Krüppel-like factor 9 upregulates E-cadherin transcription and represses breast cancer invasion and metastasis

Abstract

Aberrant expression of Krüppel-like factor 9 (KLF9) is frequently found in some types of cancer and is implicated in cancer initiation and progression. However, the effects of KLF9 on cancer metastases and the underlying mechanisms still need to be understood. Here, we found that KLF9 evidently inhibited the capabilities of migration and invasion of breast cancer cells. The expression of KLF9 was markedly decreased in breast cancer patients compared with benign tumors, and was positively correlated with the expression of E-cadherin in the tissues of breast cancer patients. Mechanistically, chromatin immunoprecipitation combined with site-directed mutagenesis-luciferase assay revealed that KLF9 activated the E-cadherin promoter by binding to GT-box elements located +84 bp and -143 bp from the TSS in the E-cadherin promoter, leading to elevated expression of E-cadherin mRNA and protein. In vivo experiments confirmed that KLF9 strongly inhibited the lung metastasis of breast cancer and increased mouse E-cadherin expression in 4T1 mouse breast cancer cells. Taken together, our findings demonstrated that KLF9 could suppress breast cancer invasion and metastasis by upregulating E-cadherin, which provided new insight into aggressive treatment of breast cancer by targeting the KLF9/E-cadherin axis.

Keywords: E-cadherin; Krüppel-like factor 9; breast cancer; metastasis.

Figure 1

KLF9 suppresses migration and invasion of breast cancer cells. (A) Scratch wound healing assays of MCF-7 cells transfected with control vector or Flag-KLF9 (left); western blotting of Flag-KLF9 in MCF-7 cells (right). Scale bar: 100 μm. The western blotting images were cropped, and the original images were supplied in supplementary files named as “(A) (right)-01” and “(A) (right)-02”. (B) Transwell invasion assays of MCF-7 cells transfected with control vector of Flag-KLF9 (left). Scale bar: 50 μm. Data quantification of the invasive cells (right). Each group repeated three times. (C) Scratch wound healing assays of T47D cells transfected with control vector or shKLF9s (sh-KLF9#1 and sh-KLF9#2) (left); western blotting of Flag-KLF9 in T47D cells (right). Scale bar: 100 μm. The western blotting images were cropped, and the original images were supplied in supplementary files named as “(C) (right)”. (D) Transwell invasion assays of T47D cells transfected with control vector of shKLF9s (sh-KLF9#1 and sh-KLF9#2) (left). Scale bar: 50 μm. Data quantification of the invasive cells (right). (E) Real-time PCR of some invasion-related genes (epithelial and mesenchymal markers or MMPs family members) in KLF9-transfected MCF-7 cells. *, P < 0.05; **, P < 0.01; ***, P < 0.001. NS, not significant. Each group repeated three times. Data are shown as the means ± SDs (P < 0.05, significant).

Figure 2

KLF9 upregulates E-cadherin expression and promoter activity. (A) RT-PCR of E-cadherin and KLF9 mRNA in MCF-7 cells transfected with gradient doses of KLF9 (left). The DNA gels were cropped, and the original images were supplied in supplementary files named as “(A) (01)”, “(A) (02)” and “(A) (03)”. Quantification of the relative mRNA levels of E-cadherin normalized against GAPDH (right). (B) Western blotting of E-cadherin and KLF9 in MCF-7 cells transfected with gradient doses of KLF9. The western blotting images were cropped, and the original images were supplied in supplementary files named as “(B) (01)” and “(B) (02)”. (C-E) Luciferase reporter gene assay of E-cadherin promoter activities in HEK 293T cells overexpressing Flag-KLF9 (C) or in MCF-7 cells transfected with gradiently increased doses of Flag-KLF9 (D) or in T47D cells transfected with shKLF9 (E). pE-cadherin-luciferase plasmid was cotransfected into cells, and empty vectors were used for control. The western blotting images included in the columns were cropped, and the original images were supplied in supplementary files named as “(C)” and “(D & E)”. Data are shown as the means ± SDs (P < 0.05, significant).

Figure 3

KLF9-binding elements are identified in the E-cadherin promoter. (A) KLF9 binding motif identified using the JASPAR database. (B) E-cadherin promoter binding sites of KLF9 predicted by the JASPAR database. (C) Top, schematic illustration of the E-cadherin promoter; bottom, ChIP PCR reactions of E-cadherin promoter regions in HEK 293T cells transfected with Flag-KLF9. Antibody IgG was used for negative control. The DNA gels were cropped, and the original images were supplied in supplementary files named as “(C) (bottom)”. (D) Site-specific mutants for GT or GC boxes in the E-cadherin promoter reporter construct combined with luciferase reporter assays determining promoter activities in MCF-7 cells treated with Flag-KLF9 or control vector. The western blotting images were cropped, and the original images were supplied in supplementary files named as “(D)”. Data are shown as the means ± SDs (P < 0.05, significant; NS, not significant).

Figure 4

E-cadherin expression is positively correlated with KLF9 in human breast cancer tissues. (A) Representative results of IHC staining showing the protein expressions of KLF9 and E-cadherin in breast tumor tissues. Antibodies against KLF9 or E-cadherin were used to detect protein levels. Scale bar: 100 μm. (B & C) KLF9 (B) and E-cadherin (C) protein levels in tissues of breast fibroadenoma (n = 5), invasive ductal carcinoma (n = 5), and triple-negative breast cancer (n = 5). (D) Linear regression analyses of association between KLF9 and E-cadherin expressions with r-square values evaluating correlation degree. (E) Chi-squared test detecting the correlation of KLF9 with E-cadherin expression. Data are shown as the means ± SDs (P < 0.05, significant).

Figure 5

KLF9 and CDH1 mRNA expression were downregulated in breast cancer tissues than in normal from two public datasets (“TCGA breast” and “Curtis breast”). A. KLF9 mRNA levels in normal and invasive lobular breast cancer tissues in the TCGA breast (normal = 61, tumor = 36) (left) and Curtis breast (normal = 144, tumor = 148) (right) datasets from Oncomine database (https://www.oncomine.org). B. CDH1 mRNA levels in normal and invasive lobular breast cancer tissues in the TCGA breast (normal = 61, tumor = 36) (top) and Curtis breast (normal = 144, tumor = 148) (bottom) datasets from Oncomine database. C. Pearson test showing the correlation of KLF9 and CDH1 mRNA levels in TCGA breast (left) and Curtis breast (right). The mRNA expression data were retrieved from Oncomine website. P < 0.05, significant.

Figure 6

KLF9 inhibits the metastasis of 4T1 cells and upregulates E-cadherin expression in BALB/c mice. (A) Conservation analysis of KLF9 binding sites on the E-cadherin promoter in human and murine species. Nucleotides marked in red represent cis-elements of GT-boxes predicted to be the potential binding sites of KLF9. (B) HE staining showing the lung metastases of the 3 groups (NS: negative control; 4T1-Ctrl: positive control; and 4T1-KLF9: case group). IHC staining showing the KLF9 (anti-KLF9 antibody, species: human and mouse) and E-cadherin (anti-E-cadherin antibody, species: mouse) expressions in mouse lungs among the three groups. Scale bar: 100 μm. NS, normal saline. (C) The percent lung metastatic area calculated by ImageJ was shown among the three groups. (D) Western blotting assessing Flag-KLF9 and the corresponding mouse E-cadherin expression in lung carcinoma tissues in control and 4T1-KLF9 groups. The western blotting images were cropped, and the original images were supplied in supplementary files named as “(D) (01)” and “(D) (02)”. Data are shown as the means ± SDs (n = 5). P < 0.05, significant.