Osteosarcoma (OSA) is the most common primary malignant musculoskeletal tumor and occurs mainly in young people (children and adolescents < 20 years of age). By inducing neovascularization, these tumors frequently lead to the development of bone and soft tissue metastases.
A large panel of mutations associated with significant variation in the expression of genes controlling cell proliferation or differentiation has been reported in OSA. Notably, suppressor genes such as p53, MDM2, and Rb were found to have major roles in tumorigenesis in OSA.1, 2 Some bone-related transcription factors also have been linked to OSA. For example, transgenic mice overexpressing c-fos in bone develop OSA.3 Osterix also may play a role in OSA tumor growth and metastasis.4, 5 Runx2 was found to be absent or nonfunctional in OSA cell lines.6 Genomic alterations on the Twist gene have been detected in high-grade pediatric OSA.7 The broad spectrum of mutations and the lack of correlation between cell differentiation and tumor grade led to focus on the defective control of cell growth processes and to establishment of strategies consisting of antiproliferative compounds. However, despite aggressive chemotherapeutic treatment strategies, the development of metastatic lesions and resistance to chemotherapy remain major mechanisms responsible for the failure of OSA treatment. Indeed, the estimated survival rate for patients with localized OSA is about 75% compared with 30% for patients with metastatic disease. Up to the present, only few genes such as HES18 and Ezrin9 were found to play a role in the progression of tumors and metastasis in OSA.
We have shown previously that statins induce antitumoral effects in OSA cells.10–12 Here we used microarray analysis to identify statin-modulated genes that potentially may be targeted to reduce tumor invasion in OSA. One identified sequence that we found to be downregulated by statins encodes cystein rich protein 61 (Cyr61; gene ID 94295). This immediate-early gene encodes a member of the CCN (cyr61/CTGF/Nov) family of secreted proteins comprising connective tissue growth factor (CTGF), nephroblastoma overexpressed (NOV), and three Wnt-induced secreted proteins (WISP1, -2, and -3). CCN proteins are associated with the extracellular matrix and cell surface.13 Here we demonstrate that Cyr61 expression correlates with in vivo aggressiveness and metastatic potential of human OSA. We also show that Cyr61 protein level increases with tumor grade in OSA patients and that Cyr61 silencing reduces in vitro and in vivo OSA cell invasion. Altogether, our results reveal that Cyr61 downregulation represents a novel potential molecular target for therapeutic intervention in metastatic OSA.
Materials and Methods
SaOS2, U2OS, MG63, K7M2, and HEK293T cells were from ATCC (Rockville, MD, USA). OHS414 and CAL7215 cells were kindly provided by Dr Fournier and Dr Price (University of California Davis, La Jolla, CA, USA) and Dr Rochet (INSERM, UMR6235, Nice, France), respectively. Cells were cultured in DMEM (Invitrogen, Paisley, Scotland) containing 10% fetal calf serum (FCS), 10,000 U/mL of penicillin, and 10,000 µg/mL of streptomycin at 37°C in humidified atmosphere containing 5% CO2 in air. Culture medium was changed three times a week.
RNA extraction and RT-qPCR
RNA was isolated using TriZol Reagent (Invitrogen, Cergy Pontoise, France). RNA (3 µg) was denaturated for 10 minutes at 70°C and then reversed transcribed at 37°C for 90 minutes using 300 U of Moloney murine leukemia virus reverse transcriptase (MMLV), 15 µg oligodT primers, and 1 mM deoxynucleoside triphosphate (dNTP) in 30 µL of total volume. Quantitative polymerase chain reaction (qPCR) was performed using a SYBR Green Master Kit (ABGen, Courtabœuf, France) supplemented with 0.5 µM of primers: human Cyr61 primers: 5'-AAA CCC GGA TTT GTG AGG T-3' and 5'-GCT GCA TTT CTT GCC CTT T-3'; mouse cyr61 primers: 5'-GGA TCT GTG AAG TGC GTC CT-3' and 5'-CTG CAT TTC TTG CCC TTT TT-3'; 18S primers: 5'-CGG CTA CCA CAT CCA AGG AA-3' and 5'-GCT GGA ATT ACC GCG GCT-3'. Thermal conditions were 15 minutes at 95°C and then 40 cycles of 95°C for 20 seconds, 58°C for 15 seconds, and 72°C for 15 seconds. All signals with threshold cycle (Ct) > 39 were set as undetermined. Relative amounts of RNA were calculated by the 2–ΔΔCt method.
Proliferation and cell death assays
Cell replication was determined using a BrdU ELISA assay (Roche, Meylan, France). A double staining with ethidium bromide and acridine orange was performed as described previously.16 Effector caspase activity was determined as described previously.11, 12
Migration and invasion assays
Wounding assay was performed according to the manufacturer's instructions (Ibidi, Martinsried, Germany). Recovery of the denuded area was computerized using an inverted microscope (Leica, Cambridge, UK). Cell migration and invasion were determined in the modified Boyden chamber assay, as described previously.11
Cell lysates were analyzed as described previously.11, 12 Anti-GAPDH and anti-Cyr61 were from Abcam (Cambridge, UK). Anti-MMP2 was from Cell Signaling Technology (Ozyme; St Quentin Fallavier, France).
Mouse metastatic model
All procedures were performed with the approval of the French Ethical Animal Committee. BALB/c mice (4-weeks old; Charles River laboratories, Arbresle, France) were injected intraperitoneally with 50 mg/kg of ketamine (Merial, Duluth, GA, USA) and 6 mg/kg of xylazine (Henry Schein, Melville, NY, USA) before injection of 106 OSA cells/20 µL of PBS in thigh muscles. At 1 month, mice were euthanized. Thigh muscle with femur were fixed in formalin and decalcified in paraformaldehyde (PFA)/EDTA before inclusion in paraffin. Lungs were fixed in formalin before inclusion in paraffin. Tissue sections (7 µm) were deparaffined in xylene and rehydrated through a graded ethanol series before hematoxylin and eosin (H&E) staining or immunostaining (anti-Cyr61 at 1:100).
Human tissue microarrays
Tissue microarrays (TMAs) composed of paraffin included 231 tissue cores that were deparaffinized and rehydrated before overnight incubation with anti-Cyr61 antibody (1:100) at 4°C. Signal, revealed using a Vectastain Elite ABC System (Vector Laboratories, Ltd., Peterborough, UK), was estimated by two observers without prior information about TMA spots.
Comparisons between data were performed using two-factor analysis of variance (ANOVA). A minimal level of p < .05 was considered significant. Statistical TMA results were presented as Bonferroni-corrected p values of Fisher's exact test.
Statin reduces Cyr61 expression in osteosarcoma cells
Using microarray,17 we compared the genomic expression pattern of SaOS2 cells cultured with or without atorvastatin (10 µM) for 6, 15, or 24 hours. We focused our attention on consistently modulated transcripts (Supplemental Table S1). We selected Cyr61 because this candidate gene showed the quickest (as soon as 6 hours after induction) and the strongest sustained inhibition of expression under statin treatment. We found that mRNA encoding Cyr61 is detectable in a panel of human OSA cells at levels ranging from 1 to 100 (Fig. 1A). To validate the microarray data, we analyzed the effect of statin supplementation on the expression of Cyr61 in a panel of human (SaOS2, U2OS, MG63, CAL72, and OHS4) and murine (K7M2) OSA cell lines. As evaluated by qPCR analysis, atorvastatin (10 µM) markedly decreased Cyr61 mRNA expression (up to 85%, p < .05 versus untreated) in all tested OSA cell lines (Fig. 1B). Immunoblot analysis confirmed that atorvastatin treatment decreased Cyr61 protein levels (Fig. 1C). Similar effects were observed using the natural compound simvastatin, the highly potent cerivastatin, and the hydrophilic statin pravastatin (data not shown). These results reveal that Cyr61 is downregulated by statins in all murine and human OSA cells tested. In following studies, we chose SaOS2 and U2OS cell lines because they expressed the highest and lowest levels of Cyr61, respectively.
Repression of Cyr61 expression by atorvastatin requires geranylgeranylation
We investigated the involvement of some key metabolites of the cholesterol synthesis pathway12 in the inhibitory effect of statin on Cyr61 level of expression. Pretreatment with geranylgeranylpyrophosphate (GGPP; 10 µM) but not with farnesylpyrophosphate (FPP; 10 µM) prevented the reduction in Cyr61 mRNA and protein levels induced by atorvastatin in SaOS2, U2OS, and K7M2 cells (Supplemental Fig. S1).
Since we showed previously that statin-induced GGPP depletion led to a relocalization and inactivation of the GTPase RhoA,12 we next evaluated the effect of statin-induced defective isoprenylation of RhoA on Cyr61 expression. Overexpression of a constitutively active mutant form of RhoA (RhoA-G14V) did not modify basal expression of Cyr61 and did not counterbalance the inhibitory effect of atorvastatin (Supplemental Fig. S1). These results indicate that the posttranslational modification geranylgeranylation of some other proteins than RhoA-GTPase plays a critical role in Cyr61 expression in osteosarcoma cells.
Establishment of modified osteosarcoma cell lines
To establish the role of Cyr61, we generated new cell lines by transduction with lentiviral vectors encoding either the full-length sequence (LV-Cyr61) or a specific shRNA (shCyr61). As expected, Cyr61 mRNA and protein levels were significantly modulated in transduced cells compared with parental cells (Supplemental Fig. S2). We also confirmed that the Cyr61 secreted by overexpressing cells was functional because it induced tubule structure formation (Supplemental Fig. S2).
Osteosarcoma cell proliferation is independent of Cyr61 expression
We first compared cell viability of transduced cells with parental cells. No modulation in cell viability could be detected by the MTT test (24 to 72 hours; data not shown) or BrdU incorporation assay (Fig. 2A). These data indicate that Cyr61 does not significantly affect osteosarcoma cell proliferation in human and murine osteosarcoma cell lines.
Osteosarcoma cell resistance to apoptosis depends on Cyr61 expression
We then investigated the influence of Cyr61 on OSA cell death. As shown in Fig. 2B, under FCS-supplemented conditions, apoptotic and necrotic indexes were increased in sh-Cyr61-transduced cells compared with parental cells. In contrast, both indexes were decreased in LV-Cyr61-transduced cells. Consistently, caspase activity and the Bax/Bcl2 ratio were increased in sh-Cyr61 transduced cells compared with parental cells. In contrast, overexpression of Cyr61 reduced caspase activity and Bax/Bcl2 ratio compared with parental cells (Fig. 2C, D). These data show that Cyr61 expression affects OSA cell death even under favorable culture conditions.
We demonstrated previously that doxorubicin, cisplatin, and methotrexate dose-dependently reduce OSA cell viability.11 In this study, we found that Cyr61 silencing increased apoptotic cell death induced by doxorubicine, cisplatine, or methotrexate, whereas LV-Cyr61 led to a slight protective effect (Fig. 3B–D). Similar effects were observed with U2OS and K7M2 cells. These data indicate that silencing Cyr61 further promotes OSA cell death induced by cytotoxic agents.
In vitro cell migration and invasion depend on Cyr61 expression
Since metastasis represents the main deleterious characteristic of OSA, we determined the invasiveness and migratory potential of transduced OSA cell lines. We found that Cyr61 silencing reduced cell wounding potency compared with parental cells, whereas overexpression favored cell wounding (Fig. 4A, B). A Boyden's chamber assay confirmed the strong relation between Cyr61 expression and migratory potential (Fig. 4C). Overexpression of Cyr61 did not counterbalance the inhibitory effect of statin on cell migration (Fig. 4D). These data show that Cyr61 expression controls human and murine OSA cell migratory potential.
Cell invasion was evaluated using Matrigel-coated inserts (Becton Dickinson, Le Pont-De-Claix, France). We found that Cyr61 silencing reduced the cell capability to invade extracellular matrix, whereas forced expression of Cyr61 increased invasiveness (Fig. 4E). These data indicate that Cyr61 controls in vitro cell migration and invasive abilities in human and murine OSA cell lines and demonstrate that reducing Cyr61 expression level attenuates OSA cell agressiveness.
We then investigated the activity of matrix metalloprotease 2 (MMP2).10, 11 We found that sh-Cyr61 decreased MMP2 protein level and activity compared with parental cells, whereas Cyr61 overexpression increased MMP2 expression and activity (Supplemental Fig. S3), implying a role for MMP2 in the control of OSA cell invasion by Cyr61.
In vivo Cyr61 expression increases with tumor grade in human osteosarcoma
To determine the potential role of Cyr61 in human OSA, we evaluated Cyr61 protein expression level in tissue microarrays. As shown in Fig. 5A, in all tumors tested, metastatic cancerous tumor cells expressed higher levels of Cyr61 than primary tumor cells. Furthermore, recurrent OSA tissues exhibited the highest levels of Cyr61. Remarkably, semiquantitative analysis indicated that Cyr61 protein expression increases with tumor grade in human OSA and correlates with OSA aggressiveness (Fig. 5B). These data strongly support a role of Cyr61 in cell invasiveness in human OSA.
Finally, we investigated the influence of Cyr61 on in vivo tumor behavior using a mouse metastatic model. Parental and transduced K7M2 cells were injected into BALB/c mice. In control mice (injected with PBS), Cyr61 signal was detected mainly in bone marrow cells. In mice injected with parental K7M2 cells, Cyr61 expression was detected in tumor tissue located in muscles. In mice injected with sh-Cyr61 and LV-Cyr61, downregulation and overexpression of Cyr61, respectively, were found in vivo during tumor development (Fig. 6A).
The incidence of pulmonary metastases then was investigated by H&E staining. Intramuscular injection of parental K7M2 cells led to the development of several pulmonary tumor foci (Fig. 6B). Overexpression of Cyr61 greatly increased the size of tumoral foci in lung tissue (Fig. 6B). In contrast, pulmonary tissue of mice injected with sh-cyr61 cells was almost exempt of metastatic tumor. Quantification of tumor area confirmed that cyr61 overexpression induced 4.5-fold larger metastases than parental cells (p < .05), whereas sh-cyr61 reduced the number and size of tumors (up to –80%, p < .05; Fig. 6C–E). These data demonstrate that Cyr61 expression in OSA cells controls in vivo invasiveness and that Cyr61 silencing strongly reduces lung metastases.
Prenylation is a posttranslational modification involving covalent addition of isoprenoids to conserved cysteine residues. It is required for the protumorigenic activity of some GTPases, including Ras and Rho-like proteins.18, 19 This has led to the development of FTase and GGTase inhibitors as anticancer drugs.20, 21 The synthetic compounds bisphosphonates, which are also able to block part of prenylation through inhibition of FPP synthase activity,22 have antitumor activity and could be beneficial for the treatment of bone metastases.23–26 Another class of drugs are the statins, which by inhibition of HMGCoA reductase activity27 act as hypocholesterolemic agents. This generates depletion in downstream isoprenoid residues such as GGPP or FPP. We demonstrated previously that statins induce apoptosis, reduce cell migration and invasion, and potentiate the action of chemotherapeutic agents in OSA cells.10–12 The in vivo application of statins as anticancer agents, however, is unclear. A large analysis of more than 60,000 patients reported that the incidence of cancers of any origin was significantly decreased in patients using statins.28 Other clinical studies have revealed that the use of statins may be associated with either increased or decreased cancer risk.29, 30 These conflicting data reinforce the need for a better understanding of the mechanisms of action of statins to refine the development of new anticancer targets.
In order to identify molecular targets of statins in OSA cells, we performed a comparative transcriptomic analysis. One of the downregulated genes was Cyr61, which encodes a secreted protein known to modulate tumor development and progression.31–33 We identified Cyr61 as a target of statins in human and murine OSA cells. A recent publication reported an inhibitory effect of simvastatin on Cyr61 expression.34 Although Cyr61 is expressed in primary osteoblasts and OSA cell lines35, 36 and was reported to promote mesenchymal stem cell proliferation and osteoblast differentiation,37 its role on OSA is unknown. The Cyr61 gene is located in 1p22.3 on chromosome 1, which is reported to be the most commonly involved chromosome in human cancers because the short arm of chromosome 1 is reported to be the target of numerous translocations.38 Osteosarcoma cells exhibit a high percentage of polysomy in chromosome 1.39 The expression of Cyr61 thus could be modified following chromosomal reorganization, although further studies are required to explore this hypothesis.
To determine the role of Cyr61 in OSA development, we increased or silenced the expression of Cyr61 in human and murine OSA cells. We did not detect significant change in cell proliferation in any human and murine OSA cell line. In contrast, Cyr61 silencing slightly increased OSA cell death and further enhanced the antineoplasic and proapoptotic effects of doxorubicin, cisplatin, and methotrexate, which are part of the anticancer strategy in OSA. We and others have already demonstrated a positive combinatory effect of statins with chemotherapeutic drugs in OSA or other cancer types.10, 40–43 This study now focuses on Cyr61 expression, independent of the presence of statins. Cyr61 has been reported as a candidate gene for drug resistance in melanoma,44 but its role as a potential target in OSA cells was unknown. Overexpression of Cyr61/CCN1 was found to amplify resistance to chemotherapeutic drugs in breast and ovarian cancer cells.45, 46 In contrast, silencing of Cyr61 in combination with anticancer agents has not been reported. Our data suggest that targeting Cyr61 could reinforce the antitumoral effect of these treatments and thereby reduce resistance to chemotherapy in these bone tumors.
The rapid and frequent development of metastatic lesions is a critical issue in OSA. Our in vitro experiments showed that Cyr61 silencing reduced both migration and invasive potencies of human and murine OSA cells, whereas overexpression led to an increase in cell migration and invasiveness. Interestingly, the effect of Cyr61 silencing on OSA migration and invasion was independent of the OSA cell genotype (alteration of Rb, p53, or ARF). These data indicate that inhibition of Cyr61 expression negatively controls in vitro human and murine OSA cell invasion and migration in a large context. Although silencing or neutralization of Cyr61 was reported to reduce cell motility and invasiveness in breast and prostate cancer,47, 48 little is known with regard to the CCN protein family in OSA. CCN3/Nov was found to be expressed at variable levels in OSA cell lines,49 which may be associated with poor prognosis and higher risk of developing metastases.50 Based on the analysis of tissue sections from patients, we found that CCN1/Cyr61 protein expression level is significantly higher in primary OSA tissue than in normal bone tissue. Furthermore, Cyr61 expression increased gradually with the incidence of metastasis, particularly in cases of recurrence, indicating that Cyr61 expression correlates with tumor invasiveness in human OSA. Future studies of the analysis of the evolution of primary and metastatic tumors are needed to conclude on the predictive value of Cyr61 expression level on tumor development in human OSA. A role for Cyr61 in OSA metastasis is strongly supported by our experiments using manipulated OSA cells in mice. Indeed, our in vivo experiments showed that upregulation of Cyr61 expression increased OSA tumor behavior dramatically by acting on pulmonary metastatic foci number and size. More important, we demonstrated that Cyr61 silencing markedly reduced OSA lung metastasis occurrence in mice. The finding that attenuation of Cyr61 reduces cancer cell progression in vivo therefore may have important implications for reducing metastatic tumor evolution in OSA.
In summary, our data demonstrate that Cyr61 expression controls cell invasion and migration in vitro and tumor invasiveness in vivo. This study supports an important role of Cyr61 in human OSA tumorigenesis and provides a molecular basis for a novel therapeutic strategy to reduce osteosarcoma malignancy.
All the authors state that they have no conflicts of interest.
We acknowledge the support of the Nice–Sophia Antipolis Transcriptome Platform of the Marseille-Nice Genopole. We thank Mrs Marty, Miss Chantrenne, and Dr Ostertag (INSERM, U606, Paris, France) for their technical contributions. We thank Drs Fournier (Novartis, Basel, Switzerland) and Price (UCSD, La Jolla, CA, USA) for providing OHS4 cells, Dr Rochet (INSERM, UMR6235, Nice, France) for the gift of CAL72 cells, and Dr Schutze (University of Wurzburg, Germany) for providing pTriEx1-Cyr61 vector. This work was supported in part by INSERM and the Association Rhumatisme et Travail, Paris, France. FL was supported by AECC project, “UTE project FIMA” agreement, RTICCC C03/10, PI042284, and PI10/01580 (to AP).