I. Developmental Biology of Female Reproductive Organs

Mechanisms underlying the organogenesis of the female reproductive tract have been the central themes of our research for 20 years. Most of the female reproductive tract, namely the oviduct, uterus, cervix, and vagina, arises from the Müllerian ducts (MDs). MDs undergo a dynamic transformation from simple tubes consisting of homogeneous epithelium and mesenchyme into distinct organs during embryogenesis. Our studies showed that the organ-specific mesenchymal cells drive the morphogenesis of MD-derived organs. The current theory is that the gradient of retinoic acid(RA) in the MD establishes nested and overlapping expression patterns of Abdominal B class HOX gene clusters (HOXA9-A13 and HOXD9-D13). The unique set of HOX transcription factors in each segment regulate organ-specific morphogenesis (Fig. 1A). We are particularly interested in the molecular mechanisms underlying the differentiation of uterine and vaginal epithelia. Utilizing genetically engineered mouse models, we demonstrated that the cell fate of MD epithelium (MDE) to become vaginal or uterine epithelial cells is determined by the expression of ΔNp63α transcription factor: When factors secreted from vaginal mesenchyme (activin A, BMP4, FGF7, and FGF10) induces ΔNp63α in MDE, ΔNp63α-positive cells differentiate into the squamous vaginal epithelium. ΔNp63α autonomously sustains vaginal epithelial-specific gene expression, including ΔNp63α (Fig. 1B). When MDE in the vaginal segment fails to express ΔNp63α, the ΔNp63α-negative cells differentiate into uterine epithelium within the vagina, forming vaginal adenosis. In the uterine segment of MD, reciprocal interactions between MDE and uterine mesenchyme maintain canonical WNT activities, which represses the expression of ΔNp63α in MDE. The uterine gland formation (adenogenesis) is also under the control of the canonical WNT pathway. Crosstalks among canonical WNT pathway signals and its downstream transcription factors (TFs) induce FOXA2, the master regulator of uterine adenogenesis, in a small subset of uterine epithelial cells (Fig. 1B). Subsequently, the FOXA2-positive epithelial cells proliferate, invade the underlying stroma, and form the uterine gland (Fig. 1C). Our current focus is the crosstalk among the TFs essential for adenogenesis. We also investigate the role of adenogenesis signaling pathways in the metastasis of endometrial cancer, as described below.

II. Pathogenesis of Endometrial Cancer (EC)

While the mortality rate of almost all cancers has been decreasing, both the disease-specific mortality rate and incidence of EC continue to increase in developed countries. The discovery and characterization of drivers of endometrial tumorigenesis are central to the ultimate goal of reducing morbidity and mortality. Recent cancer genomics studies showed that genes regulating the development of reproductive organs (e.g. CTNNB1, FOXA2, SOX17, FGFR2) are frequently mutated in EC. Hence, we study the pathogenesis of EC, focusing on the deregulation of signaling pathways that control uterine development and functions.

II-A. Tumor-Repressing Effects of Normal Endometrium

Our recent murine study demonstrated that the tissue microenvironment of normal endometrium represses the growth of neoplastic epithelial cells (Terakawa, et al., 2019 PNAS). In our mouse model, the co-mutation of Pten, Pik3ca, and Ctnnb1, the three most frequently mutated genes in EC, promoted the growth of mutant uterine epithelial cells in a dose-dependent fashion, while each mutant allele alone was insufficient to drive the overgrowth of mutant epithelial cells. Our finding provides a possible explanation for the observation that EC always carries several oncogenic mutations, even when they are of low-grade and early-stage. Meanwhile, the combination of Pten, Pik3ca, and Ctnnb1 mutations did not cause EC when ovarian activities are maintained. Ovariectomy of mice transformed the hyperplasia to metastatic EC, and the ovariectomy (OVX)-induced carcinogenesis was inhibited by treatment with progesterone (P4) or estradiol (E2) indicating that steroid hormones maintain the anti-cancer activity of normal endometrium. This finding may explain why EC occurs predominantly in postmenopausal women, in whom circulating P4 and E2 levels decline. In our mouse model, progesterone receptor (PGR) was detected in stromal and non-mutant epithelial cells but not in the mutant uterine epithelial cells, suggesting that the anti-tumor activity of P4 is mediated by surrounding normal cells. This is not surprising because P4 and E2 control the proliferation, differentiation, and regression of uterine epithelium in part via stromal cells, as demonstrated by our tissue recombination experiments utilizing mice lacking nuclear receptors for estrogen, progesterone, and androgen. We currently study the factors mediating tumor-suppressing effects of P4 and E2 utilizing EC model mice.

II-B. Functions of Mutations associated with EC recurrence

Model: Endometrial Carcinogenesis

CTNNB1 mutations that activate the canonical WNT pathway are associated with the recurrence risk of EC. In our mouse model, Ctnnb1 exon 3 mutation transforms hyperplasia lesions initiated by PTEN and PI3K mutations into EC, driving myometrial invasion and metastasis. In human EC, CTNNB1 mutations are associated with the upregulation of transcription factors (LEF1, DLX5, DLX6, and FOXA2) that regulate uterine adenogenesis (Fig. 1C). Therefore, we pursue a hypothesis that CTNNB1 mutations promote the invasion of ECs into the stroma and myometrium via activation of the adenogenesis pathway.

II-C. Molecular Mechanisms Underlying EC Risks

A steady increase in EC incidence and mortality rates in developed countries suggests that a certain proportion of EC can be prevented by lifestyle change. There are several well-established risk factors for EC, including estrogen exposure, menopause, and obesity. Since endometrial carcinogenesis is a multistep process involving complex interactions among diverse elements (e.g., immune system, gene mutations, hormones, and metabolism), it is unclear the effects of risk factors on which step/element increase EC incident. Our in vivo mouse model is ideal for addressing the molecular mechanism of the EC risk factor effect as it replicates the multistep carcinogenesis of human endometrium from the clonal expansion of initiated cells within the normal endometrial environment to the malignant transformation and metastasis of EC. We are currently assessing how dietary factors increase EC incidence by utilizing the mouse model.

III. Pathogenesis of Uterine Leiomyoma

Uterine leiomyomas (ULMs) or fibrioids are the most common neoplasm of the female reproductive tract with ~70% lifetime prevalence. The growth of ULM was long thought to be driven solely by the E2-induced proliferation of tumor cells. However, our studies utilizing the patient-derived xenograft (PDX) model revised this view by demonstrating that ULM growth is driven by P4 but not E2. We also showed that P4 increases the tumor volume by stimulating hypertrophy, proliferation, and extracellular matrix (ECM) secretion in ULM cells. E2 had no growth-promoting effect but was required to sensitize ULM cells to P4 by inducing PGR. ULM cells survived without P4 and E2, but P4 withdrawn shrunk ULM through the cell size reduction. Clinical trials of selective progesterone receptor modulators (SPRMs) confirmed our findings: SPRMs temporarily reduced the ULM size and associated symptoms, such as abnormal bleeding, but the ULM regrew, and symptoms recurred when SPRMs discontinued.

Whole-genome and whole-exome sequencing studies identified four ULM subtypes with unique genomic alterations. MED12 mutant ULM (MED12-ULM) is the most common subtype accounting for ~70 % of all ULM cases. The second most common subtype, HMGA2-overexpressing ULMs (HMGA2-ULMs), accounts for ~15 %. We demonstrated that the cellular compositions of MED12-ULM and HMGA2-ULMs are distinctive. Approximately 90% of cells in HMGA2-ULMs are tumor smooth muscle cells (T-SMC) overexpressing HMGA2, whereas MED12-ULM primarily consists of cells of two independent lineages, ~60% of T-SMCs and ~40% of tumor-associated fibroblasts (TAFs), and MED12 mutations occurred exclusively in T-SMCs. Our PDX studies determined that P4 induced proliferation of T-SMCs, whereas the growth of TAFs was stimulated by E2 alone. Surprisingly, MED12 mutant T-SMCs do not grow in the cell culture, and the cells growing in the primary culture of MED12-ULM are TAFs. The primary culture was the standard research model for ULM for decades. Thus, our studies raise questions about the accuracy of current ULM literature.

We currently investigate interactions among different cell types in ULMs utilizing the PDX model (Fig. 2). The coordinated growth of T-SMC and TAF suggests that the interactions between these two cell populations play critical roles in the pathogenesis of MED12-ULMs. Furthermore, since blood vessels are embedded in TAF-rich ECM, we are currently testing a hypothesize that TAFs regulate the growth of ULMs by controlling the function of vascular cells.

IV. Molecular Mechanism of Oocyte Quality Control

Our group investigates the functions of the p53 tumor suppressor gene family in the development and quality control of female germ cells utilizing genetically engineered mouse models. The removal of excess and damaged germ cells by p53 homologs is an evolutionally conserved mechanism throughout the animal kingdom. In female mammals, three p53 homologs, namely p53, p63, and p73, comprise a multilayered quality control system for germ cells. The ultimate goal of this project is to improve the health of women and their children through better understanding the function of p53 tumor suppressor homologs in the quality control of oocytes.

Mammalian females lose more than 90% of their germ cells during follicle formation and sexual maturation. Using oocyte-specific mutant mouse models, we showed that TAp63α, a p63 isoform exclusively expressed in immature oocytes within primordial follicles, removes excess oocytes via apoptosis. TAp63α also plays a pivotal role in removing damaged oocytes. Studies by multiple research groups, including my laboratory, established that kinases activated by DNA damage convert inactive TAp63α dimer into tetramer and induce apoptosis in immature ovarian follicles (Fig. 3). Since immature oocytes are highly susceptible to DNA damage due to TAp63α expression, genotoxic cancer therapies reduce the ovarian reserve in young female cancer patients, thereby resulting in premature ovarian insufficiency (POI). We showed that chemical DNA adducts induced by chemotherapy and DNA strand breaks induced by ionizing radiation activate TAp63a through distinct pathways (Fig. 3). Since our studies with oocyte-specific knockout mice of p73 demonstrated that chemotherapy induces oocyte apoptosis via activation of TAp73α, a p73 isoform, we currently investigate the mechanism of TAp73a activation by chemotherapeutic drugs. Based on our unpublished results, we hypothesize that TAp73α plays a central role in oocyte quality control to remove ovarian follicles with mitochondrial dysfunctions. We pursue the function of TAp73α in the oocyte quality control utilizing mutant mice of p53 gene family members.