Loss of TBK1 Induces Epithelial–Mesenchymal Transition in the Breast Cancer Cells by ERα Downregulation

Breast cancer is one of the most common cancers among females, and 50% to 80% of breast tumors express estrogen receptor α (ERα; refs. 1, 2). ERα-positive tumors are characterized by a well-differentiated phenotype and are associated with a better prognosis than ERα-negative tumors, which are correlated with disease-free survival (3, 4). The major reason for this is that ERα-positive tumors initially respond well to anti-estrogen agents such as tamoxifen (5). However, a significant proportion of ERα-positive tumors eventually become resistant to anti-estrogen therapy (6). Because of its complexity, breast cancer may be considered a broad set of diseases that includes multiple, distinct biological subtypes with diverse natural histories that present a varied spectrum of clinical, pathologic, and molecular features with different prognostic and therapeutic implications.

ERα plays an important role in breast cancer development and progression by influencing the genes and signaling pathways that are involved in cellular proliferation. Ligand-activated ERα translocates to the nucleus and then binds to the responsive element in the target gene promoter, thus stimulating gene transcription (7). The activation of ERα in the cytoplasm contributes to the activation of tyrosine kinase receptors including EGFR, IGF-IR, and HER2, as well as to the interaction of ERα with cellular kinases. This in turn leads to the activation of the MAPK and AKT pathways, which are known to enhance cell proliferation and survival (8–10). Although ERα signaling is one of the most extensively studied pathways, the role of ERα signaling in which the molecular basis of breast cancer progresses to metastasis remains poorly understood.

The epithelial–mesenchymal transition (EMT) gives cells the ability to migrate and invade; this typically reflects the plasticity of epithelial cells and is a crucial process during embryonic development and tissue organization (11). EMT has also been reported to be involved in cancer progression, specifically in the development of metastatic carcinomas (12, 13). The E-cadherin–EMT pathway is an important pathway in human breast cancer progression and may regulate the tumor progression, invasion, and metastasis in certain types of human breast cancers (14–16). Recently, it has been shown that ERα signaling regulates E-cadherin and EMT through the transcriptional repression of Slug, an E-cadherin repressor (17). Moreover, the loss of ERα leads to tamoxifen resistance, which drives cellular transdifferentiation from an epithelial to a mesenchymal phenotype, aggravating the invasiveness of the breast cancer (18, 19).

IKKϵ and TBK1 (TANK-binding kinase 1) are known as the noncanonical members of the IKK kinase family. IKKϵ and TBK1 are activated by inflammatory stimuli and regulate critical cellular processes, including inflammation, survival, proliferation, and antiviral responses (20, 21). IKKϵ and TBK1 can also promote oncogenesis (22). For example, IKKϵ is overexpressed in breast and ovarian cancers, whereas IKKϵ silencing inhibits cell growth and invasiveness in breast cancer cells (23, 24). TBK1 is involved in the RalB GTPase–mediated activation of the innate immune signaling of tumor cell survival and acts as a synthetic lethal partner of oncogenic K-Ras in K-Ras–driven lung cancer cells (25, 26). Moreover, TBK1 can directly phosphorylate the oncogenic AKT kinase in a phosphoinositide 3-kinase (PI3K)–independent manner (27). Recently, Guo and colleagues (28) showed that ERα is a bona fide substrate of IKKϵ. IKKϵ induces ERα transactivation activity and enhances ERα binding to DNA, contributing to tamoxifen resistance in breast cancer. Although IKKϵ has been studied extensively in breast cancer, the relationship between TBK1 and ERα expression status in breast cancer subtypes is poorly understood.

Here, we show that the loss of TBK1 induces EMT through the downregulation of ERα expression in ERα-positive breast cancers. We found that TBK1 was highly expressed in ERα-positive breast cancers but expressed at much lower levels in ERα-negative breast cancers. Furthermore, the loss of TBK1 expression induced EMT and enhanced tumor growth and lung metastasis by suppressing ERα expression, suggesting that TBK1 may be associated with ERα expression status and an aggressive phenotype in breast cancer.

The human breast cancer cell lines MCF-7, SK-BR-3, and MDA-MB-231 were obtained from American Type Culture Collection; ZR-75B, cloned from the ZR-75-1 cell line (29), was obtained from National Cancer Institute (NCI)/NIH. All breast cancer cell lines were cultured in a Dulbecco's Modified Eagle Medium (DMEM; WelGENE) containing 10% FBS (WelGENE) and 1% penicillin/streptomycin.

MG132 was purchased from Sigma-Aldrich. Antibodies were obtained from Santa Cruz Technology (anti-ZEB1, anti-TWIST, anti-SNAIL1, anti-SLUG, and anti-ERα), Cell Signaling Technology (anti-TBK1, anti-AKT, and anti-phospho-AKT S473), Abcam (anti-FOXO3A and anti-phospho-FOXO3A S253), and BD Biosciences (anti-E-cadherin and anti-vimentin).

TBK1 short hairpin RNA (shRNA) lentivirial vectors (Sigma-Aldrich) were used for knockdown of TBK1. For TBK1 shRNA lentivirus production, 293T cells were transfected with pLKO-shTBK1 or scrambled control pLKO-pGL2 together with the packaging plasmids encoding Δ8.9 and VSV-G. Supernatants containing lentivirus shTBK1 were collected 48 and 72 hours posttransfection. Cells were seeded at 1 × 10⁵ cells/well into 6-well plates. Cells were infected with lentiviral particles and polybrene (8 μg/mL). After infection, virus-containing medium was replaced with normal medium and then all cells were selected by puromycin (2 μg/mL).

Cells were washed twice in cold PBS and lysed in immunoprecipitation buffer (50 mmol/L Tris pH 7.4, 150 mmol/L NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 2 mmol/L EDTA, 10% glycerol) plus phosphatase and protease inhibitors (Roche). Whole-cell lysates were eluted in 5× SDS sample loading buffer. Eluted proteins were separated by SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membrane (Millipore), and detected with appropriate primary antibodies coupled with horseradish peroxidase–conjugated secondary antibody by chemiluminescence (GE Healthcare).

Cells seeded on chamber slides were fixed in ice-cold acetone for 5 minutes on ice. Cells were blocked with 5% bovine serum albumin (BSA) for 1 hour and incubated with anti-E-cadherin, anti-vimentin, anti-TBK1, anti-ERα, or anti-FOXO3A antibodies overnight at 4°C. Antibody-bound cells were detected with Alexa Fluor 488–conjugated secondary antibodies (Invitrogen). Immunofluorescence images were obtained with a Zeiss LSM META 510 confocal microscope.

Total RNA was isolated from cells and tissues using the TRIzol reagent (Invitrogen) according to the protocol provided by the manufacturer. Reverse transcription was carried out with 1 μg of purified RNA using SuperScript II reverse transcriptase (Invitrogen). The synthesized cDNA was amplified by PCR using specific primers. PCR products were visualized by electrophoresis on 1.5% agarose gels with ethidium bromide staining and analyzed with an ImageQuant LAS 4000 image analyzer (GE Healthcare). The 18S rRNA gene was used as an internal control.

Chromatin immunoprecipitation (ChIP) was carried out according to the Millipore protocol. Briefly, cells were treated with 1% formaldehyde to induce DNA–protein crosslinking, resuspended in ChIP lysis buffer (1% SDS, 10 mmol/L EDTA, and 50 mmol/L Tris-HCl, pH 8.0), and sonicated on ice to generate chromatin fragments of 150 to 500 base pairs. Soluble chromatin was incubated with Dynabeads Protein G (Invitrogen) coupled to anti-normal IgG or anti-FOXO3A antibodies. After incubation, the immunocomplexes were washed to remove nonspecifically bound material and then eluted. The eluted immunocomplexes were reverse crosslinked by heating at 65°C overnight. Samples were treated with proteinase K and RNase A, extracted with phenol/chloroform, and precipitated with ethanol. PCR was carried out using primers for human ERα promoter, which is shown in Supplementary Table S1.

A 70 μL aliquot of cells (5 × 10⁵ cells/mL) was seeded into each well of the insert (Ibidi) and plated in low 35-mm² dishes. After allowing cells to attach overnight, the culture inserts were removed, leaving 2 separated cell monolayers with a cell-free gap of approximately 500 μm. The cells were incubated with fresh culture medium and allowed to migrate for 16 hours. At 0 and 16 hours, gap areas were photographed with a light microscope at 100× magnification.

Invasion assays were conducted with BioCoat Matrigel invasion chambers (BD Biosciences) as described in the manufacturer's protocol. Cells (1 × 10⁴) were plated in the top chamber, which contained DMEM with 0.1% FBS. The bottom chamber contained DMEM with 10% FBS. After 24 hours of incubation, noninvasive cells were removed with a cotton swab. The cells that migrated through the membrane and adhered to the lower surface of the membrane were fixed with ethanol and stained with 1% toluidine blue. Wells were repeated in triplicate, and the numbers of invaded cells in each field of view were quantified for statistical analysis.

MCF-7 breast cancer cell lines stably expressing control shRNA or TBK1-specific shRNA (1 × 10⁷) were injected subcutaneously into the flanks of severe combined immunodeficient (SCID) mice. Tumor dimensions were measured per day 3. Mice were sacrificed at 34 day after injection, and tumors were surgically isolated. Tumor volume (V) was calculated by using the formula (S × S × L) × 0.5, where S and L were the short and long dimensions, respectively. For lung colonization assay, SCID mice were injected with control or TBK1-knockdowned MCF-7 cells (2 × 10⁶) into tail veins. Briefly, lungs were collected on day 60 and fixed in 10% formalin. The number of metastatic colonies on the surface of the lungs was counted under a dissection microscope. All animals were maintained according to the CHA Hospital Animal Care and Use Committee guidelines under the protocol number IACUC110004.

Human breast normal and tumor samples were obtained from the Gangnam Severance Hospital after approval by the institutional review board and the ethics committee of Gangnam Severance Hospital (IRB approval number 3-2011-0191). For immunohistochemistry, paraffin-embedded blocks prepared from breast cancer tissues were fixed and stained using anti-TBK1 or anti-ERα antibodies. Visualization was conducted with aminoethylcarbazole.

We downloaded expression data from a public database (30, 31) and extracted the expression values of the genes of interest, that is, ERBB2, ESR1, and TBK1. The NeatMap package for R (http://www.r-project.org) was used to draw a heatmap. A t test was conducted to compare the mean expression across the experimental conditions.

Breast cancer prognoses and survival rates vary greatly depending on the cancer subtype, and different subtypes require different treatments. Breast cancers can be classified according to the genetic profiles of various types of tumor cells (32). To understand the roles of TBK1 in different breast tumor subtypes, we examined public microarray datasets (GSE2034) and analyzed the expression of ERBB, ESR1, and TBK1 in patients with the following breast tumor subtypes: 95 luminal A, 25 luminal B, 34 HER2-enriched, and 55 basal-like, as well as 53 normal tissues (30). A heatmap of breast cancer microarray datasets was generated to visualize the expression levels of ERBB2, ESR1, and TBK1 across the 5 different breast cancer subtypes (Fig. 1A). Interestingly, TBK1 was downregulated in the ESR1-negative subgroup, which contained basal-like tissues, compared with the ESR1-positive subgroup, which contained luminal A and luminal B tissues. Moreover, we found higher TBK1 expression in ESR1-positive luminal tumors (P = 0.0011) and lower TBK1 expression in ESR1-negative basal-like breast tumors (P = 0.0007) than in normal human breast tissues. Notably, a significantly higher level of TBK1 expression was observed in ESR1-positive luminal tumor types than in ESR1-negative basal-like tumor types (P < 0.0001; Fig. 1B). We next tested the level of TBK1 expression in normal tissues and primary human cancer tissues, including luminal A, HER2-enriched, and basal-like subtypes. Reverse transcriptase PCR (RT-PCR) analyses showed an increased expression of TBK1 mRNA in the ESR1-positive luminal A tumor tissue but decreased expression in the ESR1-negative tumor tissue (Fig. 1C). To further investigate the relationship between TBK1 expression and ESR1 expression, we confirmed the expression level of TBK1 in various breast cancer cells. TBK1 was more highly expressed in the ESR1-positive breast cancer cells (ZR-75B, MCF-7, and T47D) than in the ESR1-negative breast cancer cells (Hs578T, MDA-MB-231, MDA-MB-435, and SK-BR-3; Fig. 1D). Taken together, these results suggest that the level of TBK1 expression is correlated with the level of ESR1 expression in breast cancer cells.

Because our analysis of the public microarray datasets and breast cancer cells showed that TBK1 was overexpressed in ERα-positive breast cancer cells, we conducted immunohistochemistry using anti-TBK1 and anti-ERα–specific antibodies to investigate the expression of TBK1 and ERα in normal, ERα-positive, and ERα-negative breast tumor tissues. The expression of TBK1 was more highly elevated in the tumor compartments of ERα-positive breast carcinomas (luminal-type) than in those of ERα-negative breast carcinomas (basal-type; Fig. 2A). We also confirmed the expression levels of TBK1 in tumors with different ERα expression statuses using public microarray datasets (GSE2034). The expression of TBK1 was significantly increased in ERα-positive breast tumors than in ERα-negative breast tumors (Fig. 2B). Next, we analyzed the distant metastasis-free survival of TBK1^high- and TBK1^low-breast cancer patient cohorts based on clinical annotations (31). Patients with low TBK1 expression levels exhibited significantly shorter distant metastasis-free survival time (P = 0.0087, GSE11121; Fig. 2C). These results suggest that TBK1 expression is strongly associated with the expression status of ERα.

It is known that ERα-negative breast cancer cells show aggressive or mesenchymal phenotypes. Because ERα-negative breast cancer cells exhibited lower TBK1 expression than did ERα-positive breast cancers, we hypothesized that reduced TBK1 expression may be associated with a mesenchymal phenotype in breast cancer cells. To determine whether reduced TBK1 expression induced EMT, we generated breast cancer cells that stably expressed TBK1-specific shRNA using a lentiviral system. A significant reduction in TBK1 protein expression was observed in both ERα-positive (ZR-75B and MCF-7) and -negative (SK-BR-3 and MDA-MB-231) breast cancer cells (Fig. 3A). TBK1 knockdown induced a morphologic change from an epithelial- to a mesenchymal-like morphology in ERα-positive breast cancer cells, whereas ERα-negative breast cancer cells did not exhibit a change in cell morphology (Fig. 3B). Next, to confirm that the morphologic change induced by TBK1 knockdown was the direct cause of EMT, we monitored the expression of epithelial and mesenchymal cell markers in TBK1-knockdowned breast cancer cells. In ERα-positive breast cancer cells, TBK1 knockdown decreased the expression of the epithelial marker CDH1 (encoding E-cadherin) but increased the expression of the mesenchymal markers CDH2 (encoding N-cadherin), VIM (encoding vimentin), and FN1 (encoding fibronectin), as confirmed by RT-PCR and immunoblotting experiments (Fig. 3C and D). Consistent with the observation above, confocal microscopy imaging showed that TBK1-knockdowned MCF-7 cells lost E-cadherin expression and acquired vimentin expression (Fig. 3E). Taken together, these results indicate that the loss of TBK1 expression induces a morphologic change from an epithelial- to mesenchymal-like phenotype.

EMT is associated with the increase of proliferative and tumorigenic potential. We examined whether tumor progression was increased when endogenous TBK1 expression was depleted in ERα-positive or -negative breast cancer cells. TBK1 knockdown increased cell growth in ERα-positive breast cancer cells but not in ERα-negative breast cancer cells (Fig. 4A). To test whether knockdown of TBK1 affected the clonogenic potential, we conducted the foci formation assay. TBK1 knockdown significantly increased foci formation in ERα-positive breast cancer cells compared with control cells (Fig. 4B). Next, to investigate the effects of depletion of endogenous TBK1 in vivo, we injected MCF-7 cells, stably expressing either TBK1-specific shRNA or control shRNA into flank of SCID mice, subcutaneously. Knockdown of TBK1 increased the ability of tumor formation of MCF-7 cells, suggesting that EMT induced by TBK1 knockdown in ERα-positive breast cancer cells enhanced the tumorigenicity in vivo (Fig. 4C). As EMT is generally associated with enhanced cell motility and invasion, we also examined cell migration and invasion in TBK1-knockdowned ERα-positive or -negative breast cancer cells. Cell migration and invasion were significantly increased in TBK1-knockdowned ERα-positive breast cancer cells (Fig. 4D and Supplementary Fig. S1; Fig. 4E and Supplementary Fig. S2). To examine whether knockdown of TBK1 enhances metastatic activity in vivo in ERα-positive breast cancer cells, we injected TBK1-knockdowned MCF-7 or control cells into tail veins of SCID mice. It is known that MCF-7 cell line does not metastasize. However, lung colonization was observed in mice injected with TBK1-knockdowned MCF-7 cells (Fig. 4F). Taken together, these results suggest that EMT induced by knockdown of TBK1 expression enhances tumor growth and lung metastasis in ERα-positive breast cancer cells.

Recently, it has been found that ERα silencing induces EMT by regulating the expression of SLUG, an E-cadherin repressor, in ERα-positive breast cancer cells (17). As shown in Fig. 3, we observed the induction of EMT through the loss of TBK1 expression in ERα-positive breast cancer cells (ZR-75B and MCF-7). We explored the possibility that TBK1 might regulate the expression of ERα. TBK1 knockdown markedly decreased the expression of ERα at both the protein and mRNA levels in ERα-positive breast cancer cells (Fig. 5A). We also confirmed the expression of TBK1 and ERα in TBK1-knockdowned ZR-75B cells using immunofluorescence. As expected, cells with intact TBK1 expression had normal ERα expression, whereas TBK1-knockdowned cells exhibited reduced ERα expression (Fig. 5B). The proteasomal degradation pathway has previously been shown to play an important role in the ERα protein turnover (33). To determine whether the reduced expression levels of the ERα protein in TBK1-knockdowned cells is related to proteasomal degradation, we examined the level of ERα protein expression in TBK1-knockdowned MCF-7 cells after treatment with the proteasome inhibitor MG132. In the control cells, MG132 treatment slightly increased the stability of the ERα protein by inhibiting the proteasomal degradation pathway. However, the reduced level of ERα protein expression in TBK1-knockdowned MCF-7 cells was not rescued by MG132 treatment (Supplementary Fig. S3). Next, we sought to confirm that the TBK1 knockdown contributed directly to the decrease in the expression of ERα target genes. We observed that the knockdown of TBK1 inhibited the expression of pS2, GREB1, and PGR in ZR-75B cells and of GREB1 and PGR in MCF-7 cells (Fig. 5C). A number of transcription factors, including ZEB1, TWIST, SNAIL1, and SLUG, regulate EMT. To investigate whether knockdown of TBK1 gene expression modulates expression of EMT-inducing transcription factors, we examined expression of ZEB1, TWIST, SNAIL1, and SLUG, master EMT transcription factors, by immunoblotting in TBK1-knockdowned ERα-positive breast cancer cells. TBK1 knockdown induced SLUG expression, but it did not affect expression of other master EMT transcription factors (Fig. 5D). These data indicate that the loss of TBK1 expression induces EMT through downregulation of ERα expression at the transcriptional level.

The transcription of the ERα gene is regulated by FOXM1, FOXO3A, and GATA3; ERα also regulates its own expression (33–36). Recently, the FOXO3A transcription factor has been shown to translocate from the nucleus to the cytoplasm after its phosphorylation by AKT, suggesting that the cytoplasmic localization of FOXO3A may be associated with the reduced expression of the ERα gene (37, 38). To determine whether FOXO3A and AKT were phosphorylated in response to TBK1 silencing, we carried out immunoblotting experiments. Knockdown of TBK1 in ZR-75B or MCF-7 cells induced the phosphorylation of FOXO3A and AKT (Fig. 6A). We next confirmed the subcellular distribution of endogenous FOXO3A by immunofluorescence staining in TBK1-knockdowned ZR-75B cells. As shown in Fig. 6B, endogenous FOXO3A was observed in both the cytoplasm and the nucleus of control cells but was predominantly detected in the cytoplasm of TBK1-knockdowned ZR-75B cells, indicating that phosphorylated FOXO3A was translocated from the nucleus to the cytoplasm. Recently, the ectopic expression of FOXO3A has been shown to induce estrogen response element–driven reporter activity and the expression of an ERα target gene by binding the ERα promoter in ZR-75 cells (38). To determine whether the observed repression of ERα by TBK1 knockdown decreased the direct binding of FOXO3A to the ERα gene locus, we conducted ChIP using the anti-FOXO3A antibody, followed by RT-PCR with the DNAs of control cells or TBK1-knockdowned ZR-75B cells. Because 2 proximal promoters (promoter A region, −1,168 to +190; promoter B region, −3,713 to −1,864) of ERα have been identified as functional in breast cancer cells (39), we designed primers to amplify putative FOXO3A-binding sites in the 4-kb ERα promoter region (Fig. 6C and Supplementary Table S2). In control cells, the −350 to −111, −624 to −251, −1,258 to −1,032, 1,834 to −1,534, or −2,208 to −1,907 promoter regions did not interact with FOXO3A, whereas the −3,219 to −2,951 promoter regions did interact with FOXO3A, indicating that endogenous FOXO3A protein was recruited to the ERα promoter. Notably, when TBK1 was knocked down in ZR-75B cells, the −3,219 to −2,951 promoter regions were no longer precipitated with the FOXO3A antibody (Fig. 6D). Thus, these results indicate that the retention of nuclear FOXO3A by TBK1 is important in regulating the transcriptional activation of the ERα promoter in ERα-positive breast cancer cells.

Breast cancer subtypes have been classified on the basis of their expression of certain molecules, such as ER, progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), which correlate with different tumor sizes, histologic grades, and incidences of lymph node metastasis (32). ERα-positive breast tumors tend to be benign and respond to hormonal manipulation, whereas ERα-negative breast tumors are more aggressive, poorly differentiated, and hormonally unresponsive (6). The hormonal receptor molecules expressed in these cells represent a widely applicable target for breast cancer treatment, which may help to explain the prognosis of patients with breast cancer subtypes. Our results here suggest that the loss of TBK1 induces EMT by downregulating the expression of ERα in ERα-positive breast cancer cells. Our analysis of different breast tumor subtypes showed that TBK1 was significantly overexpressed in ERα-positive breast tumors (luminal-type) compared with the ERα-negative tumors (basal-type). The loss of TBK1 led to morphologic changes from epithelial- to mesenchymal-like cell shape and enhanced tumor growth and lung metastasis by inhibiting the expression of ERα in ERα-positive breast cancer cells. Our findings provide evidence that TBK1 expression is correlated with breast tumor subtype with respect to ERα expression status.

IKKϵ and TBK1, the regulators of antiviral response, are noncanonical IKK family members that promote oncogenesis. For instance, IKKϵ overexpression contributes to the tumorigenic phenotype in breast and ovarian cancers (23). Knockdown of IKKϵ sensitizes breast cancer cells to tamoxifen-induced growth arrest, whereas the ectopic expression of IKKϵ exerts the opposite effect through interaction with ERα in breast cancer cells (24). In particular, TBK1 has been validated as a potential therapeutic target for increasing synthetic lethality in K-Ras–driven cancers through its relationship with mutant K-Ras (26, 27, 40, 41). Although IKKϵ has been extensively examined in breast cancer cells, the relationship between TBK1 and ERα expression status in different breast cancer subtypes remains to be elucidated. Because IKKϵ and TBK1 share similar kinase domains and functions, we hypothesized that TBK1 might also be related to ERα status in breast cancer, similar to IKKϵ. Surprisingly, TBK1 expression was closely correlated with ERα status in public microarray datasets of different breast cancer subtypes. TBK1 was overexpressed in human luminal breast cancers and ERα-positive breast cancer cells, whereas it was underexpressed in human basal-like breast cancers and ERα-negative breast cancer cells. Moreover, we found that the loss of TBK1 markedly changed the morphology of breast cancer cells to a more aggressive and basal-like phenotype. TBK1 knockdown in ERα-positive breast cancer cells also changed several hallmarks of EMT, including decreased E-cadherin expression and increased vimentin expression, along with a more aggressive phenotype characterized by increased cell migration and invasiveness. Further supporting this hypothesis, immunohistochemical experiments showed that TBK1 expression was higher in ERα-positive luminal breast tumor tissue than in ERα-negative basal-like breast tumor tissue. Thus, TBK1 expression may be a marker of human breast cancer subtype, possibly depending on their ERα expression status.

The majority of the studies of EMT in breast cancers have found correlations between aggressive phenotype, ERα status, and the expression of EMT hallmarks such as E-cadherin or vimentin expression. Recently, it was reported that ERα signaling regulates E-cadherin and EMT through the transcriptional repression of SLUG, an E-cadherin repressor (17). The siRNA-mediated knockdown of ERα caused changes in morphology, as well as increased cell motility and invasiveness, in MCF-7 human breast cancer cells; it also led to tamoxifen resistance, which drives transdifferentiation within each subtype from an EMT phenotype, thereby aggravating the invasiveness in breast cancers (18, 19). In TBK1-knockdowned ERα-positive breast cancer cells, the expression of the ERα gene was significantly decreased, and the ability of ERα to suppress SLUG expression was also markedly attenuated. Thus, the loss of TBK1 induced EMT by reducing the expression of the ERα gene. Guo and Sonenshein (38) reported that FOXO3A, an ERα transcription factor, induced endogenous ERα target gene expression by inducing ERα expression in human breast cancer cells. We found that the knockdown of TBK1 did not affect the expression level of the FOXO3A gene but rather induced the phosphorylation of the FOXO3A protein and the translocation of FOXO3A to the cytoplasm from the nucleus, decreasing the expression of ERα genes. Thus, it is likely that TBK1 suppresses the phosphorylation of FOXO3A to maintain the expression of the ERα gene in the nucleus. It was previously known that the phosphorylation of FOXO3A by AKT resulted in FOXO3A retention in the cytoplasm and the inhibition of transcription of FOXO3A target genes (38). Using a phosphoarray including 46 specific protein kinases and substrates, we found that loss of TBK1 induced the phosphorylation of AKT in ERα-positive breast cancer cells (Supplementary Fig. S4). Other studies have reported that AKT was a bona fide substrate of TBK1 that mediated TBK1-dependent signaling in oncogenesis, but these studies investigated mainly ERα-negative cells and other cancer cells, not ERα-positive breast cancer cells (26, 27). Therefore, AKT activation may not depend on TBK1 in ERα-expressing cells; this implies the involvement of an indirect pathway. Further studies will be needed to better understand this observation.

In conclusion, we have shown that the loss of TBK1 induced EMT, inhibiting the expression of the ERα gene through the phosphorylation of FOXO3A. Because breast cancer subtype is strongly associated with the ERα expression status, TBK1 may be useful as a prognostic marker of breast cancer subtype.

No potential conflicts of interest were disclosed.

Conception and design: S.-J. Kim, K.-M. Yang

Development of methodology: S.-J. Kim, K.-M. Yang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K.-M. Yang, J. Jeong

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.-M. Yang

Writing, review, and/or revision of the manuscript: S.-J. Kim, K.-M. Yang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K.-M. Yang, Y. Jung, J.-M. Lee, W. Kim, J.K. Cho

Study supervision: S.-J. Kim

The authors thank J. Gim and T. Park of Seoul National University for analyzing public microarray datasets.

This work was supported by the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (2009-0081756).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.