Fos plays an essential role in the upregulation of RANK expression in osteoclast precursors within the bone microenvironment

Osteoclasts, multinucleated giant cells responsible for bone resorption, exist along the bone surface (Chambers, 2000; Martin et al., 1998; Roodman, 1999). Osteoclasts form from monocyte–macrophage lineage precursors under the tight regulation of bone-forming osteoblasts. Osteoblasts express two cytokines, colony-stimulating factor 1 (CSF-1, also called macrophage colony-stimulating factor) and receptor activator of NF-κB ligand (RANKL; also known as tumor necrosis factor ligand superfamily member 11, Tnsf11), both of which are essential to osteoclastic differentiation (Arron and Choi, 2000; Boyle et al., 2003; Hofbauer et al., 2000; Kong et al., 1999; Lacey et al., 1998; Suda et al., 1999; Yasuda et al., 1998; Yoshida et al., 1990). Osteoclastic differentiation is severely depressed in the bone of CSF-1-mutated osteopetrotic op/op mice and RANKL-deficient (RANKL^−/−; Tnsf11^−/−) mice (Boyle et al., 2003; Kong et al., 1999; Wiktor-Jedrzejczak et al., 1990; Yoshida et al., 1990). CSF-1 is constitutively expressed, whereas RANKL is expressed inducibly by osteoblasts in response to bone-resorbing stimuli, such as parathyroid hormone and 1α, 25-dihydroxyvitamin D₃ (Suda et al., 1999; Yoshida et al., 1990). In the presence of tumor necrosis factor α (TNF-α) and transforming growth factor β (TGF-β), hematopoietic precursors from RANKL^−/− or RANK^−/− mice can differentiate into osteoclasts in vitro, suggesting the existence of alternative routes for osteoclastic differentiation (Kim et al., 2005; Kobayashi et al., 2000).

After the discovery of the RANKL signaling in osteoclastogenesis, RANKL–RANK signaling was shown to be involved in other crucial biological processes such as the development of lymph nodes, the development of medullary thymic epithelial cells and intestinal antigen-sampling M (microfold) cells, lactation, breast cancer metastasis to bone and the central fever response in inflammation (Akiyama et al., 2008; Cao et al., 2001; Fata et al., 2000; Hanada et al., 2009; Hikosaka et al., 2008; Jones et al., 2006; Knoop et al., 2009). Medullary thymic epithelial cells necessary for maintaining self-tolerance and intestinal M cells necessary for intestinal immunity have been shown to highly express RANK (Akiyama et al., 2008; Hikosaka et al., 2008; Knoop et al., 2009). Thus, RANKL–RANK interaction is now recognized to participate in biological responses beyond those in bone.

Fos, a component of activator protein 1 (AP-1), is a transcription factor essential for osteoclastogenesis (Grigoriadis et al., 1994; Wang et al., 1992). Fos-deficient (Fos^−/−) mice exhibit severe osteopetrosis because of a lack of osteoclasts. In striking contrast, F4/80-positive macrophages have been detected in bone of Fos^−/− mice (Grigoriadis et al., 1994). These results suggest that hematopoietic progenitor cells can differentiate into macrophages but not into osteoclasts in the absence of Fos, because if the F4/80-positive macrophages in bone were osteoclast precursors then Fos would be required for the differentiation of macrophages into osteoclasts. It was also reported that RANKL-induced expression of nuclear factor of activated T cells c1 (NFATc1), a master regulator of osteoclastic differentiation, is tightly regulated by Fos (Ishida et al., 2002; Matsuo et al., 2004; Takayanagi et al., 2002). Therefore, Fos in osteoclast precursors is believed to play a role in osteoclastogenesis as part of the RANKL-induced pathway.

We have examined the characteristics and behavior of osteoclast precursors in vivo. Immunohistochemical analysis revealed that the precursors exist along the bone surface in RANKL^−/− mice, as RANK-positive cells (Mizoguchi et al., 2009). We named these osteoclast precursors ‘cell-cycle-arrested quiescent osteoclast precursors’ (QOPs), because QOPs rapidly differentiate into osteoclasts in response to RANKL without cell-cycle progression. In contrast to RANKL^−/− mice, Fos^−/− mice had no RANK-positive cells in bone tissues (Mizoguchi et al., 2009). These results suggest that Fos is also necessary for the occurrence of RANK-positive cells along the bone surface.

In the present study, we examined the role of Fos in RANK expression in osteoclast precursors. We show that the bone environment is involved in the upregulation of RANK expression in osteoclast precursors, and the expression of Fos in the precursors is necessary for the upregulation. Our results suggest that osteoblast-derived factors such as CSF-1 are involved in the Fos-dependent upregulation of RANK expression in osteoclast precursors, which makes it possible that osteoclast precursors use only RANKL as an osteoclast differentiation factor under physiological conditions. The physiological importance of the upregulation of RANK expression in osteoclast precursors is discussed in detail.

Many tartrate-resistant acid phosphatase (TRAP)-positive osteoclasts were observed along the surfaces of proximal tibiae in wild-type mice but not in RANKL^−/− mice or Fos^−/− mice (Fig. 1A). RANK-positive osteoclasts or osteoclast precursors were also detected with similar distributions to TRAP-positive cells in wild-type mice (Fig. 1B). As we reported previously, RANK-positive cells existed along the bone surface as osteoclast precursors in RANKL^−/− mice, but not in Fos^−/− mice (Fig. 1B) (Mizoguchi et al., 2009). By contrast, CSF-1R-positive cells were similarly distributed in all the genotypes, and the difference in expression of CSF-1R mRNA in bone tissues of RANKL^−/− and Fos^−/− mice was not significant (Fig. 1B,C). The number of RANK-positive cells and CSF-1R-positive cells in wild-type mice are not shown in Fig. 1B, right panel, because both multinucleated osteoclasts and osteoclast precursors express RANK and CSF-1R. Most RANK-positive cells express CSF-1R in both wild-type and RANKL^−/− mice (Mizoguchi et al., 2009). Real-time PCR and western blot analyses confirmed that the mRNA and protein levels of RANK were much lower in bone tissues obtained from Fos^−/− mice than those from RANKL^−/− mice (Fig. 1C,D).

Osteoclast precursors have been shown to exist in spleen in wild-type mice (Takahashi et al., 1988). We therefore examined the distribution of RANK-positive cells in comparison with that of CSF-1R-positive cells in spleen of wild-type, RANKL^−/− and Fos^−/− mice (Fig. 2A). CSF-1R-positive cells were observed in spleen in all the genotypes of mice. By contrast, RANK-positive cells were detected in spleen in wild-type and RANKL^−/− mice but not in Fos^−/− mice (Fig. 2A). However, the level of RANK expression in spleen was much lower than that in bone in wild-type and RANKL^−/− mice, because in order to obtain the similar fluorescence intensity of the positive cells, the camera exposure period for spleen sections needed to be much longer (more than ten times) than that for bone sections. Nevertheless, RANK-positive cells were not detected in spleen, as well as bone, in Fos^−/− mice.

Medullary thymic epithelial cells and intestinal antigen-sampling M cells have been reported to express RANK in order to accomplish RANKL-ordered functions (Akiyama et al., 2008; Hikosaka et al., 2008; Knoop et al., 2009). A similar distribution of RANK-positive cells was observed in medullary thymic epithelium and in intestinal epithelium situated over Peyer's patches in wild-type, RANKL^−/− and Fos^−/− mice (Fig. 2B). These results suggest that Fos-induced upregulation of RANK expression is cell-type specific.

To confirm that cell autonomous expression of Fos in osteoclast precursors is necessary for the upregulation of RANK expression on bone surfaces, we used a bone marrow transplantation model. When wild-type bone marrow cells were inoculated into Fos^−/− mice that had been myelosuppressed with busulfan, RANK-positive cells appeared along the bone surface (Fig. 3A). Immunostainable RANK-positive cells were not observed in soft tissues around the bone (data not shown). These results suggest that Fos is required for the upregulation of RANK expression, and bone-derived factors are involved in the process.

We then examined whether RANK expression in precursors is upregulated in cocultures with osteoblasts. When wild-type and Fos^−/− spleen cells were cocultured for 1 or 5 days with wild-type osteoblasts, the expression of RANK was increased in wild-type spleen cells but not in Fos^−/− spleen cells (Fig. 3B,C). The expression of CSF-1R in Fos^−/− spleen cells was increased on day 5. RANK expression in wild-type spleen cells was also increased, even by coculturing with RANKL^−/− osteoblasts (Fig. 3B,C). Most RANK-positive cells were positive for CSF-1R (Fig. 3B). CSF-1 is reported to enhance RANK expression in osteoclast precursors (Arai et al., 1999). AFS98, an anti-CSF-1R monoclonal antibody, has been shown to inhibit the interaction between CSF-1 and CSF-1R (Sudo et al., 1995). Therefore, we added AFS98 to cocultures of wild-type osteoblasts and wild-type spleen cells. AFS98 inhibited the appearance of both CSF-1R-positive cells and RANK-positive cells (Fig. 3B,C). These results suggest that RANK is expressed by CSF-1-expressing cells. We also observed that RANK-positive cells were scarce in bone in CSF-1-deficient op/op mice (Y.N., unpublished data). Furthermore, many CSF-1R-positive cells in bone tissues were observed to be in direct contact with CSF-1-expressing osteoblastic cells in Fos^−/− mice (Fig. 3D). These results suggest that CSF-1 but not RANKL expressed by osteoblasts plays a role in the upregulation of RANK expression in osteoclast precursors.

We next examined the effect of CSF-1 stimulation on RANK expression in wild-type and Fos^−/− macrophages. It was reported that the binding of CSF-1 to CSF-1R induced the downregulation of CSF-1R mRNA expression within 6 hours in human monocytes (Sariban et al., 1989). In accordance with the previous finding, CSF-1 treatment suppressed CSF-1R expression in both wild-type and Fos^−/− macrophages (Fig. 4A, upper panel), suggesting that CSF-1R-mediated signaling is normally transduced in Fos^−/− macrophages. CSF-1 also stimulated Fos mRNA expression in wild-type macrophages (Fig. 4A, middle panel). In culture, the expression of RANK mRNA in Fos^−/− macrophages is comparable with that of wild-type macrophages. The expression of RANK mRNA was upregulated by CSF-1 in wild-type macrophages but not in Fos^−/− macrophages (Fig. 4A, lower panel). Western blot analysis confirmed that CSF-1 stimulated Fos protein expression in wild-type macrophages (Fig. 4B). Wild-type and Fos^−/− macrophages similarly expressed RANK protein at basal levels, and the CSF-1-induced upregulation of RANK protein expression was observed only in wild-type macrophages. Spleen, as well as bone, was shown to strongly express CSF-1 (Wiktor-Jedrzejczak et al., 1990). These results suggest that immunohistochemically stainable levels of RANK protein are expressed in osteoclast precursors in bone and spleen in wild-type mice but not Fos^−/− mice.

Then we examined the effect of transfection of Fos on RANK expression in Fos^−/− macrophages. Overexpression of Fos in Fos^−/− macrophages increased the level of RANK (Fig. 4C). These results suggest that upregulation of Fos expression in macrophages is crucial for the expression of RANK. Furthermore, these results raised the possibility that if RANK is highly expressed in Fos^−/− osteoclast precursors, these precursors may differentiate into osteoclasts even in the absence of Fos.

Next, spleen-derived wild-type and Fos^−/− macrophages were prepared and infected with a retrovirus carrying cDNA for RANK (pMX-RANK; Fig. 5). Phosphorylation of ERK in Fos^−/− macrophages was increased by the transfection with RANK cDNA even in the absence of RANKL, and was further enhanced in response to RANKL (Fig. 5A). Wild-type macrophages transfected with RANK cDNA spontaneously differentiated into TRAP-positive cells even in the absence of RANKL (Fig. 5B). The treatment with RANKL enhanced TRAP-positive cell formation in both RANK-transfected and empty vector-transfected macrophages, although the number of TRAP-positive multinucleated cells induced by RANKL was substantially higher in RANK-transfected cultures (Fig. 5B). By contrast, overexpression of RANK in Fos^−/− macrophages failed to induce their differentiation into TRAP-positive cells in the presence or absence of RANKL. Expression of NFATc1 was induced in response to RANKL in wild-type macrophages but not Fos^−/− macrophages transfected with and without RANK cDNA (Fig. 5C). These results suggest that Fos plays essential roles, not only in the upregulation of RANK expression in osteoclast precursors, but also in the differentiation of RANK-positive precursors into osteoclasts (Fig. 5D).

Fos has been thought to act on osteoclastic differentiation only down-stream of the RANKL–RANK signaling pathway (Grigoriadis et al., 1994; Matsuo et al., 2004). In the present study, we showed that the upregulation of RANK expression during osteoclastogenesis is cell-type specific and requires Fos. Although RANK-positive cells were observed in both spleen and bone, the expression level was much lower in spleen than in bone. The upregulation of RANK expression in osteoclast precursors occurred in the bone tissues, suggesting the bone environment to be essential: that is, the Fos-dependent upregulation of RANK expression made it possible for osteoclast precursors to use RANKL for their osteoclastic differentiation. RANK overexpression in Fos^−/− macrophages failed to induce osteoclastic differentiation. Importantly, this finding provides suggests that Fos is involved in two steps during osteoclastogenesis: induction of RANK expression in osteoclast precursors and transduction of osteoclast-inducing signals from RANK and CSF-1R.

Immunohistochemical studies have shown that RANK is normally expressed by medullary thymic epithelial cells and intestinal antigen-sampling cells in Fos^−/− mice. The basal expression of RANK in osteoclast precursors was not influenced by the presence or absence of Fos. One abnormality observed in Fos^−/− mice is the absence of any increase in RANK in osteoclast precursors. CSF-1R-positive cells were observed in spleen of Fos^−/− mice but they failed to express RANK. These results suggest that RANK expression in osteoclast precursors is upregulated in a Fos-dependent manner. The upregulation appears to be related to physiological osteoclastogenesis, and requires an increase in Fos expression. Nevertheless, overexpression of RANK in Fos^−/− macrophages did not lead to the development of osteoclasts. These results suggest that in osteoclastogenesis, Fos is required first for bone- and spleen-induced RANK production in osteoclast precursors, and second for the RANKL-induced upregulation of NFATc1 expression in osteoclast precursors (Matsuo et al., 2004).

Cells that strongly express RANK were detected in bone tissues in RANKL^−/− mice and wild-type mice but not in Fos^−/− mice. These results suggest that bone-derived factors are involved in the upregulation of RANK expression in osteoclast precursors. Arai et al. previously showed, for the first time, that CSF-1 acted on osteoclast precursors and induced RANK expression (Arai et al., 1999). We also confirmed their finding and further showed that the CSF-1-induced upregulation of RANK expression is dependent on Fos. Cells positive for CSF-1R were always detected near or in contact with CSF-1-expressing osteoblasts or bone marrow stromal cells in bone. These results suggest that CSF-1 is one of the bone-derived factors that induce the upregulation of RANK expression in osteoclast precursors.

IL-34 is a newly discovered cytokine that binds to CSF-1R and exerts its action in the same way as CSF-1 (Lin et al., 2008). On exploring the mechanism of osteoclastogenesis in op/op mice, we found that IL-34 is involved in the development of osteoclast precursors in spleen (Y.N., unpublished data). IL-34 is highly expressed in spleen but not in bone. Cells double positive for CSF-1R and RANK were detected in spleen in op/op mice as well as wild-type mice. However, the expression level of RANK in osteoclast precursors in spleen was much lower than that in bone in op/op mice and wild-type mice. These results suggest that additional factors other than CSF-1 expressed by osteoblasts are necessary for the increase in RANK in osteoclast precursors in bone tissues. Recently we have observed that noncanonical Wnt signaling in osteoclast precursors induces RANK expression (Y.K., unpublished data). Wnt5a produced by osteoblasts enhanced RANK expression in osteoclast precursors through the receptor Ror2. These results suggest that the upregulation of RANK expression in osteoclast precursors is induced by several factors expressed by osteoblasts.

Kobayashi et al. were the first to report that TNF-α stimulated osteoclastic differentiation independent of RANKL–RANK signaling (Kobayashi et al., 2000). By contrast, Li et al. showed that although the injection of TNF-α to the calvariae of RANKL^−/− mice could generate osteoclasts, the number of osteoclasts produced was very low (Li et al., 2000). These results raise the question of why RANKL cannot be replaced with TNF-α during osteoclastogenesis in vivo. Treatment of macrophages with CSF-1 upregulated the expression of RANK (Fig. 4A) but not TNF receptor type 1 and TNF receptor type 2 (data not shown). The overexpression of RANK in macrophages induced osteoclastic differentiation even in the absence of the addition of RANKL. The upregulation of RANK expression in osteoclast precursors must lower the threshold for the RANKL-induced osteoclastogenesis. These results suggest that, under physiological conditions in vivo, the upregulation of RANK expression in osteoclast precursors is an important requirement for RANKL-induced but not TNF-α-induced osteoclastogenesis.

We previously reported that osteoclasts were generated in bone in response to an injection of RANKL in RANKL^−/− mice, as they were in wild-type mice (Yamamoto et al., 2006). Osteoclasts were not observed in the soft tissues around the bone in RANKL-injected RANKL^−/− mice, indicating that the expression of RANKL is not involved in determining the correct site for osteoclastogenesis. The present study showed that osteoblasts play a role in the upregulation of RANK expression in osteoclast precursors. These results suggest that the distribution of osteoclast precursors expressing high levels of RANK determines the site for osteoclastogenesis. Thus, we have uncovered the physiological importance of the increase in RANK expression in osteoclast precursors and the role of Fos in this process. Further experiments should reveal the molecular mechanism by which Fos regulates RANK expression in osteoclast precursors.

C57BL/6 mice were obtained from Japan SLC (Tokyo, Japan). RANKL^−/− mice (C57BL/6) were generated in the laboratory of J.P. (Kong et al., 1999). Fos^−/− mice (C57BL/6) were obtained from Jackson Laboratory (Bar Harbor, ME). All experiments were conducted in accordance with the guidelines for studies with laboratory animals of the Matsumoto Dental University Experimental Animal Committee.

The antibodies used for the immunohistochemical analysis were biotin-conjugated anti-RANK and anti-CSF-1R and anti-CSF-1R from R&D Systems (Minneapolis, MN) and anti-CSF-1 (EP1179Y) from Abcam (Cambridge, UK). CSF-1R neutralizing antibody (AFS98) was obtained from eBioscience (San Diego, CA). The antibodies used for the western blot analysis were anti-RANK, anti-Fos, anti-extracellular signal-regulated kinase (ERK), anti-phosphorylated ERK and anti-NFTAc1 antibodies from Cell Signaling Technology (Danvers, MA), and anti-β-actin antibody (AC-74) from Sigma-Aldrich (St. Louis, MO).

Cultured cells and whole bone tissue homogenized with Tissue Lyser II (Qiagen, Hilden, Germany) were lysed with TRIzol reagent (Invitrogen, Carlsbad, CA), and total RNA was extracted from the lysate using a PureLink RNA Mini Kit (Ambion, Austin, TX). cDNA was synthesized from the total RNA using reverse transcriptase (ReverTra Ace; Toyobo, Osaka, Japan), and subjected to a two-step real-time PCR in Applied Biosystems StepOnePlus^TM (Applied Biosystems, Foster City, CA). Results were normalized with respect to the amount of glyceraldehyde-3-phosphate dehydrogenase (G3PDH) in the same sample. The fold-change ratios between test and control samples were calculated. All primers were purchased from TAKARA BIO (Shiga, Japan). The primer IDs were as follows: RANK, MA075536; CSF-1R, MA072804; Fos, MA05419; G3PDH, MA05371.

Paraffin-embedded 4-µm thick sections were stained for TRAP, and counterstained with Hematoxylin. For immunofluorescent staining, tissues were frozen in hexane using a cooling apparatus (PSL-1800; Tokyo Rikakikai, Tokyo, Japan) and embedded in a 5% carboxymethyl cellulose (CMC) gel. Sections, 5-µm thick, were prepared using Kawamoto's film method (Cryofilm transfer kit; Finetec, Tokyo, Japan) (Kawamoto and Shimizu, 2000). The sections and cultured cells were fixed in ice-cold 5% acetic acid in ethanol, and subjected to staining for RANK, CSF-1R and CSF-1 using specific antibodies. Biotin-conjugated antibodies were used for RANK and CSF-1R staining (R&D Systems). Anti-CSF-1R antibody was labeled with FITC using a commercially available kit (Dojindo, Kumamoto, Japan) for double staining for RANK and CSF-1R. Horseradish peroxidase (HRP)-conjugated streptavidin (PerkinElmer, Boston, MA), HRP-conjugated anti-FITC (PerkinElmer), NorthernLights (NL) 557-conjugated anti-sheep IgG (R&D Systems) and Histofine Simple Stain Mouse MAX-PO (Rabbit; Nichirei Biosciences, Tokyo, Japan) were used as the secondary antibodies. Nuclei were detected by 4,4′-diamidino-2-phenylindole (DAPI) staining (Vector Laboratories, Burlingame, CA). RANK- or CSF-1R-positive cells in bone were counted in three images of 0.135 mm² (318×425 µm) of the central area just under the growth plate with AxioVision 4.8.1 Mosaix software (Carl Zeiss, Jena, Germany). RANK- or CSF-1R-positive cells in spleen were also counted in randomly selected areas (0.135 mm²) (318×425 µm) of the red pulp region. Three images of bone and spleen were prepared from three different mice of each genotype using a microscope (Axiovert 200; Carl Zeis) with a digital camera (AxioCamHRc; Carl Zeiss). Images were captured with AxioVision 3.1 (Carl Zeiss). The construction of figures using the images was performed with Photoshop software (Adobe, San Jose, CA).

Whole bone tissue homogenized with Tissue Lyser II (Qiagen) was lysed with TRIzol reagent (Invitrogen), and protein lysates were prepared according to the Qiagen protocol. In brief, the phenol supernatant containing the protein fraction was incubated with isopropanol and subsequently incubated with 0.3 M guanidine in ethanol. Protein was dissolved using 10 M urea and 50 mM dithiothreitol (DTT). Cultured cells were lysed in 0.1% NP-40 lysis buffer [20 mM Tris (pH 7.5), 50 mM β-glycerophosphate, 150 mM NaCl, 1 mM EDTA, 25 mM NaF, 1 mM Na₃VO₄, 1× protease inhibitor cocktail (Sigma-Aldrich), 1× phosphatase inhibitor cocktail I (Sigma-Aldrich) and phosphatase inhibitor cocktail II (Sigma-Aldrich)]. Lysates were electrophoresed on a SDS-PAGE gel, transferred onto a PVDF membrane (Clear blot P membrane, Atto, Tokyo, Japan), blotted with antibodies to specific proteins, and visualized using ECL (Amersham, Piscataway, NJ).

RANKL^−/− spleen cells were used as control (wild-type) cells against Fos^−/− spleen cells for the following reasons. (1) The total number of hematopoietic progenitor cells in spleen is lower in wild-type mice than Fos^−/− and RANKL^−/− mice, because osteopetrotic mice exhibit splenic extramedullary hematopoiesis because of the lack of a bone marrow environment (Kong et al., 1999; Okada et al., 1994). (2) RANKL is not expressed by macrophages (Kong et al., 1999; Suda et al., 1999; Yasuda et al., 1998). Spleen cells were prepared from 6-week-old male Fos^−/− and RANKL^−/− mice, and layered onto a lympholyte-M (Cedarlane Laboratories, Burlington, ON, Canada) gradient. After centrifugation, mononuclear cells were collected and cultured in α-MEM (Sigma-Aldrich) containing 10% FBS (JRH Bio-sciences, Lenexa, KS) in the presence of 10⁴ units/ml CSF-1 (Kyowa Hakko Kirin, Tokyo, Japan). After 16 hours, nonadherent spleen cells were harvested. Spleen cells (1×10⁶) were incubated with 10⁴ units/ml CSF-1 for 2 days in 6-well plates, and used as spleen macrophages. Old medium was replaced with fresh medium without FBS and CSF-1. After culturing for 16 hours, spleen macrophages were further cultured with 10⁴ units/ml CSF-1 for given periods.

To isolate primary osteoblasts from either wild-type or RANKL^−/− mice, calvariae from 2-day-old mice (male and female) were separately subjected to sample preparation. Briefly, each calvaria was cut into small pieces and cultured for 5 days in type I collagen gel (cell matrix type-IA; Nitta Gelatin, Osaka, Japan) prepared in α-MEM containing 10% FBS. Osteoblasts grown from the calvariae were collected by treating the collagen gel cultures with collagenase (Wako, Osaka, Japan) and 1×10⁵ were cocultured with 5×10⁴ nonadherent spleen cells in α-MEM containing 10% FBS in 96-well plates for specific periods. Cells were cultured for 1 and 5 days without osteoclastogenesis-stimulating factors. In some experiments, an anti-CSF-1R monoclonal antibody (AFS98; 10 µg/ml) was added to the culture medium.

Spleen macrophages (2×10⁵ cells) were infected with empty pMX retrovirus, Fos-expressing retrovirus or RANK-expressing retrovirus and cultured with 10⁴ units/ml CSF-1. After 1 day, infected cells were further cultured in the presence of 10⁴ units/ml CSF-1 with or without 5 nM RANKL (GST-RANKL) for given periods. GST-RANKL was generated in the laboratory of H.Y. (Oriental Yeast, Shiga, Japan).

Bone marrow cells were prepared from 6-week-old male wild-type mice, and layered onto a lympholyte-M (Cedarlane Laboratories) gradient. After centrifugation, mononuclear cells (5×10⁶) were collected, and injected into the left cardiac ventricle of 8-week-old male Fos^−/− mice. The recipient mice were treated with busulfan (25 mg/kg/day) for 2 days before the transplantation of wild-type bone marrow mononuclear cells (Ashizuka et al., 2006). After 18 days, mice were killed, and tibiae were removed and subjected to RANK staining.

Stat View 5.0 software (SAS Institute, Inc., Cary, NC) was used for all statistical analyses. Data were evaluated by a one-way analysis of variance (ANOVA) followed by Fisher's PLSD test. Experiments were performed three times and similar results were obtained. The results were expressed as means ± s.d. for three cultures or three photographs. P<0.01 was considered statistically significant. One representative result of each experiment was shown in the manuscript unless otherwise noted.