Conditional abrogation of transforming growth factor-β receptor 1 in PTEN-inactivated endometrium promotes endometrial cancer progression in mice.

Approximately 61,380 new cases and 10,920 deaths from uterine corpus cancers, the majority of which are endometrial cancers, are projected to occur in 2017 in the United States [ 1 ]. There are two major types of endometrial cancer. Type I cancers are mostly endometrioid adenocarcinomas that are associated with excess estrogen. Type II cancers are mainly composed of serous carcinomas which are estrogen independent, with overall poorer prognoses [ 2 ]. The mechanisms of endometrial cancer development are not well defined and effective prophylactic and therapeutic approaches are needed. Therefore, understanding the molecular mechanisms underlying the pathogenesis of endometrial cancer is an essential step toward developing novel targeted therapies. Phosphatase and tensin homolog ( PTEN ), a tumor suppressor gene, is mutated in a variety of human cancers including endometrioid adenocarcinoma. PTEN suppresses the activity of phosphoinositide 3-kinase (PI3K)-AKT signaling pathway by dephosphorylating phosphatidylinositol (3, 4, 5)-trisphosphate (PIP3) that participates in AKT activation. The PI3K-AKT signaling pathway regulates diverse cellular functions, including but not limited to, growth, survival, and metabolism; its abnormal activation is associated with cancer development [ 3 ]. Heterozygous Pten mice develop neoplasms in multiple organs [ 4 ]. Conditional deletion of Pten in the mouse uterus using progesterone receptor ( Pgr )-Cre leads to endometrial cancer formation, supporting a pivotal role of PTEN in endometrial oncogenesis [ 5 ]. Cancer cells, particularly those arising from advanced malignancies, can metastasize from the primary site to secondary site(s). Metastasis, consisting of a series of events including local invasion, intravasation, circulation, extravasation, and colonization, is the major cause of morbidity and mortality in cancer patients [ 6 ]. To establish distant metastasis, cancer cells originating from the primary site acquire an enhanced ability to migrate and invade. Myometrial invasion is an important factor for the diagnosis and staging of endometrial cancer. The depth of myometrial invasion has been used as a criterion for staging endometrial cancers by the International Federation of Gynecology and Obstetrics (FIGO), where FIGO stages IA and IB refer to endometrial cancers with no or less than half myometrial invasion and those with half or more than half myometrial invasion, respectively [ 7 ]. Moreover, a higher risk of extrauterine metastases has been found in endometrial cancer patients with more than 50% myoinvasion compared to those with less than 50% myoinvasion [ 8 ]. Endometrial cancer can metastasize to other organs, with the common sites being the lymph nodes, vagina, peritoneum, and lung [ 9 , 10 ]. It has been increasingly recognized that the cancer microenvironment is critical for metastasis by promoting adhesion, survival, extracellular matrix proteolysis, cell migration/invasion, immune escape, and angiogenesis [ 11 ]. Chemokines and their receptors are important regulators of many cancer cell properties such as proliferation, invasion, apoptosis, and metastasis [ 12 ]. Moreover, tumor-associated macrophages (TAMs), a major population of infiltrating leukocytes and well-known immunosuppressive cells, have been shown to promote cancer growth, invasion, and metastasis [ 13 ]. Of note, metastasis has not been reported in the mouse model of endometrial cancer with uterine Pten depletion [ 5 , 14 ], indicating a need to create additional models to study the metastatic process and associated mechanisms of this gynecologic malignancy. Transforming growth factor beta (TGFB) signaling is known to be tumor suppressive. Many essential elements of this pathway including the ligands, receptors, and SMAD transducers are mutated and/or altered in human diseases including cancers [ 15 ]. In vitro studies suggest that TGFB signaling regulates endometrial cancer cell proliferation, survival, invasion, and metastasis [ 16 - 18 ]. However, the contribution of TGFB signaling to the pathogenesis of endometrial cancer at the organism level remains to be uncovered. Therefore, this study explores the role of TGFB signaling in endometrial cancer development and progression by creating a mouse model that harbors concurrent deletion of Tgfbr1 and Pten in the uterus.

Pten Pgr -Cre conditional knockout (termed Pten d/d ) mice develop endometrial cancer [ 5 ]. To define the function of TGFB signaling in endometrial cancer, we simultaneously ablated Pten and Tgfbr1 in the mouse uterus using Pgr -Cre. To examine the expression of Tgfbr1 in Pten d/d uteri, we performed RNAscope in situ hybridization and demonstrated the localization of Tgfbr1 mRNA to the hyperplastic uterine epithelia in Pten d/d uteri ( supplementary material , Figure S1A, B ). Positive and negative controls were depicted ( supplementary material , Figure S1C, D ). To validate these models, we demonstrated that Pten and Tgfbr1 conditional alleles were recombined in the uteri, but not the tails, of Pten d/d and/or Pten d/d ; Tgfbr1 d/d mice ( supplementary material , Figure S2A ). A significant reduction of mRNA levels of Pten and/or Tgfbr1 was detected in the uteri of Pten d/d , Tgfbr1 d/d , and Pten d/d ; Tgfbr1 d/d mice by real-time PCR analysis ( supplementary material , Figure S2B-D ). Reduction of the protein levels of PTEN in the uteri of Pten d/d ; Tgfbr1 d/d mice was demonstrated using immunohistochemistry and western blot ( supplementary material , Figure S2E-G, K ). Consistent with the loss of inhibition of PI3K-AKT pathway, phospho-AKT (pAKT) levels were increased in PTEN-depleted uteri ( supplementary material , Figure S2H-K ). Thus, we successfully created a mouse model with conditional deletion of Pten and Tgfbr1 in the uterus. Endometrial cancer affects the lifespan of Pten d/d mice, beginning around 5 months of age [ 5 ]. While all Pten d/d , Tgfbr1 d/d , Tgfbr1 f/f , Pten f/f , and Pten f/f ; Tgfbr1 f/f mice survived past 17 weeks, the Pten d/d ; Tgfbr1 d/d mice demonstrated significantly shortened lifespan ( Figure 1A ). Consistent with cancer development, the uterus/body weight ratio was increased in both Pten d/d and Pten d/d ; Tgfbr1 d/d mice versus Tgfbr1 f/f , Pten f/f , Pten f/f ; Tgfbr1 f/f , and Tgfbr1 d/d mice at both 4 and 9 weeks of age ( supplementary material , Figure S3A, B ). The uterine cancer in Pten d/d ; Tgfbr1 d/d mice was generally more hemorrhagic and/or locally invasive ( Figure 1B ). Tgfbr1 d/d mice did not develop endometrial cancer despite the observation of cystic endometrial glands and adenomyosis in 8-month-old mice (data not shown). Thus, this mouse line was not the focus of subsequent studies. To determine potential phenotypic differences in endometrial cancer development between Pten d/d and Pten d/d ; Tgfbr1 d/d mice, we performed H&E staining and immunohistochemical analyses of E-cadherin (ECAD), an epithelial marker, using uteri from Pten d/d , Pten d/d ; Tgfbr1 d/d , and corresponding control mice at various stages. Atypical endometrial hyperplasia was found at the age of 10 days (not shown) and 2 weeks in both Pten d/d and Pten d/d ; Tgfbr1 d/d mice ( supplementary material , Figure S4A-F ). Dramatic differences between Pten d/d ; Tgfbr1 d/d and Pten d/d mice were not found at 2 weeks of age, except the presence of more hyperplastic epithelia in some Pten d/d ; Tgfbr1 d/d mice ( Figure 2A-D ; supplementary material , Figure S4A-F ). At the age of 4 weeks, the atypical endometrial hyperplasia in Pten d/d mice progressed to carcinoma. In contrast, Pten d/d ; Tgfbr1 d/d mice developed more severe lesions with the presence of large foci of adenocarcinoma accompanied by degenerating cells within the central region in some mice ( Figure 2E-H ; supplementary material , Figure S4G-L ). Of note, endometrial cancers in both Pten d/d and Pten d/d ; Tgfbr1 d/d mice were positive for estrogen receptor (ER) and PGR at this stage ( supplementary material , Figure S5A-H ). However, a reduction of PGR but not ER signals was observed in Pten d/d ; Tgfbr1 d/d mice at a later stage ( Figure 2I-P ; supplementary material , Figure S5I ). At the age of 2 months, although myoinvasion could be observed in both Pten d/d and Pten d/d ; Tgfbr1 d/d mice ( supplementary material , Figure S6A-D ), Pten d/d ; Tgfbr1 d/d uteri demonstrated a haphazard glandular pattern, desmoplastic stroma, and more severe myometrial disruption ( supplementary material , Figure S6E, F ; Figure 3A-E ). Immunohistochemical analysis using a smooth muscle marker, calponin 1 (CNN1), revealed that Pten d/d mice had recognizable myometrial layers at 2 months of age ( Figure 3A, B, E ), while severely disrupted myometrial layers intermingled with epithelial cancer cells were found in approximately 69% of Pten d/d ; Tgfbr1 d/d mice ( Figure 3C, D, E ). No myometrial abnormality was found in Tgfbr1 d/d and Tgfbr1 f/f mice at the examined time stage (data not shown). To better visualize epithelial and myometrial compartments, we performed double immunofluorescence using anti-cytokeratin 8 (KRT8) and CNN1 antibodies to label the respective epithelium and smooth muscle ( Figure 3F-M ). The results revealed that cancer epithelia breached the uterine wall in Pten d/d ; Tgfbr1 d/d mice ( Figure 3J-M ) versus Pten d/d mice ( Figure 3F-I ). Thus, these results provide in vivo evidence of accelerated endometrial cancer progression and enhanced cell invasion in mice with conditional deletion of Pten and Tgfbr1 . Metastasis is a major cause of death in endometrial cancer patients [ 9 , 10 ]. In contrast to the Pten d/d mice where visible metastasis in distant organs was not observed even at 25-36 weeks of age ( supplementary material , Table S3 ), Pten d/d ; Tgfbr1 d/d mice developed distant organ metastasis, preferentially in the lung. At the advanced stage of disease development (i.e., 7-16 weeks), no grossly visible metastases were observed in the bladder, heart, kidney, and spleen ( supplementary material , Table S3 ), although microscopically detectable metastases positive for KRT8 and ER could be found in the lymph nodes of Pten d/d ; Tgfbr1 d/d and Pten d/d mice ( supplementary material , Figure S7 ). In contrast, grossly visible metastases were evident in the lungs of Pten d/d ; Tgfbr1 d/d mice ( Figure 4A ; supplementary material , Figure S8A , Table S3 ). Unlike the lungs of Pten d/d mice comprising highly organized alveoli structures ( Figure 4B ), the lungs of Pten d/d ; Tgfbr1 d/d mice demonstrated multiple metastatic sites, consisting of pathological lesions with variable differentiation status ( Figure 4C-G ). In addition, the metastasis was frequently accompanied by hemorrhage and loss of morphologically normal alveoli ( supplementary material , Figure S8A ; Figure 4C-G ). Using immunohistochemistry, we further verified that the metastatic nodules were positive for KRT8 ( Figure 4I ; supplementary material , Figure S8C ) and ER ( Figure 4K ), but were negative for PGR ( Figure 4M ). Of note, analysis of ER expression in the lungs of Pten d/d ; Tgfbr1 d/d mice demonstrated the formation of endometrial gland-like lesions within the metastatic nodules ( supplementary material , Figure S8D, E ). Furthermore, recombined Tgfbr1/Pten conditional alleles were detected in the lung metastases of Pten d/d ; Tgfbr1 d/d mice but not in the lung tissues of Pten d/d mice ( supplementary material , Figure S8F ). The metastatic nodules also expressed mucin 1 (MUC1; Figure 4O ), a protein that is frequently overexpressed in metastatic cancers [ 31 ]. Immunostaining of the lungs from Pten f/f ; Tgfbr1 f/f controls were representatively shown ( Figure 4H, J, L, N ). To confirm that pulmonary metastasis was specific to Pten d/d ; Tgfbr1 d/d mice, we analyzed age-matched lungs from Pten d/d and Pten f/f mice and did not find metastasis in these mice (data not shown). Chemoattractant cytokines, or chemokines, and their receptors play important roles in many carcinogenic events [ 32 ]. To determine the potential mechanism of metastasis resulting from ablation of TGFBR1, we compared uterine mRNA expression of chemokines including C-X-C motif ligand 1 ( Cxcl1 ), Cxcl5 , Cxcl12 , chemokine (C-C motif) ligand 2 ( Ccl2 ), and Ccl9 and a chemokine receptor Cxcr2 that are involved in metastasis [ 33 - 36 ] between Pten d/d ; Tgfbr1 d/d and Pten d/d mice, along with Pten f/f ; Tgfbr1 f/f and Pten f/f controls. Results showed that the transcript levels of Cxcl5 and its receptor Cxcr2 were increased in the uteri of Pten d/d ; Tgfbr1 d/d mice versus Pten d/d and control mice at 2 weeks of age ( Figure 5B, D ), while those of Cxcl1 ( Figure 5A ), Cxcl12 ( Figure 5C ), Ccl2 ( Figure 5E ), and Ccl9 ( Figure 5F ) were comparable between Pten d/d ; Tgfbr1 d/d and Pten d/d mice at this timepoint. To determine whether there was a correlation between serum CXCL5 levels and cancer progression and metastasis, we performed ELISA and demonstrated significantly elevated serum CXCL5 levels in Pten d/d ; Tgfbr1 d/d mice versus Pten d/d and control mice at 9 weeks of age ( Figure 5G ). Interestingly, serum CCL2 levels were also increased in 9-week-old Pten d/d ; Tgfbr1 d/d mice ( Figure 5H ). Further analysis revealed significantly upregulated mRNA levels of Ccl2 and Cxcl5 in the lung metastases of 9-week-old Pten d/d ; Tgfbr1 d/d mice versus age-matched Pten d/d mice ( Figure 5I ), suggesting the lung metastases may serve as a potential source of the elevated serum levels of these chemokines. Furthermore, CXCL5 was localized to the metastatic nodules within the lungs of Pten d/d ; Tgfbr1 d/d mice versus Pten d/d mice ( Figure 5J, K ). These results suggest a potential involvement of CXCL5/CCL2 in endometrial cancer progression in Pten d/d ; Tgfbr1 d/d mice. TGFB signaling regulates CXCL5 secretion in mammary cancer cells and recruits immune cells including TAMs to promote tumor progression [ 34 , 35 ]. TAMs can acquire a polarized M2 phenotype (i.e., M2 macrophage or alternatively activated macrophage) and promote tumor progression [ 37 ]. We therefore examined the potential involvement of TAMs in tumor progression in Pten d/d ; Tgfbr1 d/d mice by immunohistochemistry using antibodies directed to CD163 (a marker for M2 macrophages) [ 38 ]. F4/80, a well-established marker of murine macrophages [ 39 ], was used to monitor the macrophage population in the uteri. The presence of F4/80 + macrophages was revealed in the stroma of both Pten d/d and Pten d/d ; Tgfbr1 d/d tumors at 9 weeks of age ( Figure 6A, C ). Interestingly, cells positive for CD163 were readily detectable in the uteri of Pten d/d ; Tgfbr1 d/d mice versus Pten d/d mice ( Figure 6B, D ). Positive and negative controls were included ( Figure 6E, F ). Consistent with the immunohistochemical observation, western blotting showed that CD163 expression was significantly increased in Pten d/d ; Tgfbr1 d/d mice compared with Pten d/d mice ( Figure 6G, H ). Immunostaining of F4/80 and CD163 using age-matched Pten f/f and Pten f/f ; Tgfbr1 f/f mice was included in supplementary material , Figure S9 . These findings suggest a potential role of TAMs in endometrial cancer progression in Pten d/d ; Tgfbr1 d/d mice.

Mutation or inactivation of TGFB signaling components has been associated with the development of a broad array of cancers [ 40 ]. Although a putative role for TGFB signaling in the pathogenesis of human endometrial cancer has long been proposed [ 41 ], the precise function of TGFB signaling in endometrial cancer development has remained elusive. PTEN mutations have been identified in a variety of cancers. Loss of PTEN, a negative regulator of the PI3K-AKT pathway, is involved in the oncogenesis of endometrial carcinoma [ 42 ]. Conditional deletion of Pten in the mouse uterus using Pgr -Cre leads to the development of endometrial malignancy, which phenocopies many pathological characteristics of the human endometrial cancer [ 5 ]. As a result, the Pten d/d mouse model has been used to study endometrial cancer development [ 14 , 43 , 44 ]. However, Pten d/d mice do not develop visible organ metastasis, a cause of recurrence after surgical intervention for primary cancers. Recurrence of endometrial cancer resulting from either local or distant metastases remains devastating. Endometrial cancer has been reported to have the highest frequency of pulmonary metastases (20-25%) versus other gynecologic cancers [ 9 , 10 ]. Therefore, development of an endometrial cancer model with organ metastases mimicking human endometrial cancer is critical to understand metastatic initiation and progression. Little is known about the potential interplay between TGFB signaling and PTEN during endometrial carcinogenesis although their interactions have been documented in other tumors [ 45 , 46 ]. Inactivation of TGFB signaling and loss of growth inhibition are associated with the development of human endometrial cancer [ 47 , 48 ]. It has been shown that mutations of TGFB signaling components including TGFB receptors and SMADs and alteration of TGFB signaling activity reflected by changes in gene expression and/or phosphorylation of key signaling proteins play a role in the pathoetiology of human endometrial cancer [ 47 , 49 , 50 ]. TGFBR1 has a 5.6% mutation and alteration rate in human endometrial cancer [ 49 , 51 ]. As reported, cancerous human endometrium expresses lower levels of TGFBR1 protein [ 47 , 52 ]. TGFBR2 is mutated/altered in 6.5% endometrial tumors [ 49 , 51 ]. Studies from The Cancer Genome Atlas (TCGA) Research Network have also suggested alterations of several SMADs in human endometrial cancer [ 49 , 51 ]. Of particular importance was the availability of the two endometrial cancer models (i.e., Pten d/d and Pten d/d ; Tgfbr1 d/d mice) that demonstrated distinct pulmonary metastatic outcomes depending on the status of a known TGFB signaling component (i.e., TGFBR1) in the PTEN-inactivated uterus. Therefore, the aforementioned mutations and/or dysregulation of TGFB signaling components in endometrial cancer and the development of pulmonary metastasis in Pten d/d ; Tgfbr1 d/d mice make this model a valuable tool to study the pathogenesis and progression of endometrioid adenocarcinoma, the most frequent type of endometrial cancer in women. However, it needs to be pointed out that due to the expression of Pgr -Cre in uterine stromal cells and the myometrium, a potential impact of TGFBR1 and/or PTEN loss in these cellular compartments on the tumor phenotype cannot be excluded, which may represent a limitation of the current study. Future investigations are needed to address this question using Cre lines targeting distinct cellular compartments of the uterus. Metastasis is a complex process where cancer cells can disseminate via the bloodstream and/or lymphatics to establish secondary tumors at the metastatic sites. Because of the invlovment of chemokines in the process of metastasis, therapeutic strategies targeting the action of chemokines may block/attenuate metastasis. A functional link between TGFB signaling and chemokines during cancer metastasis has emerged [ 33 , 53 , 54 ]. In one study, conditional ablation of TGFBR2 in mammary fibroblasts increased the secretion of CCL2, which promoted mammary cancer progression at least partially through a TAM-dependent mechanism [ 33 ]. Enhanced production of CCL9 resulting from loss of TGFB signaling promotes immature myeloid cell infiltration and the development of invasive intestinal tumors [ 55 ]. Loss of TGFB signaling in mouse mammary epithelial cells promotes metastasis by recruiting myeloid-derived suppressor cells into tumor tissues via CXCL5-CXCR2 and CXCL12-CXCR4 axes [ 34 ]. The increased transcript levels of Cxcl5 and its receptor Cxcr2 in the uteri of Pten d/d ; Tgfbr1 d/d mice suggest a potential common mechanism of CXCL5/CCL2 upregulation in the absence of TGFB receptor-mediated signaling in epithelial cancer cells [ 34 ]. This finding together with the role of CXCL5 in cell invasion and metastasis in several types of cancers including bladder cancer [ 56 ], mammary cancer [ 34 ], and liver cancer [ 57 ] suggest that activation of the CXCL5-CXCR2 axis due to loss of TGFBR1 may contribute to the pulmonary metastasis of endometrial cancer induced by conditional abrogation of PTEN in the uterus. Importantly, we showed that serum levels of CXCL5 and CCL2 were elevated in Pten d/d ; Tgfbr1 d/d mice compared with Pten d/d mice. Since CXCL5 expression is increased in human endometrial cancer tissues compared with normal endometrium [ 58 ], it is plausible to further explore the potential of CXCL5 and CCL2 as biomarkers for endometrial cancer patients. Studies using genetically modified mice have revealed the emerging role of TAMs in the development of invasive endometrial cancer [ 59 , 60 ]. However, the involvement of uterine TGFB signaling in TAM recruitment and polarization and the contribution of TAMs to the development of endometrial cancer remain to be elucidated. At advanced stages of tumor development, CD163 + cells were readily detectable and the expression of CD163 was increased in the uteri of Pten d/d ; Tgfbr1 d/d mice versus Pten d/d mice, suggesting the involvement of TAMs in promoting endometrial cancer progression in Pten d/d ; Tgfbr1 d/d mice. Future studies are warranted to define the mechanism of action of pro-inflammatory chemokines, particularly CXCL5 and CCL2, and the function of TAMs within the tumor microenvironment in our model. It is noteworthy that TGFB signaling may act as a tumor suppressor in pre-malignant cells but as a tumor promoter during advanced tumor development [ 61 ]. That the deletion of Tgfbr1 in PTEN-inactivated uterus accelerates endometrial cancer progression also highlights the complex role of TGFB signaling in vivo , where compensatory pathways may function to promote cancer invasion/metastasis in the absence of TGFBR1. In summary, conditional deletion of Pten and Tgfbr1 leads to a disease that recapitulates invasive and lethal endometrial cancer. This new mouse model is potentially valuable for preclinical testing of targeted therapies to treat endometrial cancer with metastasis.

Mice were on a mixed C57BL/6/129 genetic background and the use of mice for this study was approved by the Institutional Animal Care and Use Committee at Texas A&M University. The Pgr -Cre and Tgfbr1 flox mice were generated previously [ 19 , 20 ]. The Pgr -Cre mice were obtained from Drs. John Lydon and Francesco DeMayo, and the Tgfbr1 flox mice were contributed by Dr. Stefan Karlsson and imported from the Matzuk laboratory at Baylor College of Medicine. The Pten flox mice were purchased from The Jackson Laboratory (stock # 006440; Bar Harbor, ME, USA) [ 21 ]. The genotypes of mice and DNA recombination were analyzed by genomic PCR ( supplementary material , Table S1 ) [ 20 - 29 ]. Tissue samples were fixed in 10% neutral buffered formalin (Sigma, St. Louis, MO, USA), embedded in paraffin wax, and cut into 5 μm thick sections for hematoxylin and eosin (H&E) staining, immunohistochemistry, or immunofluorescence as described [ 30 ]. Antibody details are presented in supplementary material , Table S2 . Western blotting was conducted as described [ 30 ] using the indicated primary antibodies ( supplementary material , Table S2 ). Quantification of western blots was performed using NIH ImageJ (version 1.50i). Data are presented as percentage, where the levels of target protein in the Pten d/d group were set to 100%. Serum CXCL5 levels were measured using Quantikine ELISA kit (R&D, Minneapolis, MN, USA) according to the manufacturer's instruction. In brief, mouse serum samples were diluted 1:20 and assayed in duplicate, along with CXCL5 controls and working standards. Upon completion of the assay, the optical density (OD) value of each well was measured by a microplate reader (BioTek, Winooski, VT, USA) at wavelengths of 450 nm and 540 nm. The OD values were corrected by subtracting readings at 540 nm from those at 450 nm. Serum CCL2 concentration was analyzed using a mouse MCP-1/CCL2 ELISA Kit (Sigma) according to the manufacturer's protocol. The concentration of each sample was calculated using an online software ( http://elisaanalysis.com ). RNAscope 2.5 HD detection reagent (brown) and mouse Tgfbr1 probe (catalog No. 406201) were purchased from Advanced Cell Diagnostics (ACD, Newark, CA, USA) and the analysis was performed according to the manufacturer's instructions. In brief, paraffin sections were deparaffinized, pretreated by boiling, and digested using protease before hybridization. Hybridization of the Tgfbr1 probe set was carried out at 40 °C for 2 h, followed by a series of amplification steps. Brown signals were developed using 3,3′-diaminobenzidine (DAB). Mouse uterine tissues were homogenized in RNA lysis tissue (RLT) buffer (Qiagen, Redwood City, CA, USA). Total RNA was isolated using an RNeasy Mini Kit (Qiagen) based on the manufacturer's protocol, with on-column DNase digestion. The resultant RNA was dissolved in ribonuclease-free water. Reverse transcription was carried out using 200 ng (uterus) or 1 μg (lung) RNA and SuperScript III Reverse Transcriptase (ThermoFisher Scientific, Waltham, MA, USA). Quantitative (real-time) PCR was conducted using a Bio-Rad Real-time PCR Detection System (Hercules, CA, USA). Each assay was performed at least in duplicate using primers listed in supplementary material , Table S1 and iTaq Universal SYBR Green Supermix (Bio-Rad) [ 29 ]. Statistical analysis was performed using GraphPad Prism (version 7.01). Data are mean ± standard error of the mean (s.e.m.). Comparisons between two means were performed using two-tailed t -tests (unpaired). Comparisons of means among multiple groups were performed by one-way analysis of variance (ANOVA) followed by Holm-Sidak pairwise comparisons. Survival curves were analyzed using the Log-rank/Mantel-Cox test. Significantly skewed data were log transformed prior to ANOVA. Statistical significance was defined as * P < 0.05, ** P < 0.01, and *** P < 0.001.

Figure S1. Localization of Tgfbr1 mRNA expression in Pten d/d uteri Figure S2. Validation of Tgfbr1 , Pten , and Tgfbr1/Pten conditional knockout mice Figure S3. Gross analysis of endometrial cancer development Figure S4. Histological analysis of endometrial cancer development Figure S5. Expression of ER and PGR in mouse uteri with conditional deletion of Pten and/or Tgfbr1 Figure S6. Histological analysis of tumor development and myometrial invasion Figure S7. Identification of endometrial cancer metastasis in lymph nodes Figure S8. Lung metastasis in Pten d/d ; Tgfbr1 d/d mice Figure S9. Expression of F4/80 and CD163 in control uteri Table S1. Primers for conventional PCR and RT-qPCR Table S2. Primary antibodies for immunostaining and western blotting Table S3 . Visible metastasis at advanced tumor stage in Pten d/d ; Tgfbr1 d/d mice versus Pten d/d mice