Introduction

Osteosarcoma (OS) is the most common malignant primary bone tumor deriving from bone-forming mesenchymal cells, which mainly affects children and adolescents and occurs in long bone extremities, such as the distal femur, the proximal tibia, or the humerus. The current treatment for OS is surgical resection and adjuvant chemotherapy. Although advances in chemotherapy protocols have improved the clinical outcome of some OS patients, the overall prognosis and patient survival still remain dissatisfying, which is strongly associated with the tumor cell response to chemotherapy, and metastatic status1, 2. A 5-year survival rate of patients with non-metastatic OS is 70%; however the five-year survival rate of patients with metastatic OS was only 30%3. Understanding the metastatic process of OS is a pre-requisite for future effective therapy.

The fibulin-3 (FBLN-3) gene, also recognized as epidermal growth factor-containing fibulin-like extracellular matrix protein 1 (EFEMP1), is a member of the fibulin family of secreted extracellular glycoproteins that is widely expressed in blood vessel walls, and in basement membranes of epithelial and endothelial cells4. Fibulins consist of 7 extracellular matrix proteins and contribute to the stabilisation of supramolecular structures such as elastic fibres and basement membranes5, 6. Fibulin family members are involved in the processes of cell morphology maintenance, growth, adhesion, and movement, indeed, both tumour suppressive and oncogenic activities have been proposed in previous researches7. Fibulin-3 also has pro- and anti-tumorigenic bioactivities, with up- or down-regulation expression pattern depending on the cancer investigated. Upregulation of fibulin-3 was found in ovarian cancer8, 9, cervical cancer10, 11, pancreatic adenocarcinoma12, and malignant gliomas13, and high fibulin-3 expression was significantly correlated with advanced tumour stage and lymph node metastasis. Functionally, fibulin-3 had the ability to promote cancer cell growth and invasion. However, in hepatocellular carcinoma14, 15, gastric cancer16, lung cancer17, 18, endometrial carcinoma19, and nasopharyngeal carcinoma20, fibulin-3 was down-regulated in cancer tissues and suppressed cancer cell growth and invasion. In our current study, we investigated the function of fibulin-3 in human osteosarcoma invasion and metastasis, and the the relationship between fibulin-3 and EMT.

Materials and Methods

Cell culture

Osteosarcoma cell lines (HOS and U-2OS) and the normal osteoblastic cell line, hFOB, were obtained from the Shanghai Institute for Biological Sciences, Chinese Academy of Sciences. All cell lines were cultured in complete growth media DMEM/F12 (Gibco BRL, Rockville, MD) supplemented with 10% FBS (Gibco BRL, Rockville, MD) at 37 °C with 5% CO2.

Isolation of HOS cell subclones

HOS cells were diluted to about 10 cells/ml and seeded into a 96-well plate with 0.1 ml/well. Thus, as far as possible, there was only one cell in each well. After 1 week at 37 °C with 5% CO2, a single clone from one well was selected and cultured as a subclone. Using a cell electrophoretic instrument (DY-100, from College of Life Science, Shandong University, China), the electrophoretic migration rates of these subclones were measured individually. Moreover, the invasive and proliferative abilities of the highly invasive and low invasive subclones were analyzed by in vitro and in vivo functional assays. All data are expressed as mean ± standard error (SE).

Osteosarcoma tissue samples

With informed consent from patients, 290 specimens were obtained from the Department of Pathology, Shandong Qilu Hospital. None of these patients had undergone preoperative radiation or chemotherapy. All patients received regular follow-up. During the study period, contact with 15 patients was lost and 37 patients died. The follow-up period was from 2005 to 2014. This study was approved by the Institutional Medical Ethics Committee of Shandong University. All methods were performed in accordance with the relevant guidelines and regulations.

Immunohistochemistry (IHC) and immunocytochemistry (ICC)

For IHC, paraffin-embedded sections were dewaxed in xylene and rehydrated in ethanol. Heat-induced epitope retrieval was then performed using a pressure cooker in 0.01 M citrate buffer at pH 6.0. The sections were incubated for 3 min after the cooker reached full pressure. For ICC, cells at 75–80% confluency were seeded into a cell culture dish containing coverslips. After 24 h, the coverslips were harvested, washed thrice with PBS, and fixed in 95% ethanol for 30 min. According to the procedure of the Streptavidin-Peroxidase Detection Kit (ZSGB-BIO), the following steps were common for both IHC and ICC. All sections and coverslips were treated with 3% hydrogen peroxide (H2O2) and goat serum for 30 min sequentially, to block endogenous peroxidase and the non-specific binding sites, and then incubated with mouse anti-human fibulin-3 antibodies (sc-33722, Santa Cruz Biotechnology, Inc) at working dilutions of 1:100 overnight at 4 °C. The sections and coverslips were then incubated with the anti-mouse biotin-conjugated secondary antibody for 30 min at room temperature, stained for 1–5 min with the enzyme substrate 3′,3-diaminobenzidine tetrahydrochloride (DAB, Sigma-Aldrich, St. Louis, MO, USA), and counterstained for 5 min with hematoxylin. Paraffin-embedded sections of human ovarian cancer specimens (fibulin-3-positive) were used as positive controls21, and the negative control was obtained by replacing the primary antibody with PBS. Brown granules in the cytoplasm or stroma were considered as positive fibulin-3 expression. Each section or coverslip was repeated in triplicate.

Immunohistochemistry (IHC) and immunocytochemistry (ICC) analysis

To assess fibulin-3 expression in IHC and ICC experiments, the stained cell percentage and staining intensity were measured. The percentage of positively stained cells was scored from 0 to 4 (score 0, 0% cells stained; score 1, 1–25%; score 2, 26–50%; score 3, 51–75%; or score 4, 76–100%), whereas the staining intensity of fibulin-3 was scored as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong)22. Taken together, the intensity and percentage scores made up the final staining score (0–7), and the scores of 0, 1–3, 4–5, and 6–7 were converted into sum indices −, +, ++, and +++, respectively. For statistical analysis, low fibulin-3 expression was defined as − or +, whereas high fibulin-3 expression was indicated by ++ or +++. Each tissue section was independently analyzed by three pathologists.

Lentivirus transfection

The pLVX-fibulin-3 vector and fibulin-3 shRNA as well as a negative control were obtained from GeneChem Inc. (Shanghai, China). According to the manufacturer’s instructions, prior to viral infection, target cells were plated at 0.5 × 105 cells per well in a 24-well plate and incubated at 37 °C in a CO2 incubator for 24 h. The cells were then infected by adding the viral stock at a multiplicity of infection (MOI) of 100. The cells were incubated overnight at 37 °C with 5% CO2; the transfection mixture was then replaced with normal complete growth medium to avoid cell toxicity. At the end of 48 h of incubation, the transfected cells were assessed by fluorescence microscopy. The transfection efficiency was confirmed by western blotting, real-time quantitative RT-PCR, and immunocytochemistry (ICC). The siRNA sequence for fibulin-3 was 5′-TGTGAGACAGCAATGCAAA-3′.

Quantitative real-time-polymerase chain reaction (qRT-PCR)

TRIzol® Reagent (Ambion™) was used to isolate total RNA from cell and tissue samples. Reverse Transcription was carried out with TaqMan® Reverse Transcription Reagents (Applied Biosystems Inc.; Thermo Fisher Scientific, Inc.). The procedure was based on the protocol provided by Invitrogen. The real-time PCR mixture volume was 25 μl including 12.5 μl SYBR Green Mix (Power SYBR® Green PCR Master Mix, Applied Biosystems Inc.), 0.2 μl cDNA, 1 μl primer pair mix (5 pmol/μl each primer), and 11.3 μl DNAse/RNAse-free H2O. The experiment was then set up with the following PCR program on ABI Prism SDS 7000 (Applied Biosystems Inc.; Thermo Fisher Scientific, Inc.): 50 °C for 2 min, 1 cycle; 95 °C for 10 min, 1 cycle; 40 cycles of 95 °C for 15 s–> 60 °C for 30 s −>72 °C for 30 s; 72 °C 10 min, 1 cycle. Finally, the results were analyzed with SDS 7000 software. Specific primers were designed by LightCycler® Probe Design software (Roche Diagnostics, Basel, Switzerland) and were synthesized by Takara Biotechnology Co., Ltd. The primer sequences were as follows: fibulin-3: 5′-ACCCTTCCCACCGTATCCA-3′, 5′-TCTGCTCTACAGTTGTGCGTCC-3′; E-cadherin: 5′-GGATTGCAAATTCCTGCCATTC-3′, 5′-AACGTTGTCCCGGGTGTCA-3′; N-cadherin: 5′-GTAGCTAATCTAACTGTGACCGATAAGG-3′, 5′-TTGGTTTGACCACGGTGACTAA-3′; vimentin: 5′-GCAGGAGGCAGAAGAATGGTA-3′, 5′-GGGACTCATTGGTTCCTTTAAGG-3′; Snail: 5′-TCGGAAGCCTAACTACAGCGA-3′, 5′-AGATGAGCATTGGCAGCGAG-3′; Slug: 5′-TGTGACAAGGAATATGTGAGCC-3′, 5′-TGAGCCCTCAGATTTGACCTG-3′; Twist: 5′-AGCAAGATTCAGACCCTCAAGCT-3′, 5′-CCTGGTAGAGGAAGTCGATGTACCT-3′; β-actin: 5′-CCACGAAACTACCTTCAACTCCA-3′, 5′-GTGATCTCCTTCTGCATCCTGTC-3′.

Western blot

Cells were lysed on ice in RIPA (radioimmunoprecipitation assay) buffer with 1 mM PMSF (phenylmethylsulfonyl fluoride). From the cell lysate, 40 μg of total protein was loaded into the wells of the SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis) gel, along with a molecular weight marker. Electrophoresis was carried out for 1–2 h at 100 V, followed by transfer to PVDF (polyvinyl difluoride) membranes, which were then blocked with 5% BSA (bovine serum albumin). The membranes were then incubated overnight at 4 °C with primary antibodies (fibulin-3 sc-33722, E-cadherin sc-8426, N-cadherin sc-59987, Vimentin sc-6260, Santa Cruz; Snail ab167609, Slug ab27568, Twist ab50887, β-catenin ab32572, C-myc ab32072, Cyclin D1 ab134175, Abcam) at working dilutions of 1:1000. After washing the membranes thrice with TBST for 5 min each, the membranes were incubated with conjugated secondary antibody diluted to 1:1000, at room temperature for 1 h. Blots were developed using the enhanced chemiluminescence method (Pierce™ ECL Western Blotting Substrate; Thermo Fisher Scientific, Inc.).

Growth curves

Cells at the logarithmic phase were collected, seeded into the wells of a 24-well plate (1 × 104 cells/well) and cultured at 37 °C with 5% CO2. Three wells were harvested every day and the cells were counted and averaged. Growth curves were then plotted according to the average cell counts of 7 consecutive days.

Soft agar colony formation assay

DMEM with 20% FBS (1.5 ml) mixed with 1.5 ml of 1.2% agar was added to 3.5 cm dishes and solidified for the bottom layer. Next, 1.5 ml of 0.7% agar was mixed with 1.5 ml DMEM (20% FBS) and 200 µl of cell suspension (containing 600 cells), and was immediately added to the above culture dishes. All the dishes were incubated for 2 weeks at 37 °C with 5% CO2. The assay was repeated in triplicate. Under an inverted microscope (Nikon Eclipse), the dish was divided into quadrants and in each quadrant, colonies with diameters of more than 2 mm were counted and the average was calculated. All the data are expressed as mean ± SE.

Cell invasion assay and migration assay

The in vitro Matrigel invasion assay was performed as described previously23. The polyvinylpyrrolidone-free polycarbonate (PVPF) membrane of Boyden chambers (BD Biosciences, Bedford, MA) were coated with 50 µl of Matrigel 1:3 diluted with serum-free media. Cell suspensions in volumes of 200 µl (2 × 105 cells) were seeded into the upper chambers, and 600 µl of serum-free culture supernatant of NIH3T3 cells was added to the lower chamber as a chemotactic factor. The Boyden chambers were then incubated at 37 °C for 24 h. The non-invading cells on the upper surface of the membrane were removed, and the cells on the lower surface were fixed with 4% paraformaldehyde, stained with hematoxylin and eosin (H&E), and counted in five random high-power fields (HPF) under an inverted microscope. The cell migration assay was simultaneously performed with the above steps, without the Matrigel coating on the membrane and the incubation time was only 12 h. The cell invasion and migration assays were both repeated in triplicate. All data are expressed as mean ± SE.

Chromatin immunoprecipitation (ChIP)-qPCR

ChIPs were performed using the EZ-ChIP kit (Millipore), according to the manufacturer’s protocols. Cells were treated with formaldehyde to crosslink the proteins to the DNA, and then sonication was performed to shear the chromatin to a manageable size (200–1000 bp). The samples were pre-cleared and immunoprecipitated with 0.2 μg/mL of desired antibodies (anti-H3K4me3, Millipore) and Protein G-conjugated agarose beads. Mock immunoprecipitations with anti-IgG served as controls. Input (total chromatin extract), mock and ChIP samples, were recovered and then analysed by real-time qPCR. Primers used for PCR detection are listed as follows: E-cadherin (−80 to +88): 5′-CCCGCCCCAGCTGTGTCATTTT-3′, 5′-AATGGTGCCCATCCACGTGG-3′; N-cadherin (−5112 to −4961): 5′-CCAAAGTGCTGGTATTCCGCTGTAAG-3′, 5′-GTGTGCTCCCAGAGTCGGGTTTGC-3′; vimentin (−116 to +91): 5′-GGTGTGGTTTCATGGGGGGAGG-3′, 5′-CCCTAAGTTTTTAATAACTCGCTAAAG-3′.

β-Catenin/TCF Reporter Assay

The TCF reporter constructs TOPflash and FOPflash are widely used to evaluate the transcriptional activity of β-catenin/TCF. TOPflash contains 3 copies of wild-type TCT/LEF binding sites driven by the thymidine kinase (TK) promoter and the luciferase open reading frame. FOPflash has mutant TCF/LEF binding sites and is used as a negative control. Both TOPflash and FOPflash were purchased from Upstate Cell Signaling Solutions (Billerica, MA). pLVX-fibulin-3 infected cells and negative control HOS-29 were plated in 6-well plates, and transfected with LipofectAMINE™ 2000 Reagent (Invitrogen) and 1.8 μg of TOPflash or FOPflash plasmids. To normalize, cells were cotransfected with 0.2 μg of the Renilla luciferase internal control vector (Promega). Then, luciferase activity was determined using the Dual-luciferase reporter assay system (Promega) according to the manufacturer’s protocol. Each experiment was triplicated independently and repeated three times. After normalizing the transfection efficiency, the results were expressed as folds compared to control.

Tumor xenografts in nude mice

BALB/C-nu/nu nude mice were purchased from the National Resource Center for Rodent Laboratory Animal of China. Each group included 5 nude mice, each of which was inoculated subcutaneously with 5.0 × 106 cells. The mice were maintained in a sterile animal facility and monitored daily for tumor growth. Every week, the tumor volumes were measured using vernier calipers, and calculated according to the formula, V = length × width2 × 0.25. After 2 months, the mice were sacrificed and the tumors were dissected and examined histologically. All data are expressed as mean ± SE. This animal experiment was approved by the Institutional Animal Care and Use Committee of Shandong University and complied with all the regulatory guidelines.

Statistical analysis

IHC data were analyzed using a χ2 test. A two-tailed t-test was used to compare the means between two sets, and a one-way analysis of variance was used to compare the means among three groups. By the Kaplan-Meier method and the log-rank test, survival curve analysis was performed to study the relationship between fibulin-3 and the prognosis of patients with osteosarcoma. The data were analyzed with SPSS software version 13.0 (SPSS Inc., Chicago, IL, USA). P < 0.05 (two-sided) was considered statistically significant.

Results

Fibulin-3 expression in human osteosarcoma tissues

Low fibulin-3 protein expression was detected in most normal tissues, however, fibulin-3 immunoreactivity was very high in osteosarcoma tissues, mainly in the stroma and in the osteosarcoma cell cytoplasm (Fig. 1). Moreover, high fibulin-3 expression was positively associated with low differentiation and lymph node metastasis (Table 1). Similar results were also observed in the qRT-PCR experiment. High fibulin-3 mRNA expression was observed in osteosarcoma tissues and was correlated with low tumor differentiation and positive nodal metastasis (Table 2). Survival analysis showed that patients with low fibulin-3 expression had much better prognosis than those with high fibulin-3 expression (log rank, P < 0.05; Fig. 1E).

Figure 1
figure 1

Expressions of fibulin-3 in human osteosarcoma tissues. Fibulin-3 expressions in osteosarcoma tissue (A,B), fibrous dysplasia tissue (C) and normal tissue (D) were measured by IHC. (E) Patients with high fibulin-3 expression (blue line) had a much worse prognosis than those with low fibulin-3 expression (green line). (Magnification ×200).

Full size image
Table 1 Protein expression of fibulun-3 in human osteosarcoma tissues.
Full size table
Table 2 mRNA expression of fibulin-3 in human osteosarcoma tissues.
Full size table

Establishment of highly invasive and low invasive subclones

Using the single cell cloning technique, 29 subclones were obtained from HOS cells. The subclone HOS-1, which had the highest migration rate (21.54 ± 0.62 μm/s) showed higher proliferative and invasive abilities, compared to the subclone HOS-29, which showed the lowest migration rate (6.84 ± 0.17 μm/s). In vivo, the subcutaneous tumor formation rate for the highly invasive subclone group was 100%, and was accompanied by rapid tumor growth. However, the tumor formation rate of the low invasive subclone group was only about 50%, with very slow tumor growth. The tumor volume for HOS-1 was 526.10 ± 37.64 mm3, much larger than that formed by HOS-29 (57.80 ± 3.65 mm3, P < 0.01). These results are shown in Fig. 2.

Figure 2
figure 2

Establishment of high invasive subclone and low invasive subclone. (A) The high invasive subclone had the higher proliferative abilities than the low invasive subclone. (B) The average counts of migrating and invading cells of high invasive subclone were both much higher than those of low invasive subclone. (C) The images of cells crossing PVPF filters without Matrigel were examined by cell migration assay. (D) The images of cells crossing PVPF filters with Matrigel were examined by cell invasion assay. (E) The tumor volumes formed by high invasive subclone were much larger than those formed by low invasive subclone after subcutaneous inoculation. (F) Tumor growths of high invasive subclone and low invasive subclone were observed continuously for 2 months. (Magnification ×200). *P < 0.05.

Full size image

Different proliferation and invasion abilities of human osteosarcoma cell lines and the normal osteoblastic cell line

Compared to the normal osteoblastic cell line hFOB, the human osteosarcoma cell lines U-2OS and HOS showed stronger proliferative abilities (Fig. 3A). In the soft agar colony formation assay, the number of colonies formed by HOS and U-2OS was also significantly greater than that formed by hFOB (Fig. 3B,C). In the cell migration and Matrigel invasion assays, the average counts of migrating and invading HOS and U-2OS cells were both much higher than those of hFOB (Fig. 3D–F). Upon comparing the two osteosarcoma cell lines, we found that HOS had stronger proliferation and invasion abilities than those of U-2OS.

Figure 3
figure 3

Different proliferation, migration and invasion abilities of human osteosarcoma cell lines and normal osteoblastic cell line. (A) The human osteosarcoma cell lines U-2OS and HOS had stronger proliferative abilities than normal osteoblastic cell line hFOB. (B) The number of colonies formed by HOS and U-2OS was also significantly greater than that formed by hFOB. (C) The colony images of human osteosarcoma cell lines HOS and U-2OS and normal osteoblastic cell line hFOB were examined by soft agar colony formation assay. (D) The images of cells migrating PVPF filters were examined by cell migration assay using Boyden chambers. (E) The images of cells invading Matrigel-coated membranes were examined by cell invasion assay using Boyden chambers. (F) The average counts of migrating and invading cells of human osteosarcoma cell lines HOS and U-2OS were both much higher than those of normal osteoblastic cell line hFOB. (Magnification ×200). *P < 0.05.

Full size image

Fibulin-3 expression in human osteosarcoma cell lines and in differently invasive subclones

As shown in Fig. 4, fibulin-3 expression was very weaker in the normal osteoblastic cell line hFOB than in the human osteosarcoma cell lines HOS and U-2OS. The strongest fibulin-3 expression was detected in HOS, which showed the highest proliferation and invasion abilities. Similar results were also observed upon comparing subclones with differing invasive abilities. Compared with the low invasive subclone HOS-29, high fibulin-3 expression was detected in the highly invasive subclone HOS-1. These results indicate that high fibulin-3 expression might be positively associated with the proliferative and invasive abilities of osteosarcoma cells.

Figure 4
figure 4

Fibulin-3 expressions in human osteosarcoma cell lines and in differently invasive subclones. fibulin-3 expressions in human osteosarcoma cell lines HOS and U-2OS and normal osteoblastic cell line hFOB were measured by ICC staining. (A) Brown granules were detected in the cell cytoplasm. (B) Negative controls were obtained by replacing the primary antibody with PBS. (C) Real time q-RT-PCR and (D) western blot were performed to detect the mRNA and protein expressions of fibulin-3 in human osteosarcoma cell lines and normal osteoblastic cell line. Fibulin-3 expressions in high invasive subclone and low invasive subclone were measured by (E) ICC staining, (F) western blot (cropped blot) (see full blots in the supplementary information) and (G) q-RT-PCR. (Magnification ×200). *P < 0.05.

Full size image

Identification of downregulated and upregulated fibulin-3 expression in lentivirus transfection systems

To further investigate the potential role of fibulin-3 in osteosarcoma cell proliferation and invasion, we decreased the expression of fibulin-3 in the highly invasive subclone HOS-1, and increased fibulin-3 expression in the low invasive subclone HOS-29 by lentivirus transfection. After viral infection, real-time q-RT-PCR, western blotting, and ICC were used to confirm the altered expression of fibulin-3 at both mRNA and protein levels, indicating the high efficiency of the lentivirus transfections (Fig. 5).

Figure 5
figure 5

Identification of downregulated and upregulated fibulin-3 expression in lentivirus transfection systems. Fibulin-3 expressions in non-infected cells, control shRNA infected cells and fibulin-3 shRNA infected cells were measured by (A) ICC staining, (B) real time q-RT-PCR and (C) Western blot. Fibulin-3 expressions in non-infected cells, control shRNA infected cells and pLVX-fibulin-3 infected cells were measured by (D) ICC staining, (E) real time q-RT-PCR and (F) Western blot (cropped blot). (Magnification ×200). *P < 0.05.

Full size image

Effect of fibulin-3 knockdown and overexpression on osteosarcoma cell proliferation

Downregulated fibulin-3 markedly inhibited cell proliferation of the highly invasive subclone HOS-1, whereas upregulated fibulin-3 significantly promoted cell proliferation of the low invasive subclone HOS-29 (Fig. 6A). In the soft agar colony formation assay, the colony forming efficiencies of fibulin-3-silenced cells were decreased, and conversely, upregulation of fibulin-3 could increase the colony forming efficiency of the low invasive subclone (Fig. 6B,C). No significant differences were observed in the uninfected and negative control groups.

Figure 6
figure 6

Effect of fibulin-3 knockdown and overexpression on osteosarcoma cell proliferation and colony formation abilities. (A) Downregulated fibulin-3 markedly inhibited cell proliferation of the highly invasive subclone HOS-1, whereas upregulated fibulin-3 significantly promoted cell proliferation of the low invasive subclone HOS-29. (B) The colony forming efficiencies of fibulin-3-silenced cells were decreased, and conversely, upregulation of fibulin-3 could increase the colony forming efficiency of the low invasive subclone. (C) The colony images of non-infected cells, control shRNA infected cells and fibulin-3 shRNA infected cells were examined by soft agar colony formation assay. (D) The colony images of non-infected cells, control shRNA infected cells and pLVX-fibulin-3 infected cells were examined by soft agar colony formation assay. (Magnification × 200). *P < 0.05.

Full size image

Effect of fibulin-3 knockdown and overexpression on osteosarcoma cell migration and invasion

As shown in Fig. 7, fibulin-3 knockdown inhibited osteosarcoma cell invasion and migration. The average counts of migrating and invading fibulin-3 shRNA infected cells were much lower than those of the negative controls and uninfected groups (P < 0.05). Meanwhile, fibulin-3 overexpression promoted the invasion and migration of osteosarcoma cells. The average counts of migrating and invading pLVX-fibulin-3 infected cells were much higher than those of the negative controls and the uninfected groups (P < 0.05). There were no significant differences between the negative controls and uninfected groups.

Figure 7
figure 7

Effect of fibulin-3 knockdown and overexpression on osteosarcoma cell migration and invasion. (A) Cell migration and invasion images of non-infected cells, control shRNA infected cells and fibulin-3 shRNA infected cells were measured by cell migration and invasion assays using Boyden chambers. (B) Cell migration and invasion images of non-infected cells, control shRNA infected cells and pLVX-fibulin-3 infected cells were measured by cell migration and invasion assays using Boyden chambers. (C) The average counts of migrating and invading fibulin-3 shRNA infected cells were much lower than those of the negative controls and non-infected groups; meanwhile, the average counts of migrating and invading pLVX-fibulin-3 infected cells were much higher than those of the negative controls and the non-infected groups. (Magnification × 200). *P < 0.05.

Full size image

Effects of fibulin-3 knockdown and overexpression on tumor growth in a xenograft model

The fibulin-3 shRNA infected cells, pLVX-fibulin-3 infected cells, negative control HOS-1, and negative control HOS-29 were each inoculated subcutaneously in 5 nude mice respectively. The tumor formation rate of the negative control HOS-1 was 100%, whereas the tumor formation rate in the fibulin-3 shRNA infected group was only 40%. Moreover, the average volumes of the tumors formed in the fibulin-3 shRNA infected group were much lower than those formed by the negative control HOS-1. Fibulin-3 knockdown inhibited tumor formation in nude mice. Simultaneously, fibulin-3 overexpression promoted tumor growth in nude mice. The tumor formation rate of pLVX-fibulin-3 infected cells was 100%, whereas the tumor formation rate of the negative control HOS-29 was only 60%. Moreover, the average volumes of the tumors formed by pLVX-fibulin-3 infected cells were much higher than those formed by the negative control HOS-29 (Fig. 8).

Figure 8
figure 8

Effects of fibulin-3 knockdown and overexpression on tumor growth in a xenograft model. (A) Tumor growths in lentivirus transfection systems were observed continuously for 8 weeks. (B) The average volumes of the tumors formed in the fibulin-3 shRNA infected group were much lower than those formed by the negative control HOS-1. (C) The average volumes of the tumors formed by pLVX-fibulin-3 infected cells were much higher than those formed by the negative control HOS-29. *P < 0.05.

Full size image

Effects of fibulin-3 on key EMT genes

EMT is significantly associated with the invasion and metastasis of various cancers. Therefore, we wondered if fibulin-3 knockdown and upregulation affected several key EMT genes including E-cadherin, N-cadherin, vimentin, Snail, Slug, and Twist. The results showed that fibulin-3 knockdown could remarkably suppress the expression of EMT markers such as N-cadherin, vimentin, Snail, Slug, and Twist; in contrast, fibulin-3 upregulation could increase the expression of EMT markers N-cadherin, vimentin, Snail, Slug, and Twist (Fig. 9A,B). In conclusion, fibulin-3 could promote the invasion and metastasis of osteosarcoma cells by inducing EMT.

Figure 9
figure 9

Effects of fibulin-3 knockdown and overexpression on EMT genes correlated to tumor progression. After lentivirus transfections, EMT markers, including E-cadherin, N-cadherin, vimentin, Snail Slug and Twist were measured by (A) Western blot (cropped blot) and (B) real-time q-RT-PCR in the lentivirus transfection systems. (C) H3K4me3 and H3K27me3 expressions were assessed by Western blot. Histone H3 was used as a loading control. H3K4me3 is associated with transcriptional activation, and H3K27me3 was responsible for turning off transcription. Ectopic expression of fibulin-3 only increased H3K4me3 modification but had no effect on H3K27me3 modification. (D) Quantitative chromatin immunoprecipitation was performed to assess H3K4me3 occupancy in pLVX-EFEMP1 cells. IgG was used as the negative control. “% input” indicated the ratio of DNA fragment of each promoter region bound by H3K4me3 to the total amount of input DNA fragment without H3K4me3 antibody pull-down. *P < 0.05.

Full size image

Fibulin-3 induced EMT via regulating the H3K4me3 modification

Histone modifications act as a significant role in regulating gene expression and chromatin accessibility, previous studies have demonstrated that histone H3 modifications occured during cancer cell EMT24,25,26. So we explored whether fibulin-3 regulated histone H3 methylation in osteosarcoma cells. H3K4me3 is associated with transcriptional activation, and H3K27me3 was responsible for turning off transcription. In our experiments, H3K4me3 and H3K27me3 were measured after overexpression of fibulin-3. At the same time, we also detected whether fibulin-3 overexpression was associated with the H3K4me3 modification at the promoters of E-cadherin, N-cadherin, and vimentin in osteosarcoma cells. The results revealed that ectopic expression of fibulin-3 only increased H3K4me3 modification but had no effect on H3K27me3 modification (Fig. 9C). We also found fibulin-3 expression was associated with increased H3K4me3 levels at region −5112 to −4961 bp of the N-cadherin promoter and −116 to +91 bp of the vimentin promoter, in contrast, the H3K4me3 level of −80 to +88 bp of the E-cadherin promoter was decreased in pLVX-fibulin-3 infected cells, by ChIP-qPCR assay (Fig. 9D). These data suggested that fibulin-3 could induce EMT through regulating H3K4me3 modification and enriching H3K4me3 to the promoters of N-cadherin and vimentin genes.

Effects of fibulin-3 on the Wnt/β-catenin pathway

Wnt signaling pathway can induce the process of EMT by inhibiting the phosphorylation of glycogen synthase kinase -3β (GSK3β) and inhibiting the degradation of β-catenin in cytoplasm. So we wondered that fibulin-3 maybe could regulate the process of EMT through the Wnt signaling pathway. Using the luciferase reporters TOPflash and FOPflash, which have been widely used to determine the transcriptional activity of β-catenin/TCF, we found that after 48 h transfection, the TOPflash reporter activity was upregulated by about 6 fold compared with FOPflash reporter in pLVX-fibulin-3 infected cells, and no obvious change was observed in the cells transfected with the empty control vector. Fibulin-3 was involved in the activation of β-catenin/TCF transcription. Then several key gene expressions of the Wnt/β-catenin pathway, including GSK3β-pS9, β-catenin, C-myc, and cyclin D1, were assessed by western blot in the lentivirus transfection system. The results showed that fibulin-3 knockdown decreased the phosphorylation level of GSK3β and then activated its kinase activity, which in turn promoted β-catenin degradation and prevented its accumulation and nuclear translocation, so as to decreased the expressions of TCF/LEF related oncogenes c-myc and cyclin-D1. Meanwhile, fibulin-3 upregulation increased the phosphorylation level of GSK3β and then inhibited its kinase activity, which subsequently prevented β-catenin degradation and promoted its accumulation and nuclear translocation, as a consequence, the expressions of TCF/LEF related oncogenes c-myc and cyclin-D1 were increased. Using Wnt signaling pathway inhibitor XAV-939 (5, 10, and 20 μmol/L) and activator LiCl (5, 10, and 20 μmol/L), we treated the pLVX-EFEMP1-infected cells and EFEMP1 shRNA-infected cells, for 48 h. We found that Wnt signaling pathway inhibitor XAV-939 could significantly inhibit the Wnt/β-catenin pathway, and EMT, both of which were activated by fibulin-3 upregulation; at the same time, Wnt signaling pathway activator LiCl could promote the Wnt/β-catenin pathway and EMT, which were deactivated by fibulin-3 knockdown. We observed concentration-dependent increases with the increase in the dose. These results indicated that fibulin-3 could promote osteosarcoma cell migration and invasion by inducing EMT and activating the Wnt/β-catenin pathway (Fig. 10).

Figure 10
figure 10

Effects of fibulin-3 on the Wnt/β-catenin signaling pathway by Western blot (cropped blot). (A) Fibulin-3 knockdown reduced the expression of GSK3β-pS9, β-catenin, C-myc, and cyclin D1; in contrast, fibulin-3 upregulation increased the expression of GSK3β-pS9, β-catenin, C-myc, and cyclin D1. (B) Wnt signaling pathway activator LiCl could promote the Wnt/β-catenin pathway and EMT, both of which were deactivated by fibulin-3 knockdown; meanwhile, Wnt signaling pathway inhibitor XAV-939 could significantly inhibit the Wnt/β-catenin pathway and EMT, both of which were activated by fibulin-4 upregulation. (C) Cells were transfected with TOPFLASH or FOPFLASH construct. After 48 h, luciferase activity was determined by the dual-luciferase system. The pRL-TK Renilla luciferase reporter construct was co-transfected in each sample to normalize transfection efficiency. The luciferase activity of the reporter is expressed relative to the activity in control cells, which is defined as 1.0. *P < 0.05.

Full size image

Discussion

In our study, we observed that high fibulin-3 expression was associated with poor prognosis of human osteosarcoma, and that fibulin-3 was over-expressed in the highly invasive osteosarcoma cell line and subclone. Fibulin-3 could promote osteosarcoma cell invasion and metastasis by inducing EMT and activating the Wnt/β-catenin pathway.

From the protein and mRNA levels, we found that fibulin-3 expression in osteosarcoma tissues and cell lines was much higher than that in normal tissues and cell lines, especially in the highly metastatic osteosarcoma cell line and subclone. Moreover, high fibulin-3 expression was detected in low differentiation and positive lymph node status. Similar results have been reported by Wang Z et al.27, EFEMP1 was specifically upregulated in osteosarcoma and promoted the migration and invasion of osteosarcoma cells via MMP-2. In ovarian cancer8, by activating AKT signaling pathway, EFEMP1 contributed to ovarian cancer invasion and metastasis as a positive regulator. In cervical cancer10, EFEMP1 promoted the angiogenesis and accelerated the growth of cervical carcinoma through a VEGF up-regulation pathway. In malignant gliomas13, fibulin-3 promoted tumor cell survival and invasion in a Notch-dependent manner. In our study, fibulin-3 exerted tumor oncogenic effects and promoted osteosarcoma cell invasion and metastasis by inducing EMT and activating the Wnt/β-catenin pathway.

The epithelial-mesenchymal transition (EMT) is a highly conserved cellular program by which epithelial cells lose their cell polarity and cell-cell adhesion, and gain migratory and invasive properties to become mesenchymal cells. EMT is essential for numerous developmental processes including mesoderm formation and neural tube formation. EMT has also been shown to occur in wound healing, in organ fibrosis and in the initiation of metastasis for cancer progression28. Our results revealed that fibulin-3 knockdown upregulated the expression of epithelial marker E-cadherin, decreased the expressions of mesenchymal marker N-cadherin and vimentin, and downregulated the expressions of transcription factor Snail (Snail-1), Slug (Snail-2), and Twist; meanwhile fibulin-3 upregulation decreased E-cadherin expression, and increased the expressions of N-cadherin, vimentin, Snail, Slug and Twist. Further ChIP-qPCR assay suggested that fibulin-3 could induce EMT through regulating H3K4me3 modification and enriching H3K4me3 to the promoters of N-cadherin and vimentin genes. These data simultaneously indicate that fibulin-3 could induce EMT by regulating H3K4me3 modification, and thereby promote the invasion and metastasis of osteosarcoma cells. In endometrial carcinoma29, EFEMP1 could remarkably decrease the expression of EMT markers such as Vimentin and Snail, and acted as a tumor suppressor. In lung cancer stem cells18, forced expression of fibulin-3 suppressed invasion and migration of lung adenocarcinoma cells and decreased the expression of EMT activators, including N-cadherin and Snail. In ovarian cancer8, the EMT PCR array results indicated that EFEMP1 knockdown suppressed EMT and inhibited ovarian cancer invasion and metastasis. In summary, the role of fibulin-3 in EMT may exhibit tissue specificity, wherein different tumor microenvironments determine the gene functions30.

The wingless (Wnt) signaling is suggested as a fundamental hierarchical pathway and plays important roles in morphogenesis, cell development, and carcinogenesis. The activated Wnt proteins are essentially involved in carcinogenesis and cancer31,32,33,34. In our study, using the luciferase reporters TOPflash and FOPflash, we found that fibulin-3 was involved in the activation of β-catenin/TCF transcription. At the same time, fibulin-3 upregulation increased several key gene expressions of the Wnt/β-catenin pathway; on the other hand, fibulin-3 knockdown decreased these gene expressions. Furthermore, these processes could be reversed by the Wnt signal pathway activators or inhibitors. So, we believed that fibulin-3 could induce EMT and activating the Wnt/β-catenin signaling pathway to promote the invasion and metastasis of osteosarcoma cells. In endometrial carcinoma29, EFEMP1 could remarkably suppress the Wnt/β-catenin target genes like Cyclin-D1 and c-Myc, which could be restored when EFEMP1 was silenced. In addition, the inhibitory effect of EFEMP1 could be blocked by XAV93920 (the inhibitor of the Wnt/β-catenin pathway) and enhanced by LiCl (the activator of the Wnt/β-catenin pathway). In lung cancer17, the expression levels of fibulin-3 and MMP-7 were inversely correlated. Fibulin-3 inhibited extracellular signal-regulated kinase (ERK) to activate glycogen synthase kinase 3β and suppress Wnt/β-catenin pathway, which induces MMP-7 expression in lung cancer cells. Collectively, fibulin-3 was involved in the Wnt/β-catenin pathway, the specific function was determined by the pro- or anti-tumorigenic bioactivities of fibulin-3 in different tumors.

In conclusion, high fibulin-3 expression was associated with poor prognosis in human osteosarcoma and the malignant phenotype of osteosarcoma cells. Fibulin-3 has the ability to promote proliferation, invasion, and metastasis of osteosarcoma cells by inducing EMT and activating the Wnt/β-catenin signaling pathway. We believe that further research on fibulin-3 can contribute to the early diagnosis and treatment of human osteosarcoma.