The authors have no conflict of interest.
Fibroblasts Express RANKL and Support Osteoclastogenesis in a COX-2-Dependent Manner After Stimulation With Titanium Particles†
Article first published online: 14 FEB 2005
Copyright © 2005 ASBMR
Journal of Bone and Mineral Research
Volume 20, Issue 7, pages 1136–1148, July 2005
How to Cite
Wei, X., Zhang, X., Zuscik, M. J., Drissi, M. H., Schwarz, E. M. and O'Keefe, R. J. (2005), Fibroblasts Express RANKL and Support Osteoclastogenesis in a COX-2-Dependent Manner After Stimulation With Titanium Particles. J Bone Miner Res, 20: 1136–1148. doi: 10.1359/JBMR.050206
- Issue published online: 4 DEC 2009
- Article first published online: 14 FEB 2005
- Manuscript Accepted: 8 FEB 2005
- Manuscript Revised: 26 JAN 2005
- Manuscript Received: 18 AUG 2004
- synovial fibroblast;
Synovial fibroblasts are possible mediators of osteolysis. Fibroblasts respond directly to titanium particles and increase RANKL expression through a COX-2/PGE2/EP4/PKA signaling pathway. Fibroblasts pretreated with titanium or PGE2 stimulated osteoclast formation, showing the functional importance of RANKL induction. Synovial fibroblasts and their activation pathways are potential targets to prevent osteolysis.
Introduction: Bone loss adjacent to the implant is a major cause of joint arthroplasty failure. Although the cellular and molecular response to microscopic wear debris particles is recognized as causative, little is known concerning role of synovial fibroblasts in these events.
Materials and Methods: Murine embryonic fibroblasts and knee synovial fibroblasts in culture stimulated with titanium particles were examined by FACS, real time RT-PCR, Northern blot, and Western blot for expressions of vascular cell adhesion molecule (VCAM)1, RANKL, cyclooxygenase (COX)-1, and COX-2, and the four prostaglandin E2 (PGE2) receptor isoforms. Experiments were performed in the presence and absence of COX inhibitors, protein kinase A (PKA) and protein kinase C (PKC) inhibitors, and various EP receptor agonists. Osteoclast formation was examined in co-cultures of pretreated glutaraldehyde-fixed fibroblasts and primary murine spleen cells treated with macrophage-colony stimulating factor (M-CSF) for 7-days.
Results: TNF-α stimulated VCAM1 expression, consistent with a synovial fibroblast phenotype. Titanium particles stimulated RANKL gene and protein expressions in fibroblasts in a dose-dependent manner. Gene expression was increased 5-fold by 4 h, and protein levels reached a maximum after 48 h. Within 1 h, titanium particles also induced COX-2 mRNA and protein levels, whereas both indomethacin and celecoxib blocked the stimulation of RANKL, suggesting a COX-2-mediated event. Furthermore, PGE2 induced RANKL gene and protein expression and rescued RANKL expression in titanium-treated cultures containing COX-2 inhibitors. Fibroblast cultures pretreated with either PGE2 or titanium particles enhanced osteoclast formation, indicating the functional importance of RANKL induction. EP4 was the most abundant PGE2 receptor isoform, EP1 and EP2 were expressed at low levels, and EP3 was absent. The EP1 selective agonist iloprost and the EP2 selective agonist butaprost minimally stimulated RANKL. In contrast, the EP2 and EP4 agonist misoprostol induced RANKL to a magnitude similar to PGE2. Finally, PKA antagonism strongly repressed RANKL stimulation by PGE2.
Conclusion: Fibroblasts respond directly to titanium particles and increase RANKL expression through a COX-2/PGE2/EP4/PKA signaling pathway. Thus, the synovial fibroblast is important mediator of osteolysis and target for therapeutic strategies.
TOTAL JOINT ARTHROPLASTY is the optimal treatment for severe arthritis involving major joints, including the hip, knee, shoulder, and elbow. Arthroplasty results in a dramatic reduction in pain and a marked improvement in function and is increasingly being used, with >1.3 million arthroplasties performed around the world each year.(1) However, in up to 20% of cases, bone loss develops adjacent to the implant and results in mechanical instability, renewed pain, and the need for revision surgery.(2,3)
The pathogenesis of prosthetic loosening is still unclear, but both biological and mechanical factors likely contribute.(1) Motion at the bearing surfaces of the prosthetic joint results in the generation of submicron particles composed of polyethylene and metallic debris. These particles are shed into the joint space and become imbedded in the surrounding tissues where >109 particles are found per gram of tissue.(4–6) Prior work has shown a direct relationship between the generation of particles, as measured by volumetric polyethylene wear, and aseptic loosening.(7,8) In response to the particles, an inflammatory membrane composed of a fibrovascular tissue containing macrophages and occasional lymphocytes develops in the joint lining and invades the prosthesis-bone interface.(9) Both in vitro and in vivo studies show that this inflammatory membrane mediates the bone loss that occurs in aseptic loosening.
Macrophages compose ∼15% of the cells in the inflammatory membrane. In situ hybridization experiments with retrieved human membranes showed that interleukin (IL)-1 expression was limited to macrophages and did not occur in other cell types.(9,10) Thus, macrophages are considered the major source of proinflammatory mediators and on stimulation with particulate debris release inflammatory cytokines that directly and indirectly stimulate bone resorption.(9,10) Moreover, recent studies show that membrane derived macrophages can form multinucleated, TRACP+ cells that resorb bone. As such, the macrophage has been extensively studied as a mediator of osteolysis.(1) In contrast, there is much more limited information regarding the role of other cells, especially fibroblasts, which compose ∼70% of the cells in the periprosthetic membrane.(9,10)
Recent information suggests that synovial fibroblasts present in the periprosthetic membrane are important targets of wear debris during osteolysis. The pathogenesis of inflammatory joint diseases, including rheumatoid arthritis, have been shown to involve the synovial fibroblast.(11,12) In inflammatory arthritis, synovial fibroblasts secrete proteases that stimulate matrix degradation and release inflammatory mediators.(13–15) Importantly, synovial fibroblasts participate in the bone resorption that characterizes inflammatory arthritis by secreting osteoclastogenic factors including RANKL.(11,12,15,16) Fibroblasts have also been shown to be directly responsive to particles and release metalloproteinases.(17) Work in our laboratory has shown that titanium particles induce prostaglandin E2 (PGE2) and IL-6 expression in fibroblasts, but this effect is absent in fibroblasts lacking cyclooxgenase (COX)-2 gene expression and in cultures containing COX-2 inhibitors.(18) Immunohistochemical examinations have confirmed COX-2 expression in fibroblasts in human periprosthetic membranes and in fibroblasts adjacent to murine calvaria treated with titanium particles.(18,19) Others have shown that fibroblasts in the periprosthetic membrane express RANKL.(20,21)
RANKL is a member of the TNF/TNF receptor (TNFR) superfamily and is present primarily as a membrane protein and is produced by osteoblasts/stromal cells.(22,23) The receptor RANK is located on macrophage/osteoclast precursors, and on activation by RANKL, stimulates the expression of genes required for osteoclast differentiation.(24) Both gain and loss of function experiments have established RANKL/RANK as essential for osteoclast formation.(25,26) The importance of RANKL for the aseptic loosening is indicated by the complete absence of wear debris-induced osteolysis in a murine model after loss of RANK-RANKL signaling.(27)
Here we perform a series of in vitro experiments that define the fibroblasts as an important target of wear debris particles. In response to micron-sized titanium particles, fibroblasts express RANKL and directly stimulate osteoclast formation. We show that this effect is dependent on the induction of COX-2 and is mediated by prostaglandins. Finally, we show that the induction of RANKL likely involves PGE2 through the EP4 receptor and protein kinase A (PKA) signaling pathway. These findings provide a mechanism for earlier work in our laboratory that showed complete absence of inflammatory osteolysis in COX-2-deficient mice using an in vivo murine calvarial model of particle-induced bone loss.(19)
MATERIALS AND METHODS
The fibroblastic cell line was previously established from embryos of mouse CBAXBL6 strain.(28) Confirmatory experiments were also performed using fibroblast-like synoviocytes obtained from the murine knee joint.(29) Fibroblasts and ST-2 cells were both cultured in DMEM medium in the presence of 10% heat-inactivated FBS, 1% penicillin, and 5% CO2 at 37°C. As a control for RANKL expression, ST-2 cells were cultured in the presence or absence of 10−8 M 1,25-dihydroxyvitamin D3 [1,25(OH)2D3; Biomol; Plymouth Meeting, PA, USA] and 10−7 M dexamethasone (Sigma, St Louis, MO, USA) for 48 h. Fibroblasts were stimulated with particles (from 5 × 105/ml to 1 × 107/ml) or PGE2 (1 × 10−6 M; Cayman Chemicals, Ann Arbor, MI, USA), IL-6 (100 ng/ml; R&D, Minneapolis, MN, USA), IL-1 (100 ng/ml; R&D), or TNF-α (10 ng/ml; R&D). In various experiments, the cyclooxygenase inhibitors indomethacin (1 μM; Cayman Chemicals) and celecoxib (10 μM; G.D. Searle, Chicago, IL, USA) or the EP receptor agonists, iloprost (2 μM), butprost (10 μM), or misoprostol (1 μM; Cayman Chemicals) were added to the cultures. The PKA and protein kinase C (PKC) inhibitors H-89 (10 μM) and Go6976 (10 μM; Calbiochem, San Diego, CA, USA) were used in some experiments.
Titanium particles were obtained from Johnson Matthey Chemicals (Ward Hill, MA, USA) and contain 1- to 3-μm-diameter particles. Particles were suspended in PBS and autoclaved for sterilization. In addition, particles were also prepared by alternately washing three times in nitric acid and sodium hydroxide solutions, and then washing three times in sterile PBS as described by Bi et al.(30) Experiments showed similar induction of RANKL with washed and unwashed particles, and the data shown use unwashed titanium particles. For use in cell culture, the particles were suspended in PBS at 1 × 108/ml. As a final check on particle size and count, the suspension was evaluated by Coulter ZM Channelizer analysis.
Untreated fibroblasts and fibroblasts treated with mouse TNF-α (R&D) for 24 h were washed and harvested in PBS containing 5 mM EDTA at 4°C. Cells (1 × 106) were stained with biotin conjugated antibody specific for anti-mouse vascular cell adhesion molecule (VCAM)-1 and fluorescein isothiocyanate (FITC)-conjugated streptavidin (Pharmingen, San Diego, CA, USA). Isotype-matched monoclonal antibodies were used as negative controls (Pharmingen). A total of 1 × 104 stained cells was assessed by a FACSCalibur cytometer and the Cell Quest plotting program (Becton Dickinson, Franklin Lakes, NJ, USA).
RT-PCR and quantitative real-time PCR
Total RNA was isolated from cell cultures at various times using the RNeasy kit (Qiagen, Valencia, CA, USA) and was reverse transcribed to cDNAs using Superscript II according to manufacturer's instructions (Invitrogen, Carlsbad, CA, USA). Primers specific for murine RANKL, COX-2, GAPDH, and the EP receptors were used. For quantitative real-time PCR, RANKL was amplified using 5′-CACCATCAGCTGAAGATAGT-3′; 5′-CCAAGATCTCTAACATGACG-3′, COX-2 was amplified using 5′-CACAGCCTACCAAAACAGCCA-3′; 5′-GCTCAGTTGAACGCCTTTTGA-3′, and GAPDH was amplified using 5′-AACGACCCCTTCATTGAC-3′; 5′-TCCACGACATACTCAGCAC-3′) primer sets. Duplicate PCR reactions were carried out in Rotor-Gene 3000 (Corbett Research) using n = 3 for each sample. SYBR green dye was used for detection of the product using the SYBR Green PCR Master Mix assay (Applied Biosystems, Warrington, UK). The standard curve used a series of duplicate dilutions of plasmid for RANKL and GAPDH cDNA. The amplification reaction was performed for 45 cycles with denaturation at 95°C for 15 s, followed by annealing at 58°C for 20 s and extension and detection at 72°C for 10 s. For EP receptor expression, primers were selected to give products of similar size for the various receptors, and annealing temperature was selected for the relative same amplification efficiency. The primers were EP1 (5′-TACATGGGATGCTCGAAACA-3′; 5′-TTTTAAGCCCGTGTGGGTAG-3′), EP2 (5′-ATGCTCCTGCTGCTTATCGT-3′; 5′-TAATGGCCAGGAGAATGAGG-3′), EP3 (5′-GGATCATGTGTGTGCTGTCC-3′; 5′-AACTGGAGACAGCGTTTGCT-3′), and EP4 (5′-CCATCGCCACATACATGAAG-3′; 5′-TGCATAGATGGCGAAGAGTG-3′). PCR amplification was performed in 95°C for 15 s, 50°C for 20 s, and 72°C for 30 s for 30 cycles. The cDNA products were separated on a 1% agarose gel.
Fifteen micrograms of total RNA was separated on a 1.2% agarose gel containing 17.5% formaldehyde, and the RNA was transferred to a Gene Screen Plus membrane (New England Nuclear, Boston, MA, USA). After RNA cross-linking to the membrane by UV light, prehybridization was performed in QuickHyb solution (Stratagene, La Jolla, CA, USA) for 60 minutes at 68°C. A mouse cDNA probe for RANKL (from Dr Lianping Xing, Department of Pathology, University of Rochester) and the murine GAPDH probe were used.(29) The probes were labeled with [32P]deoxycytidine 5′-triphosphate using a random priming kit (Stratagene, La Jolla, CA, USA) and hybridized to the blots overnight at 68°C. The blots were washed under stringent conditions and exposed to X-OMAT AR film (Kodak, Rochester, NY, USA) for autoradiography or a phosphor storage screen for PhosphorImage (Molecular Dynamics, Sunnyvale, CA, USA).
Total protein extracts were prepared from cells at various times in the presence of protease inhibitors as previously described.(31) Fifteen micrograms of the protein extract was separated by SDS-PAGE. After transfer to a polyvinylidene fluoride (PVDF) membrane, the blots were probed overnight with antibodies specific for murine COX-1, COX-2 (Cayman Chemicals), or β-actin (Sigma). After washing the blots, horseradish peroxidase (HRP)-conjugated secondary antibodies were used to detect the immune complexes using ECL plus (Amersham Biosciences, Piscataway, NJ, USA). For RANKL, IP-Western was performed as described.(32) Briefly, ∼500 μg of total protein extracts were immunoprecipitated by RANK-Fc recombinant protein (R&D) and protein A sepharose (Amersham Biosciences). The complex was boiled in 2× SDS sample buffer and separated by SDS-PAGE and assayed by Western blotting with anti-RANKL antibody (Calbiochem, San Diego, CA, USA). Control experiments were performed in ST-2 cells as previously described.(32)
PGE2 enzyme-linked immunosorbent assay
Levels of PGE2 in the culture supernatants were measured using commercially available enzyme-linked immunosorbent assay (EIA) kits (Cayman Chemicals) following the manufacturer's instructions, as we have described previously.(18,19) Spectrophotometry was performed at 405 nm using an automated microELISA plate reader (TECAN Sunrise, ResearchTriangle Park, NC, USA), and results were compared with standard concentrations analyzed on the same plate.
A mouse calvarial resorption model was used as previously described.(19,27) Briefly, 8-week-old female mice were anesthetized by peritoneal injection using 80 mg/kg of ketamine and 5-7 mg/kg of xylazine. A 1-cm midline sagittal incision was made, and 30 mg of titanium particles was placed evenly on top of the calvaria. After implantation, the skin was closed with 4-0 Ethilon suture. Sham controls had the surgical incision, but no particles were implanted. At day 3, mice were killed by cervical dislocation, and the calvaria were removed and fixed in 10% neutral buffered formalin. Samples were blocked with a 1:20 dilution of normal goat serum and incubated with a 1:20 dilution of primary antibody (Santa Cruz Biotech, Santa Cruz, CA, USA) overnight at 4°C. The sections were washed with PBS and incubated with a 1:200 dilution of secondary antibody followed by streptavidin-peroxidase complex, and the binding of antibody was visualized with aminoethyl carbazole (AEC; Zymed, San Francisco, CA, USA), according to the supplier instructions.
In vitro osteoclastogenesis assay
Fibroblasts (50,000 cells/well) were plated in 24-well plates overnight and treated with titanium particles (5 × 106/ml) or 1 μM PGE2 (Cayman Chemicals). After 24 h, cells were fixed in 2.5% glutaraldehyde for 1 minute, followed by the addition of 1.5% glycine for 1 minute. The cell layers were subsequently washed three times in PBS and covered with α-MEM containing 10% FBS. Spleen cells were isolated from 4-month-old C57BL/6 female mice. Isolated spleens were washed, minced, and purified by adding 0.83% NH4Cl in 10 mM Tris buffer (pH 7.4).(33) The cells were washed three times with PBS, suspended in α-MEM containing 10% FBS, 10 ng/ml macrophage-colony stimulating factor (M-CSF; R&D), and overlaid on prefixed fibroblasts cell layers. In the positive control group, media also contained soluble 50 ng/ml RANKL protein (R&D). Experiments were also performed in the presence and absence of RANK-Fc (50 ng/ml; R&D) to block RANKL signaling. Media with M-CSF with or without RANKL and RANK-Fc were changed every other day. After a 7-day culture, the osteoclast-like cells were detected by TRACP staining (Sigma). Multinucleated (>3 nuclei per cell) and TRACP+ (purple staining cells) were counted as osteoclasts.
Results are given as means ± SE. Comparisons were made by ANOVA and Student's t-test between data from the experimental and control groups. p values < 0.05 were considered statistically significant.
Mouse embryonic fibroblasts stimulated with TNF-α express VCAM-1
We began by assessing the ability of the murine embryonic fibroblast cell line to express a synovial fibroblast phenotype.(28) Because VCAM1 is a marker of synovial fibroblasts,(29) we examined the expression of VCAM1 by flow cytometry in embryonic fibroblast cultures in the presence and absence of TNF-α (Fig. 1). Under basal conditions, embryonic fibroblasts express low levels of VCAM1 (Fig. 1B). However, VCAM1 is markedly induced after treatment with TNF-α (Fig. 1C). Because these findings are similar to those described with synovial fibroblasts,(29) we used the embryonic fibroblast cell line as a model to study the cell response to titanium particles in vitro.
Titanium particles stimulate RANKL gene and protein expression in fibroblasts
RANKL expression was examined in fibroblast cultures to determine if these cells might directly participate in the induction of osteoclasts in response to wear debris particles. Fibroblasts were treated with titanium particles (5 × 106 particles/ml) for various times up to 24 h, and RANKL expression was determined by quantitative real-time RT-PCR. RANKL was stimulated at 1 h, and whereas peak levels occurred at 4 h (5-fold), expression remained elevated for 24 h after exposure to titanium particles (Fig. 2A). RANKL stimulation was confirmed by Northern blot after 4 and 8 h of titanium exposure (Fig. 2B). RANKL induction occurred over a broad range of particle concentrations (Fig. 2C). A small level of stimulation was observed at 5 × 105 particles and progressively increased up to a concentration of 1 × 107 particles, where maximal stimulation was observed (Fig. 2C). Because more than a 5-fold stimulation was observed with concentrations of 5 × 106 particles, this concentration was used in subsequent experiments.
To determine whether induction of RANKL gene expression is associated with an increase in protein expression, whole cell lysates were examined by immunoprecipitation-Western (IP-Western). The mouse stromal cell line ST-2 was used as a control for RANKL protein induction. After 48 h of treatment with 10−8 M 1,25(OH)2D3 and 10−6 M dexamethasone, RANKL protein was markedly enhanced in ST-2 cells (Fig. 2D, left), as previously described.(32) Whole cell extracts from fibroblast cultures treated with titanium particles for 0, 8, 24, and 48 h were similarly harvested and subjected to IP-Western for RANKL expression. Similar to ST-2 cells, RANKL protein was undetected under basal conditions but was induced by 8 h with maximal levels occurring after 48 h (Fig. 2D, right). Altogether, the findings show that titanium particles activate fibroblasts to produce RANKL, an important osteoclastogenic factor.
Titanium particles induce COX-2 and stimulate PGE2 secretion in fibroblasts
Prior work in our laboratory established that particles induce PGE2 in fibroblast cultures and suggested a COX-2-mediated effect.(18) For this reason, we examined whether titanium particles induced COX-2 gene and protein expression in fibroblast cultures (Fig. 3). Within 30 minutes after exposure to 5 × 106 titanium particles/ml, COX-2 gene expression was induced 2-fold, and a maximal 12-fold increase was present by 2 h (Fig. 3A). Under basal conditions, COX-2 protein expression was undetectable, but expression was observed within 1 h of exposure to titanium particles (Fig. 3B). Maximal levels were present by 2 h and were similar at 4 h after treatment. In contrast, COX-1 expression was constitutive and was not altered after treatment with titanium particles. β-Actin Western blot was used as a loading control and was similar in all of the extracts (Fig. 3B). Finally, COX-2 induction was associated with a massive accumulation of PGE2 in the conditioned medium (Fig. 3C).
RANKL induction in fibroblasts is mediated by PGE2
Fibroblasts were treated with titanium particles in the presence and absence of combinations of indomethacin and PGE2 to determine if PGE2 is involved in the induction of RANKL (Fig. 4A). Indomethacin was selected because it inhibits both COX-1 and COX-2 enzyme function.(34,35) As previously shown, 4 h of treatment with titanium particles induced RANKL expression. However, treatment with indomethacin completely blocked RANKL induction by titanium, whereas addition of 10−6 M PGE2 restored the expression of RANKL in cultures treated with titanium particles and indomethacin. The level of RANKL expression under these conditions was similar to that observed in fibroblast cultures treated with PGE2 alone and slightly more than that which occurred after titanium treatment (Fig. 4A).
To determine if RANKL stimulation by titanium particles is related to the increase in COX-2 expression, fibroblast cultures were treated with titanium particles for 4 h in the presence and absence of the COX-2-specific inhibitor, celecoxib. Celecoxib markedly reduced the induction of RANKL consistent with a COX-2-mediated effect (Fig. 4B). Because the interfascial membrane adjacent to loose prosthetic joints contains high levels of proinflammatory cytokines, the ability of PGE2 to induce RANKL expression in comparison with the osteoclastogenic factors IL-1, IL-6, and TNF-α was assessed (Fig. 4C).(9,36) IL-1 and IL-6 had no effect, whereas TNF-α resulted in a small increase in RANKL expression in the fibroblast cultures.
Finally, to confirm the role of PGE2 in the induction of RANKL, total protein extracts from control and PGE2-treated fibroblast cultures were examined by IP-Western blot (Fig. 4D). Twenty-four hours of treatment with PGE2 resulted in a marked induction of RANKL protein. The level of induction was comparable with the degree of protein expression that occurred in ST-2 cell cultures treated with 1,25(OH)2D3 and dexamethasone for 48 h. Altogether, the findings show that PGE2 is induced by titanium particles through COX-2 and is a powerful stimulator of RANKL in fibroblasts.
Titanium particles stimulate RANKL induction in cultures of murine knee synovial fibroblasts and in calvarial tissues in vivo
To verify the induction of RANKL expression by titanium in synovial fibroblasts, experiments were repeated in authentic fibroblast-like synoviocytes (FLSs) obtained from the murine knee joint.(29) Titanium and PGE2 both induced RANKL expression in FLS cultures between 9- and 11-fold, and the induction by titanium was inhibited by the addition of indomethacin (Fig. 5A). The real time RT-PCR findings were confirmed by Northern blot (Fig. 5B). Thus, the findings are identical to those observed in the embryonic fibroblasts.
Additionally, experiments were performed using immunohistochemistry to localize RANKL in the inflammatory tissues adjacent to the calvaria of sham-operated mice and mice 3 days after the implantation of titanium particles onto the calvarial surface. In sham-operated mice, minimal or no RANKL was observed, whereas RANKL staining was readily apparent in the soft tissues overlying the calvaria implanted with titanium particles (Fig. 5C). The staining was present in fibroblasts adjacent to the calvarial surface and was consistent with the abundant bone resorption that occured 5 days after titanium implantation in vivo and its inhibition by the RANKL antagonists, RANK-Fc and OPG.(27,37)
Titanium particle and PGE2 activated fibroblasts directly support osteoclastogenesis in vitro
To determine the functional significance of fibroblast RANKL expression, an in vitro osteoclast formation assay was performed using glutaraldehyde-fixed fibroblasts co-cultured with primary murine splenocytes in the presence of 10 ng/ml M-CSF (Fig. 6). Subconfluent fibroblast cell cultures were treated with control medium or medium containing PGE2 (10−6 M) or titanium particles (5 × 106 particles/ml). After 24 h, the cell layers were fixed with glutaraldehyde and glycine and were overlaid with primary murine splenocytes. Splenocytes cultured alone in the absence of recombinant RANKL failed to form multinucleated TRACP+ cells (Fig. 6A). Co-culture with fibroblasts treated in the presence of control medium similarly resulted in minimal osteoclast formation (Fig. 6B), whereas RANKL addition as a positive control resulted in the abundant formation of large multinucleated TRACP+ cells (Fig. 6C). When fibroblasts were treated with either titanium or PGE2 for 24 h before glutaraldehyde fixation, a significant stimulation of multinucleated TRACP+ cells was observed in the co-cultures (Figs. 6D and 6E). However, whereas the addition of recombinant RANKL resulted in enormous cells, with as many as 50 or 100 nuclei, the multinucleated TRACP+ cells that were formed in the titanium and PGE2 co-cultures were substantially smaller. The mean numbers of multinucleated TRACP+ cells that formed in 7-day splenocyte cultures are summarized in Fig. 6F. The findings show that fibroblasts treated with either titanium or PGE2 support osteoclast formation independent of the addition of recombinant RANKL.
To confirm the role of RANKL in this model of osteoclastogenesis, fibroblasts were treated with PGE2, titanium particles, or control medium and fixed with glutaraldehyde. Splenocytes were added to the fibroblast cultures in the presence or absence of the RANKL antagonist, RANK-Fc. In these experiments, RANKL was added only to the cultures containing fibroblasts treated with control medium (Fig. 7A). Whereas osteoclasts formed in all of the cultures in the absence of RANK-Fc, the addition of this factor markedly inhibited osteoclast formation (Figs. 7A and 7B).
RANKL is induced by specific EP receptors in fibroblasts
Four different receptors (EP1-4) specific for PGE2 have been described and mediate separate downstream signaling events to influence gene expression.(38–40) RT-PCR was performed to examine which of the EP receptors are expressed on fibroblasts (Fig. 8A). After 30 cycles of amplification, EP3 expression was absent, and low levels of EP1 and EP2 were observed. Substantially higher levels of EP4 expression was observed, suggesting that this is the major receptor subtype found on these cells.
Using receptor agonists with relative specificity, the functional role of the various receptors in RANKL induction was assessed (Fig. 8B). As previously observed, PGE2, which binds to all of the receptors, results in a large induction of RANKL. Iloprost, which binds to the EP1 receptor, and also has affinity for the EP3 and IP receptors, caused only a small increase in RANKL.(41–44) Similarly, the EP2 receptor agonist, butaprost,(41,45) minimally induced RANKL. However, misoprostol, an activator of both EP2 and EP4 receptors, stimulated RANKL to a magnitude similar to that observed with PGE2.(41) Altogether, the findings show that the EP4 receptor is the most abundant isoform expressed by fibroblasts and that induction of RANKL is mediated by activation of this receptor isoform.
Because the EP4 receptor activates adenylate cyclase and PKA signaling,(46) the effect of inhibition of PKA (H89, 10 μM) and PKC (Go6976, 10 μM) signaling on PGE2 effects on fibroblasts was examined (Fig. 8C). Whereas the induction of RANKL by PGE2 was strongly inhibited by the H89, weak repression was observed with PKC inhibition.
These experiments define the fibroblast as an important mediator osteolysis adjacent to joint replacements. After stimulation with titanium particles, fibroblasts synthesize COX-2, secrete prostaglandins, and express RANKL. The induction of RANKL was functionally important because glutaraldehyde-fixed fibroblasts stimulated with titanium particles directly induced osteoclast formation. The induction of RANKL by particles could be blocked by both the nonselective nonsteroidal anti-inflammatory drug indomethacin and by the COX-2-specific inhibitor celecoxib and could be compensated for by the addition of PGE2 to the cultures. Analysis of EP receptor expression and the use of EP receptor agonists implicate the EP4 receptor as the major mediator of the prostaglandin effect. Inhibition of PKA signaling caused a strong suppression of RANKL in response to PGE2, consistent with an EP4 signaling event. Finally, addition of RANK-Fc to fibroblast-splenocyte co-cultures blocked osteoclast formation. Altogether the findings suggest that fibroblasts in the periprosthetic membrane are important targets in the cell and molecular events that regulate bone loss around total joint arthroplasties. The data also account for the lack of in vivo osteolysis observed in mice genetically deficient for COX-2 or RANK and wildtype mice treated with COX-2 or RANKL inhibitors.(19,27,37)
Multiple methods were used to confirm the induction of RANKL by titanium particles, including real time RT-PCR, Northern blot, and Western blot, and showed that the stimulation of the RANKL gene is linked to an increase in protein expression. Furthermore, fibroblasts stimulated with both titanium particles and PGE2 were able to induce osteoclast formation in splenocyte cultures in the presence of M-CSF. Because RANKL is necessary for the induction of osteoclasts, this is further evidence supporting the increase in expression and activity of this important mediator. Confirmatory evidence for the role of RANKL was provided by experiments showing inhibition of osteoclast formation in RANKL, PGE-2, and titanium treated cultures in the presence of RANK-Fc.
PGE2 is the primary product of arachidonic acid metabolism in many cells.(47) The COX enzymes mediate a critical step in prostaglandin biosynthesis and include two isoforms, COX-1, which is constitutive, and COX-2, which is inducible.(47–49) Nonsteroidal anti-inflammatory drugs, which block both COX-1 and COX-2, have been used successfully to reduce inflammatory responses to a variety of different stimuli.(49) More recently, specific inhibitors for COX-2 have been introduced that have the ability to selectively inhibit the COX-2 isoform and avoid potential complications associated with the blockade of COX-1-mediated events.(48,50)
Prior work in our laboratory showed an induction in PGE2 after treatment with titanium particles and suggested a COX-2-mediated effect.(18) Here we show that both COX-2 gene and protein expressions are induced by titanium particles. The initial increase in gene expression is associated with a subsequent elevation in protein levels and PGE2 secretion within 1 h after titanium exposure. The rapid response suggests a direct effect on gene transcription and is consistent with a role for COX-2 at the apex of a response leading to RANKL induction. Further evidence for the role of COX-2 is provided by the finding that both nonselective COX inhibition and selective COX-2 inhibition blocked RANKL stimulation by particles. Finally, the findings are consistent with prior work in our laboratory showing the requirement of COX-2 for in vivo bone loss in a murine model of titanium-induced osteolysis.(19)
There are four different types of G protein-coupled receptors for PGE2.(38,39) The various EP receptors activate different downstream signaling events and are differentially expressed on various cells.(38–40) EP1 is linked with Gq/p and stimulates phospholipase C activity and results in calcium transients and enhanced PKC signaling.(51) EP2 and EP4 are coupled with Gsα and stimulate adenylate cyclase to trigger the cAMP-PKA signaling pathway.(46) In contrast, the EP3 isoforms (α,β,γ) are primarily coupled to Gi and act to inhibit cAMP-PKA signaling, and thus have effects opposite those of the EP2 and EP4 receptors.(39,40)
Relatively low levels of EP1 and EP2 expression were present, EP3 receptor expression was not observed, and EP4 was the most abundantly expressed PGE2 receptor isoform. The importance of the EP4 receptor is supported by the induction of RANKL by misoprostol (EP2 and EP4 receptor agonist) to a level similar to that observed with PGE2. In contrast, the EP1 receptor agonist, iloprost, and the EP2 specific agonist, butaprost, minimally induced RANKL.(41) Altogether the findings most strongly implicate the EP4 receptor as the major target of COX/PGE2 effects on fibroblasts. However, our data do not exclude the possible contribution of other prostaglandin metabolites in the induction of RANKL downstream of COX-2, along with other potential receptors.
For these studies, we used an embryonic fibroblast cell line derived from 18-day-old embryos.(28) The cells represent a useful model because they respond to the inflammatory mediator, TNF-α, with an induction in VCAM1 expression as measured by FACS analysis. Thus, the cells express a synovial fibroblast phenotype.(52–54) The advantage of a cell line is that it allows analysis of critical intracellular signaling events in a stable population of cells over time. Confirmatory experiments were performed using synovial-like fibroblasts derived from the murine knee joint(29) and also showed a marked induction in RANKL expression by titanium particles and PGE2. Recent work has shown that fibroblasts in human periprosthetic tissues express RANKL, and we obtained similar findings in immunohistochemical stains of murine calvarial tissues 3 days after the implantation of titanium particles. Altogether these studies show a role for COX-2/PGE-2/EP4/PKA-mediated signals in the induction of RANKL.(20,21)
The commercially available titanium particles used in the experiments have previously been established as a model to study the biological response to debris particles both in vitro and in vivo.(18,27,30,55–57) Qualitatively similar cell-mediated inflammatory responses occur with a variety of particles, including native particles isolated from periprosthetic tissues as well as commercially prepared metal, polymethylmethacrylate, and polyethylene particles.(58,59) Whereas macrophages phagocytose particles, controversy remains regarding whether phagocytosis is required to incite the inflammatory response.(57,60) There is considerably less information regarding fibroblasts.(17,58) Whereas this study focused on particle-mediated RANKL induction and showed a mechanistic role for COX-2, PGE2, and subsequent signals, the mechanisms through which particles initiate and activate a cellular response in fibroblasts remains unknown.
Osteoclast formation requires interaction between stromal cells and preosteoclasts that are derived from cells of the macrophage/monocyte lineage.(61) Murine splenocyte cultures provide abundant preosteoclasts, and in the presence of M-CSF and RANKL, these cells readily form large, TRACP+ multinucleated osteoclasts.(33) The number of nuclei and the size of the cells that form with the addition of soluble recombinant RANKL are unique to cell culture and are not found under in vivo conditions. In contrast, the TRACP+ multinucleated osteoclasts that form in fibroblast co-cultures were smaller, but more closely resembled the size of osteoclasts found in vivo. The morphology of these cells was also similar to osteoclasts produced in fibroblast co-cultures by others.(62) The smaller number of osteoclasts in the co-cultures is likely caused by reduced concentrations and/or availability of expressed RANKL on the surface of the fibroblasts.
Various stromal cell populations express RANKL. RANKL is induced in osteoblasts by PTH, vitamin D3, and IL-6(22,61,63,64) and is inhibited by TGF-β.(65) Similar to our findings in fibroblasts, RANKL is induced in murine osteoblast cultures by PGE2 primarily through an EP4 receptor mechanism.(66,67) In human periodontal ligament fibroblasts, RANKL is induced by lipopolysaccharide (LPS) through a mechanism that involves the induction of IL-1 and TNF-α.(68) However, substantially less in known about factors regulating RANKL expression in fibroblast-like synoviocytes. Cultured rheumatoid fibroblast-like synoviocytes expressed RANKL in response to vitamin D3.(16) However, whereas high levels of RANKL are present in rheumatoid synovial tissue, minimal expression occurs in osteoarthritic synovium.(16) Thus, it is likely that synovial cells and tissue are involved in the pathogenesis of inflammatory bone resorption.(11,15) This study is the first directly showing RANKL stimulation in fibroblasts in response to titanium particles and that PGE2 is a stronger stimulus than the proinflammatory cytokines, IL-1, IL-6, and TNF-α, which are more frequently associated with osteolysis.
Despite the fact that proinflammatory cytokines are highly expressed in inflammatory tissues, little is known regarding the role of these factors in RANKL induction in synovial fibroblasts. Sen et al.(69) used human rheumatoid fibroblast-like synoviocytes to study the effects of Wnt 5A on IL-6, IL-15, and RANKL expression. Whereas inhibition of Wnt signaling blocked IL-6, IL-15, and RANKL expression, the effect of IL-6 on RANKL expression was not examined. Quinn et al.(62) showed that 1,25(OH)2D3 induced RANKL in skin fibroblasts and enhanced osteoclast formation in splenocyte and fibroblast co-culture. Although the effects of other factors on RANKL expression were not directly studied, the relative ability of TNF-α, IL-l, PTH-related protein (PTHrP), IL-11, PGE2, and vitamin D to induce osteoclast formation in co-culture was examined. IL-1 and TNF-α minimally increased the number of osteoclasts and were markedly less stimulatory than PGE2. Thus, although the information is indirect, it suggests that the proinflammatory cytokines minimally induce RANKL compared with PGE2 and are consistent with our findings.
Thus, evidence is accumulating that fibroblasts can regulate bone remodeling, particularly in the setting of inflammatory joint disease.(15,53) Synovial fibroblasts are the most abundant cell type found in the inflammatory membrane adjacent to loose implants and have extensive contact with wear debris particles.(9,10) Previously we have shown that wear debris-induced osteolysis is ablated in COX-2−/− mice and in wildtype mice with COX-2 inhibition.(19) Here we further our studies by implicating the synovial fibroblast as a potential target of COX-2 inhibition. The findings define the PGE2/EP4/PKA pathway as a potentially important target to regulate particle mediated osteolysis and have relevance for inflammatory bone loss in general.
The authors would like to acknowledge the help of Christine Clark and Lin Ma. Public Health Service Award AR46545 supported the experiments.
- 12000 Anti-TNF-alpha therapy as a clinical intervention for periprosthetic osteolysis. Arthritis Res 2: 165–168., ,
- 21998 Charnley total hip arthroplasty in patients less than fifty years old. A twenty to twenty-five-year follow-up note. J Bone Joint Surg Am 80: 704–714., , , ,
- 31999 Outcome of Charnley total hip replacement across a single health region in England: The results at five years from a regional hip register. J Bone Joint Surg Br 81: 577–581., ,
- 41994 Isolation and characterization of debris in membranes around total joint prostheses. J Bone Joint Surg Am 76: 1664–1675., , , ,
- 51996 Comparison and quantitation of wear debris of failed total hip and total knee arthroplasty. J Biomed Mater Res 31: 257–263., , ,
- 61996 Characterization and comparison of wear debris from failed total hip implants of different types. J Bone Joint Surg Am 78: 1235–1243., , , ,
- 72002 Volumetric computerized tomography as a measurement of periprosthetic acetabular osteolysis and its correlation with wear. Arthritis Res 4: 59–63., , , , , , , , , , ,
- 82001 Correlation of wear debris-induced osteolysis and revision with volumetric wear-rates of polyethylene: A survey of 8 reports in the literature. Acta Orthop Scand 72: 22–28., , ,
- 91993 Production of cytokines around loosened cemented acetabular components. Analysis with immunohistochemical techniques and in situ hybridization. J Bone Joint Surg Am 75: 863–879., , , , , , ,
- 101995 Analysis of cell types and mediator production from tissues around loosening joint implants. Br J Rheumatol 34: 1127–1134., , , ,
- 112002 The molecular mechanism of osteoclastogenesis in rheumatoid arthritis. Arthritis Res 4: 281–289., , , ,
- 122001 Activated human T cells directly induce osteoclastogenesis from human monocytes: Possible role of T cells in bone destruction in rheumatoid arthritis patients. Arthritis Rheum 44: 1003–1012., , , , , , , , , , , , , , , ,
- 132003 IL-6 and matrix metalloproteinase-1 are regulated by the cyclin-dependent kinase inhibitor p21 in synovial fibroblasts. J Immunol 170: 838–845., , , , , , , , , , ,
- 142003 Cartilage degradation and invasion by rheumatoid synovial fibroblasts is inhibited by gene transfer of TIMP-1 and TIMP-3. Gene Ther 10: 234–242., , , , , , , , , , ,
- 152000 Fibroblast biology. Effector signals released by the synovial fibroblast in arthritis. Arthritis Res 2: 356–360.
- 162000 Involvement of receptor activator of nuclear factor kappaB ligand/osteoclast differentiation factor in osteoclastogenesis from synoviocytes in rheumatoid arthritis. Arthritis Rheum 43: 259–269., , , , , , , , ,
- 171996 Particulate-Induced, Prostaglandin- and Cytokine-Mediated Bone Resorption in an Experimental System and in Failed Joint Replacements. Am J Ther 3: 27–41., , , , , , , , ,
- 182003 PGE2 and IL-6 production by fibroblasts in response to titanium wear debris particles is mediated through a COX-2 dependent pathway. J Orthop Res 22: 6–12., , , , , , , , ,
- 192001 Evidence for a direct role of cyclo-oxygenase 2 in implant wear debris-induced osteolysis. J Bone Miner Res 16: 660–670., , , , , , , , ,
- 202003 Imbalance of RANKL/RANK/OPG system in interface tissue in loosening of total hip replacement. J Bone Joint Surg Br 85: 1196–1201., , , , , ,
- 212002 Fibroblasts from the inner granulation tissue of the pseudocapsule in hips at revision arthroplasty induce osteoclast differentiation, as do stromal cells. Ann Rheum Dis 61: 103–109., , , , , , , ,
- 222004 PTH differentially regulates expression of RANKL and OPG. J Bone Miner Res 19: 235–244., , , , , ,
- 232002 IGF-I regulates osteoprotegerin (OPG) and receptor activator of nuclear factor-kappaB ligand in vitro and OPG in vivo. J Clin Endocrinol Metab 87: 4273–4279., , , , , , , , ,
- 241999 Tumor necrosis factor receptor family member RANK mediates osteoclast differentiation and activation induced by osteoprotegerin ligand. Proc Natl Acad Sci USA 96: 3540–3545., , , , , , , , , , , , , , , , , , , ,
- 251998 Osteoclast differentiation factor is a ligand for osteoprotegerin/osteoclastogenesis-inhibitory factor and is identical to TRANCE/RANKL. Proc Natl Acad Sci USA 95: 3597–3602., , , , , , , , , , , , , , ,
- 262001 Role of receptor activator of nuclear factor-kappaB ligand and osteoprotegerin in bone cell biology. J Mol Med 79: 243–253.,
- 272002 In vivo RANK signaling blockade using the receptor activator of NF-kappaB: Fc effectively prevents and ameliorates wear debris-induced osteolysis via osteoclast depletion without inhibiting osteogenesis. J Bone Miner Res 17: 192–199., , , , , , , , , ,
- 281999 Malignant transformation and antineoplastic actions of nonsteroidal antiinflammatory drugs (NSAIDs) on cyclooxygenase-null embryo fibroblasts. J Exp Med 190: 451–460., , ,
- 292000 NF-kappa B regulates VCAM-1 expression on fibroblast-like synoviocytes. J Immunol 164: 5990–5997., , ,
- 302001 Adherent endotoxin on orthopedic wear particles stimulates cytokine production and osteoclast differentiation. J Bone Miner Res 16: 2082–2091., , , , , ,
- 312001 PTHrP modulates chondrocyte differentiation through AP-1 and CREB signaling. J Biol Chem 276: 11639–11647., , , , , ,
- 322000 Protein expression and functional difference of membrane-bound and soluble receptor activator of NF-kappaB ligand: Modulation of the expression by osteotropic factors and cytokines. Biochem Biophys Res Commun 275: 768–775., , , , , ,
- 332004 RANK signaling is not required for TNFalpha-mediated increase in CD11(hi) osteoclast precursors but is essential for mature osteoclast formation in TNFalpha-mediated inflammatory arthritis. J Bone Miner Res 19: 207–213., , , , ,
- 342003 Metabolic acidosis stimulates RANKL RNA expression in bone through a cyclo-oxygenase-dependent mechanism. J Bone Miner Res 18: 1317–1325.,
- 352001 PGE(2) is essential for gap junction-mediated intercellular communication between osteocyte-like MLO-Y4 cells in response to mechanical strain. Endocrinology 142: 3464–3473., , , , , ,
- 361999 Immunohistochemical evaluation of interface membranes from failed cemented and uncemented acetabular components. J Biomed Mater Res 48: 889–898., ,
- 372002 Recombinant adeno-associated virus-mediated osteoprotegerin gene therapy inhibits wear debris-induced osteolysis. J Bone Joint Surg Am 84: 1405–1412., , , , ,
- 382003 Prostaglandin E2 induced functional expression of early growth response factor-1 by EP4, but not EP2, prostanoid receptors via the phosphatidylinositol 3-kinase and extracellular signal-regulated kinases. J Biol Chem 278: 12151–12156., ,
- 392002 Prostaglandin receptors: Advances in the study of EP3 receptor signaling. J Biochem (Tokyo) 131: 781–784., ,
- 402002 Receptors and signaling mechanisms required for prostaglandin E2-mediated regulation of mast cell degranulation and IL-6 production. J Immunol 169: 4586–4593., , , , , , , ,
- 411997 Ligand binding specificities of the eight types and subtypes of the mouse prostanoid receptors expressed in Chinese hamster ovary cells. Br J Pharmacol 122: 217–224., , , , ,
- 421997 EP1 receptor blockade attenuates both spontaneous tone and PGE2-elicited contraction in guinea pig trachealis. Am J Physiol 273: L626–L633., , ,
- 431998 Prostanoid receptors of the EP3 subtype mediate inhibition of evoked [3H]acetylcholine release from isolated human bronchi. Br J Pharmacol 125: 271–276., , ,
- 442001 Prostanoid EP(1)- and TP-receptors involved in the contraction of human pulmonary veins. Br J Pharmacol 134: 1671–1678., , ,
- 451996 Anabolic effects of prostaglandins in cultured fetal rat calvariae: Structure-activity relations and signal transduction pathway. J Bone Miner Res 11: 1249–1255., ,
- 462003 EP2 and EP4 prostanoid receptor signaling. Life Sci 74: 143–153.
- 472001 Roles of cyclooxygenase (COX)-1 and COX-2 in prostanoid production by human endothelial cells: Selective up-regulation of prostacyclin synthesis by COX-2. J Immunol 167: 2831–2838., , , ,
- 482002 Distinct functions of COX-1 and COX-2. Prostaglandins Other Lipid Mediat 68–69: 165–175.
- 492000 Cyclooxygenases: Structural, cellular, and molecular biology. Annu Rev Biochem 69: 145–182., ,
- 502000 COX-2-Specific inhibitors-the emergence of a new class of analgesic and anti-inflammatory drugs. Clin Rheumatol 19: 331–343., ,
- 512001 EP1 and EP4 receptors mediate exocytosis evoked by prostaglandin E(2) in guinea-pig antral mucous cells. Exp Physiol 86: 451–460., , , , ,
- 522004 Three-dimensional expression of adhesion molecules on the superficial synovial intima of LPS-induced arthritis of the mouse knee analyzed by immuno-SEM. J Electron Microsc (Tokyo) 53: 93–97., , , , ,
- 532003 Rheumatoid fibroblast-like synoviocytes overexpress the chemokine stromal cell-derived factor 1 (CXCL12), which supports distinct patterns and rates of CD4+ and CD8+ T cell migration within synovial tissue. Arthritis Rheum 48: 2472–2482., , , , , , , , , ,
- 541999 Sustained elevated levels of VCAM-1 in cultured fibroblast-like synoviocytes can be achieved by TNF-alpha in combination with either IL-4 or IL-13 through increased mRNA stability. Am J Pathol 154: 1149–1158., , ,
- 552001 Efficacy of Etanercept for wear debris-induced osteolysis. J Bone Miner Res 16: 338–347., , ,
- 562000 Phagocytosis of wear debris by osteoblasts affects differentiation and local factor production in a manner dependent on particle composition. Biomaterials 21: 551–561., , , , , , , , , ,
- 572004 Effects of interleukin-10 on titanium particle-induced macrophage transcription factor activation and cytokine expression in vitro. J Biomed Mater Res 69A: 40–46., , , , ,
- 582001 Effects of hydroxyapatite particulate debris on the production of cytokines and proteases in human fibroblasts. J Orthop Res 19: 621–628., , , ,
- 592002 Inflammatory responses to orthopaedic biomaterials in the murine air pouch. Biomaterials 23: 517–526., , , , , , , ,
- 601999 Signaling pathways for tumor necrosis factor-alpha and interleukin-6 expression in human macrophages exposed to titanium-alloy particulate debris in vitro. J Bone Joint Surg Am 81: 603–615., , , , , ,
- 612003 Vitamin D3 supports osteoclastogenesis via functional vitamin D response element of human RANKL gene promoter. J Cell Biochem 89: 771–777., , ,
- 622000 Fibroblastic stromal cells express receptor activator of NF-kappa B ligand and support osteoclast differentiation. J Bone Miner Res 15: 1459–1466., , , ,
- 632002 IL-6, leukemia inhibitory factor, and oncostatin M stimulate bone resorption and regulate the expression of receptor activator of NF-kappa B ligand, osteoprotegerin, and receptor activator of NF-kappa B in mouse calvariae. J Immunol 169: 3353–3362., , ,
- 642003 RANKL expression is related to the differentiation state of human osteoblasts. J Bone Miner Res 18: 1088–1098., , , , , , , ,
- 652001 Transforming growth factor beta affects osteoclast differentiation via direct and indirect actions. J Bone Miner Res 16: 1787–1794., , , , , , , , , , ,
- 662002 Effects of prostaglandin E2 on gene expression in primary osteoblastic cells from prostaglandin receptor knockout mice. Bone 30: 567–573., , , ,
- 672002 Prostaglandin receptor EP2 mediates PGE2 stimulated hypercalcemia in mice in vivo. Prostaglandins Other Lipid Mediat 67: 173–180., , , ,
- 682004 Lipopolysaccharide stimulates expression of osteoprotegerin and receptor activator of NF-kappa B ligand in periodontal ligament fibroblasts through the induction of interleukin-1 beta and tumor necrosis factor-alpha. Bone 35: 629–635., , ,
- 692001 Blockade of Wnt-5A/frizzled 5 signaling inhibits rheumatoid synoviocyte activation. Arthritis Rheum 44: 772–781., , , ,