Stimulation of Osteoclast Formation by RANKL Requires Interferon Regulatory Factor-4 and Is Inhibited by Simvastatin in a Mouse Model of Bone Loss

Diseases of bone loss are a major public health problem. Here, we report the novel therapeutic action of simvastatin in osteoclastogenesis and osteoprotection, demonstrated by the ability of simvastatin to suppress osteoclast formation in vitro and in vivo. We found that in vitro, IRF4 expression is upregulated during osteoclast differentiation induced by RANKL (receptor activator of nuclear factor-κB ligand), while simvastatin blocks RANKL-induced osteoclastogenesis and decreases expression of NFATc1 (nuclear factor of activated T-cells, cytoplasmic, calcineurin-dependent 1), IRF4 and osteoclast markers. We also show that IRF4 acts in cooperation with NFATc2 and NF-κB on the promoter region of NFATc1 to accelerate its initial transcription during the early stage of osteoclastogenesis. Moreover, our study using IRF4 siRNA knockdown directly demonstrates the requirement for IRF4 in NFATc1 mRNA transcription and its necessity in RANKL-induced osteoclast differentiation. Our results suggest that the reduction in osteoclastogenesis is partly due to the inhibition of IRF4 production in RANKL-induced osteoclast differentiation. To investigate the in vivo effects of simvastatin in RANKL-treated mice, we examined the bone mineral density (BMD) of a mouse model of bone loss, and found that simvastatin significantly reduced bone loss by suppressing osteoclast numbers in vivo, even in the presence of high concentrations of RANKL. These results suggest that the depletion of osteoclasts is not due to the reduction in RANKL produced by osteoblasts in vivo. The results are consistent with the hypothesis that simvastatin blocks RANKL-induced IRF4 expression in osteoclastogenesis. We propose that the expression of IRF4 by osteoclasts could be a promising new therapeutic target in bone-loss diseases.


Introduction
RANKL/RANK signaling induces osteoclast formation and activation via several transcription factors, such as interferonregulatory factors (IRFs) [1,2], c-Fos, NF-kB and NFATc1 [3,4]. It has also been shown that NFATc1 cooperates with PU.1 on the Cathepsin K and OSCAR promoters [5,6], and forms an osteoclastspecific transcriptional complex containing AP-1 (Fos/Jun) and PU.1 for the efficient induction of osteoclast-specific genes, such as Atp6v0d2, Cathepsin K, DC-STAMP and TRAP [4,7,8]. PU.1 confers specificity to the NFATc1 response in RAW264.7 cells [9]. IRF4 and interferon consensus sequence-binding protein (ICSBP)/IRF8 are members of the IRF family, which are expressed in bone marrow-derived cells [10]. Both factors can be recruited to the IRF DNA-binding site in target genes through interaction with PU.1 [11][12][13][14]. Recently, an in vivo and in vitro study indicated that IRF8 suppresses osteoclastogenesis. 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 [15]. However, our understanding of the function of IRF4 in osteoclastogenesis remains elusive. Therefore, in this study, to dissect further these IRF4 functions in osteoclast differentiation, we focused on the transcriptional control of NFATc1 gene expression in RAW264.7 cells. Furthermore, we performed a pharmacological experiment to identify inhibitors of IRF4. Simvastatin is an orally administered competitive inhibitor of 3hydroxy-3-methyl-glutaryl-coenzyme A (HMG-CoA) reductase, an enzyme that catalyzes the conversion of HMG-CoA to mevalonic acid [16]. As effective cholesterol-lowering agents, statins have been extensively used for prevention of cardiovascular disease. Simvastatin inhibits the isoprenoids farnesyl pyrophosphate and geranylgeranyl pyrophosphate (GGPP). These isoprenoid pyrophosphates serve as essential adjuncts in the posttranslational modification of numerous key proteins that function as molecular switches, including the small GTPases RAS, RAC and RAS homologue (RHO) [17,18]. Osteoclast survival, differentiation and function require the GTPases including RAS [19][20][21], RAC [22,23] and RHO [24,25]. The membrane attachment and biological activity of these small GTPases require prenylation. The Rho family of GTPases is a large family of proteins, which includes RhoA, Rac1 and Rac2. Rho kinase (ROCK) has been shown to activate the DNA binding of IRF4 [26], while another report showed that simvastatin inhibits IRF4 gene expression via selective inhibition of ROCK in Th17 cells [27]. Therefore, in this study, we used simvastatin as an inhibitor of IRF4, and report the role of IRF4 in osteoclast differentiation in the presence of RANKL. Our study shows that IRF4 is a constituent of the signalling pathways that mediate the effect of prenylated GTPases on RANK/RANKL-dependent osteoclastogenesis in vitro and in vivo.

Reagents
Reagents were obtained from the following suppliers: Alpha-

Animal care
All experimental protocols were in accordance with the guidelines for the care and use of laboratory animals set by the Graduate School of the Institute of Health Biosciences, the University of Tokushima (Tokushima, Japan). The protocol was approved by the Committee on Animal Experiments of the University of Tokushima (permit number: 12052 and 12067). C57BL/6J female mice (4-8 weeks old; Japan SLC, Shizuoka, Japan) were maintained under controlled temperature (2362uC) and light conditions (lights on from 08:30-20:30) and fed standard rodent chow pellets with water ad libitum. All efforts were made to minimize the suffering of the animals.

Animal treatment
To evaluate the effect of chronic administration of the drug, simvastatin (10 mg/kg) or saline was injected intraperitoneally into 4-week-old female mice (n = 5/group) at 24-h intervals for 4 weeks before sacrifice.
A mouse model of bone loss was established as described [28]. Briefly, RANKL (1 mg/kg) or saline was injected intraperitoneally into 7-week-old female mice (n = 5/group). After 48 h the mice were killed and the femora were harvested for analysis. To evaluate the effect of simvastatin on this model of bone loss, simvastatin (10 mg/kg) was injected intraperitoneally 24 h before the first RANKL injection, followed by simvastatin injections at 24-h intervals for 2 days before sacrifice (n = 5).
Cell Culture RAW264.7 cells (mouse macrophage-derived cells, purchased from RIKEN Cell Bank) were cultured in plastic dishes containing a-MEM supplemented with 10% FBS in a CO 2 incubator (5% CO 2 in air) at 37uC and subcultured every 2 days.

Real time PCR and RT-PCR
Cells were cultured in 35 mm dishes in osteoclastogenic medium to ,80% confluence. RNA preparation, real time PCR analyses and RT-PCR analyses were as described previously [30,31], and were performed using primers listed in Table 1. Images were recorded using an ATTO CS analyser (ATTO, Tokyo, Japan).

Immunohistochemistry
Tissues were fixed in 4% paraformaldehyde, decalcified in 2.5% EDTA (pH 7.2) containing 0.4 M glucose at 4uC for 2 weeks, dehydrated and embedded in paraffin. Antigens were retrieved with 0.4 mg/mL proteinase K at room temperature for 5 min. After quenching of endogenous peroxidase activity with 1% H 2 O 2 in methanol, sections were incubated with an anti-TRAP polyclonal antibody (Santa Cruz Biotechnology) or anti-osteopontin antibody: (Immuno-Biological Laboratories Co., Ltd.) at 4uC overnight, washed with PBS, then incubated with peroxidaseconjugated secondary antibody according to the manufacturer's instructions (Histofine Simple Stain MAX-PO, Nichirei Bioscience). Colour was developed with 3,3-diaminobenzidine tetrahydrochloride (DAB), and haematoxylin was used as a nuclear counterstain.
Immunoprecipitation RAW264.7 cells were cultured in 100 mm dishes in osteoclastogenic medium to ,80% confluence. Immunoprecipitation was performed as described previously [33], using specific antibodies for IRF4 and IRF8.
Chromatin Immunoprecipitation (ChIP) Assay RAW264.7 cells were cultured in 100 mm dishes in osteoclastogenic medium to ,80% confluence. The Chip Assay was described previously [34]. DNA was extracted with a Wizard Genomic DNA Purification Kit (Promega KK, Tokyo, Japan). Ethanol-precipitated DNA was solubilized in water (1.0610 6 cell equivalent/30 mL). Semiquantitative PCR was performed following the method for RT-PCR, and was performed using primers listed in Table 1.

siRNA knockdown
siRNAs, chemically synthesized and purified by HPLC, were purchased from Japan Bio Services Co., Ltd. (Saitama, Japan). siRNA were performed using sequences listed in Table 2.
Transient transfection with siRNA was performed at 2-day intervals in fresh osteoclastogenic medium with HilyMAX reagent (Dojindo, Kumamoto, Japan) in accordance with the manufacturer's instructions.

Statistical analysis
Statistical analysis was performed using Student's t-test to compare two samples. Statistical analysis of comparisons between multiple groups (more than two groups) was performed using oneway and two-way ANOVA with StatPlus software (AnalystSoft). Statistical significance was set at P,0.05 for all tests. Results shown are representative examples of three independent experiments.

IRF4 increases during osteoclastogenesis
To assess the expression of IRF4 during osteoclastogenesis, we used RT-PCR and immunoblot analyses to detect IRF4 expression in RAW264.7 cells after RANKL stimulation (Fig. 1A, D; full-length blots in Fig. S1A), and showed that robust induction of NFATc1 by RANKL is a necessary and pivotal step for osteoclast differentiation characterized by enhanced expression of Jmjd3, which is an H3K27me3 demethylase. In concert, both IRF4 and NFATc1 expression were higher after RANKL stimulation. In addition, activation of EZH2-mediated H3K27 methylation increased during the later stage of osteoclastogenesis (Fig. 1A).

Epigenetic regulation of IRF4 and NFATc1 genes in osteoclastogenesis
We examined the mechanism underlying the increase in IRF4 and NFATc1 expression with RANKL. We employed a chromatin immunoprecipitation assay using anti-H3K27me3 antibody to evaluate the interaction between H3K27me3modified DNA with the IRF4 and NFATc1 promoters in RAW264.7 cells. We confirmed by ChIP analysis that H3K27 in the promoter region of IRF4 is methylated in osteoclast precursors ( Fig. 1B; full-length gels in Fig. S1B). Another study has indicated that NFATc1 is apparently epigenetically regulated by Jmjd3 in osteoclastogenesis [35,36]. Furthermore, the expression of both NFATc1 and IRF4 increase with demethylase activity (Fig. 1A, D). NFATc1 binds to its own promoter, which leads to the robust induction of NFATc1 and this autoamplification is critical for osteoclastogenesis. Fig. 1B shows that EZH2-mediated H3K27 methylation of the promoter regions of IRF4 and NFATc1 increases during the later stage of osteoclastogenesis. We consider that the methylation acts to reduce IRF4 gene activation by the second day after RANKL stimulation.   Nuclear translocation of IRF4 and NFATc1 in osteoclastogenesis RANKL stimulation resulted in substantially higher concentrations of nuclear IRF4 and NFATc1 protein after 4 days ( Fig. 1C; full-length blots in Fig. S1C).

IRF4 accelerates transcriptional activity of NFATc1
IRF4-specific siRNA was prepared, and IRF4 knockdown cells were treated with RANKL. We found that IRF4 siRNA markedly suppressed RANKL-induced osteoclast formation (Fig. 2A). The siRNA knockdown was confirmed by attenuated levels of both IRF4 mRNA and protein ( Fig. 2A; full-length blots and gels in Fig.  S2A). Real-time PCR and western blot analyses confirmed that both NFATc1 mRNA (Fig. 2B) and protein ( Fig. 2C; full-length blots in Fig. S2C) were suppressed in osteoclastogenesis. Previous studies showed that cooperation of NFATc2 and NF-kB activates the initial induction of NFATc1 [37]. In addition, our study shows that IRF4 participates in the cooperation of NFATc2 and NF-kB to activate the initial induction of NFATc1 ( Fig. 2D; fulllength gels in Fig. S2D), which may play a role in early osteoclastogenesis.

Simvastatin represses osteoclastogenesis by reducing expression of several osteoclast-specific genes
Next, we examined the previously unexplored effect of simvastatin on osteoclast differentiation in vitro and in vivo. In this study, simvastatin inhibited RANKL-induced osteoclast formation (Fig. 3A). Real-time PCR and western blot analyses confirmed that NFATc1 mRNA (Fig. 3C), IRF4 and NFATc1 protein were suppressed during simvastatin stimulation. The NF-kB inhibitor BAY11-7082 reduced the protein level of both IRF4 and NFATc1 (Fig. 3B, D; full-length blots in Fig. S3B, D). This result shows that the role of IRF4 is partly dependent on NF-kB activation in RANKL-induced osteoclast formation. In addition, we treated RAW264.7 cells with the Rho kinase/ROCK signaling inhibitor Y-27632 and found that IRF4 expression decreased after 4 days of RANKL treatment (Fig. 3E; full-length blots in Fig. S3E). RANKL-stimulated induction of the osteoclastic genes Atp6v0d2, Cathepsin K and TRAP was also severely impaired by simvastatin without affecting the expression of DC-STAMP (Fig. 3F).

In vivo effects of simvastatin on bone anomalous absorption
To prepare a mouse model of bone loss, RANKL was injected intraperitoneally into 7-wk-old female mice. Simvas- tatin was injected from 1 day before the first RANKL injection. To determine the impact of simvastatin on bone resorption, we performed high-resolution microcomputed tomography (mCT) studies, which showed that simvastatin significantly reduced RANKL-induced bone loss (Fig. 4A, B). This reduction in bone loss was not as evident in the cortical region. The rapid decrease in BMD in this model seems not only to be caused by stimulation of the final differentiation of osteoclast progenitors but also by the activation of a preexisting pool of osteoclasts. We believe that osteoclast precursors are more abundant in the bone marrow than in blood. Bone sections immunostained for tartrate-resistant acid phosphatase (TRAP) revealed that simvastatin significantly reduced the numbers of osteoclasts in bone loss model mice following intraperitoneal administration of RANKL. Osteopontin develops early in bone formation that expression is high during remodeling site and is concerned with the bone morphogenetic process. We observed increases in both bone formation and osteoblastic activity. Immunostaining for osteopontin revealed that simvastatin does not affect bone remodeling activity, while toluidine blue staining revealed a normal rate of new bone formation rate in bone loss model mice following intraperitoneal administration of RANKL.

Discussion
A clinical trial of simvastatin in postmenopausal female patients with osteoporosis [38,39] demonstrated the ability of simvastatin to increase new bone formation [40], while an in vitro study characterized the mechanisms through which simvastatin (2.5 mM) increases expression of the BMP-2 gene in bone cells [40]. Mundy and colleagues reported [40] increased trabecular bone volume in ovariectomised rats given simvastatin at a daily dose of 5-10 mg/kg for 35 days. Although the dose per body weight in the rats was higher than the lipid-lowering dose used in humans, Mundy and colleagues predicted that there would be similar effects on bone formation in humans at lipid-lowering doses. However the U.S. Food and Drug Administration (FDA) is recommending limiting the use of the highest approved dose of simvastatin (80 mg) because of the increased risk of muscle damage reported in 2011 [41].
Several animal models have been created for the study of bone loss, such as ovariectomy (OVX) and denervation. In this study, based on the fact that osteoclast differentiation and activation are mediated by RANKL, we used RANKL-treated mice as a model of bone loss. The mechanism of bone loss in this model is simple, in that excessive RANKL directly mediates the differentiation and activation of osteoclasts. The rapid decrease in bone mineral density (BMD) in this model seems not only to be caused by stimulation of the final differentiation of osteoclast progenitors but also to the activation of a preexisting pool of osteoclasts. However, the activation of osteoclasts by RANKL may be different from normal osteoclast activation by membrane-bound RANKL produced by osteoblasts. Osteoblast-bound RANKL would likely continue to stimulate osteoclasts by cell-to-cell interaction for longer than exogenous RANKL. The RANKL model is more protective of laboratory animal welfare because of the shorter experimental periods required, the lack of any requirement for anesthesia or surgery, and the lower numbers of treatments with test materials required compared with existing approaches. However, since the term osteoporosis refers to a specific form of bone-loss disease, we have avoided using this term in the title and elsewhere.
In this study, we hypothesize that simvastatin acts via IRF4 to suppress osteoclastogenesis. However, simvastatin is not an IRF4specific inhibitor, and no IRF4 inhibitors have yet been developed. Simvastatin inhibits the numerous key proteins that function as molecular switches, including the small GTPases RAS, RAC and RAS homologue (RHO), and it is reported that RAS, RAC and RHO mediate osteoclastogenesis. Because of this, we cannot conclusively prove that simvastatin acts only via IRF4, which is one limitation of this study, but our findings strongly support our hypothesis concerning the role of IRF4 in osteoclastogenesis. Simvastatin suppresses osteoclastogenesis by inhibiting the expression of NFATc1 via the disappearance of IRF4. It was previously shown that the IRF-association domain (IAD) of IRF4 allows interaction with other IRFs such as IRF8 [12,42] which suppresses osteoclastogenesis by inhibiting the function and expression of NFATc1 [15]. In contrast, in our study, IRF4 was not found to induce the association of IRF8 in osteoclastogenesis (data not shown). IRF8 has a suppressive role in TNF-a-induced osteoclastogenesis [15]. TNF-a stimulation involves activiation of the transcription factor nuclear factor-kB (NF-kB), which plays a crucial role in osteoclast differentiation. This report shows that the role of IRF8 is independent of NF-kB activation in osteoclast differentiation. The NF-kB inhibitor BAY11-7082, is one of the best-known osteoclastogenesis inhibitors, and is shown to reduce IRF4 protein levels in osteoclast differentiation (Fig. 3B). This result shows that the role of IRF4 is dependent on NF-kB activation in osteoclast differentiation. Thus, we hypothesize that the role of IRF4 and IRF8 are independent, and that the activity of the RANKL-regulated NFATc1 promoter is directly mediated by IRF4 in osteoclastogenesis.
We examined the mechanism underlying the increase in expression of IRF4 and NFATc1 with RANKL. The increase in NFATc1 and IRF4 expression and reduced H3K27me3 detection could be coincidental and not causal. De Santa et al. [43] have recently reported that Jmjd3 is activated in an NF-kB-dependent fashion, suggesting that therapeutic targeting of the NF-kB signalling pathway [44] may be rearranged by IRF4 signalling. Interestingly, in our study, the expression level of IRF4 mRNA was decreased the second day after RANKL treatment, in contrast to NFATc1 mRNA expression which continued to increase during osteoclastogenesis (Fig. 1D), and is induced by an established autoregulatory loop in which it binds to its own promoter region, leading to its robust induction [37]. By contrast, activation of EZH2-mediated H3K27 methylation increased during the later stage of osteoclastogenesis (Fig. 1A). Fig. 1B shows that EZH2mediated H3K27 methylation increased on the promoter region of IRF4 and NFATc1 during the later stage of osteoclastogenesis. We believe that methylation acts to reduce IRF4 gene activation by the second day after RANKL stimulation.
Our data identify a mechanism by which IRF4 can enhance osteoclastogenesis (depicted in Fig. 5). A detailed analysis of the mouse NFATc1 promoter indicates that IRF4 can bind to DNA elements situated next to well-known NFATc1 binding sites, including autoamplification of its own promoter [45]. We further show that IRF4 can functionally cooperate with the NFATc1 protein and that the effect of IRF4 on expression of the osteoclastic genes Atp6v0d2, Cathepsin K and TRAP can be blocked by administration of simvastatin, which interferes with NFATc1 and IRF4 activation. Taken together these data are consistent with the notion that IRF4 can function as a lineage-specific partner for NFATc2 proteins [46]. Thus, the inductive effect of IRF4 upon osteoclast activation is likely to represent one of the critical steps that can endow osteoclasts with the ability to perform their unique set of biologic responses. Regarding formation of new bone and osteoblastic activity, performed toluidine blue staining and immunostaining of osteopontin, a key protein for the bone metabolism modulator which participates in bone formation and resorption. The present results demonstrated that in the statin group, the level of osteopontin and the volume of new bone were not affected by statin. And, Our results suggest that the depletion of osteoclast numbers were not due to the reduction in RANKL production by osteoblastic activation. Since we used RANKLtreated mice, the level of RANKL in bone rapidly increases. In an earlier report, it was demonstrated that mevastatin inhibited the fusion of osteoclasts and disrupted actin ring formation [47]. This finding is in accord with our results, because RANKL is an important protein for the fusion of preosteoclast cells [48]. Tumor necrosis factor alpha, interleukin-1, and interleukin-11 are also proteins which are well known to stimulate osteoclast differentiation. However, they act in a RANK/RANKL-independent manner [49]. To elucidate further the function of statins in osteoclast differentiation, a RANK/RANKL-independent osteoclast differentiation system should be examined in future studies.
In conclusion, this study provides evidence for the hitherto unknown effects of an IRF4 inhibitor (simvastatin) in inhibiting osteoclast differentiation and action, suggesting new therapeutic possibilities for the treatment of bone loss diseases.