Osteoclasts, the sole effective bone-resorbing cells, are required for physiological bone remodeling and mediate pathological bone erosion associated with tumors and inflammatory conditions.1–3 Two key factors that drive differentiation of myeloid lineage cells into osteoclasts are macrophage colony stimulating factor (M-CSF) and receptor activator of NF-κB ligand (RANKL). M-CSF promotes proliferation and survival of myeloid cells and induces expression of RANK, the receptor for RANKL, thereby generating osteoclast precursors (OCPs) that can subsequently differentiate into osteoclasts in response to RANKL stimulation. RANKL elicits complex signaling cascades leading to induction and activation of the nuclear factor of activated T cells, cytoplasmic 1 (NFATc1) “master transcriptional regulator” of osteoclastogenesis that mediates expression of osteoclast-related genes.4 Costimulation for RANKL signaling and NFATc1 activation is provided by costimulatory receptors that utilize the immunoreceptor tyrosine-based activating motif (ITAM)-associated adaptors DNAX-activation protein 12 (DAP12) and Fc receptor gamma chain (FcRγ).5, 6 A key subsequent step required for efficient fusion of preosteoclasts into multinucleated polykaryons, cytoskeletal reorganization, and effective bone resorbing function is signaling by αVβ3 integrins and other receptors such as osteoclast-associated receptor (OSCAR) that interact with the bone surface.7, 8 Thus, appropriate integrin-matrix interactions play a key role in the late stages of osteoclast differentiation and function.
More recently it has become clear that osteoclastogenesis is restrained by transcriptional repressors that are basally expressed and functional in early myeloid lineage cells and osteoclast precursors.9 These include interferon regulatory factor 8 (IRF8) and B-cell lymphoma 6 (BCL6) that inhibit expression of Nfatc1 and downstream NFATc1 target genes, and Eos, V-maf musculoaponeurotic fibrosarcoma oncogene homolog B (MafB), and inhibitor of DNA binding (Id) that inhibit expression of microphthalmia-associated transcription factor (MITF) and NFATc1 target genes important for osteoclastogenesis. In order for osteoclast differentiation to proceed, RANKL signaling needs to overcome the barrier imposed by these repressors. This release from inhibition occurs by RANKL-mediated downregulation of expression of these repressors, thereby allowing inductive signaling and osteoclast differentiation to proceed.9 Mechanisms that mediate downregulation of transcriptional repressors during early phases of osteoclastogenesis are not known, although IRF8 and MafB expression in later phases of osteoclastogenesis is suppressed by B-lymphocyte-induced maturation protein 1 (Blimp1).10 Induction or activation of transcriptional repressors as part of an active inhibitory program that suppresses osteoclastogenesis has to our knowledge not been described.
Inflammation in conditions such as rheumatoid arthritis and periodontitis is associated with pathologic bone resorption. Inflammatory cytokines drive excessive bone loss by suppressing the anabolic functions of osteoblasts and by inducing RANKL expression on osteoblast lineage and stromal cells, thereby augmenting osteoclastogenesis. In addition, inflammatory cytokines such as tumor necrosis factor (TNF) and interleukin 1 (IL-1) can synergize with RANKL to augment signaling by RANK and directly promote osteoclastogenesis. Emerging evidence indicates that potent inflammatory factors, such as Toll-like receptor ligands and cytokines, also engage feedback inhibition mechanisms to restrain osteoclastogenesis.9, 11–13 Such feedback inhibition limits the extent of bone resorption during the acute and transient inflammation that occurs in physiological settings. In contrast, progression of bone erosion during inflammatory diseases is evidence of relatively ineffective feedback inhibition and dominance of the pro-resorptive pathways induced by inflammatory factors. Inflammation-mediated feedback inhibition typically targets early myeloid lineage cells or osteoclast precursors and redirects their differentiation toward macrophages that can participate in host defense. Potent inflammatory inhibitors of osteoclast precursors include Toll-like receptor ligands, granulocyte-macrophage colony-stimulating factor (GM-CSF), interferons, and IL-27 that work by suppressing RANK expression and downstream signaling.9
OCPs are bone marrow-derived myeloid lineage cells that enter the circulation as quiescent cells and subsequently migrate into sites of bone erosion, where they differentiate into osteoclasts.14–16 Studies with human subjects where blood cells can be readily accessed have shown that OCPs are contained in the CD14+ monocyte pool, and suggest that CD14+ cells that have differentiated to express high levels of CD16 and dendritic cell-specific transmembrane protein (DC-STAMP) exhibit higher osteoclastogenic potential.16, 17 In contrast to osteoclasts that express high levels of αVβ3 integrins, monocytes and OCPs express β2 integrins such as CD11b/CD18 (also termed αMβ2, CR3, or Mac1). CD11b expression is dynamically regulated during murine osteoclastogenesis. CD11b is initially expressed at lower levels in mouse bone marrow cells with high osteoclastogenic potential, is upregulated upon culturing with M-CSF, and is subsequently downregulated after RANKL stimulation.15, 18, 19 However, a functional role of CD11b in regulating osteoclastogenesis has not been previously investigated. In addition to interacting with extracellular matrix (ECM) components, CD11b/CD18 is ligated by a variety of inflammatory ligands, such as fibrin(ogen) and complement split products,20, 21 and thus can also potentially mediate inflammatory regulation of osteoclastogenesis. We investigated the role of CD11b in osteoclastogenesis by combining a loss-of-function approach using CD11b-deficient mice with a gain-of-function approach using high avidity ligation of CD11b/CD18 with its ligand fibrinogen. CD11b-deficient mice exhibited decreased bone mass that was associated with increased osteoclast numbers. CD11b/CD18 signaling potently inhibited osteoclast differentiation by transiently suppressing RANK expression and subsequently activating BCL6 to bind to the NFATC1 gene locus and repress NFATC1 transcription, thereby making NFATC1 refractory to osteoclastogenic signaling. These results identify CD11b as a negative regulator of the earliest stages of osteoclast differentiation and provide the first example of induction of a transcriptional repressor to suppress osteoclastogenesis in a regulated manner.
Subjects and Methods
Mice and analysis of bone phenotype
Eight- to 12-week-old C57BL/6 and CD11b-deficient mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). Age- and sex-matched BALB/c and FcRγ-deficient mice on the BALB/c genetic background were purchased from Taconic (Centre Hudson, NY, USA). Animal experiments were approved by the Hospital for Special Surgery Institutional Animal Care and Use Committee (IACUC). To evaluate bone volume and architecture by micro–computed tomography (micro-CT), mouse femurs were fixed in 70% ethanol and scanned using a Scanco micro-CT-35 instrument (Scanco Medical, Brüttisellen, Switzerland). 1.35 mm of the distal part of femurs, starting 100 µm from the growth plate, was used for trabecular bone analysis. 6 µm voxel size, 55 KVp, 0.36 degree rotation step (180 degree angular range), and a 400 ms exposure per view were used for the scans. The Scanco µCT software (HP, DECwindows Motif 1.6) was used for 3D reconstruction and viewing of images. After 3D reconstruction, volumes were segmented using a global threshold of .4g/c. These analyses measured cortical and trabecular bone volume fraction and thickness, as well as trabecular number and separation, as described in the literature.22 Bone histomorphometry was performed on mouse femurs in the Center for Bone Histomorphometry at the University of Connecticut Health Center. Five-micrometer sections were stained with tartrate-resistant acid phosphatase (TRAP) and hematoxylin for osteoclast visualization, and histomorphometric analysis was performed with the Osteomeasure system (Osteometrics Inc., Atlanta, GA, USA) using standard procedures.23 Briefly, all measurements were confined to the secondary spongiosa and restricted to an area between 400 and 2000 µm distal to the growth plate–metaphyseal junction of the distal femur. Osteoclasts were identified as TRAP+ cells that were multinucleated and adjacent to bone. As markers for osteoclast formation and bone resorption, serum TRAP5b and C-terminal telopeptide fragments of type I collagen (CTX-1) were measured with the RatLaps enzyme immunoassay (EIA) kit (Immunodiagnostic Systems) following the manufacturer's instructions.
To measure bone mineralization in vivo, mice were intraperitoneally injected with Calcein (green) at 10 µg/g (body weight) twice, 7 days apart. Two days after the second injection, femurs were collected, fixed in 80% ethanol, embedded in poly(methyl methacrylate) as described,24 and then cut into 8- to 10-µm sections. Histomorphometry analysis using OsteoII software (Bioquant, Nashville, TN, USA) was performed on trabecular bone within the tibial metaphysis. The mineral apposition rate (MAR) was determined by measuring the distance between two fluorochrome-labeled mineralization fronts. The mineralizing surface was determined by measuring the double-labeled surface and one-half of the single-labeled surface, and by then expressing this value as a percentage of total bone surface. The bone formation rate (BFR) was then expressed as MAR × mineralizing surface/total bone surface, using a surface referent.
Human and murine M-CSF, soluble RANKL (sRANKL), and human osteopontin were purchased from Peprotech (Rocky Hill, NJ, USA). Fibrinogen (F9754) and poly Arg-Gly-Asp (polyRGD) were purchased from Sigma-Aldrich (St. Louis, MO, USA), and gamma-fibrinogen (Fb) (377–395) and control scrambled peptides were purchased from ANASPEC (Fremont, CA, USA). Vitronectin and fibronectin were purchased from Millipore (Billerica, MA, USA). The 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) assay kit was purchased from Roche and cell viability was measured following the manufacturer's instructions.
Peripheral blood mononuclear cells were obtained from blood leukocyte preparations purchased from the New York Blood Center by density gradient centrifugation with Ficoll (Invitrogen, Carlsbad, CA, USA) using a protocol approved by the Hospital for Special Surgery Institutional Review Board. Monocytes were obtained from peripheral blood, using anti-CD14 magnetic beads, as recommended by the manufacturer (Miltenyi Biotec Auburn, CA, USA). Monocyte-derived OCPs that express RANK were obtained by culture for 1 day with 20 ng/mL of M-CSF (Peprotech) in α modified essential medium (α-MEM) (Invitrogen) supplemented with 10% fetal bovine serum (FBS; Hyclone), and purity of monocytes was >97%, as verified by flow cytometric analysis. For ligation with Fb, plates were coated with Fb (100 µg) or FBS (10%) for 1 hour at room temperature, and washed with phosphate-buffered saline (PBS) before cells were added. OCPs were harvested and added to Fb-coated or FBS-coated tissue culture plates for 1 hour. Cell adherence was observed for both control and Fb stimulated cells. Polymyxin B (14 µg/mL; Sigma Aldrich), which was verified to not directly stimulate macrophages but essentially completely blocked exogenous lipopolysaccharide (LPS) up to concentrations of 10 ng/mL in our system, was used to ensure that contaminating endotoxin did not contribute to the effects observed, as described.25
Human CD14+ cells were incubated with 20 ng/mL of M-CSF for 1 day to generate OCPs. For human osteoclastogenesis assays, cells were added to 96-well plates in triplicate at a seeding density of 5 × 104 cells per well. OCPs were incubated with 20 ng/mL of M-CSF and 40 ng/mL of human soluble RANKL for an additional 5 days in α-MEM supplemented with 10% FBS. Cytokines were replenished every 3 days. On day 6, cells were fixed and stained for TRAP using the Acid Phosphatase Leukocyte diagnostic kit (Sigma-Aldrich) as recommended by the manufacturer. Multinucleated (greater than three nuclei) TRAP-positive osteoclasts were counted in triplicate wells. For mouse osteoclastogenesis, bone marrow (BM) cells were flushed from femurs of mice and cultured with murine M-CSF (20 ng/mL; Peprotech) on Petri dishes in α-MEM supplemented with 10% FBS after lysis of RBCs using ACK lysis buffer (Cambrex, Walkersville, MD, USA). Then, the nonadherent cell population was recovered on the next day and cultured with murine M-CSF (20 ng/mL) for an additional 3 days. We defined this cell population as mouse OCPs. For murine osteoclastogenesis assays, we plated 2 × 104 OCPs per well in triplicate wells on a 96-well plate and added M-CSF and RANKL (100 ng/mL) for an additional 6 days, with exchange of fresh media every 3 days. For detection of actin ring formation, cells were fixed, permeabilized with 0.1% Triton X-100, and incubated with fluorescein isothiocyanate-phalloidin in a humidified chamber for 45 minutes at 37°C. After rinsing in PBS, cells were imaged using a Zeiss Axioplan microscope (Zeiss, Germany) with an attached Leica DC 200 digital camera (Leica, Switzerland).
Gene expression analysis
For real-time PCR, DNA-free RNA was obtained using the RNeasy Mini Kit from Qiagen with DNase treatment, and 1 µg of total RNA was reverse transcribed using a First Strand cDNA Synthesis kit (Fermentas, Hanover, MD, USA). Real-time PCR was performed in triplicate using the iCycler iQ thermal cycler and detection system (Applied Biosystems, Carlsbad, CA, USA) following the manufacturer's protocols. Expression of the tested gene was normalized relative to levels of GAPDH. For primary transcript analysis, relative amounts of primary transcripts were measured by real-time PCR using primer pairs that amplify exon-intron junctions or intronic sequences.
Chromatin immunoprecipitation (ChIP) was performed as described.26 Briefly, after Fb ligation, 1 × 107 human primary OCPs were fixed by adding formaldehyde directly to the medium to a final concentration of 1%. Cells were harvested, washed, and lysed. Chromatin was sheared by sonication to a length of approximately 500 base pairs using a Bioruptor sonicator (Diagenode). Sheared chromatin was precleared and then immunoprecipitated with anti-BCL6 or immunoglobulin G (IgG) control (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Immune complexes were subsequently collected and washed and DNA crosslinking was reversed by heating at 65°C overnight. Following proteinase K digestion, DNA was recovered by PCR purification kit (Qiagen) and real-time PCR was performed to detect the occupancy of BCL6. The primers used to amplify the human NFATc1 promoter (−2217/−2138) are the following: sense, 5′-CAGGAGAAGGGATTTG-3′; anti-sense, 5′-GGAGACGTTACACGGGTTT-3′.
Two one-hundreths of a nanomole (0.2 nmol) of four independent short interfering RNAs (siRNAs), specifically targeting human Bcl6, or two nontargeting control siRNAs (Dharmacon and Invitrogen) were transfected into primary human CD14+ monocytes with the Amaxa Nucleofector device set to program Y-001 using the Human Monocyte Nucleofector kit (Amaxa), as described.27
Staining for cell-surface expression of various integrins was performed using antibodies against human β2 integrin (BioLegend), human CD11b, β1, β3, and β5 integrins, colony-stimulating factor-1 receptor (c-Fms) (R&D Systems), human CD14 and CD16, and mouse CD11b (BD Parmingen, San Diego, CA, USA). A FACSCalibur flow cytometer with CELLQuest software (Becton Dickinson) was used.
Foreign body granuloma model
Mice were anesthetized using ketamine/xylazine (1 mg/0.1 mg per 20 g body weight). Under aseptic conditions, PMMA particles (catalog number 19130, Polysciences, Warrington, PA, USA) particles were deposited intramuscularly in the dorsal chest and the incision was sutured. Then, 3 or 7 days later mice were euthanized and the granulomatous tissues harvested. CD11b-positive cells were isolated from digested granulomatous tissue using anti-CD11b magnetic beads, as recommended by the manufacturer (Miltenyi Biotec).
All statistical analyses were performed with Graphpad Prism 5.0 software using the two-tailed, unpaired Mann-Whitney test (two conditions), paired t test, or one-way ANOVA with Tukey's post t test for multiple comparisons (more than two conditions); p < 0.05 was taken as statistically significant.
CD11b-deficient mice exhibit decreased bone mass and increased osteoclast numbers
We analyzed the bone phenotype of CD11b-deficient mice using micro-CT and histomorphometry. Micro-CT analysis of 12-week-old male mice revealed that the trabecular bone volume fraction (BV/TV) was significantly lower in CD11b-deficient mice compared to the wild-type mice (Fig. 1A). Accordingly, trabecular space (Th.Sp) was significantly increased in CD11b deficient mice, as compared to the wild type, whereas trabecular number (Th.N) was decreased in CD11b-deficient mice (Fig. 1A). To address whether decreased bone mass in CD11b-deficient mice was related to increased osteoclastogenesis, we analyzed bone sections that were stained with the osteoclast marker TRAP. Bone sections from 8-week-old mice showed significantly increased osteoclast numbers and osteoclast surface in CD11b-deficient mice compared to control mice (Fig. 1B). This increase in osteoclast number correlated with a trend toward increased serum TRAP levels (Supplemental Fig. S1A). Increased osteoclast-mediated bone resorptive activity was supported by increased serum CTX levels (Fig. S1A), and, consistent with the micro-CT results, histomorphometry revealed decreased BV/TV in 12-week-old CD11b-deficient mice (data not shown). CD11b-deficient mice also exhibited significantly decreased MAR and trended toward decreased bone formation rate (BFR) (Fig. 1C). CD11b and its dimerization partner CD18 are generally considered to be specifically expressed in hematopoietic cells,21 which was reflected by low expression in osteoblast cultures (Supplemental Fig. S2A), and osteoblast activity and expression of osteoblast differentiation markers in CD11b-deficient mice were comparable to those of control mice in vitro (Supplemental Fig. S2B,C). Thus, the decreased osteoblast activity in vivo in CD11b-deficient mice is unlikely to have resulted from an intrinsic osteoblast defect but instead may reflect uncoupling of osteoclast and osteoblast function, although this notion needs to be further investigated. Collectively, these results show that CD11b-deficient mice show increased osteoclastogenesis and decreased bone formation at an early age (8 weeks) that results in a clear low bone mass phenotype at 12 weeks of age. The results suggest that CD11b is a negative regulator of osteoclastogenesis.
We wished to corroborate a negative role for CD11b in osteoclastogenesis by activating CD11b signaling in a complementary gain-of-function approach. We activated CD11b signaling in mouse bone marrow-derived osteoclast precursors using the well-defined CD11b ligand Fb and the standard approach of adding cells to plates coated with Fb versus control plates coated with serum.28 Immobilized Fb undergoes a conformational change that reveals the binding epitope present in fibrin, which is present in inflamed tissues and inflammatory exudates.28 Treatment with Fb nearly completely and significantly (p < 0.0001) suppressed RANKL-induced differentiation of OCPs into multinucleated (more than three nuclei per cell) TRAP-positive osteoclasts (Fig. 2A). We also activated CD11b signaling using the specific CD11b ligand inactivated complement component 3b (iC3b).29 Ligation of CD11b by iC3b inhibited osteoclastogenesis, and this inhibition was partially reversed in CD11b-deficient OCPs (Supplemental Fig. S3A,B), further supporting a role of CD11b signaling in the inhibition of osteoclastogenesis. Fb strongly suppressed RANKL-induced expression of osteoclast-related genes such as cathepsin K and β3 integrin (Fig. 2B). These results show that Fb, a ligand for CD11b, strongly inhibits RANKL-induced osteoclastogenesis by acting directly on OCPs. To confirm that Fb was signaling through CD11b we performed experiments using CD11b-deficient OCPs. Deficiency of CD11b resulted in a minimal increase in RANKL-induced osteoclastogenesis (Fig. 2C), likely because CD11b signaling is not induced under these in vitro culture conditions in the absence of exogenous ligands. Fb-induced suppression of osteoclastogenesis was partially reversed in CD11b-deficient cells (Fig. 2C), suggesting that Fb engages additional receptors that generate inhibitory signals that cooperate with CD11b signaling to inhibit osteoclastogenesis. To test the role of Fb in suppressing osteoclastogenesis in vivo, we used a foreign body granuloma model, in which subcutaneously implanted PMMA particles become coated with fibrin(ogen) and recruit myeloid cells from the circulation to elicit a foreign body reaction.30–32 Strikingly, CD11b+ CD115+ myeloid precursors isolated from foreign body granulomas were not able to differentiate into osteoclasts in response to M-CSF and RANKL (Fig. 2D), suggesting inhibition of osteoclastogenic pathways. Taken together, these results suggest a role for Fb and CD11b signaling in suppressing osteoclastogenesis.
CD11b signaling inhibits human osteoclast differentiation
We wished to test the effects of CD11b signaling on differentiation of human blood-derived OCPs. Human cells offer the advantages of working with quiescent blood OCPs that correspond directly to cells that migrate into sites of bone erosion and differentiate into osteoclasts. Thus, experiments with human OCPs avoid potential issues related to proliferation, in vitro maturation into macrophage-like cells, and contamination with stromal cells, which arise when mouse bone marrow is cultured for several days in vitro with M-CSF in the standard protocols used to generate bone marrow–derived OCPs. In addition, experiments with human cells allow us to directly test the effects of CD11b signaling on differentiation of postmitotic OCPs, separately from any effects on proliferation and survival of progenitors that can occur in experiments with bone marrow precursors. We tested the effects of CD11b ligation on osteoclast differentiation using a validated culture system in which primary human blood-derived monocytes differentiate into osteoclasts.33 Treatment with Fb nearly completely and significantly (p < 0.0001) suppressed RANKL-induced differentiation of human OCPs into multinucleated (more than three nuclei per cell) TRAP-positive osteoclasts (Fig. 3A). Fb-mediated inhibitory effects on osteoclastogenesis were not overcome by high concentrations of RANKL (Supplemental Fig. S4A). Fb inhibited differentiation of human OCPs more strongly than murine OCPs (Fig. 2A versus Fig. 3A), and inhibition of human OCPs was observed at lower concentrations of Fb (data not shown). Fb also inhibited actin ring formation, a marker of the later stages of osteoclastogenesis (Fig. 3B). Fb strongly suppressed RANKL-induced expression of osteoclast-related genes such as cathepsin K and β3 integrin (Fig. 3C). To exclude the possibility that treatment with Fb induces apoptosis of OCPs, we examined the effect of Fb on the viability of OCPs using the MTT assay. We found that Fb treatment did not decrease the viability or attachment of OCPs (Supplemental Fig. S5A,B). These results suggested a role for integrin ligation by Fb in suppressing osteoclastogenesis, and a negative role of integrins was further supported by effective inhibition of RANKL-induced osteoclast differentiation by polyRGD, which specifically ligates integrin receptors (Fig. 3D). To confirm that Fb was inhibiting osteoclastogenesis via CD11b signaling, we used an inhibitory peptide, gamma-Fb (377–395), that specifically blocks the Fb-CD11b interaction.34 Inhibition of Fb-CD11b interactions using gamma-Fb (377–395) partially reversed the inhibitory effects of Fb on osteoclastogenesis that were observed in the presence of control scrambled peptide (Fig. 3E). Inhibition of osteoclastogenesis was specific to β2 integrins, because the β1 and β3 integrin ligands osteopontin, vitronectin, and fibronectin had minimal effects on osteoclastogenesis, and blocking antibodies to β5 integrins had no effect on Fb-induced inhibition (Fig. 3F; data not shown). Thus, Fb-CD11b signaling inhibits human osteoclastogenesis, suggesting a direct suppressive effect on osteoclast differentiation.
RANKL inhibits β2 integrin expression and induces a switch to β3 integrin expression
The aforedescribed results raise the question of how osteoclast differentiation proceeds in the face of negative signaling by CD11b/CD18 β2 integrins that are ligated even under physiological conditions by ECM molecules and counter-ligands expressed on other cell types.21, 35, 36 One paradigm that has emerged recently is that RANKL overcomes the “brakes” imposed by OCP-expressed inhibitors of osteoclast differentiation by downregulating the expression of these inhibitory molecules.9 Thus, we tested the effects of RANKL stimulation on integrin expression. We first examined whether cell surface levels of CD11b were regulated during osteoclastogenesis using flow cytometric analysis. Consistent with previous reports,18, 19 RANKL stimulation decreased cell surface CD11b expression on a subset of murine OCPs (Fig. 4A). This resulted in a decrease in the number of cells that expressed the highest levels of CD11b and a significant decrease in mean fluorescence intensity of CD11b staining, thus confirming that CD11b is downregulated as OCPs begin to respond to RANKL and differentiate down the osteoclast pathway. We next investigated the effects of RANKL on integrin expression in human OCPs. Consistent with the literature,21, 37 microarray analysis showed that human OCPs expressed β1, β2, and β5 integrins, with no detectable β3 expression (mean hybridization intensities: β1, 3295.41; β2, 13159.73; β5, 941.23; β3 < 25 and expression considered not significantly above background with seven different probe sets). This expression pattern was corroborated by flow cytometric analysis that showed high expression of β2 integrins and CD11b, with no detectable β3 integrin staining (Fig. 4B). Interestingly, expression of β2 integrins and CD11b diminished after RANKL stimulation, whereas expression of β3 integrins increased as expected (Fig. 4C). RANKL stimulation also decreased expression of β5 integrins, whereas β1 expression did not change (Fig. 4C). Two-color flow cytometry showed that cells which expressed β3 integrins after RANKL stimulation expressed significantly lower levels of β2 integrins and CD11b on their cell surface than β3-negative cells (Fig. 4D, region A versus region B), showing downregulation of β2 integrins in cells that are differentiating into osteoclasts in response to RANKL. These results indicate that RANKL induces a switch in integrin expression from β2 to β3 integrins, and suggest that RANKL-induced downregulation of β2 integrins attenuates a negative regulator of osteoclast differentiation.
Fb-CD11b signaling inhibits NFATC1 transcription
We next investigated mechanisms by which CD11b ligation inhibits human osteoclast differentiation. A key event in osteoclast differentiation is the induction of the transcription factor NFATc1, a master regulator of osteoclast genes.4 As expected, RANKL-induced NFATc1 protein expression was readily apparent 1 day after RANKL stimulation. RANKL-induced NFATc1 protein expression was essentially completely abrogated by Fb (Fig. 5A). Inhibition of NFATc1 expression by Fb persisted for up to 5 days during osteoclast differentiation cultures (Fig. 5A). Fb-mediated suppression of RANKL-induced NFATc1 expression was partially attenuated in CD11b-deficient cells (Supplemental Fig. S6), suggesting that CD11b signaling contributes to suppression of NFATc1 expression. Thus, Fb disrupts early steps in OCP differentiation leading to NFATc1 induction. These results indicate that Fb inhibits osteoclastogenesis by suppressing expression of NFATc1 (Fig. 5A), thereby preventing expression of downstream NFATc1 target genes important for osteoclast differentiation, as is shown in Figs. 2B and 3C. We next tested whether Fb inhibited NFATC1 gene expression. Fb inhibited RANKL-mediated induction of NFATc1 mRNA (Fig. 5B) and of primary unspliced NFATc1 transcripts, a direct measure of transcription rate (Fig. 5C). Thus, Fb inhibits induction of NFATC1 gene expression by RANKL.
BCL6 mediates Fb-induced inhibition of osteoclastogenesis
CD11b ligation could inhibit NFATC1 induction by suppressing RANK signaling or by inducing transcriptional repressive mechanisms that render the NFATC1 gene refractory to activation by upstream signaling pathways. We investigated both of these possibilities, because inhibitory mechanisms often cooperate to effectively suppress signaling and gene induction responses over an extended time frame.38 We first tested the effects of Fb on RANK expression and downstream signaling. Fb-stimulated OCPs exhibited substantially lower RANK mRNA 1 day after RANKL stimulation than did control OCPs; however, the differences in RANK mRNA levels in Fb-treated versus control cells resolved after 2 days of RANKL stimulation (Fig. 6A). Interestingly, downregulation of RANK expression was substantially more pronounced in human versus mouse OCPs (data not shown), which is consistent with a previous report11 and may help explain why Fb-mediated suppression of osteoclastogenesis was more effective with human OCPs (Fig. 2A, 3A, and data not shown). We investigated which Fb-induced signals mediate downregulation of RANK mRNA expression in human OCPs. CD11b/CD18 integrins that are ligated by Fb signal via ITAM-containing adaptors DAP12 and FcRγ to activate the spleen tyrosine kinase (Syk) and downstream calcium, NF-κB, and mitogen-activated protein kinase (MAPK) pathways.21 Interestingly, inhibition of Syk and calcium signaling prevented the Fb-induced downregulation of RANK mRNA, whereas inhibition of MAPKs had no effect (Fig. 6B; data not shown). These results suggest that, in addition to promoting osteoclastogenesis, especially at later stages of differentiation,5, 6, 39–41 high-intensity ITAM signaling can have a suppressive effect on early osteoclast precursors. Further support for the notion that high-avidity ligation of ITAM-associated receptors can inhibit early stages of osteoclastogenesis is provided by evidence that crosslinking of ITAM-associated Fc-gamma receptors (FcγRs) also inhibited RANK expression and osteoclastogenesis (data not shown) and by a recent report demonstrating that immune complexes that signal via FcγRs can inhibit osteoclastogenesis by a mechanism dependent on ITAM-containing FcRγ.42 In addition, Fb-induced inhibition of NFATc1 expression and osteoclastogenesis was partially reversed when the ITAM-containing FcRγ adaptor was deleted in mouse OCPs, thereby providing direct genetic evidence that ITAM signaling can inhibit osteoclast differentiation (Fig. 6C, D). Of note, this inhibitory role of FcRγ was reproducibly apparent on the BALB/c but not on the C57BL/6 genetic background. The reasons for this effect of genetic background may be related to strain-dependent strength of CD11b coupling with FcRγ and will need to be investigated in future work, but genetic background effects may explain some of the discrepancies in published reports concerning the role of ITAM signaling molecules in osteoclastogenesis.43, 44 Overall, the results show that high-avidity ligation of ITAM-associated receptors suppresses RANK expression in early-stage human OCPs, and are consistent with the hypothesis that the outcomes of ITAM signaling vary according to the timing and avidity of receptor ligation,38, 39, 45 as set forth in the discussion section.
The transient differences in human RANK expression (Fig. 6A), which did not occur in mouse OCPs, cannot fully explain the complete and sustained Fb-induced block in NFATc1 expression that was observed in human and mouse OCPs (Figs. 5A, 6C). Thus, we wished to test whether Fb could additionally block RANK signaling, which could occur over a more extended timeframe. Consistent with the literature,21, 38 Fb strongly activated similar NF-κB and MAPK signaling pathways, as did RANKL, and we were not able to detect any RANKL-induced signals above those induced by Fb (data not shown). This raised the paradox of how Fb-induced signals that are similar to those induced by RANKL can actually inhibit RANK-mediated osteoclastogenesis. Although differences in functional outcomes can be partially explained by different magnitude and kinetics of signaling,38, 39 we reasoned that Fb may also induce transcriptional repressors that block responses of the NFATC1 gene to upstream signals. One such repressor that is regulated by inflammatory signaling in immune cells is BCL6,46 and we tested the role of BCL6 in Fb-mediated inhibition of osteoclastogenesis.
Interestingly, BCL6 expression was induced by Fb ligation in human OCPs (Fig. 7A), whereas other transcriptional suppressors of osteoclastogenesis such as IRF8 and MafB were not induced (data not shown). BCL6 induction by Fb was attenuated in CD11b-deficient mouse OCPs (Supplemental Fig. S7), suggesting that CD11b signaling contributes to Fb-induced BCL6 expression. BCL6 induction by Fb was dependent on Syk, calcium, and NF-κB signaling (Fig. 7B) and was attenuated in FcγR-deficient mice (Supplemental Fig. S8), suggesting that BCL6 expression is induced by ITAM signaling. BCL6 represses Nfatc1 transcription directly by binding to the Nfatc1 locus.47 Thus, we used ChIP assays to test whether Fb regulated BCL6 binding at the NFATC1 gene locus in human OCPs. Consistent with a previous report using a murine system,47 basal BCL6 occupancy was detected at the NFATC1 locus in human OCPs and diminished after RANKL stimulation (Fig. 7C). In sharp contrast, in Fb-treated cells there was a massive recruitment of BCL6 to NFATC1 after RANKL stimulation (Fig. 7C), suggesting that BCL6 mediates inhibition of osteoclastogenesis by Fb. We tested this notion by using RNA interference to knock down BCL6 expression in primary human OCPs. Diminished expression of BCL6 resulted in an essentially complete reversal of the inhibitory effects of Fb on osteoclastogenesis (Fig. 7D); these results were corroborated using additional control siRNAs and a total of four different BCL6-specific siRNAs (data not shown). Collectively, these results suggest that Fb induces recruitment of BCL6 to NFATC1 to suppress NFATc1 expression and thereby inhibit osteoclastogenesis.
The importance of negative regulation of osteoclastogenesis in preserving bone homeostasis and preventing excessive bone resorption in inflammatory settings has been recently established.9 Relative to positive regulation of osteoclastogenesis that has been thoroughly investigated,40 little is known about signaling pathways and molecular mechanisms that restrain osteoclastogenesis. In this study, we found that CD11b and β2 integrins suppress osteoclast differentiation by acting early in the differentiation pathway to prevent induction of NFATc1 and thus of downstream osteoclast genes. CD11b limited bone resorption under physiological conditions in vivo, and activation of CD11b/β2 integrin signaling by ligands such as Fb strongly suppressed osteoclast differentiation. CD11b suppressed induction of NFATc1 by the complementary mechanisms of downregulation of RANK expression and induction of recruitment of the transcriptional repressor BCL6 to the NFATC1 gene. These findings identify CD11b as an inhibitory receptor for osteoclastogenesis and demonstrate that integrin and ITAM signaling can inhibit osteoclast differentiation, which extends the prevailing view of the function of these signaling pathways. Inducible recruitment of BCL6 to NFATC1 provides the first example of inducible recruitment of a transcriptional repressor as an inhibitory mechanism that suppresses osteoclastogenesis in response to environmental cues. The results also suggest that CD11b ligands that are highly expressed at inflammatory sites, such as fibrin(ogen), can activate feedback inhibition to limit the extent of pathological bone resorption.
The current paradigm is that the predominant role of integrins and cell-matrix interactions in osteoclastogenesis is to promote osteoclast differentiation and resorptive function.48 However, key molecules that mediate osteoclast–bone matrix interactions and promote differentiation, such as β3 integrins and OSCAR,7, 8 are not expressed in osteoclast precursors and are instead only induced and subsequently ligated several days after RANKL stimulation. In contrast, CD11b and β2 integrins that suppress osteoclastogenesis are expressed on osteoclast precursors and can be ligated prior to RANKL stimulation. As β2 and β3 integrins induce similar signaling pathways,21 the different function of these integrins may be related to the timing of integrin ligation relative to RANKL signaling. Thus, early ligation of integrins would suppress RANKL responses and osteoclastogenesis, whereas late ligation after the RANKL response has unfolded would promote terminal osteoclast differentiation and function. This view is supported by our findings that integrins only effectively suppressed osteoclastogenesis if ligated prior to RANKL stimulation (Park-Min et al., unpublished data), and by a report that β5 integrins, which are preferentially expressed in osteoclast precursors, can suppress osteoclastogenesis.49 However, it remains possible that signals transduced by β2 integrin cytoplasmic domains that are distinct from β3 integrin cytoplasmic domains can contribute to the differential functions of these receptor families.
In contrast to β3 integrins and OSCAR that selectively recognize collagen and other bone matrix components and drive differentiation of osteoclasts on bone surfaces, CD11b/CD18 (αMβ2) integrins have multiple cell-surface, soluble, and matrix ligands that are expressed under physiological and inflammatory conditions.21 For example, intercellular adhesion molecule 1 (ICAM1) expressed by cell types that interact with OCPs, such as endothelial and stromal cells, is a CD11b ligand and can potentially suppress osteoclastogenesis under physiological conditions. Because a role for CD11b signaling and the CD11b ligand Fb in restraining osteoclastogenesis under physiological conditions is supported by our findings of decreased bone mass and increased osteoclast numbers in CD11b-deficient mice under homeostatic conditions, and by skeletal manifestations such as bone cysts in human patients with congenital afibrinogenemia.50 β2 integrin-mediated inhibitory mechanisms may occur in a context- and a time-dependent manner. For example, OCPs on bone surfaces can develop into osteoclasts despite the presence of CD11b, suggesting that ligation of other noninhibitory integrins by the bone surface or distinct signals derived from bone may overcome β2 integrin-mediated inhibitory signals. Furthermore, ligation of β2 integrins in soft tissues may help explain why osteoclasts do not develop distal from bone even when RANKL is expressed, such as in breast tissue. In addition, our findings suggest that CD11b may help restrain osteoclastogenesis in inflammatory settings, where high levels of CD11b/CD18 ligands, such as fibrin(ogen) and complement split products, are expressed and would be encountered by OCPs prior to engagement of the bone surface.
Although our experiments showing suppressed osteoclastogenic potential in the foreign body granuloma system support Fb-mediated inhibition, we have not been able to further directly test this notion, because disruption of Fb-CD11b function and signaling in vivo diminishes inflammation21, 51, 52 and this compromises the ability to measure effects on inflammatory bone resorption. Regulation of both inflammation and osteoclastogenesis by CD11b suggests that CD11b serves as a “dual switch”—promoting influx of inflammatory cells and cytokine production, while at the same time limiting the amount of inflammation-associated bone resorption. Our results link CD11b to induction of BCL6 and suppression of NFATc1 expression, although the inhibitory effects of fibrinogen on NFATc1 expression were only partially reversed in CD11b-deficient cells. This is in line with results showing that Fb-mediated inhibition of osteoclastogenesis was not fully reversed in CD11b-deficient mice. Although CD11b plays a role in the inhibition of osteoclast differentiation by Fb, our results suggest that ligation of additional receptors by Fb, as has been described in other systems,28 likely generates additional inhibitory signals that cooperate with CD11b signaling to suppress osteoclastogenesis. These additional Fb-activated inhibitory receptors and signals may work to increase suppression of NFATc1 expression and to suppress pathways that cooperate with NFATc1 to induce full osteoclast differentiation.
DAP12 and FcRγ are the main ITAM-containing signaling adaptors expressed in osteoclast precursors. The prevailing view is that cross-activation of DAP12- and FcRγ-mediated ITAM signaling by RANKL is important for costimulation and effective induction of NFATc1 and osteoclast differentiation,5, 6, 38 although ITAM-mediated costimulation is not required for osteoclast differentiation under all culture conditions.41 In addition, ITAM signaling by NFATc1-induced proteins such as β3 integrins and OSCAR augments late stages of osteoclast differentiation and function. However, the in vivo role of DAP12 and FcRγ, and ITAM signaling in bone phenotype is complex, context-dependant, and not fully understood. Compelling genetic evidence has established a key role for DAP12 and FcRγ in osteoclastogenesis and bone remodeling under physiological conditions in mice, and for the FcRγ- and DAP12-associated receptors paired Ig-like receptor (PIR)-A and myeloid DAP12-associating lectin (MDL)-1 in inflammatory bone resorption in a mouse arthritis model.5, 6, 53, 54 However, DAP12 and FcRγ are not required for bone resorption at specific anatomic sites under stress conditions such as low estrogen or low calcium states.55 In addition, there are apparently paradoxical findings that DAP12-deficient patients exhibit decreased in vitro osteoclastogenesis but osteoporosis in vivo, and that the DAP12-associated receptor triggering receptor expressed on myeloid cells (TREM)-2 may have different effects on osteoclastogenesis in human versus mouse OCPs in vitro and on bone phenotype in vivo.56–59 These differences in the function of ITAM-associated receptors have proved difficult to understand, despite intensive investigation over the last decade. Our work now reveals that β2 integrins that utilize DAP12 and FcRγ for signaling can inhibit osteoclastogenesis, and suggests that ITAM signaling pathways can inhibit as well as promote osteoclast differentiation. A context-dependent negative role for ITAM signaling in osteoclastogenesis is further supported by evidence that immune complexes, which signal via FcRγ, can inhibit osteoclastogenesis.42 The timing of ligation of ITAM-associated receptors relative to RANKL stimulation is likely an important determinant of functional outcome of ITAM signaling, because early ligation of β2 integrins and FcγRs on OCPs suppresses RANKL responses, whereas late ligation of β3 integrins and OSCAR after the RANKL-induced program has been established instead promotes terminal osteoclast differentiation and function. Another potential explanation for different functional outcomes of signaling by ITAM-associated receptors in osteoclastogenesis is different avidity of receptor ligation. Differences in avidity of ITAM-associated receptor ligation are well established to result in different and even opposing functional outcomes.38 Consistent with the avidity hypothesis, low-avidity FcγR ligation tended to promote osteoclastogenesis whereas high-avidity FcγR ligation inhibited osteoclastogenesis (Lim et al., unpublished data). Our findings provide insights that suggest approaches to modulate osteoclastogenesis and bone resorption by varying the timing and intensity of ITAM receptor signaling.
Recent reports have established that differentiation of osteoclast precursors is restrained by a barrier imposed by basally expressed transcriptional repressors, and that RANK signaling overcomes this barrier by downregulating expression of these repressors. Our study advances this concept in two ways. First, our findings suggest that RANKL also downregulates expression of receptors that inhibit osteoclastogenesis, such as CD11b and β2 integrins, in order to escape from suppression and effectively promote osteoclast differentiation. Second, we have found that at least one transcriptional repressor which suppresses gene induction by osteoclastogenic signals, BCL6, does not solely function constitutively. Instead, BCL6 expression was modestly increased, and its recruitment to NFATC1 was massively induced by Fb-CD11b signaling. Thus, BCL6 expression and function are regulated by inhibitory signals that oppose RANK function, and modulation of BCL6 provides a new mechanism by which osteoclast differentiation is suppressed in response to environmental cues. Previous work has shown that Toll-like receptors (TLR) ligands, which also inhibit osteoclastogenesis, induce a genomewide redistribution of BCL6 to reprogram macrophage gene expression and promote host defense.46 Thus, Fb-induced recruitment of BCL6 to NFATC1 is likely part of an inflammatory program that diverts myeloid cell differentiation from an osteoclast fate toward an inflammatory phenotype. This would serve the dual purpose of attenuating inflammation-associated bone damage and promoting host defense. Many of the transcriptional repressors of osteoclastogenesis, such as BCL6, IRF8, Id, CCAAT/enhancer binding protein β (C/EBPβ), and MafB, are regulated by inflammatory stimuli. Thus, our work suggests that inflammatory factors regulate these repressors to induce feedback inhibition of osteoclastogenesis and opens new avenues of research into transcriptional regulation of osteoclast differentiation.
All authors state that they have no conflicts of interest.
We thank Lyudmila Lukashova in the Musculoskeletal Repair and Regeneration Core Center for micro-CT analysis, Nicholas Brownell at Hospital for Special Surgery for technical assistance, Stephen Doty and Orla O'Shea in the Analytical Microscopy Core at Hospital for Special Surgery for analysis of bone formation, and Se-Hwan Mun at the University of Connecticut Health Center for help with the histomorphometric analysis. We thank Xiaoyu Hu, Anna Yarilina, and Baohong Zhao for critical review of the manuscript. This work was supported by grants from the National Institutes of Health (KHPM and LBI).
Authors' roles: Study design: KHPM, EYL, and LBI. Study conduct: KHPM, EYL, NM, EL, SKL, CH, and PEP. Data analysis and data interpretation: KHPM, EYL, NM, EL, SKL, JAL, CH, AMM, PEP, SRG, and LBI. Drafting of the manuscript: KHPM and LBI. LBI takes responsibility for the integrity of the data analysis.