The endometrium, which is of crucial importance for reproduction, undergoes dynamic cyclic tissue remodeling. Knowledge of its molecular and cellular regulation is poor, primarily owing to a lack of study models. Here, we have established a novel and promising organoid model from both mouse and human endometrium. Dissociated endometrial tissue, embedded in Matrigel under WNT-activating conditions, swiftly formed organoid structures that showed long-term expansion capacity, and reproduced the molecular and histological phenotype of the tissue's epithelium. The supplemented WNT level determined the type of mouse endometrial organoids obtained: high WNT yielded cystic organoids displaying a more differentiated phenotype than the dense organoids obtained in low WNT. The organoids phenocopied physiological responses of endometrial epithelium to hormones, including increased cell proliferation under estrogen and maturation upon progesterone. Moreover, the human endometrial organoids replicated the menstrual cycle under hormonal treatment at both the morpho-histological and molecular levels. Together, we established an organoid culture system for endometrium, reproducing tissue epithelium physiology and allowing long-term expansion. This novel model provides a powerful tool for studying mechanisms underlying the biology as well as the pathology of this key reproductive organ.
INTRODUCTION
The endometrium, which represents the inner lining of the uterus, plays a crucial role in mammalian reproduction (Kobayashi and Behringer, 2003). The murine endometrium is composed of luminal epithelium bordering the uterine cavity, glandular epithelium embedded in stromal tissue and myometrium. The human endometrium consists of a basalis layer in contact with the myometrium and a functionalis layer adjacent to the uterine lumen. Both layers contain glandular epithelium and stromal tissue. The endometrium undergoes remarkable cyclic changes of growth, differentiation and degeneration under the coordinated control of two key ovarian hormones: estrogen and progesterone (Pedram et al., 2014; Roy and Matzuk, 2011). In humans, the functionalis layer strongly expands in response to rising estrogen levels (proliferative phase), then matures with the differentiation of secretory and ciliated cells under the influence of progesterone (secretory phase), to be finally shed following the drop in hormones (menstrual phase). In the non-menstruating mouse, proestrous, estrous, metestrous and diestrous together make up the estrous cycle. The endometrium expands during proestrous and estrous, and then starts to degenerate from metestrous through vacuolar degeneration with a decrease in size and vascularity. Little is understood regarding the molecular and cellular mechanisms underlying this dynamic remodeling activity. Moreover, disruption of this process, as occurs in endometrial atrophy or cancer, has a major impact on fertility and reproduction (Roy and Matzuk, 2011). The main obstacle to studying these processes in detail is the lack of models that are reliable, reproducible and flexible. Immortalized or carcinoma-derived endometrial cell lines do not faithfully mimic the in vivo situation, in particular because of their transformed phenotype. Primary adult endometrial cells prove difficult to maintain in long-term culture, and clinical endometrial biopsies are limited in size and not expandable. Moreover, the cultured cells quickly lose their phenotype and hormone responsiveness (Mannelli et al., 2015). In general, traditional 2D cell cultures do not faithfully represent 3D organ biology and functioning.
Over the past decade, 3D cell arrangements called organoids have been developed from manifold organs as powerful tools for studying tissue biology and disease (Clevers, 2016). Organoids are self-forming 3D reconstructions of an organ's epithelium, typically developing under defined ‘wingless-type MMTV integration site’ (WNT)-activating culture conditions. They reproduce many aspects of the tissue's epithelium histology, functionality and (patho-)biology (Barker et al., 2010; Sato et al., 2009; Clevers, 2016; Gao et al., 2014; Karthaus et al., 2014), thereby superior to the classical 2D in vitro cell cultures. Importantly, organoids show high expansion capacity with retention of phenotypical and functional properties, thus overcoming the limited availability or expandability of primary human tissue (Chua et al., 2014; Yui et al., 2012; Huch et al., 2015).
Organoids from the endometrium have not yet been reported. Here, we describe their establishment from both mouse and human endometrium. The organoids obtained show long-term expandability and reproduce multiple histological and physiological aspects of the endometrial epithelium.
RESULTS
Development of organoids from mouse endometrium
Organoid formation is typically started from tissue fragments, such as crypts from small intestine (Sato et al., 2009). We dissociated mouse endometrium and cultured the glandular-type fragments (Fig. S1A) in conditions previously shown to enable organoid formation from a diversity of tissues (Barker et al., 2010; Karthaus et al., 2014; Kessler et al., 2015; Ren et al., 2014; Sato et al., 2009, 2011). Typically, dissociated tissue is embedded in Matrigel as an extracellular matrix scaffold, and cultured in a cocktail of growth and signaling factors, essentially containing WNT activators like WNT3A; R-spondin 1 (RSPO1), which is the ligand of leucine-rich repeat-containing G-protein-coupled receptor 5 (LGR5) and acts as a WNT signaling amplifier; epithelial-cell mitogens, such as epidermal growth factor (EGF) and fibroblast growth factor 10 (FGF10); inhibitors of bone morphogenetic protein (BMP) such as Noggin, which allow long-term expansion by preventing differentiation; and antagonists of the transforming growth factor β (TGFβ) pathway such as A83-01 Alk inhibitor, which maintain epithelial-cell character. Given the well-known mitogenic effect of insulin on endometrial epithelium (Shiraga et al., 1997), our medium was further supplemented with insulin-transferrin-selenium (ITS). As a source of WNT pathway activation, we used WNT3A- and RSPO1-conditioned media (CM), both added at 25% v/v (Kessler et al., 2015). The final culture medium (for full composition, see the supplementary Materials and Methods) is referred to as 25W/25R (indicating the presence of 25% WNT3A-CM and 25% RSPO1-CM). When cultured under these conditions, endometrial fragments swiftly started to self-organize into organoid-like structures that further expanded in size (Fig. 1A). WNT pathway activation by the conditioned media was validated by expression analysis of WNT target genes (Fig. S1B). Organoids were also obtained when CM were replaced with recombinant WNT3A and RSPO1 (Fig. S1C). After 7-10 days of culture, the developed organoid structures were fragmented into mainly single cells for passaging. Organoids again formed and expanded, at present successfully achieved for more than 14 passages (i.e. more than 5 months) (Fig. 1B). In addition, cryopreserved organoids could be regrown and expanded (data not shown). Time-lapse monitoring showed organoids developing from single cells (Fig. S1D), suggestive of a clonal origin. In support of this, endometrial organoids (EMOs) generated from a 1:1 mixture of fluorescent TdTomato+ and non-fluorescent wild-type cells all looked homogeneously fluorescent or non-fluorescent (Fig. S1D).
Organoids develop from mouse endometrium and reproduce the phenotype of the epithelium. (A) Time-lapse recording of organoid development from dissociated mouse endometrium. Light-microscopic overview pictures are shown at 6 h and 120 h after seeding, with the boxed area followed over time and magnified at the indicated time points. Endometrial fragments start to self-organize into organoid structures within 24-48 h, and then further expand (72-120 h) (arrows). Scale bar: 200 µm. (B) Long-term expansion and passaging capability of mouse EMOs. Organoids that developed from triturated endometrial tissue (P0) were dissociated and re-seeded in consecutive passages in which they reformed and expanded again. Representative bright-field images are shown at the indicated passage numbers. Scale bar: 200 µm. (C) The EMOs reproduce endometrial epithelium phenotype. Immunohistochemical staining for PanCK, E-cadherin and ERα, and PAS staining of mucin (arrows) are shown in mouse endometrium (upper panels) and EMOs (lower panels). Scale bars: 50 µm. (D) TEM reveals the presence of microvilli (arrows) in the EMOs at the luminal side (L), as shown at lower (left; scale bar: 5 µm) and higher magnification (right; scale bar: 2 µm). All these experiments were performed using 25W/25R culture medium (meaning the presence of 25% WNT3A-CM and 25% RSPO1-CM; for full medium composition, see supplementary Materials and Methods).
Organoids develop from mouse endometrium and reproduce the phenotype of the epithelium. (A) Time-lapse recording of organoid development from dissociated mouse endometrium. Light-microscopic overview pictures are shown at 6 h and 120 h after seeding, with the boxed area followed over time and magnified at the indicated time points. Endometrial fragments start to self-organize into organoid structures within 24-48 h, and then further expand (72-120 h) (arrows). Scale bar: 200 µm. (B) Long-term expansion and passaging capability of mouse EMOs. Organoids that developed from triturated endometrial tissue (P0) were dissociated and re-seeded in consecutive passages in which they reformed and expanded again. Representative bright-field images are shown at the indicated passage numbers. Scale bar: 200 µm. (C) The EMOs reproduce endometrial epithelium phenotype. Immunohistochemical staining for PanCK, E-cadherin and ERα, and PAS staining of mucin (arrows) are shown in mouse endometrium (upper panels) and EMOs (lower panels). Scale bars: 50 µm. (D) TEM reveals the presence of microvilli (arrows) in the EMOs at the luminal side (L), as shown at lower (left; scale bar: 5 µm) and higher magnification (right; scale bar: 2 µm). All these experiments were performed using 25W/25R culture medium (meaning the presence of 25% WNT3A-CM and 25% RSPO1-CM; for full medium composition, see supplementary Materials and Methods).
We examined whether the organoid-forming efficiency differed according to the estrous cycle phase (as determined by vaginal smears; Caligioni, 2009; Fig. S1E). The highest number of organoids was obtained from endometrial tissue isolated at estrous (Fig. S1E), which may be due to: cell proliferation peaking at this phase; more efficient recovery of starting material given the enlarged size of the endometrium; and/or a higher proportion or activation status of the organoid-forming cells. To achieve the most efficient organoid growth for our further study, we performed subsequent experiments starting from endometrial tissue of (young) female mice in estrous. Of note, EMOs could not only be developed from female mice in their early reproductive period, but also, and in comparable numbers, from 1-year-old mice that already delivered multiple litters (Fig. S1F).
Mouse EMOs reproduce phenotypical characteristics of endometrial epithelium
Immunohistological analysis of the mouse EMOs revealed a lumen surrounded by an epithelial (pancytokeratin/PanCK+ and E-cadherin+) layer, thereby mimicking the glandular structures in the mouse endometrium in situ (Fig. 1C). In addition, several markers of mouse endometrial epithelium (such as Cxcl15, Esr1, Foxa2, Muc1 and Sprr2f) (Contreras et al., 2010; Jeong et al., 2010; Kelleher et al., 2017; Kobayashi and Behringer, 2003), which are present in the starting glandular fragments, were found to be expressed in the EMOs (Fig. S1G). Immunohistochemical analysis confirmed that the expression of Esr1-encoded estrogen receptor α (ERα) in the EMOs, as also occurring in the endometrium in vivo (Fig. 1C). In agreement with Muc1 gene expression, mucus was detected in the EMO lumen (as analyzed by Periodic Acid Schiff or PAS staining), similar to its presence in the endometrial glands in vivo (Fig. 1C). Finally, the ultrastructural property of apical microvilli in mouse endometrial epithelium (Rohdin, 1974) was also reproduced in the EMOs with microvilli directed toward the lumen (Fig. 1D), thus demonstrating the apicobasal polarity of these cells. Taken together, these data show that organoids can be developed from mouse endometrium that mimic endometrial (glandular) epithelium and display long-term growth capacity.
Influence of WNT/RSPO1 on mouse EMO growth efficiency and phenotype
Next, we determined the requirement of the core components of the generic organoid medium, i.e. WNT3A, RSPO1, EGF and Noggin, for (1) the initial organoid formation process (referred to as passage 0 or P0); and (2) the further expansion and long-term passaging. For both processes, the combined presence of all four factors was most efficient (Fig. 2A,B). Removal of RSPO1 alone did not affect organoid formation efficiency (P0) when compared with all four factors together (Fig. 2A), indicating that RSPO1 is not essential for this most efficient organoid development. In contrast, subsequent growth, expansion and passageability were markedly reduced by the individual absence of each of the four core compounds tested, and most prominently when both WNT3A and RSPO1 were omitted (Fig. 2B). Hepatocyte growth factor (HGF), transforming growth factor α (TGFα) and insulin-like growth factor (IGF) were also tested as medium supplements because of their reported stimulatory activity on endometrium epithelial cell proliferation (Gargett et al., 2008). However, these factors were found to be dispensable for EMO development and expansion (Fig. S1H).
WNT3A and RSPO1 are needed for efficient long-term expansion and culture of mouse EMOs. (A) Number of organoids developed (i.e. P0) from mouse endometrium in culture medium as specified (i.e. 25W/25R with omission of the components as indicated). *P<0.05. (B) Number of passages of the EMOs in culture medium as indicated. In the absence of WNT3A-CM (25%)+RSPO1-CM (25%), organoids can be passaged only once. *P<0.05, **P<0.01. (C) Number of organoids developed from mouse endometrium (P0) in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM as indicated. *P<0.05. (D) Number of passages of the EMOs in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM, as indicated. The conditions 25W/25R and 10W/25R (boxed) are the most efficient ones for organoid regrowth and expansion. *P<0.05, **P<0.01. Data represent mean±s.e.m. of n=3 biological replicates.
WNT3A and RSPO1 are needed for efficient long-term expansion and culture of mouse EMOs. (A) Number of organoids developed (i.e. P0) from mouse endometrium in culture medium as specified (i.e. 25W/25R with omission of the components as indicated). *P<0.05. (B) Number of passages of the EMOs in culture medium as indicated. In the absence of WNT3A-CM (25%)+RSPO1-CM (25%), organoids can be passaged only once. *P<0.05, **P<0.01. (C) Number of organoids developed from mouse endometrium (P0) in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM as indicated. *P<0.05. (D) Number of passages of the EMOs in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM, as indicated. The conditions 25W/25R and 10W/25R (boxed) are the most efficient ones for organoid regrowth and expansion. *P<0.05, **P<0.01. Data represent mean±s.e.m. of n=3 biological replicates.
To further fine-tune the EMO growth conditions, we titrated the WNT3A-/RSPO1-CM volumes added to the culture. In agreement with the results above, the presence of WNT3A (alone or together with RSPO1) enhanced organoid formation efficiency (Fig. 2C, Fig. S1I). Regarding subsequent EMO expansion and passageability, the most efficient culture conditions eventually identified were: (1) 25% WNT3A-CM together with 25% RSPO1-CM (25W/25R, i.e. as used above, referred to as ‘high WNT’); and (2) 10W/25R, in which WNT3A-CM was lowered to 10% (referred to as ‘low WNT’) (Fig. 2D). Remarkably, whereas all EMOs developing in high WNT consisted of a cell layer bordering a lumen (termed cystic organoids), low WNT conditions could generate dense EMOs (Fig. 3A). These specific morphological phenotypes were retained at passaging (Fig. 3A, Fig. S2A), and were further clearly distinguished immunohistologically (Hematoxylin and Eosin, and PanCK staining; Fig. 3B, Fig. S2A). In addition, the dense organoids showed a higher proliferative activity than the cystic EMOs, as revealed by a higher proportion of Ki67+ cells (Fig. 3B) and a faster growth rate and larger eventual size, as monitored in low cell-density conditions (Fig. 3C).
The mouse EMOs show different morphology and phenotype depending on WNT3A levels. (A) Expansion (passaging) and morphology of EMOs in 10W/25R (low WNT) and 25W/25R (high WNT) culture conditions. Representative bright-field images are shown of the indicated passages. The low WNT condition generates dense organoids, whereas the high WNT condition gives rise to cystic organoids (see also magnified view in insets and right panels). Scale bars: 200 µm. (B) Hematoxylin and Eosin staining, and immunohistochemical staining for PanCK further document the dense and cystic morphologies of the EMOs. Immunohistochemical staining for the proliferation marker Ki67 and quantification of the Ki67+ cells in the EMOs reveals higher proliferative activity in the dense (10W/25R) compared with the cystic (25W/25R) organoids. Scale bar: 50 µm. ****P<0.0001. (C) Time-lapse recording of organoid development from individual cells (at P3), confirming the faster proliferation rate in 10W/25R (upper panels) versus 25W/25R (lower panels). Light-microscopic pictures are shown at the indicated time points after seeding the dissociated cells at low density. Whereas the ‘low WNT’ organoids remain dense, the ‘high WNT’ organoids become cystic and develop a lumen (arrow). Scale bar: 200 μm. (D) Expression level of endometrial epithelium genes in EMOs obtained in 25W/25R, relative to EMOs grown in 10W/25R (set as 1, dotted line), and PAS staining for mucin together show a more differentiated phenotype in the high WNT condition. Light-pink spots outside the organoid in 10W/25R are due to Matrigel background (Mat). Data are mean±s.e.m. of n=3 biological replicates. *P<0.05. (E) Gene expression of Lgr4, Lgr5 and Lgr6 in mouse EMOs (at P4) grown in 25W/25R, presented as fold difference versus expression in 10W/25R (set as 1, dotted line). Data are mean±s.e.m. of n=3 biological replicates. *P<0.05. (F) Transferring EMOs to a higher WNT3A concentration induces a more differentiated phenotype. Dense EMOs grown in low WNT (10W/25R) until P6 were further passaged (referred to as P′) under the same conditions or in higher WNT (25W/25R). The initially dense EMOs started to develop a lumen already from P′1 in high WNT (arrow), becoming more clearly developed at P′4 (arrow). Hematoxylin and Eosin, Ki67 and mucin staining further support that high WNT induces a differentiation phenotype (i.e. formation of a lumen; decreased proliferation as also quantified in the bar graph; n=3 biological replicates, **P<0.01; and production of mucin, arrow). Scale bars: 200 µm. (G) Lgr6 expression rapidly decreases when shifting from low to high WNT (P′0 means at the end of the first culture in 25W/25R, before passaging to P′1). Data are mean±s.e.m. of n=3 biological replicates. *P<0.05.
The mouse EMOs show different morphology and phenotype depending on WNT3A levels. (A) Expansion (passaging) and morphology of EMOs in 10W/25R (low WNT) and 25W/25R (high WNT) culture conditions. Representative bright-field images are shown of the indicated passages. The low WNT condition generates dense organoids, whereas the high WNT condition gives rise to cystic organoids (see also magnified view in insets and right panels). Scale bars: 200 µm. (B) Hematoxylin and Eosin staining, and immunohistochemical staining for PanCK further document the dense and cystic morphologies of the EMOs. Immunohistochemical staining for the proliferation marker Ki67 and quantification of the Ki67+ cells in the EMOs reveals higher proliferative activity in the dense (10W/25R) compared with the cystic (25W/25R) organoids. Scale bar: 50 µm. ****P<0.0001. (C) Time-lapse recording of organoid development from individual cells (at P3), confirming the faster proliferation rate in 10W/25R (upper panels) versus 25W/25R (lower panels). Light-microscopic pictures are shown at the indicated time points after seeding the dissociated cells at low density. Whereas the ‘low WNT’ organoids remain dense, the ‘high WNT’ organoids become cystic and develop a lumen (arrow). Scale bar: 200 μm. (D) Expression level of endometrial epithelium genes in EMOs obtained in 25W/25R, relative to EMOs grown in 10W/25R (set as 1, dotted line), and PAS staining for mucin together show a more differentiated phenotype in the high WNT condition. Light-pink spots outside the organoid in 10W/25R are due to Matrigel background (Mat). Data are mean±s.e.m. of n=3 biological replicates. *P<0.05. (E) Gene expression of Lgr4, Lgr5 and Lgr6 in mouse EMOs (at P4) grown in 25W/25R, presented as fold difference versus expression in 10W/25R (set as 1, dotted line). Data are mean±s.e.m. of n=3 biological replicates. *P<0.05. (F) Transferring EMOs to a higher WNT3A concentration induces a more differentiated phenotype. Dense EMOs grown in low WNT (10W/25R) until P6 were further passaged (referred to as P′) under the same conditions or in higher WNT (25W/25R). The initially dense EMOs started to develop a lumen already from P′1 in high WNT (arrow), becoming more clearly developed at P′4 (arrow). Hematoxylin and Eosin, Ki67 and mucin staining further support that high WNT induces a differentiation phenotype (i.e. formation of a lumen; decreased proliferation as also quantified in the bar graph; n=3 biological replicates, **P<0.01; and production of mucin, arrow). Scale bars: 200 µm. (G) Lgr6 expression rapidly decreases when shifting from low to high WNT (P′0 means at the end of the first culture in 25W/25R, before passaging to P′1). Data are mean±s.e.m. of n=3 biological replicates. *P<0.05.
We then compared cystic and dense EMOs regarding expression of endometrial epithelium cell markers (see above) and found higher levels in the cystic organoids (Fig. 3D). A notable strong upregulation (∼1500-fold) was measured for Muc1 and, in agreement, mucin was observed in the lumen of the cystic EMOs (25W/25R) but not in the dense EMOs (10W/25R) (Fig. 3D). The mucin-positive and -negative phenotypes remained stable in subsequent passages (data not shown). Together, these results suggest a more differentiated endometrial epithelium phenotype of the cystic organoids.
RSPO ligands act through specific receptors, i.e. LGR4, LGR5 and/or LGR6 (de Lau et al., 2011; Wang et al., 2013). As RSPO1 is needed for efficient EMO culture, we analyzed gene expression of these receptors. Lgr5, which is typically proposed to be the receptor mediating RSPO1 activity (Carmon et al., 2011), was higher expressed in the cystic organoids (high WNT) than in the dense EMOs (low WNT) (Fig. 3E). Intriguingly, expression of Lgr6 showed the opposite pattern. The expression of Lgr4 was not different (Fig. 3E). This Lgr gene expression profile remained similar in further passages (as shown for P10; Fig. S2B), thereby supporting gene expression stability in the organoids. Finally, we examined whether transferring dense organoids from low to high WNT changed morphology and phenotype. A cystic configuration with the formation of a lumen was induced, and proliferative activity and Lgr6 expression dropped, while mucin production emerged (Fig. 3F,G).
Together, our findings strongly support that cystic EMOs grown under high WNT conditions possess a more differentiated phenotype, including a lumen-containing glandular-like shape, reduced proliferative activity and a secretory nature. The low WNT condition, however, appears to favor a more immature character.
Mouse EMOs show physiological hormone responsiveness
Endometrial biology and remodeling are predominantly regulated by estrogen and progesterone (Pedram et al., 2014). In essence, estrogen increases proliferation and growth of the endometrial tissue, whereas progesterone regulates differentiation and secretory maturation (Filant and Spencer, 2014; Gargett et al., 2008). We examined whether these specific actions were reproduced in the EMO model (starting from the cystic organoids with the more-evolved endometrial phenotype). Exposure to estrogen (17β-estradiol, E2), but not to progesterone (Prog), increased the number of proliferating cells in the EMOs (Fig. 4A). Gene expression analysis further showed specific regulation of several known E2- and Prog-responsive markers (Bagchi et al., 2005; Lee et al., 2006; Li et al., 2011; O'Sullivan et al., 2004; Surveyor et al., 1995). Egf, Igf1 and Lf (lactoferrin, also known as lactotransferrin) were significantly upregulated in response to E2, whereas Alox15, Lif and Prss28 were elevated by Prog (Fig. 4B), both of which mimic the in vivo responses. Treatment with Prog downregulated Esr1 expression as in vivo, in accordance with a reduction in the number of ERα-immunopositive cells (26.9±8.8% versus 55.7±4.5% in control EMOs; mean±s.e.m.; n=3, P<0.01). Known effects on WNT pathway components were also reproduced (Miller et al., 1998), in particular the upregulated expression of Wnt5a and Wnt7a upon Prog treatment (Fig. 4C). Moreover, Lgr5 was downregulated upon hormonal treatment, in agreement with a previous report (Sun et al., 2009), whereas Lgr4 expression showed no change (Fig. 4C). Rather unexpectedly, Muc1 expression was not upregulated by Prog (Fig. 4B). Similarly, no difference in mucus production was observed after Prog treatment, although primarily because mucin was already prominently present in the non-treated cystic EMO tested (25W/25R; Fig. 4D). Indeed, when the more immature (dense) organoids were subjected to the individual hormone treatments, mucin production was clearly stimulated by Prog, in agreement with the upregulation of Muc1 expression (Fig. 4D). The latter observations further add to the notion that dense and cystic organoids represent a dissimilar status, being more immature and more differentiated, respectively.
Mouse EMOs show physiological responses to hormones. (A) Effect of E2 and Prog treatment on cell proliferation in mouse EMOs (grown in 25W/25R), as assessed by Ki67 immunostaining (quantified in the bar graph; n=3 biological replicates, **P<0.01). EMOs were treated with E2 for 2 days (E2), with E2 followed by Prog for 2 additional days (Prog), or were not hormonally treated (control, Co). Scale bar: 50 µm. (B) Expression levels of hormone-responsive genes in EMOs treated with E2 (black) or Prog (gray), calculated as fold change versus non-treated control EMOs (set as 1, dotted line). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05, **P<0.01. (C) Expression levels of Wnt/Lgr-associated genes in EMOs treated with E2 (black) or Prog (gray), calculated as fold change versus non-treated control EMOs (set as 1, dotted line). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05, **P<0.01, ***P<0.001. (D) Effect of E2 and Prog treatment on mucin production in EMOs, grown in 25W/25R or 10W/25R, as indicated. Scale bar: 200 µm. Gene expression of Muc1 in mouse EMOs treated with E2 (black) or Prog (gray), presented as fold difference versus untreated control (set as 1, dotted line). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05.
Mouse EMOs show physiological responses to hormones. (A) Effect of E2 and Prog treatment on cell proliferation in mouse EMOs (grown in 25W/25R), as assessed by Ki67 immunostaining (quantified in the bar graph; n=3 biological replicates, **P<0.01). EMOs were treated with E2 for 2 days (E2), with E2 followed by Prog for 2 additional days (Prog), or were not hormonally treated (control, Co). Scale bar: 50 µm. (B) Expression levels of hormone-responsive genes in EMOs treated with E2 (black) or Prog (gray), calculated as fold change versus non-treated control EMOs (set as 1, dotted line). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05, **P<0.01. (C) Expression levels of Wnt/Lgr-associated genes in EMOs treated with E2 (black) or Prog (gray), calculated as fold change versus non-treated control EMOs (set as 1, dotted line). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05, **P<0.01, ***P<0.001. (D) Effect of E2 and Prog treatment on mucin production in EMOs, grown in 25W/25R or 10W/25R, as indicated. Scale bar: 200 µm. Gene expression of Muc1 in mouse EMOs treated with E2 (black) or Prog (gray), presented as fold difference versus untreated control (set as 1, dotted line). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05.
Finally, we examined whether mouse EMOs also responded to hormones in vivo. EMOs established from endometrium of TdTomato reporter mice (Fig. S3A) were transplanted under the kidney capsule of ovariectomized immunodeficient NSG mice that were sequentially treated with E2 and Prog (see Janzen et al., 2013), or were sham treated (Fig. S3B). In the absence of ovarian hormones, TdTomato+ structures were still observed 1.5 months after transplantation, showing the capacity of the EMOs to survive in vivo; however, structures remained small (Fig. S3C). In contrast, after E2 and Prog treatment, the TdTomato+ organoid grafts had expanded and assembled into an organized structure with glandular-type protuberances (Fig. S3C). Thus, the EMOs also respond to hormonal exposure in vivo, but further research is needed to fully characterize the formed tissue. Taken together, the results described above demonstrate that mouse EMOs are responsive to hormonal regulation in a manner that resembles the behavior of endometrial epithelium in vivo, thus underscoring the physiological relevance of this newly developed organoid model.
Development of organoids from human endometrium and the influence of WNT/RSPO
We then investigated whether organoids could also be established from human endometrium. Starting from the EMO growth medium as defined above (to which N-acetyl-L-cysteine and the p38 inhibitor SB202190 were additionally added as is generally applied to human organoids; Sato et al., 2011), we again titrated the concentrations of WNT3A- and RSPO1-CM. Organoids were obtained in all conditions tested from all endometrial biopsies (Fig. 5A,B). In contrast to mouse EMOs, WNT3A was not needed for further expansion and passaging of human EMOs (see 0W/25R; Fig. 5C,D), which may be due to a sufficiently active endogenous WNT system. Indeed, several WNT ligands were found to be expressed in the human EMOs (Fig. S4A), and interference with endogenous WNT ligand secretion using IWP2 diminished the abundance of regrowing EMOs (Fig. S4B). In contrast, RSPO1 was required for efficient, long-term human EMO expansion (Fig. 5C), as also defined above for mouse EMOs. In the absence of RSPO1, human organoids could no longer be passaged after P3 (see 25W/0R; Fig. 5C, Fig. S4C). Endogenous expression of RSPO ligands (RSPO1, RSPO2 and RSPO3) was indeed not found in the EMOs (data not shown). Analysis of RSPO receptors showed that human EMOs express LGR4 and LGR5, and that expression levels are high (comparable to the levels found in human small-intestinal organoids, which are well known to strongly express these genes) (Barker et al., 2013; de Lau et al., 2011) (Fig. 5E). LGR6 showed much lower expression, which is understandable in view of the general cystic nature of the human EMOs obtained (see Fig. 5A,D), and thus consistent with the finding of lower expression of Lgr6 in the cystic (versus the dense) mouse EMOs. The cystic nature of human EMOs was further obvious in immunohistological analysis, showing a glandular morphology with a PanCK+ epithelial cell layer surrounding a lumen and an E-cadherin+ pattern mimicking the in vivo cell-cell adhesion junctions of the endometrial epithelium (Fig. 5F). In addition, the glandular epithelium marker FOXA2 was found expressed in the human EMOs (Fig. S4D). Moreover, ultrastructural properties of human endometrial epithelium, in particular the presence of apical microvilli as well as of cilia (Rohdin, 1974), was recapitulated in human EMOs, thus supporting apicobasal polarity (Fig. 5F). The human EMOs also expressed ERα (Fig. 5F), as occurs in vivo, and in line with this expression, we found that addition of E2, although not strictly required, promoted the growth and expansion of the EMOs (data not shown). Therefore, E2 was further included in the human EMO culture medium, allowing successful expansion for more than 10 passages (4 months) currently (Fig. 5D).
Organoids develop from human endometrium copying the phenotype of the epithelium. (A) Organoid formation from dissociated human endometrium in culture medium with varying proportions of WNT3A- and/or RSPO1-CM, as indicated. Representative light-microscopic pictures are shown at 10 days of culture after seeding. Scale bar: 200 µm. (B) Number of organoids developed from human endometrium (i.e. P0) in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM, as indicated. Data are presented as dot plots with indications of mean±s.e.m. of n=5 biological replicates (hormonally non-treated patients). (C) Number of passages of human EMOs in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM, as indicated. Organoid regrowth and expansion is highly efficient in the absence of WNT3A (i.e. 0W/25R, boxed). Data are presented as dot plots with indications of mean±s.e.m. of n=5 biological replicates (hormonally non-treated patients). *P<0.05. (D) Organoid development from human endometrium and passaging in 0W/25R (without E2; upper panels) and in medium further optimized by addition of E2 (lower panels), showing long-term expansion and passaging capability, and maintenance of the morphological (cystic) phenotype. Representative bright-field images are shown of P0 (day 0 and day 7) and the indicated passages. Scale bars: 200 µm. (E) Gene expression of LGR4, LGR5 and LGR6 in human EMOs, presented as fold difference versus expression in organoids developed from human small intestine (set as 1, dotted line). LGR4 and LGR5 are highly expressed in the EMOs, comparable with their known high expression in small-intestinal organoids. Data are presented as dot plots with indications of mean±s.e.m. of n=9 biological replicates (hormonally non-treated patients). *P<0.05. (F) The EMOs show phenotypic characteristics of human endometrial epithelium. Hematoxylin and Eosin staining, and immunohistochemical staining for PanCK, E-cadherin and ERα (with box magnified in inset) of EMOs, and also for E-cadherin in endometrium, are shown, revealing a glandular-like morphology with a lumen bordered by epithelial cells. Cell-cell adhesion E-cadherin+ signals in the epithelial columnar-type cells are similar in endometrium and EMOs. Scale bars: 50 µm. TEM reveals the presence of microvilli and cilia in the EMOs at the luminal side (L), as indicated by arrow and arrowhead in the magnified view of the boxed region (bottom), respectively. Scale bars in TEM images: 10 µm (top panel) and 2 µm (bottom panel).
Organoids develop from human endometrium copying the phenotype of the epithelium. (A) Organoid formation from dissociated human endometrium in culture medium with varying proportions of WNT3A- and/or RSPO1-CM, as indicated. Representative light-microscopic pictures are shown at 10 days of culture after seeding. Scale bar: 200 µm. (B) Number of organoids developed from human endometrium (i.e. P0) in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM, as indicated. Data are presented as dot plots with indications of mean±s.e.m. of n=5 biological replicates (hormonally non-treated patients). (C) Number of passages of human EMOs in culture medium with varying proportions (%) of WNT3A- and/or RSPO1-CM, as indicated. Organoid regrowth and expansion is highly efficient in the absence of WNT3A (i.e. 0W/25R, boxed). Data are presented as dot plots with indications of mean±s.e.m. of n=5 biological replicates (hormonally non-treated patients). *P<0.05. (D) Organoid development from human endometrium and passaging in 0W/25R (without E2; upper panels) and in medium further optimized by addition of E2 (lower panels), showing long-term expansion and passaging capability, and maintenance of the morphological (cystic) phenotype. Representative bright-field images are shown of P0 (day 0 and day 7) and the indicated passages. Scale bars: 200 µm. (E) Gene expression of LGR4, LGR5 and LGR6 in human EMOs, presented as fold difference versus expression in organoids developed from human small intestine (set as 1, dotted line). LGR4 and LGR5 are highly expressed in the EMOs, comparable with their known high expression in small-intestinal organoids. Data are presented as dot plots with indications of mean±s.e.m. of n=9 biological replicates (hormonally non-treated patients). *P<0.05. (F) The EMOs show phenotypic characteristics of human endometrial epithelium. Hematoxylin and Eosin staining, and immunohistochemical staining for PanCK, E-cadherin and ERα (with box magnified in inset) of EMOs, and also for E-cadherin in endometrium, are shown, revealing a glandular-like morphology with a lumen bordered by epithelial cells. Cell-cell adhesion E-cadherin+ signals in the epithelial columnar-type cells are similar in endometrium and EMOs. Scale bars: 50 µm. TEM reveals the presence of microvilli and cilia in the EMOs at the luminal side (L), as indicated by arrow and arrowhead in the magnified view of the boxed region (bottom), respectively. Scale bars in TEM images: 10 µm (top panel) and 2 µm (bottom panel).
Human EMOs show physiological, cycle-mimicking responses to hormones
We examined whether the human EMOs could reproduce the menstrual cycle, in particular the proliferative, secretory and menstrual phases. Organoids were sequentially treated with the hormones that drive these phases, i.e. successive E2 and Prog exposure followed by hormone withdrawal. Treatment with E2 resulted in a higher proportion of Ki67+ cells, and subsequent administration of Prog again reduced the number of proliferating cells (Fig. 6A), consistent with the in vivo effects of these hormones (Eritja et al., 2013). Interestingly, Prog treatment induced morphological changes in EMOs that mimicked characteristics of maturation into secretory-phase endometrium, in particular increased folding and tortuosity of the glands with columnar cells displaying subnuclear vacuolation (Fig. 6B,C). On the other hand, epithelium of E2-treated EMOs showed resemblance to pseudostratified glandular epithelium of the proliferative-phase endometrium (Fig. 6C). Moreover, the proliferative phase-specific marker thyrotropin-releasing hormone (TRH) was found in EMOs after E2 treatment (comparable to in vivo) and was absent in Prog-treated EMOs, whereas the secretory phase-specific marker progestagen-associated endometrial protein (PAEP) was detected in EMOs after Prog treatment (comparable with the in vivo secretory endometrium) and absent in E2-treated EMOs (Fig. 6D) (Zieba et al., 2015). Furthermore, immunofluorescence staining detected acetylated α-tubulin+ cilia and mucin production upon Prog treatment, both of which are present in the secretory-phase endometrium (Fig. 6D) (Meseguer et al., 2001). Mucin production reflects the upregulation of MUC1 gene expression by Prog (Fig. 6E). In addition, other markers of the secretory-phase endometrium (such as ALOX15 and AQP3) (Kuokkanen et al., 2010; Ruiz-Alonso et al., 2012; Talbi et al., 2006) were upregulated by Prog treatment, whereas markers of the proliferative phase (such as ESR1) were downregulated (Fig. 6E). The extensive DNA synthesis during the proliferative phase requires expression of mitosis-regulatory factors such as the minichromosome maintenance (MCM) family, which are elevated in vivo when compared with secretory-phase endometrium (Kuokkanen et al., 2010). Accordingly, MCM2, MCM3 and MCM4 were upregulated by E2 in the EMO model (Fig. S5A). Finally, several WNT ligands were found to be differentially expressed upon E2 and Prog treatment, consistent with reports on their phase-specific expression in vivo (Punyadeera et al., 2005; Talbi et al., 2006). WNT4, WNT5A and WNT9A showed the highest differences in expression, being upregulated by Prog (Fig. 6E), whereas the WNT pathway inhibitor WIF1 was downregulated by Prog, as also found in vivo (Talbi et al., 2006). As these particular WNT ligands are associated with non-canonical WNT signaling, we analyzed the expression of non-canonical target genes. SIAH2 and CAPN1, which are reported to mediate WNT5A signaling in the non-canonical Ca2+-dependent pathway (Komiya and Habas, 2008; Topol et al., 2003), were found to be upregulated in the Prog-treated EMOs (Fig. 6E). Levels of PCDH8, a target gene of the non-canonical planar-polarity pathway, did not change.
The human EMOs show physiological responses to hormones and reproduce the menstrual cycle. (A) The effect of E2 and Prog treatment on cell proliferation (as assessed by Ki67 immunostaining) in human EMOs. Organoids were treated with E2 for 7 days (E2), with E2 followed by Prog for 7 additional days (while reducing the E2 concentration to half; Prog) or were not hormonally treated (Co). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05, **P<0.01. (B) Prog increases tortuosity in human EMOs (arrows, indicating the same organoids at day 0 and 7 of treatment, with the boxed area magnified in the inset; an additional example of tortuous EMOs is on the right). Representative bright-field images are shown. Scale bars: 200 µm. (C) Hematoxylin and Eosin staining of human EMOs treated with E2 or Prog (lower panels), and of corresponding proliferative- and secretory-phase endometrium (upper panels). The epithelium is pseudostratified in the proliferative phase and after E2 treatment, and there is increased tortuosity and subnuclear vacuolation (magnified boxes) in the secretory phase and after Prog treatment. Scale bars: 50 µm. (D) Immunostaining of phase-specific markers in human endometrium and hormone-treated EMOs. The proliferative phase marker TRH is expressed in the EMOs under E2 exposure, whereas the secretory phase marker PAEP is expressed upon Prog treatment. Cilia become visible (as analyzed by immunofluorescent staining of acetylated α-tubulin) in EMOs under Prog exposure, corresponding to their presence in the secretory phase endometrium (arrows and magnified boxes). Mucin is observed in the Prog-treated EMOs, in accordance with its presence in the secretory phase endometrium. Scale bar: 50 µm. (E) Expression levels of hormone-responsive genes, of WNT/LGR-associated genes and of non-canonical WNT target genes in human EMOs treated with Prog, calculated as fold change versus E2-treated EMOs (set as 1, dotted line). Data are mean±s.e.m. of n=4-6 biological replicates. *P<0.05. (F) Following successive E2 and Prog treatment, organoids were cultured in hormone- and Phenol Red-free medium for another 4 days, and immunostained for CC3. Hormone withdrawal leads to abundant emergence of apoptotic cells (arrows). Scale bar: 50 µm.
The human EMOs show physiological responses to hormones and reproduce the menstrual cycle. (A) The effect of E2 and Prog treatment on cell proliferation (as assessed by Ki67 immunostaining) in human EMOs. Organoids were treated with E2 for 7 days (E2), with E2 followed by Prog for 7 additional days (while reducing the E2 concentration to half; Prog) or were not hormonally treated (Co). Data are mean±s.e.m. of n=5 biological replicates. *P<0.05, **P<0.01. (B) Prog increases tortuosity in human EMOs (arrows, indicating the same organoids at day 0 and 7 of treatment, with the boxed area magnified in the inset; an additional example of tortuous EMOs is on the right). Representative bright-field images are shown. Scale bars: 200 µm. (C) Hematoxylin and Eosin staining of human EMOs treated with E2 or Prog (lower panels), and of corresponding proliferative- and secretory-phase endometrium (upper panels). The epithelium is pseudostratified in the proliferative phase and after E2 treatment, and there is increased tortuosity and subnuclear vacuolation (magnified boxes) in the secretory phase and after Prog treatment. Scale bars: 50 µm. (D) Immunostaining of phase-specific markers in human endometrium and hormone-treated EMOs. The proliferative phase marker TRH is expressed in the EMOs under E2 exposure, whereas the secretory phase marker PAEP is expressed upon Prog treatment. Cilia become visible (as analyzed by immunofluorescent staining of acetylated α-tubulin) in EMOs under Prog exposure, corresponding to their presence in the secretory phase endometrium (arrows and magnified boxes). Mucin is observed in the Prog-treated EMOs, in accordance with its presence in the secretory phase endometrium. Scale bar: 50 µm. (E) Expression levels of hormone-responsive genes, of WNT/LGR-associated genes and of non-canonical WNT target genes in human EMOs treated with Prog, calculated as fold change versus E2-treated EMOs (set as 1, dotted line). Data are mean±s.e.m. of n=4-6 biological replicates. *P<0.05. (F) Following successive E2 and Prog treatment, organoids were cultured in hormone- and Phenol Red-free medium for another 4 days, and immunostained for CC3. Hormone withdrawal leads to abundant emergence of apoptotic cells (arrows). Scale bar: 50 µm.
Subsequent hormone withdrawal (and use of medium without the estrogenic Phenol Red) led to disruption of the glandular architecture with darkening of the organoids and loss of cells at their border that were Trypan Blue positive (Fig. S5B). Accordingly, apoptotic (i.e. cleaved caspase 3- or CC3-immunopositive) cells abundantly emerged in the organoids and their lumen (Fig. 6F). Taken together, we were also able to develop organoids from human endometrium that mimic the cycle-specific epithelial characteristics of the tissue in response to hormones and that are capable of long-term propagation in culture, thereby providing a promising model for studying human endometrial biology and pathology in greater depth.
DISCUSSION
Our present study describes the realization of a long-term expandable and stable organoid culture system starting from mouse and human endometrium. The organoids replicate the epithelium of the tissue, including glandular-type organization and expression of specific markers. Moreover, the organoids reproduce functional behavior such as mucus production and responsiveness to the core regulatory hormones estrogen and progesterone. The human EMOs mimic the menstrual cycle as regulated by these hormones. Characteristics of the proliferative phase such as increased epithelial cell proliferation and expression of specific markers such as TRH (Zieba et al., 2015) are reproduced in the EMOs after E2 treatment. Subsequent Prog exposure induces features of the secretory phase, as occurs in vivo, including enhanced folding and tortuosity, formation of columnar epithelium with subnuclear vacuolation, mucus production, ciliogenesis, downregulation of ERα and expression of specific markers such as PAEP (Zieba et al., 2015). Finally, hormone withdrawal triggers a phenotype that is reminiscent of the menstrual phase, including disruption of tissue structure and shedding of dying cells. The human EMOs thus provide an interesting model with which to study mechanisms of cycle physiology and remodeling of endometrial epithelium. Furthermore, the EMOs can be expanded in long-term culture while stably maintaining their properties, thereby overcoming the limitations of existing endometrium in vitro 2D cell cultures. Organoids, which reflect the epithelial compartment of a tissue, have been shown to provide powerful tools for modeling and deciphering tissue development, regulation and disease, without the essential presence of the mesenchymal compartment of the tissue (Clevers, 2016). Nonetheless, co-culture of the epithelial EMO with endometrial stromal cells can further complement the organoid model, given the reciprocal epithelium-stroma interactions operative in the tissue (Kurita et al., 2001; Li et al., 2011). The development of endometrial epithelium organoids represents the first step toward such a (re-)combination model. Indeed, although endometrial stromal cells are readily expandable in culture (De Clercq et al., 2015), endometrial epithelial cells remained recalcitrant to robust expansion, an obstacle that is now removed. Co-culture systems within the organoid field have just started to be developed (e.g. intestinal epithelial organoids combined with mesenchymal cells; Stzepourginski et al., 2017). Endometrial stroma may be needed to induce hormone responses to their full extent. We observed specific responses in the EMOs but effects seemed to be less pronounced than those occurring in vivo (Li et al., 2011). A co-culture system from human endometrium has been reported before (Bläuer et al., 2005). Although ‘organoids’ were reported in these cultures, they clearly differ from the organoid concept currently adopted (Clevers, 2016) and from our EMO model in important aspects. Whereas in our approach organoids self-form starting from endometrial fragments or cells, Bläuer et al. (2005) dissected so-called ‘organoids’ directly from the endometrium, representing the glands without fragmentation. Characterization of the endometrial phenotype was limited, and growth and expansion of the structures was not obtained (Bläuer et al., 2005; Blauer et al., 2008), thus excluding these cultures from being useful models for detailed scrutiny. Of note, further optimization may also be achieved by substituting Matrigel (which contains, among other substances, laminin, collagen IV, entactin and proteoglycans), although widely used for organoid models and also effective in other 3D endometrial cultures (Eritja et al., 2010, 2013; Gargett and Masuda, 2010; Janzen et al., 2013; Valentijn et al., 2013), for more endometrium-typical extracellular matrix (containing, among other substances, collagens I, III, IV and VI, laminin, desmin, nidogen 1 and biglycan).
As is true for organoids of most tissues (Clevers, 2016), the WNT/RSPO1 pathway is important for efficient EMO culture. The culture conditions identified for human and mouse EMOs were comparable, except for WNT3A, which was not needed for efficient growth and expansion of human EMOs (as similarly reported for human prostate organoids; Karthaus et al., 2014). WNT ligands are endogenously expressed in the human EMO cultures, in agreement with their expression in the in vivo endometrium (Punyadeera et al., 2005; Tulac et al., 2003), which may explain the WNT3A-independent expansion. Endogenous expression of RSPO ligands was not detected in the human EMO cultures, justifying the essential addition of RSPO1. Interestingly, the level of WNT3A influenced the maturation state of the (mouse) EMOs, with high WNT level yielding a more differentiated phenotype. Although it is known that the WNT pathway is essential during embryonic development of the uterus (Dunlap et al., 2011; Franco et al., 2011; Mericskay et al., 2004; Miller et al., 1998), its exact function during adulthood remains ill defined. It has been reported that WNT signaling plays a role in endometrial growth and regression (Tulac et al., 2003), and that it is modulated by estrogen and progesterone during the cyclic remodeling (Wang et al., 2010). We found differential expression of WNT ligands upon E2 and Prog treatment in both mouse and human EMOs, comparable with previous in vivo reports. In mice, Wnt5a and Wnt7a, which are upregulated upon Prog treatment, increase at estrous when Prog levels start to rise (Miller et al., 1998), and are described as being important for glandular differentiation as well as for preparing a receptive environment for blastocyst implantation (Cha et al., 2014; Miller et al., 1998). In humans, WNT5A and WNT7A were found to be downregulated and the WNT inhibitor WIF1 was upregulated during the proliferative phase (Punyadeera et al., 2005; Talbi et al., 2006), comparable to the reversed findings in human EMOs after Prog treatment (reflecting the secretory phase), showing upregulation of WNT5A and WNT7A (although the latter not significantly) and downregulation of WIF1, respectively. However, except for a similar change of WNT3A and WNT4 being down- and upregulated in the secretory versus proliferative phase, respectively, Tulac et al. (2003) reported downregulation of WNT5A and WNT7A in the secretory phase (Tulac et al., 2003). Different experimental approaches, e.g. the analysis of global endometrium containing both epithelial and stromal cells in the reports described (Punyadeera et al., 2005; Talbi et al., 2006; Tulac et al., 2003), and the comparison of samples pooled from different patients with potential inter-individual differences (whereas we compared control and hormone-treated organoids within the same patients) may explain some of the differences observed. Interestingly, ligands and targets of the non-canonical WNT pathway were upregulated by Prog exposure (and canonical WNT3A, WNT5B and WNT6 expression was downregulated), suggesting that canonical WNT signaling shifts to non-canonical during endometrial maturation in preparation for blastocyst implantation. In addition, the WNT pathway may play a role in endometrium proliferative diseases. β-Catenin, the core transducer of the canonical WNT pathway (Reya and Clevers, 2005), is one of the most mutated genes in endometrial cancer (Jeong et al., 2009; Cancer Genome Atlas Research Network, 2013). The function and regulation of WNT in endometrial homeostasis, remodeling and disease can now be more deeply studied using the EMO model.
RSPO1 signaling, which is essential for EMO growth and expansion, is mediated by LGR receptors that were found to be expressed in the EMOs. Interestingly, Lgr5 and Lgr6 showed opposite expression patterns in the cystic versus dense mouse EMOs. The reason is not clear yet but may have to do with expression in different types or states of cells, in particular of stem cells. Indeed, LGR5 and LGR6 mark stem cells in multiple adult tissues, but their expression profile is different in different organs (Barker et al., 2013; Clevers, 2016; Kessler et al., 2015). As the stem cells of tissues are generally considered to initiate organoid development, our EMO model may also provide insight into the identity of endometrial stem cells, which remains enigmatic and poorly defined (Gargett et al., 2016). Lgr5 expression was found to be downregulated in the mouse EMOs after E2 and Prog treatment. By analogy, Lgr5 mRNA is highly expressed in pre-pubertal mouse uterus and is strongly downregulated in ovariectomized mice upon hormone injection (Sun et al., 2009). In the human EMOs, LGR4 and LGR5 gene expression did not change under hormone treatment, in agreement with a previous report (Krusche et al., 2007).
Taken together, we achieved the development of a physiological and long-term expandable organoid model from both mouse and human endometrium. This new culturing system provides a powerful tool for exploring which cell types and molecular pathways are driving the high remodeling activity of this key reproductive tissue, and how deregulation may lead to disease by comparing organoids from normal and diseased endometrium. The method offers the prospect of creating a biobank resource from clinical biopsies for profound research of endometrial (patho-)biology, also including gene-editing approaches with CRISPR/Cas9 (Matano et al., 2015). Along the same line, the EMO approach, which has the ability to increase the limited quantity of starting material, can provide a valuable platform and preclinical model for drug development and screening, as recently shown for colorectal cancer (van de Wetering et al., 2015).
MATERIALS AND METHODS
Animals
Female mice (aged 6-10 weeks or 1 year; bred in-house) were used for the study. Wild-type mice (FVB and C57B6) were originally purchased from Janvier Labs. The TdTomato (Tandem Tomato) reporter mouse model [Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo] and the NOD-Scid IL2Rgammanull (NSG) immunodeficient mice were obtained from The Jackson Laboratory. Animal experiments were approved by the KU Leuven Ethical Committee. Mice were kept in the KU Leuven animal housing facility at constant temperature, humidity and day/night cycle, with access to water and food ad libitum.
Organoid culture from mouse endometrium
Uterine horns were isolated from the mice, minced, incubated in EDTA solution and mechanically dissociated. The obtained endometrial fragments (Fig. S1A) were resuspended in 70% Matrigel/30% DMEM/F12, plated as drops and cultured in DMEM/F12 containing a cocktail of growth and signaling factors (for further details, see supplementary Materials and Methods). Outgrowing organoids were passaged every 7-10 days after trypsinization. Unless otherwise stated, organoids of low passage number (P3-P5) were used for the experiments described.
Organoid culture from human endometrium
Endometrial biopsies were obtained from hormonally non-treated patients undergoing laparoscopy for benign gynecological conditions. The study was approved by the UZ Leuven Ethical Committee (S59006) and informed written consent was obtained. Tissue samples were dissociated using collagenase IV and mechanical trituration, cells were plated in 70% Matrigel and cultured in medium as described in the supplementary Materials and Methods. Organoids were enzymatically dissociated and passaged every 14 days. Unless otherwise stated, organoids of low passage number (P3-P5) were used.
Hormonal treatment
Mouse EMOs were either treated for 2 days with E2 (1 nM) or for 2 days with E2 followed by another 2 days with Prog (50 ng/ml). Human EMOs were either treated with E2 (1 nM) for 7 days, or for 7 days with E2 followed by Prog (50 ng/ml) for another 7 days. A series was subsequently cultured in hormone- and Phenol Red-free medium for another 4-6 days. For further details, see supplementary Materials and Methods.
In vivo transplantation of mouse EMOs
Organoids grown from endometrium of TdTomato reporter mice were transplanted beneath the kidney capsule of ovariectomized NSG mice, and sequentially treated with E2 and Prog (Janzen et al., 2013) or sham treated (for time schedule, see Fig. S3B). Kidneys were removed from euthanized animals and grafts analyzed for TdTomato red fluorescence. For further details, see supplementary Materials and Methods.
Histochemical and TEM analysis
Organoids were fixed and paraffin sections subjected to Hematoxylin and Eosin or immunohistochemical staining (for antibodies, see Table S1), or subjected to TEM (see supplementary Materials and Methods). Periodic Acid Schiff (PAS) staining was performed to visualize mucin (see supplementary Materials and Methods).
Gene expression analysis by RT-qPCR
Organoid RNA was reverse-transcribed and subjected to SYBR Green-based quantitative real-time PCR (qPCR) using the forward and reverse primers described in Table S2. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and hypoxanthine phosphoribosyltransferase (Hprt) were used as housekeeping genes. Relative gene expression levels were calculated as ΔCt values (Ct ‘target’ minus Ct ‘housekeeping gene’) and in most analyses compared between ‘sample’ and ‘reference’ to express fold change, i.e. 2-(ΔCt sample - ΔCt reference). For further details, see supplementary Materials and Methods.
Statistical analysis
Statistical analyses were performed as described in detail in the supplementary Materials and Methods. All experiments were carried out using at least three biological replicates (n≥3), each analyzed in at least two (≥2) technical replicates.
Acknowledgements
We thank our other colleagues (H.V. group: Dr Mertens, Dr Willems and Y. Van Goethem; Stem Cell Institute: Dr R. Khoueiry and Dr A. Santo Ramalho Venâncio) for their valuable input and technical help. We are also very grateful to J. Laureys (Department of Clinical and Experimental Medicine, KU Leuven, Belgium) for his expert help in mouse transplantation experiments, and to the patients, staff and nurses (UZ Leuven) for providing the clinical samples. We are thankful to Dr Clevers (Hubrecht Institute, Utrecht, The Netherlands) for providing the cell lines producing WNT3A and RSPO1, as well as to the research groups that originally developed these cell lines (from Dr Nusse and Dr Kuo, respectively). We also thank InfraMouse (KU Leuven-VIB, Hercules type 3 project ZW09-03) for use of their histological instruments and microscopes. Finally, we acknowledge the expert TEM help of Dr Baatsen and K. Vints (Electron Microscopy Platform of the Centre for Human Genetics, VIB KU Leuven).
Author contributions
Conceptualization: M.B., B.C., A.F., A.V., C.M., H.V.; Methodology: M.B., B.C., M.N., N.H., H.R., M.F., H.V.; Software: M.B., H.V.; Validation: M.B., H.V.; Formal analysis: M.B., B.C., H.V.; Investigation: M.B., B.C., M.N., N.H., H.R., H.V.; Resources: M.B., M.N., N.H., A.F., F.A., D.T., C.T., A.V., C.M., M.F., H.V.; Data curation: M.B., B.C., H.V.; Writing - original draft: M.B., B.C., M.N., H.V.; Writing - review & editing: M.B., B.C., H.V.; Visualization: M.B., H.V.; Supervision: H.V.; Project administration: H.V.; Funding acquisition: H.V.
Funding
This work was supported by grants from the KU Leuven (Research Fund; GOA) and from the Fonds Wetenschappelijk Onderzoek (FWO). M.B. is a PhD Fellow supported by the GOA grant. B.C. and M.N. are supported by a PhD Fellowship from the Fonds Wetenschappelijk Onderzoek, and H.R. is supported by a PhD Fellowship from the Agentschap voor Innovatie door Wetenschap en Technologie (IWT). D.T. and M.F. are Senior Clinical Investigators of the Fonds Wetenschappelijk Onderzoek.
References
Competing interests
The authors declare no competing or financial interests.