EZH2 and Endometrial Cancer Development: Insights from a Mouse Model

Enhancer of zeste homolog 2 (EZH2), a core component of polycomb repressive complex 2, plays an important role in cancer development. As both oncogenic and tumor suppressive functions of EZH2 have been documented in the literature, the objective of this study is to determine the impact of Ezh2 deletion on the development and progression of endometrial cancer induced by inactivation of phosphatase and tensin homolog (PTEN), a tumor suppressor gene frequently dysregulated in endometrial cancer patients. To this end, we created mice harboring uterine deletion of both Ezh2 and Pten using Cre recombinase driven by the progesterone receptor (Pgr) promoter. Our results showed reduced tumor burden in Ptend/d; Ezh2d/d mice compared with that of Ptend/d mice during early carcinogenesis. The decreased Ki67 index in EZH2 and PTEN-depleted uteri versus that in PTEN-depleted uteri indicated an oncogenic role of EZH2 during early tumor development. However, mice harboring uterine deletion of both Ezh2 and Pten developed unfavorable disease outcome, accompanied by exacerbated epithelial stratification and heightened inflammatory response. The observed effect was non-cell autonomous and mediated by altered immune response evidenced by massive accumulation of intraluminal neutrophils, a hallmark of endometrial carcinoma in Ptend/d; Ezh2d/d mice during disease progression. Hence, these results reveal dual roles of EZH2 in endometrial cancer development.


Introduction
Endometrial cancer is the most common cancer in the genital tract in women, with approximately 65,570 new cases and 12,940 deaths each year in the United States [1]. Endometrial cancer is classified into two distinct types [2]. The type I cancer represents the major type (~90%) and is often companied by endometrial hyperplasia [2,3]. The type II cancer accounts for~10% of the total cases and is more aggressive than the type I cancer [2][3][4][5]. Histologically, the type I cancer is endometrioid carcinoma while the type II cancer consists of several subtypes, including serous carcinoma and clear-cell carcinoma [6]. Notably, the type I, but not the type II, endometrial cancer is related to estrogen stimulation [7]. Using molecular sequencing technologies, endometrial cancer has been classified into the following types by The Cancer Genome Atlas (TCGA) Research Network: DNA polymerase epsilon catalytic subunit (POLE) (ultramutated), microsatelliteinstability (MSI) (hypermutated), copy-number low, as well as copy-number high [8]. To facilitate the classification in clinical practice, the Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE) has been developed and validated, with the inclusion of immunohistochemical analysis of DNA mismatch repair (MMR) protein and tumor protein p53 (TP53) [9][10][11]. Interestingly, a recent report shows that a combination of tumorinfiltrating lymphocytes pattern and MMR may be used as a surrogate for the POLE mutation group [12]. ProMisE has been used in molecular diagnosis of human endometrial cancer [13].
Significant challenges remain for endometrial cancer treatment. Determining the histological subtype of endometrial cancer is an effective strategy that guides cancer treatment, with an emerging need to incorporate more molecular details into clinical interventions [14]. While surgery remains to be the most common option to treat this gynecological malignancy, new therapeutic strategies targeting actionable mutations and/or molecular pathways are potentially valuable [15,16]. Of particular importance, knowledge gaps need to be filled in areas of early cancer diagnostics, cancer risk stratification, and molecular identity-based treatment options [14].
We previously showed that uterine-specific loss of EZH2 in the mouse provokes the formation of stratified epithelia and the development of endometrial hyperplasia [30]. To determine the impact of Ezh2 deletion on the development and progression of endometrial cancer induced by PTEN inactivation, we generated mice containing double deletion of Ezh2 and Pten using Cre recombinase driven by the progesterone receptor (Pgr) promoter. Our results revealed dual roles of EZH2 in endometrial cancer development.

Animals and Ethics
Protocols involving the use of mice were approved by Texas A&M University Institutional Animal Care and Use Committee. Mice were on a mixed C57BL/6/129SvEv background and handled according to NIH guideline for the Care and Use of Laboratory Animals. The reporting of experiments followed the ARRIVE guidelines. Mice were housed in the Laboratory Animal Resources and Research (LARR) facility under a 12-h light: 12-h dark cycle. Pgr-Cre mice were generated previously [31]. Ezh2 flox/flox mice (# 022616) and Pten flox/flox mice (# 006440) were purchased from the Jackson Laboratory.

Histology, Immunohistochemistry, and Immunofluorescence
Uterine samples were fixed in 10% (v/v) neutral-buffered formalin (MilliporeSigma, Burlington, MA, USA), embedded in paraffin, and processed using the Texas A&M College of Veterinary Medicine & Biomedical Sciences Core Histology Laboratory. Sections (5 µm) were subjected to hematoxylin and eosin (H.E.) staining and Periodic Acid Schiff (PAS) staining to determine the histopathological features of the uterus/endometrial cancer. Immunohistochemistry and immunofluorescence procedures were detailed elsewhere [34]. Briefly, slides were deparaffinized, rehydrated, and boiled in sodium citrate buffer (pH = 6) to restore antigenicity. Sections were then blocked and incubated sequentially with primary antibodies (Table 1) overnight at 4 • C and biotinylated secondary antibodies (immunohistochemistry) or fluorescent secondary antibodies (immunofluorescence). For immunohistochemistry, avidin-biotin complex (# PK-6100; Vector Laboratories, Burlingame, CA, USA) and NovaRed (#SK-4800; Vector Laboratories) were used to amplify the signal and develop the slides, respectively. For immunohistochemistry, slides were mounted with Fisher mounting medium. In contrast, slides from immunofluorescence experiment were directly mounted using DAPI-containing medium to counterstain the nuclei.

Western Blot
Western blot was performed as described elsewhere [34]. Uterine tissue homogenates were prepared from mice at 14 days of age, and 30 µg of protein samples were subject to electrophoresis. Incubation of primary antibodies (Table 1) was carried out overnight at 4 • C. Western blot images were quantified using NIH Image J (version 1.52p, Bethesda, MD, USA).

Hypoxia Staining
Hypoxia staining was performed using Hypoxyprobe Plus Kit (# HP2-100Kit, Hypoxyprobe, Burlington, MA, USA) based on the manufacturer's protocol. Briefly, hypoxyprobe-1 (pimonidazole) was administered prior to sample collection. Slides were incubated at 60 • C for 20 min, deparaffinized, and then rehydrated. The slides were then treated with H 2 O 2 to inactivate endogenous peroxidase activity. The antigenicity was restored by boiling the sides in sodium citrate buffer (pH = 6). After being blocked with non-immune sera, slides were incubated with FITC-MAb1 (1:50) at 4 • C overnight. In hypoxic cells/tissues, pimonidazole is bioreductively activated to form stable adducts (PIM) detectable by immunostaining. The next day, slides were washed and then mounted with DAPI-containing medium and examined under a fluorescence microscope (Olympus, Waltham, MA, USA).

Enzyme-Linked Immunosorbent Assay (ELISA)
The levels of mouse neutrophil elastase/ELA2 or tumor necrosis factor α (TNFα) in the serum/uterine tissue homogenates were measured using Quantikine ELISA kit (R&D, Minneapolis, MN, USA) based on manufacturer's instructions. Serum samples were diluted (1:5-1:100) to meet the detection range, with at least three biological replicates per experimental group and two technical replicates per sample. Tissue homogenates were prepared and then treated with repeated freezing and thawing cycles. The optical density (OD) values were measured using a microplate reader (Bio-Rad, Hercules, CA, USA) at dual wavelengths (450 nm and 540 nm). OD values were corrected by subtracting readings at 540 nm from those at 450 nm. The concentration of ELA2 or TNFα was calculated using an online tool (http://elisaanalysis.com (accessed on 28 January 2020)).

Hormone Assays
Serum estradiol and progesterone levels from nine-week-old Pten d/d and Pten d/d ; Ezh2 d/d mice were determined using the Ligand Assay and Analysis Core (Center for Research in Reproduction, University of Virginia). Assay details are available at https: //med.virginia.edu/research-in-reproduction/ligand-assay-analysis-core/ (accessed on 2 July 2021).

Statistical Analysis
Statistical analysis was conducted using GraphPad Prism 9 (San Diego, CA, USA). Unpaired two-tailed t-test was used to compare means between two groups. One-way analysis of variance (ANOVA) and Tukey's multiple comparison test were used to compare means among multiple groups. Kaplan-Meier survival curves were analyzed using Logrank (Mantel-Cox) test. Data are means ± s.e.m. A p value of less than 0.05 was reported as statistically significant (* p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001).

Generation of Mice with Conditional Deletion of Ezh2 and Pten
Conditional ablation of PTEN using Pgr-Cre (Pten d/d ) in the mouse uterus leads to the development of endometrial cancer [20,33]. Ezh2 conditional knockout mice are free of endometrial cancer but develop stratified uterine epithelia that contain basal-like cells absent in the normal uterus [30]. To determine the role of EZH2 in endometrial cancer development, we created mice with simultaneous deletion of Ezh2 and Pten (Pten d/d ; Ezh2 d/d ) or Pten only (Pten d/d ) in the uterus using Cre-LoxP approach ( Figure 1A). Pten f/f and Pten f/f ; Ezh2 f/f mice were included as controls. Recombination of Pten and Ezh2 alleles occurred specifically in the uteri of Pten d/d and Pten d/d ; Ezh2 d/d mice but not controls ( Figure S1A). Conditional deletion of Ezh2 and Pten at the transcript levels was demonstrated using qRT-PCR ( Figure 1B,C). We also verified the ablation of PTEN and EZH2 proteins by western blot ( Figure 1D) and immunostaining ( Figure S1B-G). Of note, expression of EZH2 was increased in Pten d/d uteri compared with age-matched controls (i.e., Pten f/f and Pten f/f ; Ezh2 f/f ) ( Figure 1D). Loss of PTEN was expected to enhance PI3K/AKT activity, as PTEN inhibits PI3K-AKT pathway [38]. Indeed, increased expression of phospho-AKT (pAKT) was found in the uteri of Pten d/d and Pten d/d ; Ezh2 d/d mice versus controls (Figures 1D and S1H-J).
EZH2 is a histone methyltransferase that methylates H3K27 [23]. To determine if loss of EZH2 affected H3K27me3 levels, we examined the expression of H3K27me3 using uteri from controls, Pten d/d , and Pten d/d ; Ezh2 d/d mice at 14 days of age. Immunostaining revealed reduced levels of H3K27me3 in Pten d/d ; Ezh2 d/d uteri compared with Pten d/d uteri ( Figure S1K-M). Although EZH2 expression was increased in Pten d/d uteri ( Figure 1D), H3K27me3 levels were not altered in Pten d/d uteri in comparison with the control ( Figure S1N,O). The above results indicate successful ablation of EZH2 and PTEN in the mouse uterus.

Loss of EZH2 Reduces Tumor Burden during Early Carcinogenesis but Negatively Impacts Disease Outcome
Consistent with the documented epithelial hyperplasia and endometrial cancer development resulting from loss of PTEN [20,39], the uterine weights of Pten d/d and Pten d/d ; Ezh2 d/d mice were significantly increased compared with controls at three weeks of age ( Figure 1E). Simultaneous loss of EZH2 and PTEN reduced the uterine weight compared with Pten d/d mice ( Figure 1E). However, the uteri of Pten d/d ; Ezh2 d/d mice were larger than those of Pten d/d mice at the age of nine weeks ( Figure 1F). To assess the outcome of endometrial cancer in mice with conditional deletion of Pten and Ezh2, we generated Kaplan-Meier survival curves, which showed that Pten d/d ; Ezh2 d/d mice succumbed to death starting around two months of age ( Figure 1G). As endometrial cancer in Pten d/d mice does not substantially affect the viability up through five months of age [20,33], current results suggest that deletion of Ezh2 negatively impacts the disease outcome.
As Pten d/d ; Ezh2 d/d mice showed a reduction in uterine weight compared with Pten d/d mice during early tumor development ( Figure 1E), we sought to determine the effect of Ezh2 deletion on the proliferation of endometrial cancer cells. Immunostaining of Ki67, a cell proliferation marker, was performed using uteri from three-week-old control, Pten d/d , and Pten d/d ; Ezh2 d/d mice. Results showed reduced Ki67 index (i.e., number of Ki67-positive cells/number of total cells) in Pten d/d ; Ezh2 d/d uteri versus Pten d/d uteri (Figure 2A-E). It has been reported that retinoic acid (RA) signaling inhibits endometrial cancer cell proliferation [40]. Herein, we found that several genes associated with RA synthesis were upregulated in Pten d/d ; Ezh2 d/d uteri versus Pten d/d uteri ( Figure 2F-H). These genes encode DHRS9 that is involved in RA biosynthesis from retinaldehyde, CRABP2 that transports RA to the RA receptor, and ALDH3B2, an enzyme of the aldehyde dehydrogenase superfamily. The finding that deletion of Ezh2 in Pten d/d uteri reduced endometrial cell proliferation during early carcinogenesis suggests an oncogenic role of EZH2.

Loss of EZH2 Reduces Tumor Burden during Early Carcinogenesis but Negatively Impacts Disease Outcome
Consistent with the documented epithelial hyperplasia and endometrial cancer development resulting from loss of PTEN [20,39], the uterine weights of Pten d/d and Pten d/d ; Ezh2 d/d mice were significantly increased compared with controls at three weeks of age ( Figure 1E). Simultaneous loss of EZH2 and PTEN reduced the uterine weight compared with Pten d/d mice ( Figure 1E). However, the uteri of Pten d/d ; Ezh2 d/d mice were larger than those of Pten d/d mice at the age of nine weeks ( Figure 1F). To assess the outcome of tion [40]. Herein, we found that several genes associated with RA synthesis were upregu lated in Pten d/d ; Ezh2 d/d uteri versus Pten d/d uteri ( Figure 2F-H). These genes encode DHRS9 that is involved in RA biosynthesis from retinaldehyde, CRABP2 that transports RA to the RA receptor, and ALDH3B2, an enzyme of the aldehyde dehydrogenase superfamily. The finding that deletion of Ezh2 in Pten d/d uteri reduced endometrial cell proliferation during early carcinogenesis suggests an oncogenic role of EZH2.

Ezh2 and Pten Deletion Enhances the Accumulation of Intraluminal Neutrophils Compared with Pten Deletion Alone
To begin to understand the cellular basis of altered endometrial cancer progression in Pten d/d ; Ezh2 d/d mice, we examined the morphological/histological changes of the uterus. At one month of age, H.E. staining showed that the size of the uteri was enlarged in both shown in Figure 3K and N. CXCL5 and CXCR2 are critical for recruiting neutrophils to endometrial cancer lesions [39]. To determine whether expression of Cxcl5 and Cxcr2 by uterine epithelial cells was altered upon Ezh2 deletion, we isolated uterine epithelia from  Figure S3).  Neutrophils are critical for cancer development and metastasis [41]. To determine a timeline of the observed intraluminal neutrophil accumulation, tumor development was next examined in Pten d/d ; Ezh2 d/d mice at three weeks of age. To better visualize neutrophil infiltration, we performed immunostaining of lymphocyte antigen 6 complex, locus G (LY6G), a neutrophil marker. Although neutrophil infiltration occurred in both Pten d/d ; Ezh2 d/d and Pten d/d mice, no substantial accumulation of intraluminal neutrophils was observed in Pten d/d ; Ezh2 d/d or Pten d/d mice at this stage ( Figure 3I-N). Negative controls are shown in Figure 3K and N. CXCL5 and CXCR2 are critical for recruiting neutrophils to endometrial cancer lesions [39]. To determine whether expression of Cxcl5 and Cxcr2 by uterine epithelial cells was altered upon Ezh2 deletion, we isolated uterine epithelia from Pten d/d and Pten d/d ; Ezh2 d/d mice. Results showed that the expression levels of Cxcl5 and Cxcr2 were not statistically different between Pten d/d ; Ezh2 d/d and Pten d/d mice, despite a drastic reduction of Ezh2 expression in the epithelia from Pten d/d ; Ezh2 d/d mice ( Figure S3).
To further assess the extent of neutrophil accumulation during tumor progression, we examined the uteri of Pten d/d ; Ezh2 d/d and Pten d/d mice at nine weeks of age. While LY6Gpositive neutrophils were sparse in control uteri ( Figure 4A,D), they were increased within the epithelia in Pten d/d mice ( Figure 4B,E). Strikingly, abundant neutrophils were found in Pten d/d ; Ezh2 d/d uteri encompassing both the stroma and epithelia, despite a substantial loss of uterine epithelia in these mice ( Figures 4C,F and S4). In addition, all Pten d/d ; Ezh2 d/d mice developed the aforementioned ring-like uterine lumen when cross sections were examined ( Figure S4C). As macrophage is involved in the clearance of cellular debris, we also examined the presence of macrophage by immunostaining of F4/80. Results showed that immunoreactive signals of F4/80 were mainly localized to the stroma of control and Pten d/d uteri ( Figure 4G,H,J,K). However, F4/80-positive macrophages were accumulated in both uterine epithelia and stroma of Pten d/d ; Ezh2 d/d mice ( Figure 4I,L). Negative controls are shown in Figure 4M-O.
To further assess the extent of neutrophil accumulation during tumor progression, we examined the uteri of Pten d/d ; Ezh2 d/d and Pten d/d mice at nine weeks of age. While LY6Gpositive neutrophils were sparse in control uteri ( Figure 4A,D), they were increased within the epithelia in Pten d/d mice ( Figure 4B,E). Strikingly, abundant neutrophils were found in Pten d/d ; Ezh2 d/d uteri encompassing both the stroma and epithelia, despite a substantial loss of uterine epithelia in these mice ( Figures 4C,F and S4). In addition, all Pten d/d ; Ezh2 d/d mice developed the aforementioned ring-like uterine lumen when cross sections were examined ( Figure S4C). As macrophage is involved in the clearance of cellular debris, we also examined the presence of macrophage by immunostaining of F4/80. Results showed that immunoreactive signals of F4/80 were mainly localized to the stroma of control and Pten d/d uteri ( Figure 4G,H,J,K). However, F4/80-positive macrophages were accumulated in both uterine epithelia and stroma of Pten d/d ; Ezh2 d/d mice ( Figure 4I,L). Negative controls are shown in Figure 4M-O. To determine a potential link between the stage-dependent intraluminal accumulation of neutrophils and chronic inflammation in Pten d/d ; Ezh2 d/d mice, we measured the levels of serum ELA2, a serine proteinase produced by neutrophils during inflammation [42]. To determine a potential link between the stage-dependent intraluminal accumulation of neutrophils and chronic inflammation in Pten d/d ; Ezh2 d/d mice, we measured the levels of serum ELA2, a serine proteinase produced by neutrophils during inflammation [42]. It was found that ELA2 levels were not statistically different between Pten d/d ; Ezh2 d/d and Pten d/d mice at the age of 1 month ( Figure 5A). In contrast, serum ELA2 levels were markedly elevated in Pten d/d ; Ezh2 d/d mice at nine weeks of age, compared with age-matched Pten d/d mice and controls ( Figure 5B). In addition, we determined the levels of another important pro-inflammatory cytokine, TNFα, in uterine tissue homogenates or the serum of control, Pten d/d , and Pten d/d ; Ezh2 d/d mice. The levels of TNFα were below the limit of detection in controls. However, TNFα levels were elevated in Pten d/d and Pten d/d ; Ezh2 d/d mice, although a statistical significance between Pten d/d and Pten d/d ; Ezh2 d/d mice was not achieved due to sample variations ( Figure 5C,D). Collectively, loss of EZH2 enhanced the accumulation of intraluminal neutrophils, leading to heightened chronic inflammation.  Figure 5A). In contrast, serum ELA2 levels were mark edly elevated in Pten d/d ; Ezh2 d/d mice at nine weeks of age, compared with age-matched Pten d/d mice and controls ( Figure 5B). In addition, we determined the levels of anothe important pro-inflammatory cytokine, TNFα, in uterine tissue homogenates or the serum of control,  Figure 5C,D). Collectively, loss of EZH2 enhanced the accumulation of intraluminal neutrophils, leading to heightened chronic inflammation.

Factors Contributing to the Developmental Trajectory of Endometrial Cancer Lacking PTEN and EZH2
As our previous studies showed that conditional loss of EZH2 in the uterus elicit epithelial stratification [30], we asked the question of whether uterine epithelial stratifica tion occurred in Pten d/d ; Ezh2 d/d mice during tumor development. To approach this ques tion, we first performed immunostaining of KRT14 and ΔNp63, two basal cell markers Immunoreactive signals for both KRT14 and ΔNp63 were detectable as early as three weeks of age in Pten d/d ; Ezh2 d/d mice, but not in age-matched Pten d/d mice and controls (Fig  ure 6A-D). Supporting the immunohistochemical results, transcript levels of Krt14 and ΔNp63 were increased in uterine epithelia of Pten d/d ; Ezh2 d/d mice compared with Pten d/ mice ( Figure 6E,F). Further, Pten d/d ; Ezh2 d/d mice contained stratified epithelia positively stained for KRT14 and ΔNp63 at the age of one month ( Figures 6I,J,M,N and S2). In the severe case with massive accumulation of intraluminal neutrophils, nearly the entire uter ine lumen was surrounded by stratified epithelia (Figures 6I,M and S5), in sharp contras to age-matched Pten d/d and control mice, where minor to negligible staining of KRT14 and ΔNp63 was found ( Figure 6G,H,K,L). Immunostaining of ECAD was conducted to show the epithelial components of the uterus ( Figure 6O-R). Thus, conditional deletion of Ezh2 in Pten d/d uteri exacerbated uterine epithelial stratification that might negatively impac the disease outcome.

Factors Contributing to the Developmental Trajectory of Endometrial Cancer Lacking PTEN and EZH2
As our previous studies showed that conditional loss of EZH2 in the uterus elicits epithelial stratification [30], we asked the question of whether uterine epithelial stratification occurred in Pten d/d ; Ezh2 d/d mice during tumor development. To approach this question, we first performed immunostaining of KRT14 and ∆Np63, two basal cell markers. Immunoreactive signals for both KRT14 and ∆Np63 were detectable as early as three weeks of age in Pten d/d ; Ezh2 d/d mice, but not in age-matched Pten d/d mice and controls ( Figure 6A-D). Supporting the immunohistochemical results, transcript levels of Krt14 and ∆Np63 were increased in uterine epithelia of Pten d/d ; Ezh2 d/d mice compared with Pten d/d mice ( Figure 6E,F). Further, Pten d/d ; Ezh2 d/d mice contained stratified epithelia positively stained for KRT14 and ∆Np63 at the age of one month ( Figures 6I,J,M,N and S2). In the severe case with massive accumulation of intraluminal neutrophils, nearly the entire uterine lumen was surrounded by stratified epithelia (Figures 6I,M and S5), in sharp contrast to age-matched Pten d/d and control mice, where minor to negligible staining of KRT14 and ∆Np63 was found ( Figure 6G,H,K,L). Immunostaining of ECAD was conducted to show the epithelial components of the uterus (Figure 6O-R). Thus, conditional deletion of Ezh2 in Pten d/d uteri exacerbated uterine epithelial stratification that might negatively impact the disease outcome. Hypoxia is implicated in endometrial cancer development [39,43]. We found that hypoxia signals were mainly localized to the epithelial compartment of the uteri from Pten d/d and Pten d/d ; Ezh2 d/d mice at one month of age ( Figure 7E-L), with background levels of staining in the control uteri ( Figure 7A-D). Interestingly, reduced hypoxia was observed in Pten d/d ; Ezh2 d/d uteri compared with Pten d/d uteri ( Figure 7E-L). As relieving tumor hypoxia enhances the tumoricidal activity of neutrophils in Pten d/d mouse model [44], lower hypoxic levels in the Pten d/d ; Ezh2 d/d uteri may facilitate the debridement of cancer epithelia by neutrophils, resulting in increased intraluminal accumulation of cancer cells/debris. Hypoxia is implicated in endometrial cancer development [39,43]. We found that hypoxia signals were mainly localized to the epithelial compartment of the uteri from Pten d/d and Pten d/d ; Ezh2 d/d mice at one month of age ( Figure 7E-L), with background levels of staining in the control uteri ( Figure 7A-D). Interestingly, reduced hypoxia was observed in Pten d/d ; Ezh2 d/d uteri compared with Pten d/d uteri ( Figure 7E-L). As relieving tumor hypoxia enhances the tumoricidal activity of neutrophils in Pten d/d mouse model [44], lower hypoxic levels in the Pten d/d ; Ezh2 d/d uteri may facilitate the debridement of cancer epithelia by neutrophils, resulting in increased intraluminal accumulation of cancer cells/debris. Progesterone receptor signaling plays important roles in endometrial cancer development, and loss of PGR is linked to the development of aggressive endometrial cancer [45,46]. Immunostaining was performed to examine whether PGR expression was altered in Pten d/d ; Ezh2 d/d uteri. Results showed reduced PGR expression in the luminal epithelia of 1-month-old Pten d/d ; Ezh2 d/d mice ( Figure 8G-L) compared with age-matched Pten d/d mice and controls ( Figure 8A-F). Hormone assays showed that the levels of estrogen and progesterone were comparable between Pten d/d ; Ezh2 d/d mice and Pten d/d mice ( Figure S6), indicating that ablation of EZH2 did not affect the levels of ovarian steroid hormones. As PGR signaling interacts with estrogen signaling that promotes neutrophil recruitment [47,48], reduced PGR expression may alter estrogen action and inflammation. Collectively, these studies identified potential contributing factors to the unfavorable outcome of endometrial cancer lacking both PTEN and EZH2 ( Figure 8M). Progesterone receptor signaling plays important roles in endometrial cancer development, and loss of PGR is linked to the development of aggressive endometrial cancer [45,46]. Immunostaining was performed to examine whether PGR expression was altered in Pten d/d ; Ezh2 d/d uteri. Results showed reduced PGR expression in the luminal epithelia of 1-monthold Pten d/d ; Ezh2 d/d mice ( Figure 8G-L) compared with age-matched Pten d/d mice and controls ( Figure 8A-F). Hormone assays showed that the levels of estrogen and progesterone were comparable between Pten d/d ; Ezh2 d/d mice and Pten d/d mice ( Figure S6), indicating that ablation of EZH2 did not affect the levels of ovarian steroid hormones. As PGR signaling interacts with estrogen signaling that promotes neutrophil recruitment [47,48], reduced PGR expression may alter estrogen action and inflammation. Collectively, these studies identified potential contributing factors to the unfavorable outcome of endometrial cancer lacking both PTEN and EZH2 ( Figure 8M). Cells 2022, 11, x FOR PEER REVIEW 13 of 19

Discussion
Both PTEN and EZH2 play important roles in endometrial cancer. The mutation of PTEN gene has been identified in ~20% of human endometrial hyperplasia, suggesting its importance in early cancer development [49]. The frequency of PTEN mutation appears to be associated with the histotypes of endometrial cancer, as PTEN mutation occurs in ~40% of endometrioid cancers but only 5% of serous or clear cell endometrial cancers [19]. EZH2 is overexpressed in endometrial cancer, and its downregulation in endometrial cancer cells inhibits cell proliferation [24,50]. A more recent study has identified a correlation between overexpression of EZH2 in endometrial cancer patients and disease-free and overall survival [51]. This report has further demonstrated that silencing EZH2 in endometrial cancer cells impairs the expression of growth-related genes such as peroxiredoxin

Discussion
Both PTEN and EZH2 play important roles in endometrial cancer. The mutation of PTEN gene has been identified in~20% of human endometrial hyperplasia, suggesting its importance in early cancer development [49]. The frequency of PTEN mutation appears to be associated with the histotypes of endometrial cancer, as PTEN mutation occurs in~40% of endometrioid cancers but only 5% of serous or clear cell endometrial cancers [19]. EZH2 is overexpressed in endometrial cancer, and its downregulation in endometrial cancer cells inhibits cell proliferation [24,50]. A more recent study has identified a correlation between overexpression of EZH2 in endometrial cancer patients and disease-free and overall survival [51]. This report has further demonstrated that silencing EZH2 in endometrial cancer cells impairs the expression of growth-related genes such as peroxiredoxin 6 (PRDX6) [51].
The mechanisms underlying EZH2 action in endometrial cancer progression remain incompletely understood. However, it appears that microRNA-361/Twist axis plays an important role in mediating the role of EZH2 in driving endometrial cancer development [52]. The evidence points to the therapeutic potential of targeting EZH2. However, EZH2 may also function as a tumor suppressor in myeloma and pancreatic tumor [53,54]. It has been shown that loss of EZH2 in the mouse uterus enhances epithelial cell proliferation [55][56][57] and induces epithelial stratification [30]. Herein, we found that conditional deletion of both Ezh2 and Pten reduced cell proliferation and uterine growth during early carcinogenesis but exacerbated intraluminal neutrophil accumulation and chronic inflammation during tumor progression, leading to an unfavorable disease outcome. Current results revealed dual roles of EZH2 in the development of endometrial cancer lacking Pten, a gene frequently mutated in endometrioid carcinomas.
The uterine weights of Pten d/d ; Ezh2 d/d mice were lower than those of Pten d/d mice at three weeks of age, accompanied by reduced cell proliferation revealed by Ki67-staining. As EZH2 inhibits uterine epithelial cell proliferation and uterine growth [30,[55][56][57], our results suggest that EZH2 plays distinct roles in normal uterine epithelial cells versus malignant epithelial cells. Supporting the assumption that the role of EZH2 in PTENdepleted epithelial cells differs from that in PTEN-expressing epithelial cells, it was reported that loss of PTEN or activation of AKT switches the tumor suppressive role of EZH2 to an oncogenic function [58]. Interestingly, AKT activation is also implicated in normal and estrogen-induced uterine epithelial cell proliferation [59]. These findings support a complex, yet contextually dependent, role of EZH2 in cancer development.
Neutrophils are the first-line defenders that actively participate in host defense, tissue damage, and inflammatory disease [60]. Tumor-associated neutrophils play important roles in tumor microenvironment, where N1 neutrophils are anti-tumorigenic and N2 neutrophils are pro-tumorigenic [41,61]. The pro-tumorigenic action of neutrophils is generally associated with their effects on cancer cell invasion, extracellular matrix remodeling, and angiogenesis [61]. Although the oncogenic role of EZH2 has been documented, some in vivo experiments suggest a tumor-suppressive function of EZH2. One study showed that loss of EZH2 promotes KRas G12D -driven oncogenesis in pancreatic cancer [54]. In another report, deletion of Ezh2 accelerates Kras-driven lung adenocarcinoma in a mouse model [62]. In both cases, EZH2 appears to play a role in controlling inflammatory microenvironment [54,62]. In the present study, we found that tumor burden was reduced in Pten d/d ; Ezh2 d/d mice during early tumor development, revealing an oncogenic role of EZH2 in endometrial cancer development. However, unfavorable cancer outcomes were observed in these mice compared with Pten d/d mice. The latter effect is likely non-cell autonomous, as dysregulation of EZH2 in cancer cells is known to alter immune response [63]. Indeed, massive accumulation of intraluminal neutrophils is a hallmark of the endometrial cancer in Pten d/d ; Ezh2 d/d mice at nine weeks of age. Our finding is also consistent with a previous report that increased levels of intratumoral neutrophils correlate with a poor cancer outcome [64].
The underlying mechanisms that promote the heightened inflammation in Pten d/d ; Ezh2 d/d mice remain unclear. However, several important contributing factors were identified by the present study. First, we found reduced hypoxia in the uteri of Pten d/d ; Ezh2 d/d mice at 1 month of age. Elegant studies have demonstrated that hypoxia increases neutrophil recruitment in endometrial cancer induced by PTEN depletion, which serves to restrain the development of endometrial cancer by debridement of the malignant cells [39,44]. Interestingly, reduction of hypoxia causes attenuated neutrophil infiltration. However, these neutrophils gain more efficient capability of attacking cancer cells [44]. Loss of EZH2 limited the extent of hypoxia in Pten d/d ; Ezh2 d/d mice, likely enhancing the tumoricidal effect of neutrophils [44]. Intraluminal accumulation of cancer cells/debris would in turn stimulate neutrophil influx and cause heightened immune reactions, forming a vicious cycle and resulting in chronic inflammation and/or eliciting secondary infectious event. The exact reasons of how EZH2 ablation led to reduced hypoxia is unclear. However, increased vascularization in Pten d/d ; Ezh2 d/d uteri (Fang X and Li Q, unpublished observation) may be one of the reasons. Second, conditional deletion of Ezh2 potentiated epithelial stratification in Pten d/d mice. The uterus contains simple columnar epithelial cells expressing KRT8 but not KRT14 and p63 [65]. Current results showed that stratified epithelial markers KRT14 and ∆Np63 were expressed earlier in Pten d/d ; Ezh2 d/d uteri than Pten d/d uteri, consistent with our previous finding that loss of EZH2 in the uterus promotes the development of basal cells and stratified epithelia [30]. The intensified epithelial stratification in Pten d/d ; Ezh2 d/d uteri likely reflected the additive effect of loss of EZH2 and PTEN. Uterine epithelial stratification is a pathological event that alters the polarity and function of epithelial cells [66,67]. It is possible that epithelial stratification adversely impacts the progression of endometrial cancer due to altered epithelial cell properties. The role of epithelial stratification in endometrial cancer development in our model requires further investigation. Finally, it was found that epithelia adjacent to the uterine lumen had reduced expression of PGR in Pten d/d ; Ezh2 d/d mice at one month of age, when epithelial stratification intensified and marked accumulation of intraluminal neutrophils occurred. PGR loss has been associated with increased cell proliferation and metastasis [45,46]. PGR signaling antagonizes estrogen signaling during tumor development [68]. Estrogen is known to promote neutrophil recruitment during mammary involution or breast cancer development [47,48]. Thus, it is tempting to speculate that the reduction of PGR expression is associated with estrogen-directed neutrophil infiltration and heightened inflammation, which merits further investigation.
Endometrial cancer in Pten d/d mice is not metastatic to other organs even at 25-36 weeks of age [20,33,69]. However, dysregulation of several key regulators/signaling pathways may trigger metastasis. We have shown that conditional deletion of transforming growth factor β type 1 receptor (Tgfbr1) in Pten d/d mice promotes pulmonary metastases [33]. Lung metastasis was also reported in a mouse model where PTEN-ablated and K-ras expressed endometrial cancer cells were grafted [70]. In addition, conditional deletion of both Pten and dicer 1, ribonuclease type III (Dicer1) in the mouse uterus triggers adnexal metastasis [71]. EZH2 expression has been linked to endometrial cancer cell invasion and metastasis [50]. As mice conditionally overexpressing EZH2 are available [72], future investigations are needed to determine whether conditional overexpression of EZH2 in PTEN-depleted uteri impacts metastasis.
From a systems biology perspective, the functions of cells are achieved and coordinated by numerous genes/pathways within a highly interactive network [73]. Cancer may develop when perturbations of protein-protein interactions occur due to gene mutations [74]. Studies on protein-protein interaction networks in cancer may benefit cancer treatment by gaining a holistic view of mechanisms governing tumor development and discovering novel cancer drivers as well as therapeutic targets [75,76]. Recent studies have begun to explore protein-protein interaction networks in female reproductive cancers including endometrial cancer using an integrative computational approach [77]. Defining the interactome of endometrial cancer remains to be one of our key goals in the future.
The current study revealed dual roles of EZH2 in endometrial cancer development. We showed that ablation of EZH2 in the PTEN-inactivated endometrium reduced tumor burden during the early pathogenesis of endometrial cancer. However, these mice progressed to unfavorable disease condition, accompanied by intensified epithelial stratification, massive accumulation of intraluminal neutrophils, and heightened inflammation. These findings point to potentially unwanted effects of EZH2-targed therapy in cancer treatment. EZH2 has been implicated as an important cancer target in different types of cancers. Several EZH2 inhibitors have been developed and tested in clinical trials, such as GSK2816126 and tazemetostat [78,79]. No clinical trials of EZH2 inhibitors in endometrial cancer have been reported. The attenuation of early tumor growth upon Ezh2 deletion in the current study suggests a therapeutic benefit by targeting EZH2 in endometrial cancer. However, the heightened inflammatory response and unfavorable disease outcome during tumor development strongly suggest that caution should be taken when designing antiendometrial cancer strategies. Thus, stage-specific role of EZH2 should be considered.
It is also tempting to postulate that a combination of EZH2 inhibitors and inflammatory modulators may be a possible approach to combat this most common gynecological disease.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/cells11050909/s1. Figure S1: generation and validation of mice with conditional deletion of Ezh2 and Pten; Figure S2: a summary of uterine histological features in Pten d/d and Pten d/d ; Ezh2 d/d mice at 1 month of age; Figure S3: levels of Cxcl5, Cxcr2, and Ezh2 transcripts in uterine epithelial cells isolated from Pten d/d and Pten d/d ; Ezh2 d/d mice; Figure S4: intraluminal neutrophil infiltration in the uteri of Pten d/d and Pten d/d ; Ezh2 d/d mice; Figure S5: immunofluorescence of KRT14 and KRT8 using uteri from 1-month-old Pten d/d ; Ezh2 d/d mice; Figure S6: serum estrogen and progesterone levels in Pten d/d and Ptend /d ; Ezh2 d/d mice.