EZH2 deletion does not affect acinar regeneration but restricts progression to pancreatic cancer in mice

Abstract

Enhancer of zeste homologue 2 (EZH2) is part of the Polycomb Repressor Complex 2, which promotes trimethylation of lysine 27 on histone 3 (H3K27me3) and gene repression. EZH2 is overexpressed in many cancers, and studies in mice attributed both prooncogenic and tumor suppressive functions to EZH2 in pancreatic ductal adenocarcinoma (PDAC). EZH2 deletion enhances de novo KRAS-driven neoplasia following pancreatic injury, while increased EZH2 expression in patients with PDAC is correlated to poor prognosis, suggesting a context-dependant effect for EZH2 in PDAC progression. In this study, we examined EZH2 in pre- and early neoplastic stages of PDAC. Using an inducible model to delete the SET domain of EZH2 in adult acinar cells (EZH2ΔSET), we showed that loss of EZH2 activity did not prevent acinar cell regeneration in the absence of oncogenic KRAS (KRASG12D) nor did it increase PanIN formation following KRASG12D activation in adult mice. Loss of EZH2 did reduce recruitment of inflammatory cells and, when combined with a more aggressive PDAC model, promoted widespread PDAC progression and remodeling of the tumor microenvironment. This study suggests that expression of EZH2 in adult acinar cells restricts PDAC initiation and progression by affecting both the tumor microenvironment and acinar cell differentiation.

Keywords: Cancer; Epigenetics; Mouse models; Oncology.

Figure 1. KRAS^G12D promotes increased K27me3 enrichment in pancreatic acini.

(A) Representative images of H&E-stained pancreatic tissue from control and KRAS^G12D mice 22 days after TX gavage. Scale bar: 50 μm. (B) Heatmaps show K27me3 and K4me3 enrichment from 2 kb before the transcriptional start sites (TSS) to 2 kb after the transcriptional end site (TES) of all genes. Blue and yellow boxes indicate areas showing increased or decreased K27me3 enrichment in KRAS^G12D mice. There is reduced K27me3 at TSSs, which appears restricted in KRAS^G12D mice. (C) Comparison of called K27me3 and K4me3 enrichment at TSSs in control and KRAS^G12D acinar cells. Red dots represent genes with significantly dysregulated enrichment. Green line indicates expectation for equal enrichment between genotypes. (D) Comparison of chromatin states in control, KRAS^G12D, and KRAS^G12DEZH2^ΔSET mice 22 days after KRAS^G12D induction based on K4me3 and K27me3 enrichment. Numbers in first column indicate the percentage of each state across of the genome. Graphs show quantification of these numbers at the different gene regions. (E) Correlation between gene expression and chromatin states in control, KRAS^G12D, and KRAS^G12DEZH2^ΔSET pancreata 22 days after KRAS^G12D induction. Data represent mean ± SEM (n = 3 mice /group). Two-way ANOVA followed by Tukey’s post hoc test was performed. *P < 0.05; **P < 0.01; ***P < 0.001.

Figure 2. Loss of EZH2 methyltransferase activity increases KRAS^G12D-mediated PanIN progression.

Histological and quantitative analysis comparing KRAS^G12D and KRAS^G12DEzh2^ΔSET mice 51 days after initiating KRAS^G12D and 35 days after treatment with saline or cerulein. (A) Representative H&E images of pancreatic tissue. Box plots indicate the amount of lesion area as a percentage of the entire pancreatic tissue. Significance was measured by 1-way ANOVA followed by Tukey’s post hoc tests. (B) Representative IHC for CK19 or amylase followed by counterstaining with hematoxylin in cerulein-treated mice. Box plots compare the ratio of CK19⁺/amylase⁺ tissue. Significance was measured by 2-tailed unpaired Mann-Whitney U test. (C and D) Representative images of alcian blue histology (C) or periodic acid–Schiff (PSA) (D) histology showing advanced lesions (arrows) in saline- or cerulein-treated KRAS^G12D and KRAS^G12DEzh2^ΔSET mice. Box plots compare the stained area as a percentage of ADM/PanIN lesions. Significance was measured by 2-tailed unpaired Mann-Whitney U test. Scale bar: 100 μm. For graphs, individual mice (n = 7 mice per group) are shown and data represent mean ± minimum to maximum. *P ≤ 0.05, **P ≤ 0.01.

Figure 3. Loss of EZH2 methyltransferase activity alters the effects of KRAS^G12 on expression of genes linked to the tissue microenvironment.

(A) Volcano plot of RNA-Seq analysis between control and KRAS^G12D pancreata 22 days after TX gavage. Significantly downregulated genes are shown in blue and significantly upregulated genes in red. Significance was evaluated with FDR ≤ 0.05. (B and C) Similar Volcano plots comparing gene expression between KRAS^G12DEzh2^ΔSET and control (B) or KRAS^G12D (C) pancreatic tissue 22 days after activating KRAS^G12D (n = 3 mice). (D) KEGG pathway analysis performed on genes enriched for K27me3 and K4me3 identifies an increase in the state 4 pathways in KRAS^G12D tissue (number of pathways) including unique enrichment of downstream mediators of KRAS signaling (red arrows). (E) KEGG pathway analysis based on DEGs from RNA-Seq identified enriched pathways between KRAS^G12D (all pathways shown) or KRAS^G12DEZH2^ΔSET (top 20 pathways shown) and control tissue. Bars indicate –log₁₀ (P value), and dots indicate gene counts. Arrows indicate KRAS-related pathways unique (red) or common (black) to each data set. (F) Gene set enrichment analysis comparing enrichment of HALLMARK_KRAS_UP signaling between control, KRAS^G12D, and KRAS^G12DEzh2^ΔSET tissue 22 days following tamoxifen treatment. Normalized enrichment scores (NES) are significantly different between KRAS^G12DEzh2^ΔSET and both control and KRAS^G12D expression (n = 3).

Figure 4. EZH2 deletion alters immune cell infiltration promoted by KRAS^G12D after acute cerulein treatment.

(A) KEGG pathway analysis of DEGs between KRAS^G12D and KRAS^G12DEzh2^ΔSET pancreatic tissue 22 days after tamoxifen treatment. Bars indicate the FDR values, while black dots indicate the number of genes associated with each pathway. (B) K27me3, K4me3, and RNA tracks showing bivalency and differential K27me3 enrichment between KRAS^G12D and control or KRAS^G12DEzh2^ΔSET at Cd1d2. Red asterisks indicate K27me3 enrichment specific to KRAS^G12D mice. Tracks are an overlay of n = 3 mice. (C and D) IHC for CD3, CD8, and F4/80 (C) or IF for CD4⁺ cells (D) in pancreatic tissue from KRAS^G12D and KRAS^G12DEzh2^ΔSET mice 51 days after expressing KRAS^G12D and 35 days following cerulein treatment. Scale bar: 100 μm. Box plots compare the mean number of positive cells, and individual values (n = 5 mice per condition) are included. Data are shown as mean ± minimum to maximum. Significance was measured using a 2-tailed unpaired Mann-Whitney U test. **P ≤ 0.01. (E) Representative images of IHC for vimentin or α-SMA staining on pancreatic tissue. Scale bar: 100 μm.

Figure 5. Combined loss of MIST1 and EZH2^ΔSET promotes rapid loss of acinar tissue in the presence of KRAS^G12D.

(A) RNA-Seq analysis revealed marked increases in Ezh2 in Mist1^creERT/– KRAS^G12D pancreatic tissue 22 days after KRAS^G12D induction relative to all other genotypes and RNA tracks for Ezh2 confirm deletion of exon 16–19 (red box). Data represent mean ± minimum to maximum (n = 3 for control, Ezh2^ΔSET, KRAS^G12D, KRAS^G12DEzh2^ΔSET, and MKE and n = 2 for Mist1^creERT/– KRAS^G12D). Letters indicate statistically similar groups. ^bP ≤ 0.001. (B) Representative Western blots for EZH2, amylase, or total ERK, 60 days after KRAS^G12D induction. (C and D) RNA tracks for Ezh2 (C) and Amy1 (D). Tracks are the overlay of n = 3 mice. (E) Representative H&E-stained pancreatic sections 60 days after KRAS^G12D induction. Genotypes are indicated. Scale bar: 100 μm. (F) Box plot quantifying the percentage of lesional area in all genotypes based on H&E staining. Data are shown as mean ± minimum to maximum (n = 4 for Ezh2^ΔSET and KRAS^G12D, n = 6 for Mist1^creERT/– KRAS^G12D, n = 7 for control, n = 9 for MKE, and n = 14 for KRAS^G12DEzh2^ΔSET). Significance was measured by 1-way ANOVA followed by a Tukey’s post hoc test. Different letters indicate statistically different P values; ^bP ≤ 0.01, ^cP ≤ 0.001. (G) Higher-magnification images of H&E-stained pancreatic tissue from MKE mice. Green arrows indicate high-grade PanIN lesions and putative PDAC that is only found in these animals. Scale bar: 50 μm.

Figure 6. MKE mice exhibit extensive ductal and PanIN lesion progression.

(A and B) Representative IHC for amylase (A) or CK-19 (B) on pancreatic tissue 60 days after KRAS^G12D induction. Genotypes are indicated. Scale bar: 100 μm. (C) Quantification of amylase staining in the various genotypes based on IHC staining. Data are shown as mean ± minimum to maximum (n = 3 mice for KRAS^G12D; n = 5 mice for Ezh2^ΔSET, KRAS^G12DEzh2^ΔSET, and Mist1^creERT/– KRAS^G12D; and n = 6 mice for control and MKE). Significance was measured by 1-way ANOVA followed by a Tukey’s post hoc test. ^bP ≤ 0.001. (D) Representative immunofluorescence for SOX9 on pancreatic sections 60 days after KRAS^G12D induction. Genotypes are indicated. Nuclei are counterstained with DAPI. White arrows identity positive SOX9 cells. Scale bar: 50 μm. (E) Quantification of SOX9 staining in the different mouse lines based on IF staining. Data are shown as mean ± minimum to maximum (n = 3 mice per conditions). Significance was measured by 1-way ANOVA followed by a Tukey’s post hoc test. Different letters indicate statistically different P values. ^bP ≤ 0.001, ^cP ≤ 0.0001.

Figure 7. Acinar-specific deletion of Ezh2^ΔSET in KRAS^G12D-mediated PDAC alters the tumor microenvironment.

(A) Representative H&E staining of pancreatic tissue from Mist1^creERT/– KRAS^G12D and MKE mice 22 days after KRAS^G12D induction. Green arrows indicate ADM. Scale bar: 50 μm. (B) Principal component analysis based on RNA-Seq data 22 days after KRAS^G12D induction. (C) Volcano plot showing differentially expressed genes between Mist1^creERT/– KRAS^G12D and MKE mice 22 days after KRAS^G12D induction based on RNA-Seq. Genes with significantly lower or higher expression in MKE mice are indicated in blue and red, respectively. Significance was determined with a FDR ≤ 0.05. (D) Top 20 pathways identified by gene set enrichment analysis using GO terms based on RNA-Seq (P_adj. ≤ 0.05). (E) Gene set enrichment analysis shows increased enrichment in KEGG pathways “Regulation of inflammation response” and “ECM organization” in MKE tissue compared with Mist1^creERT/– KRAS^G12D. Similar analysis shows decreased enrichment of genes involved in “Nucleosome assembly” in MKE tissue. (F) Representative trichrome blue staining of pancreas section from control, Ezh2^ΔSET, KRAS^G12D, KRAS^G12DEzh2^ΔSET, Mist1^creERT/– KRAS^G12D, and MKE mice 60 days after KRAS^G12D induction. Scale bar: 100 μm.

Figure 8. EZH2^ΔSET deletion increases ADM in the absence of the tissue microenvironment.

(A) Experimental design for acinar cell isolation and embedding into collagen 22 days after KRAS^G12D induction. (B) Representative images of cell clusters 3 and 7 days after acinar cell isolation. Genotypes are indicated. Scale bar: 100 μm. (C) Quantification of the percentage of cell clusters with visible ADM, 1–9 days after acinar cell isolation. Fifty or more clusters were counted for each condition. Data are shown as mean ± SEM (n = 2 for Mist1^creERT/– KRAS^G12D, n = 3 mice for KRAS^G12DEzh2^ΔSET, n = 4 mice for KRAS^G12D, n = 5 mice for Ezh2^ΔSET, n = 6 mice for control, and n = 7 mice for MKE). (D) Representative images of control acinar after 7 days of treatment with increasing amounts of EZH2 inhibitor EPZ6438. Scale bar: 100 μm. (E) Quantification of 50+ acinar clusters for each condition. Data are shown as mean ± SEM. n = 3. In all cases, significance was measured by a repeated measures 1-way ANOVA followed by Dunnett’s correction. *P ≤ 0.05, **P ≤ 0.01, ***P < 0.001.

Figure 9. EZH2^ΔSET deletion has different cell autonomous roles depending on the context in which KRAS^G12D is expressed.

(A) Morphology of KRAS^G12D, KRAS^G12DEzh2^ΔSET, and MKE tissue 2 weeks after induction of CIP. Scale bar: 500 μm (left images), 100 μm (right images). (B) Representative images of organoids cultured in matrigel 1, 7, and 11 days after isolation. Cells were seeded at 5,000 cells. Genotypes are indicated. Scale bar: 2.4 mm. (C) Representative images of organoids 0, 5, and 7 days after first passage. Cells were seeded at 5,000 cells. Scale bar: 2.4 mm. (D) Quantification of organoid area 5 and 7 days after passage for KRAS^G12D, KRAS^G12DEzh2^ΔSET, and MKE cultures. Data represent mean ± 95% CI. Number of organoids assessed is indicated above each data point. Significance was measured by 2-way ANOVA followed by Tukey’s correction. ***P ≤ 0.001.