FOXA2-Interacting FOXP2 Prevents Epithelial-Mesenchymal Transition of Breast Cancer Cells by Stimulating E-Cadherin and PHF2 Transcription

Abstract

FOXP2, a member of forkhead box transcription factor family, was first identified as a language-related gene that played an important role in language learning and facial movement. In addition, FOXP2 was also suggested regulating the progression of cancer cells. In previous studies, we found that FOXA2 inhibited epithelial-mesenchymal transition (EMT) in breast cancer cells. In this study, by identifying FOXA2-interacting proteins from FOXA2-pull-down cell lysates with Mass Spectrometry Analysis, we found that FOXP2 interacted with FOXA2. After confirming the interaction between FOXP2 and FOXA2 through Co-IP and immunofluorescence assays, we showed a correlated expression of FOXP2 and FOXA2 existing in clinical breast cancer samples. The overexpression of FOXP2 attenuated the mesenchymal phenotype whereas the stable knockdown of FOXP2 promoted EMT in breast cancer cells. Even though FOXP2 was believed to act as a transcriptional repressor in most cases, we found that FOXP2 could activate the expression of tumor suppressor PHF2. Meanwhile, we also found that FOXP2 could endogenously bind to the promoter of E-cadherin and activate its transcription. This transcriptional activity of FOXP2 relied on its interaction with FOXA2. Furthermore, the stable knockdown of FOXP2 enhanced the metastatic capacity of breast cancer cells in vivo. Together, the results suggested that FOXP2 could inhibit EMT by activating the transcription of certain genes, such as E-cadherin and PHF2, in concert with FOXA2 in breast cancer cells.

Keywords: E-cadherin; FOXA2 transcription factor; FOXP2 transcription factor; PHF2; epithelial-mesenchymal transition of breast cancer.

Figure 1

Identification of FOXA2-interacting transcription factor FOXP2 in breast cancer cells. (A) The lysates of MCF-7 cells transfected with pAvi-FOXA2, pEF1-BirA, or both, were incubated with streptavidin resin. The pull-down proteins were separated by PAGE and visualized with coomassie brilliant blue staining. The position of FOXP2 was pointed out in the lane of the FOXA2+BirA sample on the gel. (B, C) The interaction between FOXA2 and FOXP2 in cells. The lysates of cells transfected with pAvi-FOXA2 and pHis-FOXP2 or pEF1-BirA, were pulled down with streptavidin resin (B) or Ni beads (C). Western blotting was performed with anti-FOXA2 antibody, anti-FOXP2 antibody, or streptavidin-conjugated HRP (SA-HRP) respectively. (D, E) Mapping the regions mediating FOXA2-FOXP2 interaction in the proteins. The lysates of cells transfected with pHis-FOXP2 plus one of expressing plasmid vectors containing Flag-tagged FOXA2 1-165aa, 166-324aa, and 409-633aa (D), or the lysates of cells transfected with pCMV-FOXA2 plus one of expressing plasmid vectors containing Flag-tagged FOXP2 1-196aa, 197-408aa, 325-463aa (E), were immunoprecipitated with anti-Flag antibody. Western blotting was performed with anti-FOXP2 antibody, anti-FOXA2 antibody, or anti-Flag antibody respectively. The top panel showed a schematic representation of the regions in FOXA2 protein (D) and FOXP2 protein (E). (F) The interaction of endogenous FOXA2 and FOXP2 was detected by the immunofluorescent microscopy in MCF-7 cells. The staining for FOXA2 (red), for FOXP2 (green), for DAPI (blue), and the merge of the FOXA2 and FOXP2 staining were shown. The cells were pictured by TE2000S (Nikon, 200×). (G) The levels of FOXP2 in the four subgroups (Basal, Her2, LumA, and LumB) of breast cancer and normal breast tissue in the clinical breast cancer samples (the BRCA data set, n=394). (H, I) The gene expression correlation analysis of FOXA2 and FOXP2 in the BRCA data set (n=394) (H) and in the Basal subgroup (n=98) (I) were executed by using ggstatsplot package through R project. The pink and green histograms represented the distribution of BRCA samples based on the levels of FOXA2 and FOXP2 expression respectively.

Figure 2

FOXP2 inhibited the EMT of breast cancer cells. (A) The changes of the morphology at day 6 post epidermal growth factor (EGF) treatment (100 ng/ml) in MCF-7 cells were pictured by TE2000S (Nikon, 200×). (B, C) The expression of FOXP2 was downregulated during EGF-induced epithelial-mesenchymal transition (EMT) in MCF-7 cells. MCF-7 cells were treated with EGF (100 ng/ml) and harvested at different time points (day 0, day 3, day 6) post the treatment. The levels of FOXP2, E-cadherin, and Vimentin were examined by Western blotting (B) and qPCR (C), respectively. (D–F) Knockdown of FOXP2 enhanced EMT in MCF-7 cells. MCF-7 cells were infected with one of lentiviral vectors expressing various FOXP2-specific shRNA (Lv-shFOXP2#1, #2, #3) or the control lentiviral vector (Lv-shCon). The stable FOXP2-knockdown MCF-7 cells were selected and harvested for protein and total RNA preparation. The levels of FOXP2, E-cadherin, and Vimentin were examined by Western blotting (D) and qPCR (E). The migration ability of Lv-shCon-infected MCF-7 cells (shCon) and Lv-shFOXP2#2-infected cells (shFOXP2) was measured by the Transwell invasion test and the statistical data are shown on the right (F). (G–I) Overexpression of FOXP2 abolished the mesenchymal phenotype of MDA-MB-231 cells. MDA-MB-231 cells were transfected with pHis-FOXP2 or the control vector pCMV-EGFP. The levels of FOXP2, E-cadherin, and Vimentin were examined by Western blotting (G) and qPCR (H). The migration ability of control vector transfected MDA-MB-231 cells (231) and FOXP2-expressing MDA-MB-231 cells (FOXP2) was measured by the Transwell invasion test and the statistical data are shown on the right (I). The asterisks indicate statistically significant changes: *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

Figure 3

FOXP2 stimulated the expression of PHF2 during MET of breast cancer cells. (A, B) The expression levels of PHF2 and FOXP2 were upregulated during CTx-induced MET of MDA-MB-468 breast cancer cells. MDA-MB-468 cells were treated with different concentrations of CTx (50 or 100 ng/ml) for 7 days. The levels of FOXP2, PHF2, E-cadherin, and Vimentin were examined by Western blotting (A) and qPCR (B), respectively. (C) The overexpression of FOXP2 stimulated PHF2 expression in MDA-MB-468 cells. MDA-MB-468 cells were transfected with pHis-FOXP2 or pCMV-EGFP. The levels of FOXP2 and PHF2 were examined by Western blotting and qPCR. (D) FOXP2 bound to the endogenous promoter of PHF2. Gene sequence analysis was performed to predict positions of putative FOXP2 binding sites in -2 kb human PHF2 promoter and the primers for ChIP assays were designed. The chromatin of MCF-7 cells was cross-linked, sonicated, and immunoprecipitated (IP) with either FOXP2 antibody or rabbit IgG. The amount of promoter DNA associated with the IP chromatin was measured by qPCR with primers specific to PHF2 promoter regions -1,838 bp to -1,693 bp and -1,519 bp to -1,403 bp. (E) FOXP2 bound to PHF2 promoter region -1,765 bp to -1,743 bp. The nuclear extracts were prepared from pHis-FOXP2-transfected cells and used for EMSAs with a FAM-labeled DNA probe synthesized from PHF2 promoter sequence -1,765 bp to -1,743 bp (PHF2 pro). The unlabeled probe (50×) or 1 μg of FOXP2 antibody (α-FOXP2) was added to the reaction to show the specificity of FOXP2/DNA complex formation. EMSAs with a FAM-labeled mutated probe were also performed (Mut pro). (F) The PHF2 promoter was activated by FOXP2 in breast cancer cells. A luciferase reporter plasmid (1.5 μg) containing the fragment of -2,000 bp to +60 bp of PHF2 promoter and loading control pRL-CMV luciferase reporter plasmid (20 ng) was transfected into MDA-MB-468 cells with different amounts of pHis-FOXP2 (0, 0.5, 1.0, and 1.5 μg, balanced with the different amounts of empty expression vector). Protein lysates were prepared at 48 h following transfection and then used to measure dual luciferase enzyme activities. The asterisks indicate statistically significant changes: *P ≤ 0.05, **P ≤ 0.01, and ***P ≤ 0.001.

Figure 4

The activation of the E-cadherin promoter by FOXP2 relied on the participation of FOXA2. (A) The E-cadherin promoter was activated by FOXP2 in breast cancer cells. A luciferase reporter plasmid (1.5 μg) containing the fragment of -733bp to +60 bp of E-cadherin promoter or a control luciferase reporter plasmid (1.5 μg) containing the fragment of -233 bp to +52 bp of E-cadherin promoter and loading control pRL-CMV luciferase reporter plasmid (20 ng) were transfected into MCF-7 cells with pHis-FOXP2 or pCMV-EGFP (1 μg). (B) FOXP2 bound to the endogenous E-cadherin promoter. Gene sequence analysis was performed to predict positions of putative FOXP2 binding sites in -1 kb human E-cadherin promoter and the primers for ChIP assays were designed. The chromatin of MCF-7 cells was cross-linked, sonicated, and immunoprecipitated (IP) with either FOXP2 antibody or rabbit IgG. The amount of promoter DNA associated with the IP chromatin was measured by qPCR with primers specific to E-cadherin promoter regions -760 bp to -610 bp and -414 bp to -280 bp. (C) FOXP2 bound to E-cadherin promoter region -701 bp to -691 bp. Nuclear extracts were prepared from pHis-FOXP2-transfected cells and used for EMSAs with a FAM-labeled DNA probe synthesized from E-cadherin promoter sequence -706 bp to −686 bp (E-cad pro). The unlabeled probe (50×) or 1 μg of FOXP2 antibody (α-FOXP2) was added to the reaction to show the specificity of FOXP2/DNA complex formation. EMSAs with a FAM-labeled mutated probe were also performed (Mut pro). (D) FOXP2 mediated the binding of FOXA2 to the E-cadherin promoter. Nuclear extracts were prepared from pHis-FOXP2-transfected, pCMV-FOXA2-transfected, or both-transfected cells (48 h) and used for EMSAs with a FAM-labeled DNA probe synthesized from E-cadherin promoter sequence -706 bp to -686 (E-cad probe). (E) Knockdown of FOXA2 abolished the activation of the E-cadherin promoter by FOXP2. A luciferase reporter plasmid (1.5 μg) containing the fragment of -733 bp to +60 bp of E-cadherin promoter and loading control pRL-CMV luciferase reporter plasmid (20 ng) were transfected into MCF-7 cells with the pHis-FOXP2 or pCMV-EGFP (1 μg) plus FOXA2 siRNA or Control siRNA (200 nM). Protein lysates were prepared at 36 h following transfection and used to measure dual luciferase enzyme activities. (F, G) The knockdown of FOXP2 and FOXA2 together showed a synergistic repression of the E-cadherin expression in breast cancer cells. The shCon MCF-7 cells and shFOXP2 MCF-7 cells were transfected with Control siRNA or FOXA2 siRNA (200 nM) and examined by qPCR for mRNA levels (F) and by Western blotting for protein levels (G). (H) Overexpression of FOXA2 enhanced the activation of the E-cadherin promoter by FOXP2 in MDA-MB-231 cells. A luciferase reporter plasmid (1.5 μg) containing the fragment of -733 bp to +60 bp of E-cadherin promoter and loading control pRL-CMV luciferase reporter plasmid (20 ng) were transfected into MDA-MB-231 cells with the pCMV-FOXA2 (1 μg), pHis-FOXP2 (1 μg), or both (balanced with the pCMV-EGFP vector). Protein lysates were prepared at 36 h following transfection and used to measure dual luciferase enzyme activities. (I, J) The overexpression of FOXP2 and FOXA2 together showed a synergistic activation of the E-cadherin expression in MDA-MB-231 cells. The MDA-MB-231 cells were transfected with FOXA2 expression vector (pCMV-FOXA2), FOXP2 expression vector (pHis-FOXP2), or both (balanced with the pCMV-EGFP vector) and examined by Western blotting for protein levels (I) and by qPCR for mRNA levels (J). The asterisks indicate statistically significant changes: **P ≤ 0.01, ***P ≤0.001.

Figure 5

Stable knockdown of FOXP2 stimulated the metastasis of breast cancer cells in vivo. The metastasis models of human breast cancer cells in nude mice were created via tail vein injection of control lentivirus-infected MCF-7 (shCon) cells or FOXP2-knockdown MCF-7 cells (shFOXP2) (2×10⁵ cells/mouse). The mice (n=12 for each group) were sacrificed at day 1, day 35, and day 60 post-injection. (A) The stable knockdown of FOXP2 in cells increased the number of circulating tumor cells in the blood of model mice. Total RNAs of blood of each mouse were isolated at day 1 and day 35 post-injection and the relative concentration of human tumor cells in blood was determined by qPCR for the mRNA levels of human specific CYCLOPHILIN over the mRNA levels of mouse specific Cyclophilin. (B, C) Increased tumor formation was found in the lungs of mice injected with shFOXP2 cells. Representative photographs of the lung of the two groups of mice day 60 post-injection were shown, and the statistical data of the tumor foci number are presented on the right (B). Total RNAs of the lung of each mouse were isolated at day 60 post-injection and the relative concentration of human tumor cells in lung was determined by qPCR for the mRNA levels of human specific CYCLOPHILIN over the mRNA levels of mouse specific Cyclophilin (C). (D) H&E staining histological images of the lung of the two groups of mice. (E) The expression of exogenous human FOXP2, E-cadherin, and PHF2 in lungs of nude mouse metastasis models of human breast cancer cells. The total RNAs of lungs of different groups were isolated and the ratio of mRNA levels of human FOXP2, E-cadherin and PHF2 over human CYCLOPHILIN were determined by qPCR. The asterisks indicate statistically significant changes: *P ≤0.05, **P ≤ 0.01.

Figure 6

The roles of FOXP2 in EMT of breast cancer cells. (A) The Oncomine boxplots of FOXP2 levels were analyzed between invasive breast tissues and normal breast tissues from the oncomine.org website. The p-value of the TCGA RNA-seq data of normal breast samples (n=61) versus invasive breast carcinoma samples (n=76) was 7.76E^-22. (B) The FOXP2-improved survival in breast cancer patients relied on the participation of FOXA2. The FOXA2^High subgroup (n=131) or the FOXA2^Low subgroup (n=131) contained either the top 1/3 or the bottom 1/3 of the collected BRCA data set (n=394), respectively. FOXP2-related survival of patients in the FOXA2^High and FOXA2^Low subgroups was fitted by the “survfit” function, and Kaplan–Meier curves were drawn by the “ggsurv” function in the R package “survival”. (C) The roles of FOXP2 in epithelial-mesenchymal transition (EMT) of breast cancer cells. FOXP2 prevented EMT of breast cancer cells by regulating the transcription of multiple EMT-related genes: FOXP2 could bind to certain promoters and stimulate the transcription of genes such as PHF2 and E-cadherin. The transcriptional activation by FOXP2 could be mediated by FOXA2. FOXP2 could also repress the transcription of certain genes such as SRPX2 and uPAR, through recruiting co-repressors such as CtBP-1.