TNFα Increases RANKL Expression via PGE2-Induced Activation of NFATc1

Tumor necrosis factor α (TNFα) is known to upregulate the expression of receptor activator of NF-κB ligand (RANKL). We investigated the role of the calcineurin/nuclear factor of activated T-cells (NFAT) signaling pathway in TNFα-induced RANKL expression in C2C12 and primary cultured mouse calvarial cells. TNFα-induced RANKL expression was blocked by the calcineurin/NFAT pathway inhibitors. TNFα increased NFAT transcriptional activity and subsequent RANKL promoter binding. Mutations in the NFAT-binding element (MT(N)) suppressed TNFα-induced RANKL promoter activity. TNFα increased prostaglandin E2 (PGE2) production, which in turn enhanced NFAT transcriptional activity and binding to the RANKL promoter. MT(N) suppressed PGE2-induced RANKL promoter activity. TNFα and PGE2 increased the expression of RANKL, NFAT cytoplasmic-1 (NFATc1), cAMP response element-binding protein (CREB), and cyclooxygenase 2 (COX2); which increment was suppressed by indomethacin, a COX inhibitor. Mutations in the CRE-like element blocked PGE2-induced RANKL promoter activity. PGE2 induced the binding of CREB to the RANKL promoter, whereas TNFα increased the binding of both CREB and NFATc1 to this promoter through a process blocked by indomethacin. The PGE2 receptor antagonists AH6809 and AH23848 blocked TNFα-induced expression of RANKL, NFATc1, and CREB; transcriptional activity of NFAT; and binding of NFATc1 or CREB to the RANKL promoter. These results suggest that TNFα-induced RANKL expression depends on PGE2 production and subsequent transcriptional activation/enhanced binding of NFATc1 and CREB to the RANKL promoter.

Tumor necrosis factor α (TNFα) is a multifunctional cytokine that regulates various cellular and biological processes such as cell proliferation, differentiation, apoptosis, immunity, and inflammation [9]. TNFα is known to directly induce bone resorption by activating mature osteoclasts and stimulating the proliferation and differentiation of osteoclast precursors or indirectly by inducing the expression of osteoclastogenic factors in stromal cells or osteoblasts [10][11][12][13].
Several mechanistic pathways have been proposed to determine how TNFα induces the expression of RANKL [14,15]. For example, p38 mitogen-activated protein kinase (MAPK) pathway activation mediates TNFα-induced RANKL expression and osteoclast differentiation in precursor bone marrow cells [16,17]. Cyclooxygenase (COX)/prostaglandin E (PGE) signaling is also considered a mechanistic pathway by which TNFα induces RANKL expression. Prostaglandin E2 (PGE 2 ) belongs to the family of prostanoid, autocrine, and paracrine lipid mediators produced by cells following injury or cytokine or growth factor stimulation [18]. PGE 2 has been described as a potent stimulator of osteoclastic bone resorption in the context of inflammatory diseases such as rheumatoid arthritis and osteomyelitis [19][20][21][22]. PGE 2 elevates RANKL expression in cultured mouse primary osteoblasts [20,23] and human periodontal fibroblasts [24] and is known to bind to any of four G protein-coupled receptors (EP1, EP2, EP3, or EP4) in various target cells [24,25]. COX is a prostaglandin endoperoxide synthase that catalyzes prostaglandin synthesis. Expression of the COX isoform cyclooxygenase 2 (COX2), which is thought to mediate inflammatory events, is rapidly induced by proinflammatory mediators [26,27]. TNFα is known to induce COX2 expression and PGE 2 production in human gingival fibroblasts via activation of the NFκB pathway [28]. Studies have also reported that TNFα increases RANKL expression through the COX2/PGE 2 /EP4/protein kinase A (PKA) signaling pathway [12,19,29].
We previously reported that the cAMP/PKA and calcineurin/NFAT signaling pathways must cooperate to induce parathyroid hormone-related protein (PTHrP)-induced RANKL expression in mouse osteoblastic cells [6]. The NFAT family comprises five members: NFAT cytoplasmic-1 (NFATc1) through NFAT5. Calcium signaling pathways dephosphorylate NFATc1 through NFATc4 via the activated calcineurin serine or threonine phosphatase. Dephosphorylated NFATs translocate to the nucleus and regulate the expression of target genes. The calcineurin/NFAT pathway plays an important role in bone resorption, and NFATc1 is a particularly critical transcription factor for osteoclast differentiation [30]. In mice, NFATc1 overexpression in osteoblasts led to increased osteoclast generation and bone resorption [31]. However, the role of the calcineurin/NFAT signaling pathway in TNFα-induced RANKL expression remains unexplored.
In the present study, we demonstrated that TNFα induces the transcriptional activation of NFAT via PGE 2 production and that activation of the calcineurin/NFAT signaling pathway is involved in the TNFα/COX2/PGE 2 -mediated induction of RANKL expression.

Calcineurin/Nuclear Factor of Activated T-Cells (NFAT) Signaling Is Involved in Tumor Necrosis Factor α (TNFα)-Induced Receptor Activator of Nuclear Factor-κB Ligand (RANKL) Expression
To confirm the effect of TNFα on RANKL expression in C2C12 cells, cells were incubated for 0, 1, 4, 6, 12, and 24 h in the presence of TNFα (10 ng/mL); subsequently, the RANKL expression patterns were examined. TNFα clearly upregulated the expression of both RANKL mRNA and protein, and TNFα-induced RANKL expression reached a peak at 24 h ( Figure 1A). Therefore, we chose an incubation period of 24 h for the following experiments.
To determine whether the calcineurin/NFAT pathway is activated by TNFα, we used an NFAT reporter assay with reporter plasmid containing an NFAT response element [32]. TNFα induced an approximately 4.5-fold increase in NFAT reporter activity ( Figure 1B). Next, we explored whether the calcineurin/NFAT pathway is involved in TNFα-induced RANKL expression. C2C12 cells were treated with TNFα for 24 h in the presence or absence of the calcineurin phosphatase inhibitors FK506 and cyclosporin A (CsA). Although FK506 did not affect the basal RANKL expression level, TNFα-induced expression of both RANKL mRNA and protein was blocked ( Figure 1C). CsA also downregulated both the basal and TNFα-induced RANKL protein levels and significantly suppressed TNFα-induced RANKL mRNA expression despite increasing the basal RANKL mRNA expression ( Figure 1C). downregulated both the basal and TNFα-induced RANKL protein levels and significantly suppressed TNFα-induced RANKL mRNA expression despite increasing the basal RANKL mRNA expression ( Figure 1C). These results indicate that calcineurin/NFAT pathway activation plays a role in TNFαmediated RANKL induction in C2C12 cells. Figure 1. Calcineurin/nuclear factor of activated T-cells (NFAT) activation is involved in tumor necrosis factor α (TNFα)-induced receptor activator of nuclear factor-κB ligand (RANKL) expression in C2C12 cells. (A) TNFα increased RANKL expression in a time-dependent manner. C2C12 cells were incubated in the presence of 10 ng/mL TNFα for the indicated time periods and subjected to quantitative reverse transcription-polymerase chain reaction (RT-PCR) and western blot analyses. Quantitative data are presented as means ± standard deviations (SD); (B) TNFα induces NFAT transcriptional activity. C2C12 cells were transfected with a reporter plasmid containing an NFAT response element, exposed to TNFα for 24 h, and subjected to a luciferase assay. Data are presented as firefly luciferase activity levels relative to Renilla activity; (C) Inhibition of the calcineurin/NFAT pathway blocked TNFα-mediated RANKL expression. C2C12 cells pretreated with FK506 (10 μg/mL) or cyclosporin A (10 μg/mL) were treated with TNFα for 24 h and subjected to RT-PCR and western blot analyses; (D) TNFα increases NFAT binding to the mouse RANKL promoter. C2C12 cells were incubated for 24 h with TNFα, after which a chromatin immunoprecipitation (ChIP) assay was performed using an antibody against NFATc1, with IgG serving as a negative control. The RANKL promoter region containing the NFAT binding element was amplified via PCR. Quantitative ChIP data were normalized to the input and are presented as values relative to vehicle-treated control samples (CON) (E) TNFα increased RANKL promoter-reporter activity in an NFAT binding elementdependent manner. C2C12 cells were transfected with a wild-type (RANKL-WT-luc) or NFATbinding site mutant (RANKL-MT(N)-luc) RANKL promoter reporter, incubated for 24 h in the presence of TNFα or NFAT overexpression vector, and subjected to a luciferase assay (* p < 0.05, compared to control; # p < 0.05, compared to the indicated pair).
Next, a chromatin immunoprecipitation (ChIP) assay was performed to investigate whether TNFα-activated NFAT could directly transactivate the RANKL gene by binding to its promoter. The PCR amplification result revealed that TNFα increased the NFATc1 binding to the RANKL promoter ( Figure 1D). To further confirm that the binding of NFAT to the RANKL promoter was functionally important in TNFα-induced RANKL expression, luciferase (luc) reporter assays were conducted. RANKL promoter-reporter contains −2174 to +1 bp of the mouse RANKL gene [4]. C2C12 cells were transfected with a reporter plasmid containing an NFAT response element, exposed to TNFα for 24 h, and subjected to a luciferase assay. Data are presented as firefly luciferase activity levels relative to Renilla activity; (C) Inhibition of the calcineurin/NFAT pathway blocked TNFα-mediated RANKL expression. C2C12 cells pretreated with FK506 (10 µg/mL) or cyclosporin A (10 µg/mL) were treated with TNFα for 24 h and subjected to RT-PCR and western blot analyses; (D) TNFα increases NFAT binding to the mouse RANKL promoter. C2C12 cells were incubated for 24 h with TNFα, after which a chromatin immunoprecipitation (ChIP) assay was performed using an antibody against NFATc1, with IgG serving as a negative control. The RANKL promoter region containing the NFAT binding element was amplified via PCR. Quantitative ChIP data were normalized to the input and are presented as values relative to vehicle-treated control samples (CON) (E) TNFα increased RANKL promoter-reporter activity in an NFAT binding element-dependent manner. C2C12 cells were transfected with a wild-type (RANKL-WT-luc) or NFAT-binding site mutant (RANKL-MT(N)-luc) RANKL promoter reporter, incubated for 24 h in the presence of TNFα or NFAT overexpression vector, and subjected to a luciferase assay (* p < 0.05, compared to control; # p < 0.05, compared to the indicated pair).
Next, a chromatin immunoprecipitation (ChIP) assay was performed to investigate whether TNFα-activated NFAT could directly transactivate the RANKL gene by binding to its promoter. The PCR amplification result revealed that TNFα increased the NFATc1 binding to the RANKL promoter ( Figure 1D). To further confirm that the binding of NFAT to the RANKL promoter was functionally important in TNFα-induced RANKL expression, luciferase (luc) reporter assays were conducted. RANKL promoter-reporter contains −2174 to +1 bp of the mouse RANKL gene [4].
When the RANKL-WT-luc reporter was used, TNFα induced an approximately three-fold increase in luciferase activity; however, the insertion of mutations into the NFAT binding element (−941 to −936 bp) partially inhibited the TNFα-mediated induction of RANKL promoter activity ( Figure 1E). NFATc1 overexpression (positive control) resulted in an approximately two-fold induction that was completely blocked by mutations in the NFAT binding element ( Figure 1E). These results indicate that TNFα-activated NFATc1 directly binds to and transactivates the RANKL promoter.
In C2C12 cells, TNFα treatment increased the expression levels of both COX2 mRNA and protein; these levels peaked at 24 h ( Figure 2A). Consistent with this result, TNFα treatment for 24 h strongly induced PGE 2 production, which was blocked by the COX inhibitor indomethacin ( Figure 2C). FK506 did not significantly decrease basal COX2 mRNA expression or PGE 2 production ( Figure 2B,C). However, the addition of FK506 resulted in approximately 40% and 20% decreases in TNFα-induced COX2 mRNA expression and PGE 2 production, respectively ( Figure 2B,C). These results suggest that the calcineurin/NFAT signaling pathway plays a non-critical role in TNFα-induced COX2/PGE 2 expression. When the RANKL-WT-luc reporter was used, TNFα induced an approximately three-fold increase in luciferase activity; however, the insertion of mutations into the NFAT binding element (−941 to −936 bp) partially inhibited the TNFα-mediated induction of RANKL promoter activity ( Figure 1E). NFATc1 overexpression (positive control) resulted in an approximately two-fold induction that was completely blocked by mutations in the NFAT binding element ( Figure 1E). These results indicate that TNFα-activated NFATc1 directly binds to and transactivates the RANKL promoter.
In C2C12 cells, TNFα treatment increased the expression levels of both COX2 mRNA and protein; these levels peaked at 24 h ( Figure 2A). Consistent with this result, TNFα treatment for 24 h strongly induced PGE2 production, which was blocked by the COX inhibitor indomethacin ( Figure 2C). FK506 did not significantly decrease basal COX2 mRNA expression or PGE2 production ( Figure 2B,C). However, the addition of FK506 resulted in approximately 40% and 20% decreases in TNFα-induced COX2 mRNA expression and PGE2 production, respectively ( Figure 2B,C). These results suggest that the calcineurin/NFAT signaling pathway plays a non-critical role in TNFα-induced COX2/PGE2 expression. (B) TNFα-mediated COX2 expression decreased following inhibition of the calcineurin/NFAT signaling pathway. C2C12 cells pretreated with the calcineurin phosphatase inhibitor FK506 (10 μg/mL) were incubated in the presence or absence of TNFα for 24 h and subjected to RT-PCR analyses; (C) TNFα-induced PGE2 production was blocked by indomethacin, a COX inhibitor. Treatment with FK506 slightly suppressed TNFα-induced PGE2 production. C2C12 cells were incubated with FK506 or indomethacin (20 μM) in the presence of TNFα for 24 h and subjected to a PGE2 parameter assay (* p < 0.05, compared to control; # p < 0.05).

Figure 2.
The calcineurin/NFAT signaling pathway mediates TNFα-induced cyclooxygenase 2 (COX2) expression and prostaglandin E2 (PGE 2 ) production in C2C12 cells. (A) TNFα stimulated COX2 mRNA and protein expression in a time-dependent manner; (B) TNFα-mediated COX2 expression decreased following inhibition of the calcineurin/NFAT signaling pathway. C2C12 cells pretreated with the calcineurin phosphatase inhibitor FK506 (10 µg/mL) were incubated in the presence or absence of TNFα for 24 h and subjected to RT-PCR analyses; (C) TNFα-induced PGE 2 production was blocked by indomethacin, a COX inhibitor. Treatment with FK506 slightly suppressed TNFα-induced PGE 2 production. C2C12 cells were incubated with FK506 or indomethacin (20 µM) in the presence of TNFα for 24 h and subjected to a PGE 2 parameter assay (* p < 0.05, compared to control; # p < 0.05).

PGE 2 Enhances RANKL Promoter Activity in an NFAT Binding Element-Dependent Manner
We next investigated whether calcineurin/NFAT signaling plays a role in PGE 2 -induced RANKL expression. When C2C12 cells were incubated in the presence of PGE 2 (50 nM) for 1, 4, 6, 12, and 24 h, the levels of RANKL mRNA and protein were slightly decreased at an early time point, but remarkably increased in a time-dependent manner after 6 h ( Figure 3A). PGE 2 also induced an approximately three-fold increase in NFAT-luc reporter activity ( Figure 3B). In addition, a ChIP assay with an NFATc1 antibody revealed that PGE 2 enhanced NFATc1 binding to the RANKL promoter ( Figure 3C). Furthermore, PGE 2 significantly increased luciferase activity when the RANKL-WT-luc reporter, but not when the RANKL-MT(N)-luc, was used ( Figure 3D). These results indicate that PGE 2 activates the calcineurin/NFAT signaling pathway and that PGE 2 -induced NFAT directly activates RANKL transcription by binding to the RANKL promoter.

PGE2 Enhances RANKL Promoter Activity in an NFAT Binding Element-Dependent Manner
We next investigated whether calcineurin/NFAT signaling plays a role in PGE2-induced RANKL expression. When C2C12 cells were incubated in the presence of PGE2 (50 nM) for 1, 4, 6, 12, and 24 h, the levels of RANKL mRNA and protein were slightly decreased at an early time point, but remarkably increased in a time-dependent manner after 6 h ( Figure 3A). PGE2 also induced an approximately three-fold increase in NFAT-luc reporter activity ( Figure 3B). In addition, a ChIP assay with an NFATc1 antibody revealed that PGE2 enhanced NFATc1 binding to the RANKL promoter ( Figure 3C). Furthermore, PGE2 significantly increased luciferase activity when the RANKL-WT-luc reporter, but not when the RANKL-MT(N)-luc, was used ( Figure 3D). These results indicate that PGE2 activates the calcineurin/NFAT signaling pathway and that PGE2-induced NFAT directly activates RANKL transcription by binding to the RANKL promoter. in the presence of PGE2 (50 nM) for the indicated time periods, followed by quantitative RT-PCR and western blot analyses; (B) PGE2 increased NFAT transcriptional activity. C2C12 cells were transfected with a reporter plasmid containing an NFAT response element, treated with PGE2 for 24 h, and subjected to a luciferase assay; (C) PGE2 induced NFAT binding to the mouse RANKL promoter. C2C12 cells were treated with PGE2 for 24 h and a ChIP assay was performed; (D) PGE2 increased RANKL promoter-reporter activity in an NFAT binding element-dependent manner. C2C12 cells were transfected with RANKL-WT-luc or RANKL-MT(N)-luc, incubated for 24 h in the presence of PGE2, and subjected to a luciferase assay (* p < 0.05, compared to control; # p < 0.05).

Cyclooxygenase (COX) Inhibitor Blocks TNFα-Induced Binding of NFATc1 and cAMP Response Element-Binding Protein (CREB) to the RANKL Promoter
Significant increases in the mRNA and protein expression of RANKL, NFATc1, cAMP response element-binding protein (CREB), and COX2 were observed in C2C12 cells incubated with PGE2 for 24 h ( Figure 4A). TNFα induced similar levels of gene expression, and this process was inhibited by indomethacin ( Figure 4A). Further treatment with PGE2 in the presence of indomethacin and TNFα partially rescued the expression of RANKL, NFATc1, CREB, and COX2. These results suggest that COX/PGE2 mediates the TNFα-induced expression of RANKL, NFATc1, and CREB.
We previously demonstrated that PTHrP-induced cAMP/PKA signaling promoted NFAT transcriptional activity and that NFAT and CREB cooperate to transactivate the gene encoding RANKL in mice [6]. Because our above data demonstrate that PGE2 enhances the transcriptional activity of NFAT and its binding to the RANKL promoter, we next investigated whether PGE2, produced in response to TNFα, would enhance the binding of CREB to the RANKL promoter and subsequent transactivation. The RANKL promoter-reporter assay demonstrated that mutations inserted in the CRE-like element (−1093 to −1086 bp) prevented PGE2 from enhancing reporter activity ( Figure 4B). In addition, a ChIP assay with a CREB antibody revealed that PGE2 enhanced the binding of CREB to the RANKL promoter ( Figure 4C). Consistent with these results, TNFα significantly increased the binding of CREB and NFATc1 to the RANKL promoter in a process blocked by

Cyclooxygenase (COX) Inhibitor Blocks TNFα-Induced Binding of NFATc1 and cAMP Response Element-Binding Protein (CREB) to the RANKL Promoter
Significant increases in the mRNA and protein expression of RANKL, NFATc1, cAMP response element-binding protein (CREB), and COX2 were observed in C2C12 cells incubated with PGE 2 for 24 h ( Figure 4A). TNFα induced similar levels of gene expression, and this process was inhibited by indomethacin ( Figure 4A). Further treatment with PGE 2 in the presence of indomethacin and TNFα partially rescued the expression of RANKL, NFATc1, CREB, and COX2. These results suggest that COX/PGE 2 mediates the TNFα-induced expression of RANKL, NFATc1, and CREB.
We previously demonstrated that PTHrP-induced cAMP/PKA signaling promoted NFAT transcriptional activity and that NFAT and CREB cooperate to transactivate the gene encoding RANKL in mice [6]. Because our above data demonstrate that PGE 2 enhances the transcriptional activity of NFAT and its binding to the RANKL promoter, we next investigated whether PGE 2 , produced in response to TNFα, would enhance the binding of CREB to the RANKL promoter and subsequent transactivation. The RANKL promoter-reporter assay demonstrated that mutations inserted in the CRE-like element (−1093 to −1086 bp) prevented PGE 2 from enhancing reporter activity ( Figure 4B). In addition, a ChIP assay with a CREB antibody revealed that PGE 2 enhanced the binding of CREB to the RANKL promoter ( Figure 4C). Consistent with these results, TNFα significantly increased the binding of CREB and NFATc1 to the RANKL promoter in a process blocked by indomethacin ( Figure 4D). These results suggest that TNFα-induced COX/PGE 2 increases the binding of both CREB and NFAT to the RANKL promoter and transcriptional activation of the RANKL gene. indomethacin ( Figure 4D). These results suggest that TNFα-induced COX/PGE2 increases the binding of both CREB and NFAT to the RANKL promoter and transcriptional activation of the RANKL gene. The RANKL promoter region containing the CREB-binding element was then amplified; (D) Indomethacin prevented TNFα-induced NFAT and CREB binding to the mouse RANKL promoter. C2C12 cells were incubated for 24 h with the indicated reagents and subjected to a ChIP assay (* p < 0.05, compared to control; # p < 0.05).

The PGE2 Receptor Antagonists AH6809 and AH23848 Blocked TNFα-Induced Activation of NFAT and Binding of NFATc1 and CREB to the RANKL Promoter
We next investigated whether PGE2 receptor antagonists would inhibit TNFα-induced RANKL expression and NFAT activation. AH6809 has the highest affinity for the EP2 receptor, but also acts as a weak antagonist against EP1 and prostaglandin D2 receptor 1 (DP1) receptors in the mouse [35]. AH23848 is a dual antagonist of the EP4 and thromboxane (TP1) receptors [36,37]. C2C12 cells were incubated with AH6809 (5 μM) in the presence of TNFα for 1, 4, 6, 12, and 24 h, and then the levels of RANKL mRNA and protein were examined. AH6809 did not significantly suppress TNFα-induced RANKL expression at an early time point (6 h or earlier), but inhibited the TNFα-mediated induction of RANKL thereafter ( Figure 5A). Incubation for 24 h in the presence of AH6809 or AH23848 blocked the TNFα-mediated induction of RANKL, NFATc1, and CREB mRNA expression in C2C12 cells ( Figure 5B). Compared with NFATc1 mRNA expression, TNFα-induced NFAT protein levels decreased moderately in the presence of AH6809 or AH23848 ( Figure 5B). However, TNFα-induced NFAT transcriptional activity was abolished by the addition of AH6809 or AH23848 ( Figure 5C). C2C12 cells were incubated with the indicated reagents for 24 h and subjected to RT-PCR and western blot analyses; (B) PGE 2 increased RANKL promoter-reporter activity in a CREB binding element-dependent manner. C2C12 cells were transfected with RANKL-WT-luc or a CREB-binding site mutant (RANKL-MT(C)-luc) RANKL promoter, incubated for 24 h in the presence of PGE 2 , and subjected to a luciferase assay; (C) PGE 2 induced CREB binding to the mouse RANKL promoter. C2C12 cells were incubated for 24 h with PGE 2 and subjected to a ChIP assay with CREB and control IgG antibodies. The RANKL promoter region containing the CREB-binding element was then amplified; (D) Indomethacin prevented TNFα-induced NFAT and CREB binding to the mouse RANKL promoter. C2C12 cells were incubated for 24 h with the indicated reagents and subjected to a ChIP assay (* p < 0.05, compared to control; # p < 0.05).

The PGE 2 Receptor Antagonists AH6809 and AH23848 Blocked TNFα-Induced Activation of NFAT and Binding of NFATc1 and CREB to the RANKL Promoter
We next investigated whether PGE 2 receptor antagonists would inhibit TNFα-induced RANKL expression and NFAT activation. AH6809 has the highest affinity for the EP2 receptor, but also acts as a weak antagonist against EP1 and prostaglandin D2 receptor 1 (DP1) receptors in the mouse [35]. AH23848 is a dual antagonist of the EP4 and thromboxane (TP1) receptors [36,37]. C2C12 cells were incubated with AH6809 (5 µM) in the presence of TNFα for 1, 4, 6, 12, and 24 h, and then the levels of RANKL mRNA and protein were examined. AH6809 did not significantly suppress TNFα-induced RANKL expression at an early time point (6 h or earlier), but inhibited the TNFα-mediated induction of RANKL thereafter ( Figure 5A). Incubation for 24 h in the presence of AH6809 or AH23848 blocked the TNFα-mediated induction of RANKL, NFATc1, and CREB mRNA expression in C2C12 cells ( Figure 5B). Compared with NFATc1 mRNA expression, TNFα-induced NFAT protein levels decreased moderately in the presence of AH6809 or AH23848 ( Figure 5B). However, TNFα-induced NFAT transcriptional activity was abolished by the addition of AH6809 or AH23848 ( Figure 5C). Furthermore, ChIP assays with antibodies to NFATc1 or CREB demonstrated that PGE 2 receptor antagonists suppressed TNFα-induced binding of NFATc1 or CREB to the RANKL promoter ( Figure 5D). These results indicate that TNFα-induced RANKL expression depends on the production of PGE 2 , which subsequently binds to EP2 or EP4 receptor and enhances the binding and transcriptional activity of NFATc1 and CREB in the mouse RANKL promoter. Furthermore, ChIP assays with antibodies to NFATc1 or CREB demonstrated that PGE2 receptor antagonists suppressed TNFα-induced binding of NFATc1 or CREB to the RANKL promoter ( Figure  5D). These results indicate that TNFα-induced RANKL expression depends on the production of PGE2, which subsequently binds to EP2 or EP4 receptor and enhances the binding and transcriptional activity of NFATc1 and CREB in the mouse RANKL promoter.

TNFα Enhances RANKL Expression in a PGE2 Production-and NFAT Activation-Dependent Manner in Primary Cultured Mouse Calvarial Cells
We next confirmed the roles of PGE2 and calcineurin/NFAT signaling with regard to TNFαregulated RANKL expression in primary cultured mouse calvarial (MC) cells. Consistent with our observations in C2C12 cells, TNFα induced both RANKL and COX2 mRNA and protein expression ( Figure 6A). PCR amplification of the DNA region containing the NFAT binding element or CRE-like region revealed that TNFα enhanced the binding of NFATc1 or CREB to the RANKL promoter ( Figure 6B). FK506-mediated inhibition of NFAT activation significantly reduced TNFα-induced

TNFα Enhances RANKL Expression in a PGE 2 Production-and NFAT Activation-Dependent Manner in Primary Cultured Mouse Calvarial Cells
We next confirmed the roles of PGE 2 and calcineurin/NFAT signaling with regard to TNFα-regulated RANKL expression in primary cultured mouse calvarial (MC) cells. Consistent with our observations in C2C12 cells, TNFα induced both RANKL and COX2 mRNA and protein expression ( Figure 6A). PCR amplification of the DNA region containing the NFAT binding element or CRE-like region revealed that TNFα enhanced the binding of NFATc1 or CREB to the RANKL promoter ( Figure 6B). FK506-mediated inhibition of NFAT activation significantly reduced TNFα-induced RANKL expression ( Figure 6C). Indomethacin, AH6809, and AH23848 also inhibited TNFα-induced RANKL mRNA expression ( Figure 6D). These results further suggest that PGE 2 mediates TNFα-induced RANKL expression through a process that involves enhanced binding of NFAT and CREB to the RANKL promoter. RANKL expression ( Figure 6C). Indomethacin, AH6809, and AH23848 also inhibited TNFα-induced RANKL mRNA expression ( Figure 6D). These results further suggest that PGE2 mediates TNFαinduced RANKL expression through a process that involves enhanced binding of NFAT and CREB to the RANKL promoter.

Discussion
In the present study, we demonstrated that TNFα-induced RANKL expression involves both PGE2-induced activation of the calcineurin/NFAT pathway and subsequent binding of NFATc1 to the RANKL promoter. The following results of our study support the involvement of PGE2 and the calcineurin/NFAT signaling pathway in TNFα-induced RANKL expression: (i) TNFα stimulated COX2 expression and PGE2 production; (ii) a COX inhibitor or EP antagonists blocked TNFα-induced RANKL expression; (iii) TNFα-induced RANKL expression was significantly downregulated by inhibition of the calcineurin/NFAT pathway; (iv) TNFα enhanced the transcriptional activity of NFATc1 and binding of NFATc1 to the RANKL promoter; (v) TNFα-induced RANKL promoterreporter activity was attenuated by mutations in the NFAT binding element; and (vi) a COX inhibitor or EP antagonists blocked TNFα-enhanced transcriptional activity and binding of NFAT to the RANKL promoter.
It has been well demonstrated that RANKL expression increases with TNFα treatment in a dosedependent manner and 10 ng/mL of TNFα was used as a representative dose in several studies [38][39][40][41][42][43]. Therefore, we adopted a concentration of 10 ng/mL in the present study.
The ability of TNFα to stimulate PGE2 production via COX2 has been well demonstrated [28,[44][45][46], and PGE2 is known to activate the cAMP/PKA signaling pathway in target cells [19]. Of the PGE2 receptor subtypes, EP4 has the highest expression in fibroblasts. Compared to control mice, EP4deficient mice exhibited reduced inflammatory responses in a collagen-induced arthritis model,

Discussion
In the present study, we demonstrated that TNFα-induced RANKL expression involves both PGE 2 -induced activation of the calcineurin/NFAT pathway and subsequent binding of NFATc1 to the RANKL promoter. The following results of our study support the involvement of PGE 2 and the calcineurin/NFAT signaling pathway in TNFα-induced RANKL expression: (i) TNFα stimulated COX2 expression and PGE 2 production; (ii) a COX inhibitor or EP antagonists blocked TNFα-induced RANKL expression; (iii) TNFα-induced RANKL expression was significantly downregulated by inhibition of the calcineurin/NFAT pathway; (iv) TNFα enhanced the transcriptional activity of NFATc1 and binding of NFATc1 to the RANKL promoter; (v) TNFα-induced RANKL promoter-reporter activity was attenuated by mutations in the NFAT binding element; and (vi) a COX inhibitor or EP antagonists blocked TNFα-enhanced transcriptional activity and binding of NFAT to the RANKL promoter.
It has been well demonstrated that RANKL expression increases with TNFα treatment in a dose-dependent manner and 10 ng/mL of TNFα was used as a representative dose in several studies [38][39][40][41][42][43]. Therefore, we adopted a concentration of 10 ng/mL in the present study.
The ability of TNFα to stimulate PGE 2 production via COX2 has been well demonstrated [28,[44][45][46], and PGE 2 is known to activate the cAMP/PKA signaling pathway in target cells [19]. Of the PGE 2 receptor subtypes, EP4 has the highest expression in fibroblasts. Compared to control mice, EP4-deficient mice exhibited reduced inflammatory responses in a collagen-induced arthritis model, fewer osteoclasts, and reduced RANKL expression in osteoblast cells [20,[47][48][49]. These reports suggest that EP4 is critical to the stimulation of bone resorption. Another report demonstrated that in MC osteoblastic cells, PGE 2 -induced cAMP production and RANKL expression are mainly mediated by EP4, although EP4 and EP2 must cooperate to ensure a full response to PGE 2 [23]. In the present study, either an EP2 or EP4 antagonist completely blocked TNFα-induced RANKL expression. In addition, both antagonists abolished TNFα-induced NFAT reporter activity and the binding of NFATc1 and CREB to the RANKL promoter in C2C12 cells. In primary cultured MC cells, both antagonists had similar partial inhibitory effects on TNFα-induced RANKL expression. These results suggest that both EP2 and EP4 are involved in PGE 2 -induced RANKL expression. However, it is not clear why the responses of primary MC cells and C2C12 cells to EP antagonists differ.
In the present study, we did not examine the receptor subtypes involved in TNFα-induction of COX2 and PGE 2 production. TNFα regulates cellular activities via the activation of either of two receptors: TNF receptor 1 (TNFR1) and TNFR2. TNFR1 is expressed ubiquitously, whereas TNFR2 expression is restricted to specific cell types, including immune cells and neurons [50]. Previous reports have demonstrated that TNFα induces COX2 expression and PGE 2 production via TNFR1 in synovial fibroblasts [51,52]. Furthermore, induction of RANKL by TNFα was abolished in TNFR1-deficient mouse gingival epithelial cells [29]. These studies suggest that the activation of TNFR1 contributes to TNFα-induced activation of COX2 and production of PGE 2 in C2C12 and mouse calvarial cells.
Although the PGE 2 -induced cAMP/PKA signaling is known to induce RANKL expression, the effector molecules downstream of PKA have not yet been clearly elucidated. Previously, PKA was shown to target CREB and thus increase the extent of CREB binding to distal enhancers of RANKL [53]. Assuming that PGE 2 /PKA signaling mediates TNFα-induced RANKL expression, we propose NFAT to be another transcription factor that acts downstream of PGE 2 to induce RANKL transcription. Similar to a previous report in which CREB and NFATc1 cooperation was necessary to induce RANKL transcription via the PTHrP/cAMP/PKA pathway [6], TNFα-induced PGE 2 enhanced the binding of CREB and NFATc1 to the mouse RANKL promoter, and mutations in either NFAT-binding element or the CRE-like element blocked PGE 2 -induced RANKL promoter activity. In addition, indomethacin and EP antagonists prevented TNFα-induced binding of NFATc1 and CREB to the RANKL promoter. These data support a role for NFATc1 as another mediator linking PGE 2 /PKA signaling to RANKL transcription.
A rich body of literature demonstrates that NFAT activation directly regulates COX expression and PGE 2 production. Notably, NFAT activation regulates COX2-encoding genes in human colon carcinoma [33]. In this study, COX2 expression and PGE 2 production were induced in response to NFAT stimuli and blocked when CsA or FK506 was used to inhibit calcineurin phosphatase activity. Lipopolysaccharide-induced NFAT activation was shown to regulate PGE 2 synthesis in dendritic cells [34], and NFAT was found to induce COX2 transcription in human glioblastoma cells [54] and to regulate constitutive COX2 expression in the renal medulla [55]. Furthermore, PTH was shown to induce COX2 transcription through cross-talk between the cAMP/PKA and calcineurin/NFAT signaling pathways in murine osteoblastic cells [56]. Therefore, we determined whether inhibition of the calcineurin/NFAT pathway would suppress COX2-induced PGE 2 production. FK506 partially suppressed TNFα-induced COX2 expression and PGE 2 production. However, functional PGE 2 inhibition by EP antagonists completely abolished the TNFα-induced transcriptional activity of NFAT, indicating that TNFα-induced NFAT activation is PGE 2 -dependent and subsequently contributes to the further induction of COX2 expression.
In the present study, TNFα-induced COX2 expression was obvious only after incubation for 4 h, although TNFα-induced RANKL expression was significant even after 1 h-incubation in C2C12 cells. In addition, PGE 2 antagonist AH6809 did not significantly suppress TNFα-induced RANKL expression at an early time point (6 h or earlier). Given the data demonstrating the difference between the early vs. late response, it is suggested that there is a PGE 2 -independent pathway in RANKL induction of TNFα in the immediate response. Considering the continuing effect of TNFα and PGE 2 on RANKL induction after 12 h, PGE 2 seems to mediate the slow, delayed, long-term RANKL expression by TNFα, which is more likely to be associated with pathologic conditions such as inflammatory bone resorption.
In conclusion, the present study demonstrates that TNFα-induced RANKL expression depends on PGE 2 production and subsequent transcriptional activation, as well as on the enhanced binding of NFATc1 and CREB to the RANKL promoter in mouse osteoblastic cells.

Cell Culture
C2C12 cells, a murine mesenchymal cell line that can be differentiated into osteoblasts [57], were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/mL of penicillin, and 100 µg/mL of streptomycin.
MC cells were isolated from the frontal and parietal bones of neonatal ICR mice, as previously described [58]. Before preparation of MC cells, animals were weaned in mouse gang cages following Institutional Animal Care and Use Committee policies. The animal study was reviewed and approved by the Special Committee on Animal Welfare, Seoul National University, Seoul, Republic of Korea (approval no. SNU-20140228-1-5). MC cells were cultured in α-minimum essential medium (α-MEM) supplemented with 10% FBS, 100 U/mL of penicillin, and 100 µg/mL of streptomycin. DMEM, α-MEM, and FBS were obtained from Hyclone (Walkersville, MD, USA).

Western Blot Analysis
The western blot analysis was performed as described previously [6]. Immune complexes were visualized using EZ-Western Lumi Pico (Daeillab Service Co., Seoul, Korea), and chemiluminescence was detected using a MicroChemi device (DNR, Jerusalem, Israel).

Chromatin Immunoprecipitation Assay
The chromatin immunoprecipitation (ChIP) assay was performed as described previously [4]. Cross-linked DNA fragments were subjected to pre-clearing with blocked protein G agarose, and immunoprecipitation was performed with an NFATc1 or CREB antibody or species-matched control IgG. After purification, DNA sequences were analyzed by PCR amplification of the mouse RANKL promoter region encompassing either the NFAT-binding element (amplified region: −1070 to −858 bp) or CRE-like element (amplified region: −1147 to −967 bp). PCR primers with the following sequences were used: NFAT-binding element forward, 5 -GCA AGC TCC AGG CCA GCC TAG-3 and reverse, 5 -CCA ATA AGA CGG CTC AGC TG-3 ; CRE-like element forward, 5 -AGG AGG CAG AGA TGG CAG AG-3 and reverse, 5 -ACA CGC GCG CGC GCA AAT A-3 .

Luciferase Reporter Assay
The luciferase reporter assay was performed as described previously [6]. In order to normalize transfection efficiency, Renilla luciferase plasmid was co-transfected with RANKL promoter reporters.

PGE 2 Assay
Cells were incubated for 24 h in medium supplemented with the indicated reagents, after which supernatants were collected. The concentration of PGE 2 in the supernatant was determined using Parameter PGE 2 (R&D Systems) according to the manufacturer's instructions.

Statistical Analysis
Statistical significance was determined using Student's t-test. For the multiple comparisons, one way ANOVA was performed. When significant main effects were detected, post hoc analyses were conducted with the least squares means error test. Differences were considered significant at p < 0.05. Data were analyzed by using the SAS program (version 9.1; SAS Institute, Cary, NC, USA).