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

Breast cancer is the most common cancer in women worldwide (1). Tumor metastasis is the leading cause of mortality associated with cancer including breast cancer (2, 3). Epithelial–mesenchymal transition (EMT), an early embryonic development program in which cells convert from the epithelial to the mesenchymal state, plays a pivotal role during malignant tumor progression and metastasis. During an EMT, epithelial cells shed their characteristics such as loss of cell–cell contacts, apical-basal polarity, and downregulation of epithelial markers, followed by acquisition of mesenchymal features including enhancement of motility and invasiveness, reorganization of cytoskeleton, elevated resistance to apoptosis, and increased expression of ECM components like fibronectin and MMPs (4, 5).

Several developmentally important transcription factors such as Snail, Twist, FOXC1, FOXC2, and LBX1 were reported to act as molecular triggers of the EMT in breast cancer and other cancers (4, 6). In addition to these transcription factors, an EMT program can also be initiated by several signal pathways, including TGF-β, Wnt, Notch, and Hedgehog (7); among them, the TGF-β signaling plays a significant role in contributing to the initiation of EMT during embryonic development and cancer pathogenesis (8). Recent studies have shown that EMT endows cancer cell with stem cell-like properties and relates to the highly aggressive basal like breast cancer subtype (9, 10).

SOX4, a member of the C subgroup of SRY-related HMG box (SOX) transcription factor family that are structurally characterized by a highly conserved HMG box domain that directly binds the minor groove of DNA helix, has been implicated in various development processes, such as embryonic cardiac, central nervous system, lymphocyte development, and differentiation, through its transcriptional activity (11–16). Besides functioning as a transcriptional factor in regulation of development processes, SOX4 has been implicated in cancer progression. Increased SOX4 expression is observed in many human tumors, including medulloblastoma, small cell lung cancer, prostate cancer, and breast cancer (17–21). Both TGF-β and Wnt signal pathways have been implicated in cancer progression and EMT (4, 5). TGF-β has been shown to induce SOX4 expression in glioma-initiating cells and glioblastoma multiforme (22, 23). SOX4 has been shown to activate the canonical Wnt signal pathway in cancer cells (24, 25). Therefore, we speculate that SOX4 might play a role in breast cancer progression and aggression through induction of an oncogenic EMT.

In this study, we showed that SOX4 was able to activate EMT program in epithelial cells, increase the number of CD44^high/CD24^low population and potentiate mammosphere-forming ability. We also showed that in mammary epithelial cells, SOX4 cooperated with the activated oncogenic Ras to endow the cells with the tumorigenicity in vivo. We further showed that human triple-negative breast cancer (TNBC) has increased SOX4 expression. These data implicate a novel role of SOX4 in inducing EMT in breast cancer progression and its close association with invasive subtypes of human breast cancer.

MCF10A, MDA-MB-231, BT549, NMuMG, MDCK, and 293T cell lines were obtained from the American Type Culture Collection, where they were characterized by DNA-fingerprinting and isozyme detection. Cells were immediately expanded and frozen such that they could be revived every 3 to 4 months. MCF10A cells were cultured as previously described (26) in DMEM/F12 supplemented with 5% horse serum, 20 ng/mL EGF, 0.5 μg/mL hydrocortisone, 100 ng/mL cholera toxin, 10 μg/mL insulin and pen/strep. MDCK and 293T cells were cultured in DMEM containing 10% FBS (Hyclone). NMuMG cells were maintained in DMEM supplemented with 10% FBS and 10 μg/mL insulin. MDA-MB-231 cells were cultured in Leibovitz's L-15 medium with 10% FBS at 37°C without CO₂. BT549 cells were cultured in RPMI-1640 medium supplemented with 10% FBS and 0.023 IU/mL insulin.

The antibodies and reagents are listed in Supplementary Materials and Methods.

Standard procedures for immunoblotting are described in Supplementary Materials and Methods.

Experiments were carried out as described in Supplementary Materials and Methods.

A description of procedures are detailed in Supplementary Materials and Methods.

Human breast tumor specimens were obtained from the Second Hospital of Jilin University, China. Tissue samples were fixed in 10% formaldehyde and embedded in paraffin. Sections were cut and stained using a conventional immunohistochemistry procedure. Briefly, the tissue sections were deparaffinized and rehydrated. Antigen retrieval was done by heating the sample in 10 mM citrate buffer (pH 6.0) at 95°C for 15 minutes. Endogenous peroxidase activity was blocked with peroxidase (DAKO), and sections were blocked with 10% goat serum. Sections were incubated with anti-SOX4 antibody (Abcam, 1:50) for 1 hour, followed by incubation with secondary antibodies (DAKO) for 30 minutes. The immunostaining was developed with 3.3′-diaminobenzidine. Finally, sections were counterstained with hematoxylin.

Experiments were carried out as described in Supplementary Materials and Methods.

Mammosphere assays were carried out as described elsewhere (27) with minor modifications. Single cells were plated at 10,000 cells/mL on 6-well ultra-low attachment plates (Corning) in serum-free DMEM/F12 supplemented with 20 ng/mL bFGF, 20 ng/mL EGF, 4 μg/mL insulin, 4 μg/mL heparin, 1 μg/mL hydrocortisone, 0.4% BSA and B27. Fresh medium was supplemented every 3 days. The mammospheres were counted at day 14.

A total of 1 × 10⁶ cells were resuspended in 100 μL PBS containing 2% FBS (FACS buffer), and then incubated on ice for 10 minutes. CD44-APC and CD24-PE (BD Biosciences) were added to cell suspension and incubated on ice for 30 minutes. Cells were washed and resuspended in 500 μL FACS buffer and analyzed using a FACS Calibur Flow Cytometer (BD Biosciences).

A total of 5 × 10⁶ cells in 100 μL PBS were injected subcutaneously into 6-week-old female BALB/c nude mice. Five mice per group were used in each experiment. Tumor volume was measured weekly using a Vernier caliper and calculated according to the formula: π/6 × length × width². Eight weeks later, the mice were sacrificed, and tumors were collected and photographed. All animal experiments were approved by the Animal Care Committee of the Northeast Normal University, China.

The protocol is described in Supplementary Materials and Methods.

Cell growth rates were assessed by the MTT assay as described in Supplementary Materials and Methods.

Experiments were carried out as described in Supplementary Materials and Methods.

Data are presented as mean ± SD. The Student t test (2-tailed) was used to determine statistically the significance of differences between groups. P < 0.05 was considered statistically significant. SOX4 expression intensities in human breast cancer samples were analyzed by χ² test. Statistical analysis was carried out using the SPSS17.0 software.

To investigate the role of SOX4 in EMT, we stably overexpressed SOX4 in the immortalized normal human mammary epithelial cell line MCF10A that lacks endogenous SOX4 expression, by using retroviral infection, as confirmed by immunoblotting (Fig. 1A) and real-time PCR (Supplementary Fig. S1). We observed that MCF10A cells transfected with vector retained their cobblestone-like morphology with tight cell–cell adhesion, whereas cells expressing exogenous SOX4 displayed an elongated fibroblast-like morphology with scattered distribution in culture (Fig. 1B). We then examined both epithelial and mesenchymal markers by immunoblotting (Fig. 1C) and immunofluorescence (Fig. 1D). As can be seen, the SOX4-expressing MCF10A cells exhibited a significant downregulation of β-catenin and complete loss of E-cadherin and occludin from cell–cell contacts; meanwhile the mesenchymal markers fibronectin, vimentin, and N-cadherin were dramatically upregulated. Real-time PCR analyses also revealed the expression of E-cadherin mRNA and concomitant induction of N-cadherin, fibronectin, and vimentin mRNAs in SOX4-expressing MCF10A cells (Fig. 1E). These morphologic and molecular changes suggested an apparent transition of the SOX4-expressing MCF10A cells from an epithelial to mesenchymal status. To further probe the possible interactions between SOX4 and other EMT-inducing transcription factors, we examined the expression of other known EMT inducers. We showed that the endogenous mRNA levels of Snail1, FOXC2, GSC, SIX1, Twist1, HOXB7, and ZEB1 were elevated in response to SOX4 overexpression, to a variable extent, whereas the ZEB2 mRNA level exhibited no detectable change (Fig. 1F). We also ectopically overexpressed other EMT inducers Snail and Twist1 in MCF10A cells. We observed that Snail and Twist1 induced EMT in MCF10A cells (Supplementary Fig. S2A), and upregulated SOX4 mRNA expression in MCF10A cells (Supplementary Fig. S2B). Typically, the EMT phenotype is usually accompanied by the acquisition of cell traits such as greater migration and more invasive ability. As shown in Fig. 1G and H, SOX4-expressing MCF10A cells dramatically increased their migratory and invasive behaviors. Similar results were observed in MDCK cells, a prototypic cell model for EMT study (Supplementary Fig. S3A–E). Together, these results show that SOX4 is a novel inducer of EMT and it promotes cell migration and invasion in MCF10A cells.

Loss of E-cadherin seems to be a critical event in EMT. We next carried out a luciferase reporter assay to investigate whether SOX4 could transcriptionally regulate the E-cadherin expression. The results showed that the relative luciferase activity of E-cadherin was not changed in 293T cells transiently cotransfected with E-cadherin promoter reporter together with SOX4 expression construct or with empty vector (Supplementary Fig. S4A), indicating that SOX4 may indirectly regulate E-cadherin transcription. In line with these observations, SOX4 increased the expression of known EMT inducers (including Snail and ZEB1) that act to directly repress E-cadherin transcription, during SOX4-induced EMT (Fig. 1F). To test the effect of SOX4 on the transcription activity of promoters of these 2 EMT inducers, we transiently transfected 293T cells with luciferase reporter constructs containing proximal promoter of ZEB1 or Snail. We found that the relative luciferase activity of ZEB1 was increased proportionally to the increasing amount of SOX4 in 293T cells (Supplementary Fig. S4B). However, the Snail promoter activity was not changed (Supplementary Fig. S4C). These findings suggest that E-cadherin transcriptional inactivation during SOX4-induced EMT may be a result of induction of ZEB1 expression, a well-known upstream repressor of E-cadherin.

We next carried out a loss-of-function assay to further study the role of SOX4 in invasive mesenchymal breast cancer cells MDA-MB-231 and BT549. The effect of SOX4 silencing on the invasive ability of these cells was investigated using transwell invasion assays. The results indicated that SOX4 knockdown prominently impaired the migration and invasion ability of these 2 breast cancer cells, but it did not affect cell proliferation and apoptosis (Supplementary Fig. S5A–G). Moreover, suppression of SOX4 expression resulted in vimentin downregulation in both cell lines (Supplementary Fig. S5A and S5B). However, no detectable changes were observed in other EMT markers (data not shown). In addition, downregulation of SOX4 did not cause significant cellular morphologic changes in both MDA-MB-231 and BT549 cells (Supplementary Fig. S5H). Thus, our loss-of-function study suggested that the suppression of SOX4 could partially reverse the EMT phenotype of MDA-MB-231 and BT549 cells.

Mammary epithelial cells undergoing an EMT program have been linked to stem cell phenotypes such as an increased CD44^high/CD24^low population and mammospheres formation ability (9). To determine whether SOX4 has the effect to lead to the stem cell phenotypes upon induction of EMT, we carried out FACS to identify CD44^high/CD24^low populations. We observed that the SOX4-expressing MCF10A cells exhibited a significant increase in the CD44^high/CD24^low stem cell population compared with vector-infected cells (Fig. 2A). Meanwhile, as evidenced in Fig. 2B and C, the SOX4-expressing MCF10A cells increased both in size and in number of mammospheres in comparison with the vector-infected cells. We thus concluded that the SOX4-induced EMT generates mesenchymal cells with stem cell-like phenotypes, a feature recently defined for EMT inducers (6, 28).

Although SOX4 was able to induce EMT, MCF10A cells expressing SOX4 were unable to form tumors when injected into nude mice (Supplementary Fig. S6A), suggesting that SOX4 itself lacks the ability to induce neoplastic transformation. It has recently been reported that EMT inducers Twist1, Twist2, and LBX1 cooperate with activated oncogenic Ras to promote even more dramatic characteristics of EMT and enhance tumorigenesis (6, 29). To investigate this possible cooperation, we coexpressed SOX4 with oncogenic H-RasV12 in MCF10A cells. As shown in Fig. 3A and Supplementary Fig. S6B, coexpression of SOX4 and oncogenic Ras led to a more dramatic morphologic change, characterized by prominently elongated spindle-shaped cells. In addition, coexpression of both SOX4 and H-RasV12 triggered a further reduction of β-catenin, relative to either SOX4 or H-RasV12 alone; however, no changes in other epithelial or mesenchymal markers were observed (Fig. 3B). Furthermore, MCF10A cells expressing SOX4+H-RasV12 exhibited higher migration (Fig. 3C) and invasive (Fig. 3D) ability than cells expressing either SOX4 or H-RasV12 alone. Meanwhile, all these cells exhibited roughly the same growth rates with no statistically significant differences (Supplementary Fig. S6C). These data implicated that SOX4 and H-RasV12 worked synergistically in inducing a more prominent EMT phenotype.

To further establish whether SOX4 is able to trigger tumorigenesis in vivo, we tested its effect in a xenograft mouse model. We detected no primary tumor growth 8 weeks after the nude mice were subcutaneously injected with MCF10A cells stably expressing SOX4 alone. Meanwhile, MCF10A cells expressing oncogenic H-RasV12 alone produced small palpable tumors. Interestingly, coexpression of both SOX4 and H-RasV12 resulted in rapid development of very large tumors (Fig. 3E and F). These results clearly indicate that SOX4 and oncogenic H-RasV12 work synergistically to induce tumorigenesis in vivo.

To evaluate the clinical relevance of SOX4 expression in human mammary carcinomas, we carried out immunohistochemistry 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 immunoreactivity. 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. 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 = 0.014), and negative PR (P = 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 = 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 clinical outcome. In particular, among all of the 12 microarray studies in Oncomine data, SOX4 was significantly upregulated in unfavorable ER⁻/PR⁻/HER2⁻ triple-negative basal-like subtype (Supplementary Table S2). These observations implicate the potential usefulness of the aberrant high SOX4 expression as a novel prognostic molecular marker for TNBC.

Several lines of evidence have implicated the involvement of TGF-β in EMT process both in embryonic development and breast cancer progression (8, 30). We next intended to identify whether TGF-β signaling is activated in SOX4-induced EMT. Our real-time PCR revealed an increased expression of TGF-β1 and TGF-β2 mRNAs in SOX4-expressing MCF10A cells (Fig. 5A). In addition, the level of phosphorylated Smad2 protein, a downstream effector of TGF-β pathway, was significantly increased in SOX4-expressing MCF10A cells (Fig. 5B left). Moreover, SOX4 silencing efficiently decreased the level of phosphorylated Smad2 and the expression of TGF-β1 and TGF-β2 mRNAs in SOX4-expressing MCF10A cells (Fig. 5C); whereas knockdown of SOX4 upregulated E-cadherin expression and partly downregulated N-cadherin expression, presumably due to the residual ectopic SOX4 (Supplementary Fig. S7A). Consistent with these changes, we observed some tight cell clusters upon SOX4 silencing (Supplementary Fig. S7B). Apparently, these experiments pointed to a reinforced TGF-β signaling upon ectopic SOX4 expression in MCF10A cells. To further validate that TGF-β signaling is responsible for the SOX4-induced EMT and the enhanced cell motility, we used a specific TGF-β receptor kinase inhibitor SB431542 to block the TGF-β signaling in SOX4-expressing MCF10A cells. We found that suppression of TGF-β signaling by the inhibitor reduced Smad2 phosphorylation level and downregulated vimentin expression (Fig. 5B right), without affecting the morphologic and molecular features of SOX4-expressing MCF10A cells undergoing EMT (data not shown). Moreover, treatment of SOX4-expressing MCF10A cells with SB431542 reduced their migration and invasive ability (Fig. 5D and E). These results suggest that the intensified TGF-β signaling induced by SOX4 promotes cell motility, and this partly contributes to EMT.

TGF-β signaling has been shown to be able to induce EMT in many epithelial cells including MCF10A and NMuMG (31, 32). In addition, recent studies have shown that SOX4 is a direct target gene of TGF-β in glioma-initiating cells and is also responsive to TGF-β in glioblastoma multiforme (22, 23). Here, we showed that SOX4 mRNA was also induced in MCF10A (Fig. 6A and B) and NMuMG (Supplementary Fig. S8) cells, in a dose- and time-dependent manner upon the addition of TGF-β to the cell culture medium. To investigate whether SOX4 is required for TGF-β–induced EMT, we knocked down SOX4 expression in MCF10A cells with short hairpin RNA (shRNA) and examined their responses to TGF-β1 treatments. MCF10A cells expressing SOX4 shRNA (MCF10A-shSOX4#2) or nontarget control shRNA (MCF10A-shCtrl) were treated with TGF-β1. We observed that SOX4 expression was induced 48 hours after TGF-β1 addition in the MCF10A-shCtrl cells, whereas MCF10A-shSOX4#2 cells exhibited a reduction in the basal expression level of SOX4 as well as in the induction of SOX4 after TGF-β1 stimulation (Fig. 6C). We also found that TGF-β1 treatments induced EMT in MCF10A-shCtrl cells, but not in MCF10A-shSOX4#2 cells (Fig. 6D). Consistent with morphologic changes, E-cadherin expression was inhibited and N-cadherin expression was increased in MCF10A-shCtrl cells after TGF-β1 treatments. However, under the same conditions E-cadherin and N-cadherin expression were not changed in MCF10A-shSOX4#2 cells (Fig. 6C). These experiments suggest that SOX4 is also involved in and required for TGF-β–induced EMT.

Increasing evidence suggests that SOX4 overexpression is associated with several human cancers including brain, lung, and breast cancers (17). Specifically, it has been reported that SOX4 is expressed in normal breast and breast cancer cells, and progestin can promote SOX4 expression and induce SOX4-mediated transcriptional activity in breast cancer cell lines (21). Moreover, miR-335 suppresses breast cancer metastasis and migration by modulating SOX4 expression (33). A recent analysis of the transcriptional profile of human normal mammary stem cells (hNMSC), aimed at prediction of biological and molecular features of breast cancers, unravels that SOX4 is upregulated in PKH26-positive cells that possess all the characteristics of hNMSCs (34). These findings implicate the involvement of SOX4 in breast cancer development. Nevertheless, how SOX4 implements its oncogenic effects during breast cancer progression remains unclear. We find in this study that SOX4 functions as a trigger for EMT to contribute to breast cancer progression, a previously unreported role of SOX4 in breast cancer. Specifically, our data show that SOX4 represses the epithelial phenotype, induces the mesenchymal phenotype, and dramatically increases the migration and invasion of MCF10A cells.

Besides breast cancer, it has been reported that ectopic overexpression of SOX4 in hepatocellular carcinoma endows an antiapoptotic effect and contributes to hepatocarcinogenesis (35). Previous study shows that the epigenetic deregulation of miRNA-129-2 leads to the oncogenic overexpression of SOX4 in endometrial cancer. Reactivation of miR-129-2 results in SOX4 downregulation and reduces the proliferation of endometrial cancer cells (36). Accumulating evidence indicates that EMT inducers such as Snail and Twist1 control cell proliferation and survival that are critical in cancer progression (37). These results suggest that SOX4 may act in a similar fashion as Snail and Twist1, exerting dual effects of potent prosurvival function and orchestrating an EMT in tumor progression. Interestingly, SOX4 is also able to promote cell-cycle arrest and apoptosis in a p53-dependent manner in HCT116 (38). The distinct physiologic functions of SOX4 in different cell lineages and cancer types highlight the importance of cell context and genetic background in various human cancer cell lines and different cancer types that determine the SOX4 function.

Moreover, we have shown in this work that SOX4 upregulates the mRNA expression of several EMT-inducing transcription factors and SOX4 expression is induced by a large number of known regulators of the EMT program, notably the Snail, Twist, Ras, and TGF-β1. Our data are in accordance with a recent study that discovers that overexpression of one of the EMT inducers upregulates a subset of other EMT-inducing transcription factors, implicating the interactions among these EMT inducers (39). E-cadherin plays a pivotal role in epithelial cell–cell adhesion. Functional loss of E-cadherin is considered a hallmark of EMT (4, 5). Our results show that SOX4 may indirectly downregulate the transcription of E-cadherin. It is reported that EMT inducers such as Twist1, Snail, Slug, ZEB1, ZEB2, FOXC1, and FOXC2 are able to transcriptionally repress E-cadherin expression either directly or indirectly (4). Esmeralda Casas and colleague show that Twist1 indirectly suppresses E-cadherin transcription to promote EMT through directly binding to the Snail2 gene promoter to activate its transcription (40). We show in this study that SOX4 indirectly downregulates the expression of E-cadherin through activating ZEB1 expression. Interestingly, search of the defined region of ZEB1 promoter with the transcription factor database (TRANSFAC) did not reveal a known SOX4 consensus-binding site, suggesting that SOX4 may not be able to bind directly to the ZEB1 promoter; rather, it may indirectly activate the ZEB1 promoter through as yet unidentified intermediate factor(s).

More importantly, our data have suggested that SOX4 is also involved in and required for TGF-β–induced EMT. We show in our study that the autocrine TGF-β signaling is activated in SOX4-induced EMT; meanwhile TGF-β signaling induces SOX4 expression in breast epithelial cells and knockdown of SOX4 blocks TGF-β1–induced EMT. Thus, we speculate that the SOX4-TGFβ-SOX4 feedback loop presumably functions as a novel signaling pathway in EMT regulation, as well as in breast cancer progression. A recent study by Christina Scheel and colleague show that autocrine signaling pathways involving TGF-β and both the canonical and noncanonical Wnt signaling are activated in EMT program induced by diverse stimuli including transcription factors such as Twist, and they propose that these signalings further maintain the EMT program, and disruption of these extracellular autocrine signalings abrogates transcription factor-induced EMT (41). In support of the involvement of additional EMT-promoting pathways, it has been documented that SOX4 activates the Wnt/β-catenin pathway in colon carcinoma and melanoma (24, 25, 42). Whether the canonical Wnt/β-catenin signaling can be activated in SOX4-induced EMT needs to be clarified.

Recent studies have shown that EMT can induce noncancer stem cells to acquire cancer stem cell (CSC)-like properties in breast cancer cells that exhibit a CD44^high/CD24^low antigenic phenotype (9). In this study, we have shown that ectopic expression of SOX4 in MCF10A cells increases the CD44^high/CD24^low subpopulation and enhances the mammosphere-forming ability, a property of the stem cells. It is reported that normal and neoplastic breast stem-like cells express some EMT inducers such as Twist1, Snail1, Snail2, and ZEB1 (9). A recent study by Chen and colleague unravels that MCF-7 cells cultured in 3D collagen scaffolds acquire the properties of CSCs, and the transcriptions of stem cell markers such as OCT4A and SOX2, and breast cancer stem cell signatures, including SOX4, JAG1, and CD49F, are significantly unregulated (43). Furthermore, SOX4 is also upregulated in PKH26-positive cells that possess all the characteristics of hNMSCs (34). All these data are supportive of a close relationship among SOX4, EMT, and CSCs. Previous study reports that the CD44^high/CD24^low subpopulation breast cancer cells enhance tumorigenicity in a xenograft model (44). However, in our xenograft model, MCF10A cells expressing SOX4 alone are unable to form tumors when injected into nude mice, suggesting that SOX4 itself lacks the ability to induce neoplastic transformation, although it promotes the generation of stem cell-like cells. Nevertheless, we have observed a cooperative action between SOX4 and activated oncoprotein Ras in MCF10A cells, resulting in the formation of large tumors in nude mice. In line with our observations, a previous work has shown that SOX4 overexpression alone in NIH3T3 cells do not increases the transforming ability of the cells, but the effect occurs when cells are cotransfected with the weakly oncogenic RHOA-Q63L (45). Thus, our findings provide evidence for the ability of SOX4 to potentiate the oncogenic effect of activated Ras in breast tumorigenesis. Interestingly, we have found that activated Ras itself upregulates the endogenous SOX4 expression, but no synergistic effects between Ras and SOX4 are observed. An earlier study suggests that Ras also induces the expression of other EMT inducer such as Twist1 in MCF10A cells (46). Our result shows that Ras upregulates Snail1 expression in mammary epithelial cells (Supplementary Fig. S6B), but these inducers do not function cooperatively with Ras either. These data seem to support the assumption that only the exogenous EMT inducers and activated Ras are able to exert the cooperative effects to promote even more dramatic characteristics of EMT and cancer progression (6, 9, 29, 41). However, the mechanisms underlying this phenomenon still remain unknown. Conclusively, SOX4 may function as a key tumor progression factor in breast cancer, rather than a tumor-initiating event for breast carcinogenesis.

TNBC is the most aggressive form of breast cancer with high histologic grade, aggressive clinical behavior, high incidence of brain and lung metastases, and the lack of an effective therapeutic target (47). The TNBC shares many clinicopathologic and molecular features with basal-like breast cancers (BLBC), a subtype of breast cancer defined by gene-expression profiling, and accounts for approximately 70% of the BLBC cases. Recent studies have linked EMT with the aggressive BLBC (47, 48). Targeting the EMT-like phenotypes would seem to represent the potential strategies for the development of novel anticancer therapeutics. Our results in this study are in favor of a strong correlation between SOX4 and TNBC, and hence point to the prospect of using SOX4 as a novel biomarker for prognosis and diagnosis of TNBC, as well as a potential molecular therapeutic target for this highly aggressive breast cancer.

No potential conflicts of interest were dislosed.

Conception and design: J. Zhang, Q. Liang, Y. Zhang, J. Lu, B. Huang

Development of methodology: J. Feng, Y. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Zhang, Q. Liang, Y. Lei, M. Yao, L. Li, J. Feng, H. Gao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Zhang, Q. Liang, Y. Lei, L. Li, Y. Zhang, B. Huang

Writing, review, and/or revision of the manuscript: J. Zhang, D.-X. Liu, J. Lu, B. Huang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Gao, J. Feng, Y. Zhang

Study supervision: B. Huang

The authors thank Torben Ørntoft, James M. Wells, Patrick Humbert, R.A. Weinberg, L. Miguel Martins, Ji-Hshiung Chen, and Daniel A. Haber for providing the plasmids mentioned in Materials and Methods.

This work is supported by the grants from the National Natural Science Foundation of China (Grant numbers 91019011, 31170719, 31071149, 31100998), the Fundamental Research Funds for Central Universities, and the Program for Introducing Talents to Universities (B07017).

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.