Osteoclasts are multinucleated cells formed by the fusion of circulating hematopoietic precursor cells of the monocytic/macrophagic lineage, which are responsible for bone resorption. Osteoclastogenesis is a multistep differentiation process under the control of the bone microenvironment, which includes stromal cells, osteoblasts, and local factors (Boyle et al., 2003). One of the key factors mediating osteoclastogenesis is the tumor necrosis factor (TNF) family member receptor activator of nuclear factor kappa B ligand (RANKL), also referred to as osteoclast differentiation factor and osteoprotegerin ligand. Transmembrane RANKL is expressed on the surface of osteoblastic/stromal cells, while the extracellular domain of RANKL can be released by osteoblastic/stromal cells and act as a soluble homotrimeric protein (Boyle et al., 2003). RANK, the cognate transmembrane high-affinity receptor for RANKL, belongs to the TNF receptor family and is expressed in osteoclastic precursor cells (Hsu et al., 1999). RANKL works in conjunction with macrophage-colony stimulating factor (M-CSF), a cytokine of the colony-stimulating factor family, which stimulates proliferation of monocytes and differentiation along the macrophagic lineage (Felix et al., 1994).
Although several intracellular pathways, such as NF-kB, have been involved in RANKL signaling (Boyle et al., 2003), the molecular events underlying changes that occur as precursor cells differentiate into mature osteoclasts are not fully elucidated. Thus, we have investigated the expression of a series of genes after treatment with M-CSF or M-CSF plus RANKL by using the c-DNA microarray technology. In particular, we have analyzed the modulation of genes encoding for cytokines or genes involved in cell survival/apoptosis and cell proliferation. These genes were selected because cell survival and apoptosis are crucial aspects of the osteoclast life cycle. Additionally, the coordination of cell cycle progression and cell differentiation is a key aspect to achieve the specialized characteristics of a terminally differentiated cell (Okahashi et al., 2001; Bharti et al., 2004; Sankar et al., 2004; Ogasawara et al., 2004; Kwak et al., 2005; Wu et al., 2005).
MATERIALS AND METHODS
Cell Cultures and Treatments
As a model system of osteoclastogenesis, human peripheral blood mononuclear cells (PBMCs) were used as previously described (Zauli et al., 2004). Briefly, human PBMCs from healthy normal donors were separated by gradient centrifugation with lymphocyte cell separation medium (Cedarlane Laboratories; Hornby, Ontario) and seeded in culture at a density of 5 × 106 cells/ml. After incubation for 18 hr, nonadherent PBMCs were removed, and the remaining adherent cells were referred to as adherent PBMCs. Cultures were maintained in RPMI medium containing 10% fetal bovine serum (FBS; Gibco BRL, Gaithersburg, MD) and pretreated with human M-CSF (50 ng/ml, PeproTech, London, UK) alone for 6 days. Cells were then cultured in the presence of human M-CSF (50 ng/ml) alone or in combination with human RANKL (50 ng/ml, Alexis Biochemicals, Lausen, Switzerland) and harvested after 12 days of treatment. Medium was replaced every 3 days. At 12 days of culture, the overall number of viable cells was scored under a microscope (10× magnification). The presence of apoptosis was evaluated at different culture times by propidium iodide staining, and flow cytometric analysis was performed as previously described (Campioni et al., 1995; Secchiero et al., 2001).
Cytochemical Staining and Bone Resorption Assay
The cells were stained with 4′,6-diamidino-2-phenylindole (DAPI) to identify nuclei. For this technique, the cells were washed with phosphate buffered saline (PBS), fixed in paraformaldehyde 4% for 10 min, permeabilized in Triton X-100 0.1% for 10 min, washed again with PBS, and incubated with 500 ng/ml DAPI (Sigma Chemicals, St. Louis, MO) in PBS for 15 min at 37°C in a dark, humidified chamber. After several washes in PBS, the coverslips were mounted on PBS/glycerin and the intercalation of DAPI was visualized by means of a fluorescence microscope.
To identify osteoclasts, cells were stained for tartrate-resistant acid phosphatase (TRAP) using a leukocyte acid phosphatase kit (387-A, Sigma Chemicals), according to the manufacturer's instructions. After the staining, cells were rinsed with PBS and photographed under a light microscope.
To assess the ability of the cells to resorb bone, PBMC were plated on 24-well plates coated with an artificial bone slide (OAAS, Osteogenic Core Technologies, Choongnam, Korea) and allowed to attach overnight. Cytokines were added starting from the next day. After 6 days of pretreatment with M-CSF (50 ng/ml) and 12 days of treatments (M-CSF or RANK-L+M-CSF), plates were processed according to the manufacturer's instructions, and resorption lacunae were visualized using a light microscope, as previously described (Zauli et al., 2004).
cDNA Microarray Analysis
RNA was isolated using a Qiagen RNeasy kit (Hilden, Germany) from cells harvested after 12 days of culture with M-CSF alone or with M-CSF plus RANKL (Alexis Biochemicals). Labeled cDNA was hybridized with a customized cDNA microarray containing 96 key apoptosis genes, 96 key cell cycle regulation genes, and 75 stress & toxicity genes together with housekeeping genes at the SuperArray core facility (the list of the genes is available at http://www.superarray.com/gene_array_product/HTML/HS-603.html; Bioscience Corporation, GEArray Service, Frederick, MD). Analysis of gene expression was done essentially as described (available at www. superarray.com/microarrays.php). Alterations imposed by the different treatments on the basal gene expression were determined as the ratio of relative gene expression compared with unstimulated cells. The expression between unstimulated and treated cultures was considered significant when a > 2- or < 0.5-fold change was observed. Some genes were selected for reverse transcriptase-polymerase chain reaction (RT-PCR) confirmation assays by SingleGene PCR kit, purchased from Bioscience Corporation, GEArray.
Scanning Electron Microscopy
For scanning electron microscopy, cell cultures were fixed in 2.5% glutaraldehyde (Sigma) in 0.1 buffered phosphate, pH 7.3 for 30 min at 4°C, and post-fixed with 1% OsO4 (Sigma) in 0.1 buffered phosphate, pH 7.3 for 1 hr at room temperature. The cells were dehydrated in ethanol and dried by the critical-point method. Cultures were sputter-coated with gold (Electron microscopy sciences, Hatfield, PA). The methods were as previously described (Robards and Wilson, 1993). Specimens were observed using a Stereoscan 430i microscope (Leica, Houston, TX).
Flow Cytometric Analysis
For flow cytometric analysis, adherent cells were harvested from culture plates by gentle scraping on ice. Surface cell staining was performed at 4°C for 40 min by incubating 3 × 105 cells in 200 μl of PBS (containing 1% bovine serum albumin and 5% human plasma) with the indicated antibodies (Abs). The level of RANK surface expression was detected by indirect staining with a primary Ab anti-RANK (Imgenex, San Diego, CA), followed by PE-conjugated antimouse secondary Ab (Immunotech, Marseille, France). Nonspecific fluorescence was assessed using normal mouse immunoglobulin G (IgG; Immunotech) followed by secondary Ab (Beckman Coulter, Marseille, France). Expression of monocytic/macrophagic markers was documented by using PE- or fluorescein isothiocyanate–conjugated anti-CD14, CD15, and CD11b (Immunotech), as previously described (Secchiero et al., 2002). Nonspecific fluorescence was assessed by incubation with irrelevant isotype-matched conjugated mAb (Immunotech). Flow cytometry analysis was performed by FACScan (Becton Dickinson; San Jose, CA).
For each set of experiments, values are reported as means ± SD. The results were evaluated by using analysis of variance with subsequent comparisons by Student's t-test. Statistical significance was defined as P < 0.05.
Characterization of Osteoclastic Differentiation From Adherent Human PBMCs
In vitro osteoclastic differentiation was induced starting from enriched populations of adherent human PBMCs, expressing high levels of the CD14 and CD11b monocytic/macrophagic markers (Fig. 1A). After 12 days of culture with RANKL plus M-CSF, the levels of surface CD14 progressively decreased (Fig. 1B). On the other hand, the levels of surface RANK, detectable already at the beginning of the culture, did not show significant differences (Fig. 1C). This finding is consistent with its role in mediating differentiation of osteoclast precursors and activation of mature osteoclasts (Boyle et al., 2003). The addition in culture of M-CSF plus RANKL for 12 days induced a progressive increase in the number of multinucleated TRAP-positive cells (Fig. 2). These cells possessed major characteristics of functional mature osteoclasts, as demonstrated by the appearance in culture of several resorption pits in plates coated with carbonated calcium phosphate (Fig. 3A) and by the presence of the ruffled border observed by electron microscopy (Fig. 3B). In parallel, the culture of CD14+ monocytes with M-CSF alone promoted the differentiation along the macrophagic lineage. As expected, no multinucleated TRAP-positive cells became apparent in culture treated with M-CSF alone (Fig. 2).
Analysis of Gene Expression Profile in Osteoclastic Cells Versus Macrophages
To identify genes potentially involved in osteoclastic differentiation, we have analyzed the gene expression profile of adherent CD14+ monocytes, cultured with M-CSF plus RANKL for 12 days, in comparison to CD14+-derived monocytes obtained from the same donors, cultured with M-CSF alone. For this purpose, RNA was extracted from cells after 12 days of culture and it was analyzed by cDNA microarray for a panel of genes, which can be subgrouped according to their cellular function, as follows (1) cytokines/chemokines and molecules related to the NF-kB signaling pathway (Table 1); (2) apoptosis related proteins (Table 2), and (3) cell cycle-related proteins (Table 3).
Table 1. Proteins related to NF-kB signaling, cytokines, and chemokinesa
M-CSF + RANKL/M-CSF
ns, genes with no signal; M-CSF, macrophage-colony stimulating factor; RANKL, receptor activator of nuclear factor kappa B ligand. Data presented in boldface type indicate genes significantly up-modulated (above the cutoff of 2-fold induction) or down-modulated (below the cutoff of 0.5-fold of reduction).
NFK light polypeptide gene enhancer in B-cells 1 (p105)
Tumor necrosis factor receptor superfamily, member 11A
Tumor necrosis factor receptor superfamily, member 9
Interleukin 1 beta
Macrophage inflammatory protein1-alpha
Small inducible cytokine A4 (homologous to mouse Mip-1b)
Homo sapiens prostaglandin-endoperoxide synthase 2
Table 2. Apoptosis related proteins
M-CSF + RANKL/M-CSF
ans, genes with no signal; M-CSF, macrophage-colony stimulating factor; RANKL, receptor activator of nuclear factor kappa B ligand. Data presented in boldface type indicate genes significantly up-modulated (above the cutoff of 2-fold induction) or down-modulated (below the cutoff of 0.5-fold of reduction).
Tumor necrosis factor receptor superfamily, member 10d
TRAIL R4/ DcR2
Tumor necrosis factor receptor superfamily, member 21
Tumor necrosis factor (ligand) superfamily, member 8
Table 3. Cell cycle-related proteins
M-CSF + RANKL/M-CSF
aM-CSF, macrophage-colony stimulating factor; RANKL, receptor activator of nuclear factor kappa B ligand. Data presented in boldface type indicate genes significantly up-modulated (above the cutoff of 2-fold induction) or down-modulated (below the cutoff of 0.5-fold of reduction).
Transformed 3T3 cell double minute 2, p53 binding protein
Antigen identified by monoclonal antibody Ki-67
S-phase kinase-associated protein 2 (p45)
The results of the comparison between the gene profile of fully differentiated osteoclasts, obtained after 12 days of treatment with M-CSF plus RANKL and the gene-profile of macrophages differentiated by the addition of M-CSF alone are shown in Figure 4. Among the group of cytokines/chemokines, the only noticeable difference concerned the up-regulation of CCL3, also known as MIP1-α, in osteoclastic cultures compared with the macrophage cultures (Fig. 4A; Table 1). In this respect, it has been recently demonstrated that the direct production of the chemokine MIP1-α by myeloma cells, in combination with the RANKL induction in bone marrow stromal cells, is critical for osteoclast activation that characterizes multiple myeloma patients (Oba et al., 2005). Another remarkable difference in gene expression between osteoclasts and macrophages was observed among the TRAF family of adaptor proteins, which are known to mediate the signaling by RANK (Wong et al., 1998). In this regard, we observed a robust activation of TRAF3 in the osteoclastic lineage (Fig. 4A; Table 1). This finding is noteworthy because different splice variants of TRAF3 have been shown to either stimulate or inhibit the NF-kB pathway (Van Eyndhoven et al., 1999; Takaori-Kondo et al., 2000), which plays a crucial role in osteoclastic differentiation (Boyle et al., 2003). Quite unexpectedly, TRAF6 was not up-regulated (Table 1) and this was not expected because TRAF6 has been shown to be functionally important for RANK-mediated osteoclast activation (Lomaga et al., 1999; Naito et al., 1999).
Among the group of genes involved in the control of cell survival/apoptosis, mature osteoclasts showed increased expression levels of both anti-apoptotic (Bcl2) and pro-apoptotic (CIDEB, PYCARD, and BAK1) genes (Da et al., 2006; Hemmati et al., 2006) (Fig. 4A; Table 2). These findings might constitute a basis to explain the apparent high susceptibility of mature osteoclasts to be induced into apoptosis. Moreover, mature osteoclasts showed an increased expression of BAG3 (Fig. 4A; Table 2), a stress gene involved in down-modulating cell adhesion and controlling cell motility (Kassis et al., 2006).
The most striking up-regulation in the genes that control the cell cycle was observed in cyclin D2, which showed approximately a 10-fold induction (Fig. 4B; Table 3). In addition to playing a noncatalytic role in regulating G1 progression, cyclin D family proteins have been shown to inhibit proliferation in chronic growth assays and to block cells from entering into S phase (Atadja et al., 1995). Cyclin A1 in complex with CDK2 peaks at the G1/S transition of the cell cycle and is required for entry into S-phase. Consistent with the notion that terminally differentiated osteoclasts are nondividing cells, cyclin A1 was down-modulated by RANKL (Fig. 4B; Table 3). Moreover, whereas CDK4 (Matsushime et al., 1992) was increased in cultures of fully differentiated osteoclasts with respect to mature macrophages (Fig. 4A; Table 3), cyclin A and the majority of cyclin dependent kinases were down-modulated or silenced: CDC2, CDC27 (required for cell division), CDC42, and MKI67 (Ki 67; Fig. 4B; Table 3).
The process leading to the formation of mature osteoclasts from circulating precursors involves a series of complex steps, including exit from the cell cycle, cell fusion, and acquisition of the specialized functions of mature osteoclasts (Boyle et al., 2003). Under physiological conditions, osteoclastic differentiation occurs in the presence of osteoblasts/stromal cells and strictly depends on their surface expression of RANKL and/or their capacity to release soluble RANKL. Previous studies have already investigated the gene expression profiling induced by RANKL in murine models of osteoclastic differentiation (Cappellen et al., 2002; Chaisson et al., 2004). In particular, it has been demonstrated that M-CSF induced genes necessary for a direct response to RANKL, such as RANK and genes related to the NF-kB pathway. On the other hand, RANKL induced 70 novel target genes, including chemokines such as RANTES; growth factors such as platelet derived growth factor-α and insulin-like growth factor-1; histamine; and α1A-adrenergic receptors. Alpha1A-adrenergic receptors appear to be primarily involved in the interaction with osteoblasts and immune cells.
Few data are available on the gene expression profile of human osteoclasts. Because conflicting data have been reported on the ability of RANKL to modulate cell cycle progression and survival/apoptosis of osteoclast precursors and mature osteoclasts (Okahashi et al., 2001; Bharti et al., 2004; Sankar et al., 2004; Wu et al., 2005), we have investigated a gene expression profile focused on genes involved in the control of cell cycle and cell survival/apoptosis in this study. As a model system, we chose adherent PBMCs cultured for 12 days either with M-CSF alone, which induces differentiation along the macrophagic lineage, or with M-CSF plus RANKL, which drives osteoclastic differentiation. Despite lacking a feeder-layer of osteoblasts or bone marrow stromal cells, we could demonstrate that monocyte-derived osteoclasts are functional. In fact, they displayed bone resorption activity and showed the specialized ruffled border, characteristics of active osteoclasts.
Of interest, RANKL up-regulated the expression of the chemokine CCL3/MIP1-α, which has recently been shown to promote osteoclastic activation in patients affected by multiple myeloma (Oba et al., 2005). Therefore, the ability of RANKL to up-regulate the expression of MIP1-α might represent a mechanism for amplifying the efficiency of osteoclastic differentiation and/or to promote their activation once the differentiation process is completed. On the other hand, it was surprising that TRAF3 rather than TRAF6 was markedly up-regulated in mature osteoclasts. In fact, TRAF6 has been shown to play an essential function in mediating RANKL signaling in preosteoclasts (Lomaga et al., 1999; Naito et al., 1999). A possible explanation for our present findings is that TRAF3 may become more important in the maintenance of osteoclastic differentiation rather than in its induction. In any case, the activation of chemokines and transcription factors observed in our model of human osteoclastogenesis is consistent with the previous study of Cappellen et al. (2002) performed in mouse bone marrow preosteoclasts.
Analysis of the genes involved in the control of cell survival/apoptosis showed that a longer-term exposure to RANKL induced the prevalent expression of pro-apoptotic genes. This finding is consistent with the findings that these cells are prone to apoptosis, which must be actively prevented likely by adhesion contacts or cell-to-cell contacts, which are not recapitulated in our in vitro model system. These findings are in line with a couple of recent studies showing that RANKL activates pro-apoptotic pathways in a murine preosteoclastic cell line (Bharti et al., 2004; Wu et al., 2005) as well as with a previous study of Lacey et al. (2000) who demonstrated that optimal survival of osteoclasts required the simultaneous presence of M-CSF and RANKL.
Analysis of the cell cycle-related genes clearly showed that a longer-term exposure to RANKL down-regulated cell cycle-related genes. Thus, although preosteoclasts undergo a transient phase of proliferation in response to RANKL (Nishikawa et al., 1998; Okahashi et al., 2001; Sankar et al., 2004; Kwak et al., 2005), their exit from the cell cycle represents an essential step in the progression of the osteoclastic differentiation pathway. It is also noteworthy that previous studies emphasized the role of a progressive rise of cyclin-dependent kinase inhibitors (CDKIs), such as p27KIP1 and p21CIP1, and a decrease of cyclin-dependent kinase 6 (CDK-6), in mediating the exit from the cell cycle, which anticipates osteoclastic differentiation (Okahashi et al., 2001; Sankar et al., 2004; Ogasawara et al., 2004; Kwak et al., 2005). A similar coordination between exit from the cell cycle and differentiation mediated by CDKIs is required in neuronal differentiation (Milani et al., 1993). On the other hand, our present findings strongly suggest the involvement of cyclin D2 up-regulation in mediating such biological effects of RANKL. In fact, cyclin D2 showed approximately a 10-fold induction in response to RANKL. In addition to playing a noncatalytic role in regulating G1 progression, cyclin D family proteins have been shown to inhibit proliferation in chronic growth assays and to block cells from entering into the S phase of the cell cycle (Atadja et al., 1995).