Abstract
Stemness and epithelial–mesenchymal transition (EMT) are two fundamental characteristics of metastasis that are controlled by diverse regulatory factors, including transcription factors. Compared with other subtypes of breast cancer, basal-type or triple-negative breast cancer (TNBC) has high frequencies of tumor relapse. However, the role of alpha-globin transcription factor CP2 (TFCP2) has not been reported as an oncogenic driver in those breast cancers. Here, we show that TFCP2 is a potent factor essential for EMT, stemness, and metastasis in breast cancer. TFCP2 directly bound promoters of EGF and TGFα to regulate their expression and stimulate autocrine signaling via EGFR. These findings indicate that TFCP2 is a new antimetastatic target and reveal a novel regulatory mechanism in which a positive feedback loop comprising EGF/TGFα and AKT can control malignant breast cancer progression.
TFCP2 is a new antimetastatic target that controls TNBC progression via a positive feedback loop between EGF/TGFα and the AKT signaling axis.
Introduction
Tumor metastasis is a multistep process that occurs through local invasion, intravasation, transport, extravasation, and colonization, resulting in the dissemination of tumor cells from their primary site to a distant site where secondary tumors are formed. In the tumor metastasis process, epithelial–mesenchymal transition (EMT), and cancer stem cell (CSC)-like characteristics make primary contributions to drive tumor initiation and development (1). During the EMT process, epithelial cells shift toward a mesenchymal phenotype, with the loss of cell–cell contacts and adhesions and an increased ability for migration and invasion during morphogenesis (2). After the EMT program is activated, cancer cells can also exhibit high plasticity by acquiring stem-like traits that endow them with the potential for tumor initiation and metastasis, all of which promote cancer progression. Hence, tumor cells that undergo EMT and possess CSC-like features have greater metastatic potential and result in poorer outcomes in patients with cancer.
TFCP2 is a member of the CP2/Grainyhead family of transcription factors that are conserved throughout metazoans and fungi. In mammals, the CP2/Grainyhead family consists of two distinct subfamilies: the first includes three Grainyhead-like factors currently termed GRHL1–3, and the other subfamily consists of three factors known as TFCP2 (CP2c), TFCP2L1 (CRTR1), and UBP1 (CP2a and CP2b; refs. 3, 4). Among these, TFCP2, also known as LSF, was first identified as a transcriptional activator factor of the Simian virus 40 late promoter in HeLa cells (5) and of the murine α-globin promoter (6). As mentioned earlier, TFCP2 regulates a diverse range of cellular and viral promoters (7, 8), is expressed in all mammalian cell types, and plays an important role in cell-cycle regulation (8). It facilitates entry into G1–S-phase of the cell cycle, promotes DNA synthesis, and functions as an antiapoptotic factor (9). Overexpression of TFCP2 may promote transformation and cancer cell survival. Recent studies suggest that TFCP2 may also play a role in the pathogenesis of colon, lung, and hepatocellular carcinoma (10–12). Moreover, TFCP2 has been shown to activate osteopontin and matrix metalloproteinase-9 expression to regulate invasion, metastasis, and angiogenesis in hepatocellular carcinoma (HCC; refs. 13, 14). Although the expression and regulatory roles of TFCP2 have been reported individually in some types of cancer, there is no evidence of TFCP2 involvement in metastatic triple-negative breast cancer (TNBC) progression.
The EGFR is a major oncogene identified in a variety of human cancers, including breast cancer (15–17). These receptors are activated by ligand binding and consequent receptor homo- and heterodimerization, which leads to activation of the kinase domain, auto- and transphosphorylation of their intracellular domains, and the initiation of signaling (18). Many different ligands, including EGF-like molecules, TGFα, and neuregulins, activate the receptor by binding to the extracellular domain and inducing the formation of receptor homodimers or heterodimers (19). Genes in the EGF signaling pathway are among the most frequently activated oncogenes in breast cancer (20). Clinical studies have shown that TNBC is more aggressive than other breast cancers and patients with TNBC have a worse prognosis than those with other breast cancer subtypes (21). More than 60% of TNBCs express EGFR, which may serve as a prognostic marker for TNBC outcomes (22). Although EGFR has been widely studied in breast cancer tumorigenesis, the mechanism underlying TFCP2-induced malignancy remains unknown. The goal of this study was to determine the role of TFCP2 in breast cancer progression, with a focus on EMT, stemness, and metastasis. Using basal-like breast cancer cell lines and NOD/SCID gamma (NSG, 5–6 weeks old) mouse models, we investigated the mechanisms by which TFCP2 affects the EGFR signaling axis positively or negatively in TNBCs.
Materials and Methods
Materials and reagents
All breast cell lines (MCF10A, MCF7, SKBR3, T47D, BT474, MDA-MB-453, MDA-MB-231, BT549, Hs578T, and MDA-MB-361) were purchased from ATCC, cultured in the indicated media, and incubated at 37°C with 5% CO2 according to standard protocols. They were routinely tested for Mycoplasma contamination using PCR methods and cultured cell lines were often treated with Plasmocin treatment (InvivoGen) to remain contamination free. Cells were used at least less than 20 passages number for 3 months, but were not independently authenticated. After each thawing, cells were confirmed as Mycoplasma negative prior to experiments. The p-EF1α (control vector), p-EF1α-CP2 [TFCP2 wild-type (WT) vector], TFCP2 D153A, and LSFdn vectors were received from Department of Life Science, Hanyang University (Seoul, South Korea), and SNAI1 (pBabe puro Snail) and pGL3-basic vectors were purchased from Addgene. EGF (#AFL236) and TGFα (#239-A) were purchased from R&D Systems, Inc. Antibodies and inhibitors' information are listed in Supplementary Table S1.
Transfection
p-EF1α, p-EF1α-CP2, TFCP2 D153A, LSFdn, and SNAI1 vectors or siRNAs were transfected into the appropriate cells using Lipofectamine Reagents (Invitrogen) according to the manufacturer's instructions. Cells were harvested 48 hours after transfection for subsequent experiments. All siRNAs were purchased from Genolution Pharmaceuticals, Inc.
Scratch, soft agar, morphology (collagen coating), and transwell assays
For the wound-healing (scratch) assay, cells were seeded in 35-mm cell culture dishes and cultured until 80%–90% confluent. Afterwards, a 200-μL pipet tip was used to make a scratch wound across the middle of the cell monolayer. Images were taken with Olympus IX71 Fluorescence Microscope (Olympus) immediately after and 24 hours after the scratch was made. The rate of cell migration from at least three independent experiments was calculated with ImageJ. To examine anchorage-independent growth, cells were suspended in 0.4% agar in growth medium and analyzed as described previously (23). For morphologic analysis, cells were seeded at medium confluency directly into Corning BioCoat Collagen I-Coated Plates (Corning Inc.) and photographed 24 hours later. Migration assays were performed using Boyden Chambers (Corning Inc.). Cells (2 × 104) in 200 μL of serum-free medium were seeded into the top chamber, and 800 μL of medium with 10% FBS was added to the bottom chamber as a chemoattractant. Migratory cells were stained using a Diff-Quick Kit (Thermo Fisher Scientific) then imaged and counted. The invasion assays were carried out in accordance with the migration assays except that each transwell chamber was coated with growth factor–reduced Matrigel (BD Biosciences). All experiments were performed in triplicate.
Spheroid assays
For the sphere formation assays, sphere size was determined using Motic Images Plus 2.0 Software (Motic) in three randomly chosen visual fields each day until day 4 after the cells were seeded. For colony formation assays, single-cell suspensions containing breast cells were plated in 96-well cell culture plates and colony formation was observed at different time points for 2 weeks. Colonies were photographed and their diameter was measured (24).
IHC analysis
For IHC experiments, paraffin-embedded tissue sections were deparaffinized in xylene and then rehydrated in a graded series of ethanol (95%, 90%, 80%, and 70%), followed by PBS treatment. Epitopes were retrieved with 20 mg/mL proteinase K in PBS containing 0.1% Triton X-100. Sections were incubated overnight with appropriate primary antibodies at 4°C and then processed as described previously (25).
RNA preparation and qPCR
Total RNA was prepared using TRIzol Reagent (Invitrogen), and RNA quality was measured using a NanoDrop Spectrophotometer (ND1000, NanoDrop Technologies). qRT-PCR was performed using a KAPA SYBR FAST qPCR Kit (KAPA Biosystems) according to the manufacturer's procedures. Reactions were performed in a Rotor Gene Q Instrument (Qiagen), and the results were expressed as the fold-change relative to the control sample calculated using the ΔΔCt method. β-actin served as an internal normalization control. All primers were purchased from DNA Macrogen. All primers used in this study are listed in Supplementary Tables S2 and S3.
Western blotting
Total cellular protein was extracted in cold lysis buffer [Tris–HCl (40 mmol/L, pH 8.0), NaCl (120 mmol/L), and Nonidet-P40 (0.1%)] enriched with protease inhibitors and was quantified using a BSA assay. Protein lysates were separated by SDS-PAGE and transferred onto nitrocellulose membranes (Amersham). The membranes were then blocked in PBST containing 5% milk and probed with the indicated primary antibody followed by the corresponding horseradish peroxidase–conjugated secondary antibody. Finally, the protein bands were detected using chemiluminescence (Amersham) according to the manufacturer's instructions.
Flow cytometry analysis
To assess cell death, cells were incubated in the listed conditions for the desired time, after which, they were labeled with propidium iodide (Sigma, 50 ng/mL), incubated for 20 minutes at 4°C, and analyzed immediately. To detect the cancer stemness marker CD44 and CD24, 1 × 106 of TFCP2 knockdown cells were harvested by trypsin digestion, washed, and resuspended in PBS. An R-phycoerythrin–conjugated anti-CD44 mAb and FITC-conjugated anti-CD24 antibody (Miltenyi Biotec Inc.) were used for detection. All data were analyzed using CellQuest Software (BD Biosciences) and repeated three times.
ELISA
Cellular protein was prepared as Western blotting. The concentrations of total and phosphorylated EGFR were measured using human EGFR (pY1068) and total EGFR ELISA Kits (Abcam), respectively, according to the manufacturer's instructions.
Chromatin immunoprecipitation assays
Prior to performing chromatin immunoprecipitation (ChIP) experiments, cells were cross-linked with 4% paraformaldehyde. ChIP assays were performed using an EZ-ChIP Kit (EMD Millipore) according to the manufacturer's instructions. Immunoprecipitation was performed using an anti-TFCP2 antibody or a rabbit isotype control IgG (Upstate Biotechnology). PCR was performed using primers specific to the EGF and TGFα gene promoter regions shown in Supplementary Table S3.
Construction of luciferase reporter plasmids
The human EGF and TGFα promoter regions were obtained by PCR, using genomic DNA from MDA-MB-231 cells as a template. The promoters of EGF and TGFα were generated by PCR using primers shown in Supplementary Table S3. They were subcloned into XhoI and HindIII sites of pGL3-basic vector (Addgene). All constructs were verified by sequencing.
Luciferase reporter assay
HEK293T cells were seeded in a 24-well plate, at 60%–70% confluence and cells were cotransfected using Lipofectamine 2000 according to the manufacturer's manual. In brief, each well was transfected with 300 ng of reporter constructs and 300 ng of pRL-CMV-Renilla plasmid (Promega) for 48 hours. Luciferase activity was measured using a Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer's instructions and normalized to Renilla luciferase activity. All experiments were performed in triplicate.
In vivo xenografts and metastasis assays
All animal procedures were performed according to the guidelines of the Institutional Animal Care and Use Committee of Academia Sinica. NSG mice (5–6 weeks old) were obtained from Orient Bio. For these experiments, 40 μL of metastatic MDA-MB-231-LM1 breast cancer cells (1 × 106), which were derived from lung lesions after colonization (shCtrl-LM1 and shTFCP2-LM1), were injected into the fourth mammary fat pad of NSG mice. Twice per week, the mice were weighed and tumor size was determined using a digital caliper. Tumor volumes were assessed by measuring the length (l) and width (w) and calculated using the following formula: (shortest diameter2) × (longest diameter/2). Mice were sacrificed 8–12 weeks after injection, and the lungs were removed and fixed in 9% paraformaldehyde. Detectable tumor nodules on the surface of the entire lung were counted to calculate the metastatic index. Tumor tissues were homogenized, and the expression of target genes was analyzed by Western blotting and qRT-PCR.
Immunofluorescence
For immunofluorescence staining, cells were plated onto glass coverslips, fixed with 4% paraformaldehyde, and permeabilized with 0.1% Triton in PBS. Cells then incubated overnight at 4°C with the appropriate primary antibody. The following day, Alexa Fluor 488–conjugated anti-rabbit or anti-mouse and Alexa Fluor 546–conjugated anti-rabbit or anti-mouse (Molecular Probes) secondary antibodies were used to visualize the proteins. Cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (Sigma). Immunostained cells were observed using an IX71 Fluorescence Microscope (Olympus).
Human tissue microarrays
Breast tissue microarray samples were obtained from US Biomax (BR1101, BR20814, and BR1509). Healthy specimens were also included in each of the array blocks. Samples were reviewed by a pathologist to confirm the diagnosis of breast carcinoma, histologic grade, and tumor purity. TFCP2, EGF, and TGFα levels were graded as 0, 1, 2, or 3 according to the intensity scores.
Gene set enrichment analysis, dataset evaluation, and Kaplan–Meier analysis
Gene set enrichment analysis (GSEA) was performed on diverse gene signatures by comparing gene sets from either the Molecular Signature Database (MSigDB) or published gene signatures. To analyze the expression of TFCP2, TFCP2L1, and UBP1 in breast invasive carcinomas, previously published microarray data under accession codes GSE41313, GSE1456, GSE20713, GSE7513, GSE2603, and GSE25055 were reanalyzed. To examine the prognostic value of TFCP2, EGF, and TGFα, patient samples were divided into two groups (low and high expression) for each gene, which were analyzed using the Kaplan–Meier plot program (http://kmplot.com/analysis/) as described previously (26).
Statistical analysis
All experimental data are presented as the mean ± SD of at least three independent experiments. Statistical analyses were performed using an unpaired two-tailed parametric Student t test. Multiple group comparisons were made by ANOVA using PRISM 8.0 Software (GraphPad). Variances were confirmed to be similar between groups that were being statistically compared, and P < 0.05 was considered significant. No samples were excluded from the analysis. The investigators were not blinded to allocation during experiments and outcome assessments.
Results
TFCP2 is upregulated and associated with poor survival in patients with breast cancer
To explore the association of CP2 family transcription factors in breast cancer, we utilized an online database screening system to investigate their expression in normal and breast cancer tissues using gene expression profiling interactive analysis (27). Assessment of the dataset showed that TFCP2 and UBP1 expression is comparatively higher in breast tumors than in normal tissues, but not TFCP2L1 (Fig. 1A). To verify further, additionally, GSEA was performed with data from the MSigDB, which showed that TFCP2 expression is well-correlated with aggressive basal subtypes of breast cancer compared with the luminal type (Fig. 1B; Supplementary Fig. S1) while UBP1 does not seem so, suggesting the possible involvement of TFCP2 in basal type breast tumors. Consistent data were also observed in basal and luminal breast cancer subtypes using a Gene Expression Omnibus (GEO) dataset (28) and Gene Expression-Based Outcome for Breast Cancer Online (29) datasets (Fig. 1C and D). In agreement with the online database findings, quantification of TFCP2, TFCP2L1, and UBP1 mRNA expression shows that the basal-type cell lines expressed high levels of TFCP2 compared with the luminal cell lines (Fig. 1E). After validating the specificity of TFCP2 levels in malignant breast cancer, we confirmed its expression in human tissue samples using tissue microarray analysis. Notably, increased TFCP2 expression was observed in basal and high-grade breast carcinomas (Fig. 1F and G). Furthermore, Kaplan–Meier survival analysis (30) determined that high expression of TFCP2 correlates with poor survival in patients with breast cancer independently among basal, luminal, and HER2 subtypes (Fig. 1H). These observations strongly suggest that TFCP2 is primarily associated with malignant breast cancer.
TFCP2 enhances EMT, metastasis, and stemness in breast cancer
To discover the pathologic mechanism of TFCP2 upregulation in breast cancer, we screened hallmarks of cancer with the GSEA database in breast cancer. Gene ontology enrichment analysis showed that TFCP2 is positively correlated with signature gene sets relating to EMT and cancer stemness (Fig. 2A). When we ectopically introduced TFCP2 in luminal MCF7 and SKBR3 breast cancer cells to examine EMT, MCF7 and SKBR3 cells showed a reduction in epithelial features and an increase in mesenchymal features as evidenced by the elongation of cells on the collagen-coated surface, enhanced cell invasion and migration abilities and the induction of fibronectin (FN), N-cadherin (CDH2), and vimentin (VIM) expression (Fig. 2B–E; Supplementary Fig. S2A–S2D). Similar results were also observed in MCF10A normal breast epithelial cells with TFCP2 overexpression (Supplementary Fig. S2E–S2H). Prior to the TFCP2 overexpression experiments in MCF7, additional experiments with TFCP2 family members TFCP2L1 and UBP1 were performed, which confirmed that silencing of these genes did not affect the migration and invasion capabilities of Hs578T and MDA-MB-453 cells, which expresses high level of TFCP2L1 and UBP1, respectively (Supplementary Fig. S3A and S3B). In addition, TFCP2 knockdown (with siRNA#1 and siRNA#2) mitigated these effects in both MDA-MB-231 and BT549 basal-type breast cancer cells (Supplementary Fig. S3C–S3H). Recent studies have documented that the acquisition of CSC traits occurs with the EMT program (31, 32) and the induction of EMT can induce many of the defining characteristics of stem cells, including self-renewal (33). In accordance with these studies and our GSEA analysis data (Fig. 2A), we tried to determine the involvement of TFCP2 in the acquisition of stemness in breast cancer cells. The sphere-forming or clonal efficiency of breast cancer cells was dramatically reduced after TFCP2 knockdown in both EGF-treated (a well-known EMT inducer) MCF7 and SKBR3 cells in a sphere-permissive medium (Fig. 2F–I). The inhibitory effect of TFCP2 on stem-like breast cells was observed until several passages, as evidenced by limiting dilution sphere-forming assays with MCF7 cells (Supplementary Fig. S3I). We also screened several CSC-related transcription factors, such as SOX2, NANOG, and OCT4. qRT-PCR and immunofluorescence data showed that OCT4 was prominently affected along with CD44 after TFCP2 knockdown in spheres formed from MCF7 or SKBR3 cells (Supplementary Fig. S3J and S3K). EMT-related transcription factors have been reported to boost the CD44+/CD24− subpopulation, as observed in breast CSCs (34). In agreement, flow cytometry analysis revealed that the percentage of CD44+ and CD24− cells were decreased in both MCF7 and SKBR3 cells after TFCP2 knockdown (Fig. 2J). In addition, we also performed sphere formation assay and checked percentage of CD44+/CD24− by FACS analysis in TFCP2 overexpression system using MCF7 cells. We found that overexpressed TFCP2 can induce cancer stemness in those MCF7 cells (Supplementary Fig. S3L and S3M). On the other hand, interestingly, silencing of TFCP2 did not affect breast cancer cell proliferation, as confirmed by soft agar–independent cell growth and cell death assays (Supplementary Fig. S3N–S3P). These results showed that TFCP2 has the potential to regulate EMT and the CSC phenotype rather than cell proliferation and death in breast cancer cells.
Silencing TFCP2 expression blocks EMT and stemness in vivo
We further investigated the effect of TFCP2 knockdown on metastasis in a breast cancer mouse models by injecting LM1-MDA-MB-231 cells into the fat pad (Fig. 3A). As shown in the knockdown experiments, diminished TFCP2 expression reduced the formation of lung metastatic foci with inhibition of EMT- and CSC-related transcription factors and markers such as FN, CDH2, VIM, OCT4, and CD44 (Fig. 3B–J). However, tumor growth remained unaffected in these xenografts after TFCP2 knockdown (Supplementary Fig. S4A–S4C). Measurement of Ki67 and cleaved caspase 3 in shTFCP2-injected mouse tissues confirmed that TFCP2 did not induce any cell death in breast tumors (Supplementary Fig. S4D and S4E). These data further revealed a critical role of TFCP2 in maintaining mesenchymal features to sustain EMT and metastasis with stemness in basal-type breast cancer without affecting growth or cell death.
Because TFCP2 knockdown affects SNAI1 expression more effectively than the other EMT transcription factors tested, we hypothesized that TFCP2 mediated metastatic activity through the SNAI1 pathway. We knocked down TFCP2 with siRNA in SNAI1-overexpressing MDA-MB-231 basal-type breast cancer cells. Consistent with previous experiments, exogenous SNAI1 expression recovered TFCP2-mediated inhibition of migration, invasion, and expression of the mesenchymal markers FN, CDH2, and VIM in those cells (Supplementary Fig. S4F–S4H). Similar effects were observed when MCF7 cells in spheroids were treated with similar conditions. When overexpressing SNAI1, MCF7 cells with TFCP2 knockdown regained the ability to form spheres and exhibited a stemness phenotype along with CD44+ cells (Supplementary Fig. S4I–S4L). Thus, we suggest that SNAI1 drives TFCP2-induced EMT and stemness, thereby sustaining metastatic programming in basal-type breast cancer cells.
TFCP2 regulates EGFR signaling activation in breast cancer
To achieve mechanism selectivity, GSEA analysis was performed to investigate which signaling pathways are involved in the TFCP2-induced prometastatic effects in basal-type breast cancer. We found that there is a highly positive correlation between TFCP2 expression and EGFR signaling along with its downstream signaling pathway (Fig. 4A). Next, we performed ELISA and Western blot assays to determine whether TFCP2 regulates EGFR activity. Both assays showed that TFCP2 enhanced EGFR protein level activity based on the TFCP2 silencing and overexpression experiments; however, the EGFR mRNA expression levels did not change (Fig. 4B–D; Supplementary Fig. S5A and S5B). EGFR is widely involved in a variety of cellular processes, including proliferation, motility, and survival, and can be activated by a variety of polypeptide ligands, such as EGF and HBEGF. Because TFCP2 regulates EGFR activity, we assessed the expression level of several EGFR ligands, including EGF, TGFα, epiregulin (EREG), amphiregulin (AREG), epithelial mitogen (EPGN), heparin-binding EGF (HBEGF), and betacellulin (BTC), all of which can activate EGFR. Silencing and overexpressing TFCP2 revealed that TFCP2 more profoundly affects EGF and TGFα than all the other ligands tested in MDA-MB-231 and MCF7 cells, respectively (Fig. 4E; Supplementary Fig. S5C). Subsequently, we hypothesized that if there is any correlation between TFCP2 and both the EGF and TGFα ligands. We screened breast cancer cohorts through GEO databases and found a positive correlation of TFCP2 with both EGF and TGFα (Fig. 4F). Luciferase assays confirmed the binding of TFCP2 on EGF and TGFα promoters, suggesting their role in TFCP2-induced EGFR signaling (Fig. 4G). To further confirm this correlation, we predicted TFCP2-binding sites to EGF and TGFα promoters using the JASPAR online tool (http://jaspar.genereg.net) and performed ChIP assays using MDA-MB-231 and Hs578T cell lines. We found that TFCP2 can directly bind to the F2 fragment site of the EGF promoter and F1-4 fragment sites of the TGFα promoter in both cells (Fig. 4H and I). In addition, when TFCP2 D153A (which mutated aspartate (D) to alanine (A) at the TFCP2 DNA binding site), LSF dominant negative (LSFdn, double amino acid substitution mutant of LSF that is unable to bind DNA, initially named 234QL/236KE; ref. 9), and TFCP2 WT constructs were overexpressed in MCF7 cells, we observed that only TFCP2 WT significantly upregulated EGF and TGFα mRNA expression along with EGFR protein activity; however, there was no change observed in cells overexpressing TFCP2 D153A or LSFdn, which was similar to the levels in the control cells (Supplementary Fig. S5D and S5E). ChIP assay also confirmed that neither TFCP2 D153A nor LSFdn can bind to EGF or TGFα promoters (Supplementary Fig. S5F). It is worth mentioning that inhibiting EGFR signaling in TFCP2-overexpressing MCF7 cells suppressed signature genes related to EMT and CSC, which further suggests that TFCP2-induced EGFR activation regulates both phenotypes in breast cancer (Supplementary Fig. S5G).
Rescue experiments were performed in both MDA-MB-231 and Hs578T cell lines with silenced TFCP2 in the presence or absence of recombinant EGF/TGFα. Interestingly, EMT-related SNAI1, CDH2, and VIM protein expression; migration; and invasion were decreased upon silencing TFCP2; these decreases were rescued again with the treatment of EGF or TGFα in MDA-MB-231 and Hs578T cells (Fig. 4J; Supplementary Fig. S5H). Similar effects were observed in TFCP2-overexpressing MCF7 cells with or without EGF/TGFα silencing (Supplementary Fig. S5I). Taken together, these results indicated that TFCP2 can directly upregulate EGF and TGFα expression to activate the EGFR signaling pathway in metastatic breast cancer cells.
A positive feedback loop exists between TFCP2 and the EGFR/PI3K/AKT axis
Because TFCP2 could regulate EGF/TGFα expression to activate EGFR activation, we investigated the upstream regulator of TFCP2. To this end, we examined TFCP2 expression in the presence of various signaling pathway inhibitors, such as U0126 (a MEK1/2 inhibitor), a JNK inhibitor, SB203580 (a p38 MAP kinase), a JAK inhibitor, WP1066 (a STAT3 inhibitor), LY294002 (a PI3 kinase/AKT inhibitory), and PP2 (a SRC inhibitor). Prior to this experiment, the efficacy of the inhibitors was confirmed by Western blot analysis (Supplementary Fig. S6A). These analyses revealed that inhibiting the AKT pathway downregulates TFCP2 mRNA levels to a greater extent than inhibition of other pathways in MDA-MB-231 cells (Fig. 5A). To further confirm that the AKT pathway was blocked in the same cells with AKT-targeted siRNA or LY294002, the TFCP2 protein level was measured and shown to be dramatically reduced as observed by Western blot analysis (Fig. 5B; Supplementary Fig. S6B). As we found above, EGFR can be activated through TFCP2, and we next asked whether AKT could regulate EGFR through TFCP2. Knockdown or pharmacologic blockade of AKT attenuated EGFR and TFCP2 activity in MDA-MB-231 cells, while this attenuation was rescued in the presence of TFCP2 overexpression. The expression of SNAI1, FN, CDH2, CD44, and OCT4 in MDA-MB-231 cells was also affected by LY294002 treatment, which was rescued with TFCP2 overexpression (Fig. 5C and D; Supplementary Fig. S6C). This result supported the view that AKT serves an essential upstream regulator of TFCP2, thereby inducing EGFR activation to promote metastatic activity. Earlier, we observed that concomitant treatment with recombinant EGF/TGFα increases TFCP2 expression in TFCP2-silenced MDA-MB-231 cells compared with that in cells with TFCP2 knockdown alone (Fig. 4J). Hence, we hypothesized whether there is any positive feedback loop such as EGFR activation can also regulate TFCP2 expression in breast cancer. To verify this, we analyzed TFCP2 expression in the presence of either si-EGFR or an EGFR inhibitor (AG1478). Both approaches decreased the protein activity and mRNA expression of TFCP2 in MDA-MB-231 cells (Fig. 5E; Supplementary Fig. S6D and S6E), suggesting potential involvement of EGFR in TFCP2 induction. Similar observations were also made in BT549 cells by inhibiting EGFR on TFCP2 levels (Supplementary Fig. S6F). To evaluate whether the EGFR/PI3K/AKT axis is a true upstream candidate of TFCP2 feedback regulation, we treated MCF7 cells with recombinant EGF/TGFα in the presence or absence of LY294002. TFCP2 expression was greatly enhanced in the presence of EGF/TGFα alone, but this effect was reduced when LY294002 was added to the cells (Fig. 5F). These data indicated that TFCP2 expression is upregulated by the EGFR/PI3K/AKT axis in metastatic breast cancer cells through a positive feedback loop.
Correlation between TFCP2 and EGF/TGFα expression in patient samples
To evaluate whether our in vitro findings are correlated with human breast cancer and metastasis formation, we assessed the prognostic value of TFCP2 expression with EGF and TGFα in breast cancer patient samples. To this end, we performed IHC analysis to examine the coexpression of TFCP2, EGF, and TGFα in tumors derived from patients with breast carcinoma (n = 150). Tissue array data showed that expression of TFCP2, EGF, and TGFα was significantly higher in tumor tissues than in matched normal tissues (Fig. 6A). The increased TFCP2 expression correlated well with the augmented expression of EGF and TGFα (Fig. 6B). Furthermore, Kaplan–Meier survival analysis revealed that the survival time of patients with high TFCP2 and either EGF or TGFα expression was shorter than those with low expression of these genes (Fig. 6C and D). In addition, high expression of TFCP2, EGF, and TGFα was found to be significantly associated with poor outcomes in patients with breast cancer, as shown in Fig. 6E. Overall, clinical dataset analysis indicates that TFCP2 levels are positively correlated with EGFR signaling in breast tumors, and higher TFCP2 levels are associated with reduced metastasis-free survival in breast tumors and an increased probability of poor overall survival in EGF/TGFα-high tumors.
Discussion
Here, we demonstrated that TFCP2 functions as an activator of prometastatic transcription factors by directly regulating the expression of the EGFR ligands EGF and TGFα, resulting in EGFR activation; this signaling cascade is a critical determinant of oncogenic EGFR signaling, leading to poor long-term survival in patients with breast cancer. Although various studies have shown that EMT affects the migratory potential of metastatic breast cancer cells and subsequent metastasis, many critical factors regarding how these tumor cells access this fundamental cellular event remain unknown. In this study, we found that the transcription factor, TFCP2, promotes breast cancer cells to acquire increased migration/invasion abilities and CSC traits via the EGFR signaling pathway, but has no effect on tumor growth and apoptosis. As we know, EGFR overexpression is frequently observed in TNBC (16, 35–37), the aggressive behavior of TNBC and the lack of established clinical treatment targets create a major challenge in treating these patients. In this article, we showed that high levels of TFCP2 expression in TNBC cells increased the activation of the EGFR signaling pathway. In this pathway, ligands play several significant roles at different levels to promote invasion and metastasis (38). Furthermore, our ChIP assay data suggested that TFCP2 directly binds to the promoters of the EGFR ligands EGF and TGFα (Fig. 4H and I). Dysregulation of the EGFR pathway via overexpression or constitutive activation can promote processes related to tumor progression, such EMT and stemness, and is associated with poor prognosis in breast malignancies (39–41). Hence, inhibiting TFCP2 expression, which can suppress EGF and TGFα expression, may be effective at preventing EGFR activation in TNBC.
In a previous study, TFCP2 was shown to be overexpressed in HCC and to target FN1 and TJP1 to regulate HCC metastasis (42). Moreover, TFCP2 can affect pancreatic cancer cell growth, invasion, and migration (43). However, in these studies, the mechanism by which TFCP2 expression is regulated was not fully addressed. In this study, we found that TFCP2 expression was mediated by the EGFR/PI3K/AKT axis in TNBC, which caused a positive feedback loop (Fig. 5). However, there are some limitations to this study. First, we identified a signaling pathway that can regulate TFCP2 expression, but the specific regulators of TFCP2 expression remain unclear. Second, although TFCP2 induced both EGF and TGFα expression to activate EGFR, we could not distinguish which ligand was more important for EGFR signaling. Third, TFCP2 regulated the EGF/TGFα/EGFR axis in breast cancer; however, we do not know whether this also occurs in other human malignancies. Further studies are required to determine the role of TFCP2 in other cancer types.
In summary, we describe the functions of TFCP2 as an oncogenic driver in TNBC. In addition, we identified the underlying mechanisms whereby TFCP2 regulates EMT and CSC activity in breast cancer through the EGFR/AKT axis. In turn, TFCP2 expression is also regulated by the EGFR/AKT axis (Fig. 6F). Enhanced expression of TFCP2 was associated with poor patient survival and therefore lends itself as a potential target for metastatic breast cancer treatment.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: Y. Zhao, C.G. Kim, S.-J. Lee
Development of methodology: Y. Zhao, N. Uddin, S.-J. Lee
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Zhao, J.-H. Kang, N.K. Kaushik, S.-J. Lee
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Zhao, N.K. Kaushik, S.H. Son, S.-J. Lee
Writing, review, and/or revision of the manuscript: Y. Zhao, N. Kaushik, S.H. Son, S.-J. Lee
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Zhao, J.-H. Kang, M.-J. Kim, S.-J. Lee
Study supervision: C.G. Kim, S.-J. Lee
Acknowledgments
We are thankful to all the research participants who participated in this study. This study was supported by the Bio & Medical Technology Development Program of the National Research Foundation funded by the Korean government (MSIT; 2019M3E5D1A01069361 to S.-J. Lee).
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.