BMP-2 Differentially Modulates FGF-2 Isoform Effects in Osteoblasts From Newborn Transgenic Mice

Abstract

We previously generated separate lines of transgenic mice that specifically overexpress either the Fibroblast growth factor (FGF)-2 low-molecular-mass isoform (Tg^LMW) or the high-mass isoforms (Tg^HMW) in the osteoblast lineage. Vector/control (Tg^Vector) mice were also made. Here we report the use of isolated calvarial osteoblasts (COBs) from those mice to investigate whether the FGF-2 protein isoforms differentially modulate bone formation in vitro. Our hypothesis states that FGF-2 isoforms specifically modulate bone morphogenetic protein 2 (BMP-2) function and subsequently bone differentiation genes and their related signaling pathways. We found a significant increase in alkaline phosphatase-positive colonies in Tg^LMW COBs compared with Tg^Vector controls. BMP-2 treatment significantly increased mineralized colonies in Tg^Vector and Tg^LMW COBs. BMP-2 caused a further significant increase in mineralized colonies in Tg^LMW COBs compared with Tg^Vector COBs but did not increase alkaline phosphatase-positive colonies in Tg^HMW COBs. Time-course studies showed that BMP-2 caused a sustained increase in phosphorylated mothers against decapentaplegic-1/5/8 (Smad/1/5/8), runt-related transcription factor-2 (Runx-2), and osterix protein in Tg^LMW COBs. BMP-2 caused a sustained increase in phospho-p38 MAPK in Tg^Vector but only a transient increase in Tg^LMW and Tg^HMW COBs. BMP-2 caused a transient increase in phospho-p44/42 MAPK in Tg^Vector COBs and no increase in Tg^LMW COBs, but a sustained increase was found in Tg^HMW COBs. Basal expression of FGF receptor 1 protein was significantly increased in Tg^LMW COBs relative to Tg^Vector COBs, and although BMP-2 caused a transient increase in FGF receptor 1 expression in Tg^Vector COBs and Tg^HMW COBs, there was no further increase Tg^LMW COBs. Interestingly, although basal expression of FGF receptor 2 was similar in COBs from all genotypes, BMP-2 treatment caused a sustained increase in Tg^LMW COBs but decreased FGF receptor 2 in Tg^Vector COBs and Tg^HMW COBs.

Osteoblast differentiation and bone formation are regulated by many local factors, and, among those, bone morphogenetic proteins (BMPs) are the most potent. BMPs, synthesized and released by osteoblasts in the bone matrix (1, 2), signal through heterotetrameric complexes formed by two receptor type II and 2 receptor type I subunits (3). BMP-2 specific mothers against decapentaplegic (R-Smads; Smad 1/5/8) are the major transducer for BMP receptors, and, once activated, they form heteromeric complexes with Smad 4 for nuclear translocation to regulate target genes by interacting with various transcription factors (3). Disorders in BMP signaling have been implicated in multiple bone diseases including tumor metastasis, brachydactyly type A2, and osteoarthritis (3, 4). It is known that BMP-2 controls the expression and function of runt-related transcription factor-2 (Runx-2; Cbfa1) through Smad signaling (5, 6) in which Runx-2 is an essential protein for osteoblast maturation (7). Indeed, in humans, heterozygous loss of Runx-2 function segregates with abnormal skeletal phenotypes (8, 9), and murine Runx-2 haploinsufficiency causes similar phenotypes with a primary defect in intramembranous bone formation (7, 10). Moreover, BMP-2 strongly increases osterix, the secondary regulator of osteoblastogenesis (11–13) that is specifically expressed in the osteoblasts and osteocytes of all developing bones, in which it likely acts downstream of Runx-2 (14).

Fibroblast growth factor (FGF)-2 potently stimulates proliferation of osteoblast precursor and is therefore anabolic in experimental mice (15), whereas Fgf2^−/− mice show significantly decreased bone mass and bone formation (16). The FGF-2 gene encodes multiple protein isoforms from alternative translation start sites (17, 18). The human FGF-2 high-molecular-mass isoforms 22, 23, and 24 kDa contain a nuclear localization signal in the amino- terminus extension to confer an intracrine function (18–20). The low-molecular-mass 18-kDa FGF-2 protein is exported from cells, presumably conferring a different function vs the high-molecular-mass (HMW) FGF protein isoforms. In rodents, there are two nuclear isoforms of 19 kDa and 21 kDa and then a low-molecular-mass, 17.5-kDa FGF-2 protein that is exported from cells and stored in the extracellular matrix (21–23).

Previously we reported that FGF-2 null mice developed osteopenia with aging (16), demonstrating that FGF-2 is an important regulator of bone formation in vivo. Alternatively, global transgenic overexpression of the cDNA that encodes for all FGF-2 protein isoforms resulted in decreased endochondral and intramembranous bone formation (24). We have also shown that overexpression of the HMW/nuclear isoforms preferentially mediate augmented osteoblastic ROS17/2.8 cell proliferation (25), whereas the low-molecular-weight (LMW)/exported isoform is involved in the prostaglandin F2α (PGF2α) effects found in mouse primary calvarial osteoblasts (26–28). In addition, we demonstrated that the anabolic effects of PTH on bone are, at least in part, mediated by FGF-2 (29). In other studies we reported that BMP-2 increased FGF-2 mRNA and protein in osteoblasts from Fgf2^+/+ mice (30). We showed that the BMP-2 induced nuclear accumulation of Runx-2 was markedly impaired in osteoblasts from Fgf2^−/− mice, whereas an increased Runx-2 accumulation and nuclear localization were found after BMP-2 treatment in Fgf2^−/− osteoblasts transfected with a retroviral Fgf2 construct (31). Thus, it is evident that a reciprocal regulation of FGF-2 and BMP-2 exists in osteoblasts. However, so far it is unclear whether the LMW and HMW FGF-2 isoforms selectively modulate the effect of BMP-2-induced bone formation or the regulation of the BMP-2 downstream signaling in osteoblasts.

Therefore, the aim of this in vitro study was to explore the potential role of FGF-2 isoforms for BMP-2 signaling in osteoblasts differentiation. For this purpose, we used primary calvarial osteoblasts isolated from different lines of transgenic (Tg) mice that selectively overexpress either the LMW isoforms of FGF-2 (Tg^LMW) or the HMW isoforms of FGF-2 (Tg^HMW).

Materials and Methods

Animals

Generation of mice expressing HMW and LMW isoforms of FGF-2

Transgenic mouse lines overexpressing specific FGF-2 isoforms, regulated by collagen type I 5′ elements for osteoblast specificity, were developed as previously described (32), and they are as follows: Tg^LMW, which specifically overexpresses low-molecular-mass human FGF-2 (18 kDa); Tg^HMW, which specifically overexpresses the three high-molecular-mass human FGF-2 isoforms (22, 23, and 24 kDa); and Tg^Vector, which overexpresses the transgene cassette without FGF-2 coding sequences. All mice were bred and housed in the transgenic facility in the Center for Laboratory Animal Care at the University of Connecticut Health Center.

Primary calvarial osteoblasts (COBs)

Newborn mice were killed by CO₂ narcosis and cervical dislocation. The Animal Care and Use Committee of the University of Connecticut Health Center approved all animal protocols. Because the mice were newborn, it was not possible to identify the specific sex. Cells were digested from calvariae of newborn control Tg^Vector as well as Tg^LMW and Tg^HMW mice in a solution of 1 mL trypsin as previously described (16). Cells were pooled and then plated in 100-mm dishes and cultured to confluence in DMEM without phenol red (Sigma Chemical Co, St Louis, Missouri) containing 10% heat-inactivated-fetal calf serum (HIFCS), penicillin (100 U/mL), and streptomycin (50 μg/mL) in a humidified atmosphere of 5% CO₂ at 37°C. Then cultured cells were shipped frozen in medium plus 10% dimethylsulfoxide to the University of Camerino (Camerino, Italy), where they were thawed and cultured immediately upon receipt to perform the in vitro experiments described below. No significant effects on cell morphology and viability were observed.

Alkaline phosphatase (ALP) assay and Alizarin red S histochemical staining

COBs from Tg^Vector mice and the two FGF-2 transgenic lines (described above) were plated at 5 × 10⁴ cells/well in 6-well culture plates and cultured for 6 days in differentiation media (DMEM, 10% HIFCS, 8 mM β-glycerophosphate, 50 μg/mL ascorbic acid, 100 nM dexamethasone) in the presence or absence of BMP-2 (50 ng/mL) or vehicle. ALP staining was per formed with a commercial kit (Sigma-Aldrich, Milano, Italy) according to the manufacturer's instructions. Using Alizarin red S, COBs cultured on 6-well culture dishes containing coverslips, were stained on day 14 for assessing the mineralized matrix. The medium was removed, and the cell layers were rinsed with PBS and fixed in 4% paraformaldehyde diluted in PBS for 25 minutes at room temperature. Then cultures were washed with PBS and stained with 2% Alizarin red S (pH 7.2; Sigma-Aldrich) for 20 minutes at 37°C. Cultures were observed at light microscope level. Then stain was desorbed and the collected solutions were distributed as 100 μL/well on 96-well plates for absorbance reading at 590 nm by spectrophotometry (Tecan Infinite reader; Tecan, Milano, Italy).

Western blotting

COBs from Tg^Vector mice and the two FGF-2 transgenic lines (described above) were plated at 10 × 10⁶ cells/well in 6-well culture dishes and grown in appropriate medium with 10% HIFCS, penicillin, and streptomycin until confluence. Cells were serum deprived for 24 hours and stimulated with BMP-2 (50 ng/mL) or vehicle for selected time periods. Proteins were extracted with cell lysis buffer (Cell Signaling Celbio, Milano, Italy), and the concentration was determined by the BCA protein assay reagent (Pierce, Celbio, Italy). After SDS-PAGE on 12% gels, proteins were transferred to polyvinyl difluoride (PVDF) membranes (Amersham Biosciences Europe Gmbh, Cologno Monzese, Milano, Italy). The next steps were performed by enhanced chemiluminescence Advance Western blotting detection kit (Amersham Biosciences Europe). Briefly, membranes were blocked with Advance blocking agent in PBS containing 0.1% Tween 20 (PBS-T) for 1 hour at room temperature. Then, membranes were incubated for 2 hours at room temperature with the following primary antibodies: rabbit antiphospho-Smads 1/5/8, rabbit antiphospho-p38, rabbit antiphospho-p44/42, rabbit antiphospho-Akt, rabbit anti-Akt antibodies (Cell Signaling), all diluted at 1:800; and rabbit anti-Pebp2aA (Runx-2), rabbit antiosterix, rabbit anti-C-term-FGFR1 (Flg C-15), and rabbit anti-C-term-FGFR2 (Bek C-17) antibodies (Santa Cruz Biotechnology, Tebu-Bio, Milano, Italy), all diluted 1:200. After washing with PBS-T, the blots were incubated with horseradish peroxidase-conjugated donkey antirabbit IgG antibody (Amersham Biosciences Europe) diluted 1:100 000 in blocking solution for 1 hour at room temperature. After further washing with PBS-T, the immunoreactive bands were visualized using luminol reagents and Hyperfilm-enhanced chemiluminescence film (Amersham Biosciences Europe) in accordance with the manufacturer's instructions. To normalize the bands, the filters were stripped and reprobed with a mouse anti-α-tubulin antibody (Sigma-Aldrich). Band densities were quantified densitometrically.

RNA interference

To specifically study the effects of BMP-2 treatments on FGF receptor-1 (FGFR1) and FGFR2, COBs from Tg^Vector and the two FGF-2 transgenic lines (described above) were plated in 6-well culture dishes at a density of 10 × 10⁶ cells/well and grown for 5 days in DMEM containing 10% HIFCS, penicillin, and streptomycin. Other cultures were transfected with FGFR1 small interfering RNA (siRNA) or with FGFR2 siRNA according to siRNA transfection protocol from the manufacturer (Santa Cruz Biotechnology). A control siRNA (nonhomologous to any known gene sequence) was used as a negative control. The levels of FGFR1 and FGFR2 were analyzed by Western blotting using a rabbit anti-C-term-FGFR1 (Flg C-15) and a rabbit anti-C-term-FGFR2 antibody (Bek C-17) antibody (Santa Cruz Biotechnology), and the specificity of the silencing was confirmed in 3 independent experiments. Cells were 24 hours serum deprived before the addition of BMP-2 (50 ng/mL) or vehicle for 2 hours. Then proteins were extracted and phospho-Smads 1/5/8 and phospho-p44/42 as well as Runx-2 and osterix levels were analyzed by Western blotting as described above. To normalize the bands, filters were stripped and reprobed with a monoclonal anti-α-tubulin (Sigma-Aldrich). Band densities were quantified using densitometry.

Statistical analysis

All experiments were repeated at least 3 times with overlapping results. An unpaired 2-tailed Student's t test was used to test for significant differences between 2 groups with a minimum threshold of P < .05.

Results

BMP-2 effects on bone mineralization marker expression in COBs from Tg^Vector and FGF-2 transgenic mouse lines

We evaluated whether the overexpression of specific FGF-2 isoforms differentially modulated bone formation in vitro or influenced the BMP-2 regulation of bone differentiation markers that control matrix mineralization. After 6 days of treatment, vehicle-only-treated cultures showed a significantly higher level of ALP-positive colonies in Tg^LMW COBs compared with Tg^HMW COBs (Figure 1). Treatment with BMP-2 caused a further significant increase in ALP-positive colonies in both Tg^Vector and Tg^LMW COBs. In contrast, no significant changes between BMP-2-treated and -untreated cultures were observed in Tg^HMW COBs, as shown in the graphic derived from a pool of 3 independent experiments (Figure 1, A and B). Alizarin red staining of untreated cultures after 14 days revealed a significant decrease in calcified nodules in Tg^HMW COBs compared with Tg^Vector and Tg^LMW COBs (Figure 2). Moreover, BMP-2 treatment caused a significant increase in calcified nodules only in Tg^Vector and Tg^LMW COBs (Figure 2, A and B).

Figure 1.

ALP staining. COBs from Tg^Vector mice and FGF-2 transgenic mice were grown for 6 days on 6-well culture dishes in differentiation medium with BMP-2 (50 ng/mL) or vehicle. Colony area was measured by NIH Image (National Institutes of Health, Bethesda, Maryland), and values are reported as means ± SD and statistically analyzed with a t test. *, P < .05 vs the corresponding vehicle-treated COBs; #, P < .05 vs the corresponding BMP-2-treated COBs (A) were in accordance with ALP staining observed with a light microscope. Bar, 100 μm (B).

Figure 2.

Alizarin red S staining. COBs from Tg^Vector, Tg^LMW, and Tg^HMW mice were plated on coverslips in 6-well culture dishes and then growth for 14 days in differentiation medium in presence of BMP-2 (50 ng/mL) or vehicle. The results show a quantitative analysis of Alizarin red staining in which values are reported as means ± SD and statistically analyzed with a t test. #, P < .05 vs the corresponding vehicle-treated Tg^Vector and Tg^LMW COBs; *, P < .05 vs the corresponding vehicle-treated COBs and were in accordance with the Alizarin red S staining observed with a light microscope. Bar, 100 μm (B).

Expression and BMP-2 regulation of the intracellular signaling molecules in COBS from Tg^Vector, Tg ^LMW, or Tg^HMW mice

The results shown in Figure 1 demonstrated that overexpression of the FGF-2 isoforms differentially influenced the expression of bone formation markers in both untreated and BMP-2-treated conditions. Therefore, we determined whether the FGF-2 isoforms influenced the BMP-2-activated downstream intracellular signaling molecules. Those pathways included the classically BMP-2 activated Smad signaling pathways (2) as well as p38 MAPK and ERKs p44/42 that were reported to be significant in FGF-2 (15) and BMP-2 signal cascades (33). The basal expression of phospho-Smad 1/5/8, phospho-p38 and phospho-p44/42 in 6-day cultured and 24-hour serum-deprived COBs from Tg^Vector and the FGF-2 transgenic mouse lines are shown in Figure 3. COBs isolated from Tg^HMW overexpression mice showed significantly up-regulated levels of phospho-Smad 1/5/8 level compared with Tg^Vector COBs. As shown in Figure 4, A and B, treatment with BMP-2 increased phospho-Smad 1/5/8 levels in Tg^Vector and in Tg^LMW at 30 minutes and 2 and 6 hours. Conversely, in Tg^HMW COBs (Figure 4C), the increase was evidenced only at 30 minutes and 2 hours. We observed no differences in basal phospho-p38 expression in COBs from any of the FGF-2 transgenic mice (Figure 3). However, as shown in Figure 4A, phospho-p38 was increased by BMP-2 in Tg^Vector COBs at all time points tested, with the maximal up-regulation at 2 hours, although phospho-p38 was increased only at 2 hours in Tg^LMW (Figure 4B) and in Tg^HMW COBs, as shown in Figure 4C. The basal level of phospho-p44/42, observed in Tg^LMW COBs, was comparable with that found in Tg^Vector COBs (Figure 3) and was unchanged by treatment with BMP-2 (Figure 4B). In contrast, BMP-2 up-regulated phospho-p44/42 levels at 2 hours in Tg^Vector COBs (Figure 4A), and then BMP-2 also increased phospho-p44/42 levels in Tg^HMW COBs at all time points (Figure 4C).

Figure 3.

COBs from Tg^Vector, Tg^LMW, and Tg^HMW mice were serum deprived for 24 hours. Proteins were extracted, subjected to SDS-PAGE, transferred to a PVDF membrane, and then probed with the rabbit antiphospho-Smad 1/5/8, the rabbit antiphospho-p38, the rabbit antiphospho-p44/42, the rabbit anti-Runx-2, and the rabbit anti-osterix antibodies; next, the filters were stripped and reprobed with the mouse anti-α-tubulin antibody. Results from 3 independent experiments are shown as means ± SD. *, P < .05 vs Tg^Vector COBs.

Figure 4.

COBs isolated from Tg^Vector (A), Tg^LMW (B), and Tg^HMW (C) mice were serum deprived for 24 hours and then treated with BMP-2 (50 ng/mL) or vehicle for 30 minutes, 2 hours, and 6 hours. Proteins were extracted, subjected to SDS-PAGE, transferred to a PVDF membrane, and probed with antibodies to either rabbit antiphospho-Smads (1, 5, 8), rabbit antiphospho-p38, rabbit antiphospho-p44/42, rabbit anti-Runx-2, or rabbit anti-osterix; then the filters were stripped and reprobed with the mouse anti-α-tubulin antibody for quantitative analysis. Results from 3 independent experiments are shown as means ± SD. *, P < .05 vs the corresponding unstimulated COBs.

We also examined the levels of the osteogenic master transcription factors Runx-2 and osterix. The levels of both proteins were decreased in Tg^LMW COBs compared with Tg^Vector COBs (Figure 3). As shown in Figure 4A, BMP-2 treatment increased Runx-2 only at 6 hours; however, BMP-2 significantly increased osterix protein during the entire treatment period in Tg^Vector COBs. In addition, Runx-2 and osterix were strongly and stably increased by BMP-2 in Tg^LMW COBs (Figure 4B). In contrast, in Tg^HMW COBs, BMP-2 down-regulated Runx-2 at 6 hours, and, in a time dependent manner, it also reduced osterix synthesis (Figure 4C).

Because recent studies demonstrated that the requirement of protein kinase B (Akt) for BMP-2 stimulated osteoblast differentiation (34), we also examined its expression. Interestingly, a consistent and stable increase of phospho-Akt was observed only in Tg^LMW COBs (Figure 5).

Figure 5.

COBs from Tg^Vector, Tg^LMW, and Tg^HMW mice were serum deprived for 24 hours and treated with BMP-2 (50 ng/mL) or vehicle for 30 minutes, 2 hours, and 6 hours. Proteins were extracted, subjected to SDS-PAGE, transferred to a PVDF membrane, and probed with the rabbit antiphospho-Akt and the rabbit anti-Akt; then the filters were stripped and reprobed with the mouse anti-α-tubulin antibody. Results from 3 independent experiments are shown as means ± SD. *, P < .05 vs the corresponding unstimulated COBs.

Expression and regulation of the FGFR1 and the FGFR2 by BMP-2 in COBs from Tg^Vector, Tg^LMW, and Tg^HMW mice

We first evaluated whether there was FGF-2 isoform-specific regulation of the FGFR1 and FGFR2 proteins and whether BMP-2 differentially regulated FGFR1 and FGFR2 protein synthesis in FGF-2 isoform-specific COBs. As shown in Figure 6, in 24-hour serum-derived cultures, a higher level of FGFR1 was detected in Tg^LMW COBs compared with Tg^Vector COBs, whereas the receptor level was similar in Tg^HMW COBs and in Tg^Vector COBs. Interestingly, treatment with BMP-2 at 6 hours significantly increased the FGFR1 level in Tg^Vector COBs but did not affect the receptor level in Tg^LMW COBs at any time point. However, a strong up-regulation of FGFR1 was observed in Tg^HMW COBs incubated for 30 minutes and 2 hours with BMP-2. In contrast to FGFR1, as shown in Figure 6, basal expression of FGFR2 was similar in Tg^LMW COBs and Tg^HMW COBs compared with Tg^Vector COBs in which FGFR2 protein was decreased by BMP-2 at 30 minutes in Tg^Vector COBs and at 6 hours in Tg^HMW COBs. Interestingly, a significant increase in FGFR2 protein was observed in Tg^LMW COBs for the entire time course of BMP-2 treatment.

Figure 6.

COBs from Tg^Vector, Tg^LMW, and Tg^HMW mice were serum deprived for 24 hours and treated with BMP-2 (50 ng/mL) or vehicle for 30 minutes, 2 hours, and 6 hours. Proteins were extracted, subjected to SDS-PAGE, transferred to a PVDF membrane, and probed with the rabbit anti-C-term-FGFR1 antibody (Flg C-15) or with the rabbit anti anti-C-term-FGFR2 antibody (Bek C-17); filters were stripped and reprobed with the mouse anti-α-tubulin antibody. Results from 3 independent experiments are shown as means ± SD. *, P < .05 vs corresponding unstimulated control; #, P < .05 vs Tg^Vector unstimulated COBs.

FGFR2 and FGFR1 knockdown in FGF-2 transgenic COBs

To analyze the potential involvement of the FGFR2 in BMP-2-modulated intracellular signaling proteins, we conducted FGFR2 knockdown experiments in BMP-2-treated, FGF-2 transgenic COBs (Figure 6). COBs isolated from Tg^Vector, FGF-2 Tg^LMW, and Tg^HMW mice were cultured in either 2-hour vehicle-treated or BMP-2-treated conditions (Figure 7). siRNA against the FGFR2 message efficiently and specifically silenced (>80%) FGFR2 protein expression in Tg^Vector COBs, in Tg^LMW COBs, and in Tg^HMW COBs (Supplemental Data, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). Depletion of FGFR2 strongly decreased phospho-Smad 1/5/8 in vehicle-treated and in BMP-2-treated Tg^Vector COBs (Figure 7A) and partially inhibited the BMP-2-induced increase of phospho-Smad in Tg^LMW COBs (Figure 7B); moreover, the FGFR2 siRNA totally inhibited the BMP-2-induced increase of phospho-Smads in Tg^HMW COBs (Figure 7C). Lack of FGFR2 also caused a decrease of phospho-p44/42 in vehicle-treated and BMP-2-treated Tg^Vector COBs and Tg^LMW COBs (Figure 7, A and B). Interestingly, phospho-p44/42 was increased in vehicle-treated FGFR2 knocked-down Tg^HMW COBs, which was unchanged by treatment with BMP-2 (Figure 7C). In FGFR2 knocked-down Tg^Vector COBs, treatment with BMP-2 decreased Runx-2 and osterix protein level (Figure 7A). FGFR2 knockdown also decreased both Runx-2 and osterix proteins in Tg^LMW COBs, in Tg^HMW vehicle-treated, and BMP-2-treated COBs (Figure 7, B and C).

Figure 7.

Depletion of FGFR2 and FGFR1 expression by siRNA in COBs from Tg^Vector (A), Tg^LMW (B), and Tg^HMW (C) mice. COBs were transfected with FGFR1 or FGFR2 siRNA. After transfection, as described above, cells were serum deprived for 24 hours and treated with BMP-2 (50 ng/mL) or vehicle for 2 hours. Proteins were extracted for Western blotting analysis and probed with antibodies to either rabbit antiphospho-Smads (Smads 1, 5, 8), rabbit antiphospho-p44/42, rabbit anti-Runx-2, or rabbit anti-osterix; then the filters were stripped and reprobed with the mouse anti-α-tubulin antibody. Results from 3 independent experiments are shown as means ± SD and statistically analyzed with a t test. *, P < .05 vs unstimulated COBs expressing the receptor; §, P < .05 vs unstimulated COBs expressing the receptor; #, P < .05 vs BMP-2-treated COBs expressing the receptor.

We also conducted similar FGFR1 knockdown experiments on FGF-2 transgenic COBs (Supplemental Data). As shown in Figure 7, the use of FGFR1 siRNA in Tg^Vector COBs and in Tg^LMW COBs decreased phospho-Smad in both the presence and absence of BMP-2 (Figure 7, A and B). Conversely, in Tg^HMW COBs, the decrease of phospho-Smad levels was observed only in vehicle-treated cells (Figure 7C). Moreover, the depletion of FGFR1 caused a decrease of phospho-p44/42 levels in Tg^Vector COBs and in Tg^LMW COBs, both under vehicle-treated and BMP-2-treated conditions (Figure 7, A and B). In addition, FGFR1 depletion blocked the up-regulation of phospho-p44/42 induced by BMP-2 in Tg^HMW COBs (Figure 7C). The levels of Runx-2 and osterix in the absence and the presence of BMP-2 in Tg^Vector COBs, lacking FGFR1, were similar to that found in Tg COBs expressing FGFR1 (Figure 7A). Finally, in Tg^LMW COBs, Runx-2 and osterix proteins were increased only in vehicle-treated conditions (Figure 7B), whereas in Tg^HMW COBs, decreased Runx-2 and osterix was observed in the presence and absence of BMP-2 (Figure 7C).

Discussion

In this study we evaluated whether the FGF-2 isoform-specific regulation of bone formation was regulated by interaction with BMP-2 that, in turn, regulates bone differentiation markers to control matrix mineralization. Our results clearly demonstrate a differential effect from the specific expression of the FGF-2 isoforms on bone formation, in which the exported LMW isoform enhanced matrix deposition, whereas the HMW isoforms had an inhibitory effect. These results are consistent with our previous observations of increased bone formation in vivo in Tg^LMW (32) and decreased bone formation in vivo in Tg^HMW mice (35).

The results also show a differential response for BMP-2 in FGF-2 isoform COBs in which BMP-2 enhances the anabolic effects of the LMW FGF-2 isoform but could not rescue the inhibitory effects of the HMW FGF-2 isoform on osteoblast differentiation and mineralization. The major signaling pathway for BMP-2 action occurs via the activation of phospho-Smad, but BMPs also activate the MAPK signaling pathway (2). In the present study, BMP-2 activated phospho-Smad 1/5/8 in Tg^Vector COBs as well as in Tg^LMW and in Tg^HMW COBs. Therefore, regulation of Smads does not mediate the differential effects of BMP-2 on bone formation in FGF-2 isoform mice, suggesting modulation through other signaling molecules. Indeed, we previously demonstrated that BMP-2 caused a similar induction of phospho-Smad in COBs from Fgf2^+/+ and Fgf2^−/− mice (30). These data extend those results in a very significant manner.

Both the p38 MAPK and ERKs p44/42 were reported to be important in BMP-2-induced osteoblast differentiation (33). In particular, the p38 MAPK pathway plays a positive role in BMP-2-induced osterix expression (36). The fact that BMP-2 increased phospho-p38 at all time points in Tg^Vector COBs but not in COBs from Tg^LMW or Tg^HMW mice suggests that p38 does not play a significant role in BMP-2 effects on bone formation in FGF-2 isoform-specific transgenic mice. Indeed, we found that phospho-p38 proteins were comparably regulated by BMP-2 in COBs from Fgf2^+/+ and Fgf2^−/− mice (30).

The function of MAPK in osteoblasts is still controversial. Indeed, it was reported that ERK activation stimulates Runx-2 activity and osteoblast differentiation (37, 38). Likewise, constitutively active MAPK kinase-1, a MAPK activator, induced Runx-2 stimulation (39), whereas FGF-2 also increased the stabilization, acetylation, and activation of Runx-2. Previously this function was restricted to ERK (40). In this context, we previously reported that PTH increased Runx-2 only in Fgf2^+/+ COBs (29). Other reports demonstrated that persistent ERK signaling is associated with inhibition of bone formation (14, 41, 42). Moreover, it was proposed that the pathogenesis of achrondroplasia may be mediated by inhibition of Smad 1/5/8 activity through increased MAPK phosphorylation (43). The new results presented here suggest that the marked and stable increase of phospho-p44/42 by BMP-2, observed only in Tg^HMW COBs, negatively affected the osteogenic master transcription factors, Runx-2 and osterix, which were found to be strongly decreased by these effectors. However, the robust and stable up-regulation of Runx-2 and osterix by BMP-2 that we found in Tg^LMW COBs, consistent with other reports (34, 44), could be partly correlated with the increase of phospho-Akt that occurred only in Tg^LMW COBs.

In support of this hypothesis, our recent demonstration that BMP-2 increased phospho-Akt in COBs from Fgf2^+/+ mice but decreased these phospho-proteins in COBs from Fgf2^−/− mice emphasizes the potential role of Akt in the BMP-2/FGF-2 signal (45). In COBs from vector mice, although Runx-2 was slightly increased only after 6 hours of treatment, osterix was up-regulated during the entire time period. The former result is consistent with previous studies, indicating that BMP-2-induced osterix mRNA expression is not dependent on Runx-2 (46). Thus, the results in toto add new insight to our previous data about the importance of the LMW FGF-2 isoform in mediating bone remodeling.

Current models hold that most of the biological functions for FGF-2 are mediated by binding to and activation of its specific tyrosine kinase receptors in the cell membrane, whereas other effects involve the translocation of the receptors into the nucleus (15, 26–28). Our data show, for the first time, differential basal expression of FGFR1, but not FGFR2, in COBs from FGF-2 isoform genotype mice. We also demonstrate, for the first time, that BMP-2 differentially regulates FGFR1 and FGFR2 synthesis in COBs. Moreover, the specific receptor regulation by BMP-2 is dependent on the overexpression of the specific FGF-2 isoforms. BMP-2 did not increase FGFR1 but did increase FGFR2 protein in Tg^LMW COBs, suggesting that FGFR2 but not FGFR1 may be important in the enhanced anabolic effect of BMP-2 on bone formation in Tg^LMW COBs. A role for FGFR2 in BMP-2-modulated intracellular signaling proteins and bone formation is further supported by the results from siRNA depletion of FGFR2, which showed a marked decreased of Runx-2 and osterix protein levels in vehicle-treated and BMP-2-treated Tg^LMW COBs. These results are consistent with previous studies that identified FGFR2 as an inducer of osteogenic differentiation of mouse calvarial osteoblasts (47, 48). Similarly, we observed a decrease of Runx-2 and osterix in the Tg^HMW COBs that were FGFR2 depleted. The notable increase of phospho-p44/42, which was found in the FGFR2-silenced Tg^HMW COBs that were either vehicle or BMP-2 treated, seems to confirm our previous observation that their activation is not involved in the regulation of Runx-2 and osterix; indeed, both proteins were down-regulated in these Tg COBs, with or without BMP-2 treatment.

Based on our data, FGFR1 may participate in the osteoblast differentiation process by playing a negative role in the untreated condition and a positive function in BMP-2-induced osteoblast maturation. Indeed, although we found that FGFR1 ablation in Tg^LMW COBs increases the basal level of Runx-2 and osterix proteins, the results also showed that the lack of FGFR1 blocks the increase of Runx-2 and osterix induced by BMP-2. In summary, this study, consistent with our previous in vivo studies showing the importance of the exported 18-kDa FGF-2 isoform in osteoblast differentiation (32), provides new knowledge regarding the influence of FGF-2 isoforms on signaling pathways implicated in bone remodeling processes. We previously reported that low doses of FGF-2 in combination with BMP-2 increases mineralization more than BMP-2 alone (49). The actual intriguing evidence, in which differential osteoblast responses to BMP-2 are clearly dependent on both the FGF-2 isoform species and the specific expression of FGFRs, will prove to be very useful in bone remodeling research, and, more importantly, it will provide additional insights into potential therapeutic applications of BMP-2.