CTGF enhances the motility of breast cancer cells via an integrin-αvβ3–ERK1/2-dependent S100A4-upregulated pathway

Metastasis, the major cause of mortality for cancer patients, is a complex and multi-stage process in which secondary tumors are formed in distant sites (Van't Veer and Weigelt, 2003). Typically, the development of metastasis involves several steps that comprise cellular transformation and tumor growth, angiogenesis and lymphangiogenesis, entry of cancer cells into the circulation by intravasation, anchorage and/or attachment in the target organ, invasion of the target organ by extravasation, and proliferation within the organ parenchyma (Hanahan and Weinberg, 2000). The migratory ability of a cancer cell is important for many of these steps, and therefore is correlated with tumor metastasis. In particular, several secreted proteins, including vascular endothelial growth factor (VEGF), PC-cell-derived growth factor (PCDGF/GP88, progranulin), epidermal growth factor (EGF), and stromal-cell-derived factor 1 (CXCL12) confer ameliorated migratory ability to breast cancer cells and are correlated with more advanced-stages of breast cancer (Darash-Yahana et al., 2004; Miralem et al., 2001; Price et al., 1999; Tangkeangsirisin and Serrero, 2004). These molecules have been proposed as therapeutic targets (Barrett-Lee, 2005). Identification and molecular characterization of new molecules that are involved in tumor progression, therefore, have important clinical implications.

In order to invade, a tumor cell must undergo major changes in shape. Cellular motility depends on localized actin polymerization at the leading edge of the cells (Kirfel et al., 2004), and the polymerization and depolymerization of actin filaments must be under dynamic control. Simultaneously, paxillin and vinculin interact at focal contacts of the actin stress fibers, providing a link to the extracellular matrix (e.g. fibronectin and vitronectin). These cytoskeletal changes enable the invading cell to pass through the stromal cells, extracellular matrix and endothelial cell layer. Integrins, paxillin, selectins, transmembrane receptor tyrosine kinases, phospholipids, focal adhesion kinases (FAKs), GTPases and the S100 calcium binding protein A4 (S100A4) calcium binding protein have been described as being involved in regulating the organization of the actin cytoskeleton (Kim and Helfman, 2003; Turner, 2000; Brunton et al., 2004). Some of these molecules have also been implicated in the malignant phenotype of certain carcinomas, such as ERBB-2, in mammary carcinomas (Mariani et al., 2005). ERBB2 increases the potential for metastasis by upregulating the expression of the prometastatic S100A4 in medulloblastoma (Hernan et al., 2003).

Connective tissue growth factor (CTGF, also known as CCN2) belongs to the CCN family (Bork, 1993). This family consists of six members, CTGF, NOVH, CYR61, WISP1, WISP2 and WISP3 (Perbal, 2004) that all possess an N-terminal signal peptide identifying them as secreted proteins. CCN proteins probably carry out their biological activity through binding and activating of the cell surface integrins, accompanied by activation of Akt and/or MAPK signal cascades (Perbal, 2004). The biological properties of CCN proteins involve the stimulation of cellular proliferation, migration, adhesion, extracellular matrix formation, and also the regulation of angiogenesis and tumorigenesis (Lau and Lam, 1999). Overexpression of CTGF, WISP1, and CYR61 in breast tumor cells have been linked to tumor size and lymph node metastasis (Xie et al., 2001), suggesting that these CCN proteins are involved in the progression of breast cancer. Recently, we reported that CYR61 influences the resistance to chemotherapeutic-agent-induced apoptosis in human breast cancer MCF-7 cells (Lin et al., 2004). However, the biological activities of CTGF in breast cancer have not yet been explored. CTGF serves as an angiogenic factor in collaboration with matrix metalloproteinases (Kondo et al., 2002). Use of in vitro and in vivo selection models and large-scale microarray analysis has revealed that CTGF crucial for the formation of osteolytic bone metastasis in breast cancer (Kang et al., 2003; Minn et al., 2005). Thus, CTGF plays an important role in breast cancer progression. However, the precise role of CTGF in breast cancer metastasis is still unknown.

In this study, we demonstrate that CTGF can modulate the cytoskeletal reorganization and in vitro migratory behavior of breast cancer cells. We show that activation of ERK1/2 through integrin αvβ3 confers the enhanced cellular motility. Microarray and reverse transcription (RT)-PCR analysis revealed that the crucial prometastatic S100A4 is significantly upregulated in cells and breast tumors that overexpress CTGF. Our results support a new mechanism, in which integrin-αvβ3–ERK1/2-dependent upregulation of S100A4 contributes to CTGF-enhanced migratory ability.

CTGF levels are correlated with most of the advanced stages of breast cancers (Xie et al., 2001). Hence, we investigated expression levels of CTGF by northern blot analysis in several adeno/ductal carcinoma of human breast cancer cell lines including BT474, BT483, T47D, MDA-MB-453, MCF-7, MDA231 and MDA-MB-435. The breast cancer cell lines BT474, BT483, T47D, MDA-MB-453 and MCF-7 expressed extremely low or undetectable levels of CTGF mRNA, whereas expression was high in MDA231 and MDA-MB-435 (Fig. 1A). No correlation was found between CTGF expression and their histological types. Since cell migration is crucial for cancer cell invasiveness, we next tested the cellular migratory ability by a wound healing migration assay. MDA231 and MDA-MB-435 cells that exuberantly expressed CTGF mRNA exhibited a higher level of migratory ability than the cell lines with a lower level of CTGF mRNA (Fig. 1B). To assess whether CTGF was directly involved in the regulation of breast cancer cell motility, both non-migrating MCF-7 and migrating MDA231 cells were employed to generate the CTGF-overexpression (MCF-7/CTGF) and antisense-CTGF expression (MDA231/AS) cells, respectively. The CTGF expression levels in vector control cells and stable transfectants was compared using RT-PCR and western blotting. Both mRNA and protein levels of CTGF were significantly higher in MCF-7/CTGF than in MCF-7/neo cells (Fig. 1C). However, expression of CTGF was dramatically inhibited by antisense-CTGF orientation in MDA231/AS cells. Since, CTGF has already been reported to act as a mitogen in endothelial cells (Shimo et al., 1999), we sought to characterize the cellular growth rate of control cells and transfectants, by performing the MTT assay 1-6 days after cell seeding (Fig. 1D). No appreciable difference in cell growth ability was evident among these cells (Fig. 1D), suggesting that CTGF does not have any mitogenic effect in human breast cancer cells. Furthermore, the migratory ability of these transfectants was analyzed using a Boyden-chamber migration assay. As shown in Fig. 1E, the overexpression of CTGF increased the migratory ability by approximately 1.7-fold in MCF-7 cells. Knockdown of CTGF expression inhibited the migratory ability by approximately 60% in MDA231 cells (Fig. 1F).

Transfection of MCF-7 cells with CTGF led to a change in cellular morphology (Fig. 2a,e). MCF-7/CTGF cells became more spindle-like with long, protruding filamentous processes. Double staining of the MCF-7/neo cells using antibodies conjugated to Texas-Red-conjugated phalloidin and monoclonal anti-paxillin (the latter as a marker for focal adhesions) revealed few, if any, actin stress fibers, whereas paxillin was diffusely distributed in the cells (Fig. 2b,c,d). The MCF-7/CTGF cells exhibited numerous stress fibers (Fig. 2f) with patches of paxillin staining appearing at the leading edges of actin stress fibers (Fig. 2f,g,h). By contrast, overexpression of CTGF changed the MDA231 cell morphology from spindle-like to epithelial-like (Fig. 2i,m). In MDA231/neo cells, a plethora of both actin stress fibers and paxillin were evident, and paxillin was macroaggregated at focal adhesions connected to actin stress fibers (Fig. 2i,j,k). Reducing the level of endogenous CTGF by transfection with CTGF/AS significantly decreased the amount of actin stress fibers and paxillin (Fig. 2n,o,p). These results support the hypothesis that CTGF promotes actin contractility events and the dynamic regulation of focal contact structures in breast cancer cells.

Since CTGF is a secreted protein, we explored the mechanism by which CTGF-mediated outside-in signals involved in CTGF-induced cellular migration using purified Fc-tagged recombinant CTGF (rCTGF) and the conditioned medium (CM) derived from MDA231 (MDA231/CM) to treat MCF-7 cells. Both rCTGF and MDA231/CM effectively enhanced the MCF-7 cell migratory ability (Fig. 3A,B). CTGF-neutralizing antibody reversed the MDA231/CM-induced migration in a dose-dependent manner (Fig. 3B). Similar effects were also found in wild-type MDA231 cells, and the migratory ability of MDA231 was dose dependently inhibited by treatment with CTGF-neutralizing antibody (Fig. 3C). Simultaneously, cells changed from a spindle-like to an epithelial-like morphology by treatment with CTGF-neutralizing antibody in MDA231 cells (Fig. 3Da,e,i). Immunofluorescence staining showed that both actin stress fibers and paxillin-containing focal adhesions were disrupted in CTGF-neutralizing antibody-treated MDA231 cells (Fig. 3D).

Integrins αvβ3, αIIbβ3, αMβ2 and α5β1 are classical cell-surface receptors of CTGF (Babic et al., 1999; Jedsadayanmata et al., 1999; Schober et al., 2002; Weston et al., 2003). Among these integrins, αvβ3 is the only one that associates with the invasiveness in human breast cell lines and advanced tumor progression (Berry et al., 2004; Damjanovich et al., 1997; Felding-Habermann et al., 2001; Gui et al., 1996; Jones et al., 1995). Treatment with integrin-αvβ3_-blocking antibody dose dependently reversed the CTGF-induced migratory ability (Fig. 3E), suggesting that integrin αvβ3 is a key receptor for CTGF-mediated signaling pathway. Moreover, CTGF expression could upregulate integrin β3 expression and integrin αvβ3 formation (supplementary material Fig. S1), suggesting the amplification loop strengthens the CTGF-αvβ3 signaling pathway. Taken together, our results indicate that CTGF promotes cytoskeleton rearrangement and migratory ability by an autocrine/paracrine outside-in signaling manner through binding to integrin αvβ3 in human breast cancer cells.

To investigate the possible signaling pathways involved in CTGF function, we used western blotting to detect the phosphorylation status of Akt and mitogen-activated protein kinases, which are crucial signaling pathways in cancer progression (Sebolt-Leopold and Herrera, 2004; Vivanco and Sawyers, 2002). The phosphorylation status of most of the kinases was not altered, whereas ERK1/2 phosphorylation was enhanced in MCF-7/CTGF cells and decreased in MDA231/AS cells (Fig. 4A). To explore whether ERK1/2 is involved in CTGF-induced cell migration, we used the MEK1-specific inhibitor PD98059 and a constitutively active MEK1 mutant to modulate the ERK1/2 phosphorylation status. The CTGF-enhanced migratory ability was reduced when CTGF-mediated ERK1/2 phosphorylation was inhibited by PD98059 inhibitor in MCF-7/CTGF cells (Fig. 4B). However, transfection with active mutant MEK1 caused ERK1/2 re-phosphorylation in MDA231/AS cells, leading to the enhancement of migratory ability (Fig. 4C). These results indicated that the ERK1/2 pathway is essential for CTGF-mediated cell migration.

The recombinant CTGF protein also dramatically induced ERK1/2 phosphorylation within 10-60 minutes (Fig. 4D). Treatment with integrin-αvβ3-blocking antibody inhibited the rCTGF-induced ERK1/2 phosphorylation in MCF-7 cells (Fig. 4E). Under similar circumstances, integrin-αvβ3-blocking antibody diminished the CTGF-induced ERK1/2 activation in MCF-7/CTGF cells (Fig. 4F). Altogether, these results indicate that the integrin-αvβ3–ERK1/2 pathway is crucial for CTGF-induced ERK1/2 activation and subsequent cellular migration.

Activated ERK1/2 regulates membrane protrusions and focal adhesion turnover through various substrates, such as cytoplasmic proteins – including MCLK, paxillin and FAK – or nuclear proteins – including Elk-1 and Ets (Deak et al., 1998; Frodin and Gammeltoft, 1999; Fukunaga and Hunter, 1997; Hunger-Glaser et al., 2003; Klemke et al., 1997; Waskiewicz et al., 1997). Therefore, we examined the subcellular localization of phosphorylated (P)-ERK1/2 using western blotting and immunofluorescence staining. CTGF overexpression promoted their activation and nuclear translocation in MCF-7 cells (Fig. 4G,H), while consistently decreasing the levels of P-ERK1/2 in the nucleus of MDA231/AS cells (Fig. 4G,H). Thus, CTGF can promote ERK1/2 phosphorylation and nuclear translocation, which then may activate gene expression.

The question remained as to which gene was the possible downstream effecter that contributed to CTGF-mediated ERK1/2-dependent cell migratory effect. Since CTGF induced P-ERK1/2 nuclear translocation, transcriptional regulation might occur. We used a cDNA microarray to identify the genetic expression profile (Tables 1 and 2). S100A4 (Fig. 5A) and E-cadherin are of particular interest because they function as regulators of metastasis of human tumors, and have been implicated in cytoskeleton-membrane interaction, migratory behaviors and malignancy in cancer cells (Cho et al., 2003; Cui et al., 2004; Davies et al., 1996; Ebralidze et al., 1989; Glenney, Jr et al., 1989; Hernan et al., 2003; Jiang, 1996; Kim and Helfman, 2003; Masiakowski and Shooter, 1988; Mazzucchelli, 2002; Missiaglia et al., 2004; Nikitenko et al., 2000; Platt-Higgins et al., 2000; Rudland et al., 2000). To confirm the expression levels of S100A4 and E-cadherin in the stable transfactants and their control cells, we assayed S100A4 and E-cadherin levels by RT-PCR and western blotting. The expression of E-cadherin mRNA (data not shown) and protein was undetectable both in MDA231/neo and MDA231/AS cells (supplementary material Fig. S2); moreover, E-cadherin level was not altered under ectopic expression of CTGF in MCF-7 cells (supplementary material Fig. S2), indicating that the deregulation of E-cadherin mRNA in microarray data (Table 1) is a false-positive and without biological significance. However, both the mRNA and protein expression levels of S100A4 were upregulated in MCF-7/CTGF cells but downregulated in MDA231/AS cells (Fig. 5B). We also found that S100A4 was overexpressed in MDA231 and MDA435 cells, suggesting a strong association with CTGF expression and cellular aggressiveness (supplementary material Fig. S3). Subsequently, treatment with the MEK1 inhibitor PD98059 effectively abolished the CTGF-induced S100A4 upregulation (Fig. 5C), whereas transfection with the constitutively active MEK1 mutant significantly increased S100A4 expression in MDA231/AS cell (Fig. 5D). These results strongly suggest that CTGF-mediated S100A4 upregulation depend on the ERK1/2 signaling pathway.

To ascertain the role of S100A4 in CTGF-mediated cell migration, we generated the pcDNA4-S100A4 and pcDNA4-AS-S100A4 constructs to establish the doubly transfected stable transfectants of MDA231/AS and MCF-7/CTGF cells, respectively. S100A4 expression levels were reduced by AS-S100A4 in MCF-7/CTGF cells, which also decreased the migratory ability (Fig. 5E). Restoring levels of S100A4 in MDA231/AS cells significantly enhanced the migratory ability (Fig. 5F). Immunofluorescence staining further showed that transfection with S100A4 significantly promoted re-polymerization of actin stress fibers and aggregation of focal adhesion complexes in MDA231/AS cells (Fig. 5G). These results indicate that S100A4 is a crucial downstream effecter involved in CTGF-mediated cytoskeleton rearrangement and cellular migration.

To investigate whether the CTGF-induced S100A4 expression plays a causal role in tumor metastasis, we injected MDA231/neo/pcDNA4, MDA231/neo/S100A4, MDA231/AS/pcDNA4, and MDA231/AS/S100A4 cells intravenously into the tail vein of nude mice and measured the metastatic colonization. Whereas tumors expressing the AS-CTGF formed very few macroscopically visible metastases in the lungs, tumors that co-expressed AS-CTGF and S100A4 formed a large number of metastases (Fig. 5H). In the MDA231/neo/pcDNA4 group, the mean lung nodule number was 39; in the MDA231/AS/pcDNA4 group, the mean lung nodule number was 9. The numbers of visible metastatic nodules in mice injected with MDA231/neo/pcDNA4 cells were higher than those in MDA231/AS/pcDNA4-bearing mice (Table 3). In the MDA231/AS/S100A4 group, the mean lung nodule number was 24 (Table 3), indicating that transfection with S100A4 dramatically increased the mean number of metastatic nodules and the occurrence of lung metastasis in MDA231/AS tumor-bearing mice. Together, these data demonstrate that CTGF-mediated tumor metastasis requires the expression of S100A4

To establish whether expression levels of CTGF and S100A4 are correlated in clinical breast cancer, 24 primary human breast cancer specimens and their non-tumor counterparts were collected, and analyzed semi-quantitatively by RT-PCR and quantitatively by densitometry (Fig. 6). Non-tumor cells displayed extremely low or undetectable levels of CTGF and S100A4 expression. Conversely, all tumor tissues expressed both CTGF and S100A4 (Fig. 6A). Based on our observations in breast cancer cells, the expression levels of CTGF and S100A4 were found to closely correlate in-patient samples (Fig. 6B, r²=0.65, P=0.0012). These data strongly support the hypothesis that CTGF signaling upregulates the expression of S100A4 in primary human breast cancer.

In this study, we demonstrate that the molecular mechanism by which CTGF-confers cellular metastatic ability is mediated by S100A4 upregulation by integrin αvβ3 and/or ERK1/2. This is based on the following evidence. First, overexpression of CTGF appreciably increases the migratory ability of MCF-7 cells. Conversely, knockdown of CTGF abolishes the migratory ability of MDA231 cells. Second, CTGF expression leads to morphological alterations and formation of F-actin and focal adhesions. Third, ERK1/2 activation is essential for the CTGF-mediated migratory effects. Fourth, blockade of the CTGF–integrin-αvβ3 axis attenuates CTGF-induced ERK1/2 activation and subsequent cellular migration. Fifth, the prometastatic S100A4 gene is regulated by the signaling cascades of CTGF–integrin-αvβ3–ERK1/2 and contributes to the metastatic ability. Finally, CTGF expression levels are correlated with S100A4 expression levels in primary human breast tumors.

When CTGF-overexpressing cells were treated with the MEK1 inhibitor PD98059 their migratory ability was inhibited. Moreover, constitutively active MEK1 also prevented the migratory ability in MDA231/AS cells. Thus, our data suggest that the ERK1/2 signaling pathway is crucial for CTGF-induced cell migration. In support of our observations, others have demonstrated that constitutive activation of ERK1/2 frequently occurs in a variety of cancers, including lung cancer, cervical cancer and breast tumors (Adjei, 2005; Branca et al., 2004; Milde-Langosch et al., 2005). Constitutive ERK1/2 activation has also been observed in different breast cancer cells (Santen et al., 2002), and activation of ERK1/2 leads to cell migration (Krueger et al., 2001). Mutationally activated receptor tyrosine kinases – such as Ras – or binding of growth factors – such as transforming growth factor alpha (TGFα), epidermal growth factor (EGF), vascular endothelial growth factor-A (VEGF-A), platelet-derived growth factor beta (PDGFβ) and heregulin – to their cognate receptors may induce ERK1/2 activation (Seton-Rogers et al., 2004; Gollob et al., 2006). We presently found that CTGF induces cell migration in an ERK1/2-dependent manner. Thus, ERK1/2 signaling, which is vital for breast cancer cell migration, is regulated by CTGF and contributes to the increased migratory ability. Activated ERK1/2 regulates membrane protrusions and focal adhesion turnover via various substrates. The cytoplasmic substrates of P-ERK1/2 include several protein kinases, such as RSK1, MSK1, MNK1/2, myosin light chain kinase (MCLK) and FAK (Deak et al., 1998; Frodin and Gammeltoft, 1999; Fukunaga and Hunter, 1997; Hunger-Glaser et al., 2003; Klemke et al., 1997; Waskiewicz et al., 1997), the protease calpain (Glading et al., 2004) and paxillin (Liu et al., 2002). However, ERK1/2 translocation into the nucleus after activation contributes to transcriptional regulation. The key mechanism is believed to involve the further phosphorylation and activation of key transcription factors such as Ets, CREB, Jun, and Myc (Santen et al., 2002). These events are likely to be mediated through stimulation of transcription in the nucleus. In our study, CTGF-mediated ERK1/2 phosphorylation is predominantly located in the nucleus, indicating that CTGF-mediated signaling is possibly regulated by ERK1/2-dependent transcriptional regulation, rather than by paxillin-FAK interaction in breast cancer cells.

Since CTGF expression promotes ERK1/2 activation and subsequent cellular migration, we investigated the origin of the CTGF-mediated signaling pathway. Although CTGF is a secreted protein, it carries out its biological activities via intracellular transport (Wahab et al., 2001). Thus, clarification of the origin of CTGF-mediated signaling is important for further in vitro studies that may contribute to therapeutic approaches. According to our data, rCTGF rapidly and transiently induces ERK1/2 phosphorylation. Moreover, CTGF-neutralizing antibody and integrin-αvβ3-blocking antibody are dose dependently inhibitory. These results indicate that the outside-in signaling is important for CTGF-mediated biological activities in breast cancer cells. CTGF exerts a range of diverse functions, many of which involve the binding to cell-surface integrins, such as integrins αvβ3, αIIbβ3, αMβ2 and α5β1 (Babic et al., 1999; Jedsadayanmata et al., 1999; Schober et al., 2002; Weston et al., 2003). Of those, integrin αvβ3 is the only integrin highly correlated with breast cancer metastasis and its progression (Taddei et al., 2003). CTGF induces adhesion and activation of rat hepatic stellate cells by C-terminally binding to integrin αvβ3 (Gao and Brigstock, 2004). Moreover, this integrin, which is important in osteoclast attachment to bone, is highly expressed in breast cancer cells and bone (Liapis et al., 1996). Although antibody against integrin αvβ3 interfered with CTGF-induced ERK1/2 activity and migration, it is possible that integrin-αvβ3-mediated adhesion is required as a parallel pathway for ERK1/2 activity and migration. Additionally, ectopic expression of CTGF induced integrin β3 expression (supplementary material Fig. S1A) and integrin αvβ3 formation in MCF-7 cells (supplementary material Fig. S1B,C). Also, knockdown of CTGF reduced expression of integrin β₃ (Table 1 and supplementary material Fig. S1A) and formation of integrin αvβ3 in MDA231 cells (supplementary material Fig. S1B,C). This regulation suggests an amplification loop within the CTGF–integrin-αvβ3 signaling pathway induced by CTGF expression. All of the above suggests that the integrin αvβ3 is the most crucially involved in breast cancer metastasis. Our study also shows that the CTGF-mediated signaling pathway and the cell migratory effects are mainly activated through integrin αvβ3.

S100A4, the crucial molecule downstream the CTGF-mediated signaling pathway, is closely associated with metastasis in other breast cancer cells, as well as in rodent and in human cancer specimens (Ambartsumian et al., 1996; Ambartsumian et al., 2001; Cho et al., 2003; Cui et al., 2004; Hernan et al., 2003; Mazzucchelli, 2002; Missiaglia et al., 2004; Platt-Higgins et al., 2000). Elevated levels of S100A4 have been found in metastatic cancers in mice and in humans (Ebralidze et al., 1989; Nikitenko et al., 2000). Its expression is also strongly correlated with the demise of breast cancer patients (Rudland et al., 2000). Increased levels of rat or human S100A4 result in metastatic capability of initially benign mammary tumor cells in rats (Ambartsumian et al., 1996; Davies et al., 1993; Davies et al., 1996; Lloyd et al., 1998). Although the precise function of S100A4 is not entirely clear, its interaction with cytoskeletal moieties and its early role in EMT indicates that S100A4 expression is involved in the migratory process (Okada et al., 1997). This may result from its role in the reorganization of cytoskeletal components, such as nonmuscle myosin II (Ford et al., 1995; Kriajevska et al., 1994), actin (Watanabe et al., 1993) and tropomycin (Takenaga et al., 1994), which indicates that S100A4 is involved in motility by interacting with various components of the cytoskeleton (Okada et al., 1997). In addition, S100A4 can act as an angiogenic factor (Ambartsumian et al., 2001), and can negatively regulate bone mineralization and osteoblast differentiation (Duarte et al., 2003). In endothelial cells, S100A4 activates transcription of matrix metallopeptidases MMP11, MMP13 and MMP14 (Schmidt-Hansen et al., 2004a; Schmidt-Hansen et al., 2004b). Accordingly, we hypothesize that CTGF-regulated S100A4 expression not only confers increased metastasis, but also regulates angiogenesis, bone homeostasis and the transcription of MMPs. Together, our study and studies by others provide the evidence that S100A4 is a crucial factor in modulating motility in human breast cancer cells, and that its expression can be regulated by CTGF.

In conclusion, we provide a detailed mechanism describing the potential role of CTGF in migratory behavior via an integrin-αvβ3–P-ERK1/2-dependent upregulation of S100A4. This is the first identification of a new molecular mechanism regulating the CTGF-driven prometastatic pathway, possibly providing a new target for therapeutic intervention in metastatic breast cancer.

The human breast cancer cell lines MCF-7 and MDA231 were obtained from the American Type Culture Collection. BT474, BT483, MDA-MB-453, MDA231 and MDA-MB-435 cells were provided generously by the National Health Research Institute (Taipei, Taiwan). These cell lines were maintained in minimum essential medium (MEM) supplemented with 10% fetal bovine serum (FBS; Gibco BRL, Gaithersburg, MD), 2 mM L-glutamine (Life Technologies, Carlsbad, CA), 100 μg/ml streptomycin and 100 U/ml penicillin in a humidified 5% CO₂ atmosphere. Monoclonal mouse anti-human CTGF antibody was purchased from R&D Systems (Minneapolis, MN). Antibodies against integrin αvβ3, α-tubulin, P-ERK1/2 and ERK1/2 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Polyclonal rabbit anti-human S100A4 antibody was purchased from Abnova Biotechnology (Taipei, Taiwan). Anti-paxillin antibody was from BD Biosciences (Franklin Lakes, NJ). Goat anti-rabbit or anti-mouse IgG conjugated with either fluorescein isothiocynate (FITC) or tetramethylrhodamine isothiocyanate (TRITC) were purchased from Jackson Immunoresearch (West Grove, PA). Diamidino phenylindole dimethylsulfoxide (DAPI) was purchased from Molecular Probes (Eugene, OR). The constitutively active MEK1 mutant construct (substitution of the regulatory phosphorylation sites S218D and S222D) as described previously (Brunet et al., 1994) was gift from Ruey-Hwa Chen.

Migration of human breast cancer cells through polycarbonate filters was examined in 24-well modified Boyden chambers (pore size 8 μm). The lower wells of the chamber were loaded with 1 ml of MEM. Breast cancer cells (50,000 cells per 100 μl) were placed in the upper wells. After 6 hours of incubation, cells on the lower surface of the filter were fixed and stained with Crystal Violet and counted using a microscope (type 090-135.001; Leica Microsystems, Wetzlar, Germany). Incubations were performed in triplicate and each experiment was repeated at least three times.

A 1.2-kb pair sense- or antisense-orientation cDNA fragment of human CTGF was each cloned into the vector pcDNA3 to produce sense- and antisense-oriented CTGF expression vectors. Furthermore, 0.4 kb sense- or antisense cDNA fragments of human S100A4 were each cloned into the vector pcDNA4 to produce sense-oriented (pcDNA4-S100A4) and antisense-oriented (pcDNA4-AS-S100A4) S100A4 expression vectors. Cells were transfected with plasmids using the lipofectin transfection reagent (Invitrogen, Carlsbad, CA). Transfected cells were grown in an atmosphere of 5% CO₂ at 37°C in MEM supplemented with 10% FBS. After 48 hours of transfection with pcDNA3-based plasmids, cells were trypsinized and replated in MEM containing 10% FBS 800 μg/ml G418. G418-resistant clones (MCF-7/neo, MCF-7/CTGF, MDA231/neo, and MDA231/AS) were selected and pooled for further studies. Doubly transfected cells harboring another pcDNA4-based plasmid were trypsinized after 48 hours of transfection, and replated in MEM supplemented with 10% FBS, 50 μg/ml G418 and 500 ug/ml zeocin. Zeocin-resistant clones were selected, expanded and pooled for further studies.

Proteins in the total cell lysate (40 μg of protein) were separated on 10% SDS-PAGE and electrotransferred to a polyvinylidene difluoride membrane (Immobilon-P membrane; Millipore, Bedford, MA). After the blot was blocked in a solution of 5% skimmed milk, 0.1% Tween 20 and PBS, membrane-bound proteins were probed with primary antibodies against CTGF, S100A4, P-ERK1/2, ERK1/2, SP1 (Santa Cruz Biotechnology, Santa Cruz, CA). The membrane was washed and then incubated with horseradish peroxidase-conjugated secondary antibodies for 30 minutes. Antibody-bound protein bands were detected with enhanced chemiluminescence reagents (Amersham Pharmacia Biotech, Piscataway, NJ) and photographed with Kodak X-Omat Blue autoradiography film (Perkin Elmer Life Sciences, Boston, MA).

Total RNA was isolated using Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions, and 1 μg was reverse transcribed into single-stranded cDNA with M-MLV reverse transcriptase and random hexamers (Promega, Madison, WI). Amplification of cDNAs was performed by PCR with specific primer pairs. Primers were, for CTGF 5′-ACTGTCCCGGAGACAATGAC-3′ (forward) and 5′-TGCTCCTAAAGCCACACCTT-3′ (reverse), for S100A4 5′-TCTCTCCTCAGCGCTTCTTC-3′ (forward) and 5′-CTTCCTGGGCTGCTTATCTG-3′ (reverse), for GAPDH 5′-ACCCAGAAGACTGTGGATGG-3′ (forward) and 5′-GTCCACCACCCTGTTGCTGT-3′ (reverse). PCR conditions were: denaturing once at 95°C (10 minutes), 95°C (1 minute), 52°C (1 minute) and 72°C (1 minute) for 30 cycles, once at 72°C (10 minutes). PCR products were analysed using agarose gel electrophoresis.

For immunofluorescence staining, all fixation and staining procedures were conducted at room temperature. Cultures were grown in a Lab-Tek four-chamber slide apparatus (Nalge NUNC, Rochester, NY), fixed with cold 4% paraformaldehyde for 15 minutes, washed three times with phosphate-buffered saline (PBS) and permeabilized with 0.1% Triton X-100 in PBS for 5 minutes. After washing three times with 0.05% Tween 20 in PBS (PBST), cultures were blocked in 5% nonfat milk (Carnation) in PBST for 30 minutes. Cells were incubated with primary antibodies against paxillin or P-ERK1/2) for 1 hour and incubated at room temperature with Texas-Red-conjugated phalloidin (Molecular Probes) as specified in each experiment. After washing, the cells were incubated with secondary antibody (1:100) and DAPI (1:5000) for 60 minutes. The control samples included those treated with either secondary antibody alone or with pre-immune mouse serum. The slides were mounted with VectorShield (Vector Laboratories, Burlingame, CA) and examined under a Nikon Eclipse E600 upright microscope equipped with fluorescent devices.

Total RNA was isolated using Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions and 10 μg were separated by electrophoresis on 1.2% agarose. RNA was transferred to Hybond-XL membranes (Amersham Pharmacia Biotech, Piscataway, NJ), then hybridized with cDNA probe and randomly labeled with α[³²P]dATP (3000 Ci/mmol/l; DuPont/NEN, Boston, MA). Probes were labeled using the Prime-It II Random Primer kit (Stratagene, La Jolla, CA). Membranes were stripped and reprobed with cDNA for GAPDH mRNA and exposed on Kodak X-OMAT LS film (Eastman Kodak, Rochester, NY) was exposed for various periods of time to the blots.

Cells were washed once with PBS and harvested in 0.05% trypsin/0.025% EDTA. Detached cells were washed with PBS containing 1% FCS and 1% penicillin-streptomycin (wash buffer), and resuspended in wash buffer (10⁶ cells per 100 μl). Cells were then incubated with anti-integrin αvβ3 monoclonal antibodies (Santa Cruz Biotechnology) or respective isotype controls and FITC-conjugated secondary antibodies (Sigma-Aldrich) for 30 minutes on ice. The labeled cells were washed in wash buffer, fixed in PBS containing 1% paraformaldehyde and then analyzed on a FACSVantage (BD Biosciences).

The Agilent human 1 cDNA microarray (Agilent Technology) containing 18,564 spots of 13,574 different genes was used in this study. Fifteen micrograms of purified total RNA were converted to cDNA using a 3DNATM Array 50 Expression Array Detection Kit (Genisphere). RNA was labeled with Cy3, and RNA from Universal Human Reference RNA was labeled with Cy5. Correspondingly synthesized cDNA products were combined and concentrated by ethanol precipitation and suspended in hybridization buffer. Labeled cDNA was hybridized to Agilent human 1 cDNA microarray (Agilent Technologies) at 65°C for 17 hours. After hybridization, slides were washed in 5×SSC with 0.01% SDS at room temperature for 5 minutes and 0.06×SSC at room temperature for 2 minutes. Washed microarrays were then hybridized with Cy3 and Cy5 dendrimers in formamide-based buffer at 53°C for 3 hours. After hybridization with dendrimers, slides were washed in 2×SSC with 0.01% SDS at 42°C for 15 minutes, 2×SSC at room temperature for 10 minutes, and 0.2×SSC at room temperature for 10 minutes. Washed microarrays were scanned with a Virtek fluorescence reader (Virtek, CA) at 532 nm and 635 nm for Cy3 and Cy5, respectively. Scanned images were analyzed by Array-Pro image acquisition software (Media Cybernetics); an image analysis algorithm was used to quantify signal and background intensity for each target spot (gene). The mean intensity of the spot area was computed as the microarray raw data. The data were further analyzed with software package S-Plus 6.1 (Venables, 2002).

Cells were washed and resuspended in PBS. Subsequently, a single-cell suspension containing 10⁶ cells in 0.1 ml of PBS was injected into the lateral tail vein of 7-week-old female nude mice (supplied by the animal center in the College of Medicine, National Taiwan University, Taipei, Taiwan). Mice were killed after 18 weeks. (Our preliminary study in this animal model indicated that MDA231 cells developed numerous lung metastasis nodules by 18 weeks.) All organs were examined for metastasis formation. Lungs were removed, weighed and fixed in 10% formalin. Lung tumor colonies were counted under a dissecting microscope. All animal work was performed usinmg protocols approved by the Institutional Animal Care and Use Committee of the College of Medicine, National Taiwan University.

Cell monolayers were wounded 24 hours after plating by scratching with a pipette tip. Debris was removed by washing. Images were taken at 20× magnification (Nikon Ph1 DL; NA 0.4) with a Nikon TMS microscope equipped with a Nikon F-601 camera. Distances between wound edges were measured at five sites/image (n=3). Alternatively, wounded monolayers were subjected to time-lapse studies for 3 hours at 10× magnification using a Zeiss Axiovert 200 microscope equipped with the AxioCam digital system.

To purify the recombinant CTGF protein (rCTGF) protein, 293T cells were used to express rCTGF. Protein A Sepharose 4 Fast Flow was used with an Atka-fast protein liquid chromoatography system (Amersham Pharmacia, Freiburg, Germany). Columns were equilibrated with PBS (pH 7.0), the supernatant (15-45 ml) was applied at a flow rate of 2 ml/minute. Columns were washed with ten column volumes of PBS and the protein was eluted with elution buffer (0.1 M glycine pH 2.5). Directly thereafter, eluted fractions were neutralized using 3 M Tris-HCl pH 8. The rCTGF protein was desalted using a PD-10 column (Amersham Pharmacia) and the purity of the protein checked by silver staining and western blotting.

Anti-CTGF antibody was raised in rabbits by immunization with a synthetic CTGF peptide composed of 20 amino acids (aa 243 to aa 263; EADLEENIKKGKKCIRTPKIS). This sequence is shared with CTGF (also known as Fisp12 in mouse) (the mouse homologue of), but not CYR61, NOV, WISP1 (also known as Elm1 in mouse), WISP2 (also known as rCop1 in mouse) or WISP3. Anti-CTGF antibody was purified from the serum as previously described (Shimo et al., 1999).

We thank the anonymous reviewers for helpful comments. This work was supported by grants from the Department of Industrial Technology, Ministry of Economic Affairs, Taipei, Taiwan (95-EC-17-A-19S1-016), the National Science Council, Taiwan (NSC95-2314-B-002-318-MY3, NSC94-2320-B-002-012, and NSC94-2323-B-002-010), and the National Taiwan University Hospital (NTUH-94M04). We thank Tung-Tien Sun (Department of Urology, New York University School of Medicine, NY) for revising this paper. Ruey-Hwa Chen is thanked for providing the active MEK1 plasmid.