Rho GDP dissociation inhibitor β (Rho-GDIβ), an inhibitor of Rho GTPases, is primarily expressed by hematopoietic cells but is also found in epithelial cancer cells. Recently, we have identified Rho-GDIβ as a target of the transcription factor Ets1. Here, we show that, in breast cancer cells, Ets1 regulates Rho-GDIβ expression and binds to the upstream region of the Rho-GDIβ gene. Furthermore, in primary breast cancer, Rho-GDIβ is coexpressed with Ets1. Studying the function of Rho-GDIβ in breast cancer, we found that a Rho-GDIβ–specific small interfering RNA increased cellular migration but also decreased the expression of cyclooxygenase-2 (Cox-2) oncogene as shown by microarray, quantitative reverse transcription-PCR, and Western blot analyses. Further studies revealed that Rho-GDIβ regulates Cox-2 gene at least partly on the transcriptional level, most likely by activating nuclear factor of activated T cells 1 (NFAT-1). Vav-1, an interaction partner of Rho-GDIβ, was also found to interfere with Cox-2 expression and NFAT-1 cellular distribution, suggesting a cooperative action of Rho-GDIβ and Vav-1 on Cox-2 expression. To explore the importance of Rho-GDIβ for the survival of breast cancer patients, two cohorts, including 263 and 117 patients, were analyzed for clinical outcome in relation to Rho-GDIβ RNA and protein levels, respectively. Expression of Rho-GDIβ was not associated with either disease-free or overall survival in the two patient population. Our data suggest that the expression of Rho-GDIβ in breast cancer is neither beneficial nor disadvantageous to the patient. This may be the net effect of two opposing activities of Rho-GDIβ, one that suppresses tumor progression by inhibiting migration and the other that stimulates it by enhancing Cox-2 expression. [Cancer Res 2007;67(22):10694–702]
The family of Rho GDP dissociation inhibitors (Rho-GDI) consist of three members: Rho-GDIα, Rho-GDIβ, and Rho-GDIγ (1–3). Whereas Rho-GDIα is ubiquitously expressed, the expression of Rho-GDIβ and Rho-GDIγ is more limited to certain cell types. Rho-GDIs are inhibitors of Rho GTPases. Rho GTPases cycle between an active, membrane-bound GTP-bound form and an inactive, cytosolic GDP-bound form. Rho-GDIs bind to and stabilize the GDP-bound form, thereby keeping the Rho protein inactive. Rho proteins, such as RhoA, Rac1, and CDC42, are important regulators of cellular migration and adhesion (4, 5). They regulate the dynamics of actin, allowing stress fibers, lamellipodia, and filopodia to be formed. They are also involved in epithelial-mesenchymal transition during cancer progression (6) and overexpressed in breast cancer (7).
Rho-GDIβ is primarily expressed in hematopoietic cells (8) but is also found in epithelial cancer cells (9–11). In bladder and lung cancer, Rho-GDIβ suppresses invasion and metastasis (9, 10), and in bladder cancer, Rho-GDIβ is a positive prognostic factor (12). If truncated, Rho-GDIβ can promote metastasis of colon cancer (13). Rho-GDIβ is also part of a 76-gene signature that predicts distant metastasis in lymph node–negative breast cancer (14).
We have identified Rho-GDIβ as a target gene of the transcription factor Ets1 and its regulator PKCα in breast cancer cells (15). Like Rho-GDIβ, Ets1 is primarily expressed in hematopoietic cells but is also found in epithelial cancer cells (16–18), where it is associated with increased expression of matrix metalloproteinases and invasive behavior. Like Ets1, PKCα is able to promote tumor progression (19, 20). Rho-GDIα and Rho-GDIβ are substrates of the protein kinase Src (21), which plays a key role in the progression of breast cancer cells (22). The observation that two tumor promoter proteins regulate Rho-GDIβ prompted us to analyze the function and the prognostic value of Rho-GDIβ protein in breast cancer. We identified tumor promoter protein cyclooxygenase-2 (Cox-2) as a new target of Rho-GDIβ, suggesting that Rho-GDIβ may stimulate tumor progression.
Materials and Methods
Cell lines, small interfering RNA, and transient transfection. MDA-MB-231, MCF-7, SKBR-3 breast cancer, and Jurkat T cells were maintained in RPMI 1640 (Invitrogen) supplemented with 10% FCS (Biochrom) in the absence of antibiotics. For RNA interference studies, cells were transfected with small interfering RNA (siRNA) by electroporation as described (23), grown for 3 days, and lysed for RNA or protein analysis. The siRNAs used were purchased from MWG Biotech and are listed in Supplementary Table S1.
Quantitative reverse transcription-PCR. Preparation of RNA, cDNA synthesis, and PCR were carried out essentially as described (23). The relative RNA level of each gene of interest was calculated relative to RNA levels of the housekeeping genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or HPRT. The primers were purchased from MWG Biotech and are listed in Supplementary Table S2.
Microarrays. In three independent transfection experiments, MDA-MB-231 cells were transfected with siRβ or siLuc (control), RNA was isolated, and the six RNAs were analyzed by using Affymetrix probe type HG-U133A arrays. The analyses were carried out as described (23). Normalization of the raw data was done by using the Affymetrix MAS5.0 software, which uses the Tukey's biweight method to estimate the amount of variation in the data. For each replicate, the signal log ratio between sample (siRβ) and control (siLuc) was estimated. The signal log ratio (log to the basis 2) was converted into fold induction (siRβ/siLuc), and the average fold induction ± SD was calculated from the values of the three replicates.
Protein extraction and Western blot analyses. Cytosolic and nuclear protein extracts were prepared as described (23). To extract proteins from tumor tissue, 100 to 300 mg of tissue were pulverized in a Microdismembrator U (Sartorius), and the pulverized tissue was rehydrated in 200 μL TBS containing 1% Triton X-100 and rotated overnight at 4°C. The homogenate was centrifuged at 100,000 × g for 1 h at 4°C, and the supernatant was collected for further analysis. Protein amounts were measured by using the bicinchoninic acid protein assay kit (Pierce).
Western blot analysis was done as described previously (24). The Rho-GDIβ protein was detected by incubating the protein blot with polyclonal rabbit anti-D4-GDI antibody (BD PharMingen) at a dilution of 1:5,000. For the detection of the Cox-2 protein, a monoclonal mouse anti-Cox-2 antibody from Cayman was used at a dilution of 1:200. Antibodies recognizing nuclear factor of activated T cells 1 (NFAT-1), Ets1, or extracellular signal-regulated kinase 1/2 (ERK1/2) were from BD Transduction Laboratories (1:2,500), Santa Cruz Biotechnology, Inc. (C20, 1:2,000), or Cell Signaling (1:1,000), respectively. GAPDH-specific (1:5,000; Ambion) and ERK1/2-specific antibodies were used to check for equal protein loading (25). As GAPDH is also abundant in the nucleus (26), we used the GAPDH-specific antibody also to control protein loading of nuclear extracts. Secondary antibodies conjugated with horseradish peroxidase specific for mouse or rabbit primary antibodies were purchased from Cell Signaling or GE-Amersham. Peroxidase activity was visualized by chemiluminescence using enhanced chemiluminescence (ECL) Plus (GE-Amersham) followed by exposure to Hyperfilm ECL (GE-Amersham).
Chromatin immunoprecipitation assay. The chromatin immunoprecipitation (ChIP) assay was done by using a ChIP kit (Lake Placid Biologicals) following the instructions of the manufacturer. After fixation of MDA-MB-231 cells with formaldehyde, chromatin was sheared by 10 cycles of sonication (10 s each) by using a Branson Sonifier 250 (duty cycle, 60%; output, 5). Between sonications, cell extracts were kept on ice for 1 min. Following immunoprecipitation by using 10 μL of anti-Ets1 antibody (C20, Santa Cruz Biotechnology) or no antibody (control), DNA was eluted, ethanol precipitated, and dissolved in 15 μL DNase-free water. Specific sequences between −121 and −197 or −775 and −878 (relative to the translational start site) of the Rho-GDIβ upstream gene region were amplified by conventional PCR by mixing 0.5 μL of the DNA solution with 3.5 μL water, 2× GoTaq (Promega), and 1 μL each of a forward and the corresponding reverse primer (10 pmol/μL). The sequences of the primers used are shown in Supplementary Table S2. Conservative PCR was carried out in a Biometra TGradient PCR machine.
Cox-2 promoter assay. MDA-MB-231 cells were first electroporated with siRNA as described above. After 3 days of incubation in 60-mm dishes, the medium was removed and cells were transfected with the p274-Cox-2 promoter firefly luciferase construct (27) by using transfectin (Bio-Rad). Cox-2 promoter DNA (5 μg) and 15 μL transfectin were separately mixed with 500 μL RPMI 1640 each, and the solutions were combined and incubated at room temperature for 20 min before this mixture was added together with 2 mL of fresh medium/serum to the cells. Following incubation overnight, cells were lysed and analyzed for luciferase activity by using a Luciferase Assay System kit (Promega) and a Sirius luminometer (Berthold Detection Systems).
Immunocytochemistry/immunohistochemistry. Cells transfected with siRNA were grown on slides (SuperFrostPlus, Menzel) for 3 days, washed with PBS, dried, and fixed in 4% buffered formaldehyde for 10 min. After washing once with PBS, slides were air dried and rehydrated in PBS containing hydrogen peroxide (10:1, v/v). After incubation of cells in this solution for 30 min, they were treated with citrate buffer (29.4 g trisodium citrate dihydrate/L, pH 6.0) at 95°C for 45 min and washed briefly with PBS. After blocking with a blocking solution (Zytomed) for 10 min, slides were incubated with the anti-D4-GDI antibody (diluted 1:500) at 37°C for 1 h. Detection of the primary antibody was achieved by using a biotinylated secondary antibody/streptavidin horse peroxidase conjugate-based assay (Zytomed) and following the manufacturer's instructions. The antibody complexes were visualized by using an AEC substrate kit (Zytomed). Reaction was stopped by rinsing the slides with water. Nuclei were stained by hematoxylin.
Immunohistochemical staining of paraffin sections was carried out as follows. Sections were incubated at 60°C for 1 h and washed twice in xylene for 10 min, once in 100% ethanol for 5 min, twice in 96% ethanol for 5 min, and once in 80% and 70% ethanol 5 min each. After washing the slides in water for 15 min, slides were treated with PBS containing hydrogen peroxide and then heat treated in citrate buffer, blocked, and incubated with antibodies as described above. The strength of the immunoreaction was evaluated by determining the immunoreactive score (IRS). The IRS was calculated by multiplying staining intensity (0 = no staining, 1 = weak, 2 = moderate, 3 = strong) by the percentage of stained tumor cells (0 = no cells stained, 1 = ≤10% of cells stained, 2 = 11–50% of cells stained, 3 = 51–80% of cells stained, 4 >81% of cells stained).
Migration assays. Cell migration was studied by seeding transfected cells (2 × 105) on a tissue culture insert (ThinCerts; pore size, 8 μm; Greiner) inserted into a well of a six-well plate. After 48 h, the number of cells on the underside of the insert was determined by microscopic counting (15 fields per insert).
Breast cancer biopsies. Two hundred sixty-three tumor samples from patients with unilateral operable breast cancer who underwent resection of their primary tumor between 1987 and 1997 were selected by availability of frozen tissue in the tumor bank in the Department of Chemical Endocrinology, Radboud University Nijmegen Medical Centre (Nijmegen, the Netherlands). Coded tumor tissues were used in accordance with the Code of Conduct of the Federation of Medical Scientific Societies in the Netherlands (“Code for Proper Secondary Use of Human Tissue in the Netherlands”),5
Statistical methods. Two-sided Pearson correlation coefficient was used to analyze the relationship between the relative expression of Rho-GDIβ and the relative expression of other genes in breast cancer samples. The t test was used to compare Cox-2 promoter activities in Rho-GDIβ–deficient cells versus control cells. The Mann-Whitney U test was applied for comparing age, menopausal status, and estrogen and progesterone receptor status with Rho-GDIβ expression. The relationship between nodal status, tumor type, and histologic grade and Rho-GDIβ RNA levels was analyzed by the Kruskal-Wallis test and that between tumor size and Rho-GDIβ RNA levels by the Spearman correlation test. In immunohistochemical analysis, the χ2 test was used to test for association between the clinicopathologic factors and Rho-GDIβ immunoreactivity (IRS of 0 was coded 0; IRS > 0 was coded 1). Survival analysis was carried out with the Kaplan-Meier method. For comparison of survival curves, the log-rank test was used. All statistical analyses were done with Statistical Package for the Social Sciences 12.0 software (SPSS, Inc.). P < 0.05 was considered significant.
Ets1 regulates Rho-GDIβ expression in breast cancer cells. We have previously shown that Rho-GDIβ RNA levels are reduced when MDA-MB-231 cells were transfected with an Ets1-specific siRNA (siE1) or by a PKCα-specific siRNA (siPα; ref. 15). Here, we show that the same siRNAs also decrease the protein level of Rho-GDIβ as determined by Western blot analysis (Fig. 1A). Furthermore, a second Ets1-specific siRNA (siE1#2), which was as effective as siE1 in suppressing Ets1 expression, also decreased Rho-GDIβ RNA levels, as measured by Q-RT-PCR (Fig. 1B). In another approach, we analyzed samples that were previously used to show that calphostin C, an inhibitor of PKCα, abrogates Ets1 expression for Rho-GDIβ protein expression (24). We found that Rho-GDIβ was eliminated along with Ets1 (data not shown). To analyze whether Ets1 directly regulates Rho-GDIβ expression, we did ChIP assays. Searching for potential Ets binding sites in the first 1,000 bp upstream of the translational start site of the human Rho-GDIβ gene, we found two clusters of GGAA(T) motives: one between −87 and −165 and a second one between −716 and −930 relative to the translational start site (Supplementary Fig. S1). As found with the University of California at Santa Cruz Genome Browser, these sequences overlap with genomic regions that are conserved among the human, rabbit, and dog Rho-GDIβ gene. The conserved sequences include the GGAA(T) motives at positions −165, −159, −135, −764, −822, and −849. ChIP assays were done with chromatin from MDA-MB-231 cells in the presence or absence of an Ets1-specific antibody, which has already been successfully used for ChIP analyses (28, 29), followed by PCR amplification of Rho-GDIβ–specific sequences between −121 and −197 and −775 and −878. We observed that the presence of the anti-Ets1 antibody slightly increased the amount of precipitated Rho-GDIβ–specific DNA fragments that contained sequences between −121 and −197 (Fig. 1C). A much stronger effect of the anti-Ets1 antibody on the amount of precipitated Rho-GDIβ–specific DNA was found for fragments harboring the sequence between −775 and −878. This suggests that Ets1 binds to one or more of the GGAA/T motives that are located between −716 and −930 relative to the translational start site of the Rho-GDIβ gene. Collectively, these data suggest that Ets1 regulates Rho-GDIβ expression in MDA-MB-231 cells.
Ets1 expression correlates with Rho-GDIβ expression in breast cancer cell lines and primary breast cancer. We next compared the RNA levels of Ets1 and Rho-GDIβ among breast cancer cell lines (MDA-MB-231, SKBR-3, and MCF-7 cells). We found that the Rho-GDIβ level was much lower in cell lines that produce less Ets1 RNA (Fig. 2A). We extended this study to primary breast carcinomas. Again, Rho-GDIβ expression correlated well with that of Ets1 (r = 0.81; P < 0.001; Fig. 2B). In contrast, no correlation was observed between Rho-GDIβ RNA level and the level of either Ets2 or Esx (Fig. 2C and D), two other Ets genes commonly expressed in breast cancer cells (17, 18). These results suggest that Ets1 may be of general importance for the regulation of Rho-GDIβ expression in breast cancer cells.
Rho-GDIβ inhibits migration of breast cancer cells. To analyze the function of Rho-GDIβ in breast cancer cells, we used a Rho-GDIβ–specific siRNA, siRβ, which strongly down-regulated the Rho-GDIβ mRNA and protein level without affecting the expression of Rho-GDIα or Rho-GDIγ (Fig. 3A and B). We also examined the effect of siRβ on Rho-GDIβ expression by immunocytochemistry. In the presence of siRβ, overall immunoreactivity to the Rho-GDIβ–specific antibody dropped and the number of strongly stained cells was reduced from approximately 15% to 8% (Fig. 3C).
As a natural inhibitor of Rho GTPases, Rho-GDIβ may inhibit migration of breast cancer cells. We compared the ability of siRβ-transfected and control siRNA (siLuc)-treated MDA-MB-231 cells to migrate through an 8-μm filter in a Boyden chamber. We found that siRβ increased migration of MDA-MB-231 cells by ∼3-fold (Fig. 3D). Interestingly, a similar effect was observed when the Ets1-specific siRNA siE1 was used. These data suggest that Rho-GDIβ inhibits migration of MDA-MB-231 cells.
Rho-GDIβ regulates Cox-2 expression. To explore the possibility that Rho-GDIβ may affect gene expression in breast cancer, we transfected MDA-MB-231 cells with either siLuc or siRβ and analyzed changes in gene expression by cDNA microarray analyses by using Affymetrix HG-U233A gene chips. The data resulting from three independent transfection experiments are deposited in Gene Expression Omnibus as series GSE8087. Genes whose expression in all three experiments was found to be different in siRβ-treated versus siLuc-treated cells by at least 2-fold are listed in Supplementary Table S3 and Fig. 4A. In addition to Rho-GDIβ, these group of genes included GTP-binding protein 9 (GTPBP9), Cox-2, and collagen type IV α2 (COL4A2), all of which showed reduced expression in the presence of siRβ. The expression of two genes, FLJ21424 and AKAP350 (A kinase anchor protein 9), was increased in the presence of siRβ. By using Q-RT-PCRs, we could verify reduced RNA levels for GTPBP9, GTPBP9 isotype 1, Cox-2, and COL4A2 in the siRβ samples (Fig. 4A) but failed to show increased levels for FLJ21424 (data not shown). The failure to confirm the increased expression of FLJ21424 by Q-RT-PCR might be the consequence of the existence of a specific splicing variant, which might have been detected by the microarray analysis but might have been missed by Q-RT-PCR. To check for out-of-target effects of siRβ, we rerun the transfection experiments with a second Rho-GDIβ–specific siRNA (siRβ2) and analyzed the RNAs by Q-RT-PCR. Although siRβ2 failed to interfere with the expression of GTPBP9, GTPBP isotype 1, or COL4A2, it affected Rho-GDIβ and Cox-2 expression by a similar extent as in siRβ (Fig. 4A). By Western blot analysis, we could also show that knockdown of Rho-GDIβ reduces Cox-2 protein levels (Fig. 4A). We next compared Rho-GDIβ and Cox-2 levels in primary breast cancer by analyzing 45 breast cancer samples by Q-RT-PCR. It was found that Cox-2 levels significantly correlate with those of Rho-GDIβ (Supplementary Fig. S2). Collectively, these data suggest that Cox-2 is a target gene of Rho-GDIβ in MDA-MB-231 breast cancer cells.
To analyze the mechanism(s) underlying Rho-GDIβ–depending Cox-2 regulation, we did promoter assays with a p274-Cox-2 luciferase promoter construct containing the sequences between −170 and +104 (relative to the transcriptional start site) of the human Cox-2 gene (27). The activities of this promoter in siRβ- and siLuc-treated MDA-MB-231 cells were compared. In the presence of siRβ, Cox-2 promoter activity was significantly lower (Fig. 4B), suggesting that Rho-GDIβ at least partly regulates Cox-2 on the transcriptional level. No effect of siRβ was observed on a control promoter (data not shown). One of the major regulators of the −170/+104 Cox-2 promoter fragment in breast cancer cells is the transcription factor NFAT-1 (30). NFAT activity depends on dephosphorylation (31). Dephosphorylation results in masking of a nuclear export signaling sequence within the NFAT-1 protein, allowing this protein to stay in the nucleus. We investigated the presence of NFAT-1 protein in nuclear and cytosolic extracts from siRβ- and siLuc-treated MDA-MB-231 cells. Three different NFAT-1–specific bands, I, II, and III, representing different phosphorylated species (31, 32), were observed in nuclear as well as cytosolic extracts from siLuc-treated control cells (Fig. 4C). Of these bands, bands II and III, which most likely represent less phosphorylated forms, disappeared when nuclear extracts, but not cytosolic extracts, from siRβ-treated cells were analyzed. These data suggest that less phosphorylated forms of NFAT-1 are excluded from the nucleus when Rho-GDIβ is suppressed. Reprobing with anti-Rho-GDIβ antibody confirmed that Rho-GDIβ was strongly suppressed in siRβ-treated cells. To show that the nuclear extract is enriched with nuclear proteins, we reprobed the blot with an antibody specific to Ets1. This nuclear protein was more abundant in the nuclear extract than in the cytosolic extract. In contrast, cytoplasmic Rho-GDIβ was more prominent in the cytosolic extract.
Interestingly, Rho-GDIβ is able to cooperate with another hematopoietic protein, the Rho GTPase guanine nucleotide exchange factor Vav-1, which leads to activation of NFAT in Jurkat T cells (33). In the same cells, NFAT is a major regulator of the −170/+104 Cox-2 promoter fragment (27). To analyze the importance of Vav-1 for Cox-2 expression in MDA-MB-231 cells, we suppressed Vav-1 expression in these breast cancer cells by Vav-1–specific siRNAs siV#1 and siV#2. Both siRNAs were equally effective in reducing Vav-1 expression and also able to decrease Cox-2 expression (Fig. 4D). Furthermore, like siRβ, siV#1 prevented the accumulation of NFAT protein species II and III in the nucleus (Fig. 4C). Importantly, siV#1 did not change Rho-GDIβ protein levels, indicating that the effect of siV#1 on NFAT and Cox-2 was not the consequence of reduced Rho-GDIβ expression levels. We also found that, in primary breast cancer, Vav-1 expression significantly correlates with Rho-GDIβ expression (Supplementary Fig. S2). These data warrant a model by which Rho-GDIβ and Vav-1 cooperate in breast cancer cells to induce nuclear translocation of specific NFAT-1 forms to activate Cox-2 gene transcription.
Rho-GDIβ mRNA levels do not correlate with prognosis. To explore the importance of tumoral Rho-GDIβ expression for the survival of breast cancer patients, we analyzed a cohort of 259 patients with invasive breast cancer for Rho-GDIβ RNA expression in fresh-frozen breast cancer samples and compared these RNA levels with clinicopathologic data and with the patients' survival data. Although we found that Rho-GDIβ RNA levels were significantly lower in tumors that grew larger (Supplementary Table S4), Kaplan-Meier survival curves showed that the likelihood of a patient to develop a relapse was the same, irrespective of whether the breast cancer produced Rho-GDIβ at higher or lower levels (Fig. 5).
Rho-GDIβ protein expression does not have an effect on patient's survival. We next analyzed Rho-GDIβ protein levels in primary breast cancers. We prepared protein extracts from 87 invasive breast carcinomas and analyzed them by Western blot analysis for the Rho-GDIβ protein. Forty-three percent of the carcinoma samples were found to express Rho-GDIβ protein at the same or at a higher level as MDA-MB-231 cells (Fig. 6A). Rho-GDIβ protein expression in primary breast cancer was also studied by immunohistochemistry. Rho-GDIβ–specific staining was observed in the cytoplasm of tumor as well as of stromal cells. An analysis of a total of 117 primary breast cancer samples revealed that in 74 (63%) no Rho-GDIβ–specific immunoreactivity (IRS = 0) in the tumor cells could be observed. Of the remaining 37% breast cancer samples, 13 (11%) expressed Rho-GDIβ at low levels (IRS 2 or 3), 11 (9.4%) at medium levels (IRS 4 or 6), and 19 (16.2%) at high levels (IRS > 8). For the statistical analysis, tumors showing an IRS of 0 were defined as Rho-GDIβ–negative breast cancers and those with IRS > 0 as Rho-GDIβ–positive cancers. When we compared the Rho-GDIβ protein expression pattern with the classic clinicopathologic data, we could not find that the presence or absence of Rho-GDIβ coincided with a particular status of one of the clinicopathologic factors (Supplementary Table S5). Survival analyses using Kaplan-Meier statistics showed no significant differences in overall survival (P = 0.534) between patients with Rho-GDIβ–positive and Rho-GDIβ–negative tumors (Fig. 6C,, right). In disease-free survival, there was a tendency that patients with Rho-GDIβ–positive tumors may have a better prognosis (Fig. 6C , left), but this tendency was not significant (P = 0.149). It was still not significant (P = 0.069 or 0.057) when the Rho-GDIβ–negative group also included those breast cancer samples that showed Rho-GDIβ IRS values of 2 to 3 or IRS values of 2 to 4, respectively. These data agree with the results obtained by the analysis of the Rho-GDIβ RNA expression showing that Rho-GDIβ has no prognostic value in breast cancer.
Our data suggest that Rho-GDIβ protein is expressed in ∼40% of all breast cancers. As an inhibitor of Rho GTPases, Rho-GDIβ may simply act as an antimigratory protein as shown with T24 bladder cancer cells (9). In fact, we found that Rho-GDIβ inhibits migration of MDA-MB-231 cells. Exerting such function, Rho-GDIβ may be a tumor-suppressing protein. However, the clinical outcome of breast cancer patients is not improved when breast cancer cells express Rho-GDIβ. Interestingly, Rho-GDIβ is coexpressed with the transcription factor Ets1, a tumor-promoting protein (16). RNA interference or other means to suppress Ets1 expression as well as ChIP analyses revealed that this coexpression is not a coincidence but the consequence of a regulatory action of Ets1 on Rho-GDIβ. Ets1-specific and Rho-GDIβ–specific siRNA had also similar positive effects on cellular migration. This effect of Ets1-specific siRNA was unexpected as Ets1 has been previously reported to promote migration (34, 35). The discrepancy may be explained by the use of different inhibitors, siRNA versus Ets1-DNA binding domain. The Ets1-DNA binding domain, which acts as a transdominant-negative form of Ets1, may also affect the activity of other Ets proteins (36). Alternatively, the effect of Ets1 on migration may be cell type specific. Depending on which other migration-relevant genes Ets1 may target in a given cell, migration may be increased or decreased by Ets1.
The regulation of Rho-GDIβ by the oncogene Ets1 may imply that Rho-GDIβ may also have tumor-promoting functions. In support of this notion, Rho-GDIβ has been reported to regulate the expression of integrin β1, an integrin that plays a role in breast cancer progression (37). In our study, we identified Cox-2 as a target of Rho-GDIβ. There is strong evidence that Cox-2 promotes tumor progression. Elevated Cox-2 protein levels are linked to decreased survival of breast cancer patients (38). In addition, Cox-2 stimulates metastasis of breast cancer to bone and lung (39, 40) and enhances invasion of breast cancer cells (30, 41). Our data suggest that Rho-GDIβ regulates the expression of Cox-2 in breast cancer cells at least partly on the transcriptional level. We present evidence that Rho-GDIβ targets NFAT-1. As NFAT is a major regulator of Cox-2 gene expression in breast cancer and T-leukemia cells (27, 30), NFAT is likely to mediate the effect of Rho-GDIβ on Cox-2 transcription. We could further show that another Rho GTPase-regulating protein, Vav-1, regulates Cox-2 gene expression and also targets NFAT-1 in MDA-MB-231 cells. Vav-1 has been reported to bind to NFAT (42) and to cooperate with Rho-GDIβ to activate NFAT in T cells (33). Like Cox-2, Vav-1 and NFAT-1 have been associated with cancer progression. Proto-oncogene Vav-1 has been shown to regulate proliferation of pancreas cancer cells and to contribute to unfavorable prognosis (43), whereas NFAT has been shown to increase the invasiveness of breast and colon cancer cells (32). Interestingly, like Rho-GDIβ and Ets1, Vav-1 is primarily expressed in hematopoietic cells (44). Hence, it is tempting to hypothesize that breast cancer cells have adopted a mechanism allowing the expression and coordinate interaction of these hematopoietic factors to promote tumor progression. Besides Rho-GDIβ and Vav-1, other proteins, including α6β4 integrin and AKT, have been found to regulate NFAT-1 activity in breast cancer cells (32, 45). In addition to NFAT-1, transcription factors, such nuclear factor-κB, CAAT/enhancer binding protein δ, and transforming growth factor-β–regulated Smads, have also been reported to activate Cox-2 transcription (39, 46), suggesting a complex interaction network that controls Cox-2 activity.
In summary, the stimulatory effect of Rho-GDIβ on the Cox-2 oncogene on the one hand and its antimigratory activity on the other hand may even out and explain why Rho-GDIβ does not have an effect on the clinical outcome of breast cancer patients. However, in individual tumors, the one or the other function of Rho-GDIβ may prevail and Rho-GDIβ expression may be either of advantage or disadvantage for the breast cancer patient.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Grant support: Bundesministerium für Bildung und Forschung grant NBL3 FKZ13/11.
We thank Katayoun Sheikheleslamy, Gareth Palidwor, and Pearl Campbell for doing the DNA microarray experiments and computational analyses and Miguel A. Íñiguez for providing us with the p274-Cox-2 promoter construct.