Stage-specific functions of leukemia/lymphoma-related factor (LRF) in the transcriptional control of osteoclast development

Abstract

Cell fate determination is tightly regulated by transcriptional activators and repressors. Leukemia/lymphoma-related factor (LRF; encoded by Zbtb7a), known as a POK (POZ/BTB and Krüppel) family transcriptional repressor, is induced during the development of bone-resorbing osteoclasts, but the physiological significance of LRF in bone metabolism and the molecular mechanisms underlying the transcriptional regulation of osteoclastogenesis by LRF have not been elucidated. Here we show that LRF negatively regulates osteoclast differentiation by repressing nuclear factor of activated T cells c1 (NFATc1) induction in the early phase of osteoclast development, while positively regulating osteoclast-specific genes by functioning as a coactivator of NFATc1 in the bone resorption phase. The stage-specific distinct functions of LRF were demonstrated in two lines of conditional knockout mice in which LRF was deleted in the early or late phase of osteoclast development. Thus, this study shows that LRF plays stage-specific distinct roles in osteoclast differentiation, exemplifying the delicate transcriptional regulation at work in lineage commitment.

Osteoclasts are responsible for both physiological and pathological bone resorption, and an accurate understanding of the molecular mechanisms of osteoclast differentiation is thus crucially important for the development of therapeutic strategies against bone and joint diseases (1–3). Receptor activator of nuclear factor-κB (NF-κB) ligand (RANKL) is an essential cytokine that induces the differentiation of monocyte/macrophage lineage cells into osteoclasts in the presence of macrophage colony-stimulating factor (M-CSF) (2, 3). RANKL promotes osteoclastogenesis through the induction and autoamplification of the key transcription factor nuclear factor of activated T cells c1 (NFATc1), which transcriptionally regulates most of the osteoclast-specific genes required for the bone resorbing activity, including Ctsk (encoding cathepsin K), Mmp-9, and Clcn7 (encoding chloride channel 7) (3–6). The autoamplification of NFATc1 is a hallmark event in the early phase of osteoclast development, in which NFATc1 is preferentially recruited to its own promoter, thus enabling the autoamplification of expression (3, 7, 8). To achieve efficient NFATc1 autoamplification, transcription factors such as NF-κB and c-Fos are required. Osteoclast differentiation is negatively regulated by the transcription factors IFN regulatory factor-8 (IRF-8), v-maf musculoaponeurotic fibrosarcoma oncogene family protein B (MafB), and B-cell lymphoma 6 (Bcl-6), mainly through the inhibition of NFATc1 activity and expression (9–11). Thus, NFATc1 expression is controlled by a delicate balance between positive and negative transcriptional regulators during osteoclastogenesis.

Leukemia/lymphoma-related factor (LRF, also called Pokemon: POK erythroid myeloid ontogenic factor), which is encoded by the Zbtb7a gene, is a member of the POK (POZ/BTB and Krüppel) family of transcriptional repressors (12, 13). LRF is involved in the oncogenesis of T- and B-cell lymphoma, prostate, breast, non–small-cell lung, and ovarian cancers (14–18). LRF exerts its oncogenic effect by suppressing the expression of the tumor suppressor genes Arf and Rb (14, 19). LRF is implicated not only in oncogenesis, but also in diverse biological processes such as cell survival and lineage fate decisions in hematopoietic cells (20–22).

In the skeletal system, osteoclast-derived zinc finger (OCZF), a rat homolog of LRF, was originally identified as an osteoclast-specific protein in a screening performed with monoclonal antibodies (23). Recently, mice overexpressing LRF in osteoclasts were shown to exhibit an osteoporotic phenotype due to the increased number of osteoclasts (24). However, the physiological function of LRF in bone remodeling has not been demonstrated, because global deletion of LRF results in embryonic lethality (14). Thus, we investigated the function of LRF in osteoclastogenesis by disrupting Zbtb7a at the early and late stages of osteoclast differentiation using Mx1- and Ctsk-Cre mice, respectively. The distinct phenotypes of the two conditional knockout mice revealed that LRF plays certain stage-specific roles in the transcriptional program of osteoclast development.

Results

Physiological and Ectopic Expression of LRF During Osteoclastogenesis.

We examined the expression and localization of the LRF protein during osteoclastogenesis. LRF was only slightly expressed in osteoclast precursor cells, but was markedly induced in bone marrow-derived monocyte/macrophage precursor cells (BMMs) stimulated with RANKL (Fig. S1A). LRF accumulated in the nuclei as BMMs underwent differentiation into osteoclasts (Fig. S1B), suggesting that it has a role in gene regulation. To examine the effect of the ectopic expression of LRF at the early and late stages of osteoclast differentiation, we infected BMMs at distinct time points with a retroviral vector carrying the Zbtb7a gene (pMX-LRF-IRES-EGFP). When BMMs were infected with the LRF-expressing retrovirus, the formation of tartrate-resistant acid phosphatase (TRAP)-positive multinucleated cells (MNCs) was significantly impaired in the EGFP⁺ cells (Fig. 1A). However, this suppressive effect was not observed when the cells had been stimulated with RANKL for 2 d before the retroviral infection and the survival of the mature osteoclasts was not influenced by this ectopic expression of LRF in the late stage of osteoclast development (Fig. 1B and Fig. S2). These results suggest that LRF negatively regulates osteoclast differentiation at the early but not the late stage of osteoclastogenesis. It has been reported (24), however, that overexpression of LRF under the Ctsk promoter results in a prolonged survival of osteoclasts. These inconsistent in vitro results suggest that in vivo loss-of-function studies will be required for a clear understanding of the physiological function of LRF.

Fig. 1.

Effect of ectopic expression of LRF on osteoclastogenesis and generation of two types of stage-specific conditional knockout mice. (A) Effect of retroviral expression of LRF at an early stage of osteoclastogenesis. TRAP⁺ multinucleated cells were counted among the EGFP⁺ (infected) cells. (B) Effect of retroviral expression of LRF at a late stage of osteoclastogenesis. (C) Expression of LRF at the protein (Left) and mRNA level (Right) in Zbtb7a^Flox/Flox and Zbtb7a^{Flox/FloxMx1cre+} cells during osteoclastogenesis. (D) Expression of LRF at the protein (Left) and mRNA level (Right) in Zbtb7a^Flox/Flox and Zbtb7a^Flox/−Ctsk^Cre/+ cells during osteoclastogenesis. **P < 0.01.

Differentiation Stage-Specific Disruption of LRF.

To explore the physiological role of LRF in the osteoclast lineage, we investigated LRF conditional knockout mice. Because LRF exerted stage-specific effects in vitro, we generated two types of LRF conditional knockout mice, Zbtb7a^{Flox/FloxMx1cre+} (20) and Zbtb7a^Flox/−Ctsk^Cre/+ mice, by crossing Zbtb7a^Flox/Flox with Mx1-Cre transgenic mice and Zbtb7a^Flox/− with Ctsk-Cre knock-in (Ctsk^Cre/+) mice, respectively. In the Zbtb7a^{Flox/FloxMx1cre+} mice, the Zbtb7a gene is deleted upon polyinosinic-polycytidylic acid (poly I:C) treatment in various cell types, including immature hematopoietic cells, which allowed us to examine the effect of LRF depletion at the very early stage of osteoclast development. In fact, the expression of both the LRF protein and mRNA was undetectable at the stage of osteoclast precursor cells (time 0) in the Zbtb7a^{Flox/FloxMx1cre+} cells (Fig. 1C). In the Zbtb7a^Flox/−Ctsk^Cre/+ mice, the Zbtb7a gene was deleted at the later stage of osteoclast lineage cells expressing cathepsin K. We found the LRF expression level in Zbtb7a^Flox/−Ctsk^Cre/+ cells to be markedly decreased at 48 and 72 h after RANKL stimulation (Fig. 1D).

Increased Osteoclast Number and Low Bone Mass in Zbtb7a^{Flox/FloxMx1cre+} Mice.

We analyzed the bone phenotype of Zbtb7a^{Flox/FloxMx1cre+} mice, which had received poly I:C injection at the age of 21 d. The bone volume and the trabecular number were significantly reduced and trabecular separation was increased in the Zbtb7a^{Flox/FloxMx1cre+} mice (Fig. 2A and Fig. S3A), without any changes in cortical thickness and bone mineral density (Fig. S4). Bone morphometric analysis indicated an increase in the osteoclast number and eroded surface (Fig. 2 B and C), but the parameters for osteoblastic bone formation were normal in the Zbtb7a^{Flox/FloxMx1cre+} mice (Fig. S3 B and C). Wild-type mice engrafted with bone marrow cells derived from the Zbtb7a^{Flox/FloxMx1cre+} mice also exhibited a low bone mass phenotype (Fig. S5). These results indicate that the low bone mass phenotype in the Zbtb7a^{Flox/FloxMx1cre+} mice is caused by hematopoietic cells including osteoclast precursor cells. Thus, LRF in osteoclast precursor cells negatively regulates the osteoclast number in vivo.

Fig. 2.

Osteoporotic phenotype of Zbtb7a^{Flox/FloxMx1cre+} mice. (A) Images of the proximal femurs of the control (Zbtb7a^Flox/Flox) and Zbtb7a^{Flox/FloxMx1cre+} mice (Upper, longitudinal view; Lower, axial view of the metaphyseal region) and the parameters on microcomputed tomography (μCT) analysis. (B) Histological analysis of the proximal tibiae of the control and Zbtb7a^{Flox/FloxMx1cre+} mice (TRAP staining). (C) Parameters for osteoclastic bone resorption determined by bone morphometric analysis. *P < 0.05, **P < 0.01.

Enhanced Osteoclastogenesis and NFATc1 Expression by the Deletion of LRF in Osteoclast Precursor Cells.

We examined osteoclast differentiation in Zbtb7a^{Flox/FloxMx1cre+} BMMs stimulated with RANKL in the presence of M-CSF. Because the number of T and B cells in the bone marrow differed between the Zbtb7a^Flox/Flox and Zbtb7a^{Flox/FloxMx1cre+} mice (20), we sorted and cultured CD4⁻CD8⁻B220⁻ cells in the presence of M-CSF for 2 d, which were then used as BMMs. The TRAP⁺ MNC number was markedly increased in the Zbtb7a^{Flox/FloxMx1cre+} cells compared with the control cells (Fig. 3A), but there was a slight decrease in bone resorbing activity in Zbtb7a^{Flox/FloxMx1cre+} osteoclasts when the same number of mature osteoclasts were seeded (Fig. 3B). Therefore, the increase in bone resorption in the Zbtb7a^{Flox/FloxMx1cre+} mice was caused by the increased number of osteoclasts, not by an increase in their activity. To examine the influence of abnormalities of T or B cells on osteoclastogenesis, we analyzed osteoclast formation after depleting the T or B cells from bone marrow cells. Depletion and reconstitution of either T or B cells did not affect osteoclastogenesis in Zbtb7a^{Flox/FloxMx1cre+} cells, suggesting that LRF functions in a cell-autonomous manner (Fig. S6). We examined the ratio of the osteoclast precursor cells among the bone marrow cells. The percentage of c-kit⁺c-fms⁺ cells in the CD11b^lo/−CD3ε⁻B220⁻ population (25) was comparable between the control and Zbtb7a^{Flox/FloxMx1cre+} mice, indicating that the proportion of osteoclast precursor cells in the non-T non-B cells in the bone marrow was unchanged (Fig. 3C). In addition, there was no significant difference in the number or proliferation rate of CD11b⁺ cells cultured in the presence of M-CSF for 2 d (Fig. 3D).

Fig. 3.

LRF suppresses the early stage of osteoclast differentiation through inhibition of NFATc1 expression. (A) Osteoclast differentiation of the control (Zbtb7a^Flox/Flox) and Zbtb7a^{Flox/FloxMx1cre+} cells (Left). TRAP⁺ multinucleated cells (TRAP⁺ MNC) were counted (Right). (B) Bone-resorbing activity of mature osteoclasts derived from the control and Zbtb7a^{Flox/FloxMx1cre+} BMMs. (C) The percentage of c-kit⁺c-fms⁺ cells in the CD11b^lo/−CD3ε⁻B220⁻ population in the bone marrow of the control and Zbtb7a^{Flox/FloxMx1cre+} mice. (D) Proportion of CD11b⁺ osteoclast precursor cells (Left) and the BrdU incorporation (Right) in the control and Zbtb7a^{Flox/FloxMx1cre+} BMMs. (E) NFATc1 expression at the protein (Left) and mRNA level (Right) in the control and Zbtb7a^{Flox/FloxMx1cre+} cells during osteoclastogenesis. **P < 0.01, ***P < 0.005.

In Zbtb7a^{Flox/FloxMx1cre+} cells, the induction of NFATc1 expression was accelerated at both the protein and mRNA levels (Fig. 3E). It is noteworthy that the expression of NFATc1 was detected even in the absence of RANKL stimulation (Fig. 3E), suggesting that LRF inhibits osteoclastogenesis through a suppression of NFATc1 expression. These results demonstrate that LRF deficiency in osteoclast precursor cells results in enhanced osteoclast differentiation by promoting NFATc1 expression without affecting the generation of osteoclast precursor cells.

Increased Bone Mass in Zbtb7a^Flox/−Ctsk^Cre/+ Mice.

Unlike the Zbtb7a^{Flox/FloxMx1cre+} mice, the Zbtb7a^Flox/−Ctsk^Cre/+ mice were shown to have an increased trabecular bone volume and trabecular number (Fig. 4A and Fig. S7A). Bone morphometric analysis indicated a marked decrease in the eroded surface, which is a bone resorption marker, but there was no significant difference in either the osteoclast number or surface (Fig. 4 B and C). There was no obvious abnormality in the parameters for osteoblastic bone formation, such as the bone formation rate or osteoblast surface (Fig. S7 B and C). These results suggested that the high bone mass phenotype of the Zbtb7a^Flox/−Ctsk^Cre/+ mice was caused by a decrease in osteoclastic bone-resorbing activity.

Fig. 4.

Increased bone mass in Zbtb7a^Flox/−Ctsk^Cre/+ mice. (A) Images of the proximal femurs of the control (Zbtb7a^Flox/+Ctsk^Cre/+) and Zbtb7a^Flox/−Ctsk^Cre/+ mice (Upper, longitudinal view; Lower, axial view of the metaphyseal region) and the parameters on μCT analysis. (B) Histological analysis of the proximal tibiae of the control and Zbtb7a^Flox/−Ctsk^Cre/+ mice (TRAP staining). (C) Parameters for osteoclastic bone resorption determined by bone morphometric analysis. *P < 0.05, **P < 0.01.

We obtained bone marrow cells from the Zbtb7a^Flox/−Ctsk^Cre/+ mice and subjected them to an in vitro osteoclast formation and bone resorption assay on dentine. RANKL induced a comparable number of TRAP⁺ MNCs in the control and Zbtb7a^Flox/−Ctsk^Cre/+ cells (Fig. 5A), but the number and area of the resorption pits were markedly decreased in Zbtb7a^Flox/−Ctsk^Cre/+ osteoclasts, even though the same number of osteoclasts were seeded (Fig. 5B). The impaired bone resorption activity in Zbtb7a^Flox/−Ctsk^Cre/+ osteoclasts may be explained by the decrease in the mRNA expression of the genes required for bone resorption including Ctsk, Mmp9, and Clcn7 may influence (Fig. 5C), although we cannot rigorously rule out the other possibilities. The NFATc1 induction was observed to be equal in the control and Zbtb7a^Flox/−Ctsk^Cre/+ cells (Fig. 5D), consistent with the result indicating the normal formation of TRAP⁺ MNCs (Fig. 5A). LRF suppresses erythroblast apoptosis by inhibiting the expression of the proapoptotic protein Bim (encoded by Bcl2l11) (21). Bim also regulates osteoclast apoptosis (26), but there was no increase in Bim expression or the apoptosis rate in Zbtb7a^Flox/−Ctsk^Cre/+ cells (Fig. 5E). There was no significant difference in the number or the proliferation rate between the control and Zbtb7a^Flox/−Ctsk^Cre/+ cells 48 h after RANKL stimulation (Fig. 5F). Although LRF overexpression reportedly led to prolonged osteoclast survival in a gain-of-function study (24), these results in a loss-of-function study suggest that LRF expression at the late stage of osteoclastogenesis is not required for the survival of osteoclast lineage cells, but is instead essential for bone resorption activity.

Fig. 5.

LRF promotes the bone resorption activity of mature osteoclasts. (A) Osteoclast differentiation of the control (Zbtb7a^Flox/Flox) and Zbtb7a^Flox/−Ctsk^Cre/+ cells (Left). TRAP⁺ MNCs were counted (Right). (B) Bone-resorbing activity of mature osteoclasts derived from the control and Zbtb7a^Flox/−Ctsk^Cre/+ BMMs. (C) Expression of osteoclast-specific genes in the control and Zbtb7a^Flox/−Ctsk^Cre/+ cells during osteoclastogenesis. (D) NFATc1 expression at the protein (Left) and mRNA (Right) in the control and Zbtb7a^Flox/−Ctsk^Cre/+ cells during osteoclastogenesis. (E) Expression of Bcl2l11 (Left) and the apoptosis rate (Right) in the control and Zbtb7a^Flox/−Ctsk^Cre/+ cells 48 h after RANKL stimulation. (F) Proportion of CD11b⁺ cells (Left) and BrdU incorporation (Right) in control and Zbtb7a^Flox/−Ctsk^Cre/+ cells 48 h after RANKL stimulation. **P < 0.01.

Positive and Negative Regulation by LRF.

Previous studies demonstrated that IRF-8, MafB, and Bcl-6 potently inhibit the expression and function of NFATc1 (9–11). However, the expression of such antiosteoclastogenic factors was not reduced in Zbtb7a^{Flox/FloxMx1cre+} cells (Fig. S8A), raising the possibility that LRF directly inhibits NFATc1 expression at an early stage of osteoclastogenesis. A computational search on the upstream sequence of the transcription start site of the Nfatc1 gene revealed that several LRF-binding sites (14) were located in the promoter region required for the NFATc1 autoamplification (Fig. 6A). In contrast, a smaller number of LRF-binding sites were found in the promoters of genes expressed in mature bone-resorbing osteoclasts, such as Ctsk, Calcr, Mmp9, and Clcn7, all of which are known to be transcriptional targets of NFATc1 (Fig. 6A and Fig. S8B).

Fig. 6.

Possible mechanism for the biphasic role of LRF in osteoclastogenesis. (A) Schematic view of the putative promoter regions in osteoclast-specific genes. The binding sites of LRF and NFAT on the 5′flanking region of the genes are shown as black and green rectangles, respectively. (B) Physical interaction between NFATc1 and LRF or its truncated mutants. Schematic illustration of LRF, ΔPOZ-LRF, and ΔZF-LRF (Upper). IP, immunoprecipitation; IB, immunoblot. (C) Effect of LRF and its truncated mutants on the NFATc1-mediated activation of the Nfatc1 (P1 luc), Ctsk (Ctsk luc), and Calcr (CTR luc) promoters. (D) Recruitment of LRF to the promoters of the Nfatc1, Ctsk, and Calcr genes at early (day 0) and late (day 3) stages of osteoclastogenesis (Upper). Recruitment of corepressors to the Nfatc1 promoter in Zbtb7a^{Flox/FloxMx1cre+} BMMs (Lower). (E) Working model for the stage-specific functions of LRF during osteoclastogenesis. At the early stage of osteoclast development, LRF blocks NFATc1 expression by acting as a transcriptional repressor. At the late stage of osteoclastogenesis, LRF acts as a coactivator of NFATc1 to promote the expression of genes required for the bone-resorbing activity of osteoclasts.

LRF physically interacted with NFATc1 in the late phase of osteoclastogenesis (Fig. S8C). NFATc1 coimmunoprecipitated with full-length LRF and a mutant, ΔZF-LRF, which lacks the zinc finger domain required for DNA binding, but not with a mutant, ΔPOZ-LRF, lacking the POZ domain (Fig. 6B). Thus, LRF interacts with NFATc1 through the POZ domain. LRF suppressed the activation of the Nfatc1-P1 promoter by NFATc1 and this suppression was also observed without NFATc1 overexpression (Fig. 6C). ΔZF-LRF failed to suppress the Nfatc1-P1 promoter activity enhanced by NFATc1, suggesting that LRF acts as a transcriptional repressor by binding to the promoter through its zinc finger domain (Fig. 6C). Indeed, LRF resided at the proximal Nfatc1 promoter before RANKL stimulation, but not in the late phase of differentiation (Fig. 6D). Previous reports showed that LRF binds to a corepressor complex composed of histone deacetylases (HDACs), N-CoR and SMRT through the POZ domain (19, 27). At the early stage of osteoclastogenesis, HDAC1, HDAC2, HDAC3, N-CoR, SMRT, and B-CoR were recruited to the Nfatc1 promoter, but the recruitment of most of these components, if not all, was severely impaired in the LRF deficiency (Fig. 6D). The expression of Jmjd3, which is reported to be a histone demethylase regulating osteoclastogenesis (28), was not altered by the LRF expression (Fig. S8D). These results suggest that LRF repressed transcriptional activity of NFATc1 by recruiting a corepressor complex (Fig. 6E).

LRF was associated with the promoters of genes expressed by mature osteoclasts including Ctsk and Calcr in the late phase of differentiation (Fig. 6D). Activation of these promoters by NFATc1 was further enhanced by the expression of LRF as well as ΔZF-LRF (Fig. 6C), but not by the expression of ΔPOZ-LRF. There were four and eight NFAT-binding sites, but only two and no LRF-binding sites, in the Ctsk and Calcr promoter regions examined here, respectively (Fig. 6A). These results suggest that LRF functions as a transcription coactivator by binding to NFATc1 through its POZ domain in the regulation of genes required for bone resorbing activity, and this activity is not dependent on the direct DNA binding of LRF through the zinc finger domain.

Discussion

This study revealed that mice lacking LRF in osteoclast precursor cells exhibit an osteoporotic phenotype due to an increased osteoclast number, whereas mice lacking LRF in cells in the more advanced stage of osteoclastogenesis have an increased bone mass due to impaired osteoclastic bone resorption. LRF thus plays a biphasic role in the transcriptional regulation of osteoclastogenesis: LRF represses osteoclast differentiation by acting as a transcriptional repressor of the Nfatc1 gene in the early phase of osteoclast development and promotes bone resorption, at least in part, by acting as a coactivator of NFATc1 in controlling the genes required for the bone-resorbing activity of osteoclasts. The phenotype caused by the gene deletion in the early phase could be influenced by the deletion in the late phase that inevitably occurs, but the LRF deletion in the early phase resulted in a very high expression of NFATc1, which sufficiently functions even without cooperation with LRF. LRF regulates B versus T lymphoid lineage fate decision in the bone marrow through inhibiting the Notch pathway (20). Because Notch signaling is known to repress osteoclastogenesis (29), it is unlikely that LRF inhibits osteoclast differentiation through suppression of Notch function. Recent studies showed that LRF positively regulates transcription by the specific recognition of its DNA-binding sequence (18, 30, 31). It remains to be determined how LRF exerts positive and negative effects on the activity of each promoter, but it is interesting to note that the relative number of LRF- and NFAT-binding sites is markedly different between the promoters that are respectively inhibited and activated by LRF (Fig. 6A and Fig. S8B). Because the negative regulation by LRF is dependent on DNA binding and positive regulation is dependent on the interaction with NFATc1, it is reasonable to hypothesize that LRF functions as a repressor when the promoter contains a large number of LRF-binding sites and as an activator when the promoter contains considerably more NFAT- than LRF-binding sites (Fig. 6E).

Cell lineage commitment is strictly controlled so as to maintain the homeostasis of the biological systems of multicellular organisms. To this end, the cell differentiation program is tightly regulated by multiple signaling pathways, and negative feedback regulation is extremely important to avoid excessive cell differentiation. The negative regulators of osteoclastogenesis IRF-8, MafB, and Bcl-6, are highly expressed in the early phase of development, but the expression is blocked by various repressors, including Blimp-1, as the development proceeds (32). Unlike other antiosteoclastogenic factors, LRF is expressed at a relatively low level in osteoclast precursor cells, but it is induced during osteoclastogenesis, ultimately coming to function as a positive regulator of different target genes. The unique stage-specific regulation of osteoclastogenesis by LRF reveals an intricate transcriptional network in bone homeostasis and may provide a novel molecular basis for therapeutic strategies against bone and joint diseases.

Materials and Methods

Mice and Analysis of the Bone Phenotype.

The generation of Zbtb7a^Flox/Flox, Mx1-Cre transgenic (003556; The Jackson Laboratory), and Ctsk-Cre knock-in (Ctsk^Cre/+) mice was previously described (20, 33). All of the animal experiments were approved by the institutional animal care and use committee and conformed to relevant guidelines and laws. All mice were backcrossed with C57BL/6 mice more than six times and maintained under specific pathogen-free conditions. For the induction of Cre recombinase expression in hematopoietic stem cells, poly I:C (Invitrogen) was administered into mice, as described, with minor modifications (34). Briefly, Zbtb7a^Flox/Flox and Zbtb7a^{Flox/FloxMx1cre+} mice were injected intraperitoneally with 0.25 mL of 1 mg mL⁻¹ poly I:C every other day starting at postnatal day 21 for a total of three doses. The femurs of 9-wk-old male Zbtb7a^{Flox/FloxMx1cre+} mice (n = 8), 12-wk-old female Zbtb7a^Flox/−Ctsk^Cre/+ mice (n = 8), and their littermate controls were subjected to 3D microcomputed tomography (μCT) analysis. The tibiae of these mice were subjected to histomorphometric analysis. The methods for μCT and histomorphometric analyses were previously described (35).

In Vitro Assays for Osteoclast Differentiation and Function.

The methods for in vitro osteoclast differentiation and bone resorption assay were previously described (36). Briefly, bone marrow cells cultured with 10 ng mL⁻¹ M-CSF (R&D Systems) for 2 d were used as BMMs, which were further cultured with 50 ng mL⁻¹ (unless otherwise indicated) of RANKL (Peprotech) in the presence of M-CSF (10 ng mL⁻¹) for 3 d. For the analysis of osteoclastogenesis in the Zbtb7a^{Flox/FloxMx1cre+} cells, CD4⁻CD8⁻B220⁻ cells isolated using an AutoMACS cell separator were cultured in the presence of M-CSF for 2 d and used as BMMs. For the bone resorption assay, the osteoclasts generated on collagen-coated dishes were harvested 3 d after RANKL stimulation by trypsinization and cultured for 2 d on dentin slices (Wako) in the presence of indicated concentrations of RANKL. The resorbed areas were measured using an image analyzing software, ImageJ (NIH Image).

Statistical Analysis.

Each series of experiments was repeated at least three times. All data are expressed as the mean ± SEM (n = 5). Statistical analysis was performed using Student's t test and ANOVA followed by a Bonferroni test when applicable.

Detailed procedures for the immunofluorescence staining, the retroviral gene transfer, immunoblot, and immunoprecipitation analyses, the quantitative PCR analysis, the flow cytometric analysis, the luciferase reporter gene assay, and the ChIP assay are described in SI Materials and Methods.