|Home | About | Journals | Submit | Contact Us | Français|
B.Z. performed most of the experiments with significant assistance from M.T. X.W. conducted the histological analysis. A.Y., T.K., X.H., and T.T. assisted with the experiments. K.O. provided the Irf8−/− mice. B.Z. and M.T. designed the project and wrote the manuscript. Y.C. provided recombinant RANKL and contributed to manuscript preparation. L.B.I. oversaw the bone marrow chimera and human cell experiments and contributed to manuscript revision. H.T. oversaw the inflammatory bone destruction experiments and provided critical advice for the experiments. M.T. and R.K. supervised the project.
Bone metabolism results from a balance between osteoclast-driven bone resorption and osteoblast-mediated bone formation. Diseases such as periodontitis and rheumatoid arthritis are characterized by increased bone destruction due to enhanced osteoclastogenesis1,2. Here we report that interferon regulatory factor 8 (IRF8), a transcription factor expressed in immune cells, is a key regulatory molecule for osteoclastogenesis. IRF8 expression in osteoclast precursors was downregulated during the initial phase of osteoclast differentiation induced by receptor activator of nuclear factor κB ligand (RANKL, also called TRANCE, ODF, and OPGL), which is encoded by the Tnfsf11 gene. Mice deficient in IRF8 exhibited severe osteoporosis due to increased numbers of osteoclasts, and enhanced bone destruction following lipopolysaccharide (LPS) administration. Irf8–/– osteoclast precursors underwent increased osteoclastogenesis in response to RANKL and tumor necrosis factor α (TNFα). IRF8 suppressed osteoclastogenesis by inhibiting the function and expression of nuclear factor of activated T cells c1 (NFATc1). Our results show that IRF8 inhibits osteoclast formation under physiological and pathological conditions, and suggest a model where downregulation of inhibitory factors like IRF8 contributes to RANKL-mediated osteoclastogenesis.
Osteoclasts are multinucleated giant cells derived from the monocyte/macrophage lineage. Their differentiation is triggered by RANKL in the presence of macrophage colony-stimulating factor (M-CSF, encoded by Csf-1 gene), which is produced by osteoblasts1,2. RANKL induces intracellular signals via its receptor RANK, and upregulates the expression of various genes, such as Nfatc1,fos, Oscar, Ctsk, and Calcr that encode proteins for NFATc1, c-Fos, OSCAR, cathepsin K and calcitonin receptor, respectively3–5. Numerous studies have focused on these upregulated genes and their roles in osteoclastogenesis. On the other hand, the expression levels of various genes are simultaneously downregulated during osteoclastogenesis6. The biological significance of the downregulated expression of these genes following RANK activation, however, has not been fully elucidated.
To identify genes that show reduced expression levels in response to RANK signaling, we performed a genome-wide screening of mRNAs from osteoclast precursors and osteoclasts using a DNA microarray technique (data not shown). Among the identified genes, expression of the transcription factor Irf8 [also called interferon consensus sequence binding protein (ICSBP)] was found to be downregulated during the initial phase of osteoclastogenesis triggered by RANKL (data not shown). IRF8 is known to be specifically expressed in immune cells, including monocytes/macrophages, B lymphocytes, and activated T lymphocytes7–9. It is a member of the IRF family and has been shown to regulate myeloid cell development by interacting with the Ets family transcription factors10,11. Hence, Irf8–/– mice show an increased number of myeloid progenitors, defects in macrophage function including impaired IL-12 production, developmental defects in CD8α+ and plasmacytoid dendritic cells, and systemic expansion of the granulocyte population, which frequently leads to a fatal blast crisis12–14.
Using RT-PCR and immunoblot analysis, IRF8 expression was detected in bone marrow-derived macrophages (BMMs) and spleen-derived macrophages, which are capable of differentiating into osteoclasts (Fig. 1a,b). As previously shown3, NFATc1 was modestly expressed at a basal level in unstimulated BMMs and NFATc1 expression was strongly induced after RANKL stimulation (Fig. 1b); robust induction of NFATc1 by RANKL is a necessary and pivotal step for osteoclast differentiation characterized by enhanced expression of osteoclast marker genes such as Oscar, Itgb3 that encodes integrin β3 protein, and Ctsk3. In contrast, IRF8 expression decreased after RANKL stimulation (Fig. 1a,b), resulting in substantially lower nuclear IRF8 protein 24 h after RANKL addition (Fig. 1b). We hypothesized that IRF8 downregulation may be required for osteoclast differentiation and that IRF8 may inhibit early stages of osteoclastogenesis.
To examine the role of IRF8 in osteoclastogenesis, we constructed a retroviral vector, pMX-Irf8-IRES-EGFP, which was engineered to express both IRF8 and enhanced green fluorescence protein (EGFP). Macrophage-like osteoclast precursors were transduced with retroviral particles encoding Irf8 or control viral particles that lacked the Irf8 sequence (pMX-IRES-EGFP). IRF8–overexpressing precursors failed to differentiate into osteoclasts in response to RANKL stimulation (Fig. 1c). Furthermore, these cells were able to phagocytize zymosan particles, whereas multinucleated osteoclasts failed to do so (Fig. 1d). These results suggest that IRF8 inhibits osteoclastogenesis from precursor cells, which instead retain the characteristics of phagocytic macrophages.
Next, we analyzed the Irf8–/– mice for abnormal bone phenotypes. Radiographic and microcomputed tomographic analyses showed that these mice had severe osteoporosis accompanied by dramatic decreases in trabecular bone volume, number, and thickness, as well as the number of bone nodules (Fig. 2a, b). Histomorphometric analysis also revealed reduced bone mass in the Irf8–/– mice (data not shown). Importantly, increases in the osteoclast number and surface area were observed in these mice (Fig. 2c). There were no significant differences between wild-type and Irf8–/– mice in the serum level of osteoprotegerin (OPG encoded by Tnfrsf11b gene), a decoy RANKL receptor that inhibits osteoclastogenesis, or in RANKL levels in primary cultured osteoblasts (Supplementary Fig. 1 online). Moreover, an increased rate of bone formation accompanying the accelerated bone resorption rate was observed in the Irf8–/– mice, suggesting that their osteoporosis was caused by enhanced bone turnover and remodeling (Fig. 2d). Together, these observations indicated that IRF8 plays a suppressive role in osteoclastogenesis during in vivo bone remodeling. To address whether the bone phenotype could be explained by Irf8 deficiency in osteoclast precursors, or whether a cell autonomous effect of Irf8 deficiency in osteoblasts or bone marrow stromal cells could play a role, we established chimeric mouse models in which either wild-type or Irf8–/– littermate bone marrow was transplanted into lethally irradiated wild-type recipients. Consistent with the global Irf8 deficient mice, Irf8–/– bone marrow chimeric mice showed severe high turnover osteoporosis (Fig. 2e–h). IRF8 is mainly expressed in hematopoietic cells7–9. Indeed, IRF8 expression was not detected in primary calvarial osteoblasts (data not shown). Furthermore, primary Irf8–/– osteoblast differentiation and matrix calcification were not affected compared to wild-type cells (Supplementary Fig. 2 online). Thus, the osteoporosis resulted from enhanced osteoclastogenesis, which is a cell autonomous consequence of Irf8 deficiency in osteoclast precursors but not osteoblasts.
On the basis of these results, we examined the differentiation potential of osteoclast precursors obtained from Irf8–/– mice. When wild-type or Irf8–/– osteoclast precursors were cocultured with wild-type or Irf8–/– primary osteoblasts in the presence of active vitamin D3 (an inducer of RANKL expression in osteoblasts), a greater number of osteoclasts formed from the Irf8–/– precursors than from the wild-type cells, independent of whether the osteoblasts were from Irf8–/– or wild-type mice (Fig. 3a). These data further support the notion that there is no difference in osteoclastogenic activity between wild-type and Irf8–/– osteoblasts, and that Irf8 deficiency in osteoclast precursors plays a decisive role in promoting osteoclast differentiation. Indeed, macrophage cultures prepared from Irf8–/– mice also exhibited augmented osteoclastogenesis in the presence of RANKL and M-CSF (Fig. 3b), which led to an increase in resorption on dentin slices in vitro (data not shown). On the other hand, retrovirus-mediated reconstitution of IRF8 expression in Irf8–/– macrophages inhibited RANKL-induced osteoclastogenesis (Fig. 3c). These results demonstrate that IRF8 in osteoclast precursors is involved in the inhibition of osteoclastogenesis in vitro and in vivo. Furthermore, we found that IRF8 expression was also decreased during RANKL-induced human osteoclastogenesis, although with slower kinetics (Fig. 3d). Silencing of Irf8 mRNA in human osteoclast precursors by siRNAs resulted in enhanced osteoclast differentiation (Fig. 3e–f), indicating the function of IRF8 in osteoclastogenesis is well conserved in humans and mice.
Consistent with the results regarding osteoclastogenesis, the mRNA expression profiles of various osteoclast markers were more strongly upregulated by RANKL stimulation in macrophages prepared from Irf8–/– mice compared with those obtained from wild-type mice (Supplementary Fig. 3 online). However, Rank and c-fms mRNA expression and RANK cell surface protein expression were comparable in wild-type and Irf8–/– osteoclast precursors (Supplementary Fig. 4a online and data not shown). During osteoclast precursor generation from bone marrow cells, M-CSF treatment did not lead to a greater increase in Rank mRNA expression in Irf8–/– cells than in control cells (Supplementary Fig. 4b online). These results suggest that enhanced osteoclastogenesis in Irf8–/– macrophages is not due to changes in these receptor levels.
Expression of Nfatc1 gene and the osteoclast marker Acp5 gene that encodes tartrate-resistant acid phosphatase (TRAP) protein in Irf8–/– precursors was induced by concentrations of RANKL that were 5%–25% of those required for the expression of these genes in wild-type precursors (Fig. 4a). Furthermore, overexpression of IRF8 in the precursor cells repressed the RANKL-induced expression of Nfatc1 and Acp5 mRNAs (Fig. 4b).
These observations led us to examine the effect of IRF8 on the transcriptional activity of NFATc1, which was previously reported to interact with IRF815. For these experiments, we employed luciferase reporter plasmids driven by three copies of the NFATc1 binding site from the human IL-2 distal promoter (p3x Nfatc1-Luc) or by the mouse Acp5 promoter (pAcp5-Luc). Overexpression of Nfatc1 gene activated these promoters, whereas simultaneous expression of Irf8 gene reduced the activities of the promoters to control levels (Fig. 4c), indicating that IRF8 inhibits the transcriptional activity of NFATc1. When GST-IRF8 proteins were incubated with nuclear lysates containing FLAG-hemagglutinin-tagged NFATc1 (FH-NFATc1), anti-FLAG antibodies resulted in the coimmunoprecipitation of GST-IRF8 and FH-NFATc1, suggesting that IRF8 physically interacts with NFATc1 (Fig. 4d). Furthermore, association of endogenous IRF8 and NFATc1 was identified by co-immunoprecipitation from nuclear extracts of human monocytic cells (Supplementary Fig. 5 online). We further examined the effects of IRF8 on the binding of NFATc1 to its target DNA elements in an electrophoretic mobility shift assay (EMSA); NFATc1-DNA complexes were detected, which was confirmed by the addition of competitive probes or anti-NFATc1 antibodies. Increased levels of GST-IRF8 or nuclear lysates containing excess IRF8, however, significantly decreased the binding of NFATc1 to the probes (Fig. 4e, f), demonstrating the inhibitory effect of IRF8 on NFATc1 binding to its target DNA elements.
We then attempted to examine the roles of IRF8 in the processes underlying pathological bone destruction. Because several members of the IRF family of transcription factors, including IRF8, have been demonstrated to play crucial roles in toll-like receptor (TLR) signaling in response to such microbial components as LPS and unmethylated CpG DNA16,17, we examined the role of IRF8 in the bone destruction observed during TLR-mediated inflammation. Administration of LPS to the calvarial periosteum resulted in enhanced osteoclast formation in wild-type mice, whereas more extensive bone destruction was observed in Irf8–/– mice (Fig. 4g). These results suggest that IRF8 is a critical negative regulator of osteoclastogenesis and a mediator of the maintenance of bone integrity during inflammatory bone destruction. We also examined the effects of TLR ligands on osteoclastogenesis in wild-type and Irf8–/– osteoclast precursors. As previously reported18, LPS at high doses completely inhibited RANKL-induced osteoclastogenesis in wild-type cell cultures. In contrast, corresponding high doses of LPS, and also of the TLR3 ligand poly(I:C) and the TLR9 ligand CpG DNA, only partially inhibited osteoclast differentiation in Irf8–/– cell cultures (Fig. 4h and Supplementary Fig. 6 on line). Irf8-deficient cells were almost completely refractory to the inhibitory effects of the TLR2 ligand peptidoglycan (Supplementary Fig. 6a). These results suggest a potential inhibitory role of IRF8 in the regulation of osteoclastogenesis by TLRs, although TLRs can also activate IRF8-independent inhibitory mechanisms. We further found that Ifnα/β expression was not diminished in Irf8-deficient cells, and that IFN-γ completely inhibited osteoclastogenesis in Irf8–/– macrophages, similar to wild-type cells (data not shown), thus suggesting that IRF8 can inhibit osteoclastogenesis independently of IFN-γ.
Finally, we examined the effect of TNFα on osteoclastogenesis using Irf8–/– precursors, because TNFα is a critical mediator of inflammation induced by TLRs19–21 and has been suggested to be able to induce osteoclastogenesis22–24. Consistent with previous reports, TNFα induced the development of a small number of osteoclasts in cultures of wild-type precursor cells (Fig. 4i). Notably, osteoclastogenesis was enhanced in cultures of Irf8–/– precursor cells treated with TNFα (Fig. 4i). The mRNA expression levels of Nfatc1 and Acp5 in the Irf8–/– osteoclast precursors were also augmented by TNFα (Fig. 4j), indicating that IRF8 also plays a suppressive role in TNFα–induced osteoclastogenesis.
Our data provide a mechanism by which IRF8 suppresses osteoclastogenesis (depicted in Supplementary Fig. 7 on line). In osteoclast precursors, abundant IRF8 interacts with basally expressed NFATc1 to suppress its transcriptional activity and thus prevent its activation of target genes, including autoamplification of its own promoter. Stimulation of osteoclast precursors with RANKL results in the activation of NF-κB and AP-1 that bind to the Nfatc1 promoter to induce its activity. At the same time, RANKL induces the downregulation of IRF8, thereby releasing NFATc1 from IRF8-mediated suppression and augmenting NFATc1-mediated auto-amplification of its own expression. Together, these mechanisms result in robust NFATc1 expression and induction of downstream genes required for osteoclast differentiation.
Bone erosion that occurs in the setting of infection and chronic inflammation, termed inflammatory osteolysis, contributes to the pathogenesis of infectious and inflammatory diseases such as rheumatoid arthritis25. Inflammatory bone erosion is driven by microbial products such as LPS and inflammatory cytokines including TNFα that activate osteoclastogenesis directly or indirectly via activation of stromal cells and osteoblasts25. Our findings show that IRF8 plays an important role in attenuating LPS-induced inflammatory bone resorption and LPS- and TNFα-induced osteoclastogenesis. This homeostatic role of IRF8 may be important during acute infections and also in chronic inflammatory conditions such as rheumatoid arthritis. Identification of additional factors and mechanisms that augment IRF8 expression or function may represent a fruitful approach to therapeutic suppression of inflammatory bone erosion.
Overall, our findings support a model of RANKL-induced NFATc1 expression and osteoclast differentiation that involves cooperation between RANKL-mediated induction of positive regulators of osteoclastogenesis and suppression of negative regulators of osteoclast differentiation such as IRF8.
The Irf8–/– mice (C57BL/6) used in this study have been described previously12. Tetracycline hydrochloride (20 mg/kg; Sigma) and, 72 h later, calcein (10 mg/kg; Wako) were injected subcutaneously into 8-week-old wild-type and Irf8–/– mice. The mice were then euthanized 32 h after the second injection. The LPS-induced model of bone loss has been described previously26, except that 12.5 mg/kg LPS (Sigma) was used in the present study. The mice were subjected to histomorphometric and microradiographic examinations as described previously27. All mice were born and maintained under specific pathogen-free conditions. All animal experiments were approved by and conducted according to the guidelines of the Showa University Animal Care and Use Committee (approval number: 17079).
Donor bone marrow cells from wild-type or Irf8–/– littermate mice (on C57/BL6 background) were harvested and one-fourth of total bone marrow cells from each donor were injected intravenously via tail vein into each of the irradiated wild-type recipients. Recipient mice (3-week-old C57/BL6 mice) were obtained from Jackson Laboratory and were lethally irradiated with a single dose of 875 rads 1 day prior to transplantation. Chimeric mice were sacrificed 8 weeks after bone marrow transplantation. The experiments using chimeric mice were approved by the Hospital for Special Surgery Institutional Animal Care and Use Committee.
The method used to analyze osteoclast differentiation in vitro has been described previously28. Briefly, mouse bone marrow or spleen cells were cultured with 50 ng/ml M-CSF for 3 days. The obtained BMMs or spleen-derived macrophages were further stimulated with 150 ng/ml RANKL in the presence of 50 ng/ml M-CSF for 3–4 days. We also cocultured bone marrow cells, BMMs, or spleen-derived macrophages with primary osteoblasts derived from mouse calvaria in the presence of 108 M 1,25 (OH)2D3 and 10−6 M prostaglandin E2 for 6 days. Media were changed every 2 days. Generation of osteoclasts was evaluated by TRAP staining and its activity assays28. Methods used for the phagocytosis assay and labeling of actin have been described previously28.
Human fresh peripheral blood mononuclear cells (PBMCs) were obtained from whole blood from disease-free volunteers. Human CD14+ monocytes were then purified from PBMCs with anti-CD14 magnetic beads (Miltenyi Biotec) and were cultured in α-MEM medium with 10% FBS (Hyclone) and 20 ng/ml of M-CSF (Peprotech) for 2 days to induce osteoclast precursors that were further stimulated with RANKL in the presence of 20 ng/ml of M-CSF for 3–6 days to generate human osteoclasts. The experiments using human cells were approved by the Hospital for Special Surgery Institutional Review Board. Short interfering RNAs (siRNAs) specifically targeting human Irf8 or control siRNAs (Dharmacon, 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). Two different targeting siRNAs and control siRNAs were used with comparable results in a total of 6 independent experiments with different blood donors.
Irf8 cDNA was prepared and amplified in RT-PCRs using RNA from BMMs. The coding region of Irf8 was PCR-amplified using the following primers: 5’-GCAGGATGTGTGACCGGAAC-3’ (sense) and 5’-ACTGAGGCTTAGACGGTGAT-3’ (antisense). The amplified PCR fragment was subcloned into a pCR-Blunt vector to produce pCR-Blunt-Irf8 using a Zero Blunt PCR Cloning kit (Invitrogen). The pMX-Irf8-IRES-EGFP retroviral vector was constructed by inserting a 1.3-kb EcoRI/EcoRI DNA fragment encoding Irf8 from pCR-Blunt-Irf8 into the same site in the pMX-IRES-EGFP vector. The 1.3-kb BamHI/XhoI DNA fragment encoding IRF8 from pMX-Irf8-IRES-EGFP was then inserted into the same site in the pMX-puro vector to construct the pMX-Irf8-puro retroviral vector. The pcDNA3-Irf8 expression vector was constructed by inserting a 1.3-kb BamHI/NotI DNA fragment encoding IRF8 from pMX-Irf8-IRES-EGFP into the same site in the pcDNA3 vector. pcDNA3-Nfatc1 was constructed by inserting an EcoRI/EcoRI DNA fragment encoding NFATc1 from pMX-Nfatc1-IRES-EGFP 3 into the same site in the pcDNA3 vector. The pFH-Nfatc1, p3×Nfatc-Luc, and pAcp5-Luc vectors were previously described3,29.
Statistical analysis was performed using Student t-tests (p < 0.05 was taken as statistically significant) and all data are presented as the mean ± s.e.m. Results are representative of more than four individual experiments.
Details of the methods including retroviral gene transduction, GeneChip, RT-PCR, Northern blot assay, Real-time PCR, Luciferase reporter assay, preparation of GST fusion proteins, immunoprecipitation and immunoblot analyses, and EMSA are described in Supplementary Methods on line.
We are grateful to Dr. M. Asagiri (Tokyo Medical and Dental University) for great technical assistance. We thank Dr. T. Kitamura (University of Tokyo) for providing the retrovirus expression system. We also thank Dr. A. Mochizuki and all members of the Department of Biochemistry, School of Dentistry, Showa University for valuable discussion. This work is supported in part by High-Tech Research Center Project for Private Universities from the Ministry of Education, Culture, Sports, Science and Technology, Japan, 2005–2009, and by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (20390474 to M.T.), and by grants from the NIH (AR053843 and DE19381 to Y.C., and DE019420 and AR46713 to L.B.I.).
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.