Elevated Response to Type I IFN Enhances RANKL-Mediated Osteoclastogenesis in Usp18-Knockout Mice

Maintenance of bone mass is mediated by the complementary processes of osteoclastic resorption and osteoblastic formation within local bone-remodeling sites (1–3). A loss of balance between osteoclastic bone resorption and osteoblastic bone formation is often the cause of bone disease; for example, tipping the balance in favor of osteoclasts results in a pathological bone resorption that leads to diseases such as autoimmune arthritis and postmenopausal osteoporosis (4–6). Osteoclasts are multinucleated cells (MNCs) generated from monocyte/macrophage precursor cells through the stepwise differentiation processes that involve the regulation of cytokines and transcription factors. Bone marrow precursors are matured to preosteoclasts by M-CSF, resulting in the expression of receptor activator of NF-κB (RANK) (7–11). The interaction of RANK on preosteoclasts with the RANK ligand (RANKL) on osteoblasts recruits the TNFR-associated factor protein family, including TNFR-associated factor 6, leading to activation of inhibitor of NF-κB kinase and JNK (12–14). Consequently, transcription factors, such as NF-κB and AP-1, are activated (15–17), in addition to a key transcription factor NFATc1, which is for the expression of the multiple genes responsible for osteoclastic functioning (16, 18, 19). In addition, the RANK–RANKL interaction triggers cell–cell fusion and osteoclast differentiation. The multinucleated osteoclasts attach to the bone surface using α_vβ₃ integrin, and the activated osteoclasts secrete protons and lytic enzymes, such as cathepsin K and tartrate-resistant acid phosphatase (TRAP), to degrade the organic bone matrix (12).

Although the RANK–RANKL interaction is crucial for osteoclastogenesis, the interaction induces another signal that negatively regulates osteoclast activation. It was shown that IFN-β^−/− and IFNAR1^−/− mice exhibited osteopenia with enhanced osteoclastogenesis, indicating a negative effect of IFN signaling on osteoclast differentiation (20). The induction of c-Fos upon the RANK–RANKL interaction mediates the expression of the IFN-β gene in osteoclast precursor cells; interestingly, IFN-β, in turn, inhibits c-Fos, forming a type of autoregulatory circuit (20).

IFN-stimulated gene (ISG)15 is an ubiquitin-like protein whose expression and conjugation to target proteins (ISGylation) increase greatly upon stimulation by type I IFN (21). Previous studies showed that IFN-stimulated ISGylation exhibits antiviral activities against several viruses, including influenza A and B, lymphocytic choriomeningitis virus, and Sindbis virus (22, 23). USP18 (also known as UBP43) was originally identified as a deconjugating enzyme for ISGylation (24) and functions as a negative regulator of type I IFN signaling, independent of its enzyme activity (25). In the absence of USP18, cells exhibit an enhanced and prolonged STAT1 phosphorylation and an elevated expression of ISGs in response to IFN-α/β (25). Although hypersensitivity to type I IFN was consistent in all genetic backgrounds, phenotypes of Usp18-knockout mice were influenced by genetic background; mice with mixed genetic background of C57/B6 and 129 exhibited a decreased lifespan with growth retardation primarily due to brain abnormalities (26). Although Usp18-knockout mice were embryonic lethal at embryonic day 15.5 on the C57/B6 background, a majority of homozygous knockout mice on the FVB background exhibited normal growth and lifespan (21). Mice lacking Usp18 are more resistant to viral and bacterial infections as a result of the elevated response to type I IFN (27, 28), and bone marrow cells from Usp18-deficient mice are sensitive to type I IFN–induced apoptosis (29). Analysis of the underlying mechanism revealed that the negative regulation of type I IFN signaling is independent of its deISGylating activity (30); instead, USP18 interacts directly with IFNAR2, a specific subunit of the type I IFNR, and excludes the association of JAK with IFNAR2. Because USP18 is one of the ISGs (25), it is believed that USP18 is a negative-feedback inhibitor specific to type I IFN signaling. The knock-in mouse model expressing the enzymatically inactive Usp18 further validated its function as a negative regulator of type I IFN, independent of its activity (31). Interestingly, these mice were not hypersensitive to type I IFN; however, they showed increased resistance against vaccinia virus and influenza B virus infections with increased cellular ISGylation level. Other negative regulators of JAK/STAT signaling are protein inhibitor of activated STAT and suppressor of cytokine signaling (SOCS). Intriguingly, the downregulation of protein inhibitor of activated STAT 3 significantly enhanced RANK-mediated osteoclastogenesis in RAW264.7 cells (32). The deletion of SOCS3 in mice (Socs3^−/Δvav) leads to exacerbated bone destruction, reduced basal trabecular bone volume, and an enhanced number of osteoclasts (33).

In the course of preparing bone marrow cells, we found that femurs from Usp18-deficient mice are more fragile than those from wild-type littermates; a morphological analysis of Usp18-deficient femurs showed osteopenia phonotypes. Based on this finding, we examined the effect of a USP18 deficiency on RANKL-mediated osteoclastogenesis.

FVB.129-Usp18^tm1Dzh mice were described previously (28). A majority of Usp18-knockout mice on FVB and 129 mixed background showed normal growth, and ∼10% of knockout mice showed minor growth retardation. We used age- and size-matched mice for the analyses. All mouse experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee at Sookmyung Women’s University. Mice were maintained in a specific pathogen–free environment and used at 7 wk old. Bone marrow cells from the femurs of Usp18^+/+ and Usp18^−/− mice were harvested and cultured in α-MEM (Welgene) supplemented with 10% FBS and 30 ng/ml M-CSF (PeproTech) for 1 d. Suspended cells were harvested and further cultured with M-CSF (30 ng/ml) in α-MEM containing 10% FBS. After 3 d, differentiated bone marrow–derived macrophages (BMMs) were treated with 200 ng/ml recombinant soluble RANKL (sRANKL; PeproTech) and 30 ng/ml M-CSF for osteoclast differentiation.

The following Abs were used in this study: caspase 3, cleaved caspase 3, phospho-ERK, ERK, phospho-JNK, JNK, phospho-p38, p38, phospho-Stat1, Stat1, and β-actin (all from Cell Signaling); c-Fos and NFATc1 (Santa Cruz Biotechnology); and α-tubulin (Sigma-Aldrich). Ab against USP18 was described previously (30).

Femurs isolated from 7-wk-old littermates of Usp18^+/+ and Usp18^−/− mice were fixed in 4% paraformaldehyde at 4°C and subjected to microcomputed tomography (microCT) analysis. Scanned femurs were analyzed using a SkyScan 1172 scanner (SkyScan). One millimeter–thick images obtained from distal femoral area, starting from 1 mm below the growth plate, were analyzed using the CT-Analyser program (SkyScan) to estimate bone volume and bone parameters. A three-dimensional (3D) model of images was constructed using CTVol software (SkyScan).

Femurs were isolated from 7-wk-old littermates of Usp18^+/+ and Usp18^−/− mice and fixed in 4% paraformaldehyde, followed by decalcification in 14% EDTA for 2 wk. After decalcification, femurs were processed for paraffin embedding and sectioned to 5 μm thickness. For H&E staining, rehydrated slides were stained with hematoxylin followed by eosin (Sigma-Aldrich). For TRAP staining, rehydrated slides were stained with TRAP staining solution (1 mg Fast Red, 0.1 M sodium acetate, 50 mM sodium tartrate, and 10 μg Naphthol AS-MX in distilled water) for 30 min at 37°C. TRAP-stained slides were stained with hematoxylin (Sigma-Aldrich). The stained slides were observed using an IX71 microscope (Olympus).

Total RNA was isolated from 1 × 10⁶ cells using TRIzol reagent (Life Technologies), and 2.5 μg the total RNA was used to synthesize cDNA. A reverse-transcription reaction was performed using a First-Strand cDNA Synthesis Kit (Fermentas), according to the manufacturer’s instructions. Sequences of the primers used for PCR are listed in Supplemental Table I.

PCR was performed under the following conditions: after an initial activation at 95°C for 5 min, 30 cycles of denaturation at 95°C for 15 s, annealing at 58°C for 30 s, and extension at 72°C for 30 s were done. PCR products were electrophoresed on 1% agarose gel, followed by ethidium bromide staining. For quantitative real-time RT-PCR, PCR amplification was performed with 2× Power SYBR Green PCR Master Mix with 1 pmol primers, 125 ng cDNA, and nuclease-free water, according to the manufacturer’s protocol (Applied Biosystems). PCR conditions included an initial activation at 95°C for 15 min, denaturation at 95°C for 15 s, annealing at 60°C for 30 s, and extension at 72°C for 30 s. PCR reaction was repeated for 40 cycles. For relative quantification, the expressions of target genes were normalized to the expression of β-actin, cyclophilin, or GAPDH.

Bone marrow cells from the femurs of 6-wk-old ICR mice were differentiated to BMMs by treatment with M-CSF (30 ng/ml) for 3 d. BMMs were transfected with 30 nM control small interfering RNA (siRNA) or Usp18-specific siRNA using HiPerFect reagent (QIAGEN). One day later, cells were treated with sRANKL (200 ng/ml) and M-CSF (30 ng/ml) and cultured for 5 d until used for analyses. Control siRNA and Usp18-specific siRNA were purchased from Santa Cruz Biotechnology (sc-37007 for control and sc-60866 for Usp18). For the stable knockdown of Usp18, RAW 264.7 cells were transfected with a plasmid construct containing control short hairpin RNA (shRNA) or Usp18-specific shRNA (34). Single colonies were isolated by culturing cells in a 96-well plate and tested for the knockdown of Usp18.

Bone marrow cells from the femurs of Usp18^+/+ and Usp18^−/− mice were differentiated to BMMs, as described above. BMMs were then seeded and cultured on dentine discs (Immunodiagnostic Systems) for 9 d in the presence of 30 ng/ml M-CSF and 200 ng/ml RANKL. Cells were removed by sonication, and resorption pits in the dentine discs were analyzed using confocal microscopy and ImageJ software.

Sera were isolated from the blood of 7-wk-old littermates of Usp18^+/+ and Usp18^−/− mice. Cytokines in the serum samples were measured with a Mouse Magnetic Luminex Screening Assay (R&D Systems).

Statistical differences between the test and control samples were determined nonparametrically by the Mann–Whitney U test using GraphPad Prism 5 software (GraphPad). Data are presented as mean ± SD.

Because we observed that the bones of Usp18-deficient mice were more fragile than those of the wild-type littermates, we suspected that the Usp18 deficiency caused the osteopenia phenotype. To examine this, we compared the femurs from Usp18^+/+ and Usp18^−/− male littermates using microCT analysis. Usp18-deficient femurs showed a significantly reduced trabecular bone mass compared with the femurs of wild-type littermates (Fig. 1A). In addition, several bone parameters, such as bone volume/total volume, trabecular thickness, and the numbers of trabeculae were decreased in Usp18-deficient femurs compared with wild-type femurs (Fig. 1B). The distances between trabeculae tended to be greater in the femurs of Usp18^−/− mice compared with those of wild-type mice (Fig. 1B). The 3D models of femur images generated from the microCT images clearly revealed a reduced trabecular bone mass in Usp18-deficient femurs (Fig. 1C). The femurs of female Usp18-deficient mice also showed a significantly reduced trabecular bone mass (Supplemental Fig. 1A). Consistently, microCT analyses revealed reduced cortical bone mass, shown by representative images, and reduced cortical volume and thickness in Usp18-deficient femurs compared with wild-type femurs from male and female mice (Fig. 1D, 1E, Supplemental Fig. 1B). Therefore, it is clear that Usp18 deficiency caused osteopenia in male and female mice.

To examine whether the osteopenia phenotype of Usp18-deficient bones was the result of increased bone resorption, we analyzed femurs from Usp18^+/+ and Usp18^−/− mice using TRAP-activity staining to measure the extent of osteoclast activation. The intensity of TRAP staining was increased in femurs of Usp18^−/− mice compared with those of Usp18^+/+ mice (Fig. 1D). When TRAP/hematoxylin-stained sections were analyzed using OsteoMeasure histomorphometry software (OsteoMetrics), osteoclast activation was clearly enhanced in Usp18-deficient mice compared with wild-type controls. Several parameters reflecting osteoclast activation, such as eroded surface per bone surface, osteoclast surface per bone surface, number of osteoclasts per bone perimeter, and number of osteoclasts per total area, were increased ≥2-fold in Usp18-deficient bones (Fig. 1E). Taken together, these results indicate that a Usp18 deficiency in mice caused osteopenia with an increase in osteoclastic bone resorption.

Because Usp18-deficient bones exhibited osteopenia, we questioned whether Usp18 is directly involved in osteoclastogenesis. We first evaluated the expression of Usp18 over the course of osteoclastogenesis. Bone marrow cells from Usp18^+/+ and Usp18^−/− mice were treated with M-CSF to generate macrophages. BMMs were treated with RANKL, and the expression of Usp18 mRNA and protein were measured over time. The basal expression of Usp18 was almost undetectable without RANKL treatment; however, mRNA and protein were induced 24 h after treatment with RANKL in wild-type BMMs (Fig. 2A, 2B), indicating that Usp18 is indeed expressed during RANKL-mediated osteoclastogenesis. We then examined RANKL-mediated osteoclast differentiation using the same in vitro system, whereby the differentiation was visualized by staining for TRAP activity and F-actin ring (Fig. 2C, 2D). When TRAP⁺ MNCs containing three or more nuclei were counted, 5–6-fold more TRAP⁺ MNCs were detected from osteoclasts differentiated from Usp18-deficient BMMs compared with those from wild-type controls (Fig. 2D); furthermore, we obtained consistent results with BMMs from female mice in the in vitro osteoclastogenesis assay. The numbers of differentiated osteoclasts (TRAP⁺ MNCs) were greatly increased in BMMs that had been prepared from Usp18^−/− female mice (Supplemental Fig. 2A). In our previous studies (27), as well as in the current experiment, we did not notice any difference in cell proliferation between BMMs derived from wild-type and Usp18-deficient mice (data not shown). We also checked for possible differences in cell death rate by measuring the conversion of caspase 3 to its shorter activated form. When BMMs were treated with sRANKL, Usp18-deficient BMMs exhibited slightly higher levels of caspase 3 and the activated form of cleaved caspase 3 compared with wild-type BMMs (Supplemental Fig. 2B). Thus, the osteopenia phenotype observed in Usp18-knockout mice is not due to increased osteoclast cell number as a result of altered proliferation or lifespan of Usp18-deficient BMMs. Taken together, these data indicate that an increased tendency of the bone marrow cells lacking Usp18 to be differentiated to osteoclasts by RANKL, rather than an increase in osteoclast cell number, might be closely linked to the osteopenia phenotype of the mice.

To examine whether a Usp18 deficiency enhances osteoclast differentiation via RANKL signaling, we analyzed the RANKL-induced activation of signaling cascades and gene expression in osteoclast precursors derived from the bone marrow of Usp18-deficient and wild-type mice by treating bone marrow cells with M-CSF for 3 d prior to the RANKL treatment. The administration of RANKL to BMMs induced a rapid phosphorylation of ERK, p38, and JNK in wild-type and Usp18-deficient BMMs; furthermore, the patterns and intensities of the phosphorylation for all of the kinases were similar in both genotypes (Fig. 3A). The protein levels of c-Fos, which were increased at early time points (3–6 h) after RANKL treatment, showed no significant difference between wild-type and Usp18-deficient BMMs. However, the levels of NFATc1, which are increased later (24–48 h) after RANKL administration, were significantly elevated in Usp18-deficient BMMs (Fig. 3B). In addition, the expression of marker genes for osteoclast differentiation, such as cathepsin K and TRAP, was clearly increased in Usp18-deficient BMMs compared with BMMs from wild-type littermates (Fig. 3C). Interestingly, the expression of iNOS, a well-known target gene of type I IFN, was dramatically increased in Usp18-deficient BMMs (Fig. 3B), indicating a hyperresponsiveness to type I IFN (most probably IFN-β) that is generated during RANKL-mediated osteoclast differentiation. The elevated expression of NFATc1, cathepsin K, and TRAP was also clear in separate experiments in which BMMs from female wild-type and Usp18-deficient mice were treated with RANKL (Supplemental Fig. 2C). Taken together, these results suggest that a Usp18 deficiency caused enhanced activation of RANKL signaling at late time points in osteoclast precursors, resulting in increased differentiation of Usp18-deficient BMMs to TRAP⁺ MNCs.

Although we obtained evidence that elevated RANKL-mediated signaling and transcription might cause enhanced osteoclast differentiation of osteoclast precursor cells from Usp18-deficient mice, we wanted to confirm whether this phenomenon is consistent in other systems. We adapted two independent systems to deplete Usp18 expression, by the transient knockdown of Usp18 in BMMs derived from femurs of ICR mice and by stable knockdown of Usp18 in RAW264.7 mouse macrophage cells, and measured RANKL-mediated osteoclast differentiation and expression of RANKL target genes. When we transfected control or Usp18 siRNA into BMMs from ICR mice and induced RANKL-mediated osteoclastogenesis, we detected a ≥2-fold increase in TRAP⁺ MNCs from the osteoclasts that were differentiated from Usp18-knockdown BMMs compared with the control (Fig. 4A); accordingly, mRNA expression of NFATc1, TRAP, and cathepsin K was increased in osteoclasts from Usp18-knockdown BMMs (Fig. 4B). We achieved consistent results with the stable knockdown of Usp18 in RAW264.7 cells. We generated RAW264.7 cells in which control shRNA or Usp18 shRNA was stably expressed and validated the efficient removal of Usp18 mRNA (Supplemental Fig. 3); these cells were then used for a RANKL-mediated osteoclastogenesis assay. Compared with control cells, Usp18-depleted RAW264.7 cells showed a greatly increased number of differentiated TRAP⁺ MNCs (Fig. 4C). Staining against F-actin ring showed a similar pattern to TRAP activity staining. mRNAs for TRAP, cathepsin K, and NFATc1 were also increased during RANKL-mediated osteoclast differentiation of Usp18-knockdown RAW264.7 cells (Fig. 4D). Together with the data from Usp18-knockout mice and cells, these results strongly suggest a causal relationship between Usp18 deficiency and increased RANKL-mediated osteoclastogenesis.

Because we detected increased osteoclastogenesis of bone marrow cells in the absence of Usp18, we questioned whether this mechanism is linked to the reduced bone mass in Usp18-deficient mice, which might be caused by osteoclastic bone resorption. To evaluate the outcome of increased osteoclastogenesis, BMMs from wild-type and Usp18-deficient mice were cultured on dentine discs in the presence of M-CSF and RANKL for 9 d, and resorption pits were analyzed. Consistent with the increased osteoclastogenesis, dentine discs with Usp18-deficient BMMs showed greatly increased resorbed areas (Fig. 5A). Image analyses revealed ∼2- and 1.8-fold increases in resorption depth and resorption area, respectively, from the dentine discs cultured with Usp18-deficient BMMs compared with those cultured with wild-type BMMs (Fig. 5B). These results indicated that the osteopenia phenotype of Usp18-deficient bones resulted from increased osteoclastogenesis and osteoclastic bone resorption.

Because Usp18 is known to negatively regulate type I IFN signaling, and Usp18-deficient cells are hypersensitive to IFN-α/β (25, 30), we questioned whether increased RANKL-mediated osteoclast differentiation in Usp18-deficient BMMs might be related to hypersensitivity to type I IFN. We examined RANKL-induced activation of JAK/STAT signaling in osteoclast precursors from Usp18-deficient mice and wild-type mice. When we analyzed the tyrosine phosphorylation of Stat1 as an indicator of the activation of JAK/STAT signaling during RANKL-mediated osteoclastogenesis, we detected a dramatic increase in phosphorylated Stat1 in Usp18-deficient BMMs, whereas it was very weak in wild-type BMMs (Fig. 6A), indicating a hyperactivation of JAK/STAT signaling in Usp18-deficient BMMs. Next, we measured the expression of several cytokine genes (IP-10, IL-6, IL-15, TNF-α, and IFN-β) that are known to be associated with osteoclastogenesis (35). Among them, it was shown that the expression of IP-10, IL-6 and IL-10 was highly elevated in Usp18-deficient BMMs upon IFN-β treatment (36). Upon RANKL administration, the expression of IP-10, IL-6, and IL-15 was increased in Usp18-deficient BMMs, and the expression levels were significantly higher in Usp18-deficient BMMs compared with wild-type BMMs (Fig. 6B). TNF-α was not inducible by RANKL; however, mRNA levels were significantly higher in Usp18-deficient BMMs compared with wild-type BMMs. Interestingly, the expression of IFN-β was also elevated in Usp18-deficient BMMs upon RANKL treatment compared with wild-type BMMs. Thus, the elevated and prolonged activation of Stat1 in Usp18-deficient BMMs upon RANKL treatment (Fig. 6A) might be a combinatorial effect from the intrinsic hypersensitivity of Usp18-deficient cells and the increased autocrine/paracrine secretion of IFN-β. These findings suggest that several cytokine genes were considerably overexpressed during RANKL-mediated osteoclastogenesis in Usp18-deficient BMMs and that an abnormal increase in cytokines might be responsible for triggering enhanced osteoclast differentiation.

Finally, we measured the amounts of five cytokines (TNF-α, IL-1β, IP-10, IL-6, and M-CSF) in the blood sera collected from age-matched wild-type and Usp18-deficient mice. TNF-α and IL-1β were not detectable in the mice of either genotype. The median serum levels of IL-6 and M-CSF in Usp18-deficient mice were ∼1.6- and 1.8-fold higher, respectively, than those of wild-type controls (Fig. 6C). Of special significance, IP-10 levels of Usp18-deficient mice were 7-fold higher than those of wild-type controls (Fig. 6C). Based on the sum of these results, elevated cytokine levels in the serum might be an additional cause of the osteopenia phenotype of Usp18-knockout mice.

Osteoporosis is a pathological loss of bone matrix resulting from a predominance of osteoclastic bone resorption over osteoblastic bone formation (1). We observed a severe osteopenia phenotype in male and female Usp18-knockout mice and investigated its possible cause. We identified the following possible sources. Serum levels of inflammatory cytokines, such as IP-10, IL-6, and M-CSF, were relatively higher in Usp18-deficient mice than in wild-type mice. Especially, median IP-10 levels were 7-fold higher in the serum of Usp18-deficient mice. Most importantly, RANKL-induced osteoclast differentiation (MNC formation) was markedly increased in Usp18-deficient osteoclast precursor cells. In addition, bone-resorption activity was clearly higher in Usp18-deficient BMMs compared with wild-type BMMs. The combinatorial effects of these causes could generate a severe osteopenia phenotype in Usp18-knockout mice. Although we did not detect a critical difference in the proliferation rate of wild-type and Usp18-deficient BMMs in vitro, we cannot completely rule out the possibility that an increase in BMM number might contribute, at least in part, to the osteopenia phenotype in vivo because Usp18-deficient mice showed a tendency toward increased levels of M-CSF compared with wild-type controls.

The most interesting observation from this study was enhanced RANKL-mediated osteoclastogenesis in Usp18-knockout BMMs. In our previous studies, Usp18 was identified as a negative regulator of type I IFN signaling (25). The representative consequence of Usp18 deficiency was an elevated and prolonged activation of JAK/STAT signaling upon IFN-α/β administration, which resulted in a tremendous overexpression of ISGs in many of the cell types tested, including BMMs (25). When BMMs were treated with the same dose of type I IFN, Usp18-deficient BMMs showed prolonged and hyperphosphorylation of Stat1 (27). In accordance with previous studies, microarray analyses of BMMs treated with IFN-β in vitro exhibited tremendously elevated expression of IFN-responsive genes in Usp18-deficient BMMs compared with the wild-type control (36). In bone biology, IFN-β is known to act as an inhibitor of RANKL-induced osteoclast differentiation by interfering with the expression of c-Fos (14). During osteoclast differentiation, RANKL mediates c-Fos expression in preosteoclasts; c-Fos then induces IFN-β and IFN-β, which in turn, reduces the expression of c-Fos, forming a type of autoregulatory loop to restrict osteoclast overactivation (14). In our study, basal Usp18 levels were very low in cultured BMMs; however, RANKL increased mRNA and protein levels of Usp18 expression at 24–48-h after treatment, most probably via IFN-β signaling. Based on the negative effect of IFN-β on RANKL-mediated osteoclast differentiation (14) and the enhancement of IFN-β signaling without Usp18 in BMMs (25), one could expect that Usp18 deficiency may strongly inhibit RANKL-mediated osteoclastogenesis of BMMs due to the increased negative effect of IFN-β. On the contrary, however, Usp18 deficiency increased RANKL-mediated osteoclast differentiation in the following three independent experimental settings: BMMs from Usp18-deficient mice, BMMs from ICR mice with siRNA-mediated knockdown of Usp18, and RAW264.7 cells with a stable shRNA-mediated knockdown of Usp18. Obviously, RANKL signaling induced a tremendous hyperactivation of JAK/STAT signaling and the overexpression of cytokine mRNAs in Usp18-deficient BMMs, both of which are known to promote osteoclast differentiation (35). Although IFN-β limits the excessive activation of osteoclast differentiation upon RANKL signaling in normal situations, exaggeration of IFN-β signaling as a result of Usp18 deficiency tips the balance in the opposite direction via upregulation of inflammatory cytokines.

IP-10 is expressed in human osteoclasts, is significantly increased during osteoclast differentiation (37), and was identified as a bone-erosive factor in experimental arthritis (38). RANKL promotes IP-10 expression in preosteoclasts, and IP-10 mediates RANKL expression in CD4⁺ T cells in the synovium. It was proposed that this positive cross-talk between IP-10 and RANKL, or other cytokines, such as TNF-α, is responsible for inflammation and bone erosion, whereby CD4⁺ T cells and macrophages are recruited into the synovium (39). In human circulating monocytes serving as early progenitors of osteoclasts, STAT1 and IP-10 were identified as candidate genes whose expression is responsible for the increase in the differentiation of peak bone mass at the monocyte stage (39). In addition to increased expression during RANKL-mediated osteoclastogenesis, serum levels of IP-10 were abnormally high (over 7-fold) in Usp18-knockout mice compared with normal littermates. Serum levels of M-CSF and IL-6 were also elevated in Usp18-knockout mice, although the increase was less significant (<2-fold). Therefore, elevated levels of serum cytokines, especially IP-10, may also contribute to the osteopenia phenotype of Usp18-knockout mice.

In summary, our analysis of the osteopenia phenotype of Usp18-knockout mice revealed some unexpected results. Although IFN-β plays a negative-feedback regulatory role in RANKL-induced osteoclast differentiation, the results show that an uncontrolled activation of IFN signaling in Usp18-knockout mice accelerated osteoclast differentiation with a high induction of osteoclastogenic cytokines.

The authors have no financial conflicts of interest.