Conditional abrogation of Atm in osteoclasts extends osteoclast lifespan and results in reduced bone mass

Ataxia-telangiectasia mutated (ATM) kinase is a central component involved in the signal transduction of the DNA damage response (DDR) and thus plays a critical role in the maintenance of genomic integrity. Although the primary functions of ATM are associated with the DDR, emerging data suggest that ATM has many additional roles that are not directly related to the DDR, including the regulation of oxidative stress signaling, insulin sensitivity, mitochondrial homeostasis, and lymphocyte development. Patients and mice lacking ATM exhibit growth retardation and lower bone mass; however, the mechanisms underlying the skeletal defects are not fully understood. In the present study, we generated mutant mice in which ATM is specifically inactivated in osteoclasts. The mutant mice did not exhibit apparent developmental defects but showed reduced bone mass due to increased osteoclastic bone resorption. Osteoclasts lacking ATM were more resistant to apoptosis and showed a prolonged lifespan compared to the controls. Notably, the inactivation of ATM in osteoclasts resulted in enhanced NF-κB signaling and an increase in the expression of NF-κB-targeted genes. The present study reveals a novel function for ATM in regulating bone metabolism by suppressing the lifespan of osteoclasts and osteoclast-mediated bone resorption.

and increased bone resorption, the potential cell-autonomous functions of ATM in osteoblasts and osteoclasts are not yet fully understood.
Osteoclasts are multinucleated cells that are essential for bone resorption and remodeling in vivo [26][27][28] . The progenitors of osteoclasts are derived from the monocyte/macrophage cell lineage and express the receptor activator of NF-κ B (RANK). Binding of RANK with RANK ligand (RANKL), a membrane-bound ligand expressed on osteoblasts and osteocytes, triggers intracellular signaling, which ultimately leads to osteoclast differentiation. Past studies have revealed that the differentiation and activity of osteoclasts are tightly regulated by various hormones and cytokines produced by immune cells and osteoblasts. However, despite the intricate mechanisms involved in maintaining skeletal homeostasis in vivo, the activity of osteoclasts is often dysregulated, resulting in overt bone resorption and reduced bone mass, as observed in skeletal disorders such as osteoporosis, bone metastasis, and arthritic bone destruction [29][30][31] . Therefore, it is clinically relevant and crucial to learn more about the regulatory mechanisms behind the differentiation and activity of osteoclasts.
In the present study, we generated mutant mice in which Atm is specifically abrogated under the control of the cathepsin K promoter (Atm Ctsk mice) to elucidate the potential function of ATM in regulating osteoclast activity. These mice did not exhibit overt abnormalities but showed reduced bone mass due to increased bone resorption. Osteoclasts lacking ATM were more resistant to apoptosis and had an extended lifespan compared to control cells. Of note, the NF-κ B pathway, a critical regulator of osteoclast activity and longevity 32 , was significantly enhanced in osteoclasts derived from Atm Ctsk mice compared to those from control mice. These results show an unexpected role of ATM as a negative regulator of osteoclast longevity through suppressing NF-κ B signaling in vivo and may shed novel light on the biology of ATM and skeletal homeostasis.

Results
Atm Ctsk mice exhibit reduced bone volume. In a preliminary experiment, we found that ATM is activated at a higher level in mature osteoclasts than in bone marrow-derived macrophages (BMMs), indicating that ATM is involved in a later stage of osteoclastogenesis or in the maintenance of osteoclast functions (Fig. 1A). The activation of ATM in mature osteoclasts was, in part, dependent on ROS production as a treatment with a potent ROS scavenger N-acetyl cysteine suppressed the phosphorylation of ATM ( Supplementary Fig. 1A). On the other hand, DDR was not markedly induced in osteoclasts under the present experimental settings ( Supplementary Fig. 1B). These results suggest that the activation of ATM is, at least in part, dependent on ROS production but not derived from DSB in mature osteoclasts.
To elucidate the potential role of ATM in osteoclasts, we generated mutant mice in which Atm is abrogated under the control of a Ctsk (encodes the osteoclast-specific proteolytic enzyme cathepsin K) promoter. Atm Ctsk mice did not show any gross defects under unchallenged conditions (data not shown). Atm flox/+ , Atm flox/flox , and Atm flox/+ /Ctsk-Cre mice exhibited no apparent pathological phenotype and were used as control (Ctrl) animals in the present study. Western blot analysis showed that ATM was expressed in both BMMs and mature osteoclasts from Ctrl mice. The expression of ATM was also observed in Atm Ctsk mouse-derived BMMs; however, this Bone morphometric analysis of the femurs collected from 3-and 9-week-old Ctrl and Atm Ctsk mice. BV/TV, bone volume/tissue volume; Tb.Th, trabecular thickness; Tb.N, trabecular number; Tb.Sp, trabecular space. n = 7 and 6 for 3-week-old Ctrl and Atm Ctsk mice, respectively, and n = 5 mice for each genotype at 9 weeks old. expression was markedly suppressed when the cells were induced into osteoclasts, demonstrating that the Atm alleles were successfully excised specifically in osteoclasts (Fig. 1B). Although Atm Ctsk mice showed no significant difference in body size or weight (data not shown), the length of their long bones was approximately 5% shorter compared to Ctrl mice at 9 weeks of age (Fig. 1C). We next performed micro-computed tomography (μ CT) analysis and found that Atm Ctsk mice exhibited reduced bone mass compared to Ctrl mice, as highlighted by decreased BV/TV, Tb.Th, and Tb.N, and increased Tb.Sp values (Fig. 1D,E). Structural histomorphometric parameters (BV/TV, Tb.Th, Tb.N, and Tb.Sp) of the tibial sections at the secondary spongiosa showed a similar trend ( Fig. 2A,B). Notably, we observed a significant increase in the osteoclast number in Atm Ctsk mice at the edge of primary spongiosa (Fig. 2D). On the other hand, there was no significant differences in the bone formative or resorptive parameters in histomorphometric, inducing Oc.S/BS, Oc.N/BS, ES/BS, Ob.S/BS, MAR, and BFR/ BS, in the secondary spongiosa (Fig. 2B), or in the thickness of the growth plates (Fig. 2D). These observations indicate that Atm Ctsk mice exhibit reduced bone mass primarily due to an increase in osteoclastic bone resorption at the edge of primary spongiosa, whereas bone resorption at the secondary spongiosa is not overtly enhanced.
BMMs derived from Atm Ctsk mice exhibit no apparent defects in differentiation. The lower bone mass and increased osteoclast number observed in Atm Ctsk mice at the primary spongiosa suggested that osteoclastogenesis was enhanced in these mice. BMMs collected from Ctrl and Atm Ctsk mice were incubated in the presence of soluble RANKL (sRANKL) and macrophage colony-stimulating factor (M-CSF) for 3-5 d and stained for tartrate-resistant acid phosphatase (TRAcP). Contrary to our expectations, there was no difference in the number of multinucleated osteoclasts between Ctrl and Atm Ctsk mouse-derived BMMs at any time period (Fig. 3). We also found no changes in the expression levels of genes involved in osteoclastogenesis (including Nfatc1, Fos, Acp5, and Dcstamp) between Ctrl and Atm Ctsk osteoclasts (data not shown). These observations show that conditional abrogation of Atm under the control of the Ctsk gene promoter does not have any marked impact on osteoclast differentiation in vitro, as has been previously described 20,25 . Osteoclasts lacking ATM exhibit prolonged survival. Because BMMs derived from Atm Ctsk mice were not defective in osteoclast differentiation, we next asked if the lack of ATM had any impact on the lifespan of osteoclasts. Tibial sections prepared from Ctrl and Atm Ctsk mice were stained for apoptotic cells using the terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) method and also stained for TRAcP. We enumerated the TRAcP-positive cells and assessed whether these cells contained nuclei that were positive for TUNEL. As shown in Fig. 4A, we found that the ratio of TRAcP-positive multinucleated cells containing at least one TUNEL-positive nucleus was significantly higher in Atm Ctsk mice than in Ctrl mice, indicating that osteoclasts in Atm Ctsk mice are less prone to apoptosis than those in Ctrl mice. To confirm this finding, we next induced osteoclasts in vitro using BMMs prepared from Ctrl and Atm Ctsk mice and further cultured these cells under sRANKL-and M-CSF-depleted conditions for 8 h to induce apoptosis. At the end of the incubation, the cells were stained for TRAcP, and the number of surviving cells was enumerated. As shown in Fig. 4B, we found that more TRAcP-positive multinucleated cells remained on the plates with Atm Ctsk mouse-derived cells than with control cells, indicating that osteoclasts derived from Atm Ctsk BMMs were more resistant to growth factor-deprived conditions. In parallel with these findings, the amount of resorption, as evaluated with a pit formation assay, was increased in Atm Ctsk osteoclasts compared to Ctrl osteoclasts, which was statistically significant at day 10 of incubation (Fig. 4C). This observation indicates that a lack of ATM in osteoclasts does not enhance the bone resorption activity per se, but the prolonged lifespan results in an overall increase in the amount of resorption. Furthermore, the protein level of cleaved caspase-3, a marker for apoptotic cells, was decreased in Atm Ctsk cells compared to control cells under growth factor-deprived conditions, suggesting that osteoclasts lacking ATM are more resistant to apoptosis under the current experimental conditions (Fig. 4D).

NF-κB activity is enhanced in osteoclasts lacking ATM.
To further elucidate the potential mechanism behind the pro-apoptotic effect of ATM in osteoclasts, we next asked whether the lack of ATM in osteoclasts affects the activity of the NF-κ B pathway, a signaling pathway critically involved in osteoclast survival 28,32 . Western blots using nuclear extracts from osteoclasts showed an increase in the level of p65 (encoded by Rela), one of the major components of the NF-κ B complex, in Atm Ctsk osteoclasts compared to Ctrl osteoclasts (Fig. 5A). Consistent with this, immunostaining for p65 revealed nuclear accumulation of p65 in Atm Ctsk osteoclasts, which reflects increased NF-κ B activity in the absence of ATM (Fig. 5B). Furthermore, there was an increase in the transcripts for anti-apoptotic genes, including Il1a, Bcl2, and Bcl2l1 (encodes Bcl-XL), which are all regulated by the NF-κ B pathway (Fig. 5C). In parallel with these findings, we also confirmed the increased levels of phosphorylated-Iκ B and phosphorylated-p65 (Ser536) (Fig. 5D, both of which reflect the activity of the NF-κ B pathway. Of note, we found that the levels of both phosphorylated-Iκ B and phosphorylated-p65 (Ser536) decreased after sRANKL-and M-CSF-depletion, yet the expression of these molecules remained higher in Atm Ctsk osteoclasts than in Ctrl cells (Fig. 5D). In addition, as have been previously described 33,34 , inhibition of p65 activity by a proteasome inhibitor MG-132 significantly suppressed the survival of both Atm Ctsk and Ctrl osteoclasts (Fig. 5E), indicating that the survival of osteoclasts are highly dependent on the activity of the NF-κ B pathway. Taken together, these observations suggest that the abrogation of Atm in osteoclasts extends the lifespan of osteoclasts most likely by enhancing NF-κ B activity.

Discussion
Our data show that the conditional abrogation of Atm in osteoclasts leads to decreased bone mass, potentially due to prolonged osteoclast survival. We found that fewer osteoclasts in Atm Ctsk mice stained positive for TUNEL compared to those in Ctrl mice, indicating that a lack of ATM renders osteoclasts less prone to apoptosis. Consistently, osteoclasts from Atm Ctsk mice were more resistant to apoptosis under sRANKL-and M-CSF-deprived conditions than those from Ctrl mice. Furthermore, we found that nuclear accumulation of p65 was enhanced in Atm Ctsk osteoclasts, suggesting that the prolonged longevity of Atm Ctsk osteoclasts was due to increased activation of the NF-κ B pathway. Taken together, the present study reveals an unexpected role for ATM in regulating osteoclast longevity and bone volume.
The observation that the loss of ATM promotes the survival of mature osteoclasts was somewhat unexpected. Because ATM plays a protective role against genomic damage and the accumulation of intracellular ROS, we initially assumed that osteoclasts lacking ATM would be more prone to apoptosis and hence would have a shorter lifespan. In the developing central nervous system, radiation-induced cell death is largely dependent on ATM activity, and it has been shown that the loss of ATM renders neurons more resistant to apoptosis 35 . In a similar vein, our data suggest that the induction of apoptosis under physiological conditions is in part dependent on ATM activity in osteoclasts. However, a lack of ATM in osteoclasts does not have marked effects on the efficiency of bone resorption under physiological conditions in adult mice.
As a potential mechanism behind the enhanced survival of osteoclasts lacking ATM, we observed an increase in the activity of NF-κ B in Atm Ctsk osteoclasts. Because the activation of NF-κ B can be triggered by ATM 36,37 , our data may appear counterintuitive. Nonetheless, a recent study has shown that the loss of ATM can induce NF-κ B signaling, at least in certain cell types. Hathcock et al. demonstrated that systemic abrogation of ATM promotes the development of B-cell lymphoma in mice lacking T cells and that the NF-κ B-pathway was significantly enhanced in these lymphoma cells 38 . The study also showed that the survival of the ATM-deficient lymphoma cells was dependent on NF-κ B activity using pharmacological inhibitors of IKK. Although it is not fully understood how the loss of ATM enhances NF-κ B activity in lymphocytes, the study supports the idea that ATM can both positively and negatively regulate NF-κ B activity in a context-or cell-dependent manner. Given that the primary function of ATM is to induce DDR upon DNA double-strand break, it is also possible that the extended life-span was derived from the failure to induce DDR-mediated apoptosis. However, our data indicate that DDR is not noticeably induced in mature osteoclasts ( Supplementary Fig. 1B), arguing against this hypothesis. How ATM is activated in mature osteoclasts and how ATM negatively regulates the activity of NF-κ B remain to be addressed. Our data suggest that ROS production is partially responsible for the activation of ATM 39 . On the other hand, we did not find any marked DDR in mature osteoclasts, indicating that DSB is not the major contributor for ATM activation. The potential mechanisms behind the suppression of NF-κ B activity by ATM may include the inhibition of p65 activity via the phosphorylation of Ser547 by ATM 40 ; the reciprocal inhibition of NF-κ B and p53, where the abrogation of the ATM-p53 axis reversely enhances NF-κ B activity 41 ; the competitive binding of ATM and NF-κ B to protein phosphatase 2A, which exhibits inhibitory activity towards both ATM and NF-κ B 42,43 . In the present study we found the levels of the phosphorylation of p65 (Ser536) and Iκ B were increased in Atm Ctsk osteoclasts compared to that in Ctrl (Fig. 5D). This observation indicates that the NF-κ B pathway is activated through the conventional pathway which involves phosphorylation and degradation of Iκ B in Atm Ctsk osteoclasts. On the other hand, it remains to be clarified what the direct substrates of ATM in osteoclasts and whether other factors than phosphorylation of Iκ B and p65 are involved in the activation of the NF-κ B pathway in osteoclasts. Screening of the potential ATM substrates in osteoclasts using mass spectrometry may help to address these issues.
There are limitations in the present study. First, most importantly, because all of the animal experiments were performed under unchallenged conditions, it is not clear whether the results of the present study reflect the pathological conditions of skeletal diseases with enhanced bone resorption. Second, because we used mice of mixed genetic background (129Sv and C57BL/6) for morphometric analyses the data potentially have a large dispersion. Nevertheless, the data from 3 week-old mice showed a similar trend with those of 9 week-old mice, indicating that our results were consistent among different sets of animals. Of note, a recent study has shown that the Atm protein and transcript levels decline with aging in splenocytes 44 . Accordingly, our preliminary observations indicate that this is also the case for bone marrow cells and osteoclasts (data not shown). Because a loss of ATM in osteoclasts results in enhanced bone resorption, it is tempting to speculate that the decrease in ATM activity is causally related to the pathogenesis of senile osteoporosis.
In summary, the present study identifies ATM as a novel negative regulator of osteoclastic bone resorption. The loss of ATM in osteoclasts leads to enhanced NF-κ B activity and promotes their survival, leading to an increase in overall bone resorption. Our data therefore reveal an unexpected role for ATM, a critical molecule involved in the maintenance of genome integrity, in osteoclast longevity and bone resorption. Methods Mice. The Atm flox/flox mice and Ctsk-Cre transgenic mice were generated as previously described 16,45 . The mice used in the present study are of a mixed genetic background (129Sv and C57BL/6). All animal experiments were approved by the Institutional Animal Care and Use Committee of the Keio University School of Medicine (approval number, 09101) and performed in accordance with the approved guidelines. Atm flox/flox mice and Ctsk-Cre transgenic mice were kindly provided by Dr. Fred Alt (Harvard Univ.) and Dr. Shigeaki Kato (Tokyo Univ.), respectively. Histology and morphometric analyses were performed using male mice. Bone marrow cells were collected from both male and female mice.
μCT and histomorphometric analysis. Three-dimensional images of the distal femurs were reconstructed using a μ CT scanner (R_mCT2, Rigaku, Yokohama, Japan) and the 3D software TRI/3D-BON (Ratoc System Engineering, Tokyo, Japan). The ethanol-fixed tibiae were embedded in glycol methacrylate resin and sectioned into 5-μ m slices. For histomorphometric analyses, an area 1.2 mm above the growth plate (1.62-2.34 mm 2 in size) at the proximal metaphysis was evaluated. For the assessment of osteoclasts, tibial sections were stained for TRAcP and with van Gieson's solution and Alcian blue. All of the measurements were performed in a blinded manner.
Osteoclast culture. BMMs were prepared as previously described 46 and used as osteoclast precursors. The BMMs were cultured in the presence of sRANKL (50 ng/ml) and M-CSF (3.3 × 10 3 U/ml) for 5 d to induce osteoclastogenesis. TRAcP staining was performed as previously described 46 . For the osteoclast survival assay, osteoclasts induced in vitro were cultured for 8 h in the absence of sRANKL and M-CSF as previously described 47,48 . At the end of the incubation, the cells remaining on the dishes were stained for TRAcP, and the TRAcP-positive multinucleated cells (> 3 nuclei) were enumerated. To examine the inhibitory effect of MG-132 on osteoclast survival, mature osteoclasts formed in vitro were treated with MG-132 (0, 3, and 10 μ M) for 18 h. The surviving cells were fixed and stained for TRAcP.

RNA isolation and quantitative RT-PCR.
Total RNAs were extracted from cells using Sepasol reagent (Nacalai Tesque, Kyoto, Japan) and used for RT-PCR analyses. The nucleotide sequences of the oligos used in the present study are presented in Supplementary Table 1. The relative gene expression was determined by SYBR Green-based real-time PCR using a 7300/7500 Fast real-time PCR system (Applied Biosystems, Foster City, CA) and a LightCycler II (Roche, Risch-Rotkreuz, Switzerland). The expression levels were normalized to the Gapdh transcript expression levels.
Resorption pit assay. Osteoclast activity was evaluated using an Osteo Assay Surface (Corning, Corning, NY) according to the manufacturer's instructions. Stripwells were scanned using a flatbed scanner, and osteoclast-formed pits were analyzed using ImageJ software (http://imagej.nih.gov/ij/index.html).
Immunofluorescence microscopy. The cells were fixed with 4% paraformaldehyde for 15 min and permeabilized in 0.3% Triton X-100 for 10 min. The fixed cells were incubated with a primary antibody overnight at 4 °C and subsequently with a secondary antibody (AlexaFluor 488-conjugated anti-rabbit IgG or AlexaFluor 546/555-conjugated anti-mouse IgG, Invitrogen, Carlsbad, CA) for 1 h at 37 °C. The nuclei were stained with DAPI (Vectashield HardSet Mounting Medium with DAPI, Vector Laboratories, Burlingame, CA). Images were captured using a FluoView FV10i confocal microscope (Olympus, Tokyo, Japan) and the Olympus FluoView Viewer software (Ver. 3.1). The acquired images were processed with Adobe Photoshop CS6 (Adobe Systems, San Jose, CA) for contrast and black balance correction.

Evaluation of apoptotic multinucleated cells. The sections of the tibiae collected from 3-week-old
Atm Ctsk and Ctrl mice were stained for TRAcP and TUNEL. TUNEL staining was performed using an In situ Apoptosis Detection Kit (Takara Bio, Shiga, Japan) according to the manufacturer's instructions. The nuclei were counterstained with methyl green. At least 50 TRACP-positive osteoclasts in each section were evaluated for the presence of TUNEL-positive nuclei, and the ratio of TUNEL-positive multinucleated cells was calculated.
Statistics. All statistical analyses were performed using GraphPad Prism 7 (GraphPad Software, La Jolla, CA).
Two-tailed Student's t-tests for 2 samples assuming equal variances were used to calculate the p values except for those in Fig. 5E. One-way ANOVA was used to calculate the p values in Fig. 5E. Values of p < 0.05 were considered significant. The values are presented as the mean ± SD.