Ras Promotes Transforming Growth Factor-β (TGF-β)-induced Epithelial-Mesenchymal Transition via a Leukotriene B4 Receptor-2-linked Cascade in Mammary Epithelial Cells*

Background: The mechanism by which oncogenic Ras contributes to epithelial-mesenchymal transition (EMT) is not clearly defined. Results: The “BLT2-ROS-NF-κB”-linked cascade lies downstream of Ras and promotes EMT. Conclusion: Activation of leukotriene B4 receptor-2 (BLT2)-linked pathway contributes to EMT in mammary epithelial cells. Significance: These findings suggest that BLT2 is a novel regulator of EMT in mammary epithelial cells. Inflammation and inflammatory mediators are inextricably linked with epithelial-mesenchymal transition (EMT) through complex pathways in the tumor microenvironment. However, the mechanism by which inflammatory mediators, such as the lipid inflammatory mediators, eicosanoids, contribute to EMT is largely unknown. In the present study we observed that BLT2, leukotriene B4 receptor-2, is markedly up-regulated by oncogenic Ras and promotes EMT in response to transforming growth factor-β (TGF-β) in mammary epithelial cells. Blockade of BLT2 by the BLT2 inhibitor LY255283 or by siRNA reduced EMT induced by Ras in the presence of TGF-β. In addition, stimulation of BLT2 by the addition of a BLT2 ligand, such as leukotriene B4, restored EMT in the presence of TGF-β in human immortalized mammary epithelial MCF-10A cells. We further searched BLT2 downstream components and identified reactive oxygen species and nuclear factor κB as critical components that contribute to EMT. Taken together, these results demonstrate for the first time that a BLT2-linked inflammatory pathway contributes to EMT. This provides valuable insight into the mechanism of EMT in mammary epithelial cells. In addition, considering the implications of EMT with the stemness of cancer cells, our finding may contribute to a better understanding of tumor progression.

mesenchymal cell in response to stress or specific growth factors. This process is characterized by down-regulation of epithelial marker proteins, such as E-cadherin and zonula occludens-1 (ZO-1), and up-regulation of mesenchymal marker proteins, such as vimentin, N-cadherin, and fibronectin (1). It is currently thought that a subset of cancerous cells undergoes EMT to enable them to escape from surrounding cells, penetrate into neighboring lymph or blood vessels, and passage to distant sites to form secondary metastases (2,3).
Transforming growth factor-␤ (TGF-␤) is one of the key growth factors involved in driving EMT. Increased expression of TGF-␤ is observed in breast cancers, where it is associated with advanced stage, increased recurrence, and poor prognosis (4). Depending on the cellular context, TGF-␤ can induce EMT in cooperation with oncogenic Ras (5)(6)(7). Elevated expression of the oncogenic Ras protein has been found in 60 -70% of human primary breast carcinomas, and thus Ras expression has been suggested as a marker of tumor aggressiveness in breast cancer (8). By synergistically promoting EMT in the presence of TGF-␤, oncogenic Ras has been shown to collaborate with TGF-␤ to promote cell migration in vitro and tumor invasion and metastasis in vivo (5,9). The mechanism by which oncogenic Ras contributes to EMT is not yet understood. Previously, we have shown that BLT2 lies downstream of Ras and mediates oncogenic Ras-induced transformation and invasion (10 -12). The levels of leukotriene B 4 (LTB 4 ), one of the local lipid mediators in the inflammatory microenvironment, and its receptor, BLT2, are markedly up-regulated by oncogenic Ras and mediate Ras-associated tumorigenic activities (10 -12). Expression of BLT2 in ovarian and breast cancer tissue is increased in advanced stages and is associated with poor clinical outcome (13)(14)(15). Furthermore, autocrine or paracrine BLT2 signaling mediates the invasiveness and metastasis of ovarian, bladder, and breast cancer cells (14,16,17).
Despite these observations implicating BLT2 as a potential mediator for aggressive metastatic cancer, its mechanism of action in EMT is not characterized. In the present study we found that BLT2 lies downstream of Ras and collaborates with TGF-␤ to induce EMT in mammary epithelial cells. We further examined BLT2 downstream components and identified reac-tive oxygen species (ROS) and NF-B as critical components that contribute to EMT.
For real-time quantitative PCR analysis, total RNA extracted from cells using the easy-BLUE RNA extraction kit (Intron) was subjected to RT with Moloney murine leukemia virus reverse transcriptase (Beams Biotechnology). The BLT1, BLT2, and GAPDH cDNAs were amplified as described previously (15) with SYBR Green I-based real-time PCR using a LightCycler 480 instrument (Roche Applied Science). The quantification data were analyzed with LightCycler software 3.3 (Roche Diagnostics). The melting curves were analyzed to ensure the specificity of amplification. Primers (forward and reverse, respectively) for real-time PCR analysis were 5Ј-CCTGAAAAGGATG-CAGAAGC-3Ј and 5Ј-AAAAAGGGAGCAGTGAGCAA-3Ј for human BLT1 and 5Ј-CTTCTCATCGGGCATCACAG-3Ј and 5Ј-ATCCTTCTGGGCCTACAGGT-3Ј for human BLT2.
Three-dimensional Collagen Assay-Depending on the final volume required, calculated amounts of 10 ϫ phosphate-buffered saline (PBS), 5 N sodium hydroxide (NaOH), and type I rat tail collagen (BD Biosciences; final concentration 2 mg/ml) were prepared. The pH was adjusted to 7.4 Ϯ 0.05, and the solutions were kept on ice for up to 30 min before use. Wells of a 24-well dish were pre-coated with 200 l of collagen solution. Cells were harvested, washed 3 times with PBS, and resuspended at 3 ϫ 10 4 cells/ml in the collagen solution. A 300-l volume of cell suspension was laid over the bottom layer. After gelling, complete culture medium was added and changed every 2-3 days. Morphology was assessed on day 9 by image capture at ϫ20 magnifications.
Invasion Assay-To monitor the invasion potential of MCF-10A and MCF-10A/Hras cells, BD BioCoat TM Matrigel TM Invasion Chambers (BD Biosciences) were used. Cells (5 ϫ 10 4 ) were removed from culture plates with trypsin-EDTA, washed in DMEM/F-12 containing 0.5% serum, and seeded on top of the rehydrated Matrigel inserts. As a chemoattractant, DMEM/ F-12 containing 5% serum was added to the lower chamber, and plates were incubated at 37°C for 48 h. The membranes were fixed in methanol and stained with hematoxylin and eosin (H&E) stain. The contents of the upper membrane surface were then removed, and invasive cells were microscopically counted in four random high power fields per filter; each sample was assayed in triplicate, and the assays were performed three times.
Flow Cytometric Analysis of BLT2-For quantitation of BLT2 expression, cells were incubated in 60-mm dishes for 24 h. The cells were detached by treatment with trypsin, washed with PBS, and fixed in 2% paraformaldehyde and then permeabilized with 0.1% Triton X-100 for 10 min. After exposure to 2% BSA for 30 min, they were incubated for 1 h at room temperature with rabbit polyclonal antibodies to BLT2 (1:100 dilution; Cayman Chemical, Ann Arbor, MI), washed 3 times with PBS, and incubated for 30 min at room temperature with fluorescein isothiocyanate-conjugated goat antibodies to rabbit immunoglobulin G (1:200 dilution; Molecular Probes, Eugene, OR). The cells (10,000 per sample) were then subjected to flow cytometry using a FACSCalibur instrument and Cell Quest software (BD Biosciences) for determination of mean fluorescence intensity.
Immunofluorescence Staining of E-cadherin and Vimentin-Cells were cultured as a monolayer on glass coverslips (24 ϫ 24 mm), washed twice with PBS, fixed with 4% paraformaldehyde in PBS for 10 min, permeabilized with 0.1% Triton X-100 for 10 min, and washed 3 times in PBS. They were blocked with 2% BSA in PBS for 30 min at room temperature and incubated for 60 min with primary antibodies (1:100 dilution for E-cadherin, 1:100 dilution for vimentin; Santa Cruz Biotechnology, Santa Cruz, CA) and then with fluorescein isothiocyanate-or tetramethylrhodamine isothiocyanate-conjugated secondary antibodies (1:200 dilution; Molecular Probes) for 45 min. Each step was performed in PBS containing 2% BSA, and the cells were washed 3 times for 5 min with the same solution between steps. The integrity of the nuclei was confirmed by DAPI (Sigma) staining. The coverslips were mounted on 25 ϫ 75-mm microslides using 90% glycerol, examined with a confocal laserscanning microscope (510META; Carl Zeiss, Oberkochen, Germany).
Western Blotting-Adherent cells were collected and lysed with buffer (40 mM Tris⅐HCl, pH 8.0, 120 mM NaCl, 0.1% Non-idet-P40, 100 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate, 2 g/ml leupeptin, 2 g/ml aprotinin). Proteins were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane. The membrane was blocked with 5% nonfat dry milk in Trisbuffered saline and incubated with primary antibodies against E-cadherin, vimentin, p-IB-␣, IB-␣,and ␣-tubulin for 1 h at room temperature. Blots were developed with a peroxidaseconjugated secondary antibody, and proteins were visualized by enhanced chemiluminescence (Amersham Biosciences).
Measurement of ROS-Intracellular hydrogen peroxide (H 2 O 2 ) was measured in cells plated in 60-mm dishes. Actively growing cells were incubated for 10 min in the dark at 37°C with 20 M 2Ј,7Ј-dichlorofluorescein diacetate, after which dichlorofluorescein fluorescence was measured by flow cytometry with a FACSCalibur (BD Biosciences).
Luciferase Reporter Assay for NF-B Activity-MCF-10A and MCF-10A/Hras cells were transfected with 1.4 g of NF-B luciferase reporter gene (28) using Lipofectamine reagent (Invitrogen). To monitor variations in cell numbers and transfection efficiency, MCF-10A and MCF-10A/Hras cells were cotransfected with 0.6 g pSV40-␤-galactosidase, a eukaryotic expression vector containing the Escherichia coli ␤-galactosidase (lacZ) structural gene under the control of the SV40 promoter. After 24 h in complete medium, starvation with 0.5% horse serum for 3 h, and treatment with TGF-␤ for 48 h, the cells were harvested and assayed for luciferase activity as described (28).
Data Analysis and Statistics-The data are representative of three independent experiments. Results are presented as the means Ϯ S.D. Analyses were performed with Student's t test using SigmaPlot 8.0 software. p values less than 0.05 were considered significant.

Ras Collaborates with TGF-␤ to Potentiate EMT in Mam-
mary Epithelial Cells-It has been shown that Ras collaborates with TGF-␤ to induce EMT in several epithelial cell models (5)(6)(7)29). In the present study we used MCF-10A immortalized human mammary epithelial cells, which have a round cobblestone-like morphology (Fig. 1A). In addition, we used MCF-10A/Hras cells expressing oncogenic H-Ras, which exhibit a partial morphological alteration from round cobblestone-like to fibroblast-like cells. Treatment of MCF-10A/Hras cells with TGF-␤ further generated a fibroblast-like elongated shape and induced a loss of cell-cell contact, which is typical of EMT (Fig.  1A), concomitant with a decrease in the expression of Ecadherin and an increase in that of vimentin, as examined by semiquantitative RT-PCR and immunoblot analysis (Fig. 1B). The same sets of cells were evaluated for morphology in threedimensional culture. Cells that have undergone EMT are known to exhibit a branched morphology in three-dimensional matrix. As shown in Fig. 1C, after treatment with TGF-␤, MCF-10A/Hras cells exhibited highly branched structures, typical of EMT. EMT was also shown to cause increased motility and invasiveness, thus promoting tumor metastasis (30). We, therefore, determined invasiveness using a modified Boyden chamber assay. MCF-10A/Hras cells stimulated with TGF-␤ were more invasive than TGF-␤-stimulated MCF-10A cells (Fig.  1D). Together, these results suggest that Ras promotes EMT in response to TGF-␤ in MCF-10A cells.
Blockade of BLT2 Significantly Reduced EMT Promoted by Ras-To elucidate the downstream components of Ras signaling in EMT induced in MCF-10A/Hras cells in response to TGF-␤, we first analyzed BLT2 expression. In agreement with a previous report (10), expression of BLT2, but not BLT1, was elevated by Ras, as detected by semiquantitative RT-PCR and quantitative real-time PCR analysis ( Fig. 2A). Consistent with these results, the abundance of BLT2 protein as determined by flow cytometry was higher in MCF-10A/Hras cells than in MCF-10A cells (Fig. 2B). In contrast, TGF-␤ treatment did not affect transcript levels of BLT2 (Fig. 2C). We next examined whether BLT2 plays any role in EMT triggered by Ras. Treatment of MCF-10A/Hras cells with BLT2 inhibitor LY255283 or depletion of BLT2 with a BLT2-specific small interfering RNA (siBLT2) greatly reduced the morphological alterations (Fig. 2,  D and H) and the changes in epithelial and mesenchymal marker expression (Fig. 2, E and I). In contrast, the BLT1 inhibitor U75302 had no effect on morphology and epithelial/mesenchymal marker expression (Fig. 2, D and E). Similarly, in three-dimensional culture, development of the branched morphology of MCF-10A/Hras cells in response to TGF-␤ was dra-matically inhibited by LY255283 or siBLT2 knockdown (Fig. 2, F and J). Next, we examined whether BLT2 contributed to the increased invasion observed in MCF-10A/Hras cells in response to TGF-␤. Application of LY255283 or siBLT2 knockdown reduced the invasiveness of MCF-10A/Hras cells in response to TGF-␤ (Fig. 2, G and K). Together, these results suggest that BLT2 lies downstream of Ras and contributes to EMT.
BLT2 Cooperates with TGF-␤ to Induce EMT in Mammary Epithelial Cells-To further examine the contribution of BLT2 in signaling the EMT in mammary epithelial cells, we examined whether BLT2 could collaborate with TGF-␤ to drive EMT in MCF-10A cells. For this test, MCF-10A cells were treated with LTB 4 , an agonist of BLT2. Treating MCF-10A cells with LTB 4 alone had no significant effect on either morphology (Fig. 3A) or EMT marker expression (Fig. 3, B and C). In addition, TGF-␤ treatment alone partially decreased expression of E-cadherin and partly increased expression of vimentin (Fig. 3, B and C). However, co-treatment of the cells with LTB 4 and TGF-␤ markedly promoted the change in morphology, with cells adapting a spindle-like fibroblastic shape (Fig. 3A) and EMT marker expression (Fig. 3, B and C). To confirm the collaboration between BLT2 and TGF-␤, we transiently transfected MCF-10A cells with BLT2 expression plasmid. The addition of TGF-␤ to these BLT2-transfected cells markedly reduced levels of E-cadherin and enhanced levels of vimentin (Fig. 3D). Also, in three-dimensional collagen culture, TGF-␤ induced branching in BLT2-overexpressing MCF-10A cells (Fig. 3E). Moreover, overexpression of BLT2 in MCF-10A cells markedly enhanced TGF-␤-induced invasiveness (Fig. 3F). Together, these results suggest that BLT2 and TGF-␤ can collaborate to induce EMT in mammary epithelial cells.
"Nox1-ROS" Lied Downstream of BLT2 and Contributed to EMT-ROS play important roles in EMT and the mobility of many cell types (31). Also, our previous observations suggested that ROS production lies downstream of BLT2 in Ras signaling leading to transformation or during cancer progression (11,12,16). We, therefore, examined whether ROS functions down-stream of BLT2 to promote EMT in MCF-10A/Hras cells. First, the ROS levels in MCF-10A and MCF-10A/Hras cells were examined. With the use of the redox-sensitive fluorescent probe dichlorofluorescein, we found that intracellular ROS production was increased in MCF-10A/Hras cells compared with MCF-10A cells (Fig. 4A). However, this increase in ROS levels in MCF-10A/Hras cells was reduced by depletion of BLT2 by siRNA (Fig. 4B), suggesting that BLT2 mediates ROS production in MCF-10A/Hras cells.
We recently reported that BLT2 up-regulates NADPH oxidase (Nox) (12,15,17,32). Therefore, we determined whether BLT2 plays a role in the Nox up-regulation in MCF-10A/Hras cells. Clearly, the level of Nox1 mRNA, but not that of Nox4 mRNA, in MCF-10A/Hras cells was reduced by LY255283 (Fig. or control (siCont) siRNAs for 48 h, after which the amounts of BLT2 mRNAs were determined by semiquantitative RT-PCR (left panels). MCF-10A/Hras cells were transfected with BLT2 or control siRNAs for 24 h before TGF-␤ (5 ng/ml) treatment for 48 h. The cells were visualized using an Olympus BX51 microscope at ϫ20 magnification (right panels). I, MCF-10A/Hras cells were treated as in H (right panels), after which the expressions of E-cadherin and vimentin were determined by semiquantitative RT-PCR analysis or immunoblot analysis. J, MCF-10A/Hras cells were transfected with BLT2 or control siRNAs for 24 h. The cells were then propagated in three-dimensional cultures for 9 days in the absence or presence of TGF-␤ (5 ng/ml), at which point alterations in morphology were visualized using an Olympus BX51 microscope at ϫ20 magnification. Scale bars, 100 m. K, MCF-10A/Hras cells were transfected with BLT2 or control siRNAs for 24 h and then assayed for invasiveness in the presence of TGF-␤ (5 ng/ml) for 48 h. All quantitative data are shown as the mean Ϯ S.D. of three independent experiments. *, p Ͻ 0.05; †, p Ͻ 0.01; ‡, p Ͻ 0.005. AUGUST 8, 2014 • VOLUME 289 • NUMBER 32

JOURNAL OF BIOLOGICAL CHEMISTRY 22155
4C, left panels) or by siBLT2 knockdown (Fig. 4C, right panels), whereas U75302 had no such effect (data not shown). Interestingly, TGF-␤ was found to induce Nox4 up-regulation (Fig. 4, C  and D). Overexpression of BLT2 markedly enhanced Nox1 mRNA, but not Nox4 mRNA (Fig. 4D), suggesting that Nox1 functions downstream of BLT2. Consistent with these results, depletion of Nox1 by siRNA resulted in marked inhibition of ROS generation in MCF-10A/Hras cells (Fig. 4E). It also markedly attenuated the changes in epithelial and mesenchymal marker expression induced by TGF-␤ treatment (Fig. 4F). Together, these results suggest that "'Nox1-derived ROS production" lies downstream of BLT2 and promotes EMT in the presence of TGF-␤.

NF-B Lies Downstream of BLT2-ROS and Contributes to EMT-
The transcription factor NF-B has been identified as a central mediator of EMT in a mouse model of breast cancer progression (33). In addition, increased production of ROS has been shown to activate NF-B (34). We, therefore, determined whether NF-B contributes to EMT triggered by Ras in the presence of TGF-␤. To do this, MCF-10A/Hras cells were pretreated with the NF-B inhibitor Bay11-7082 before stimulation with TGF-␤. As expected, this reduced the changes in E-cadherin and vimentin expression in MCF-10A/Hras cells in the presence of TGF-␤ (Fig. 5A). We next examined whether NF-B lies downstream of the BLT2-ROS cascade in MCF-10A/Hras cells. Immunoblot analysis of p-IB-␣ and IB-␣ revealed that NF-B activation was markedly enhanced in MCF-10A/Hras cells compared with in MCF-10A cells (Fig.  5B). And depletion of BLT2 by siRNA resulted in marked attenuation of the NF-B activation (Fig. 5C). To further study the role of the BLT2-ROS cascade in the activation of NF-B, MCF-10A cells were stimulated by increasing the concentration of LTB 4 alone or in combination with TGF-␤, and NF-B activation was measured by immunoblot analysis of p-IB-␣ and IB-␣. As shown in Fig. 5D, the addition of LTB 4 enhanced the phosphorylation of IB-␣ and degradation of IB-␣ in response to TGF-␤. Moreover, BLT2 overexpression enhanced the phosphorylation of IB-␣ and degradation of IB-␣ in response to TGF-␤ (Fig. 5E). We also determined the activity of an NF-B luciferase reporter gene and found that BLT2 depletion by siBLT2 in MCF-10A/Hras cells inhibited NF-B luciferase activity in response to TGF-␤ (Fig. 5F). In addition, BLT2 stimulation by either treatment with LTB 4 or overexpressing BLT2 markedly enhanced NF-B luciferase activity in response to TGF-␤ (Fig. 5, G and H). Finally, we found that depletion of Nox1 with siRNA markedly inhibited the phosphorylation of IB-␣ and degradation of IB-␣ (Fig. 5I). Together, these results suggest that NF-B activation lies downstream of the

DISCUSSION
In this study we have shown that BLT2 lies downstream of Ras and contributes to EMT through collaboration with the TGF-␤ signaling pathway in human mammary epithelial cells. In addition, we analyzed the key components mediating the BLT2 signaling to EMT and found that the Nox1-ROS-NF-B cascade appears to be downstream of BLT2. Our findings provide important insights into the mechanism of EMT in mammary epithelial cells.
Elevated amounts of the oncogenic Ras protein have been found in 60 -70% of primary breast carcinomas, and thus Ras expression has been suggested to be a marker of tumor aggressiveness in breast cancer (6,8). Previous reports have suggested that oncogenic Ras cooperates with TGF-␤ to induce EMT in both mammary and keratinocyte-derived tumors and to drive tumor metastasis (35). Despite these reports implicating Ras in EMT, the downstream pathway of Ras mediating EMT has not been previously characterized. The results of the present study clearly demonstrate that BLT2 up-regulation lies downstream of Ras and contributes to EMT through collaboration with the TGF-␤ signaling pathway in mammary epithelial cells (Fig. 2). In support of BLT2 as a mediator of EMT, treatment with LTB 4 or overexpression of BLT2 induced enhanced EMT in the presence of TGF-␤ (Fig. 3). These results suggest that BLT2 and TGF-␤ somehow collaborate to fully induce EMT in mammary epithelial cells. To further examine the contributing role of BLT2 in EMT in mammary epithelial cells, we tested another cell line, EpH4, mouse mammary epithelial cells. Previously, EpH4 cells were shown to undergo EMT by cooperation between TGF-␤ and Ras (6). Using EpH4 and EpRas cells (EpH4 cells expressing oncogenic Ras) (6), we obtained results that were quite similar to those in MCF-10A and MCF-10A/Hras cells. For example, the expression of BLT2 was elevated in EpRas cells compared with EpH4 control cells (Fig. 6A). Also, the depletion of BLT2 with siBLT2 significantly attenuated the TGF-␤-induced changes in protein levels of E-cadherin and vimentin in EpRas cells (Fig. 6B), together again supporting the role of BLT2 in EMT in mammary epithelial cells.
We evaluated the signaling pathways involved in the stimulation of EMT by BLT2 in response to TGF-␤. Previous studies have shown that TGF-␤ drives EMT predominantly via the Smad2/3 signaling pathway (36,37). In this study we found that TGF-␤-mediated phosphorylation of Smad2/3 was not affected by altering BLT2 (Fig. 6C), suggesting that the BLT2 cascade is independent of Smad2/3 phosphorylation. In a previous report TGF-␤ was shown to collaborate with activated mutants of Ras, which activate constitutive extracellular signal-regulated kinases (ERK), to induce EMT in numerous epithelial cell types (6). In the MCF-10A system described here, we found that PD98059, an inhibitor of ERK, attenuated TGF-␤-induced EMT. However, the BLT2 inhibitor LY255283 had no effect on the phospho-ERK levels of MCF-10A/Hras cells (data not shown), suggesting that the ERK pathway does not function downstream of BLT2 in EMT. We propose that NF-B lies downstream of BLT2 and promotes TGF-␤-induced EMT. We found that TGF-␤-induced EMT was attenuated by the selective NF-B inhibitor Bay11-7082 (Fig. 5A), and activation of NF-B was suppressed by BLT2 knockdown (Fig. 5, C and F). In addition, previous studies have shown that the transcription factor NF-B is a central mediator of EMT in a mouse model of breast cancer progression (33). In support of BLT2 as a mediator of NF-B activation, overexpression of BLT2 induced enhanced NF-B activation in the presence of TGF-␤ (Fig. 5, E and H). Furthermore, co-stimulation with TGF-␤ and LTB 4 increased NF-B activation compared with either factor alone (Fig. 5, D  and G). Therefore, these results suggest that synergistic regulation of NF-B activation is an important component of the collaborative effect of oncogenic Ras and TGF-␤ signaling in mammary epithelial cells and that NF-B activation in response to both TGF-␤ and BLT2 may be a key convergent point with a role in the inflammatory accentuation of EMT. Consistent with the role of NF-B in EMT, activation of NF-B has been shown to promote the transcription of the EMT regulators (e.g. snail and ZEB1/2). For example, it was reported that the activation of Snail transcription requires NF-B; the Snail promoter has functional NF-B sites (38), and the inhibition of NF-B significantly reduced Snail transcription (39,40). In addition, Chua et al. (41) reported that NF-B suppresses the expression of E-cadherin and induces the expression of the mesenchymespecific gene vimentin in MCF-10A cells through ZEB1/2. Consistent with those reports, we observed that inhibition of NF-B with Bay11-7082 significantly diminished TGF-␤-induced snail expression (Fig. 6D) and attenuated TGF-␤-induced Ecadherin suppression (Fig. 5A) in MCF-10A/Hras cells. Because our results (Fig. 5, C-H) demonstrated that NF-B activation lies downstream of BLT2 in the signaling to EMT, we next examined whether the inhibition or stimulation of BLT2 affects the expression of snail. BLT2 knockdown with siRNA clearly reduced snail expression in MCF-10A/Hras cells in the presence of TGF-␤ (Fig. 6E), and BLT2 transfection followed by LTB 4 treatment in MCF-10A cells further increased the expres- sion of snail (Fig. 6F). These findings support a role of BLT2-NFB in the activation of EMT regulators such as snail.
Our studies suggest that Nox1-derived ROS generation lies upstream of NF-B activation in the BLT2-signaling cascade to EMT. We observed that Nox1 lies downstream of BLT2 and Nox1-induced ROS generation accentuates TGF-␤-driven EMT via NF-B activation. These results are consistent with previous reports suggesting that ROS are involved in EMT (42). We have shown that knockdown of Nox1 by siRNA attenuates ROS generation as well as TGF-␤-induced EMT. Boudreau et al. (43) recently reported that Nox4 contributes to NADPH oxidase-dependent ROS production, which may be critical for the progression of EMT in breast epithelial cells. They demonstrated that phosphorylation of Smad3 positively regulates the human Nox4 promoter in breast epithelial cells. Similarly, we found that the level of Nox4 mRNA, but not that of Nox1 mRNA, was markedly increased by exposure to TGF-␤ (Fig. 4,  C and D). However, we showed that the TGF-␤-induced upregulation of Nox4 mRNA was not inhibited by knockdown of BLT2 (Fig. 4C). It is possible that BLT2-induced Nox1 and TGF-␤-induced Nox4 collaborate to fully induce EMT in mammary epithelial cells. For example, BLT2-Nox1-derived ROS and TGF-␤-Nox4-derived ROS may additively stimulate NF-B activation. In support of this idea, we found that depletion of both Nox1 and Nox4 by the appropriate siRNA led to a further dramatic inhibition of NF-B activation (Fig. 6G) and changes in EMT markers (e.g. E-cadherin and vimentin) (Fig.  6H).
In any event, based on our observations, we propose a model that links the Ras and BLT2 signaling pathways and leads to accentuation of TGF-␤-induced EMT in mammary epithelial cells. As summarized in Fig. 5J, our data reveal that up-regulation of BLT2 by oncogenic Ras results in the generation of ROS via Nox1 that together with NF-B are probable downstream components of the BLT2-Nox1-ROS cascade, with potential roles in EMT associated with the collaboration of Ras with TGF-␤. Identification of this novel mechanism may contribute to a better understanding of the mechanism of EMT.