NFAT Induces Breast Cancer Cell Invasion by Promoting the Induction of Cyclooxygenase-2*

The NFAT (nuclear factor of activated T cells) family of transcription factors plays a fundamental role in the transcriptional regulation of the immune response. However, NFATs are ubiquitously expressed, and recent evidence points to their important functions in human epithelial cells and carcinomas. Specifically, NFAT has been shown to be active in human breast and colon carcinoma cells and to promote their invasion through Matrigel. The mechanisms by which NFAT promotes invasion have not been defined. To identify NFAT target genes that induce carcinoma invasion, we have established stable breast cancer cell lines that inducibly express transcriptionally active NFAT. Gene expression profiling by cDNA microarray of cells induced to express NFAT revealed up-regulation of cyclooxygenase-2 (COX-2). Increased NFAT expression and activity induced COX-2 expression as well as prostaglandin E2 synthesis. This induction was more prominent when NFAT was activated by phorbol 12-myristate 13-acetate and calcium ionophore ionomycin and was blocked by the NFAT antagonist cyclosporin A. Breast cancer cells with elevated COX-2 expression showed increased invasion through Matrigel, and this was reduced in cells treated with COX-2 inhibitors. Conversely, loss of NFAT1 protein expression using small interfering RNA led to a reduction in COX-2 transcription and reduced invasion. Similarly, Matrigel invasion was reduced in cells in which COX-2 expression was reduced using specific siRNA. These findings demonstrate that NFAT promotes breast cancer cell invasion through the induction of COX-2 and the synthesis of prostaglandins.

NFAT 2 was first identified in T cells as a rapidly inducible transcription factor that binds to the distal antigen receptor response element of the human interleukin-2 (IL-2) promoter and is involved in the regulation of inducible genes such as cytokines and cell surface receptors in immune cells (1)(2)(3). However, subsequent studies have revealed that NFAT is also expressed and is active in many other cell types and tissues, and it regulates the expression of genes related to cell cycle progression, angiogenesis, tumorigenesis, and cell differentiation (4 -8). The mechanisms by which NFAT controls these cellular processes remain largely undefined.
The NFAT family of transcription factors comprises four classical members: NFAT1, NFAT2, NFAT3, and NFAT4 (2). All are calciumresponsive and are regulated by the calcium/calcineurin signaling path-way (9,10). A recently identified member, NFAT5, is distinct from NFAT1-4 as it is calcium-insensitive and is regulated by osmotic stress and integrins (11). In resting cells, NFATs are phosphorylated at a cluster of serine residues located in the regulatory domain, effectively masking a nuclear localization signal, thereby retaining NFAT in an inactive conformation in the cytoplasm (2,12). Upon stimulation with agonists that elicit an increase in intracellular calcium, NFATs are dephosphorylated by the phosphatase calcineurin and translocate to the nucleus. Here they are transcriptionally active by binding to the promoter regions of target genes (13,14). When the cells return to their unstimulated state, NFAT becomes rephosphorylated and is exported out of the nucleus (2). Classical NFATs typically interact with other transcription factors such as AP-1 (15,16) and GATA4 (17) to activate transcription. Adjacent NFAT and AP-1 binding sites are present in the promoter region of inducible genes including IL-2 and cyclooxygenase-2 (COX-2) (18,19).
Cyclooxygenases convert arachidonic acid produced from membrane phospholipids by phospholipase A 2 to prostaglandin H 2 (PGH 2 ). Prostaglandin endoperoxide is then converted to biologically active prostaglandins (PGD 2 , PGE 2 , PGF 2␣ ), prostacyclin (PGI 2 ), and thromboxanes (TxA 2 ) by tissue-specific synthases or reductases (20). Following their synthesis, these prostanoids are secreted and bind to G protein-coupled membrane receptors in target cells in an autocrine or paracrine fashion, thereby triggering downstream signaling events (21). Prostaglandins are important regulators of numerous cellular processes including cell proliferation, inflammation, and angiogenesis (22).
Cyclooxygenase-2 catalyzes the formation of prostaglandin E 2 (PGE 2 ). COX-2 is distinct from the other isoform, COX-1, which is considered a housekeeping enzyme and is expressed constitutively in most tissues. Conversely, COX-2 is normally expressed at very low or undetectable levels and is rapidly induced at sites of inflammation and proliferation in response to stimuli such as growth factors and tumor promoters (23). COX-2 expression and PGE 2 levels are elevated in a variety of human cancers (24 -26) and are associated with increased angiogenesis, tumor invasion, and resistance to apoptosis (27)(28)(29)(30). Similarly, overexpression of COX-2 has been shown to induce cancer formation in transgenic mice (31,32). Several epidemiological studies have indicated that continuous users of aspirin and other nonsteroidal antiinflammatory drugs, which inhibit COX activity, have reduced risk or mortality from cancer (33)(34)(35). Moreover, COX-2-specific inhibitors have been shown to suppress tumor growth in animal models of human cancer (36,37).
A link between NFAT activity and COX-2 is evident from previous studies. NFAT has been reported to regulate COX-2 expression in human T lymphocytes (19). Putative NFAT recognition sequences are present in the human COX-2 proximal promoter, and deletion analysis has shown that they are important for its transcriptional activation (19). A recent study also demonstrated that these sites are essential for the induction of COX-2 by NFAT in colon carcinoma cells (38). The con-sequence of NFAT-mediated COX-2 induction for cancer cell phenotypes has not been established.
The significance of NFAT for cancer development or progression to metastasis has to date not been investigated. Our previous studies on NFAT revealed that it plays an essential role in promoting migration and invasion of breast and colon carcinoma cells (39). To identify the downstream NFAT target genes that are important for invasion, we have analyzed the gene expression profile of breast cancer cells that express NFAT1. We detected a significant up-regulation of COX-2 in these cells. We show that activation of NFAT increases COX-2 expression and PGE 2 synthesis. Inactivation of NFAT by cyclosporin A (CsA) or siRNA significantly diminished COX-2 expression. Expression of COX-2 promoted invasion through Matrigel, and this was reduced by the COX-2 inhibitor NS-398 or with siRNA. Together, these results provide the first direct evidence that NFAT promotes breast cancer cell invasion through the induction of COX-2.

EXPERIMENTAL PROCEDURES
Cell Lines and Reagents-The human cell lines MDA-MB-435 and MDA-MB-231 were obtained from American Type Culture Collection (Manassas, VA) and maintained in Dulbecco's modified Eagle's medium with 1 g/ml glucose, L-glutamine, and sodium pyruvate (Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum (Nova-Tech, Grand Island, NE). The estrogen-independent breast cancer cell line SUM-159-PT has been described (40) and was maintained in Ham's F-12 medium with L-glutamine (Cambrex, Walkersville, MD) supplemented with 5% fetal bovine serum, 1 g/ml hydrocortisone, and 5 g/ml insulin (Sigma). SUM.N1-16 and 435.N1-23 cells with inducible NFAT1 expression were derived from SUM-159-PT and MDA-MB-435 cells, respectively, and generated using the tetracycline-regulated expression system from Invitrogen. HA-tagged NFAT1 was subcloned into pcDNA4/TO/Myc-His and was co-transfected into SUM-159-PT cells with pcDNA6/TR. Positive clones were selected and maintained in medium with 20 g/ml blasticidin and 0.1 mg/ml zeocin (InvivoGen, San Diego, CA). 435.N1-23 cells were prepared by transfecting HA-NFAT1 in pcDNA4/TO/Myc-His into MDA-MB-435 cells that have constitutive expression of tetracycline repressors from pcDNA6/TR. Positive clones were selected and maintained in culture medium containing 10 g/ml blasticidin and 0.4 mg/ml zeocin. Expression of NFAT1 was induced with 1 g/ml doxycycline (Clontech) for 16 -24 h at 37°C. SUM.N1 siRNA-4 and SUM.N1 siRNA-17 cells that express NFAT1 siRNA were prepared by transfecting SUM-159-PT cells with NFAT1 siRNA (see below). Stable clones were selected with 20 g/ml blasticidin and 0.5 mg/ml Geneticin (Invitrogen). Cells were treated with PMA (100 nM; Alexis Biochemicals, San Diego, CA) and ionomycin (100 nM; CalBiochem) for 16 -24 h. Cyclosporin A (CalBiochem) was used at 10 M and NS-398 (Cayman Chemical Co., Ann Arbor, MI) at 50 -100 M. Anti-HA antibody was purified in-house from the 12CA5 hybridoma. Monoclonal anti-COX-2 was from Cayman Chemical Co., and anti-NFAT1 antibody was from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-␤-actin antibody was from Sigma.
Immunoblotting-Total cell lysates were prepared in ice-cold radioimmune precipitation assay lysis buffer (50 mM Tris HCl, pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.2 mM phenylmethylsulfonyl fluoride, and 2 mM sodium orthovanadate) supplemented with protease inhibitor mixture from Sigma. The lysates were clarified by centrifugation at 12,000 ϫ g for 10 min, and protein concentration was determined by the Bradford assay (Bio-Rad). Protein lysates were denatured, resolved by SDS-polyacrylamide gel electrophoresis, and transferred onto nitrocellulose membranes. The protein blots were blocked with 5% nonfat milk and incubated with the appropriate primary antibodies for 2 h at room temperature or overnight at 4°C. Signals were developed by using the SuperSignal West Pico chemiluminescent substrate from Pierce.
Real-time RT-PCR-Total RNA samples were extracted from the cultured cells using TRIzol (Invitrogen) and were reverse transcribed into cDNA using Taqman reverse transcriptase and oligo(dT) 16 (Roche Applied Science) according to the manufacturer's instructions. Quantitative real-time PCR was performed using the SYBR Green PCR master mix in an ABI Prism 7700 sequence detector (both from Applied Biosystems, Foster City, CA). The reactions were carried out with a polymerase-activating step of 95°C for 10 min followed by 40 cycles of a two-step cycling program (95°C for 15 s; 60°C for 1 min) for NFAT1 detection or a three-step cycling program (95°C for 15 s; 57°C for 30 s, 72°C for 45 s) for analyzing COX-2 transcription. For murine NFAT1, the primers were 5Ј-CGG AGT CCA AGG TTG TGT TCA-3Ј (sense) and 5Ј-TGT GGC TGA CTT CGT TTC CTC-3Ј (antisense). For human NFAT1, the primers were 5Ј-TGC ATC TAA CCC CAT CGA GTG-3Ј (sense) and 5Ј-TGA GGA TCA TTT GCT GGC C-3Ј (antisense). For glyceraldehyde-3-phosphate dehydrogenase, the primers were 5Ј-GCA AAT TCC ATG GCA CCG T-3Ј (sense) and 5Ј-TCG CCC CAC TTG ATT TTG G-3Ј (antisense). For COX-2, the primers were 5Ј-CAA AAG CTG GGA AGC CTT CTC TAA CC-3Ј (sense) and 5Ј-GCC CAG CCC GTT GGT GAA AG-3Ј (antisense). The PCR products were analyzed on 1 or 1.5% agarose gels to ensure the specificity of amplification.
Transfection and Luciferase Assays-All cell lines were transfected using the TransIT-LT1 transfection reagent from Mirus Bio Corporation (Madison, WI). Luciferase reporter constructs were transiently cotransfected into the cells with pCS2-(n)-␤-gal. Cells were then left untreated or treated overnight with PMA/ionomycin with or without cyclosporin A. Total cell lysates were prepared 24 h after transfection. Luciferase and ␤-galactosidase activities were determined using Promega's luciferase assay system and Galacton-Plus from Tropix (Bedford, MA), respectively, in a MicroLumat LB 96 P luminometer (Berthold Analytical Instruments, Nashua, NH).
Matrigel Invasion Assays-Invasion assays were performed essentially as described previously (39). Briefly, Transwell chambers with 8-m pore filters (Corning, Acton, MA) were coated with 1-5 g of Matrigel (BD Biosciences). Cells were harvested by trypsinization, resuspended in serum-free Dulbecco's modified Eagle's medium con-taining 0.1% bovine serum albumin, and added (1.0 -1.5 ϫ 10 5 cells/ assay) in triplicates to Transwell chambers. The cells were allowed to invade the Matrigel-coated filters at 37°C toward NIH 3T3-conditioned medium in the lower compartment. After 4 -8 h, cells that had invaded to the lower surface of the filter were fixed and stained with crystal violet or 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) and counted. PGE 2 Measurements-PGE 2 levels in the culture medium of treated cells were determined using a PGE 2 -monoclonal enzyme immunoassay kit from Cayman Chemical Co. according to the manufacturer's protocol. To collect culture medium supernatants for the assay, growth medium of the cells was replaced with serum-free medium, and after overnight incubation at 37°C, cell culture medium was spiked with 10 g/ml arachidonic acid and collected 30 min later. The samples were clarified by centrifugation and analyzed in triplicates. The amount of PGE 2 in the culture medium was determined by referring to the signal intensities obtained from a set of PGE 2 standards.

RESULTS
Our previous study demonstrated that the ␣6␤4 integrin, a tumorassociated antigen, up-regulates NFAT activity in breast carcinoma. We showed that although NFAT5 increases cell migration, NFAT1 promotes both migration and invasion (39). To identify the target genes induced by NFAT that are responsible for promoting migration and invasion, we established clones of MDA-MB-435 and SUM-159-PT cells that inducibly express NFAT1 upon stimulation with tetracycline or the analog doxycycline. As shown in Fig. 1A, NFAT1 expression was induced in doxycycline-treated clones of MDA-MB-435 (435.N1-23) and SUM-159-PT (SUM.N1-16) cells. Increased NFAT1 expression induction was also confirmed at the mRNA level by real-time RT-PCR (Fig. 1B). To investigate whether the induced NFAT1 was transcriptionally active, we transfected an IL-2 luciferase reporter plasmid. As predicted, the induced NFAT1 was functional at driving NFAT-dependent transcription of IL-2 (Fig. 1C). When the stable transfectants were allowed to invade Matrigel in an in vitro invasion assay, doxycyclinetreated cells were significantly more invasive compared with their untreated controls (Fig. 1D), consistent with previous data (39).
Using cDNA prepared from untreated and doxycycline-treated 435.N1-23 and SUM.N1-16 cells, we analyzed the gene expression profile induced subsequent to NFAT1 expression. A 6.06-fold (435.N1-23) and 3.05-fold (SUM.N1-16) induction of COX-2 was observed (data not shown). NFAT has been shown to bind to the COX-2 promoter and regulate its transcription in immune cells (19), and a recent study has reported the regulation of COX-2 by NFAT in human colon carcinoma (38). However, the relevance of NFAT in COX-2 regulation and function in breast cancer cell signaling or responses has not been determined. To address this question, we analyzed the expression of COX-2 in doxycycline-treated cells. COX-2 transcription measured by RT-PCR was increased upon NFAT1 expression induced by doxycycline ( Fig.  2A). At the protein level, untreated SUM.N1-16 cells revealed low to undetectable levels of both NFAT1 and COX-2. After doxycycline treatment, NFAT1 expression significantly increased and was accompanied by a small but reproducible elevation in COX-2 protein (Fig. 2B). Stimulation with the NFAT agonists PMA and ionomycin led to a dramatic increase in COX-2 expression, likely because of the activation of endogenous NFAT. Moreover, combined treatment of doxycycline and PMA/ ionomycin led to an even greater increase in NFAT1 and COX-2 expression (Fig. 2B). Next we determined whether the induction of COX-2 by NFAT1 is mediated through the transcriptional activation of the COX-2 promoter. SUM.N1-16 and MDA-MB-231 cells were transfected with the pIL2-luc or the human COX-2 promoter luciferase reporter construct P2-274. Cells stimulated with PMA/ionomycin showed a basal low level of NFAT activity when compared with untreated control (Fig. 3A, left  panel). In contrast, induction of NFAT1 expression with doxycycline significantly increased the transcription of both IL-2 (Fig. 3A, left panel) and COX-2 (right panel). Maximum induction of IL-2 and COX-2 transcription was revealed when induced NFAT1 was activated by PMA/ ionomycin. Blocking the activation of NFAT with CsA significantly diminished transcription from both reporters (Fig. 3A).
As MDA-MB-231 cells have high levels of endogenous NFAT, we also examined the NFAT-driven IL-2 and COX-2 transcription in these cells. Activation of endogenous NFAT by PMA/ionomycin significantly increased transcriptions from both the IL-2 and COX-2 promoters (Fig.   FIGURE 2. Activation of NFAT1 induces COX-2. 435.N1-23 and SUM.N1-16 cells were induced with Dox (1 g/ml) for 24 h. A, total RNA was analyzed by real-time PCR analysis of COX-2 transcription. B, immunoblotting (IB) of the total lysates prepared from control or Dox-treated SUM.N1-16 cells with or without stimulation with 100 nM PMA and ionomycin (Ion.). Antibodies used were specific to anti-NFAT1, anti-COX-2, or anti-actin. n.s., nonspecific bands immunoreactive against the COX-2 antibody.

3B). Again, NFAT inhibition by CsA reduced the induction of transcriptional activation of both reporters.
To further investigate the regulation of COX-2 by NFAT, we have developed clones of SUM-159-PT cells in which NFAT expression is reduced by siRNA. Real-time RT-PCR analysis demonstrated that two distinct clones expressed significantly less NFAT1 transcripts compared with parental SUM-159-PT cells. This was confirmed by immunoblotting with anti-NFAT1 (Fig. 4A). Reduced NFAT1 expression with siRNA also markedly attenuated transcription from the IL-2 promoter in cells treated with PMA/ionomycin compared with cells transfected with control siRNA (Fig. 4B). Most importantly, reduced expression of NFAT by siRNA also led to a quantitative reduction of COX-2 promoter luciferase activity (Fig. 4C). Finally, to demonstrate that loss of NFAT1 translates into loss of invasion, we measured Matrigel invasion in control SUM-159-PT cells compared with the stable siRNA clones. As predicted, reduced expression of NFAT1 by siRNA resulted in not only a loss of COX-2 expression but also a marked loss of invasion (Fig. 4D).
To extend the above findings, we transiently transfected COX-2 into MDA-MB-435 and SUM-159-PT cells and measured Matrigel invasion. In both cell lines, COX-2 expression increased invasion (Fig. 5A). COX-2 expression was confirmed by immunoblotting. We also used a loss-of-function approach and constructed COX-2 siRNA to silence COX-2 expression. The efficacy of silencing endogenous COX-2 was determined by immunoblotting (Fig. 5B). Next, SUM.N1-16 cells were transfected with either control or COX-2 siRNA, induced with doxycycline to express NFAT1, followed by Matrigel assays. When NFAT1 expression was induced in control transfected cells, as already demonstrated, this led to a reproducible increase in invasion. However, in the presence of COX-2 siRNA, invasion was reduced to control levels (Fig.  5C, left panel). COX-2 siRNA also significantly blunted Matrigel invasion of MDA-MB-231 cells (Fig. 5C, right panel). These results demonstrate directly that at least one mechanism by which NFAT promotes invasion is through the induction of COX-2. PGE 2 is the major product of COX-2. To further define the role of COX-2 induction by NFAT1 in the invasion phenotype, we assayed PGE 2 levels in the culture medium of SUM.N1 cells. In control cells, low levels of PGE 2 were detected, but these levels significantly increased when cells were stimulated by PMA/ionomycin to activate endogenous NFAT (Fig. 6A). Moreover, treatment with NS-398, a potent and specific COX-2 inhibitor, blocked PGE 2 induction by PMA/ionomycin. More importantly, NFAT1 expression also stimulated PGE 2 production, and this was greatly augmented in the presence of PMA/ionomycin (Fig. 6A). Again, this was inhibited by NS-398. We also examined the production of PGE 2 in MDA-MB-231 cells, and found that these cells produce high levels of NS-398-sensitive PGE 2 in response to PMA/ionomycin, likely because of the high endogenous expression of NFAT1 (Fig. 6B).
SUM.N1-16 cells with induced NFAT1 expression are more invasive (see Fig. 1). That invasion is specifically mediated by COX-2 induction and PGE 2 production was revealed by performing Matrigel invasion assays in the presence of increasing amounts of NS-398. As shown in Fig. 6C, invasion stimulated by NFAT1 expression was blocked by the same concentrations of NS-398 that blocked COX-2 activity and reduced PGE 2 levels.

DISCUSSION
Carcinoma invasion is defined as the penetration of tumor cells into adjacent tissues and is one of the hallmarks of malignant tumors. In contrast to benign tumors, invasive tumors have the potential to metastasize because of their access to the vasculature. For this reason, much work has gone into understanding the mechanisms by which epithelial cells acquire this invasive phenotype. Numerous changes at the levels of signaling and gene transcription coordinate the acquisition of motility and invasion. In our previous studies, we showed that the NFAT transcription factor represents one such mechanism, and it is both necessary and sufficient to promote carcinoma invasion in both breast and colon cancer cells (39). However, what has remained elusive is the nature of the genes induced by this transcription factor, which are directly responsible for promoting cancer cell invasion. To address this problem, we developed breast cancer cell lines with elevated inducible NFAT1 expression. As predicted in our previous work (39), inducible expression of NFAT led to increased transcriptional activity with a concomitant increase in invasion through Matrigel. Gene expression profiling by microarray analysis revealed that COX-2 is one of the most consistently up-regulated genes induced by NFAT in breast cancer cells.
Induction of COX-2 by NFAT has been reported in human T lymphocytes (19). This issue has also been investigated in nonimmune cells. For example, NFAT apparently increases COX-2 expression in vascular smooth muscle cells when induced by platelet-derived growth factor and in balloon injury-induced neointima formation in rat carotid artery (42). Similarly, a recent study also provides evidence for COX-2 expression induced by NFAT in colon carcinoma (38). However, our study is the first to demonstrate induction of COX-2 by NFAT in breast cancer cells, and more importantly, we directly demonstrate that COX-2 is one of the genes that is responsible for NFAT-driven Matrigel invasion.
To probe the mechanism by which NFAT induces COX-2, we used a COX-2 promoter reporter construct that harbors two putative NFAT binding sites. The distal site at nucleotide Ϫ105 to Ϫ97 has no adjacent AP-1 binding site, whereas the proximal NFAT site at nucleotides Ϫ76 to Ϫ68 has a highly homologous AP-1 consensus sequence at Ϫ61 to Ϫ67. These elements have been shown to bind NFAT and are important for COX-2 transcription in various cells (19,38). In human T cells, the proximal NFAT1 site is crucial for regulation of COX-2 transcription (19). In our experiments in breast cancer cells, induction of NFAT1 expression in the presence of the activators PMA/ionomycin resulted in a dramatic increase in NFAT transcriptional activity. Concomitant with this increase was a significant induction of COX-2 message and protein. Similar results, albeit less prominent, were obtained in cells not induced to express NFAT1, and this is likely because of the activation of endogenous NFAT1 and AP-1, known to be expressed in these cells. It is well known that the basic leucine-zipper protein AP-1, composed of Fos and Jun in mammalian cells, is activated by phorbol esters such as PMA (43). In the presence of the calcineurin antagonist cyclosporin A, NFAT activity was diminished, also concomitant with a decrease in COX-2. We therefore conclude that in breast cancer cell lines, NFAT and AP-1 cooperate by binding to the COX-2 promoter, leading to de novo gene transcription and protein expression. The net effect was an increase in PGE 2 synthesis, which we also showed was COX-2-and NFAT-dependent.
We also evaluated COX-2 regulation in breast cancer cells by silencing NFAT1 expression using siRNA. Reduced expression of NFAT1 by siRNA resulted in reduced NFAT1 expression, transcriptional activity, and invasion, reinforcing the notion that NFAT is both necessary and sufficient to promote invasion. Moreover, reduced NFAT1 also diminished COX-2 induction, and because COX-2 siRNA also effectively blocked invasion in two distinct breast cancer cell lines, whereas COX-2 over-expression increased invasion, we conclude that COX-2 is both necessary and sufficient to drive Matrigel invasion.
Our results are in agreement with several lines of evidence showing that COX-2 is intimately associated with tumor progression and metastasis. There is a strong correlation between the high levels of COX-2 expression and increased invasive potential of cancer cells (44,45). Long-term use of nonsteroidal anti-inflammatory drugs has been shown to lower the risk of cancer development such as colorectal and prostate cancers (35,46). Moreover, COX-2 inhibitors show potent effects in reducing tumor growth and metastasis in animal models of cancer (47,48). Reduced tumorigenesis has also been observed in COX-2 null mice (49,50). PGE 2 is the major product of COX-2 (20) and has been shown to increase the metastatic potential of cancer cells (28,30). Our data demonstrate that NFAT induction and activation result in elevated levels of PGE 2 . As expected, the COX-2 inhibitor NS-398, a nonsteroidal anti-inflammatory drug, significantly diminished PGE 2 production. Although the induction of NFAT expression promoted invasion, cells treated with NS-398 displayed reduced invasion in a dosage-dependent manner. This result further indicates that NFAT promotes cancer invasion by induction of COX-2 and increased PGE 2 production.
The exact mechanism by which COX-2 and PGE 2 enhance tumor invasion is still poorly understood. PGE 2 has been shown to regulate aromatase expression and activity in breast cancer, thus affecting local estrogen synthesis (51). Estrogen is strongly associated with breast tumorigenesis by stimulating cell proliferation (52). Immunohistochemical analysis of tissue array specimens of invasive breast cancers also reveals that increased COX-2 expression is invariably associated with elevated expression of matrix metalloproteinase-2 (53). Induction of metalloproteinase-2 by PGE 2 has been reported to enhance the invasion of pancreatic cancer cells (30). Studies on transgenic mice suggest that PGE 2 signaling via the G protein-linked E-series prostanoid-2 receptor of PGE 2 , which is highly expressed in breast cancer, induces mammary hyperplasia through the cAMP-dependent up-regulation of amphiregulin (54). A recent study also shows that PGE 2 binds to the E-series prostanoid-2 receptor to activate ␤-catenin signaling that stimulates the proliferation of colon cancer cells (55). COX-2 expression has also been shown to increase the survival of intestinal epithelial cells (56) and to be pro-angiogenic in endothelial cells (44). For these reasons, COX-2 inhibitors such as rofecoxib (Vioxx) and celecoxib (Celebrex) were until recently enrolled into clinical trials for colorectal cancer therapy. Our studies add a new dimension to the pleiotropic effects induced by COX-2 by demonstrating that COX-2 can promote invasion in an NFAT-dependent manner. . Induction of COX-2 by NFAT1 increases PGE 2 production and promotes invasion. A, SUM.N1-16 cells were cultured with or without 1 g/ml Dox. After 24 h incubation, the culture medium was replaced with serum-free medium, and the cells were either left untreated or stimulated overnight with 100 nM PMA and ionomycin (Ion). NS-398 was added at 50 M. The culture medium was collected after 24 h, and the production of PGE 2 was determined by a PGE 2 monoclonal enzyme immunoassay kit as described under "Experimental Procedures." B, PGE 2 production was assayed in the culture medium of MDA-MB-231 cells with or without overnight stimulation with PMA and ionomycin in the absence or presence of NS-398. C, SUM.N1-16 cells were treated with the indicated amounts of NS-398 for 2 h prior to incubation with 1 g/ml Dox. After 24 h, cells were harvested, and invasion was determined by Matrigel assay. The data shown are the representative of two independent experiments with identical results. Bars, S.D. The asterisks denote statistically significant differences in PGE 2 production between treated and untreated cells (unpaired Student's t test, p Ͻ 0.02).
To conclude, our results show that NFAT promotes breast cancer invasion through the induction of COX-2. To our knowledge, to date this is only the second time that a gene induced by NFAT in cancer cells has been directly demonstrated to be responsible for promoting invasion in vitro. Recently, Chen and O'Connor (57) showed that autotaxin/ ENPP2, an enzyme that catalyzes the production of lysophosphatidic acid, is also induced by NFAT in breast cancer cells and promotes invasion. Although NFAT has been shown to regulate COX-2 in various cells, and high expression of COX-2 is correlated with increased metastatic potential, no studies have investigated the role of COX-2 as an invasion-promoting gene in the NFAT pathway. We speculate that this may be clinically significant because the identification and characterization of NFAT target genes such as COX-2 may provide new targets and information to design better anticancer drugs that directly inhibit NFAT and its downstream targets.