Two-Way Development of the Genetic Model for Endometrial Tumorigenesis in Mice: Current and Future Perspectives.

In addition to pursuing tissue specificity in vivo by developing novel Cre transgenic mice with the promoter of endometrium-specific genes, researchers have also attempted to achieve the same goal using an alternative approach. It is a cell-based carcinogenesis model comprising two steps: 1 ) ex vivo culture of murine primary endometrial cells followed by gene transduction to reconstitute EC-specific mutation and 2 ) allograft implantation of transduced cells into immunocompromised mice for in vivo monitoring of tumor development. An obvious advantage of this ex vivo/in vivo hybrid approach is that the cell and tissue type specificity is warranted by physical separation of the uterus and that the outcome is apparent within a short period. On the other hand, the caveat is that the authentic uterine microenvironment, including the proficient immune response, is not thoroughly provided, potentially affecting the outcome depending on the genetic alterations tested in this model. At any rate, no model is perfect by itself. Ideally, they should complement one another to move EC research forward. Below, we have illustrated recent efforts, including ours, to develop cell-based ex vivo/in vivo hybrid models for endometrial carcinogenesis. To achieve uterine-specific genetic engineering ex vivo , the uterus was isolated from wild-type (WT) mice. After enzymatic and mechanical dissociation, primary murine epithelial cells were transiently prepared and expanded in a Petri dish. Genetic engineering, such as the introduction of myr - Akt (encoding a membrane-bound constitutive active form of Akt) or sh Pten (encoding short hairpin RNA against Pten ) into primary endometrial cells, is conducted using lentiviral vectors. By implanting transduced endometrial cells, together with the WT uterine stromal cells from neonates, into the renal capsule of severe combined immunodeficiency (SCID) mice, adenocarcinoma development was achieved ( Memarzadeh et al., 2010 ). This study demonstrated that the generation of GEM is dispensable for the induction of EC in murine cells. Similarly, epithelial cells isolated from the uterus of GEM also develop tumors. After isolation of the uterus from Pten fl/fl mice, epithelial cells were separated from cultured stromal cells using flow cytometry. After infecting these cells with lentivirus encoding Cre (LV- Cre ), the resultant Pten -null epithelial cells and WT uterine stroma were jointly implanted into the renal capsule of oophorectomized SCID mice. With estrogen supplementation, adenocarcinoma-like tumors eventually develop ( Janzen et al., 2013 ). These two studies suggest the importance of cooperation between epithelia and stroma for the development of Pten deletion-driven EC, although whether co-transplanted uterine stromal cells are dispensable for tumor development is yet to be investigated. In addition, it remains to be seen whether tumors can develop even in the subcutis, where there is a lower blood supply. Organoid culture is an emerging technology that enables the self-renewal and differentiation of normal epithelial stem cells in a matrix-assisted three-dimensional structure supplemented with distinct stem cell niche factors. As organoids are maintained in serum-free media, epithelial cells are selectively propagated in response to EGF, while non-epithelial cells are quickly and automatically depleted from the culture. Thus, flow cytometry-based sorting is not necessary for the purification of epithelial cells. It was first demonstrated that murine intestinal stem cells marked by Lgr5 could infinitely proliferate in vitro , giving rise to all differentiated lineages of the small intestine ( Sato et al., 2009 ). Organoid culture technology has been extensively applied to various cancer research fields, including carcinogenesis, drug resistance, and drug discovery ( Drost and Clevers, 2018 ). We previously developed an efficient in vitro gene transduction method using a lentivirus ( Maru et al., 2016 ), which is based on the Matrigel bilayer organoid culture (MBOC) protocol ( Maru et al., 2019a ). This technical innovation enabled us to establish the first murine organoid-based carcinogenesis model with WT normal cells in the ex vivo/in vivo setting ( Figure 1 ). It recapitulated multi-step colon carcinogenesis in the subcutis of nude mice without generating GEM. Similarly, multiple genetic alterations were reconstituted in murine and human intestinal organoids using CRISPR/Cas9 technology to generate organoid models of multi-step colorectal carcinogenesis ( Drost et al., 2015 ; O'rourke et al., 2017 ). Although tumor development from murine organoids is achieved by one or two genetic alterations, it remains technically challenging to induce full-blown tumors from normal human organoids even with three to four genetic alterations. This suggests that murine cells might have a lower threshold for tumorigenesis, presumably reflecting a shorter lifespan in mice. Organoid-based ex vivo/in vivo hybrid carcinogenesis model. Primary cells were isolated and subjected to the Matrigel bilayer organoid culture method. Cancer-specific recurrent genetic alterations are reconstituted in murine organoids by lentiviral infection. These transduced organoids were inoculated into the subcutis of nude mice to monitor tumor development for 8–10 weeks. The effects of particular genetic alterations on tumorigenic potential were evaluated in terms of the incidence, size, and histology of the tumors. Induced subcutaneous tumors and mouse-derived tumor organoids can be analyzed using many downstream assays. Using this approach, we demonstrated that reconstitution of the common genetic alterations in each organ could considerably and generally recapitulate multi-step carcinogenesis independently of the tissue-specific microenvironment. For example, mutant Kras in intestinal organoids markedly accelerates tumorigenesis caused by Apc knockdown alone ( Onuma et al., 2013 ). In organoids from the hepatobiliary tract ( Ochiai et al., 2019 ) and pancreas ( Matsuura et al., 2020 ), full-blown tumors are developed not by mutant Kras alone but by concurrent inactivation of the p53 or Rb pathways. These outcomes are consistent with the results from earlier in vivo GEM studies with identical genetic alterations, if the tumorigenic potential was evaluated based on the tumor development rate, size, and histology. Therefore, these findings point toward the notion that these models could, at least partly, substitute and complement GEMs in evaluating the pathological relevance of reconstituted genetic interactions. Ex vivo/in vivo hybrid carcinogenesis models using murine FT organoids have been recently described ( Maru et al., 2021b ). Although Trp53 loss alone is not sufficient for tumor development in FT organoids, the concurrent introduction of common genetic aberrations in ovarian high-grade serous carcinoma leads to the development of tumors of various grades and histological features. For example, adding Pik3ca H1047R or Kras G12D to p53-null FT organoids leads to the development of adenocarcinoma and CS, respectively. Drug sensitivity varies among mouse-derived tumor organoids (MDTO) with different genetic aberrations. Given the ectopic nature as a tumor developing site and the lack of the right immune response, the subcutis of Balb/c background nude mice may not be ideal for C57bl/6J background endometrial organoids for tumor development. However, we initially did not pay much attention to this situation because the purpose of inoculating transduced organoids in nude mice was to detect cellular transformation, which was supposed to be achieved by genetic engineering alone. Intriguingly, a detailed analysis of MDTO revealed that this artificial setting supported Kras -driven carcinogenesis from pancreatic organoids in a way that recapitulates pancreatic carcinogenesis in humans ( Matsuura et al., 2020 ). For example, organoids acquired genetic or epigenetic changes after subcutaneous inoculation, which were, in turn, positively selected by nude mice for more malignant tumors. Moreover, mutation signatures in MDTO were similar to those observed in human pancreatic cancer. Based on these findings, it is tempting to speculate that the hybrid carcinogenesis models might depend on both ex vivo and in vivo parts. Both would further cooperate for tumorigenesis. Thus, we updated our view on ex vivo/in vivo hybrid carcinogenesis models that had initially emphasized the ex vivo part. Murine endometrial organoids recapitulate the physiological response of the endometrial epithelium to hormones, including estrogen and progesterone ( Boretto et al., 2017 ). However, it remains unclear whether carcinogenesis could also be recapitulated. We recently aimed to establish an ex vivo/in vivo hybrid carcinogenesis model using endometrial organoids ( Maru et al., 2021a ). Unlike in earlier in vivo studies using GEM, inactivation of Pten in endometrial organoids has only a marginal impact on tumorigenesis, even when combined with mutant Kras . These results reconfirmed the relevance of the uterine stroma in endometrial carcinogenesis. We are currently testing many other combinations of genetic alterations that could potentially cooperate for EC development. We have already generated multiple tumors, and detailed information will be reported elsewhere. Intriguingly, the propagation of endometrial organoids was unexpectedly halted after the lentiviral introduction of the pLKO.1-puro vector and sh Luc (shRNA against luciferase ), an empty backbone vector and the negative control shRNA, respectively ( Maru et al., 2021a ). These phenomena have never been observed in any other tissue-derived organoids under almost the same culture conditions ( Onuma et al., 2013 ; Ochiai et al., 2019 ; Matsuura et al., 2020 ; Maru et al., 2021b ). In addition, endometrial organoids with potentially advantageous alterations, such as Cre -mediated Kras G12D induction or shRNA-mediated Pten knockdown, continued to proliferate over many passages after lentiviral infection. Together, these observations raise the possibility that endometrial organoids may be extremely sensitive to DNA damage triggered by genome integration following lentiviral infection, which could be overcome by oncogenic signals. Alternatively, this might imply that the currently adopted culture conditions require further optimization for endometrial organoids. Notably, Cdkn2a knockdown or Trp53 deletion in Kras G12D -expressing endometrial organoids leads to CS or monophasic sarcoma development via extensive epithelial-mesenchymal transition ( Maru et al., 2021a ). Re-inoculating MDTO into the subcutis of nude mice revealed that the transition from carcinoma to CS might be unidirectional. Considering that identical genetic aberrations in pancreatobiliary organoids always result in the development of adenocarcinoma ( Ochiai et al., 2019 ; Matsuura et al., 2020 ), these findings, together with the results of a study with FT organoids, suggest that gynecological organs, in general, may be predisposed to CS due to some inherent epigenetic status. Detailed examination of organoids in each step of the models revealed that spontaneous deletion of the Kras WT allele frequently occurs in MDTOs with Kras G12D and sh Cdkn2a knockdown, but not in those with Kras G12D and Trp53 deletion. These findings are in line with the notion that the Kras WT , which is normally regarded as an oncogene, serves as a relative tumor suppressor gene in the presence of Kras G12D ( Zhang et al., 2001 ), suggesting that Cdkn2a suppression might be less tumorigenic than Trp53 loss in the context of Kras -driven tumorigenesis in gynecological organs. It remains to be seen whether tumor development can be achieved in GEM models with identical genetic alterations.

Mouse genetic models that recapitulate EC development to varying extents in in vivo or ex vivo/in vivo hybrid settings have been developed. Precise modeling of EC and mutual comparison will contribute to both basic and translational research on EC. We hope that this review article will help researchers in the field to develop novel EC models.

The development of genetic mouse models for EC has been extended in two different directions. One conventional method is to pursue the generation of GEMs with the highest specificity for genetic engineering of the uterine epithelium. To this end, the two Cre lines, Ltf-iCre ( Daikoku et al., 2014 ) and BAC -Sprr2f-Cre ( Cuevas et al., 2019 ), are currently the most sophisticated in terms of accurate genetic engineering of the endometrium. As the cooperation between epithelial cells and the microenvironment, such as stromal cells and immune cells, is also integrated, these in vivo GEM models will likely contribute to the elucidation of the molecular mechanisms underlying EC initiation and progression. Hence, it is unlikely that there is a growing demand for the development of novel superior Cre lines in generating EC models. Researchers would rather generate more GEMs with diverse combinations of genetic alterations using these available Cre lines. Most previous studies mainly focused on histological features but not genomic, transcriptomic or proteomic features concerning the similarity between induced mouse uterine tumors and human EC. Now that public databases of so-called Omics-based analysis of human tumors have become available, future studies will definitely need to include these data in assessing the inter-species similarity of the tumors. Another new way is to develop ex vivo/in vivo hybrid models based on genetic engineering of primary endometrial cells or organoids. Given the rapid application of organoids and CRISPR/Cas9 technology in many research fields ( Izumiya et al., 2021 ), it is probable that an increasing number of organoid-based models for EC will be developed. However, only a few such models have been documented for FRT organs to date ( Zhang et al., 2019 ; Lohmussaar et al., 2020 ; Maru et al., 2021a ; Maru et al., 2021b ), including ours being the first and only EC model ( Maru et al., 2021a ). Although this study described the development of a novel model for uterine CS, it belongs to a minor category of type II EC, and no organoid-based model for type I EC has been developed so far. However, in an ongoing project, we are developing new organoid-based type I EC models. These results will also shed light on the mechanisms by which the outcomes could vary between in vivo and ex vivo/in vivo hybrid models for EC. One of the obvious shortcomings of the current ex vivo/in vivo hybrid carcinogenesis model is the absence of a proficient immune response and physiological tissue-specific microenvironment. To overcome this limitation associated with the subcutis of nude mice, we recently established an orthotopic gallbladder (GB) cancer model using syngeneic mice ( Kato et al., 2021 ). Specifically, we generated genetically engineered organoids in vitro using LV- Cre and CRISPR/Cas9 technology. The organoids were inoculated into the dorsal skin of a syngeneic WT mouse, which developed subcutaneous tumors in several weeks. Subsequently, minced tumor fragments were directly sutured to the outer surface of the GB of another syngeneic WT mouse. This novel two-step model enabled detailed analysis of tumor-infiltrating immune cells and evaluation of drug responses in more physiological settings. Notably, the therapeutic effects of immune checkpoint inhibitors will become feasible. In our earlier studies, the subcutis of nude mice was selected as the target site for organoid inoculation. Still, it was mainly for technical convenience and did not necessarily mean using nude mice is required. Indeed, we confirmed that transduced pancreatic organoids could develop in the subcutis of syngeneic mice, albeit smaller in size ( Matsuura et al., 2020 ). Moreover, orthotopic implantation of or tumor organoids or fragments derived from nude mice efficiently developed tumors in the pancreas. Similarly, we obtained subcutaneous tumors in syngeneic mice with a subset of endometrial organoids expressing Kras G12D and shRNAs targeting certain tumor suppressor genes. Therefore, the application of the abovementioned two-step approach to transduced endometrial organoids that are proven tumorigenic in nude mice might be worth testing and should be further pursued. Such efforts will probably lead to significant improvements in the current organoid-based EC model as a preclinical model. Considering the feasibility of organoid culture of patient-derived normal endometrium and endometriosis ( Nikolakopoulou and Turco, 2021 ), establishing an organoid-based carcinogenesis model in humans might also be plausible. Patient-derived tumor organoids (PDTO) from diverse types of cancer ( Bleijs et al., 2019 ; Maru et al., 2019b ; Maru and Hippo, 2019 ), including EC ( Boretto et al., 2019 ; Maru et al., 2019c ; Alzamil et al., 2021 ), have been recently established. Collectively, integrated analyses of the two-way mouse models of EC, namely in vivo GEM models and ex vivo/in vivo hybrid carcinogenesis models, and PDTO as a human model, will probably accelerate research on EC in many aspects, such as the elucidation of the mechanisms underlying carcinogenesis and development of new therapeutic strategies.

Endometrial cancer (EC) is the most common gynecological malignancy and the fourth most common cancer in developed countries ( Yamagami et al., 2017 ; Siegel et al., 2020 ). Risk factors for EC include age, obesity, diabetes, and exposure to unopposed estrogen or tamoxifen. The 5-year overall survival rate is between 74 and 90% for patients with stage I or II EC according to the International Federation of Gynecology and Obstetrics (FIGO) staging, whereas that of patients with EC in stage IV EC is 20–26% ( Morice et al., 2016 ). Patients with the following hereditary diseases are susceptible to EC development: Cowden’s disease, Peutz–Jeghers syndrome, and Lynch syndrome, which are caused by germline mutations in PTEN ( Blumenthal and Dennis, 2008 ), LKB1 ( Hearle et al., 2006 ), and mismatch repair genes ( Meyer et al., 2009 ), respectively. According to the traditional classification, EC is roughly divided into two distinct subtypes based on clinical and pathological characteristics. Type I EC, comprising 80% of all ECs, is mainly associated with estrogen dependency, endometrial hyperplasia, and a favorable prognosis. The predominant histological subtype is endometrioid carcinoma, harboring mutations most frequently in PTEN , followed by PIK3CA , KRAS , ARID1A , and CTNNB1 . Type II EC, comprising various minor subtypes including serous carcinoma, clear cell carcinoma, and carcinosarcoma (CS), is associated with estrogen independence, endometrial atrophy, and poor prognoses. Mutations in TP53 , PIK3CA , PPP2R1A , and ERBB2 amplification are among the most frequent genetic alterations ( Kandoth et al., 2013 ; Murali et al., 2014 ). Based on comprehensive genomic profiling of EC, four molecular subtypes have been recently proposed: DNA polymerase epsilon (POLE) ultramutated type, microsatellite instability (MSI) hypermutated type, copy number low type, and copy number high type. Patients with the first two subtypes have a significantly better prognosis than those with the latter two subtypes ( Kandoth et al., 2013 ). As a therapeutic option for EC, surgery is primarily selected if the tumor is locally confined. Surgery is followed by adjuvant radiotherapy or chemotherapy for high-risk patients, such as those suffering from recurrent and metastatic disease, Stage III–IV disease, or type II EC. In these cases, cisplatin, carboplatin, and paclitaxel are the representative agents ( Morice et al., 2016 ). Although molecular targeted therapies, including anti-angiogenesis agents and immune checkpoint inhibitors, are effective in a subset of patients with advanced or recurrent disease ( Lee, 2021 ), treatment options for EC are still limited, underscoring the need for the development of new therapeutic options. Cancer models that accurately mimic the biological properties of human tumors are essential for efficient drug discovery and elucidation of the mechanisms of carcinogenesis. To this end, various murine in vivo models for EC have been established using the genetically engineered mouse (GEM) approach ( Van Nyen et al., 2018 ). Earlier studies have mainly investigated the effect of Pten deletion, while the relevance of other genetic alterations remains to be elucidated. In this review article, we review the status of genetic models for EC. Specifically, we stratified GEMs by assessing a series of Cre transgenic mice generated to achieve uterine-specific genetic engineering. We also illustrate recent advances in the development of cell-based ex vivo/in vivo hybrid models for EC and provide future perspectives in EC modeling.