MAP3K4/CBP-regulated H2B acetylation controls epithelial-mesenchymal transition in trophoblast stem cells

Abstract

Epithelial stem cells self-renew while maintaining multipotency, but the dependence of stem cell properties on maintenance of the epithelial phenotype is unclear. We previously showed that trophoblast stem (TS) cells lacking the protein kinase MAP3K4 maintain properties of both stemness and epithelial-mesenchymal transition (EMT). Here, we show that MAP3K4 controls the activity of the histone acetyltransferase CBP, and that acetylation of histones H2A and H2B by CBP is required to maintain the epithelial phenotype. Combined loss of MAP3K4/CBP activity represses expression of epithelial genes and causes TS cells to undergo EMT while maintaining their self-renewal and multipotency properties. The expression profile of MAP3K4-deficient TS cells defines an H2B acetylation-regulated gene signature that closely overlaps with that of human breast cancer cells. Taken together, our data define an epigenetic switch that maintains the epithelial phenotype in TS cells and reveals previously unrecognized genes potentially contributing to breast cancer.

Figure 1. TS^KI4 Cells Deficient in MAP3K4 Activity Maintain Self-renewal, Multipotency, and Developmental Potency While Exhibiting Properties of EMT

(A–D) Expression of TS cell stemness genes in TS^KI4 cells measured by qRT-PCR. (E–G) Maintenance of multipotency in TS^KI4 cells demonstrated by induction of trophoblast differentiation markers measured by qRT-PCR. Differentiation was induced by FGF4 withdrawal for the indicated number of days. (H–N) Developmental potency of TS^WT and TS^KI4 cells demonstrated by injection of GFP positive TS^WT and TS^KI4 cells into wild-type blastocysts. (H) Data shows the number of GFP positive chimeras per cell type injected expressed as a percent of total embryos examined. A and B are two independent TS^KI4 cell clones. (I) Sites of incorporation of GFP positive TS^WT and TS^KI4 cells in E6.5 conceptuses. (J, K, and L) Three representative whole embryo E6.5 TS^KI4 chimeras and (M) 18 µm section from E6.5 TS^KI4 chimera are shown. Black bar represents 100 µm. EPC, ectoplacental cone; EXE, extraembryonic ectoderm; GC, giant cells. (N) Representative E9.5 TS^WT cell chimera. Red asterisks indicate location of embryos. (O–V) Mesenchymal phenotype of TS^KI4 cells compared to epithelial morphology of TS^WT cells. (O) Phase microscopy of TS^WT cells. (P) Immunostaining with anti-E-cadherin antibody (red) shows strong staining around the cellular periphery in TS^WT cells with nuclear DAPI stain (blue). (Q) Cortical actin staining (green) and nuclear DAPI stain (blue) in TS^WT cells. (R) Enlarged inset of region indicated by arrow in (Q) showing peripheral cortical actin in TS^WT cells. (S) Phase microscopy of TS^KI4 cells. Black bar represents 100 µm. (T) Relocalization and loss of E-cadherin (red) from the periphery of TS^KI4 cells with nuclear DAPI stain (blue). (U) Filamentous actin staining (green) and nuclear DAPI stain (blue) in TS^KI4 cells. White bar represents 50 µm. (V) Enlarged inset of region indicated by arrow in (U) showing filamentous actin in TS^KI4 cells. White bar represents 10 µm. White arrow indicates site for insets in (R) and (V). (W) Reduced protein expression of E-cadherin and vinculin in TS^KI4 cells shown by Western blotting. (X) Invasion through Matrigel by undifferentiated TS^WT cells (0), TS^WT cells differentiated for four days by FGF4 removal (4), or two independent TS^KI4 cell clones (A and B) cultured in the presence of FGF4. Fold change is relative to TS^WT cells. (Y) Representative E9.5 decidua from blastocyst injected with TS^WT cells. (Z-BB) Three representative E9.5 deciduas from blastocysts injected with TS^KI4 cells. Arrowheads indicate sites of defective decidualization. (A–G) and (X) represent the mean ± range of two independent experiments.

Figure 2. Differentiation of TS^WT Cells Induces EMT

(A) Differentiation of TS cells results in increased invasiveness. Invasion through Matrigel by undifferentiated TS cells (0) or TS cells differentiated by FGF4 withdrawal for the indicated number of days (2,3,4,5) is shown. Data are the mean ± range of two independent experiments performed in duplicate. Undifferentiated cells (B,C), or invasive cells isolated from the bottom of Matrigel-coated transwell chambers (D,E) plated on glass coverslips. (B,D) Cells were stained for actin (green) and nuclei (blue). (C,E) Actin staining alone shows peripheral cortical actin in undifferentiated cells (C) and filamentous actin in invasive trophoblasts (E). White bar represents 50 µm. (F) Reduced Cdh1 message (E-cadherin) in invasive trophoblasts (T^INV) compared to undifferentiated TS cells (TS^WT) and TS cells differentiated for four days (T^DIFF) measured by qRT-PCR. Data normalized to TS^WT samples represent the mean ± SEM of three independent experiments. (G) Reduced E-cadherin protein in T^INV cells shown by Western blotting. (H) Phase microscopy shows mesenchymal morphology of TS^Snail cells. Black bar represents 100 µm. (I) Filamentous actin staining (green) and nuclear DAPI stain (blue) in TS^Snail cells. (J) Filamentous actin staining alone in TS^Snail cells is shown. (K) Increased invasiveness of TS^KI4 and TS^Snail cells relative to TS^WT cells is shown. (L) Reduced Cdh1 message in TS^Snail cells compared to TS^WT cells is measured by qRT-PCR. (M–Q) Expression of EMT-inducing genes in TS^WT, TS^KI4, and TS^Snail cells measured by qRT-PCR is shown. Data show the mean ± range of two independent experiments.

Figure 3. Gene Expression Profiling of EMT in TS^KI4 Compared to T^INV Cells

(A) Loss of MAP3K4 activity in TS ^KI4 cells results in significant changes in gene expression. Heat map compares gene expression changes in TS^KI4 cells (1785 genes), T^DIFF cells (5706 genes), and T^INV cells (6641 genes) relative to TS^WT cells. Significantly changed genes have a Benjamini-Hochberg p-value < 0.05 and log₂ ratio ≥ abs(1). Arrowheads indicate overlapping genes. (B,C) Venn diagrams show the overlap of upregulated (B) and downregulated (C) genes in TS^KI4 cells (blue), T^DIFF cells (yellow), and T^INV cells (red). (D) 25% of the overlapping genes between TS^KI4 and T^INV EMT models control cell adhesion and motility. Diagram depicts T^INV cell log₂ ratios from (A) for genes that regulate adhesion and motility having shared changes between TS^KI4 and T^INV cells. Circle diameter depicts changes in TS^KI4 cell log₂ ratios.

Figure 4. Selective Loss of H2A and H2B Acetylation in Undifferentiated TS^KI4 Cells

(A) Acetylation of histones H2A, H2B, H3 and H4 is decreased upon induction of TS cell differentiation via withdrawal of FGF4. Western blotting of lysates from TS^WT or TS^KI4 cells differentiated for the indicated days is shown. (B,C) TS^KI4 cells exhibit selective loss of acetylation on histones H2A and H2B and no change H3 methylation. Western blots were performed as described for (A). (D) TS^Snail cells exhibit selective loss of acetylation on histones H2A and H2B. (A–D) Results are representative of two to three independent experiments.

Figure 5. Reduction of H2BK5Ac on Select Promoters in TS^KI4 Cells Contributes to Repression of the Epithelial Phenotype

(A) Peak H2BK5Ac density occurs within 1kb of the TSS. Plot depicts the distribution of read tags around the TSS of mouse RefSeq genes. Solid red line represents the read tag distribution density of all genes in TS^WT cells, and dotted blue line represents TS^KI4 cells. Data was pooled from two independent experiments sequenced in duplicate. (B) Loss of H2BK5Ac occurs near the TSS of 3515 genes in TS^KI4 cells. Scatter plot of RefSeq genes with read tag counts above 20 reads per 2kb for genes identified from H2BK5Ac ChIP-seq of TS^WT (x-axis) and TS^KI4 (y-axis) cells. Solid line represents the data normalization reference line for delineating genes with significant read count differences between TS^WT and TS^KI4 cells. Significance is based upon Benjamini-Hochberg FDR adjusted p-value ≤ 0.05. Green and red dots represent 3515 genes with a significant decrease and 648 genes with a significant increase in read tag density between TS^WT and TS^KI4 cells, respectively. Gray dots represent all non-significantly changed genes. Dotted line is a 45° reference line. (C) Bar graph summarizes all genes in (B) with significantly altered H2BK5Ac density in TS^KI4 cells. Data for (B and C) were pooled as described in (A). (D) Reduced H2BK5Ac near the TSS of specific genes in TS^KI4 cells. Plots of H2BK5Ac density compare the normalized coverage depth (NCD) of read tags around the TSS for indicated genes with loss of H2BK5Ac (p-values ≤ 1e⁻¹⁸) between TS^WT and TS^KI4 cells. The x-axis for each subplot is the distance to TSS, and the y-axis is the NCD of read tags. (E) Validation of reduced H2BK5Ac on specific genes in both TS^KI4 and T^INV cells shown by ChIP qRT-PCR of TS^WT, TS^KI4 and T^INV cells. Data show the mean ± range of two independent experiments performed in duplicate. (F) Genes with increased H2BK5Ac near the TSS in TS^KI4 cells plotted as described in (D). (G) Plots of H2BK5Ac density comparing the NCD for stem cell markers between TS^WT and TS^KI4 cells. (H) Changes in H2BK5Ac correlates with altered gene expression in TS^KI4 cells. Scatter plot of 6722 RefSeq genes with read tag counts above 50 reads per 1kb for genes identified by H2BK5Ac ChIP-seq (x-axis) and mRNA-seq (y-axis) in TS^KI4 cells. Read tag counts were converted to log₂ ratio values for comparison. Solid red line represents the linear regression line of the data with a Pearson correlation coefficient of 0.62 (p-value < 10⁻¹⁶). Labeled dots are genes with a four-fold change in both mRNA abundance and H2BK5Ac enrichment.

Figure 6. TS^KI4 Cells and Claudin-low Breast Cancer With Properties of EMT and Stemness Show Loss of H2BK5Ac on Shared Genes

(A) TS^KI4 cells cluster with the CL subtype of breast cancer cells. Heat map compares expression of 22 breast cancer EMT genes in TS^KI4 cells and 52 breast cancer cell lines. TS^KI4 cell and CL breast cancer cluster is outlined in red. Red and green arrows indicate shared upregulated and downregulated genes respectively. (B) Venn diagram depicts the intersection between genes elevated and downregulated in CL cell lines compared to TS^KI4 cells. (C) Plot shows log₂ ratios (y-axis) of TS^KI4 cells for intersecting genes depicted in (B). (D) Reduced H2BK5Ac on specific genes in both TS^KI4 and TS^Snail cells from the intersecting downregulated genes in (C) measured by qRT-PCR of ChIP samples is shown. (E) Reduced H2BK5Ac on specific genes in CL breast cancer lines compared to normal HMEC breast cells as measured by ChIP-qRT-PCR. (D,E) Data are the mean ± range of two independent experiments performed in duplicate. (F) CL human tumors show the highest expression of TS^KI4 genes among breast cancer subtypes. Mean expression of TS^KI4 cell upregulated genes across the subtypes of breast cancer in the UNC337 data set. p-value was calculated by comparing gene expression means across all subtypes using an ANOVA test. Each + represents a distinct tumor sample within the data set. (G) Venn diagram of intersecting genes elevated and downregulated in CL human tumors and TS^KI4 cells.

Figure 7. CBP Expression is Required for MAP3K4-dependent Regulation of the Epithelial Phenotype in TS Cells

(A) HAT activity is significantly decreased in TS^KI4 cells. Activity of one or three µg of nuclear lysate isolated from TS^WT or TS^KI4 cells is measured. Significance of change in HAT activity was evaluated by an unpaired Student’s T-test, *p-value < 0.05. (B) MAP3K4/JNK increases HAT activity of CBP measured as incorporation of [^3H]-acetyl-CoA in counts per minute (CPM) in 293 cells co-expressing CBP with the constitutively active kinase domain of MAP3K4 or kinase-inactive (KI). Significance of change in HAT activity was evaluated by an unpaired Student’s T-test. ***p-value < 0.001; *p-value < 0.05. Data represents the mean ± SEM of three independent experiments performed in triplicate. (C) MAP3K4 and JNK promote phosphorylation of CBP as measured by kinase assay. γ³²P-ATP labeled proteins were visualized by autoradiography and quantified by phosphorimaging. Data are representative of two independent experiments. (D,E) qRT-PCR shows reduced expression of CBP and p300 in TS^WT cells infected with unique shRNAs for CBP (D) or p300 (E) compared to control virus-infected cells (Vec). Data represents the mean ± range of two independent experiments performed in triplicate. (F, G, and H) shRNA knockdown of CBP results in loss of the epithelial phenotype. Phase microscopy shows morphology of control virus-infected cells (F) compared to mesenchymal morphology of cells expressing CBP shRNAs (G,H). Black bar represents 100 µm. (I) Increased invasiveness of shCBP expressing cells compared to control virus-infected cells. Data shown represents the mean ± range of two independent experiments performed in triplicate. (J,K) Selective loss of H2A/H2BAc in CBP knockdown cells. Western blotting of lysates from TS^WT cells infected with control virus, four independent CBP shRNAs (J) or two independent p300 shRNAs is shown. Results are representative of two independent experiments.