Estrogen receptor alpha regulates uterine epithelial lineage specification and homeostasis.

EEO were established from global Esr1 null ( Esr1 −/− ) mice 12 and wild-type (WT) littermate control mice on postnatal days (PND) 15, 30, and 60 ( Figure 1 A). Epithelial cells were isolated through enzymatic digestion and cultured in a basement membrane extract (BME) under WNT-activating conditions using described STAR Methods . 22 Efficient establishment of EEO was observed from both genotypes at all ages under these conditions; however, distinct differences in morphology were noted between genotypes ( Figure 1 A). EEO derived from the uteri of WT mice were spherical with a discernible lumen, whereas those generated from Esr1 −/− mice were larger, denser, and lobular ( Figure 1 A). ESR1 was abundant in WT EEO but absent in EEO developed from Esr1 −/− epithelium ( Figure S1 ). The EEO derived from PND 30 uteri of both genotypes could be expanded for 30 passages over a 6-month period ( Figure 1 B). Thus, ESR1 absence does not affect long-term organoid expansion and phenotype based on morphology. Figure 1 Derivation and characterization of Esr1 −/− endometrial epithelial organoids (A) Brightfield images of organoids established from WT and Esr1 −/− females at postnatal days (PND) 15, 30, and 60. (Scale bars, 500 μm). (B) Brightfield images demonstrating long-term expansion potential and gross morphology of WT and Esr1 −/− organoids at passage 30. (Scale bars, 500 μm). (C) Representative images from the organoid formation assay from single cells expanded for 20 days (scale bars, 500 μm). (D and E) Quantification of average organoid diameter (D) and total organoid number (E). (F and G) Immunofluorescent localization of Ki67 (F) and ESR1 (G) (Scale bars, 100 μm). All experiments were performed on passage 3 organoids (n ≥ 3 mice per genotype). Data are presented as mean ± S.E.M. Solid dots represent individual drops in technical replicates. ∗p < 0.05 and ∗∗∗p < 0.001 (Student’s t test). Derivation and characterization of Esr1 −/− endometrial epithelial organoids (A) Brightfield images of organoids established from WT and Esr1 −/− females at postnatal days (PND) 15, 30, and 60. (Scale bars, 500 μm). (B) Brightfield images demonstrating long-term expansion potential and gross morphology of WT and Esr1 −/− organoids at passage 30. (Scale bars, 500 μm). (C) Representative images from the organoid formation assay from single cells expanded for 20 days (scale bars, 500 μm). (D and E) Quantification of average organoid diameter (D) and total organoid number (E). (F and G) Immunofluorescent localization of Ki67 (F) and ESR1 (G) (Scale bars, 100 μm). All experiments were performed on passage 3 organoids (n ≥ 3 mice per genotype). Data are presented as mean ± S.E.M. Solid dots represent individual drops in technical replicates. ∗p < 0.05 and ∗∗∗p < 0.001 (Student’s t test). Passage 2 WT and Esr1 −/− EEO from PND 30 uteri were dissociated into single cells ( Figure S2 ) and 5,000 cells were placed into equal amounts of BME. After 20 days of culture, 12.1% of WT cells, but only 1.2% of Esr1 −/− cells, formed EEO ( Figures 1 C–1E). Although fewer EEO formed from Esr1 −/− uterine epithelial cells ( Figure 1 E), the organoids were larger in diameter ( Figures 1 C and 1D) and contained greater numbers of Ki67-positive proliferative cells (41.9% vs. 26.2%) ( Figure 1 F). Histologically, the WT EEO contained a single layer of columnar-type epithelial cells, whereas the cells of Esr1 −/− EEO were multilayered ( Figures 1 F and 1G). The transcriptome was assessed by RNA-seq using passage 3 EEO established from the uterine epithelium of PND 30 WT and Esr1 −/− mice. As illustrated in Figure 2 A, 2,729 transcripts were classified as differentially expressed genes (DEGs) using specific criteria (Log2 fold change >2, FDR 1) ( Table S1 ). mRNA transcripts of genes were increased (2,016) and decreased (713) in Esr1 −/− as compared to WT EEO ( Figure S3 A). Gene set enrichment analysis of the increased genes revealed biological processes associated with keratinization, keratinocyte differentiation, epidermal cell differentiation, and epithelium development ( Figure 2 A). In contrast, genes with decreased mRNA transcript abundance were enriched for pathways involved in glycoprotein and carbohydrate metabolic processes, protein glycosylation, regulation of cell migration, transmembrane transport, and tissue development. Network analysis determined the relationship between the top biological processes enriched in Esr1 −/− EEO and identified genes belonging to the epidermal differentiation complex (EDC) ( Figure S3 B). The EDC family of genes is required for the terminal differentiation of basal epithelial cells and includes S100 genes, small proline-rich ( Sprr ) genes, late cornified envelope ( Lce ) genes, and filaggrin-like genes ( Flg , Rptn , and Tchh ). 29 , 30 , 31 The mRNA transcripts corresponding to those EDC genes were much more abundant in Esr1 −/− than WT EEO ( Figures S3 C–S3F). Additionally, many of the DEGs with increased mRNA levels regulate keratinocyte differentiation and encode main structural components of the cornified envelope in squamous epithelium ( Bcl11b , Ivl , Lor ) 32 , 33 , 34 ( Figure S3 G). Figure 2 Absence of Esr1 alters EEO gene expression RNA was isolated from passage 3 organoids established from individual PND 30 mice. (A) Heatmap of DEGs and visualization of biological process GO terms associated with genes differentially expressed in Esr1 −/− and WT organoids based on RNA-seq analysis. (B) Heatmap of differential keratin expression in Esr1 −/− organoids. (C) Immunofluorescent localization for basal (KRT5/p63) or luminal (KRT7) cell markers, and for the uterine gland-specific marker FOXA2. (Scale bars, 100 μm). Absence of Esr1 alters EEO gene expression RNA was isolated from passage 3 organoids established from individual PND 30 mice. (A) Heatmap of DEGs and visualization of biological process GO terms associated with genes differentially expressed in Esr1 −/− and WT organoids based on RNA-seq analysis. (B) Heatmap of differential keratin expression in Esr1 −/− organoids. (C) Immunofluorescent localization for basal (KRT5/p63) or luminal (KRT7) cell markers, and for the uterine gland-specific marker FOXA2. (Scale bars, 100 μm). Upstream transcription factor binding analysis of DEGs was performed using ChEA3. 35 The top transcription factor was predicted as ESR1 for the genes with higher expression in WT EEO. Interestingly, the same analysis performed on DEGs that increased in Esr1 −/− EEO uncovered significant enrichment for p63 ( Figure S3 H). p63 is an identity switch for Müllerian duct epithelium to be cervicovaginal versus uterine 9 and directly regulates the expression of basal keratin genes. 36 , 37 Indeed, expression of many basal keratins ( Krt1 , Krt5 , Krt6 , Krt10 , Krt13 , Krt14 , Krt15 , Krt16 , Krt17 , Krt75 , and Krt78) was much more abundant in Esr1 −/− EEO ( Figure 2 B), whereas expression of luminal keratins ( Krt7 and Krt18) 38 was much lower or absent in Esr1 −/− EEO ( Figure 2 B). As illustrated in Figure 2 C, Esr1 −/− EEO possessed high levels of basal KRT5 and p63 proteins, but not luminal KRT7, in their multilayered cells. In contrast, WT EEO contained a single layered epithelium that expressed luminal KRT7 with no evidence of p63 ( Figure 2 C). The Esr1 −/− organoid phenotype resembles the multilayered organoids developed from cervicovaginal tissues, as both were multilayered with similar patterns of p63, Ki67, and KRT5 expression 24 , 27 ( Figures 1 A–1F and 2 C). FOXA2, a transcription factor specific to GE cells of the mouse and human uterus, 3 is also present in all cells of the cervical epithelium. 39 Nuclear FOXA2 was present in all cells of the Esr1 −/− EEO ( Figure 2 C), but only in a few cells in the WT EEO. Western blot analysis of EEO confirmed the increase of FOXA2 in Esr1 −/− EEO ( Figure S1 ). Thus, the gene expression signature of the Esr1 −/− EEO signifies cervicovaginal type of epithelium, which is a multilayered stratified squamous epithelium that contains a mitotically active basal layer and a superficial layer of cornified cells with basal-type keratins. 40 Stromal-derived paracrine factors impact epithelial proliferation, differentiation, and function in the female reproductive tract and other epitheliomesenchymal organs. 6 , 7 , 8 Elegant tissue recombination studies using cells from the uterus and vagina discovered that the mesenchyme/stroma instructs the undetermined multipotent female reproductive tract epithelium to differentiate into uterine (simple columnar) or cervicovaginal (stratified squamous with basal cells) epithelium. 4 , 5 , 10 Thus, an in vitro co-culture system was established to study effects of stromal-derived factors on epithelial cell growth and differentiation into organoids ( Figure 3 A). All experiments used WT stroma, as we were not able to obtain sufficient quantities of Esr1 −/− stroma cells. Uterine epithelia from WT or Esr1 −/− null mice were placed into BME on a transwell cell culture insert that was then placed in a well containing WT stroma cells or no stroma cells as a control ( Figure 3 A). Over a 20 days growth period, the WT epithelial cells formed EEO with normal growth and single layer morphology, and the presence of WT stromal cells had no discernible effects on the morphology of WT EEO ( Figures 3 B and 3C). As expected, Esr1 −/− epithelial cells formed larger, denser, and lobular structures in the absence of stroma ( Figures 3 B and 3C) that expressed cervicovaginal-type epithelium markers ( Trp63 , Krt5 , and Krt14 ) ( Figure 3 D). In contrast, Esr1 −/− epithelial cells cultured with WT stroma formed EEO that exhibited morphology more similar to EEO from WT ( Esr1 +/+ ) uterine epithelium ( Figure 3 B) and lacked expression of cervicovaginal-type epithelium markers ( Trp63 , Krt5 , and Krt14 ) ( Figure 3 D). Thus, the ESR1-expressing WT stroma produces paracrine factors that impact growth and differentiation of Esr1 −/− epithelia. Figure 3 Stromal-epithelial co-culture impacts Esr1 −/− organoid development (A) Experimental design. Epithelial cells were seeded in Cultrex on transwell inserts over a bed of stromal fibroblasts adhered to the bottom of 6 well plates and cultured for 20 days (n ≥ 5 mice per genotype). (B) Representative brightfield images at day 20 of culture (scale bars, 500 μm). (C) Quantification of the number of organoids at day 20 of culture. Solid dots represent average organoid number per biological replicate. (D) RT-qPCR analysis for the basal cell markers Trp63 , Krt5 , and Krt14 (n = 3 mice per genotype/treatment). Data are presented as fold change compared to WT EEO cultured in the absence of stromal fibroblasts. ∗p < 0.05 and ∗∗∗p < 0.001 (Student’s t test). Stromal-epithelial co-culture impacts Esr1 −/− organoid development (A) Experimental design. Epithelial cells were seeded in Cultrex on transwell inserts over a bed of stromal fibroblasts adhered to the bottom of 6 well plates and cultured for 20 days (n ≥ 5 mice per genotype). (B) Representative brightfield images at day 20 of culture (scale bars, 500 μm). (C) Quantification of the number of organoids at day 20 of culture. Solid dots represent average organoid number per biological replicate. (D) RT-qPCR analysis for the basal cell markers Trp63 , Krt5 , and Krt14 (n = 3 mice per genotype/treatment). Data are presented as fold change compared to WT EEO cultured in the absence of stromal fibroblasts. ∗p < 0.05 and ∗∗∗p < 0.001 (Student’s t test). Based on our in vitro 3D organoid observations, studies of uterine epithelium were performed in WT, Esr1 −/− , and Esr1 conditional knockout (cKO) mice ( Figure 4 A). In the Pgr Cre model, 41 the Cre is active by PND 3 in the uterus and deletes floxed genes in both uterine epithelium and stroma ( Figure 4 A). On PND 60, no basal keratins (KRT5 and KRT14) or p63-expressing cells were observed in the uteri of either WT, Esr1 −/− , or Pgr Cre/+ Esr1 f/f females ( Figures 4 B–4D). As expected, endometrial organoids established from Pgr Cre/+ Esr1 f/f mice developed into large, dense, and lobular structures ( Figure 4 A), replicating the phenotype of Esr1 −/− EEO. Figure 4 Epithelial Esr1 deletion results in a stratified uterine epithelial layer (A) Top panel: uteri of PND 60 Esr1 f/f (WT), Esr1 −/− , Pgr Cre/+ Esr1 f/f , and Foxa2 Cre/+ Esr1 f/f mice (scale bars, 1 cm). Middle panel: ESR1 immunolocalization in uterine cross-sections (scale bars, 100 μm). Bottom panel: gross morphology of passage 3 endometrial epithelial organoids (EEO) (scale bars = 500 μm). (B–D) Immunolocalization of the basal markers KRT5 (B), KRT14 (C), and p63 (D) (scale bars, 250 μm). For (B), (C), and (D), panels on the left show uterine cross-sections and dashed panels on the right are insets of luminal and glandular epithelium. White and black arrows indicate the presence of basal cells in the uterine glandular epithelium of Foxa2 Cre/+ Esr1 f/f mice (n ≥ 3 mice/genotype). LE, luminal epithelium; GE, glandular epithelium; S, stroma; M, mesometrial; AM, antimesometrial. Epithelial Esr1 deletion results in a stratified uterine epithelial layer (A) Top panel: uteri of PND 60 Esr1 f/f (WT), Esr1 −/− , Pgr Cre/+ Esr1 f/f , and Foxa2 Cre/+ Esr1 f/f mice (scale bars, 1 cm). Middle panel: ESR1 immunolocalization in uterine cross-sections (scale bars, 100 μm). Bottom panel: gross morphology of passage 3 endometrial epithelial organoids (EEO) (scale bars = 500 μm). (B–D) Immunolocalization of the basal markers KRT5 (B), KRT14 (C), and p63 (D) (scale bars, 250 μm). For (B), (C), and (D), panels on the left show uterine cross-sections and dashed panels on the right are insets of luminal and glandular epithelium. White and black arrows indicate the presence of basal cells in the uterine glandular epithelium of Foxa2 Cre/+ Esr1 f/f mice (n ≥ 3 mice/genotype). LE, luminal epithelium; GE, glandular epithelium; S, stroma; M, mesometrial; AM, antimesometrial. The lack of epithelial proliferation in the Esr1 −/− mouse uterus 11 , 12 may hinder epithelial differentiation toward the basal cell lineage. Moreover, several studies have indicated that epithelial cell proliferation and differentiation are regulated by paracrine signaling from the stroma. 4 , 13 To determine the impact of disrupted stromal-epithelial crosstalk on epithelial differentiation in the absence of epithelial Esr1 , the epithelial-specific Wnt7a Cre 14 and GE-specific Foxa2 Cre 42 mouse models were used to conditionally delete Esr1 in endometrial epithelial cells. Wnt7a Cre is active in the Müllerian duct epithelium of the embryo, whereas Foxa2 Cre should be active only in the differentiating and developing GE. 43 , 44 , 45 , 46 , 47 Indeed, Wnt7a Cre/+ Esr1 f/f deletes Esr1 in only the uterine LE and GE cells but not the stroma or myometrium. 14 As expected, ESR1 was undetectable in the glands of adult Foxa2 Cre/+ Esr1 f/f mice but was observed in the pseudostratified LE and stroma ( Figure 4 A). Foxa2 Cre/+ Esr1 f/f females failed to achieve pregnancies and displayed complete infertility in a 6-month breeding trial ( Table S2 ). On PND 60, the uterus of Foxa2 Cre/+ Esr1 f/f mice contained areas of squamous metaplasia with a multilayered epithelium only in the glands, whereas the glands of the uterus from WT, Esr1 −/− , or Pgr Cre/+ Esr1 f/f females were normal with a simple columnar-type epithelium ( Figures 4 A–4D). The glands of Foxa2 Cre/+ Esr1 f/f uteri exhibiting squamous metaplasia contained KRT5, KRT14, and p63-positive basal cells ( Figures 4 B–4D). Further, EEO generated from epithelial cells isolated from uteri of Foxa2 Cre/+ Esr1 f/f mice displayed a mixture of phenotypes including larger, dense, multilayered, and lobular organoids typical of Esr1 −/− EEO and smaller, spherical, and single-layered EEO typical of WT mice ( Figure 4 A). Adult Wnt7a Cre/+ Esr1 f/f uteri also possessed KRT5-positive cells, but they were present in both the LE and glands ( Figure S4 A). Organoids generated from the uterine epithelial cells of Wnt7a Cre/+ Esr1 f/f mice displayed the same morphology as Pgr Cre/+ Esr1 f/f and Esr1 −/− EEO and were dense, multilayered, and lobular ( Figure S4 C). Areas of the LE and all GE of uteri from Wnt7a Cre/+ Esr1 f/f mice were positive for FOXA2, which is atypical as FOXA2 is not present in the uterine LE of WT mice 46 ( Figure S4 B). This is consistent with previous observations in Wnt7a Cre/+ Esr1 f/f mice where FOXA2 is upregulated in the ESR1-negative uterine luminal and vaginal epithelium. 48 In WT mice, FOXA2 is expressed in the upper layers of the stratified squamous-type epithelium of the cervix 39 and vagina. 48 Collective in vivo and in vitro results support the idea that Esr1 expression in the uterine epithelium has a role in LE lineage specification, homeostasis, and developmental plasticity. Normal epithelial differentiation and gland development is observed in Esr1 −/− mice from birth to PND 30; however, those mice lose glands and have reduced uterine cell proliferation between PNDs 30 and 60 in contrast to continued adenogenesis and proliferation observed in WT uteri. 11 , 12 Indeed, KRT5 and p63-expressing basal cells are not observed in PND 30 Foxa2 Cre/+ Esr1 f/f ( Foxa2 cKO) mice, but are observed at PND 60 ( Figures 4 B–4D, 5 A, and 5B). In mice, ligand-dependent ESR1 signaling is crucial for uterine epithelial proliferation and function after puberty. 11 , 14 , 15 These collective results support the idea that E2 acts via the stroma to alter differentiation of the Esr1 −/− epithelium in the uterus. To address that hypothesis, WT and Foxa2 Cre/+ Esr1 f/f females were ovariectomized (OVX) on PND 30 and treated for 10 days with either a sham implant as a control or E2 ( Figure 5 C). No evidence of epithelial stratification, basal keratins (KRT5, KRT14), or p63-positive cells was observed in the uteri of control or E2-treated WT mice ( Figures 5 D and 5F). In contrast, uteri from E2-treated Foxa2 Cre/+ Esr1 f/f mice displayed KRT5, KRT14, and p63-positive stratified cells in the glands. Thus, E2 causes abnormal epithelial differentiation in ESR1-negative epithelial cells of the uterus via paracrine factors emanating from the ESR1-positive stroma. Figure 5 Basal cell differentiation in the uterine epithelium is mediated by E2 actions on the stroma (A and B) Immunolocalization of p63 (A) and KRT5 (B) in uterine cross-sections of Esr1 f/f (WT) and Foxa2 Cre/+ Esr1 f/f mice. Histological analysis was performed on PND 30 (top panels) and PND 60 (bottom panels) females (scale bars, 100 μm). White and black arrows indicate the presence of basal cells in the uterine glandular epithelium. LE, luminal epithelium; GE, glandular epithelium; S, stroma. (C) Experimental design. Esr1 f/f (WT) and Foxa2 Cre/+ Esr1 f/f females were ovariectomized (OVX) at PND 30 and supplemented with E2 (20 μg/pellet) for 10 days. (D–F) Immunolocalization of the basal markers KRT5 (D), KRT14 (E), and p63 (F). White and black arrows indicate the presence of basal cells in the uterine glandular epithelium (scale bars, 100 μm). (G) Experimental design for the co-culture experiment. Stromal cells isolated WT females were co-cultured with organoids established from individual WT or Esr1 −/− mice (n = 3 females per genotype). Co-cultures were treated with vehicle (100% Ethanol) or 100 nM E2 for 20 days. (H) RT-qPCR for the basal cell markers Trp63 , Krt5 , and Krt14 . Gene expression data were normalized to Actb and Ppia mRNA and are presented as mean relative expression in comparison to the value of an appropriate control sample using the 2 −ΔCt method ±S.E.M. ∗, p < 0.05 (Student’s t test) (n ≥ 3 mice per genotype). Basal cell differentiation in the uterine epithelium is mediated by E2 actions on the stroma (A and B) Immunolocalization of p63 (A) and KRT5 (B) in uterine cross-sections of Esr1 f/f (WT) and Foxa2 Cre/+ Esr1 f/f mice. Histological analysis was performed on PND 30 (top panels) and PND 60 (bottom panels) females (scale bars, 100 μm). White and black arrows indicate the presence of basal cells in the uterine glandular epithelium. LE, luminal epithelium; GE, glandular epithelium; S, stroma. (C) Experimental design. Esr1 f/f (WT) and Foxa2 Cre/+ Esr1 f/f females were ovariectomized (OVX) at PND 30 and supplemented with E2 (20 μg/pellet) for 10 days. (D–F) Immunolocalization of the basal markers KRT5 (D), KRT14 (E), and p63 (F). White and black arrows indicate the presence of basal cells in the uterine glandular epithelium (scale bars, 100 μm). (G) Experimental design for the co-culture experiment. Stromal cells isolated WT females were co-cultured with organoids established from individual WT or Esr1 −/− mice (n = 3 females per genotype). Co-cultures were treated with vehicle (100% Ethanol) or 100 nM E2 for 20 days. (H) RT-qPCR for the basal cell markers Trp63 , Krt5 , and Krt14 . Gene expression data were normalized to Actb and Ppia mRNA and are presented as mean relative expression in comparison to the value of an appropriate control sample using the 2 −ΔCt method ±S.E.M. ∗, p < 0.05 (Student’s t test) (n ≥ 3 mice per genotype). In vitro co-culture studies were then conducted to determine E2 effects on ESR1-negative uterine epithelial cell differentiation ( Figure 5 G). Epithelial cells were isolated from either WT or Esr1 −/− uteri and placed in BME on cell culture inserts placed into wells containing stromal cells from WT uteri ( Figure 5 G). Organoid-stroma co-cultures were treated with 100 nM E2 or vehicle as a control for 20 days. Similar to results presented in Figure 3 , both WT and Esr1 −/− epithelial cells co-cultured with WT stroma formed EEO with single-layered cell morphology ( Figures S5 A and S5B). Treatment of co-cultures with E2 increased expression of basal cell markers ( Trp63 , Krt5 , and Krt14 ) in Esr1 -/- EEO but not WT EEO ( Figure 5 H). These results support the idea that E2 promotes basal cell differentiation and stratification of ESR1-negative uterine epithelial cells via paracrine factors produced by WT stromal cells.

These studies provide novel insights into the developmental plasticity of epithelia in the developing postnatal uterus and essential role of Esr1 in regulating epithelial cell differentiation, maintaining luminal epithelial cell identity, and restricting cellular plasticity of specified uterine LE and GE cells. These findings also reinforce the importance of paracrine factors from the stroma in directing development, differentiation, and homeostasis of the epithelium in the uterus. These findings have important implications for how estrogens, ESR1 agonists, and ESR1 antagonists impact uterine function and the etiology of common endometrial-based diseases including infertility, uterine squamous metaplasia, adenomyosis, adenocarcinoma, and endometriosis. Previous studies found that knockouts of Esr1 , both global and cell type specific, impact postnatal uterine morphogenesis and growth as well as adult function through cell-specific actions. Uterine epithelial-specific conditional knockout mice defined differential cell type-specific roles for ESR1 and its contribution to epithelial function. 14 , 15 While Esr1 is not necessary for uterine development, growth, and adenogenesis during the first month of life, maintenance of glands and cell proliferation is compromised in the uterus of global Esr1 −/− mice during the second month after puberty. 11 , 12 The findings of the presented studies strongly support the idea that Esr1 has a critical and unexplored role in uterine epithelial homeostasis and plasticity based on in vivo studies of the uterus and in vitro studies of organoids. Organoids are biomimetic 3D mini-organs that can be developed from single cells or groups of cells in vitro using culture conditions involving BME and growth factors. Endometrial epithelial-based organoids are long-term expandable and retain properties of the original tissue while remaining genomically stable, which makes them useful tools to study mechanisms underlying epithelial development and function. 49 Mouse and human EEO established in BME and cultured under WNT-activating conditions form spheroids with a single layer of columnar-type epithelium and a central lumen. Human EEO contain both major LE and GE cell types as well as proliferative, stem, and secretory cell subtypes that are typical of endometrial epithelium of the cycling uterus. 22 , 50 In the present studies, epithelial cells from WT mice also formed spheroids with a single layer of epithelial cells that closely resembles in vivo epithelium of the uterus with both GE (FOXA2 positive, ESR1 positive) and LE cells (FOXA2 negative, ESR1 positive) as well as proliferating cell subtypes. In contrast, EEO established from uterine epithelial cells of Esr1 −/− mice more closely resembled those of the stratified squamous-type epithelium of the ectocervix or vagina 24 with basal cells (p63, KRT5, and Ki67), parabasal cells (p63, KRT5), and terminally differentiated basal cells ( Krt5 , Flg , Lor , and Bc11b ). Indeed, organoids derived from the epithelia of the mouse cervix and vagina also form stratified multilayered structures with basal and proliferative cells. 24 , 27 Transcriptome analysis of Esr1 −/− EEO revealed a significant upregulation of genes associated with basal epithelium differentiation, particularly those involved in keratinization and epidermal development. Notably, the expression of several epidermal differentiation complex (EDC) gene families, including the small proline-rich ( Sprr ), late cornified envelope ( Lce ), S100 fused-type protein, and S100 genes, as well as the main structural components of the cornified envelope in squamous epithelium, involucrin (Ivl) and loricrin (Lor) , was markedly increased in Esr1 −/− EEO. 29 , 30 These genes encode proteins that participate in the cornification of stratified squamous epithelium present in the cervix and vagina 32 , 33 and present in cervical organoids. 24 Enrichment of binding sites for p63, a known identity switch for Müllerian duct epithelium to be cervicovaginal versus uterine, was present in DEGs of Esr1 −/− EEO. 9 , 36 , 37 The shift to a basal cell phenotype was confirmed by the loss of luminal keratins ( Krt7 and Krt18 ) and gain of basal keratins ( Krt1 , Krt5 , Krt6 , Krt10 , Krt13 , Krt14 , Krt15 , Krt16 , Krt17 , Krt75 , and Krt78 ) in Esr1 −/− EEO. A recent study in breast cancer epithelial cells found a strong negative correlation between basal keratin expression and Esr1 levels, 51 in which the loss of ESR1-bound enhancers triggers upregulation of basal keratins. 52 Indeed, Krt5 , Krt6 , Krt16 , and Krt17 are overexpressed in Esr1 mutant cells that undergo rapid clonal expansion and acquire stem-like properties in tumors. 53 Thus, ESR1 has a regulatory role in uterine epithelial lineage determination and homeostasis. Cellular plasticity is a phenomenon in which cells change their identity or phenotype outside of conventional cell lineage determination pathways or tissue homeostasis and has been extensively documented in cancer. 34 , 54 Notably, cancer cells undergo changes in cell plasticity involving basal and luminal epithelial cells originating from shared bipotential progenitors, resulting in highly proliferative and invasive tumor phenotypes. 52 , 55 The multipotent epithelial cells lining the Müllerian duct and newborn female reproductive tract can differentiate into either uterine luminal-type or cervicovaginal basal-type epithelium. 1 , 2 , 4 The molecular mechanisms responsible for the loss of cellular identity through epithelial plasticity remain unclear; 34 however, recent studies have shown that the deletion of the tumor suppressor gene Arid1a induces basal differentiation and tumorigenesis in breast cancer cells by downregulating ESR1 target genes. 56 , 57 Similarly, disruptions in the interaction between high-order assemblies of transcription factors and ESR1-bound distal enhancers have been linked to the endocrine resistance-mediated upregulation of basal invasive markers in breast cancers following treatment with Esr1 antagonists. 52 These findings underscore the intrinsic role of ESR1 in regulating luminal epithelial lineage determination and homeostasis in the uterus. However, additional studies will be required to fully understand the cell-intrinsic mechanisms of ESR1-mediated uterine epithelial differentiation. In contrast to global Esr1 null mice, epithelial-specific ablation of Esr1 resulted in the appearance of basal cells and squamous metaplasia in the glands or necks of glands near the lumen of the adult uterus. However, basal cell differentiation was not detected in the epithelium of epithelial-specific Esr1 -ablated mice on PND 30. 11 , 12 Thus, ESR1 has a biological role in the maintenance of the differentiated epithelial state in adult uteri. The emergence of basal cells in the uterine epithelium of the epithelial-specific Esr1 cKO mice indicates that E2 acts via WT stroma to produce paracrine factors that alter differentiation of the ESR1-negative glandular epithelium and LE cells near the glands. Furthermore, these data suggest that the absence of stratification and basal epithelial signatures in the uterus of Esr1 −/− and Pgr Cre/+ Esr1 f/f females is the result of the lack of E2/ESR1 signaling in the stroma. The influence of the uterine mesenchyme/stroma on epithelial lineage determination and development has been long recognized. During neonatal development, the mesenchyme surrounding the Müllerian duct epithelium secretes paracrine factors (BMP4, ACTA, and FGF7/10) that activate p63 expression by mediating signaling pathways (SMAD4, RUNX1, and MAPK) in the epithelium. 5 , 6 , 7 , 8 Similarly, in the transition zone between the endo- and ectocervix of the adult mouse reproductive tract, the stroma drives the differential proliferation of specific epithelial cell lineages. 24 The stromal-epithelial organoid co-culture system employed here supports a requirement for factors produced by the endometrial stroma governing epithelial differentiation, homeostasis, and plasticity of Esr1 −/− epithelial cells. Studies in breast cancer have provided evidence for a causal link between stromal signaling and the regulation of basal keratins ( Krt14 , Krt16 , and Krt17 ) in Esr1 mutant epithelium. 51 Here, E2 treatment of OVX mice resulted in the development of basal cells specifically in areas of Esr1 deletion in the epithelium, whereas there was no evidence of basal cells in WT mice treated with E2. Similarly, treatment of co-cultures with E2 increased the expression of basal cell marker genes only in EEO lacking ESR1. These collective findings indicate that ESR1 has rather unexplored but important biological roles in mediating epithelial differentiation and cell lineage specification in response to paracrine factors produced by the stroma in response to E2. In summary, our findings reveal that ESR1 regulates a gene regulatory network critical to maintain specified epithelial cell fate and plasticity in the uterus, as the absence of epithelial ESR1 leads to basal cell differentiation and squamous metaplasia in the adult uterus via paracrine signals produced by ESR1-positive stromal cells. Future studies should focus on unraveling the complex biology and nature of ESR1, E2, and stroma signals in the regulation of cellular plasticity and epithelial differentiation and homeostasis in the uterus. Those studies are important, because loss of ESR1 in the epithelium is linked to tumorigenesis, basal cell differentiation, and acquired endocrine resistance in several types of cancers. 58 , 59 , 60 , 61 Indeed, proper Esr1 expression in the epithelium is likely critical to maintaining determined epithelial cell fate and homeostasis to prevent the development of endometrial hyperplasia and adenocarcinoma 34 , 52 , . 58 , 59 , 60 , 61 While these studies provide a thorough characterization of ESR1’s role in uterine epithelial development, there are certain limitations to consider. In vitro studies utilizing EEO from global Esr1 -null mice may not fully capture complex in vivo cell-cell interactions that take place in the tissue microenvironment of the developing uterus. Additionally, only WT stroma was evaluated for the co-culture system due to limitations in consistently obtaining confluent Esr1 −/− stroma. Consequently, the study could not directly assess ESR1-independent E2 effects on stromal-epithelial interaction. Such interactions could provide additional information regarding the role of stromal-derived factors on basal cell differentiation in the absence of epithelial ESR1. Finally, the study did not use a human endometrial epithelial model to investigate whether the signaling identified in the mouse is conserved in humans.

The female reproductive tract develops from the Müllerian duct and contains different types of epithelia in the oviduct, uterus, cervix, and vagina. 1 In mice, development of the uterus is only completed after birth. 1 , 2 During the first week, the multipotent epithelium of the uterus becomes specified into a simple columnar type of luminal epithelium (LE), and the mesenchyme stratifies into stroma and myometrium. During the second week, the glandular epithelial (GE) cells arise from LE stem progenitor cells and tubular gland morphogenesis is initiated. 3 Tissue recombination studies established the importance of mesenchyme/stroma to instruct epithelial differentiation in the female reproductive tract. 4 , 5 Specifically, paracrine factors from mesenchymal cells in the caudal region of the Müllerian duct trigger expression of the transcription factor p63 ( Trp63 ) in the epithelium, 6 , 7 , 8 leading to the differentiation of the cervix and vagina with a multilayered stratified type of epithelium containing basal cells. 9 Of note, the simple columnar-type luminal epithelium of the uterus does not express Trp63 . Trp63 null mice lack basal cells in the cervicovaginal epithelium, supporting the idea that p63 is a master regulator of epithelial cell lineage specification in the Müllerian duct. 10 The role of estrogen receptor alpha ( Esr1 ) in postnatal epithelial differentiation and homeostasis in the uterus is not fully understood. Esr1 is expressed in all cells of the myometrium, stroma, and epithelium of the adult uterus. 11 , 12 Although Esr1 is dispensable for embryonic Müllerian duct patterning and differentiation into the uterus, 12 , 13 it is essential for postnatal uterine gland maintenance and control of epithelial apoptosis in the adult uterus. 11 , 14 , 15 Interestingly, Esr1 is not expressed in the developing epithelium of the uterus during the first week after birth. 16 However, neonatal exposure to the diethylstilbestrol, a potent ESR1 agonist, altered differentiation of the uterus and induced squamous metaplasia in a p63-dependent manner. 10 Intriguingly, Trp63 expression is augmented and expanded in cervicovaginal epithelium in the absence of Esr1 . 17 While these data support the idea that Esr1 may regulate postnatal uterine epithelial development, further interrogation of the impact of Esr1 is required to explain the previous findings. Organoid culture is a valuable model system for investigating the cell autonomous regulation of epithelial differentiation and physiology. 18 , 19 Organoids derived from different areas of the mouse reproductive tract have been established, including the oviduct, 20 , 21 uterus, 22 , 23 cervix, 24 , 25 and vagina. 26 , 27 These organoids maintain phenotypic and morphological features of their organs of origin. 20 , 23 , 24 , 27 Characterization and analysis of organoid formation in these models enable the identification of biomarkers and mutations associated with organ development and function, 23 , 28 which can be leveraged to uncover the mechanisms underlying normal and abnormal epithelial development in the female reproductive tract. Here, a comprehensive set of studies involving endometrial epithelial organoids (EEO), co-culture of uterine stromal and epithelial cells, and mouse genetic models of Esr1 deletion were employed to understand the biological role of ESR1 in uterine epithelium development. The findings reveal that loss of Esr1 in the epithelium of the uterus results in the activation of a basal differentiation program that is mediated by E2-dependent paracrine factors from the stroma.

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Mouse monoclonal anti-CDH1 (IF; 1:500) BD Biosciences Cat# 610182; RRID: AB_397581 Rabbit polyclonal anti-ESR1 (IF, WB; 1:2000, 1:1000) Abcam Cat# ab3575; RRID: AB_303921 Rabbit monoclonal anti-FOXA2 (IF, WB; 1:500, 1:1000) Abcam Cat# ab108422, RRID: AB_11157157 Rabbit polyclonal anti-KI67 (IF, 1:500) Abcam Cat# ab15580; RRID: AB_443209 Rabbit monoclonal anti-KRT5 (IF, 1:500) Abcam Cat# ab52635; RRID: AB_869890 Mouse monoclonal ani-KRT7 (IF, 1:500) Santa Cruz Cat# sc-23876; RRID: AB_2265604 Rabbit polyclonal anti-KRT14 (IF, 1:500) LS Bio Cat# ls-B3916; RRID: AB_10662336 Rabbit monoclonal anti-p63 (IF, 1:500) Cell Signaling Cat# 39692; RRID: AB_2799159 Rabbit polyclonal anti-p63 (WB, 1:1000) Abcam Cat# ab53039; RRID: AB_881860 Rabbit polyclonal anti-TUBA1B (WB, 1:10000) Proteintech Cat# 11224-1-AP; RRID: AB_2210206 Biotinylated goat anti-rabbit IgG (IHC, 1:1000) Vector Laboratories Cat# ba-1000; RRID: AB_2313606 Streptavidin-HRP (IHC, 1:1000) Invitrogen Cat# 434323; RRID: AB_2619743 HRP-conjugated goat anti-rabbit IgG (WB, 1:5000) Fisher Cat# 31460; RRID: AB_228341 Alexa Fluor 488, goat anti-rat IgG (IF 1:500) Jackson ImmunoResearch Cat# 112-545-143; RRID: AB_2338361 Alexa Fluor 647, goat anti-rabbit IgG (IF 1:500) Jackson ImmunoResearch Cat# 111-605-144; RRID: AB_2338078 Chemicals, peptides, and recombinant proteins Advanced DMEM/F12 Gibco Cat# 12634010 B27 Supplement Gibco Cat# 12587010 Insulin-transferrin-selenium Gibco Cat# 41400045 Primocin InvivoGen Cat# Ant-pm Glutamax Gibco Cat# 35050061 A83-01 Gibco Cat# 21041025 Murine EGF BioGems Cat# 9094360 Murine FGF-10 Peprotech Cat# AF-315-09 Murine R-spondin1 Peprotech Cat# 450-61 Murine Noggin Peprotech Cat# 315-32 Murine Wnt3a Peprotech Cat# AF-250-38 Nicotinamide Peprotech Cat# AF-315-20 N2 BioGems Cat# 9899208 DMEM/F12, no phenol red Life Technologies Cat# 17502048 Trypsin Sigma Cat# T4799 Hank’s Balanced Salt Solution Gibco Cat# 14175 Soybean trypsin inhibitor Gibco Cat# 17075029 DNAse I Roche Cat# 10104159001 Cultrex BME R&D Systems Cat# 3445-005-01 Dimethyl sulfoxide Fisher Cat# D2650 Fetal bovine serum Sigma Cat# D1391 Collagenase V Sigma Cat# C9263 Easystrainer (70 μm) Greiner Bio-One Cat# 542070 Culture flask (25 cm 2 ) Greiner Bio-One Cat# 690195 DMEM/F12 Gibco Cat# 11320033 Antibiotic-Antimycotic (100X) Gibco Cat# 15240062 TrypLE, no phenol red Gibco Cat# 12563011 Flowmi cell strainers (40 μm) Sigma Cat# BAH136800040 Trypsin-EDTA (0.25%) ThermoFisher Cat# 25200056 Transwell inserts (0.4 μm PET membrane) Corning Cat# 353090 17β-estradiol Sigma Cat# E1024 Paraformaldehyde (96%) Fisher Cat# AC416780030 EM grade paraformaldehyde (16%) Electron Microscopy Science Cat# 15710 Agarose Fisher Cat# BP1356 Sodium citrate dihydrate Fisher Cat# BP327 Hydrogen Peroxide, 30% Fisher Cat# H325-500 Normal goat serum Invitrogen Cat# 01-6201 Permount Fisher Cat# SP15-500 Hematoxylin solution, Harris modified Sigma Cat# HHS32 Methanol Fisher Cat# A452-4 Phosphatase Inhibitor Cocktail 3 Sigma Cat# P0044 Phosphatase Inhibitor Cocktail 2 Sigma Cat# P5726 Protease Inhibitor Cocktail Sigma Cat# 11697498001 TRIzol Reagent Invitrogen Cat# 15596026 Critical commercial assays DAB ImmPACT substrate kit Vector Laboratories Cat# SK-4100 Hoechst 33342 Invitrogen Cat# H3570 ProLong™ Diamond Antifade Mountant Invitrogen Cat# 36961 Bradford reagent Bio-Rad Cat# 50000006 Mini-PROTEAN TGX stain-free gels (4-20%) Bio-Rad Cat# 4568096 SuperSignal™ West Pico PLUS Thermo Scientific Cat# 34580 Direct-zol RNA Miniprep Plus Kits Zymo Research Cat# R2070 RNase-Free DNase I Set Qiagen Cat# 79254 iScript™ cDNA Synthesis Kit Bio-Rad Cat# 1708890 SsoAdvanced Universal SYBR Green Supermix Bio-Rad Cat# 1725274 Deposited data Transcriptome data from this study Gene Expression Omnibus GEO: GSE232655 Experimental models: Organisms/strains Mice: Esr1 -/- Kindly provided by Dr. Dennis Lubahn (University of Missouri, Columbia, MO) N/A Mice: Pgr Cre Kindly provided by Dr. Francesco Demayo (NIEHS, Durham, NC) and Dr. John Lydon (Baylor College of Medicine, Houston, TX). N/A Mice: Foxa2 Cre Kindly provided by Dr. Heiko Lickert (Helmholtz Zentrum München, Institute of Stem Cell Research, Neuherberg, Germany) N/A Mice: Wnt7a Cre The Jackson Laboratory JAX stock #036637 Mice: Esr1 f/f The Jackson Laboratory JAX stock #032173 Oligonucleotides Primers for Realtime PCR, see Table S4 This paper N/A Software and algorithms ImageJ Schneider et al., 2012 62 https://imagej.nih.gov/ij/ GraphPad Prism v 9.5.1 GraphPad Software www.graphpad.com FastQC v0.11.7. The Babraham Institute https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ Trim Galore! V 0.6.6 The Babraham Institute https://github.com/FelixKrueger/TrimGalore STAR v 2.7.10b Dobin et al., 2013 63 https://github.com/alexdobin/STAR featureCounts v 2.0.3 Liao et al., 2014 64 https://subread.sourceforge.net/ edgeR robust Robinson et al., 2010 65 https://support.bioconductor.org/p/79149/ ShinyGO 0.77 Ge et al., 2020 66 http://bioinformatics.sdstate.edu/go/ ChEA3 Keenan et al., 2019 35 https://maayanlab.cloud/chea3/ Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Andrew M. Kelleher ( [email protected] ). This study did not generate new unique reagents. Floxed Esr1 mice 67 were crossed with Pgr Cre , 41 Foxa2 Cre , 42 or Wnt7a Cre 14 mice to generate conditional knockout animals. Pgr Cre mice 41 were provided by Dr. Francesco Demayo (National Institute of Environmental Health Sciences, Durham, North Carolina) and Dr. John Lydon (Baylor College of Medicine, Houston, Texas). The Foxa2 Cre mice 42 were provided by Dr. Heiko Lickert (Helmholtz Zentrum München, Institute of Stem Cell Research, Neuherberg, Germany). Esr1 -/- mice 12 were provided by Dr. Dennis Lubahn (University of Missouri, Columbia, Missouri). Wnt7a Cre 14 and Floxed Esr1 ( Esr1 f/f ) mice were obtained from The Jackson Laboratory (Jax stock #036637 and #032173, respectively). Day of birth was considered postnatal day (PND) 0 in all studies. All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Missouri, Columbia and were conducted according to the NIH Guide for the Care and Use of Laboratory Animals. Uteri were removed from mice and digested in 1% trypsin (Sigma, T4799) in calcium and magnesium-free HBSS (Gibco, 14175-095) for 45 minutes at 4°C followed by 45 minutes at 37°C using an orbital shaker. The digestion was stopped with 1% soybean trypsin inhibitor (Gibco, 17075-029) in HBSS with 5 mM MgCl2 and 0.1 mg/mL DNAse-1 (Roche, 10104159001). For PND 15, Pasteur pipettes were carefully forged to create a slightly smaller diameter than the uterus, and sheets of epithelium retrieved by gentle suction using a mouth pipette. For PNDs 30 and 60, the digested uterine horns were gently squeezed with forceps. The epithelial tube sheets from each uterus were rinsed gently with HBSS and placed into 1.5 mL Eppendorf tubes. After centrifugation (300 x g for 3 minutes), the supernatant was removed, and epithelium pellet resuspended in base organoid media ( Table S3 ). Following another round of centrifugation, the epithelium pellet was resuspended in 80% Cultrex (R&D Systems, 3445-005-01) and 20% organoid expansion media ( Table S3 ). EEO were passaged (1:3 ratio) every 7-10 days. To passage the EEO, Cultrex/EEO drops were detached from the cell culture plate by gentle pipetting and transferred into 1.5 mL Eppendorf tubes. After centrifugation (300 x g for 3 minutes), the supernatant was removed, replaced by 1.5 mL of HBSS, and pipetted up and down to dissociate the pellet. Following an additional centrifugation step, the cell pellet was resuspended in Cultrex and plated in 12-well plates (25 μL drops; 4 drops per well). Cultrex drops were incubated at 37°C for 15 min prior to adding 800 μL of organoid expansion media per well. Organoid cryopreservation was performed with freezing medium consisting of 10% DMSO in fetal bovine serum (FBS, Sigma, F0926) as described previously. 50 All experiments were performed using passage 3 EEO. Brightfield images were captured on a Leica DMi8 inverted microscope and Leica K8 camera using Leica Application Suite X (LAS X). Stromal cells were isolated from the uterus of adult WT females. First, the epithelium was removed from the uterus of 4-5 females as described previously. Uteri were then pooled, minced with scissors, transferred into a 15 mL Falcon tube, and digested in 10 mL of HBSS containing 0.1 mg/mL collagenase V (Sigma, C-9263) and 0.1 mg/mL DNAse-1 at 37°C for 30 minutes on an orbital shaker (80 RPM). The digested tissue was then filtered through a 70 mm nylon filter (Greiner Bio-One, 542070), and the filtrate containing stromal cells transferred to 25 cm 2 flasks (Greiner Bio-One, 690195). Stromal cells were cultured in DMEM/F12 (Gibco, 11320-033) with 10% FBS and 1% antibiotic-antimycotic (Anti-Anti, Gibco, 15240-062) at 37°C in a humidified 5% CO 2 environment. After 6 h, the media was changed to remove unattached and dead cells. After reaching 80% confluency, cells were utilized for co-culture experiments. For fertility studies, individual adult (6-8 weeks of age) Esr1 f/f (WT) and Foxa2 Cre/+ Esr1 f/f females were placed with a CD-1 male mouse of proven fertility for six months, and the number of litters and pups born during that period were recorded. For the E2 treatment study, WT and Foxa2 Cre/+ Esr1 f/f females were ovariectomized (OVX) at PND 30 and implanted with a sham implant as a control or a pellet containing 20 μg 17β-estradiol. 68 All mice were collected on PND 40 for analysis. Uterine gross morphology was recorded, and tissues were fixed in 4% paraformaldehyde for subsequent histological analysis. Passage 2 EEO embedded in Cultrex were transferred into a 5 mL Eppendorf tube. The EEO suspension was centrifuged (300 x g for 3 minutes at 4°C), the supernatant was removed, and the pellet was resuspended in 5 mL HBSS. Following an additional centrifugation step, the supernatant was removed, and the pellet was resuspended in 1 mL TrypLE (Gibco, 12563011) in a 1.5 mL Eppendorf tube. The cells were incubated in TrypLE for 10 minutes at 37°C on an orbital shaker. Following digestion, the cells were centrifuged at 300 x g for 3 minutes at 4°C, the supernatant was removed, and the pellet was resuspended in 500 mL base organoid media via gentle pipetting to create a single cell suspension. The cell resuspension was passed through a 40 mm cell strainer (Flowmi, Sigma, BAH-136800040), and filtered cells pelleted by centrifugation at 300 x g for 3 minutes at 4°C. Cell number and viability were assessed with a Countess™ II FL Automated Cell Counter (Invitrogen, AMQAX2000) prior to plating to ensure homogenous single-cell suspensions across samples ( Figure S2 ). All organoid formation assays were conducted using at least three biological (mice) and three technical (wells) replicates (5000 cells/25 μL drops; 2 drops per well). Organoid formation efficiency and diameter were analyzed using ImageJ, 62 and the data are presented as the mean ± SEM. Statistical differences between two groups were determined with Student's t-test (GraphPad Prism 9), and statistical significance was defined as p  < 0.05. All co-culture experiments were performed using stromal cells from WT uteri, and EEO established from epithelium isolated from individual PND 30 uteri. For stroma seeding, stromal fibroblasts were detached from flasks using 0.25% trypsin (Gibco, 25200056), washed with HBSS, recovered by centrifugation, resuspended in 1.5 mL stromal cell medium (DMEM/F12 with 10% FBS and 1% Anti-Anti), and plated into 6-well plates (150K cells per well). For organoid establishment, uterine epithelial cells were isolated as described above. Briefly, after the second round of centrifugation, cell pellets were resuspended in 200 mL of 80% Cultrex: 20 % expansion media mix and divided into two 6-transwell inserts (0.4 μm pores, Corning, 353090), each containing 4 drops of 25 mL Cultrex mix. Inserts/Cultrex drops were then incubated at 37°C for 15 min. Next, the transwell inserts/Cultrex drops were placed into 6-well plates that either contained uterine stromal cells or did not contain any cells. For E2 treatment, vehicle (100% ethanol) or 100 nM E2 (17β-estradiol, Sigma, E1024; diluted in 100% ethanol) were added to phenol red-free organoid expansion media ( Table S3 ) prior to culture. A total of 3 mL of media was added to the co-culture [1.5 mL per well (bottom) and 1.5 mL per insert (top)]. Fresh expansion media supplemented with vehicle or E2 was added to each well and insert every 72 h. New primary uterine stromal fibroblasts were isolated and seeded every 5 days, and EEO were passage into a new insert every 6-8 days. Co-cultures were documented every 5 days by brightfield microscopy. Following 20 days in culture, EEO were collected for RNA analysis. EEO derived from 3-7 females per genotype (WT or Esr1 -/- ) were used for each co-culture experiment. Images were taken with a Leica DMi8 inverted microscope and Leica K8 camera using Leica Application Suite X (LAS X). EEO number and diameter were analyzed using ImageJ and the data are presented as the mean ± SEM. Statistical differences between two groups were determined with Student’s t-test (GraphPad Prism 9) and statistical significance was defined as p  < 0.05. At least four Cultrex drops were examined per mouse, timepoint, and genotype. EEOs cultured in 12-well plates were washed with HBSS, fixed using warm 4% EM grade paraformaldehyde (Electron Microscopy Sciences, 15710) for 15 min at room temperature, stained with hematoxylin for 10 min, and washed with HBSS twice to remove excess staining. EEOs were then embedded using 1.5 mL of 2% agar (Fisher, BP1356) in HBSS per well and incubated overnight at 4°C with a wet pad to let blocks solidify. EEO blocks were carefully removed from the wells and placed in tissue cassettes. For tissue processing, uteri from 3-6 females were collected per time point, genotype, and treatment. Uteri were fixed in 4% paraformaldehyde in PBS overnight. EEO blocks and fixed uteri were dehydrated in ethanol, embedded in paraffin wax, and sectioned (7 μm). Sections were mounted on slides, baked for 30 min at 60°C, deparaffinized in xylene, and rehydrated in a graded alcohol series. Deparaffinized sections were subjected to antigen retrieval by incubating sections in 10 mM citrate buffer (pH 6.0) at 95°C for 15 min, followed by cooling to room temperature. For peroxidase-based staining, sections were incubated with 5% H 2 O 2 diluted in methanol for 12 minutes. All slides were blocked with 2% (v/v) normal goat serum (Invitrogen, 01-6201) in PBS at room temperature for 1 h and incubated with primary antibodies (see key resources table ) overnight at 4°C in 1% BSA diluted in PBS. For peroxidase-based staining, the slides were washed in PBS and incubated with biotinylated secondary goat anti-rabbit antibody (1:1000 dilution; Vector Labs, Catalog # BA-1000) for 1 h at room temperature, followed by incubation for 45 min with Streptavidin HRP diluted in PBS (1:1000 dilution; Invitrogen, 434323). Signal was developed using Vector Labs DAB ImmPACT staining according to the manufacturer's instructions (Vector Labs, SK-4100). Sections were lightly counterstained with hematoxylin before affixing coverslips with Permount (Fisher, SP15-500). Immunofluorescence visualization was performed with Alexa 488 or Alexa 647-conjugated secondary antibodies (1:500 dilution; Jackson ImmunoResearch, #112-545-143, #111-605-144). Sections were counterstained with Hoechst 33342 (2 μg/mL; Invitrogen, H3570) before affixing coverslips with ProLong™ Diamond Antifade Mountant (Invitrogen, 36961). Images were taken with a Leica DM6 B upright microscope and Leica K8 camera using Leica Application Suite X (LAS X). EEOs from 3 wells (12 Cultrex drops) were harvested and the Cultrex was removed. The resulting cell pellet was lysed in Cell Lysis Buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) containing: 1x phosphatase inhibitor cocktail (Sigma-Aldrich, P0044, P5726) and 1x complete protease inhibitor cocktail (Roche, 11697498001). The lysates were incubated at 4°C for 30 minutes on an orbital shaker and clarified by centrifugation at 21,000 RCF in a microfuge for 15 minutes at 4°C. Protein concentrations in the lysates were determined using Bradford reagent (Bio-Rad, 50000006). The lysates were run on stain free 4-20% Mini-PROTEAN TGX Precast gels (Bio-Rad, 456-8096) and transferred to a nitrocellulose membrane. The membranes were blocked with 5% nonfat milk in TBST for 1 h at room temperature. After washing, the membranes were incubated with primary antibodies (see key resources table ) in 1% nonfat milk made in TBST overnight at 4°C, followed by anti-rabbit HRP-conjugated IgG (1:5000; Thermo Fisher, 31460) for 1 h at room temperature. Western blot signals were detected using an ECL detection reagent (Thermo Fisher, 34580). Samples were collected at the indicated time points and total RNA was isolated using the Direct-zol kit (Zymo Research, R2070) according to the manufacturer's instructions. To eliminate genomic DNA contamination, RNA was treated with DNase I (Qiagen, 79254) during RNA isolation. The quantity and purity of total RNA were determined using a Nanodrop spectrophotometer (Fisher, 840274200). Total RNA was reverse transcribed to synthesize cDNAs for each sample using an iScript™ cDNA Synthesis Kit (Bio-Rad, 1708890). The cDNA samples were subjected to qPCR using gene-specific primers ( Table S4 ). Briefly, real-time qPCR amplification of cDNAs was carried out in a reaction mixture (10 μL) containing SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, 1725274) and primers (250 nM each). A CFX384 Touch Thermal Cycler (Bio- Rad, 1851138) was employed to perform PCR using a two-step protocol with initial denaturation/enzyme activation at 95°C for 30 s, followed by 40 cycles of denaturation at 95°C for 5 s, and annealing/extension for 30 s at 60°C. Samples were analyzed in duplicates and melting curve analyses were performed to ensure specific amplification of the targeted amplicon. Target gene expression was normalized to the expression of Actb and Ppia mRNA . Relative expression was determined in comparison to the value of an appropriate control sample. All experiments were performed at least three times with independent biological replicates. Statistical differences between two groups were determined using Student’s t-test (GraphPad Prism 9) and statistical significance was defined as p  < 0.05. Biological replicates were generated by pooling total RNA from 3 wells of EEOs per female (n=3 females/genotype/treatment). RNA was isolated using a Direct-zol kit (Zymo Research, R2070) according to the manufacturer’s instructions. The quality and concentration of RNA were determined using a Fragment Analyzer (Agilent, Santa Clara, CA, USA). Libraries were prepared by the University of Missouri DNA core using an Illumina TruSeq mRNA kit (Illumina Inc., San Diego, CA, USA) and sequenced (paired end 150; 30 million read pairs) using an Illumina NovaSeq 6000. Quality of raw data before and after trimming was assessed using FastQC v0.11.7. Adapters and low-quality bases (Phred score < 20) were trimmed from reads using Trim Galore! V0.6.6. The reads were then mapped to the mouse reference genome (GRCm39) using STAR v2.7.10b 63 with transcript annotation index in R v4.2.2. Raw read counts per gene were quantified with featureCounts v2.0.3. 64 Differential expression analysis was conducted using edgeR robust , a method demonstrably robust in conducting exact tests of significance suitable for small counts and limited numbers of biological replicates. 65 Differentially expressed genes (DEG) between WT and KO were determined at Log2 fold change > 2, FDR  1 and carried forward for further analyses. Heatmaps were visualized using the “pheatmap” package in R v4.2.2. Enriched gene ontology biological processes and network analysis were determined with ShinyGO. 66 Transcription factors with binding sites enriched in annotated promoter regions of DEG were determined with ChEA3 using a database of publicly available ChIP-seq experiments derived from literature. 35 All western blots with quantification were performed with at least three independent biological replicates and analyzed using Image Lab Software (Bio-Rad). The data are presented as the mean ± SEM, as determined from at least three independent experiments. Following the Shapiro-WILK test for normality, the data were analyzed using Student's t-test (GraphPad Prism 9). A p-value of less than 0.05 was considered statistically significant. Statistical analyses for the genomic experiments were performed using standard genomic statistical tests as described above.