Impaired osteoclast formation in bone marrow cultures of Fgf2 null mice in response to parathyroid hormone.

Fibroblast growth factor (FGF)-2 and parathyroid hormone (PTH) are potent inducers of osteoclast (OCL) formation, and PTH increases FGF-2 mRNA and protein expression in osteoblasts. To elucidate the role of endogenous FGF-2 in PTH responses, we examined PTH-induced OCL formation in bone marrow cultures from wild type and mice with a disruption of the Fgf2 gene. FGF-2-induced OCL formation was similar in marrow culture from both genotypes. In contrast, PTH-stimulated OCL formation in bone marrow cultures or co-cultures of osteoblast-spleen cells from Fgf2-/mice was significantly impaired. PTH increased RANKL mRNA expression in osteoblasts cultures from both genotypes. After 6 days of treatment, osteoprotegerin protein in cell supernatants was 40-fold higher in vehicle-treated and 30-fold higher in PTH-treated co-cultures of osteoblast and spleen cells from Fgf2-/mice compared with Fgf2+/+ mice. However, a neutralizing antibody to osteoprotegerin did not rescue reduced OCL formation in response to PTH. Injection of PTH caused hypercalcemia in Fgf2+/+ but not Fgf2-/mice. We conclude that PTH stimulates OCL formation and bone resorption in mice in part by endogenous FGF-2 synthesis by osteoblasts. Because RANKL- and interleukin-11-induced OCL formation was also reduced in bone marrow cultures from Fgf2-/mice, we further conclude that endogenous FGF-2 is necessary for maximal OCL formation by multiple bone resorbing factors.

Bone is a complex organ that has multiple functions in the body. It provides a structural framework, is a storehouse for calcium, and is the site of hematopoiesis. To maintain these functions, bone constantly remodels. Bone resorption is accomplished by osteoclasts, which remove bone, whereas osteoblasts are the cells that form bone (1). Basic fibroblast growth factor (bFGF or FGF-2), 1 a potent stimulator of osteoblast replication (2), is produced by osteoblasts (3,4) and stored in the extracellular matrix (5). We reported previously that FGF-2 treatment increases bone resorption in mouse calvarial cultures (6) and increases osteoclast formation in murine bone marrow cultures (7). However, the physiologic role of endogenous FGF-2 in osteoclast formation has not been studied.
Osteoclasts are multinucleated giant cells that are capable of removing both the mineral and organic component of bone (8). Osteoclasts, granulocytes, and macrophages are believed to be derived from a common hematopoietic progenitor cell (9), and studies have shown that production of osteoclasts from progenitor cells is regulated by growth factors, cytokines, and hormones (7 ,10 -15). Osteoclast differentiation is enhanced by interactions between marrow progenitor cells and either mesenchymal stromal cells (16) or osteoblastic cells (17). FGF-2, which is also produced by stromal cells (18), could be important in osteoclast formation in response to hormones and cytokines.
Parathyroid hormone (PTH) is a potent bone resorbing agent (19) and inducer of osteoclast formation that plays a central role in maintaining normal serum calcium (20). In recent studies, we have shown that PTH increases Fgf2 gene expression in murine calvarial organ cultures and in murine osteoblastic cells (21). We hypothesized that endogenous FGF-2 may be important in PTH-induced osteoclast formation. Thus, the development of mice with disruption of the Fgf2 gene allowed us to assess the role of endogenous FGF-2 in PTH-induced osteoclastogenesis. Fgf2Ϫ/Ϫ mice were developed in a Black Swiss x 129Sv background (22). These mice are viable and reproduce. Northern blot analysis of tissues from the Fgf2Ϫ/Ϫ mice revealed no wild type or truncated Fgf2 mRNA. Western blot analysis showed that none of the three isoforms of FGF-2 protein were expressed in the Fgf2Ϫ/Ϫ mice. These mice were found to have decreased vascular smooth muscle contractility, low blood pressure, thrombocytosis (22), neurologic deficits (23,24), and delayed wound healing (24). In recent studies, we have shown that disruption of the Fgf2 gene also results in decreased bone mass and bone formation (25).
In this report we assessed the role of endogenous FGF-2 in PTH-induced osteoclast formation by comparing the ability of PTH to stimulate the formation of OCL from precursor cells in bone marrow cultures and co-culture of osteoblast and spleen cells from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice. We show that the ability of PTH to increase OCL formation in vitro, to resorb bone, and to increase serum calcium in vivo is markedly impaired in Fgf2Ϫ/Ϫ mice; we therefore conclude that endogenous FGF-2 is an important local factor in PTH-induced OCL formation and bone resorption in mice. Because OCL formation was also markedly reduced in RANKL-and IL-11-treated marrow stromal cultures from Fgf2Ϫ/Ϫ mice, we conclude that endogenous FGF-2 is important for maximal OCL formation in response to PTH, as well as other stimulators of OCL formation. Because the ability of PTH to regulate OPG mRNA and protein expression in osteoblasts from Fgf2Ϫ/Ϫ mice was also impaired, but OCL formation by PTH was still impaired in the presence of a neutralizing antibody to OPG, we further conclude that other signaling pathways are necessary for maximal OCL formation in the Fgf2Ϫ/Ϫ mice.

EXPERIMENTAL PROCEDURES
Mice were bred and housed in the transgenic facility in the Center for Laboratory Animal Care at the University of Connecticut Health Center. Mice were sacrificed by CO 2 narcosis and cervical dislocation. The Animal Care Committee of the University of Connecticut Health Center approved all animal protocols.
Mouse Bone Marrow Cultures-Mouse bone marrow cells were isolated by a modification of previously published methods (7). Tibiae and femurs from 8 -12-week-old littermate Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice were dissected free of adhering tissue. Bone ends were removed, and the marrow cavity was flushed with ␣ minimal essential medium (␣MEM; Invitrogen) without serum by slowly injecting medium into one end of the bone using a sterile 25-gauge needle. Marrow cells were collected into tubes, washed twice with ␣MEM, and cultured in ␣MEM containing 10% heat-inactivated fetal calf serum (HIFCS; Invitrogen). Cells were plated in 24-multiwell plates (1 ϫ 10 6 ) cells/well. Cultures were fed every 3 days by replacing 80% of the medium with fresh medium. Effectors were added at the beginning of the culture and with each medium change. Cells for tartrate-resistant acid phosphatase (TRAP) staining were fixed on day 7 of culture with 2.5% glutaraldehyde for 30 min. TRAP staining was performed with a commercial kit (Sigma). OCL were defined as TRAP positive multinucleated cells that contained greater than three nuclei. In some experiments, we determined whether PTH-induced OCL formation in marrow cultures would be rescued by a neutralizing mouse OPG antibody or the cytotoxic ligand TRAIL that blocks the anti-osteoclastic activity of OPG. TRAIL was a gift from Dr. W. Dougall Anderson (Immunex Corporation, Seattle, WA). Anti-mouse OPG antibody was purchased from R&D Systems, Inc., Minneapolis, MN.
Co-culture of Calvarial Osteoblasts and Spleen Cells-Osteoblastic cells were prepared from calvariae of 7-9-week-old Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice by sequential collagenase digestion (35). Spleen cells were simultaneously harvested from splenic tissue of these mice. Osteoblastic cells (5 ϫ 10 4 cells/well) and spleen cells (1 ϫ 10 6 cells/well) were co-cultured for 7 days in ␣MEM containing 10% HIFCS in the absence or presence of agonists. At the end of the culture, cells were fixed and stained for TRAP.
Resorbed Pit Formation-Bone resorption was assayed by measuring the ability of cultured bone marrow cells to form resorption pits on devitalized bovine cortical bone slices (4.4 ϫ 4.4 ϫ 0.2 mm) by previously described methods (7,36,37). Briefly, trypsinized bone marrow cells that had been cultured for 7 days with or without effectors were allowed to settle onto the surface of bone slices for 90 min in phosphate-buffered saline. Bone slices were rinsed vigorously and incubated for 24 h at 37°C in ␣MEM (0.7 g of sodium bicarbonate per liter) and 10% HIFCS. After the incubation, cells were fixed with 2.5% glutaraldehyde in phosphate-buffered saline for 30 min, stained for TRAP, and counterstained with 1% toluidine blue in 1% borax to observe resorption pits. Pits on these bone slices were quantitated using reflected light microscopy.
mRNA Isolation and Northern Blot Analysis-In co-culture experiments, cells were cultured in 10% serum in the presence or absence of the effector for the entire culture period. For Northern blot analysis, osteoblastic cells (3.5 ϫ 10 5 cells/well) and spleen cells (50 ϫ 10 6 cells/well) were co-cultured in the absence or presence of PTH for 7 days. Total RNA was extracted from cells by the acid guanidinium isothiocyanate extraction and cesium chloride ultracentrifugation methods (38,39). For Northern analysis, 20 g of total RNA was denatured and fractionated on a 0.8% agarose/1.1 M formaldehyde gel, transferred to filters by capillary blotting. and fixed to the filter by UV irradiation. After a 4-h prehybridization, filters were hybridized overnight with a [ 32 P]cDNA probe for the mRNAs of interest. Bands were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Northern blots were quantitated by autoradiography and densitometry. The mouse RANKL cDNA was a gift from Dr. W. Dougall Anderson (Immunex Corporation, Seattle, WA). The mouse OPG cDNA was a gift from Dr. W. Boyle (AMGEN, Thousand Oaks, CA).
Enzyme-linked Immunoabsorbent Assay for OPG-To measure OPG protein (40) in cell supernatants, osteoblastic cells (3.5 ϫ 10 5 cells/well) and spleen cells (50 ϫ 10 6 cells/well) were co-cultured in the absence or presence of PTH for 7 days. Confluent osteoblasts/spleen cells cultured for 7 days in the absence or presence of PTH were harvested, and cell supernatants were clarified by microfuge centrifugation (14,000 rpm, 10 min at 4°C). OPG protein was determined using a Quantikine M mouse OPG immunoassay kit according to the manufacturer's instructions (R & D Systems, Inc., Minneapolis, MN).
In Vivo PTH Administration-To examine serum calcium levels in Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice in response to PTH, 12-14-week-old mice were weighed and injected over the calvariae with vehicle (1 mM HCl with 1 mg/ml bovine serum albumin) or 80 or 320 g/kg body weight PTH (Bachem, Torrence, CA) according to the protocol of Mohan et al. (41). Mice were sacrificed 4 h later, and venous blood was obtained by cavernous sinus puncture. Total serum calcium was measured by a calorimetric assay (42).
Statistics-Significant differences among groups within each experiment were determined by analysis of variance followed by post hoc Scheffe test. When multiple experiments were pooled, statistical differences between genotypes were determined by Student's t test. well in cultures from Fgf2ϩ/ϩ mice. In contrast, fewer OCLs formed in bone marrow cultures from Fgf2Ϫ/Ϫ mice that were treated with PTH. Pooling of data from four separate experiments showed that 50% fewer OCLs were formed in response to PTH in marrow cultures from Fgf2Ϫ/Ϫ mice (58 Ϯ 12 in Fgf2Ϫ/Ϫ cells versus 116 Ϯ 18 in Fgf2ϩ/ϩ cells). We also examined the dose response effect of PTH (0.01-10 nM) on the number of OCLs that were present in 7-day Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ bone marrow cultures (Fig. 2). Few OCLs formed in unstimulated bone marrow cultures from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice. PTH caused a dose-dependent increase in the number of OCLs per well in bone marrow cultures from Fgf2ϩ/ϩ mice. In contrast, OCL formation at all concentrations of PTH was greatly reduced in bone marrow cultures from Fgf2Ϫ/Ϫ mice.

Effect of the Combination of FGF-2 and PTH on Osteoclastlike Cell Formation in Mouse Bone Marrow Cultures from
Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice-To assess whether there was an intrinsic abnormality in the ability of osteoclast precursors to form osteoclasts, we examined the effect of simultaneous treatment of bone marrow cultures with FGF-2 plus PTH on OCL formation. In the presence of FGF-2 (10 nM) and PTH (10 nM), the number of OCL in the cultures from the Fgf2Ϫ/Ϫ mice were similar to that observed in bone marrow cultures from Fgf2ϩ/ϩ mice treated with PTH alone (Fig. 3).
Effects of Co-culture of Osteoblasts and Spleen Cells from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice on Osteoclast Formation in Response to FGF-2 and PTH-To further examine the mechanism of the reduction in number of OCLs in Fgf2Ϫ/Ϫ mice, co-culture experiments were performed. Calvarial cells (fraction 2-5) from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice were used as osteoblast (support) cells, and spleen cells were used as a source of osteoclast precursors. Calvarial cells were plated at (5 ϫ 10 4 ) cells/well with (1 ϫ 10 6 ) spleen cells/well from either Fgf2ϩ/ϩ or Fgf2Ϫ/Ϫ mice and cultured in ␣MEM containing 10% HIFCS for 7 days. FGF-2 (10 nM) or PTH (10 nM) were added at the start of the experiment and with each media change, and OCL formation was determined by TRAP staining. Treatment with FGF-2 increased OCL formation in co-cultures of osteoblasts and spleen cells from either genotype (Fig. 4). PTH-increased OCL formation was similar in co-cultures of Fgf2ϩ/ϩ osteoblasts with spleen cells of either genotype. In contrast, PTH-induced OCL formation was significantly reduced in co-cultures of osteoblasts from Fgf2Ϫ/Ϫ mice with spleen cells of either genotype (Fig. 4).

Effect of the Combination of PTH and RANKL on Osteoclastlike Cell Formation in Mouse Bone Marrow Cultures from
Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice-To determine whether endogenous FGF-2 was specific for OCL formation by PTH, we examined the effect of RANKL to increase OCL formation in both genotypes. We also examined the effect of simultaneous treatment of bone marrow cultures with PTH and RANKL on OCL formation. PTH (10 nM) and RANKL (30 ng/ml) significantly increased OCL formation in bone marrow cultures from Fgf2ϩ/ϩ, mice and there was a further significant increase in the number of OCLs when the two effectors were administered simultaneously (Fig. 5A). In contrast, OCL formation by PTH and RANKL was significantly reduced in marrow stromal cells from the Fgf2Ϫ/Ϫ mice, and there was no additive effect when they were administered simultaneously. We examined whether FGF-2 would reverse the reduced OCL formation in response to RANKL. As shown in Fig 5B, OCL formation in marrow cultures from the Fgf2Ϫ/Ϫ mice was still significantly reduced in response to a very high concentration of RANKL (100 ng/ml). However, in the presence of both FGF-2 and RANKL, similar numbers of OCLS were formed in marrow cultures from both genotypes.

Effect of the Combination of PTH and IL-11 on Osteoclastlike Cell Formation in Mouse Bone Marrow Cultures from
Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice-In previous studies, the cytokine IL-11 was reported to be important in OCL formation (43). We, therefore, determined the effect of single versus simultaneous treatment of bone marrow cultures with PTH (10 nM) and IL-11 (10 Ϫ9 M) on OCL formation in bone marrow stromal cells from both genotypes. As shown in Fig. 6A, PTH and IL-11 significantly increased OCL formation in marrow cultures from Fgf2ϩ/ϩ mice when administered either as single agents or in combination, but there was no additive effect when the two effectors were administered simultaneously. OCL formation by PTH and IL-11 was significantly reduced in marrow stromal cells from the Fgf2Ϫ/Ϫ mice, and there was no additive effect when both effectors were administered simultaneously. We determined whether FGF-2 also reversed the reduced OCL formation in response to IL-11 in marrow cultures from the Fgf2Ϫ/Ϫ mice. As shown in Fig. 6B, addition of FGF-2 (10 nM) in combination with IL-11 reversed the reduced OCL formation observed in marrow cultures from Fgf2Ϫ/Ϫ mice. Expression of RANKL and OPG mRNA in Response to PTH in Co-culture of Osteoblasts and Spleen Cells from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice-Studies have shown that PTH-mediated regulation of RANKL and OPG is critical for OCL formation (1,33). We therefore compared the effects of PTH on RANKL and OPG mRNA expression in co-culture of osteoblasts and spleen cells from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice. Calvarial cells (fraction 2-5) from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice were used as osteoblast (support) cells, and spleen cells were used as a source of osteoclast precursors. Calvarial cells were plated at (3.5 ϫ 10 5 ) cells/well with (50 ϫ 10 6 ) spleen cells/well from either Fgf2ϩ/ϩ or Fgf2Ϫ/Ϫ mice and cultured in ␣MEM containing 10% HIFCS for 7 days. Vehicle or PTH (10 nM) was added at the start of the experiment and with each media change, and total RNA was extracted after 7 days. As shown in Fig. 7 treatment with PTH increased RANKL mRNA expression in cells from both genotypes. PTH caused a 42% reduction in OPG mRNA in co-cultures from Fgf2ϩ/ϩ mice. Basal OPG mRNA was markedly higher in vehicle-treated co-cultures from Fgf2Ϫ/Ϫ mice and was not decreased by PTH (Fig. 7). We also examined OPG mRNA expression in calvarial osteoblasts from both genotypes that were treated with vehicle or PTH (10 nM) for 2 h. As shown in Fig. 8 treatment with PTH caused a 30% reduction in OPG mRNA in calvarial osteoblasts cells from Fgf2ϩ/ϩ but had no effect on OPG mRNA in calvarial osteoblasts from Fgf2Ϫ/Ϫ mice.
Regulation of OPG Protein by PTH in Co-cultures of Osteoblasts and Spleen Cells from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice-To further examine whether reduced OCL formation in cultures from Fgf2Ϫ/Ϫ mice was in part because of changes in the expression of OPG protein, we examined the effects of PTH treatment on OPG protein in osteoblast/spleen cell co-cultures from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice. Cultures were treated for 7 days with or without PTH (10 nM). As shown in Fig. 9, basal OPG protein level was 40-fold higher in vehicle-treated cultures from Fgf2Ϫ/Ϫ mice, and although there was a 10-fold reduction in PTH-treated cultures from Fgf2Ϫ/Ϫ mice, these levels were still 30-fold higher compared with Fgf2ϩ/ϩ mice.
Effect of OPG on FGF-2 and PTH Osteoclast-like Cell Formation in Mouse Bone Marrow Cultures from Fgf2ϩ/ϩ Mice-We determined the effect of treatment of bone marrow stromal cells with OPG (10 ng/ml) on the ability of PTH (10 nM) and FGF-2 (1 nM) to induce OCL formation in marrow cultures from Fgf2ϩ/ϩ mice. As shown in Fig. 10, PTH and FGF-2 significantly increased OCL formation in marrow cultures from Fgf2ϩ/ϩ mice. However treatment with OPG significantly reduced PTH-mediated OCL formation and completely abrogated FGF-2-induced OCL formation.

Effect of Neutralizing Antibody to OPG on PTH-induced Osteoclast-like Cell Formation in Mouse Bone Marrow
Cultures from Fgf2ϩ/ϩ Mice-Because OPG mRNA and protein were increased in osteoblasts from Fgf2Ϫ/Ϫ mice, we determined whether a neutralizing antibody to OPG would reverse the reduced OCL formation observed in PTH-treated marrow stromal cultures from Fgf2Ϫ/Ϫ mice. As shown in Fig. 11, pretreatment with OPG antibody (2 g/ml) enhanced PTH-induced OCL formation in marrow stromal cultures from Fgf2ϩ/ϩ mice but did not rescue the reduced OCL formation in response to PTH in marrow stromal cultures from Fgf2Ϫ/Ϫ mice. We also examined whether TRAIL (100 g/ml) would rescue the reduced OCL formation in response to PTH and found that TRAIL did not rescue the reduced OCL formation in response to PTH (data not shown).
Injection of PTH over the Calvariae of Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice-Because we observed reduced OCL formation in vitro in response to PTH, we examined whether there were differences in the response to PTH in vivo. We utilized the protocol of Mohan et al. (41) to compare the ability of PTH to increase serum calcium in mice of both genotypes. 12-to 14-week-old mice were weighed and injected with vehicle or PTH (80 or 320 g/kg body weight) above the right hemicalvariae. Mice were sacrificed 4 h later, and blood samples were collected for measurement of serum calcium. As shown in Fig. 12, there was a significant increase in calcium in serum obtained from Fgf2ϩ/ ϩmice that were treated with PTH (320 g/kg) but no change in serum calcium from similarly treated Fgf2Ϫ/Ϫ mice. Serum calcium was not increased by 80 g/kg PTH in either genotype.
Formation of Bone Resorption Pits by Cultured Bone Marrow Cells from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ Mice-We determined pit formation by OCLs in 7-day bone marrow cultures from both genotypes that were treated with FGF-2 (10 nM) or PTH (10 nM). Quantitation of the formation of bone resorption pits in response to FGF-2 (10 nM) and PTH (10 nM) alone or in combination is shown in Fig. 13. OCLs in bone marrow cultures from Fgf2ϩ/ϩ mice that were treated with FGF-2 or PTH singly or

FIG. 4. Effect of co-culture of calvarial osteoblasts and spleen cells from Fgf2؉/؉ and Fgf2؊/؊ on OCL formation in response to FGF-2 or PTH.
A, calvarial osteoblasts from Fgf2ϩ/ϩ or Fgf2Ϫ/Ϫ were plated at 5 ϫ 10 4 cells/well with spleen cells 1 ϫ 10 6 /well from either Fgf2ϩ/ϩ or Fgf2Ϫ/Ϫ mice and cultured in ␣MEM for 7 days. FGF-2 (10 nM) or PTH (10 nM) were added as single agents or in combination at the start of the experiment and with each media change, and OCL formation was determined by TRAP staining. Values are the mean Ϯ S.E. for four determinations/group. *, significantly different from control (p Ͻ 0.01). a, significantly different from PTH (Fgf2ϩ/ϩ) (p Ͻ 0.01).
in combination significantly increased resorption pits on cortical bovine bone slices. OCLs in bone marrow cultures from Fgf2Ϫ/Ϫ mice that were treated with FGF-2 formed a similar number of pits. In contrast, OCL from Fgf2Ϫ/Ϫ mice that were treated with PTH formed 56% fewer pits than similar cultures from Fgf2ϩ/ϩ mice. DISCUSSION This study demonstrates that disruption of the Fgf2 gene in mice results in markedly decreased OCL formation in response to PTH. In earlier studies, we showed that treatment of marrow cultures with FGF-2 significantly increased OCL formation (7). We also reported that PTH, which is a potent inducer of OCL formation, is also a stimulator of FGF-2 mRNA and protein expression in osteoblasts (21). We therefore hypothesized that disruption of the Fgf2 gene would result in reduced OCL formation in response to PTH. The data in Figs. 1-6 show that disruption of the Fgf2 gene markedly reduced OCL formation in bone marrow cultures in response to PTH. The fact that exogenous FGF-2 caused a similar increase in the number of OCLs formed in bone marrow cultures from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice suggests that there is no intrinsic abnormality in the ability of the marrow osteoclast precursor cells from the Fgf2 mutant mice to form OCLs.
Studies have shown that OCL formation requires or is enhanced by interactions between osteoclast progenitor cells derived from hematopoietic lineage and mesenchymal stromal cells, which contain osteoblast precursors, in the presence of a stimulator of bone resorption (16,17). We therefore compared the ability of PTH to induce OCL formation in co-cultures of primary calvarial-enriched osteoblasts (support cells) and spleen cells (osteoclast precursors) from Fgf2ϩ/ϩ versus Fgf2Ϫ/Ϫ mice. The co-culture experiments (Fig. 4) show that when spleen cells from either genotype were co-cultured with osteoblasts from Fgf2ϩ/ϩ mice, in the presence of PTH, a similar number of OCLs were formed. In contrast, OCL formation was significantly impaired in response to PTH when osteoblasts were derived from Fgf2Ϫ/Ϫ mice. These results suggest an osteoblastic defect is present, which results in reduced osteoclast formation in the Fgf2Ϫ/Ϫ mice in response to PTH.
Marrow stroma regulates hematopoietic cell proliferation and differentiation, including macrophage/myeloid progenitors (44) from which osteoclasts are derived (16), through cell to cell interactions, as well as production of local factors including FGF-2 (45)(46)(47). FGF-2 is a potent mitogen for stromal fibroblasts that produce and respond to FGF-2 in an autocrine/ paracrine manner (48). It is possible that in the absence of endogenous FGF-2, there is a decrease in the number of osteoblast support cells. This is supported by our observation that there was a significant reduction in DNA content and thymidine incorporation in calvarial osteoblasts, as well as colony formation in bone marrow cultures from Fgf2Ϫ/Ϫ in comparison with Fgf2ϩ/ϩ mice (25).
It is also possible that endogenous FGF-2 is important in the proliferation of the OCL hematopoietic precursor population. In our previous studies we showed that FGF-2 stimulated OCL proliferation, as well as their differentiation from progenitor cells (7). Other studies have shown that FGF-2 promotes hematopoietic cell development by enhancing the colony stimulating activity of IL-3 on primitive hematopoietic progenitors in vitro (49). We reported in our earlier studies that in methylcellulose culture of bone marrow cells from FGF-2-deficient mice, there was a significant decrease in colony formation in response to the hematopoietic cytokine IL-3 (22). This observation is noteworthy, because Hattersley et al. (12) reported that IL-3 stimulated OCL precursor proliferation from murine hematopoietic cells but not their differentiation. However, addi-tion of IL-3 to PTH did not reverse reduced OCL formation in bone marrow cells from the Fgf2Ϫ/Ϫ mice (data not shown).
To assess whether reduced OCL formation was a generalized phenomena in the Fgf2Ϫ/Ϫ marrow cells and not specific for PTH, we examined the effect of RANKL and IL-11 to induce OCL formation in bone marrow cultures from both genotypes. Similar to our observation with PTH, there was a marked reduction in OCL formation in the Fgf2Ϫ/Ϫ marrow cultures in response to both effectors that was rescued by exogenous FGF-2. These results suggest a role for endogenous FGF-2 in OCL formation in response to multiple effectors.
RANKL has been identified as the factor expressed on osteoblast/stromal cells that transduces a signal via cell-cell interaction to osteoclast progenitors resulting in their differentiation (31). It is believed that RANKL functions to induce differentiation of osteoclast progenitors but not their proliferation (31). Because FGF-2 is produced, stored, and released from bone marrow stromal cells (48), it could induce osteoclastogenesis in part by regulating the expression of RANKL on osteoblast/stromal cells. Studies have shown that OPG inhibits OCL formation (30), and PTH and FGF-2 both down-regulate OPG mRNA expression in osteoblasts (33,34). We hypothesized that because PTH increased Fgf2 gene expression in osteoblasts (21), local FGF-2 production might be important in PTH/FGF-2/RANKL/OPG signaling. We therefore examined whether the expression of RANKL and OPG mRNA levels in bone osteoblasts from Fgf2ϩ/ϩ and Fgf2Ϫ/Ϫ mice were similar in response to treatment with PTH. As shown in Fig. 7, PTH increased RANKL mRNA expression in osteoblasts from Fgf2Ϫ/Ϫ mice suggesting that diminished RANKL expression is not the mechanism for reduced OCL formation in bone marrow cultures from Fgf2Ϫ/Ϫ mice. This is further supported by the studies in Fig. 5, A and B showing reduced OCL formation in bone marrow cultures from Fgf2Ϫ/Ϫ mice treated with exogenous RANKL.
Because previous studies showed that OPG can reduce OCL formation in vitro and in vivo (1), we examined OPG mRNA and protein in both genotypes. In co-culture experiments, cells were treated with vehicle or PTH for the entire culture period. As shown in Figs. 7 and 8, although PTH reduced OPG mRNA levels in osteoblasts from Fgf2ϩ/ϩ mice, it did not reduce OPG mRNA in osteoblasts from the Fgf2Ϫ/Ϫ mice. We also examined OPG protein (Fig. 9) in co-cultures of osteoblasts and spleen cells from both genotypes. Basal OPG protein level was 40-fold higher in vehicle treated co-cultures utilizing osteoblasts from the Fgf2Ϫ/Ϫ mice and was only reduced 10-fold by PTH. Previous studies have shown that OPG blocks osteoclas-togenesis by altering cell-to-cell signaling (30) and by inhibiting the differentiation of osteoclasts (32,50). FGF-2 was reported previously to down-regulate OPG mRNA expression in osteoblasts (34,51). Therefore we postulated that in the absence of endogenous FGF-2, OPG mRNA and protein would be increased in osteoblasts and that a relative increase in OPG in the Fgf2Ϫ/Ϫ mice contributed to reduced OCL formation. However, an antibody to OPG did not reverse reduced OCL formation in response to PTH. Furthermore a very high concentration of RANKL did not overcome the reduced OCL formation. FIG. 7. Expression of RANKL and OPG mRNA in response to PTH in co-cultures of calvarial osteoblasts and spleen cells from Fgf2؉/؉ and Fgf2؊/؊ mice. Calvarial osteoblasts from Fgf2ϩ/ϩ or Fgf2Ϫ/Ϫ were plated at 3.5 ϫ 10 5 cells/well with spleen cells (50 ϫ 10 6 cells/well) from either Fgf2ϩ/ϩ or Fgf2Ϫ/Ϫ mice and cultured in ␣MEM for 7 days. Vehicle or PTH (10 nM) was added at the start of the experiment and with each media change. Total RNA was extracted after 7 days. Twenty micrograms of total RNA was analyzed for RANKL and OPG mRNAs by Northern blotting. The filters were stripped and reprobed for GAPDH mRNA that was utilized for normalization of data. The numbers represent the ratio of the optical density of the gene of interest normalized to the optical density of GAPDH.
FIG. 8. Effect of PTH on OPG mRNA expression in primary calvarial osteoblasts from Fgf2؉/؉ and Fgf2؊/؊ mice. Calvarial osteoblasts from Fgf2ϩ/ϩ or Fgf2Ϫ/Ϫ were plated at 5000 cells/cm 6 in 6-well plates and cultured in ␣MEM for 7 days. Cells were serumdeprived for 24 h and then treated with vehicle or PTH (10 nM) for 2 h. Total RNA was extracted as described under "Experimental Procedures." Twenty micrograms of total RNA was analyzed for OPG mRNA by Northern blotting. The filters were stripped and reprobed for GAPDH mRNA that was utilized for normalization of data. The numbers represent the ratio of the optical density of the gene of interest normalized to the optical density of GAPDH. These results suggest that increased OPG alone does not mediate the reduced OCL formation in the Fgf2Ϫ/Ϫ mice and that additional mechanisms may be involved. Interestingly although counter-intuitive, OPG levels were found to be higher in postmenopausal women with osteoporosis, which suggest that OPG levels are regulated by age-related factors (52,53).
We previously published that mice with a disruption of the Fgf2 gene develop decreased bone mass and bone formation with age (25). We also indicated in our previous report (25) that although osteoclast number was not significantly decreased, osteoclast surface/bone surface was reduced by 48% in the proximal tibiae of 8-month-old Fgf2Ϫ/Ϫ mice. Our observation that Fgf2Ϫ/Ϫ mice also have decreased OCL formation is reminiscent of the low turnover state observed in the SAMP-6 mouse in which both osteoblast and osteoclast formation are reduced (43). Previous studies by Kodama et al. (54) showed that reduced OCL formation in response to 1.25(OH) 2 D 3 observed in bone marrow stromal cells from the SAMP-6 mouse could be reversed by co-administration of IL-11. We therefore assessed whether addition of IL-11 could enhance PTH-induced OCL formation in the Fgf2Ϫ/Ϫ mice. We observed that IL-11induced OCL formation was also significantly impaired, and there was no additive or synergistic effect of the combination of IL-11 and PTH to increase OCL formation in bone marrow stromal cells from the Fgf2Ϫ/Ϫ mice.
Because PTH-induced osteoclastogenesis was impaired in the in vitro studies using the Fgf2Ϫ/Ϫ mice, we reasoned that we might observe an alteration in the response to PTH in vivo. We, therefore, compared the effect of subcutaneous injection of PTH on serum calcium levels in mice of both genotypes. As shown in Fig. 12, a significant increase in calcium was observed in serum obtained from Fgf2ϩ/ϩ mice that were treated with a high concentration of PTH, but no change in serum calcium from similarly treated Fgf2Ϫ/Ϫ mice was observed. These data are consistent with impaired response of osteoclasts in the Fgf2Ϫ/Ϫ mice. We reported previously (7) that FGF-2 increased resorbed pit area, and these results were confirmed by other investigators (55). Because PTH did not increase serum calcium in Fgf2Ϫ/Ϫ mice, we reasoned that endogenous FGF-2 might also influenced osteoclast activity. We found that PTHinduced OCLs from the Fgf2Ϫ/Ϫ mice formed significantly fewer resorption pits on bone slices and that the addition of FGF-2 increased bone pit resorption to the level observed by PTH in the Fgf2ϩ/ϩ cultures (Fig. 13). Thus endogenous FGF-2 is required for both maximal OCL formation and OCL activity.
In summary, we have shown that OCL formation is impaired in Fgf2Ϫ/Ϫ mice in response to PTH, an important stimulator of OCL formation and an important regulator of FGF-2 production in osteoblasts. We show that the ability of PTH to increase OCL formation and bone resorption in vitro and to increase serum calcium in vivo is markedly impaired in Fgf2Ϫ/Ϫ mice. We therefore conclude that endogenous FGF-2 is an important local factor in PTH-induced OCL formation and bone resorption in mice. Because OCL formation was also markedly reduced in RANKL-and IL-11-treated marrow stromal cultures, we conclude that endogenous FGF-2 is important for maximal OCL formation in response to PTH, as well as other stimulators of OCL formation. PTH did not down-regulate OPG mRNA and protein expression in osteoblasts from Fgf2Ϫ/Ϫ mice, and an FIG. 11. Effect of a neutralizing OPG antibody on PTH (10 nM)-induced OCL formation in mouse bone marrow cultures from Fgf2؉/؉ mice. Bone marrow stromal cells were pretreated with OPG antibody (2 g/ml) for 2 h prior to the addition of PTH at the start of the experiment and with each change of culture media. TRAP stain for OCLs was performed as described under "Experimental Procedures." Values are the mean Ϯ S.E. for six determinations per group. *, significantly different from control (p Ͻ 0.01). a, significantly different from PTH in Fgf2ϩ/ϩ (p Ͻ 0.01).
FIG. 12. Effect of PTH (1-34) on serum calcium in Fgf2؉/؉ and Fgf2؊/؊ mice. Mice were weighed and injected subcutaneously over the calvariae with either vehicle or PTH (80 or 320 g/kg body weight). At 4 h, mice (n ϭ 3) from each group were sacrificed, and blood samples were collected for measurement of serum calcium. *, significantly different from vehicle in Fgf2ϩ/ϩ mice (p Ͻ 0.03).
FIG. 13. Formation and quantification of bone resorption pits by cultured bone marrow cells from Fgf2؉/؉ and Fgf2؊/؊ mice in response to FGF-2 (10 nM) or PTH (10 nM) alone or in combination. Cells were incubated on bovine cortical bone slices for 24 h as described under "Experimental Procedures." Pits on these bone slices were counted using reflected light microscopy. Values are the mean Ϯ S.E. for eight determinations/group. *, significantly different from control (p Ͻ 0.01). a, significantly different from PTH (Fgf2ϩ/ϩ) (p Ͻ 0.01) antibody to OPG did not rescue reduced OCL formation in response to PTH. We conclude that a non-RANKL/OPG signaling pathway is important for maximal OCL formation in Fgf2Ϫ/Ϫ mice.