RBP-J–Regulated miR-182 Promotes TNF-α–Induced Osteoclastogenesis

This article is featured in In This Issue, p.4837

Osteoclasts, multinucleated giant cells derived from the monocyte/macrophage lineage, are responsible for bone resorption. As the exclusive bone-degrading cells, osteoclasts play an indispensable role in physiological bone development, remodeling, and repair. Osteoclastogenesis is physiologically triggered by RANKL in the presence of M-CSF– and ITAM-mediated costimulation. Upon stimulation by these factors, a broad range of signaling cascades is activated, such as NF-κB pathways, protein tyrosine kinases and calcium signaling, and MAPK pathways. These signaling cascades lead to induction of the key transcription factor NFATc1 that functions in concert with other positive regulators, such as c-Fos and B lymphocyte–induced maturation protein-1 (Blimp1), to drive osteoclast differentiation (1–9). Recent evidence has made it clear that the process of osteoclast differentiation is also delicately controlled by a braking system, in which negative regulators, such as IFN regulatory factor 8 (Irf8), v-maf avian musculoaponeurotic fibrosarcoma oncogene homolog B, and B cell lymphoma, restrain the numbers of osteoclasts that are generated to prevent excessive bone resorption that leads to bone loss (10). Inflammation promotes osteoclastogenesis; thus, osteoclasts also function as pathogenic cells leading to excessive bone resorption that is commonly associated with inflammatory bone diseases, such as rheumatoid arthritis (RA), periodontitis, and periprosthetic osteolysis. The inflammatory cytokine TNF-α plays a major role, primarily in synergy with RANKL, in promoting pathologic osteoclastogenesis and bone resorption in these inflammatory diseases (2, 9, 11, 12). However, compared with RANKL, TNF-α alone does not effectively induce osteoclast differentiation. The mechanisms that restrain TNF-α–induced osteoclastogenesis are much less understood than those that promote osteoclastogenesis in response to RANKL (2, 13).

Recently, we discovered that transcription factor RBP-J functions as a novel osteoclastogenic repressor and plays a critical role in inhibiting TNF-α–induced osteoclast differentiation and bone resorption (13). RBP-J functions as a central transcription factor that receives input from several signaling pathways, including the canonical Notch pathway and Wnt–β-catenin and NF-κB pathways in a context-dependent manner to regulate cell differentiation, survival, and many other cellular responses and activities (13–19). Distinct from most negative regulators of osteoclast differentiation, a unique feature of RBP-J is that it plays a prominent and selective role in inhibiting TNF-α–induced osteoclastogenesis, with minimal effects on RANKL-induced osteoclastogenesis (13). Recent genetic studies revealed that RBPJ allelic polymorphisms are linked with susceptibility to RA (20–22). In parallel, RBP-J expression levels are lower in osteoclast precursors isolated from the synovial fluid of RA patients compared with those from healthy donors (19). These studies establish the critical role of RBP-J in restraining TNF-α–mediated inflammatory osteoclastogenesis and support a role for RBP-J in RA disease pathogenesis. Therefore, elucidation of the targets of RBP-J action and mechanisms of its function has the potential to identify novel therapeutic targets for treating excessive osteoclastogenesis and inflammatory bone erosion. The molecular mechanisms by which RBP-J limits TNF-α–induced osteoclast differentiation are not fully understood.

MicroRNAs (miRNAs) are a family of small evolutionarily conserved noncoding ssRNAs containing ∼22 nt that are derived from longer transcribed precursor transcripts. miRNAs repress gene expression by targeting specific mRNAs. They bind specific mRNAs via imperfect complementary binding but with a perfect base pairing between the miRNA seed region (nucleotides 2–7 of the miRNA) and the targeted sequences of mRNAs. miRNAs regulate gene expression at the posttranscriptional level by promoting degradation or inhibiting translation of specific target mRNAs or a combination of both mechanisms. miRNAs account for ∼3% of the human genome but regulate ∼90% of protein-coding genes (23–29). The last decade of studies demonstrated the importance of miRNAs in various biological and pathological settings. As potential therapeutic targets or biomarkers, miRNAs have been gaining much clinical attention (e.g., in immunity, cancers, neurologic diseases, and metabolic disorders) in recent years (23, 28, 30–32). The investigation of the role of miRNAs in bone biology and diseases is just emerging. A few miRNAs were reported to play important roles in RANKL-induced osteoclast differentiation (33–37). Nonetheless, it remains unclear how TNF-α mobilizes miRNA expression during inflammatory osteoclast differentiation and, conversely, how miRNAs regulate TNF-α–induced osteoclastogenesis in an inflammatory setting.

Although miRNAs are involved in almost all major cellular functions, miRNA expression is highly specific to cell and tissue types and is correlated with various stresses or disease settings (23). The expression patterns and roles of miRNAs involved in TNF-α stimulation are much underexplored so far and remain unknown in osteoclast biology. To address these questions and further explore the molecular mechanisms by which RBP-J limits TNF-α–induced osteoclast differentiation, we took advantage of a miRNA sequencing (miRNA-seq) technique and performed expression-profile analysis of miRNAs in response to TNF-α stimulation during osteoclast differentiation using wild-type control and RBP-J–deficient bone marrow–derived macrophages (BMMs). We profiled the miRNAs specifically involved in TNF-α–induced osteoclast differentiation. Furthermore, we focused on miR-182, which is the most abundant miRNA induced by TNF-α and whose expression is repressed by RBP-J during osteoclast differentiation. The important function of miR-182 in cancer, cell growth and cell fate, and T lymphocyte expansion was recently appreciated (38–46). In the current study, we identified miR-182 as a novel and critical positive regulator of inflammatory osteoclastogenesis mediated by TNF-α, with minimal effects on RANKL-induced osteoclastogenesis. Moreover, miR-182 is negatively controlled by RBP-J. We also identified two miR-182 targets, Maml1 and Foxo3a, as negative regulators of TNF-α–induced osteoclast differentiation. Therefore, our study uncovered a novel regulatory network, in which miR-182 functions as an important new node that receives inputs from RBP-J and TNF-α signaling and positively regulates inflammatory osteoclastogenesis. Suppression of miR-182 by RBP-J serves as an important mechanism that restrains TNF-α–induced osteoclastogenesis.

Rbpj^ΔM/ΔM [Rbpj^flox/floxLysMcre(+)] mice were described previously (13). Age- and gender- matched mice with a LysMcre(+) genotype (hereafter referred to as wild-type control) were used as controls. Eight-week-old male C57/BL6 mice (The Jackson Laboratory) were used in Maml1- and Foxo3a-knockdown experiments. All mouse experiments were approved by the Institutional Animal Care and Use Committee of the Hospital for Special Surgery.

Murine TNF-α and soluble RANKL were purchased from PeproTech. mirVana miRNA inhibitor mmu-miR-182-5p (Assay ID: MH13088; cat. no. 4464084), mirVana miRNA Inhibitor, Inhibitor Negative Control #1 (cat. no. 4464076), and mirVana miRNA mimic, Mimic Negative Control #1 (cat. no. 4464058) were purchased from Life Technologies. SignalSilence Foxo3a siRNA I (Mouse Specific) (cat. no. 8620S) was purchased from Cell Signaling; Silencer Negative Control No. 1 siRNA (AM4611; Life Technologies) was used as a control small interfering RNA (siRNA). SMARTpool: siGENOME Mouse Maml1 siRNA (cat. no. M-059179-02-0005) and corresponding control siRNA were purchased from Dharmacon.

Murine bone marrow cells were harvested from tibiae and femora and cultured overnight in Petri dishes (BD Biosciences) in α-MEM (Invitrogen) with 10% FBS (Invitrogen). Except where stated, CMG14-12 supernatant, which contained the equivalent of 20 ng/ml rM-CSF, was used as a source of M-CSF, as described (13), in experiments. Nonadherent cells were replated into tissue culture dishes and cultured in the same medium for 3 d to obtain BMMs, which are capable of differentiating to osteoclasts, and, thus, were used as osteoclast precursors. The attached BMMs were scraped, seeded at a density of 4.5 × 10⁴/cm², and cultured in α-MEM with 10% FBS and CMG14-12 supernatant for 1 d. Except where stated, the cells at this time point were used for the basal condition. The cells were treated or not with TNF-α (40 ng/ml) or RANKL (40 ng/ml) for the times indicated in the figure legends. Culture media were exchanged every 3 d. For osteoclast-differentiation assay in 96-well plates, four replicate wells/condition were used. TRAP staining was performed with an acid phosphatase leukocyte diagnostic kit (Sigma-Aldrich), in accordance with the manufacturer’s instructions.

MicroRNA antagomirs and mimics were applied to silence and overexpress microRNA expression, respectively. Antagomirs were used at concentrations of 40 nM, whereas mimics were used at concentrations of 10 nM. siRNAs were used at the concentrations indicated in the figures. Antagomirs, mimics, siRNAs, or their corresponding control oligonucleotides were transfected into murine BMMs using TransIT-TKO Transfection Reagent (Mirus), in accordance with the manufacturer’s instructions.

BMMs were transfected with miR-182 mimics (10 nM) or antagomirs (40 nM) or corresponding controls for 24 h. Then cells were replated and seeded at a density of 4.5 × 10⁴/cm² and cultured in α-MEM with 10% FBS and CMG14-12 supernatant for the times shown in the figure legends. The cells attached to the plate at 4 h after reseeding was regarded as time 0. Cell proliferation and viability were analyzed by a CyQUANT Cell Proliferation Assay Kit (Invitrogen) or a Cell Proliferation Kit I (MTT assay; Roche), following the manufacturers’ instructions. Data analyzed by these two methods showed similar results. Relative proliferation/viability levels of the cells at each time point in the MTT assay were calculated as the ratio of the absorbance [A_550nm−A_690nm] at each indicated time point/corresponding time 0.

Pit-formation assay was performed as previously described (13). BMMs were transfected with miR-182 mimics (10 nM), antagomirs (40 nM), Foxo3a siRNAs (80 nM), Maml1 siRNAs (80 nM), or corresponding controls for 24 h. Then cells were replated on dentin slices (4 mm diameter, 0.2 mm thick; 5 × 10⁴ cells/slice) in 96-well culture plates with 200 μl culture medium/well. The cells were cultured for 10 d with TNF-α, with or without RANKL priming in the presence of M-CSF, with media exchanges every 3 d. The dentin slices were washed with water, and pits formed by mature osteoclasts on the dentin slices were stained with 1% toluidine blue O (Sigma-Aldrich). The total resorptive pit area on each slice was analyzed by BIOQUANT OSTEO II software. The relative pit area (% of slice) was calculated as the percentage of total pit area relative to the dentin slice area.

For quantification of microRNA, total RNA was isolated, and the small RNA fraction was enriched with the mirVana miRNA Isolation Kit (Life Technologies), according to the manufacturer’s instructions. For quantitative RT-PCR analysis of miRNA, cDNA was prepared from total RNA with the TaqMan microRNA Reverse Transcription Kit (Applied Biosystems). TaqMan MicroRNA assays were used according to the manufacturer’s recommendations (Applied Biosystems) for real-time PCR. The TaqMan U6 snRNA assay (Applied Biosystems) was used for normalization of expression values.

For quantification of mRNA, reverse-transcription and real-time PCR were performed as previously described (13). The primers for real-time PCR were as follows: Acp5: 5′-ACGGCTACTTGCGGTTTC-3′ and 5′-TCCTTGGGAGGCTGGTC-3′; Ctsk: 5′-AAGATATTGGTGGCTTTGG-3′ and 5′-ATCGCTGCGTCCCTCT-3′; Itgb3: 5′-CCGGGGGACTTAATGAGACCACTT-3′ and 5′-ACGCCCCAAATCCCACCCATACA-3′; and Gapdh: 5′-ATCAAGAAGGTGGTGAAGCA-3′ and 5′-AGACAACCTGGTCCTCAGTGT-3′.

Total cell extracts were obtained using lysis buffer containing 150 mM Tris-HCl (pH 6.8), 6% SDS, 30% glycerol, and 0.03% Bromophenol Blue; 10% 2-ME was added immediately before harvesting cells. Cell lysates were fractionated on 7.5% SDS-PAGE, transferred to Immobilon-P membranes (Millipore), and incubated with specific Abs. Western Lightning plus-ECL (PerkinElmer) was used for detection. Densitometry was performed using ImageJ software (National Institutes of Health). NFATc1 Ab was from BD Biosciences; Blimp1, GAPDH, and p38α Abs were from Santa Cruz Biotechnology; Foxo3 Ab was from Cell Signaling; and Maml1 Ab was from Bethyl Laboratories. p-Akt (serine 473) Ab was obtained from Cell Signaling.

Total RNA was isolated, and the small RNA fraction was enriched with the mirVana miRNA Isolation Kit (Life Technologies), according to the manufacturer’s instructions. miRNA libraries were constructed per the Illumina TruSeq Small RNA Library preparation kit. High-throughput sequencing was performed using the Illumina HiSeq 1500. miRNA-seq reads were aligned to the mouse miRNA sequences in the miRBase database (release 21) using miRDeep2. Mature miRNA values were normalized by library size (corresponding to cpm mapped miRNA reads). miRNAs with values < 5 cpm in all conditions were cut off. miRNA-seq data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE72966 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE72966).

A paired t test was used to calculate significance between percentage difference in the number of TRAP⁺ cells relative to control in at least three independent experiments. For quantitative real-time PCR (qPCR) experiments, the fold changes in response to TNF-α were pooled, and the Student t test was applied using a lognormal distribution. In the case of more than two groups of samples, one-way ANOVA followed by the Tukey post hoc test was used to calculate the significance of differences between any groups of samples. Statistical analysis was carried out using GraphPad Prism. The p values < 0.05 were taken as statistically significant. Unless indicated, all data are presented as the mean ± SD.

The mechanisms that prevent TNF-α from effectively inducing osteoclast differentiation have remained an enigma. As shown in Fig. 1A (column 1 and 3 of a heat map depicting data derived from RNA-seq GSE53218) and the literature (13), TNF-α alone has a very weak capacity for inducing the expression of osteoclastogenic regulators, such as NFATc1 and Blimp1, and osteoclast marker genes, such as calcitonin receptor, Oscar, integrin β3, TRAP, and cathepsin K. In contrast, the induction of all of these positive osteoclastic genes was dramatically enhanced, whereas the expression of negative regulator Irf8 was diminished, in RBP-J–deficient cells (Fig. 1A; compare columns 3 and 4). This is in alignment with a previous study that RBP-J deficiency enables TNF-α to effectively induce osteoclast differentiation (13). These data, together with our previous findings (13), demonstrate that RBP-J is a key inhibitory regulator of inflammatory osteoclastogenesis and bone resorption induced by TNF-α; however, the underlying molecular mechanisms are not fully understood.

We reasoned that RBP-J regulates the expression of target genes important in osteoclastogenesis and wondered whether RBP-J could regulate the expression of miRNAs that control TNF-α–induced osteoclastogenesis, a possibility that has not been previously considered. We first performed global profiling of miRNA expression using a genome-wide approach, high-throughput sequencing of microRNAs (microRNA-seq), to determine the miRNAs that are regulated by TNF-α and RBP-J in osteoclast precursors. BMMs obtained from Rbpj^ΔM/ΔM mice [Rbpj^flox/floxLysMcre(+) mice in which RBP-J was specifically deleted in myeloid macrophages] and wild-type control mice (13) were used as osteoclast precursors. Surprisingly, TNF-α significantly changed the expression of only a small number of miRNAs during osteoclast differentiation (refer to 2Materials and Methods for cutoff criteria) (Fig. 1B). By overlapping miRNA-seq data from two independent experiments, we found that, of 1915 miRNAs that were expressed in osteoclast precursors, TNF-α treatment for 24 h induced the expression of only 27 miRNAs and suppressed 12 miRNAs by ≥1.2-fold (Supplemental Fig. 1). Thus, TNF-α regulated the expression of only ∼2% of miRNAs in BMMs/osteoclast precursors, indicating a limited and selective regulatory pattern of miRNA expression by long-term exposure to TNF-α stimulation in this cell type. Furthermore, we wished to identify miRNAs that are induced by TNF-α but significantly suppressed by the negative regulator RBP-J during osteoclast differentiation. We hypothesized that this group of miRNAs suppressed by RBP-J would have important roles in promoting TNF-α–induced osteoclast differentiation. We screened for microRNAs that were induced ≥1.2-fold by TNF-α in control and RBP-J–deficient cells. From this list, we then searched for microRNAs that were expressed ≥1.2-fold greater in TNF-α–stimulated RBP-J–deficient cells than in control cells. After overlapping lists derived from two independent miRNA-seq datasets, we came up with a list of six microRNAs that fit our criteria for TNF-α–induced and RBP-J–suppressed miRNAs (Fig. 1C). In this group, miRNA-182-5p stood out as being the most highly expressed in TNF-α–stimulated RBP-J–deficient cells. We then validated our miRNA-seq data with qPCR (Fig. 1D) and found that TNF-α–induced miR-182 was enhanced significantly by RBP-J deficiency. In addition, in silico analysis of the promoter region of miR-182 revealed three potential RBP-J binding sites within 3 kb upstream of its transcription start site (Supplemental Fig. 2A), suggesting that miR-182 is likely a direct target suppressed by RBP-J. The induction pattern of miR-182 expression by TNF-α in the absence of RBP-J is similar to that of the osteoclastogenic genes (Fig. 1A) and correlates with the dramatically enhanced osteoclastogenesis in RBP-J–deficient cells (13), which led us to question whether miR-182 is a positive regulator of osteoclast differentiation. To test this hypothesis, we applied a gain-of-function approach by overexpressing an miR-182 mimic and a loss-of-function approach by inhibiting miR-182 function with a specific antagomir.

We first examined whether miR-182 contributed to the enhanced osteoclast differentiation in response to TNF-α that is caused by RBP-J deficiency. Because miR-182 expression is increased by RBP-J deficiency, we took advantage of an miR-182 antagomir to suppress miR-182 expression. We confirmed the significant downregulation of miR-182 by the antagomir (Supplemental Fig. 2B). As shown in Fig. 2A, Rbpj^ΔM/ΔM BMMs transfected with antagomir control formed large TRAP⁺ multinucleated cells (MNCs) in response to TNF-α (Fig. 2A), as expected from our previous work (13). However, inhibition of miR-182 by antagomir significantly suppressed the number and size of TRAP⁺ MNCs induced by TNF-α in RBP-J–deficient cell cultures (Fig. 2A, 2B). In parallel with the decreased number and size of TRAP⁺ MNCs, inhibition of miR-182 substantially suppressed the induction of osteoclast marker genes Acp5 (encoding TRAP), CtsK (encoding cathepsin K), and Itgb3 (encoding β3) in RBP-J–deficient cells relative to those transfected with control antagomir (Fig. 2C), indicating that increased expression of miR-182 contributes significantly to RBP-J deficiency–enhanced osteoclastogenesis. Furthermore, we investigated whether miR-182 affected the expression of regulators that control osteoclast differentiation. Our previous work showed that RBP-J inhibits induction of positive osteoclastogenic regulators and attenuates downregulation of negative regulator Irf8. In the current study, we found that inhibition of miR-182 led to a dramatic decrease in the induction of NFATc1 and Blimp1 in RBP-J–deficient cells (Fig. 2D, 2E). Downregulation of Irf8 was not consistently or significantly affected by miR-182 (data not shown), suggesting that miR-182 is mainly involved in the RBP-J–regulated NFATc1/Blimp1 axis. Inhibition of miR-182 by the antagomir did not affect cell proliferation/viability (Fig. 2F). Notably, in parallel with the suppression of osteoclast differentiation by miR-182 antagomirs, the resorptive pits generated by the RBP-J–deficient osteoclasts on dentin slices in the presence of TNF-α were drastically inhibited by miR-182 antagomirs (Fig. 2G). Collectively, these results indicate that miR-182 plays an important role in enhancing TNF-α–induced osteoclastogenesis in RBP-J–deficient cells. Our data also suggest that miR-182, as a downstream factor restrained by RBP-J, is a positive regulator of TNF-α–induced osteoclast differentiation.

To test our hypothesis that miR-182 functions as a positive regulator in TNF-α–mediated inflammatory osteoclastogenesis, we transfected wild-type BMMs with an miR-182 mimic to increase miR-182 levels (Supplemental Fig. 2C). Consistent with the literature, TNF-α stimulation alone was unable to effectively induce osteoclast differentiation in wild-type cells (Fig. 3A, 3B). Strikingly, forced expression of miR-182 increased the number and size of TRAP⁺ MNCs formed upon stimulation by TNF-α (Fig. 3A, 3B). In parallel, the expression of osteoclast marker genes Acp5, CtsK, and Itgb3 was significantly upregulated in the wild-type BMMs transfected with the miR-182 mimic in response to TNF-α relative to cells transfected with the mimic control (Fig. 3C). Furthermore, we examined whether increasing the expression of miR-182 could induce the expression of key osteoclastogenic regulators Blimp1 and NFATc1 in wild-type control cells upon stimulation with TNF-α alone. Indeed, Blimp1 and NFATc1 levels were enhanced by the miR-182 mimic (Fig. 3D, 3E), whereas cell proliferation and viability were not affected (Fig. 3F). The relative phosphorylation levels of Akt, which is known to be involved in the cell survival pathway, were not significantly changed by miR-182 mimic or antagomir in response to TNF-α (Supplemental Fig. 2D). Importantly, consistent with the enhanced osteoclast differentiation, the increased expression of miR-182 in wild-type BMMs dramatically promoted the level of osteoclastic resorption on dentin slices (Fig. 3G), indicating that the osteoclasts derived from the miR-182 mimic–transfected precursors are functional and possess a bone-resorptive capability. These results indicate that increased expression of miR-182 significantly promotes TNF-α–induced osteoclast differentiation in wild-type BMMs.

Given the positive role of miR-182 in TNF-α–mediated inflammatory osteoclastogenesis, we reasoned whether miR-182 played a role in RANKL-induced homeostatic osteoclastogenesis as well. miR-182 expression was gradually induced by TNF-α or RANKL (Supplemental Fig. 2E, 2F). Although RANKL induced miR-182 expression to an even greater level than TNF-α, neither forced expression nor inhibition of miR-182 consistently dramatically affected RANKL-induced in vitro osteoclast differentiation (Supplemental Fig. 2G). Thus, miR-182 appears to play a selective and prominent role in TNF-α–induced osteoclastogenesis. The preference of regulation of TNF-α signaling by miR-182 is similar to that by RBP-J, supporting the idea that RBP-J and miR-182 function in the same axis regulating osteoclast differentiation.

After highlighting the importance of miR-182 in the setting of inflammatory osteoclast differentiation driven by TNF-α, we set out to determine the direct targets of miR-182. We performed two parallel RNA-seq experiments, one with an miR-182 gain-of-function approach using a miR-182 mimic and the other with an miR-182 loss-of-function approach using an antagomir, to first identify genes regulated by miR-182 in osteoclast precursors. In general, miRNAs have modest effects on target mRNA expression, and our first-level screen was for genes that were upregulated ≥1.1-fold by the miR-182 antagomir and suppressed ≥1.1-fold by the miR-182 mimic. With this approach, we obtained a list of 383 genes, which contained miR-182 potential targets. Next, we took advantage of previous literature that experimentally identified miR-182 targets by pull-down assay of biotin-labeled miR-182 (44). We then overlapped these miR-182 targets (44) with the genes regulated by miR-182 in osteoclast precursors obtained from our RNA-seq experiments and found an overlap of 32 miR-182 direct targets (Fig. 4A). In this group of genes, Foxo3a (encoding Foxo3), a member of the Foxo family of transcription factors, was reported to be a direct target of miR-182 (45). Foxo proteins regulate RANKL-induced osteoclastogenesis (47, 48), and FOXO3 activity is associated with outcomes in infectious and inflammatory diseases, including RA (49, 50), which suggest its possible relevance in TNF-α–driven osteoclastogenesis. We also selected Maml1 as a gene of interest because of its role as a cotranscriptional regulator that is essential for RBP-J signaling (51), which suggests a potential function of Maml1 in osteoclast biology. The miR-182 seed region was identified in the 3′ untranslated region of the Maml1 and Foxo3a genes, and it is well conserved between mice and humans (Fig. 4B). To validate the effect of miR-182 on the protein expression of Maml1 and Foxo3, we transfected BMMs with the miR-182 mimic or antagomir and their corresponding controls. Maml1 and Foxo3 expression levels were decreased in cells transfected with the miR-182 mimic compared with control. In parallel, the expression of Maml1 and Foxo3 increased in the cells transfected with miR-182 antagomir relative to its control (Fig. 4C). These results indicate that Maml1 and Foxo3 are miR-182 targets in osteoclast precursors.

We then asked whether Maml1 and Foxo3 are functionally important in TNF-α–mediated inflammatory osteoclastogenesis. To address this question, we knocked down Maml1 or Foxo3a in BMMs (Supplemental Fig. 3). Upon TNF-α stimulation, BMMs transfected with Maml1 siRNA formed a greater number of large TRAP⁺ MNCs relative to the cells transfected with the control siRNA (Fig. 5A). Moreover, knockdown of Maml1 markedly enhanced the expression levels of osteoclast marker genes Acp5, CtsK, and Itgb3 during TNF-α–induced osteoclastogenesis (Fig. 5B). Knockdown of Foxo3a was insufficient for TNF-α alone to induce osteoclastogenesis; however, with a brief RANKL priming, Foxo3a knockdown significantly promoted TNF-α–induced osteoclast differentiation (Fig. 5C) and the expression of osteoclast marker genes Acp5, CtsK, and Itgb3 (Fig. 5D). Furthermore, knockdown of Maml1 or Foxo3a markedly enhanced resorptive pit formation induced by TNF-α (Fig. 5E, 5F). These results identified miR-182 targets Maml1 and Foxo3 as two novel negative regulators involved in TNF-α–mediated osteoclast differentiation. Our findings highlight that miR-182 functions as a key positive regulator in TNF-α–mediated inflammatory osteoclastogenesis through its inhibition of multiple targets that suppress osteoclastogenesis in the inflammatory setting.

miRNAs have emerged as important regulators of various biological processes and disease settings (23, 28, 30–32). Targeting miRNAs as a new treatment approach showed therapeutic potential in several diseases. For example, a phase II clinical trial using an anti-miRNA approach showed promising therapeutic efficacy in the treatment of hepatitis C virus infection (52), highlighting the promising potential of targeting miRNAs in diseases. Although altered miRNA expression in synovial tissue and synovial fibroblasts in rheumatoid arthritis was reported (53), the role of miRNAs in inflammatory diseases associated with bone destruction, including inflammatory osteoclastogenesis and osteolysis, is underexplored. In the current study, we applied next-generation deep sequencing to profile miRNAs (miRNA-seq) in an inflammatory setting of osteoclastogenesis driven by the proinflammatory cytokine TNF-α. This miRNA-seq data provided the first genome-wide dataset of miRNA profiling in TNF-α–induced osteoclastogenesis, to our knowledge. Our study showed that TNF-α–regulated miRNAs in osteoclast precursors are quite different from the existing few datasets (54–56). The variation could be explained by different cell and tissue types, as well as the doses of and exposure times to TNF-α. This underscores the context dependency of miRNA expression (23) and the importance of identifying miRNAs in different scenarios. We also used miRNA-seq to screen the miRNAs regulated by the key osteoclastogenic inhibitor RBP-J to fill a gap between RBP-J signaling and its regulation of miRNAs, which was unknown in the context of osteoclastogenesis. By overlapping these datasets, we identified miR-182 as a novel key osteoclastogenic miRNA that is negatively regulated by RBP-J and preferentially promotes TNF-α–induced osteoclastogenesis using loss- and gain-of–miR-182–function approaches. Our data identified an important miRNA player, miR-182, with novel osteoclastogenic function that adds a new link between TNF-α signaling and RBP-J signaling. Moreover, we demonstrated that two key targets of miR-182, Maml1 and Foxo3, are new negative regulators of inflammatory osteoclastogenesis. Overall, this newly described RBP-J–miR-182–Maml1/Foxo3 axis identified an miRNA-controlled regulatory pathway in inflammatory osteoclastogenesis, provided important mechanisms by which RBP-J inhibits inflammatory osteoclastogenesis, and uncovered a cross-talk connected by an miRNA between TNF-α signaling and RBP-J signaling (Supplemental Fig. 4).

Investigation of osteoclast differentiation is focused on understanding the signaling and function of the key osteoclastogenic cytokine RANKL, which is nonredundant under physiological conditions. Predominant lines of investigation are based upon the hypothesis that RANKL elicits novel cellular signaling events and responses that are not induced by similar cytokines, such as TNF-α. Instead, our previous work and this study introduce the alternative and not mutually exclusive concept that cytokines, such as TNF-α, are subject to brakes or feedback-inhibitory mechanisms that limit their osteoclastogenic potential. A key driver of these negative-regulatory mechanisms is the transcription factor RBP-J, and the current study shows that RBP-J applies the brakes, at least in part, by regulating transcription of the gene encoding miR-182, which, in turn, regulates expression of Foxo3 and Maml1. These findings add a new regulatory circuit to negative regulation of osteoclastogenesis that is orchestrated by RBP-J. Interestingly, the miR-182–Foxo3/Maml axis is more effective in restraining the osteoclastogenic functions of the inflammatory cytokine TNF-α relative to RANKL. This finding suggests that targeting the miR-182–Foxo3/Maml1 axis may be preferentially effective in suppressing osteoclastogenesis at local inflammatory sites while exerting minimal effects on physiological bone remodeling (similar to what we observed with RBP-J) (13). Our attempts to test this hypothesis in vivo have been limited by technical issues related to delivery of antagomirs to osteoclast precursors and difficulty in obtaining knockout mice because they harbor infections. Future work is needed to test the role of the miR-182–Foxo3/Maml1 axis in pathological inflammatory osteoclastogenesis in vivo.

Recognition of the biological importance of miR-182 is just recently emerging. miR-182 was shown to play key roles in cell differentiation, apoptosis, tumor development, and immune responses (38–46). It is differentially expressed in various cell types in given conditions and tissues/organs, including the skeletal system (Supplemental Fig. 2H) (38–46, 57). A recent publication noted that miR-182 negatively regulates osteoblast differentiation (57), which brought attention to the role of miR-182 in bone biology. Foxo1 and Foxo3 are well-defined miR-182 targets in several biological settings (42, 45). We found that Foxo3 is also a key miR-182 target in TNF-α–mediated osteoclast differentiation. Despite different functions for different Foxo family members, Foxo1, Foxo3, and Foxo4 proteins were reported to regulate RANKL-induced osteoclast differentiation (47, 48). To our knowledge, our study is the first to investigate the role of Foxo3 in TNF-α–mediated osteoclastogenesis. Interestingly, a brief RANKL-priming period is needed to reveal the role of Foxo3 in TNF-α–induced osteoclast differentiation, suggesting that other negative regulator(s) presumably play an important role in limiting the early phase of osteoclastogenesis induced by TNF-α. The other miR-182 target, Maml1, is likely an inhibitory candidate during the early phase because its deletion turned on osteoclast differentiation induced by TNF-α without a need for RANKL priming. Therefore, the two miR-182 targets (Maml1 and Foxo3) likely work coordinately to restrain inflammatory osteoclastogenesis.

In the current study, we identified two key targets of miR-182, as well as RBP-J as an important upstream inhibitory regulator of miR-182 in response to TNF-α. RBP-J has several well-conserved potential binding sites in the upstream regulatory region of the gene encoding miR-182 precursor transcript, and RBP-J deficiency significantly increases miR-182 expression, indicating that miR-182 is a downstream target of RBP-J. Furthermore, our results show that miR-182 plays a key role in RBP-J–mediated osteoclast differentiation by contributing to the regulation of NFATc1 and Blimp1. In addition to these positive osteoclastogenic regulators, we found previously that RBP-J also suppresses negative regulator Irf8 (13). In this study, miR-182 seems to regulate osteoclast differentiation independently of Irf8 (B. Zhao, unpublished observations). Therefore, our data identified a key RBP-J–regulated osteoclastogenic miRNA and provided an important mechanism for RBP-J–mediated inhibition of osteoclastogenesis. Interestingly, transcriptional coactivator Maml1 is also involved in Notch/RBP-J signaling, primarily through interaction with RBP-J to mediate its transcriptional activity (51). Downregulation of Maml1 by miR-182 may attenuate its cotranscriptional activity for RBP-J, which may contribute to RBP-J–mediated suppression of osteoclastogenesis. Thus, miR-182 seems to be involved in a reciprocal regulatory circuit with RBP-J signaling in TNF-α–induced osteoclastogenesis and forms a new regulatory network in which miR-182 is centered to receive an activating signal from TNF-α stimulation and an inhibitory signal from RBP-J, as well as to suppress two downstream targets: Foxo3 and Maml1 (Supplemental Fig. 4). Suppression of miR-182 by RBP-J serves as an important mechanism that restrains TNF-α–induced osteoclastogenesis. Taken together with the genetic evidence that RBPJ and FOXO3 are closely associated with human inflammatory diseases (20–22, 49, 50), such as RA, our findings establishing a key role for miR-182 in TNF-α–induced osteoclastogenesis and RBP-J–mediated inhibition of osteoclastogenesis support the biological importance of the newly identified RBP-J–miR-182–Maml1/Foxo3 axis and suggest its therapeutic implications for inflammatory osteoclastogenesis and bone lysis.

We thank Drs. Baosen Jia, Kazuki Inoue, and Shiaoching Gong for valuable discussions and technical support.

The authors have no financial conflicts of interest.