SOX4 Induces Epithelial–Mesenchymal Transition and Contributes to Breast Cancer Progression

Published OnlineFirst July 11, 2012; DOI: 10.1158/0008-5472.CAN-12-1045

Cancer Tumor and Stem Cell Biology Research

Jianchao Zhang1, Qian Liang2, Yang Lei1, Min Yao3, Lili Li2, Xiaoge Gao1, Jingxin Feng2, Yu Zhang1, Hongwen Gao3, Dong-Xu Liu4, Jun Lu1, and Baiqu Huang2

Abstract Epithelial–mesenchymal transition (EMT) is a developmental program, which is associated with breast cancer progression and metastasis. Here, we report that ectopic overexpression of SOX4 in immortalized human mammary epithelial cells is sufficient for acquisition of mesenchymal traits, enhanced cell migration, and invasion, along with epithelial stem cell properties defined by the presence of a CD44high/CD24low cell subpopulation. SOX4 positively regulated expression of known EMT inducers, also activating the TGF-b pathway to contribute to EMT. SOX4 itself was induced by TGF-b in mammary epithelial cells and was required for TGF-b–induced EMT. Murine xenograft experiments showed that SOX4 cooperated with oncogenic Ras to promote tumorigenesis in vivo. Finally, in clinical specimens of human breast cancer, we found that SOX4 was abnormally overexpressed and correlated with the triple-negative breast cancer subtype (ER /PR /HER2 ). Our findings define an important function for SOX4 in the progression of breast cancer by orchestrating EMT, and they implicate this gene product as a marker of poor prognosis in this disease. Cancer Res; 72(17); 4597–608. 2012 AACR.

Introduction to act as molecular triggers of the EMT in breast cancer and Breast cancer is the most common cancer in women world- other cancers (4, 6). In addition to these transcription factors, wide (1). Tumor metastasis is the leading cause of mortality an EMT program can also be initiated by several signal path- b associated with cancer including breast cancer (2, 3). Epithe- ways, including TGF- , Wnt, Notch, and Hedgehog (7); among b fi lial–mesenchymal transition (EMT), an early embryonic devel- them, the TGF- signaling plays a signi cant role in contrib- opment program in which cells convert from the epithelial to uting to the initiation of EMT during embryonic development the mesenchymal state, plays a pivotal role during malignant and cancer pathogenesis (8). Recent studies have shown that tumor progression and metastasis. During an EMT, epithelial EMT endows cancer cell with stem cell-like properties and cells shed their characteristics such as loss of cell–cell contacts, relates to the highly aggressive basal like breast cancer subtype apical-basal polarity, and downregulation of epithelial mar- (9, 10). kers, followed by acquisition of mesenchymal features includ- SOX4, a member of the C subgroup of SRY-related HMG box ing enhancement of motility and invasiveness, reorganization (SOX) transcription factor family that are structurally charac- of cytoskeleton, elevated resistance to apoptosis, and increased terized by a highly conserved HMG box domain that directly expression of ECM components like fibronectin and MMPs binds the minor groove of DNA helix, has been implicated in (4, 5). various development processes, such as embryonic cardiac, Several developmentally important transcription factors central nervous system, lymphocyte development, and differ- – such as Snail, Twist, FOXC1, FOXC2, and LBX1 were reported entiation, through its transcriptional activity (11 16). Besides functioning as a transcriptional factor in regulation of development processes, SOX4 has been implicated in cancer pro- Authors' Affiliations: 1The Institute of Genetics and Cytology; 2The Key gression. Increased SOX4 expression is observed in many Laboratory of Molecular Epigenetics of Ministry of Education (MOE), human tumors, including medulloblastoma, small cell lung Northeast Normal University; 3Department of Pathology, The Second Hospital of Jilin University, Changchun, China; and 4The Liggins Institute, cancer, prostate cancer, and breast cancer (17–21). Both TGF-b the University of Auckland, Auckland, New Zealand and Wnt signal pathways have been implicated in cancer J. Zhang and Q. Liang contributed equally to this work. progression and EMT (4, 5). TGF-b has been shown to induce Corresponding Authors: Baiqu Huang, The Key Laboratory of Molecular SOX4 expression in glioma-initiating cells and glioblastoma Epigenetics of Ministry of Education (MOE), Northeast Normal University, multiforme (22, 23). SOX4 has been shown to activate the Changchun 130024, China. Phone: 86-431-85099798; Fax: 86-431- canonical Wnt signal pathway in cancer cells (24, 25). There- 85099768; E-mail: [email protected]; and Jun Lu, The Institute of Genetics and Cytology, Northeast Normal University, 5268 Renmin fore, we speculate that SOX4 might play a role in breast cancer Street, Changchun 130024, China. Phone: 86-431-85098729; Fax: 86- progression and aggression through induction of an oncogenic 431-85099768; E-mail: [email protected] EMT. doi: 10.1158/0008-5472.CAN-12-1045 In this study, we showed that SOX4 was able to activate 2012 American Association for Cancer Research. EMT program in epithelial cells, increase the number of

www.aacrjournals.org 4597

Zhang et al.

CD44high/CD24low population and potentiate mammo- incubation with secondary antibodies (DAKO) for 30 min- sphere-forming ability. We also showed that in mammary utes. The immunostaining was developed with 3.30-diami- epithelial cells, SOX4 cooperated with the activated onco- nobenzidine. Finally, sections were counterstained with genic Ras to endow the cells with the tumorigenicity in vivo. hematoxylin. We further showed that human triple-negative breast cancer (TNBC) has increased SOX4 expression. These data impli- Transwell migration and invasion assays cate a novel role of SOX4 in inducing EMT in breast cancer Experiments were carried out as described in Supplemen- progression and its close association with invasive subtypes tary Materials and Methods. of human breast cancer. Mammosphere formation assays Materials and Methods Mammosphere assays were carried out as described else- Cell culture where (27) with minor modifications. Single cells were plated MCF10A, MDA-MB-231, BT549, NMuMG, MDCK, and at 10,000 cells/mL on 6-well ultra-low attachment plates 293T cell lines were obtained from the American Type (Corning) in serum-free DMEM/F12 supplemented with 20 Culture Collection, where they were characterized by ng/mL bFGF, 20 ng/mL EGF, 4 mg/mL insulin, 4 mg/mL DNA-fingerprinting and isozyme detection. Cells were heparin, 1 mg/mL hydrocortisone, 0.4% BSA and B27. Fresh immediately expanded and frozen such that they could be medium was supplemented every 3 days. The mammospheres revived every 3 to 4 months. MCF10A cells were cultured as were counted at day 14. previously described (26) in DMEM/F12 supplemented with 5% horse serum, 20 ng/mL EGF, 0.5 mg/mL hydrocortisone, Flow cytometry 100 ng/mL cholera toxin, 10 mg/mL insulin and pen/strep. A total of 1 106 cells were resuspended in 100 mL PBS MDCK and 293T cells were cultured in DMEM containing containing 2% FBS (FACS buffer), and then incubated on ice for 10% FBS (Hyclone). NMuMG cells were maintained in 10 minutes. CD44-APC and CD24-PE (BD Biosciences) were DMEM supplemented with 10% FBS and 10 mg/mL insulin. added to cell suspension and incubated on ice for 30 minutes. MDA-MB-231 cells were cultured in Leibovitz's L-15 medium Cells were washed and resuspended in 500 mL FACS buffer with 10% FBS at 37 CwithoutCO2. BT549 cells were and analyzed using a FACS Calibur Flow Cytometer (BD cultured in RPMI-1640 medium supplemented with 10% Biosciences). FBS and 0.023 IU/mL insulin. Xenograft mouse experiments Antibodies and reagents A total of 5 106 cells in 100 mL PBS were injected The antibodies and reagents are listed in Supplementary subcutaneously into 6-week-old female BALB/c nude mice. Materials and Methods. Five mice per group were used in each experiment. Tumor volume was measured weekly using a Vernier caliper and Immunoblotting calculated according to the formula: p/6 length width2. Standard procedures for immunoblotting are described in Eight weeks later, the mice were sacrificed, and tumors were Supplementary Materials and Methods. collected and photographed. All animal experiments were approved by the Animal Care Committee of the Northeast Immunofluorescence Normal University, China. Experiments were carried out as described in Supplemen- tary Materials and Methods. Luciferase reporter assays The protocol is described in Supplementary Materials and Plasmids, viral production, and infection of target cells Methods. A description of procedures are detailed in Supplementary Materials and Methods. Cell proliferation assays Cell growth rates were assessed by the MTT assay as Human breast tumor specimens and described in Supplementary Materials and Methods. immunohistochemistry Human breast tumor specimens were obtained from the RT-PCR and real-time PCR analysis Second Hospital of Jilin University, China. Tissue samples Experiments were carried out as described in Supplemen- were fixed in 10% formaldehyde and embedded in paraffin. tary Materials and Methods. Sections were cut and stained using a conventional immunohistochemistry procedure. Briefly,thetissuesectionswere Statistical analysis deparaffinized and rehydrated. Antigen retrieval was done Data are presented as mean SD. The Student t test (2- by heating the sample in 10 mM citrate buffer (pH 6.0) at tailed) was used to determine statistically the significance of 95C for 15 minutes. Endogenous peroxidase activity was differences between groups. P < 0.05 was considered statisti- blocked with peroxidase (DAKO), and sections were blocked cally significant. SOX4 expression intensities in human breast with 10% goat serum. Sections were incubated with cancer samples were analyzed by c2 test. Statistical analysis anti-SOX4 antibody (Abcam, 1:50) for 1 hour, followed by was carried out using the SPSS17.0 software.

4598 Cancer Res; 72(17) September 1, 2012 Cancer Research

SOX4 Promotes EMT and Breast Cancer Progression

Results promoters of these 2 EMT inducers, we transiently transfected Exogenous SOX4 expression induced EMT in human 293T cells with luciferase reporter constructs containing prox- mammary epithelial cells imal promoter of ZEB1 or Snail. We found that the relative To investigate the role of SOX4 in EMT, we stably over- luciferase activity of ZEB1 was increased proportionally to expressed SOX4 in the immortalized normal human mammary the increasing amount of SOX4 in 293T cells (Supplementary epithelial cell line MCF10A that lacks endogenous SOX4 Fig. S4B). However, the Snail promoter activity was not chan- expression, by using retroviral infection, as confirmed by ged (Supplementary Fig. S4C). These findings suggest that immunoblotting (Fig. 1A) and real-time PCR (Supplementary E-cadherin transcriptional inactivation during SOX4-induced Fig. S1). We observed that MCF10A cells transfected with EMT may be a result of induction of ZEB1 expression, a well- vector retained their cobblestone-like morphology with tight known upstream repressor of E-cadherin. cell–cell adhesion, whereas cells expressing exogenous SOX4 We next carried out a loss-of-function assay to further study displayed an elongated fibroblast-like morphology with scat- the role of SOX4 in invasive mesenchymal breast cancer cells tered distribution in culture (Fig. 1B). We then examined both MDA-MB-231 and BT549. The effect of SOX4 silencing on the epithelial and mesenchymal markers by immunoblotting (Fig. invasive ability of these cells was investigated using transwell 1C) and immunofluorescence (Fig. 1D). As can be seen, the invasion assays. The results indicated that SOX4 knockdown SOX4-expressing MCF10A cells exhibited a significant down- prominently impaired the migration and invasion ability of regulation of b-catenin and complete loss of E-cadherin and these 2 breast cancer cells, but it did not affect cell proliferation occludin from cell–cell contacts; meanwhile the mesenchymal and apoptosis (Supplementary Fig. S5A–G). Moreover, sup- markers fibronectin, vimentin, and N-cadherin were dramat- pression of SOX4 expression resulted in vimentin downregula- ically upregulated. Real-time PCR analyses also revealed the tion in both cell lines (Supplementary Fig. S5A and S5B). expression of E-cadherin mRNA and concomitant induction However, no detectable changes were observed in other EMT of N-cadherin, fibronectin, and vimentin mRNAs in SOX4- markers (data not shown). In addition, downregulation of expressing MCF10A cells (Fig. 1E). These morphologic and SOX4 did not cause significant cellular morphologic changes molecular changes suggested an apparent transition of the in both MDA-MB-231 and BT549 cells (Supplementary Fig. SOX4-expressing MCF10A cells from an epithelial to mesen- S5H). Thus, our loss-of-function study suggested that the chymal status. To further probe the possible interactions suppression of SOX4 could partially reverse the EMT pheno- between SOX4 and other EMT-inducing transcription factors, type of MDA-MB-231 and BT549 cells. we examined the expression of other known EMT inducers. We showed that the endogenous mRNA levels of Snail1, FOXC2, SOX4-mediated EMT generated stem cell-like cells GSC, SIX1, Twist1, HOXB7, and ZEB1 were elevated in response Mammary epithelial cells undergoing an EMT program have to SOX4 overexpression, to a variable extent, whereas the been linked to stem cell phenotypes such as an increased ZEB2 mRNA level exhibited no detectable change (Fig. 1F). CD44high/CD24low population and mammospheres formation We also ectopically overexpressed other EMT inducers Snail ability (9). To determine whether SOX4 has the effect to lead to and Twist1 in MCF10A cells. We observed that Snail and the stem cell phenotypes upon induction of EMT, we carried Twist1 induced EMT in MCF10A cells (Supplementary Fig. out FACS to identify CD44high/CD24low populations. We S2A), and upregulated SOX4 mRNA expression in MCF10A cells observed that the SOX4-expressing MCF10A cells exhibited (Supplementary Fig. S2B). Typically, the EMT phenotype is a significant increase in the CD44high/CD24low stem cell pop- usually accompanied by the acquisition of cell traits such as ulation compared with vector-infected cells (Fig. 2A). Mean- greater migration and more invasive ability. As shown in Fig. while, as evidenced in Fig. 2B and C, the SOX4-expressing 1G and H, SOX4-expressing MCF10A cells dramatically MCF10A cells increased both in size and in number of mammo- increased their migratory and invasive behaviors. Similar spheres in comparison with the vector-infected cells. We thus results were observed in MDCK cells, a prototypic cell model concluded that the SOX4-induced EMT generates mesenchy- for EMT study (Supplementary Fig. S3A–E). Together, these mal cells with stem cell-like phenotypes, a feature recently results show that SOX4 is a novel inducer of EMT and it defined for EMT inducers (6, 28). promotes cell migration and invasion in MCF10A cells. Loss of E-cadherin seems to be a critical event in EMT. We SOX4 cooperated with activated oncogenic Ras to next carried out a luciferase reporter assay to investigate promote tumorigenesis whether SOX4 could transcriptionally regulate the E-cadherin Although SOX4 was able to induce EMT, MCF10A cells expression. The results showed that the relative luciferase expressing SOX4 were unable to form tumors when injected activity of E-cadherin was not changed in 293T cells transiently into nude mice (Supplementary Fig. S6A), suggesting that SOX4 cotransfected with E-cadherin promoter reporter together itself lacks the ability to induce neoplastic transformation. It with SOX4 expression construct or with empty vector (Sup- has recently been reported that EMT inducers Twist1, Twist2, plementary Fig. S4A), indicating that SOX4 may indirectly and LBX1 cooperate with activated oncogenic Ras to promote regulate E-cadherin transcription. In line with these observa- even more dramatic characteristics of EMT and enhance tions, SOX4 increased the expression of known EMT inducers tumorigenesis (6, 29). To investigate this possible cooperation, (including Snail and ZEB1) that act to directly repress E- we coexpressed SOX4 with oncogenic H-RasV12 in MCF10A cadherin transcription, during SOX4-induced EMT (Fig. 1F). cells. As shown in Fig. 3A and Supplementary Fig. S6B, coex- To test the effect of SOX4 on the transcription activity of pression of SOX4 and oncogenic Ras led to a more dramatic

www.aacrjournals.org Cancer Res; 72(17) September 1, 2012 4599

Zhang et al.

Figure 1. Ectopic expression of SOX4 in MCF10A cells induced an EMT program. A, immunoblotting assessment of the ectopic SOX4 protein expression after retroviral infection in MCF10A cells. B, morphologic change of MCF10A cells expressing SOX4 or empty vector. Scale bar, 100 mm. C, immunoblotting analysis of expression of the epithelial markers E-cadherin, b-catenin, and occludin, and the mesenchymal markers fibronectin, vimentin, and N-cadherin. D, immunofluorescence staining for the epithelial and mesenchymal markers. Scale bar, 100 mm. E, the expression of E-cadherin, N-cadherin, fibronectin, and vimentin mRNAs were assessed by real-time PCR. F, mRNA expression levels of known EMT inducers were assessed by real-time PCR. Error bars represent the mean SD of triplicate experiments. G and H, migration (1 d; G) and invasion (2 d; H) assays in stable MCF10A cells. The mean was derived from cell counts of 5 fields, and each experiment was repeated 3 times (, P < 0.001, compared with the control). Representative images of migrated and invaded cells are shown.

morphologic change, characterized by prominently elongated in other epithelial or mesenchymal markers were observed spindle-shaped cells. In addition, coexpression of both SOX4 (Fig. 3B). Furthermore, MCF10A cells expressing SOX4þH- and H-RasV12 triggered a further reduction of b-catenin, RasV12 exhibited higher migration (Fig. 3C) and invasive relative to either SOX4 or H-RasV12 alone; however, no changes (Fig. 3D) ability than cells expressing either SOX4 or H-RasV12

4600 Cancer Res; 72(17) September 1, 2012 Cancer Research

SOX4 Promotes EMT and Breast Cancer Progression

tochemistry staining of SOX4 in 149 human breast tumor samples representing different subtypes and several normal human breast tissues. Seventy-two percent (107 of 149) of breast cancer samples exhibited positive SOX4 immunoreac- tivity. Among the 107 SOX4 positive tumor samples, 46 (43%) cases showed strong nuclear staining, and 34 (32%) moderate nuclear staining, whereas nonneoplastic breast tissues only had faint nuclear staining of SOX4 (Fig. 4A–E). Further analysis revealed that high level of SOX4 expression occurred in 56.9% of III grade, 40.9% of PR and 38.6% of ER breast cancer samples. þ þ In contrast, 14.3% of I/II grade, 16.4% of PR and 19.7% of ER breast cancer samples showed high expression of SOX4 (Fig. 4F). These data showed that SOX4 overexpression was significantly correlated with several poor prognostic parameters such as high tumor grade malignancy (P < 0.001), negative ER (P ¼ 0.014), and negative PR (P ¼ 0.001; Table 1). Moreover, we observed that high level of SOX4 expression was associated with the highly aggressive TNBC. As shown in Fig. 4F, 45.7% of TNBC samples showed abnormally high SOX4 expression compared with that of 24.3% in non-TNBC samples (P ¼ 0.009). To further determine the relevance of SOX4 expression with breast cancer progression, we carried out an extensive analysis of the expression profile of SOX4 using the Oncomine cancer microarray database. Increased expression of SOX4 mRNA in mammary carcinoma in contrast to normal breast tissues was observed in 15 of 19 microarray studies (Supplementary Table S1). Further analyses revealed that the expression levels of SOX4 were also significantly correlated to higher tumor grades and stages; and increased SOX4 expression predicted an unfavorable Figure 2. SOX4-induced EMT generated stem cell-like cells. A, FACS clinical outcome. In particular, among all of the 12 micro- analysis of cell-surface markers CD44 and CD24 in MCF-10A cells array studies in Oncomine data, SOX4 was significantly high expressing SOX4 or empty vector. Percentages of mean CD44 / upregulated in unfavorable ER /PR /HER2 triple-negative CD24low subpopulation SD based on triplicate experiments are indicated. B, phase contrast images of mammospheres formation. Scale basal-like subtype (Supplementary Table S2). These obser- bar, 100 mm. C, quantification of mammosphere numbers formed from vations implicate the potential usefulness of the aberrant 3 independent experiments (error bar, mean SD; , P < 0.05, compared high SOX4 expression as a novel prognostic molecular with the control). marker for TNBC. alone. Meanwhile, all these cells exhibited roughly the same Activation of TGF-b was necessary for SOX4-induced cell growth rates with no statistically significant differences (Sup- motility plementary Fig. S6C). These data implicated that SOX4 and H- Several lines of evidence have implicated the involvement RasV12 worked synergistically in inducing a more prominent of TGF-b in EMT process both in embryonic development EMT phenotype. and breast cancer progression (8, 30). We next intended to To further establish whether SOX4 is able to trigger tumor- identify whether TGF-b signaling is activated in SOX4- igenesis in vivo, we tested its effect in a xenograft mouse model. induced EMT. Our real-time PCR revealed an increased We detected no primary tumor growth 8 weeks after the nude expression of TGF-b1 and TGF-b2 mRNAs in SOX4-expres- mice were subcutaneously injected with MCF10A cells stably sing MCF10A cells (Fig. 5A). In addition, the level of phos- expressing SOX4 alone. Meanwhile, MCF10A cells expressing phorylated Smad2 protein, a downstream effector of TGF-b oncogenic H-RasV12 alone produced small palpable tumors. pathway, was significantly increased in SOX4-expressing Interestingly, coexpression of both SOX4 and H-RasV12 MCF10A cells (Fig. 5B left). Moreover, SOX4 silencing effi- resulted in rapid development of very large tumors (Fig. 3E ciently decreased the level of phosphorylated Smad2 and the and F). These results clearly indicate that SOX4 and oncogenic expression of TGF-b1 and TGF-b2 mRNAs in SOX4-expres- H-RasV12 work synergistically to induce tumorigenesis in vivo. sing MCF10A cells (Fig. 5C); whereas knockdown of SOX4 upregulated E-cadherin expression and partly downregu- High expression of SOX4 was correlated with invasive lated N-cadherin expression, presumably due to the residual breast cancer and highly aggressive TNBC ectopic SOX4 (Supplementary Fig. S7A). Consistent with To evaluate the clinical relevance of SOX4 expression in these changes, we observed some tight cell clusters upon human mammary carcinomas, we carried out immunohis- SOX4 silencing (Supplementary Fig. S7B). Apparently, these

www.aacrjournals.org Cancer Res; 72(17) September 1, 2012 4601

Zhang et al.

Figure 3. SOX4 cooperated with activated oncogenic Ras to promote tumorigenesis. MCF10A cells were infected with empty vector, retroviral vector encoding SOX4, or sequentially infected with retroviral vector encoding H-RasV12 (Ras) and SOX4 as indicated. A, phase contrast images of the cell morphology. Scale bar, 100 mm. B, immunoblotting analysis of expression of the epithelial and mesenchymal markers. C and D, migration (1 d; C) and invasion (2 d; D) assays. The mean was derived from cell counts of 5 ﬁelds, and each experiment was repeated 3 times. Representative images of migrated or invaded cells are also shown. E, stable MCF10A cells were subcutaneously injected into BALB/c female nude mice (n ¼ 5 for each experimental group). F, individual tumor volume was measured according to the formula: p/6 length width2 at week 8 after injection.

experiments pointed to a reinforced TGF-b signaling upon reduced their migration and invasive ability (Fig. 5D and E). ectopic SOX4 expression in MCF10A cells. To further vali- These results suggest that the intensiﬁed TGF-b signaling date that TGF-b signaling is responsible for the SOX4- induced by SOX4 promotes cell motility, and this partly induced EMT and the enhanced cell motility, we used a contributes to EMT. speciﬁcTGF-b receptor kinase inhibitor SB431542 to block TGF-b signaling has been shown to be able to induce EMT in the TGF-b signaling in SOX4-expressing MCF10A cells. We many epithelial cells including MCF10A and NMuMG (31, 32). found that suppression of TGF-b signaling by the inhibitor In addition, recent studies have shown that SOX4 is a direct reduced Smad2 phosphorylation level and downregulated target gene of TGF-b in glioma-initiating cells and is also vimentin expression (Fig. 5B right), without affecting the responsive to TGF-b in glioblastoma multiforme (22, 23). Here, morphologic and molecular features of SOX4-expressing we showed that SOX4 mRNA was also induced in MCF10A MCF10A cells undergoing EMT (data not shown). Moreover, (Fig. 6A and B) and NMuMG (Supplementary Fig. S8) cells, in a treatment of SOX4-expressing MCF10A cells with SB431542 dose- and time-dependent manner upon the addition of TGF-b

4602 Cancer Res; 72(17) September 1, 2012 Cancer Research

SOX4 Promotes EMT and Breast Cancer Progression

Figure 4. SOX4 expression was associated with invasive subtypes of human breast cancer. Representative images of the immunohistochemical staining of SOX4 in breast cancer samples. A, normal human breast tissue. B, negative tumor staining. C, weak nuclear staining. D, moderate nuclear staining. E, strong nuclear staining. Scale bar, 100 mm. F, percentages of human breast cancer samples with high level of SOX4 expression in respective tumor subtypes and in different tumor grades. Corresponding P values analyzed by c2 test are indicated.

to the cell culture medium. To investigate whether SOX4 is MCF10A-shSOX4#2 cells (Fig. 6C). These experiments suggest required for TGF-b–induced EMT, we knocked down SOX4 that SOX4 is also involved in and required for TGF-b–induced expression in MCF10A cells with short hairpin RNA (shRNA) EMT. and examined their responses to TGF-b1 treatments. MCF10A cells expressing SOX4 shRNA (MCF10A-shSOX4#2) or nontar- Discussion get control shRNA (MCF10A-shCtrl) were treated with TGF-b1. Increasing evidence suggests that SOX4 overexpression is We observed that SOX4 expression was induced 48 hours after associated with several human cancers including brain, lung, TGF-b1 addition in the MCF10A-shCtrl cells, whereas and breast cancers (17). Speciﬁcally, it has been reported that MCF10A-shSOX4#2 cells exhibited a reduction in the basal SOX4 is expressed in normal breast and breast cancer cells, and expression level of SOX4 as well as in the induction of SOX4 progestin can promote SOX4 expression and induce SOX4- after TGF-b1 stimulation (Fig. 6C). We also found that TGF-b1 mediated transcriptional activity in breast cancer cell lines treatments induced EMT in MCF10A-shCtrl cells, but not in (21). Moreover, miR-335 suppresses breast cancer metastasis MCF10A-shSOX4#2 cells (Fig. 6D). Consistent with morpho- and migration by modulating SOX4 expression (33). A recent logic changes, E-cadherin expression was inhibited and N- analysis of the transcriptional proﬁle of human normal mam- cadherin expression was increased in MCF10A-shCtrl cells mary stem cells (hNMSC), aimed at prediction of biological and after TGF-b1 treatments. However, under the same conditions molecular features of breast cancers, unravels that SOX4 is E-cadherin and N-cadherin expression were not changed in upregulated in PKH26-positive cells that possess all the

www.aacrjournals.org Cancer Res; 72(17) September 1, 2012 4603

Zhang et al.

Table 1. Association between SOX4 expression and tumor subtypes in invasive ductal carcinomas of breast

No. with high-level Percentage with Total no. expression for high SOX4 Tumor type of cases SOX4 expression (%) P Estrogen receptor 149 0.014 Negative 88 34 38.6 Positive 61 12 19.7 Progesterone receptor 149 0.001 Negative 88 36 40.9 Positive 61 10 16.4 ERBB/HER2 149 0.046 Negative 79 30 38.0 Positive 70 16 22.9 Triple negative 149 0.009 Yes 46 21 45.7 No 103 25 24.3 Histologic grade 149 <0.001 Grade III 58 33 56.9 Grade I/II 91 13 14.3

characteristics of hNMSCs (34). These findings implicate the recent study that discovers that overexpression of one of the involvement of SOX4 in breast cancer development. Neverthe- EMT inducers upregulates a subset of other EMT-inducing less, how SOX4 implements its oncogenic effects during breast transcription factors, implicating the interactions among these cancer progression remains unclear. We find in this study that EMT inducers (39). E-cadherin plays a pivotal role in epithelial SOX4 functions as a trigger for EMT to contribute to breast cell–cell adhesion. Functional loss of E-cadherin is considered cancer progression, a previously unreported role of SOX4 in a hallmark of EMT (4, 5). Our results show that SOX4 may breast cancer. Specifically, our data show that SOX4 represses indirectly downregulate the transcription of E-cadherin.Itis the epithelial phenotype, induces the mesenchymal phenotype, reported that EMT inducers such as Twist1, Snail, Slug, ZEB1, and dramatically increases the migration and invasion of ZEB2, FOXC1, and FOXC2 are able to transcriptionally repress MCF10A cells. E-cadherin expression either directly or indirectly (4). Esmer- Besides breast cancer, it has been reported that ectopic alda Casas and colleague show that Twist1 indirectly sup- overexpression of SOX4 in hepatocellular carcinoma endows presses E-cadherin transcription to promote EMT through an antiapoptotic effect and contributes to hepatocarcinogen- directly binding to the Snail2 gene promoter to activate its esis (35). Previous study shows that the epigenetic deregulation transcription (40). We show in this study that SOX4 indirectly of miRNA-129-2 leads to the oncogenic overexpression of SOX4 downregulates the expression of E-cadherin through activating in endometrial cancer. Reactivation of miR-129-2 results in ZEB1 expression. Interestingly, search of the defined region SOX4 downregulation and reduces the proliferation of endo- of ZEB1 promoter with the transcription factor database metrial cancer cells (36). Accumulating evidence indicates that (TRANSFAC) did not reveal a known SOX4 consensus-binding EMT inducers such as Snail and Twist1 control cell prolifer- site, suggesting that SOX4 may not be able to bind directly to ation and survival that are critical in cancer progression (37). the ZEB1 promoter; rather, it may indirectly activate the ZEB1 These results suggest that SOX4 may act in a similar fashion as promoter through as yet unidentified intermediate factor(s). Snail and Twist1, exerting dual effects of potent prosurvival More importantly, our data have suggested that SOX4 is function and orchestrating an EMT in tumor progression. also involved in and required for TGF-b–induced EMT. We Interestingly, SOX4 is also able to promote cell-cycle arrest show in our study that the autocrine TGF-b signaling is and apoptosis in a p53-dependent manner in HCT116 (38). The activated in SOX4-induced EMT; meanwhile TGF-b signaling distinct physiologic functions of SOX4 in different cell lineages induces SOX4 expression in breast epithelial cells and and cancer types highlight the importance of cell context and knockdown of SOX4 blocks TGF-b1–induced EMT. Thus, genetic background in various human cancer cell lines and we speculate that the SOX4-TGFb-SOX4 feedback loop pre- different cancer types that determine the SOX4 function. sumably functions as a novel signaling pathway in EMT Moreover, we have shown in this work that SOX4 upregu- regulation,aswellasinbreast cancer progression. A recent lates the mRNA expression of several EMT-inducing transcrip- study by Christina Scheel and colleague show that autocrine tion factors and SOX4 expression is induced by a large number signaling pathways involving TGF-b andboththecanonical of known regulators of the EMT program, notably the Snail, and noncanonical Wnt signaling are activated in EMT Twist, Ras, and TGF-b1. Our data are in accordance with a program induced by diverse stimuli including transcription

4604 Cancer Res; 72(17) September 1, 2012 Cancer Research

SOX4 Promotes EMT and Breast Cancer Progression

Figure 5. Activation of TGF-b was necessary for SOX4-induced cell motility. A, expression of TGF-b1 and TGF-b2 mRNAs was determined by real-time PCR in MCF-10A cells expressing empty vector or SOX4. B, left, immunoblots of p-Smad2 and total Smad2/3 protein. Right, SOX4- MCF10A cells were treated with 10 mmol/L SB431542 (SB43) for 24 hours. Immunoblots of p-Smad2, total Smad2/3 protein, and vimentin. C, SOX4-MCF10A cells were stably infected with SOX4 shRNA (shSOX4#2) or nontarget vector shRNA (shCtrl). Immunoblotting of SOX4, p-Smad2, and total Smad2/3 protein (left). Real-time PCR analysis of the expression of TGF-b1 and TGF-b2 mRNAs (right). D and E, migration (1 d; D) and invasion (2 d; E) assays of SB431542-treated SOX4- MCF10A cells. The mean was calculated from cell counts of 5 ﬁelds, and each experiment was repeated 3 times (, P < 0.001, compared with the DMSO treated). Representative images of migrated or invaded cells are shown.

factors such as Twist, and they propose that these signalings expression of SOX4 in MCF10A cells increases the CD44high further maintain the EMT program, and disruption of these /CD24low subpopulation and enhances the mammosphere- extracellular autocrine signalings abrogates transcription forming ability, a property of the stem cells. It is reported that factor-induced EMT (41). In support of the involvement of normal and neoplastic breast stem-like cells express some additional EMT-promoting pathways, it has been documen- EMT inducers such as Twist1, Snail1, Snail2, and ZEB1 (9). A ted that SOX4 activates the Wnt/b-catenin pathway in colon recent study by Chen and colleague unravels that MCF-7 cells carcinoma and melanoma (24, 25, 42). Whether the canon- cultured in 3D collagen scaffolds acquire the properties of ical Wnt/b-catenin signaling can be activated in SOX4- CSCs, and the transcriptions of stem cell markers such as induced EMT needs to be clariﬁed. OCT4A and SOX2, and breast cancer stem cell signatures, Recent studies have shown that EMT can induce noncancer including SOX4, JAG1, and CD49F, are signiﬁcantly unregulated stem cells to acquire cancer stem cell (CSC)-like properties in (43). Furthermore, SOX4 is also upregulated in PKH26-positive breast cancer cells that exhibit a CD44high/CD24low antigenic cells that possess all the characteristics of hNMSCs (34). All phenotype (9). In this study, we have shown that ectopic these data are supportive of a close relationship among SOX4,

www.aacrjournals.org Cancer Res; 72(17) September 1, 2012 4605

Zhang et al.

Figure 6. SOX4 was necessary for TGF-b–induced EMT. A and B, SOX4 mRNA expression levels in MCF10A cells treated with activated TGF-b1 at indicated time and concentrations. Error bar represents mean SD of triplicate assays. C, immunoblotting of SOX4, E-cadherin, and N-cadherin after 48-hour treatment of TGF-b1 (10 ng/mL) in MCF10A cells with shCtrl and shSOX4#2. D, morphology of MCF10A cells with shCtrl and shSOX4#2 after TGF-b1 treatment as in C. Scale bar, 100 mm.

EMT, and CSCs. Previous study reports that the CD44high progression factor in breast cancer, rather than a tumor- /CD24low subpopulation breast cancer cells enhance tumori- initiating event for breast carcinogenesis. genicity in a xenograft model (44). However, in our xenograft TNBC is the most aggressive form of breast cancer with high model, MCF10A cells expressing SOX4 alone are unable to form histologic grade, aggressive clinical behavior, high incidence of tumors when injected into nude mice, suggesting that SOX4 brain and lung metastases, and the lack of an effective ther- itself lacks the ability to induce neoplastic transformation, apeutic target (47). The TNBC shares many clinicopathologic although it promotes the generation of stem cell-like cells. and molecular features with basal-like breast cancers (BLBC), a Nevertheless, we have observed a cooperative action between subtype of breast cancer defined by gene-expression profiling, SOX4 and activated oncoprotein Ras in MCF10A cells, resulting and accounts for approximately 70% of the BLBC cases. Recent in the formation of large tumors in nude mice. In line with our studies have linked EMT with the aggressive BLBC (47, 48). observations, a previous work has shown that SOX4 over- Targeting the EMT-like phenotypes would seem to represent expression alone in NIH3T3 cells do not increases the trans- the potential strategies for the development of novel antican- forming ability of the cells, but the effect occurs when cells are cer therapeutics. Our results in this study are in favor of a cotransfected with the weakly oncogenic RHOA-Q63L (45). strong correlation between SOX4 and TNBC, and hence point Thus, our findings provide evidence for the ability of SOX4 to to the prospect of using SOX4 as a novel biomarker for potentiate the oncogenic effect of activated Ras in breast prognosis and diagnosis of TNBC, as well as a potential tumorigenesis. Interestingly, we have found that activated Ras molecular therapeutic target for this highly aggressive breast itself upregulates the endogenous SOX4 expression, but no cancer. synergistic effects between Ras and SOX4 are observed. An Disclosure of Potential Conflicts of Interest earlier study suggests that Ras also induces the expression of No potential conflicts of interest were dislosed. other EMT inducer such as Twist1 in MCF10A cells (46). Our result shows that Ras upregulates Snail1 expression in mam- Authors' Contributions mary epithelial cells (Supplementary Fig. S6B), but these Conception and design: J. Zhang, Q. Liang, Y. Zhang, J. Lu, B. Huang Development of methodology: J. Feng, Y. Zhang inducers do not function cooperatively with Ras either. These Acquisition of data (provided animals, acquired and managed patients, data seem to support the assumption that only the exogenous provided facilities, etc.): J. Zhang, Q. Liang, Y. Lei, M. Yao, L. Li, J. Feng, H. Gao EMT inducers and activated Ras are able to exert the coop- Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Zhang, Q. Liang, Y. Lei, L. Li, Y. Zhang, B. Huang erative effects to promote even more dramatic characteristics Writing, review, and/or revision of the manuscript: J. Zhang, D.-X. Liu, J. Lu, of EMT and cancer progression (6, 9, 29, 41). However, the B. Huang Administrative, technical, or material support (i.e., reporting or orga- mechanisms underlying this phenomenon still remain nizing data, constructing databases): X. Gao, J. Feng, Y. Zhang unknown. Conclusively, SOX4 may function as a key tumor Study supervision: B. Huang

4606 Cancer Res; 72(17) September 1, 2012 Cancer Research

SOX4 Promotes EMT and Breast Cancer Progression

Acknowledgments the Fundamental Research Funds for Central Universities, and the Program for The authors thank Torben rntoft, James M. Wells, Patrick Humbert, R.A. Introducing Talents to Universities (B07017). The costs of publication of this article were defrayed in part by the payment of Weinberg, L. Miguel Martins, Ji-Hshiung Chen, and Daniel A. Haber for providing advertisement the plasmids mentioned in Materials and Methods. page charges. This article must therefore be hereby marked in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Grant Support This work is supported by the grants from the National Natural Science Received March 19, 2012; revised June 17, 2012; accepted June 25, 2012; Foundation of China (Grant numbers 91019011, 31170719, 31071149, 31100998), published OnlineFirst July 11, 2012.

References 1. Bray F, McCarron P, Parkin DM. The changing global patterns of 21. Graham JD, Hunt SM, Tran N, Clarke CL. Regulation of the expression female breast cancer incidence and mortality. Breast Cancer Res and activity by progestins of a member of the SOX gene family of 2004;6:229–39. transcriptional modulators. J Mol Endocrinol 1999;22:295–304. 2. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. 22. Ikushima H, Todo T, Ino Y, Takahashi M, Miyazawa K, Miyazono K. Cell 2011;144:646–74. Autocrine TGF-beta signaling maintains tumorigenicity of glioma-ini- 3. Gupta GP, Massague J. Cancer metastasis: building a framework. Cell tiating cells through Sry-related HMG-box factors. Cell Stem Cell 2006;127:679–95. 2009;5:504–14. 4. Polyak K, Weinberg RA. Transitions between epithelial and mesen- 23. Lin B, Madan A, Yoon JG, Fang X, Yan X, Kim TK, et al. Massively chymal states: acquisition of malignant and stem cell traits. Nat Rev parallel signature sequencing and bioinformatics analysis identifies Cancer 2009;9:265–73. up-regulation of TGFBI and SOX4 in human glioblastoma. PloS One 5. Acloque H, Adams MS, Fishwick K, Bronner-Fraser M, Nieto MA. 2010;5:e10210. Epithelial-mesenchymal transitions: the importance of changing cell 24. Sinner D, Kordich JJ, Spence JR, Opoka R, Rankin S, Lin SC, et al. state in development and disease. J Clin Invest 2009;119:1438–49. Sox17 and Sox4 differentially regulate beta-catenin/T-cell factor activ- 6. Yu M, Smolen GA, Zhang J, Wittner B, Schott BJ, Brachtel E, et al. A ity and proliferation of colon carcinoma cells. Mol Cell Biol 2007;27: developmentally regulated inducer of EMT, LBX1, contributes to 7802–15. breast cancer progression. Genes Dev 2009;23:1737–42. 25. Lee AK, Ahn SG, Yoon JH, Kim SA. Sox4 stimulates b-catenin activity 7. Huber MA, Kraut N, Beug H. Molecular requirements for epithelial- through induction of CK2. Oncology Rep 2011;25:559–65. mesenchymal transition during tumor progression. Curr Opin Cell Biol 26. Debnath J, Muthuswamy SK, Brugge JS. Morphogenesis and onco- 2005;17:548–58. genesis of MCF-10A mammary epithelial acini grown in three-dimen- 8. Xu J, Lamouille S, Derynck R. TGF-beta-induced epithelial to mesen- sional basement membrane cultures. Methods 2003;30:256–68. chymal transition. Cell Res 2009;19:156–72. 27. Dontu G, Abdallah WM, Foley JM, Jackson KW, Clarke MF, Kawamura 9. Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, et al. The MJ, et al. In vitro propagation and transcriptional profiling of human epithelial-mesenchymal transition generates cells with properties of mammary stem/progenitor cells. Genes Dev 2003;17:1253–70. stem cells. Cell 2008;133:704–15. 28. Zeng Q, Li W, Lu D, Wu Z, Duan H, Luo Y, et al. CD146, an epithelial- 10. Sarrio D, Rodriguez-Pinilla SM, Hardisson D, Cano A, Moreno- mesenchymal transition inducer, is associated with triple-negative Bueno G, Palacios J. Epithelial-mesenchymal transition in breast breast cancer. Proc Natl Acad Sci U S A 2012;109:1127–32. cancer relates to the basal-like phenotype. Cancer Res 2008;68: 29. Ansieau S, Bastid J, Doreau A, Morel AP, Bouchet BP, Thomas C, et al. 989–97. Induction of EMT by twist proteins as a collateral effect of tumor- 11. Schepers GE, Teasdale RD, Koopman P. Twenty pairs of sox: extent, promoting inactivation of premature senescence. Cancer Cell homology, and nomenclature of the mouse and human sox transcrip- 2008;14:79–89. tion factor gene families. Dev Cell 2002;3:167–70. 30. Miyazono K. Transforming growth factor-beta signaling in epithelial- 12. Wegner M. From head to toes: the multiple facets of Sox proteins. mesenchymal transition and progression of cancer. Proc Jpn Acad Ser Nucleic Acids Res 1999;27:1409–20. B Phys Biol Sci 2009;85:314–23. 13. van Houte L, van Oers A, van de Wetering M, Dooijes D, Kaptein R, 31. Maeda M, Johnson KR, Wheelock MJ. Cadherin switching: essential Clevers H. The sequence-specific high mobility group 1 box of TCF-1 for behavioral but not morphological changes during an epithelium-to- adopts a predominantly alpha-helical conformation in solution. J Biol mesenchyme transition. J Cell Sci 2005;118:873–87. Chem 1993;268:18083–7. 32. Vincent T, Neve EP, Johnson JR, Kukalev A, Rojo F, Albanell J, et al. A 14. Schilham MW, Oosterwegel MA, Moerer P, Ya J, de Boer PA, van de SNAIL1-SMAD3/4 transcriptional repressor complex promotes TGF- Wetering M, et al. Defects in cardiac outflow tract formation and pro-B- beta mediated epithelial-mesenchymal transition. Nat Cell Biol lymphocyte expansion in mice lacking Sox-4. Nature 1996;380:711–4. 2009;11:943–50. 15. Cheung M, Abu-Elmagd M, Clevers H, Scotting PJ. Roles of Sox4 in 33. Tavazoie SF, Alarcon C, Oskarsson T, Padua D, Wang Q, Bos PD, et al. central nervous system development. Brain Res Mol Brain Res Endogenous human microRNAs that suppress breast cancer metas- 2000;79:180–91. tasis. Nature 2008;451:147–52. 16. van de Wetering M, Oosterwegel M, van Norren K, Clevers H. Sox-4, an 34. Pece S, Tosoni D, Confalonieri S, Mazzarol G, Vecchi M, Ronzoni S, Sry-like HMG box protein, is a transcriptional activator in lymphocytes. et al. Biological and molecular heterogeneity of breast cancers corre- EMBO J 1993;12:3847–54. lates with their cancer stem cell content. Cell 2010;140:62–73. 17. Penzo-Mendez AI. Critical roles for SoxC transcription factors in 35. Hur W, Rhim H, Jung CK, Kim JD, Bae SH, Jang JW, et al. SOX4 development and cancer. Int J Biochem Cell Biol 2010;42:425–8. overexpression regulates the p53-mediated apoptosis in hepatocel- 18. Lee CJ, Appleby VJ, Orme AT, Chan WI, Scotting PJ. Differential lular carcinoma: clinical implication and functional analysis in vitro. expression of SOX4 and SOX11 in medulloblastoma. J Neurooncol Carcinogenesis 2010;31:1298–307. 2002;57:201–14. 36. Huang YW, Liu JC, Deatherage DE, Luo J, Mutch DG, Goodfellow PJ, 19. Castillo SD, Matheu A, Mariani N, Carretero J, Lopez-Rios F, Lovell- et al. Epigenetic repression of microRNA-129-2 leads to overexpres- Badge R, et al. Novel transcriptional targets of the SRY-HMG box sion of SOX4 oncogene in endometrial cancer. Cancer Res 2009;69: transcription factor SOX4 link its expression to the development of 9038–46. small cell lung cancer. Cancer Res 2012;72:176–86. 37. Foubert E, De Craene B, Berx G. Key signalling nodes in mammary 20. Liu P, Ramachandran S, Ali Seyed M, Scharer CD, Laycock N, Dalton gland development and cancer. The Snail1-Twist1 conspiracy in WB, et al. Sex-determining region Y box 4 is a transforming oncogene malignant breast cancer progression. Breast Cancer Res 2010; in human prostate cancer cells. Cancer Res 2006;66:4011–9. 12:206.

www.aacrjournals.org Cancer Res; 72(17) September 1, 2012 4607

Zhang et al.

38. Pan X, Zhao J, Zhang WN, Li HY, Mu R, Zhou T, et al. Induction of SOX4 modeling of cancer and anti-cancer drugs. Biomaterials 2012;33: by DNA damage is critical for p53 stabilization and function. Proc Natl 1437–44. Acad Sci U S A 2009;106:3788–93. 44. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. 39. Taube JH, Herschkowitz JI, Komurov K, Zhou AY, Gupta S, Yang J, Prospective identiﬁcation of tumorigenic breast cancer cells. Proc Natl et al. Core epithelial-to-mesenchymal transition interactome gene- Acad Sci U S A 2003;100:3983–8. expression signature is associated with claudin-low and metaplas- 45. Medina PP, Castillo SD, Blanco S, Sanz-Garcia M, Largo C, Alvarez tic breast cancer subtypes. Proc Natl Acad Sci U S A 2010;107: S, et al. The SRY-HMG box gene, SOX4, is a target of gene ampli- 15449–54. ﬁcation at chromosome 6p in lung cancer. Hum Mol Genet 2009;18: 40. Casas E, Kim J, Bendesky A, Ohno-Machado L, Wolfe CJ, Yang J. 1343–52. Snail2 is an essential mediator of Twist1-induced epithelial mesen- 46. Liu M, Casimiro MC, Wang C, Shirley LA, Jiao X, Katiyar S, et al. chymal transition and metastasis. Cancer Res 2011;71:245–54. p21CIP1 attenuates Ras- and c-Myc-dependent breast tumor epithe- 41. Scheel C, Eaton EN, Li SH, Chaffer CL, Reinhardt F, Kah KJ, et al. lial mesenchymal transition and cancer stem cell-like gene expression Paracrine and autocrine signals induce and maintain mesenchymal in vivo. Proc Natl Acad Sci U S A 2009;106:19035–9. and stem cell states in the breast. Cell 2011;145:926–40. 47. Carey L, Winer E, Viale G, Cameron D, Gianni L. Triple-negative breast 42. Cai H, Ni A, Li W, Li J. Inhibition of melanoma cell proliferation cancer: disease entity or title of convenience? Nat Rev Clin Oncol by targeting Wnt/b-catenin pathway through Sox4 RNA interference. 2010;7:683–92. J Huazhong U Sci-Med 2011;31:565–9. 48. Peddi PF, Ellis MJ, Ma C. Molecular basis of triple negative breast 43. Chen L, Xiao Z, Meng Y, Zhao Y, Han J, Su G, et al. The enhancement of cancer and implications for therapy. Int J Breast Cancer 2012;2012: cancer stem cell properties of MCF-7 cells in 3D collagen scaffolds for 217185.

4608 Cancer Res; 72(17) September 1, 2012 Cancer Research

SOX4 Induces Epithelial−Mesenchymal Transition and Contributes to Breast Cancer Progression

Jianchao Zhang, Qian Liang, Yang Lei, et al.

Cancer Res 2012;72:4597-4608. Published OnlineFirst July 11, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-12-1045

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2012/07/12/0008-5472.CAN-12-1045.DC1

Cited articles This article cites 48 articles, 16 of which you can access for free at: http://cancerres.aacrjournals.org/content/72/17/4597.full#ref-list-1

Citing articles This article has been cited by 5 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/72/17/4597.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/72/17/4597. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.