Osteosarcoma is the most frequent, nonhematopoietic, primary malignant tumor of bone. Histopathologically, osteosarcoma is characterized by complex mixtures of different cell types with bone formation. The role of environmental factors in the formation of such a complicated tissue structure as osteosarcoma remains to be elucidated. Here, a newly established murine osteosarcoma model was used to clarify the roles of environmental factors such as fibroblast growth factor-2 (Fgf2) or leukemia-inhibitory factor (Lif) in the maintenance of osteosarcoma cells in an immature state. These factors were highly expressed in tumor environmental stromal cells, rather than in osteosarcoma cells, and they potently suppressed osteogenic differentiation of osteosarcoma cells in vitro and in vivo. Further investigation revealed that the hyperactivation of extracellular signal–regulated kinase (Erk)1/2 induced by these factors affected in the process of osteosarcoma differentiation. In addition, Fgf2 enhanced both proliferation and migratory activity of osteosarcoma cells and modulated the sensitivity of cells to an anticancer drug. The results of the present study suggest that the histology of osteosarcoma tumors which consist of immature tumor cells and pathologic bone formations could be generated dependent on the distribution of such environmental factors. The combined blockade of the signaling pathways of several growth factors, including Fgf2, might be useful in controlling the aggressiveness of osteosarcoma. Mol Cancer Res; 10(3); 454–68. ©2012 AACR.

Osteosarcoma is the most frequent, nonhematopoietic, primary malignant tumor of bone. Osteosarcoma is characterized by aggressive local growth and systemic hematogenous dissemination, which together confer a generally poor prognosis (1, 2). The identification of osteoid (dense, amorphous intercellular material of immature bone) production by tumor cells was one of the critical findings for the diagnosis of osteosarcoma (3). Histopathologically, osteosarcoma is characterized by complex mixtures of different cell types with bone formation (3). Previous reports have suggested that the response to therapies and prognosis of malignant tumors are greatly influenced by the differentiation state of tumor cells and by histologic subtypes (4, 5). Moreover, the extracellular matrix produced by tumor cells protects tumors from apoptosis induced by anticancer agents (6, 7). Considering these points, the difficulty of osteosarcoma treatment might be related to the tissue heterogeneity within osteosarcoma tumors.

Malignant tumors are composed of transformed neoplastic cells and tumor stroma containing a variety of extracellular matrix components and cell types such as fibroblasts, endothelial cells, and hematopoietic cells (8, 9). Accumulated evidence has clarified that cytokines, chemokines, and growth factors released from tumor stroma play critical roles in the promotion of tumor progression (10, 11). However, it remains to be elucidated how osteosarcoma microenvironments could contribute to osteosarcoma progression or the formation of such a complicated histopathology as the mixture of immature cells and osteogenic differentiation areas. Clarification of environmental effects may provide important clues towards the development of novel therapeutic approaches as well as the elucidation of osteosarcoma pathogenesis.

Previously, we developed a novel osteosarcoma mouse model by overexpressing c-MYC in bone marrow stromal cells (BMSC) derived from mice with a homozygous deletion of the Ink4a/Arf locus (12). We isolated highly tumorigenic cells (designated AX cells) from the BMSCs using single cell cloning. Inoculation of AX cells into syngeneic C57BL/6 mice results in the development of lethal osteosarcoma with metastatic lesions in various organs, including lung, which mimics human osteoblastic osteosarcoma. Although AX cells are maintained in an immature state in vitro, they generate mature bone and tumors that contain structures with heterogeneous differentiation patterns in vivo.

In the present study, fibroblast growth factor-2 (Fgf2), leukemia-inhibitory factor (Lif), and insulin-like growth factor 1 (Igf1) were focused on as the factors highly expressed in the tumor environment rather than in osteosarcoma cells. Further investigation revealed that the hyperphosphorylation of extracellular signal–regulated kinase (Erk)1/2 in AX cells induced by Fgf2 or Lif potently suppressed osteogenic differentiation. Erk1/2 hyperactivation was detected at the immature areas of tumors, suggesting that it plays key roles in maintaining immaturity in osteosarcoma cells in vivo. In addition, Fgf2 afforded a growth advantage and the enhancement of cellular motility, as well as resistance to Adriamycin, to AX cells, indicating the contribution of Fgf2 to tumor progression and refractoriness to treatment.

Our results suggest that the complex histology of osteosarcoma is generated, depending on the distribution of environmental growth factors.

Cell culture

Mouse osteosarcoma AX cells were established as previously described (12) and were cultured in Iscove's Modified Dulbecco's Medium (IMDM; Invitrogen) supplemented with 20% FBS. Human osteosarcoma cell lines (SAOS2, U2OS, and SJSA1) were purchased from American Type Culture Collection (ATCC) and cultured in Dulbecco's Modified Eagle's Medium (DMEM) high glucose (Invitrogen) supplemented with 10% FBS.

Differentiation assays

For the induction of osteogenesis, cells were spread on 8-well chamber culture slides (BD Biosciences) with 80% confluency. The next day, the culture medium was changed to IMDM supplemented with 10% FBS, 20 mmol/L β-glycerophosphate (Sigma-Aldrich), 50 ng/mL thyroxine (Sigma-Aldrich), 1 nmol/L dexamethasone (Sigma-Aldrich), and 0.5 μmol/L ascorbate-2-phosphate (Sigma-Aldrich). This culture medium was designated osteogenic medium (OG medium) in the article. In some assays, human FGF2 (Peprotech), mouse Lif (ESGRO; 1,000 U = 1 ng; Millipore), mouse Fgf1 (R&D), or mouse Igf1 (R&D) was added to osteogenic medium at the indicated concentrations. For the inhibition of Fgf receptor (Fgfr) or Mek, cells were cultured in osteogenic medium supplemented with Fgf2 or Lif plus each inhibitor, PD173074 (Sigma-Aldrich), and SU5402 (Calbiochem) for Fgfr inhibition, or U0216 (Calbiochem) and PD98059 (Cayman Chemical) for Mek inhibition, 1 hour after the pretreatment of each inhibitor alone in osteogenic medium. The medium was changed every third day. Cells were fixed in 4% paraformaldehyde for 20 minutes at room temperature and then stained with Alizarin Red S (Sigma-Aldrich) at pH 4.3.

Flow cytometry

Subcutaneous AX-derived osteosarcoma tumors were excised from euthanized mice. They were minced and incubated in IMDM supplemented with 2% FBS, 50 U/mL collagenase I (Worthington), 100 U/mL collagenase IV (Worthington), 2.5 U/mL collagenase XIV (Sigma-Aldrich), and 20 U/mL DNaseI (Sigma-Aldrich) for 2 hours at 37°C. The cell suspension was washed with PBS and filtered through a 40-μm cell strainer (BD Biosciences), and single-cell suspensions were prepared. For blocking of Fc receptors, the suspensions were treated with anti-mouse CD16/CD32 antibody (1:100; e-Bioscience) for 15 minutes on ice. Then they were incubated with phycoerythrin (PE)-conjugated antibodies to CD45, F4/80, and APC-conjugated Sca1 antibody (e-Bioscience). The labeled cells were analyzed with FACSCalibur and Cell Quest software (BD Biosciences). At least 10,000 live cells were examined. GFP-positive, GFP-negative-plus-F4/80-positive, or GFP/CD45-negative-plus-Sca1-positive fractions were sorted by Vantage SE flow cytometer (BD Biosciences).

Cell proliferation assay

Cells were transferred to 96-well tissue culture plates and cultured in IMDM supplemented with 10% FBS, Adriamycin, and/or the indicated growth factors. Cell proliferation was measured with the use of a CellTiter-Glo cell proliferation assay kit (Promega). All assays were conducted in triplicate.

RNA interference

Knockdown of Fgfr1 and Fgfr2 was carried out using premade Stealth select siRNA (FGFR1MSS204294, 204295, 274341, FGFR2MSS204296, 204298) purchased from Invitrogen. The control siRNA for firefly luciferase (GL2); 5′-CGUACGCGGAAUACUUCGATT-3′ was purchased from Japan Bio Services. A total of 50 nmol/L of each siRNA was transfected to AX cells according to the manufacturer's instruction and continuously treated during differentiation assays.

Immunoblot analysis

Cells were lysed with Laemmli sample buffer (BioRad) supplemented with β-mercaptoethanol (Sigma-Aldrich), and immunoblot analyses were conducted according to standard procedures with primary antibodies to phospho-Erk1/2 (Cell Signaling Technologies), Erk1/2 (Cell Signaling Technologies), phospho-Akt (Cell Signaling Technologies), Akt (Cell Signaling Technologies), Fgfr1 (Epitomics), Fgfr2 (Santa Cruz Biotechnology), and α-Tubulin (Sigma-Aldrich). In some assays, human FGF2, mouse Fgf1, mouse Lif, or mouse Igf1 was added and incubated for the indicated times. For the inhibition of Fgfr or Mek, cells were cultured with each factor and each inhibitor 1 hour after pretreatment with each inhibitor alone.

Tumor xenograft model

All animal care and procedures were conducted in accordance with the guidelines of Keio University (Tokyo, Japan). All mice were purchased from SLC. For xenograft experiments, AX cells were suspended in PBS and then injected subcutaneously into syngeneic 8-week-old female C57BL/6 mice. For intrabone marrow injection, AX cells were injected into a tiny hole made with a 23G needle at the intercondylar fossa of a femur. To overexpress Fgf2 and Lif in tumors in vivo, 2 million AX cells were mixed with Fgf2 polyhedra (125 ng of Fgf2 per an injection; refs. 13, 14) or Lif polyhedra (25 ng of Lif per an injection; ref. 15) suspended in PBS containing 50 ng/mL Fgf2 or 1,000 U/mL Lif, respectively and subcutaneously injected. Two weeks later, mice were euthanized and tumors were subjected to immunohistochemistry. To investigate the effect of a combined treatment of PD173074 (Sigma-Aldrich) and Adriamycin, 2 million AX cells were subcutaneously inoculated (day 0). Eleven days later, PD173074 was administered intraperitoneally at 10 mg/kg every other day 6 times in total. Eight mg/kg of Adriamycin was intravenously injected on day 14. On day 22, all mice were euthanized and tumors were subjected to analyses.

Gene expression profiling

The gene expression profiling of AX cells sorted from tumors was carried out using the 3D-Gene Chip (Toray) as previously described (12).

Real-time reverse transcriptase PCR analysis

Total RNA was extracted with RNeasy Mini Spin columns (Qiagen) and was subjected to reverse transcriptase by the Prime Script RT-PCR kit (Takara).

Real-time reverse transcriptase (RT)-PCR analysis was conducted using SYBR Premix Ex TaqII and Thermal Cycler Dice (Takara). Data are represented as means ± SD from 3 independent experiments. The sequences of the primers used in this study are shown in Supplementary Table S1.

Immunohistochemistry

Immunohistochemical analysis was conducted using standard methods. Deparaffinized sections were stained with antibodies to GFP, Fgf2, Lif, Runx2, Fgfr2 (Santa Cruz Biotechnology), Igf1 (R&D), Fgfr1 (Epitomics), Fgfr3 (Assay Biotechnology), CD31 (Abcam), or phospho-Erk1/2. For anti-mouse primary antibodies, the pretreatment was carried out by MOM kit (Vector Laboratories) to avoid nonspecific staining. Immune complexes were detected with the use of Histofine (Nichirei Bioscience) and Simple Stain kit (Nichirei Bioscience). For immunofluorescence staining, deparaffinized sections were stained with rabbit polyclonal anti-GFP (Santa Cruz Biotechnology) or mouse monoclonal anti-GFP antibody (MBL) followed by Alexa488-conjugated anti-rabbit IgG or anti-mouse IgG secondary antibody (Invitrogen), respectively, anti-F4/80 antibody followed by Alexa555-conjugated anti-rat IgG secondary antibody (Invitrogen), anti-Fgf2 antibody followed by Alexa555-conjugated anti-rabbit IgG secondary antibody (Invitrogen), or an APC-conjugated Sca1 antibody (e-Bioscience). TOTO3 (Invitrogen) was used for nuclear staining. Prepared samples were observed with confocal microscopy (LSM510; Zeiss) and analyzed with an LSM image browser (Zeiss).

Quantitative protein analysis of Fgf2, Lif, and Vegfa

To acquire tumor homogenate, tumors and adjacent host tissues were suspended with an equivalent volume of cell lysis buffer (MBL) and mashed with Biomasher (Nippi). The concentration of Fgf2, Lif, and Vegfa in the lysates was quantified by Bio-Plex Pro Mouse Cytokine 9-plex Assay G2 (BioRad) according to the manufacturer's instruction.

Migration assay

Transwell migration assays were conducted with Transwell membrane filter inserts with 8-μm pores (BD Biosciences) in 24-well tissue culture plates. AX cells (50,000 cells) were replated onto the upper chamber in 500 μL of serum-free IMDM supplemented with 50 ng/mL of Fgf2, 2,000 U/mL of Lif, 10 ng/mL of Igf1, and the mixture of each factor. The chamber was placed in 750 μL of IMDM supplemented with 1% FBS with the same concentration of each factor as the upper chamber. After 4 hours of incubation at 37°C in 5% CO2, cells on the upper surface of the filter were removed by wiping with a moist cotton swab. The cells migrated through the filter and adhered to its lower surface, where they were fixed, stained by a Diff-Quick kit (Sysmex) and counted under the microscope. Each assay was conducted in triplicate.

Immature osteosarcoma cells were located in the periphery of tumors

We recently developed a mouse osteosarcoma model (12). Highly tumorigenic AX cells, which overexpressed enhanced GFP (EGFP) in addition to c-MYC, were isolated from the BMSCs by single cell cloning. Upon subcutaneous injection into syngeneic C57BL/6 mice, AX cells developed lethal metastatic osteosarcoma (12). Tumors consisted of both immature cells and the region of mature bone formation (Fig. 1A). Notably, immature cells were accumulated in the peripheral part of the tumor, whereas bone formations were apparent inside the tumor (Fig. 1A). We speculated that the differentiation state of tumor cells was affected by the components of the tumor microenvironment, resulting in the formation of histologic heterogeneity of osteosarcoma.

Figure 1.

Histology of mouse osteosarcoma. A, Alizarin Red S (left) and hematoxylin–eosin (H&E) staining of mouse osteosarcoma (OS) serial sections. Boxed regions are shown at a higher magnification in the right. P, immature cells were enriched at the peripheral part of osteosarcoma; I, osteogenesis with marked osteoid production was apparent mainly inside the tumor. B, osteosarcoma tumors were dissected and minced into single-cell suspension and analyzed for GFP, CD45, and Sca1 positivity by flow cytometry. C, CD45- and F4/80-positive hematopoietic cells in osteosarcoma were analyzed by flow cytometry. D, morphology of sorted cells. GFP-positive osteosarcoma cells and F4/80-positive hematopoietic cells were cytospun and stained with May–Giemsa. The primary culture of Sca1-positive fibroblasts is also shown. E, confocal microscopic analysis of GFP- and F4/80-expressing cells in an osteosarcoma tumor. TOTO3 was used for nuclear staining. F, immunofluorescence analysis of GFP and Sca1 expression in an osteosarcoma tumor. Sca1-positive cells were located around the GFP-positive tumor cells. SSC, side scatter; FSC, forward scatter.

Figure 1.

Histology of mouse osteosarcoma. A, Alizarin Red S (left) and hematoxylin–eosin (H&E) staining of mouse osteosarcoma (OS) serial sections. Boxed regions are shown at a higher magnification in the right. P, immature cells were enriched at the peripheral part of osteosarcoma; I, osteogenesis with marked osteoid production was apparent mainly inside the tumor. B, osteosarcoma tumors were dissected and minced into single-cell suspension and analyzed for GFP, CD45, and Sca1 positivity by flow cytometry. C, CD45- and F4/80-positive hematopoietic cells in osteosarcoma were analyzed by flow cytometry. D, morphology of sorted cells. GFP-positive osteosarcoma cells and F4/80-positive hematopoietic cells were cytospun and stained with May–Giemsa. The primary culture of Sca1-positive fibroblasts is also shown. E, confocal microscopic analysis of GFP- and F4/80-expressing cells in an osteosarcoma tumor. TOTO3 was used for nuclear staining. F, immunofluorescence analysis of GFP and Sca1 expression in an osteosarcoma tumor. Sca1-positive cells were located around the GFP-positive tumor cells. SSC, side scatter; FSC, forward scatter.

Close modal

First, we tried to identify the cellular components of osteosarcoma. Single-cell suspensions from AX-derived osteosarcoma tumors were stained with PE-conjugated anti-CD45 antibody and analyzed by flow cytometry (Fig. 1B). Three fractions were identified in osteosarcoma tumors: GFP-positive AX cells, CD45-positive hematopoietic cells, and the remaining fraction without GFP or CD45 expression (Fig. 1B). Because the majority of CD45-positive hematopoietic cells also expressed a mature macrophage marker, F4/80 (Fig. 1C), F4/80-positive cells were sorted and used for further analyses (Fig. 1D). These macrophages highly express genes related to the M2 phenotype such as Mmp1, Tnfα, Il10, and Arginase, suggesting that they seem to be tumor-associated macrophages (16, 17 and data not shown). Immunofluorescence analysis revealed that macrophages gathered around and inside the osteosarcoma tumors (Fig. 1E). GFP-positive AX cells were small and exhibited a high nuclear/cytosolic ratio (Fig. 1D). The third fraction, without GFP or CD45 expression, contained a heterogeneous population of cells composed mainly of fibroblastic stromal cells, as well as debris. To enrich the live fibroblasts, Stem cell antigen 1 (Sca1) was used, which was suggested to be expressed in the multipotent dermal fibroblasts (18, 19). Live fibroblasts were successfully enriched with an APC-conjugated Sca-1 antibody and these were located around the tumors (Fig. 1B, D, and F). Collectively, AX-derived tumors were consisted of fibroblasts and macrophages as well as osteosarcoma cells, and the interfacial regions of tumors were composed of immature tumor cells.

Environmental factors promote osteosarcoma tumor cell proliferation and suppress osteogenesis

Detailed observations of mouse subcutaneous osteosarcoma indicated that tumor cells that did not directly contact surrounding nontumor cells also exhibited immature states (Fig. 1A). Thus, to elucidate the environmental effects on the generation of histologic heterogeneity of osteosarcoma, we focused on the soluble factors, rather than the molecules involved in direct cell to cell contact. The gene expression profile of tumor cells sorted from AX-derived osteosarcoma tumors was used for the initial screening. Given that we intended to exclude the effects of factors released from tumor cells, we focused on secreted soluble factors whose gene expression levels were extremely low in tumor cells (Supplementary Table S2). Among these factors the mRNA levels of Fgf1, Fgf2, Igf1, Igf2, Lif, and hepatocyte growth factor (Hgf) were evaluated in tumor cells and sorted environmental components by real-time RT-PCR analysis. Fgf1, Fgf2, Lif, and Igf1 exhibited markedly higher expression in macrophages or fibroblasts than tumor cells (Fig. 2A and data not shown) and they were subjected to further analyses. These factors afforded a growth advantage to AX cells in vitro (Fig. 2B). Fgf1, Fgf2, and Lif significantly enhanced cell growth and Fgf2 exhibited the most potent effect. Igf1 also showed a slight tendency to promote cell growth, albeit without statistical significance. Importantly, the mixture of Fgf2, Lif, and Igf1 possessed additive effect compared with treatment of each factor alone.

Figure 2.

Environmental growth factors support osteosarcoma cell growth and suppress osteogenesis. A, real-time RT-PCR analyses of Fgf1, Fgf2, Lif, and Igf1 genes in AX cells and sorted cells from osteosarcoma tumors. Results are shown as the ratio of the abundance of each mRNA to that of glyceraldehyde-3-phosphate dehydrogenase (Gapdh; n = 3). B, cell proliferation was analyzed with or without the indicated factors. Data are shown as the ratio of the measurement of each time point to that of day 0 and represented as means ± SD from 3 independent experiments. *, P < 0.05; **, P < 0.005; NS, not significant (the Student t test). C, osteogenesis was evaluated by Alizarin Red S staining of cells in osteogenic medium with or without the indicated factors for 7 days. OG, osteogenic culture condition.

Figure 2.

Environmental growth factors support osteosarcoma cell growth and suppress osteogenesis. A, real-time RT-PCR analyses of Fgf1, Fgf2, Lif, and Igf1 genes in AX cells and sorted cells from osteosarcoma tumors. Results are shown as the ratio of the abundance of each mRNA to that of glyceraldehyde-3-phosphate dehydrogenase (Gapdh; n = 3). B, cell proliferation was analyzed with or without the indicated factors. Data are shown as the ratio of the measurement of each time point to that of day 0 and represented as means ± SD from 3 independent experiments. *, P < 0.05; **, P < 0.005; NS, not significant (the Student t test). C, osteogenesis was evaluated by Alizarin Red S staining of cells in osteogenic medium with or without the indicated factors for 7 days. OG, osteogenic culture condition.

Close modal

Next, the effects of these factors on the osteogenic ability of AX cells were evaluated. After 7 days of culture in the osteogenic medium, AX cells produced mineralized materials and exhibited osteogenesis, as evaluated by the intracellular calcium deposition with Alizarin Red S staining (Fig. 2C), whereas under the control nonosteogenic medium AX cells did not exhibit osteogenesis. The supplementation of Fgf2 or Lif into the osteogenic medium potently suppressed these osteogenic processes resulting in negative staining for Alizarin Red S (Fig. 2C). Fgf1 also exhibited suppressive effect on osteogenesis, although the higher dose was required than Fgf2. The effect of Igf1 was modest and osteogenic ability in AX cells was still maintained at 1 μg/mL of Igf1. Importantly, these factors exhibited an additive effect and the combination of each factor at an insufficient dose to inhibit osteogenesis could potently block the differentiation process (Fig. 2C). These findings indicated that several environmental factors, such as Fgf1, Fgf2, or Lif may contribute to both the suppression of osteogenic differentiation and the enhancement of proliferation in osteosarcoma cells.

The treatment of Fgf2 hyperactivated Erk1/2 in AX cells

To investigate the molecular mechanisms that affect osteogenic ability, the downstream signaling pathways of receptors of Fgf, Lif, and Igf1 were analyzed by Western blotting. Previous reports indicated that the Map/Erk kinase (Mek)-Erk1/2 pathway or phosphoinositide 3-kinase (PI3K)-Akt pathways were activated by treatment with these factors (20–22). Although the phosphorylation level of Erk1/2 in AX cells increased after treatment with Fgf1, Fgf2, Lif, and Igf1, Fgf2 was the most potent activator among them (Fig. 3A and B; Supplementary Fig. S1A). The Erk1/2 phosphorylation was enhanced 20 minutes after Fgf2 treatment, and continued more than 6 hours, whereas in the cases of Fgf1, Lif, or Igf1 treatment, this activation returned to the same level as the normal culture condition in 6 hours (Fig. 3C; Supplementary Fig. S1A). Moreover, the phosphorylation level in AX cells treated with Igf1 was weaker compared with those treated with other factors. Stat3 was also phosphorylated by Lif treatment, as previously reported (21). Akt was constitutively phosphorylated even under normal culture conditions. Its phosphorylation was markedly enhanced by treatment with Igf1 (Fig. 3D). In contrast, treatment with Fgf1, Fgf2, and Lif scarcely affected the phosphorylation status of Akt in AX cells (Fig. 3D; Supplementary Fig. S1B and data not shown). Moreover, Fgf2 supplemented with Fgfr inhibitors did not change Akt activation (Supplementary Fig. S1C). These results indicated that Fgf1, Fgf2, and Lif potently induced the phosphorylation of Erk1/2 without hyperactivating Akt, whereas Igf1 was mainly involved in the activation of the PI3K-Akt pathway in AX cells.

Figure 3.

Fgf and Lif potently activated Erk1/2 in AX cells. A, Western blot analysis of phospho-Stat3, Stat3, phospho-Erk1/2, and Erk1/2 in AX cells treated with Fgf2 (20 ng/mL) and Lif (2,000 U/mL) for the indicated times. α-Tubulin was used as a loading control. B, phospho-Erk1/2 and Erk1/2 were assessed by immunoblotting. AX cells were treated with Fgf2 (20 ng/mL) or Igf1 (100 ng/mL) for the indicated times. α-Tubulin was used as a loading control. C, quantification of the phosphorylation level of Erk2 in AX cells treated with indicated factors for 6 hours. Results are shown by the ratio of the density of phospho-Erk2 (p-Erk2) to that of α-tubulin (n = 3). *, P < 0.05 (the Student t test). D, Western blot analysis of phospho-Akt and Akt in AX cells treated with Fgf2 (20 ng/mL) or Igf1 (100 ng/mL) for the indicated times. For Akt phosphorylation, photographs with both short- and long-time exposure are shown. α-Tubulin was used as a loading control.

Figure 3.

Fgf and Lif potently activated Erk1/2 in AX cells. A, Western blot analysis of phospho-Stat3, Stat3, phospho-Erk1/2, and Erk1/2 in AX cells treated with Fgf2 (20 ng/mL) and Lif (2,000 U/mL) for the indicated times. α-Tubulin was used as a loading control. B, phospho-Erk1/2 and Erk1/2 were assessed by immunoblotting. AX cells were treated with Fgf2 (20 ng/mL) or Igf1 (100 ng/mL) for the indicated times. α-Tubulin was used as a loading control. C, quantification of the phosphorylation level of Erk2 in AX cells treated with indicated factors for 6 hours. Results are shown by the ratio of the density of phospho-Erk2 (p-Erk2) to that of α-tubulin (n = 3). *, P < 0.05 (the Student t test). D, Western blot analysis of phospho-Akt and Akt in AX cells treated with Fgf2 (20 ng/mL) or Igf1 (100 ng/mL) for the indicated times. For Akt phosphorylation, photographs with both short- and long-time exposure are shown. α-Tubulin was used as a loading control.

Close modal

Fgf2 reversibly inhibited the mRNA expression of osteogenic genes in AX cells

Given that Fgf2 enhanced cell growth and potently suppressed osteogenesis of AX cells in vitro (Fig. 2), its function was further assessed in osteosarcoma tissues. Immunofluorescence analyses revealed that Fgf2 was distributed mainly in the surrounding tissues and interfacial regions between tumor and stromal cells (Fig. 4A). Consistent with the gene expression patterns (Fig. 2A), Lif and Igf1 were also enriched in the surrounding tissues rather than inside the tumors (Supplementary Fig. S2A and S2B).

Figure 4.

Fgf2 reversibly suppressed osteogenesis in AX cells. A, Fgf2 localization in osteosarcoma tumors was assessed by immunofluorescence analysis. Tumor cells were identified by GFP staining. TOTO3 was used for nuclear staining. BF, bright field. B, immunohisto-chemistry of serial sections of an AX-derived osteosarcoma tumor for GFP, Fgfr1, Fgfr2, and Fgfr3. C, real-time RT-PCR analysis of Fgfr genes in the indicated cells. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). D, AX cells were cultured in osteogenic (OG) medium supplemented with Fgf2 (20 ng/mL) according to the indicated schedule, and the expression of osteogenic genes was assessed by real-time RT-PCR analysis. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). *, P < 0.05; **, P < 0.005; NS, not significant (the Student t test). E, immunohistochemistry of serial sections of an AX-derived osteosarcoma tumor for Fgf2 and Runx2. Boxed regions are shown at a higher magnification in the right.

Figure 4.

Fgf2 reversibly suppressed osteogenesis in AX cells. A, Fgf2 localization in osteosarcoma tumors was assessed by immunofluorescence analysis. Tumor cells were identified by GFP staining. TOTO3 was used for nuclear staining. BF, bright field. B, immunohisto-chemistry of serial sections of an AX-derived osteosarcoma tumor for GFP, Fgfr1, Fgfr2, and Fgfr3. C, real-time RT-PCR analysis of Fgfr genes in the indicated cells. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). D, AX cells were cultured in osteogenic (OG) medium supplemented with Fgf2 (20 ng/mL) according to the indicated schedule, and the expression of osteogenic genes was assessed by real-time RT-PCR analysis. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). *, P < 0.05; **, P < 0.005; NS, not significant (the Student t test). E, immunohistochemistry of serial sections of an AX-derived osteosarcoma tumor for Fgf2 and Runx2. Boxed regions are shown at a higher magnification in the right.

Close modal

We also carried out intrabone marrow injections of AX cells to generate orthotopic osteosarcoma model (Supplementary Fig. S2C). The intrabone tumor also showed bone differentiation accompanied by osteoid formations inside the tumors, and Fgf2 or Lif was stained in the surrounding regions of tumor cells. These factors appeared to be abundantly present inside normal bone marrow, suggesting that Fgf2 and Lif also modulate osteogenic differentiation of osteosarcoma cells in a bone microenvironment where human osteosarcoma tumors usually originate.

Quantitative protein analysis of Fgf2 and Lif revealed that the concentration from subcutaneous tumor homogenate was 668.1 ± 182.6 pg/mL (n = 3) and 39.12 ± 9.87 pg/ml (n = 3), respectively. Although these concentrations were lower than those we used in the assays in vitro, these factors were possibly concentrated into higher values in the limited areas where tumor cells closely contact with adjacent stromal cells that produce these factors.

Previous reports suggested that Fgf1 and Fgf2 bind with high affinity to Fgfr1c, 2b, 2c, and 3c and Fgfr1c and 2c, respectively (23). Immunohistochemical staining for Fgfrs revealed that Fgfr1 and 3 were expressed both in tumor and stromal cells, and Fgfr2, which was thought to be a mesenchymal Fgfr2c isofom, was strongly positive in stromal cells compared with tumor cells (Fig. 4B). These findings were largely compatible with the gene expression patterns (Fig. 4C). Interestingly, the gene expression levels of Fgfr2 and 3 were different between AX cells and those sorted from tumors. These results might reflect to the altered response of AX cells to the ligands after tumor development in vivo. In addition, the immunohistochemical staining of Fgfr1 indicated that the expression level of Fgfr1 in the fibroblasts surrounding the tumors was variable (Fig. 4B), which might be the cause of a large error bar of Fgfr1 expression in the sorted fibroblasts (Fig. 4C). It is possible that fibroblasts surrounding tumors contain heterogeneous populations which play different roles responding to the ligands.

To investigate the mechanisms underlying the suppression of osteogenesis by Fgf2, we conducted real-time RT-PCR analyses of AX cells cultured in osteogenic medium supplemented with Fgf2 and evaluated the expression levels of genes that are involved in osteogenesis. The expression of all osteogenic genes examined, including Runx2, Sp7, Alp, Bglap1, and Ibsp, was downregulated after 4 days' culture supplemented with Fgf2 compared with those in osteogenic medium alone (Fig. 4D). Notably, the downregulated expression of osteogenic genes could be reverted by withdrawal of Fgf2 from the conditioned medium in a time-dependent manner. These results suggested that the suppression of osteogenesis by Fgf2 was reversible in AX cells and that, after the cancellation of the Fgf2 effect, the osteogenic process in osteosarcoma tumors could be restored to the previous levels. Nuclear Runx2 was strongly positive inside the tumors, whereas it was negative at the peripheral Fgf2-rich areas (Fig. 4E), suggesting that the suppression of osteogenic gene expressions by Fgf2 could also occur in vivo.

FGF2 inhibits osteogenesis in a human osteosarcoma cell line

Next, we investigated the effects of FGF2 on human osteosarcoma cell lines including U2OS, SAOS2, and SJSA1. Growth of these 3 cells was slightly enhanced by the treatment with FGF2, although the effect was modest compared with AX cells (Fig. 5A). Whereas U2OS and SJSA1 did not show any osteogenic differentiation, SAOS2 exhibited osteogenesis which was suppressed by FGF2 treatment similar to AX cells (Fig. 5B). In addition, the gene expression levels of osteogenic markers were attenuated by FGF2 in SAOS2 and U2OS cells (Fig. 5C). The phosphorylation level of ERK1/2 was enhanced in all osteosarcoma cell lines examined (Fig. 5D). Collectively, some characteristic effects of FGF2 on our established osteosarcoma mouse model were also observed in other osteosarcoma systems.

Figure 5.

Fgf2 suppressed osteogenesis in human osteosarcoma cells. A, cell proliferation was analyzed with or without 20 ng/mL of FGF2. Data are shown as the ratio of the measurement of each time point to that of day 0 and represented as means ± SD from 3 independent experiments. *, P < 0.05; **, P < 0.005; NS, not significant (the Student t test). B, osteogenesis was evaluated by Alizarin Red S staining of cells in osteogenic medium with or without FGF2 for 14 days. OG, osteogenic culture condition. C, human osteosarcoma cells were cultured with or without FGF2 (20 ng/mL) for 1 day. The expression of osteogenic genes was assessed by real-time RT-PCR analysis. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). *, P < 0.05 (the Student t test). D, phospho-ERK1/2 and ERK1/2 were assessed by immunoblotting. Osteosarcoma cells were treated with FGF2 (20 ng/mL) for the indicated times. α-Tubulin was used as a loading control.

Figure 5.

Fgf2 suppressed osteogenesis in human osteosarcoma cells. A, cell proliferation was analyzed with or without 20 ng/mL of FGF2. Data are shown as the ratio of the measurement of each time point to that of day 0 and represented as means ± SD from 3 independent experiments. *, P < 0.05; **, P < 0.005; NS, not significant (the Student t test). B, osteogenesis was evaluated by Alizarin Red S staining of cells in osteogenic medium with or without FGF2 for 14 days. OG, osteogenic culture condition. C, human osteosarcoma cells were cultured with or without FGF2 (20 ng/mL) for 1 day. The expression of osteogenic genes was assessed by real-time RT-PCR analysis. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). *, P < 0.05 (the Student t test). D, phospho-ERK1/2 and ERK1/2 were assessed by immunoblotting. Osteosarcoma cells were treated with FGF2 (20 ng/mL) for the indicated times. α-Tubulin was used as a loading control.

Close modal

The suppression of osteogenesis by Fgf2 in osteosarcoma was dependent on Erk1/2 phosphorylation

Fgf2 highly activates Erk1/2 (Fig. 3A) and potently suppresses osteogenesis (Fig. 2C) in AX cells in vitro. Immunohistochemical staining of serial sections of AX-derived osteosarcoma tumors revealed that Erk1/2 was highly activated mainly in the interfacial regions between tumors and host tissues (Fig. 6A). Because Erk1/2 was activated mainly in the regions where immature tumor cells were found, we speculated that such a hyperactivation of Erk1/2 by environmental factors might play critical roles in the suppression of osteogenesis in osteosarcoma tumor tissues.

Figure 6.

Osteogenesis in AX cells was rescued by the inhibition of Erk1/2 phosphorylation. A, immunohistochemistry of serial sections of an AX-derived osteosarcoma tumor for phospho-Erk1/2. Tumor cells were identified by GFP positivity. B, AX cells were cultured under osteogenic medium supplemented with or without Fgf2 (20 ng/mL), Fgfr inhibitors (SU5402, 10 μg/mL; PD173074, 1 μmol/L), or Mek inhibitors (U0126, 50 μmol/L; PD98059, 50 μmol/L) for the indicated times. Phosphorylation of Erk1/2 was assessed by Western blotting. α-Tubulin was used as a loading control. C, osteogenesis was evaluated by Alizarin Red S staining of AX cells in osteogenic medium supplemented with or without Fgf2 (20 ng/mL) or Fgfr inhibitors (SU5402 or PD173074) at the indicated concentrations for 7 days. D, osteogenesis was evaluated by Alizarin Red S staining of AX cells in osteogenic medium supplemented with or without Fgf2 or Mek inhibitors (PD98059 or U0126) at the indicated concentrations for 7 days. Suppression of osteogenesis by Lif (2,000 U/mL) in AX cells was also assessed by the same method. E, AX cells were cultured in osteogenic medium supplemented with or without Fgf2 (20 ng/mL), SU5402 (SU, Fgfr inhibitor, 10 μg/mL), PD173074 (PD, Fgfr inhibitor, 1 μmol/L), or U0126 (Mek inhibitor, 50 μmol/L) for 1 day. The expression of osteogenic genes was assessed by real-time RT-PCR analysis. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). *, P < 0.05; **, P < 0.005 (the Student t test). F, immunohistochemistry of serial sections of an AX-derived osteosarcoma tumor. Black arrowheads and red arrowheads indicate the locations of Fgf2 polyhedra and osteogenic differentiation, respectively. Boxed regions are shown at a higher magnification in the bottom.

Figure 6.

Osteogenesis in AX cells was rescued by the inhibition of Erk1/2 phosphorylation. A, immunohistochemistry of serial sections of an AX-derived osteosarcoma tumor for phospho-Erk1/2. Tumor cells were identified by GFP positivity. B, AX cells were cultured under osteogenic medium supplemented with or without Fgf2 (20 ng/mL), Fgfr inhibitors (SU5402, 10 μg/mL; PD173074, 1 μmol/L), or Mek inhibitors (U0126, 50 μmol/L; PD98059, 50 μmol/L) for the indicated times. Phosphorylation of Erk1/2 was assessed by Western blotting. α-Tubulin was used as a loading control. C, osteogenesis was evaluated by Alizarin Red S staining of AX cells in osteogenic medium supplemented with or without Fgf2 (20 ng/mL) or Fgfr inhibitors (SU5402 or PD173074) at the indicated concentrations for 7 days. D, osteogenesis was evaluated by Alizarin Red S staining of AX cells in osteogenic medium supplemented with or without Fgf2 or Mek inhibitors (PD98059 or U0126) at the indicated concentrations for 7 days. Suppression of osteogenesis by Lif (2,000 U/mL) in AX cells was also assessed by the same method. E, AX cells were cultured in osteogenic medium supplemented with or without Fgf2 (20 ng/mL), SU5402 (SU, Fgfr inhibitor, 10 μg/mL), PD173074 (PD, Fgfr inhibitor, 1 μmol/L), or U0126 (Mek inhibitor, 50 μmol/L) for 1 day. The expression of osteogenic genes was assessed by real-time RT-PCR analysis. Results are shown as the ratio of the abundance of each mRNA to that of Gapdh. (n = 3). *, P < 0.05; **, P < 0.005 (the Student t test). F, immunohistochemistry of serial sections of an AX-derived osteosarcoma tumor. Black arrowheads and red arrowheads indicate the locations of Fgf2 polyhedra and osteogenic differentiation, respectively. Boxed regions are shown at a higher magnification in the bottom.

Close modal

To verify this notion, osteogenic ability under the blockade of Fgfr or Mek signaling pathways was analyzed with each inhibitor. Erk1/2 was strongly phosphorylated by treatment with Fgf2 in AX cells, within 60 minutes (Fig. 6B). Supplementation of an Fgfr inhibitor, PD173074, SU5402, or a Mek inhibitor, U0126, almost completely blocked this activation process. Another Mek inhibitor; PD98059 also suppressed the activation of Erk1/2; however, this effect was less potent. Next, whether these inhibitors could rescue the suppression of osteogenesis by Fgf2 was examined. Reflecting the potent inhibition of the Fgfr signaling pathway as shown in Fig. 6B, supplementation with SU5402 or PD173074 successfully rescued the osteogenic differentiation suppressed by Fgf2 (Fig. 6C). Treatment with DMSO alone (1:200) did not affect the osteogenic process in AX cells. The knockdown of both Fgfr1 and Fgfr2 expressions by siRNA also appeared to rescue the osteogenic differentiation suppressed by Fgf2 (Supplementary Fig. S3A and S3B). In addition, osteogenic ability blocked by Fgf2 in AX cells was restored by supplementation with U0126 (Fig. 6D). The other Mek inhibitor, PD98059, was not able to rescue osteogenesis. This appeared to reflect an insufficient level of Mek inhibition by PD98059 compared with U0126 in AX cells (Fig. 6B). Of note, AX cells did not show osteogenesis under treatment with each inhibitor alone, without osteogenic medium (data not shown). Also, importantly, the suppression of osteogenesis by Lif, which activates Erk1/2 more weakly than Fgf2 (Fig. 3A), could be restored by treatment with PD98059 (Fig. 6D). Consistent with these findings, the treatment of U0126 at a lower concentration compared with those in the Fgf2 experiment could also rescue osteogenic ability of AX cells inhibited by Lif treatment (Fig. 6D).

Next, it was further examined whether the suppression of osteogenic gene expression by Fgf2 treatment could be reverted through supplementation with Fgfr or Mek inhibitors. The expression levels of Runx2, Sp7, and Alp, all of which critically involve in osteogenesis, were downregulated upon 1 day of treatment with Fgf2 (Fig. 6E). Notably, supplementation with SU5402, PD173074, or U0126 could restore the expression of all genes. Collectively, Fgf2 suppressed the osteogenic process in AX cells at the mRNA level and the mechanism might be dependent on the hyperactivation of Erk1/2.

To further evaluate the earlier findings in vivo, AX cells were coinjected with Fgf2 or Lif polyhedra which were the cubic protein microcrystals continuously releasing each factor for more than 1 month (13, 14, 15). Fgf2 and Lif were successfully enriched in the surrounding areas of each polyhedra where Erk also appeared to be hyperphosphorylated (Supplementary Fig. S4A and S4B). GFP-negative fibroblastic cells proliferated in the polyhedron abundant regions. Notably, Alizarin Red S positive osteogenic differentiation was not found near around the Fgf2 or Lif polyhedra (Fig. 6F; Supplementary Fig. S4B). These findings suggest that Fgf2 and Lif also hyperactivate Erk1/2 and suppress osteogenesis of osteosarcoma cells in vivo as well.

Fgf2 enhanced the motility of AX cells and afforded chemoresistance

As shown in the earlier analyses, environmental factors promoted cellular proliferation and suppressed osteogenic differentiation of osteosarcoma cells. The impacts of these factors on osteosarcoma tumor progression were further assessed.

At first, cellular motility was evaluated by Transwell migration assays with Transwell membrane filter inserts (24, 25). Although AX cells exhibited high basal levels of motility, with more than 300 cells migrating through Transwell filters over 4 hours, this motility was enhanced to levels 1.5 times that of the basal level in the presence of Fgf2 (Fig. 7A and B). Given that the proliferative effect due to cell division could be ignored in a 4-hour observation period (data not shown), the enhancement of cell motility by Fgf2 was significant. Supplementation with Lif or Igf1 also tended to promote cellular motility (Fig. 7B). Notably, the motility was significantly enhanced with the combined treatment of these factors compared with that of each factor alone. The invasion ability of AX cells was not enhanced in the presence of Fgf2 or Lif, although both factors showed a slight tendency to enhance the invasion ability of AX cells (data not shown). These findings indicated that Fgf2 could also affect the cellular motility of osteosarcoma cells.

Figure 7.

Fgf2 enhanced migration of osteosarcoma cells and afforded chemoresistance. A, the microscopic findings of migrated AX cells through Transwell membranes. B, migrated cells were counted and data are represented as means ± SD from 3 independent experiments. *, P < 0.05; **, P < 0.001 (the Student t test). C, cell proliferation of AX cells 2 days after treatment with Adriamycin (ADR) at the indicated concentrations was analyzed in the presence or absence of Fgf2 (20 ng/mL). Data are shown as the ratio of the measurement in the presence of each concentration of Adriamycin to that in the absence of Adriamycin and represented as means ± SDs from 3 independent experiments. *, P < 0.05; NS, not significant (the Student t test). D, a macroscopic finding of osteosarcoma tumors developed in mice with or without PD and/or Adriamycin treatment. E, tumor weight of AX-derived tumors developed in mice treated with or without PD and/or Adriamycin. F, immunohistochemistry of AX-derived osteosarcoma tumors in mice treated with each drug. Boxed regions are shown at a higher magnification in the bottom. G, the proposed model for osteosarcoma progression. Microenvironmental stromal cells release factors such as Fgf2, which maintain the immature state of osteosarcoma cells. Osteosarcoma cells inside the tumor exhibit osteogenesis accompanied by the withdrawal of the effects of environmental factors. In addition, these factors can enhance proliferation and motility of osteosarcoma cells. Stroma remodeling by the proliferation of fibroblasts and invasion of macrophages supports the dissemination of osteosarcoma cells. H, immunohistochemistry of an AX-derived osteosarcoma tumor for CD31. Both, a peripheral and an intratumor central region are shown. I, immunofluorescence analysis of Ki-67 expression in the peripheral or central part of an AX-derived osteosarcoma tumor. Tumor cells were identified by GFP staining. TOTO3 was used for nuclear staining.

Figure 7.

Fgf2 enhanced migration of osteosarcoma cells and afforded chemoresistance. A, the microscopic findings of migrated AX cells through Transwell membranes. B, migrated cells were counted and data are represented as means ± SD from 3 independent experiments. *, P < 0.05; **, P < 0.001 (the Student t test). C, cell proliferation of AX cells 2 days after treatment with Adriamycin (ADR) at the indicated concentrations was analyzed in the presence or absence of Fgf2 (20 ng/mL). Data are shown as the ratio of the measurement in the presence of each concentration of Adriamycin to that in the absence of Adriamycin and represented as means ± SDs from 3 independent experiments. *, P < 0.05; NS, not significant (the Student t test). D, a macroscopic finding of osteosarcoma tumors developed in mice with or without PD and/or Adriamycin treatment. E, tumor weight of AX-derived tumors developed in mice treated with or without PD and/or Adriamycin. F, immunohistochemistry of AX-derived osteosarcoma tumors in mice treated with each drug. Boxed regions are shown at a higher magnification in the bottom. G, the proposed model for osteosarcoma progression. Microenvironmental stromal cells release factors such as Fgf2, which maintain the immature state of osteosarcoma cells. Osteosarcoma cells inside the tumor exhibit osteogenesis accompanied by the withdrawal of the effects of environmental factors. In addition, these factors can enhance proliferation and motility of osteosarcoma cells. Stroma remodeling by the proliferation of fibroblasts and invasion of macrophages supports the dissemination of osteosarcoma cells. H, immunohistochemistry of an AX-derived osteosarcoma tumor for CD31. Both, a peripheral and an intratumor central region are shown. I, immunofluorescence analysis of Ki-67 expression in the peripheral or central part of an AX-derived osteosarcoma tumor. Tumor cells were identified by GFP staining. TOTO3 was used for nuclear staining.

Close modal

Finally, the influence of Fgf2 on the chemotherapeutic treatment of osteosarcoma cells was examined because previous reports suggested that Fgfr signaling affects the sensitivity of anticancer agents (20, 26, 27). Adriamycin is a well-known agent for osteosarcoma treatment and is commonly combined with several other chemotherapeutic agents (1). Cell proliferation assays under treatment with Adriamycin revealed that supplementation with Fgf2 slightly but significantly attenuated the cytotoxic effect of Adriamycin up to 80 ng/mL compared with that in control cells. (Fig. 7C). To further evaluate whether Fgfr signaling pathway could be a potential target for osteosarcoma, a combined treatment of the Fgfr inhibitor (PD173074) together with Adriamycin was conducted in vivo. Treatment of PD173074 or Adriamycin alone did not exhibit any antitumor effect (Fig. 7D and E). However, primary tumors in mice treated with both PD173074 and Adriamycin were significantly smaller than those in control mice or mice treated with each single agent. All tumors equally exhibited mineralized mature bone formation (Supplementary Fig. S5). Tumors in mice treated with a combination of PD173074 and Adriamycin appeared to have more osteoid formation even at the peripheral part of tumors compared with other groups (Fig. 7F). Osteogenic differentiation process might be enhanced by the combination therapy. In contrast, treatment with PD173074 or Adriamycin alone did not show any apparent microscopic difference. Therefore, although the combined treatment could not cease the primary tumor growth or metastatic process (data not shown), the inhibition of Fgfr signaling might be effective in terms of the sensitization of osteosarcoma cells to conventional chemotherapies in vivo. Collectively, these results indicated that Fgf2 moderately contributes to the enhancement of cellular motility and the reduction of sensitivity to anticancer drugs in addition to maintaining immaturity in osteosarcoma cells.

The histopathology of osteosarcoma includes a complex mixture of immature tumor cells and mature bone formation (3). It remains to be elucidated how these complicated structures are generated during osteosarcoma progression. The present study focused on the role of soluble factors such as Fgf2 and Lif released from nontumor cells in the osteosarcoma microenvironment.

Although the roles of Fgf2 or Lif on osteogenic differentiation are complicated and have not been fully clarified yet, both factors seem to possess similar potential for osteogenesis. The targeted disruption of each signaling pathway in mice resulted in a significant reduction in bone mass (28–30). In contrast, transgenic mice of Fgf2 exhibited a reduction in osteoblast differentiation and abnormal bone formation (31, 32). Fgf2 has also been found to promote proliferation and differentiation in osteoblasts and to increase bone formation (33–35). However, sustained exposure to a high concentration of Fgf2 inhibited osteogenesis by suppressing the production of bone materials (36–38). Similarly, Lif stimulated bone formation during the early proliferative stages of osteogenic process but inhibited it at later stages in osteoblast culture systems in vitro (39). Thus, both factors seem to show opposite impacts on osteoblast proliferation or differentiation at distinct stages of maturation, and the nonphysiologic exposure to both factors might result in the inhibition of osteogenesis (37, 39). AX cells are believed to have originated from transformed osteochondral progenitor cells (12). We suggest that Fgf2 and Lif mainly exert effects on proliferation and osteogenic suppression in AX cells in which the apoptotic program has been fully disrupted.

The physiologic roles of Erk activation in osteoblast differentiation also remain controversial and contradictory results have been observed, depending on the differential cell context or cell culture systems (40–42). A recent report showed that the cellular kinase activity of Erk1/2 was significantly downregulated during the differentiation of MC-3T3-E1 osteoblasts cultured for 7 days in the presence of ascorbic acid and β-glycerophosphate, which we used in our osteogenic culture system (43). Importantly, although both Fgf2 and Lif highly activated Erk1/2 in AX cells, the level of activation was more potent upon treatment with Fgf2 than with Lif (Fig. 3A). The suppression of osteogenic differentiation by Lif in AX cells was rescued by the Mek inhibitor PD98059 at 50 nmol/L, which was not able to restore the osteogenic ability inhibited by Fgf2 (Fig. 6D). Importantly, the restoration of osteogenesis suppressed by Lif treatment was also observed with another Mek inhibitor: U0126 at a lower concentration than that required for restoration in Fgf2 treatment (Fig. 6D). Moreover, Erk1/2 was more weakly phosphorylated by Igf1 than Fgf2 or Lif, and the high concentration of Igf1 also blocked osteogenic activity (Fig. 2C). Also importantly, osteogenic differentiation of AX cells occurs in the areas where Erk1/2 activation was absent in vivo (Fig. 6F; Supplementary Fig. S4B). Collectively, these findings suggested that the suppression of osteogenesis in AX cells might be dependent on the strength of activation of Erk1/2.

Fgf2 mediates various cellular events and promotes tumor progression (20, 44, 45). It is also suggested to involve in differentiation and/or cell death in some tumors such as Ewing tumor (46, 47). Our results indicated that Fgf2 enhanced cellular proliferation and motility in AX cells (Figs. 2B and 7A and B). Importantly, Fgfr1 was highly expressed not only in tumor cells but also in stromal fibroblasts (Fig. 4B and C). In addition, nontumor fibroblastic cells clustered in the Fgf2 enriched areas generated by Fgf2 polyhedra (Supplementary Fig. S4A). It is possible that Fgf2 released from stromal fibroblasts enhances proliferation and motility in the fibroblasts themselves as well as tumor cells. Although further evaluation will be needed, such a positive feedback loop in tumor stromal cells mediated by Fgf2 might contribute to remodeling tumor surroundings into suitable environments for tumor progression (Fig. 7G). Angiogenesis is one of the important roles of Fgf2 in tumor progression (20, 48, 49). In our established osteosarcoma tumors, tumor vasculatures were equally abundant both in peripheral and inside regions (Fig. 7H). In fact, AX cells produce Vegfa (data not shown) and the concentration measured using tumor homogenate was high [4132.4 ± 1438.1 pg/mL (n = 3)]. Therefore, the role of Fgf2 in angiogenesis in our established model might be limited.

We wonder whether the enhancement of osteogenesis by withdrawal of each environmental factor alone could contribute to osteosarcoma cell death, resulting in an effective therapeutic approach. Terminal differentiation accompanied by apoptosis is known to control osteoblast life span and bone formation (50). However, osteosarcoma cells appear to lapse into a differentiation block and AX cells may possibly return to an immature state from an osteogenic phase in the presence of Fgf2 (Fig. 4D). Consistent with this notion, a significant portion of osteosarcoma cells exhibited Ki-67 positivity, which is a marker for proliferative activity, in the region of bone formation in addition to the peripheral region of tumors (Fig. 7I). In view of therapeutic approaches to osteosarcoma, modulation of sensitivity to anticancer agents by Fgf2 might be more important (Fig. 7C, D, and E). Fgf2 might contribute to the formation of chemoresistant areas inside the tumors, resulting in residual disease in vivo.

Finally, some tumor malignant phenotypes such as cellular proliferation and motility were more enhanced by the mixture of several environmental factors than Fgf2 alone (Figs. 2B and 7B). Therefore, the combined blockade of signaling pathways of several growth factors, including Fgf2, may be useful in controlling the aggressiveness of osteosarcoma.

No potential conflicts of interest were disclosed.

The authors thank I. Ishimatsu, T. Suzuki, and S. Hyashi for technical assistance and K. Arai for secretarial assistance.

This work was partly supported by grants from the Ministry of Education, Science, Sports, and Culture of Japan (to T. Shimizu and H. Saya) and a grant from the National Institute of Biomedical Innovation, Japan (to H. Saya).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Bielack
SS
,
Kempf-Bielack
B
,
Delling
G
,
Exner
GU
,
Flege
S
,
Helmke
K
, et al
Prognostic factors in high-grade osteosarcoma of the extremities or trunk: an analysis of 1,702 patients treated on neoadjuvant cooperative osteosarcoma study group protocols
.
J Clin Oncol
2002
;
20
:
776
90
.
2.
Clark
JCM
,
Dass
CR
. 
A review of clinical and molecular prognostic factors in osteosarcoma
.
J Cancer Res Clin Oncol
2008
;
134
:
281
97
.
3.
Fletcher
CDM
,
Unni
KK
,
Mertens
F
editors
. 
Osteogenic tumours: WHO classification tumours of soft tissue and bone
.
Lyon, France:
IARC Press;
2002
.
4.
Gazdar
AF
,
Kadoyama
C
,
Venzon
D
,
Park
JG
,
Tsai
CM
,
Linnoila
RI
, et al
Association between histological type and neuroendocrine differentiation on drug sensitivity of lung cancer cell lines
.
J Natl Cancer Inst Monogr
1992
;
13
:
191
6
.
5.
Silverberg
SG
. 
Histologic grading of ovarian carcinoma: a review and proposal
.
Int J Gynecol Pathol
2000
;
19
:
7
15
.
6.
Sethi
T
,
Rintoul
RC
,
Moore
SM
,
MacKinnon
AC
,
Salter
D
,
Choo
C
, et al
Extracellular matrix proteins protect small cell lung cancer cells against apoptosis: a mechanism for small cell lung cancer growth and drug resistance in vivo
.
Nat Med
1999
;
5
:
662
8
.
7.
Uhm
JH
,
Dooley
NP
,
Kyritsis
AP
,
Rao
JS
,
Gladson
CL
. 
Vitronectin, a glioma-derived extracellular matrix protein, protects tumor cells from apoptotic death
.
Clin Cancer Res
1999
;
5
:
1587
94
.
8.
McAllister
SS
,
Weinberg
RA
. 
Tumor-host interactions: a far-reaching relationship
.
J Clin Oncol
2010
;
28
:
4022
8
.
9.
Bissell
MJ
,
Radisky
D
. 
Putting tumors in context
.
Nat Rev Cancer
2001
;
1
:
46
54
.
10.
DeWever
O
,
Demetter
P
,
Mareel
M
,
Bracke
M
. 
Stromal myofibroblasts are drivers of invasive cancer growth
.
Int J Cancer
2008
;
123
:
2229
38
.
11.
Seruga
B
,
Zhang
H
,
Bernatein
LJ
,
Tannock
IF
. 
Cytokines and their relationship to the symptoms and outcome of cancer
.
Nat Rev Cancer
2008
;
8
:
887
99
.
12.
Shimizu
T
,
Ishikawa
T
,
Sugihara
E
,
Kuninaka
S
,
Miyamoto
T
,
Mabuchi
Y
, et al
c-MYC overexpression with loss of Ink4a/Arf transforms bone marrow stromal cells into osteosarcoma accompanied by loss of adipogenesis
.
Oncogene
2010
;
29
:
5687
99
.
13.
Ijiri
H
,
Coulibaly
F
,
Nishimura
G
,
Nakai
D
,
Chiu
E
,
Takenaka
C
, et al
Structure-based targeting of bioactive proteins into cypovirus polyhedra and application to immobilized cytokines for mammalian cell culture
.
Biomaterials
2009
;
30
:
4297
308
.
14.
Mori
H
,
Shukunami
C
,
Furuyama
A
,
Notsu
H
,
Nishizaki
Y
,
Hiraki
Y
. 
Immobilization of bioactive fibroblast growth factor-2 into cubic proteinous microcrystals (bombyx mori cypovirus polyhedra) that are insoluble in a physiological cellular environment
.
J Biol Chem
2007
;
282
:
17289
96
.
15.
Nishishita
N
,
Ijiri
H
,
Takenaka
C
,
Kobayashi
K
,
Goto
K
,
Kotani
E
, et al
The use of leukemia inhibitory factor immobilized on virus-derived polyhedra to support the proliferation of mouse embryonic and induced pluripotent stem cells
.
Biomaterials
2011
;
32
:
3555
63
.
16.
Kubota
Y
,
Takubo
K
,
Shimizu
T
,
Ohno
H
,
Kishi
K
,
Shibuya
M
, et al
M-CSF inhibition selectively targets pathological angiogenesis and lymphangiogenesis
.
J Exp Med
2009
;
206
:
1089
102
.
17.
Noonan
DM
,
Barbaro
ADL
,
Vannini
N
,
Mortara
L
,
Albini
A
. 
Inflammation, inflammatory cells and angiogenesis: decisions and indecisions
.
Cancer Metastasis Rev
2008
;
27
:
31
40
.
18.
Holmes
C
,
Stanford
WL
. 
Stem cell antigen-1: expression, function, and enigma
.
Stem Cells
2007
;
25
:
1339
47
.
19.
Toma
JG
,
Akhavan
M
,
Fernandes
KJL
,
Barnabe-Heider
F
,
Sadikot
A
,
Kaplan
DR
, et al
Isolation of multipotent adult stem cells from the dermis of mammalian skin
.
Nat Cell Biol
2001
;
3
:
778
84
.
20.
Turner
N
,
Grose
R
. 
Fibroblast growth factor signaling: from development to cancer
.
Nat Rev Cancer
2010
;
10
:
116
29
.
21.
Heinrich
PC
,
Behrmann
I
,
Haan
S
,
Hermanns
HM
,
Muller-Newen
G
,
Schanper
F
. 
Principles of interleukin (IL)-6-type cytokine signaling and its regulation
.
Biochem J
2003
;
374
:
1
20
.
22.
Rikhof
B
,
de Jong
S
,
Suurmeijer
AJH
,
Meijer
C
,
van der Graaf
WTA
. 
The insulin-like growth factor system and sarcomas
.
J Pathol
2008
;
217
:
469
82
.
23.
Mohammadi
M
,
Olsen
SK
,
Ibrahimi
OA
. 
Structural basis for fibroblast growth factor receptor activation
.
Cytokine & Growth Factor Rev
2005
;
16
:
107
37
.
24.
Kamura
S
,
Matsumoto
Y
,
Fukushi
J
,
Fujiwara
T
,
Iida
K
,
Okada
Y
, et al
Basic fibroblast growth factor in the bone microenvironment enhances cell motility and invasion of Ewing's sarcoma family of tumours by activating the FGFR1-PI3K-Rac1 pathway
.
Br J Cancer
2010
;
103
:
370
81
.
25.
Lai
JP
,
Chien
J
,
Strome
SE
,
Staub
J
,
Montoya
DP
,
Greene
EL
, et al
HSulf-1 modulates HGF-mediated tumor cell invasion and signaling in head and neck squamous carcinoma
.
Oncogene
2003
;
23
:
1439
47
.
26.
Karajannis
MA
,
Vincent
L
,
DiRenzo
R
,
Shmelkov
SV
,
Zhang
F
,
Feldman
EJ
, et al
Activation of FGFR1β signaling pathway promotes survival, migration and resistance to chemotherapy in acute myeloid leukemia cells
.
Leukemia
2006
;
20
:
979
86
.
27.
Kwabi-Addo
B
,
Ozen
M
,
Ittmann
M
. 
The roles of fibroblast growth factors and their receptors in prostate cancer
.
Endocr Relat Cancer
2004
;
11
:
709
24
.
28.
Montero
A
,
Okada
Y
,
Tomita
M
,
Ito
M
,
Tsurukami
H
,
Nakamura
T
, et al
Disruption of the fibroblast growth factor-2 gene results in decrease bone mass and bone formation
.
J Clin Invest
2000
;
105
:
1085
93
.
29.
Marie
PJ
. 
Fibroblast growth factor signaling controlling osteoblast differentiation
.
Gene
2003
;
316
:
23
32
.
30.
Ware
CB
,
Horowitz
MC
,
Renshaw
BR
,
Hunt
JS
,
Liggitt
D
,
Koblar
SA
, et al
Targeted disruption of the low-affinity leukemia inhibitory factor receptor gene causes placental, skeletal, neural and metabolic defects and results in perinatal death
.
Development
1995
;
121
:
1283
99
.
31.
Soube
T
,
Naganawa
T
,
Xiao
L
,
Okada
Y
,
Tanaka
Y
,
Ito
M
, et al
Over-expression of fibroblast growth factor-2 causes defective bone mineralization and osteopenia in transgenic mice
.
J Cell Biochem
2005
;
95
:
83
94
.
32.
Coffin
JD
,
Florkiewicz
RZ
,
Neumann
J
,
Mort-Hopkins
T
,
Dorn
GW
,
Lightfoot
P
, et al
Abnormal bone growth and selective translational regulation in basic fibroblast growth factor (FGF-2) transgenic mice
.
Mol Biol Cell
1995
;
6
:
1861
73
.
33.
Pitaru
S
,
Kotev-Emeth
S
,
Noff
D
,
Kaffuler
S
,
Savion
N
. 
Effect of basic fibroblast growth factor on the growth and differentiation of adult stromal bone marrow cells: Enhanced development of mineralized bone-like tissue in culture
.
J Bone Miner Res
1993
;
8
:
19
29
.
34.
Nagai
H
,
Tsukuda
R
,
Mayahara
H
. 
Effects of basic fibroblast growth factor (bFGF) on bone formation in growing rats
.
Bone
1995
;
16
:
367
73
.
35.
Nakamura
T
,
Harada
K
,
Tamura
M
,
Shibanushi
T
,
Nigi
H
,
Tagawa
M
, et al
Stimulation of endosteal bone formation by systemic injections of recombinant basic fibroblast growth factor in rats
.
Endocrinology
1995
;
136
:
1276
84
.
36.
Hurley
MM
,
Abreu
C
,
Harrison
JR
,
Lichtler
AC
,
Raisz
LG
,
Kream
BE
. 
Basic fibroblast growth factor inhibits Type I Collagen gene expression in osteoblastic MC-3T3-E1 cells
.
J Biol Chem
1993
;
268
:
5588
93
.
37.
Debiais
F
,
Hott
M
,
Graulet
AM
,
Marie
PJ
. 
The effects of fibroblast growth factor-2 on human neonatal calvaria osteoblastic cells are differentiation stage specific
.
J Bone Mier Res
1998
;
13
:
645
55
.
38.
Canalis
E
,
Centrella
M
,
McCarthy
T
. 
Effects of basic fibroblast growth factor on bone formation in vitro
.
J Clin Invest
1988
;
81
:
1572
7
.
39.
Malaval
L
,
Aubin
JE
. 
Biphasic effects of leukemia inhibitory factor on osteoblastic differentiation
.
J Cell Biochem Suppl
2001
;
Suppl
36
:
63
70
.
40.
Ge
C
,
Xiao
G
,
Jiang
D
,
Franceschi
RT
. 
Critical role of the extracellular signal-regulated kinase-MAPK pathway in osteoblast differentiation and skeletal development
.
J Cell Biol
2007
;
176
:
709
18
.
41.
Schindeler
A
,
Little
DG
. 
Ras-MAPK signaling in osteogenic differentiation: friend or foe?
J Bone Miner Res
2006
;
21
:
1331
8
.
42.
Kawasaki
T
,
Niki
Y
,
Miyamoto
T
,
Horiuchi
K
,
Matsumoto
M
,
Aizawa
M
, et al
The effect of timing in the administration of hepatocyte growth factor to modulate BMP-2-induced osteoblast differentiation
.
Biomaterials
2010
;
31
:
1191
8
.
43.
Neto
AHC
,
Queiroz
KC
,
Milani
R
,
Paredes-Gamero
EJ
,
Justo
GZ
,
Peppelenbosch
MP
, et al
Profiling the changes in signaling pathways in ascorbic acid/β-glycerophosphate-induced osteoblastic differentiation
.
J Cell Biochem
2011
;
112
:
71
7
.
44.
Bikfalvi
A
,
Klein
S
,
Pintucci
G
,
Rifkin
DB
. 
Biological roles of fibroblast growth factor-2
.
Endocr Rev
1997
;
18
:
26
45
.
45.
Yu
PJ
,
Ferrari
G
,
Galloway
AC
,
Mignatti
P
,
Pintucci
G
. 
Basic fibroblast growth factor (FGF-2): The high molecular weight forms come of age
.
J Cell Biochem
2007
;
100
:
1100
8
.
46.
Sturla
LM
,
Westwood
G
,
Selby
PJ
,
Lewis
IJ
,
Burchill
SA
. 
Induction of cell death by basic fibroblast growth factor in Ewing's sarcoma
.
Cancer Res
2000
;
60
:
6160
70
.
47.
Kim
MS
,
Kim
CJ
,
Jung
HS
,
Seo
MR
,
Juhnn
YS
,
Shin
HY
, et al
Fibroblast growth factor 2 induces differentiation and apoptosis of Askin tumor cells
.
J Pathol
2004
;
202
:
103
12
.
48.
Nissen
LJ
,
Cao
R
,
Hedlund
EM
,
Wang
Z
,
Zhao
X
,
Wetterskog
D
, et al
Angiogenic factors FGF2 and PDGF-BB synergistically promote murine tumor neovascularization and metastasis
.
J Clin Invest
2007
;
117
:
2766
77
.
49.
Haugsten
EM
,
Wiedlocha
A
,
Olsnes
S
,
Wescha
J
. 
Roles of fibroblast growth factor receptors in carcinogenesis
.
Mol Cancer Res
2010
;
8
:
1439
52
.
50.
Manolagas
SC
. 
Birth and death of bone cells: basic regulatory mechanisms and implications for the pathogenesis and treatment of osteoporosis
.
Endocr Rev
2000
;
21
:
115
37
.