PBS) and sonicated for 5 min.
Articular chondrocyte preparation
Chondrocytes were generated from C57BL/6J mice as described previously.32 Cells (3.5×104 cells/cm2) were cultured for 7 days in complete Dulbecco's modified Eagle Medium (DMEM) (10% fetal bovine serum (FBS)). Chondrocyte stimulations with crystals were performed in serum-free DMEM. For chondrocyte mineralisation analysis, cells were cultured for 3 days in complete Fitton-Jackson Modified (BJGb) medium (Gibco) (10% FBS, 50 µg/mL ascorbic acid, 20 mM β-glycerol phosphate), stimulated or not with 10 ng/mL of IL-6 (Gibco PMC0064) and treated or not with different inhibitors: Syk kinase inhibitor piceatannol (100 µM, Calbiochem 527948), PI3 kinase inhibitor Wortmannin (20 µM, Sigma W1628), anti-IL-6 receptor (8 µg/mL, R&D System AF1830) or Jak2 inhibitor AG490 (50 µM, Calbiochem 658401). Medium was changed for the last 4 days.
Fourier transform infrared spectroscopy analysis
Fourier transform infrared spectroscopy (FTIR) was used for in vitro (chondrocytes) and in vivo (dissected ectopic calcifications in knee joints) analyses. The mineral phase was evaluated by FTIR Bruker Vector 22 (BruckerSpectrospin, Wissembourg), as previously described.33 ,34
Crystal detection from chondrocyte cultures
Articular chondrocytes cultured for 7 days were washed in PBS and crystal deposition analysed as previously described.35
Calcium phosphate crystal stimulation
Chondrocytes were primed overnight with100 ng/mL Pam3Cys, where indicated, and stimulated with BCP or CPPD crystals or with 10 ng/mL IL-6. In some experiments, cells were treated with: piceatannol (100 µM), Wortmannin (20 µM), anti-IL-6 receptor (8 µg/mL), AG490 (50 µM) or cucurbitacin (10 µM, Calbiochem 238590). Supernatants were collected for cytokine ELISAs, and cells placed in TRIZOL for qRT-PCR analysis.
Human cartilage explants experiments
Macroscopically intact knee cartilage from four patients with OA (Kellgren-Lawrence score of 4, mean age 73±10 years) was obtained from the Orthopedics Department (CHUV, Lausanne, Switzerland) at the time of joint replacement, with the approval of the hospital ethical committee and patients’ written informed consent. Cartilage disks 6 mm diameter (9–20 disks/patient) were divided in halves, and each half was stimulated for 24 h in DMEM supplemented with 20 µg/mL ascorbic acid. Explants were stimulated with 500 µg/mL HA crystals in presence or absence of 5 µg/mL Actemra (tocilizumab, Roche) or Ilaris (canakinumab, Novartis). Supernatants were collected for ELISAs. Proteoglycans were examined by histology in formol-fixed Safranin-O-stained cartilage sections (0=normal to 4=completely degraded cartilage). IL-6 analysis (% of IL-6 positive cells out of three independent fields) was performed by immunohistochemistry using an anti-IL-6 antibody (US Biological Life Sciences).
Supernatants were assayed using murine or human IL-6, TNF-α, IL-1β and monocyte chemoattractant protein-1 (MCP-1) ELISA kits (eBioscience) following the manufacturer's protocol. Results were read at 450 nm using the Spectrax M5e (Molecular devices).
All experiments were performed with triplicates and reproduced independently at least twice. Statistical analysis was performed using the Student's t test, one-way or two-way analysis of variance (ANOVA) test corrected with post hoc tests for multiple comparisons, or linear regression, where appropriate. Data was analysed with GraphPad Prism software (GraphPad, San Diego, California, USA).
BCP crystals induce IL-6 secretion by primary murine chondrocytes through Syk kinase, PI3 kinase, Jak2 and Stat3 signalling
Murine chondrocytes exposed to CPPD crystals and BCP crystals (CA, HA and OCP) secreted high amounts of IL-6 (figure 1A), while IL-1β and TNF-α remained undetectable. HA-induced IL-6 secretion was dose-dependent (figure 1B) and time-dependent (figure 1C). In line with the ELISA results, qRT-PCR analysis revealed significantly increased IL-6 gene expression (6× compared with control) in HA-stimulated chondrocytes, whereas IL-1α, IL-1β and TNF-α expression was not modulated (figure 1D). IL-6 gene expression was modulated in a time-dependent manner (figure 1E). Additionally, the expression of the catabolic genes Mmp-13 and Mmp-3 and of Adamts-4 and Adamts-5 was strongly upregulated upon HA stimulation (figure 1F). Finally, HA-induced IL-6 secretion was abrogated by piceatannol and Wortmannin (figure 1G), both having no cytotoxic effects (results not shown). In addition, HA-induced IL-6 secretion was significantly diminished by a blocking IL-6 receptor antibody (anti-IL-6R) but not by an isotype matched control antibody (figure 1G and result not shown). Consistent with this latter result, AG490 and cucurbitacin also inhibited HA-induced IL-6 secretion (figure 1G).
HA crystals induce matrix degradation in human cartilage explants by IL-6 dependent mechanisms
Since BCP crystals led to massive secretion of IL-6 by murine articular chondrocytes and enhanced IL-6 expression in the meniscectomy model (see online supplementary figure S2), we investigated the role of HA-induced IL-6 in human cartilage catabolism. Explants were cultured in the presence or absence of HA crystals for 24 h. We observed loss of Safranin-O staining in HA-stimulated explants compared with unstimulated ones (figure 2A, left), confirmed by Safranin-O scoring (figure 2C, left). We next studied the role of IL-6 in HA crystal-induced matrix degradation by adding an IL-6 receptor inhibitor (tocilizumab) to the culture and we observed a significant protection against cartilage degradation as well as matrix degeneration (figure 2A, middle). As a control, we incubated HA-stimulated explants with an anti-IL-1β antibody (canakinumab). This isotype-matched control antibody had no effect on proteoglycan depletion (figure 2A, right). Safranin-O scoring confirmed the protective effect of IL-6 blockade and no effect of IL-1β inhibition (figure 2C middle and right). These results strongly suggested that IL-6 is involved in cartilage degradation.
We also checked if HA crystals induced IL-6 in explants. HA crystals increased chondrocyte IL-6 staining in the transitional zone of cartilage, where pictures were taken (figure 2B, D left). This was inhibited by tocilizumab, but not by canakinumab (figure 2B, D middle and right). Finally, as in the murine model, we found a positive correlation between Safranin-O loss and IL-6 expression (figure 2E). The importance of IL-6 in cartilage matrix degradation is further strengthened by the finding that when HA crystals were added to cartilage explants, we failed to detect any secreted IL-1β or TNF-α but just IL-6, as we previously found in murine chondrocytes. Second, HA-mediated IL-6 release was reduced in explants exposed to IL-6 pathway blockade but not by IL-1β inhibition (results not shown). Taken together, these findings strongly suggest that HA-induced cartilage degradation is IL-6 dependent.
IL-6 increases BCP formation by chondrocytes
As cartilage IL-6 secretion correlated with matrix degradation, and was induced by BCP crystals, we investigated if IL-6, in turn, could induce calcium-containing crystal formation in chondrocyte cultures. After 7 days of culture, Alizarin red staining (figure 3A, black arrows) in IL-6-stimulated chondrocytes was significantly increased compared with unstimulated cells. This was confirmed by spectrophotometric quantification of Alizarin red, after acidic extraction of crystals from the entire cell monolayer (figure 3B). Interestingly, IL-6-induced crystal formation was abrogated by the addition of piceatannol, Wortmannin, an anti-IL-6R antibody and AG490 (figure 3A, B), the same inhibitors that were able to decrease BCP crystal-induced IL-6 secretion. These inhibitory effects could not be attributed to cytotoxic effects as lactate dehydrogenase (LDH) activity was similar or lower to controls for all the tested inhibitors (figure 3C).
FTIR analysis of the calcium-containing crystals produced by IL-6-stimulated chondrocytes showed the presence of BCP crystals, specifically OCP crystals (phosphate bands at 1112/cm and 1030/cm and phosphate deformation at 600/cm and 560/cm) (figure 3F). In addition to Alizarin red staining, as a control of IL-6 biological activity, we measured MCP-1 secretion, known to be induced by IL-6. Its secretion was increased by IL-6 and significantly decreased by the different inhibitors (figure 3D).
Finally, we investigated the underlying mechanism by which IL-6 promotes calcium-containing crystal formation in chondrocytes by qRT-PCR (figure 3E). IL-6 was able to upregulate the expression of three genes involved in the calcification process (Ank, Anx5 and Pit-1). No modulation was detected for genes codifying for other proteins involved in crystal formation (Pit-2, Pc-1 and Tnap). Furthermore, IL-6 increased Coll10 and Runx2, markers of hypertrophic chondrocytes, and Sox9, marker of early chondrocytic differentiation. Therefore, IL-6 effect on mineralisation cannot simply be explained in terms of impact on the global chondrocytic differentiation.
What links biomechanical stress and inflammatory responses to cartilage breakdown in OA is not clearly understood, but there is persuasive evidence that calcification of joint structures could be involved. We demonstrated that in the murine meniscectomy model of OA, calcium-containing crystal deposits were formed in the articular space prior to cartilage degradation (see online supplementary figure S1). These crystal deposits were previously identified as hypertrophic calcification,36 ectopic bone,37 mineralised area38 and heterotopic cartilage,39 but their chemical characterisation and their role in OA had never been determined. Starting from 1 month after OA induction, microCT-scan revealed multiple deposits within the joint, detached from bone, so that they could not be considered as osteophytes. These structures were composed of bone-like and cartilage-like tissues, showed catabolic and anabolic features typical of OA cartilage40 and contained CA crystals (see online supplementary figure S2). These structures resembled human synovial osteochondromatosis, a condition associated with OA,41–43 characterised by chondroid metaplasia and osteocartilaginous mineralised bodies in the capsule. A process similar to chondroid metaplasia may be involved, whereby fibroblasts undergo metaplastic transformation to chondrocytes, which then calcify.
Calcium phosphate crystals can have multiple effects on chondrocytes, including iNOS gene expression, NO production,20 Mmp-13 induction,22 intracellular calcium oscillations44 and apoptosis.45 We show here that BCP crystals (in the form of HA) additionally induced Mmp-3, Adamts-4 and Adamts-5 expression in chondrocytes. Moreover, these microcrystals strongly upregulated IL-6 at the transcriptional and translational levels, while IL-1α, IL-1β and TNF-α cytokines remained undetectable.
We suggest that IL-6 secretion upon crystal stimulation is a result of IL-6 directly induced by BCP (direct pathway) plus IL-6 induced via ligation to its IL-6R, expressed on chondrocytes27 ,46 (autocrine pathway). We further found that pharmacological inhibition of the kinases Syk and PI3 and of the signalling molecules Jak2 and Stat3 completely abrogated BCP-induced IL-6 secretion, suggesting that these molecules are implicated in the IL-6 direct and in the IL-6 autocrine pathways. Interestingly, in the presence of saturating concentrations of the neutralising anti-IL-6R antibody, we observed only partial inhibition of crystal-induced IL-6, whereas complete inhibition occurred with Jak and Stat inhibitors. This latter result suggests that IL6-independent, but Jak, Stat-dependent pathways induced by other gp130 signalling cytokines such as Oncostatin M could be involved.47 We then analysed the consequences of IL-6 blockade in human cartilage explants. HA crystal-stimulated explants showed increased proteoglycan loss and IL-6 expression. Treatment of HA crystal-stimulated explants with an IL-6 inhibitor led to restoration of proteoglycan synthesis and IL-6 expression. These findings suggest that chondrocytes secrete IL-6 that, in turn, binds to its receptor resulting in an autocrine loop. In contrast, addition of an isotype matched IL-1β inhibitor blocking antibody did not exert any protective role with respect to HA crystal-induced proteoglycan loss and IL-6 expression.
IL-1β and TNF-α cytokines, as in BCP-stimulated chondrocytes, remained undetectable. Although IL-1 has been proposed to be a key catabolic cytokine in OA,48 we were unable to confirm its importance in our previous and current studies. Mice deficient for IL-1α, IL-1β and the adaptor molecule MyD88 were not protected from experimental OA (Nasi et al29 in preparation). These experiments suggest that IL-1β might not be involved in OA pathogenesis whereas IL-6 could be a key cytokine in cartilage degradation induced by crystal stress.
To further support a key role of IL-6 in OA pathogenesis, we have shown here that IL-6 is able to enhance in vitro BCP crystal formation by primary murine chondrocytes and that, indeed, IL-6 inhibitors block in vitro crystal generation. It is interesting to note that OCP crystals, considered a precursor phase of apatite crystals and representing a recent mineralisation process, were the only crystals detected in our chondrocytes after 1 week cell culture. Later on, OCP crystals are normally converted to apatite (CA) crystals, which are indeed the crystals we found in our crystal deposits 2 months after meniscectomy in the in vivo experiments The promineralisation effect of IL-6 cannot simply result from IL-6-effects on chondrocyte differentiation as IL-6 increased Coll10 and Runx2, two markers of hypertrophic chondrocytes, and Sox9, an early chondrocytic differentiation marker. We next hypothesise IL-6 may directly modulate one or several of the crucial proteins involved in chondrocyte mineralisation. Indeed, IL-6 was able to upregulate Ank, Anx5 and Pit-1 gene expression. Ank increases ePPi (that can be hydrolysed in Pi by Tnap) whereas Anx5 and Pit1 concentrate Ca2+ and Pi, respectively, into matrix vesicles and overexpression of Ank, or Anx5, or Pit1 was shown to induce BCP crystal formation.49–51 Therefore IL-6 upregulation of Ank, or Anx5, or Pit1 genes could act in concert to promote BCP crystal formation in chondrocytes. Whether IL-6 inhibitors block calcification through direct modulation of these genes involved in calcification remains to be clarified.
In conclusion, our results strongly suggest that BCP crystal-induced stress and IL-6 production are interlinked key factors in OA pathogenesis (figure 4). Therefore, inhibition of calcification or of IL-6 signalling pathway represents possible therapeutic approaches for OA treatment.
The authors thank Professor Frédéric Lioté and Dr Hang-Korng Ea (Hospital Lariboisière and Uniersité Paris Diderot, UFR de Médecine) for helpful discussions, Véronique Chobaz (Laboratory of Rheumatology, CHUV, Lausanne-CH) for her excellent technical support, Prof Brigitte Haeberli-Jolles (DAL, CHUV, Lausanne-CH) for kindly providing human cartilage, Dr Florence Morgenthaler (CIF, University of Lausanne-CH) for her constant support in micro CT-scan analysis and Dr Kempf and Dr Bianchi (CNRS-Université de Lorraine, Nancy, France) for Ank experiments. The authors also thank Dr Peter van Lent (University of Nijmegen, The Netherlands) for his generous help with the anti-VDIPEN staining experiment.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
Contributors SN, AS and NB designed, performed and evaluated all experiments. CC synthesised calcium-containing crystals; MD performed FTIR analysis; the entire work was supervised by NB. The figures were prepared and the manuscript was written by SN, AS and NB. All authors discussed and commented on the manuscript.
Funding This work was supported by the Fonds National Suisse de la recherche scientifique (grant 310030-130085/1 and 310030-153010/1), the Institute of Arthritis Research, the Fondation RMR, the Fondation Jean and Linette Warnery.
Competing interests None declared.
Patient consent Obtained.
Ethics approval CHUV Ethical Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.