The Department of Immunology at Erasmus University Medical Center receives royalties for commercialized antibodies used in this study.
S100A8 enhances osteoclastic bone resorption in vitro through activation of Toll-like receptor 4: Implications for bone destruction in murine antigen-induced arthritis†
Version of Record online: 27 APR 2011
Copyright © 2011 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 63, Issue 5, pages 1365–1375, May 2011
How to Cite
Grevers, L. C., de Vries, T. J., Vogl, T., Abdollahi-Roodsaz, S., Sloetjes, A. W., Leenen, P. J. M., Roth, J., Everts, V., van den Berg, W. B. and van Lent, P. L. E. M. (2011), S100A8 enhances osteoclastic bone resorption in vitro through activation of Toll-like receptor 4: Implications for bone destruction in murine antigen-induced arthritis. Arthritis & Rheumatism, 63: 1365–1375. doi: 10.1002/art.30290
- Issue online: 27 APR 2011
- Version of Record online: 27 APR 2011
- Accepted manuscript online: 17 FEB 2011 12:07PM EST
- Manuscript Accepted: 1 FEB 2011
- Manuscript Received: 31 AUG 2010
- European Union Sixth Framework Programme. Grant Number: AutoCure project grant RI000101
- Dutch Arthritis Association. Grant Number: 07-2-301
- Top Institute Pharma. Grant Number: project D1-101
- German Ministry of Education and Research. Grant Number: BMBF project AID-NET
Rheumatoid arthritis, which is associated with elevated levels of S100A8 and S100A9, is characterized by severe bone erosions caused by enhanced osteoclast formation and activity. The aim of the present study was to investigate the role of S100A8 and S100A9 in osteoclastic bone destruction in murine antigen-induced arthritis (AIA).
Bone destruction was analyzed in the arthritic knee joints of S100A9-deficient mice in which S100A8 protein expression was also lacking, and in wild-type (WT) controls. Osteoclast precursors from S100A9-deficient and WT mice were differentiated into osteoclasts in vitro. Additionally, precursors were stimulated with S100A8, S100A9, or S100A8/A9 during osteoclastogenesis. Receptor involvement was investigated using an anti–receptor for advanced glycation end products (anti-RAGE)–blocking antibody, soluble RAGE, or Toll-like receptor 4 (TLR-4)–deficient osteoclast precursors. The formation of osteoclasts and actin rings, the regulation of osteoclast markers, and bone resorption were analyzed.
Bone erosions and cathepsin K staining were significantly suppressed in S100A9-deficient mice after AIA induction. However, osteoclast precursors from S100A9-deficient mice developed normally into functional osteoclasts, which excludes a role for intrinsic S100A8/A9. In contrast to the results observed with S100A9 and S100A8/A9, the addition of S100A8 during osteoclastogenesis resulted in stimulation of osteoclast formation in conjunction with enhanced actin ring formation and increased bone resorption. Analysis of the putative receptor for S100A8 in osteoclastogenesis revealed that osteoclast differentiation and function could not be inhibited by blocking RAGE, whereas the increase in osteoclast numbers and enhanced bone resorption were completely abrogated using TLR-4–deficient osteoclast precursors.
These results demonstrate that S100A8 stimulated osteoclast formation and activity and suggest that both S100A8 and TLR-4 are important factors in mediating osteoclastic bone destruction in experimental arthritis.
Bone erosions are an important hallmark of rheumatoid arthritis (RA) and typically result from enhanced formation and activity of osteoclasts in affected joints. Osteoclasts originate from hematopoietic precursors of the monocyte/macrophage lineage that differentiate into multinucleated osteoclasts under the influence of macrophage colony-stimulating factor (M-CSF) and RANKL (1).
The interdisciplinary research field of osteoimmunology, combining osteoclast biology and immunology, has contributed significantly to the understanding of bone loss in RA, in which the inflamed synovium is an active site of interplay between immune and bone cells. Arthritic synovium is enriched with activated macrophages, which are strong producers of inflammation mediators that also drive osteoclastogenesis, such as interleukin-1 and tumor necrosis factor α (2).
The “alarmins” S100A8 and S100A9, which are key players in the initiation and amplification of inflammatory responses (3), are the most up-regulated proteins present in RA synovial fluid, and they are significantly correlated with arthritis severity and joint degradation (4–6). S100 proteins are low molecular weight calcium-binding proteins produced mainly by monocytes, macrophages, and neutrophils and are secreted upon activation. In general, S100A8 and S100A9 are coexpressed and form a heterodimer complex (S100A8/A9) in which S100A8 is the active component and S100A9 stabilizes S100A8 to prevent its degradation. Besides heterodimers, homodimers can also be formed and may exert functions other than the heterodimer complex.
Targeted deletion of the S100A9 gene results in loss of expression of both S100A8 and S100A9 protein, probably due to rapid degradation of S100A8 because its binding partner S100A9 is absent (7, 8). S100 proteins are implicated in various intracellular and extracellular functions. Intracellularly, they maintain cell homeostasis by binding calcium and regulating cell migration and tubulin polymerization. Once these proteins are secreted, they can mediate inflammatory responses by exerting cytokine-like functions, resulting in the attraction and activation of leukocytes (9, 10).
The receptor through which S100A8 and S100A9 exert their effect is still an area of debate. Promotion of tumor cell growth by S100A8/A9 was shown to be mediated by the receptor for advanced glycation end products (RAGE) (11). Recently, S100A8 and the S100A8/A9 complex were identified as endogenous ligands of Toll-like receptor 4 (TLR-4) (12). TLRs belong to the family of pattern recognition receptors, recognizing both exogenous pathogen-associated molecular patterns and endogenous damage-associated molecular patterns, to which the alarmins S100A8 and S100A9 belong (13).
Both TLR-4 and RAGE are important for osteoclast differentiation and function. RAGE-deficient mice display increased bone mineral density and decreased bone resorptive activity in vivo. Bone marrow–derived precursors exhibit impaired osteoclast formation and reduced bone resorptive activity in vitro (14). Local injections of the bacterial endotoxin lipopolysaccharide (LPS) strongly increase osteoclast numbers and bone resorption; this is blocked in C3H/HeJ mice carrying a single point mutation of an amino acid in the cytoplasmic domain of TLR-4, resulting in impaired intracellular signal transduction (15).
Using S100A9-deficient mice in which S100A8 protein expression was also lacking, we previously showed that S100A8 and S100A9 are important mediators of matrix metalloproteinase–mediated cartilage destruction in antigen-induced arthritis (AIA) (16). Prominent expression of S100A8 and S100A9 in osteoclasts was previously reported (17), but the function of these proteins in osteoclasts and their role in arthritic bone loss have not been investigated. The aim of the present study was to investigate the role of S100A8 and S100A9 in osteoclast-mediated bone destruction during AIA as well as their role in osteoclastogenesis and the activity of osteoclasts.
MATERIALS AND METHODS
S100A9-deficient mice were generated as previously described (7) and backcrossed to the C57BL/6 background for 10 generations. TLR-4–deficient mice on the C57BL/6 background were provided by Professor S. Akira (Research Institute for Microbial Diseases, Osaka University, Osaka, Japan). Control C57BL/6 mice were purchased from Janvier-Elevage. Mice were housed in filter-top cages, and a standard diet and water were provided ad libitum. All animal studies were approved by the institutional research ethics committee.
Induction of AIA.
Mice were immunized with 100 μg of methylated bovine serum albumin (mBSA; Sigma-Aldrich) emulsified in 100 μl of Freund's complete adjuvant (CFA; Difco). Injections were divided over both flanks and footpads of the fore legs. Heat-killed Bordetella pertussis (National Institute for Public Health and the Environment [RIVM]) was administered intraperitoneally as an additional adjuvant. Two subcutaneous booster injections with, in total, 50 μg of mBSA/CFA were administered in the neck region 1 week after the initial immunization. Three weeks after these injections were given, AIA was induced by injecting 60 μg of mBSA in 6 μl of phosphate buffered saline (PBS) directly into the knee joint, resulting in chronic arthritis.
Total knee joints were isolated, fixed in 4% formalin for 4 days, decalcified in formic acid, dehydrated, and embedded in paraffin. Frontal sections (7 μm) were prepared and stained with hematoxylin and eosin to study bone erosion, which was graded on a scale ranging from 0 (no erosion) to 3 (complete bone loss). Osteoclast activity was visualized by immunohistochemistry for the osteoclast activity marker cathepsin K. Sections were deparaffinized, rehydrated, and treated with 3% H2O2 in methanol for 15 minutes. Subsequently, sections were preincubated with 10% casein in PBS for 30 minutes and thereafter were incubated for 1 hour with the primary antibody rabbit anti–cathepsin K (Abcam). As a control, sections were incubated with normal rabbit IgG. Additionally, sections were incubated with the secondary antibody biotinylated goat anti-rabbit IgG, and binding was detected using the Vectastain Elite ABC–horseradish peroxidase kit (Vector). Peroxidase was developed with diaminobenzidine, and sections were counterstained with hematoxylin for 1 minute.
Bone marrow isolation and osteoclastogenesis.
Bone marrow was isolated from the femora and tibiae by flushing the marrow cavity with culture medium (α-minimal essential medium [Invitrogen] supplemented with 5% fetal calf serum, 100 units/ml penicillin, and 100 μg/ml streptomycin), using a sterile syringe with a 21-gauge needle. Single-cell suspensions were obtained by gently aspirating the cell clumps through the syringe. Bone marrow cells were washed and plated into 96-well plates at a density of 1 × 105/well in 150-μl culture medium containing 30 ng/ml of recombinant murine M-CSF (R&D Systems) with or without 40 ng/ml of recombinant murine RANKL (RANKL-TEC; R&D Systems). Additionally, cells were seeded on 650-μm–thick bovine cortical bone slices. Culture media were replaced every 3 days. S100A8 protein was measured in supernatants of S100A9-deficient cultures by enzyme-linked immunosorbent assay (ELISA) as described previously (12). Recombinant murine S100A8, S100A9, or the S100A8/A9 heterodimer, purified as previously described (18), was added to cell cultures from day 4 onward. LPS contamination was tested by Limulus amebocyte cell lysate assay. LPS was not detectable (approximate mean ± SEM sensitivity 0.7 ± 0.5 pg LPS/μg protein), indicating that the maximal possible contamination of S100 protein preparations was <1.2 pg LPS/μg protein. Purified Escherichia coli LPS (1.0 μg/ml; Sigma-Aldrich) was used as a control where indicated. RAGE ligation was blocked by incubation with a specific RAGE-blocking antibody (166 μg/ml; R&D Systems) 1 hour prior to the addition of S100A8, according to the manufacturer's protocol. Additionally, 40 μg/ml of recombinant human soluble RAGE, generated as previously reported (19), was preincubated with S100A8, 1 hour prior to the addition to osteoclast cultures.
Filamentous actin (F-actin) staining and confocal microscopy.
Actin ring formation in osteoclasts cultured on bone was analyzed using confocal microscopy. Osteoclast membranes were visualized by CD44 staining, as previously described (20). Bone slices were washed in PBS and fixed in 4% PBS-buffered paraformaldehyde for 10 minutes. Nonspecific binding was blocked for 30 minutes with 10% normal goat serum, followed by 1-hour incubation with rat anti-CD44 (IM7.8.1; Cedarlane). Bound antibody was visualized with Alexa Fluor 647–conjugated goat anti-rat IgG (Invitrogen). F-actin was stained using Alexa Fluor 488–conjugated phalloidin (Invitrogen). Nuclei were stained with propidium iodide (Sigma-Aldrich). Image stacks were generated with a Leica TCS-SP2 confocal laser scanning microscope, using an argon laser (for Alexa Fluor 488 and propidium iodide) and a helium laser (for Alexa Fluor 647).
Fluorescence-activated cell sorting analysis.
Six different bone marrow subsets were analyzed by 2-color flow cytometry, as previously described (21). Briefly, freshly isolated bone marrow cells were washed in 1× PBS/1% BSA and labeled with biotinylated monoclonal antibody ER-MP12, recognizing CD31. After washing, cells were labeled with fluorescein isothiocyanate–conjugated ER-MP20, recognizing lymphocyte antigen 6 complex, locus C (Ly-6C), and streptavidin–phycoerythrin conjugate (BioLegend). Fluorescence was analyzed using a fluorescence-activated cell sorter (FACSCalibur; BD Biosciences) and FlowJo software (Tree Star).
Tartrate-resistant acid phosphatase (TRAP) staining and cell count.
After 6 days or 9 days of culture, the number of TRAP-positive multinucleated cells was determined. Cells were washed in PBS, fixed in 4% PBS-buffered formaldehyde for 10 minutes, and stained for TRAP activity, using the Leukocyte Acid Phosphatase Kit (Sigma-Aldrich) according to the manufacturer's protocol. Only cells with ≥3 nuclei were considered to be osteoclasts. Based on the number of nuclei per cell, each osteoclast was assigned to 1 of 3 groups (3–5 nuclei, 6–10 nuclei, or >10 nuclei), because the number of nuclei may reflect the maturity of the osteoclast (22).
Osteoclast resorption assay.
To determine resorption pit formation by osteoclasts differentiated on bone slices, cells were lysed in water, and cell remnants were mechanically removed by sonicating the bone slices in 10% ammonia for 30 minutes. The slices were thoroughly washed and incubated in a 10% saturated alum (Kal[SO4]2 · 12H2O) solution for 10 minutes. Finally, slices were washed again, and resorption pits were stained with Coomassie brilliant blue (PhastGel Blue R-350; GE Healthcare). Five micrographs per slice from predetermined positions were obtained with a digital camera mounted on an inverted light microscope. The percentage of resorbed bone area was quantified using the QWin image analysis system (Leica).
RNA isolation and quantitative polymerase chain reaction (qPCR).
Total RNA of cultured cells was isolated using TRIzol reagent and reverse transcribed into complementary DNA (cDNA) with Moloney murine leukemia virus reverse transcriptase, oligo(dT) primers, and dNTP (Invitrogen). Quantitative real-time PCR was performed using the ABI PRISM 7000 Sequence Detection System (Applied Biosystems). The qPCR amplification protocol was as follows: 2 minutes at 50°C, 10 minutes at 95°C, followed by 40 cycles of 15 seconds at 95°C and 1 minute at 60°C, with data collection during the last 30 seconds. Product specificity was confirmed by postamplification dissociation curve analysis. Quantitative PCR was performed in a total volume of 20 μl, containing 4 μl of cDNA, 1.2 μl of forward primer (5 μM), 1.2 μl of reverse primer (5 μM), 10 μl of SYBR Green Master Mix (Applied Biosystems), and 3.6 μl of distilled water. Samples were normalized for GAPDH expression by calculating the difference in the threshold cycle (ΔCt = Ct GAPDH − Ct target gene). Relative gene expression was calculated as 2.
Statistical analysis was performed using GraphPad Prism version 4.0 (GraphPad Software). Student's 2-tailed t-test was performed for comparisons between wild-type (WT) and knockout mice regarding the number of TRAP-positive multinucleated cells and for analysis of actin ring formation. Bone resorption levels were analyzed by Mann-Whitney U test. Results are expressed as the mean ± SEM. P values less than 0.05 were considered significant.
Reduced bone erosions in S100A8/A9-deficient mice with AIA.
To study the involvement of S100A8 and S100A9 in osteoclast-mediated bone destruction during experimental arthritis, total knee joints of S100A9-deficient and WT mice were isolated 7 days and 21 days after induction of AIA and processed for histologic analysis. Seven and 21 days after AIA induction, bone erosions were significantly suppressed in S100A9-deficient mice compared with WT controls (58% and 56% reduction, respectively) (Figure 1A). Additionally, immunohistochemical staining for the osteoclast activity marker cathepsin K displayed significantly decreased numbers of osteoclasts in the knee joints of S100A9-deficient mice 7 days after AIA induction (Figure 1B and C). Twenty-one days after AIA induction, the number of osteoclasts within the knee joints was strongly decreased in both WT and knockout mice, and the differences were no longer significant (Figure 1B).
Normal development of S100A8/A9-deficient osteoclast precursors into functional osteoclasts.
To determine whether intrinsic lack of S100A8 and S100A9 affects the differentiation and function of osteoclasts, bone marrow was isolated from S100A9-deficient mice and WT mice and differentiated toward osteoclasts under the influence of M-CSF and RANKL. First, we used ELISA to measure S100A8 protein production by S100A9-deficient osteoclasts. Consistent with studies demonstrating the absence of S100A8 protein expression in the myeloid cells of S100A9-deficient mice (7, 8), S100A8 was not detected in supernatants of S100A9-deficient osteoclast cultures (data not shown). Next, we determined the composition of osteoclast precursors in the bone marrow, using flow cytometry. Using monoclonal antibodies recognizing CD31 (ER-MP12) and Ly-6C (ER-MP20) (21), 6 subpopulations were discriminated, of which early blast cells (CD31high/Ly-6C−), myeloblasts (CD31+/Ly-6C+), and monocytes (CD31−/Ly-6Chigh) are the populations that give rise to osteoclasts (23). Compared with WT mice, S100A9-deficient mice displayed increased percentages of myeloblasts, monocytes, and granulocytes (mean ± SEM 6.5 ± 0.3% versus 8.1 ± 0.2%, 6.7 ± 0.3% versus 8.3 ± 0.4%, and 23.5 ± 0.9% versus 27.4 ± 0.7%, respectively, for WT versus S100A9-deficient bone marrow cells), whereas no differences were detected in the composition of the remaining subpopulations (Figure 2A).
Despite subtle differences in bone marrow composition, in vitro osteoclastogenesis led to the formation of comparable osteoclast numbers in the cultures of S100A9-deficient mice and WT mice, and no differences were observed in the average number of nuclei per cell (Figures 2B and D). Furthermore, messenger RNA (mRNA) expression levels of the osteoclast markers nuclear factor of activated T cells c1 (NF-ATc1), dendritic cell–specific transmembrane protein (DC-STAMP), TRAP, calcitonin receptor, and cathepsin K were similar in control and S100A9-deficient osteoclasts (data not shown). Finally, osteoclast function was investigated by analysis of resorption pit formation 6 days and 9 days after osteoclast differentiation on bone slices. As shown in Figures 2C and D, no differences were observed in the formation of resorption pits generated by bone marrow–derived osteoclasts from S100A9-deficient mice and WT controls.
Stimulation of osteoclast formation and bone resorption by S100A8.
The previous results indicated that intrinsic S100A8 and S100A9 production by precursors is not required for normal osteoclast differentiation and function. In RA, however, S100 proteins are produced in excessive amounts by activated macrophages and polymorphonuclear cells, which may affect differentiation of precursors attracted to the site of inflammation. Therefore, we investigated the effect of S100A8, S100A9, and the S100A8/A9 complex on osteoclast differentiation and function. The addition of S100A8 on day 4 of differentiation resulted in significantly enhanced, dose-dependent formation of TRAP-positive multinucleated cells (Figure 3A). Next, mRNA levels of osteoclast markers were determined by real-time PCR, which showed that the expression of osteoclast markers was not changed after stimulation with S100A8 (Figure 3B). To determine whether the changes in osteoclast differentiation also affected function, osteoclasts were differentiated on bone in the presence of S100A8. After 6 days of culture, the percentage of bone resorbed by S100A8-stimulated osteoclasts was dose-dependently increased as compared with unstimulated controls and reached significance with doses of 1.0 μg/ml and 5.0 μg/ml S100A8 (Figures 3C and D).
In contrast, the addition of S100A9 or the S100A8/A9 heterodimer to osteoclastogenesis cultures resulted in only a minor increase in the formation of TRAP-positive multinucleated cells containing 3–5 nuclei per cell, whereas expression levels of osteoclast markers and resorption pit formation on bone were not affected by these stimuli (data not shown). Because these results suggest that stimulation of osteoclast formation and bone resorption is specific for S100A8, further research was focused on investigating the effects of S100A8 at a dose of 1.0 μg/ml.
Enhanced actin ring formation in S100A8-stimulated osteoclasts.
In the presence of S100A8, a marked change in osteoclast morphology was observed. Stimulated cells proved to be more stretched out than unstimulated controls (Figure 3D), suggesting a role for S100A8 in the organization of the osteoclast cytoskeleton. Actin ring formation is highly characteristic for osteoclasts and is a prerequisite for bone resorption (24). To study the effect of S100A8 stimulation on actin ring formation, F-actin in osteoclasts cultured on bone was stained using Alexa Fluor 488–conjugated phalloidin and analyzed with confocal microscopy. The changes in osteoclast shape when culture was performed on plastic were not present when osteoclasts were cultured on bone. Osteoclast size was shown to be comparable for control and S100A8-stimulated osteoclasts (Figures 4A and B). However, the number of actin rings per cell in S100A8-stimulated osteoclasts was significantly increased compared with unstimulated controls (Figures 4A and C).
Mediation of S100A8-stimulated osteoclast differentiation and function by TLR-4.
The receptor through which S100A8 mediates its effect is still a matter of debate. Involvement of RAGE has been demonstrated, whereas it was recently shown that S100A8 mediates its effects in macrophages via TLR-4. To study the role of RAGE, osteoclast precursors from WT C57BL/6 mice were stimulated with S100A8 in the presence of a specific RAGE-blocking antibody or soluble RAGE. Blockade of RAGE, however, did not suppress S100A8-enhanced formation of TRAP-positive multinucleated cells and bone resorption (Figure 5), indicating that, on osteoclasts, RAGE is not the receptor via which S100A8 mediates its effects.
Next, we investigated the role of TLR-4 in S100A8-stimulated osteoclast formation. Bone marrow from TLR-4–deficient mice and their WT littermates (TRL-4 intact) was isolated and differentiated into osteoclasts in the presence or absence of S100A8. Stimulation with 1.0 μg/ml of purified E coli LPS was used as a positive control for TLR-4 activation. The composition of bone marrow subpopulations, as identified by ER-MP12/ER-MP20 staining, was similar for WT and TLR-4–deficient mice (data not shown). On day 6 of culture, the formation of TRAP-positive multinucleated cells was analyzed, RNA was isolated, and resorption pit formation on bone was quantified. Stimulation of TLR-4–intact mouse osteoclasts with S100A8 increased the formation of TRAP-positive multinucleated cells (Figure 6A), and predominantly of cells containing 3–5 and 6–10 nuclei. S100A8-stimulated osteoclasts had an increased tendency to be more stretched out. In contrast, S100A8 stimulation of TLR-4–deficient mouse osteoclasts altered neither the number nor the shape of osteoclasts (Figures 6A and B). Finally, analysis of bone resorption capacity showed that bone resorption levels were significantly increased in S100A8-stimulated TLR-4–intact mouse osteoclasts, whereas this effect was completely blocked in S100A8-stimulated TLR-4–deficient mouse osteoclasts (Figure 6C).
In the present study, we demonstrated for the first time that the alarmin S100A8 stimulates both osteoclast formation and osteoclast function. Because S100A8 levels are significantly increased in arthritic joints and are strongly correlated with joint destruction, these results indicate that S100A8 is an important factor in mediating osteoclastic bone destruction in RA.
The observation that both S100A8 and S100A9 are strongly expressed in osteoclasts suggested that these proteins might play an important role in the maturation and function of osteoclasts and skeletal development (17). This idea was further supported by the finding of significantly reduced bone erosion levels after AIA induction in S100A9-deficient mice compared with WT controls. Immunolocalization of the osteoclast marker cathepsin K showed significantly reduced osteoclast numbers in S100A9-deficient mice 7 days after AIA induction. Twenty-one days after AIA induction, osteoclast numbers were strongly reduced, and the difference between WT and knockout mice was no longer significant, indicating that in this arthritis model, mediation of osteoclastic bone destruction is transient and seems to occur primarily in the early phase of disease.
To further investigate the role of S100A8 and S100A9 in osteoclasts, we first studied the role of these S100 proteins in normal osteoclast differentiation and function. Analysis of bone marrow subpopulations revealed increased percentages of 2 fractions that contain osteoclast precursors (myeloblasts and monocytes) in S100A9-deficient mice. Despite these differences and the lack of both S100A8 and S100A9 protein expression, osteoclast precursors from S100A9-deficient mice developed normally into TRAP-positive multinucleated osteoclasts, and the bone resorption capacity of these osteoclasts was not affected. The possibility of differences in maturation kinetics was excluded, because different time points after the start of differentiation were analyzed, and none of these time points displayed differences between S100A9-deficient and WT osteoclasts. These findings are in accordance with those from previous studies, demonstrating subtle differences in bone marrow composition that do not result in abnormal development of S100A9-deficient mice; furthermore, no abnormalities in bone phenotype have been described (7, 8). From these findings, it can be concluded that S100A8 and S100A9 are not required for normal physiologic development of osteoclasts, and the increased number of osteoclast precursors present within the bone marrow compartment appears to be irrelevant for bone mass levels in vivo.
In RA, proliferation of inflamed synovial lining cells and infiltration of activated macrophages result in the formation of pannus that covers the surface of cartilage and bone, establishing close interactions between immune and bone cells. Activated macrophages in the pannus tissue produce multiple factors that control both immunity and bone homeostasis, creating an imbalance between bone formation and bone resorption in favor of the latter, resulting in severe bone destruction (2, 25). Of the various factors produced in the inflamed synovium, S100A8 and S100A9 are the most up-regulated, and they are produced in excessive amounts (4–6). In the S100A8/A9 heterodimer complex, S100A8 is the active component, and S100A9 stabilizes S100A8 to prevent its degradation (7). Recently, we demonstrated that in murine macrophages, S100A8, independent of complex formation with S100A9, up-regulated activating Fcγ receptors and metalloproteinases, which are important mediators of cartilage destruction in arthritis (26, 27). Additionally, we demonstrated that S100A9 and the heterodimer complex were much less potent (16, 28). Because bone erosion levels were strongly reduced in arthritic S100A9-deficient mice that also lacked S100A8 protein expression, we hypothesize that exogenous S100A8, produced by activated macrophages, stimulates osteoclast differentiation and the function of precursors that are attracted to the site of inflammation.
Our hypothesis is supported by the finding that stimulation of osteoclastogenesis with S100A8 resulted in dose-dependently increased osteoclast numbers and bone resorption, whereas stimulation with S100A9 or the S100A8/A9 complex affected neither osteoclast formation nor osteoclast function; these observations are similar to our previous findings in macrophages where S100A8 was also more potent than S100A9 and the heterodimer complex (28). The stimulatory effects of S100A8 cannot be explained by the induction of genes known to be important for the differentiation and function of osteoclasts, such as NF-ATc1 (29, 30), DC-STAMP (31), TRAP (32), and cathepsin K (33), because their expression was not regulated by S100A8 stimulation. However, we did notice a marked change in the shape of the osteoclasts. Osteoclast differentiation in the presence of S100A8 resulted in osteoclasts that appeared to be more stretched out compared with unstimulated control osteoclasts.
Regulation of cell morphology is a complex process involving components of the cytoskeleton, including actin microfilaments, microtubules, and intermediate filaments (34). Attachment of osteoclasts to the bone matrix induces their activation and leads to reorganization of the actin cytoskeleton, formation of sealing zones, and a polarized ruffled membrane juxtaposed to bone. The formation of actin rings is highly characteristic for osteoclasts and is related to their capacity to resorb bone (24, 35). It was previously demonstrated that S100A9, in complex with S100A8, promoted polymerization of microtubules, and that S100A9-deficient granulocytes contained significantly less polymerized tubulin (36). Although binding of S100 proteins to other components of the cytoskeleton, such as keratin, vimentin, and F-actin, has been reported (37–39), the functional consequences of these interactions have not been thoroughly investigated.
Because the differentiation and bone-resorptive capacity of S100A8/A9-deficient osteoclasts were normal, it is not likely that osteoclast morphology is regulated by direct interactions of S100A8 with cytoskeletal components. However, the clear changes in the shape of S100A8-stimulated osteoclasts did motivate us to further investigate whether external osteoclast activation with S100A8 could stimulate actin ring formation in osteoclasts.
Culturing osteoclasts on bone was different from culturing osteoclasts on plastic, because the differences in morphology were less prominent, and osteoclast size was comparable when cells were cultured on bone. These findings stress earlier studies on the impact of mineral substrates on osteoclastogenesis, which showed increased osteoclast formation of osteopontin-deficient cells when culture was performed on plastic that was no longer present when culture was performed on bone (40). However, F-actin staining demonstrated that S100A8-stimulated osteoclasts contained significantly more actin rings per cell. Because bone resorption occurs only when a sealing zone is formed and the actin ring is present, creating the acidic environment crucial for cathepsin K activity (41–43), this suggests that S100A8 enhanced the bone resorptive capacity per osteoclast.
Finally, we investigated which receptor regulated the S100A8-mediated effects on osteoclasts. Putative receptors described for S100A8 are N-glycans (44), RAGE (11), and TLR-4 (12). Although binding of S100A8 to each of these receptors has been demonstrated, functional relevance has been observed only for its ligation to RAGE and TLR-4. To investigate the role of RAGE in S100A8 stimulation of osteoclasts, RAGE ligation was blocked either using a specific blocking antibody or by preincubation of S100A8 with soluble RAGE. However, neither method suppressed the increased formation of osteoclasts and enhanced bone resorption, suggesting that another receptor mediates the stimulatory effects of S100A8 on osteoclasts.
Another potential candidate for an S100A8 receptor is TLR-4, a crucial receptor involved in mediating joint destruction in experimental arthritis (45, 46). S100A8 was shown to be an endogenous ligand for TLR-4, inducing intracellular translocation of myeloid differentiation factor 88 and activating interleukin-1 receptor–associated kinase and NF-κB (12), the latter being crucial for osteoclast differentiation and function (47, 48). Using TLR-4–deficient osteoclasts, we clearly demonstrated complete blockade of the S100A8-mediated increase in osteoclast formation and bone resorption.
In conclusion, the results of this study demonstrated that S100A8, but not S100A9 or the S100A8/A9 heterodimer, stimulates osteoclast differentiation, thus increasing the number of osteoclasts. S100A8 also appeared to stimulate osteoclast activity by enhancing the formation of actin rings, likely resulting in an increased bone resorptive capacity. Taken together, these effects ultimately lead to enhanced resorption induced by S100A8, which was shown to be mediated by TLR-4. The reduced bone erosion levels in arthritic S100A9-deficient mice, together with our in vitro findings, indicated that both TLR-4 and S100A8 are important factors in mediating osteoclastic bone destruction in experimental arthritis.
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. Ms Grevers had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Grevers, van den Berg, van Lent.
Acquisition of data. Grevers, Vogl, Sloetjes, van den Berg, van Lent.
Analysis and interpretation of data. Grevers, de Vries, Vogl, Abdollahi-Roodsaz, Sloetjes, Leenen, Roth, Everts, van den Berg, van Lent.
- 26Coordinate expression of activating Fcγ receptors I and III and inhibiting Fcγ receptor type II in the determination of joint inflammation and cartilage destruction during immune complex–mediated arthritis. Arthritis Rheum 2003; 48: 255–65., , , , , , et al.
- 28S100A8 causes a shift toward expression of activatory Fcγ receptors on macrophages via Toll-like receptor 4 and regulates Fcγ receptor expression in synovium during chronic experimental arthritis. Arthritis Rheum 2010; 62: 3353–64., , , , , , et al.
- 46Shift from Toll-like receptor 2 (TLR-2) toward TLR-4 dependency in the erosive stage of chronic streptococcal cell wall arthritis coincident with TLR-4–mediated interleukin-17 production. Arthritis Rheum 2008; 58: 3753–64., , , , , , et al.
- 48NF-κB p50 and p52 expression is not required for RANK-expressing osteoclast progenitor formation but is essential for RANK- and cytokine-mediated osteoclastogenesis. J Bone Miner Res 2002; 17: 1200–10., , , , , , et al.