FOXA2 suppresses endometrial carcinogenesis and epithelial-mesenchymal transition by regulating enhancer activity

Abstract

FOXA2 encodes a transcription factor mutated in 10% of endometrial cancers (ECs), with a higher mutation rate in aggressive variants. FOXA2 has essential roles in embryonic and uterine development. However, FOXA2's role in EC is incompletely understood. Functional investigations using human and mouse EC cell lines revealed that FOXA2 controls endometrial epithelial gene expression programs regulating cell proliferation, adhesion, and endometrial-epithelial transition. In live animals, conditional inactivation of Foxa2 or Pten alone in endometrial epithelium did not result in ECs, but simultaneous inactivation of both genes resulted in lethal ECs with complete penetrance, establishing potent synergism between Foxa2 and PI3K signaling. Studies in tumor-derived cell lines and organoids highlighted additional invasion and cell growth phenotypes associated with malignant transformation and identified key mediators, including Myc and Cdh1. Transcriptome and cistrome analyses revealed that FOXA2 broadly controls gene expression programs through modification of enhancer activity in addition to regulating specific target genes, rationalizing its tumor suppressor functions. By integrating results from our cell lines, organoids, animal models, and patient data, our findings demonstrated that FOXA2 is an endometrial tumor suppressor associated with aggressive disease and with shared commonalities among its roles in endometrial function and carcinogenesis.

Keywords: Cancer; Molecular pathology; Oncology; Tumor suppressors.

Figure 1. FOXA2 is the endometrium-specific FOXA and is downregulated in ECs.

(A) Tissue sections of normal human endometrium (secretory/HSecE, proliferative/HProE) and prostate (HProstate) immunostained for FOXA1 or FOXA2. FOXA1 is expressed in the prostate (luminal/secretory and basal layers) but not endometrium, whereas FOXA2 is expressed in endometrial epithelial cells and only weakly and heterogeneously in the basal layer of prostate glands. The 6 immunostains were performed under identical conditions, antibody concentrations, and chromogen incubation times and are thus directly comparable to one another. Insets for HProstate are higher magnifications of smaller boxed areas. Scale bars: 100 μm. (B) Western blot of immortalized endometrium (HEuEC) and prostate cancer (LNCaP) cell lines confirmed mutually exclusive patterns of FOXA1 and FOXA2 expression. (C) mRNA expression levels by qRT-PCR (n = 3) and correlation to protein expression by Western blot in 1 immortalized normal endometrial cell line (HEuEC) and 13 EC cell lines. FOXA2 was undetectable in 7/13 and significantly downregulated in 4/13 of the EC cell lines. Data shown as mean ± SEM. (D) Western blot analysis of estrogen receptor α (ERα) and progesterone receptor-A/B (PR-A/B) in EC cell lines. (E) Expression of FOXA2, ERα, and PR in normal human endometrium and grade 1–3 human endometrial carcinomas; representative images. Scale bars: 100 μm. (F) FOXA2 expression levels per H-scores in normal endometrium (n = 5) and ECs (grade 1, n = 16; grade 2, n = 23; grade 3, n = 21). Box-and-whisker plot represents medians with minimum and maximum values. P value was determined by 2-tailed Mann-Whitney U tests compared with normal cases. Data are shown as mean ± SEM; **P < 0.01, ***P < 0.001, ****P < 0.0001. (G) Scatter plot shows correlation analysis of FOXA2 with ERα and PR in normal endometrium and ECs. FOXA2, ERα, and PR expression levels were determined by H-score. Pearson’s r determined correlation between data groups.

Figure 2. FOXA2 suppresses EC cell proliferation and enhances cell adhesion.

(A) FOXA2 reexpression after lentiviral transduction in Ishikawa (ISK) cells (ISK-FOXA2). (B) Immunofluorescence shows expected FOXA2 nuclear localization, DAPI counterstain. Scale bars: 50 μm. (C) Cell growth analysis showing FOXA2-mediated growth suppression; confluency normalized to t₀ (n = 3). Data shown as mean ± SEM; 2-tailed t test. (D) Wound-healing assay at t₀ and 60 hours. Scale bars: 250 μm. (E) Wound closure per gap distance (n = 3). Data shown as mean ± SEM; 2-tailed t test. (F) Cell cycle analysis (n = 3). Peaks in blue, yellow, and green show percentage cells in G0/G1, S, and G2/M phase. (G) Xenografts after s.c. injection of 1 million cells in left/right flanks of NOD scid gamma females (n = 4). Tumors harvested 35 days after injection. (H) Growth curves of ISK-EV and ISK-FOXA2 xenografts per caliper measurements (n = 4). Data shown as mean ± SEM; 2-tailed t test. (I) Endpoint xenograft weights at day 35 (n = 4, same tumors shown in G). Data shown as mean ± SEM; 2-tailed t test. (J) Gene Ontology (GO) enrichment analysis of 89 significantly upregulated genes by RNA-Seq (≥4-fold and P < 0.0001) in ISK-FOXA2 cells. GO pathways plotted by gene ratio. Dots sized in proportion to gene numbers in the GO term colored by FDR value per inset. (K) Relative change in mRNA expression of AGR2, CNTN1, and PCDHA1 genes (n = 3). Data shown as mean ± SEM; multiple 2-tailed t tests. (L) Cell adhesion assays, representative phase contrast images of ISK-EV and ISK-FOXA2 cells adhering to extracellular matrix protein panel after 16 hours. Scale bars: 50 μm. (M) Quantitative analysis, cell adhesion assays (n = 9). Data shown as mean ± SEM; multiple 2-tailed t tests. For all panels, *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Figure 3. FOXA2 knockdown promotes EC cell migration and invasion.

(A) Western blot shows effective lentiviral shRNA-mediated knockdown of FOXA2 in HEC-1-B cells (one of the EC lines with high endogenous FOXA2 expression, Figure 1C). (B) Immunostaining of FOXA2 (red) with actin control (green) showing expected reduction of nuclear FOXA2 signal in FOXA2^KD HEC-1-B cells. Nuclei were stained with DAPI (blue). Scale bars: 50 μm. (C) RNA expression analysis, volcano plot showing differentially expressed significant genes (P <0.05) in FOXA2^KD HEC-1-B cells compared with empty vector control (1713 upregulated genes [red, fold change ≥ 2] and 2439 downregulated genes [green, fold change ≤ –2]). Dotted vertical lines represent log₂ (fold change) threshold of ±1 and dotted horizontal line represents P value threshold of 0.05. Selected genes are shown. (D) Protein expression by Western blot in control and FOXA2^KD HEC-1-B cells. (E) Indirect immunofluorescence in control and FOXA2^KD HEC-1-B cells; representative images. Nuclei were stained with DAPI (blue). Scale bars: 50 μm. (F) Cell migration and invasion assays for control and FOXA2^KD HEC-1-B cells. Cells that migrated/invaded from the upper chamber of a Transwell to its lower chamber without (migration) or with a growth factor gradient (invasion) are visualized by crystal violet staining; representative images. Scale bars: 200 μm. (G) Quantitative analysis of migrated or invaded control and FOXA2^KD HEC-1-B cells (n = 3). Data shown as mean ± SEM; *P < 0.05; ***P < 0.001, 2-tailed t test. (H) Western blot documenting lentivirus-mediated enforced CDH1 (E-cadherin) expression in FOXA2^KD HEC-1-B cells. (I) E-cadherin expression (red) by immunofluorescence in FOXA2^KD HEC-1-B cells with CDH1 enforced expression. Nuclei were stained with DAPI (blue). Scale bars: 50 μm. (J) Cell invasion and migration assays of FOXA2^KD HEC-1-B cells with or without CDH1 reconstitution. Scale bars: 200 μm. (K) Quantitative analysis of migrated and invaded FOXA2^KD HEC-1-B cells with or without CDH1 reconstitution (n = 3). Data shown as mean ± SEM; ***P < 0.001, 2-tailed t test.

Figure 4. Foxa2 functions as a tumor suppressor that cooperates with Pten in an in vivo conditional KO mouse model.

(A) Gross images of uteri from control, BAC-Sprr2f-Cre Foxa2^fl/fl (Foxa2), Pten^fl/fl (Pten), or Foxa2^fl/flPten^fl/fl (Foxa2/Pten) mice at 3 to 12 months of age. Scale bar: 1 mm. (B) Survival analysis of Foxa2/Pten (n = 45), Pten (n = 20), and Foxa2 (n = 30), and littermate control (n = 25) mice. ****P < 0.0001, log-rank test. (C) Uterine weights of Foxa2/Pten (n = 23), Foxa2 (n = 19), Pten (n = 13), and littermate control (n = 18) mice at necropsy. **P < 0.01; ****P < 0.0001, 1-way ANOVA, Tukey’s multiple-comparison test. (D) Uterine horn lengths of Foxa2/Pten (n = 28), Foxa2 (n = 20), Pten (n = 22), and littermate control (n = 20) mice. ****P <0.0001, 1-way ANOVA, Tukey’s multiple-comparison test. (E) Control, Foxa2, Pten, and Foxa2/Pten uteri (transverse sections) at 9 months of age immunostained with pan-cytokeratin (green) to label epithelial cells and αSMA (red) to label myometrium. Representative confocal tile-scan images are shown. Scale bars: 500 μm. (F–M) H&E-stained sections of Foxa2/Pten mouse cancers. (F) Well-differentiated adenocarcinoma. (G) Well-differentiated adenocarcinoma with papillary architecture. (H) Well-differentiated adenocarcinoma with squamous differentiation; inset shows p63 immunostain. sq, squamous differentiation. (I) Full-thickness myometrial invasion with higher magnifications of insets. (J) Infiltration into adjacent adipose tissue. (K–M) Abdominal and distant metastases marked by dashed black lines. Scale bars: 100 μm.

Figure 5. Synergism between Foxa2 and Pten in 3D organoid growth.

(A) Representative phase contrast images of control, Foxa2, Pten, and Foxa2/Pten mouse endometrial epithelial organoids at day 2 to day 8. Right panel shows E-cadherin, cytokeratin (CK), and GM130 expression on day 8 organoids by indirect immunofluorescence. Nuclei were stained with DAPI (blue). Scale bars: 50 μm. (B and C) Comparison of diameter and proliferation among organoids of differing genotypes (n = 3). Data shown as mean ± SEM; *P < 0.05; **P < 0.01; ****P < 0.0001, 2-tailed t test.

Figure 6. Foxa2 reconstitution in FP cells suppresses cell growth and tumor phenotypes.

(A) Brightfield and indirect immunofluorescence images of FP cell line derived from Foxa2/Pten mouse uterine tumor. Scale bars: 50 μm. (B) Western blot shows enforced Foxa2 expression after lentiviral transduction in FP cells (FP-Foxa2). For controls, transduction was performed with empty vector (FP-EV). (C) Immunofluorescence of FOXA2 (red) validates nuclear localization in the FP-Foxa2 cells. Scale bar: 50 μm. (D) Growth comparison of FP-EV and FP-Foxa2 cells by real-time live cell imaging (n = 3). Data shown as mean ± SEM; ****P < 0.0001, 2-tailed t test. (E) Comparison of cell cycle analysis in FP-EV and FP-Foxa2 cells by flow cytometry (n = 3). The peaks in blue and yellow show percentage of cells in G0/G1 and S phase. (F) Gross images of xenograft tumors after s.c. injection of 1 million FP-EV and FP-Foxa2 cells in the left and right flanks of NOD scid gamma female mice (n = 4). Tumors were harvested 35 days after cell injection. (G) Growth curves of FP-EV and FP-Foxa2 xenografts per caliper measurements (n = 4). Data represent mean ± SEM; *P < 0.05, 2-tailed t test. (H) Endpoint xenograft tumor weights at day 35 (n = 4, same tumors shown in F). Data represent mean ± SEM; *P < 0.05, 2-tailed t test. (I) Immunostaining of actin filaments (red) in FP-EV and FP-Foxa2 organoids counterstained with DAPI (blue). Representative midsections of Z-stack images are shown. Scale bars: 50 μm. (J) Quantitative analysis of invasive or round-shaped FP-EV and FP-Foxa2 organoids (n = 5). Data shown as mean ± SEM; ****P < 0.0001, 2-tailed t test. (K) Comparison of FP-EV and FP-Foxa2 organoid proliferation (n = 4). Data shown as mean ± SEM; *P < 0.05, 2-tailed t test.

Figure 7. Transcriptional reprogramming by FOXA2.

(A) Venn diagram representing H3K27ac peaks. (B) Average coverage plot and heatmap representation of ChIP-Seq signals ±4 kb around acetyl histone (H3K27ac) peaks. (C) Average coverage plot and heatmap representation of ChIP-Seq signals ±4 kb around transcriptional start sites. (D) Genome browser representation of H3K27ac peaks in the MYC gene. (E) Western blot analysis of H3 and H3K27ac protein expression in ISK-EV, ISK-FOXA2, FP-EV, and FP-Foxa2 cells. (F and G) Gene set enrichment analysis (GSEA). Genes shaded with red in the heatmap are upregulated in parental ISK cells (in comparison to ISK-FOXA2). Genes shaded with blue in the heatmap are upregulated in ISK-FOXA2 (in comparison with parental ISK cells).