Identification and Characterization of the Precursors Committed to Osteoclasts Induced by TNF-Related Activation-Induced Cytokine/Receptor Activator of NF-κB Ligand¹ Free

Osteoclasts are multinucleated giant cells responsible for bone resorption, and it is generally accepted that osteoblasts control osteoclast differentiation and bone-resorbing activity through cell-cell physical contact (1, 2, 3). Osteoblasts produce a membrane-bound cytokine known as TNF-related activation-induced cytokine (TRANCE).⁴ TRANCE is identical with the molecules widely called receptor activator of NF-κB ligand (RANKL), osteoprotegerin (OPG) ligand, osteoclast differentiation factor (ODF), TNFSF11, and CD254, which trigger osteoclast differentiation from osteoclast precursors in the presence of M-CSF (4, 5, 6, 7).

A receptor for TRANCE, called RANK, is expressed on osteoclast precursors and activates intracellular signaling pathways by recruiting TNFR-associated factor proteins such as TNFR-associated factors 1, 2, 3, 5, and 6 (5, 8, 9). Intracellular signal mediators such as TGF-β-activated kinase 1, p38 MAPK, and JNK activate transcription factors such as NF-ATc1 (identical with NF-AT2), c-Fos, and NF-κB, which regulate the expression of genes related to osteoclastogenesis (4, 10, 11, 12, 13, 14, 15).

Endogenous cytokines and hormones regulate osteoclast differentiation induced by TRANCE, while OPG (identical with osteoclastogenesis-inhibitory factor) is a decoy receptor for TRANCE that inhibits osteoclast differentiation by a mechanism that interrupts TRANCE-RANK interaction (16, 17, 18). IL-4, IFN-γ, and GM-CSF are known to inhibit osteoclast differentiation by direct actions toward osteoclast precursors, whereas IL-1, PGE₂, and TGF-β promote osteoclast differentiation induced by TRANCE (19, 20, 21, 22, 23, 24, 25, 26, 27).

Many investigators have explored the mechanism of osteoclast differentiation using a culture system of mouse bone marrow-derived macrophages (BMMs), in which BMMs are prepared from bone marrow cells by treatment with M-CSF, and then differentiate into osteoclasts in the presence of M-CSF and TRANCE within 72 h (28, 29). We previously reported that BMMs possess physiological functions typical of macrophages, including phagocytosis and production of TNF-α in response to LPS (30). However, during differentiation into osteoclasts, the cells lost those functions, while they obtained a bone-resorbing function, suggesting that characteristics of BMMs are dramatically changed during the process of differentiation into osteoclasts.

Osteoclast precursors are derived from hemopoietic stem cells through a monocyte/macrophage lineage (31, 32, 33). Arai et al. (34) reported that M-CSF induces osteoclast precursors expressing RANK that possess a bipotentiality to differentiate into osteoclasts and macrophages. Furthermore, Li et al. (35) showed that BMMs possess the potential to differentiate into not only osteoclasts, but also dendritic cells. These observations suggest that osteoclast precursors have a multipotential capacity to differentiate into several cell types, including osteoclasts and dendritic cells. Therefore, it has been speculated that osteoclast precursors lose their multipotential differentiation capacity after their cell fate is committed to osteoclasts during osteoclastogenesis caused by TRANCE stimulation, although it is not known when or how the cell-fate commitment occurs.

In the present study, to explore the process of cell-fate commitment of BMMs to osteoclasts, we first determined when osteoclast precursors are committed to differentiate into osteoclasts in vitro. Subsequently, we characterized the committed cells in regard to their differentiation capacity and cellular functions. Our results provide new information about cellular characteristics during osteoclastogenesis.

Human M-CSF (Leucoprol) and fluorescein-conjugated zymosan A (Saccharomyces cerevisiae) Bio Particles were purchased from Kyowa Hakko Kogyo and Molecular Probes, respectively. Alpha-modified MEM (α-MEM) and LPS from Escherichia coli (O55:B5) were purchased from Sigma-Aldrich. Soluble forms of human TRANCE and human OPG were produced by insect cells, and purified by affinity chromatography, as described previously (4, 36). Recombinant human TGF-β, mouse IL-4, mouse GM-CSF, and mouse IFN-γ were purchased from R&D Systems. Rabbit polyclonal Abs against phospho-IκB (9241) and IκB (9242) were purchased from Cell Signaling Technology.

Five- to 6-wk-old mice (ddY strain) were obtained from Saitama Experimental Animals. To obtain BMMs, mouse bone marrow cells were collected from the tibiae and femora of the mice, and cultured for 3 days in α-MEM containing 10% FBS and M-CSF (50 ng/ml) in 100-mm-diameter type I collagen-coated culture dishes (IWAKI-Asahi Glass) (1 × 10⁷ cells/10 ml/dish). To promote the efficiency of osteoclast differentiation, human TGF-β (1 ng/ml) was added to the culture medium together with M-CSF, according to the previous reports (25, 26, 27, 37, 38, 39). After culturing for 3 days, floating cells were gently removed by rinsing with PBS, and cells remaining attached to the culture plates were collected by treatment with trypsin-EDTA (Invitrogen Life Technologies) and used as BMMs. To obtain committed cells, BMMs were further cultured in the presence of M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml) for 24 h. Then, attached cells were harvested by treatment with trypsin-EDTA and used as committed cells. To induce osteoclast differentiation, BMMs and committed cells were stimulated with M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml) for 72 h (BMMs) or 48 h (committed cells). Then, osteoclast formation was evaluated by quantifying the tartrate-resistant acid phosphatase (TRAP) activity, a marker enzyme of osteoclasts, or by counting TRAP-positive multinucreated cells containing three or more nuclei. To induce immature dendritic cell differentiation, BMMs were cultured in the presence of M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml) with GM-CSF (10 ng/ml) for 72 h. Maturation of immature dendritic cells was induced by treatment with LPS (1 μg/ml) for additional 24 h. Mature dendritic cells were detected by the method described below. All procedures were performed according to the Showa University Animal Care and Use Committee Guidelines (permission number of this experiment: 15042).

FITC-conjugated zymosan A Bio Particles were added to the cultures of BMMs and committed cells (250 mg/5 ml/100-mm culture dish or 5 μg/0.2 ml/96-well culture plate) for 60 min, after which they were rinsed with PBS twice to remove the particles that were not incorporated into the cells. In some experiments, cells were collected by treatment with trypsin-EDTA and cultured in 96-well culture plates (1 × 10⁴ cells/0.2 ml/well) (Corning Glass) in the presence of M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml) for 72 h (BMMs) or 48 h (committed cells). The cells were then fixed with Formalin and treated with rhodamine-conjugated phalloidin (Molecular Probes), which binds to F-actin. Using UV illuminated microscopy, F-actin and zymosan particles in the osteoclasts were visualized as red and green dots, respectively. Finally, the cells were stained for TRAP.

BMMs and committed cells were cultured on dentin slices (4 mm in diameter, 0.2 mm in thickness; 5 × 10⁴ cells/slice) in 96-well culture plates (0.2 ml/well) for 24 h with M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml). The dentin slices were then transferred to 48-well culture plates (0.4 ml/well) using forceps to culture the cells in an adequate volume of medium. Cells on dentin slices were further cultured in the presence of M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml) for 72 h (BMMs) or 48 h (committed cells). After wiping the cells off dentin slices with cotton, the slices were immersed in toluidine blue O (Sigma-Aldrich) to stain resorption pits formed by mature osteoclasts.

BMMs and committed cells were cultured in 60-mm-diameter dishes for the indicated periods with LPS (1 μg/ml) in the presence of M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml). Then, total cell lysates were isolated, separated by SDS-PAGE, and transferred onto Immobilon-P membranes (Millipore). The membranes were blocked with 5% nonfat milk in TBST (150 mM NaCl, 20 mM Tris (pH 7.4), and 0.1% Tween 20), then subjected to immunostaining with anti-phospho IκB (1/1000) or anti-IκB Ab (1/1000), followed by secondary HRP-conjugated Ab (1/5000). The membranes were developed using an ECL detection kit (Amersham Biosciences).

Total RNA from the cells in culture dishes (60-mm diameter) was prepared using TRIzol solution (Invitrogen Life Technologies). First-strand cDNA was synthesized for PCR using Superscript II (Invitrogen Life Technologies) and subjected to amplification with Taq polymerase (Sigma-Aldrich) using the following specific PCR primers: mouse FcγRIII, 5′-TGACACCCCATCCATCCTAT-3′ (forward) and 5′-TATGCCATCAACCCTTAGCC-3′ (reverse); FcγRII, 5′-TGATTTCTGACTGGCTGCTG-3′ (forward) and 5′-CCAATGCCAAGGGAGACTAA-3′ (reverse); CD14, 5′-CTGATCTCAGCCCTCTGTCC-3′ (forward) and 5′-GCAAAGCCAGAGTTCCTGAC-3′ (reverse); lysozyme, 5′-ACTGCTCAGGCCAAGGTCTA-3′ (forward) and 5′-GCCCTGTTTCTGCTGAAGTC-3′ (reverse); TLR4, 5′-ACCTGGCTGGTTTACACGTC-3′ (forward) and 5′-CAGGCTGTTTGTTCCCAAAT-3′ (reverse); NF-ATc1, 5′-TCATCCTGTCCAACACCAAA-3′ (forward) and 5′-TTGCGGAAAGGTGGTATCTC-3′ (reverse); cathepsin K, 5′-CTTCCAATACGTGCAGCAGA-3′ (forward) and 5′-AGCCACCAATATCTTGCACC-3′ (reverse); IL-1R, 5′-GAATGACCCTGGCTTGTGTT-3′ (forward) and 5′-CGTGACGTTGCAGATCAGTT-3′ (reverse); osteoclast-associated receptor (OSCAR), 5′-ACTCCTGGGATCAACGTGAC-3′ (forward) and 5′-GATAGCACATAGGGGGCAGA-3′ (reverse); α-actin, 5′-TGGAGAGAGTCAAGCCTGGT-3′ (forward) and 5′-AGGTCCGCTTAACCCATCTT-3′ (reverse); TRAP, 5′-GAGAACGGTGTGGGCTATGT-3′ (forward) and 5′-CTGTGGGATCAGTTGGTGTG-3′ (reverse); RANK, 5′-TGCAGCTCAACAAGGATACG-3′ (forward) and 5′-ACCATCTTCTCCTCCCGAGT-3′ (reverse); integrin α_V, 5′-ACACTTTGGGCTGTGGAATC-3′ (forward) and 5′-CGCCACTTAAGAAGCACCTC-3′ (reverse); integrin β₃, 5′-GACCACAGTGGGAGTCCTGT-3′ (forward) and 5′-GAGGGTCGGTAATCCTCCTC-3′ (reverse); and integrin β₅, 5′-CGGAACCTACCTCTCAGCAG-3′ (forward) and 5′-TGCTTCCTCACTTCCTCGTT-3′ (reverse).

For Northern blot analysis, cells in 60-mm-diameter culture dishes were treated with TRANCE (150 ng/ml) or LPS (1 μg/ml) for indicated time periods and then subjected to total RNA isolation using TRIzol. The cDNA probes encoding mouse TRAP, TNF-α, and GAPDH were labeled with [³²P]dCTP using a cDNA labeling kit (Amersham Biosciences). Total RNA (20 μg) was electrophoresed on 1.5% agarose-formaldehyde gels, transferred to a nylon membrane, and hybridized with ³²P-labeled cDNA probes. After the final wash, the membranes were exposed to x-ray film (BioMax).

To detect dendritic cells, cells were stained with FITC-labeled anti-CD11c Ab (557400; BD Pharmingen) and PE-conjugated anti-CD86 Ab (553692; BD Pharmingen). To detect dead cells, BMMs and committed cells were stained with propidium iodide (PI). To detect RANK on BMMs, cells were stained with rat anti-RANK Ab, followed by PE-conjugated anti-rat IgG. Anti-RANK mAb was produced by hybridoma cells prepared from rat immunized with mouse RANK protein (40). The expression of each molecule on the cells and dead cells was analyzed using a FACSCalibur (BD Pharmingen).

To visualize the nuclei, BMMs and committed cells were fixed and stained with DAPI (Sigma-Aldrich), and visualized by UV illumination.

BMMs (2 × 10⁵ cells) or committed cells (2 × 10⁵ cells) were suspended in 2 ml of α-MEM medium containing 10% FBS, M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml). The cell suspension was loaded onto 1 ml of 2.3% methylcellulose medium (STEMPRO; Invitrogen Life Technologies) containing M-CSF (50 ng/ml), TGF-β (1 ng/ml), and TRANCE (150 ng/ml) in a 14-ml polypropylene round-bottom tube (BD Falcon; 2059). The cells were cultured for 72 h (BMMs) or 48 h (committed cells), and then harvested by centrifugation and stained for TRAP.

Cells in 96-well culture plates were rinsed with PBS and dissolved with 150 μl of lysis buffer (50 mM acetic acid buffer (pH 5.0), containing 1% sodium tartrate and 0.1% Triton X-100). The cell lysates were briefly sonicated to dissolve the cell constituents well, and then 30 μl of cell lysate was mixed with 100 μl of p-nitrophenyl phosphate solution (1 mg/ml in 50 mM acetic acid buffer (pH 5.0), containing 1% sodium tartrate) and incubated at 37°C for 30 min. After the addition of 70 μl of 1 M NaOH, absorbance was measured at 405 nm.

After culture, the cells on dentin slices were fixed in a fixative containing 2% formaldehyde and 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.3) for 2 h at 4°C. After fixation, the specimens were decalcified in 10% ethylenediamine tetracetic acid disodium solution (pH 7.3) for 2 wk. They were then postfixed with 1.5% potassium ferrocyanide-reduced 1% osmium tetroxide for 30 min at 4°C. The specimens were then block stained with 1% uranyl acetate in 10% ethanol, dehydrated through a graded ethanol series, and embedded in Quetol 812 (Nissin EM). Ultrathin sections were cut using a diamond knife on a Reichert-Jung OmU-4, stained with uranyl acetate and lead citrate, and examined with an H-7000 transmission electron microscope (Hitachi) at 75 kV.

Freshly isolated BMMs, obtained from bone marrow cell cultures, as described in Materials and Methods, were spindle shaped and adhered to the culture plates (Fig. 1,A). In the absence of exogenous stimuli, BMMs were able to remain as macrophages (data not shown). The BMMs expressed RANK, a TRANCE receptor (Fig. 1,B). After 24-h treatment with TRANCE, the cell shape became round, and the expression level of TRAP mRNA, a marker of osteoclasts, was slightly increased. However, cells were still negatively stained for TRAP, suggesting that TRAP activity is too low to be detected (Fig. 1, A, C, and D). TRAP-positive cells, which were constituted with a large number of mononuclear cells and a small number of multinucleated osteoclasts, were detected at 48 h (Fig. 1,A), and the number of multinucleated osteoclasts increased thereafter (Fig. 1, A and E). When GM-CSF was added to the cultures together with TRANCE, osteoclast differentiation was strongly inhibited (Fig. 1,E), and cells expressing CD11c and CD86 (B7-2), dendritic cell markers, were formed in the cultures at 72 h (Fig. 1, F and G). These results indicate that GM-CSF inhibits osteoclast differentiation, whereas it promotes dendritic cell differentiation even in the presence of TRANCE.

From these results, we speculated that once BMMs have become committed to osteoclast differentiation, committed cells do not differentiate into dendritic cells even in the presence of GM-CSF. Next, GM-CSF was added to BMM cultures at 6, 12, and 24 h after the addition of TRANCE (Fig. 2). When GM-CSF was added at 6 and 12 h, osteoclast differentiation was partially inhibited (Fig. 2). However, the addition of GM-CSF at 24 h following the addition of TRANCE did not inhibit osteoclast differentiation, and the number of dendritic cells was significantly decreased in the cultures (Fig. 2). These results indicate that the fate of BMMs committed to osteoclast differentiation occurs within 24 h of stimulation with TRANCE. Based on these findings, we used BMMs treated with TRANCE for 24 h as committed cells to differentiate into osteoclasts in the following experiments.

Because GM-CSF failed to inhibit osteoclast differentiation by committed cells, we tested the effects of other inhibitory factors of osteoclastogenesis, such as IFN-γ, IL-4, and the p38 MAPK inhibitor SB203580. As shown in Fig. 3, these factors inhibited osteoclast differentiation by noncommitted BMMs, but demonstrated no such effect toward committed cells (Fig. 3). These results suggest that the inhibitory factors interrupt the process of commitment into osteoclasts by TRANCE stimulation.

To examine the role of TRANCE in osteoclastogenesis by committed cells, TRANCE stimulation was blocked by OPG, a decoy receptor of TRANCE (Fig. 4). When OPG was added to the committed cell cultures, most of the cells became detached from the culture plates and floated in the medium within 24 h (Fig. 4,A). Furthermore, staining of the committed cells with DAPI revealed nuclear condensation, a typical phenotype of cell death (Fig. 4,B). Analysis of cells stained with PI by flow cytometry indicated cell death of the committed cells (Fig. 4,C). In contrast, the addition of OPG to the BMM cultures inhibited osteoclast differentiation, but did not induce cell death (Fig. 4, A–C). These results suggest that TRANCE is required for the survival of committed cells. Subsequently, we added LPS and GM-CSF in place of TRANCE and M-CSF to investigate their effect toward cell survival (Fig. 4,D). The committed cells differentiated into osteoclasts in the presence of LPS and GM-CSF and in the absence of TRANCE and M-CSF (Fig. 4 D). Thus, LPS and GM-CSF can be substituted for TRANCE and M-CSF after the commitment of BMMs to osteoclasts.

It has been reported that adhesion signals are required for osteoclast differentiation (41); therefore, we examined whether committed cells require adhesion to culture plates to differentiate into osteoclasts. To interrupt the attachment of committed cells to the plates, fresh BMMs and committed cells were cultured on methylcellulose. The fresh BMMs did not differentiate (Fig. 5, upper left panel), whereas the committed cells efficiently differentiated into multinucleated osteoclasts (Fig. 5, upper right panel), which formed resorption pits when they were transferred onto dentin slices (Fig. 5, lower right panel). These results suggest that committed cells differentiate into osteoclasts and fuse in a manner independent of adhesion signals.

We previously reported that BMMs did not differentiate into osteoclasts after incorporating zymosan particles by phagocytosis (30). In the present study, we examined the effects of zymosan phagocytosis on the differentiation of committed cells into osteoclasts (Fig. 6). BMMs as well as committed cells incorporated the FITC-labeled zymosan particles by phagocytosis within 1 h (Fig. 6,A). However, osteoclasts did not incorporate the zymosan particles (Fig. 6,A). As we reported previously, BMMs containing zymosan particles failed to differentiate into osteoclasts (Fig. 6,B). However, committed cells containing zymosan particles differentiated into osteoclasts and formed resorption pits on dentin slices (Fig. 6,B). The incorporated particles were clearly observed as green dots inside actin rings in the osteoclasts by UV illumination (Fig. 6,B). In addition, electron-microscopic examination showed the zymosan particles in the cytoplasm of the bone-resorbing osteoclasts (Fig. 6 C). These results suggest that phagocytosis of zymosan particles does not have an effect on osteoclast differentiation after the commitment.

To examine the characteristics of gene expression by the committed cells, we compared the mRNA expression in BMMs (0 h), committed cells (24 h), and osteoclasts (48–72 h) (Fig. 7,A). The expression levels of FcγRII/III, CD14, lysozyme, and β₅ integrin were decreased in concert with osteoclastogenesis, whereas the expression levels of RANK, TRAP, α_v and β₃ integrins, NF-ATc1, cathepsin K, IL-1R, and OSCAR were increased. In committed cells, the expression of these mRNAs was seen (Fig. 7,A). Because TLR4, a receptor of LPS, was expressed throughout the period of osteoclastogenesis (Fig. 7,A), we also examined the activation of intracellular signaling and TNF-α mRNA expression following LPS stimulation (Fig. 7, B and C). Phosphorylation and the following degradation of IκB were observed in BMMs as well as committed cells in response to LPS (Fig. 7,B). However, although TNF-α mRNA was strongly expressed in BMMs, it was only slightly detected in committed cells (Fig. 7 C). These results suggest that mRNA expression in committed cells is regulated differently from that in BMMs.

Exogenous TGF-β is known to promote osteoclast differentiation induced by TRANCE and affects maturation of dendritic cells (42). We confirmed that TGF-β did not affect immature dendritic cell formation, but reduced the mRNA expression levels of IL-12 induced by LPS (data not shown). This prompted us to examine whether commitment into osteoclasts is modulated by exogenous TGF-β (Fig. 8). We conducted a time-course study to determine when committed cells are formed in the absence and presence of TGF-β using phagocytosis assay because only committed cells, but not BMMs nor osteoclasts, are able to form zymosan-containing osteoclasts (Fig. 6). Zymosan particles were added to BMM cultures at 0, 12, 24, 48, 72, and 96 h after TRANCE stimulation in the absence or presence of TGF-β. All cells were cultured for 120 h, and the number of zymosan-containing osteoclasts formed in the cultures was counted (Fig. 8). Addition of zymosans at 24 h resulted in maximal formation of zymosan-containing osteoclasts in the presence of TGF-β. However, in the absence of TGF-β, addition of zymosans at 48 h caused maximal formation of zymosan-containing osteoclasts. These results suggest that although committed cells are formed in the absence of exogenous TGF-β, a longer period is required for the formation of these committed cells.

In the present study, we found evidence for the existence of a new phase during osteoclastogenesis in which cells possess unique characteristics (Fig. 9). First, we showed that GM-CSF inhibits osteoclast differentiation and induces dendritic cell differentiation from BMMs. However, GM-CSF failed to inhibit osteoclast differentiation after 24 h of TRANCE stimulation, suggesting that the cell-fate decision for differentiation of BMMs to osteoclasts occurred during this period. After the cell fate of BMMs was committed to osteoclasts, no bone-resorbing function was observed, and TRAP activity was very weak. However, committed cells possessed a phagocytic function, which is a typical characteristic of macrophages. These results suggest that cell fate of BMMs becomes committed to osteoclastogenesis in the early period of the differentiation process.

It has been reported that IL-4 and IFN-γ affect osteoclast differentiation only during the initial period of cultures (19, 43, 44). Consistent with these reports, IL-4 and IFN-γ failed to inhibit osteoclast differentiation from committed cells. These results indicate that those inhibitory factors are able to inhibit the commitment of BMMs to differentiate into osteoclasts, but not the subsequent differentiation process after commitment has been decided. SB203580, a p38 MAPK inhibitor, also failed to inhibit osteoclast differentiation from committed cells. Matsumoto et al. (11) reported that p38 MAPK-mediated signaling plays crucial roles in the osteoclast differentiation. Li et al. (35) also suggested that the up-regulation of p38 MAPK is important for osteoclastogenesis involved in the induction of NF-ATc1 activation. Thus, it is proposed that the activation of p38 MAPK is important in the early period of osteoclast differentiation process.

Among the factors tested, only OPG inhibited osteoclast differentiation from committed cells, suggesting that it is a unique factor that can stop osteoclast differentiation even after commitment. OPG is known to induce apoptosis of mature osteoclasts by blocking TRANCE stimulation (45, 46). We found that OPG also induced cell death of committed cells, which suggests that the survival of committed cells, as well as mature osteoclasts, largely depends on TRANCE stimulation.

Although the combination of LPS and GM-CSF was found to be a substitute for TRANCE and M-CSF after commitment, that could not induce the commitment of BMMs into osteoclasts (data not shown). Therefore, TRANCE and M-CSF are considered to be essential molecules involved in the commitment to osteoclasts. However, after commitment, cells do not require TRANCE and M-CSF if other molecules such as LPS and GM-CSF assist in their survival. This may be a pathologically important phenomenon, because LPS has been shown to promote bone destruction by increasing the number of osteoclasts in vivo (47, 48).

In agreement with previous reports, fresh BMMs failed to differentiate into osteoclasts on methylcellulose medium (41). In contrast to BMMs, committed cells successfully differentiated into osteoclasts and formed multinucleated giant cells when cultured on methylcellulose medium. These results suggest that adhesion is an essential requirement for commitment, but not for subsequent differentiation after commitment and cell fusion. Miyamoto et al. (41) suggested that adhesion signals from α_vβ₃ integrins are required for osteoclast differentiation. In contrast, Sago et al. (49) and Shinar et al. (50) have reported that osteoclast precursors express α_vβ₅ integrins, while mature osteoclasts express α_vβ₃ integrins. Our results suggested that β₅ integrin is predominantly expressed over β₃ integrin in the early period of differentiation. Thus, cell adhesion signals through integrins such as α_v, β₃, and β₅ will play important roles in commitment and the differentiation following it.

Zymosan particles added to the cultures were efficiently incorporated by the BMMs as well as the committed cells. Following their incorporation, BMMs failed to differentiate into osteoclasts, whereas the committed cells continued onto osteoclastogenesis. Furthermore, osteoclasts on dentin that contained zymosan particles formed a ruffled border and resorption pits, suggesting that zymosan particles did not disturb bone-resorbing functions. We previously reported that TLR signals stimulated by bacterial constituents, such as LPS, peptidoglycan, and CpG DNA, inhibited osteoclast differentiation from BMMs (30). Thus, the constituents of zymosan might inhibit osteoclast differentiation via TLRs. Similar to IL-4 and IFN-γ, signals from TLRs may not inhibit osteoclast differentiation after commitment.

Analysis of mRNA expression levels in committed cells by RT-PCR revealed that levels of mRNA typical for BMMs were decreased, while those typical for osteoclasts were increased. These results suggest that transcriptional regulation dramatically changes during cell-fate commitment. Furthermore, we consistently found that LPS strongly induced TNF-α mRNA expression in BMMs, but not in committed cells, although the intracellular signaling pathway was activated similarly in BMMs and committed cells. These findings suggest that commitment to osteoclastogenesis involves changes in mRNA expression.

Several reports have shown biphasic effects of osteoclastogenesis regulatory factors by adding those to osteoclast formation cultures in the early and late periods. Zou and Bar-Shavit (51) reported that when LPS was added to cultures from the beginning, osteoclast formation was inhibited; however, when added during a later period, LPS induced osteoclastogenesis, even in the absence of TRANCE. In addition, Ishida and Amano (52) found that osteocalcin fragments in bone matrix enhanced osteoclast maturation at the late stage of osteoclast differentiation. Together, those observations suggest that cellular characteristics are different between the early and late periods of culture, which is consistent with our conclusion that cellular characteristics are significantly different between before and after commitment.

Rivollier et al. (53) have reported that human immature dendritic cells are able to differentiate into osteoclasts when the cells were cultured in the presence of M-CSF and TRANCE. In agreement with this, we confirmed that mouse immature dendritic cells also differentiated into osteoclasts in the presence of M-CSF and TRANCE (data not shown). However, mature dendritic cells prepared from immature dendritic cells by stimulation with LPS did not differentiate into osteoclasts (data not shown). Thus, cell-fate commitment, similar to that in differentiation of osteoclasts that we reported in this work, may also occur in the process of differentiation of immature dendritic cells into mature dendritic cells.

TGF-β has been shown to play essential and accelerative roles in osteoclast differentiation (25, 26, 27, 37, 38, 39, 54). Our results showed that the time of commitment is shortened by exogenous TGF-β, suggesting that signals via TGF-β promote gene expression related to the commitment. In regard to this, Yan et al. (37) have reported that TGF-β promotes RANK expression in RAW264.7 cells that are able to differentiate into osteoclasts. Fuller et al. (26) have reported that the expression level of RANK is up-regulated during osteoclast differentiation, and we also confirmed this. Thus, up-regulation of RANK gene expression by TGF-β is speculated to accelerate the commitment. Further analysis of gene expression regulated by TGF-β in osteoclast precursors will reveal the roles of TGF-β in cell-fate commitment.

Miyamoto et al. (55) showed that overexpression of c-Fos in osteoclast precursors induced osteoclast formation, but not dendritic cell formation even in the presence of GM-CSF. In contrast, NF-ATc1, another transcription factor, has also been shown to be an important molecule that induces osteoclast differentiation (13, 56, 57). Indeed, the expression level of NF-ATc1 mRNA was already increased in committed cells. Thus, cell-fate commitment into osteoclastogenesis seems to be accompanied by down-regulation of genes that regulate dendritic cell differentiation, as well as up-regulation that regulates osteoclast differentiation. The identification of key molecules that determine cell-fate commitment might facilitate elucidation of the road map for osteoclastogenesis.

In summary, we showed that commitment to osteoclastogenesis is accompanied by dramatic changes in cellular characteristics. Our results give a better understanding of the committed phase during osteoclast differentiation (Fig. 9), and may promote the elucidation of regulatory system responsible for cell-fate determination to osteoclasts and dendritic cells.

We thank Dr. Atsushi Yamada for providing helpful advice and excellent technical support.

The authors have no financial conflict of interest.