Introduction
Breast cancer is the most frequently diagnosed cancer among females [
1]. Metastasis is the leading cause of death in breast cancer patients. To develop distant metastatic disease, the epithelial cancer cells need to undergo a morphological change into a mesenchymal phenotype to acquire migration and invasion ability [
2]. Epithelial-mesenchymal transition (EMT) is a program of switching polarized epithelial cells to migratory mesenchymal cells [
2]. The aberrant activation of the EMT program is implicated in tumor progression, metastasis and acquisition of therapeutic resistance [
2,
3]. EMT-transcription factors (EMT-TFs) such as TWIST1, SNAIL1, SNAIL2, ZEB1, ZEB2, Goosecoid and FOXC2, can trigger the EMT program by directly or indirectly suppressing E-cadherin expression [
4-
7]. TWIST1 is a member of the basic helix-loop-helix (bHLH) transcription factor family [
8] and plays a pivotal role in the regulation of embryogenesis, gastrulation and mesoderm formation during early embryonic development [
9,
10].
TWIST1 was found to be frequently activated in a wide array of human cancers and is associated with poor prognosis [
3,
11]. TWIST1 induces the EMT program by downregulating E-cadherin expression through indirect effects on the
CDH1 promoter [
3].
Breast cancer is a heterogeneous disease. Based on their gene expression profiles, breast cancers can be classified into distinct molecular subtypes: normal breast-like, luminal A, luminal B, human epidermal growth factor receptor 2 (HER2)-enriched, and basal-like [
12]. Basal-like breast cancer (BLBC) is less likely to express estrogen receptor (ER), progesterone receptor (PR) and HER2, which are also characteristics of triple-negative breast cancer (TNBC). Thus, BLBC shares many features with TNBC, and the two terms are often used interchangeably [
13]. In addition to the triple-negative marker status, BLBC is characterized by the expression of basal markers such as cytokeratins (CK) 5/6, CK14, CK17 and epithelial growth factor receptor (EGFR) in the clinic [
12]. TNBC and/or BLBC is recognized as a particularly aggressive subtype and receives less benefit from targeted therapy [
12]. Therefore, there is an urgent need to elucidate the molecular pathogenesis of TNBC and/or BLBC and develop effective systemic therapies, especially molecular-targeted therapy.
Recent reports have revealed that TNBC/BLBC is a group of heterogeneous tumors [
14]. BLBC also can be divided into extraordinarily diverse basal-like A and basal-like B subtypes [
15]. The basal-like A cells have either luminal-like or basal-like epithelial morphology, while the basal-like B cells appear poorly differentiated and possess more mesenchymal characteristics [
15]. Thus, the basal-like B subtype is more aggressive than the basal-like A subtype [
15]. Due to the heterogeneity of BLBC, it is important to identify the critical regulatory factors that are associated with aggressive phenotype of BLBC. It is well known that various embryonic and mesenchymal EMT-TFs, including SNAIL1 [
16], SNAIL2 [
17] TWIST1 [
18], and Forkhead box (FOX) transcription factor superfamily members FOXC1 [
19], FOXC2 [
4] and FOXQ1 [
20], contribute to the aggressive phenotype of BLBC.
The mesenchymal regulator FOXF2 belongs to the FOX transcription factor superfamily [
21]. It is specifically expressed in the mesenchyme adjacent to the epithelium in organs derived from the splanchnic mesoderm [
22], and plays an important role in tissue homeostasis through regulating epithelium-mesenchyme interaction to maintain epithelium polarity [
22]. Our previous clinical study demonstrated that the under-expression of
FOXF2 is associated with early-onset metastasis and poor prognosis of patients with TNBC, but not the prognosis of non-TNBC patients [
23]. This result suggests that FOXF2 deficiency is involved in TNBC/BLBC metastasis through regulating EMT. Recent studies have indicated that FOXF2 is a potential tumor suppressor in both prostate cancer [
24] and breast cancer [
25]. However, the role of FOXF2 in breast cancer metastasis and the underlying molecular mechanisms remain largely unknown.
In this study, we identified FOXF2 as a novel EMT-suppressing transcription factor in BLBC and demonstrated that it directly represses the transcription of TWIST1. FOXF2 deficiency promotes metastasis of BLBC cells through transcriptional upregulation of TWIST1 and activation of EMT.
Materials and methods
Cell culture
The human breast cancer cell lines MDA-MB-231, BT549, MCF-7, BT474, ZR-75-30, SKBR-3 and MDA-MB-453, immortalized non-tumorigenic basal-like mammary epithelial cell lines MCF-10A and HBL100 were obtained from American Type Culture Collection (Manassas, VA, USA). MDA-MB-231-luc-D3H2LN (231-Luc), a MDA-MB-231 subclone expressing luciferase, was obtained from Caliper Life Sciences (Hopkinton, MA, USA). MCF-10A cells were maintained in DMEM-F12 medium (Invitrogen, Gaithersburg, MD, USA) supplemented with 5% horse serum, 20 ng/mL EGF, 500 ng/mL hydrocortisone, 10 μg/mL insulin and 100 ng/mL cholera toxin. The other cells were cultured in DMEM-F12 (MCF-7 and MDA-MB-453) or RPMI 1640 (MDA-MB-231, BT549, BT474, ZR-75-30, SKBR-3 and HBL100) medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen), 100 units/mL penicillin, and 100 mg/mL streptomycin (Invitrogen). All cell lines were incubated in a humidified incubator at 37°C with 5% CO2 and grown into logarithmic phase and/or 80% confluence for the experiments.
Lentiviral transduction of shRNA and transfection of interfering RNA and plasmids
To obtain the stable FOXF2-knockdown cells, cells were infected with the lentiviral vector expressing a short hairpin RNA (shRNA) targeting the sequence GTCCTCAACTTCAATGGGATT in the coding sequence region of human FOXF2 gene (shFOXF2; Sigma-Aldrich, St Louis, MO, USA; TRCN0000013959). A shRNA non-targeting human and mouse gene was used as control (shControl; Sigma-Aldrich; SHC002). The cells were selected in 2 μg/mL puromycin to establish stable expressing shRNAs cells. For the rescue experiment, the stable FOXF2-knockdown cells were transfected with shRNA-resistant full-length FOXF2 plasmid (Genechem, Nanjin, China), which has three nucleotide mismatches within the target sequence of the shFOXF2 (GTCCTCAATTTTAACGGGATT). To silence the expression of TWIST1, the cells were transiently transfected with two independent small interfering RNAs (siRNAs) targeting the coding sequence region of human TWIST1 gene GGATCAAACTGGCCTGCAA (si-TW#1) and GAACACCTTTAGAAATAAA (si-TW#2; RiboBio Co., Guangzhou, China) or a negative control siRNA (si-Control) using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s specifications. The human full-length FOXF2 cDNA was amplified from a pEV-FOXF2 plasmid, which was a gift from Professor Peter Carlsson (University of Gothenburg, Department of Cell and Molecular Biology, Sweden), and then was subcloned into the pcDNA3.1-HA vector (Invitrogen). The human full-length TWIST1 cDNA (Genechem) was subcloned into the pcDNA3.1 vector (Invitrogen). The cells were transfected with pcDNA3.1-HA-FOXF2 (FOXF2), pcDNA3.1-TWIST1 (TWIST1) or vector control (Invitrogen) using Lipofectamine 2000.
Reverse transcription quantitative polymerase chain reaction
Total RNA isolation from cultured cells or tissues, reverse transcription (RT) reactions, quantitative PCR (qPCR), and the quantification of target gene expression were performed as previously described [
23].
Immunoblot, immunohistochemistry, and immunofluorescence
The protocols for protein analysis by immunoblot, immunohistochemistry, and immunofluorescence were performed as previously described [
26]. The primary antibodies were anti-E-cadherin, anti-N-cadherin, anti-vimentin (BD Biosciences, Bedford, MA, USA), anti-FOXF2 (Abnova, Aachen, Germany), anti-TWIST1, anti-HA tag (Abcam, Cambridge, MA, USA), and anti-β-actin (Sigma-Aldrich).
Chromatin immunoprecipitation assay
Chromatin immunoprecipitation (ChIP) assay was performed using a ChIP assay kit (EMD Millipore, Billerica, MA, USA) according to the manufacturer’s instructions. Anti-HA antibody (Abcam) was used for immunoprecipitation to enrich the promoter fragments containing putative FOXF2 binding site of target genes in MCF-10A cells constitutively expressing HA-tagged FOXF2 (MCF-10A-FOXF2). The isotype IgG (Abcam) was used as a negative control. The primers for the amplification of the TWIST1 promoter region containing a putative FOXF2 binding site from −2120 to −2113 relative to the transcription start site (TSS) were 5′-AGATTTCCTTTACACTTTACCC-3′ (forward) and 5′-GCGAGTGTTATTTCTCCAGCGA-3′ (reverse). The primers for the amplification of TWIST1 exon 1 (+408 to +841), which served as a negative control, were 5′-CAGCGAGGAAGAGCCAGACCG-3′ (forward) and 5′-GGAGGACCTGGTAGAGGAAGT-3′ (reverse).
Luciferase reporter assay
The TWIST1 promoter region with (−2140 to +27) or without (−2057 to +27) FOXF2 binding sites was amplified from human genomic DNA using the primers 5′-AGTCCAATCATTCGATCTC-3′ (forward) and 5′-CTGCAGACTTGGAGGCT-3′ (reverse), or 5′-ATAGCTGAAGTGGAAAAGG-3′ (forward) and 5′-CTGCAGACTTGGAGGCT-3′ (reverse) with Kpn I and Bgl II restriction endonuclease recognition sites at the 5′-ends, respectively. The fragments of the TWIST1 promoter were cloned into the luciferase reporter gene plasmid pGL3-Basic (Promega, Madison, WI, USA). The pGL3 reporter and pRL-TK plasmid were transiently co-transfected into MCF-10A and BT549 cells for 48 h. The luciferase activity of pGL3-TWIST1 (−2140/+27) or pGL3-TWIST1 (−2057/+27) was normalized to Renilla luciferase activity.
Cell proliferation
Cell proliferation ability was assessed using a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Cells were seeded at a density of 2 × 103 cells/well in 96-well plates. On day 1, 2, 3, 4 and 5, the cells were incubated with 10 μL MTT solution (5 mg/mL in phosphate-buffered saline (PBS)) at 37°C for 4 h. After removal of the medium, 100 μL DMSO was added to each well and absorbance was measured at 570 nm.
Cell invasion and migration assay
Cell invasion and migration ability
in vitro were assessed using Matrigel-coated and non-Matrigel-coated transwell inserts (BD Biosciences) as described previously [
26].
Xenograft tumor assay
Twenty female 4- to 6-week-old severe combined immunodeficiency (SCID) mice were randomly divided into two groups for the xenograft tumor assay. Then, 5 × 10
6 231-Luc cells infected with shFOXF2 (231-Luc-shFOXF2) or shControl (231-Luc-shControl) were washed and harvested in 0.1 mL PBS and injected into the left lower abdominal mammary fat pad of the SCID mice [
27]. After 22 days, all mice were anesthetized using isoflurance and injected intraperitoneally with 150 mg/kg D-luciferin/PBS. The bioluminescent imaging was performed using a Xenogen IVIS system (Caliper Life Sciences, Hopkinton, MA, USA). The data of living image were expressed as photon flux (photons per second). The tumor volumes were measured by caliper at the indicated time and calculated using the formula (length x width
2)/2 [
28]. The survival time of mice was defined as the time interval between the day of cell injection and the time of death or sacrifice. The xenograft tumors, livers and lungs were harvested from xenograft mice at sacrifice. The tissues were prepared for paraffin-embedded tissue section and hematoxylin and eosin (H&E) staining or immunohistochemistry staining. Two independent experiments were performed with xenograft mice. In one experiment, the mice died naturally to allow survival time calculations. In the second experiment, the mice were sacrificed for dissection on day 21 after cell injection. The mouse xenograft tumor assays were performed in accordance with protocols approved by the Animal Ethics Committee of Tianjin Medical University Cancer Institute and Hospital (TMUCIH; Tianjin, China. No. 058).
Breast cancer tissue specimens
A total of 34 primary TNBC tissue specimens were obtained from patients who underwent breast surgery in TMUCIH. The use of these specimens was approved by the Institutional Review Board of TMUCIH, and written consent was obtained from all participants. All cases were followed up over 3 years, and 30 cases were followed up over 5 years. Disease-free survival (DFS) was defined as the time interval between primary surgery and any relapse (local-regional, contra-lateral and/or distant), or the terminal time of follow-up without any relapse events.
Independent data sets for validation of gene expression levels in breast cancer cell lines and breast cancer tissues
The
FOXF2 mRNA expression pattern in different subtypes of breast cell lines was validated using the Gene expression-based Outcome for Breast cancer Online (GOBO) data sets [
29].
Statistical analysis
Data from in vitro and in vivo experiments were presented as mean ± standard deviation (SD). Student’s t test was used to compare differences between the experimental group and the control group. Fisher’s exact test was used to compare the difference of metastatic incidence between 231-Luc-shFOXF2 mice and the 231-Luc-shControl mice. Spearman’s rank correlation was used to analyze the correlation between FOXF2 and TWIST1 mRNA levels in breast cancer tissues. Survival plots were created using Kaplan-Meier analysis, and log-rank test was used to assess statistical significance. P <0.05 was considered statistically significant.
Discussion
The FOX transcription factors have tissue-specific expression patterns and play critical roles in embryogenesis and tissue development through regulating tissue-specific gene expression and cell differentiation [
30]. Recent evidence has demonstrated that the deregulation of FOX factors is correlated with carcinogenesis and tumor progression [
21]. It is worth noting that several FOX transcription factors link to the biological characteristics of breast cancer. FOXA1 is highly expressed in luminal subtype breast cancer [
31], and its deficiency promotes the metastasis of this subtype cancer [
32]. FOXC1 [
19], FOXC2 [
4], and FOXQ1 [
20] are highly expressed in basal-like subtype breast cancer, but are less present in luminal subtype cancer. The ectopic expression of these mesenchymal FOX factors in normal mammary epithelial cells is able to induce EMT and lead to an aggressive phenotype. FOXF2 is a mesenchymal transcription factor and specifically expressed in the mesenchyme adjacent to the epithelium and controls mesenchymal differentiation to maintain the tissue homeostasis via epithelio-mesenchymal interactions [
22]. FOXF2 was found highly expressed in benign prostatic hyperplasia and the transition zone of the normal prostate, where less suffer prostate cancer [
33,
34]. In this study, we showed that FOXF2 is highly expressed in most basal-like breast cells. Since basal-like breast cells are characterized by the basal gene expression and with mesenchymal characteristics [
15], the expression pattern of FOXF2 in basal-like breast cells is consistent with its nature of mesenchymal transcription factors. We also found that FOXF2 deficiency promoted the metastasis of BLBC cells, which is consistent with our previous report that the under-expression of FOXF2 is associated with early-onset metastasis and poor prognosis of TNBC patients [
23]. Our finding indicates that FOXF2 may play a role in controlling the differentiation state of basal-like breast cells and inhibits the aggressive property of BLBC.
It is well known that a variety of embryonic and mesenchymal transcriptional factors function as EMT activators and are involved in the aggressive behavior of BLBC, including SNAIL1 [
16], SNAIL2 [
17], TWIST1 [
18], and FOX transcription factor family members FOXC1 [
19], FOXC2 [
4] and FOXQ1 [
20]. Most EMT-TFs were identified as activators of EMT and promoters of BLBC metastasis, while EMT suppressors are rarely identified. FOXA2 is one such rare EMT suppressor [
35]. In this study, we showed that FOXF2 is a novel EMT-suppressing transcription factor in basal-like breast cells. FOXF2 deficiency enhanced the metastatic ability of basal-like mammary epithelial cells and BLBC cells by inducing an EMT phenotype. We observed that the depletion of
FOXF2 results in E-cadherin and β-catenin loss from the cell membrane. It is known that during EMT progression, the loss of E-cadherin expression from adherent junction results in the release of β-catenin into the cytoplasm. In addition, we observed that the depletion of
FOXF2 leads to a significant increase in
MMPs expression. Mesenchymal cells and the cancer cells undergoing EMT secrete extracellular matrix (ECM)-degrading enzymes to facilitate invasion and motility [
36]. Matrix metalloproteinase (MMP)2 [
37], MMP3 [
38] and MMP9 [
39] are able to induce EMT, and they trigger a positive regulatory feedback loop that stabilizes EMT. Therefore, the translocation of β-catenin translocation and upregulation of MMPs caused by FOXF2 deficiency could lead to initiate vicious cycle of EMT.
TWIST1 is recognized as a critical modulator in EMT, and its overexpression is able to trigger EMT phenotype and tumor metastasis [
3]. Thus, clarifying the regulatory mechanism of TWIST1 expression would provide insight into the regulatory pathway of TWIST1-induced EMT programming and aid in developing molecular targets to inhibit tumor metastasis. It has been reported that HIF-1 [
40] and p65 [
41] transcriptionally activate
TWIST1 expression to mediate hypoxia-induced and tumor necrosis factor alpha (TNF-α)-induced EMT and tumor metastasis, respectively. miR-720 [
42] and miR-106b [
43] suppress TWIST1 expression by posttranscriptional regulatory mechanisms. In this study, we identified FOXF2 as a transcriptional suppressor of TWIST1. Therefore, we speculate that the two embryonic/mesenchymal transcription factors might act cooperatively to maintain tissue homeostasis through balancing the differentiation or dedifferentiation of mesenchymal/myoepithelial cells. FOXF2 deficiency could lead to more dedifferentiation of myoepithelial cells and an aggressive phenotype in BLBC via a TWIST1-mediated pathway. This novel finding provided an unanticipated regulatory pathway of TWIST-induced EMT and metastasis in BLBC.
In EMT programming, pleiotropic EMT-TFs form an interaction network to enable them to work in concert to regulate the EMT phenotype. The zinc-finger transcription factors SNAIL1 [
44], SNAIL2 [
45], ZEB1 [
46] and ZEB2 [
47] directly bind to the E-boxes of the
CDH1 promoter to repress its transcription. TWIST1 [
3], Goosecoid [
48], and FOXC2 [
4] trigger EMT by indirectly suppressing
CDH1 transcription. TWIST1 directly regulates
SNAIL2 transcription, and SNAIL2 is an essential mediator of TWIST1-induced EMT [
49]. TWIST1 can also induce FOXC2 and SNAIL2 expression [
7]. SNAIL1 has been found to activate ZEB1 [
2], FOXC2 and SNAIL2 expression [
7]. In this study, we found that FOXF2 not only negatively regulated
TWIST1 expression but also regulated the expression of
SNAIL1 and
SNAIL2,
ZEB1,
ZEB2 and
GSC in basal-like cell lines. Interestingly, FOXF2 regulated
TWIST1 expression in all the three cell lines, but only regulated
SNAIL2 in MCF-10A and BT-549 cells, regulated
ZEB1 and
GSC in MDA-MB-231 and BT-549 cells, regulated
ZEB2 in MCF-10A and MDA-MB-231 cells. Therefore, in addition to activating
TWIST1 transcription, FOXF2 deficiency may induce EMT through orchestrating different EMT-TFs expression in the different basal-like breast cells.
In addition to BLBC, we also tested the effect of FOXF2 on luminal breast cancer cells. However, we observed that the ectopic expression of FOXF2 could not significantly change the phenotype of luminal breast cancer cells, even slightly enhance the migration and invasion, and slightly increase the transcription of TWIST1. Our results suggest that there are different functions of FOXF2 in basal-like and luminal cells. Based on our current results, we speculate that mesenchymal FOXF2 might be induced expression in luminal cells that had undergone transdifferentiation into basal-like cells, and when FOXF2 depletion during the progression of the cells adapted mesenchymal characteristics could drive these cells to possess more mesenchymal stem-like plasticity and aggressive phenotype by activating the transcription of target gene TWIST1 and upregulating the expression of other EMT-TFs. The regulatory mechanisms of the different expression pattern and functions of FOXF2 in basal-like and luminal breast cancer cells should be further deeply investigated. Since FOXF2 plays an essential role in the metastasis of BLBC cells, the exogenous administration of FOXF2 or interference of FOXF2-TWIST1 pathway may be a promising therapeutic strategy for improving the outcome of BLBC patients.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
QSW participated in the study design, performed experiments, analyzed and interpreted data, and wrote the manuscript. PZK participated in the study design and performed experiments. XQL and FY contributed to acquisition of data and revising the manuscript critically for important intellectual content. YMF conceived the study, interpreted data, and wrote the manuscript. All authors read and approved the final manuscript.