βig-h3 Promotes Human Osteosarcoma Cells Metastasis by Interacting with Integrin α2β1 and Activating PI3K Signaling Pathway

Osteosarcoma, the most common primary bone tumor in children and young adolescents, is characterized by local invasion and distant metastasis. But the detailed mechanisms of osteosarcoma metastasis are not well known. In the present study, we found that βig-h3 promotes metastatic potential of human osteosarcoma cells in vitro and in vivo. Furthermore, βig-h3 co-localized with integrin α2β1 in osteosarcoma cells. But βig-h3 did not change integrin α2β1 expression in Saos-2 cells. Interaction of βig-h3 with integrin α2β1 mediates metastasis of human osteosarcoma cells. The second FAS1 domain of βig-h3 but not the first FAS1 domain, the third FAS1 domain or the fourth FAS1 domain mediates human osteosarcoma cells metastasis, which is the α2β1 integrin-interacting domain. We further demonstrated that PI3K/AKT signaling pathway is involved in βig-h3-induced human osteosarcoma cells metastasis process. Together, these results reveal βig-h3 enhances the metastasis potentials of human osteosarcoma cells via integrin α2β1-mediated PI3K/AKT signal pathways. The discovery of βig-h3-mediated pathway helps us to understand the mechanism of human osteosarcoma metastasis and provides evidence for the possibility that βig-h3 can be a potential therapeutic target for osteosarcoma treatment.


Introduction
Osteosarcoma is a high-grade malignant bone neoplasm that occurs primarily in children and young adolescents. It occurs with an incidence of approximately three cases per million people per year [1]. The principles of treatment of osteosarcoma have undergone dramatic improves in the past 20 years. Multi-agent chemotherapy increased the 5-year overall survival of patients with localized disease to between 60% and 78% [2]. The survival of patients with metastatic osteosarcoma, however, remains poor with survival rates ranging from 11% to 20% [3,4]. This outcome suggested that 80% of the patients had metastasis at the time of presentation. Hence, a novel strategy that would efficiently inhibit osteosarcoma metastasis is highly desirable.
Tumor metastasis consists of a trail of complex procedures, all of which must be successfully completed to result in clinically detectable metastatic tumors at distal tissues [5,6]. To complete the process, primary cancer cells have to attach to extracellular matrix (ECM) components, invade through the basement membrane, intravasate into the circulation, and extravasate to distal tissues [7,8]. The entire process regulated by interactions between cancer cells and ECM. As a major component of the tumor microenvironment, ECM proteins potentially affect the metastasis process [9]. Thus, molecular alterations of the ECM proteins in the tumor microenvironment have a considerable impact on the metastatic process during tumorigenesis.
Transforming growth factor (TGF)-b-inducible gene-h3 (big-h3), which also called TGFBI, RGD-CAP, and MP78/70, is widely expressed in various types of tumor cells [10][11][12]. The big-h3 protein was initially identified by differential screening of a cDNA library produced from A549 human lung adenocarcinoma cells treated with TGF-b [13]. The protein consists of 683 amino acids, four fasciclin-1 (FAS1) homologous domains and an RGD motif at the C-terminus [14]. The FAS1 domains are homologous to fasciclin-1 in Drosophila and well conserved in several proteins from different species. FAS1 domain motif containing proteins, including big-h3, participate in cellular function via interactions with various integrins, including integrin a3b1, integrin avb3, and integrin avb5 [15][16][17]. As an ECM protein, big-h3 is involved in cell proliferation, migration, apoptosis and differentiation, and might function as either a promoter or an inhibitor of carcinogenesis, depending on cells and tumor types [18][19][20][21]. The gain or loss of expression of big-h3 might be involved in tumor formation and acquisition of a metastatic phenotype in human cancer. Although, previous studies have reported that big-h3 is required for apoptosis in human osteosarcoma cells [22], it is not clear yet whether big-h3 is involved in osteosarcoma metastasis.
This study sought to examine whether big-h3 expression could influence osteosarcoma cells metastasis and to determine the molecular mechanism by which this occurred, in an effort to elucidate the role of big-h3 in the regulation of osteosarcoma metastasis. In the present study, we showed that big-h3 promotes adhesion, invasion and migration of human osteosarcoma cells. big-h3 mediates human osteosarcoma cells metastasis through interacting with integrin a2b1, and then activates downstream PI3K/AKT signaling pathway. Furthermore, we identified that only the second FAS1domain of big-h3 was involved in osteosarcoma cells metastasis.

Downregulation of big-h3 decreases adhesion, invasion and migration of human osteosarcoma cells in vitro
As an ECM protein, big-h3 is involved in cell proliferation, migration, invasion, apoptosis and tumorigenesis [18][19][20][21]. To test the role of big-h3 in human osteosarcoma cells, small interfering RNAs against big-h3 (big-h3 siRNA) were transfected into the human osteosarcoma cell lines, Saos-2 cells and MG63 cells, for 48 hours to knockdown big-h3 mRNA and protein expression. Silencer negative control siRNAs (control siRNA) were also used as a negative control. As compared with control siRNA treated cells, the big-h3 siRNA could effectively decrease the mRNA and protein expression of big-h3 in Saos-2 cells and MG63 cells (P,0.05, Figure 1A and 1B). Cell function assays demonstrated that the amounts of cell adhesion were significantly decreased after big-h3 siRNA treatment in Saos-2 cells and MG63 cells (57.6%611.9% and 52.3%69.4%, respectively) (P,0.05, Figure 1C). In addition, the abilities of cells to invade through Transwell chambers were decreased after transfected with big-h3 siRNA in Saos-2 cells and MG63 cells. (37.1%618.5% and 31.2%612.8%, respectively) (P,0.05, Figure 1D). Similarly, treatment with big-h3 siRNA also decreased the amounts of cell migration in Saos-2 cells and MG63 cells (39.2%615.3% and 45.4%610.7%, respectively) (P,0.05, Figure 1E). This finding suggests that big-h3 may enhance adhesion, invasion and migration potential of human osteosarcoma cells.

big-h3 promotes metastasis of human osteosarcoma cells in vivo
To extend these studies in vivo, Saos-2 cells stably expressing GFP were established by lentiviral infection. Then the big-h3 vector and a control vector were stably transfected into Saos-2 cells which were stably expressing GFP. As compared with control vector-treated cells, the big-h3 vector could stably increase the mRNA and protein expression of big-h3 in Saos-2 cells (P,0.05, Figure 2A and 2B). Moreover, we observed stable expression of GFP and a strong correlation between GFP fluorescence signals and cell number in both control vector-treated cells and big-h3 vector-treated cells ( Figure 2C and Figure 2D). Furthermore, Saos-2 cells expressing the GFP were injected into immunodeficient mice through the tail vein. GFP fluorescence imaging was used to monitor the presence of Saos-2 cells. Due to size restrictions imposed by mouse capillaries, human tumor cells are rarely able to pass from the venous to the arterial system by way of the lung. Cells that failed to metastasize were not able to survive. Detectable GFP fluorescence signals indicated that cells had succeeded in metastasizing [23][24]. We found that the GFP signal in the group of big-h3 vector-transfected cells was significantly higher than the GFP signal in the group of control vector-transfected cells in the lung (P,0.05, Figure 2E). Therefore, the result indicated that big-h3 significantly promotes metastasis of human osteosarcoma cells in vivo.
big-h3 immunoprecipitates with a2b1 integrin in human osteosarcoma cells Integrins are cell surface adhesive receptors composed of a and b chain heterocomplexes, which play an essential role in osteosarcoma metastasis [25][26][27][28]. It has been reported that both a2 and b1 subunits are expressed in human osteosarcoma cells and they serve as bidirectional transducers of extracellular and intracellular signals in tumor metastasis processes [29][30][31]. Accordingly, we hypothesized that big-h3 might interact with integrin a2b1 to affect the metastasis ability of osteosarcoma cells. Immunofluorescent double staining was performed to examine cellular distribution of integrin a2b1 and big-h3 in Saos-2 cells. The results showed co-localizations of big-h3 with integrin a2 and integrin b1 subunits on the cell membrane ( Figure 3A). To further confirm this result, co-immunoprecipitation assay were performed to detect the interaction of big-h3 with integrin a2 and integrin b1 subunits. integrin a2 and integrin b1 subunits were found to coimmunoprecipitate with endogenous big-h3 in Saos-2 cells lysates, ( Figure 3B-3D), indicating that big-h3 and integrin a2b1 interact in their native conformations. To elucidate the effects of big-h3 on integrin a2b1 expression, we tested the protein expressions of integrin a2 and integrin b1 subunits in big-h3 siRNA transfected Saos-2 cells. We found that there were no significant expression modifications of integrin a2 and integrin b1 in big-h3 siRNA transfected cells compared with control cells ( Figure 3E). This finding suggested that big-h3 did not change integrin a2b1 expression in Saos-2 cells.
Interaction of big-h3 with integrin a2b1 mediates metastasis of human osteosarcoma cells To identify whether integrin a2b1 is involved in big-h3 mediated human osteosarcoma metastasis, the function blocking antibodies, mouse anti-human integrin a2 mAb (P1E6) and mouse anti-human integrin b1 mAb (6S6) were used. We found that big-h3 siRNA markedly reduced the amounts of cell adhesion in blank control group (p,0.05, Figure 4A). However, the addition of P1E6 and 6S6, alone or combination reduced the amounts of cell adhesion in control siRNA transfected cells to levels comparable with that in big-h3 siRNA transfected cells. big-h3 siRNA did not further reduce the amounts of cell adhesion after blocking integrin a2b1 (p.0.05, Figure 4A). The result indicated that integrin a2b1 is involved in big-h3 mediated cell adhesion. In addition, invasion and migration assay were performed. We found that big-h3 siRNA markedly reduced the amounts of cell invasion and migration in blank control groups (p,0.05, Figure 4B and 4C). However, the addition of P1E6 and 6S6, alone or combination reduced the amounts of cell invasion and migration in control siRNA transfected cells to levels comparable with that in big-h3 siRNA transfected cells. big-h3 siRNA did not further reduce the amounts of cell invasion and migration after blocking integrin a2b1 (p.0.05, Figure 4B and 4C). The above results indicate that integrin a2b1 is required for big-h3 mediates metastasis of human osteosarcoma cells.

The second FAS1 domain of big-h3 promotes human osteosarcoma cells metastasis
The big-h3 protein consists of four FAS1 homologous domains and an RGD motif at the C-terminus [14]. The FAS1 domains are homologous to fasciclin-1 in Drosophila and participate in cellular big-h3 Promotes Osteosarcoma Cells Metastasis function via interactions with various integrins [15][16][17]. To identify which FAS1domain mediated human osteosarcoma cells metastasis, we cloned the total gene of big-h3 (WT) and its four segments of highly conserved sequence, the first FAS1domain (D-I), the second FAS1domain (D-II), the third FAS1domain (D-III) and the fourth FAS1domain (D-IV) and then we transfected them into Saos-2 cells. We found that mRNA of the four FAS1domains of big-h3 were overexpressed in Saos-2 cells respectively (P,0.05, Figure 5B). Cell adhesion, invasion and migration assay demonstrated that overexpression of the second FAS1domain promoted the amounts of cell adhesion, invasion and migration to levels comparable with that in the total gene of big-h3 overexpressed cells (P,0.05, Figure 5C-5E). However, overexpression of the first FAS1 domain, the third FAS1 domain and the fourth FAS1domain were not able to promote cell adhesion, invasion and migration in Saos-2 cells (P.0.05, Figure 5C-5E). These results suggested that only the second FAS1domain, but not the first FAS1domain, the third FAS1domain or the fourth FAS1domain of big-h3 was involved in osteosarcoma cells metastasis big-h3 induces human osteosarcoma cells metastasis by activating PI3K signaling pathway Currently, the identities of integrin a2b1-associated signaling molecules that are responsible for mediating human osteosarcoma cells metastasis in response to big-h3 are unclear. To determine the signaling pathways that contribute to human osteosarcoma cells metastasis induced by big-h3, an examination was conducted into the effects of big-h3 on the phosphorylation status of AKT. Knockdown of big-h3 was found to decrease phosphorylation of AKT in Saos-2 cells ( Figure 6A). To further test whether PI3K is involved in big-h3 mediated Saos-2 cells metastasis, LY294002, a reversible inhibitor of PI3K was employed. big-h3 siRNA markedly reduced phosphorylation of AKT in control group. However, big-h3 siRNA did not further reduce phosphorylation of AKT following LY294002 treatment ( Figure 6B). The result suggested that activity of PI3K is required for big-h3 induced phosphorylation of AKT. Moreover, the addition of P1E6 and 6S6, alone or combination reduced the levels of phosphorylation of AKT in control siRNA transfected cells to levels comparable with that in big-h3 siRNA transfected cells ( Figure 6C). The result indicated that integrin a2b1 is involved in big-h3 induced phosphorylation of AKT. In addition, adhesion, invasion and big-h3 Promotes Osteosarcoma Cells Metastasis migration assay were performed. We found that big-h3 siRNA markedly reduced the amounts of cell adhesion, invasion and migration of Saos-2 cells in control groups (p,0.05, Figure 6D-6F). However, the addition of LY294002 reduced the amounts of cell adhesion, invasion and migration in control siRNA transfected cells to levels comparable with that in big-h3 siRNA transfected cells. big-h3 siRNA did not further reduce the amounts of cell adhesion, invasion and migration following LY294002 treatment (p.0.05, Figure 6D-6F). The above results indicated that PI3K/

Discussion
Osteosarcoma is the most common primary bone tumor in children and young adolescents. Although significant improvements in the treatment of patients with osteosarcoma recently, patients with metastatic osteosarcoma still have very poor prognosis [1,2]. Compounding the problem is that the molecular basis underlying metastatic osteosarcoma is poorly understood. Progression of osteosarcoma is thought to owing to cells attaching to ECM, invading through the basement membrane and migrating to distant tissues [3,4]. Thus, selectively blocking these metastatic abilities, through targeted therapy of key molecules should be an attractive strategy to inhibit osteosarcoma metastasis.
big-h3, an ECM protein mainly induced by TGF-b, was first identified in the human lung adenocarcinoma cell line A549 [13]. It is expressed in many tumor cells and tissues including the liver, lung, prostate and kidney [10][11][12]. Although its roles are largely unknown, it has been suggested that it is involved in the regulation of many aspects of tumor cell processes, including cell adhesion, spreading, invasion, proliferation and apoptosis [18][19][20][21]. In the present study, the effects of big-h3 on cell adhesion, invasion and migration were determined in osteosarcoma cells. The cell adhesion assay revealed that knockdown of big-h3 counteracted the adhesion of osteosarcoma cells to matrigel. Moreover, knockdown of big-h3 effectively inhibited the cell invasion and migration of osteosarcoma cells using transwell chamber and wound healing assay. We further discovered that big-h3 significantly promoted metastasis of human osteosarcoma cells in vivo using lung metastasis experiment. These results indicated that big-h3 acts as a major contributor to metastatic potential of osteosarcoma.
It has been suggested that biochemical signals of big-h3 can be transmitted across the plasma membrane through integrins to regulate various cellular functions, including adhesion, invasion, migration, survival, growth and differentiation [15][16][17]. Integrins, Through the interaction with the basement membrane, integrins can mediate cell adhesion and invasion [25][26][27][28]. The overexpression of integrin a2b1 has been reported to be associated with poor overall survival in patients with osteosarcoma [29][30][31]. In this study, big-h3 was found to colocalize and co-immunoprecipitate with integrin a2b1 in osteosarcoma cells. These results demonstrate that big-h3 and integrin a2b1 at least are in proximity, if not directly associated in osteosarcoma cells. Even though the interaction of big-h3 and integrins has been largely described in many other cellular lines [15][16][17], the exact mechanisms that link big-h3 to integrin a2b1 have not been reported yet. Our results further show that blocking the functions of integrin a2b1 with antibodies specific for integrin a2 and b1 reduces cell adhesion, invasion and migration in control siRNA transfected cells. However, no significant inhibitory effect is obtained in big-h3 siRNA transfected cells. These results indicate that the enhancing effect of big-h3 on cell metastasis potential is mediated through integrin a2b1. We also demonstrated that the expression levels of integrin a2 and integrin b1 are not influenced by the expression levels of big-h3 in osteosarcoma cells. That means the enhancing effect of big-h3 is not mediated through the overexpression of integrin a2b1. It is known that cells can change the conformation of their integrins in response to cellular stimulation in a process often termed ''integrin activation''. This conformational change mediates events such as cell migration, platelet aggregation, and assembly of ECM [32,33]. From the above results, we speculate that the positive effect of big-h3 is mediated through the upregulation of integrin a2b1 activity.
big-h3 contains four repetitive highly conserved FAS1 domains and a C-terminal arginyl-glycyl-aspartic acid (RGD) motif [14]. Previous studies reported that big-h3 mediates cell functions through these motifs that interact with different integrins on various cell types. The second and the fourth FAS1 domains of big-h3 mediate corneal epithelial cell adhesion by interacting with integrin a3b1. However, all four FAS1 domains of big-h3 mediate fibroblastic cell adhesion by interacting with the integrin avb5 [15][16][17]. To gain insight into the molecular mechanisms, we assessed the effect of four FAS1 domain proteins on big-h3induced metastasis of osteosarcoma cells. The results showed that only the second FAS1 domain displayed comparable cell metastasis potential to the full length big-h3 protein, indicating the existence of an integrin a2b1 -interacting motif in the second FAS1 domain of big-h3 in osteosarcoma cells.
Integrins influence cell behavior not only by providing a docking site for ECM proteins at the cell surface, but also by acting to active signaling pathway regarding cell growth, survival and migration [34,35]. To gain insight into integrin signaling mechanisms by which big-h3 promotes metastasis potential of osteosarcoma cells, we investigated the activation of integrin downstream molecules. Various studies have suggested that integrins and their ligands collaborate closely with growth factors in transducing signals through the phosphoinositide 3-kinase (PI3K)-AKT pathway [36][37][38]. In our study, knockdown of big-h3 caused reduction of AKT phosphorylation. At the same time, inhibition of PI3K resulted in inhibition of big-h3-mediated AKT phosphorylation. These results indicated that big-h3 may act via PI3K-dependent pathways to activate AKT. In addition, inhibition of PI3K resulted in inhibition of big-h3-mediated cell adhesion, invasion and migration in osteosarcoma cells. These data highlight the involvement of PI3K/AKT signaling pathway as an intracellular pathway of big-h3-mediated effects on metastasis of human osteosarcoma. Multiple intracellular pathways lead to PI3K/AKT signaling pathway activation. Ligand binding to integrins causes FAK phosphorylation, which in turn activates the PI3K/AKT signaling pathway and activation of PI3K/AKT signaling pathway by integrins has been described in other tumor cell types [39][40][41]. Our study presents that big-h3 interacted with integrin a2b1. And blocking of integrin a2b1 with specific antibodies dismissed phosphorylating effect of big-h3 on AKT. These results indicated that integrin a2b1 is involved in big-h3 induced PI3K/AKT pathway activation in osteosarcoma cells. PI3K/AKT signaling is a promising therapeutic target for metastatic osteosarcoma [42,43]. Therefore, this study revealed that downregulation of big-h3 might contribute to the anti-metastatic therapy of human osteosarcoma through inhibiting PI3K/AKT signaling.
It is well recognized that the prognosis of patients with osteosarcoma are very poor due to the incurable nature of distant metastasis. Thus, preventing human osteosarcoma metastasis is an all-important issue nowadays. The results of this study identified that big-h3 increases the adhesion, invasion and migration of human osteosarcoma cells via interacting with integrin a2b1 and activating PI3K/AKT signaling pathway. The discovery of big-h3-mediated pathway helps us to understand the mechanism of human osteosarcoma metastasis and provides evidence for the possibility that big-h3 can be a potential therapeutic target for osteosarcoma treatment.

Cell culture
The human osteosarcoma cell lines, Saos-2 and MG-63, were purchased from the American Type Culture Collection. Cells were Figure 6. big-h3 induces human osteosarcoma cells metastasis by activating PI3K signaling pathway. (A) Expression levels of phosphorylated AKT (p-AKT) and AKT were analyzed in control siRNA or big-h3 siRNA transfected Saos-2 cells. (B) Expression levels p-AKT and AKT were analyzed in control siRNA or big-h3 siRNA transfected Saos-2 cells which were incubated with PI3K inhibitor LY294002. (C) Expression levels of p-AKT and AKT were analyzed in control siRNA or big-h3 siRNA transfected Saos-2 cells which were incubated with P1E6 and 6S6, alone or combination. The amounts of cell adhesion (D), invasion (E) and migration (F) were tested in control siRNA or big-h3 siRNA transfected Saos-2 cells which were incubated with LY294002. Scale = 100 mm. Bars represent the mean of triplicate samples; error bars represent standard deviation. Data are representative of three independent experiments. *P,0.05 by Student's t test. doi:10.1371/journal.pone.0090220.g006 big-h3 Promotes Osteosarcoma Cells Metastasis PLOS ONE | www.plosone.org cultured in RPMI 1640 medium (Gibco, Grand Island, NE, USA), supplemented with 10% FBS, 1% penicillin/streptomycin and 2%L-glutamine at 37uC in a humidified atmosphere of 5% CO 2 .

Gene silencing
The sense sequence for big-h3 small interfering RNAs (siRNAs) was 59-CCUUUACGAGACCCUGGGATT-39 and the antisense sequence was 59-UCCCAGGGUCUCGUAAAGGTT-39 (Ambion, Austin, TX, USA). big-h3 siRNA was suspended in serum-free DMEM with LipofectAMINE 2000 reagent (Invitrogen, Carlsbad, CA, USA) for 20 min. The mixture was then aliquoted into 6-well plates containing pre-plated Saos-2 or MG63 cells to a final concentration of 40 nM of siRNA per well. Fortyeight hours later, cells were lysed for Western blot analysis. Silencer negative control siRNA (control siRNA) was used as a negative control under similar conditions (Ambion, Austin, TX, USA).

Western blot analysis
Protein expression was analyzed by Western blotting as previously described [39]. In brief, cells were lysed in 1% noctyl-p-D-glucopyrano-side (OG) buffer (20 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% OG, 1 mM EDTA, 10 g/ml leupeptin, 2 g/ml aprotinin,1 mM PMSF). BCA Protein Assay Kit (Pierce Biotechnology, Rockford, IL, USA) was employed to determine the total protein density and equal amounts of proteins were separated by 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and then electransferred to polyvinylidene fluoride (PVDF) microporous membrane (Millipore, Boston, MA, USA). After blocking with 5% non-fat milk, the membrane was incubated for 2 h at room temperature with the designated antibody. Immunodetection was performed by using the Western-Light chemiluminescent detection system (Applied Biosystems, Foster, CA, USA).

Real-time quantitative PCR
Total RNA was extracted from the cells with TRIzol reagents (Invitrogen, Carlsbad, CA, USA) and reverse transcribed into cDNA with a ReverTra Ace-a kit (Toyobo, Shanghai, China). All primers and probes were synthesized by Shanghai Sangon Co., and their sequences were as follows: big-h3: forward primer, 59-CATTGAGAACAGCTGCATCG- 39

Cell adhesion assay
The wells of a 96-well culture plate were coated with Matrigel (BD Biosciences, San Jose, CA, USA) at a concentration of 5 mg/ ml and incubated at 4uC overnight. The coated wells were blocked with PBS containing 2% BSA for 30 min and then washed with PBS. Cells suspended in serum-free medium containing 0.1% BSA were added to the wells (2610 4 /well) and incubated at 37uC, 5% CO 2 for 30-60 min with or without antibodies. After removing medium and non-attached cells, 0.2% crystal violet was added for 10 min. The plate was gently washed with tap water and dried in air for 24 h. Then, 0.1 ml 5% SDS/50% ethanol was added for 20 min and then the plates were read at 540 nm.

Cell invasion assay
The chemotactic cell invasion assay was performed using 24well transwell units with an 8-mm pore size polycarbonate filter (Millipore, MA, USA). Each lower compartment of the transwell contained 600 ml of 0.5% FBS as chemo-attractant or 0.5% BSA as negative control in RPMI 1640. The upper side of a polycarbonate filter was coated with Matrigel (5 mg/ml in cold medium) to form a continuous thin layer. Prior to addition of the cell suspension, the dried layer of matrigel matrix was rehydrated with medium without FBS for 2 h at room temperature. Cells (1610 5 ) pre-incubated with antibodies were suspended in 0.1 ml RPMI 1640 containing 0.1% BSA and added into the upper compartment of the transwell unit and incubated for 24 h or 18 h at 37uC in a humidified atmosphere containing 5% CO 2 . Cells remaining in the upper compartment were completely removed with gentle swabbing. The number of cells that had invaded through the filter into the lower compartment was determined using a colorimetric crystal violet assay.

Cell migration assay
Cells (2610 6 ) were plated in six-well plates and cultured to approximately 90% confluence. The cells were scraped with a pipette tip, washed several times in serum-free medium, and then examined under a phase contrast microscope (Olympus, Tokyo, Japan). The cells were re-fed with 10% FBS medium for 24 h or 18 h, and images were obtained.

Immunofluorescence assay
Cells were allowed to attach for 8 h to glass coverslips. They were then fixed in 3.7% formaldehyde in PBS, permeabilized with 0.5% Triton X-100 and blocked with 1% BSA (Fraction V) in PBS for 1 h. Coverslips were incubated with the indicated antibodies at a 1:500 dilution, or with rhodamine-phalloidin (Molecular Probes, Eugene, OR, USA) at a 1:40 dilution in PBS for 20 min. Antibody-treated cells were washed in PBS and incubated with FITC or Texas red-conjugated secondary antibodies (Pierce Biotechnology, Rockford, IL, USA) at a 1:500 dilution in PBS for 1 h. Cell nuclei were stained with DAPI (Vector Labs, Burlingame, CA). Finally, the cells were mounted using glycerol. Cells probed with rhodamine-phalloidin were washed and immediately mounted and observed by FV1000 laser scanning confocal microscopy (Olympus, Tokyo, Japan).

Co-Immunoprecipitation (Co-IP) assay
The interactions between big-h3 and integrin a2b1 in native cells were detected by ProFoundTM Mammalian Co-Immunoprecipi-tation Kit (Pierce Biotechnology, Rockford, IL, USA), according to the manufacturer's instructions. Briefly, Cells were lysed in 1% OG buffer. BCA Protein Assay Kit (Pierce Biotechnology, Rockford, IL, USA) was then used to determine the total protein density. Fraction of the lysates was saved as ''input''. Then, aliquots of lysates (1 mL) were immunoprecipitated by 25 mL of protein A-agarose that was pre-bound with 2 mg of anti-big-h3, anti-Integrin a2, anti-Integrinb1, or anti-IgG antibodies, followed by four washes with the co-immunoprecip-big-h3 Promotes Osteosarcoma Cells Metastasis itation buffer. Immune complexes were eluted from coupling gel with Elution buffer and resolved by 10% SDS-PAGE. Integrin a2, Integrinb1 or big-h3 in the immune complexes was detected by Western blot using, respectively, anti-integrin a2 anti-integrin b1 or anti-big-h3 antibodies (BD Biosciences, San Jose, CA, USA). anti-IgG were used as the negative control.

Cells stably expressing GFP
The cells were infected with a reconstructed pWPT-GFP-puro lentiviral vector, which contains a puromycin selection marker as previously reported [44]. For selection, 2 mg/ml puromycin was added to the medium, and after 14 days, the medium was changed to medium without puromycin. In this manner, the cells stably expressing GFP were established.

Lung metastasis experiment
NOD/SCID immunodeficient mice were used for experimental lung metastasis experiment. Saos-2 human osteosarcoma cells expressing GFP were trypsinized and washed with PBS. Subsequently, 16106 cells in 0.2 ml PBS were injected into the lateral tail vein. After 7 days, GFP fluorescence imaging was performed using the Xenogenin vivo Imaging System (IVIS 200, Xenogen, Alameda, CA, USA). GFP fluorescence images were analyzed with Igor image analysis software (Wavemetrics, Lake Oswego, OR, USA). The regions of interest were drawn over the signals, and the GFP fluorescence images were quantified in units of maximum photons per second per centimeter squared per steradian (p/s/ cm2/s).

Statistical analysis
Statistical analysis was performed using SPSS 13.0 statistical software. The results were expressed as mean values 6 SD. And the Student's t-test or one-way ANOVA were used to evaluate the statistical significance in the groups. The differences were considered significant when P,0.05.