In vitro models of the human endometrium: evolution and application for women’s health

Abstract

The endometrium is the inner lining of the uterus that undergoes complex regeneration and differentiation during the human menstrual cycle. The process of endometrial shedding, regeneration, and differentiation is driven by ovarian steroid hormones and prepares the endometrium and intrauterine environment for embryo implantation and pregnancy establishment. Endometrial glands and their secretions are essential for pregnancy establishment, and cross talk between the glandular epithelium and stromal cells appears vital for decidualization and placental development. Despite being crucial, the biology of the human endometrium during pregnancy establishment and most of pregnancy is incomplete, given the ethical and practical limitations of obtaining and studying endometrium from pregnant women. As such, in vitro models of the human endometrium are required to fill significant gaps in understanding endometrial biology. This review is focused on the evolution and development of in vitro three-dimensional models of the human endometrium and provides insight into the challenges and promises of those models to improve women’s reproductive health.

This review describes the evolution of in vitro models of human endometrium and potential applications for the study of endometrial function and dysfunction in women.

Introduction

Successful implantation and pregnancy rely on a developmentally competent embryo and a receptive uterine environment. As such, synchronized development and adequate cross talk between the embryo and the endometrium of the uterus is crucial [1, 2]. The inner surface of the endometrium is lined with a single layer of pseudostratified epithelium, the luminal epithelium (LE), which serves as the site for initial attachment of the embryo trophectoderm. Invaginations of the epithelium form coiled and slightly branched glands that extend from the lumen to the inner circular layer of the myometrium. Glandular epithelia (GE) secrete and produce factors that regulate embryo survival and placental growth as well as impact stromal cell decidualization [3, 4]. Stromal cells and an extracellular matrix (ECM) support the LE and GE, and the endometrium is also invested with blood vessels and immune cells. Our purpose here is to review the development of in vitro three-dimensional (3D) models of the human endometrium, with a focus on the epithelium, and highlight the challenges and promises of endometrial organoid models for studies of uterine biology and pregnancy.

The endometrium undergoes complex regeneration and differentiation during the menstrual cycle that is mostly unique to humans but also present in some Old World primates, the spiny mouse, the elephant shrew, and some bat species [5, 6]. In the humans, there are two primary layers of the endometrium—the upper functionalis and lower basalis. The functionalis layer of the endometrium is shed during menstruation due to the withdrawal of ovarian steroid hormones. However, this layer of the endometrium regenerates during the proliferative phase of the menstrual cycle that involves regrowth of the glands, stroma, and blood vessels. Following ovulation and under control of progesterone (P4), the glands undergo a secretory transformation, becoming increasingly tortuous and secretory, and the stromal cells begin to decidualize. If pregnancy does not occur, then the functionalis layer is again shed during menstruation and the cycle repeats. Of note, both shedding and reepithelialization appear to take place simultaneously [1].

Uterine glands and their secretions provide essential nutrients for fetoplacental growth during pregnancy [7], particularly during the first trimester, as well as influence stromal cell decidualization and facilitate embryo implantation [4]. Mice that lack glands in their uterus are infertile and experience recurrent pregnancy loss due to defects in embryo implantation and stromal cell decidualization [8–10]. Uterine gland secretions are perturbed in women with clinical infertility during both the proliferative and secretory phases [11–15]. Stromal cell decidualization is critical for pregnancy as decidual cells regulate placenta development and the maternal immune system [16]. Disrupted decidualization is associated with reproductive pathologies such as endometriosis and recurrent pregnancy loss [17–23], implantation failure [24, 25], and development of preeclampsia [26]. It is difficult to impossible to study the basic aspects of endometrial regeneration and differentiation as well as progression of reproductive pathologies in vivo in women due to ethical and practical limitations. While animal models have provided some important insights, common laboratory rodent models do not completely recapitulate the biology of the human endometrium. A key difference is the absence of spontaneous cyclic stromal cell decidualization in mice and rats. Therefore, in vitro models of the human endometrium are required to understand its biology and function.

Over the past few decades, in vitro studies of human endometrium have utilized two-dimensional (2D) cell culture methods involving growth and analysis of individual endometrial cell types in plastic dishes that are either untreated or perhaps coated with a layer of ECM such as collagen. Of note, those 2D cultures of the endometrial epithelium are insufficient models and often not hormonally responsive. Instead, 3D cultures offer more appropriate physiological response to hormones and gene expression [27, 28]. Most 2D cell culture systems lack ECM, which leads to reduced mechanical and biochemical cues for epithelial cells to polarize and exhibit proper apical and basolateral surfaces and tight junctions, and thus, vectorial secretion to communicate with other cell types. Further, the ECM in the human endometrium changes across the menstrual cycle [29, 30]. The ECM is important, as it has an important instructive role in epithelial differentiation and polarity [31, 32].

Since the 1980s, 3D cell culture models for the human endometrium were developed and refined to investigate epithelial function, stromal cell decidualization, and embryo–endometrium interactions (Table 1 and Figure 1). Building on a substantial increase in the 3D spheroid and organoid research for other organs and tissues, the first endometrial epithelial organoids (EEOs) were reported in 2017 [33, 34].

Table 1.

Summary of key 3D human endometrial culture studies.

Cell typesa	Matrix	Model design	Reference
Primary endometrial cells	Collagen	Spheroids formed from endometrial fragments and embedded in collagen	[42]
Primary EEC	Matrigel	Gland fragments suspended in Matrigel were seeded on top of hardened Matrigel	[43]
Primary ESC and EEC	Collagen I Matrigel	EEC rest on Matrigel with stromal cells embedded in collagen	[46]
Immortalized endometrial stromal cell lines (HESC)	Collagen I	HESC embedded in collagen	[54]
Primary ESC and EEC	Collagen Matrigel	EEC rest on Matrigel with stromal cells embedded in collagen	[47]
Primary ESC and EEC	Collagen I	Endometrial cells seeded in collagen	[55]
Primary ESC and EEC	Matrigel	(1) Epithelial cells grown on Matrigel-coated insert; (2) epithelial cells grown on Matrigel-coated insert, with stromal cells grown on plastic below; and (3) epithelial cells grown on top of stromal cells embedded in Matrigel	[56]
Primary ESC and EEC	Collagen Matrigel	Epithelial cells grown on Matrigel-coated insert, and stromal cells grown in collagen in basal compartment	[59]
Primary ESC and EEC; HEC1A and Ishikawa; and HESC	Fibrin-agarose	(1) Primary ESC mixed with fibrin-agarose in insert, with primary EEC seeded on top; (2) HESC mixed with fibrin-agarose in insert. After 24 h, HEC1A or Ishikawa cells are seeded on top	[60]
Ishikawa, RL95-2, KLE, and EN-1078D	Reconstituted basement membrane (rBM)	Cell lines mixed with rBM and overlaid onto already polymerized rBM.	[61]
HEC1A, Ishikawa, and RL95-2	Matrigel	Matrigel droplets containing cells were seeded onto plate. Cell culture medium containing cells of the same cell lines were added, and cells spread onto the Matrigel droplets and plated	[62]
Primary EEC	Matrigel	Primary EEC seeded in Matrigel droplets and cultured in a defined, WNT-activating medium	[33]
Primary EEC	Matrigel	Primary EEC seeded in Matrigel droplets and cultured in a defined, WNT-activating medium	[34]

^aDefinitions: EEC, primary endometrial epithelial cells; ESC, primary endometrial stromal cells.

Figure 1.

The 3D cell culture models of the endometrium and pregnancy. Earlier models suspended endometrial stromal cells in collagen, Matrigel or fibrin-agarose gels, and overlaid epithelial cells, forming a lumenal-like epithelial monolayer. Endometrial epithelial organoids have been developed that can be passaged and cryopreserved, and they also exhibit hormonal responses similar to the in vivo environment. Scaffold-free models allow cells to self-organize, where compact endometrial stromal cells are surrounded by a layer of epithelial cells. Future 3D models will incorporate matrices, like those of the polyethylene glycol (PEG) hydrogel and other alternative matrices that can be manipulated to reflect the in vivo endometrial ECM and incorporate multiple endometrial cell types.

Historical perspectives

Understanding the complex regeneration and differentiation of the human endometrium during the menstrual cycle, as well as the interactions between different endometrial cell types and the implanting blastocyst, has long been a challenge for reproductive biologists. Prior to the late 1980s, most cultures of human endometrial cells were performed in 2D [35–39]. These studies found a reduced or lack of response to ovarian steroid hormones when primary endometrial epithelial cells were isolated and cultured 2D on plastic. Inspired by the studies of the mammary and thyroid glands [40, 41], a number of pioneering studies established that 3D culture of human endometrial epithelial cells is critical to maintain hormone responsiveness and differentiation. Although serum was needed for cell attachment to occur in monolayer cultures, it diminished the hormone response of cells [42]. Kirk and Alvarez [42] found that nonenzymatic digested endometrial tissue fragments floated in culture medium and formed hollow epithelial spheroids that detached from the tissue and remained free floating. The external apical surface of the vesicles displayed microvilli, the internal basal surface presented a basal lamina associated with collagen fibers, and the nuclei were basally located nuclei indicative of cell polarity [42]. When surrounded in rat tail collagen, the spheroids formed gland-like structures. Similar to recent 3D organoids [33, 34], the spheroids contained a polarized epithelium wherein the apical surface formed a lumen. Although some spheroids degraded the collagen and the epithelial cells formed a 2D monolayer after 10 or so days, the unattached floating spheroids could be maintained for a longer period.

Another early in vitro model of human endometrial glands was constructed by separating glands from endometrial tissue and surrounding them with Matrigel [43]. Matrigel is a solubilized basement membrane preparation extracted from the Engelbreth–Holm–Swarm mouse sarcoma, a tumor rich in ECM proteins, including laminin (a major component), collagen IV, heparan sulfate proteoglycans, and entactin/nidogen, and a number of growth factors [44]. The gland–Matrigel mix was seeded into culture wells coated with Matrigel. Over a period of weeks, the gland fragments formed small colonies and then spheroids. The generated spheroids or organoids contained columnar-type epithelium that was polarized with microvilli and tight junctions at the apical surface. The isolated, fragmented glands could be propagated in culture and maintained long term while displaying similarities in architecture and behavior to glands in vivo. The cells were cultured in media supplemented with hydrocortisone, prostaglandin E2 (PGE2), and estrogen (E2) to mimic the proliferative phase, but responsiveness of these cells in culture to other steroid hormones was not determined.

A representative model of the human endometrium needs to contain both endometrial stromal and epithelial cells, as stromal–epithelial communication is important for epithelial morphogenesis and function [45]. However, deducing the correct matrix conditions for both cell types to grow and function has been challenging. Collagen I is present throughout the endometrial stroma [30], but it does not maintain the 3D structures of epithelial glands in the long-term culture [42]. In contrast, Matrigel promoted continuous epithelial growth and organoid formation [43]. The chief components of Matrigel are structural ECM proteins that present cultured cells with adhesive peptide sequences as well as growth factors, such as transforming growth factor beta (TGFB) and epidermal growth factor (EGF), which prevent differentiation and/or promote proliferation of many cell types including stem cells. Of note, the exact composition of Matrigel varies from lot to lot.

Another study combined isolated stroma and epithelia cells with collagen I and Matrigel [46]. Stromal cells were first embedded in collagen I, and Matrigel was overlaid onto which gland fragments were seeded [46]. The result was the development of polarized epithelial cells with cilia and some apparent gland formation. Using this developed model [46], the effect of the antiprogestin, mifepristone or RU486, and the emergency contraceptive, levonorgestrel, on endometrial receptivity and blastocyst attachment were investigated [47, 48]. Human blastocyst attachment was inhibited by mifepristone [47] in concert with the modulation of endometrial receptivity markers [48]. These studies were able to apply an established model as well as incorporate human blastocysts, highlighting the capabilities of in vitro models. As such, 3D models of the endometrium are useful as a drug discovery tool including the development and testing of contraceptives.

Unfortunately, the previous models [46] were limited with respect to culture length, because the gel shrank over time, likely due to the release of matrix degrading proteolytic enzymes from the stromal and/or epithelial cells. This highlights a major difficulty in the long-term coculturing of stromal and epithelial cells, as matrix metalloproteinases (MMPs) and serine proteases are released by the cells that remodel the ECM during the menstrual cycle [49–51]. Indeed, MMPs 1, 2, 3, and 9 are released from the human endometrial cells and can be regulated by both hormones and locally produced factors such as interleukin 1 alpha (IL1A) and tumor necrosis factor alpha (TNFA) [49, 50, 52, 53]. Schutte and Taylor [54] mixed an immortalized human endometrial stromal cell (HESC) line with collagen I. Following gelation, the cells were treated with different hormones to reflect the early and late secretory phases, decidualization, and hormone withdrawal to mimic menstruation. Stromal cells in this 3D structure decidualized, but within 24 h of hormone withdrawal, there was degradation and deterioration of the collagen matrix and tissue construct. Furthermore, collagenase and MMP2 activity increased as normally occurs during menstruation.

Another method to investigate the relationship between stromal cells and glands was developed in which epithelial and stromal cells were seeded together in a collagen I gel, or epithelial cells in a collagen I gel were treated with stromal cell-conditioned medium [55]. In both approaches, the epithelial cells formed gland-like spheroids. Interestingly, the epithelial cells were able to form spheroids after first being cultured in 2D on plastic. The medium used to culture cells in 2D contained EGF, thereby establishing potential beneficial effects of stromal cell-derived factors on the epithelium. Although not addressed in that study, the stability of the collagen gel was likely due its formulation and its inherent rigidity, which may have prevented the cells from compete matrix degradation.

Another variation on in vitro endometrial cultures involved the use of a coculture model wherein epithelial cells are grown on Matrigel-coated cell culture inserts with underlying stromal cells on plastic, within the Matrigel, or within the insert in Matrigel placed underneath the layer of epithelial cells [56] (Figure 1). These studies found that stromal cells had a paracrine effect on the epithelial cells and controlled their growth and differentiation. The Matrigel provided a basement membrane in this model; however, this model does not account for the gradient effects of stromal cells deeper in the endometrium that are not as close to the basement membrane of the epithelia. In a similar model from the same group, stromal cells were cultured with the Ishikawa endometrial epithelial adenocarcinoma cell line [57]. These findings suggested that stromal cell-derived factors could influence the differentiation of endometrial cancer cells to resemble a more functionally normal epithelium.

While stromal cells regulate epithelial growth and differentiation, it is also evident that epithelial cells influence stromal cells. One coculture model [58] was utilized wherein collagen-embedded stromal cells were placed in the bottom compartment and epithelial cells were seeded on top of Matrigel in a cell culture insert to examine the influence epithelial cells on stromal cell decidualization [59]. Results from that model suggested that P4 indirectly impacts stromal cell decidualization via epithelial cells. Wang et al. [60] developed a novel model using both primary cells and cell lines to investigate early human implantation. Primary endometrial stromal cells were suspended in a fibrin-agarose gel, and epithelial cells were seeded onto the solidified gel containing these stromal cells. In this model, the primary epithelial cells formed a polarized monolayer on top of the gel. Interestingly, epithelial cells that contaminated the stromal cell fraction formed gland-like spheroids in the gel. In a separate culture system by this same group, immortalized HESCs were suspended in the fibrin-agarose matrix, allowed to solidify and incubated for 24 h, after which endometrial epithelial adenocarcinoma cell lines, HEC1A or Ishikawa, were seeded above and cocultured for 2–3 days [60]. Spheroids made from the human placental choriocarcinoma JAR cell line were seeded on top of the coculture system. The HEC1A cells became columnar and were partly polarized, whereas Ishikawa cells did not polarize. The JAR spheroids displayed greater attachment to the Ishikawa cells compared to the HEC1A cells, and attachment increased when HESCs were present.

Another study compared the growth and morphology of different endometrial cancer lines in 2D and 3D [61]. The cell lines Ishikawa, RL95-2, KLE, and EN-1078D were used based on varying degrees of inherent differentiation. The cells were mixed with a recombinant basement membrane and overlaid onto an already polymerized recombinant basement membrane. The well-differentiated Ishikawa and moderately differentiated RL95-2 formed gland-like structures, whereas the poorly differentiated KLE and EN-1078D formed spheroids. Ishikawa cells formed structures with a central lumen and displayed apicobasal polarity. Of note, the endometrial cancer cell lines proliferated more slowly in 3D compared to the 2D cell culture. Similarly, the differences in these cell lines were utilized to examine embryo implantation in which the three endometrial carcinoma cell lines (HEC1A, RL95-2, and Ishikawa) were separately mixed with Matrigel and plated as droplets [62]. After these droplets solidified, cells of the same cell lines were added and then cultured for 4 days. Following this, AC-IM88 trophoblast cells were seeded into the well and the coculture was maintained for a further 5 days. Cultures with endometrial epithelial cancer cell lines that were less polarized and less differentiated, such as RL95-2, were more strongly invaded by AC-IM88 cells [62]. This 3D model highlighted how the changes in endometrial epithelial cell polarity and differentiation may influence trophoblast invasion, but the models relied on cancer cells rather than the normal noncancerous cells.

These studies highlight the challenge in developing a 3D model of the endometrium, most notably the difficulty in expanding and the long-term culture of primary epithelial cells. Further, the primary endometrial epithelial and stromal cells degrade the commonly used collagen matrix at a faster rate than the cells are able to form their own matrix [42, 46]. Further, the effects of culture conditions on steroid hormone responsiveness were not determined [43], and the cells were not able to be passaged. It is evident that factors released from the primary epithelial cells, epithelial cell lines, and stromal cells have a reciprocal effect on the other cell type [55–57] and, as such, suggests that a defined medium in which the cells are grown may improve their growth and behavior. Although Matrigel is not ideal, it has the most promising influence on in vitro epithelial cell growth and is a major component of recently developed 3D organoid cultures.

Endometrial epithelial organoids

A breakthrough in development of functional endometrial epithelial cell cultures occurred in 2017. Two groups used knowledge from stem cell-derived organoid cultures of other organs to develop 3D EEOs from the human and mouse endometrium [33, 34] (Figure 2A). In this approach, human endometrial epithelial cells were isolated from endometrial biopsies or term decidua, seeded in Matrigel, and cultured in a chemically defined medium. The medium included a collection of growth factors used for long-term culture and expansion of organoids from other human organs, such as the intestine, including the wingless-type MMTV integration site family, member (WNT) activator R-spondin 1 (RSPO1), epithelial cell mitogens EGF, and fibroblast growth factor 10 (FGF10), differentiation inhibitor Noggin, and TGFB antagonist A83-01 [33, 34] (Table 2). The EEOs could be grown from single cells, expanded long term, and could be passaged, cryopreserved, thawed, and regrown [63]. The 3D EEO system overcame the limited expansion of primary endometrial epithelial cells in the 2D culture. Establishment and preservation of EEOs from multiple donors allows for biobanking and comparative studies, which was previously impossible using 2D culture.

Figure 2.

Human endometrial epithelial organoids (EEOs) characteristics. (A) Organoids form in Matrigel under WNT-activating conditions and express FOXA2, a marker of the GE. (Scale bars: left: 500 μm, right: 75 μm.). (B) Hormone-responsive genes in EEOs are determined by bulk RNA-seq analysis following treatment with E2 or E2 and MPA treatment [61]. (C) Single-cell RNA-seq analysis discovered that human EEOs are composed of many different cell types that respond to hormone treatment [61].

Table 2.

Human endometrial epithelial organoid media culture components.

Component	Function	Reference(s)
Estradiol-17β	Promotes growth and expansion of endometrial cells	[34]
A83-01	Alk3/4/5 chemical inhibitor that blocks TGFB pathway to inhibit cell differentiation	[32, 34, 63, 64]
CHIR99021	Glycogen synthase kinase 3 beta (GSK3B) inhibitor	[64]
EGF	Growth factor mitogen	[33, 34, 63, 64]
FGF10	Growth factor mitogen	[33, 34, 63]
HGF	Growth factor mitogen	[32, 63]
N2 supplement	Contains insulin, transferrin, progesterone, putrescine, selenium. Promotes cell differentiation and survival	[33, 34, 63, 64]
N-acetyl-L-cysteine	Antioxidant that prevents apoptosis	[33, 34, 63]
Nicotinamide	Amide derivative of vitamin B3 used in the differentiation of embryonic stem and induced pluripotent stem cells	[33, 34, 63]
Noggin	Secreted protein that binds and inactivates members of the TGFB superfamily, such as BMP4, to inhibit cell differentiation	[33, 34, 63, 64]
PGE2	Prostaglandin that signals via cAMP	[64]
RSPO1a	Activator of the canonical WNT signaling pathway by acting as a ligand for LGR4-6 receptors that promotes cell proliferation and inhibits TGFB signaling	[33, 34, 63, 64]
SB202190	Chemical inhibitor of P38 MAPK that promotes cell growth	[34]
WNT3Aa,b	Secreted protein that activates WNT signaling pathways and mediates self-renewal of the stem cells	[34]
Y-27632c	Chemical inhibitor of ROCK inhibitor that enhances cell survival and renewal	[33, 34, 63]

^aDepending on the experiment, obtained from either cell lines or in recombinant form.

^bNot needed for further expansion and passaging of human EEOs.

^cOnly needed for either initial organoid formation following seeding or thawing.

The EEOs generated using that approach with Matrigel contain a mixture of epithelial cell types, including proliferative, ciliated, unciliated, stem, and secretory [63, 64] (Figure 2C). They demonstrated clonal capacity and long-term expandability of the organoids [33, 34], suggesting that they form from epithelial stem/progenitor cells. While some stem cell markers have been proposed, such as cadherin 1 or E-cadherin (CDH1), stage-specific embryonic antigen-1 (SSEA-1), and leucine repeat containing G protein-coupled receptor 5 (LGR5) [65–67], the development of the endometrial organoid field presents an opportunity to functionally test these markers. For example, EEOs can be generated from SSEA-1-negative cells [33]. Thus, EEOs provide a promising avenue for the further investigation of endometrial epithelial stem/progenitor cells. Of note, the EEOs are heterogeneous and contain a mixture of LE and GE cells, based on the expression of forkhead box A2 (FOXA2), a gland-specific marker [63, 68], as well as their subtypes including ciliated and non-ciliated as well as secretory cells (Figure 2C). This observation is not unexpected, as those different major endometrial epithelial cell types appear to primarily originate from the same precursor both during fetal and neonatal development as well as during the proliferative phase of the menstrual cycle and postpartum regeneration [4, 69].

Remodeling of the human endometrium is dynamic and predominantly controlled by the ovarian steroid hormones, E2 and P4. For the EEOs to display this same dynamic nature and recapitulate the proliferation, differentiation, and secretory nature of endometrial epithelia, they must be responsive to those hormones. Estrogen treatment of EEOs, designed to mimic the proliferative phase of the menstrual cycle, increased expression of the established E2-responsive genes including progesterone receptor (PGR), estrogen receptor alpha (ESR1), and olfactomedin 4 (OLFM4) [33, 34, 63] (Figure 2B). Ciliogenesis and the number of ciliated endometrial epithelial cells are the greatest during the proliferative phase of the menstrual cycle and then decrease during the secretory phase as P4 levels rise [70, 71]. As such, E2 treatment of EEOs increased the number of ciliated cells in the EEOs [33, 34, 63, 64]. Thus, EEOs are useful to understand how E2 impacts the endometrial epithelial development and function.

In order to mimic the secretory phase of the cycle, EEOs were developed and then treated with P4 or medroxyprogesterone acetate (MPA), a nonmetabolizable progestin [33, 34, 63]. A hallmark of epithelial secretory transformation is P4-induced loss of PGR expression, which occurred in the EEOs treated with P4 or MPA for at least 6 days [33, 34, 63]. Treatment of EEOs with MPA in the presence of E2 increased the expression of known P4-responsive genes, including hydroxysteroid 17-beta dehydrogenase 2 (HSD17B2), leukemia inhibitory factor (LIF), progestogen-associated endometrial protein (PAEP), and secreted phosphoprotein 1 (SPP1) (Figure 2B). SPP1 is localized to endometrial epithelial cells and increased during the secretory phase [72], and PAEP is secreted by the glands during the secretory phase [73, 74]. Similarly, LIF is found in the secretory phase glands and it is decreased in infertile women [75, 76]. Single cell analysis of EEOs treated with MPA, E2, and/or cyclic adenosine monophosphate (cAMP) decreased the number of stem cells but increased the number of secretory and epithelial cells, highlighting the potential contribution of these hormones and molecules to the differentiation of putative stem/progenitor cells within the EEOs [63]. Similar to effects on decidualizing stromal cells, cAMP treatment of EEOs along with E2 and MPA substantially increased their secretory transformation based on PAEP and SPP1 expression [33, 63]. As such, EEOs are able to respond appropriately to ovarian steroid hormones based on the analysis of cell types in the EEOs as well as hormone-responsive gene expression. Future studies are needed to refine culture techniques to ensure appropriate responsiveness of the EEOs to other ovarian hormones.

Understanding patient donor effects and reproductive pathologies by using EEOs

In vitro models of the human endometrium are expected to provide useful information regarding patient-specific responses to therapeutic treatments for infertility or reproductive pathologies, such as endometriosis and endometrial cancer [34, 77]. While less investigated in endometrial epithelial cells, organoids derived from other human tissues retain a “memory” of the donor from which they came, particularly those with a pathology. With respect to the uterus, stromal cells from the endometrium of women with recurrent pregnancy loss have reduced decidualization capacity [78]. Similarly, in vitro decidualization of stromal cells is impaired in women with a previous pregnancy complicated by severe preeclampsia [26]. Gene expression is disrupted in stromal cells cultured from women with endometriosis compared to controls [19]. Furthermore, EEOs from two different donors without endometrial pathologies exhibited differences in gene expression responses to ovarian steroid hormones [63]. Therefore, patient-derived EEOs may represent the variability that is present in the human population and allow for further investigations in particular reproductive pathologies in vitro.

Recently, EEOs were developed from ectopic endometriosis samples and were found to form a lower number of organoids than those derived from healthy endometrium [77]. The endometriosis-derived organoids were clonal and could be formed from ectopic endometriosis lesion implantation sites of all endometriosis clinical stages (I–IV). Matched eutopic and ectopic endometrium from patients were also used to generate EEOs. Ectopic endometrial organoids were epithelial in origin and expressed the epithelial cell markers cytokeratin 7 (KRT7) and CDH1 as well as ESR1 and PGR hormone receptors. Like organoids derived from healthy endometrium, organoids from endometriosis samples contained polarized, secretory epithelial cells with cilia. Interestingly, organoids from endometriotic lesions had altered signaling pathways such as PI3K-AKT and WNT compared to those organoids from healthy tissue. Moreover, severe stage IV endometriosis-derived organoids possessed endometrial cancer-associated mutations, indicating an involvement of cancer driver genes in the progression of endometriosis. This study highlighted that diseased endometrial tissue retains its donor origin phenotype when cultured as organoids, presenting a promising avenue for understanding endometriosis.

Endometrial cancer is a common gynecologic cancer that is typically treated by hysterectomy, highlighting the need for better treatments and therapeutic approaches that avoid removal of an entire organ [79]. A number of studies have derived organoids from endometrial cancer [33, 77, 80–82]. Turco et al. [33] derived organoids from endometrial tumors and adjacent noncancerous endometrial tissue. The tumor organoids resembled the primary tumor for a FIGO grade I endometrioid carcinoma [33]. The cancer organoids were SRY-box transcription factor 17 (SOX17)-positive, suggesting that they were of glandular origin and displayed nuclear pleomorphism, a disorganized epithelium, and an irregular basement membrane [33]. A more comprehensive analysis of endometrial cancer organoids was recently reported wherein organoids were established from different grades and stages of endometrial cancer [77]. The efficiency of endometrial cancer organoid formation was much less than for organoids derived from other endometrial conditions, and they displayed limited ability to expand in culture. In some of the cancer-derived organoid lines, noncancerous cells overwhelmed the culture. In fact, the medium used for normal EEOs had to be modified to enhance the establishment and propagation of cancer organoids [77]. As previously observed, the cancer organoids recapitulated the immunohistology of the primary tumor [33]. Somewhat predictably, endometrial cancer organoids derived from lower grade or stage primary tumors demonstrated more gland-like features with a more defined lumen than higher grade cancer organoids [77]. Certainly, endometrial cancer-derived organoids can be used to test potential therapeutics for endometrial cancer as well as patient-specific drug sensitivity [80–82].

Organoids have also been derived from potential precancerous pathologies, such as three subtypes of hyperplastic endometrium: simple benign, complex atypical, and polyp [77]. Organoids derived from these subtypes of hyperplastic endometrium mimicked the in vivo disease phenotype and genetics. Although once considered a sterile environment, emerging data established that the endometrium has a microbiome [83, 84]. Thus, EEOs offer an opportunity to study epithelial–microbiota interactions [85]. In one model, the endometrial HEC1A cancer cell line was attached to collagen-coated beads and formed a single layer of cells which had microvilli and expressed mucins but were non-ciliated. The 3D organoids were challenged with microbial infection by three species, Lactobacillus crispatus, Gardnerella vaginalis, and Neisseria gonorrhoeae, and responded appropriately with the production of pro-inflammatory cytokines and chemokines. Additionally, microbes were able to colonize the cells and, following infection with gonococci, produced pro-inflammatory cytokines and chemokines [85]. This 3D model recapitulated in vivo the host–microbe interactions that were not previously observed in the 2D cultures of the human endometrium. Finally, scaffold-free 3D aggregates of endometrial stromal and epithelial cells have recently been used to investigate the effects of the increased androgens observed in polycystic ovarian syndrome [86]. These recent organoid and endometrial culture studies highlight the utility of 3D cell culture engineering approaches to advance our understanding of endometrial biology and pathology that impact women’s reproductive health.

The future

Understanding the interactions between the epithelial and stromal cells within the endometrium is of great importance and is essential to understand endometrial biology and develop strategies to treat reproductive pathologies and female infertility. Most current established models of the endometrium utilize only one cell type and use Matrigel, which unfortunately can exhibit substantial batch variation and is likely not homologous to the ECM within the endometrium that separates epithelial and stromal cells. Future endometrial 3D culture models may consider refining the EEO expansion and culture medium to accurately reflect the proliferative, secretory, and menstrual cell phenotypes. The current, most commonly used medium includes factors that maintain the epithelial or “stem-like” nature of the cells (Table 2) that may only be necessary when mimicking the earlier regenerative, proliferative phase of the menstrual cycle. Removal of these factors to induce clear differentiation of the EEOs or glands as it occurs during the secretory phase in response to P4 may be required to truly represent this phase. In coculture models where stromal cells are present, the growth factors, FGF10 and hepatocyte growth factor (HGF), could also be potentially removed.

To maintain the histoarchitecture and relationships between the different cell types of the endometrium as well as the native ECM, slices of full thickness endometrial tissue were cultured in a collagen gel [87]. In response to E2 and P4, there was an expected decrease in Ki67 cell proliferation marker expression as well as ESR1 and PGR, reflecting normal hormonal responses in utero. Interestingly, E2 and P4 treatment elicited changes in the cellular architecture of the tissue slices similar to those seen in the secretory phase of the menstrual cycle, including the appearance of decidualized endometrial stromal cells. The advantage of such a model is that the full thickness tissue could allow for the examination of the basalis and functionalis layers of the endometrium. While culturing whole tissue maintains the histoarchitecture and natural ECM produced by cells, the cells cannot be passaged and used for multiple experiments. As such, the development of a matrix or scaffold suitable for the needs of both the endometrial stromal and epithelial cells to enable functional studies is essential.

Abbas et al. [88] incorporated both endometrial epithelial and stromal cells into a 3D porous scaffold. In this method, a type I collagen aqueous slurry, derived from bovine dermal collagen, was frozen and then sublimed to remove ice crystals, leaving behind a porous structure. This collagen scaffold was then placed into media containing stromal cells, continuously rotated to allow for cell adherence, and then transferred to cell culture plates. After 2 days, EEO fragments were seeded onto the scaffolds. The stromal cells were able to proliferate and cluster within the scaffold pockets and, importantly, deposited their own ECM. The EEOs were first formed in Matrigel and then mechanically disrupted into fragments for seeding onto the 3D collagen scaffold. In this system, the EEOs formed a polarized surface layer that appeared similar to LE in vivo. The hormone responsiveness of the model was only examined in cultures of EEOs alone or stromal cells alone and, as such, the interaction of the two cell types under hormonal stimulation could not be observed [88].

Modifiable PEG hydrogels with ECM-binding and integrin-binding peptides provide a promising avenue to replace Matrigel in 3D cultures of endometrial cells [89, 90]. Endometrial stromal cells are able to differentiate and decidualize when encapsulated in PEG hydrogels in response to hormonal stimulation with MPA and cAMP, displaying a change in the morphology and secretion of the decidual markers, insulin-like growth factor binding protein 1 (IGFBP1) and prolactin (PRL) [90]. In cocultures, stromal cells were encapsulated within the hydrogel, and the epithelial cells that were seeded on top formed a monolayer. Interestingly, stromal cells appeared to migrate upward toward the epithelial monolayer and exhibited more elongation when closer to the monolayer at certain matrix densities. Modulation of hydrogel stiffness resulted in changes in cell behavior, highlighting that the matrix conditions have an integral role in developing a representative model of the basalis and functionalis layers of the endometrium. Furthermore, secreted proteins could be measured and changes in cytokines and growth factor levels were seen in response to decidualization hormones [90].

Allowing epithelial and stromal cells to self-assemble in a scaffold-free environment presents another method used to investigate intercellular interactions. Recently, endometrial epithelial and stromal cells were combined in agarose molds and then aggregated to form organoids with a center containing stromal cells and an outer layer of epithelial cells [86]. Hormone treatment of these cells performed to mimic the proliferative phase of the menstrual cycle resulted in the outer layer of cells secreting collagen and mucins as well as expressing appropriate sex hormone receptors and Ki67 proliferation marker. Taken together, this clearly suggested that the cells self-assembled so that the stromal cells were situated in the center, while the epithelial cells formed the outer layer. Unlike current 3D organoid models [33, 34, 63], the apical side of the epithelial cells in this model were facing the culture medium and the basal side faced the stromal cells. As such, the polarity of the epithelial cells in this model may not reflect that of endometrial epithelial cells in vivo [86].

Development of EEOs provides a platform to use CRISPR/Cas9 genome editing to better understand the biological roles of genes in endometrial epithelial cell function. The consequential effect of the absence of particular genes and therefore the presence or absence of secreted proteins on stromal cell differentiation or embryo implantation could be evaluated using a gene editing approach. CRISPR/Cas9 genome editing technology has been used on multiple different types of organoids derived from different tissue types [91]. CRISPR has been used in organoids, particularly those involving disease-causing mutation; for example, mutations in the gene coding for cystic fibrosis could be repaired in organoids derived from cystic fibrosis [92]. Furthermore, this approach could be used to investigate the role of specific gene mutations in endometrial epithelial pathologies [93].

To examine early implantation in humans, a human trophoblast stem cell (TSC) line [94] that has characteristics of trophectodermal cells [95] was employed. These TSCs formed compact spheroids when seeded into round-bottomed wells with methylcellulose. To mimic early embryo attachment to the uterine LE in vitro, TSC spheroids were seeded onto monolayers of ECC1 and Ishikawa endometrial epithelial cancer cell lines as well as primary epithelial cells from fertile and infertile women. In this system, adhesion of TSC spheroids was reduced in epithelial cells from infertile women. This model highlights a novel way to examine embryo attachment to cells in the endometrium of infertile women that may be altered or dysregulated.

A critical need to understand defective placental development early in pregnancy and its downstream implications led to the recent development of trophoblast organoids [96, 97]. In one study, villous cytotrophoblasts (CTBs) were isolated from pooled first trimester placentas and seeded in Matrigel with a defined medium that promotes the stem-like features of organoids. The cells initially formed small clusters that developed into irregularly shaped organoids. These organoids could be passaged, cryopreserved, and thawed. The CTB organoids could be differentiated and used to model placental development. Further, Turco et al. [97] developed trophoblast organoids that are able to differentiate into syncytiotrophoblast and extravillous trophoblast. Villous-like structures could be observed and the organoids secreted placenta-specific factors [97]. Although like many 3D endometrial models, the villi were inside-out and secretion was directed to the center of the structure. These pioneering models of trophoblasts are a potential component in a coculture model of early pregnancy that could be used to examine the elusive early pregnancy events in trophoblast-endometrial interactions critical for pregnancy establishment.

Conclusions

The human endometrium experiences significant changes during the menstrual cycle and pregnancy as well as during the development of gynecological pathologies such as endometriosis and endometrial cancer. Disruption of normal endometrial development and differentiation creates an unsuitable uterine environment with negative effects on fertility; they also likely cause later pregnancy complications if pregnancy is established. Obstacles to examining endometrial function and early pregnancy in humans in vivo have led to the development of in vitro models. The first 2D cell monolayer cultures progressed to primary endometrial cells and endometrial cell lines cultured in various ECM to the current wave of the transformative EEOs. Breakthroughs in organoid culture systems enable, for the first time, endometrial epithelial cells to be greatly expanded and biobanked, while displaying the hormone-responsive signatures seen in vivo, something that was previously greatly challenging. Future research to adapt these models will involve developing representative matrices of the endometrium and incorporating multiple endometrial cell types as well as trophoblast cultures to mimic early pregnancy. In summary, recent advances represent an exciting period for endometrial biology, and combining established knowledge with advances in bioengineering is crucial to understand the complexity of uterine function and female fertility.

Conflict of interest

The author declares there is no conflict of interest.