Fibroblast Growth Factor (FGF)-2 Directly Stimulates Mature Osteoclast Function through Activation of FGF Receptor 1 and p42/p44 MAP Kinase*

We previously reported that fibroblast growth factor-2 (FGF-2) acts not only on osteoblasts to stimulate osteoclastic bone resorption indirectly but also on mature osteoclasts directly. In this study, we investigated the mechanism of this direct action of FGF-2 on mature osteoclasts using mouse and rabbit osteoclast culture systems. FGF-2 stimulated pit formation resorbed by isolated rabbit osteoclasts moderately from low concentrations (≥10−12 m), whereas at high concentrations (≥10−9 m) it showed stimulation on pit formation resorbed by unfractionated bone cells very potently. FGF-2 (≥10−12 m) also increased cathepsin K and MMP-9 mRNA levels in mouse and rabbit osteoclasts. Among FGF receptors (FGFR1 to 4) only FGFR1 was detected on isolated mouse osteoclasts, whereas all FGFRs were identified on mouse osteoblasts. FGF-2 (≥10−12 m) up-regulated the phosphorylation of cellular proteins, including p42/p44 mitogen-activated protein (MAP) kinase, and increased the kinase activity of immunoprecipitated FGFR1 in mouse osteoclasts. The stimulation of FGF-2 on mouse and rabbit osteoclast functions was abrogated by PD-98059, a specific inhibitor of p42/p44 MAP kinase. These results strongly suggest that FGF-2 acts directly on mature osteoclasts through activation of FGFR1 and p42/p44 MAP kinase, causing the stimulation of bone resorption at physiological or pathological concentrations.

We previously reported that fibroblast growth factor-2 (FGF-2) acts not only on osteoblasts to stimulate osteoclastic bone resorption indirectly but also on mature osteoclasts directly. In this study, we investigated the mechanism of this direct action of FGF-2 on mature osteoclasts using mouse and rabbit osteoclast culture systems. FGF-2 stimulated pit formation resorbed by isolated rabbit osteoclasts moderately from low concentrations (>10 ؊12 M), whereas at high concentrations (>10 ؊9 M) it showed stimulation on pit formation resorbed by unfractionated bone cells very potently. FGF-2 (>10 ؊12 M) also increased cathepsin K and MMP-9 mRNA levels in mouse and rabbit osteoclasts. Among FGF receptors (FGFR1 to 4) only FGFR1 was detected on isolated mouse osteoclasts, whereas all FGFRs were identified on mouse osteoblasts. FGF-2 (>10 ؊12 M) upregulated the phosphorylation of cellular proteins, including p42/p44 mitogen-activated protein (MAP) kinase, and increased the kinase activity of immunoprecipitated FGFR1 in mouse osteoclasts. The stimulation of FGF-2 on mouse and rabbit osteoclast functions was abrogated by PD-98059, a specific inhibitor of p42/p44 MAP kinase. These results strongly suggest that FGF-2 acts directly on mature osteoclasts through activation of FGFR1 and p42/p44 MAP kinase, causing the stimulation of bone resorption at physiological or pathological concentrations.
Among many growth factors regulating bone metabolism, fibroblast growth factor-2 (FGF-2 or basic FGF) 1 is recognized as a potent mitogen for a variety of mesenchymal cells (1). Several genetic diseases with severe impairment of bone and cartilage formation, such as achondroplasia (2)(3)(4) and thanatophoric dysplasia type II (5), have recently been shown to be caused by mutations of FGF receptors (FGFRs). In bone tissues, FGF-2 is produced by cells of osteoblastic lineage, is accumulated in bone matrix, and acts as an autocrine/paracrine factor for bone cells (6 -10). We and others have reported that the exogenous application of FGF-2 has stimulatory effects on bone formation in several in vivo models as a pharmacological action of FGF-2 (11)(12)(13)(14)(15)(16)(17). On the other hand, in vitro studies revealed that high concentrations of FGF-2 (10 Ϫ9 -10 Ϫ8 M) stimulated osteoclastogenesis in bone marrow culture (18) and bone resorption in bone organ cultures (19,20). This stimulatory effect of FGF-2 on bone resorption is known to be mediated at least in part by cyclooxygenase-2 (COX-2) induction and prostaglandin production (18,20), which cause the expression of osteoclast differentiation factor (RANKL/ODF), a key membrane-associated molecule that regulates osteoclast differentiation, in osteoblastic cells (21). Other than this indirect action through the mediation of osteoblasts, we recently reported that FGF-2 acts directly on mature osteoclasts to stimulate bone resorption (22).
There are four structurally related high affinity receptors (FGFR1 to 4) belonging to receptor tyrosine kinases (RTKs) that have an intrinsic protein-tyrosine kinase activity and elicit tyrosine autophosphorylation of the receptor (23,24). Because it is located downstream of the autophosphorylation of FGFRs, mitogen-activated protein (MAP) kinase has been reported to be the major signaling pathway in neuronal and endothelial cells (25)(26)(27). In osteoblasts, MAP kinase activation followed by the autophosphorylation of FGFR1 and 2 is also involved in FGF-2 signaling (28,29). Osteoclastic bone resorption is regulated by two different steps: one is the recruitment and differentiation of osteoclasts and the other is the activation of mature osteoclast function. Although a number of signaling pathways through the mediation of osteoblasts for osteoclast differentiation have been clarified, little is known about the signaling to stimulate mature osteoclast function directly. A recent study of random sequence analysis of PCR-amplified cDNA clones identified 14 distinct kinase-related genes in purified rabbit mature osteoclasts (30). Eight of these genes were identified as RTKs: Tie, c-Kit, Fms, Met, Axl, Tyro3, INS-R, and FGFR1.
In this study, we investigated the molecular mechanism whereby FGF-2 stimulates mature osteoclast function using mouse and rabbit osteoclast culture systems. Studies on the signaling pathway were performed using isolated mouse osteoclasts; however, for those on the resorbing activity, isolated rabbit osteoclasts were used because mouse osteoclasts do not have enough potency to resorb bone after being isolated.

EXPERIMENTAL PROCEDURES
Materials-Neonatal, 5-week-old, and 8-week-old male ddY mice were purchased from the Shizuoka Laboratories Animal Center (Shizuoka, Japan). 10-day-old male Japanese white rabbits were purchased from Saitama Experimental Animal Co. (Saitama, Japan). Human recombinant FGF-2 was kindly provided by Kaken Pharmaceutical Co. Ltd. (Chiba, Japan) and NS-398 by Taisho Pharmaceutical Co. Ltd. (Tokyo, Japan). ␣-modified minimum essential medium (␣MEM) was purchased from Life Technologies, Inc. (Rockville, MD), and fetal bovine serum (FBS) was from the Cell Culture Laboratory (Cleveland, OH). Macrophage colony-stimulating factor (M-CSF) was from Austral Biologicals (San Ramon, CA). Bacterial collagenase, 1,25(OH) 2 vitamin D 3 , and ISOGEN were purchased from Wako Pure Chemicals Co. (Osaka, Japan), and dispase from Nitta Gelatin Co. (Osaka). Polyclonal rabbit antibody against phosphotyrosine was obtained from UBI (Lake Placid, NY), and monoclonal mouse antibody against p60 v-src (monoclonal antibody 327) was obtained from Oncogene Research Products (Cambridge, MA). This antibody recognizes specifically both p60 v-src and p60 c-src and has been used to determine the expression of p60 c-src in various primary cells and clonal cell lines. Polyclonal rabbit antibodies against mouse FGFR1 through 4 and nonimmune IgG, as well as blocking peptides for respective antibodies, were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Polyclonal rabbit antibodies against phospho-p44/42 MAP kinase, phospho-p38 MAP kinase, phospho-c-Jun N-terminal protein kinase (JNK), and 2Ј-amino-3Ј-methoxyflanone (PD-98059) were obtained from New England BioLabs, Inc. Resorbed Pit Formation Assay by Purified Mature Osteoclasts and Unfractionated Bone Cells from Rabbit Long Bones-Purified mature osteoclasts were prepared using 10-day-old rabbits as described previously (31). Briefly, long bones from 10-day-old rabbits were minced with scissors and agitated with a vortex mixer. An aliquot of unfractionated bone cells was seeded onto 0.24% collagen gel (Nitta Gelatin, Tokyo) coated on 100-mm tissue culture dishes and incubated. 2 h later, nonadherent cells were washed off and osteoclasts were then removed from the gels with 0.1% collagenase solution (Wako Pure Chemicals Co.). By staining with tartrate-resistant acid phosphatase (TRAP), we ascertained that more than 99% of isolated cells were pure osteoclasts. Purified osteoclasts (150 cells/well) or unfractionated bone cells (5 ϫ 10 4 cells/well) were cultured on a dentine slice placed in each well of 96-well dishes containing 0.1 ml of ␣MEM/5% FBS. FGF-2 (10 Ϫ11 M), NS-398 (1 M), PD-98059 (1, 3, 10, and 30 M), and SB-203580 (30 M) were added 1 h after the seeding. After 24 h of culture, cells on the dentine slices were removed with 1 N NH 4 OH solution, and stained with 0.5% toluidine blue for 1 min. The total area of pits was estimated under a light microscope with a micrometer to assess osteoclastic bone resorption using an image analyzer (System Supply Co., Nagano, Japan).
Isolation of Rabbit Osteoclasts on Plastic Dishes-Rabbit osteoclasts were isolated on plastic dishes as described previously (32). Briefly, the above unfractionated bone cells from rabbit long bones were plated on 100-mm tissue culture dishes at 1 ϫ 10 8 living cells per dish. After an overnight culture, the cells were incubated with 0.001% Pronase and 0.02% EDTA in PBS for 10 min at room temperature. By this incubation, cells other than osteoclasts became detached from the dishes and were washed off. After the Pronase E and EDTA treatment, more than 99% of the adherent cells were ascertained to be osteoclasts by TRAPpositive results and multinucleated. After purification, osteoclasts were further incubated for 2 h with ␣MEM containing 5% FBS.
Isolation of Mouse Osteoclasts on Plastic Dishes-Osteoclasts were prepared using ddY mice as described previously (33). Briefly, primary mouse osteoblastic cells were prepared from calvariae of neonatal ddY mice and bone marrow cells prepared from tibiae of 8-week-old male ddY mice. Osteoblastic cells (2 ϫ 10 6 cells/dish) and bone marrow cells (2 ϫ 10 7 cells/dish) were co-cultured in 100-mm tissue culture dishes containing ␣MEM with 10% FBS and 1,25(OH) 2 vitamin D 3 (10 Ϫ8 M) for 7 days with a medium change every 2 days. After 7 days of culture, 2-4 ϫ 10 4 osteoclasts/dish were usually yielded. The dishes were then treated with 0.001% Pronase and 0.02% EDTA in PBS for 10 min to remove osteoblastic cells. More than 99% of the adherent cells prepared were TRAP-positive and multinucleated. These cells were incubated for 2 h with ␣MEM containing 10% FBS.
Northern Blot Analysis for Cathepsin K, Matrix Metalloproteinase (MMP)-9, and MMP-14 -Total RNA from isolated mouse or rabbit osteoclasts was extracted using ISOGEN, and 5 g of total RNA was run on a 1.2% agarose-2.2 M formaldehyde gel, transferred to a nitrocellulose membrane by positive pressure, and fixed to the membrane by ultraviolet irradiation. After 1 h of prehybridization in GMC buffer (0.5 M Na 2 HPO 4 , 1% bovine serum albumin, 1 mM EDTA, and 7% SDS, pH 7.2) at 60°C, filters were hybridized overnight in GMC buffer at 65°C with a random primer [ 32 P]dCTP-labeled cDNA probe for cathepsin K, MMP-9, or MMP-14. A cDNA fragment from rabbit osteoclasts was used as a probe for cathepsin K (34). cDNA probes for MMP-9 and -14 were generated by RT-PCR with the total RNA from mouse osteoclasts. The primers for MMP-9 were: sense, 5Ј-CTGTCCAGACCAAGGGTA-CAGCCT-3Ј; antisense, 5Ј-GTGGTATAGTGGGACACATAGTGG-3Ј. The primers for MMP-14 were: sense, 5Ј-GAGATCAAGGCCAATGT-TCGGAGG-3Ј; antisense, 5Ј-TTAGATCCTCATTTTGGACAGTCC-3Ј. PCR consisted of 25 cycles of denaturation at 95°C for 30 s, annealing at 60°C for 45 s, and extension at 72°C for 60 s. The PCR products for MMP-9 and -14 were 263 bp and 382 bp, respectively. Filters were washed in 1ϫ SSC (0.15 M NaCl, 15 mM Na 3 citrate, pH 7.0)/0.1% SDS twice for 15 min at 65°C, then once for 15 min in 0.1ϫ SSC/0.1% SDS at 65°C. Signals were quantitated by densitometry (Bio-Rad Laboratories, Richmond, CA), and optical densities for cathepsin K, MMP-9, and MMP-14 were normalized to the corresponding values for G3PDH.
Analysis of Osteoclast Survival-To obtain viable mouse osteoclasts formed in the co-culture, a collagen gel culture was performed as described previously (32). Briefly, primary mouse osteoblastic cells and bone marrow cells were co-cultured on 100-mm culture dishes coated with 0.2% collagen gel matrix (Nitta Gelatin) in ␣MEM containing 10% FBS and 1,25(OH) 2 vitamin D 3 (10 Ϫ8 M) for 6 days, with a medium change every 2 days, and for 1 additional day in ␣MEM/10% FBS. After culture for 7 days, dishes were treated with 4 ml of 0.2% bacterial collagenase in ␣MEM for 20 min at 37°C. Cells released from the dishes were collected by centrifugation at 1000 rpm for 5 min and suspended in 5 ml of ␣MEM containing 10% FBS. An aliquot of crude osteoclast preparation (0.1 ml) was replaced in 24-well dishes, and further cultured. After incubation for 2 h, the plates were treated with 0.001% Pronase E and 0.02% EDTA for 10 min to remove osteoblastic cells. After purification, osteoclasts were cultured in the presence or absence of FGF-2 (10 Ϫ11 or 10 Ϫ8 M) or M-CSF (2000 units/ml) for various periods up to 48 h, fixed with citrate-acetone-formaldehyde fixative for 30 s, and stained with trypan blue and TRAP. Trypan blue-negative and TRAPpositive osteoclasts were counted.
Immunoprecipitation and Western Blot Analysis-Mouse osteoclasts and osteoblasts were washed twice with ice-cold PBS and lysed with TNE buffer (10 mM Tris-HCl, 150 mM NaCl, 1% Nonidet P-40, 1 mM EDTA, 10 mM NaF, 2 mM Na 3 VO 4 , 1 mM aminoethyl-benzenesulfonyl fluoride, and 10 g/ml aprotinin). The protein concentration in the cell lysate was measured using a Protein Assay Kit II (Bio-Rad). Immunoprecipitation was performed using antibodies either noncovalently bound or conjugated to protein G-Sepharose (Life Technologies, Inc.). Equivalent amounts (100 g) of cell lysates were incubated with coupled antibody for 4 h at 4°C, and the beads were washed three times with a lysis buffer and boiled in 3ϫ Laemmli sample buffer before electrophoresis. Each immunoprecipitant was electrophoresed by 8% SDS-PAGE, and transferred to a nitrocellulose membrane. After blockage of nonspecific binding with 5% skim milk, membranes were incu-bated with polyclonal anti-mouse FGFR1, 2, 3, and 4 or nonimmune IgG. Immunoreactive bands were stained using the ECL chemiluminescence reaction (Amersham Pharmacia Biotech) following the manufacturer's instructions. After this visualizing, the antibodies on the membrane were stripped in a buffer consisting of 62.5 mM Tris-HCl (pH 6.7), 2% SDS, and 100 mM 2-mercaptoethanol at 50°C for 40 min. To ascertain the specificity of these blots, the stripping membrane was further immunoreacted with each polyclonal anti-FGFR and respective blocking peptide, and the immunoreactive bands were again visualized under the same conditions as above. The immunoreactivity to each anti-FGFR was not lost by this stripping procedure.
Assay for Tyrosine Phosphorylation of Cellular Proteins-Osteoblastic cells and bone marrow cells were co-cultured on 100-mm tissue culture dishes in ␣MEM containing 10% FBS and 1,25(OH) 2 vitamin D 3 (10 Ϫ8 M) for 6 days, with a medium change at 2 days, and then for 1 more day in ␣MEM/0.1% FBS. After mouse and rabbit osteoclasts were isolated as described above, they were precultured for 2 h with ␣MEM/ 0.1% FBS and treated with FGF-2 (10 Ϫ12 M) for various periods (2-30 min). The cells were quickly washed twice with ice-cold PBS and lysed with TNE buffer. Cell lysates containing equal amounts of protein (10 g) were subjected to 8% SDS-PAGE, and proteins separated in the gel were subsequently electrotransferred onto nitrocellulose membranes. After blocking with 5% bovine serum albumin, the membranes with mouse osteoclast lysates were incubated with polyclonal rabbit antibody against phosphotyrosine (UBI, Lake Placid, NY) and subsequently with peroxidase-conjugated anti-rabbit IgG antibody. Phosphotyrosinecontaining proteins were visualized using the ECL chemiluminescence reaction following the manufacture's instructions. After the antibody was stripped from the membrane, to block nonspecific binding, membranes were incubated with 5% skim milk and then with monoclonal mouse antibody against p60 v-src , polyclonal rabbit antibodies against phospho-p44/42 MAP kinase, -JNK, and -p38 MAP kinase, and the immunoreactive bands were visualized as described above.
In Vitro Kinase Assay-Isolated mouse osteoclasts were incubated with and without FGF-2 (10 Ϫ12 M) for various periods (1-10 min). The cells were quickly washed twice with ice-cold PBS and lysed with TNE buffer, and equal amounts of protein (100 g) were immunoprecipitated with 1 g of polyclonal rabbit anti-FGFR1. The immune complex was washed three times with TNE buffer and three times with kinase buffer (20 mM HEPES-NaOH (pH 7.4), and 10 mM MgCl 2 ); the samples were then resuspended in 60 l of kinase buffer with 1 Ci (37 kBq) of [␥-32 P]ATP and incubated for 15 min at 30°C. The reaction was stopped by adding 20 l of 4ϫ sample buffer (250 mM Tris-HCl (pH 6.8), 8 mM EDTA, 12% SDS, 500 mM 2-mercaptoethanol, 15% glycerol, and 0.01% bromphenol blue) and subjected to 10% SDS-PAGE under reducing conditions followed by autoradiography.
Statistical Analysis-Means of groups were compared by ANOVA and significance of differences was determined by post-hoc testing using Bonferroni's method.

Direct and Indirect Effects of FGF-2 on Isolated Rabbit
Osteoclast Function-To examine the direct action of FGF-2 on osteoclasts, the effect of FGF-2 on resorbed pit formation on a dentine slice by purified osteoclasts was compared with that by unfractionated bone cells from rabbit long bones (Fig. 1). FGF-2 at 10 Ϫ12 -10 Ϫ8 M stimulated resorbed pit formation by isolated mature osteoclasts with a maximal effect of 1.9-fold at 10 Ϫ11 M, and no further stimulations were observed at higher concentrations (Fig. 1A). This stimulation was not due to the increase in the number of osteoclasts but to the activation of each osteoclast function, because the area of each pit (the total pit area per number of pits) was similarly increased by FGF-2 (Ն10 Ϫ12 M, data not shown). Because previous reports have shown that the bone resorptive effect of FGF-2 is mediated at least in part by COX-2 induction in osteoblastic cells (18, 20 -22), the contribution of COX-2 to the direct action was examined by adding NS-398 (1 M), a specific inhibitor of COX-2, to the culture of isolated osteoclasts. NS-398 did not alter the FGF-2 action on isolated osteoclasts, indicating that the direct action is not mediated by COX-2 induction or by PG production (Fig. 1A). On the other hand, FGF-2 at 10 Ϫ9 and 10 Ϫ8 M further stimulated resorbed pit formation by unfractionated bone cells up to 7.5-fold (Fig. 1B). This stimulatory effect of high concen-trations of FGF-2 on pit formation by unfractionated bone cells was 70 -80% inhibited by NS-398. Because we have previously shown that COX-2 is expressed in cells of osteoblastic lineage but not in cells of osteoclastic lineage (22), the target cells of FGF-2 to induce COX-2 are those of osteoblastic lineage as reported previously (20,21). These results confirm our previous report (22) and suggest that FGF-2 at low concentrations (Ն10 Ϫ12 M) moderately stimulates bone resorption through its direct action on osteoclasts, whereas at high concentrations (Ն10 Ϫ9 M) it potently stimulates bone resorption through its indirect action mediated by COX-2 induction in osteoblastic cells.
Effects of FGF-2 on mRNA Levels of Proteinases in Isolated Mouse and Rabbit Osteoclasts-Because mouse osteoclasts do not have enough potency to resorb bone after being isolated, functional analysis was carried out by measuring mRNA levels of proteinases, which have been reported to be produced by osteoclasts (Fig. 2, upper panel). Cathepsin K is an osteoclastselective cysteine proteinase that plays a key role in matrix degradation during bone resorption (34 -37). Among MMPs, important proteinases of matrix degradation, MMP-9 and MMP-14 (MT1-MMP) have also been reported to be produced by osteoclasts (38 -41). Northern blot analysis of the dose response of FGF-2 revealed that at 10 Ϫ12 -10 Ϫ8 M it stimulated steady-state mRNA levels of cathepsin K and MMP-9, but not that of MMP-14, at 3 h of culture. Similar regulation of cathepsin K and MMP-9 mRNA levels were seen in isolated rabbit osteoclasts, although FGF-2 effects were not as strong as those

FIG. 1. Dose response of effects of FGF-2 on resorbed pit formation by isolated osteoclasts (A) and unfractionated bone cells (B) from rabbit long bones in the presence or absence of NS-398.
Bone cells were extracted from long bones of 10-day-old rabbits and were seeded onto collagen gel. Nonadherent cells were washed off, and osteoclasts were then removed from the gels. Purified osteoclasts (Ͼ99% in purity, 150 cells/well) or unfractionated bone cells (5 ϫ 10 4 cells/well) were cultured on a dentine slice in the presence or absence of FGF-2 (10 Ϫ16 -10 Ϫ8 M) and/or NS-398 (1 M). After 24 h of culture, the total area of pits on the dentine slice was measured. Data are expressed as means (symbols) Ϯ S.E. (error bars) for 8 cultures/group. *p Ͻ 0.01, significant stimulation by FGF-2; #p Ͻ 0.01, significant inhibition by NS-398. on mouse osteoclasts (Fig. 2, lower panel). This may be because the basal expression levels of these proteinases in the control cultures were much higher in rabbit osteoclasts than in mouse osteoclasts. In both cultures, the maximum stimulation was seen at 10 Ϫ11 or 10 Ϫ10 M, and no further stimulation was observed at higher concentrations, showing a good correspondence with the effect of FGF-2 on pit formation by isolated rabbit osteoclasts (Fig. 1A).
Effects of FGF-2 on the Survival of Isolated Mouse Osteoclasts-To investigate the effect of FGF-2 (10 Ϫ11 M) on their survival, isolated mouse osteoclasts were cultured in a plastic dish for up to 48 h (Fig. 3). The survival rates decreased with time similarly in the control and FGF-2-treated cultures. At 24 h 27% and 32% of initially surviving cells still adhered to the dish in control and FGF-2-treated cultures, respectively, and by 48 h all cells had died in both cultures. Similar results were seen when a higher concentration of FGF-2 (10 Ϫ8 M) was used (data not shown). On the contrary, M-CSF (2000 units/ml), a positive control, maintained the survival of osteoclasts: the survival rates were 76% at 24 h and 21% at 48 h, as reported previously (42).

FGF Receptors (FGFR1-4) on Mouse Osteoclasts and Osteoblasts-
The molecular mechanism of the signal transduction of FGF-2 in osteoclasts was further investigated using isolated mouse osteoclasts. mRNA and protein levels of FGFRs on osteoclasts were studied and compared with those on osteoblasts from neonatal mouse calvariae by RT-PCR and Western blotting analyses, respectively. Only FGFR1 was detected on osteoclasts, whereas all FGFR1 through 4 were identified on osteoblasts both in mRNA and protein levels (Fig. 4). This difference in distribution of FGFRs between osteoclasts and osteoblasts might explain the difference of affinities and concentrations of FGF-2 affecting these cells as seen in bone resorptive activity in rabbit cell cultures (Fig. 1, A and B). Fig. 5A shows the time course of effects of FGF-2 on tyrosine phosphorylation of cellular proteins in isolated mouse osteoclasts. Several proteins were selectively phosphorylated by FGF-2 (10 Ϫ12 M) as early as 2 min.

Phosphorylation of FGFR1 and Intracellular Proteins in Mouse and Rabbit Osteoclasts-
The c-Src signal in each lane indicates a quantitative internal control. Western blot analyses using antibodies against specific proteins related to MAP kinase revealed that phosphorylation of p42/p44 MAP kinase was induced at 5 min, reached maximum at 10 min, and was maintained for more than 30 min (Fig.  5A). Phosphorylations of p38 and JNK MAP kinases were slightly induced just at 10 min. To investigate the autophosphorylation of FGFR1 by FGF-2, kinase activity of immunoprecipitated FGFR1 was examined by in vitro kinase assay. FGF-2 induced the kinase activity of FGFR1 at 1 min, which reached maximum at 2 min, and decreased considerably after 10 min (Fig. 5B). Similar regulation of tyrosine phosphorylation of cellular proteins by FGF-2 was observed in isolated rabbit osteoclasts, and intracellular proteins were phosphorylated at 2 min (Fig. 5C). However, further studies on signaling molecules in rabbit osteoclasts could not be carried out, because antibodies against rabbit proteins were not available.
Functional Relevance of MAP Kinase Activation in Rabbit and Mouse Osteoclasts-To examine the functional relevance of Mouse osteoclasts were isolated from the co-culture of mouse osteoblastic cells and bone marrow cells, and rabbit osteoclasts were from 10-day-old rabbit long bones. More than 99% of isolated mouse and rabbit cells were ascertained to be osteoclasts by TRAP staining. After incubation for 2 h, cells were cultured in the presence or absence of FGF-2 (10 Ϫ14 -10 Ϫ8 M) for 3 h. Steady-state mRNA levels were examined by Northern blot analysis. The number under each band is the treated/control ratio of the intensity of each band normalized to that of G3PDH measured by densitometry.  4. mRNA and protein levels of FGF receptors (FGFR1-4) on mouse osteoblasts (OB) and osteoclasts (OC). Steady-state mRNA was analyzed by RT-PCR. Total RNA was extracted from isolated mouse osteoclasts and neonatal mouse calvarial osteoblasts as described under "Experimental Procedures." The PCR products for FGFR1, 2, 3, and 4 were 856, 373, 635, and 550 bp, respectively. Similar results were obtained in four other separate experiments, and the increases in template amounts or cycles did not reveal the expressions of FGFR2-4 in osteoclasts. Protein levels were analyzed by immunoprecipitation and immunoblotting with antibodies against FGFR1-4 as well as nonimmune IgG. Cellular proteins extracted with TNE buffer were immunoprecipitated, subjected to SDS-PAGE, and immunoblotted with polyclonal anti-mouse FGFR antibodies or nonimmune IgG as described under "Experimental Procedures." To confirm the specificity of these blots, stripped membranes were immunoreacted with each polyclonal anti-FGFR and respective blocking peptide. the activation of p42/p44 and p38 MAP kinases by FGF-2 in osteoclasts, PD-98059, a specific inhibitor of the upstream kinase of p42/p44 MAP kinase (43,44), and SB-203580, a specific inhibitor of p38 MAP kinase (45,46), were added to the cultures of rabbit and mouse osteoclasts. PD-98059 dose dependently inhibited the stimulation of FGF-2 on pit formation resorbed by isolated rabbit osteoclasts to the levels of the control culture, while SB-203580 (30 M) did not affect the FGF-2 stimulation (Fig. 6A). PD-98059 also inhibited the FGF-2 stimulation on cathepsin K and MMP-9 mRNA levels in isolated mouse osteoclasts, and this inhibition was not seen by SB-203580 nor NS-398 (Fig. 6B). Although PD-98059 at the highest concentration (30 M) did not decrease the resorbed pit formation or proteinase mRNA levels in the control culture, inhibitors of src kinase, herbimysin (1 M) and PP-1 (10 M), abrogated both of these osteoclast functions not only in FGF-2-stimulated cultures but also in control cultures (data not shown), suggesting the essential role of src kinase signaling in the basal function of osteoclasts. DISCUSSION In the present study, we confirmed our previous report that FGF-2 directly stimulated the bone resorptive activity of rabbit osteoclasts and further demonstrated that it induced the expression of proteinases in mouse and rabbit osteoclasts. These actions were mediated by the autophosphorylation of FGFR1, the only subtype of FGFRs expressed on osteoclasts, and the subsequent phosphorylation of cellular proteins, including p42/ p44 MAP kinase.
Although it is ideal to use a single culture system for functional and molecular analyses, two different osteoclast cultures were employed in this study: one is the culture of isolated rabbit osteoclasts and the other is that of mouse osteoclasts. The isolated rabbit osteoclasts are capable of resorbing dentine and maintaining their survival on dentine slices even in the absence of bone-derived osteoblastic/stromal cells (31).

FIG. 5. Effects of FGF-2 on phosphorylation of cellular proteins and autophosphorylation of FGFR1 in isolated osteoclasts.
A, Western blotting of phosphotyrosine proteins and MAP kinase-related proteins in isolated mouse osteoclasts. Mouse osteoclasts formed in the co-culture were cultured with and without FGF-2 (10 Ϫ12 M) for various periods (2-30 min) and lysed with TNE buffer. 10 g of cell lysates was subjected to 8% SDS-PAGE and immunoblotted with polyclonal antibodies against phosphotyrosine, p60 v-src , phospho-p44/42 MAP kinase, phospho-p38 MAP kinase, and phospho-JNK as described under "Experimental Procedures." B, tyrosine kinase activity of immunoprecipitated FGFR1 in isolated mouse osteoclasts. Isolated mouse osteoclasts were cultured with and without FGF-2 (10 Ϫ12 M) for various periods (1-10 min) and lysed with TNE buffer, and 100 g of cell lysates was immunoprecipitated with polyclonal anti-FGFR1 antibody. The samples were incubated in kinase buffer with [␥-32 P]ATP and subjected to SDS-PAGE. C, Western blotting of phosphotyrosine proteins in isolated rabbit osteoclasts. Rabbit osteoclasts were cultured with and without FGF-2 (10 Ϫ12 M) for 2 min and lysed with TNE buffer. 10 g of cell lysates was subjected to 8% SDS-PAGE and immunoblotted with polyclonal antibodies against phosphotyrosine. Thereby, direct actions of osteotropic hormones and local factors on mature osteoclasts in vitro can be precisely estimated without the influence of nonosteoclastic cells. However, the lack of molecular information about nucleotide and protein sequences expressed in rabbits and antibody availability for rabbit proteins restricts studies in the rabbit osteoclast culture system. To overcome this disadvantage, the mouse osteoclast culture system may aid researchers, because much molecular information has already been accumulated. Isolated mouse osteoclasts, on the other hand, essentially require the presence of bone-derived osteoblastic/stromal cells for their bone resorbing activity and survival. Indeed, as shown in Fig. 3, only a part of isolated mouse osteoclasts remained alive after 24 h of culture. Recently, mouse osteoclasts formed from cultured bone marrow cells in the presence of M-CSF: M-CSF-dependent bone marrow macrophages (M-BMM) (47) and M-CSF-dependent bone marrow cells (MDBM cells) (48), have been reported to exhibit the potency to resorb the dentine slice even in the absence of osteoblastic/stromal cells. However, tumor necrosis factor-␣ (TNF-␣) or RANKL/ODF, in addition to M-CSF, is essential for these mouse osteoclasts to form resorbed pits. Although our preliminary studies revealed that FGF-2 (10 Ϫ11 M) increased the resorbed pit formation by mouse osteoclasts both in M-BMM and MDBM cell cultures in the presence of TNF-␣ and M-CSF, the effects in both cultures were weaker (1.3-to 1.4-fold over the culture with TNF-␣ and M-CSF alone) than seen in the rabbit osteoclast culture. Because both M-BMM and MDBM cell cultures still contain osteoclast precursors in much higher concentrations than the rabbit osteoclast culture, we cannot deny the possibility that these stimulations by FGF-2 may not be due to the direct action on mature osteoclast function but to the action on osteoclast differentiation. In addition, FGF-2 might affect the signaling of M-CSF through some cross-talk mechanism, because both their receptors, FGFR1 and Fms, are RTKs. In fact, we have reported that the direct action of FGF-2 on osteoclast precursor differentiation was inhibitory and that the tyrosine phosphorylation of several cellular proteins induced by M-CSF was inhibited by FGF-2 using the osteoclast precursor cell line C7 cell culture (49). Hence, these mouse osteoclast culture systems appear not to be suitable for this study that investigated the direct action of FGF-2 on mature osteoclasts. Given the above circumstances, we properly used the rabbit and mouse osteoclast cultures to study the function of FGF-2 on bone resorbing activity and its molecular mechanism, respectively.
Another issue is the relationship between the stimulation of pit formation and the up-regulation of proteinases. Because a previous report showed that cathepsin K antisense oligodeoxynucleotide inhibited the resorbed pit formation by isolated rabbit osteoclasts using the same system as this study (36), the induction of cathepsin K in osteoclasts is likely to contribute to the FGF-2 stimulation of osteoclastic bone resorption. On the other hand, BB94, a nonselective MMP inhibitor (50), did not affect the FGF-2 stimulation on resorbed pit formation on the dentine slice by rabbit osteoclasts but inhibited that on the dentine slice coated with collagen. 2 These results indicate that MMPs are important for the migration of osteoclasts through the unmineralized osteoid to reach the mineralized bone surface, but not for the bone resorbing activity of osteoclasts as previously reported (41).
Signaling pathways through RTKs on osteoclasts were studied on M-CSF, which stimulates motility and cytoplasmic spreading in osteoclasts (51). FGF-2 and M-CSF, although receptors of both are RTKs expressed on osteoclasts (30), showed different actions on osteoclast function. FGF-2 did not maintain the survival of osteoclasts but M-CSF did (Fig. 3). Contrarily, M-CSF itself did not stimulate resorbing activity of osteoclasts but FGF-2 did. Regarding signal transduction, M-CSF induced the autophosphorylation of its receptor, Fms, and c-src-dependent tyrosine phosphorylation of selected proteins, including Grb2-binding protein. c-Src, a ubiquitous cellular tyrosine kinase, which is highly expressed in osteoclasts, is essential for osteoclasts to form a ruffled border and to resorb bone (52), and the contribution of c-src kinase to FGF-2 signaling has been suggested in endothelial cells and fibroblasts (27,53,54). In this study, inhibitors of the src family kinases, herbimysin and PP1, abrogated the osteoclast function in control cultures as well as in FGF-2-stimulated cultures. Hence, we assume that the src kinase signal may be essential for the basal osteoclast function, whereas p42/p44 MAP kinase is the major pathway for the FGF-2 action. To our knowledge, this study is the first indicating that the activation of p42/p44 MAP kinase causes the stimulation of osteoclast function.
Regarding the physiological relevance of the direct action of FGF-2 on osteoclasts, we recently reported that endogenous FGF-2 in the synovial fluid contributes to joint destruction in rheumatoid arthritis patients (55). The concentration of FGF-2 in the synovial fluid was positively correlated to the severity of joint destruction in these patients. However, the concentrations of FGF-2 were lower, on the order of 10 Ϫ13 -10 Ϫ11 M, than other cytokines such as interleukin-6 and soluble interleukin-6 receptor, on the order of 10 Ϫ11 -10 Ϫ10 M. These levels of FGF-2 are not enough to induce COX-2 in osteoblastic cells (18 -22) but possibly affect mature osteoclasts directly. Although these effects are small compared with the COX-2-mediated effects, they occur at a concentration of FGF-2 that is likely to be important in vivo. Thus, FGF-2 in the synovial fluid might play a role in the final step of osteoclastic bone resorption in rheumatoid arthritis joint destruction that is preceded by recruitment and differentiation of osteoclasts by other factors. Other than the well-known pharmacological action of FGF-2 on bone formation, endogenous FGF-2 might function in the pathogenesis of bone resorptive diseases through its direct action on osteoclasts. Further studies will reveal the contribution of FGF-2 to the pathophysiology of osteopenic disorders like rheumatoid arthritis.